linux/tools/perf/util/annotate-data.c

1829 lines
46 KiB
C
Raw Permalink Normal View History

/* SPDX-License-Identifier: GPL-2.0 */
/*
* Convert sample address to data type using DWARF debug info.
*
* Written by Namhyung Kim <namhyung@kernel.org>
*/
#include <stdio.h>
#include <stdlib.h>
#include <inttypes.h>
#include <linux/zalloc.h>
#include "annotate.h"
#include "annotate-data.h"
#include "debuginfo.h"
#include "debug.h"
#include "dso.h"
#include "dwarf-regs.h"
#include "evsel.h"
#include "evlist.h"
#include "map.h"
#include "map_symbol.h"
#include "sort.h"
#include "strbuf.h"
#include "symbol.h"
perf annotate: Add --data-type option Support data type annotation with new --data-type option. It internally uses type sort key to collect sample histogram for the type and display every members like below. $ perf annotate --data-type ... Annotate type: 'struct cfs_rq' in [kernel.kallsyms] (13 samples): ============================================================================ samples offset size field 13 0 640 struct cfs_rq { 2 0 16 struct load_weight load { 2 0 8 unsigned long weight; 0 8 4 u32 inv_weight; }; 0 16 8 unsigned long runnable_weight; 0 24 4 unsigned int nr_running; 1 28 4 unsigned int h_nr_running; ... For simplicity it prints the number of samples per field for now. But it should be easy to show the overhead percentage instead. The number at the outer struct is a sum of the numbers of the inner members. For example, struct cfs_rq got total 13 samples, and 2 came from the load (struct load_weight) and 1 from h_nr_running. Similarly, the struct load_weight got total 2 samples and they all came from the weight field. I've added two new flags in the symbol_conf for this. The annotate_data_member is to get the members of the type. This is also needed for perf report with typeoff sort key. The annotate_data_sample is to update sample stats for each offset and used only in annotate. Currently it only support stdio output mode, TUI support can be added later. Committer testing: With the perf.data from the previous csets, a very simple, short duration one: # perf annotate --data-type Annotate type: 'struct list_head' in [kernel.kallsyms] (1 samples): ============================================================================ samples offset size field 1 0 16 struct list_head { 0 0 8 struct list_head* next; 1 8 8 struct list_head* prev; }; Annotate type: 'char' in [kernel.kallsyms] (1 samples): ============================================================================ samples offset size field 1 0 1 char ; # Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Cc: linux-toolchains@vger.kernel.org Cc: linux-trace-devel@vger.kernel.org Link: https://lore.kernel.org/r/20231213001323.718046-15-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2023-12-12 16:13:20 -08:00
#include "symbol_conf.h"
#include "thread.h"
perf annotate-data: Handle RSP if it's not the FB register In some cases, the stack pointer on x86 (rsp = reg7) is used to point variables on stack but it's not the frame base register. Then it should handle the register like normal registers (IOW not to access the other stack variables using offset calculation) but it should not assume it would have a pointer. Before: ----------------------------------------------------------- find data type for 0x7c(reg7) at tcp_getsockopt+0xb62 CU for net/ipv4/tcp.c (die:0x7b5f516) frame base: cfa=0 fbreg=6 no pointer or no type check variable "zc" failed (die: 0x7b9580a) variable location: base=reg7, offset=0x40 type='struct tcp_zerocopy_receive' size=0x40 (die:0x7b947f4) After: ----------------------------------------------------------- find data type for 0x7c(reg7) at tcp_getsockopt+0xb62 CU for net/ipv4/tcp.c (die:0x7b5f516) frame base: cfa=0 fbreg=6 found "zc" in scope=3/3 (die: 0x7b957fc) type_offset=0x3c variable location: base=reg7, offset=0x40 type='struct tcp_zerocopy_receive' size=0x40 (die:0x7b947f4) Note that the type-offset was properly calculated to 0x3c as the variable starts at 0x40. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240412183310.2518474-5-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-04-12 11:33:10 -07:00
/* register number of the stack pointer */
#define X86_REG_SP 7
static void delete_var_types(struct die_var_type *var_types);
perf annotate-data: Add debug messages Add a new debug option "type-profile" to enable the detailed info during the type analysis especially for instruction tracking. You can use this before the command name like 'report' or 'annotate'. $ perf --debug type-profile annotate --data-type Committer testing: First get some memory events: $ perf mem record ls Then, without data-type profiling debug: $ perf annotate --data-type | head Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ And with only data-type profiling: $ perf --debug type-profile annotate --data-type | head ----------------------------------------------------------- find_data_type_die [1e67] for reg13873052 (PC) offset=0x150e2 in dl_main CU die offset: 0x29cd3 found PC-rel by addr=0x34020 offset=0x20 ----------------------------------------------------------- find_data_type_die [2e] for reg12 offset=0 in __GI___readdir64 CU die offset: 0x137a45 frame base: cfa=1 fbreg=-1 found "__futex" in scope=2/2 (die: 0x137ad5) 0(reg12) type=int (die:2a) ----------------------------------------------------------- find_data_type_die [52] for reg5 offset=0 in __memmove_avx_unaligned_erms CU die offset: 0x1124ed no variable found Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:00 -07:00
#define pr_debug_dtp(fmt, ...) \
do { \
if (debug_type_profile) \
pr_info(fmt, ##__VA_ARGS__); \
else \
pr_debug3(fmt, ##__VA_ARGS__); \
} while (0)
void pr_debug_type_name(Dwarf_Die *die, enum type_state_kind kind)
perf annotate-data: Add debug messages Add a new debug option "type-profile" to enable the detailed info during the type analysis especially for instruction tracking. You can use this before the command name like 'report' or 'annotate'. $ perf --debug type-profile annotate --data-type Committer testing: First get some memory events: $ perf mem record ls Then, without data-type profiling debug: $ perf annotate --data-type | head Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ And with only data-type profiling: $ perf --debug type-profile annotate --data-type | head ----------------------------------------------------------- find_data_type_die [1e67] for reg13873052 (PC) offset=0x150e2 in dl_main CU die offset: 0x29cd3 found PC-rel by addr=0x34020 offset=0x20 ----------------------------------------------------------- find_data_type_die [2e] for reg12 offset=0 in __GI___readdir64 CU die offset: 0x137a45 frame base: cfa=1 fbreg=-1 found "__futex" in scope=2/2 (die: 0x137ad5) 0(reg12) type=int (die:2a) ----------------------------------------------------------- find_data_type_die [52] for reg5 offset=0 in __memmove_avx_unaligned_erms CU die offset: 0x1124ed no variable found Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:00 -07:00
{
struct strbuf sb;
char *str;
perf annotate-data: Improve debug message with location info To verify it found the correct variable, let's add the location expression to the debug message. $ perf --debug type-profile annotate --data-type ... ----------------------------------------------------------- find data type for 0xaf0(reg15) at schedule+0xeb CU for kernel/sched/core.c (die:0x1180523) frame base: cfa=0 fbreg=6 found "rq" in scope=3/4 (die: 0x11b6a00) type_offset=0xaf0 variable location: reg15 type='struct rq' size=0xfc0 (die:0x11892e2) ----------------------------------------------------------- find data type for 0x7bc(reg3) at tcp_get_info+0x62 CU for net/ipv4/tcp.c (die:0x7b5f516) frame base: cfa=0 fbreg=6 offset: 1980 is bigger than size: 760 check variable "sk" failed (die: 0x7b92b2c) variable location: reg3 type='struct sock' size=0x2f8 (die:0x7b63c3a) ----------------------------------------------------------- ... The first case is fine. It looked up a data type in r15 with offset of 0xaf0 at schedule+0xeb. It found the CU die and the frame base info and the variable "rq" was found in the scope 3/4. Its location is the r15 register and the type size is 0xfc0 which includes 0xaf0. But the second case is not good. It looked up a data type in rbx (reg3) with offset 0x7bc. It found a CU and the frame base which is good so far. And it also found a variable "sk" but the access offset is bigger than the type size (1980 vs. 760 or 0x7bc vs. 0x2f8). The variable has the right location (reg3) but I need to figure out why it accesses beyond what it's supposed to. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240412183310.2518474-2-namhyung@kernel.org [ Fix the build on 32-bit by casting Dwarf_Word to (long) in pr_debug_location() ] Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-04-12 11:33:07 -07:00
Dwarf_Word size = 0;
perf annotate-data: Add debug messages Add a new debug option "type-profile" to enable the detailed info during the type analysis especially for instruction tracking. You can use this before the command name like 'report' or 'annotate'. $ perf --debug type-profile annotate --data-type Committer testing: First get some memory events: $ perf mem record ls Then, without data-type profiling debug: $ perf annotate --data-type | head Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ And with only data-type profiling: $ perf --debug type-profile annotate --data-type | head ----------------------------------------------------------- find_data_type_die [1e67] for reg13873052 (PC) offset=0x150e2 in dl_main CU die offset: 0x29cd3 found PC-rel by addr=0x34020 offset=0x20 ----------------------------------------------------------- find_data_type_die [2e] for reg12 offset=0 in __GI___readdir64 CU die offset: 0x137a45 frame base: cfa=1 fbreg=-1 found "__futex" in scope=2/2 (die: 0x137ad5) 0(reg12) type=int (die:2a) ----------------------------------------------------------- find_data_type_die [52] for reg5 offset=0 in __memmove_avx_unaligned_erms CU die offset: 0x1124ed no variable found Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:00 -07:00
if (!debug_type_profile && verbose < 3)
return;
perf annotate-data: Support general per-cpu access This is to support per-cpu variable access often without a matching DWARF entry. For some reason, I cannot find debug info of per-cpu variables sometimes. They have more complex pattern to calculate the address of per-cpu variables like below. 2b7d: mov -0x1e0(%rbp),%rax ; rax = cpu 2b84: mov -0x7da0f7e0(,%rax,8),%rcx ; rcx = __per_cpu_offset[cpu] * 2b8c: mov 0x34870(%rcx),%rax ; *(__per_cpu_offset[cpu] + 0x34870) Let's assume the rax register has a number for a CPU at 2b7d. The next instruction is to get the per-cpu offset' for that cpu. The offset -0x7da0f7e0 is 0xffffffff825f0820 in u64 which is the address of the '__per_cpu_offset' array in my system. So it'd get the actual offset of that CPU's per-cpu region and save it to the rcx register. Then, at 2b8c, accesses using rcx can be handled same as the global variable access. To handle this case, it should check if the offset of the instruction matches to the address of '__per_cpu_offset'. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-20-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:11 -07:00
switch (kind) {
case TSR_KIND_INVALID:
pr_info("\n");
return;
case TSR_KIND_PERCPU_BASE:
pr_info(" percpu base\n");
return;
perf annotate-data: Handle ADD instructions There are different patterns for percpu variable access using a constant value added to the base.  2aeb:  mov    -0x7da0f7e0(,%rax,8),%r14 # r14 = __per_cpu_offset[cpu]  2af3:  mov    $0x34740,%rax # rax = address of runqueues * 2afa:  add    %rax,%r14 # r14 = &per_cpu(runqueues, cpu)  2bfd:  cmpl   $0x0,0x10(%r14) # cpu_rq(cpu)->has_blocked_load  2b03:  je     0x2b36 At the first instruction, r14 has the __per_cpu_offset. And then rax has an immediate value and then added to r14 to calculate the address of a per-cpu variable. So it needs to track the immediate values and ADD instructions. Similar but a little different case is to use "this_cpu_off" instead of "__per_cpu_offset" for the current CPU. This time the variable address comes with PC-rel addressing. 89: mov $0x34740,%rax # rax = address of runqueues * 90: add %gs:0x7f015f60(%rip),%rax # 19a78 <this_cpu_off> 98: incl 0xd8c(%rax) # cpu_rq(cpu)->sched_count Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-21-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:12 -07:00
case TSR_KIND_CONST:
pr_info(" constant\n");
return;
case TSR_KIND_POINTER:
pr_info(" pointer");
/* it also prints the type info */
break;
case TSR_KIND_CANARY:
pr_info(" stack canary\n");
return;
perf annotate-data: Support general per-cpu access This is to support per-cpu variable access often without a matching DWARF entry. For some reason, I cannot find debug info of per-cpu variables sometimes. They have more complex pattern to calculate the address of per-cpu variables like below. 2b7d: mov -0x1e0(%rbp),%rax ; rax = cpu 2b84: mov -0x7da0f7e0(,%rax,8),%rcx ; rcx = __per_cpu_offset[cpu] * 2b8c: mov 0x34870(%rcx),%rax ; *(__per_cpu_offset[cpu] + 0x34870) Let's assume the rax register has a number for a CPU at 2b7d. The next instruction is to get the per-cpu offset' for that cpu. The offset -0x7da0f7e0 is 0xffffffff825f0820 in u64 which is the address of the '__per_cpu_offset' array in my system. So it'd get the actual offset of that CPU's per-cpu region and save it to the rcx register. Then, at 2b8c, accesses using rcx can be handled same as the global variable access. To handle this case, it should check if the offset of the instruction matches to the address of '__per_cpu_offset'. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-20-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:11 -07:00
case TSR_KIND_TYPE:
default:
break;
}
perf annotate-data: Improve debug message with location info To verify it found the correct variable, let's add the location expression to the debug message. $ perf --debug type-profile annotate --data-type ... ----------------------------------------------------------- find data type for 0xaf0(reg15) at schedule+0xeb CU for kernel/sched/core.c (die:0x1180523) frame base: cfa=0 fbreg=6 found "rq" in scope=3/4 (die: 0x11b6a00) type_offset=0xaf0 variable location: reg15 type='struct rq' size=0xfc0 (die:0x11892e2) ----------------------------------------------------------- find data type for 0x7bc(reg3) at tcp_get_info+0x62 CU for net/ipv4/tcp.c (die:0x7b5f516) frame base: cfa=0 fbreg=6 offset: 1980 is bigger than size: 760 check variable "sk" failed (die: 0x7b92b2c) variable location: reg3 type='struct sock' size=0x2f8 (die:0x7b63c3a) ----------------------------------------------------------- ... The first case is fine. It looked up a data type in r15 with offset of 0xaf0 at schedule+0xeb. It found the CU die and the frame base info and the variable "rq" was found in the scope 3/4. Its location is the r15 register and the type size is 0xfc0 which includes 0xaf0. But the second case is not good. It looked up a data type in rbx (reg3) with offset 0x7bc. It found a CU and the frame base which is good so far. And it also found a variable "sk" but the access offset is bigger than the type size (1980 vs. 760 or 0x7bc vs. 0x2f8). The variable has the right location (reg3) but I need to figure out why it accesses beyond what it's supposed to. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240412183310.2518474-2-namhyung@kernel.org [ Fix the build on 32-bit by casting Dwarf_Word to (long) in pr_debug_location() ] Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-04-12 11:33:07 -07:00
dwarf_aggregate_size(die, &size);
perf annotate-data: Add debug messages Add a new debug option "type-profile" to enable the detailed info during the type analysis especially for instruction tracking. You can use this before the command name like 'report' or 'annotate'. $ perf --debug type-profile annotate --data-type Committer testing: First get some memory events: $ perf mem record ls Then, without data-type profiling debug: $ perf annotate --data-type | head Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ And with only data-type profiling: $ perf --debug type-profile annotate --data-type | head ----------------------------------------------------------- find_data_type_die [1e67] for reg13873052 (PC) offset=0x150e2 in dl_main CU die offset: 0x29cd3 found PC-rel by addr=0x34020 offset=0x20 ----------------------------------------------------------- find_data_type_die [2e] for reg12 offset=0 in __GI___readdir64 CU die offset: 0x137a45 frame base: cfa=1 fbreg=-1 found "__futex" in scope=2/2 (die: 0x137ad5) 0(reg12) type=int (die:2a) ----------------------------------------------------------- find_data_type_die [52] for reg5 offset=0 in __memmove_avx_unaligned_erms CU die offset: 0x1124ed no variable found Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:00 -07:00
strbuf_init(&sb, 32);
die_get_typename_from_type(die, &sb);
str = strbuf_detach(&sb, NULL);
perf annotate-data: Improve debug message with location info To verify it found the correct variable, let's add the location expression to the debug message. $ perf --debug type-profile annotate --data-type ... ----------------------------------------------------------- find data type for 0xaf0(reg15) at schedule+0xeb CU for kernel/sched/core.c (die:0x1180523) frame base: cfa=0 fbreg=6 found "rq" in scope=3/4 (die: 0x11b6a00) type_offset=0xaf0 variable location: reg15 type='struct rq' size=0xfc0 (die:0x11892e2) ----------------------------------------------------------- find data type for 0x7bc(reg3) at tcp_get_info+0x62 CU for net/ipv4/tcp.c (die:0x7b5f516) frame base: cfa=0 fbreg=6 offset: 1980 is bigger than size: 760 check variable "sk" failed (die: 0x7b92b2c) variable location: reg3 type='struct sock' size=0x2f8 (die:0x7b63c3a) ----------------------------------------------------------- ... The first case is fine. It looked up a data type in r15 with offset of 0xaf0 at schedule+0xeb. It found the CU die and the frame base info and the variable "rq" was found in the scope 3/4. Its location is the r15 register and the type size is 0xfc0 which includes 0xaf0. But the second case is not good. It looked up a data type in rbx (reg3) with offset 0x7bc. It found a CU and the frame base which is good so far. And it also found a variable "sk" but the access offset is bigger than the type size (1980 vs. 760 or 0x7bc vs. 0x2f8). The variable has the right location (reg3) but I need to figure out why it accesses beyond what it's supposed to. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240412183310.2518474-2-namhyung@kernel.org [ Fix the build on 32-bit by casting Dwarf_Word to (long) in pr_debug_location() ] Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-04-12 11:33:07 -07:00
pr_info(" type='%s' size=%#lx (die:%#lx)\n",
str, (long)size, (long)dwarf_dieoffset(die));
perf annotate-data: Add debug messages Add a new debug option "type-profile" to enable the detailed info during the type analysis especially for instruction tracking. You can use this before the command name like 'report' or 'annotate'. $ perf --debug type-profile annotate --data-type Committer testing: First get some memory events: $ perf mem record ls Then, without data-type profiling debug: $ perf annotate --data-type | head Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ And with only data-type profiling: $ perf --debug type-profile annotate --data-type | head ----------------------------------------------------------- find_data_type_die [1e67] for reg13873052 (PC) offset=0x150e2 in dl_main CU die offset: 0x29cd3 found PC-rel by addr=0x34020 offset=0x20 ----------------------------------------------------------- find_data_type_die [2e] for reg12 offset=0 in __GI___readdir64 CU die offset: 0x137a45 frame base: cfa=1 fbreg=-1 found "__futex" in scope=2/2 (die: 0x137ad5) 0(reg12) type=int (die:2a) ----------------------------------------------------------- find_data_type_die [52] for reg5 offset=0 in __memmove_avx_unaligned_erms CU die offset: 0x1124ed no variable found Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:00 -07:00
free(str);
}
perf annotate-data: Improve debug message with location info To verify it found the correct variable, let's add the location expression to the debug message. $ perf --debug type-profile annotate --data-type ... ----------------------------------------------------------- find data type for 0xaf0(reg15) at schedule+0xeb CU for kernel/sched/core.c (die:0x1180523) frame base: cfa=0 fbreg=6 found "rq" in scope=3/4 (die: 0x11b6a00) type_offset=0xaf0 variable location: reg15 type='struct rq' size=0xfc0 (die:0x11892e2) ----------------------------------------------------------- find data type for 0x7bc(reg3) at tcp_get_info+0x62 CU for net/ipv4/tcp.c (die:0x7b5f516) frame base: cfa=0 fbreg=6 offset: 1980 is bigger than size: 760 check variable "sk" failed (die: 0x7b92b2c) variable location: reg3 type='struct sock' size=0x2f8 (die:0x7b63c3a) ----------------------------------------------------------- ... The first case is fine. It looked up a data type in r15 with offset of 0xaf0 at schedule+0xeb. It found the CU die and the frame base info and the variable "rq" was found in the scope 3/4. Its location is the r15 register and the type size is 0xfc0 which includes 0xaf0. But the second case is not good. It looked up a data type in rbx (reg3) with offset 0x7bc. It found a CU and the frame base which is good so far. And it also found a variable "sk" but the access offset is bigger than the type size (1980 vs. 760 or 0x7bc vs. 0x2f8). The variable has the right location (reg3) but I need to figure out why it accesses beyond what it's supposed to. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240412183310.2518474-2-namhyung@kernel.org [ Fix the build on 32-bit by casting Dwarf_Word to (long) in pr_debug_location() ] Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-04-12 11:33:07 -07:00
static void pr_debug_location(Dwarf_Die *die, u64 pc, int reg)
{
ptrdiff_t off = 0;
Dwarf_Attribute attr;
Dwarf_Addr base, start, end;
Dwarf_Op *ops;
size_t nops;
if (!debug_type_profile && verbose < 3)
return;
if (dwarf_attr(die, DW_AT_location, &attr) == NULL)
return;
while ((off = dwarf_getlocations(&attr, off, &base, &start, &end, &ops, &nops)) > 0) {
perf annotate-data: Fix off-by-one in location range check The location list will have entries with half-open addressing like [start, end) which means it doesn't include the end address. So it should skip entries at the end address and match to the next entry. An example location list looks like this (from readelf -wo): 00237876 ffffffff8110d32b (base address) 0023787f v000000000000000 v000000000000002 views at 00237868 for: ffffffff8110d32b ffffffff8110d4eb (DW_OP_reg3 (rbx)) <<<--- 1 00237885 v000000000000002 v000000000000000 views at 0023786a for: ffffffff8110d4eb ffffffff8110d50b (DW_OP_reg14 (r14)) <<<--- 2 0023788c v000000000000000 v000000000000001 views at 0023786c for: ffffffff8110d50b ffffffff8110d7c4 (DW_OP_reg3 (rbx)) 00237893 v000000000000000 v000000000000000 views at 0023786e for: ffffffff8110d806 ffffffff8110d854 (DW_OP_reg3 (rbx)) 0023789a v000000000000000 v000000000000000 views at 00237870 for: ffffffff8110d876 ffffffff8110d88e (DW_OP_reg3 (rbx)) The first entry at 0023787f has [8110d32b, 8110d4eb) (omitting the ffffffff at the beginning), and the second one has [8110d4eb, 8110d50b). Fixes: 2bc3cf575a162a2c ("perf annotate-data: Improve debug message with location info") Reviewed-by: Masami Hiramatsu <mhiramat@kernel.org> Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240816235840.2754937-3-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-16 16:58:32 -07:00
if (reg != DWARF_REG_PC && end <= pc)
perf annotate-data: Improve debug message with location info To verify it found the correct variable, let's add the location expression to the debug message. $ perf --debug type-profile annotate --data-type ... ----------------------------------------------------------- find data type for 0xaf0(reg15) at schedule+0xeb CU for kernel/sched/core.c (die:0x1180523) frame base: cfa=0 fbreg=6 found "rq" in scope=3/4 (die: 0x11b6a00) type_offset=0xaf0 variable location: reg15 type='struct rq' size=0xfc0 (die:0x11892e2) ----------------------------------------------------------- find data type for 0x7bc(reg3) at tcp_get_info+0x62 CU for net/ipv4/tcp.c (die:0x7b5f516) frame base: cfa=0 fbreg=6 offset: 1980 is bigger than size: 760 check variable "sk" failed (die: 0x7b92b2c) variable location: reg3 type='struct sock' size=0x2f8 (die:0x7b63c3a) ----------------------------------------------------------- ... The first case is fine. It looked up a data type in r15 with offset of 0xaf0 at schedule+0xeb. It found the CU die and the frame base info and the variable "rq" was found in the scope 3/4. Its location is the r15 register and the type size is 0xfc0 which includes 0xaf0. But the second case is not good. It looked up a data type in rbx (reg3) with offset 0x7bc. It found a CU and the frame base which is good so far. And it also found a variable "sk" but the access offset is bigger than the type size (1980 vs. 760 or 0x7bc vs. 0x2f8). The variable has the right location (reg3) but I need to figure out why it accesses beyond what it's supposed to. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240412183310.2518474-2-namhyung@kernel.org [ Fix the build on 32-bit by casting Dwarf_Word to (long) in pr_debug_location() ] Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-04-12 11:33:07 -07:00
continue;
if (reg != DWARF_REG_PC && start > pc)
break;
pr_info(" variable location: ");
switch (ops->atom) {
case DW_OP_reg0 ...DW_OP_reg31:
pr_info("reg%d\n", ops->atom - DW_OP_reg0);
break;
case DW_OP_breg0 ...DW_OP_breg31:
pr_info("base=reg%d, offset=%#lx\n",
ops->atom - DW_OP_breg0, (long)ops->number);
break;
case DW_OP_regx:
pr_info("reg%ld\n", (long)ops->number);
break;
case DW_OP_bregx:
pr_info("base=reg%ld, offset=%#lx\n",
(long)ops->number, (long)ops->number2);
break;
case DW_OP_fbreg:
pr_info("use frame base, offset=%#lx\n", (long)ops->number);
break;
case DW_OP_addr:
pr_info("address=%#lx\n", (long)ops->number);
break;
default:
pr_info("unknown: code=%#x, number=%#lx\n",
ops->atom, (long)ops->number);
break;
}
break;
}
}
static void pr_debug_scope(Dwarf_Die *scope_die)
{
int tag;
if (!debug_type_profile && verbose < 3)
return;
pr_info("(die:%lx) ", (long)dwarf_dieoffset(scope_die));
tag = dwarf_tag(scope_die);
if (tag == DW_TAG_subprogram)
pr_info("[function] %s\n", dwarf_diename(scope_die));
else if (tag == DW_TAG_inlined_subroutine)
pr_info("[inlined] %s\n", dwarf_diename(scope_die));
else if (tag == DW_TAG_lexical_block)
pr_info("[block]\n");
else
pr_info("[unknown] tag=%x\n", tag);
}
bool has_reg_type(struct type_state *state, int reg)
{
return (unsigned)reg < ARRAY_SIZE(state->regs);
}
perf annotate-data: Implement instruction tracking If it failed to find a variable for the location directly, it might be due to a missing variable in the source code. For example, accessing pointer variables in a chain can result in the case like below: struct foo *foo = ...; int i = foo->bar->baz; The DWARF debug information is created for each variable so it'd have one for 'foo'. But there's no variable for 'foo->bar' and then it cannot know the type of 'bar' and 'baz'. The above source code can be compiled to the follow x86 instructions: mov 0x8(%rax), %rcx mov 0x4(%rcx), %rdx <=== PMU sample mov %rdx, -4(%rbp) Let's say 'foo' is located in the %rax and it has a pointer to struct foo. But perf sample is captured in the second instruction and there is no variable or type info for the %rcx. It'd be great if compiler could generate debug info for %rcx, but we should handle it on our side. So this patch implements the logic to iterate instructions and update the type table for each location. As it already collected a list of scopes including the target instruction, we can use it to construct the type table smartly. +---------------- scope[0] subprogram | | +-------------- scope[1] lexical_block | | | | +------------ scope[2] inlined_subroutine | | | | | | +---------- scope[3] inlined_subroutine | | | | | | | | +-------- scope[4] lexical_block | | | | | | | | | | *** target instruction ... Image the target instruction has 5 scopes, each scope will have its own variables and parameters. Then it can start with the innermost scope (4). So it'd search the shortest path from the start of scope[4] to the target address and build a list of basic blocks. Then it iterates the basic blocks with the variables in the scope and update the table. If it finds a type at the target instruction, then returns it. Otherwise, it moves to the upper scope[3]. Now it'd search the shortest path from the start of scope[3] to the start of scope[4]. Then connect it to the existing basic block list. Then it'd iterate the blocks with variables for both scopes. It can repeat this until it finds a type at the target instruction or reaches to the top scope[0]. As the basic blocks contain the shortest path, it won't worry about branches and can update the table simply. The final check will be done by find_matching_type() in the next patch. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-15-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:06 -07:00
static void init_type_state(struct type_state *state, struct arch *arch)
{
memset(state, 0, sizeof(*state));
INIT_LIST_HEAD(&state->stack_vars);
if (arch__is(arch, "x86")) {
state->regs[0].caller_saved = true;
state->regs[1].caller_saved = true;
state->regs[2].caller_saved = true;
state->regs[4].caller_saved = true;
state->regs[5].caller_saved = true;
state->regs[8].caller_saved = true;
state->regs[9].caller_saved = true;
state->regs[10].caller_saved = true;
state->regs[11].caller_saved = true;
state->ret_reg = 0;
perf annotate-data: Handle RSP if it's not the FB register In some cases, the stack pointer on x86 (rsp = reg7) is used to point variables on stack but it's not the frame base register. Then it should handle the register like normal registers (IOW not to access the other stack variables using offset calculation) but it should not assume it would have a pointer. Before: ----------------------------------------------------------- find data type for 0x7c(reg7) at tcp_getsockopt+0xb62 CU for net/ipv4/tcp.c (die:0x7b5f516) frame base: cfa=0 fbreg=6 no pointer or no type check variable "zc" failed (die: 0x7b9580a) variable location: base=reg7, offset=0x40 type='struct tcp_zerocopy_receive' size=0x40 (die:0x7b947f4) After: ----------------------------------------------------------- find data type for 0x7c(reg7) at tcp_getsockopt+0xb62 CU for net/ipv4/tcp.c (die:0x7b5f516) frame base: cfa=0 fbreg=6 found "zc" in scope=3/3 (die: 0x7b957fc) type_offset=0x3c variable location: base=reg7, offset=0x40 type='struct tcp_zerocopy_receive' size=0x40 (die:0x7b947f4) Note that the type-offset was properly calculated to 0x3c as the variable starts at 0x40. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240412183310.2518474-5-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-04-12 11:33:10 -07:00
state->stack_reg = X86_REG_SP;
}
}
perf annotate-data: Implement instruction tracking If it failed to find a variable for the location directly, it might be due to a missing variable in the source code. For example, accessing pointer variables in a chain can result in the case like below: struct foo *foo = ...; int i = foo->bar->baz; The DWARF debug information is created for each variable so it'd have one for 'foo'. But there's no variable for 'foo->bar' and then it cannot know the type of 'bar' and 'baz'. The above source code can be compiled to the follow x86 instructions: mov 0x8(%rax), %rcx mov 0x4(%rcx), %rdx <=== PMU sample mov %rdx, -4(%rbp) Let's say 'foo' is located in the %rax and it has a pointer to struct foo. But perf sample is captured in the second instruction and there is no variable or type info for the %rcx. It'd be great if compiler could generate debug info for %rcx, but we should handle it on our side. So this patch implements the logic to iterate instructions and update the type table for each location. As it already collected a list of scopes including the target instruction, we can use it to construct the type table smartly. +---------------- scope[0] subprogram | | +-------------- scope[1] lexical_block | | | | +------------ scope[2] inlined_subroutine | | | | | | +---------- scope[3] inlined_subroutine | | | | | | | | +-------- scope[4] lexical_block | | | | | | | | | | *** target instruction ... Image the target instruction has 5 scopes, each scope will have its own variables and parameters. Then it can start with the innermost scope (4). So it'd search the shortest path from the start of scope[4] to the target address and build a list of basic blocks. Then it iterates the basic blocks with the variables in the scope and update the table. If it finds a type at the target instruction, then returns it. Otherwise, it moves to the upper scope[3]. Now it'd search the shortest path from the start of scope[3] to the start of scope[4]. Then connect it to the existing basic block list. Then it'd iterate the blocks with variables for both scopes. It can repeat this until it finds a type at the target instruction or reaches to the top scope[0]. As the basic blocks contain the shortest path, it won't worry about branches and can update the table simply. The final check will be done by find_matching_type() in the next patch. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-15-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:06 -07:00
static void exit_type_state(struct type_state *state)
{
struct type_state_stack *stack, *tmp;
list_for_each_entry_safe(stack, tmp, &state->stack_vars, list) {
list_del(&stack->list);
free(stack);
}
}
perf annotate-data: Add dso->data_types tree To aggregate accesses to the same data type, add 'data_types' tree in DSO to maintain data types and find it by name and size. It might have different data types that happen to have the same name, so it also compares the size of the type. Even if it doesn't 100% guarantee, it reduces the possibility of mis-handling of such conflicts. And I don't think it's common to have different types with the same name. Committer notes: Very few cases on the Linux kernel, but there are some different types with the same name, unsure if there is a debug mode in libbpf dedup that warns about such cases, but there are provisions in pahole for that, see: "emit: Notice type shadowing, i.e. multiple types with the same name (enum, struct, union, etc)" https://git.kernel.org/pub/scm/devel/pahole/pahole.git/commit/?id=4f332dbfd02072e4f410db7bdcda8d6e3422974b $ pahole --compile > vmlinux.h $ rm -f a ; make a cc a.c -o a $ grep __[0-9] vmlinux.h union irte__1 { struct map_info__1; struct map_info__1 { struct map_info__1 * next; /* 0 8 */ $ drivers/iommu/amd/amd_iommu_types.h 'union irte' include/linux/dmar.h 'struct irte' include/linux/device-mapper.h: union map_info { void *ptr; }; include/linux/mtd/map.h: struct map_info { const char *name; unsigned long size; resource_size_t phys; <SNIP> kernel/events/uprobes.c: struct map_info { struct map_info *next; struct mm_struct *mm; unsigned long vaddr; }; Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Cc: linux-toolchains@vger.kernel.org Cc: linux-trace-devel@vger.kernel.org Link: https://lore.kernel.org/r/20231213001323.718046-5-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2023-12-12 16:13:10 -08:00
/*
* Compare type name and size to maintain them in a tree.
* I'm not sure if DWARF would have information of a single type in many
* different places (compilation units). If not, it could compare the
* offset of the type entry in the .debug_info section.
*/
static int data_type_cmp(const void *_key, const struct rb_node *node)
{
const struct annotated_data_type *key = _key;
struct annotated_data_type *type;
type = rb_entry(node, struct annotated_data_type, node);
if (key->self.size != type->self.size)
return key->self.size - type->self.size;
return strcmp(key->self.type_name, type->self.type_name);
perf annotate-data: Add dso->data_types tree To aggregate accesses to the same data type, add 'data_types' tree in DSO to maintain data types and find it by name and size. It might have different data types that happen to have the same name, so it also compares the size of the type. Even if it doesn't 100% guarantee, it reduces the possibility of mis-handling of such conflicts. And I don't think it's common to have different types with the same name. Committer notes: Very few cases on the Linux kernel, but there are some different types with the same name, unsure if there is a debug mode in libbpf dedup that warns about such cases, but there are provisions in pahole for that, see: "emit: Notice type shadowing, i.e. multiple types with the same name (enum, struct, union, etc)" https://git.kernel.org/pub/scm/devel/pahole/pahole.git/commit/?id=4f332dbfd02072e4f410db7bdcda8d6e3422974b $ pahole --compile > vmlinux.h $ rm -f a ; make a cc a.c -o a $ grep __[0-9] vmlinux.h union irte__1 { struct map_info__1; struct map_info__1 { struct map_info__1 * next; /* 0 8 */ $ drivers/iommu/amd/amd_iommu_types.h 'union irte' include/linux/dmar.h 'struct irte' include/linux/device-mapper.h: union map_info { void *ptr; }; include/linux/mtd/map.h: struct map_info { const char *name; unsigned long size; resource_size_t phys; <SNIP> kernel/events/uprobes.c: struct map_info { struct map_info *next; struct mm_struct *mm; unsigned long vaddr; }; Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Cc: linux-toolchains@vger.kernel.org Cc: linux-trace-devel@vger.kernel.org Link: https://lore.kernel.org/r/20231213001323.718046-5-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2023-12-12 16:13:10 -08:00
}
static bool data_type_less(struct rb_node *node_a, const struct rb_node *node_b)
{
struct annotated_data_type *a, *b;
a = rb_entry(node_a, struct annotated_data_type, node);
b = rb_entry(node_b, struct annotated_data_type, node);
if (a->self.size != b->self.size)
return a->self.size < b->self.size;
return strcmp(a->self.type_name, b->self.type_name) < 0;
}
/* Recursively add new members for struct/union */
static int __add_member_cb(Dwarf_Die *die, void *arg)
{
struct annotated_member *parent = arg;
struct annotated_member *member;
Dwarf_Die member_type, die_mem;
perf annotate-data: Set bitfield member offset and size properly The bitfield members might not have DW_AT_data_member_location. Let's use DW_AT_data_bit_offset to set the member offset correct. Also use DW_AT_bit_size for the name like in a C program. Before: Annotate type: 'struct sk_buff' (1 samples) Percent Offset Size Field - 100.00 0 232 struct sk_buff { + 0.00 0 24 union ; + 0.00 24 8 union ; + 0.00 32 8 union ; 0.00 40 48 char[] cb; + 0.00 88 16 union ; 0.00 104 8 long unsigned int _nfct; 100.00 112 4 unsigned int len; 0.00 116 4 unsigned int data_len; 0.00 120 2 __u16 mac_len; 0.00 122 2 __u16 hdr_len; 0.00 124 2 __u16 queue_mapping; 0.00 126 0 __u8[] __cloned_offset; 0.00 0 1 __u8 cloned; 0.00 0 1 __u8 nohdr; 0.00 0 1 __u8 fclone; 0.00 0 1 __u8 peeked; 0.00 0 1 __u8 head_frag; 0.00 0 1 __u8 pfmemalloc; 0.00 0 1 __u8 pp_recycle; 0.00 127 1 __u8 active_extensions; + 0.00 128 60 union ; 0.00 188 4 sk_buff_data_t tail; 0.00 192 4 sk_buff_data_t end; 0.00 200 8 unsigned char* head; After: Annotate type: 'struct sk_buff' (1 samples) Percent Offset Size Field - 100.00 0 232 struct sk_buff { + 0.00 0 24 union ; + 0.00 24 8 union ; + 0.00 32 8 union ; 0.00 40 48 char[] cb + 0.00 88 16 union ; 0.00 104 8 long unsigned int _nfct; 100.00 112 4 unsigned int len; 0.00 116 4 unsigned int data_len; 0.00 120 2 __u16 mac_len; 0.00 122 2 __u16 hdr_len; 0.00 124 2 __u16 queue_mapping; 0.00 126 0 __u8[] __cloned_offset; 0.00 126 1 __u8 cloned:1; 0.00 126 1 __u8 nohdr:1; 0.00 126 1 __u8 fclone:2; 0.00 126 1 __u8 peeked:1; 0.00 126 1 __u8 head_frag:1; 0.00 126 1 __u8 pfmemalloc:1; 0.00 126 1 __u8 pp_recycle:1; 0.00 127 1 __u8 active_extensions; + 0.00 128 60 union ; 0.00 188 4 sk_buff_data_t tail; 0.00 192 4 sk_buff_data_t end; 0.00 200 8 unsigned char* head; Commiter notes: Collect some data: root@number:~# perf mem record -a --ldlat 5 -- ping -s 8193 -f 192.168.86.1 Memory events are enabled on a subset of CPUs: 16-27 PING 192.168.86.1 (192.168.86.1) 8193(8221) bytes of data. .^C --- 192.168.86.1 ping statistics --- 13881 packets transmitted, 13880 received, 0.00720409% packet loss, time 8664ms rtt min/avg/max/mdev = 0.510/0.599/7.768/0.115 ms, ipg/ewma 0.624/0.593 ms [ perf record: Woken up 8 times to write data ] [ perf record: Captured and wrote 14.877 MB perf.data (46785 samples) ] root@number:~# root@number:~# perf evlist cpu_atom/mem-loads,ldlat=5/P cpu_atom/mem-stores/P dummy:u root@number:~# perf evlist -v cpu_atom/mem-loads,ldlat=5/P: type: 10 (cpu_atom), size: 136, config: 0x5d0 (mem-loads), { sample_period, sample_freq }: 4000, sample_type: IP|TID|TIME|ADDR|CPU|PERIOD|IDENTIFIER|DATA_SRC|WEIGHT_STRUCT, read_format: ID|LOST, disabled: 1, inherit: 1, freq: 1, precise_ip: 3, sample_id_all: 1, { bp_addr, config1 }: 0x7 cpu_atom/mem-stores/P: type: 10 (cpu_atom), size: 136, config: 0x6d0 (mem-stores), { sample_period, sample_freq }: 4000, sample_type: IP|TID|TIME|ADDR|CPU|PERIOD|IDENTIFIER|DATA_SRC|WEIGHT_STRUCT, read_format: ID|LOST, disabled: 1, inherit: 1, freq: 1, precise_ip: 3, sample_id_all: 1 dummy:u: type: 1 (software), size: 136, config: 0x9 (PERF_COUNT_SW_DUMMY), { sample_period, sample_freq }: 1, sample_type: IP|TID|TIME|ADDR|CPU|IDENTIFIER|DATA_SRC|WEIGHT_STRUCT, read_format: ID|LOST, inherit: 1, exclude_kernel: 1, exclude_hv: 1, mmap: 1, comm: 1, task: 1, mmap_data: 1, sample_id_all: 1, exclude_guest: 1, mmap2: 1, comm_exec: 1, ksymbol: 1, bpf_event: 1 root@number:~# Ok, now lets see what changes from before this patch to after it: root@number:~# perf annotate --data-type > /tmp/before Apply the patch, build: root@number:~# perf annotate --data-type > /tmp/after The first hunk of the diff, for a glib data structure, in userspace, look at those bitfields: root@number:~# diff -u10 /tmp/before /tmp/after | head -20 --- /tmp/before 2024-08-20 17:29:58.306765780 -0300 +++ /tmp/after 2024-08-20 17:33:13.210582596 -0300 @@ -163,22 +163,22 @@ Annotate type: 'GHashTable' in /usr/lib64/libglib-2.0.so.0.8000.3 (1 samples): ============================================================================ Percent offset size field 100.00 0 96 GHashTable { 0.00 0 8 gsize size; 0.00 8 4 gint mod; 100.00 12 4 guint mask; 0.00 16 4 guint nnodes; 0.00 20 4 guint noccupied; - 0.00 0 4 guint have_big_keys; - 0.00 0 4 guint have_big_values; + 0.00 24 1 guint have_big_keys:1; + 0.00 24 1 guint have_big_values:1; 0.00 32 8 gpointer keys; 0.00 40 8 guint* hashes; 0.00 48 8 gpointer values; root@number:~# As advertised :-) Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240815223823.2402285-1-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-15 15:38:23 -07:00
Dwarf_Word size, loc, bit_size = 0;
Dwarf_Attribute attr;
struct strbuf sb;
int tag;
if (dwarf_tag(die) != DW_TAG_member)
return DIE_FIND_CB_SIBLING;
member = zalloc(sizeof(*member));
if (member == NULL)
return DIE_FIND_CB_END;
strbuf_init(&sb, 32);
die_get_typename(die, &sb);
perf annotate-data: Show typedef names properly The die_get_typename() would resolve typedef and get to the original type. But sometimes the original type is a struct without name and it makes the output confusing and hard to read. This is a diff of perf report -s type before and after the change. New types such as atomic{,64}_t and sigset_t appeared and the portion of unnamed struct was reduced. Also u32, u64 and size_t were splitted from the base types. --- b 2024-08-01 17:02:34.307809952 -0700 +++ a 2024-08-07 14:17:05.245853999 -0700 - 2.40% long unsigned int + 2.26% long unsigned int - 1.56% unsigned int + 1.27% unsigned int - 0.98% struct - 0.79% long long unsigned int + 0.58% long long unsigned int + 0.36% struct + 0.27% atomic64_t + 0.22% u32 + 0.21% u64 + 0.19% atomic_t + 0.13% size_t - 0.08% struct seqcount_spinlock + 0.08% seqcount_spinlock_t + 0.08% sigset_t + 0.08% __poll_t Let's use the typedef name directly and the resolved to get the size of the type. Committer testing: root@x1:~# diff -u before after | head -30 --- before 2024-08-08 09:35:13.917325041 -0300 +++ after 2024-08-08 09:37:35.312257905 -0300 @@ -10,25 +10,27 @@ # ........ ......... # 79.40% (unknown) - 2.28% union 1.96% (stack operation) - 1.24% struct + 1.87% pthread_mutex_t 0.99% u32[] - 0.92% unsigned int 0.77% struct task_struct + 0.75% U32 0.75% struct pcpu_hot 0.63% struct qspinlock + 0.61% atomic_t 0.59% struct list_head - 0.58% int 0.53% struct cfs_rq 0.51% BYTE* - 0.48% unsigned char + 0.48% BYTE 0.48% long unsigned int 0.46% struct rq 0.41% struct worker 0.41% struct memcg_vmstats_percpu + 0.41% pthread_cond_t 0.37% _Bool + 0.36% int root@x1:~# Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240807223129.1738004-1-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-07 15:31:29 -07:00
__die_get_real_type(die, &member_type);
if (dwarf_tag(&member_type) == DW_TAG_typedef)
die_get_real_type(&member_type, &die_mem);
else
die_mem = member_type;
if (dwarf_aggregate_size(&die_mem, &size) < 0)
size = 0;
perf annotate-data: Set bitfield member offset and size properly The bitfield members might not have DW_AT_data_member_location. Let's use DW_AT_data_bit_offset to set the member offset correct. Also use DW_AT_bit_size for the name like in a C program. Before: Annotate type: 'struct sk_buff' (1 samples) Percent Offset Size Field - 100.00 0 232 struct sk_buff { + 0.00 0 24 union ; + 0.00 24 8 union ; + 0.00 32 8 union ; 0.00 40 48 char[] cb; + 0.00 88 16 union ; 0.00 104 8 long unsigned int _nfct; 100.00 112 4 unsigned int len; 0.00 116 4 unsigned int data_len; 0.00 120 2 __u16 mac_len; 0.00 122 2 __u16 hdr_len; 0.00 124 2 __u16 queue_mapping; 0.00 126 0 __u8[] __cloned_offset; 0.00 0 1 __u8 cloned; 0.00 0 1 __u8 nohdr; 0.00 0 1 __u8 fclone; 0.00 0 1 __u8 peeked; 0.00 0 1 __u8 head_frag; 0.00 0 1 __u8 pfmemalloc; 0.00 0 1 __u8 pp_recycle; 0.00 127 1 __u8 active_extensions; + 0.00 128 60 union ; 0.00 188 4 sk_buff_data_t tail; 0.00 192 4 sk_buff_data_t end; 0.00 200 8 unsigned char* head; After: Annotate type: 'struct sk_buff' (1 samples) Percent Offset Size Field - 100.00 0 232 struct sk_buff { + 0.00 0 24 union ; + 0.00 24 8 union ; + 0.00 32 8 union ; 0.00 40 48 char[] cb + 0.00 88 16 union ; 0.00 104 8 long unsigned int _nfct; 100.00 112 4 unsigned int len; 0.00 116 4 unsigned int data_len; 0.00 120 2 __u16 mac_len; 0.00 122 2 __u16 hdr_len; 0.00 124 2 __u16 queue_mapping; 0.00 126 0 __u8[] __cloned_offset; 0.00 126 1 __u8 cloned:1; 0.00 126 1 __u8 nohdr:1; 0.00 126 1 __u8 fclone:2; 0.00 126 1 __u8 peeked:1; 0.00 126 1 __u8 head_frag:1; 0.00 126 1 __u8 pfmemalloc:1; 0.00 126 1 __u8 pp_recycle:1; 0.00 127 1 __u8 active_extensions; + 0.00 128 60 union ; 0.00 188 4 sk_buff_data_t tail; 0.00 192 4 sk_buff_data_t end; 0.00 200 8 unsigned char* head; Commiter notes: Collect some data: root@number:~# perf mem record -a --ldlat 5 -- ping -s 8193 -f 192.168.86.1 Memory events are enabled on a subset of CPUs: 16-27 PING 192.168.86.1 (192.168.86.1) 8193(8221) bytes of data. .^C --- 192.168.86.1 ping statistics --- 13881 packets transmitted, 13880 received, 0.00720409% packet loss, time 8664ms rtt min/avg/max/mdev = 0.510/0.599/7.768/0.115 ms, ipg/ewma 0.624/0.593 ms [ perf record: Woken up 8 times to write data ] [ perf record: Captured and wrote 14.877 MB perf.data (46785 samples) ] root@number:~# root@number:~# perf evlist cpu_atom/mem-loads,ldlat=5/P cpu_atom/mem-stores/P dummy:u root@number:~# perf evlist -v cpu_atom/mem-loads,ldlat=5/P: type: 10 (cpu_atom), size: 136, config: 0x5d0 (mem-loads), { sample_period, sample_freq }: 4000, sample_type: IP|TID|TIME|ADDR|CPU|PERIOD|IDENTIFIER|DATA_SRC|WEIGHT_STRUCT, read_format: ID|LOST, disabled: 1, inherit: 1, freq: 1, precise_ip: 3, sample_id_all: 1, { bp_addr, config1 }: 0x7 cpu_atom/mem-stores/P: type: 10 (cpu_atom), size: 136, config: 0x6d0 (mem-stores), { sample_period, sample_freq }: 4000, sample_type: IP|TID|TIME|ADDR|CPU|PERIOD|IDENTIFIER|DATA_SRC|WEIGHT_STRUCT, read_format: ID|LOST, disabled: 1, inherit: 1, freq: 1, precise_ip: 3, sample_id_all: 1 dummy:u: type: 1 (software), size: 136, config: 0x9 (PERF_COUNT_SW_DUMMY), { sample_period, sample_freq }: 1, sample_type: IP|TID|TIME|ADDR|CPU|IDENTIFIER|DATA_SRC|WEIGHT_STRUCT, read_format: ID|LOST, inherit: 1, exclude_kernel: 1, exclude_hv: 1, mmap: 1, comm: 1, task: 1, mmap_data: 1, sample_id_all: 1, exclude_guest: 1, mmap2: 1, comm_exec: 1, ksymbol: 1, bpf_event: 1 root@number:~# Ok, now lets see what changes from before this patch to after it: root@number:~# perf annotate --data-type > /tmp/before Apply the patch, build: root@number:~# perf annotate --data-type > /tmp/after The first hunk of the diff, for a glib data structure, in userspace, look at those bitfields: root@number:~# diff -u10 /tmp/before /tmp/after | head -20 --- /tmp/before 2024-08-20 17:29:58.306765780 -0300 +++ /tmp/after 2024-08-20 17:33:13.210582596 -0300 @@ -163,22 +163,22 @@ Annotate type: 'GHashTable' in /usr/lib64/libglib-2.0.so.0.8000.3 (1 samples): ============================================================================ Percent offset size field 100.00 0 96 GHashTable { 0.00 0 8 gsize size; 0.00 8 4 gint mod; 100.00 12 4 guint mask; 0.00 16 4 guint nnodes; 0.00 20 4 guint noccupied; - 0.00 0 4 guint have_big_keys; - 0.00 0 4 guint have_big_values; + 0.00 24 1 guint have_big_keys:1; + 0.00 24 1 guint have_big_values:1; 0.00 32 8 gpointer keys; 0.00 40 8 guint* hashes; 0.00 48 8 gpointer values; root@number:~# As advertised :-) Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240815223823.2402285-1-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-15 15:38:23 -07:00
if (dwarf_attr_integrate(die, DW_AT_data_member_location, &attr))
dwarf_formudata(&attr, &loc);
perf annotate-data: Set bitfield member offset and size properly The bitfield members might not have DW_AT_data_member_location. Let's use DW_AT_data_bit_offset to set the member offset correct. Also use DW_AT_bit_size for the name like in a C program. Before: Annotate type: 'struct sk_buff' (1 samples) Percent Offset Size Field - 100.00 0 232 struct sk_buff { + 0.00 0 24 union ; + 0.00 24 8 union ; + 0.00 32 8 union ; 0.00 40 48 char[] cb; + 0.00 88 16 union ; 0.00 104 8 long unsigned int _nfct; 100.00 112 4 unsigned int len; 0.00 116 4 unsigned int data_len; 0.00 120 2 __u16 mac_len; 0.00 122 2 __u16 hdr_len; 0.00 124 2 __u16 queue_mapping; 0.00 126 0 __u8[] __cloned_offset; 0.00 0 1 __u8 cloned; 0.00 0 1 __u8 nohdr; 0.00 0 1 __u8 fclone; 0.00 0 1 __u8 peeked; 0.00 0 1 __u8 head_frag; 0.00 0 1 __u8 pfmemalloc; 0.00 0 1 __u8 pp_recycle; 0.00 127 1 __u8 active_extensions; + 0.00 128 60 union ; 0.00 188 4 sk_buff_data_t tail; 0.00 192 4 sk_buff_data_t end; 0.00 200 8 unsigned char* head; After: Annotate type: 'struct sk_buff' (1 samples) Percent Offset Size Field - 100.00 0 232 struct sk_buff { + 0.00 0 24 union ; + 0.00 24 8 union ; + 0.00 32 8 union ; 0.00 40 48 char[] cb + 0.00 88 16 union ; 0.00 104 8 long unsigned int _nfct; 100.00 112 4 unsigned int len; 0.00 116 4 unsigned int data_len; 0.00 120 2 __u16 mac_len; 0.00 122 2 __u16 hdr_len; 0.00 124 2 __u16 queue_mapping; 0.00 126 0 __u8[] __cloned_offset; 0.00 126 1 __u8 cloned:1; 0.00 126 1 __u8 nohdr:1; 0.00 126 1 __u8 fclone:2; 0.00 126 1 __u8 peeked:1; 0.00 126 1 __u8 head_frag:1; 0.00 126 1 __u8 pfmemalloc:1; 0.00 126 1 __u8 pp_recycle:1; 0.00 127 1 __u8 active_extensions; + 0.00 128 60 union ; 0.00 188 4 sk_buff_data_t tail; 0.00 192 4 sk_buff_data_t end; 0.00 200 8 unsigned char* head; Commiter notes: Collect some data: root@number:~# perf mem record -a --ldlat 5 -- ping -s 8193 -f 192.168.86.1 Memory events are enabled on a subset of CPUs: 16-27 PING 192.168.86.1 (192.168.86.1) 8193(8221) bytes of data. .^C --- 192.168.86.1 ping statistics --- 13881 packets transmitted, 13880 received, 0.00720409% packet loss, time 8664ms rtt min/avg/max/mdev = 0.510/0.599/7.768/0.115 ms, ipg/ewma 0.624/0.593 ms [ perf record: Woken up 8 times to write data ] [ perf record: Captured and wrote 14.877 MB perf.data (46785 samples) ] root@number:~# root@number:~# perf evlist cpu_atom/mem-loads,ldlat=5/P cpu_atom/mem-stores/P dummy:u root@number:~# perf evlist -v cpu_atom/mem-loads,ldlat=5/P: type: 10 (cpu_atom), size: 136, config: 0x5d0 (mem-loads), { sample_period, sample_freq }: 4000, sample_type: IP|TID|TIME|ADDR|CPU|PERIOD|IDENTIFIER|DATA_SRC|WEIGHT_STRUCT, read_format: ID|LOST, disabled: 1, inherit: 1, freq: 1, precise_ip: 3, sample_id_all: 1, { bp_addr, config1 }: 0x7 cpu_atom/mem-stores/P: type: 10 (cpu_atom), size: 136, config: 0x6d0 (mem-stores), { sample_period, sample_freq }: 4000, sample_type: IP|TID|TIME|ADDR|CPU|PERIOD|IDENTIFIER|DATA_SRC|WEIGHT_STRUCT, read_format: ID|LOST, disabled: 1, inherit: 1, freq: 1, precise_ip: 3, sample_id_all: 1 dummy:u: type: 1 (software), size: 136, config: 0x9 (PERF_COUNT_SW_DUMMY), { sample_period, sample_freq }: 1, sample_type: IP|TID|TIME|ADDR|CPU|IDENTIFIER|DATA_SRC|WEIGHT_STRUCT, read_format: ID|LOST, inherit: 1, exclude_kernel: 1, exclude_hv: 1, mmap: 1, comm: 1, task: 1, mmap_data: 1, sample_id_all: 1, exclude_guest: 1, mmap2: 1, comm_exec: 1, ksymbol: 1, bpf_event: 1 root@number:~# Ok, now lets see what changes from before this patch to after it: root@number:~# perf annotate --data-type > /tmp/before Apply the patch, build: root@number:~# perf annotate --data-type > /tmp/after The first hunk of the diff, for a glib data structure, in userspace, look at those bitfields: root@number:~# diff -u10 /tmp/before /tmp/after | head -20 --- /tmp/before 2024-08-20 17:29:58.306765780 -0300 +++ /tmp/after 2024-08-20 17:33:13.210582596 -0300 @@ -163,22 +163,22 @@ Annotate type: 'GHashTable' in /usr/lib64/libglib-2.0.so.0.8000.3 (1 samples): ============================================================================ Percent offset size field 100.00 0 96 GHashTable { 0.00 0 8 gsize size; 0.00 8 4 gint mod; 100.00 12 4 guint mask; 0.00 16 4 guint nnodes; 0.00 20 4 guint noccupied; - 0.00 0 4 guint have_big_keys; - 0.00 0 4 guint have_big_values; + 0.00 24 1 guint have_big_keys:1; + 0.00 24 1 guint have_big_values:1; 0.00 32 8 gpointer keys; 0.00 40 8 guint* hashes; 0.00 48 8 gpointer values; root@number:~# As advertised :-) Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240815223823.2402285-1-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-15 15:38:23 -07:00
else {
/* bitfield member */
if (dwarf_attr_integrate(die, DW_AT_data_bit_offset, &attr) &&
dwarf_formudata(&attr, &loc) == 0)
loc /= 8;
else
loc = 0;
if (dwarf_attr_integrate(die, DW_AT_bit_size, &attr) &&
dwarf_formudata(&attr, &bit_size) == 0)
size = (bit_size + 7) / 8;
}
member->type_name = strbuf_detach(&sb, NULL);
/* member->var_name can be NULL */
perf annotate-data: Set bitfield member offset and size properly The bitfield members might not have DW_AT_data_member_location. Let's use DW_AT_data_bit_offset to set the member offset correct. Also use DW_AT_bit_size for the name like in a C program. Before: Annotate type: 'struct sk_buff' (1 samples) Percent Offset Size Field - 100.00 0 232 struct sk_buff { + 0.00 0 24 union ; + 0.00 24 8 union ; + 0.00 32 8 union ; 0.00 40 48 char[] cb; + 0.00 88 16 union ; 0.00 104 8 long unsigned int _nfct; 100.00 112 4 unsigned int len; 0.00 116 4 unsigned int data_len; 0.00 120 2 __u16 mac_len; 0.00 122 2 __u16 hdr_len; 0.00 124 2 __u16 queue_mapping; 0.00 126 0 __u8[] __cloned_offset; 0.00 0 1 __u8 cloned; 0.00 0 1 __u8 nohdr; 0.00 0 1 __u8 fclone; 0.00 0 1 __u8 peeked; 0.00 0 1 __u8 head_frag; 0.00 0 1 __u8 pfmemalloc; 0.00 0 1 __u8 pp_recycle; 0.00 127 1 __u8 active_extensions; + 0.00 128 60 union ; 0.00 188 4 sk_buff_data_t tail; 0.00 192 4 sk_buff_data_t end; 0.00 200 8 unsigned char* head; After: Annotate type: 'struct sk_buff' (1 samples) Percent Offset Size Field - 100.00 0 232 struct sk_buff { + 0.00 0 24 union ; + 0.00 24 8 union ; + 0.00 32 8 union ; 0.00 40 48 char[] cb + 0.00 88 16 union ; 0.00 104 8 long unsigned int _nfct; 100.00 112 4 unsigned int len; 0.00 116 4 unsigned int data_len; 0.00 120 2 __u16 mac_len; 0.00 122 2 __u16 hdr_len; 0.00 124 2 __u16 queue_mapping; 0.00 126 0 __u8[] __cloned_offset; 0.00 126 1 __u8 cloned:1; 0.00 126 1 __u8 nohdr:1; 0.00 126 1 __u8 fclone:2; 0.00 126 1 __u8 peeked:1; 0.00 126 1 __u8 head_frag:1; 0.00 126 1 __u8 pfmemalloc:1; 0.00 126 1 __u8 pp_recycle:1; 0.00 127 1 __u8 active_extensions; + 0.00 128 60 union ; 0.00 188 4 sk_buff_data_t tail; 0.00 192 4 sk_buff_data_t end; 0.00 200 8 unsigned char* head; Commiter notes: Collect some data: root@number:~# perf mem record -a --ldlat 5 -- ping -s 8193 -f 192.168.86.1 Memory events are enabled on a subset of CPUs: 16-27 PING 192.168.86.1 (192.168.86.1) 8193(8221) bytes of data. .^C --- 192.168.86.1 ping statistics --- 13881 packets transmitted, 13880 received, 0.00720409% packet loss, time 8664ms rtt min/avg/max/mdev = 0.510/0.599/7.768/0.115 ms, ipg/ewma 0.624/0.593 ms [ perf record: Woken up 8 times to write data ] [ perf record: Captured and wrote 14.877 MB perf.data (46785 samples) ] root@number:~# root@number:~# perf evlist cpu_atom/mem-loads,ldlat=5/P cpu_atom/mem-stores/P dummy:u root@number:~# perf evlist -v cpu_atom/mem-loads,ldlat=5/P: type: 10 (cpu_atom), size: 136, config: 0x5d0 (mem-loads), { sample_period, sample_freq }: 4000, sample_type: IP|TID|TIME|ADDR|CPU|PERIOD|IDENTIFIER|DATA_SRC|WEIGHT_STRUCT, read_format: ID|LOST, disabled: 1, inherit: 1, freq: 1, precise_ip: 3, sample_id_all: 1, { bp_addr, config1 }: 0x7 cpu_atom/mem-stores/P: type: 10 (cpu_atom), size: 136, config: 0x6d0 (mem-stores), { sample_period, sample_freq }: 4000, sample_type: IP|TID|TIME|ADDR|CPU|PERIOD|IDENTIFIER|DATA_SRC|WEIGHT_STRUCT, read_format: ID|LOST, disabled: 1, inherit: 1, freq: 1, precise_ip: 3, sample_id_all: 1 dummy:u: type: 1 (software), size: 136, config: 0x9 (PERF_COUNT_SW_DUMMY), { sample_period, sample_freq }: 1, sample_type: IP|TID|TIME|ADDR|CPU|IDENTIFIER|DATA_SRC|WEIGHT_STRUCT, read_format: ID|LOST, inherit: 1, exclude_kernel: 1, exclude_hv: 1, mmap: 1, comm: 1, task: 1, mmap_data: 1, sample_id_all: 1, exclude_guest: 1, mmap2: 1, comm_exec: 1, ksymbol: 1, bpf_event: 1 root@number:~# Ok, now lets see what changes from before this patch to after it: root@number:~# perf annotate --data-type > /tmp/before Apply the patch, build: root@number:~# perf annotate --data-type > /tmp/after The first hunk of the diff, for a glib data structure, in userspace, look at those bitfields: root@number:~# diff -u10 /tmp/before /tmp/after | head -20 --- /tmp/before 2024-08-20 17:29:58.306765780 -0300 +++ /tmp/after 2024-08-20 17:33:13.210582596 -0300 @@ -163,22 +163,22 @@ Annotate type: 'GHashTable' in /usr/lib64/libglib-2.0.so.0.8000.3 (1 samples): ============================================================================ Percent offset size field 100.00 0 96 GHashTable { 0.00 0 8 gsize size; 0.00 8 4 gint mod; 100.00 12 4 guint mask; 0.00 16 4 guint nnodes; 0.00 20 4 guint noccupied; - 0.00 0 4 guint have_big_keys; - 0.00 0 4 guint have_big_values; + 0.00 24 1 guint have_big_keys:1; + 0.00 24 1 guint have_big_values:1; 0.00 32 8 gpointer keys; 0.00 40 8 guint* hashes; 0.00 48 8 gpointer values; root@number:~# As advertised :-) Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240815223823.2402285-1-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-15 15:38:23 -07:00
if (dwarf_diename(die)) {
if (bit_size) {
if (asprintf(&member->var_name, "%s:%ld",
dwarf_diename(die), (long)bit_size) < 0)
member->var_name = NULL;
} else {
member->var_name = strdup(dwarf_diename(die));
}
if (member->var_name == NULL) {
free(member);
return DIE_FIND_CB_END;
}
}
member->size = size;
member->offset = loc + parent->offset;
INIT_LIST_HEAD(&member->children);
list_add_tail(&member->node, &parent->children);
perf annotate-data: Show typedef names properly The die_get_typename() would resolve typedef and get to the original type. But sometimes the original type is a struct without name and it makes the output confusing and hard to read. This is a diff of perf report -s type before and after the change. New types such as atomic{,64}_t and sigset_t appeared and the portion of unnamed struct was reduced. Also u32, u64 and size_t were splitted from the base types. --- b 2024-08-01 17:02:34.307809952 -0700 +++ a 2024-08-07 14:17:05.245853999 -0700 - 2.40% long unsigned int + 2.26% long unsigned int - 1.56% unsigned int + 1.27% unsigned int - 0.98% struct - 0.79% long long unsigned int + 0.58% long long unsigned int + 0.36% struct + 0.27% atomic64_t + 0.22% u32 + 0.21% u64 + 0.19% atomic_t + 0.13% size_t - 0.08% struct seqcount_spinlock + 0.08% seqcount_spinlock_t + 0.08% sigset_t + 0.08% __poll_t Let's use the typedef name directly and the resolved to get the size of the type. Committer testing: root@x1:~# diff -u before after | head -30 --- before 2024-08-08 09:35:13.917325041 -0300 +++ after 2024-08-08 09:37:35.312257905 -0300 @@ -10,25 +10,27 @@ # ........ ......... # 79.40% (unknown) - 2.28% union 1.96% (stack operation) - 1.24% struct + 1.87% pthread_mutex_t 0.99% u32[] - 0.92% unsigned int 0.77% struct task_struct + 0.75% U32 0.75% struct pcpu_hot 0.63% struct qspinlock + 0.61% atomic_t 0.59% struct list_head - 0.58% int 0.53% struct cfs_rq 0.51% BYTE* - 0.48% unsigned char + 0.48% BYTE 0.48% long unsigned int 0.46% struct rq 0.41% struct worker 0.41% struct memcg_vmstats_percpu + 0.41% pthread_cond_t 0.37% _Bool + 0.36% int root@x1:~# Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240807223129.1738004-1-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-07 15:31:29 -07:00
tag = dwarf_tag(&die_mem);
switch (tag) {
case DW_TAG_structure_type:
case DW_TAG_union_type:
perf annotate-data: Show typedef names properly The die_get_typename() would resolve typedef and get to the original type. But sometimes the original type is a struct without name and it makes the output confusing and hard to read. This is a diff of perf report -s type before and after the change. New types such as atomic{,64}_t and sigset_t appeared and the portion of unnamed struct was reduced. Also u32, u64 and size_t were splitted from the base types. --- b 2024-08-01 17:02:34.307809952 -0700 +++ a 2024-08-07 14:17:05.245853999 -0700 - 2.40% long unsigned int + 2.26% long unsigned int - 1.56% unsigned int + 1.27% unsigned int - 0.98% struct - 0.79% long long unsigned int + 0.58% long long unsigned int + 0.36% struct + 0.27% atomic64_t + 0.22% u32 + 0.21% u64 + 0.19% atomic_t + 0.13% size_t - 0.08% struct seqcount_spinlock + 0.08% seqcount_spinlock_t + 0.08% sigset_t + 0.08% __poll_t Let's use the typedef name directly and the resolved to get the size of the type. Committer testing: root@x1:~# diff -u before after | head -30 --- before 2024-08-08 09:35:13.917325041 -0300 +++ after 2024-08-08 09:37:35.312257905 -0300 @@ -10,25 +10,27 @@ # ........ ......... # 79.40% (unknown) - 2.28% union 1.96% (stack operation) - 1.24% struct + 1.87% pthread_mutex_t 0.99% u32[] - 0.92% unsigned int 0.77% struct task_struct + 0.75% U32 0.75% struct pcpu_hot 0.63% struct qspinlock + 0.61% atomic_t 0.59% struct list_head - 0.58% int 0.53% struct cfs_rq 0.51% BYTE* - 0.48% unsigned char + 0.48% BYTE 0.48% long unsigned int 0.46% struct rq 0.41% struct worker 0.41% struct memcg_vmstats_percpu + 0.41% pthread_cond_t 0.37% _Bool + 0.36% int root@x1:~# Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240807223129.1738004-1-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-07 15:31:29 -07:00
die_find_child(&die_mem, __add_member_cb, member, &die_mem);
break;
default:
break;
}
return DIE_FIND_CB_SIBLING;
}
static void add_member_types(struct annotated_data_type *parent, Dwarf_Die *type)
{
Dwarf_Die die_mem;
die_find_child(type, __add_member_cb, &parent->self, &die_mem);
}
static void delete_members(struct annotated_member *member)
{
struct annotated_member *child, *tmp;
list_for_each_entry_safe(child, tmp, &member->children, node) {
list_del(&child->node);
delete_members(child);
zfree(&child->type_name);
zfree(&child->var_name);
free(child);
}
perf annotate-data: Add dso->data_types tree To aggregate accesses to the same data type, add 'data_types' tree in DSO to maintain data types and find it by name and size. It might have different data types that happen to have the same name, so it also compares the size of the type. Even if it doesn't 100% guarantee, it reduces the possibility of mis-handling of such conflicts. And I don't think it's common to have different types with the same name. Committer notes: Very few cases on the Linux kernel, but there are some different types with the same name, unsure if there is a debug mode in libbpf dedup that warns about such cases, but there are provisions in pahole for that, see: "emit: Notice type shadowing, i.e. multiple types with the same name (enum, struct, union, etc)" https://git.kernel.org/pub/scm/devel/pahole/pahole.git/commit/?id=4f332dbfd02072e4f410db7bdcda8d6e3422974b $ pahole --compile > vmlinux.h $ rm -f a ; make a cc a.c -o a $ grep __[0-9] vmlinux.h union irte__1 { struct map_info__1; struct map_info__1 { struct map_info__1 * next; /* 0 8 */ $ drivers/iommu/amd/amd_iommu_types.h 'union irte' include/linux/dmar.h 'struct irte' include/linux/device-mapper.h: union map_info { void *ptr; }; include/linux/mtd/map.h: struct map_info { const char *name; unsigned long size; resource_size_t phys; <SNIP> kernel/events/uprobes.c: struct map_info { struct map_info *next; struct mm_struct *mm; unsigned long vaddr; }; Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Cc: linux-toolchains@vger.kernel.org Cc: linux-trace-devel@vger.kernel.org Link: https://lore.kernel.org/r/20231213001323.718046-5-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2023-12-12 16:13:10 -08:00
}
static struct annotated_data_type *dso__findnew_data_type(struct dso *dso,
Dwarf_Die *type_die)
{
struct annotated_data_type *result = NULL;
struct annotated_data_type key;
struct rb_node *node;
struct strbuf sb;
char *type_name;
Dwarf_Word size;
strbuf_init(&sb, 32);
if (die_get_typename_from_type(type_die, &sb) < 0)
strbuf_add(&sb, "(unknown type)", 14);
type_name = strbuf_detach(&sb, NULL);
perf annotate-data: Show typedef names properly The die_get_typename() would resolve typedef and get to the original type. But sometimes the original type is a struct without name and it makes the output confusing and hard to read. This is a diff of perf report -s type before and after the change. New types such as atomic{,64}_t and sigset_t appeared and the portion of unnamed struct was reduced. Also u32, u64 and size_t were splitted from the base types. --- b 2024-08-01 17:02:34.307809952 -0700 +++ a 2024-08-07 14:17:05.245853999 -0700 - 2.40% long unsigned int + 2.26% long unsigned int - 1.56% unsigned int + 1.27% unsigned int - 0.98% struct - 0.79% long long unsigned int + 0.58% long long unsigned int + 0.36% struct + 0.27% atomic64_t + 0.22% u32 + 0.21% u64 + 0.19% atomic_t + 0.13% size_t - 0.08% struct seqcount_spinlock + 0.08% seqcount_spinlock_t + 0.08% sigset_t + 0.08% __poll_t Let's use the typedef name directly and the resolved to get the size of the type. Committer testing: root@x1:~# diff -u before after | head -30 --- before 2024-08-08 09:35:13.917325041 -0300 +++ after 2024-08-08 09:37:35.312257905 -0300 @@ -10,25 +10,27 @@ # ........ ......... # 79.40% (unknown) - 2.28% union 1.96% (stack operation) - 1.24% struct + 1.87% pthread_mutex_t 0.99% u32[] - 0.92% unsigned int 0.77% struct task_struct + 0.75% U32 0.75% struct pcpu_hot 0.63% struct qspinlock + 0.61% atomic_t 0.59% struct list_head - 0.58% int 0.53% struct cfs_rq 0.51% BYTE* - 0.48% unsigned char + 0.48% BYTE 0.48% long unsigned int 0.46% struct rq 0.41% struct worker 0.41% struct memcg_vmstats_percpu + 0.41% pthread_cond_t 0.37% _Bool + 0.36% int root@x1:~# Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240807223129.1738004-1-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-07 15:31:29 -07:00
if (dwarf_tag(type_die) == DW_TAG_typedef)
die_get_real_type(type_die, type_die);
perf annotate-data: Add dso->data_types tree To aggregate accesses to the same data type, add 'data_types' tree in DSO to maintain data types and find it by name and size. It might have different data types that happen to have the same name, so it also compares the size of the type. Even if it doesn't 100% guarantee, it reduces the possibility of mis-handling of such conflicts. And I don't think it's common to have different types with the same name. Committer notes: Very few cases on the Linux kernel, but there are some different types with the same name, unsure if there is a debug mode in libbpf dedup that warns about such cases, but there are provisions in pahole for that, see: "emit: Notice type shadowing, i.e. multiple types with the same name (enum, struct, union, etc)" https://git.kernel.org/pub/scm/devel/pahole/pahole.git/commit/?id=4f332dbfd02072e4f410db7bdcda8d6e3422974b $ pahole --compile > vmlinux.h $ rm -f a ; make a cc a.c -o a $ grep __[0-9] vmlinux.h union irte__1 { struct map_info__1; struct map_info__1 { struct map_info__1 * next; /* 0 8 */ $ drivers/iommu/amd/amd_iommu_types.h 'union irte' include/linux/dmar.h 'struct irte' include/linux/device-mapper.h: union map_info { void *ptr; }; include/linux/mtd/map.h: struct map_info { const char *name; unsigned long size; resource_size_t phys; <SNIP> kernel/events/uprobes.c: struct map_info { struct map_info *next; struct mm_struct *mm; unsigned long vaddr; }; Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Cc: linux-toolchains@vger.kernel.org Cc: linux-trace-devel@vger.kernel.org Link: https://lore.kernel.org/r/20231213001323.718046-5-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2023-12-12 16:13:10 -08:00
dwarf_aggregate_size(type_die, &size);
/* Check existing nodes in dso->data_types tree */
key.self.type_name = type_name;
key.self.size = size;
perf dso: Add reference count checking and accessor functions Add reference count checking to struct dso, this can help with implementing correct reference counting discipline. To avoid RC_CHK_ACCESS everywhere, add accessor functions for the variables in struct dso. The majority of the change is mechanical in nature and not easy to split up. Committer testing: 'perf test' up to this patch shows no regressions. But: util/symbol.c: In function ‘dso__load_bfd_symbols’: util/symbol.c:1683:9: error: too few arguments to function ‘dso__set_adjust_symbols’ 1683 | dso__set_adjust_symbols(dso); | ^~~~~~~~~~~~~~~~~~~~~~~ In file included from util/symbol.c:21: util/dso.h:268:20: note: declared here 268 | static inline void dso__set_adjust_symbols(struct dso *dso, bool val) | ^~~~~~~~~~~~~~~~~~~~~~~ make[6]: *** [/home/acme/git/perf-tools-next/tools/build/Makefile.build:106: /tmp/tmp.ZWHbQftdN6/util/symbol.o] Error 1 MKDIR /tmp/tmp.ZWHbQftdN6/tests/workloads/ make[6]: *** Waiting for unfinished jobs.... This was updated: - symbols__fixup_end(&dso->symbols, false); - symbols__fixup_duplicate(&dso->symbols); - dso->adjust_symbols = 1; + symbols__fixup_end(dso__symbols(dso), false); + symbols__fixup_duplicate(dso__symbols(dso)); + dso__set_adjust_symbols(dso); But not build tested with BUILD_NONDISTRO and libbfd devel files installed (binutils-devel on fedora). Add the missing argument: symbols__fixup_end(dso__symbols(dso), false); symbols__fixup_duplicate(dso__symbols(dso)); - dso__set_adjust_symbols(dso); + dso__set_adjust_symbols(dso, true); Signed-off-by: Ian Rogers <irogers@google.com> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Andi Kleen <ak@linux.intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ben Gainey <ben.gainey@arm.com> Cc: Changbin Du <changbin.du@huawei.com> Cc: Chengen Du <chengen.du@canonical.com> Cc: Colin Ian King <colin.i.king@gmail.com> Cc: Dima Kogan <dima@secretsauce.net> Cc: Ilkka Koskinen <ilkka@os.amperecomputing.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: James Clark <james.clark@arm.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: K Prateek Nayak <kprateek.nayak@amd.com> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Leo Yan <leo.yan@linux.dev> Cc: Li Dong <lidong@vivo.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Paran Lee <p4ranlee@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Song Liu <song@kernel.org> Cc: Sun Haiyong <sunhaiyong@loongson.cn> Cc: Thomas Richter <tmricht@linux.ibm.com> Cc: Tiezhu Yang <yangtiezhu@loongson.cn> Cc: Yanteng Si <siyanteng@loongson.cn> Cc: zhaimingbing <zhaimingbing@cmss.chinamobile.com> Link: https://lore.kernel.org/r/20240504213803.218974-6-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-05-04 14:38:01 -07:00
node = rb_find(&key, dso__data_types(dso), data_type_cmp);
perf annotate-data: Add dso->data_types tree To aggregate accesses to the same data type, add 'data_types' tree in DSO to maintain data types and find it by name and size. It might have different data types that happen to have the same name, so it also compares the size of the type. Even if it doesn't 100% guarantee, it reduces the possibility of mis-handling of such conflicts. And I don't think it's common to have different types with the same name. Committer notes: Very few cases on the Linux kernel, but there are some different types with the same name, unsure if there is a debug mode in libbpf dedup that warns about such cases, but there are provisions in pahole for that, see: "emit: Notice type shadowing, i.e. multiple types with the same name (enum, struct, union, etc)" https://git.kernel.org/pub/scm/devel/pahole/pahole.git/commit/?id=4f332dbfd02072e4f410db7bdcda8d6e3422974b $ pahole --compile > vmlinux.h $ rm -f a ; make a cc a.c -o a $ grep __[0-9] vmlinux.h union irte__1 { struct map_info__1; struct map_info__1 { struct map_info__1 * next; /* 0 8 */ $ drivers/iommu/amd/amd_iommu_types.h 'union irte' include/linux/dmar.h 'struct irte' include/linux/device-mapper.h: union map_info { void *ptr; }; include/linux/mtd/map.h: struct map_info { const char *name; unsigned long size; resource_size_t phys; <SNIP> kernel/events/uprobes.c: struct map_info { struct map_info *next; struct mm_struct *mm; unsigned long vaddr; }; Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Cc: linux-toolchains@vger.kernel.org Cc: linux-trace-devel@vger.kernel.org Link: https://lore.kernel.org/r/20231213001323.718046-5-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2023-12-12 16:13:10 -08:00
if (node) {
result = rb_entry(node, struct annotated_data_type, node);
free(type_name);
return result;
}
/* If not, add a new one */
result = zalloc(sizeof(*result));
if (result == NULL) {
free(type_name);
return NULL;
}
result->self.type_name = type_name;
result->self.size = size;
INIT_LIST_HEAD(&result->self.children);
perf annotate: Add --data-type option Support data type annotation with new --data-type option. It internally uses type sort key to collect sample histogram for the type and display every members like below. $ perf annotate --data-type ... Annotate type: 'struct cfs_rq' in [kernel.kallsyms] (13 samples): ============================================================================ samples offset size field 13 0 640 struct cfs_rq { 2 0 16 struct load_weight load { 2 0 8 unsigned long weight; 0 8 4 u32 inv_weight; }; 0 16 8 unsigned long runnable_weight; 0 24 4 unsigned int nr_running; 1 28 4 unsigned int h_nr_running; ... For simplicity it prints the number of samples per field for now. But it should be easy to show the overhead percentage instead. The number at the outer struct is a sum of the numbers of the inner members. For example, struct cfs_rq got total 13 samples, and 2 came from the load (struct load_weight) and 1 from h_nr_running. Similarly, the struct load_weight got total 2 samples and they all came from the weight field. I've added two new flags in the symbol_conf for this. The annotate_data_member is to get the members of the type. This is also needed for perf report with typeoff sort key. The annotate_data_sample is to update sample stats for each offset and used only in annotate. Currently it only support stdio output mode, TUI support can be added later. Committer testing: With the perf.data from the previous csets, a very simple, short duration one: # perf annotate --data-type Annotate type: 'struct list_head' in [kernel.kallsyms] (1 samples): ============================================================================ samples offset size field 1 0 16 struct list_head { 0 0 8 struct list_head* next; 1 8 8 struct list_head* prev; }; Annotate type: 'char' in [kernel.kallsyms] (1 samples): ============================================================================ samples offset size field 1 0 1 char ; # Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Cc: linux-toolchains@vger.kernel.org Cc: linux-trace-devel@vger.kernel.org Link: https://lore.kernel.org/r/20231213001323.718046-15-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2023-12-12 16:13:20 -08:00
if (symbol_conf.annotate_data_member)
add_member_types(result, type_die);
perf annotate-data: Add dso->data_types tree To aggregate accesses to the same data type, add 'data_types' tree in DSO to maintain data types and find it by name and size. It might have different data types that happen to have the same name, so it also compares the size of the type. Even if it doesn't 100% guarantee, it reduces the possibility of mis-handling of such conflicts. And I don't think it's common to have different types with the same name. Committer notes: Very few cases on the Linux kernel, but there are some different types with the same name, unsure if there is a debug mode in libbpf dedup that warns about such cases, but there are provisions in pahole for that, see: "emit: Notice type shadowing, i.e. multiple types with the same name (enum, struct, union, etc)" https://git.kernel.org/pub/scm/devel/pahole/pahole.git/commit/?id=4f332dbfd02072e4f410db7bdcda8d6e3422974b $ pahole --compile > vmlinux.h $ rm -f a ; make a cc a.c -o a $ grep __[0-9] vmlinux.h union irte__1 { struct map_info__1; struct map_info__1 { struct map_info__1 * next; /* 0 8 */ $ drivers/iommu/amd/amd_iommu_types.h 'union irte' include/linux/dmar.h 'struct irte' include/linux/device-mapper.h: union map_info { void *ptr; }; include/linux/mtd/map.h: struct map_info { const char *name; unsigned long size; resource_size_t phys; <SNIP> kernel/events/uprobes.c: struct map_info { struct map_info *next; struct mm_struct *mm; unsigned long vaddr; }; Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Cc: linux-toolchains@vger.kernel.org Cc: linux-trace-devel@vger.kernel.org Link: https://lore.kernel.org/r/20231213001323.718046-5-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2023-12-12 16:13:10 -08:00
perf dso: Add reference count checking and accessor functions Add reference count checking to struct dso, this can help with implementing correct reference counting discipline. To avoid RC_CHK_ACCESS everywhere, add accessor functions for the variables in struct dso. The majority of the change is mechanical in nature and not easy to split up. Committer testing: 'perf test' up to this patch shows no regressions. But: util/symbol.c: In function ‘dso__load_bfd_symbols’: util/symbol.c:1683:9: error: too few arguments to function ‘dso__set_adjust_symbols’ 1683 | dso__set_adjust_symbols(dso); | ^~~~~~~~~~~~~~~~~~~~~~~ In file included from util/symbol.c:21: util/dso.h:268:20: note: declared here 268 | static inline void dso__set_adjust_symbols(struct dso *dso, bool val) | ^~~~~~~~~~~~~~~~~~~~~~~ make[6]: *** [/home/acme/git/perf-tools-next/tools/build/Makefile.build:106: /tmp/tmp.ZWHbQftdN6/util/symbol.o] Error 1 MKDIR /tmp/tmp.ZWHbQftdN6/tests/workloads/ make[6]: *** Waiting for unfinished jobs.... This was updated: - symbols__fixup_end(&dso->symbols, false); - symbols__fixup_duplicate(&dso->symbols); - dso->adjust_symbols = 1; + symbols__fixup_end(dso__symbols(dso), false); + symbols__fixup_duplicate(dso__symbols(dso)); + dso__set_adjust_symbols(dso); But not build tested with BUILD_NONDISTRO and libbfd devel files installed (binutils-devel on fedora). Add the missing argument: symbols__fixup_end(dso__symbols(dso), false); symbols__fixup_duplicate(dso__symbols(dso)); - dso__set_adjust_symbols(dso); + dso__set_adjust_symbols(dso, true); Signed-off-by: Ian Rogers <irogers@google.com> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Andi Kleen <ak@linux.intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ben Gainey <ben.gainey@arm.com> Cc: Changbin Du <changbin.du@huawei.com> Cc: Chengen Du <chengen.du@canonical.com> Cc: Colin Ian King <colin.i.king@gmail.com> Cc: Dima Kogan <dima@secretsauce.net> Cc: Ilkka Koskinen <ilkka@os.amperecomputing.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: James Clark <james.clark@arm.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: K Prateek Nayak <kprateek.nayak@amd.com> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Leo Yan <leo.yan@linux.dev> Cc: Li Dong <lidong@vivo.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Paran Lee <p4ranlee@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Song Liu <song@kernel.org> Cc: Sun Haiyong <sunhaiyong@loongson.cn> Cc: Thomas Richter <tmricht@linux.ibm.com> Cc: Tiezhu Yang <yangtiezhu@loongson.cn> Cc: Yanteng Si <siyanteng@loongson.cn> Cc: zhaimingbing <zhaimingbing@cmss.chinamobile.com> Link: https://lore.kernel.org/r/20240504213803.218974-6-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-05-04 14:38:01 -07:00
rb_add(&result->node, dso__data_types(dso), data_type_less);
perf annotate-data: Add dso->data_types tree To aggregate accesses to the same data type, add 'data_types' tree in DSO to maintain data types and find it by name and size. It might have different data types that happen to have the same name, so it also compares the size of the type. Even if it doesn't 100% guarantee, it reduces the possibility of mis-handling of such conflicts. And I don't think it's common to have different types with the same name. Committer notes: Very few cases on the Linux kernel, but there are some different types with the same name, unsure if there is a debug mode in libbpf dedup that warns about such cases, but there are provisions in pahole for that, see: "emit: Notice type shadowing, i.e. multiple types with the same name (enum, struct, union, etc)" https://git.kernel.org/pub/scm/devel/pahole/pahole.git/commit/?id=4f332dbfd02072e4f410db7bdcda8d6e3422974b $ pahole --compile > vmlinux.h $ rm -f a ; make a cc a.c -o a $ grep __[0-9] vmlinux.h union irte__1 { struct map_info__1; struct map_info__1 { struct map_info__1 * next; /* 0 8 */ $ drivers/iommu/amd/amd_iommu_types.h 'union irte' include/linux/dmar.h 'struct irte' include/linux/device-mapper.h: union map_info { void *ptr; }; include/linux/mtd/map.h: struct map_info { const char *name; unsigned long size; resource_size_t phys; <SNIP> kernel/events/uprobes.c: struct map_info { struct map_info *next; struct mm_struct *mm; unsigned long vaddr; }; Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Cc: linux-toolchains@vger.kernel.org Cc: linux-trace-devel@vger.kernel.org Link: https://lore.kernel.org/r/20231213001323.718046-5-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2023-12-12 16:13:10 -08:00
return result;
}
static bool find_cu_die(struct debuginfo *di, u64 pc, Dwarf_Die *cu_die)
{
Dwarf_Off off, next_off;
size_t header_size;
if (dwarf_addrdie(di->dbg, pc, cu_die) != NULL)
return cu_die;
/*
* There are some kernels don't have full aranges and contain only a few
* aranges entries. Fallback to iterate all CU entries in .debug_info
* in case it's missing.
*/
off = 0;
while (dwarf_nextcu(di->dbg, off, &next_off, &header_size,
NULL, NULL, NULL) == 0) {
if (dwarf_offdie(di->dbg, off + header_size, cu_die) &&
dwarf_haspc(cu_die, pc))
return true;
off = next_off;
}
return false;
}
enum type_match_result {
PERF_TMR_UNKNOWN = 0,
PERF_TMR_OK,
PERF_TMR_NO_TYPE,
PERF_TMR_NO_POINTER,
PERF_TMR_NO_SIZE,
PERF_TMR_BAD_OFFSET,
PERF_TMR_BAIL_OUT,
};
static const char *match_result_str(enum type_match_result tmr)
{
switch (tmr) {
case PERF_TMR_OK:
return "Good!";
case PERF_TMR_NO_TYPE:
return "no type information";
case PERF_TMR_NO_POINTER:
return "no/void pointer";
case PERF_TMR_NO_SIZE:
return "type size is unknown";
case PERF_TMR_BAD_OFFSET:
return "offset bigger than size";
case PERF_TMR_UNKNOWN:
case PERF_TMR_BAIL_OUT:
default:
return "invalid state";
}
}
static bool is_pointer_type(Dwarf_Die *type_die)
{
int tag = dwarf_tag(type_die);
return tag == DW_TAG_pointer_type || tag == DW_TAG_array_type;
}
static bool is_compound_type(Dwarf_Die *type_die)
{
int tag = dwarf_tag(type_die);
return tag == DW_TAG_structure_type || tag == DW_TAG_union_type;
}
/* returns if Type B has better information than Type A */
static bool is_better_type(Dwarf_Die *type_a, Dwarf_Die *type_b)
{
Dwarf_Word size_a, size_b;
Dwarf_Die die_a, die_b;
/* pointer type is preferred */
if (is_pointer_type(type_a) != is_pointer_type(type_b))
return is_pointer_type(type_b);
if (is_pointer_type(type_b)) {
/*
* We want to compare the target type, but 'void *' can fail to
* get the target type.
*/
if (die_get_real_type(type_a, &die_a) == NULL)
return true;
if (die_get_real_type(type_b, &die_b) == NULL)
return false;
type_a = &die_a;
type_b = &die_b;
}
/* bigger type is preferred */
if (dwarf_aggregate_size(type_a, &size_a) < 0 ||
dwarf_aggregate_size(type_b, &size_b) < 0)
return false;
if (size_a != size_b)
return size_a < size_b;
/* struct or union is preferred */
if (is_compound_type(type_a) != is_compound_type(type_b))
return is_compound_type(type_b);
/* typedef is preferred */
if (dwarf_tag(type_b) == DW_TAG_typedef)
return true;
return false;
}
/* The type info will be saved in @type_die */
static enum type_match_result check_variable(struct data_loc_info *dloc,
Dwarf_Die *var_die,
Dwarf_Die *type_die, int reg,
int offset, bool is_fbreg)
{
Dwarf_Word size;
bool needs_pointer = true;
perf annotate-data: Show typedef names properly The die_get_typename() would resolve typedef and get to the original type. But sometimes the original type is a struct without name and it makes the output confusing and hard to read. This is a diff of perf report -s type before and after the change. New types such as atomic{,64}_t and sigset_t appeared and the portion of unnamed struct was reduced. Also u32, u64 and size_t were splitted from the base types. --- b 2024-08-01 17:02:34.307809952 -0700 +++ a 2024-08-07 14:17:05.245853999 -0700 - 2.40% long unsigned int + 2.26% long unsigned int - 1.56% unsigned int + 1.27% unsigned int - 0.98% struct - 0.79% long long unsigned int + 0.58% long long unsigned int + 0.36% struct + 0.27% atomic64_t + 0.22% u32 + 0.21% u64 + 0.19% atomic_t + 0.13% size_t - 0.08% struct seqcount_spinlock + 0.08% seqcount_spinlock_t + 0.08% sigset_t + 0.08% __poll_t Let's use the typedef name directly and the resolved to get the size of the type. Committer testing: root@x1:~# diff -u before after | head -30 --- before 2024-08-08 09:35:13.917325041 -0300 +++ after 2024-08-08 09:37:35.312257905 -0300 @@ -10,25 +10,27 @@ # ........ ......... # 79.40% (unknown) - 2.28% union 1.96% (stack operation) - 1.24% struct + 1.87% pthread_mutex_t 0.99% u32[] - 0.92% unsigned int 0.77% struct task_struct + 0.75% U32 0.75% struct pcpu_hot 0.63% struct qspinlock + 0.61% atomic_t 0.59% struct list_head - 0.58% int 0.53% struct cfs_rq 0.51% BYTE* - 0.48% unsigned char + 0.48% BYTE 0.48% long unsigned int 0.46% struct rq 0.41% struct worker 0.41% struct memcg_vmstats_percpu + 0.41% pthread_cond_t 0.37% _Bool + 0.36% int root@x1:~# Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240807223129.1738004-1-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-07 15:31:29 -07:00
Dwarf_Die sized_type;
perf annotate-data: Handle RSP if it's not the FB register In some cases, the stack pointer on x86 (rsp = reg7) is used to point variables on stack but it's not the frame base register. Then it should handle the register like normal registers (IOW not to access the other stack variables using offset calculation) but it should not assume it would have a pointer. Before: ----------------------------------------------------------- find data type for 0x7c(reg7) at tcp_getsockopt+0xb62 CU for net/ipv4/tcp.c (die:0x7b5f516) frame base: cfa=0 fbreg=6 no pointer or no type check variable "zc" failed (die: 0x7b9580a) variable location: base=reg7, offset=0x40 type='struct tcp_zerocopy_receive' size=0x40 (die:0x7b947f4) After: ----------------------------------------------------------- find data type for 0x7c(reg7) at tcp_getsockopt+0xb62 CU for net/ipv4/tcp.c (die:0x7b5f516) frame base: cfa=0 fbreg=6 found "zc" in scope=3/3 (die: 0x7b957fc) type_offset=0x3c variable location: base=reg7, offset=0x40 type='struct tcp_zerocopy_receive' size=0x40 (die:0x7b947f4) Note that the type-offset was properly calculated to 0x3c as the variable starts at 0x40. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240412183310.2518474-5-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-04-12 11:33:10 -07:00
if (reg == DWARF_REG_PC)
needs_pointer = false;
perf annotate-data: Handle RSP if it's not the FB register In some cases, the stack pointer on x86 (rsp = reg7) is used to point variables on stack but it's not the frame base register. Then it should handle the register like normal registers (IOW not to access the other stack variables using offset calculation) but it should not assume it would have a pointer. Before: ----------------------------------------------------------- find data type for 0x7c(reg7) at tcp_getsockopt+0xb62 CU for net/ipv4/tcp.c (die:0x7b5f516) frame base: cfa=0 fbreg=6 no pointer or no type check variable "zc" failed (die: 0x7b9580a) variable location: base=reg7, offset=0x40 type='struct tcp_zerocopy_receive' size=0x40 (die:0x7b947f4) After: ----------------------------------------------------------- find data type for 0x7c(reg7) at tcp_getsockopt+0xb62 CU for net/ipv4/tcp.c (die:0x7b5f516) frame base: cfa=0 fbreg=6 found "zc" in scope=3/3 (die: 0x7b957fc) type_offset=0x3c variable location: base=reg7, offset=0x40 type='struct tcp_zerocopy_receive' size=0x40 (die:0x7b947f4) Note that the type-offset was properly calculated to 0x3c as the variable starts at 0x40. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240412183310.2518474-5-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-04-12 11:33:10 -07:00
else if (reg == dloc->fbreg || is_fbreg)
needs_pointer = false;
perf annotate-data: Handle RSP if it's not the FB register In some cases, the stack pointer on x86 (rsp = reg7) is used to point variables on stack but it's not the frame base register. Then it should handle the register like normal registers (IOW not to access the other stack variables using offset calculation) but it should not assume it would have a pointer. Before: ----------------------------------------------------------- find data type for 0x7c(reg7) at tcp_getsockopt+0xb62 CU for net/ipv4/tcp.c (die:0x7b5f516) frame base: cfa=0 fbreg=6 no pointer or no type check variable "zc" failed (die: 0x7b9580a) variable location: base=reg7, offset=0x40 type='struct tcp_zerocopy_receive' size=0x40 (die:0x7b947f4) After: ----------------------------------------------------------- find data type for 0x7c(reg7) at tcp_getsockopt+0xb62 CU for net/ipv4/tcp.c (die:0x7b5f516) frame base: cfa=0 fbreg=6 found "zc" in scope=3/3 (die: 0x7b957fc) type_offset=0x3c variable location: base=reg7, offset=0x40 type='struct tcp_zerocopy_receive' size=0x40 (die:0x7b947f4) Note that the type-offset was properly calculated to 0x3c as the variable starts at 0x40. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240412183310.2518474-5-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-04-12 11:33:10 -07:00
else if (arch__is(dloc->arch, "x86") && reg == X86_REG_SP)
needs_pointer = false;
/* Get the type of the variable */
if (__die_get_real_type(var_die, type_die) == NULL)
return PERF_TMR_NO_TYPE;
/*
* Usually it expects a pointer type for a memory access.
* Convert to a real type it points to. But global variables
perf annotate-data: Support stack variables Local variables are allocated in the stack and the location list should look like base register(s) and an offset. Extend the die_find_variable_by_reg() to handle the following expressions * DW_OP_breg{0..31} * DW_OP_bregx * DW_OP_fbreg Ususally DWARF subprogram entries have frame base information and use it to locate stack variable like below: <2><43d1575>: Abbrev Number: 62 (DW_TAG_variable) <43d1576> DW_AT_location : 2 byte block: 91 7c (DW_OP_fbreg: -4) <--- here <43d1579> DW_AT_name : (indirect string, offset: 0x2c00c9): i <43d157d> DW_AT_decl_file : 1 <43d157e> DW_AT_decl_line : 78 <43d157f> DW_AT_type : <0x43d19d7> I found some differences on saving the frame base between gcc and clang. The gcc uses the CFA to get the base so it needs to check the current frame's CFI info. In this case, stack offset needs to be adjusted from the start of the CFA. <1><1bb8d>: Abbrev Number: 102 (DW_TAG_subprogram) <1bb8e> DW_AT_name : (indirect string, offset: 0x74d41): kernel_init <1bb92> DW_AT_decl_file : 2 <1bb92> DW_AT_decl_line : 1440 <1bb94> DW_AT_decl_column : 18 <1bb95> DW_AT_prototyped : 1 <1bb95> DW_AT_type : <0xcc> <1bb99> DW_AT_low_pc : 0xffffffff81bab9e0 <1bba1> DW_AT_high_pc : 0x1b2 <1bba9> DW_AT_frame_base : 1 byte block: 9c (DW_OP_call_frame_cfa) <------ here <1bbab> DW_AT_call_all_calls: 1 <1bbab> DW_AT_sibling : <0x1bf5a> While clang sets it to a register directly and it can check the register and offset in the instruction directly. <1><43d1542>: Abbrev Number: 60 (DW_TAG_subprogram) <43d1543> DW_AT_low_pc : 0xffffffff816a7c60 <43d154b> DW_AT_high_pc : 0x98 <43d154f> DW_AT_frame_base : 1 byte block: 56 (DW_OP_reg6 (rbp)) <---------- here <43d1551> DW_AT_GNU_all_call_sites: 1 <43d1551> DW_AT_name : (indirect string, offset: 0x3bce91): foo <43d1555> DW_AT_decl_file : 1 <43d1556> DW_AT_decl_line : 75 <43d1557> DW_AT_prototyped : 1 <43d1557> DW_AT_type : <0x43c7332> <43d155b> DW_AT_external : 1 Also it needs to update the offset after finding the type like global variables since the offset was from the frame base. Factor out match_var_offset() to check global and local variables in the same way. The type stats are improved too: Annotate data type stats: total 294, ok 160 (54.4%), bad 134 (45.6%) ----------------------------------------------------------- 30 : no_sym 32 : no_mem_ops 51 : no_var 14 : no_typeinfo 7 : bad_offset Reviewed-by: Ian Rogers <irogers@google.com> Cc: Stephane Eranian <eranian@google.com> Cc: Masami Hiramatsu <mhiramat@kernel.org> Link: https://lore.kernel.org/r/20240117062657.985479-9-namhyung@kernel.org Signed-off-by: Namhyung Kim <namhyung@kernel.org>
2024-01-16 22:26:56 -08:00
* and local variables are accessed directly without a pointer.
*/
if (needs_pointer) {
if (!is_pointer_type(type_die) ||
__die_get_real_type(type_die, type_die) == NULL)
return PERF_TMR_NO_POINTER;
}
perf annotate-data: Show typedef names properly The die_get_typename() would resolve typedef and get to the original type. But sometimes the original type is a struct without name and it makes the output confusing and hard to read. This is a diff of perf report -s type before and after the change. New types such as atomic{,64}_t and sigset_t appeared and the portion of unnamed struct was reduced. Also u32, u64 and size_t were splitted from the base types. --- b 2024-08-01 17:02:34.307809952 -0700 +++ a 2024-08-07 14:17:05.245853999 -0700 - 2.40% long unsigned int + 2.26% long unsigned int - 1.56% unsigned int + 1.27% unsigned int - 0.98% struct - 0.79% long long unsigned int + 0.58% long long unsigned int + 0.36% struct + 0.27% atomic64_t + 0.22% u32 + 0.21% u64 + 0.19% atomic_t + 0.13% size_t - 0.08% struct seqcount_spinlock + 0.08% seqcount_spinlock_t + 0.08% sigset_t + 0.08% __poll_t Let's use the typedef name directly and the resolved to get the size of the type. Committer testing: root@x1:~# diff -u before after | head -30 --- before 2024-08-08 09:35:13.917325041 -0300 +++ after 2024-08-08 09:37:35.312257905 -0300 @@ -10,25 +10,27 @@ # ........ ......... # 79.40% (unknown) - 2.28% union 1.96% (stack operation) - 1.24% struct + 1.87% pthread_mutex_t 0.99% u32[] - 0.92% unsigned int 0.77% struct task_struct + 0.75% U32 0.75% struct pcpu_hot 0.63% struct qspinlock + 0.61% atomic_t 0.59% struct list_head - 0.58% int 0.53% struct cfs_rq 0.51% BYTE* - 0.48% unsigned char + 0.48% BYTE 0.48% long unsigned int 0.46% struct rq 0.41% struct worker 0.41% struct memcg_vmstats_percpu + 0.41% pthread_cond_t 0.37% _Bool + 0.36% int root@x1:~# Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240807223129.1738004-1-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-07 15:31:29 -07:00
if (dwarf_tag(type_die) == DW_TAG_typedef)
die_get_real_type(type_die, &sized_type);
else
sized_type = *type_die;
/* Get the size of the actual type */
if (dwarf_aggregate_size(&sized_type, &size) < 0)
return PERF_TMR_NO_SIZE;
/* Minimal sanity check */
if ((unsigned)offset >= size)
return PERF_TMR_BAD_OFFSET;
return PERF_TMR_OK;
}
struct type_state_stack *find_stack_state(struct type_state *state,
int offset)
{
struct type_state_stack *stack;
list_for_each_entry(stack, &state->stack_vars, list) {
if (offset == stack->offset)
return stack;
if (stack->compound && stack->offset < offset &&
offset < stack->offset + stack->size)
return stack;
}
return NULL;
}
void set_stack_state(struct type_state_stack *stack, int offset, u8 kind,
Dwarf_Die *type_die)
{
int tag;
Dwarf_Word size;
if (dwarf_aggregate_size(type_die, &size) < 0)
size = 0;
tag = dwarf_tag(type_die);
stack->type = *type_die;
stack->size = size;
stack->offset = offset;
perf annotate-data: Support general per-cpu access This is to support per-cpu variable access often without a matching DWARF entry. For some reason, I cannot find debug info of per-cpu variables sometimes. They have more complex pattern to calculate the address of per-cpu variables like below. 2b7d: mov -0x1e0(%rbp),%rax ; rax = cpu 2b84: mov -0x7da0f7e0(,%rax,8),%rcx ; rcx = __per_cpu_offset[cpu] * 2b8c: mov 0x34870(%rcx),%rax ; *(__per_cpu_offset[cpu] + 0x34870) Let's assume the rax register has a number for a CPU at 2b7d. The next instruction is to get the per-cpu offset' for that cpu. The offset -0x7da0f7e0 is 0xffffffff825f0820 in u64 which is the address of the '__per_cpu_offset' array in my system. So it'd get the actual offset of that CPU's per-cpu region and save it to the rcx register. Then, at 2b8c, accesses using rcx can be handled same as the global variable access. To handle this case, it should check if the offset of the instruction matches to the address of '__per_cpu_offset'. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-20-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:11 -07:00
stack->kind = kind;
switch (tag) {
case DW_TAG_structure_type:
case DW_TAG_union_type:
perf annotate-data: Handle ADD instructions There are different patterns for percpu variable access using a constant value added to the base.  2aeb:  mov    -0x7da0f7e0(,%rax,8),%r14 # r14 = __per_cpu_offset[cpu]  2af3:  mov    $0x34740,%rax # rax = address of runqueues * 2afa:  add    %rax,%r14 # r14 = &per_cpu(runqueues, cpu)  2bfd:  cmpl   $0x0,0x10(%r14) # cpu_rq(cpu)->has_blocked_load  2b03:  je     0x2b36 At the first instruction, r14 has the __per_cpu_offset. And then rax has an immediate value and then added to r14 to calculate the address of a per-cpu variable. So it needs to track the immediate values and ADD instructions. Similar but a little different case is to use "this_cpu_off" instead of "__per_cpu_offset" for the current CPU. This time the variable address comes with PC-rel addressing. 89: mov $0x34740,%rax # rax = address of runqueues * 90: add %gs:0x7f015f60(%rip),%rax # 19a78 <this_cpu_off> 98: incl 0xd8c(%rax) # cpu_rq(cpu)->sched_count Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-21-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:12 -07:00
stack->compound = (kind != TSR_KIND_POINTER);
break;
default:
stack->compound = false;
break;
}
}
struct type_state_stack *findnew_stack_state(struct type_state *state,
perf annotate-data: Support general per-cpu access This is to support per-cpu variable access often without a matching DWARF entry. For some reason, I cannot find debug info of per-cpu variables sometimes. They have more complex pattern to calculate the address of per-cpu variables like below. 2b7d: mov -0x1e0(%rbp),%rax ; rax = cpu 2b84: mov -0x7da0f7e0(,%rax,8),%rcx ; rcx = __per_cpu_offset[cpu] * 2b8c: mov 0x34870(%rcx),%rax ; *(__per_cpu_offset[cpu] + 0x34870) Let's assume the rax register has a number for a CPU at 2b7d. The next instruction is to get the per-cpu offset' for that cpu. The offset -0x7da0f7e0 is 0xffffffff825f0820 in u64 which is the address of the '__per_cpu_offset' array in my system. So it'd get the actual offset of that CPU's per-cpu region and save it to the rcx register. Then, at 2b8c, accesses using rcx can be handled same as the global variable access. To handle this case, it should check if the offset of the instruction matches to the address of '__per_cpu_offset'. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-20-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:11 -07:00
int offset, u8 kind,
Dwarf_Die *type_die)
{
struct type_state_stack *stack = find_stack_state(state, offset);
if (stack) {
perf annotate-data: Support general per-cpu access This is to support per-cpu variable access often without a matching DWARF entry. For some reason, I cannot find debug info of per-cpu variables sometimes. They have more complex pattern to calculate the address of per-cpu variables like below. 2b7d: mov -0x1e0(%rbp),%rax ; rax = cpu 2b84: mov -0x7da0f7e0(,%rax,8),%rcx ; rcx = __per_cpu_offset[cpu] * 2b8c: mov 0x34870(%rcx),%rax ; *(__per_cpu_offset[cpu] + 0x34870) Let's assume the rax register has a number for a CPU at 2b7d. The next instruction is to get the per-cpu offset' for that cpu. The offset -0x7da0f7e0 is 0xffffffff825f0820 in u64 which is the address of the '__per_cpu_offset' array in my system. So it'd get the actual offset of that CPU's per-cpu region and save it to the rcx register. Then, at 2b8c, accesses using rcx can be handled same as the global variable access. To handle this case, it should check if the offset of the instruction matches to the address of '__per_cpu_offset'. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-20-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:11 -07:00
set_stack_state(stack, offset, kind, type_die);
return stack;
}
stack = malloc(sizeof(*stack));
if (stack) {
perf annotate-data: Support general per-cpu access This is to support per-cpu variable access often without a matching DWARF entry. For some reason, I cannot find debug info of per-cpu variables sometimes. They have more complex pattern to calculate the address of per-cpu variables like below. 2b7d: mov -0x1e0(%rbp),%rax ; rax = cpu 2b84: mov -0x7da0f7e0(,%rax,8),%rcx ; rcx = __per_cpu_offset[cpu] * 2b8c: mov 0x34870(%rcx),%rax ; *(__per_cpu_offset[cpu] + 0x34870) Let's assume the rax register has a number for a CPU at 2b7d. The next instruction is to get the per-cpu offset' for that cpu. The offset -0x7da0f7e0 is 0xffffffff825f0820 in u64 which is the address of the '__per_cpu_offset' array in my system. So it'd get the actual offset of that CPU's per-cpu region and save it to the rcx register. Then, at 2b8c, accesses using rcx can be handled same as the global variable access. To handle this case, it should check if the offset of the instruction matches to the address of '__per_cpu_offset'. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-20-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:11 -07:00
set_stack_state(stack, offset, kind, type_die);
list_add(&stack->list, &state->stack_vars);
}
return stack;
}
/* Maintain a cache for quick global variable lookup */
struct global_var_entry {
struct rb_node node;
char *name;
u64 start;
u64 end;
u64 die_offset;
};
static int global_var_cmp(const void *_key, const struct rb_node *node)
{
const u64 addr = (uintptr_t)_key;
struct global_var_entry *gvar;
gvar = rb_entry(node, struct global_var_entry, node);
if (gvar->start <= addr && addr < gvar->end)
return 0;
return gvar->start > addr ? -1 : 1;
}
static bool global_var_less(struct rb_node *node_a, const struct rb_node *node_b)
{
struct global_var_entry *gvar_a, *gvar_b;
gvar_a = rb_entry(node_a, struct global_var_entry, node);
gvar_b = rb_entry(node_b, struct global_var_entry, node);
return gvar_a->start < gvar_b->start;
}
static struct global_var_entry *global_var__find(struct data_loc_info *dloc, u64 addr)
{
struct dso *dso = map__dso(dloc->ms->map);
struct rb_node *node;
perf dso: Add reference count checking and accessor functions Add reference count checking to struct dso, this can help with implementing correct reference counting discipline. To avoid RC_CHK_ACCESS everywhere, add accessor functions for the variables in struct dso. The majority of the change is mechanical in nature and not easy to split up. Committer testing: 'perf test' up to this patch shows no regressions. But: util/symbol.c: In function ‘dso__load_bfd_symbols’: util/symbol.c:1683:9: error: too few arguments to function ‘dso__set_adjust_symbols’ 1683 | dso__set_adjust_symbols(dso); | ^~~~~~~~~~~~~~~~~~~~~~~ In file included from util/symbol.c:21: util/dso.h:268:20: note: declared here 268 | static inline void dso__set_adjust_symbols(struct dso *dso, bool val) | ^~~~~~~~~~~~~~~~~~~~~~~ make[6]: *** [/home/acme/git/perf-tools-next/tools/build/Makefile.build:106: /tmp/tmp.ZWHbQftdN6/util/symbol.o] Error 1 MKDIR /tmp/tmp.ZWHbQftdN6/tests/workloads/ make[6]: *** Waiting for unfinished jobs.... This was updated: - symbols__fixup_end(&dso->symbols, false); - symbols__fixup_duplicate(&dso->symbols); - dso->adjust_symbols = 1; + symbols__fixup_end(dso__symbols(dso), false); + symbols__fixup_duplicate(dso__symbols(dso)); + dso__set_adjust_symbols(dso); But not build tested with BUILD_NONDISTRO and libbfd devel files installed (binutils-devel on fedora). Add the missing argument: symbols__fixup_end(dso__symbols(dso), false); symbols__fixup_duplicate(dso__symbols(dso)); - dso__set_adjust_symbols(dso); + dso__set_adjust_symbols(dso, true); Signed-off-by: Ian Rogers <irogers@google.com> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Andi Kleen <ak@linux.intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ben Gainey <ben.gainey@arm.com> Cc: Changbin Du <changbin.du@huawei.com> Cc: Chengen Du <chengen.du@canonical.com> Cc: Colin Ian King <colin.i.king@gmail.com> Cc: Dima Kogan <dima@secretsauce.net> Cc: Ilkka Koskinen <ilkka@os.amperecomputing.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: James Clark <james.clark@arm.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: K Prateek Nayak <kprateek.nayak@amd.com> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Leo Yan <leo.yan@linux.dev> Cc: Li Dong <lidong@vivo.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Paran Lee <p4ranlee@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Song Liu <song@kernel.org> Cc: Sun Haiyong <sunhaiyong@loongson.cn> Cc: Thomas Richter <tmricht@linux.ibm.com> Cc: Tiezhu Yang <yangtiezhu@loongson.cn> Cc: Yanteng Si <siyanteng@loongson.cn> Cc: zhaimingbing <zhaimingbing@cmss.chinamobile.com> Link: https://lore.kernel.org/r/20240504213803.218974-6-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-05-04 14:38:01 -07:00
node = rb_find((void *)(uintptr_t)addr, dso__global_vars(dso), global_var_cmp);
if (node == NULL)
return NULL;
return rb_entry(node, struct global_var_entry, node);
}
static bool global_var__add(struct data_loc_info *dloc, u64 addr,
const char *name, Dwarf_Die *type_die)
{
struct dso *dso = map__dso(dloc->ms->map);
struct global_var_entry *gvar;
Dwarf_Word size;
if (dwarf_aggregate_size(type_die, &size) < 0)
return false;
gvar = malloc(sizeof(*gvar));
if (gvar == NULL)
return false;
gvar->name = name ? strdup(name) : NULL;
if (name && gvar->name == NULL) {
free(gvar);
return false;
}
gvar->start = addr;
gvar->end = addr + size;
gvar->die_offset = dwarf_dieoffset(type_die);
perf dso: Add reference count checking and accessor functions Add reference count checking to struct dso, this can help with implementing correct reference counting discipline. To avoid RC_CHK_ACCESS everywhere, add accessor functions for the variables in struct dso. The majority of the change is mechanical in nature and not easy to split up. Committer testing: 'perf test' up to this patch shows no regressions. But: util/symbol.c: In function ‘dso__load_bfd_symbols’: util/symbol.c:1683:9: error: too few arguments to function ‘dso__set_adjust_symbols’ 1683 | dso__set_adjust_symbols(dso); | ^~~~~~~~~~~~~~~~~~~~~~~ In file included from util/symbol.c:21: util/dso.h:268:20: note: declared here 268 | static inline void dso__set_adjust_symbols(struct dso *dso, bool val) | ^~~~~~~~~~~~~~~~~~~~~~~ make[6]: *** [/home/acme/git/perf-tools-next/tools/build/Makefile.build:106: /tmp/tmp.ZWHbQftdN6/util/symbol.o] Error 1 MKDIR /tmp/tmp.ZWHbQftdN6/tests/workloads/ make[6]: *** Waiting for unfinished jobs.... This was updated: - symbols__fixup_end(&dso->symbols, false); - symbols__fixup_duplicate(&dso->symbols); - dso->adjust_symbols = 1; + symbols__fixup_end(dso__symbols(dso), false); + symbols__fixup_duplicate(dso__symbols(dso)); + dso__set_adjust_symbols(dso); But not build tested with BUILD_NONDISTRO and libbfd devel files installed (binutils-devel on fedora). Add the missing argument: symbols__fixup_end(dso__symbols(dso), false); symbols__fixup_duplicate(dso__symbols(dso)); - dso__set_adjust_symbols(dso); + dso__set_adjust_symbols(dso, true); Signed-off-by: Ian Rogers <irogers@google.com> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Andi Kleen <ak@linux.intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ben Gainey <ben.gainey@arm.com> Cc: Changbin Du <changbin.du@huawei.com> Cc: Chengen Du <chengen.du@canonical.com> Cc: Colin Ian King <colin.i.king@gmail.com> Cc: Dima Kogan <dima@secretsauce.net> Cc: Ilkka Koskinen <ilkka@os.amperecomputing.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: James Clark <james.clark@arm.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: K Prateek Nayak <kprateek.nayak@amd.com> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Leo Yan <leo.yan@linux.dev> Cc: Li Dong <lidong@vivo.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Paran Lee <p4ranlee@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Song Liu <song@kernel.org> Cc: Sun Haiyong <sunhaiyong@loongson.cn> Cc: Thomas Richter <tmricht@linux.ibm.com> Cc: Tiezhu Yang <yangtiezhu@loongson.cn> Cc: Yanteng Si <siyanteng@loongson.cn> Cc: zhaimingbing <zhaimingbing@cmss.chinamobile.com> Link: https://lore.kernel.org/r/20240504213803.218974-6-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-05-04 14:38:01 -07:00
rb_add(&gvar->node, dso__global_vars(dso), global_var_less);
return true;
}
void global_var_type__tree_delete(struct rb_root *root)
{
struct global_var_entry *gvar;
while (!RB_EMPTY_ROOT(root)) {
struct rb_node *node = rb_first(root);
rb_erase(node, root);
gvar = rb_entry(node, struct global_var_entry, node);
zfree(&gvar->name);
free(gvar);
}
}
bool get_global_var_info(struct data_loc_info *dloc, u64 addr,
perf annotate-data: Support general per-cpu access This is to support per-cpu variable access often without a matching DWARF entry. For some reason, I cannot find debug info of per-cpu variables sometimes. They have more complex pattern to calculate the address of per-cpu variables like below. 2b7d: mov -0x1e0(%rbp),%rax ; rax = cpu 2b84: mov -0x7da0f7e0(,%rax,8),%rcx ; rcx = __per_cpu_offset[cpu] * 2b8c: mov 0x34870(%rcx),%rax ; *(__per_cpu_offset[cpu] + 0x34870) Let's assume the rax register has a number for a CPU at 2b7d. The next instruction is to get the per-cpu offset' for that cpu. The offset -0x7da0f7e0 is 0xffffffff825f0820 in u64 which is the address of the '__per_cpu_offset' array in my system. So it'd get the actual offset of that CPU's per-cpu region and save it to the rcx register. Then, at 2b8c, accesses using rcx can be handled same as the global variable access. To handle this case, it should check if the offset of the instruction matches to the address of '__per_cpu_offset'. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-20-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:11 -07:00
const char **var_name, int *var_offset)
{
struct addr_location al;
struct symbol *sym;
u64 mem_addr;
/* Kernel symbols might be relocated */
mem_addr = addr + map__reloc(dloc->ms->map);
addr_location__init(&al);
sym = thread__find_symbol_fb(dloc->thread, dloc->cpumode,
mem_addr, &al);
if (sym) {
*var_name = sym->name;
/* Calculate type offset from the start of variable */
*var_offset = mem_addr - map__unmap_ip(al.map, sym->start);
} else {
*var_name = NULL;
}
addr_location__exit(&al);
if (*var_name == NULL)
return false;
return true;
}
static void global_var__collect(struct data_loc_info *dloc)
{
Dwarf *dwarf = dloc->di->dbg;
Dwarf_Off off, next_off;
Dwarf_Die cu_die, type_die;
size_t header_size;
/* Iterate all CU and collect global variables that have no location in a register. */
off = 0;
while (dwarf_nextcu(dwarf, off, &next_off, &header_size,
NULL, NULL, NULL) == 0) {
struct die_var_type *var_types = NULL;
struct die_var_type *pos;
if (dwarf_offdie(dwarf, off + header_size, &cu_die) == NULL) {
off = next_off;
continue;
}
die_collect_global_vars(&cu_die, &var_types);
for (pos = var_types; pos; pos = pos->next) {
const char *var_name = NULL;
int var_offset = 0;
if (pos->reg != -1)
continue;
if (!dwarf_offdie(dwarf, pos->die_off, &type_die))
continue;
if (!get_global_var_info(dloc, pos->addr, &var_name,
&var_offset))
continue;
if (var_offset != 0)
continue;
global_var__add(dloc, pos->addr, var_name, &type_die);
}
delete_var_types(var_types);
off = next_off;
}
}
bool get_global_var_type(Dwarf_Die *cu_die, struct data_loc_info *dloc,
u64 ip, u64 var_addr, int *var_offset,
Dwarf_Die *type_die)
{
perf annotate-data: Support general per-cpu access This is to support per-cpu variable access often without a matching DWARF entry. For some reason, I cannot find debug info of per-cpu variables sometimes. They have more complex pattern to calculate the address of per-cpu variables like below. 2b7d: mov -0x1e0(%rbp),%rax ; rax = cpu 2b84: mov -0x7da0f7e0(,%rax,8),%rcx ; rcx = __per_cpu_offset[cpu] * 2b8c: mov 0x34870(%rcx),%rax ; *(__per_cpu_offset[cpu] + 0x34870) Let's assume the rax register has a number for a CPU at 2b7d. The next instruction is to get the per-cpu offset' for that cpu. The offset -0x7da0f7e0 is 0xffffffff825f0820 in u64 which is the address of the '__per_cpu_offset' array in my system. So it'd get the actual offset of that CPU's per-cpu region and save it to the rcx register. Then, at 2b8c, accesses using rcx can be handled same as the global variable access. To handle this case, it should check if the offset of the instruction matches to the address of '__per_cpu_offset'. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-20-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:11 -07:00
u64 pc;
int offset;
const char *var_name = NULL;
struct global_var_entry *gvar;
struct dso *dso = map__dso(dloc->ms->map);
Dwarf_Die var_die;
perf dso: Add reference count checking and accessor functions Add reference count checking to struct dso, this can help with implementing correct reference counting discipline. To avoid RC_CHK_ACCESS everywhere, add accessor functions for the variables in struct dso. The majority of the change is mechanical in nature and not easy to split up. Committer testing: 'perf test' up to this patch shows no regressions. But: util/symbol.c: In function ‘dso__load_bfd_symbols’: util/symbol.c:1683:9: error: too few arguments to function ‘dso__set_adjust_symbols’ 1683 | dso__set_adjust_symbols(dso); | ^~~~~~~~~~~~~~~~~~~~~~~ In file included from util/symbol.c:21: util/dso.h:268:20: note: declared here 268 | static inline void dso__set_adjust_symbols(struct dso *dso, bool val) | ^~~~~~~~~~~~~~~~~~~~~~~ make[6]: *** [/home/acme/git/perf-tools-next/tools/build/Makefile.build:106: /tmp/tmp.ZWHbQftdN6/util/symbol.o] Error 1 MKDIR /tmp/tmp.ZWHbQftdN6/tests/workloads/ make[6]: *** Waiting for unfinished jobs.... This was updated: - symbols__fixup_end(&dso->symbols, false); - symbols__fixup_duplicate(&dso->symbols); - dso->adjust_symbols = 1; + symbols__fixup_end(dso__symbols(dso), false); + symbols__fixup_duplicate(dso__symbols(dso)); + dso__set_adjust_symbols(dso); But not build tested with BUILD_NONDISTRO and libbfd devel files installed (binutils-devel on fedora). Add the missing argument: symbols__fixup_end(dso__symbols(dso), false); symbols__fixup_duplicate(dso__symbols(dso)); - dso__set_adjust_symbols(dso); + dso__set_adjust_symbols(dso, true); Signed-off-by: Ian Rogers <irogers@google.com> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Andi Kleen <ak@linux.intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ben Gainey <ben.gainey@arm.com> Cc: Changbin Du <changbin.du@huawei.com> Cc: Chengen Du <chengen.du@canonical.com> Cc: Colin Ian King <colin.i.king@gmail.com> Cc: Dima Kogan <dima@secretsauce.net> Cc: Ilkka Koskinen <ilkka@os.amperecomputing.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: James Clark <james.clark@arm.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: K Prateek Nayak <kprateek.nayak@amd.com> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Leo Yan <leo.yan@linux.dev> Cc: Li Dong <lidong@vivo.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Paran Lee <p4ranlee@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Song Liu <song@kernel.org> Cc: Sun Haiyong <sunhaiyong@loongson.cn> Cc: Thomas Richter <tmricht@linux.ibm.com> Cc: Tiezhu Yang <yangtiezhu@loongson.cn> Cc: Yanteng Si <siyanteng@loongson.cn> Cc: zhaimingbing <zhaimingbing@cmss.chinamobile.com> Link: https://lore.kernel.org/r/20240504213803.218974-6-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-05-04 14:38:01 -07:00
if (RB_EMPTY_ROOT(dso__global_vars(dso)))
global_var__collect(dloc);
gvar = global_var__find(dloc, var_addr);
if (gvar) {
if (!dwarf_offdie(dloc->di->dbg, gvar->die_offset, type_die))
return false;
*var_offset = var_addr - gvar->start;
return true;
}
/* Try to get the variable by address first */
if (die_find_variable_by_addr(cu_die, var_addr, &var_die, &offset) &&
perf annotate-data: Handle RSP if it's not the FB register In some cases, the stack pointer on x86 (rsp = reg7) is used to point variables on stack but it's not the frame base register. Then it should handle the register like normal registers (IOW not to access the other stack variables using offset calculation) but it should not assume it would have a pointer. Before: ----------------------------------------------------------- find data type for 0x7c(reg7) at tcp_getsockopt+0xb62 CU for net/ipv4/tcp.c (die:0x7b5f516) frame base: cfa=0 fbreg=6 no pointer or no type check variable "zc" failed (die: 0x7b9580a) variable location: base=reg7, offset=0x40 type='struct tcp_zerocopy_receive' size=0x40 (die:0x7b947f4) After: ----------------------------------------------------------- find data type for 0x7c(reg7) at tcp_getsockopt+0xb62 CU for net/ipv4/tcp.c (die:0x7b5f516) frame base: cfa=0 fbreg=6 found "zc" in scope=3/3 (die: 0x7b957fc) type_offset=0x3c variable location: base=reg7, offset=0x40 type='struct tcp_zerocopy_receive' size=0x40 (die:0x7b947f4) Note that the type-offset was properly calculated to 0x3c as the variable starts at 0x40. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240412183310.2518474-5-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-04-12 11:33:10 -07:00
check_variable(dloc, &var_die, type_die, DWARF_REG_PC, offset,
/*is_fbreg=*/false) == PERF_TMR_OK) {
var_name = dwarf_diename(&var_die);
*var_offset = offset;
goto ok;
}
perf annotate-data: Support general per-cpu access This is to support per-cpu variable access often without a matching DWARF entry. For some reason, I cannot find debug info of per-cpu variables sometimes. They have more complex pattern to calculate the address of per-cpu variables like below. 2b7d: mov -0x1e0(%rbp),%rax ; rax = cpu 2b84: mov -0x7da0f7e0(,%rax,8),%rcx ; rcx = __per_cpu_offset[cpu] * 2b8c: mov 0x34870(%rcx),%rax ; *(__per_cpu_offset[cpu] + 0x34870) Let's assume the rax register has a number for a CPU at 2b7d. The next instruction is to get the per-cpu offset' for that cpu. The offset -0x7da0f7e0 is 0xffffffff825f0820 in u64 which is the address of the '__per_cpu_offset' array in my system. So it'd get the actual offset of that CPU's per-cpu region and save it to the rcx register. Then, at 2b8c, accesses using rcx can be handled same as the global variable access. To handle this case, it should check if the offset of the instruction matches to the address of '__per_cpu_offset'. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-20-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:11 -07:00
if (!get_global_var_info(dloc, var_addr, &var_name, var_offset))
return false;
pc = map__rip_2objdump(dloc->ms->map, ip);
/* Try to get the name of global variable */
if (die_find_variable_at(cu_die, var_name, pc, &var_die) &&
perf annotate-data: Handle RSP if it's not the FB register In some cases, the stack pointer on x86 (rsp = reg7) is used to point variables on stack but it's not the frame base register. Then it should handle the register like normal registers (IOW not to access the other stack variables using offset calculation) but it should not assume it would have a pointer. Before: ----------------------------------------------------------- find data type for 0x7c(reg7) at tcp_getsockopt+0xb62 CU for net/ipv4/tcp.c (die:0x7b5f516) frame base: cfa=0 fbreg=6 no pointer or no type check variable "zc" failed (die: 0x7b9580a) variable location: base=reg7, offset=0x40 type='struct tcp_zerocopy_receive' size=0x40 (die:0x7b947f4) After: ----------------------------------------------------------- find data type for 0x7c(reg7) at tcp_getsockopt+0xb62 CU for net/ipv4/tcp.c (die:0x7b5f516) frame base: cfa=0 fbreg=6 found "zc" in scope=3/3 (die: 0x7b957fc) type_offset=0x3c variable location: base=reg7, offset=0x40 type='struct tcp_zerocopy_receive' size=0x40 (die:0x7b947f4) Note that the type-offset was properly calculated to 0x3c as the variable starts at 0x40. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240412183310.2518474-5-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-04-12 11:33:10 -07:00
check_variable(dloc, &var_die, type_die, DWARF_REG_PC, *var_offset,
/*is_fbreg=*/false) == PERF_TMR_OK)
goto ok;
return false;
ok:
/* The address should point to the start of the variable */
global_var__add(dloc, var_addr - *var_offset, var_name, type_die);
return true;
}
perf annotate-data: Copy back variable types after move In some cases, compilers don't set the location expression in DWARF precisely. For instance, it may assign a variable to a register after copying it from a different register. Then it should use the register for the new type but still uses the old register. This makes hard to track the type information properly. This is an example I found in __tcp_transmit_skb(). The first argument (sk) of this function is a pointer to sock and there's a variable (tp) for tcp_sock. static int __tcp_transmit_skb(struct sock *sk, struct sk_buff *skb, int clone_it, gfp_t gfp_mask, u32 rcv_nxt) { ... struct tcp_sock *tp; BUG_ON(!skb || !tcp_skb_pcount(skb)); tp = tcp_sk(sk); prior_wstamp = tp->tcp_wstamp_ns; tp->tcp_wstamp_ns = max(tp->tcp_wstamp_ns, tp->tcp_clock_cache); ... So it basically calls tcp_sk(sk) to get the tcp_sock pointer from sk. But it turned out to be the same value because tcp_sock embeds sock as the first member. The sk is located in reg5 (RDI) and tp is in reg3 (RBX). The offset of tcp_wstamp_ns is 0x748 and tcp_clock_cache is 0x750. So you need to use RBX (reg3) to access the fields in the tcp_sock. But the code used RDI (reg5) as it has the same value. $ pahole --hex -C tcp_sock vmlinux | grep -e 748 -e 750 u64 tcp_wstamp_ns; /* 0x748 0x8 */ u64 tcp_clock_cache; /* 0x750 0x8 */ And this is the disassembly of the part of the function. <__tcp_transmit_skb>: ... 44: mov %rdi, %rbx 47: mov 0x748(%rdi), %rsi 4e: mov 0x750(%rdi), %rax 55: cmp %rax, %rsi Because compiler put the debug info to RBX, it only knows RDI is a pointer to sock and accessing those two fields resulted in error due to offset being beyond the type size. ----------------------------------------------------------- find data type for 0x748(reg5) at __tcp_transmit_skb+0x63 CU for net/ipv4/tcp_output.c (die:0x817f543) frame base: cfa=0 fbreg=6 scope: [1/1] (die:81aac3e) bb: [0 - 30] var [0] -0x98(stack) type='struct tcp_out_options' size=0x28 (die:0x81af3df) var [5] reg8 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg2 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg1 type='int' size=0x4 (die:0x818059e) var [5] reg4 type='struct sk_buff*' size=0x8 (die:0x8181360) var [5] reg5 type='struct sock*' size=0x8 (die:0x8181a0c) <<<--- the first argument ('sk' at %RDI) mov [19] reg8 -> -0xa8(stack) type='unsigned int' size=0x4 (die:0x8180ed6) mov [20] stack canary -> reg0 mov [29] reg0 -> -0x30(stack) stack canary bb: [36 - 3e] mov [36] reg4 -> reg15 type='struct sk_buff*' size=0x8 (die:0x8181360) bb: [44 - 63] mov [44] reg5 -> reg3 type='struct sock*' size=0x8 (die:0x8181a0c) <<<--- calling tcp_sk() var [47] reg3 type='struct tcp_sock*' size=0x8 (die:0x819eead) <<<--- new variable ('tp' at %RBX) var [4e] reg4 type='unsigned long long' size=0x8 (die:0x8180edd) mov [58] reg4 -> -0xc0(stack) type='unsigned long long' size=0x8 (die:0x8180edd) chk [63] reg5 offset=0x748 ok=1 kind=1 (struct sock*) : offset bigger than size <<<--- access with old variable final result: offset bigger than size While it's a fault in the compiler, we could work around this issue by using the type of new variable when it's copied directly. So I've added copied_from field in the register state to track those direct register to register copies. After that new register gets a new type and the old register still has the same type, it'll update (copy it back) the type of the old register. For example, if we can update type of reg5 at __tcp_transmit_skb+0x47, we can find the target type of the instruction at 0x63 like below: ----------------------------------------------------------- find data type for 0x748(reg5) at __tcp_transmit_skb+0x63 ... bb: [44 - 63] mov [44] reg5 -> reg3 type='struct sock*' size=0x8 (die:0x8181a0c) var [47] reg3 type='struct tcp_sock*' size=0x8 (die:0x819eead) var [47] copyback reg5 type='struct tcp_sock*' size=0x8 (die:0x819eead) <<<--- here mov [47] 0x748(reg5) -> reg4 type='unsigned long long' size=0x8 (die:0x8180edd) mov [4e] 0x750(reg5) -> reg0 type='unsigned long long' size=0x8 (die:0x8180edd) mov [58] reg4 -> -0xc0(stack) type='unsigned long long' size=0x8 (die:0x8180edd) chk [63] reg5 offset=0x748 ok=1 kind=1 (struct tcp_sock*) : Good! <<<--- new type found by insn track: 0x748(reg5) type-offset=0x748 final result: type='struct tcp_sock' size=0xa98 (die:0x819eeb2) Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240821232628.353177-5-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-21 16:26:28 -07:00
static bool die_is_same(Dwarf_Die *die_a, Dwarf_Die *die_b)
{
return (die_a->cu == die_b->cu) && (die_a->addr == die_b->addr);
}
/**
* update_var_state - Update type state using given variables
* @state: type state table
* @dloc: data location info
* @addr: instruction address to match with variable
* @insn_offset: instruction offset (for debug)
* @var_types: list of variables with type info
*
* This function fills the @state table using @var_types info. Each variable
* is used only at the given location and updates an entry in the table.
*/
perf annotate-data: Implement instruction tracking If it failed to find a variable for the location directly, it might be due to a missing variable in the source code. For example, accessing pointer variables in a chain can result in the case like below: struct foo *foo = ...; int i = foo->bar->baz; The DWARF debug information is created for each variable so it'd have one for 'foo'. But there's no variable for 'foo->bar' and then it cannot know the type of 'bar' and 'baz'. The above source code can be compiled to the follow x86 instructions: mov 0x8(%rax), %rcx mov 0x4(%rcx), %rdx <=== PMU sample mov %rdx, -4(%rbp) Let's say 'foo' is located in the %rax and it has a pointer to struct foo. But perf sample is captured in the second instruction and there is no variable or type info for the %rcx. It'd be great if compiler could generate debug info for %rcx, but we should handle it on our side. So this patch implements the logic to iterate instructions and update the type table for each location. As it already collected a list of scopes including the target instruction, we can use it to construct the type table smartly. +---------------- scope[0] subprogram | | +-------------- scope[1] lexical_block | | | | +------------ scope[2] inlined_subroutine | | | | | | +---------- scope[3] inlined_subroutine | | | | | | | | +-------- scope[4] lexical_block | | | | | | | | | | *** target instruction ... Image the target instruction has 5 scopes, each scope will have its own variables and parameters. Then it can start with the innermost scope (4). So it'd search the shortest path from the start of scope[4] to the target address and build a list of basic blocks. Then it iterates the basic blocks with the variables in the scope and update the table. If it finds a type at the target instruction, then returns it. Otherwise, it moves to the upper scope[3]. Now it'd search the shortest path from the start of scope[3] to the start of scope[4]. Then connect it to the existing basic block list. Then it'd iterate the blocks with variables for both scopes. It can repeat this until it finds a type at the target instruction or reaches to the top scope[0]. As the basic blocks contain the shortest path, it won't worry about branches and can update the table simply. The final check will be done by find_matching_type() in the next patch. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-15-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:06 -07:00
static void update_var_state(struct type_state *state, struct data_loc_info *dloc,
u64 addr, u64 insn_offset, struct die_var_type *var_types)
{
Dwarf_Die mem_die;
struct die_var_type *var;
int fbreg = dloc->fbreg;
int fb_offset = 0;
if (dloc->fb_cfa) {
if (die_get_cfa(dloc->di->dbg, addr, &fbreg, &fb_offset) < 0)
fbreg = -1;
}
for (var = var_types; var != NULL; var = var->next) {
if (var->addr != addr)
continue;
/* Get the type DIE using the offset */
if (!dwarf_offdie(dloc->di->dbg, var->die_off, &mem_die))
continue;
if (var->reg == DWARF_REG_FB || var->reg == fbreg) {
int offset = var->offset;
struct type_state_stack *stack;
if (var->reg != DWARF_REG_FB)
offset -= fb_offset;
stack = find_stack_state(state, offset);
if (stack && stack->kind == TSR_KIND_TYPE &&
!is_better_type(&stack->type, &mem_die))
continue;
findnew_stack_state(state, offset, TSR_KIND_TYPE,
&mem_die);
pr_debug_dtp("var [%"PRIx64"] -%#x(stack)",
insn_offset, -offset);
perf annotate-data: Support general per-cpu access This is to support per-cpu variable access often without a matching DWARF entry. For some reason, I cannot find debug info of per-cpu variables sometimes. They have more complex pattern to calculate the address of per-cpu variables like below. 2b7d: mov -0x1e0(%rbp),%rax ; rax = cpu 2b84: mov -0x7da0f7e0(,%rax,8),%rcx ; rcx = __per_cpu_offset[cpu] * 2b8c: mov 0x34870(%rcx),%rax ; *(__per_cpu_offset[cpu] + 0x34870) Let's assume the rax register has a number for a CPU at 2b7d. The next instruction is to get the per-cpu offset' for that cpu. The offset -0x7da0f7e0 is 0xffffffff825f0820 in u64 which is the address of the '__per_cpu_offset' array in my system. So it'd get the actual offset of that CPU's per-cpu region and save it to the rcx register. Then, at 2b8c, accesses using rcx can be handled same as the global variable access. To handle this case, it should check if the offset of the instruction matches to the address of '__per_cpu_offset'. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-20-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:11 -07:00
pr_debug_type_name(&mem_die, TSR_KIND_TYPE);
} else if (has_reg_type(state, var->reg) && var->offset == 0) {
struct type_state_reg *reg;
perf annotate-data: Copy back variable types after move In some cases, compilers don't set the location expression in DWARF precisely. For instance, it may assign a variable to a register after copying it from a different register. Then it should use the register for the new type but still uses the old register. This makes hard to track the type information properly. This is an example I found in __tcp_transmit_skb(). The first argument (sk) of this function is a pointer to sock and there's a variable (tp) for tcp_sock. static int __tcp_transmit_skb(struct sock *sk, struct sk_buff *skb, int clone_it, gfp_t gfp_mask, u32 rcv_nxt) { ... struct tcp_sock *tp; BUG_ON(!skb || !tcp_skb_pcount(skb)); tp = tcp_sk(sk); prior_wstamp = tp->tcp_wstamp_ns; tp->tcp_wstamp_ns = max(tp->tcp_wstamp_ns, tp->tcp_clock_cache); ... So it basically calls tcp_sk(sk) to get the tcp_sock pointer from sk. But it turned out to be the same value because tcp_sock embeds sock as the first member. The sk is located in reg5 (RDI) and tp is in reg3 (RBX). The offset of tcp_wstamp_ns is 0x748 and tcp_clock_cache is 0x750. So you need to use RBX (reg3) to access the fields in the tcp_sock. But the code used RDI (reg5) as it has the same value. $ pahole --hex -C tcp_sock vmlinux | grep -e 748 -e 750 u64 tcp_wstamp_ns; /* 0x748 0x8 */ u64 tcp_clock_cache; /* 0x750 0x8 */ And this is the disassembly of the part of the function. <__tcp_transmit_skb>: ... 44: mov %rdi, %rbx 47: mov 0x748(%rdi), %rsi 4e: mov 0x750(%rdi), %rax 55: cmp %rax, %rsi Because compiler put the debug info to RBX, it only knows RDI is a pointer to sock and accessing those two fields resulted in error due to offset being beyond the type size. ----------------------------------------------------------- find data type for 0x748(reg5) at __tcp_transmit_skb+0x63 CU for net/ipv4/tcp_output.c (die:0x817f543) frame base: cfa=0 fbreg=6 scope: [1/1] (die:81aac3e) bb: [0 - 30] var [0] -0x98(stack) type='struct tcp_out_options' size=0x28 (die:0x81af3df) var [5] reg8 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg2 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg1 type='int' size=0x4 (die:0x818059e) var [5] reg4 type='struct sk_buff*' size=0x8 (die:0x8181360) var [5] reg5 type='struct sock*' size=0x8 (die:0x8181a0c) <<<--- the first argument ('sk' at %RDI) mov [19] reg8 -> -0xa8(stack) type='unsigned int' size=0x4 (die:0x8180ed6) mov [20] stack canary -> reg0 mov [29] reg0 -> -0x30(stack) stack canary bb: [36 - 3e] mov [36] reg4 -> reg15 type='struct sk_buff*' size=0x8 (die:0x8181360) bb: [44 - 63] mov [44] reg5 -> reg3 type='struct sock*' size=0x8 (die:0x8181a0c) <<<--- calling tcp_sk() var [47] reg3 type='struct tcp_sock*' size=0x8 (die:0x819eead) <<<--- new variable ('tp' at %RBX) var [4e] reg4 type='unsigned long long' size=0x8 (die:0x8180edd) mov [58] reg4 -> -0xc0(stack) type='unsigned long long' size=0x8 (die:0x8180edd) chk [63] reg5 offset=0x748 ok=1 kind=1 (struct sock*) : offset bigger than size <<<--- access with old variable final result: offset bigger than size While it's a fault in the compiler, we could work around this issue by using the type of new variable when it's copied directly. So I've added copied_from field in the register state to track those direct register to register copies. After that new register gets a new type and the old register still has the same type, it'll update (copy it back) the type of the old register. For example, if we can update type of reg5 at __tcp_transmit_skb+0x47, we can find the target type of the instruction at 0x63 like below: ----------------------------------------------------------- find data type for 0x748(reg5) at __tcp_transmit_skb+0x63 ... bb: [44 - 63] mov [44] reg5 -> reg3 type='struct sock*' size=0x8 (die:0x8181a0c) var [47] reg3 type='struct tcp_sock*' size=0x8 (die:0x819eead) var [47] copyback reg5 type='struct tcp_sock*' size=0x8 (die:0x819eead) <<<--- here mov [47] 0x748(reg5) -> reg4 type='unsigned long long' size=0x8 (die:0x8180edd) mov [4e] 0x750(reg5) -> reg0 type='unsigned long long' size=0x8 (die:0x8180edd) mov [58] reg4 -> -0xc0(stack) type='unsigned long long' size=0x8 (die:0x8180edd) chk [63] reg5 offset=0x748 ok=1 kind=1 (struct tcp_sock*) : Good! <<<--- new type found by insn track: 0x748(reg5) type-offset=0x748 final result: type='struct tcp_sock' size=0xa98 (die:0x819eeb2) Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240821232628.353177-5-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-21 16:26:28 -07:00
Dwarf_Die orig_type;
reg = &state->regs[var->reg];
if (reg->ok && reg->kind == TSR_KIND_TYPE &&
!is_better_type(&reg->type, &mem_die))
continue;
perf annotate-data: Copy back variable types after move In some cases, compilers don't set the location expression in DWARF precisely. For instance, it may assign a variable to a register after copying it from a different register. Then it should use the register for the new type but still uses the old register. This makes hard to track the type information properly. This is an example I found in __tcp_transmit_skb(). The first argument (sk) of this function is a pointer to sock and there's a variable (tp) for tcp_sock. static int __tcp_transmit_skb(struct sock *sk, struct sk_buff *skb, int clone_it, gfp_t gfp_mask, u32 rcv_nxt) { ... struct tcp_sock *tp; BUG_ON(!skb || !tcp_skb_pcount(skb)); tp = tcp_sk(sk); prior_wstamp = tp->tcp_wstamp_ns; tp->tcp_wstamp_ns = max(tp->tcp_wstamp_ns, tp->tcp_clock_cache); ... So it basically calls tcp_sk(sk) to get the tcp_sock pointer from sk. But it turned out to be the same value because tcp_sock embeds sock as the first member. The sk is located in reg5 (RDI) and tp is in reg3 (RBX). The offset of tcp_wstamp_ns is 0x748 and tcp_clock_cache is 0x750. So you need to use RBX (reg3) to access the fields in the tcp_sock. But the code used RDI (reg5) as it has the same value. $ pahole --hex -C tcp_sock vmlinux | grep -e 748 -e 750 u64 tcp_wstamp_ns; /* 0x748 0x8 */ u64 tcp_clock_cache; /* 0x750 0x8 */ And this is the disassembly of the part of the function. <__tcp_transmit_skb>: ... 44: mov %rdi, %rbx 47: mov 0x748(%rdi), %rsi 4e: mov 0x750(%rdi), %rax 55: cmp %rax, %rsi Because compiler put the debug info to RBX, it only knows RDI is a pointer to sock and accessing those two fields resulted in error due to offset being beyond the type size. ----------------------------------------------------------- find data type for 0x748(reg5) at __tcp_transmit_skb+0x63 CU for net/ipv4/tcp_output.c (die:0x817f543) frame base: cfa=0 fbreg=6 scope: [1/1] (die:81aac3e) bb: [0 - 30] var [0] -0x98(stack) type='struct tcp_out_options' size=0x28 (die:0x81af3df) var [5] reg8 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg2 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg1 type='int' size=0x4 (die:0x818059e) var [5] reg4 type='struct sk_buff*' size=0x8 (die:0x8181360) var [5] reg5 type='struct sock*' size=0x8 (die:0x8181a0c) <<<--- the first argument ('sk' at %RDI) mov [19] reg8 -> -0xa8(stack) type='unsigned int' size=0x4 (die:0x8180ed6) mov [20] stack canary -> reg0 mov [29] reg0 -> -0x30(stack) stack canary bb: [36 - 3e] mov [36] reg4 -> reg15 type='struct sk_buff*' size=0x8 (die:0x8181360) bb: [44 - 63] mov [44] reg5 -> reg3 type='struct sock*' size=0x8 (die:0x8181a0c) <<<--- calling tcp_sk() var [47] reg3 type='struct tcp_sock*' size=0x8 (die:0x819eead) <<<--- new variable ('tp' at %RBX) var [4e] reg4 type='unsigned long long' size=0x8 (die:0x8180edd) mov [58] reg4 -> -0xc0(stack) type='unsigned long long' size=0x8 (die:0x8180edd) chk [63] reg5 offset=0x748 ok=1 kind=1 (struct sock*) : offset bigger than size <<<--- access with old variable final result: offset bigger than size While it's a fault in the compiler, we could work around this issue by using the type of new variable when it's copied directly. So I've added copied_from field in the register state to track those direct register to register copies. After that new register gets a new type and the old register still has the same type, it'll update (copy it back) the type of the old register. For example, if we can update type of reg5 at __tcp_transmit_skb+0x47, we can find the target type of the instruction at 0x63 like below: ----------------------------------------------------------- find data type for 0x748(reg5) at __tcp_transmit_skb+0x63 ... bb: [44 - 63] mov [44] reg5 -> reg3 type='struct sock*' size=0x8 (die:0x8181a0c) var [47] reg3 type='struct tcp_sock*' size=0x8 (die:0x819eead) var [47] copyback reg5 type='struct tcp_sock*' size=0x8 (die:0x819eead) <<<--- here mov [47] 0x748(reg5) -> reg4 type='unsigned long long' size=0x8 (die:0x8180edd) mov [4e] 0x750(reg5) -> reg0 type='unsigned long long' size=0x8 (die:0x8180edd) mov [58] reg4 -> -0xc0(stack) type='unsigned long long' size=0x8 (die:0x8180edd) chk [63] reg5 offset=0x748 ok=1 kind=1 (struct tcp_sock*) : Good! <<<--- new type found by insn track: 0x748(reg5) type-offset=0x748 final result: type='struct tcp_sock' size=0xa98 (die:0x819eeb2) Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240821232628.353177-5-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-21 16:26:28 -07:00
orig_type = reg->type;
reg->type = mem_die;
perf annotate-data: Support general per-cpu access This is to support per-cpu variable access often without a matching DWARF entry. For some reason, I cannot find debug info of per-cpu variables sometimes. They have more complex pattern to calculate the address of per-cpu variables like below. 2b7d: mov -0x1e0(%rbp),%rax ; rax = cpu 2b84: mov -0x7da0f7e0(,%rax,8),%rcx ; rcx = __per_cpu_offset[cpu] * 2b8c: mov 0x34870(%rcx),%rax ; *(__per_cpu_offset[cpu] + 0x34870) Let's assume the rax register has a number for a CPU at 2b7d. The next instruction is to get the per-cpu offset' for that cpu. The offset -0x7da0f7e0 is 0xffffffff825f0820 in u64 which is the address of the '__per_cpu_offset' array in my system. So it'd get the actual offset of that CPU's per-cpu region and save it to the rcx register. Then, at 2b8c, accesses using rcx can be handled same as the global variable access. To handle this case, it should check if the offset of the instruction matches to the address of '__per_cpu_offset'. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-20-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:11 -07:00
reg->kind = TSR_KIND_TYPE;
reg->ok = true;
pr_debug_dtp("var [%"PRIx64"] reg%d",
insn_offset, var->reg);
perf annotate-data: Support general per-cpu access This is to support per-cpu variable access often without a matching DWARF entry. For some reason, I cannot find debug info of per-cpu variables sometimes. They have more complex pattern to calculate the address of per-cpu variables like below. 2b7d: mov -0x1e0(%rbp),%rax ; rax = cpu 2b84: mov -0x7da0f7e0(,%rax,8),%rcx ; rcx = __per_cpu_offset[cpu] * 2b8c: mov 0x34870(%rcx),%rax ; *(__per_cpu_offset[cpu] + 0x34870) Let's assume the rax register has a number for a CPU at 2b7d. The next instruction is to get the per-cpu offset' for that cpu. The offset -0x7da0f7e0 is 0xffffffff825f0820 in u64 which is the address of the '__per_cpu_offset' array in my system. So it'd get the actual offset of that CPU's per-cpu region and save it to the rcx register. Then, at 2b8c, accesses using rcx can be handled same as the global variable access. To handle this case, it should check if the offset of the instruction matches to the address of '__per_cpu_offset'. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-20-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:11 -07:00
pr_debug_type_name(&mem_die, TSR_KIND_TYPE);
perf annotate-data: Copy back variable types after move In some cases, compilers don't set the location expression in DWARF precisely. For instance, it may assign a variable to a register after copying it from a different register. Then it should use the register for the new type but still uses the old register. This makes hard to track the type information properly. This is an example I found in __tcp_transmit_skb(). The first argument (sk) of this function is a pointer to sock and there's a variable (tp) for tcp_sock. static int __tcp_transmit_skb(struct sock *sk, struct sk_buff *skb, int clone_it, gfp_t gfp_mask, u32 rcv_nxt) { ... struct tcp_sock *tp; BUG_ON(!skb || !tcp_skb_pcount(skb)); tp = tcp_sk(sk); prior_wstamp = tp->tcp_wstamp_ns; tp->tcp_wstamp_ns = max(tp->tcp_wstamp_ns, tp->tcp_clock_cache); ... So it basically calls tcp_sk(sk) to get the tcp_sock pointer from sk. But it turned out to be the same value because tcp_sock embeds sock as the first member. The sk is located in reg5 (RDI) and tp is in reg3 (RBX). The offset of tcp_wstamp_ns is 0x748 and tcp_clock_cache is 0x750. So you need to use RBX (reg3) to access the fields in the tcp_sock. But the code used RDI (reg5) as it has the same value. $ pahole --hex -C tcp_sock vmlinux | grep -e 748 -e 750 u64 tcp_wstamp_ns; /* 0x748 0x8 */ u64 tcp_clock_cache; /* 0x750 0x8 */ And this is the disassembly of the part of the function. <__tcp_transmit_skb>: ... 44: mov %rdi, %rbx 47: mov 0x748(%rdi), %rsi 4e: mov 0x750(%rdi), %rax 55: cmp %rax, %rsi Because compiler put the debug info to RBX, it only knows RDI is a pointer to sock and accessing those two fields resulted in error due to offset being beyond the type size. ----------------------------------------------------------- find data type for 0x748(reg5) at __tcp_transmit_skb+0x63 CU for net/ipv4/tcp_output.c (die:0x817f543) frame base: cfa=0 fbreg=6 scope: [1/1] (die:81aac3e) bb: [0 - 30] var [0] -0x98(stack) type='struct tcp_out_options' size=0x28 (die:0x81af3df) var [5] reg8 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg2 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg1 type='int' size=0x4 (die:0x818059e) var [5] reg4 type='struct sk_buff*' size=0x8 (die:0x8181360) var [5] reg5 type='struct sock*' size=0x8 (die:0x8181a0c) <<<--- the first argument ('sk' at %RDI) mov [19] reg8 -> -0xa8(stack) type='unsigned int' size=0x4 (die:0x8180ed6) mov [20] stack canary -> reg0 mov [29] reg0 -> -0x30(stack) stack canary bb: [36 - 3e] mov [36] reg4 -> reg15 type='struct sk_buff*' size=0x8 (die:0x8181360) bb: [44 - 63] mov [44] reg5 -> reg3 type='struct sock*' size=0x8 (die:0x8181a0c) <<<--- calling tcp_sk() var [47] reg3 type='struct tcp_sock*' size=0x8 (die:0x819eead) <<<--- new variable ('tp' at %RBX) var [4e] reg4 type='unsigned long long' size=0x8 (die:0x8180edd) mov [58] reg4 -> -0xc0(stack) type='unsigned long long' size=0x8 (die:0x8180edd) chk [63] reg5 offset=0x748 ok=1 kind=1 (struct sock*) : offset bigger than size <<<--- access with old variable final result: offset bigger than size While it's a fault in the compiler, we could work around this issue by using the type of new variable when it's copied directly. So I've added copied_from field in the register state to track those direct register to register copies. After that new register gets a new type and the old register still has the same type, it'll update (copy it back) the type of the old register. For example, if we can update type of reg5 at __tcp_transmit_skb+0x47, we can find the target type of the instruction at 0x63 like below: ----------------------------------------------------------- find data type for 0x748(reg5) at __tcp_transmit_skb+0x63 ... bb: [44 - 63] mov [44] reg5 -> reg3 type='struct sock*' size=0x8 (die:0x8181a0c) var [47] reg3 type='struct tcp_sock*' size=0x8 (die:0x819eead) var [47] copyback reg5 type='struct tcp_sock*' size=0x8 (die:0x819eead) <<<--- here mov [47] 0x748(reg5) -> reg4 type='unsigned long long' size=0x8 (die:0x8180edd) mov [4e] 0x750(reg5) -> reg0 type='unsigned long long' size=0x8 (die:0x8180edd) mov [58] reg4 -> -0xc0(stack) type='unsigned long long' size=0x8 (die:0x8180edd) chk [63] reg5 offset=0x748 ok=1 kind=1 (struct tcp_sock*) : Good! <<<--- new type found by insn track: 0x748(reg5) type-offset=0x748 final result: type='struct tcp_sock' size=0xa98 (die:0x819eeb2) Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240821232628.353177-5-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-21 16:26:28 -07:00
/*
* If this register is directly copied from another and it gets a
* better type, also update the type of the source register. This
* is usually the case of container_of() macro with offset of 0.
*/
if (has_reg_type(state, reg->copied_from)) {
struct type_state_reg *copy_reg;
copy_reg = &state->regs[reg->copied_from];
/* TODO: check if type is compatible or embedded */
if (!copy_reg->ok || (copy_reg->kind != TSR_KIND_TYPE) ||
!die_is_same(&copy_reg->type, &orig_type) ||
!is_better_type(&copy_reg->type, &mem_die))
continue;
copy_reg->type = mem_die;
pr_debug_dtp("var [%"PRIx64"] copyback reg%d",
insn_offset, reg->copied_from);
pr_debug_type_name(&mem_die, TSR_KIND_TYPE);
}
}
}
}
/**
* update_insn_state - Update type state for an instruction
* @state: type state table
* @dloc: data location info
* @cu_die: compile unit debug entry
* @dl: disasm line for the instruction
*
* This function updates the @state table for the target operand of the
* instruction at @dl if it transfers the type like MOV on x86. Since it
* tracks the type, it won't care about the values like in arithmetic
* instructions like ADD/SUB/MUL/DIV and INC/DEC.
*
* Note that ops->reg2 is only available when both mem_ref and multi_regs
* are true.
*/
perf annotate-data: Implement instruction tracking If it failed to find a variable for the location directly, it might be due to a missing variable in the source code. For example, accessing pointer variables in a chain can result in the case like below: struct foo *foo = ...; int i = foo->bar->baz; The DWARF debug information is created for each variable so it'd have one for 'foo'. But there's no variable for 'foo->bar' and then it cannot know the type of 'bar' and 'baz'. The above source code can be compiled to the follow x86 instructions: mov 0x8(%rax), %rcx mov 0x4(%rcx), %rdx <=== PMU sample mov %rdx, -4(%rbp) Let's say 'foo' is located in the %rax and it has a pointer to struct foo. But perf sample is captured in the second instruction and there is no variable or type info for the %rcx. It'd be great if compiler could generate debug info for %rcx, but we should handle it on our side. So this patch implements the logic to iterate instructions and update the type table for each location. As it already collected a list of scopes including the target instruction, we can use it to construct the type table smartly. +---------------- scope[0] subprogram | | +-------------- scope[1] lexical_block | | | | +------------ scope[2] inlined_subroutine | | | | | | +---------- scope[3] inlined_subroutine | | | | | | | | +-------- scope[4] lexical_block | | | | | | | | | | *** target instruction ... Image the target instruction has 5 scopes, each scope will have its own variables and parameters. Then it can start with the innermost scope (4). So it'd search the shortest path from the start of scope[4] to the target address and build a list of basic blocks. Then it iterates the basic blocks with the variables in the scope and update the table. If it finds a type at the target instruction, then returns it. Otherwise, it moves to the upper scope[3]. Now it'd search the shortest path from the start of scope[3] to the start of scope[4]. Then connect it to the existing basic block list. Then it'd iterate the blocks with variables for both scopes. It can repeat this until it finds a type at the target instruction or reaches to the top scope[0]. As the basic blocks contain the shortest path, it won't worry about branches and can update the table simply. The final check will be done by find_matching_type() in the next patch. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-15-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:06 -07:00
static void update_insn_state(struct type_state *state, struct data_loc_info *dloc,
Dwarf_Die *cu_die, struct disasm_line *dl)
{
if (dloc->arch->update_insn_state)
dloc->arch->update_insn_state(state, dloc, cu_die, dl);
}
perf annotate-data: Implement instruction tracking If it failed to find a variable for the location directly, it might be due to a missing variable in the source code. For example, accessing pointer variables in a chain can result in the case like below: struct foo *foo = ...; int i = foo->bar->baz; The DWARF debug information is created for each variable so it'd have one for 'foo'. But there's no variable for 'foo->bar' and then it cannot know the type of 'bar' and 'baz'. The above source code can be compiled to the follow x86 instructions: mov 0x8(%rax), %rcx mov 0x4(%rcx), %rdx <=== PMU sample mov %rdx, -4(%rbp) Let's say 'foo' is located in the %rax and it has a pointer to struct foo. But perf sample is captured in the second instruction and there is no variable or type info for the %rcx. It'd be great if compiler could generate debug info for %rcx, but we should handle it on our side. So this patch implements the logic to iterate instructions and update the type table for each location. As it already collected a list of scopes including the target instruction, we can use it to construct the type table smartly. +---------------- scope[0] subprogram | | +-------------- scope[1] lexical_block | | | | +------------ scope[2] inlined_subroutine | | | | | | +---------- scope[3] inlined_subroutine | | | | | | | | +-------- scope[4] lexical_block | | | | | | | | | | *** target instruction ... Image the target instruction has 5 scopes, each scope will have its own variables and parameters. Then it can start with the innermost scope (4). So it'd search the shortest path from the start of scope[4] to the target address and build a list of basic blocks. Then it iterates the basic blocks with the variables in the scope and update the table. If it finds a type at the target instruction, then returns it. Otherwise, it moves to the upper scope[3]. Now it'd search the shortest path from the start of scope[3] to the start of scope[4]. Then connect it to the existing basic block list. Then it'd iterate the blocks with variables for both scopes. It can repeat this until it finds a type at the target instruction or reaches to the top scope[0]. As the basic blocks contain the shortest path, it won't worry about branches and can update the table simply. The final check will be done by find_matching_type() in the next patch. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-15-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:06 -07:00
/*
* Prepend this_blocks (from the outer scope) to full_blocks, removing
* duplicate disasm line.
*/
static void prepend_basic_blocks(struct list_head *this_blocks,
struct list_head *full_blocks)
{
struct annotated_basic_block *first_bb, *last_bb;
last_bb = list_last_entry(this_blocks, typeof(*last_bb), list);
first_bb = list_first_entry(full_blocks, typeof(*first_bb), list);
if (list_empty(full_blocks))
goto out;
/* Last insn in this_blocks should be same as first insn in full_blocks */
if (last_bb->end != first_bb->begin) {
pr_debug("prepend basic blocks: mismatched disasm line %"PRIx64" -> %"PRIx64"\n",
last_bb->end->al.offset, first_bb->begin->al.offset);
goto out;
}
/* Is the basic block have only one disasm_line? */
if (last_bb->begin == last_bb->end) {
list_del(&last_bb->list);
free(last_bb);
goto out;
}
/* Point to the insn before the last when adding this block to full_blocks */
last_bb->end = list_prev_entry(last_bb->end, al.node);
out:
list_splice(this_blocks, full_blocks);
}
static void delete_basic_blocks(struct list_head *basic_blocks)
{
struct annotated_basic_block *bb, *tmp;
list_for_each_entry_safe(bb, tmp, basic_blocks, list) {
list_del(&bb->list);
free(bb);
}
}
/* Make sure all variables have a valid start address */
static void fixup_var_address(struct die_var_type *var_types, u64 addr)
{
while (var_types) {
/*
* Some variables have no address range meaning it's always
* available in the whole scope. Let's adjust the start
* address to the start of the scope.
*/
if (var_types->addr == 0)
var_types->addr = addr;
var_types = var_types->next;
}
}
static void delete_var_types(struct die_var_type *var_types)
{
while (var_types) {
struct die_var_type *next = var_types->next;
free(var_types);
var_types = next;
}
}
/* should match to is_stack_canary() in util/annotate.c */
static void setup_stack_canary(struct data_loc_info *dloc)
{
if (arch__is(dloc->arch, "x86")) {
dloc->op->segment = INSN_SEG_X86_GS;
dloc->op->imm = true;
dloc->op->offset = 40;
}
}
/*
* It's at the target address, check if it has a matching type.
* It returns PERF_TMR_BAIL_OUT when it looks up per-cpu variables which
* are similar to global variables and no additional info is needed.
*/
static enum type_match_result check_matching_type(struct type_state *state,
struct data_loc_info *dloc,
Dwarf_Die *cu_die,
perf annotate-data: Update stack slot for the store When checking the match variable at the target instruction, it might not have any information if it's a first write to a stack slot. In this case it could spill a register value into the stack so the type info is in the source operand. But currently it's hard to get the operand from the checking function. Let's process the instruction and retry to get the type info from the stack if there's no information already. This is an example of __tcp_transmit_skb(). The instructions are <__tcp_transmit_skb>: 0: nopl 0x0(%rax, %rax, 1) 5: push %rbp 6: mov %rsp, %rbp 9: push %r15 b: push %r14 d: push %r13 f: push %r12 11: push %rbx 12: sub $0x98, %rsp 19: mov %r8d, -0xa8(%rbp) ... It cannot find any variable at -0xa8(%rbp) at this point. ----------------------------------------------------------- find data type for -0xa8(reg6) at __tcp_transmit_skb+0x19 CU for net/ipv4/tcp_output.c (die:0x817f543) frame base: cfa=0 fbreg=6 scope: [1/1] (die:81aac3e) bb: [0 - 19] var [0] -0x98(stack) type='struct tcp_out_options' size=0x28 (die:0x81af3df) var [5] reg8 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg2 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg1 type='int' size=0x4 (die:0x818059e) var [5] reg4 type='struct sk_buff*' size=0x8 (die:0x8181360) var [5] reg5 type='struct sock*' size=0x8 (die:0x8181a0c) chk [19] reg6 offset=-0xa8 ok=0 kind=0 fbreg : no type information no type information And it was able to find the type after processing the 'mov' instruction. ----------------------------------------------------------- find data type for -0xa8(reg6) at __tcp_transmit_skb+0x19 CU for net/ipv4/tcp_output.c (die:0x817f543) frame base: cfa=0 fbreg=6 scope: [1/1] (die:81aac3e) bb: [0 - 19] var [0] -0x98(stack) type='struct tcp_out_options' size=0x28 (die:0x81af3df) var [5] reg8 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg2 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg1 type='int' size=0x4 (die:0x818059e) var [5] reg4 type='struct sk_buff*' size=0x8 (die:0x8181360) var [5] reg5 type='struct sock*' size=0x8 (die:0x8181a0c) chk [19] reg6 offset=-0xa8 ok=0 kind=0 fbreg : retry <<<--- here mov [19] reg8 -> -0xa8(stack) type='unsigned int' size=0x4 (die:0x8180ed6) chk [19] reg6 offset=-0xa8 ok=0 kind=0 fbreg : Good! found by insn track: -0xa8(reg6) type-offset=0 final result: type='unsigned int' size=0x4 (die:0x8180ed6) Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240821232628.353177-4-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-21 16:26:27 -07:00
struct disasm_line *dl,
Dwarf_Die *type_die)
perf annotate-data: Implement instruction tracking If it failed to find a variable for the location directly, it might be due to a missing variable in the source code. For example, accessing pointer variables in a chain can result in the case like below: struct foo *foo = ...; int i = foo->bar->baz; The DWARF debug information is created for each variable so it'd have one for 'foo'. But there's no variable for 'foo->bar' and then it cannot know the type of 'bar' and 'baz'. The above source code can be compiled to the follow x86 instructions: mov 0x8(%rax), %rcx mov 0x4(%rcx), %rdx <=== PMU sample mov %rdx, -4(%rbp) Let's say 'foo' is located in the %rax and it has a pointer to struct foo. But perf sample is captured in the second instruction and there is no variable or type info for the %rcx. It'd be great if compiler could generate debug info for %rcx, but we should handle it on our side. So this patch implements the logic to iterate instructions and update the type table for each location. As it already collected a list of scopes including the target instruction, we can use it to construct the type table smartly. +---------------- scope[0] subprogram | | +-------------- scope[1] lexical_block | | | | +------------ scope[2] inlined_subroutine | | | | | | +---------- scope[3] inlined_subroutine | | | | | | | | +-------- scope[4] lexical_block | | | | | | | | | | *** target instruction ... Image the target instruction has 5 scopes, each scope will have its own variables and parameters. Then it can start with the innermost scope (4). So it'd search the shortest path from the start of scope[4] to the target address and build a list of basic blocks. Then it iterates the basic blocks with the variables in the scope and update the table. If it finds a type at the target instruction, then returns it. Otherwise, it moves to the upper scope[3]. Now it'd search the shortest path from the start of scope[3] to the start of scope[4]. Then connect it to the existing basic block list. Then it'd iterate the blocks with variables for both scopes. It can repeat this until it finds a type at the target instruction or reaches to the top scope[0]. As the basic blocks contain the shortest path, it won't worry about branches and can update the table simply. The final check will be done by find_matching_type() in the next patch. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-15-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:06 -07:00
{
Dwarf_Word size;
perf annotate-data: Update stack slot for the store When checking the match variable at the target instruction, it might not have any information if it's a first write to a stack slot. In this case it could spill a register value into the stack so the type info is in the source operand. But currently it's hard to get the operand from the checking function. Let's process the instruction and retry to get the type info from the stack if there's no information already. This is an example of __tcp_transmit_skb(). The instructions are <__tcp_transmit_skb>: 0: nopl 0x0(%rax, %rax, 1) 5: push %rbp 6: mov %rsp, %rbp 9: push %r15 b: push %r14 d: push %r13 f: push %r12 11: push %rbx 12: sub $0x98, %rsp 19: mov %r8d, -0xa8(%rbp) ... It cannot find any variable at -0xa8(%rbp) at this point. ----------------------------------------------------------- find data type for -0xa8(reg6) at __tcp_transmit_skb+0x19 CU for net/ipv4/tcp_output.c (die:0x817f543) frame base: cfa=0 fbreg=6 scope: [1/1] (die:81aac3e) bb: [0 - 19] var [0] -0x98(stack) type='struct tcp_out_options' size=0x28 (die:0x81af3df) var [5] reg8 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg2 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg1 type='int' size=0x4 (die:0x818059e) var [5] reg4 type='struct sk_buff*' size=0x8 (die:0x8181360) var [5] reg5 type='struct sock*' size=0x8 (die:0x8181a0c) chk [19] reg6 offset=-0xa8 ok=0 kind=0 fbreg : no type information no type information And it was able to find the type after processing the 'mov' instruction. ----------------------------------------------------------- find data type for -0xa8(reg6) at __tcp_transmit_skb+0x19 CU for net/ipv4/tcp_output.c (die:0x817f543) frame base: cfa=0 fbreg=6 scope: [1/1] (die:81aac3e) bb: [0 - 19] var [0] -0x98(stack) type='struct tcp_out_options' size=0x28 (die:0x81af3df) var [5] reg8 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg2 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg1 type='int' size=0x4 (die:0x818059e) var [5] reg4 type='struct sk_buff*' size=0x8 (die:0x8181360) var [5] reg5 type='struct sock*' size=0x8 (die:0x8181a0c) chk [19] reg6 offset=-0xa8 ok=0 kind=0 fbreg : retry <<<--- here mov [19] reg8 -> -0xa8(stack) type='unsigned int' size=0x4 (die:0x8180ed6) chk [19] reg6 offset=-0xa8 ok=0 kind=0 fbreg : Good! found by insn track: -0xa8(reg6) type-offset=0 final result: type='unsigned int' size=0x4 (die:0x8180ed6) Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240821232628.353177-4-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-21 16:26:27 -07:00
u32 insn_offset = dl->al.offset;
int reg = dloc->op->reg1;
perf annotate-data: Update debug messages In check_matching_type(), it'd be easier to display the typename in question if it's available. For example, check out the line starts with 'chk'. ----------------------------------------------------------- find data type for 0x10(reg0) at cpuacct_charge+0x13 CU for kernel/sched/build_utility.c (die:0x137ee0b) frame base: cfa=1 fbreg=7 scope: [3/3] (die:13d9632) bb: [c - 13] var [c] reg5 type='struct task_struct*' size=0x8 (die:0x1381230) mov [c] 0xdf8(reg5) -> reg0 type='struct css_set*' size=0x8 (die:0x1385c56) chk [13] reg0 offset=0x10 ok=1 kind=1 (struct css_set*) : Good! <<<--- here found by insn track: 0x10(reg0) type-offset=0x10 final result: type='struct css_set' size=0x250 (die:0x1385b0e) Another example: ----------------------------------------------------------- find data type for 0x8(reg0) at menu_select+0x279 CU for drivers/cpuidle/governors/menu.c (die:0x7b0fe79) frame base: cfa=1 fbreg=7 scope: [2/2] (die:7b11010) bb: [273 - 277] bb: [279 - 279] chk [279] reg0 offset=0x8 ok=0 kind=0 cfa : no type information scope: [1/2] (die:7b10cbc) bb: [0 - 64] ... mov [26a] imm=0xffffffff -> reg15 bb: [273 - 277] bb: [279 - 279] chk [279] reg0 offset=0x8 ok=1 kind=1 (long long unsigned int) : no/void pointer <<<--- here final result: no/void pointer Also change some places to print negative offsets properly. Before: ----------------------------------------------------------- find data type for 0xffffff40(reg6) at __tcp_transmit_skb+0x58 After: ----------------------------------------------------------- find data type for -0xc0(reg6) at __tcp_transmit_skb+0x58 Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240821232628.353177-3-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-21 16:26:26 -07:00
int offset = dloc->op->offset;
const char *offset_sign = "";
perf annotate-data: Update stack slot for the store When checking the match variable at the target instruction, it might not have any information if it's a first write to a stack slot. In this case it could spill a register value into the stack so the type info is in the source operand. But currently it's hard to get the operand from the checking function. Let's process the instruction and retry to get the type info from the stack if there's no information already. This is an example of __tcp_transmit_skb(). The instructions are <__tcp_transmit_skb>: 0: nopl 0x0(%rax, %rax, 1) 5: push %rbp 6: mov %rsp, %rbp 9: push %r15 b: push %r14 d: push %r13 f: push %r12 11: push %rbx 12: sub $0x98, %rsp 19: mov %r8d, -0xa8(%rbp) ... It cannot find any variable at -0xa8(%rbp) at this point. ----------------------------------------------------------- find data type for -0xa8(reg6) at __tcp_transmit_skb+0x19 CU for net/ipv4/tcp_output.c (die:0x817f543) frame base: cfa=0 fbreg=6 scope: [1/1] (die:81aac3e) bb: [0 - 19] var [0] -0x98(stack) type='struct tcp_out_options' size=0x28 (die:0x81af3df) var [5] reg8 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg2 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg1 type='int' size=0x4 (die:0x818059e) var [5] reg4 type='struct sk_buff*' size=0x8 (die:0x8181360) var [5] reg5 type='struct sock*' size=0x8 (die:0x8181a0c) chk [19] reg6 offset=-0xa8 ok=0 kind=0 fbreg : no type information no type information And it was able to find the type after processing the 'mov' instruction. ----------------------------------------------------------- find data type for -0xa8(reg6) at __tcp_transmit_skb+0x19 CU for net/ipv4/tcp_output.c (die:0x817f543) frame base: cfa=0 fbreg=6 scope: [1/1] (die:81aac3e) bb: [0 - 19] var [0] -0x98(stack) type='struct tcp_out_options' size=0x28 (die:0x81af3df) var [5] reg8 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg2 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg1 type='int' size=0x4 (die:0x818059e) var [5] reg4 type='struct sk_buff*' size=0x8 (die:0x8181360) var [5] reg5 type='struct sock*' size=0x8 (die:0x8181a0c) chk [19] reg6 offset=-0xa8 ok=0 kind=0 fbreg : retry <<<--- here mov [19] reg8 -> -0xa8(stack) type='unsigned int' size=0x4 (die:0x8180ed6) chk [19] reg6 offset=-0xa8 ok=0 kind=0 fbreg : Good! found by insn track: -0xa8(reg6) type-offset=0 final result: type='unsigned int' size=0x4 (die:0x8180ed6) Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240821232628.353177-4-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-21 16:26:27 -07:00
bool retry = true;
perf annotate-data: Update debug messages In check_matching_type(), it'd be easier to display the typename in question if it's available. For example, check out the line starts with 'chk'. ----------------------------------------------------------- find data type for 0x10(reg0) at cpuacct_charge+0x13 CU for kernel/sched/build_utility.c (die:0x137ee0b) frame base: cfa=1 fbreg=7 scope: [3/3] (die:13d9632) bb: [c - 13] var [c] reg5 type='struct task_struct*' size=0x8 (die:0x1381230) mov [c] 0xdf8(reg5) -> reg0 type='struct css_set*' size=0x8 (die:0x1385c56) chk [13] reg0 offset=0x10 ok=1 kind=1 (struct css_set*) : Good! <<<--- here found by insn track: 0x10(reg0) type-offset=0x10 final result: type='struct css_set' size=0x250 (die:0x1385b0e) Another example: ----------------------------------------------------------- find data type for 0x8(reg0) at menu_select+0x279 CU for drivers/cpuidle/governors/menu.c (die:0x7b0fe79) frame base: cfa=1 fbreg=7 scope: [2/2] (die:7b11010) bb: [273 - 277] bb: [279 - 279] chk [279] reg0 offset=0x8 ok=0 kind=0 cfa : no type information scope: [1/2] (die:7b10cbc) bb: [0 - 64] ... mov [26a] imm=0xffffffff -> reg15 bb: [273 - 277] bb: [279 - 279] chk [279] reg0 offset=0x8 ok=1 kind=1 (long long unsigned int) : no/void pointer <<<--- here final result: no/void pointer Also change some places to print negative offsets properly. Before: ----------------------------------------------------------- find data type for 0xffffff40(reg6) at __tcp_transmit_skb+0x58 After: ----------------------------------------------------------- find data type for -0xc0(reg6) at __tcp_transmit_skb+0x58 Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240821232628.353177-3-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-21 16:26:26 -07:00
if (offset < 0) {
offset = -offset;
offset_sign = "-";
}
perf annotate-data: Update stack slot for the store When checking the match variable at the target instruction, it might not have any information if it's a first write to a stack slot. In this case it could spill a register value into the stack so the type info is in the source operand. But currently it's hard to get the operand from the checking function. Let's process the instruction and retry to get the type info from the stack if there's no information already. This is an example of __tcp_transmit_skb(). The instructions are <__tcp_transmit_skb>: 0: nopl 0x0(%rax, %rax, 1) 5: push %rbp 6: mov %rsp, %rbp 9: push %r15 b: push %r14 d: push %r13 f: push %r12 11: push %rbx 12: sub $0x98, %rsp 19: mov %r8d, -0xa8(%rbp) ... It cannot find any variable at -0xa8(%rbp) at this point. ----------------------------------------------------------- find data type for -0xa8(reg6) at __tcp_transmit_skb+0x19 CU for net/ipv4/tcp_output.c (die:0x817f543) frame base: cfa=0 fbreg=6 scope: [1/1] (die:81aac3e) bb: [0 - 19] var [0] -0x98(stack) type='struct tcp_out_options' size=0x28 (die:0x81af3df) var [5] reg8 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg2 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg1 type='int' size=0x4 (die:0x818059e) var [5] reg4 type='struct sk_buff*' size=0x8 (die:0x8181360) var [5] reg5 type='struct sock*' size=0x8 (die:0x8181a0c) chk [19] reg6 offset=-0xa8 ok=0 kind=0 fbreg : no type information no type information And it was able to find the type after processing the 'mov' instruction. ----------------------------------------------------------- find data type for -0xa8(reg6) at __tcp_transmit_skb+0x19 CU for net/ipv4/tcp_output.c (die:0x817f543) frame base: cfa=0 fbreg=6 scope: [1/1] (die:81aac3e) bb: [0 - 19] var [0] -0x98(stack) type='struct tcp_out_options' size=0x28 (die:0x81af3df) var [5] reg8 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg2 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg1 type='int' size=0x4 (die:0x818059e) var [5] reg4 type='struct sk_buff*' size=0x8 (die:0x8181360) var [5] reg5 type='struct sock*' size=0x8 (die:0x8181a0c) chk [19] reg6 offset=-0xa8 ok=0 kind=0 fbreg : retry <<<--- here mov [19] reg8 -> -0xa8(stack) type='unsigned int' size=0x4 (die:0x8180ed6) chk [19] reg6 offset=-0xa8 ok=0 kind=0 fbreg : Good! found by insn track: -0xa8(reg6) type-offset=0 final result: type='unsigned int' size=0x4 (die:0x8180ed6) Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240821232628.353177-4-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-21 16:26:27 -07:00
again:
perf annotate-data: Update debug messages In check_matching_type(), it'd be easier to display the typename in question if it's available. For example, check out the line starts with 'chk'. ----------------------------------------------------------- find data type for 0x10(reg0) at cpuacct_charge+0x13 CU for kernel/sched/build_utility.c (die:0x137ee0b) frame base: cfa=1 fbreg=7 scope: [3/3] (die:13d9632) bb: [c - 13] var [c] reg5 type='struct task_struct*' size=0x8 (die:0x1381230) mov [c] 0xdf8(reg5) -> reg0 type='struct css_set*' size=0x8 (die:0x1385c56) chk [13] reg0 offset=0x10 ok=1 kind=1 (struct css_set*) : Good! <<<--- here found by insn track: 0x10(reg0) type-offset=0x10 final result: type='struct css_set' size=0x250 (die:0x1385b0e) Another example: ----------------------------------------------------------- find data type for 0x8(reg0) at menu_select+0x279 CU for drivers/cpuidle/governors/menu.c (die:0x7b0fe79) frame base: cfa=1 fbreg=7 scope: [2/2] (die:7b11010) bb: [273 - 277] bb: [279 - 279] chk [279] reg0 offset=0x8 ok=0 kind=0 cfa : no type information scope: [1/2] (die:7b10cbc) bb: [0 - 64] ... mov [26a] imm=0xffffffff -> reg15 bb: [273 - 277] bb: [279 - 279] chk [279] reg0 offset=0x8 ok=1 kind=1 (long long unsigned int) : no/void pointer <<<--- here final result: no/void pointer Also change some places to print negative offsets properly. Before: ----------------------------------------------------------- find data type for 0xffffff40(reg6) at __tcp_transmit_skb+0x58 After: ----------------------------------------------------------- find data type for -0xc0(reg6) at __tcp_transmit_skb+0x58 Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240821232628.353177-3-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-21 16:26:26 -07:00
pr_debug_dtp("chk [%x] reg%d offset=%s%#x ok=%d kind=%d ",
insn_offset, reg, offset_sign, offset,
perf annotate-data: Support general per-cpu access This is to support per-cpu variable access often without a matching DWARF entry. For some reason, I cannot find debug info of per-cpu variables sometimes. They have more complex pattern to calculate the address of per-cpu variables like below. 2b7d: mov -0x1e0(%rbp),%rax ; rax = cpu 2b84: mov -0x7da0f7e0(,%rax,8),%rcx ; rcx = __per_cpu_offset[cpu] * 2b8c: mov 0x34870(%rcx),%rax ; *(__per_cpu_offset[cpu] + 0x34870) Let's assume the rax register has a number for a CPU at 2b7d. The next instruction is to get the per-cpu offset' for that cpu. The offset -0x7da0f7e0 is 0xffffffff825f0820 in u64 which is the address of the '__per_cpu_offset' array in my system. So it'd get the actual offset of that CPU's per-cpu region and save it to the rcx register. Then, at 2b8c, accesses using rcx can be handled same as the global variable access. To handle this case, it should check if the offset of the instruction matches to the address of '__per_cpu_offset'. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-20-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:11 -07:00
state->regs[reg].ok, state->regs[reg].kind);
perf annotate-data: Fix percpu pointer check In check_matching_type(), it checks the type state of the register in a wrong order. When it's the percpu pointer, it should check the type for the pointer, but it checks the CFA bit first and thought it has no type in the stack slot. This resulted in no type info. ----------------------------------------------------------- find data type for 0x28(reg1) at hrtimer_reprogram+0x88 CU for kernel/time/hrtimer.c (die:0x18f219f) frame base: cfa=1 fbreg=7 ... add [72] percpu 0x24500 -> reg1 pointer type='struct hrtimer_cpu_base' size=0x240 (die:0x18f6d46) bb: [7a - 7e] bb: [80 - 86] (here) bb: [88 - 88] vvv chk [88] reg1 offset=0x28 ok=1 kind=4 cfa : no type information no type information Here, instruction at 0x72 found reg1 has a (percpu) pointer and got the correct type. But when it checks the final result, it wrongly thought it was stack variable because it checks the cfa bit first. After changing the order of state check: ----------------------------------------------------------- find data type for 0x28(reg1) at hrtimer_reprogram+0x88 CU for kernel/time/hrtimer.c (die:0x18f219f) frame base: cfa=1 fbreg=7 ... (here) vvvvvvvvvv chk [88] reg1 offset=0x28 ok=1 kind=4 percpu ptr : Good! found by insn track: 0x28(reg1) type-offset=0x28 final type: type='struct hrtimer_cpu_base' size=0x240 (die:0x18f6d46) Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240821065408.285548-3-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-20 23:54:08 -07:00
if (!state->regs[reg].ok)
goto check_non_register;
if (state->regs[reg].kind == TSR_KIND_TYPE) {
perf annotate-data: Show typedef names properly The die_get_typename() would resolve typedef and get to the original type. But sometimes the original type is a struct without name and it makes the output confusing and hard to read. This is a diff of perf report -s type before and after the change. New types such as atomic{,64}_t and sigset_t appeared and the portion of unnamed struct was reduced. Also u32, u64 and size_t were splitted from the base types. --- b 2024-08-01 17:02:34.307809952 -0700 +++ a 2024-08-07 14:17:05.245853999 -0700 - 2.40% long unsigned int + 2.26% long unsigned int - 1.56% unsigned int + 1.27% unsigned int - 0.98% struct - 0.79% long long unsigned int + 0.58% long long unsigned int + 0.36% struct + 0.27% atomic64_t + 0.22% u32 + 0.21% u64 + 0.19% atomic_t + 0.13% size_t - 0.08% struct seqcount_spinlock + 0.08% seqcount_spinlock_t + 0.08% sigset_t + 0.08% __poll_t Let's use the typedef name directly and the resolved to get the size of the type. Committer testing: root@x1:~# diff -u before after | head -30 --- before 2024-08-08 09:35:13.917325041 -0300 +++ after 2024-08-08 09:37:35.312257905 -0300 @@ -10,25 +10,27 @@ # ........ ......... # 79.40% (unknown) - 2.28% union 1.96% (stack operation) - 1.24% struct + 1.87% pthread_mutex_t 0.99% u32[] - 0.92% unsigned int 0.77% struct task_struct + 0.75% U32 0.75% struct pcpu_hot 0.63% struct qspinlock + 0.61% atomic_t 0.59% struct list_head - 0.58% int 0.53% struct cfs_rq 0.51% BYTE* - 0.48% unsigned char + 0.48% BYTE 0.48% long unsigned int 0.46% struct rq 0.41% struct worker 0.41% struct memcg_vmstats_percpu + 0.41% pthread_cond_t 0.37% _Bool + 0.36% int root@x1:~# Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240807223129.1738004-1-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-07 15:31:29 -07:00
Dwarf_Die sized_type;
perf annotate-data: Update debug messages In check_matching_type(), it'd be easier to display the typename in question if it's available. For example, check out the line starts with 'chk'. ----------------------------------------------------------- find data type for 0x10(reg0) at cpuacct_charge+0x13 CU for kernel/sched/build_utility.c (die:0x137ee0b) frame base: cfa=1 fbreg=7 scope: [3/3] (die:13d9632) bb: [c - 13] var [c] reg5 type='struct task_struct*' size=0x8 (die:0x1381230) mov [c] 0xdf8(reg5) -> reg0 type='struct css_set*' size=0x8 (die:0x1385c56) chk [13] reg0 offset=0x10 ok=1 kind=1 (struct css_set*) : Good! <<<--- here found by insn track: 0x10(reg0) type-offset=0x10 final result: type='struct css_set' size=0x250 (die:0x1385b0e) Another example: ----------------------------------------------------------- find data type for 0x8(reg0) at menu_select+0x279 CU for drivers/cpuidle/governors/menu.c (die:0x7b0fe79) frame base: cfa=1 fbreg=7 scope: [2/2] (die:7b11010) bb: [273 - 277] bb: [279 - 279] chk [279] reg0 offset=0x8 ok=0 kind=0 cfa : no type information scope: [1/2] (die:7b10cbc) bb: [0 - 64] ... mov [26a] imm=0xffffffff -> reg15 bb: [273 - 277] bb: [279 - 279] chk [279] reg0 offset=0x8 ok=1 kind=1 (long long unsigned int) : no/void pointer <<<--- here final result: no/void pointer Also change some places to print negative offsets properly. Before: ----------------------------------------------------------- find data type for 0xffffff40(reg6) at __tcp_transmit_skb+0x58 After: ----------------------------------------------------------- find data type for -0xc0(reg6) at __tcp_transmit_skb+0x58 Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240821232628.353177-3-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-21 16:26:26 -07:00
struct strbuf sb;
strbuf_init(&sb, 32);
die_get_typename_from_type(&state->regs[reg].type, &sb);
pr_debug_dtp("(%s)", sb.buf);
strbuf_release(&sb);
/*
* Normal registers should hold a pointer (or array) to
* dereference a memory location.
*/
if (!is_pointer_type(&state->regs[reg].type)) {
if (dloc->op->offset < 0 && reg != state->stack_reg)
goto check_kernel;
return PERF_TMR_NO_POINTER;
}
/* Remove the pointer and get the target type */
perf annotate-data: Show typedef names properly The die_get_typename() would resolve typedef and get to the original type. But sometimes the original type is a struct without name and it makes the output confusing and hard to read. This is a diff of perf report -s type before and after the change. New types such as atomic{,64}_t and sigset_t appeared and the portion of unnamed struct was reduced. Also u32, u64 and size_t were splitted from the base types. --- b 2024-08-01 17:02:34.307809952 -0700 +++ a 2024-08-07 14:17:05.245853999 -0700 - 2.40% long unsigned int + 2.26% long unsigned int - 1.56% unsigned int + 1.27% unsigned int - 0.98% struct - 0.79% long long unsigned int + 0.58% long long unsigned int + 0.36% struct + 0.27% atomic64_t + 0.22% u32 + 0.21% u64 + 0.19% atomic_t + 0.13% size_t - 0.08% struct seqcount_spinlock + 0.08% seqcount_spinlock_t + 0.08% sigset_t + 0.08% __poll_t Let's use the typedef name directly and the resolved to get the size of the type. Committer testing: root@x1:~# diff -u before after | head -30 --- before 2024-08-08 09:35:13.917325041 -0300 +++ after 2024-08-08 09:37:35.312257905 -0300 @@ -10,25 +10,27 @@ # ........ ......... # 79.40% (unknown) - 2.28% union 1.96% (stack operation) - 1.24% struct + 1.87% pthread_mutex_t 0.99% u32[] - 0.92% unsigned int 0.77% struct task_struct + 0.75% U32 0.75% struct pcpu_hot 0.63% struct qspinlock + 0.61% atomic_t 0.59% struct list_head - 0.58% int 0.53% struct cfs_rq 0.51% BYTE* - 0.48% unsigned char + 0.48% BYTE 0.48% long unsigned int 0.46% struct rq 0.41% struct worker 0.41% struct memcg_vmstats_percpu + 0.41% pthread_cond_t 0.37% _Bool + 0.36% int root@x1:~# Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240807223129.1738004-1-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-07 15:31:29 -07:00
if (__die_get_real_type(&state->regs[reg].type, type_die) == NULL)
return PERF_TMR_NO_POINTER;
dloc->type_offset = dloc->op->offset;
perf annotate-data: Show typedef names properly The die_get_typename() would resolve typedef and get to the original type. But sometimes the original type is a struct without name and it makes the output confusing and hard to read. This is a diff of perf report -s type before and after the change. New types such as atomic{,64}_t and sigset_t appeared and the portion of unnamed struct was reduced. Also u32, u64 and size_t were splitted from the base types. --- b 2024-08-01 17:02:34.307809952 -0700 +++ a 2024-08-07 14:17:05.245853999 -0700 - 2.40% long unsigned int + 2.26% long unsigned int - 1.56% unsigned int + 1.27% unsigned int - 0.98% struct - 0.79% long long unsigned int + 0.58% long long unsigned int + 0.36% struct + 0.27% atomic64_t + 0.22% u32 + 0.21% u64 + 0.19% atomic_t + 0.13% size_t - 0.08% struct seqcount_spinlock + 0.08% seqcount_spinlock_t + 0.08% sigset_t + 0.08% __poll_t Let's use the typedef name directly and the resolved to get the size of the type. Committer testing: root@x1:~# diff -u before after | head -30 --- before 2024-08-08 09:35:13.917325041 -0300 +++ after 2024-08-08 09:37:35.312257905 -0300 @@ -10,25 +10,27 @@ # ........ ......... # 79.40% (unknown) - 2.28% union 1.96% (stack operation) - 1.24% struct + 1.87% pthread_mutex_t 0.99% u32[] - 0.92% unsigned int 0.77% struct task_struct + 0.75% U32 0.75% struct pcpu_hot 0.63% struct qspinlock + 0.61% atomic_t 0.59% struct list_head - 0.58% int 0.53% struct cfs_rq 0.51% BYTE* - 0.48% unsigned char + 0.48% BYTE 0.48% long unsigned int 0.46% struct rq 0.41% struct worker 0.41% struct memcg_vmstats_percpu + 0.41% pthread_cond_t 0.37% _Bool + 0.36% int root@x1:~# Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240807223129.1738004-1-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-07 15:31:29 -07:00
if (dwarf_tag(type_die) == DW_TAG_typedef)
die_get_real_type(type_die, &sized_type);
else
sized_type = *type_die;
/* Get the size of the actual type */
perf annotate-data: Show typedef names properly The die_get_typename() would resolve typedef and get to the original type. But sometimes the original type is a struct without name and it makes the output confusing and hard to read. This is a diff of perf report -s type before and after the change. New types such as atomic{,64}_t and sigset_t appeared and the portion of unnamed struct was reduced. Also u32, u64 and size_t were splitted from the base types. --- b 2024-08-01 17:02:34.307809952 -0700 +++ a 2024-08-07 14:17:05.245853999 -0700 - 2.40% long unsigned int + 2.26% long unsigned int - 1.56% unsigned int + 1.27% unsigned int - 0.98% struct - 0.79% long long unsigned int + 0.58% long long unsigned int + 0.36% struct + 0.27% atomic64_t + 0.22% u32 + 0.21% u64 + 0.19% atomic_t + 0.13% size_t - 0.08% struct seqcount_spinlock + 0.08% seqcount_spinlock_t + 0.08% sigset_t + 0.08% __poll_t Let's use the typedef name directly and the resolved to get the size of the type. Committer testing: root@x1:~# diff -u before after | head -30 --- before 2024-08-08 09:35:13.917325041 -0300 +++ after 2024-08-08 09:37:35.312257905 -0300 @@ -10,25 +10,27 @@ # ........ ......... # 79.40% (unknown) - 2.28% union 1.96% (stack operation) - 1.24% struct + 1.87% pthread_mutex_t 0.99% u32[] - 0.92% unsigned int 0.77% struct task_struct + 0.75% U32 0.75% struct pcpu_hot 0.63% struct qspinlock + 0.61% atomic_t 0.59% struct list_head - 0.58% int 0.53% struct cfs_rq 0.51% BYTE* - 0.48% unsigned char + 0.48% BYTE 0.48% long unsigned int 0.46% struct rq 0.41% struct worker 0.41% struct memcg_vmstats_percpu + 0.41% pthread_cond_t 0.37% _Bool + 0.36% int root@x1:~# Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240807223129.1738004-1-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-07 15:31:29 -07:00
if (dwarf_aggregate_size(&sized_type, &size) < 0 ||
(unsigned)dloc->type_offset >= size)
return PERF_TMR_BAD_OFFSET;
return PERF_TMR_OK;
}
perf annotate-data: Fix percpu pointer check In check_matching_type(), it checks the type state of the register in a wrong order. When it's the percpu pointer, it should check the type for the pointer, but it checks the CFA bit first and thought it has no type in the stack slot. This resulted in no type info. ----------------------------------------------------------- find data type for 0x28(reg1) at hrtimer_reprogram+0x88 CU for kernel/time/hrtimer.c (die:0x18f219f) frame base: cfa=1 fbreg=7 ... add [72] percpu 0x24500 -> reg1 pointer type='struct hrtimer_cpu_base' size=0x240 (die:0x18f6d46) bb: [7a - 7e] bb: [80 - 86] (here) bb: [88 - 88] vvv chk [88] reg1 offset=0x28 ok=1 kind=4 cfa : no type information no type information Here, instruction at 0x72 found reg1 has a (percpu) pointer and got the correct type. But when it checks the final result, it wrongly thought it was stack variable because it checks the cfa bit first. After changing the order of state check: ----------------------------------------------------------- find data type for 0x28(reg1) at hrtimer_reprogram+0x88 CU for kernel/time/hrtimer.c (die:0x18f219f) frame base: cfa=1 fbreg=7 ... (here) vvvvvvvvvv chk [88] reg1 offset=0x28 ok=1 kind=4 percpu ptr : Good! found by insn track: 0x28(reg1) type-offset=0x28 final type: type='struct hrtimer_cpu_base' size=0x240 (die:0x18f6d46) Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240821065408.285548-3-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-20 23:54:08 -07:00
if (state->regs[reg].kind == TSR_KIND_POINTER) {
pr_debug_dtp("percpu ptr");
/*
* It's actaully pointer but the address was calculated using
* some arithmetic. So it points to the actual type already.
*/
*type_die = state->regs[reg].type;
dloc->type_offset = dloc->op->offset;
/* Get the size of the actual type */
if (dwarf_aggregate_size(type_die, &size) < 0 ||
(unsigned)dloc->type_offset >= size)
return PERF_TMR_BAIL_OUT;
return PERF_TMR_OK;
}
if (state->regs[reg].kind == TSR_KIND_CANARY) {
pr_debug_dtp("stack canary");
/*
* This is a saved value of the stack canary which will be handled
* in the outer logic when it returns failure here. Pretend it's
* from the stack canary directly.
*/
setup_stack_canary(dloc);
return PERF_TMR_BAIL_OUT;
}
if (state->regs[reg].kind == TSR_KIND_PERCPU_BASE) {
u64 var_addr = dloc->op->offset;
int var_offset;
pr_debug_dtp("percpu var");
if (dloc->op->multi_regs) {
int reg2 = dloc->op->reg2;
if (dloc->op->reg2 == reg)
reg2 = dloc->op->reg1;
if (has_reg_type(state, reg2) && state->regs[reg2].ok &&
state->regs[reg2].kind == TSR_KIND_CONST)
var_addr += state->regs[reg2].imm_value;
}
if (get_global_var_type(cu_die, dloc, dloc->ip, var_addr,
&var_offset, type_die)) {
dloc->type_offset = var_offset;
return PERF_TMR_OK;
}
/* No need to retry per-cpu (global) variables */
return PERF_TMR_BAIL_OUT;
}
check_non_register:
if (reg == dloc->fbreg) {
struct type_state_stack *stack;
pr_debug_dtp("fbreg");
stack = find_stack_state(state, dloc->type_offset);
perf annotate-data: Update stack slot for the store When checking the match variable at the target instruction, it might not have any information if it's a first write to a stack slot. In this case it could spill a register value into the stack so the type info is in the source operand. But currently it's hard to get the operand from the checking function. Let's process the instruction and retry to get the type info from the stack if there's no information already. This is an example of __tcp_transmit_skb(). The instructions are <__tcp_transmit_skb>: 0: nopl 0x0(%rax, %rax, 1) 5: push %rbp 6: mov %rsp, %rbp 9: push %r15 b: push %r14 d: push %r13 f: push %r12 11: push %rbx 12: sub $0x98, %rsp 19: mov %r8d, -0xa8(%rbp) ... It cannot find any variable at -0xa8(%rbp) at this point. ----------------------------------------------------------- find data type for -0xa8(reg6) at __tcp_transmit_skb+0x19 CU for net/ipv4/tcp_output.c (die:0x817f543) frame base: cfa=0 fbreg=6 scope: [1/1] (die:81aac3e) bb: [0 - 19] var [0] -0x98(stack) type='struct tcp_out_options' size=0x28 (die:0x81af3df) var [5] reg8 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg2 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg1 type='int' size=0x4 (die:0x818059e) var [5] reg4 type='struct sk_buff*' size=0x8 (die:0x8181360) var [5] reg5 type='struct sock*' size=0x8 (die:0x8181a0c) chk [19] reg6 offset=-0xa8 ok=0 kind=0 fbreg : no type information no type information And it was able to find the type after processing the 'mov' instruction. ----------------------------------------------------------- find data type for -0xa8(reg6) at __tcp_transmit_skb+0x19 CU for net/ipv4/tcp_output.c (die:0x817f543) frame base: cfa=0 fbreg=6 scope: [1/1] (die:81aac3e) bb: [0 - 19] var [0] -0x98(stack) type='struct tcp_out_options' size=0x28 (die:0x81af3df) var [5] reg8 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg2 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg1 type='int' size=0x4 (die:0x818059e) var [5] reg4 type='struct sk_buff*' size=0x8 (die:0x8181360) var [5] reg5 type='struct sock*' size=0x8 (die:0x8181a0c) chk [19] reg6 offset=-0xa8 ok=0 kind=0 fbreg : retry <<<--- here mov [19] reg8 -> -0xa8(stack) type='unsigned int' size=0x4 (die:0x8180ed6) chk [19] reg6 offset=-0xa8 ok=0 kind=0 fbreg : Good! found by insn track: -0xa8(reg6) type-offset=0 final result: type='unsigned int' size=0x4 (die:0x8180ed6) Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240821232628.353177-4-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-21 16:26:27 -07:00
if (stack == NULL) {
if (retry) {
pr_debug_dtp(" : retry\n");
retry = false;
/* update type info it's the first store to the stack */
update_insn_state(state, dloc, cu_die, dl);
goto again;
}
return PERF_TMR_NO_TYPE;
perf annotate-data: Update stack slot for the store When checking the match variable at the target instruction, it might not have any information if it's a first write to a stack slot. In this case it could spill a register value into the stack so the type info is in the source operand. But currently it's hard to get the operand from the checking function. Let's process the instruction and retry to get the type info from the stack if there's no information already. This is an example of __tcp_transmit_skb(). The instructions are <__tcp_transmit_skb>: 0: nopl 0x0(%rax, %rax, 1) 5: push %rbp 6: mov %rsp, %rbp 9: push %r15 b: push %r14 d: push %r13 f: push %r12 11: push %rbx 12: sub $0x98, %rsp 19: mov %r8d, -0xa8(%rbp) ... It cannot find any variable at -0xa8(%rbp) at this point. ----------------------------------------------------------- find data type for -0xa8(reg6) at __tcp_transmit_skb+0x19 CU for net/ipv4/tcp_output.c (die:0x817f543) frame base: cfa=0 fbreg=6 scope: [1/1] (die:81aac3e) bb: [0 - 19] var [0] -0x98(stack) type='struct tcp_out_options' size=0x28 (die:0x81af3df) var [5] reg8 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg2 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg1 type='int' size=0x4 (die:0x818059e) var [5] reg4 type='struct sk_buff*' size=0x8 (die:0x8181360) var [5] reg5 type='struct sock*' size=0x8 (die:0x8181a0c) chk [19] reg6 offset=-0xa8 ok=0 kind=0 fbreg : no type information no type information And it was able to find the type after processing the 'mov' instruction. ----------------------------------------------------------- find data type for -0xa8(reg6) at __tcp_transmit_skb+0x19 CU for net/ipv4/tcp_output.c (die:0x817f543) frame base: cfa=0 fbreg=6 scope: [1/1] (die:81aac3e) bb: [0 - 19] var [0] -0x98(stack) type='struct tcp_out_options' size=0x28 (die:0x81af3df) var [5] reg8 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg2 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg1 type='int' size=0x4 (die:0x818059e) var [5] reg4 type='struct sk_buff*' size=0x8 (die:0x8181360) var [5] reg5 type='struct sock*' size=0x8 (die:0x8181a0c) chk [19] reg6 offset=-0xa8 ok=0 kind=0 fbreg : retry <<<--- here mov [19] reg8 -> -0xa8(stack) type='unsigned int' size=0x4 (die:0x8180ed6) chk [19] reg6 offset=-0xa8 ok=0 kind=0 fbreg : Good! found by insn track: -0xa8(reg6) type-offset=0 final result: type='unsigned int' size=0x4 (die:0x8180ed6) Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240821232628.353177-4-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-21 16:26:27 -07:00
}
if (stack->kind == TSR_KIND_CANARY) {
setup_stack_canary(dloc);
return PERF_TMR_BAIL_OUT;
}
perf annotate-data: Check kind of stack variables I sometimes see ("unknown type") in the result and it was because it didn't check the type of stack variables properly during the instruction tracking. The stack can carry constant values (without type info) and if the target instruction is accessing the stack location, it resulted in the "unknown type". Maybe we could pick one of integer types for the constant, but it doesn't really mean anything useful. Let's just drop the stack slot if it doesn't have a valid type info. Here's an example how it got the unknown type. Note that 0xffffff48 = -0xb8. ----------------------------------------------------------- find data type for 0xffffff48(reg6) at ... CU for ... frame base: cfa=0 fbreg=6 scope: [2/2] (die:11cb97f) bb: [37 - 3a] var [37] reg15 type='int' size=0x4 (die:0x1180633) bb: [40 - 4b] mov [40] imm=0x1 -> reg13 var [45] reg8 type='sigset_t*' size=0x8 (die:0x11a39ee) mov [45] imm=0x1 -> reg2 <--- here reg2 has a constant bb: [215 - 237] mov [218] reg2 -> -0xb8(stack) constant <--- and save it to the stack mov [225] reg13 -> -0xc4(stack) constant call [22f] find_task_by_vgpid call [22f] return -> reg0 type='struct task_struct*' size=0x8 (die:0x11881e8) bb: [5c8 - 5cf] bb: [2fb - 302] mov [2fb] -0xc4(stack) -> reg13 constant bb: [13b - 14d] mov [143] 0xd50(reg3) -> reg5 type='struct task_struct*' size=0x8 (die:0xa31f3c) bb: [153 - 153] chk [153] reg6 offset=0xffffff48 ok=0 kind=0 fbreg <--- access here found by insn track: 0xffffff48(reg6) type-offset=0 type='G<EF>^K<F6><AF>U' size=0 (die:0xffffffffffffffff) Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240502060011.1838090-7-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-05-01 23:00:11 -07:00
if (stack->kind != TSR_KIND_TYPE)
return PERF_TMR_NO_TYPE;
perf annotate-data: Check kind of stack variables I sometimes see ("unknown type") in the result and it was because it didn't check the type of stack variables properly during the instruction tracking. The stack can carry constant values (without type info) and if the target instruction is accessing the stack location, it resulted in the "unknown type". Maybe we could pick one of integer types for the constant, but it doesn't really mean anything useful. Let's just drop the stack slot if it doesn't have a valid type info. Here's an example how it got the unknown type. Note that 0xffffff48 = -0xb8. ----------------------------------------------------------- find data type for 0xffffff48(reg6) at ... CU for ... frame base: cfa=0 fbreg=6 scope: [2/2] (die:11cb97f) bb: [37 - 3a] var [37] reg15 type='int' size=0x4 (die:0x1180633) bb: [40 - 4b] mov [40] imm=0x1 -> reg13 var [45] reg8 type='sigset_t*' size=0x8 (die:0x11a39ee) mov [45] imm=0x1 -> reg2 <--- here reg2 has a constant bb: [215 - 237] mov [218] reg2 -> -0xb8(stack) constant <--- and save it to the stack mov [225] reg13 -> -0xc4(stack) constant call [22f] find_task_by_vgpid call [22f] return -> reg0 type='struct task_struct*' size=0x8 (die:0x11881e8) bb: [5c8 - 5cf] bb: [2fb - 302] mov [2fb] -0xc4(stack) -> reg13 constant bb: [13b - 14d] mov [143] 0xd50(reg3) -> reg5 type='struct task_struct*' size=0x8 (die:0xa31f3c) bb: [153 - 153] chk [153] reg6 offset=0xffffff48 ok=0 kind=0 fbreg <--- access here found by insn track: 0xffffff48(reg6) type-offset=0 type='G<EF>^K<F6><AF>U' size=0 (die:0xffffffffffffffff) Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240502060011.1838090-7-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-05-01 23:00:11 -07:00
*type_die = stack->type;
/* Update the type offset from the start of slot */
dloc->type_offset -= stack->offset;
return PERF_TMR_OK;
}
if (dloc->fb_cfa) {
struct type_state_stack *stack;
u64 pc = map__rip_2objdump(dloc->ms->map, dloc->ip);
int fbreg, fboff;
pr_debug_dtp("cfa");
if (die_get_cfa(dloc->di->dbg, pc, &fbreg, &fboff) < 0)
fbreg = -1;
if (reg != fbreg)
return PERF_TMR_NO_TYPE;
stack = find_stack_state(state, dloc->type_offset - fboff);
perf annotate-data: Update stack slot for the store When checking the match variable at the target instruction, it might not have any information if it's a first write to a stack slot. In this case it could spill a register value into the stack so the type info is in the source operand. But currently it's hard to get the operand from the checking function. Let's process the instruction and retry to get the type info from the stack if there's no information already. This is an example of __tcp_transmit_skb(). The instructions are <__tcp_transmit_skb>: 0: nopl 0x0(%rax, %rax, 1) 5: push %rbp 6: mov %rsp, %rbp 9: push %r15 b: push %r14 d: push %r13 f: push %r12 11: push %rbx 12: sub $0x98, %rsp 19: mov %r8d, -0xa8(%rbp) ... It cannot find any variable at -0xa8(%rbp) at this point. ----------------------------------------------------------- find data type for -0xa8(reg6) at __tcp_transmit_skb+0x19 CU for net/ipv4/tcp_output.c (die:0x817f543) frame base: cfa=0 fbreg=6 scope: [1/1] (die:81aac3e) bb: [0 - 19] var [0] -0x98(stack) type='struct tcp_out_options' size=0x28 (die:0x81af3df) var [5] reg8 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg2 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg1 type='int' size=0x4 (die:0x818059e) var [5] reg4 type='struct sk_buff*' size=0x8 (die:0x8181360) var [5] reg5 type='struct sock*' size=0x8 (die:0x8181a0c) chk [19] reg6 offset=-0xa8 ok=0 kind=0 fbreg : no type information no type information And it was able to find the type after processing the 'mov' instruction. ----------------------------------------------------------- find data type for -0xa8(reg6) at __tcp_transmit_skb+0x19 CU for net/ipv4/tcp_output.c (die:0x817f543) frame base: cfa=0 fbreg=6 scope: [1/1] (die:81aac3e) bb: [0 - 19] var [0] -0x98(stack) type='struct tcp_out_options' size=0x28 (die:0x81af3df) var [5] reg8 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg2 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg1 type='int' size=0x4 (die:0x818059e) var [5] reg4 type='struct sk_buff*' size=0x8 (die:0x8181360) var [5] reg5 type='struct sock*' size=0x8 (die:0x8181a0c) chk [19] reg6 offset=-0xa8 ok=0 kind=0 fbreg : retry <<<--- here mov [19] reg8 -> -0xa8(stack) type='unsigned int' size=0x4 (die:0x8180ed6) chk [19] reg6 offset=-0xa8 ok=0 kind=0 fbreg : Good! found by insn track: -0xa8(reg6) type-offset=0 final result: type='unsigned int' size=0x4 (die:0x8180ed6) Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240821232628.353177-4-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-21 16:26:27 -07:00
if (stack == NULL) {
if (retry) {
pr_debug_dtp(" : retry\n");
retry = false;
/* update type info it's the first store to the stack */
update_insn_state(state, dloc, cu_die, dl);
goto again;
}
return PERF_TMR_NO_TYPE;
perf annotate-data: Update stack slot for the store When checking the match variable at the target instruction, it might not have any information if it's a first write to a stack slot. In this case it could spill a register value into the stack so the type info is in the source operand. But currently it's hard to get the operand from the checking function. Let's process the instruction and retry to get the type info from the stack if there's no information already. This is an example of __tcp_transmit_skb(). The instructions are <__tcp_transmit_skb>: 0: nopl 0x0(%rax, %rax, 1) 5: push %rbp 6: mov %rsp, %rbp 9: push %r15 b: push %r14 d: push %r13 f: push %r12 11: push %rbx 12: sub $0x98, %rsp 19: mov %r8d, -0xa8(%rbp) ... It cannot find any variable at -0xa8(%rbp) at this point. ----------------------------------------------------------- find data type for -0xa8(reg6) at __tcp_transmit_skb+0x19 CU for net/ipv4/tcp_output.c (die:0x817f543) frame base: cfa=0 fbreg=6 scope: [1/1] (die:81aac3e) bb: [0 - 19] var [0] -0x98(stack) type='struct tcp_out_options' size=0x28 (die:0x81af3df) var [5] reg8 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg2 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg1 type='int' size=0x4 (die:0x818059e) var [5] reg4 type='struct sk_buff*' size=0x8 (die:0x8181360) var [5] reg5 type='struct sock*' size=0x8 (die:0x8181a0c) chk [19] reg6 offset=-0xa8 ok=0 kind=0 fbreg : no type information no type information And it was able to find the type after processing the 'mov' instruction. ----------------------------------------------------------- find data type for -0xa8(reg6) at __tcp_transmit_skb+0x19 CU for net/ipv4/tcp_output.c (die:0x817f543) frame base: cfa=0 fbreg=6 scope: [1/1] (die:81aac3e) bb: [0 - 19] var [0] -0x98(stack) type='struct tcp_out_options' size=0x28 (die:0x81af3df) var [5] reg8 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg2 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg1 type='int' size=0x4 (die:0x818059e) var [5] reg4 type='struct sk_buff*' size=0x8 (die:0x8181360) var [5] reg5 type='struct sock*' size=0x8 (die:0x8181a0c) chk [19] reg6 offset=-0xa8 ok=0 kind=0 fbreg : retry <<<--- here mov [19] reg8 -> -0xa8(stack) type='unsigned int' size=0x4 (die:0x8180ed6) chk [19] reg6 offset=-0xa8 ok=0 kind=0 fbreg : Good! found by insn track: -0xa8(reg6) type-offset=0 final result: type='unsigned int' size=0x4 (die:0x8180ed6) Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240821232628.353177-4-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-21 16:26:27 -07:00
}
if (stack->kind == TSR_KIND_CANARY) {
setup_stack_canary(dloc);
return PERF_TMR_BAIL_OUT;
}
perf annotate-data: Check kind of stack variables I sometimes see ("unknown type") in the result and it was because it didn't check the type of stack variables properly during the instruction tracking. The stack can carry constant values (without type info) and if the target instruction is accessing the stack location, it resulted in the "unknown type". Maybe we could pick one of integer types for the constant, but it doesn't really mean anything useful. Let's just drop the stack slot if it doesn't have a valid type info. Here's an example how it got the unknown type. Note that 0xffffff48 = -0xb8. ----------------------------------------------------------- find data type for 0xffffff48(reg6) at ... CU for ... frame base: cfa=0 fbreg=6 scope: [2/2] (die:11cb97f) bb: [37 - 3a] var [37] reg15 type='int' size=0x4 (die:0x1180633) bb: [40 - 4b] mov [40] imm=0x1 -> reg13 var [45] reg8 type='sigset_t*' size=0x8 (die:0x11a39ee) mov [45] imm=0x1 -> reg2 <--- here reg2 has a constant bb: [215 - 237] mov [218] reg2 -> -0xb8(stack) constant <--- and save it to the stack mov [225] reg13 -> -0xc4(stack) constant call [22f] find_task_by_vgpid call [22f] return -> reg0 type='struct task_struct*' size=0x8 (die:0x11881e8) bb: [5c8 - 5cf] bb: [2fb - 302] mov [2fb] -0xc4(stack) -> reg13 constant bb: [13b - 14d] mov [143] 0xd50(reg3) -> reg5 type='struct task_struct*' size=0x8 (die:0xa31f3c) bb: [153 - 153] chk [153] reg6 offset=0xffffff48 ok=0 kind=0 fbreg <--- access here found by insn track: 0xffffff48(reg6) type-offset=0 type='G<EF>^K<F6><AF>U' size=0 (die:0xffffffffffffffff) Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240502060011.1838090-7-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-05-01 23:00:11 -07:00
if (stack->kind != TSR_KIND_TYPE)
return PERF_TMR_NO_TYPE;
perf annotate-data: Check kind of stack variables I sometimes see ("unknown type") in the result and it was because it didn't check the type of stack variables properly during the instruction tracking. The stack can carry constant values (without type info) and if the target instruction is accessing the stack location, it resulted in the "unknown type". Maybe we could pick one of integer types for the constant, but it doesn't really mean anything useful. Let's just drop the stack slot if it doesn't have a valid type info. Here's an example how it got the unknown type. Note that 0xffffff48 = -0xb8. ----------------------------------------------------------- find data type for 0xffffff48(reg6) at ... CU for ... frame base: cfa=0 fbreg=6 scope: [2/2] (die:11cb97f) bb: [37 - 3a] var [37] reg15 type='int' size=0x4 (die:0x1180633) bb: [40 - 4b] mov [40] imm=0x1 -> reg13 var [45] reg8 type='sigset_t*' size=0x8 (die:0x11a39ee) mov [45] imm=0x1 -> reg2 <--- here reg2 has a constant bb: [215 - 237] mov [218] reg2 -> -0xb8(stack) constant <--- and save it to the stack mov [225] reg13 -> -0xc4(stack) constant call [22f] find_task_by_vgpid call [22f] return -> reg0 type='struct task_struct*' size=0x8 (die:0x11881e8) bb: [5c8 - 5cf] bb: [2fb - 302] mov [2fb] -0xc4(stack) -> reg13 constant bb: [13b - 14d] mov [143] 0xd50(reg3) -> reg5 type='struct task_struct*' size=0x8 (die:0xa31f3c) bb: [153 - 153] chk [153] reg6 offset=0xffffff48 ok=0 kind=0 fbreg <--- access here found by insn track: 0xffffff48(reg6) type-offset=0 type='G<EF>^K<F6><AF>U' size=0 (die:0xffffffffffffffff) Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240502060011.1838090-7-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-05-01 23:00:11 -07:00
*type_die = stack->type;
/* Update the type offset from the start of slot */
dloc->type_offset -= fboff + stack->offset;
return PERF_TMR_OK;
}
check_kernel:
perf dso: Add reference count checking and accessor functions Add reference count checking to struct dso, this can help with implementing correct reference counting discipline. To avoid RC_CHK_ACCESS everywhere, add accessor functions for the variables in struct dso. The majority of the change is mechanical in nature and not easy to split up. Committer testing: 'perf test' up to this patch shows no regressions. But: util/symbol.c: In function ‘dso__load_bfd_symbols’: util/symbol.c:1683:9: error: too few arguments to function ‘dso__set_adjust_symbols’ 1683 | dso__set_adjust_symbols(dso); | ^~~~~~~~~~~~~~~~~~~~~~~ In file included from util/symbol.c:21: util/dso.h:268:20: note: declared here 268 | static inline void dso__set_adjust_symbols(struct dso *dso, bool val) | ^~~~~~~~~~~~~~~~~~~~~~~ make[6]: *** [/home/acme/git/perf-tools-next/tools/build/Makefile.build:106: /tmp/tmp.ZWHbQftdN6/util/symbol.o] Error 1 MKDIR /tmp/tmp.ZWHbQftdN6/tests/workloads/ make[6]: *** Waiting for unfinished jobs.... This was updated: - symbols__fixup_end(&dso->symbols, false); - symbols__fixup_duplicate(&dso->symbols); - dso->adjust_symbols = 1; + symbols__fixup_end(dso__symbols(dso), false); + symbols__fixup_duplicate(dso__symbols(dso)); + dso__set_adjust_symbols(dso); But not build tested with BUILD_NONDISTRO and libbfd devel files installed (binutils-devel on fedora). Add the missing argument: symbols__fixup_end(dso__symbols(dso), false); symbols__fixup_duplicate(dso__symbols(dso)); - dso__set_adjust_symbols(dso); + dso__set_adjust_symbols(dso, true); Signed-off-by: Ian Rogers <irogers@google.com> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Andi Kleen <ak@linux.intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ben Gainey <ben.gainey@arm.com> Cc: Changbin Du <changbin.du@huawei.com> Cc: Chengen Du <chengen.du@canonical.com> Cc: Colin Ian King <colin.i.king@gmail.com> Cc: Dima Kogan <dima@secretsauce.net> Cc: Ilkka Koskinen <ilkka@os.amperecomputing.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: James Clark <james.clark@arm.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: K Prateek Nayak <kprateek.nayak@amd.com> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Leo Yan <leo.yan@linux.dev> Cc: Li Dong <lidong@vivo.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Paran Lee <p4ranlee@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Song Liu <song@kernel.org> Cc: Sun Haiyong <sunhaiyong@loongson.cn> Cc: Thomas Richter <tmricht@linux.ibm.com> Cc: Tiezhu Yang <yangtiezhu@loongson.cn> Cc: Yanteng Si <siyanteng@loongson.cn> Cc: zhaimingbing <zhaimingbing@cmss.chinamobile.com> Link: https://lore.kernel.org/r/20240504213803.218974-6-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-05-04 14:38:01 -07:00
if (dso__kernel(map__dso(dloc->ms->map))) {
u64 addr;
perf annotate-data: Support general per-cpu access This is to support per-cpu variable access often without a matching DWARF entry. For some reason, I cannot find debug info of per-cpu variables sometimes. They have more complex pattern to calculate the address of per-cpu variables like below. 2b7d: mov -0x1e0(%rbp),%rax ; rax = cpu 2b84: mov -0x7da0f7e0(,%rax,8),%rcx ; rcx = __per_cpu_offset[cpu] * 2b8c: mov 0x34870(%rcx),%rax ; *(__per_cpu_offset[cpu] + 0x34870) Let's assume the rax register has a number for a CPU at 2b7d. The next instruction is to get the per-cpu offset' for that cpu. The offset -0x7da0f7e0 is 0xffffffff825f0820 in u64 which is the address of the '__per_cpu_offset' array in my system. So it'd get the actual offset of that CPU's per-cpu region and save it to the rcx register. Then, at 2b8c, accesses using rcx can be handled same as the global variable access. To handle this case, it should check if the offset of the instruction matches to the address of '__per_cpu_offset'. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-20-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:11 -07:00
/* Direct this-cpu access like "%gs:0x34740" */
if (dloc->op->segment == INSN_SEG_X86_GS && dloc->op->imm &&
arch__is(dloc->arch, "x86")) {
pr_debug_dtp("this-cpu var");
addr = dloc->op->offset;
if (get_global_var_type(cu_die, dloc, dloc->ip, addr,
&offset, type_die)) {
dloc->type_offset = offset;
return PERF_TMR_OK;
}
return PERF_TMR_BAIL_OUT;
perf annotate-data: Support general per-cpu access This is to support per-cpu variable access often without a matching DWARF entry. For some reason, I cannot find debug info of per-cpu variables sometimes. They have more complex pattern to calculate the address of per-cpu variables like below. 2b7d: mov -0x1e0(%rbp),%rax ; rax = cpu 2b84: mov -0x7da0f7e0(,%rax,8),%rcx ; rcx = __per_cpu_offset[cpu] * 2b8c: mov 0x34870(%rcx),%rax ; *(__per_cpu_offset[cpu] + 0x34870) Let's assume the rax register has a number for a CPU at 2b7d. The next instruction is to get the per-cpu offset' for that cpu. The offset -0x7da0f7e0 is 0xffffffff825f0820 in u64 which is the address of the '__per_cpu_offset' array in my system. So it'd get the actual offset of that CPU's per-cpu region and save it to the rcx register. Then, at 2b8c, accesses using rcx can be handled same as the global variable access. To handle this case, it should check if the offset of the instruction matches to the address of '__per_cpu_offset'. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-20-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:11 -07:00
}
/* Access to global variable like "-0x7dcf0500(,%rdx,8)" */
perf annotate-data: Support general per-cpu access This is to support per-cpu variable access often without a matching DWARF entry. For some reason, I cannot find debug info of per-cpu variables sometimes. They have more complex pattern to calculate the address of per-cpu variables like below. 2b7d: mov -0x1e0(%rbp),%rax ; rax = cpu 2b84: mov -0x7da0f7e0(,%rax,8),%rcx ; rcx = __per_cpu_offset[cpu] * 2b8c: mov 0x34870(%rcx),%rax ; *(__per_cpu_offset[cpu] + 0x34870) Let's assume the rax register has a number for a CPU at 2b7d. The next instruction is to get the per-cpu offset' for that cpu. The offset -0x7da0f7e0 is 0xffffffff825f0820 in u64 which is the address of the '__per_cpu_offset' array in my system. So it'd get the actual offset of that CPU's per-cpu region and save it to the rcx register. Then, at 2b8c, accesses using rcx can be handled same as the global variable access. To handle this case, it should check if the offset of the instruction matches to the address of '__per_cpu_offset'. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-20-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:11 -07:00
if (dloc->op->offset < 0 && reg != state->stack_reg) {
addr = (s64) dloc->op->offset;
if (get_global_var_type(cu_die, dloc, dloc->ip, addr,
perf annotate-data: Support general per-cpu access This is to support per-cpu variable access often without a matching DWARF entry. For some reason, I cannot find debug info of per-cpu variables sometimes. They have more complex pattern to calculate the address of per-cpu variables like below. 2b7d: mov -0x1e0(%rbp),%rax ; rax = cpu 2b84: mov -0x7da0f7e0(,%rax,8),%rcx ; rcx = __per_cpu_offset[cpu] * 2b8c: mov 0x34870(%rcx),%rax ; *(__per_cpu_offset[cpu] + 0x34870) Let's assume the rax register has a number for a CPU at 2b7d. The next instruction is to get the per-cpu offset' for that cpu. The offset -0x7da0f7e0 is 0xffffffff825f0820 in u64 which is the address of the '__per_cpu_offset' array in my system. So it'd get the actual offset of that CPU's per-cpu region and save it to the rcx register. Then, at 2b8c, accesses using rcx can be handled same as the global variable access. To handle this case, it should check if the offset of the instruction matches to the address of '__per_cpu_offset'. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-20-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:11 -07:00
&offset, type_die)) {
pr_debug_dtp("global var");
perf annotate-data: Support general per-cpu access This is to support per-cpu variable access often without a matching DWARF entry. For some reason, I cannot find debug info of per-cpu variables sometimes. They have more complex pattern to calculate the address of per-cpu variables like below. 2b7d: mov -0x1e0(%rbp),%rax ; rax = cpu 2b84: mov -0x7da0f7e0(,%rax,8),%rcx ; rcx = __per_cpu_offset[cpu] * 2b8c: mov 0x34870(%rcx),%rax ; *(__per_cpu_offset[cpu] + 0x34870) Let's assume the rax register has a number for a CPU at 2b7d. The next instruction is to get the per-cpu offset' for that cpu. The offset -0x7da0f7e0 is 0xffffffff825f0820 in u64 which is the address of the '__per_cpu_offset' array in my system. So it'd get the actual offset of that CPU's per-cpu region and save it to the rcx register. Then, at 2b8c, accesses using rcx can be handled same as the global variable access. To handle this case, it should check if the offset of the instruction matches to the address of '__per_cpu_offset'. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-20-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:11 -07:00
dloc->type_offset = offset;
return PERF_TMR_OK;
perf annotate-data: Support general per-cpu access This is to support per-cpu variable access often without a matching DWARF entry. For some reason, I cannot find debug info of per-cpu variables sometimes. They have more complex pattern to calculate the address of per-cpu variables like below. 2b7d: mov -0x1e0(%rbp),%rax ; rax = cpu 2b84: mov -0x7da0f7e0(,%rax,8),%rcx ; rcx = __per_cpu_offset[cpu] * 2b8c: mov 0x34870(%rcx),%rax ; *(__per_cpu_offset[cpu] + 0x34870) Let's assume the rax register has a number for a CPU at 2b7d. The next instruction is to get the per-cpu offset' for that cpu. The offset -0x7da0f7e0 is 0xffffffff825f0820 in u64 which is the address of the '__per_cpu_offset' array in my system. So it'd get the actual offset of that CPU's per-cpu region and save it to the rcx register. Then, at 2b8c, accesses using rcx can be handled same as the global variable access. To handle this case, it should check if the offset of the instruction matches to the address of '__per_cpu_offset'. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-20-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:11 -07:00
}
return PERF_TMR_BAIL_OUT;
}
}
return PERF_TMR_UNKNOWN;
perf annotate-data: Implement instruction tracking If it failed to find a variable for the location directly, it might be due to a missing variable in the source code. For example, accessing pointer variables in a chain can result in the case like below: struct foo *foo = ...; int i = foo->bar->baz; The DWARF debug information is created for each variable so it'd have one for 'foo'. But there's no variable for 'foo->bar' and then it cannot know the type of 'bar' and 'baz'. The above source code can be compiled to the follow x86 instructions: mov 0x8(%rax), %rcx mov 0x4(%rcx), %rdx <=== PMU sample mov %rdx, -4(%rbp) Let's say 'foo' is located in the %rax and it has a pointer to struct foo. But perf sample is captured in the second instruction and there is no variable or type info for the %rcx. It'd be great if compiler could generate debug info for %rcx, but we should handle it on our side. So this patch implements the logic to iterate instructions and update the type table for each location. As it already collected a list of scopes including the target instruction, we can use it to construct the type table smartly. +---------------- scope[0] subprogram | | +-------------- scope[1] lexical_block | | | | +------------ scope[2] inlined_subroutine | | | | | | +---------- scope[3] inlined_subroutine | | | | | | | | +-------- scope[4] lexical_block | | | | | | | | | | *** target instruction ... Image the target instruction has 5 scopes, each scope will have its own variables and parameters. Then it can start with the innermost scope (4). So it'd search the shortest path from the start of scope[4] to the target address and build a list of basic blocks. Then it iterates the basic blocks with the variables in the scope and update the table. If it finds a type at the target instruction, then returns it. Otherwise, it moves to the upper scope[3]. Now it'd search the shortest path from the start of scope[3] to the start of scope[4]. Then connect it to the existing basic block list. Then it'd iterate the blocks with variables for both scopes. It can repeat this until it finds a type at the target instruction or reaches to the top scope[0]. As the basic blocks contain the shortest path, it won't worry about branches and can update the table simply. The final check will be done by find_matching_type() in the next patch. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-15-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:06 -07:00
}
/* Iterate instructions in basic blocks and update type table */
static enum type_match_result find_data_type_insn(struct data_loc_info *dloc,
struct list_head *basic_blocks,
struct die_var_type *var_types,
Dwarf_Die *cu_die,
Dwarf_Die *type_die)
perf annotate-data: Implement instruction tracking If it failed to find a variable for the location directly, it might be due to a missing variable in the source code. For example, accessing pointer variables in a chain can result in the case like below: struct foo *foo = ...; int i = foo->bar->baz; The DWARF debug information is created for each variable so it'd have one for 'foo'. But there's no variable for 'foo->bar' and then it cannot know the type of 'bar' and 'baz'. The above source code can be compiled to the follow x86 instructions: mov 0x8(%rax), %rcx mov 0x4(%rcx), %rdx <=== PMU sample mov %rdx, -4(%rbp) Let's say 'foo' is located in the %rax and it has a pointer to struct foo. But perf sample is captured in the second instruction and there is no variable or type info for the %rcx. It'd be great if compiler could generate debug info for %rcx, but we should handle it on our side. So this patch implements the logic to iterate instructions and update the type table for each location. As it already collected a list of scopes including the target instruction, we can use it to construct the type table smartly. +---------------- scope[0] subprogram | | +-------------- scope[1] lexical_block | | | | +------------ scope[2] inlined_subroutine | | | | | | +---------- scope[3] inlined_subroutine | | | | | | | | +-------- scope[4] lexical_block | | | | | | | | | | *** target instruction ... Image the target instruction has 5 scopes, each scope will have its own variables and parameters. Then it can start with the innermost scope (4). So it'd search the shortest path from the start of scope[4] to the target address and build a list of basic blocks. Then it iterates the basic blocks with the variables in the scope and update the table. If it finds a type at the target instruction, then returns it. Otherwise, it moves to the upper scope[3]. Now it'd search the shortest path from the start of scope[3] to the start of scope[4]. Then connect it to the existing basic block list. Then it'd iterate the blocks with variables for both scopes. It can repeat this until it finds a type at the target instruction or reaches to the top scope[0]. As the basic blocks contain the shortest path, it won't worry about branches and can update the table simply. The final check will be done by find_matching_type() in the next patch. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-15-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:06 -07:00
{
struct type_state state;
struct symbol *sym = dloc->ms->sym;
struct annotation *notes = symbol__annotation(sym);
struct annotated_basic_block *bb;
enum type_match_result ret = PERF_TMR_UNKNOWN;
perf annotate-data: Implement instruction tracking If it failed to find a variable for the location directly, it might be due to a missing variable in the source code. For example, accessing pointer variables in a chain can result in the case like below: struct foo *foo = ...; int i = foo->bar->baz; The DWARF debug information is created for each variable so it'd have one for 'foo'. But there's no variable for 'foo->bar' and then it cannot know the type of 'bar' and 'baz'. The above source code can be compiled to the follow x86 instructions: mov 0x8(%rax), %rcx mov 0x4(%rcx), %rdx <=== PMU sample mov %rdx, -4(%rbp) Let's say 'foo' is located in the %rax and it has a pointer to struct foo. But perf sample is captured in the second instruction and there is no variable or type info for the %rcx. It'd be great if compiler could generate debug info for %rcx, but we should handle it on our side. So this patch implements the logic to iterate instructions and update the type table for each location. As it already collected a list of scopes including the target instruction, we can use it to construct the type table smartly. +---------------- scope[0] subprogram | | +-------------- scope[1] lexical_block | | | | +------------ scope[2] inlined_subroutine | | | | | | +---------- scope[3] inlined_subroutine | | | | | | | | +-------- scope[4] lexical_block | | | | | | | | | | *** target instruction ... Image the target instruction has 5 scopes, each scope will have its own variables and parameters. Then it can start with the innermost scope (4). So it'd search the shortest path from the start of scope[4] to the target address and build a list of basic blocks. Then it iterates the basic blocks with the variables in the scope and update the table. If it finds a type at the target instruction, then returns it. Otherwise, it moves to the upper scope[3]. Now it'd search the shortest path from the start of scope[3] to the start of scope[4]. Then connect it to the existing basic block list. Then it'd iterate the blocks with variables for both scopes. It can repeat this until it finds a type at the target instruction or reaches to the top scope[0]. As the basic blocks contain the shortest path, it won't worry about branches and can update the table simply. The final check will be done by find_matching_type() in the next patch. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-15-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:06 -07:00
init_type_state(&state, dloc->arch);
list_for_each_entry(bb, basic_blocks, list) {
struct disasm_line *dl = bb->begin;
BUG_ON(bb->begin->al.offset == -1 || bb->end->al.offset == -1);
perf annotate-data: Implement instruction tracking If it failed to find a variable for the location directly, it might be due to a missing variable in the source code. For example, accessing pointer variables in a chain can result in the case like below: struct foo *foo = ...; int i = foo->bar->baz; The DWARF debug information is created for each variable so it'd have one for 'foo'. But there's no variable for 'foo->bar' and then it cannot know the type of 'bar' and 'baz'. The above source code can be compiled to the follow x86 instructions: mov 0x8(%rax), %rcx mov 0x4(%rcx), %rdx <=== PMU sample mov %rdx, -4(%rbp) Let's say 'foo' is located in the %rax and it has a pointer to struct foo. But perf sample is captured in the second instruction and there is no variable or type info for the %rcx. It'd be great if compiler could generate debug info for %rcx, but we should handle it on our side. So this patch implements the logic to iterate instructions and update the type table for each location. As it already collected a list of scopes including the target instruction, we can use it to construct the type table smartly. +---------------- scope[0] subprogram | | +-------------- scope[1] lexical_block | | | | +------------ scope[2] inlined_subroutine | | | | | | +---------- scope[3] inlined_subroutine | | | | | | | | +-------- scope[4] lexical_block | | | | | | | | | | *** target instruction ... Image the target instruction has 5 scopes, each scope will have its own variables and parameters. Then it can start with the innermost scope (4). So it'd search the shortest path from the start of scope[4] to the target address and build a list of basic blocks. Then it iterates the basic blocks with the variables in the scope and update the table. If it finds a type at the target instruction, then returns it. Otherwise, it moves to the upper scope[3]. Now it'd search the shortest path from the start of scope[3] to the start of scope[4]. Then connect it to the existing basic block list. Then it'd iterate the blocks with variables for both scopes. It can repeat this until it finds a type at the target instruction or reaches to the top scope[0]. As the basic blocks contain the shortest path, it won't worry about branches and can update the table simply. The final check will be done by find_matching_type() in the next patch. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-15-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:06 -07:00
pr_debug_dtp("bb: [%"PRIx64" - %"PRIx64"]\n",
bb->begin->al.offset, bb->end->al.offset);
list_for_each_entry_from(dl, &notes->src->source, al.node) {
u64 this_ip = sym->start + dl->al.offset;
u64 addr = map__rip_2objdump(dloc->ms->map, this_ip);
/* Skip comment or debug info lines */
if (dl->al.offset == -1)
continue;
perf annotate-data: Implement instruction tracking If it failed to find a variable for the location directly, it might be due to a missing variable in the source code. For example, accessing pointer variables in a chain can result in the case like below: struct foo *foo = ...; int i = foo->bar->baz; The DWARF debug information is created for each variable so it'd have one for 'foo'. But there's no variable for 'foo->bar' and then it cannot know the type of 'bar' and 'baz'. The above source code can be compiled to the follow x86 instructions: mov 0x8(%rax), %rcx mov 0x4(%rcx), %rdx <=== PMU sample mov %rdx, -4(%rbp) Let's say 'foo' is located in the %rax and it has a pointer to struct foo. But perf sample is captured in the second instruction and there is no variable or type info for the %rcx. It'd be great if compiler could generate debug info for %rcx, but we should handle it on our side. So this patch implements the logic to iterate instructions and update the type table for each location. As it already collected a list of scopes including the target instruction, we can use it to construct the type table smartly. +---------------- scope[0] subprogram | | +-------------- scope[1] lexical_block | | | | +------------ scope[2] inlined_subroutine | | | | | | +---------- scope[3] inlined_subroutine | | | | | | | | +-------- scope[4] lexical_block | | | | | | | | | | *** target instruction ... Image the target instruction has 5 scopes, each scope will have its own variables and parameters. Then it can start with the innermost scope (4). So it'd search the shortest path from the start of scope[4] to the target address and build a list of basic blocks. Then it iterates the basic blocks with the variables in the scope and update the table. If it finds a type at the target instruction, then returns it. Otherwise, it moves to the upper scope[3]. Now it'd search the shortest path from the start of scope[3] to the start of scope[4]. Then connect it to the existing basic block list. Then it'd iterate the blocks with variables for both scopes. It can repeat this until it finds a type at the target instruction or reaches to the top scope[0]. As the basic blocks contain the shortest path, it won't worry about branches and can update the table simply. The final check will be done by find_matching_type() in the next patch. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-15-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:06 -07:00
/* Update variable type at this address */
update_var_state(&state, dloc, addr, dl->al.offset, var_types);
if (this_ip == dloc->ip) {
ret = check_matching_type(&state, dloc,
perf annotate-data: Update stack slot for the store When checking the match variable at the target instruction, it might not have any information if it's a first write to a stack slot. In this case it could spill a register value into the stack so the type info is in the source operand. But currently it's hard to get the operand from the checking function. Let's process the instruction and retry to get the type info from the stack if there's no information already. This is an example of __tcp_transmit_skb(). The instructions are <__tcp_transmit_skb>: 0: nopl 0x0(%rax, %rax, 1) 5: push %rbp 6: mov %rsp, %rbp 9: push %r15 b: push %r14 d: push %r13 f: push %r12 11: push %rbx 12: sub $0x98, %rsp 19: mov %r8d, -0xa8(%rbp) ... It cannot find any variable at -0xa8(%rbp) at this point. ----------------------------------------------------------- find data type for -0xa8(reg6) at __tcp_transmit_skb+0x19 CU for net/ipv4/tcp_output.c (die:0x817f543) frame base: cfa=0 fbreg=6 scope: [1/1] (die:81aac3e) bb: [0 - 19] var [0] -0x98(stack) type='struct tcp_out_options' size=0x28 (die:0x81af3df) var [5] reg8 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg2 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg1 type='int' size=0x4 (die:0x818059e) var [5] reg4 type='struct sk_buff*' size=0x8 (die:0x8181360) var [5] reg5 type='struct sock*' size=0x8 (die:0x8181a0c) chk [19] reg6 offset=-0xa8 ok=0 kind=0 fbreg : no type information no type information And it was able to find the type after processing the 'mov' instruction. ----------------------------------------------------------- find data type for -0xa8(reg6) at __tcp_transmit_skb+0x19 CU for net/ipv4/tcp_output.c (die:0x817f543) frame base: cfa=0 fbreg=6 scope: [1/1] (die:81aac3e) bb: [0 - 19] var [0] -0x98(stack) type='struct tcp_out_options' size=0x28 (die:0x81af3df) var [5] reg8 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg2 type='unsigned int' size=0x4 (die:0x8180ed6) var [5] reg1 type='int' size=0x4 (die:0x818059e) var [5] reg4 type='struct sk_buff*' size=0x8 (die:0x8181360) var [5] reg5 type='struct sock*' size=0x8 (die:0x8181a0c) chk [19] reg6 offset=-0xa8 ok=0 kind=0 fbreg : retry <<<--- here mov [19] reg8 -> -0xa8(stack) type='unsigned int' size=0x4 (die:0x8180ed6) chk [19] reg6 offset=-0xa8 ok=0 kind=0 fbreg : Good! found by insn track: -0xa8(reg6) type-offset=0 final result: type='unsigned int' size=0x4 (die:0x8180ed6) Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240821232628.353177-4-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-21 16:26:27 -07:00
cu_die, dl, type_die);
pr_debug_dtp(" : %s\n", match_result_str(ret));
perf annotate-data: Implement instruction tracking If it failed to find a variable for the location directly, it might be due to a missing variable in the source code. For example, accessing pointer variables in a chain can result in the case like below: struct foo *foo = ...; int i = foo->bar->baz; The DWARF debug information is created for each variable so it'd have one for 'foo'. But there's no variable for 'foo->bar' and then it cannot know the type of 'bar' and 'baz'. The above source code can be compiled to the follow x86 instructions: mov 0x8(%rax), %rcx mov 0x4(%rcx), %rdx <=== PMU sample mov %rdx, -4(%rbp) Let's say 'foo' is located in the %rax and it has a pointer to struct foo. But perf sample is captured in the second instruction and there is no variable or type info for the %rcx. It'd be great if compiler could generate debug info for %rcx, but we should handle it on our side. So this patch implements the logic to iterate instructions and update the type table for each location. As it already collected a list of scopes including the target instruction, we can use it to construct the type table smartly. +---------------- scope[0] subprogram | | +-------------- scope[1] lexical_block | | | | +------------ scope[2] inlined_subroutine | | | | | | +---------- scope[3] inlined_subroutine | | | | | | | | +-------- scope[4] lexical_block | | | | | | | | | | *** target instruction ... Image the target instruction has 5 scopes, each scope will have its own variables and parameters. Then it can start with the innermost scope (4). So it'd search the shortest path from the start of scope[4] to the target address and build a list of basic blocks. Then it iterates the basic blocks with the variables in the scope and update the table. If it finds a type at the target instruction, then returns it. Otherwise, it moves to the upper scope[3]. Now it'd search the shortest path from the start of scope[3] to the start of scope[4]. Then connect it to the existing basic block list. Then it'd iterate the blocks with variables for both scopes. It can repeat this until it finds a type at the target instruction or reaches to the top scope[0]. As the basic blocks contain the shortest path, it won't worry about branches and can update the table simply. The final check will be done by find_matching_type() in the next patch. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-15-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:06 -07:00
goto out;
}
/* Update type table after processing the instruction */
update_insn_state(&state, dloc, cu_die, dl);
if (dl == bb->end)
break;
}
}
out:
exit_type_state(&state);
return ret;
perf annotate-data: Implement instruction tracking If it failed to find a variable for the location directly, it might be due to a missing variable in the source code. For example, accessing pointer variables in a chain can result in the case like below: struct foo *foo = ...; int i = foo->bar->baz; The DWARF debug information is created for each variable so it'd have one for 'foo'. But there's no variable for 'foo->bar' and then it cannot know the type of 'bar' and 'baz'. The above source code can be compiled to the follow x86 instructions: mov 0x8(%rax), %rcx mov 0x4(%rcx), %rdx <=== PMU sample mov %rdx, -4(%rbp) Let's say 'foo' is located in the %rax and it has a pointer to struct foo. But perf sample is captured in the second instruction and there is no variable or type info for the %rcx. It'd be great if compiler could generate debug info for %rcx, but we should handle it on our side. So this patch implements the logic to iterate instructions and update the type table for each location. As it already collected a list of scopes including the target instruction, we can use it to construct the type table smartly. +---------------- scope[0] subprogram | | +-------------- scope[1] lexical_block | | | | +------------ scope[2] inlined_subroutine | | | | | | +---------- scope[3] inlined_subroutine | | | | | | | | +-------- scope[4] lexical_block | | | | | | | | | | *** target instruction ... Image the target instruction has 5 scopes, each scope will have its own variables and parameters. Then it can start with the innermost scope (4). So it'd search the shortest path from the start of scope[4] to the target address and build a list of basic blocks. Then it iterates the basic blocks with the variables in the scope and update the table. If it finds a type at the target instruction, then returns it. Otherwise, it moves to the upper scope[3]. Now it'd search the shortest path from the start of scope[3] to the start of scope[4]. Then connect it to the existing basic block list. Then it'd iterate the blocks with variables for both scopes. It can repeat this until it finds a type at the target instruction or reaches to the top scope[0]. As the basic blocks contain the shortest path, it won't worry about branches and can update the table simply. The final check will be done by find_matching_type() in the next patch. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-15-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:06 -07:00
}
static int arch_supports_insn_tracking(struct data_loc_info *dloc)
{
if ((arch__is(dloc->arch, "x86")) || (arch__is(dloc->arch, "powerpc")))
return 1;
return 0;
}
perf annotate-data: Implement instruction tracking If it failed to find a variable for the location directly, it might be due to a missing variable in the source code. For example, accessing pointer variables in a chain can result in the case like below: struct foo *foo = ...; int i = foo->bar->baz; The DWARF debug information is created for each variable so it'd have one for 'foo'. But there's no variable for 'foo->bar' and then it cannot know the type of 'bar' and 'baz'. The above source code can be compiled to the follow x86 instructions: mov 0x8(%rax), %rcx mov 0x4(%rcx), %rdx <=== PMU sample mov %rdx, -4(%rbp) Let's say 'foo' is located in the %rax and it has a pointer to struct foo. But perf sample is captured in the second instruction and there is no variable or type info for the %rcx. It'd be great if compiler could generate debug info for %rcx, but we should handle it on our side. So this patch implements the logic to iterate instructions and update the type table for each location. As it already collected a list of scopes including the target instruction, we can use it to construct the type table smartly. +---------------- scope[0] subprogram | | +-------------- scope[1] lexical_block | | | | +------------ scope[2] inlined_subroutine | | | | | | +---------- scope[3] inlined_subroutine | | | | | | | | +-------- scope[4] lexical_block | | | | | | | | | | *** target instruction ... Image the target instruction has 5 scopes, each scope will have its own variables and parameters. Then it can start with the innermost scope (4). So it'd search the shortest path from the start of scope[4] to the target address and build a list of basic blocks. Then it iterates the basic blocks with the variables in the scope and update the table. If it finds a type at the target instruction, then returns it. Otherwise, it moves to the upper scope[3]. Now it'd search the shortest path from the start of scope[3] to the start of scope[4]. Then connect it to the existing basic block list. Then it'd iterate the blocks with variables for both scopes. It can repeat this until it finds a type at the target instruction or reaches to the top scope[0]. As the basic blocks contain the shortest path, it won't worry about branches and can update the table simply. The final check will be done by find_matching_type() in the next patch. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-15-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:06 -07:00
/*
* Construct a list of basic blocks for each scope with variables and try to find
* the data type by updating a type state table through instructions.
*/
static enum type_match_result find_data_type_block(struct data_loc_info *dloc,
Dwarf_Die *cu_die,
Dwarf_Die *scopes,
int nr_scopes,
Dwarf_Die *type_die)
perf annotate-data: Implement instruction tracking If it failed to find a variable for the location directly, it might be due to a missing variable in the source code. For example, accessing pointer variables in a chain can result in the case like below: struct foo *foo = ...; int i = foo->bar->baz; The DWARF debug information is created for each variable so it'd have one for 'foo'. But there's no variable for 'foo->bar' and then it cannot know the type of 'bar' and 'baz'. The above source code can be compiled to the follow x86 instructions: mov 0x8(%rax), %rcx mov 0x4(%rcx), %rdx <=== PMU sample mov %rdx, -4(%rbp) Let's say 'foo' is located in the %rax and it has a pointer to struct foo. But perf sample is captured in the second instruction and there is no variable or type info for the %rcx. It'd be great if compiler could generate debug info for %rcx, but we should handle it on our side. So this patch implements the logic to iterate instructions and update the type table for each location. As it already collected a list of scopes including the target instruction, we can use it to construct the type table smartly. +---------------- scope[0] subprogram | | +-------------- scope[1] lexical_block | | | | +------------ scope[2] inlined_subroutine | | | | | | +---------- scope[3] inlined_subroutine | | | | | | | | +-------- scope[4] lexical_block | | | | | | | | | | *** target instruction ... Image the target instruction has 5 scopes, each scope will have its own variables and parameters. Then it can start with the innermost scope (4). So it'd search the shortest path from the start of scope[4] to the target address and build a list of basic blocks. Then it iterates the basic blocks with the variables in the scope and update the table. If it finds a type at the target instruction, then returns it. Otherwise, it moves to the upper scope[3]. Now it'd search the shortest path from the start of scope[3] to the start of scope[4]. Then connect it to the existing basic block list. Then it'd iterate the blocks with variables for both scopes. It can repeat this until it finds a type at the target instruction or reaches to the top scope[0]. As the basic blocks contain the shortest path, it won't worry about branches and can update the table simply. The final check will be done by find_matching_type() in the next patch. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-15-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:06 -07:00
{
LIST_HEAD(basic_blocks);
struct die_var_type *var_types = NULL;
u64 src_ip, dst_ip, prev_dst_ip;
enum type_match_result ret = PERF_TMR_UNKNOWN;
perf annotate-data: Implement instruction tracking If it failed to find a variable for the location directly, it might be due to a missing variable in the source code. For example, accessing pointer variables in a chain can result in the case like below: struct foo *foo = ...; int i = foo->bar->baz; The DWARF debug information is created for each variable so it'd have one for 'foo'. But there's no variable for 'foo->bar' and then it cannot know the type of 'bar' and 'baz'. The above source code can be compiled to the follow x86 instructions: mov 0x8(%rax), %rcx mov 0x4(%rcx), %rdx <=== PMU sample mov %rdx, -4(%rbp) Let's say 'foo' is located in the %rax and it has a pointer to struct foo. But perf sample is captured in the second instruction and there is no variable or type info for the %rcx. It'd be great if compiler could generate debug info for %rcx, but we should handle it on our side. So this patch implements the logic to iterate instructions and update the type table for each location. As it already collected a list of scopes including the target instruction, we can use it to construct the type table smartly. +---------------- scope[0] subprogram | | +-------------- scope[1] lexical_block | | | | +------------ scope[2] inlined_subroutine | | | | | | +---------- scope[3] inlined_subroutine | | | | | | | | +-------- scope[4] lexical_block | | | | | | | | | | *** target instruction ... Image the target instruction has 5 scopes, each scope will have its own variables and parameters. Then it can start with the innermost scope (4). So it'd search the shortest path from the start of scope[4] to the target address and build a list of basic blocks. Then it iterates the basic blocks with the variables in the scope and update the table. If it finds a type at the target instruction, then returns it. Otherwise, it moves to the upper scope[3]. Now it'd search the shortest path from the start of scope[3] to the start of scope[4]. Then connect it to the existing basic block list. Then it'd iterate the blocks with variables for both scopes. It can repeat this until it finds a type at the target instruction or reaches to the top scope[0]. As the basic blocks contain the shortest path, it won't worry about branches and can update the table simply. The final check will be done by find_matching_type() in the next patch. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-15-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:06 -07:00
/* TODO: other architecture support */
if (!arch_supports_insn_tracking(dloc))
return PERF_TMR_BAIL_OUT;
perf annotate-data: Implement instruction tracking If it failed to find a variable for the location directly, it might be due to a missing variable in the source code. For example, accessing pointer variables in a chain can result in the case like below: struct foo *foo = ...; int i = foo->bar->baz; The DWARF debug information is created for each variable so it'd have one for 'foo'. But there's no variable for 'foo->bar' and then it cannot know the type of 'bar' and 'baz'. The above source code can be compiled to the follow x86 instructions: mov 0x8(%rax), %rcx mov 0x4(%rcx), %rdx <=== PMU sample mov %rdx, -4(%rbp) Let's say 'foo' is located in the %rax and it has a pointer to struct foo. But perf sample is captured in the second instruction and there is no variable or type info for the %rcx. It'd be great if compiler could generate debug info for %rcx, but we should handle it on our side. So this patch implements the logic to iterate instructions and update the type table for each location. As it already collected a list of scopes including the target instruction, we can use it to construct the type table smartly. +---------------- scope[0] subprogram | | +-------------- scope[1] lexical_block | | | | +------------ scope[2] inlined_subroutine | | | | | | +---------- scope[3] inlined_subroutine | | | | | | | | +-------- scope[4] lexical_block | | | | | | | | | | *** target instruction ... Image the target instruction has 5 scopes, each scope will have its own variables and parameters. Then it can start with the innermost scope (4). So it'd search the shortest path from the start of scope[4] to the target address and build a list of basic blocks. Then it iterates the basic blocks with the variables in the scope and update the table. If it finds a type at the target instruction, then returns it. Otherwise, it moves to the upper scope[3]. Now it'd search the shortest path from the start of scope[3] to the start of scope[4]. Then connect it to the existing basic block list. Then it'd iterate the blocks with variables for both scopes. It can repeat this until it finds a type at the target instruction or reaches to the top scope[0]. As the basic blocks contain the shortest path, it won't worry about branches and can update the table simply. The final check will be done by find_matching_type() in the next patch. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-15-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:06 -07:00
prev_dst_ip = dst_ip = dloc->ip;
for (int i = nr_scopes - 1; i >= 0; i--) {
Dwarf_Addr base, start, end;
LIST_HEAD(this_blocks);
if (dwarf_ranges(&scopes[i], 0, &base, &start, &end) < 0)
break;
pr_debug_dtp("scope: [%d/%d] ", i + 1, nr_scopes);
pr_debug_scope(&scopes[i]);
perf annotate-data: Implement instruction tracking If it failed to find a variable for the location directly, it might be due to a missing variable in the source code. For example, accessing pointer variables in a chain can result in the case like below: struct foo *foo = ...; int i = foo->bar->baz; The DWARF debug information is created for each variable so it'd have one for 'foo'. But there's no variable for 'foo->bar' and then it cannot know the type of 'bar' and 'baz'. The above source code can be compiled to the follow x86 instructions: mov 0x8(%rax), %rcx mov 0x4(%rcx), %rdx <=== PMU sample mov %rdx, -4(%rbp) Let's say 'foo' is located in the %rax and it has a pointer to struct foo. But perf sample is captured in the second instruction and there is no variable or type info for the %rcx. It'd be great if compiler could generate debug info for %rcx, but we should handle it on our side. So this patch implements the logic to iterate instructions and update the type table for each location. As it already collected a list of scopes including the target instruction, we can use it to construct the type table smartly. +---------------- scope[0] subprogram | | +-------------- scope[1] lexical_block | | | | +------------ scope[2] inlined_subroutine | | | | | | +---------- scope[3] inlined_subroutine | | | | | | | | +-------- scope[4] lexical_block | | | | | | | | | | *** target instruction ... Image the target instruction has 5 scopes, each scope will have its own variables and parameters. Then it can start with the innermost scope (4). So it'd search the shortest path from the start of scope[4] to the target address and build a list of basic blocks. Then it iterates the basic blocks with the variables in the scope and update the table. If it finds a type at the target instruction, then returns it. Otherwise, it moves to the upper scope[3]. Now it'd search the shortest path from the start of scope[3] to the start of scope[4]. Then connect it to the existing basic block list. Then it'd iterate the blocks with variables for both scopes. It can repeat this until it finds a type at the target instruction or reaches to the top scope[0]. As the basic blocks contain the shortest path, it won't worry about branches and can update the table simply. The final check will be done by find_matching_type() in the next patch. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-15-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:06 -07:00
src_ip = map__objdump_2rip(dloc->ms->map, start);
again:
/* Get basic blocks for this scope */
if (annotate_get_basic_blocks(dloc->ms->sym, src_ip, dst_ip,
&this_blocks) < 0) {
/* Try previous block if they are not connected */
if (prev_dst_ip != dst_ip) {
dst_ip = prev_dst_ip;
goto again;
}
pr_debug_dtp("cannot find a basic block from %"PRIx64" to %"PRIx64"\n",
src_ip - dloc->ms->sym->start,
dst_ip - dloc->ms->sym->start);
continue;
}
prepend_basic_blocks(&this_blocks, &basic_blocks);
/* Get variable info for this scope and add to var_types list */
die_collect_vars(&scopes[i], &var_types);
fixup_var_address(var_types, start);
/* Find from start of this scope to the target instruction */
ret = find_data_type_insn(dloc, &basic_blocks, var_types,
cu_die, type_die);
if (ret == PERF_TMR_OK) {
char buf[64];
perf annotate-data: Update debug messages In check_matching_type(), it'd be easier to display the typename in question if it's available. For example, check out the line starts with 'chk'. ----------------------------------------------------------- find data type for 0x10(reg0) at cpuacct_charge+0x13 CU for kernel/sched/build_utility.c (die:0x137ee0b) frame base: cfa=1 fbreg=7 scope: [3/3] (die:13d9632) bb: [c - 13] var [c] reg5 type='struct task_struct*' size=0x8 (die:0x1381230) mov [c] 0xdf8(reg5) -> reg0 type='struct css_set*' size=0x8 (die:0x1385c56) chk [13] reg0 offset=0x10 ok=1 kind=1 (struct css_set*) : Good! <<<--- here found by insn track: 0x10(reg0) type-offset=0x10 final result: type='struct css_set' size=0x250 (die:0x1385b0e) Another example: ----------------------------------------------------------- find data type for 0x8(reg0) at menu_select+0x279 CU for drivers/cpuidle/governors/menu.c (die:0x7b0fe79) frame base: cfa=1 fbreg=7 scope: [2/2] (die:7b11010) bb: [273 - 277] bb: [279 - 279] chk [279] reg0 offset=0x8 ok=0 kind=0 cfa : no type information scope: [1/2] (die:7b10cbc) bb: [0 - 64] ... mov [26a] imm=0xffffffff -> reg15 bb: [273 - 277] bb: [279 - 279] chk [279] reg0 offset=0x8 ok=1 kind=1 (long long unsigned int) : no/void pointer <<<--- here final result: no/void pointer Also change some places to print negative offsets properly. Before: ----------------------------------------------------------- find data type for 0xffffff40(reg6) at __tcp_transmit_skb+0x58 After: ----------------------------------------------------------- find data type for -0xc0(reg6) at __tcp_transmit_skb+0x58 Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240821232628.353177-3-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-21 16:26:26 -07:00
int offset = dloc->op->offset;
const char *offset_sign = "";
if (offset < 0) {
offset = -offset;
offset_sign = "-";
}
if (dloc->op->multi_regs)
snprintf(buf, sizeof(buf), "reg%d, reg%d",
dloc->op->reg1, dloc->op->reg2);
else
snprintf(buf, sizeof(buf), "reg%d", dloc->op->reg1);
perf annotate-data: Update debug messages In check_matching_type(), it'd be easier to display the typename in question if it's available. For example, check out the line starts with 'chk'. ----------------------------------------------------------- find data type for 0x10(reg0) at cpuacct_charge+0x13 CU for kernel/sched/build_utility.c (die:0x137ee0b) frame base: cfa=1 fbreg=7 scope: [3/3] (die:13d9632) bb: [c - 13] var [c] reg5 type='struct task_struct*' size=0x8 (die:0x1381230) mov [c] 0xdf8(reg5) -> reg0 type='struct css_set*' size=0x8 (die:0x1385c56) chk [13] reg0 offset=0x10 ok=1 kind=1 (struct css_set*) : Good! <<<--- here found by insn track: 0x10(reg0) type-offset=0x10 final result: type='struct css_set' size=0x250 (die:0x1385b0e) Another example: ----------------------------------------------------------- find data type for 0x8(reg0) at menu_select+0x279 CU for drivers/cpuidle/governors/menu.c (die:0x7b0fe79) frame base: cfa=1 fbreg=7 scope: [2/2] (die:7b11010) bb: [273 - 277] bb: [279 - 279] chk [279] reg0 offset=0x8 ok=0 kind=0 cfa : no type information scope: [1/2] (die:7b10cbc) bb: [0 - 64] ... mov [26a] imm=0xffffffff -> reg15 bb: [273 - 277] bb: [279 - 279] chk [279] reg0 offset=0x8 ok=1 kind=1 (long long unsigned int) : no/void pointer <<<--- here final result: no/void pointer Also change some places to print negative offsets properly. Before: ----------------------------------------------------------- find data type for 0xffffff40(reg6) at __tcp_transmit_skb+0x58 After: ----------------------------------------------------------- find data type for -0xc0(reg6) at __tcp_transmit_skb+0x58 Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240821232628.353177-3-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-21 16:26:26 -07:00
pr_debug_dtp("found by insn track: %s%#x(%s) type-offset=%#x\n",
offset_sign, offset, buf, dloc->type_offset);
perf annotate-data: Implement instruction tracking If it failed to find a variable for the location directly, it might be due to a missing variable in the source code. For example, accessing pointer variables in a chain can result in the case like below: struct foo *foo = ...; int i = foo->bar->baz; The DWARF debug information is created for each variable so it'd have one for 'foo'. But there's no variable for 'foo->bar' and then it cannot know the type of 'bar' and 'baz'. The above source code can be compiled to the follow x86 instructions: mov 0x8(%rax), %rcx mov 0x4(%rcx), %rdx <=== PMU sample mov %rdx, -4(%rbp) Let's say 'foo' is located in the %rax and it has a pointer to struct foo. But perf sample is captured in the second instruction and there is no variable or type info for the %rcx. It'd be great if compiler could generate debug info for %rcx, but we should handle it on our side. So this patch implements the logic to iterate instructions and update the type table for each location. As it already collected a list of scopes including the target instruction, we can use it to construct the type table smartly. +---------------- scope[0] subprogram | | +-------------- scope[1] lexical_block | | | | +------------ scope[2] inlined_subroutine | | | | | | +---------- scope[3] inlined_subroutine | | | | | | | | +-------- scope[4] lexical_block | | | | | | | | | | *** target instruction ... Image the target instruction has 5 scopes, each scope will have its own variables and parameters. Then it can start with the innermost scope (4). So it'd search the shortest path from the start of scope[4] to the target address and build a list of basic blocks. Then it iterates the basic blocks with the variables in the scope and update the table. If it finds a type at the target instruction, then returns it. Otherwise, it moves to the upper scope[3]. Now it'd search the shortest path from the start of scope[3] to the start of scope[4]. Then connect it to the existing basic block list. Then it'd iterate the blocks with variables for both scopes. It can repeat this until it finds a type at the target instruction or reaches to the top scope[0]. As the basic blocks contain the shortest path, it won't worry about branches and can update the table simply. The final check will be done by find_matching_type() in the next patch. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-15-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:06 -07:00
break;
}
if (ret == PERF_TMR_BAIL_OUT)
break;
perf annotate-data: Implement instruction tracking If it failed to find a variable for the location directly, it might be due to a missing variable in the source code. For example, accessing pointer variables in a chain can result in the case like below: struct foo *foo = ...; int i = foo->bar->baz; The DWARF debug information is created for each variable so it'd have one for 'foo'. But there's no variable for 'foo->bar' and then it cannot know the type of 'bar' and 'baz'. The above source code can be compiled to the follow x86 instructions: mov 0x8(%rax), %rcx mov 0x4(%rcx), %rdx <=== PMU sample mov %rdx, -4(%rbp) Let's say 'foo' is located in the %rax and it has a pointer to struct foo. But perf sample is captured in the second instruction and there is no variable or type info for the %rcx. It'd be great if compiler could generate debug info for %rcx, but we should handle it on our side. So this patch implements the logic to iterate instructions and update the type table for each location. As it already collected a list of scopes including the target instruction, we can use it to construct the type table smartly. +---------------- scope[0] subprogram | | +-------------- scope[1] lexical_block | | | | +------------ scope[2] inlined_subroutine | | | | | | +---------- scope[3] inlined_subroutine | | | | | | | | +-------- scope[4] lexical_block | | | | | | | | | | *** target instruction ... Image the target instruction has 5 scopes, each scope will have its own variables and parameters. Then it can start with the innermost scope (4). So it'd search the shortest path from the start of scope[4] to the target address and build a list of basic blocks. Then it iterates the basic blocks with the variables in the scope and update the table. If it finds a type at the target instruction, then returns it. Otherwise, it moves to the upper scope[3]. Now it'd search the shortest path from the start of scope[3] to the start of scope[4]. Then connect it to the existing basic block list. Then it'd iterate the blocks with variables for both scopes. It can repeat this until it finds a type at the target instruction or reaches to the top scope[0]. As the basic blocks contain the shortest path, it won't worry about branches and can update the table simply. The final check will be done by find_matching_type() in the next patch. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-15-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:06 -07:00
/* Go up to the next scope and find blocks to the start */
prev_dst_ip = dst_ip;
dst_ip = src_ip;
}
delete_basic_blocks(&basic_blocks);
delete_var_types(var_types);
return ret;
}
/* The result will be saved in @type_die */
static int find_data_type_die(struct data_loc_info *dloc, Dwarf_Die *type_die)
{
struct annotated_op_loc *loc = dloc->op;
Dwarf_Die cu_die, var_die;
Dwarf_Die *scopes = NULL;
perf annotate-data: Update debug messages In check_matching_type(), it'd be easier to display the typename in question if it's available. For example, check out the line starts with 'chk'. ----------------------------------------------------------- find data type for 0x10(reg0) at cpuacct_charge+0x13 CU for kernel/sched/build_utility.c (die:0x137ee0b) frame base: cfa=1 fbreg=7 scope: [3/3] (die:13d9632) bb: [c - 13] var [c] reg5 type='struct task_struct*' size=0x8 (die:0x1381230) mov [c] 0xdf8(reg5) -> reg0 type='struct css_set*' size=0x8 (die:0x1385c56) chk [13] reg0 offset=0x10 ok=1 kind=1 (struct css_set*) : Good! <<<--- here found by insn track: 0x10(reg0) type-offset=0x10 final result: type='struct css_set' size=0x250 (die:0x1385b0e) Another example: ----------------------------------------------------------- find data type for 0x8(reg0) at menu_select+0x279 CU for drivers/cpuidle/governors/menu.c (die:0x7b0fe79) frame base: cfa=1 fbreg=7 scope: [2/2] (die:7b11010) bb: [273 - 277] bb: [279 - 279] chk [279] reg0 offset=0x8 ok=0 kind=0 cfa : no type information scope: [1/2] (die:7b10cbc) bb: [0 - 64] ... mov [26a] imm=0xffffffff -> reg15 bb: [273 - 277] bb: [279 - 279] chk [279] reg0 offset=0x8 ok=1 kind=1 (long long unsigned int) : no/void pointer <<<--- here final result: no/void pointer Also change some places to print negative offsets properly. Before: ----------------------------------------------------------- find data type for 0xffffff40(reg6) at __tcp_transmit_skb+0x58 After: ----------------------------------------------------------- find data type for -0xc0(reg6) at __tcp_transmit_skb+0x58 Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240821232628.353177-3-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-21 16:26:26 -07:00
int reg, offset = loc->offset;
int ret = -1;
int i, nr_scopes;
perf annotate-data: Support stack variables Local variables are allocated in the stack and the location list should look like base register(s) and an offset. Extend the die_find_variable_by_reg() to handle the following expressions * DW_OP_breg{0..31} * DW_OP_bregx * DW_OP_fbreg Ususally DWARF subprogram entries have frame base information and use it to locate stack variable like below: <2><43d1575>: Abbrev Number: 62 (DW_TAG_variable) <43d1576> DW_AT_location : 2 byte block: 91 7c (DW_OP_fbreg: -4) <--- here <43d1579> DW_AT_name : (indirect string, offset: 0x2c00c9): i <43d157d> DW_AT_decl_file : 1 <43d157e> DW_AT_decl_line : 78 <43d157f> DW_AT_type : <0x43d19d7> I found some differences on saving the frame base between gcc and clang. The gcc uses the CFA to get the base so it needs to check the current frame's CFI info. In this case, stack offset needs to be adjusted from the start of the CFA. <1><1bb8d>: Abbrev Number: 102 (DW_TAG_subprogram) <1bb8e> DW_AT_name : (indirect string, offset: 0x74d41): kernel_init <1bb92> DW_AT_decl_file : 2 <1bb92> DW_AT_decl_line : 1440 <1bb94> DW_AT_decl_column : 18 <1bb95> DW_AT_prototyped : 1 <1bb95> DW_AT_type : <0xcc> <1bb99> DW_AT_low_pc : 0xffffffff81bab9e0 <1bba1> DW_AT_high_pc : 0x1b2 <1bba9> DW_AT_frame_base : 1 byte block: 9c (DW_OP_call_frame_cfa) <------ here <1bbab> DW_AT_call_all_calls: 1 <1bbab> DW_AT_sibling : <0x1bf5a> While clang sets it to a register directly and it can check the register and offset in the instruction directly. <1><43d1542>: Abbrev Number: 60 (DW_TAG_subprogram) <43d1543> DW_AT_low_pc : 0xffffffff816a7c60 <43d154b> DW_AT_high_pc : 0x98 <43d154f> DW_AT_frame_base : 1 byte block: 56 (DW_OP_reg6 (rbp)) <---------- here <43d1551> DW_AT_GNU_all_call_sites: 1 <43d1551> DW_AT_name : (indirect string, offset: 0x3bce91): foo <43d1555> DW_AT_decl_file : 1 <43d1556> DW_AT_decl_line : 75 <43d1557> DW_AT_prototyped : 1 <43d1557> DW_AT_type : <0x43c7332> <43d155b> DW_AT_external : 1 Also it needs to update the offset after finding the type like global variables since the offset was from the frame base. Factor out match_var_offset() to check global and local variables in the same way. The type stats are improved too: Annotate data type stats: total 294, ok 160 (54.4%), bad 134 (45.6%) ----------------------------------------------------------- 30 : no_sym 32 : no_mem_ops 51 : no_var 14 : no_typeinfo 7 : bad_offset Reviewed-by: Ian Rogers <irogers@google.com> Cc: Stephane Eranian <eranian@google.com> Cc: Masami Hiramatsu <mhiramat@kernel.org> Link: https://lore.kernel.org/r/20240117062657.985479-9-namhyung@kernel.org Signed-off-by: Namhyung Kim <namhyung@kernel.org>
2024-01-16 22:26:56 -08:00
int fbreg = -1;
int fb_offset = 0;
bool is_fbreg = false;
perf annotate-data: Check variables in every scope Sometimes it matches a variable in the inner scope but it fails because the actual access can be on a different type. Let's try variables in every scope and choose the best one using is_better_type(). I have an example with update_blocked_averages(), at first it found a variable (__mptr) but it's a void pointer. So it moved on to the upper scope and found another variable (cfs_rq). $ perf --debug type-profile annotate --data-type --stdio ... ----------------------------------------------------------- find data type for 0x140(reg14) at update_blocked_averages+0x2db CU for kernel/sched/fair.c (die:0x12dd892) frame base: cfa=1 fbreg=7 found "__mptr" (die: 0x13022f1) in scope=4/4 (die: 0x13022e8) failed: no/void pointer variable location: base=reg14, offset=0x140 type='void*' size=0x8 (die:0x12dd8f9) found "cfs_rq" (die: 0x1301721) in scope=3/4 (die: 0x130171c) type_offset=0x140 variable location: reg14 type='struct cfs_rq' size=0x1c0 (die:0x12e37e5) final type: type='struct cfs_rq' size=0x1c0 (die:0x12e37e5) IIUC the scope is like below: 1: update_blocked_averages 2: __update_blocked_fair 3: for_each_leaf_cfs_rq_safe 4: list_entry -> (container_of) The container_of is implemented like: #define container_of(ptr, type, member) ({ \ void *__mptr = (void *)(ptr); \ static_assert(__same_type(*(ptr), ((type *)0)->member) || \ __same_type(*(ptr), void), \ "pointer type mismatch in container_of()"); \ ((type *)(__mptr - offsetof(type, member))); }) That's why we see the __mptr variable first but it failed since it has no type information. Then for_each_leaf_cfs_rq_safe() is defined as #define for_each_leaf_cfs_rq_safe(rq, cfs_rq, pos) \ list_for_each_entry_safe(cfs_rq, pos, &rq->leaf_cfs_rq_list, \ leaf_cfs_rq_list) Note that the access was 0x140(r14). And the cfs_rq has leaf_cfs_rq_list at the 0x140. So it converts the list_head pointer to a pointer to struct cfs_rq here. $ pahole --hex -C cfs_rq vmlinux | grep 140 struct cfs_rq struct list_head leaf_cfs_rq_list; /* 0x140 0x10 */ Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240816235840.2754937-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-16 16:58:38 -07:00
bool found = false;
u64 pc;
perf annotate-data: Add debug messages Add a new debug option "type-profile" to enable the detailed info during the type analysis especially for instruction tracking. You can use this before the command name like 'report' or 'annotate'. $ perf --debug type-profile annotate --data-type Committer testing: First get some memory events: $ perf mem record ls Then, without data-type profiling debug: $ perf annotate --data-type | head Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ And with only data-type profiling: $ perf --debug type-profile annotate --data-type | head ----------------------------------------------------------- find_data_type_die [1e67] for reg13873052 (PC) offset=0x150e2 in dl_main CU die offset: 0x29cd3 found PC-rel by addr=0x34020 offset=0x20 ----------------------------------------------------------- find_data_type_die [2e] for reg12 offset=0 in __GI___readdir64 CU die offset: 0x137a45 frame base: cfa=1 fbreg=-1 found "__futex" in scope=2/2 (die: 0x137ad5) 0(reg12) type=int (die:2a) ----------------------------------------------------------- find_data_type_die [52] for reg5 offset=0 in __memmove_avx_unaligned_erms CU die offset: 0x1124ed no variable found Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:00 -07:00
char buf[64];
enum type_match_result result = PERF_TMR_UNKNOWN;
perf annotate-data: Update debug messages In check_matching_type(), it'd be easier to display the typename in question if it's available. For example, check out the line starts with 'chk'. ----------------------------------------------------------- find data type for 0x10(reg0) at cpuacct_charge+0x13 CU for kernel/sched/build_utility.c (die:0x137ee0b) frame base: cfa=1 fbreg=7 scope: [3/3] (die:13d9632) bb: [c - 13] var [c] reg5 type='struct task_struct*' size=0x8 (die:0x1381230) mov [c] 0xdf8(reg5) -> reg0 type='struct css_set*' size=0x8 (die:0x1385c56) chk [13] reg0 offset=0x10 ok=1 kind=1 (struct css_set*) : Good! <<<--- here found by insn track: 0x10(reg0) type-offset=0x10 final result: type='struct css_set' size=0x250 (die:0x1385b0e) Another example: ----------------------------------------------------------- find data type for 0x8(reg0) at menu_select+0x279 CU for drivers/cpuidle/governors/menu.c (die:0x7b0fe79) frame base: cfa=1 fbreg=7 scope: [2/2] (die:7b11010) bb: [273 - 277] bb: [279 - 279] chk [279] reg0 offset=0x8 ok=0 kind=0 cfa : no type information scope: [1/2] (die:7b10cbc) bb: [0 - 64] ... mov [26a] imm=0xffffffff -> reg15 bb: [273 - 277] bb: [279 - 279] chk [279] reg0 offset=0x8 ok=1 kind=1 (long long unsigned int) : no/void pointer <<<--- here final result: no/void pointer Also change some places to print negative offsets properly. Before: ----------------------------------------------------------- find data type for 0xffffff40(reg6) at __tcp_transmit_skb+0x58 After: ----------------------------------------------------------- find data type for -0xc0(reg6) at __tcp_transmit_skb+0x58 Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240821232628.353177-3-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-21 16:26:26 -07:00
const char *offset_sign = "";
perf annotate-data: Add debug messages Add a new debug option "type-profile" to enable the detailed info during the type analysis especially for instruction tracking. You can use this before the command name like 'report' or 'annotate'. $ perf --debug type-profile annotate --data-type Committer testing: First get some memory events: $ perf mem record ls Then, without data-type profiling debug: $ perf annotate --data-type | head Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ And with only data-type profiling: $ perf --debug type-profile annotate --data-type | head ----------------------------------------------------------- find_data_type_die [1e67] for reg13873052 (PC) offset=0x150e2 in dl_main CU die offset: 0x29cd3 found PC-rel by addr=0x34020 offset=0x20 ----------------------------------------------------------- find_data_type_die [2e] for reg12 offset=0 in __GI___readdir64 CU die offset: 0x137a45 frame base: cfa=1 fbreg=-1 found "__futex" in scope=2/2 (die: 0x137ad5) 0(reg12) type=int (die:2a) ----------------------------------------------------------- find_data_type_die [52] for reg5 offset=0 in __memmove_avx_unaligned_erms CU die offset: 0x1124ed no variable found Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:00 -07:00
if (dloc->op->multi_regs)
perf annotate-data: Improve debug message with location info To verify it found the correct variable, let's add the location expression to the debug message. $ perf --debug type-profile annotate --data-type ... ----------------------------------------------------------- find data type for 0xaf0(reg15) at schedule+0xeb CU for kernel/sched/core.c (die:0x1180523) frame base: cfa=0 fbreg=6 found "rq" in scope=3/4 (die: 0x11b6a00) type_offset=0xaf0 variable location: reg15 type='struct rq' size=0xfc0 (die:0x11892e2) ----------------------------------------------------------- find data type for 0x7bc(reg3) at tcp_get_info+0x62 CU for net/ipv4/tcp.c (die:0x7b5f516) frame base: cfa=0 fbreg=6 offset: 1980 is bigger than size: 760 check variable "sk" failed (die: 0x7b92b2c) variable location: reg3 type='struct sock' size=0x2f8 (die:0x7b63c3a) ----------------------------------------------------------- ... The first case is fine. It looked up a data type in r15 with offset of 0xaf0 at schedule+0xeb. It found the CU die and the frame base info and the variable "rq" was found in the scope 3/4. Its location is the r15 register and the type size is 0xfc0 which includes 0xaf0. But the second case is not good. It looked up a data type in rbx (reg3) with offset 0x7bc. It found a CU and the frame base which is good so far. And it also found a variable "sk" but the access offset is bigger than the type size (1980 vs. 760 or 0x7bc vs. 0x2f8). The variable has the right location (reg3) but I need to figure out why it accesses beyond what it's supposed to. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240412183310.2518474-2-namhyung@kernel.org [ Fix the build on 32-bit by casting Dwarf_Word to (long) in pr_debug_location() ] Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-04-12 11:33:07 -07:00
snprintf(buf, sizeof(buf), "reg%d, reg%d", dloc->op->reg1, dloc->op->reg2);
perf annotate-data: Add debug messages Add a new debug option "type-profile" to enable the detailed info during the type analysis especially for instruction tracking. You can use this before the command name like 'report' or 'annotate'. $ perf --debug type-profile annotate --data-type Committer testing: First get some memory events: $ perf mem record ls Then, without data-type profiling debug: $ perf annotate --data-type | head Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ And with only data-type profiling: $ perf --debug type-profile annotate --data-type | head ----------------------------------------------------------- find_data_type_die [1e67] for reg13873052 (PC) offset=0x150e2 in dl_main CU die offset: 0x29cd3 found PC-rel by addr=0x34020 offset=0x20 ----------------------------------------------------------- find_data_type_die [2e] for reg12 offset=0 in __GI___readdir64 CU die offset: 0x137a45 frame base: cfa=1 fbreg=-1 found "__futex" in scope=2/2 (die: 0x137ad5) 0(reg12) type=int (die:2a) ----------------------------------------------------------- find_data_type_die [52] for reg5 offset=0 in __memmove_avx_unaligned_erms CU die offset: 0x1124ed no variable found Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:00 -07:00
else if (dloc->op->reg1 == DWARF_REG_PC)
perf annotate-data: Improve debug message with location info To verify it found the correct variable, let's add the location expression to the debug message. $ perf --debug type-profile annotate --data-type ... ----------------------------------------------------------- find data type for 0xaf0(reg15) at schedule+0xeb CU for kernel/sched/core.c (die:0x1180523) frame base: cfa=0 fbreg=6 found "rq" in scope=3/4 (die: 0x11b6a00) type_offset=0xaf0 variable location: reg15 type='struct rq' size=0xfc0 (die:0x11892e2) ----------------------------------------------------------- find data type for 0x7bc(reg3) at tcp_get_info+0x62 CU for net/ipv4/tcp.c (die:0x7b5f516) frame base: cfa=0 fbreg=6 offset: 1980 is bigger than size: 760 check variable "sk" failed (die: 0x7b92b2c) variable location: reg3 type='struct sock' size=0x2f8 (die:0x7b63c3a) ----------------------------------------------------------- ... The first case is fine. It looked up a data type in r15 with offset of 0xaf0 at schedule+0xeb. It found the CU die and the frame base info and the variable "rq" was found in the scope 3/4. Its location is the r15 register and the type size is 0xfc0 which includes 0xaf0. But the second case is not good. It looked up a data type in rbx (reg3) with offset 0x7bc. It found a CU and the frame base which is good so far. And it also found a variable "sk" but the access offset is bigger than the type size (1980 vs. 760 or 0x7bc vs. 0x2f8). The variable has the right location (reg3) but I need to figure out why it accesses beyond what it's supposed to. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240412183310.2518474-2-namhyung@kernel.org [ Fix the build on 32-bit by casting Dwarf_Word to (long) in pr_debug_location() ] Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-04-12 11:33:07 -07:00
snprintf(buf, sizeof(buf), "PC");
perf annotate-data: Add debug messages Add a new debug option "type-profile" to enable the detailed info during the type analysis especially for instruction tracking. You can use this before the command name like 'report' or 'annotate'. $ perf --debug type-profile annotate --data-type Committer testing: First get some memory events: $ perf mem record ls Then, without data-type profiling debug: $ perf annotate --data-type | head Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ And with only data-type profiling: $ perf --debug type-profile annotate --data-type | head ----------------------------------------------------------- find_data_type_die [1e67] for reg13873052 (PC) offset=0x150e2 in dl_main CU die offset: 0x29cd3 found PC-rel by addr=0x34020 offset=0x20 ----------------------------------------------------------- find_data_type_die [2e] for reg12 offset=0 in __GI___readdir64 CU die offset: 0x137a45 frame base: cfa=1 fbreg=-1 found "__futex" in scope=2/2 (die: 0x137ad5) 0(reg12) type=int (die:2a) ----------------------------------------------------------- find_data_type_die [52] for reg5 offset=0 in __memmove_avx_unaligned_erms CU die offset: 0x1124ed no variable found Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:00 -07:00
else
perf annotate-data: Improve debug message with location info To verify it found the correct variable, let's add the location expression to the debug message. $ perf --debug type-profile annotate --data-type ... ----------------------------------------------------------- find data type for 0xaf0(reg15) at schedule+0xeb CU for kernel/sched/core.c (die:0x1180523) frame base: cfa=0 fbreg=6 found "rq" in scope=3/4 (die: 0x11b6a00) type_offset=0xaf0 variable location: reg15 type='struct rq' size=0xfc0 (die:0x11892e2) ----------------------------------------------------------- find data type for 0x7bc(reg3) at tcp_get_info+0x62 CU for net/ipv4/tcp.c (die:0x7b5f516) frame base: cfa=0 fbreg=6 offset: 1980 is bigger than size: 760 check variable "sk" failed (die: 0x7b92b2c) variable location: reg3 type='struct sock' size=0x2f8 (die:0x7b63c3a) ----------------------------------------------------------- ... The first case is fine. It looked up a data type in r15 with offset of 0xaf0 at schedule+0xeb. It found the CU die and the frame base info and the variable "rq" was found in the scope 3/4. Its location is the r15 register and the type size is 0xfc0 which includes 0xaf0. But the second case is not good. It looked up a data type in rbx (reg3) with offset 0x7bc. It found a CU and the frame base which is good so far. And it also found a variable "sk" but the access offset is bigger than the type size (1980 vs. 760 or 0x7bc vs. 0x2f8). The variable has the right location (reg3) but I need to figure out why it accesses beyond what it's supposed to. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240412183310.2518474-2-namhyung@kernel.org [ Fix the build on 32-bit by casting Dwarf_Word to (long) in pr_debug_location() ] Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-04-12 11:33:07 -07:00
snprintf(buf, sizeof(buf), "reg%d", dloc->op->reg1);
perf annotate-data: Add debug messages Add a new debug option "type-profile" to enable the detailed info during the type analysis especially for instruction tracking. You can use this before the command name like 'report' or 'annotate'. $ perf --debug type-profile annotate --data-type Committer testing: First get some memory events: $ perf mem record ls Then, without data-type profiling debug: $ perf annotate --data-type | head Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ And with only data-type profiling: $ perf --debug type-profile annotate --data-type | head ----------------------------------------------------------- find_data_type_die [1e67] for reg13873052 (PC) offset=0x150e2 in dl_main CU die offset: 0x29cd3 found PC-rel by addr=0x34020 offset=0x20 ----------------------------------------------------------- find_data_type_die [2e] for reg12 offset=0 in __GI___readdir64 CU die offset: 0x137a45 frame base: cfa=1 fbreg=-1 found "__futex" in scope=2/2 (die: 0x137ad5) 0(reg12) type=int (die:2a) ----------------------------------------------------------- find_data_type_die [52] for reg5 offset=0 in __memmove_avx_unaligned_erms CU die offset: 0x1124ed no variable found Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:00 -07:00
perf annotate-data: Update debug messages In check_matching_type(), it'd be easier to display the typename in question if it's available. For example, check out the line starts with 'chk'. ----------------------------------------------------------- find data type for 0x10(reg0) at cpuacct_charge+0x13 CU for kernel/sched/build_utility.c (die:0x137ee0b) frame base: cfa=1 fbreg=7 scope: [3/3] (die:13d9632) bb: [c - 13] var [c] reg5 type='struct task_struct*' size=0x8 (die:0x1381230) mov [c] 0xdf8(reg5) -> reg0 type='struct css_set*' size=0x8 (die:0x1385c56) chk [13] reg0 offset=0x10 ok=1 kind=1 (struct css_set*) : Good! <<<--- here found by insn track: 0x10(reg0) type-offset=0x10 final result: type='struct css_set' size=0x250 (die:0x1385b0e) Another example: ----------------------------------------------------------- find data type for 0x8(reg0) at menu_select+0x279 CU for drivers/cpuidle/governors/menu.c (die:0x7b0fe79) frame base: cfa=1 fbreg=7 scope: [2/2] (die:7b11010) bb: [273 - 277] bb: [279 - 279] chk [279] reg0 offset=0x8 ok=0 kind=0 cfa : no type information scope: [1/2] (die:7b10cbc) bb: [0 - 64] ... mov [26a] imm=0xffffffff -> reg15 bb: [273 - 277] bb: [279 - 279] chk [279] reg0 offset=0x8 ok=1 kind=1 (long long unsigned int) : no/void pointer <<<--- here final result: no/void pointer Also change some places to print negative offsets properly. Before: ----------------------------------------------------------- find data type for 0xffffff40(reg6) at __tcp_transmit_skb+0x58 After: ----------------------------------------------------------- find data type for -0xc0(reg6) at __tcp_transmit_skb+0x58 Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240821232628.353177-3-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-21 16:26:26 -07:00
if (offset < 0) {
offset = -offset;
offset_sign = "-";
}
perf annotate-data: Add debug messages Add a new debug option "type-profile" to enable the detailed info during the type analysis especially for instruction tracking. You can use this before the command name like 'report' or 'annotate'. $ perf --debug type-profile annotate --data-type Committer testing: First get some memory events: $ perf mem record ls Then, without data-type profiling debug: $ perf annotate --data-type | head Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ And with only data-type profiling: $ perf --debug type-profile annotate --data-type | head ----------------------------------------------------------- find_data_type_die [1e67] for reg13873052 (PC) offset=0x150e2 in dl_main CU die offset: 0x29cd3 found PC-rel by addr=0x34020 offset=0x20 ----------------------------------------------------------- find_data_type_die [2e] for reg12 offset=0 in __GI___readdir64 CU die offset: 0x137a45 frame base: cfa=1 fbreg=-1 found "__futex" in scope=2/2 (die: 0x137ad5) 0(reg12) type=int (die:2a) ----------------------------------------------------------- find_data_type_die [52] for reg5 offset=0 in __memmove_avx_unaligned_erms CU die offset: 0x1124ed no variable found Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:00 -07:00
pr_debug_dtp("-----------------------------------------------------------\n");
perf annotate-data: Update debug messages In check_matching_type(), it'd be easier to display the typename in question if it's available. For example, check out the line starts with 'chk'. ----------------------------------------------------------- find data type for 0x10(reg0) at cpuacct_charge+0x13 CU for kernel/sched/build_utility.c (die:0x137ee0b) frame base: cfa=1 fbreg=7 scope: [3/3] (die:13d9632) bb: [c - 13] var [c] reg5 type='struct task_struct*' size=0x8 (die:0x1381230) mov [c] 0xdf8(reg5) -> reg0 type='struct css_set*' size=0x8 (die:0x1385c56) chk [13] reg0 offset=0x10 ok=1 kind=1 (struct css_set*) : Good! <<<--- here found by insn track: 0x10(reg0) type-offset=0x10 final result: type='struct css_set' size=0x250 (die:0x1385b0e) Another example: ----------------------------------------------------------- find data type for 0x8(reg0) at menu_select+0x279 CU for drivers/cpuidle/governors/menu.c (die:0x7b0fe79) frame base: cfa=1 fbreg=7 scope: [2/2] (die:7b11010) bb: [273 - 277] bb: [279 - 279] chk [279] reg0 offset=0x8 ok=0 kind=0 cfa : no type information scope: [1/2] (die:7b10cbc) bb: [0 - 64] ... mov [26a] imm=0xffffffff -> reg15 bb: [273 - 277] bb: [279 - 279] chk [279] reg0 offset=0x8 ok=1 kind=1 (long long unsigned int) : no/void pointer <<<--- here final result: no/void pointer Also change some places to print negative offsets properly. Before: ----------------------------------------------------------- find data type for 0xffffff40(reg6) at __tcp_transmit_skb+0x58 After: ----------------------------------------------------------- find data type for -0xc0(reg6) at __tcp_transmit_skb+0x58 Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240821232628.353177-3-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-21 16:26:26 -07:00
pr_debug_dtp("find data type for %s%#x(%s) at %s+%#"PRIx64"\n",
offset_sign, offset, buf,
dloc->ms->sym->name, dloc->ip - dloc->ms->sym->start);
/*
* IP is a relative instruction address from the start of the map, as
* it can be randomized/relocated, it needs to translate to PC which is
* a file address for DWARF processing.
*/
pc = map__rip_2objdump(dloc->ms->map, dloc->ip);
/* Get a compile_unit for this address */
if (!find_cu_die(dloc->di, pc, &cu_die)) {
perf annotate-data: Add debug messages Add a new debug option "type-profile" to enable the detailed info during the type analysis especially for instruction tracking. You can use this before the command name like 'report' or 'annotate'. $ perf --debug type-profile annotate --data-type Committer testing: First get some memory events: $ perf mem record ls Then, without data-type profiling debug: $ perf annotate --data-type | head Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ And with only data-type profiling: $ perf --debug type-profile annotate --data-type | head ----------------------------------------------------------- find_data_type_die [1e67] for reg13873052 (PC) offset=0x150e2 in dl_main CU die offset: 0x29cd3 found PC-rel by addr=0x34020 offset=0x20 ----------------------------------------------------------- find_data_type_die [2e] for reg12 offset=0 in __GI___readdir64 CU die offset: 0x137a45 frame base: cfa=1 fbreg=-1 found "__futex" in scope=2/2 (die: 0x137ad5) 0(reg12) type=int (die:2a) ----------------------------------------------------------- find_data_type_die [52] for reg5 offset=0 in __memmove_avx_unaligned_erms CU die offset: 0x1124ed no variable found Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:00 -07:00
pr_debug_dtp("cannot find CU for address %"PRIx64"\n", pc);
ann_data_stat.no_cuinfo++;
return -1;
}
reg = loc->reg1;
offset = loc->offset;
perf annotate-data: Improve debug message with location info To verify it found the correct variable, let's add the location expression to the debug message. $ perf --debug type-profile annotate --data-type ... ----------------------------------------------------------- find data type for 0xaf0(reg15) at schedule+0xeb CU for kernel/sched/core.c (die:0x1180523) frame base: cfa=0 fbreg=6 found "rq" in scope=3/4 (die: 0x11b6a00) type_offset=0xaf0 variable location: reg15 type='struct rq' size=0xfc0 (die:0x11892e2) ----------------------------------------------------------- find data type for 0x7bc(reg3) at tcp_get_info+0x62 CU for net/ipv4/tcp.c (die:0x7b5f516) frame base: cfa=0 fbreg=6 offset: 1980 is bigger than size: 760 check variable "sk" failed (die: 0x7b92b2c) variable location: reg3 type='struct sock' size=0x2f8 (die:0x7b63c3a) ----------------------------------------------------------- ... The first case is fine. It looked up a data type in r15 with offset of 0xaf0 at schedule+0xeb. It found the CU die and the frame base info and the variable "rq" was found in the scope 3/4. Its location is the r15 register and the type size is 0xfc0 which includes 0xaf0. But the second case is not good. It looked up a data type in rbx (reg3) with offset 0x7bc. It found a CU and the frame base which is good so far. And it also found a variable "sk" but the access offset is bigger than the type size (1980 vs. 760 or 0x7bc vs. 0x2f8). The variable has the right location (reg3) but I need to figure out why it accesses beyond what it's supposed to. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240412183310.2518474-2-namhyung@kernel.org [ Fix the build on 32-bit by casting Dwarf_Word to (long) in pr_debug_location() ] Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-04-12 11:33:07 -07:00
pr_debug_dtp("CU for %s (die:%#lx)\n",
dwarf_diename(&cu_die), (long)dwarf_dieoffset(&cu_die));
perf annotate-data: Add debug messages Add a new debug option "type-profile" to enable the detailed info during the type analysis especially for instruction tracking. You can use this before the command name like 'report' or 'annotate'. $ perf --debug type-profile annotate --data-type Committer testing: First get some memory events: $ perf mem record ls Then, without data-type profiling debug: $ perf annotate --data-type | head Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ And with only data-type profiling: $ perf --debug type-profile annotate --data-type | head ----------------------------------------------------------- find_data_type_die [1e67] for reg13873052 (PC) offset=0x150e2 in dl_main CU die offset: 0x29cd3 found PC-rel by addr=0x34020 offset=0x20 ----------------------------------------------------------- find_data_type_die [2e] for reg12 offset=0 in __GI___readdir64 CU die offset: 0x137a45 frame base: cfa=1 fbreg=-1 found "__futex" in scope=2/2 (die: 0x137ad5) 0(reg12) type=int (die:2a) ----------------------------------------------------------- find_data_type_die [52] for reg5 offset=0 in __memmove_avx_unaligned_erms CU die offset: 0x1124ed no variable found Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:00 -07:00
if (reg == DWARF_REG_PC) {
if (get_global_var_type(&cu_die, dloc, dloc->ip, dloc->var_addr,
&offset, type_die)) {
dloc->type_offset = offset;
perf annotate-data: Add debug messages Add a new debug option "type-profile" to enable the detailed info during the type analysis especially for instruction tracking. You can use this before the command name like 'report' or 'annotate'. $ perf --debug type-profile annotate --data-type Committer testing: First get some memory events: $ perf mem record ls Then, without data-type profiling debug: $ perf annotate --data-type | head Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ And with only data-type profiling: $ perf --debug type-profile annotate --data-type | head ----------------------------------------------------------- find_data_type_die [1e67] for reg13873052 (PC) offset=0x150e2 in dl_main CU die offset: 0x29cd3 found PC-rel by addr=0x34020 offset=0x20 ----------------------------------------------------------- find_data_type_die [2e] for reg12 offset=0 in __GI___readdir64 CU die offset: 0x137a45 frame base: cfa=1 fbreg=-1 found "__futex" in scope=2/2 (die: 0x137ad5) 0(reg12) type=int (die:2a) ----------------------------------------------------------- find_data_type_die [52] for reg5 offset=0 in __memmove_avx_unaligned_erms CU die offset: 0x1124ed no variable found Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:00 -07:00
perf annotate-data: Improve debug message with location info To verify it found the correct variable, let's add the location expression to the debug message. $ perf --debug type-profile annotate --data-type ... ----------------------------------------------------------- find data type for 0xaf0(reg15) at schedule+0xeb CU for kernel/sched/core.c (die:0x1180523) frame base: cfa=0 fbreg=6 found "rq" in scope=3/4 (die: 0x11b6a00) type_offset=0xaf0 variable location: reg15 type='struct rq' size=0xfc0 (die:0x11892e2) ----------------------------------------------------------- find data type for 0x7bc(reg3) at tcp_get_info+0x62 CU for net/ipv4/tcp.c (die:0x7b5f516) frame base: cfa=0 fbreg=6 offset: 1980 is bigger than size: 760 check variable "sk" failed (die: 0x7b92b2c) variable location: reg3 type='struct sock' size=0x2f8 (die:0x7b63c3a) ----------------------------------------------------------- ... The first case is fine. It looked up a data type in r15 with offset of 0xaf0 at schedule+0xeb. It found the CU die and the frame base info and the variable "rq" was found in the scope 3/4. Its location is the r15 register and the type size is 0xfc0 which includes 0xaf0. But the second case is not good. It looked up a data type in rbx (reg3) with offset 0x7bc. It found a CU and the frame base which is good so far. And it also found a variable "sk" but the access offset is bigger than the type size (1980 vs. 760 or 0x7bc vs. 0x2f8). The variable has the right location (reg3) but I need to figure out why it accesses beyond what it's supposed to. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240412183310.2518474-2-namhyung@kernel.org [ Fix the build on 32-bit by casting Dwarf_Word to (long) in pr_debug_location() ] Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-04-12 11:33:07 -07:00
pr_debug_dtp("found by addr=%#"PRIx64" type_offset=%#x\n",
perf annotate-data: Add debug messages Add a new debug option "type-profile" to enable the detailed info during the type analysis especially for instruction tracking. You can use this before the command name like 'report' or 'annotate'. $ perf --debug type-profile annotate --data-type Committer testing: First get some memory events: $ perf mem record ls Then, without data-type profiling debug: $ perf annotate --data-type | head Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ And with only data-type profiling: $ perf --debug type-profile annotate --data-type | head ----------------------------------------------------------- find_data_type_die [1e67] for reg13873052 (PC) offset=0x150e2 in dl_main CU die offset: 0x29cd3 found PC-rel by addr=0x34020 offset=0x20 ----------------------------------------------------------- find_data_type_die [2e] for reg12 offset=0 in __GI___readdir64 CU die offset: 0x137a45 frame base: cfa=1 fbreg=-1 found "__futex" in scope=2/2 (die: 0x137ad5) 0(reg12) type=int (die:2a) ----------------------------------------------------------- find_data_type_die [52] for reg5 offset=0 in __memmove_avx_unaligned_erms CU die offset: 0x1124ed no variable found Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:00 -07:00
dloc->var_addr, offset);
pr_debug_type_name(type_die, TSR_KIND_TYPE);
found = true;
goto out;
}
}
/* Get a list of nested scopes - i.e. (inlined) functions and blocks. */
nr_scopes = die_get_scopes(&cu_die, pc, &scopes);
perf annotate-data: Support stack variables Local variables are allocated in the stack and the location list should look like base register(s) and an offset. Extend the die_find_variable_by_reg() to handle the following expressions * DW_OP_breg{0..31} * DW_OP_bregx * DW_OP_fbreg Ususally DWARF subprogram entries have frame base information and use it to locate stack variable like below: <2><43d1575>: Abbrev Number: 62 (DW_TAG_variable) <43d1576> DW_AT_location : 2 byte block: 91 7c (DW_OP_fbreg: -4) <--- here <43d1579> DW_AT_name : (indirect string, offset: 0x2c00c9): i <43d157d> DW_AT_decl_file : 1 <43d157e> DW_AT_decl_line : 78 <43d157f> DW_AT_type : <0x43d19d7> I found some differences on saving the frame base between gcc and clang. The gcc uses the CFA to get the base so it needs to check the current frame's CFI info. In this case, stack offset needs to be adjusted from the start of the CFA. <1><1bb8d>: Abbrev Number: 102 (DW_TAG_subprogram) <1bb8e> DW_AT_name : (indirect string, offset: 0x74d41): kernel_init <1bb92> DW_AT_decl_file : 2 <1bb92> DW_AT_decl_line : 1440 <1bb94> DW_AT_decl_column : 18 <1bb95> DW_AT_prototyped : 1 <1bb95> DW_AT_type : <0xcc> <1bb99> DW_AT_low_pc : 0xffffffff81bab9e0 <1bba1> DW_AT_high_pc : 0x1b2 <1bba9> DW_AT_frame_base : 1 byte block: 9c (DW_OP_call_frame_cfa) <------ here <1bbab> DW_AT_call_all_calls: 1 <1bbab> DW_AT_sibling : <0x1bf5a> While clang sets it to a register directly and it can check the register and offset in the instruction directly. <1><43d1542>: Abbrev Number: 60 (DW_TAG_subprogram) <43d1543> DW_AT_low_pc : 0xffffffff816a7c60 <43d154b> DW_AT_high_pc : 0x98 <43d154f> DW_AT_frame_base : 1 byte block: 56 (DW_OP_reg6 (rbp)) <---------- here <43d1551> DW_AT_GNU_all_call_sites: 1 <43d1551> DW_AT_name : (indirect string, offset: 0x3bce91): foo <43d1555> DW_AT_decl_file : 1 <43d1556> DW_AT_decl_line : 75 <43d1557> DW_AT_prototyped : 1 <43d1557> DW_AT_type : <0x43c7332> <43d155b> DW_AT_external : 1 Also it needs to update the offset after finding the type like global variables since the offset was from the frame base. Factor out match_var_offset() to check global and local variables in the same way. The type stats are improved too: Annotate data type stats: total 294, ok 160 (54.4%), bad 134 (45.6%) ----------------------------------------------------------- 30 : no_sym 32 : no_mem_ops 51 : no_var 14 : no_typeinfo 7 : bad_offset Reviewed-by: Ian Rogers <irogers@google.com> Cc: Stephane Eranian <eranian@google.com> Cc: Masami Hiramatsu <mhiramat@kernel.org> Link: https://lore.kernel.org/r/20240117062657.985479-9-namhyung@kernel.org Signed-off-by: Namhyung Kim <namhyung@kernel.org>
2024-01-16 22:26:56 -08:00
if (reg != DWARF_REG_PC && dwarf_hasattr(&scopes[0], DW_AT_frame_base)) {
Dwarf_Attribute attr;
Dwarf_Block block;
/* Check if the 'reg' is assigned as frame base register */
if (dwarf_attr(&scopes[0], DW_AT_frame_base, &attr) != NULL &&
dwarf_formblock(&attr, &block) == 0 && block.length == 1) {
switch (*block.data) {
case DW_OP_reg0 ... DW_OP_reg31:
fbreg = dloc->fbreg = *block.data - DW_OP_reg0;
perf annotate-data: Support stack variables Local variables are allocated in the stack and the location list should look like base register(s) and an offset. Extend the die_find_variable_by_reg() to handle the following expressions * DW_OP_breg{0..31} * DW_OP_bregx * DW_OP_fbreg Ususally DWARF subprogram entries have frame base information and use it to locate stack variable like below: <2><43d1575>: Abbrev Number: 62 (DW_TAG_variable) <43d1576> DW_AT_location : 2 byte block: 91 7c (DW_OP_fbreg: -4) <--- here <43d1579> DW_AT_name : (indirect string, offset: 0x2c00c9): i <43d157d> DW_AT_decl_file : 1 <43d157e> DW_AT_decl_line : 78 <43d157f> DW_AT_type : <0x43d19d7> I found some differences on saving the frame base between gcc and clang. The gcc uses the CFA to get the base so it needs to check the current frame's CFI info. In this case, stack offset needs to be adjusted from the start of the CFA. <1><1bb8d>: Abbrev Number: 102 (DW_TAG_subprogram) <1bb8e> DW_AT_name : (indirect string, offset: 0x74d41): kernel_init <1bb92> DW_AT_decl_file : 2 <1bb92> DW_AT_decl_line : 1440 <1bb94> DW_AT_decl_column : 18 <1bb95> DW_AT_prototyped : 1 <1bb95> DW_AT_type : <0xcc> <1bb99> DW_AT_low_pc : 0xffffffff81bab9e0 <1bba1> DW_AT_high_pc : 0x1b2 <1bba9> DW_AT_frame_base : 1 byte block: 9c (DW_OP_call_frame_cfa) <------ here <1bbab> DW_AT_call_all_calls: 1 <1bbab> DW_AT_sibling : <0x1bf5a> While clang sets it to a register directly and it can check the register and offset in the instruction directly. <1><43d1542>: Abbrev Number: 60 (DW_TAG_subprogram) <43d1543> DW_AT_low_pc : 0xffffffff816a7c60 <43d154b> DW_AT_high_pc : 0x98 <43d154f> DW_AT_frame_base : 1 byte block: 56 (DW_OP_reg6 (rbp)) <---------- here <43d1551> DW_AT_GNU_all_call_sites: 1 <43d1551> DW_AT_name : (indirect string, offset: 0x3bce91): foo <43d1555> DW_AT_decl_file : 1 <43d1556> DW_AT_decl_line : 75 <43d1557> DW_AT_prototyped : 1 <43d1557> DW_AT_type : <0x43c7332> <43d155b> DW_AT_external : 1 Also it needs to update the offset after finding the type like global variables since the offset was from the frame base. Factor out match_var_offset() to check global and local variables in the same way. The type stats are improved too: Annotate data type stats: total 294, ok 160 (54.4%), bad 134 (45.6%) ----------------------------------------------------------- 30 : no_sym 32 : no_mem_ops 51 : no_var 14 : no_typeinfo 7 : bad_offset Reviewed-by: Ian Rogers <irogers@google.com> Cc: Stephane Eranian <eranian@google.com> Cc: Masami Hiramatsu <mhiramat@kernel.org> Link: https://lore.kernel.org/r/20240117062657.985479-9-namhyung@kernel.org Signed-off-by: Namhyung Kim <namhyung@kernel.org>
2024-01-16 22:26:56 -08:00
break;
case DW_OP_call_frame_cfa:
dloc->fb_cfa = true;
if (die_get_cfa(dloc->di->dbg, pc, &fbreg,
perf annotate-data: Support stack variables Local variables are allocated in the stack and the location list should look like base register(s) and an offset. Extend the die_find_variable_by_reg() to handle the following expressions * DW_OP_breg{0..31} * DW_OP_bregx * DW_OP_fbreg Ususally DWARF subprogram entries have frame base information and use it to locate stack variable like below: <2><43d1575>: Abbrev Number: 62 (DW_TAG_variable) <43d1576> DW_AT_location : 2 byte block: 91 7c (DW_OP_fbreg: -4) <--- here <43d1579> DW_AT_name : (indirect string, offset: 0x2c00c9): i <43d157d> DW_AT_decl_file : 1 <43d157e> DW_AT_decl_line : 78 <43d157f> DW_AT_type : <0x43d19d7> I found some differences on saving the frame base between gcc and clang. The gcc uses the CFA to get the base so it needs to check the current frame's CFI info. In this case, stack offset needs to be adjusted from the start of the CFA. <1><1bb8d>: Abbrev Number: 102 (DW_TAG_subprogram) <1bb8e> DW_AT_name : (indirect string, offset: 0x74d41): kernel_init <1bb92> DW_AT_decl_file : 2 <1bb92> DW_AT_decl_line : 1440 <1bb94> DW_AT_decl_column : 18 <1bb95> DW_AT_prototyped : 1 <1bb95> DW_AT_type : <0xcc> <1bb99> DW_AT_low_pc : 0xffffffff81bab9e0 <1bba1> DW_AT_high_pc : 0x1b2 <1bba9> DW_AT_frame_base : 1 byte block: 9c (DW_OP_call_frame_cfa) <------ here <1bbab> DW_AT_call_all_calls: 1 <1bbab> DW_AT_sibling : <0x1bf5a> While clang sets it to a register directly and it can check the register and offset in the instruction directly. <1><43d1542>: Abbrev Number: 60 (DW_TAG_subprogram) <43d1543> DW_AT_low_pc : 0xffffffff816a7c60 <43d154b> DW_AT_high_pc : 0x98 <43d154f> DW_AT_frame_base : 1 byte block: 56 (DW_OP_reg6 (rbp)) <---------- here <43d1551> DW_AT_GNU_all_call_sites: 1 <43d1551> DW_AT_name : (indirect string, offset: 0x3bce91): foo <43d1555> DW_AT_decl_file : 1 <43d1556> DW_AT_decl_line : 75 <43d1557> DW_AT_prototyped : 1 <43d1557> DW_AT_type : <0x43c7332> <43d155b> DW_AT_external : 1 Also it needs to update the offset after finding the type like global variables since the offset was from the frame base. Factor out match_var_offset() to check global and local variables in the same way. The type stats are improved too: Annotate data type stats: total 294, ok 160 (54.4%), bad 134 (45.6%) ----------------------------------------------------------- 30 : no_sym 32 : no_mem_ops 51 : no_var 14 : no_typeinfo 7 : bad_offset Reviewed-by: Ian Rogers <irogers@google.com> Cc: Stephane Eranian <eranian@google.com> Cc: Masami Hiramatsu <mhiramat@kernel.org> Link: https://lore.kernel.org/r/20240117062657.985479-9-namhyung@kernel.org Signed-off-by: Namhyung Kim <namhyung@kernel.org>
2024-01-16 22:26:56 -08:00
&fb_offset) < 0)
fbreg = -1;
break;
default:
break;
}
perf annotate-data: Add debug messages Add a new debug option "type-profile" to enable the detailed info during the type analysis especially for instruction tracking. You can use this before the command name like 'report' or 'annotate'. $ perf --debug type-profile annotate --data-type Committer testing: First get some memory events: $ perf mem record ls Then, without data-type profiling debug: $ perf annotate --data-type | head Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ And with only data-type profiling: $ perf --debug type-profile annotate --data-type | head ----------------------------------------------------------- find_data_type_die [1e67] for reg13873052 (PC) offset=0x150e2 in dl_main CU die offset: 0x29cd3 found PC-rel by addr=0x34020 offset=0x20 ----------------------------------------------------------- find_data_type_die [2e] for reg12 offset=0 in __GI___readdir64 CU die offset: 0x137a45 frame base: cfa=1 fbreg=-1 found "__futex" in scope=2/2 (die: 0x137ad5) 0(reg12) type=int (die:2a) ----------------------------------------------------------- find_data_type_die [52] for reg5 offset=0 in __memmove_avx_unaligned_erms CU die offset: 0x1124ed no variable found Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:00 -07:00
pr_debug_dtp("frame base: cfa=%d fbreg=%d\n",
dloc->fb_cfa, fbreg);
perf annotate-data: Support stack variables Local variables are allocated in the stack and the location list should look like base register(s) and an offset. Extend the die_find_variable_by_reg() to handle the following expressions * DW_OP_breg{0..31} * DW_OP_bregx * DW_OP_fbreg Ususally DWARF subprogram entries have frame base information and use it to locate stack variable like below: <2><43d1575>: Abbrev Number: 62 (DW_TAG_variable) <43d1576> DW_AT_location : 2 byte block: 91 7c (DW_OP_fbreg: -4) <--- here <43d1579> DW_AT_name : (indirect string, offset: 0x2c00c9): i <43d157d> DW_AT_decl_file : 1 <43d157e> DW_AT_decl_line : 78 <43d157f> DW_AT_type : <0x43d19d7> I found some differences on saving the frame base between gcc and clang. The gcc uses the CFA to get the base so it needs to check the current frame's CFI info. In this case, stack offset needs to be adjusted from the start of the CFA. <1><1bb8d>: Abbrev Number: 102 (DW_TAG_subprogram) <1bb8e> DW_AT_name : (indirect string, offset: 0x74d41): kernel_init <1bb92> DW_AT_decl_file : 2 <1bb92> DW_AT_decl_line : 1440 <1bb94> DW_AT_decl_column : 18 <1bb95> DW_AT_prototyped : 1 <1bb95> DW_AT_type : <0xcc> <1bb99> DW_AT_low_pc : 0xffffffff81bab9e0 <1bba1> DW_AT_high_pc : 0x1b2 <1bba9> DW_AT_frame_base : 1 byte block: 9c (DW_OP_call_frame_cfa) <------ here <1bbab> DW_AT_call_all_calls: 1 <1bbab> DW_AT_sibling : <0x1bf5a> While clang sets it to a register directly and it can check the register and offset in the instruction directly. <1><43d1542>: Abbrev Number: 60 (DW_TAG_subprogram) <43d1543> DW_AT_low_pc : 0xffffffff816a7c60 <43d154b> DW_AT_high_pc : 0x98 <43d154f> DW_AT_frame_base : 1 byte block: 56 (DW_OP_reg6 (rbp)) <---------- here <43d1551> DW_AT_GNU_all_call_sites: 1 <43d1551> DW_AT_name : (indirect string, offset: 0x3bce91): foo <43d1555> DW_AT_decl_file : 1 <43d1556> DW_AT_decl_line : 75 <43d1557> DW_AT_prototyped : 1 <43d1557> DW_AT_type : <0x43c7332> <43d155b> DW_AT_external : 1 Also it needs to update the offset after finding the type like global variables since the offset was from the frame base. Factor out match_var_offset() to check global and local variables in the same way. The type stats are improved too: Annotate data type stats: total 294, ok 160 (54.4%), bad 134 (45.6%) ----------------------------------------------------------- 30 : no_sym 32 : no_mem_ops 51 : no_var 14 : no_typeinfo 7 : bad_offset Reviewed-by: Ian Rogers <irogers@google.com> Cc: Stephane Eranian <eranian@google.com> Cc: Masami Hiramatsu <mhiramat@kernel.org> Link: https://lore.kernel.org/r/20240117062657.985479-9-namhyung@kernel.org Signed-off-by: Namhyung Kim <namhyung@kernel.org>
2024-01-16 22:26:56 -08:00
}
}
retry:
perf annotate-data: Support stack variables Local variables are allocated in the stack and the location list should look like base register(s) and an offset. Extend the die_find_variable_by_reg() to handle the following expressions * DW_OP_breg{0..31} * DW_OP_bregx * DW_OP_fbreg Ususally DWARF subprogram entries have frame base information and use it to locate stack variable like below: <2><43d1575>: Abbrev Number: 62 (DW_TAG_variable) <43d1576> DW_AT_location : 2 byte block: 91 7c (DW_OP_fbreg: -4) <--- here <43d1579> DW_AT_name : (indirect string, offset: 0x2c00c9): i <43d157d> DW_AT_decl_file : 1 <43d157e> DW_AT_decl_line : 78 <43d157f> DW_AT_type : <0x43d19d7> I found some differences on saving the frame base between gcc and clang. The gcc uses the CFA to get the base so it needs to check the current frame's CFI info. In this case, stack offset needs to be adjusted from the start of the CFA. <1><1bb8d>: Abbrev Number: 102 (DW_TAG_subprogram) <1bb8e> DW_AT_name : (indirect string, offset: 0x74d41): kernel_init <1bb92> DW_AT_decl_file : 2 <1bb92> DW_AT_decl_line : 1440 <1bb94> DW_AT_decl_column : 18 <1bb95> DW_AT_prototyped : 1 <1bb95> DW_AT_type : <0xcc> <1bb99> DW_AT_low_pc : 0xffffffff81bab9e0 <1bba1> DW_AT_high_pc : 0x1b2 <1bba9> DW_AT_frame_base : 1 byte block: 9c (DW_OP_call_frame_cfa) <------ here <1bbab> DW_AT_call_all_calls: 1 <1bbab> DW_AT_sibling : <0x1bf5a> While clang sets it to a register directly and it can check the register and offset in the instruction directly. <1><43d1542>: Abbrev Number: 60 (DW_TAG_subprogram) <43d1543> DW_AT_low_pc : 0xffffffff816a7c60 <43d154b> DW_AT_high_pc : 0x98 <43d154f> DW_AT_frame_base : 1 byte block: 56 (DW_OP_reg6 (rbp)) <---------- here <43d1551> DW_AT_GNU_all_call_sites: 1 <43d1551> DW_AT_name : (indirect string, offset: 0x3bce91): foo <43d1555> DW_AT_decl_file : 1 <43d1556> DW_AT_decl_line : 75 <43d1557> DW_AT_prototyped : 1 <43d1557> DW_AT_type : <0x43c7332> <43d155b> DW_AT_external : 1 Also it needs to update the offset after finding the type like global variables since the offset was from the frame base. Factor out match_var_offset() to check global and local variables in the same way. The type stats are improved too: Annotate data type stats: total 294, ok 160 (54.4%), bad 134 (45.6%) ----------------------------------------------------------- 30 : no_sym 32 : no_mem_ops 51 : no_var 14 : no_typeinfo 7 : bad_offset Reviewed-by: Ian Rogers <irogers@google.com> Cc: Stephane Eranian <eranian@google.com> Cc: Masami Hiramatsu <mhiramat@kernel.org> Link: https://lore.kernel.org/r/20240117062657.985479-9-namhyung@kernel.org Signed-off-by: Namhyung Kim <namhyung@kernel.org>
2024-01-16 22:26:56 -08:00
is_fbreg = (reg == fbreg);
if (is_fbreg)
offset = loc->offset - fb_offset;
/* Search from the inner-most scope to the outer */
for (i = nr_scopes - 1; i >= 0; i--) {
perf annotate-data: Check variables in every scope Sometimes it matches a variable in the inner scope but it fails because the actual access can be on a different type. Let's try variables in every scope and choose the best one using is_better_type(). I have an example with update_blocked_averages(), at first it found a variable (__mptr) but it's a void pointer. So it moved on to the upper scope and found another variable (cfs_rq). $ perf --debug type-profile annotate --data-type --stdio ... ----------------------------------------------------------- find data type for 0x140(reg14) at update_blocked_averages+0x2db CU for kernel/sched/fair.c (die:0x12dd892) frame base: cfa=1 fbreg=7 found "__mptr" (die: 0x13022f1) in scope=4/4 (die: 0x13022e8) failed: no/void pointer variable location: base=reg14, offset=0x140 type='void*' size=0x8 (die:0x12dd8f9) found "cfs_rq" (die: 0x1301721) in scope=3/4 (die: 0x130171c) type_offset=0x140 variable location: reg14 type='struct cfs_rq' size=0x1c0 (die:0x12e37e5) final type: type='struct cfs_rq' size=0x1c0 (die:0x12e37e5) IIUC the scope is like below: 1: update_blocked_averages 2: __update_blocked_fair 3: for_each_leaf_cfs_rq_safe 4: list_entry -> (container_of) The container_of is implemented like: #define container_of(ptr, type, member) ({ \ void *__mptr = (void *)(ptr); \ static_assert(__same_type(*(ptr), ((type *)0)->member) || \ __same_type(*(ptr), void), \ "pointer type mismatch in container_of()"); \ ((type *)(__mptr - offsetof(type, member))); }) That's why we see the __mptr variable first but it failed since it has no type information. Then for_each_leaf_cfs_rq_safe() is defined as #define for_each_leaf_cfs_rq_safe(rq, cfs_rq, pos) \ list_for_each_entry_safe(cfs_rq, pos, &rq->leaf_cfs_rq_list, \ leaf_cfs_rq_list) Note that the access was 0x140(r14). And the cfs_rq has leaf_cfs_rq_list at the 0x140. So it converts the list_head pointer to a pointer to struct cfs_rq here. $ pahole --hex -C cfs_rq vmlinux | grep 140 struct cfs_rq struct list_head leaf_cfs_rq_list; /* 0x140 0x10 */ Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240816235840.2754937-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-16 16:58:38 -07:00
Dwarf_Die mem_die;
int type_offset = offset;
if (reg == DWARF_REG_PC) {
if (!die_find_variable_by_addr(&scopes[i], dloc->var_addr,
perf annotate-data: Check variables in every scope Sometimes it matches a variable in the inner scope but it fails because the actual access can be on a different type. Let's try variables in every scope and choose the best one using is_better_type(). I have an example with update_blocked_averages(), at first it found a variable (__mptr) but it's a void pointer. So it moved on to the upper scope and found another variable (cfs_rq). $ perf --debug type-profile annotate --data-type --stdio ... ----------------------------------------------------------- find data type for 0x140(reg14) at update_blocked_averages+0x2db CU for kernel/sched/fair.c (die:0x12dd892) frame base: cfa=1 fbreg=7 found "__mptr" (die: 0x13022f1) in scope=4/4 (die: 0x13022e8) failed: no/void pointer variable location: base=reg14, offset=0x140 type='void*' size=0x8 (die:0x12dd8f9) found "cfs_rq" (die: 0x1301721) in scope=3/4 (die: 0x130171c) type_offset=0x140 variable location: reg14 type='struct cfs_rq' size=0x1c0 (die:0x12e37e5) final type: type='struct cfs_rq' size=0x1c0 (die:0x12e37e5) IIUC the scope is like below: 1: update_blocked_averages 2: __update_blocked_fair 3: for_each_leaf_cfs_rq_safe 4: list_entry -> (container_of) The container_of is implemented like: #define container_of(ptr, type, member) ({ \ void *__mptr = (void *)(ptr); \ static_assert(__same_type(*(ptr), ((type *)0)->member) || \ __same_type(*(ptr), void), \ "pointer type mismatch in container_of()"); \ ((type *)(__mptr - offsetof(type, member))); }) That's why we see the __mptr variable first but it failed since it has no type information. Then for_each_leaf_cfs_rq_safe() is defined as #define for_each_leaf_cfs_rq_safe(rq, cfs_rq, pos) \ list_for_each_entry_safe(cfs_rq, pos, &rq->leaf_cfs_rq_list, \ leaf_cfs_rq_list) Note that the access was 0x140(r14). And the cfs_rq has leaf_cfs_rq_list at the 0x140. So it converts the list_head pointer to a pointer to struct cfs_rq here. $ pahole --hex -C cfs_rq vmlinux | grep 140 struct cfs_rq struct list_head leaf_cfs_rq_list; /* 0x140 0x10 */ Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240816235840.2754937-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-16 16:58:38 -07:00
&var_die, &type_offset))
continue;
} else {
/* Look up variables/parameters in this scope */
if (!die_find_variable_by_reg(&scopes[i], pc, reg,
perf annotate-data: Check variables in every scope Sometimes it matches a variable in the inner scope but it fails because the actual access can be on a different type. Let's try variables in every scope and choose the best one using is_better_type(). I have an example with update_blocked_averages(), at first it found a variable (__mptr) but it's a void pointer. So it moved on to the upper scope and found another variable (cfs_rq). $ perf --debug type-profile annotate --data-type --stdio ... ----------------------------------------------------------- find data type for 0x140(reg14) at update_blocked_averages+0x2db CU for kernel/sched/fair.c (die:0x12dd892) frame base: cfa=1 fbreg=7 found "__mptr" (die: 0x13022f1) in scope=4/4 (die: 0x13022e8) failed: no/void pointer variable location: base=reg14, offset=0x140 type='void*' size=0x8 (die:0x12dd8f9) found "cfs_rq" (die: 0x1301721) in scope=3/4 (die: 0x130171c) type_offset=0x140 variable location: reg14 type='struct cfs_rq' size=0x1c0 (die:0x12e37e5) final type: type='struct cfs_rq' size=0x1c0 (die:0x12e37e5) IIUC the scope is like below: 1: update_blocked_averages 2: __update_blocked_fair 3: for_each_leaf_cfs_rq_safe 4: list_entry -> (container_of) The container_of is implemented like: #define container_of(ptr, type, member) ({ \ void *__mptr = (void *)(ptr); \ static_assert(__same_type(*(ptr), ((type *)0)->member) || \ __same_type(*(ptr), void), \ "pointer type mismatch in container_of()"); \ ((type *)(__mptr - offsetof(type, member))); }) That's why we see the __mptr variable first but it failed since it has no type information. Then for_each_leaf_cfs_rq_safe() is defined as #define for_each_leaf_cfs_rq_safe(rq, cfs_rq, pos) \ list_for_each_entry_safe(cfs_rq, pos, &rq->leaf_cfs_rq_list, \ leaf_cfs_rq_list) Note that the access was 0x140(r14). And the cfs_rq has leaf_cfs_rq_list at the 0x140. So it converts the list_head pointer to a pointer to struct cfs_rq here. $ pahole --hex -C cfs_rq vmlinux | grep 140 struct cfs_rq struct list_head leaf_cfs_rq_list; /* 0x140 0x10 */ Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240816235840.2754937-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-16 16:58:38 -07:00
&type_offset, is_fbreg, &var_die))
continue;
}
pr_debug_dtp("found \"%s\" (die: %#lx) in scope=%d/%d (die: %#lx) ",
dwarf_diename(&var_die), (long)dwarf_dieoffset(&var_die),
i+1, nr_scopes, (long)dwarf_dieoffset(&scopes[i]));
/* Found a variable, see if it's correct */
perf annotate-data: Check variables in every scope Sometimes it matches a variable in the inner scope but it fails because the actual access can be on a different type. Let's try variables in every scope and choose the best one using is_better_type(). I have an example with update_blocked_averages(), at first it found a variable (__mptr) but it's a void pointer. So it moved on to the upper scope and found another variable (cfs_rq). $ perf --debug type-profile annotate --data-type --stdio ... ----------------------------------------------------------- find data type for 0x140(reg14) at update_blocked_averages+0x2db CU for kernel/sched/fair.c (die:0x12dd892) frame base: cfa=1 fbreg=7 found "__mptr" (die: 0x13022f1) in scope=4/4 (die: 0x13022e8) failed: no/void pointer variable location: base=reg14, offset=0x140 type='void*' size=0x8 (die:0x12dd8f9) found "cfs_rq" (die: 0x1301721) in scope=3/4 (die: 0x130171c) type_offset=0x140 variable location: reg14 type='struct cfs_rq' size=0x1c0 (die:0x12e37e5) final type: type='struct cfs_rq' size=0x1c0 (die:0x12e37e5) IIUC the scope is like below: 1: update_blocked_averages 2: __update_blocked_fair 3: for_each_leaf_cfs_rq_safe 4: list_entry -> (container_of) The container_of is implemented like: #define container_of(ptr, type, member) ({ \ void *__mptr = (void *)(ptr); \ static_assert(__same_type(*(ptr), ((type *)0)->member) || \ __same_type(*(ptr), void), \ "pointer type mismatch in container_of()"); \ ((type *)(__mptr - offsetof(type, member))); }) That's why we see the __mptr variable first but it failed since it has no type information. Then for_each_leaf_cfs_rq_safe() is defined as #define for_each_leaf_cfs_rq_safe(rq, cfs_rq, pos) \ list_for_each_entry_safe(cfs_rq, pos, &rq->leaf_cfs_rq_list, \ leaf_cfs_rq_list) Note that the access was 0x140(r14). And the cfs_rq has leaf_cfs_rq_list at the 0x140. So it converts the list_head pointer to a pointer to struct cfs_rq here. $ pahole --hex -C cfs_rq vmlinux | grep 140 struct cfs_rq struct list_head leaf_cfs_rq_list; /* 0x140 0x10 */ Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240816235840.2754937-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-16 16:58:38 -07:00
result = check_variable(dloc, &var_die, &mem_die, reg, type_offset, is_fbreg);
if (result == PERF_TMR_OK) {
perf annotate-data: Improve debug message with location info To verify it found the correct variable, let's add the location expression to the debug message. $ perf --debug type-profile annotate --data-type ... ----------------------------------------------------------- find data type for 0xaf0(reg15) at schedule+0xeb CU for kernel/sched/core.c (die:0x1180523) frame base: cfa=0 fbreg=6 found "rq" in scope=3/4 (die: 0x11b6a00) type_offset=0xaf0 variable location: reg15 type='struct rq' size=0xfc0 (die:0x11892e2) ----------------------------------------------------------- find data type for 0x7bc(reg3) at tcp_get_info+0x62 CU for net/ipv4/tcp.c (die:0x7b5f516) frame base: cfa=0 fbreg=6 offset: 1980 is bigger than size: 760 check variable "sk" failed (die: 0x7b92b2c) variable location: reg3 type='struct sock' size=0x2f8 (die:0x7b63c3a) ----------------------------------------------------------- ... The first case is fine. It looked up a data type in r15 with offset of 0xaf0 at schedule+0xeb. It found the CU die and the frame base info and the variable "rq" was found in the scope 3/4. Its location is the r15 register and the type size is 0xfc0 which includes 0xaf0. But the second case is not good. It looked up a data type in rbx (reg3) with offset 0x7bc. It found a CU and the frame base which is good so far. And it also found a variable "sk" but the access offset is bigger than the type size (1980 vs. 760 or 0x7bc vs. 0x2f8). The variable has the right location (reg3) but I need to figure out why it accesses beyond what it's supposed to. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240412183310.2518474-2-namhyung@kernel.org [ Fix the build on 32-bit by casting Dwarf_Word to (long) in pr_debug_location() ] Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-04-12 11:33:07 -07:00
if (reg == DWARF_REG_PC) {
pr_debug_dtp("addr=%#"PRIx64" type_offset=%#x\n",
perf annotate-data: Check variables in every scope Sometimes it matches a variable in the inner scope but it fails because the actual access can be on a different type. Let's try variables in every scope and choose the best one using is_better_type(). I have an example with update_blocked_averages(), at first it found a variable (__mptr) but it's a void pointer. So it moved on to the upper scope and found another variable (cfs_rq). $ perf --debug type-profile annotate --data-type --stdio ... ----------------------------------------------------------- find data type for 0x140(reg14) at update_blocked_averages+0x2db CU for kernel/sched/fair.c (die:0x12dd892) frame base: cfa=1 fbreg=7 found "__mptr" (die: 0x13022f1) in scope=4/4 (die: 0x13022e8) failed: no/void pointer variable location: base=reg14, offset=0x140 type='void*' size=0x8 (die:0x12dd8f9) found "cfs_rq" (die: 0x1301721) in scope=3/4 (die: 0x130171c) type_offset=0x140 variable location: reg14 type='struct cfs_rq' size=0x1c0 (die:0x12e37e5) final type: type='struct cfs_rq' size=0x1c0 (die:0x12e37e5) IIUC the scope is like below: 1: update_blocked_averages 2: __update_blocked_fair 3: for_each_leaf_cfs_rq_safe 4: list_entry -> (container_of) The container_of is implemented like: #define container_of(ptr, type, member) ({ \ void *__mptr = (void *)(ptr); \ static_assert(__same_type(*(ptr), ((type *)0)->member) || \ __same_type(*(ptr), void), \ "pointer type mismatch in container_of()"); \ ((type *)(__mptr - offsetof(type, member))); }) That's why we see the __mptr variable first but it failed since it has no type information. Then for_each_leaf_cfs_rq_safe() is defined as #define for_each_leaf_cfs_rq_safe(rq, cfs_rq, pos) \ list_for_each_entry_safe(cfs_rq, pos, &rq->leaf_cfs_rq_list, \ leaf_cfs_rq_list) Note that the access was 0x140(r14). And the cfs_rq has leaf_cfs_rq_list at the 0x140. So it converts the list_head pointer to a pointer to struct cfs_rq here. $ pahole --hex -C cfs_rq vmlinux | grep 140 struct cfs_rq struct list_head leaf_cfs_rq_list; /* 0x140 0x10 */ Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240816235840.2754937-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-16 16:58:38 -07:00
dloc->var_addr, type_offset);
perf annotate-data: Improve debug message with location info To verify it found the correct variable, let's add the location expression to the debug message. $ perf --debug type-profile annotate --data-type ... ----------------------------------------------------------- find data type for 0xaf0(reg15) at schedule+0xeb CU for kernel/sched/core.c (die:0x1180523) frame base: cfa=0 fbreg=6 found "rq" in scope=3/4 (die: 0x11b6a00) type_offset=0xaf0 variable location: reg15 type='struct rq' size=0xfc0 (die:0x11892e2) ----------------------------------------------------------- find data type for 0x7bc(reg3) at tcp_get_info+0x62 CU for net/ipv4/tcp.c (die:0x7b5f516) frame base: cfa=0 fbreg=6 offset: 1980 is bigger than size: 760 check variable "sk" failed (die: 0x7b92b2c) variable location: reg3 type='struct sock' size=0x2f8 (die:0x7b63c3a) ----------------------------------------------------------- ... The first case is fine. It looked up a data type in r15 with offset of 0xaf0 at schedule+0xeb. It found the CU die and the frame base info and the variable "rq" was found in the scope 3/4. Its location is the r15 register and the type size is 0xfc0 which includes 0xaf0. But the second case is not good. It looked up a data type in rbx (reg3) with offset 0x7bc. It found a CU and the frame base which is good so far. And it also found a variable "sk" but the access offset is bigger than the type size (1980 vs. 760 or 0x7bc vs. 0x2f8). The variable has the right location (reg3) but I need to figure out why it accesses beyond what it's supposed to. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240412183310.2518474-2-namhyung@kernel.org [ Fix the build on 32-bit by casting Dwarf_Word to (long) in pr_debug_location() ] Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-04-12 11:33:07 -07:00
} else if (reg == DWARF_REG_FB || is_fbreg) {
pr_debug_dtp("stack_offset=%#x type_offset=%#x\n",
perf annotate-data: Check variables in every scope Sometimes it matches a variable in the inner scope but it fails because the actual access can be on a different type. Let's try variables in every scope and choose the best one using is_better_type(). I have an example with update_blocked_averages(), at first it found a variable (__mptr) but it's a void pointer. So it moved on to the upper scope and found another variable (cfs_rq). $ perf --debug type-profile annotate --data-type --stdio ... ----------------------------------------------------------- find data type for 0x140(reg14) at update_blocked_averages+0x2db CU for kernel/sched/fair.c (die:0x12dd892) frame base: cfa=1 fbreg=7 found "__mptr" (die: 0x13022f1) in scope=4/4 (die: 0x13022e8) failed: no/void pointer variable location: base=reg14, offset=0x140 type='void*' size=0x8 (die:0x12dd8f9) found "cfs_rq" (die: 0x1301721) in scope=3/4 (die: 0x130171c) type_offset=0x140 variable location: reg14 type='struct cfs_rq' size=0x1c0 (die:0x12e37e5) final type: type='struct cfs_rq' size=0x1c0 (die:0x12e37e5) IIUC the scope is like below: 1: update_blocked_averages 2: __update_blocked_fair 3: for_each_leaf_cfs_rq_safe 4: list_entry -> (container_of) The container_of is implemented like: #define container_of(ptr, type, member) ({ \ void *__mptr = (void *)(ptr); \ static_assert(__same_type(*(ptr), ((type *)0)->member) || \ __same_type(*(ptr), void), \ "pointer type mismatch in container_of()"); \ ((type *)(__mptr - offsetof(type, member))); }) That's why we see the __mptr variable first but it failed since it has no type information. Then for_each_leaf_cfs_rq_safe() is defined as #define for_each_leaf_cfs_rq_safe(rq, cfs_rq, pos) \ list_for_each_entry_safe(cfs_rq, pos, &rq->leaf_cfs_rq_list, \ leaf_cfs_rq_list) Note that the access was 0x140(r14). And the cfs_rq has leaf_cfs_rq_list at the 0x140. So it converts the list_head pointer to a pointer to struct cfs_rq here. $ pahole --hex -C cfs_rq vmlinux | grep 140 struct cfs_rq struct list_head leaf_cfs_rq_list; /* 0x140 0x10 */ Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240816235840.2754937-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-16 16:58:38 -07:00
fb_offset, type_offset);
perf annotate-data: Improve debug message with location info To verify it found the correct variable, let's add the location expression to the debug message. $ perf --debug type-profile annotate --data-type ... ----------------------------------------------------------- find data type for 0xaf0(reg15) at schedule+0xeb CU for kernel/sched/core.c (die:0x1180523) frame base: cfa=0 fbreg=6 found "rq" in scope=3/4 (die: 0x11b6a00) type_offset=0xaf0 variable location: reg15 type='struct rq' size=0xfc0 (die:0x11892e2) ----------------------------------------------------------- find data type for 0x7bc(reg3) at tcp_get_info+0x62 CU for net/ipv4/tcp.c (die:0x7b5f516) frame base: cfa=0 fbreg=6 offset: 1980 is bigger than size: 760 check variable "sk" failed (die: 0x7b92b2c) variable location: reg3 type='struct sock' size=0x2f8 (die:0x7b63c3a) ----------------------------------------------------------- ... The first case is fine. It looked up a data type in r15 with offset of 0xaf0 at schedule+0xeb. It found the CU die and the frame base info and the variable "rq" was found in the scope 3/4. Its location is the r15 register and the type size is 0xfc0 which includes 0xaf0. But the second case is not good. It looked up a data type in rbx (reg3) with offset 0x7bc. It found a CU and the frame base which is good so far. And it also found a variable "sk" but the access offset is bigger than the type size (1980 vs. 760 or 0x7bc vs. 0x2f8). The variable has the right location (reg3) but I need to figure out why it accesses beyond what it's supposed to. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240412183310.2518474-2-namhyung@kernel.org [ Fix the build on 32-bit by casting Dwarf_Word to (long) in pr_debug_location() ] Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-04-12 11:33:07 -07:00
} else {
perf annotate-data: Check variables in every scope Sometimes it matches a variable in the inner scope but it fails because the actual access can be on a different type. Let's try variables in every scope and choose the best one using is_better_type(). I have an example with update_blocked_averages(), at first it found a variable (__mptr) but it's a void pointer. So it moved on to the upper scope and found another variable (cfs_rq). $ perf --debug type-profile annotate --data-type --stdio ... ----------------------------------------------------------- find data type for 0x140(reg14) at update_blocked_averages+0x2db CU for kernel/sched/fair.c (die:0x12dd892) frame base: cfa=1 fbreg=7 found "__mptr" (die: 0x13022f1) in scope=4/4 (die: 0x13022e8) failed: no/void pointer variable location: base=reg14, offset=0x140 type='void*' size=0x8 (die:0x12dd8f9) found "cfs_rq" (die: 0x1301721) in scope=3/4 (die: 0x130171c) type_offset=0x140 variable location: reg14 type='struct cfs_rq' size=0x1c0 (die:0x12e37e5) final type: type='struct cfs_rq' size=0x1c0 (die:0x12e37e5) IIUC the scope is like below: 1: update_blocked_averages 2: __update_blocked_fair 3: for_each_leaf_cfs_rq_safe 4: list_entry -> (container_of) The container_of is implemented like: #define container_of(ptr, type, member) ({ \ void *__mptr = (void *)(ptr); \ static_assert(__same_type(*(ptr), ((type *)0)->member) || \ __same_type(*(ptr), void), \ "pointer type mismatch in container_of()"); \ ((type *)(__mptr - offsetof(type, member))); }) That's why we see the __mptr variable first but it failed since it has no type information. Then for_each_leaf_cfs_rq_safe() is defined as #define for_each_leaf_cfs_rq_safe(rq, cfs_rq, pos) \ list_for_each_entry_safe(cfs_rq, pos, &rq->leaf_cfs_rq_list, \ leaf_cfs_rq_list) Note that the access was 0x140(r14). And the cfs_rq has leaf_cfs_rq_list at the 0x140. So it converts the list_head pointer to a pointer to struct cfs_rq here. $ pahole --hex -C cfs_rq vmlinux | grep 140 struct cfs_rq struct list_head leaf_cfs_rq_list; /* 0x140 0x10 */ Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240816235840.2754937-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-16 16:58:38 -07:00
pr_debug_dtp("type_offset=%#x\n", type_offset);
}
if (!found || is_better_type(type_die, &mem_die)) {
*type_die = mem_die;
dloc->type_offset = type_offset;
found = true;
perf annotate-data: Improve debug message with location info To verify it found the correct variable, let's add the location expression to the debug message. $ perf --debug type-profile annotate --data-type ... ----------------------------------------------------------- find data type for 0xaf0(reg15) at schedule+0xeb CU for kernel/sched/core.c (die:0x1180523) frame base: cfa=0 fbreg=6 found "rq" in scope=3/4 (die: 0x11b6a00) type_offset=0xaf0 variable location: reg15 type='struct rq' size=0xfc0 (die:0x11892e2) ----------------------------------------------------------- find data type for 0x7bc(reg3) at tcp_get_info+0x62 CU for net/ipv4/tcp.c (die:0x7b5f516) frame base: cfa=0 fbreg=6 offset: 1980 is bigger than size: 760 check variable "sk" failed (die: 0x7b92b2c) variable location: reg3 type='struct sock' size=0x2f8 (die:0x7b63c3a) ----------------------------------------------------------- ... The first case is fine. It looked up a data type in r15 with offset of 0xaf0 at schedule+0xeb. It found the CU die and the frame base info and the variable "rq" was found in the scope 3/4. Its location is the r15 register and the type size is 0xfc0 which includes 0xaf0. But the second case is not good. It looked up a data type in rbx (reg3) with offset 0x7bc. It found a CU and the frame base which is good so far. And it also found a variable "sk" but the access offset is bigger than the type size (1980 vs. 760 or 0x7bc vs. 0x2f8). The variable has the right location (reg3) but I need to figure out why it accesses beyond what it's supposed to. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240412183310.2518474-2-namhyung@kernel.org [ Fix the build on 32-bit by casting Dwarf_Word to (long) in pr_debug_location() ] Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-04-12 11:33:07 -07:00
}
} else {
pr_debug_dtp("failed: %s\n", match_result_str(result));
perf annotate-data: Add debug messages Add a new debug option "type-profile" to enable the detailed info during the type analysis especially for instruction tracking. You can use this before the command name like 'report' or 'annotate'. $ perf --debug type-profile annotate --data-type Committer testing: First get some memory events: $ perf mem record ls Then, without data-type profiling debug: $ perf annotate --data-type | head Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ And with only data-type profiling: $ perf --debug type-profile annotate --data-type | head ----------------------------------------------------------- find_data_type_die [1e67] for reg13873052 (PC) offset=0x150e2 in dl_main CU die offset: 0x29cd3 found PC-rel by addr=0x34020 offset=0x20 ----------------------------------------------------------- find_data_type_die [2e] for reg12 offset=0 in __GI___readdir64 CU die offset: 0x137a45 frame base: cfa=1 fbreg=-1 found "__futex" in scope=2/2 (die: 0x137ad5) 0(reg12) type=int (die:2a) ----------------------------------------------------------- find_data_type_die [52] for reg5 offset=0 in __memmove_avx_unaligned_erms CU die offset: 0x1124ed no variable found Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:00 -07:00
}
perf annotate-data: Check variables in every scope Sometimes it matches a variable in the inner scope but it fails because the actual access can be on a different type. Let's try variables in every scope and choose the best one using is_better_type(). I have an example with update_blocked_averages(), at first it found a variable (__mptr) but it's a void pointer. So it moved on to the upper scope and found another variable (cfs_rq). $ perf --debug type-profile annotate --data-type --stdio ... ----------------------------------------------------------- find data type for 0x140(reg14) at update_blocked_averages+0x2db CU for kernel/sched/fair.c (die:0x12dd892) frame base: cfa=1 fbreg=7 found "__mptr" (die: 0x13022f1) in scope=4/4 (die: 0x13022e8) failed: no/void pointer variable location: base=reg14, offset=0x140 type='void*' size=0x8 (die:0x12dd8f9) found "cfs_rq" (die: 0x1301721) in scope=3/4 (die: 0x130171c) type_offset=0x140 variable location: reg14 type='struct cfs_rq' size=0x1c0 (die:0x12e37e5) final type: type='struct cfs_rq' size=0x1c0 (die:0x12e37e5) IIUC the scope is like below: 1: update_blocked_averages 2: __update_blocked_fair 3: for_each_leaf_cfs_rq_safe 4: list_entry -> (container_of) The container_of is implemented like: #define container_of(ptr, type, member) ({ \ void *__mptr = (void *)(ptr); \ static_assert(__same_type(*(ptr), ((type *)0)->member) || \ __same_type(*(ptr), void), \ "pointer type mismatch in container_of()"); \ ((type *)(__mptr - offsetof(type, member))); }) That's why we see the __mptr variable first but it failed since it has no type information. Then for_each_leaf_cfs_rq_safe() is defined as #define for_each_leaf_cfs_rq_safe(rq, cfs_rq, pos) \ list_for_each_entry_safe(cfs_rq, pos, &rq->leaf_cfs_rq_list, \ leaf_cfs_rq_list) Note that the access was 0x140(r14). And the cfs_rq has leaf_cfs_rq_list at the 0x140. So it converts the list_head pointer to a pointer to struct cfs_rq here. $ pahole --hex -C cfs_rq vmlinux | grep 140 struct cfs_rq struct list_head leaf_cfs_rq_list; /* 0x140 0x10 */ Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240816235840.2754937-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-16 16:58:38 -07:00
pr_debug_location(&var_die, pc, reg);
perf annotate-data: Check variables in every scope Sometimes it matches a variable in the inner scope but it fails because the actual access can be on a different type. Let's try variables in every scope and choose the best one using is_better_type(). I have an example with update_blocked_averages(), at first it found a variable (__mptr) but it's a void pointer. So it moved on to the upper scope and found another variable (cfs_rq). $ perf --debug type-profile annotate --data-type --stdio ... ----------------------------------------------------------- find data type for 0x140(reg14) at update_blocked_averages+0x2db CU for kernel/sched/fair.c (die:0x12dd892) frame base: cfa=1 fbreg=7 found "__mptr" (die: 0x13022f1) in scope=4/4 (die: 0x13022e8) failed: no/void pointer variable location: base=reg14, offset=0x140 type='void*' size=0x8 (die:0x12dd8f9) found "cfs_rq" (die: 0x1301721) in scope=3/4 (die: 0x130171c) type_offset=0x140 variable location: reg14 type='struct cfs_rq' size=0x1c0 (die:0x12e37e5) final type: type='struct cfs_rq' size=0x1c0 (die:0x12e37e5) IIUC the scope is like below: 1: update_blocked_averages 2: __update_blocked_fair 3: for_each_leaf_cfs_rq_safe 4: list_entry -> (container_of) The container_of is implemented like: #define container_of(ptr, type, member) ({ \ void *__mptr = (void *)(ptr); \ static_assert(__same_type(*(ptr), ((type *)0)->member) || \ __same_type(*(ptr), void), \ "pointer type mismatch in container_of()"); \ ((type *)(__mptr - offsetof(type, member))); }) That's why we see the __mptr variable first but it failed since it has no type information. Then for_each_leaf_cfs_rq_safe() is defined as #define for_each_leaf_cfs_rq_safe(rq, cfs_rq, pos) \ list_for_each_entry_safe(cfs_rq, pos, &rq->leaf_cfs_rq_list, \ leaf_cfs_rq_list) Note that the access was 0x140(r14). And the cfs_rq has leaf_cfs_rq_list at the 0x140. So it converts the list_head pointer to a pointer to struct cfs_rq here. $ pahole --hex -C cfs_rq vmlinux | grep 140 struct cfs_rq struct list_head leaf_cfs_rq_list; /* 0x140 0x10 */ Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240816235840.2754937-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-16 16:58:38 -07:00
pr_debug_type_name(&mem_die, TSR_KIND_TYPE);
}
perf annotate-data: Check variables in every scope Sometimes it matches a variable in the inner scope but it fails because the actual access can be on a different type. Let's try variables in every scope and choose the best one using is_better_type(). I have an example with update_blocked_averages(), at first it found a variable (__mptr) but it's a void pointer. So it moved on to the upper scope and found another variable (cfs_rq). $ perf --debug type-profile annotate --data-type --stdio ... ----------------------------------------------------------- find data type for 0x140(reg14) at update_blocked_averages+0x2db CU for kernel/sched/fair.c (die:0x12dd892) frame base: cfa=1 fbreg=7 found "__mptr" (die: 0x13022f1) in scope=4/4 (die: 0x13022e8) failed: no/void pointer variable location: base=reg14, offset=0x140 type='void*' size=0x8 (die:0x12dd8f9) found "cfs_rq" (die: 0x1301721) in scope=3/4 (die: 0x130171c) type_offset=0x140 variable location: reg14 type='struct cfs_rq' size=0x1c0 (die:0x12e37e5) final type: type='struct cfs_rq' size=0x1c0 (die:0x12e37e5) IIUC the scope is like below: 1: update_blocked_averages 2: __update_blocked_fair 3: for_each_leaf_cfs_rq_safe 4: list_entry -> (container_of) The container_of is implemented like: #define container_of(ptr, type, member) ({ \ void *__mptr = (void *)(ptr); \ static_assert(__same_type(*(ptr), ((type *)0)->member) || \ __same_type(*(ptr), void), \ "pointer type mismatch in container_of()"); \ ((type *)(__mptr - offsetof(type, member))); }) That's why we see the __mptr variable first but it failed since it has no type information. Then for_each_leaf_cfs_rq_safe() is defined as #define for_each_leaf_cfs_rq_safe(rq, cfs_rq, pos) \ list_for_each_entry_safe(cfs_rq, pos, &rq->leaf_cfs_rq_list, \ leaf_cfs_rq_list) Note that the access was 0x140(r14). And the cfs_rq has leaf_cfs_rq_list at the 0x140. So it converts the list_head pointer to a pointer to struct cfs_rq here. $ pahole --hex -C cfs_rq vmlinux | grep 140 struct cfs_rq struct list_head leaf_cfs_rq_list; /* 0x140 0x10 */ Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240816235840.2754937-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-16 16:58:38 -07:00
if (!found && loc->multi_regs && reg == loc->reg1 && loc->reg1 != loc->reg2) {
reg = loc->reg2;
goto retry;
}
perf annotate-data: Check variables in every scope Sometimes it matches a variable in the inner scope but it fails because the actual access can be on a different type. Let's try variables in every scope and choose the best one using is_better_type(). I have an example with update_blocked_averages(), at first it found a variable (__mptr) but it's a void pointer. So it moved on to the upper scope and found another variable (cfs_rq). $ perf --debug type-profile annotate --data-type --stdio ... ----------------------------------------------------------- find data type for 0x140(reg14) at update_blocked_averages+0x2db CU for kernel/sched/fair.c (die:0x12dd892) frame base: cfa=1 fbreg=7 found "__mptr" (die: 0x13022f1) in scope=4/4 (die: 0x13022e8) failed: no/void pointer variable location: base=reg14, offset=0x140 type='void*' size=0x8 (die:0x12dd8f9) found "cfs_rq" (die: 0x1301721) in scope=3/4 (die: 0x130171c) type_offset=0x140 variable location: reg14 type='struct cfs_rq' size=0x1c0 (die:0x12e37e5) final type: type='struct cfs_rq' size=0x1c0 (die:0x12e37e5) IIUC the scope is like below: 1: update_blocked_averages 2: __update_blocked_fair 3: for_each_leaf_cfs_rq_safe 4: list_entry -> (container_of) The container_of is implemented like: #define container_of(ptr, type, member) ({ \ void *__mptr = (void *)(ptr); \ static_assert(__same_type(*(ptr), ((type *)0)->member) || \ __same_type(*(ptr), void), \ "pointer type mismatch in container_of()"); \ ((type *)(__mptr - offsetof(type, member))); }) That's why we see the __mptr variable first but it failed since it has no type information. Then for_each_leaf_cfs_rq_safe() is defined as #define for_each_leaf_cfs_rq_safe(rq, cfs_rq, pos) \ list_for_each_entry_safe(cfs_rq, pos, &rq->leaf_cfs_rq_list, \ leaf_cfs_rq_list) Note that the access was 0x140(r14). And the cfs_rq has leaf_cfs_rq_list at the 0x140. So it converts the list_head pointer to a pointer to struct cfs_rq here. $ pahole --hex -C cfs_rq vmlinux | grep 140 struct cfs_rq struct list_head leaf_cfs_rq_list; /* 0x140 0x10 */ Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240816235840.2754937-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-16 16:58:38 -07:00
if (!found && reg != DWARF_REG_PC) {
result = find_data_type_block(dloc, &cu_die, scopes,
perf annotate-data: Check variables in every scope Sometimes it matches a variable in the inner scope but it fails because the actual access can be on a different type. Let's try variables in every scope and choose the best one using is_better_type(). I have an example with update_blocked_averages(), at first it found a variable (__mptr) but it's a void pointer. So it moved on to the upper scope and found another variable (cfs_rq). $ perf --debug type-profile annotate --data-type --stdio ... ----------------------------------------------------------- find data type for 0x140(reg14) at update_blocked_averages+0x2db CU for kernel/sched/fair.c (die:0x12dd892) frame base: cfa=1 fbreg=7 found "__mptr" (die: 0x13022f1) in scope=4/4 (die: 0x13022e8) failed: no/void pointer variable location: base=reg14, offset=0x140 type='void*' size=0x8 (die:0x12dd8f9) found "cfs_rq" (die: 0x1301721) in scope=3/4 (die: 0x130171c) type_offset=0x140 variable location: reg14 type='struct cfs_rq' size=0x1c0 (die:0x12e37e5) final type: type='struct cfs_rq' size=0x1c0 (die:0x12e37e5) IIUC the scope is like below: 1: update_blocked_averages 2: __update_blocked_fair 3: for_each_leaf_cfs_rq_safe 4: list_entry -> (container_of) The container_of is implemented like: #define container_of(ptr, type, member) ({ \ void *__mptr = (void *)(ptr); \ static_assert(__same_type(*(ptr), ((type *)0)->member) || \ __same_type(*(ptr), void), \ "pointer type mismatch in container_of()"); \ ((type *)(__mptr - offsetof(type, member))); }) That's why we see the __mptr variable first but it failed since it has no type information. Then for_each_leaf_cfs_rq_safe() is defined as #define for_each_leaf_cfs_rq_safe(rq, cfs_rq, pos) \ list_for_each_entry_safe(cfs_rq, pos, &rq->leaf_cfs_rq_list, \ leaf_cfs_rq_list) Note that the access was 0x140(r14). And the cfs_rq has leaf_cfs_rq_list at the 0x140. So it converts the list_head pointer to a pointer to struct cfs_rq here. $ pahole --hex -C cfs_rq vmlinux | grep 140 struct cfs_rq struct list_head leaf_cfs_rq_list; /* 0x140 0x10 */ Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240816235840.2754937-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-16 16:58:38 -07:00
nr_scopes, type_die);
if (result == PERF_TMR_OK) {
perf annotate-data: Implement instruction tracking If it failed to find a variable for the location directly, it might be due to a missing variable in the source code. For example, accessing pointer variables in a chain can result in the case like below: struct foo *foo = ...; int i = foo->bar->baz; The DWARF debug information is created for each variable so it'd have one for 'foo'. But there's no variable for 'foo->bar' and then it cannot know the type of 'bar' and 'baz'. The above source code can be compiled to the follow x86 instructions: mov 0x8(%rax), %rcx mov 0x4(%rcx), %rdx <=== PMU sample mov %rdx, -4(%rbp) Let's say 'foo' is located in the %rax and it has a pointer to struct foo. But perf sample is captured in the second instruction and there is no variable or type info for the %rcx. It'd be great if compiler could generate debug info for %rcx, but we should handle it on our side. So this patch implements the logic to iterate instructions and update the type table for each location. As it already collected a list of scopes including the target instruction, we can use it to construct the type table smartly. +---------------- scope[0] subprogram | | +-------------- scope[1] lexical_block | | | | +------------ scope[2] inlined_subroutine | | | | | | +---------- scope[3] inlined_subroutine | | | | | | | | +-------- scope[4] lexical_block | | | | | | | | | | *** target instruction ... Image the target instruction has 5 scopes, each scope will have its own variables and parameters. Then it can start with the innermost scope (4). So it'd search the shortest path from the start of scope[4] to the target address and build a list of basic blocks. Then it iterates the basic blocks with the variables in the scope and update the table. If it finds a type at the target instruction, then returns it. Otherwise, it moves to the upper scope[3]. Now it'd search the shortest path from the start of scope[3] to the start of scope[4]. Then connect it to the existing basic block list. Then it'd iterate the blocks with variables for both scopes. It can repeat this until it finds a type at the target instruction or reaches to the top scope[0]. As the basic blocks contain the shortest path, it won't worry about branches and can update the table simply. The final check will be done by find_matching_type() in the next patch. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-15-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:06 -07:00
ann_data_stat.insn_track++;
perf annotate-data: Check variables in every scope Sometimes it matches a variable in the inner scope but it fails because the actual access can be on a different type. Let's try variables in every scope and choose the best one using is_better_type(). I have an example with update_blocked_averages(), at first it found a variable (__mptr) but it's a void pointer. So it moved on to the upper scope and found another variable (cfs_rq). $ perf --debug type-profile annotate --data-type --stdio ... ----------------------------------------------------------- find data type for 0x140(reg14) at update_blocked_averages+0x2db CU for kernel/sched/fair.c (die:0x12dd892) frame base: cfa=1 fbreg=7 found "__mptr" (die: 0x13022f1) in scope=4/4 (die: 0x13022e8) failed: no/void pointer variable location: base=reg14, offset=0x140 type='void*' size=0x8 (die:0x12dd8f9) found "cfs_rq" (die: 0x1301721) in scope=3/4 (die: 0x130171c) type_offset=0x140 variable location: reg14 type='struct cfs_rq' size=0x1c0 (die:0x12e37e5) final type: type='struct cfs_rq' size=0x1c0 (die:0x12e37e5) IIUC the scope is like below: 1: update_blocked_averages 2: __update_blocked_fair 3: for_each_leaf_cfs_rq_safe 4: list_entry -> (container_of) The container_of is implemented like: #define container_of(ptr, type, member) ({ \ void *__mptr = (void *)(ptr); \ static_assert(__same_type(*(ptr), ((type *)0)->member) || \ __same_type(*(ptr), void), \ "pointer type mismatch in container_of()"); \ ((type *)(__mptr - offsetof(type, member))); }) That's why we see the __mptr variable first but it failed since it has no type information. Then for_each_leaf_cfs_rq_safe() is defined as #define for_each_leaf_cfs_rq_safe(rq, cfs_rq, pos) \ list_for_each_entry_safe(cfs_rq, pos, &rq->leaf_cfs_rq_list, \ leaf_cfs_rq_list) Note that the access was 0x140(r14). And the cfs_rq has leaf_cfs_rq_list at the 0x140. So it converts the list_head pointer to a pointer to struct cfs_rq here. $ pahole --hex -C cfs_rq vmlinux | grep 140 struct cfs_rq struct list_head leaf_cfs_rq_list; /* 0x140 0x10 */ Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240816235840.2754937-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-16 16:58:38 -07:00
found = true;
perf annotate-data: Implement instruction tracking If it failed to find a variable for the location directly, it might be due to a missing variable in the source code. For example, accessing pointer variables in a chain can result in the case like below: struct foo *foo = ...; int i = foo->bar->baz; The DWARF debug information is created for each variable so it'd have one for 'foo'. But there's no variable for 'foo->bar' and then it cannot know the type of 'bar' and 'baz'. The above source code can be compiled to the follow x86 instructions: mov 0x8(%rax), %rcx mov 0x4(%rcx), %rdx <=== PMU sample mov %rdx, -4(%rbp) Let's say 'foo' is located in the %rax and it has a pointer to struct foo. But perf sample is captured in the second instruction and there is no variable or type info for the %rcx. It'd be great if compiler could generate debug info for %rcx, but we should handle it on our side. So this patch implements the logic to iterate instructions and update the type table for each location. As it already collected a list of scopes including the target instruction, we can use it to construct the type table smartly. +---------------- scope[0] subprogram | | +-------------- scope[1] lexical_block | | | | +------------ scope[2] inlined_subroutine | | | | | | +---------- scope[3] inlined_subroutine | | | | | | | | +-------- scope[4] lexical_block | | | | | | | | | | *** target instruction ... Image the target instruction has 5 scopes, each scope will have its own variables and parameters. Then it can start with the innermost scope (4). So it'd search the shortest path from the start of scope[4] to the target address and build a list of basic blocks. Then it iterates the basic blocks with the variables in the scope and update the table. If it finds a type at the target instruction, then returns it. Otherwise, it moves to the upper scope[3]. Now it'd search the shortest path from the start of scope[3] to the start of scope[4]. Then connect it to the existing basic block list. Then it'd iterate the blocks with variables for both scopes. It can repeat this until it finds a type at the target instruction or reaches to the top scope[0]. As the basic blocks contain the shortest path, it won't worry about branches and can update the table simply. The final check will be done by find_matching_type() in the next patch. Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-15-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:06 -07:00
}
}
out:
perf annotate-data: Update debug messages In check_matching_type(), it'd be easier to display the typename in question if it's available. For example, check out the line starts with 'chk'. ----------------------------------------------------------- find data type for 0x10(reg0) at cpuacct_charge+0x13 CU for kernel/sched/build_utility.c (die:0x137ee0b) frame base: cfa=1 fbreg=7 scope: [3/3] (die:13d9632) bb: [c - 13] var [c] reg5 type='struct task_struct*' size=0x8 (die:0x1381230) mov [c] 0xdf8(reg5) -> reg0 type='struct css_set*' size=0x8 (die:0x1385c56) chk [13] reg0 offset=0x10 ok=1 kind=1 (struct css_set*) : Good! <<<--- here found by insn track: 0x10(reg0) type-offset=0x10 final result: type='struct css_set' size=0x250 (die:0x1385b0e) Another example: ----------------------------------------------------------- find data type for 0x8(reg0) at menu_select+0x279 CU for drivers/cpuidle/governors/menu.c (die:0x7b0fe79) frame base: cfa=1 fbreg=7 scope: [2/2] (die:7b11010) bb: [273 - 277] bb: [279 - 279] chk [279] reg0 offset=0x8 ok=0 kind=0 cfa : no type information scope: [1/2] (die:7b10cbc) bb: [0 - 64] ... mov [26a] imm=0xffffffff -> reg15 bb: [273 - 277] bb: [279 - 279] chk [279] reg0 offset=0x8 ok=1 kind=1 (long long unsigned int) : no/void pointer <<<--- here final result: no/void pointer Also change some places to print negative offsets properly. Before: ----------------------------------------------------------- find data type for 0xffffff40(reg6) at __tcp_transmit_skb+0x58 After: ----------------------------------------------------------- find data type for -0xc0(reg6) at __tcp_transmit_skb+0x58 Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240821232628.353177-3-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-21 16:26:26 -07:00
pr_debug_dtp("final result: ");
perf annotate-data: Check variables in every scope Sometimes it matches a variable in the inner scope but it fails because the actual access can be on a different type. Let's try variables in every scope and choose the best one using is_better_type(). I have an example with update_blocked_averages(), at first it found a variable (__mptr) but it's a void pointer. So it moved on to the upper scope and found another variable (cfs_rq). $ perf --debug type-profile annotate --data-type --stdio ... ----------------------------------------------------------- find data type for 0x140(reg14) at update_blocked_averages+0x2db CU for kernel/sched/fair.c (die:0x12dd892) frame base: cfa=1 fbreg=7 found "__mptr" (die: 0x13022f1) in scope=4/4 (die: 0x13022e8) failed: no/void pointer variable location: base=reg14, offset=0x140 type='void*' size=0x8 (die:0x12dd8f9) found "cfs_rq" (die: 0x1301721) in scope=3/4 (die: 0x130171c) type_offset=0x140 variable location: reg14 type='struct cfs_rq' size=0x1c0 (die:0x12e37e5) final type: type='struct cfs_rq' size=0x1c0 (die:0x12e37e5) IIUC the scope is like below: 1: update_blocked_averages 2: __update_blocked_fair 3: for_each_leaf_cfs_rq_safe 4: list_entry -> (container_of) The container_of is implemented like: #define container_of(ptr, type, member) ({ \ void *__mptr = (void *)(ptr); \ static_assert(__same_type(*(ptr), ((type *)0)->member) || \ __same_type(*(ptr), void), \ "pointer type mismatch in container_of()"); \ ((type *)(__mptr - offsetof(type, member))); }) That's why we see the __mptr variable first but it failed since it has no type information. Then for_each_leaf_cfs_rq_safe() is defined as #define for_each_leaf_cfs_rq_safe(rq, cfs_rq, pos) \ list_for_each_entry_safe(cfs_rq, pos, &rq->leaf_cfs_rq_list, \ leaf_cfs_rq_list) Note that the access was 0x140(r14). And the cfs_rq has leaf_cfs_rq_list at the 0x140. So it converts the list_head pointer to a pointer to struct cfs_rq here. $ pahole --hex -C cfs_rq vmlinux | grep 140 struct cfs_rq struct list_head leaf_cfs_rq_list; /* 0x140 0x10 */ Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240816235840.2754937-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-16 16:58:38 -07:00
if (found) {
pr_debug_type_name(type_die, TSR_KIND_TYPE);
ret = 0;
} else {
switch (result) {
case PERF_TMR_NO_TYPE:
case PERF_TMR_NO_POINTER:
pr_debug_dtp("%s\n", match_result_str(result));
ann_data_stat.no_typeinfo++;
break;
case PERF_TMR_NO_SIZE:
pr_debug_dtp("%s\n", match_result_str(result));
ann_data_stat.invalid_size++;
break;
case PERF_TMR_BAD_OFFSET:
pr_debug_dtp("%s\n", match_result_str(result));
ann_data_stat.bad_offset++;
break;
case PERF_TMR_UNKNOWN:
case PERF_TMR_BAIL_OUT:
case PERF_TMR_OK: /* should not reach here */
default:
pr_debug_dtp("no variable found\n");
ann_data_stat.no_var++;
break;
}
ret = -1;
perf annotate-data: Add debug messages Add a new debug option "type-profile" to enable the detailed info during the type analysis especially for instruction tracking. You can use this before the command name like 'report' or 'annotate'. $ perf --debug type-profile annotate --data-type Committer testing: First get some memory events: $ perf mem record ls Then, without data-type profiling debug: $ perf annotate --data-type | head Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ And with only data-type profiling: $ perf --debug type-profile annotate --data-type | head ----------------------------------------------------------- find_data_type_die [1e67] for reg13873052 (PC) offset=0x150e2 in dl_main CU die offset: 0x29cd3 found PC-rel by addr=0x34020 offset=0x20 ----------------------------------------------------------- find_data_type_die [2e] for reg12 offset=0 in __GI___readdir64 CU die offset: 0x137a45 frame base: cfa=1 fbreg=-1 found "__futex" in scope=2/2 (die: 0x137ad5) 0(reg12) type=int (die:2a) ----------------------------------------------------------- find_data_type_die [52] for reg5 offset=0 in __memmove_avx_unaligned_erms CU die offset: 0x1124ed no variable found Annotate type: 'struct rtld_global' in /usr/lib64/ld-linux-x86-64.so.2 (1 samples): ============================================================================ samples offset size field 1 0 4336 struct rtld_global { 0 0 0 struct link_namespaces* _dl_ns; 0 2560 8 size_t _dl_nns; 0 2568 40 __rtld_lock_recursive_t _dl_load_lock { 0 2568 40 pthread_mutex_t mutex { 0 2568 40 struct __pthread_mutex_s __data { 0 2568 4 int __lock; $ Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20240319055115.4063940-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-03-18 22:51:00 -07:00
}
free(scopes);
return ret;
}
/**
* find_data_type - Return a data type at the location
* @dloc: data location
*
* This functions searches the debug information of the binary to get the data
* type it accesses. The exact location is expressed by (ip, reg, offset)
* for pointer variables or (ip, addr) for global variables. Note that global
* variables might update the @dloc->type_offset after finding the start of the
* variable. If it cannot find a global variable by address, it tried to find
* a declaration of the variable using var_name. In that case, @dloc->offset
* won't be updated.
*
* It return %NULL if not found.
*/
struct annotated_data_type *find_data_type(struct data_loc_info *dloc)
{
struct dso *dso = map__dso(dloc->ms->map);
Dwarf_Die type_die;
/*
* The type offset is the same as instruction offset by default.
* But when finding a global variable, the offset won't be valid.
*/
dloc->type_offset = dloc->op->offset;
dloc->fbreg = -1;
if (find_data_type_die(dloc, &type_die) < 0)
perf annotate: Cache debuginfo for data type profiling In find_data_type(), it creates and deletes a debug info whenver it tries to find data type for a sample. This is inefficient and it most likely accesses the same binary again and again. Let's add a single entry cache the debug info structure for the last DSO. Depending on sample data, it usually gives me 2~3x (and sometimes more) speed ups. Note that this will introduce a little difference in the output due to the order of checking stack operations. It used to check the stack ops before checking the availability of debug info but I moved it after the symbol check. So it'll report stack operations in DSOs without debug info as unknown. But I think it's ok and better to have the checking near the caching logic. Committer testing: root@x1:~# perf mem record -a sleep 5s root@x1:~# perf evlist cpu_atom/mem-loads,ldlat=30/P cpu_atom/mem-stores/P dummy:u root@x1:~# diff -u before after --- before 2024-08-08 09:33:53.880780784 -0300 +++ after 2024-08-08 09:35:13.917325041 -0300 @@ -81,8 +81,8 @@ # Overhead Data Type # ........ ......... # - 55.43% (unknown) - 11.61% (stack operation) + 55.56% (unknown) + 11.48% (stack operation) 4.93% struct pcpu_hot 3.26% unsigned int 2.48% struct Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240805234648.1453689-1-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-05 16:46:48 -07:00
return NULL;
perf annotate: Cache debuginfo for data type profiling In find_data_type(), it creates and deletes a debug info whenver it tries to find data type for a sample. This is inefficient and it most likely accesses the same binary again and again. Let's add a single entry cache the debug info structure for the last DSO. Depending on sample data, it usually gives me 2~3x (and sometimes more) speed ups. Note that this will introduce a little difference in the output due to the order of checking stack operations. It used to check the stack ops before checking the availability of debug info but I moved it after the symbol check. So it'll report stack operations in DSOs without debug info as unknown. But I think it's ok and better to have the checking near the caching logic. Committer testing: root@x1:~# perf mem record -a sleep 5s root@x1:~# perf evlist cpu_atom/mem-loads,ldlat=30/P cpu_atom/mem-stores/P dummy:u root@x1:~# diff -u before after --- before 2024-08-08 09:33:53.880780784 -0300 +++ after 2024-08-08 09:35:13.917325041 -0300 @@ -81,8 +81,8 @@ # Overhead Data Type # ........ ......... # - 55.43% (unknown) - 11.61% (stack operation) + 55.56% (unknown) + 11.48% (stack operation) 4.93% struct pcpu_hot 3.26% unsigned int 2.48% struct Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240805234648.1453689-1-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-05 16:46:48 -07:00
return dso__findnew_data_type(dso, &type_die);
}
perf annotate-data: Add dso->data_types tree To aggregate accesses to the same data type, add 'data_types' tree in DSO to maintain data types and find it by name and size. It might have different data types that happen to have the same name, so it also compares the size of the type. Even if it doesn't 100% guarantee, it reduces the possibility of mis-handling of such conflicts. And I don't think it's common to have different types with the same name. Committer notes: Very few cases on the Linux kernel, but there are some different types with the same name, unsure if there is a debug mode in libbpf dedup that warns about such cases, but there are provisions in pahole for that, see: "emit: Notice type shadowing, i.e. multiple types with the same name (enum, struct, union, etc)" https://git.kernel.org/pub/scm/devel/pahole/pahole.git/commit/?id=4f332dbfd02072e4f410db7bdcda8d6e3422974b $ pahole --compile > vmlinux.h $ rm -f a ; make a cc a.c -o a $ grep __[0-9] vmlinux.h union irte__1 { struct map_info__1; struct map_info__1 { struct map_info__1 * next; /* 0 8 */ $ drivers/iommu/amd/amd_iommu_types.h 'union irte' include/linux/dmar.h 'struct irte' include/linux/device-mapper.h: union map_info { void *ptr; }; include/linux/mtd/map.h: struct map_info { const char *name; unsigned long size; resource_size_t phys; <SNIP> kernel/events/uprobes.c: struct map_info { struct map_info *next; struct mm_struct *mm; unsigned long vaddr; }; Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Cc: linux-toolchains@vger.kernel.org Cc: linux-trace-devel@vger.kernel.org Link: https://lore.kernel.org/r/20231213001323.718046-5-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2023-12-12 16:13:10 -08:00
static int alloc_data_type_histograms(struct annotated_data_type *adt, int nr_entries)
{
int i;
size_t sz = sizeof(struct type_hist);
sz += sizeof(struct type_hist_entry) * adt->self.size;
/* Allocate a table of pointers for each event */
adt->histograms = calloc(nr_entries, sizeof(*adt->histograms));
if (adt->histograms == NULL)
return -ENOMEM;
/*
* Each histogram is allocated for the whole size of the type.
* TODO: Probably we can move the histogram to members.
*/
for (i = 0; i < nr_entries; i++) {
adt->histograms[i] = zalloc(sz);
if (adt->histograms[i] == NULL)
goto err;
}
adt->nr_histograms = nr_entries;
return 0;
err:
while (--i >= 0)
zfree(&(adt->histograms[i]));
zfree(&adt->histograms);
return -ENOMEM;
}
static void delete_data_type_histograms(struct annotated_data_type *adt)
{
for (int i = 0; i < adt->nr_histograms; i++)
zfree(&(adt->histograms[i]));
zfree(&adt->histograms);
adt->nr_histograms = 0;
}
perf annotate-data: Add dso->data_types tree To aggregate accesses to the same data type, add 'data_types' tree in DSO to maintain data types and find it by name and size. It might have different data types that happen to have the same name, so it also compares the size of the type. Even if it doesn't 100% guarantee, it reduces the possibility of mis-handling of such conflicts. And I don't think it's common to have different types with the same name. Committer notes: Very few cases on the Linux kernel, but there are some different types with the same name, unsure if there is a debug mode in libbpf dedup that warns about such cases, but there are provisions in pahole for that, see: "emit: Notice type shadowing, i.e. multiple types with the same name (enum, struct, union, etc)" https://git.kernel.org/pub/scm/devel/pahole/pahole.git/commit/?id=4f332dbfd02072e4f410db7bdcda8d6e3422974b $ pahole --compile > vmlinux.h $ rm -f a ; make a cc a.c -o a $ grep __[0-9] vmlinux.h union irte__1 { struct map_info__1; struct map_info__1 { struct map_info__1 * next; /* 0 8 */ $ drivers/iommu/amd/amd_iommu_types.h 'union irte' include/linux/dmar.h 'struct irte' include/linux/device-mapper.h: union map_info { void *ptr; }; include/linux/mtd/map.h: struct map_info { const char *name; unsigned long size; resource_size_t phys; <SNIP> kernel/events/uprobes.c: struct map_info { struct map_info *next; struct mm_struct *mm; unsigned long vaddr; }; Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Cc: linux-toolchains@vger.kernel.org Cc: linux-trace-devel@vger.kernel.org Link: https://lore.kernel.org/r/20231213001323.718046-5-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2023-12-12 16:13:10 -08:00
void annotated_data_type__tree_delete(struct rb_root *root)
{
struct annotated_data_type *pos;
while (!RB_EMPTY_ROOT(root)) {
struct rb_node *node = rb_first(root);
rb_erase(node, root);
pos = rb_entry(node, struct annotated_data_type, node);
delete_members(&pos->self);
delete_data_type_histograms(pos);
zfree(&pos->self.type_name);
perf annotate-data: Add dso->data_types tree To aggregate accesses to the same data type, add 'data_types' tree in DSO to maintain data types and find it by name and size. It might have different data types that happen to have the same name, so it also compares the size of the type. Even if it doesn't 100% guarantee, it reduces the possibility of mis-handling of such conflicts. And I don't think it's common to have different types with the same name. Committer notes: Very few cases on the Linux kernel, but there are some different types with the same name, unsure if there is a debug mode in libbpf dedup that warns about such cases, but there are provisions in pahole for that, see: "emit: Notice type shadowing, i.e. multiple types with the same name (enum, struct, union, etc)" https://git.kernel.org/pub/scm/devel/pahole/pahole.git/commit/?id=4f332dbfd02072e4f410db7bdcda8d6e3422974b $ pahole --compile > vmlinux.h $ rm -f a ; make a cc a.c -o a $ grep __[0-9] vmlinux.h union irte__1 { struct map_info__1; struct map_info__1 { struct map_info__1 * next; /* 0 8 */ $ drivers/iommu/amd/amd_iommu_types.h 'union irte' include/linux/dmar.h 'struct irte' include/linux/device-mapper.h: union map_info { void *ptr; }; include/linux/mtd/map.h: struct map_info { const char *name; unsigned long size; resource_size_t phys; <SNIP> kernel/events/uprobes.c: struct map_info { struct map_info *next; struct mm_struct *mm; unsigned long vaddr; }; Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Cc: linux-toolchains@vger.kernel.org Cc: linux-trace-devel@vger.kernel.org Link: https://lore.kernel.org/r/20231213001323.718046-5-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2023-12-12 16:13:10 -08:00
free(pos);
}
}
/**
* annotated_data_type__update_samples - Update histogram
* @adt: Data type to update
* @evsel: Event to update
* @offset: Offset in the type
* @nr_samples: Number of samples at this offset
* @period: Event count at this offset
*
* This function updates type histogram at @ofs for @evsel. Samples are
* aggregated before calling this function so it can be called with more
* than one samples at a certain offset.
*/
int annotated_data_type__update_samples(struct annotated_data_type *adt,
struct evsel *evsel, int offset,
int nr_samples, u64 period)
{
struct type_hist *h;
if (adt == NULL)
return 0;
if (adt->histograms == NULL) {
int nr = evsel->evlist->core.nr_entries;
if (alloc_data_type_histograms(adt, nr) < 0)
return -1;
}
if (offset < 0 || offset >= adt->self.size)
return -1;
h = adt->histograms[evsel->core.idx];
h->nr_samples += nr_samples;
h->addr[offset].nr_samples += nr_samples;
h->period += period;
h->addr[offset].period += period;
return 0;
}
static void print_annotated_data_header(struct hist_entry *he, struct evsel *evsel)
{
struct dso *dso = map__dso(he->ms.map);
int nr_members = 1;
int nr_samples = he->stat.nr_events;
int width = 7;
const char *val_hdr = "Percent";
if (evsel__is_group_event(evsel)) {
struct hist_entry *pair;
list_for_each_entry(pair, &he->pairs.head, pairs.node)
nr_samples += pair->stat.nr_events;
}
printf("Annotate type: '%s' in %s (%d samples):\n",
perf dso: Add reference count checking and accessor functions Add reference count checking to struct dso, this can help with implementing correct reference counting discipline. To avoid RC_CHK_ACCESS everywhere, add accessor functions for the variables in struct dso. The majority of the change is mechanical in nature and not easy to split up. Committer testing: 'perf test' up to this patch shows no regressions. But: util/symbol.c: In function ‘dso__load_bfd_symbols’: util/symbol.c:1683:9: error: too few arguments to function ‘dso__set_adjust_symbols’ 1683 | dso__set_adjust_symbols(dso); | ^~~~~~~~~~~~~~~~~~~~~~~ In file included from util/symbol.c:21: util/dso.h:268:20: note: declared here 268 | static inline void dso__set_adjust_symbols(struct dso *dso, bool val) | ^~~~~~~~~~~~~~~~~~~~~~~ make[6]: *** [/home/acme/git/perf-tools-next/tools/build/Makefile.build:106: /tmp/tmp.ZWHbQftdN6/util/symbol.o] Error 1 MKDIR /tmp/tmp.ZWHbQftdN6/tests/workloads/ make[6]: *** Waiting for unfinished jobs.... This was updated: - symbols__fixup_end(&dso->symbols, false); - symbols__fixup_duplicate(&dso->symbols); - dso->adjust_symbols = 1; + symbols__fixup_end(dso__symbols(dso), false); + symbols__fixup_duplicate(dso__symbols(dso)); + dso__set_adjust_symbols(dso); But not build tested with BUILD_NONDISTRO and libbfd devel files installed (binutils-devel on fedora). Add the missing argument: symbols__fixup_end(dso__symbols(dso), false); symbols__fixup_duplicate(dso__symbols(dso)); - dso__set_adjust_symbols(dso); + dso__set_adjust_symbols(dso, true); Signed-off-by: Ian Rogers <irogers@google.com> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Andi Kleen <ak@linux.intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ben Gainey <ben.gainey@arm.com> Cc: Changbin Du <changbin.du@huawei.com> Cc: Chengen Du <chengen.du@canonical.com> Cc: Colin Ian King <colin.i.king@gmail.com> Cc: Dima Kogan <dima@secretsauce.net> Cc: Ilkka Koskinen <ilkka@os.amperecomputing.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: James Clark <james.clark@arm.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: K Prateek Nayak <kprateek.nayak@amd.com> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Leo Yan <leo.yan@linux.dev> Cc: Li Dong <lidong@vivo.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Paran Lee <p4ranlee@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Song Liu <song@kernel.org> Cc: Sun Haiyong <sunhaiyong@loongson.cn> Cc: Thomas Richter <tmricht@linux.ibm.com> Cc: Tiezhu Yang <yangtiezhu@loongson.cn> Cc: Yanteng Si <siyanteng@loongson.cn> Cc: zhaimingbing <zhaimingbing@cmss.chinamobile.com> Link: https://lore.kernel.org/r/20240504213803.218974-6-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-05-04 14:38:01 -07:00
he->mem_type->self.type_name, dso__name(dso), nr_samples);
if (evsel__is_group_event(evsel)) {
struct evsel *pos;
int i = 0;
perf annotate-data: Support --skip-empty option The --skip-empty option is to hide dummy events in a group. Like other output mode in 'perf report' and 'perf annotate', the data-type profiling output should support the option. Committer testing: With dummy: root@number:~# perf annotate --stdio --group --data-type --skip-empty | head -24 Annotate type: 'pthread_mutex_t' in /usr/lib64/libc.so.6 (50 samples): event[0] = cpu_atom/mem-loads,ldlat=30/P event[1] = cpu_atom/mem-stores/P event[2] = dummy:u ============================================================================ Percent offset size field 100.00 100.00 0.00 0 40 pthread_mutex_t { 100.00 100.00 0.00 0 40 struct __pthread_mutex_s __data { 45.21 84.54 0.00 0 4 int __lock; 0.00 0.00 0.00 4 4 unsigned int __count; 0.00 1.83 0.00 8 4 int __owner; 5.19 10.65 0.00 12 4 unsigned int __nusers; 49.61 2.97 0.00 16 4 int __kind; 0.00 0.00 0.00 20 2 short int __spins; 0.00 0.00 0.00 22 2 short int __elision; 0.00 0.00 0.00 24 16 __pthread_list_t __list { 0.00 0.00 0.00 24 8 struct __pthread_internal_list* __prev; 0.00 0.00 0.00 32 8 struct __pthread_internal_list* __next; }; }; 0.00 0.00 0.00 0 0 char[] __size; 45.21 84.54 0.00 0 8 long int __align; }; Skipping it: root@number:~# perf annotate --stdio --group --data-type --skip-empty | head -24 Annotate type: 'pthread_mutex_t' in /usr/lib64/libc.so.6 (50 samples): event[0] = cpu_atom/mem-loads,ldlat=30/P event[1] = cpu_atom/mem-stores/P ============================================================================ Percent offset size field 100.00 100.00 0 40 pthread_mutex_t { 100.00 100.00 0 40 struct __pthread_mutex_s __data { 45.21 84.54 0 4 int __lock; 0.00 0.00 4 4 unsigned int __count; 0.00 1.83 8 4 int __owner; 5.19 10.65 12 4 unsigned int __nusers; 49.61 2.97 16 4 int __kind; 0.00 0.00 20 2 short int __spins; 0.00 0.00 22 2 short int __elision; 0.00 0.00 24 16 __pthread_list_t __list { 0.00 0.00 24 8 struct __pthread_internal_list* __prev; 0.00 0.00 32 8 struct __pthread_internal_list* __next; }; }; 0.00 0.00 0 0 char[] __size; 45.21 84.54 0 8 long int __align; }; Annotate type: 'pthread_mutexattr_t' in /usr/lib64/libc.so.6 (1 samples): root@number:~# Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240807061713.1642924-1-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-06 23:17:13 -07:00
nr_members = 0;
for_each_group_evsel(pos, evsel) {
if (symbol_conf.skip_empty &&
evsel__hists(pos)->stats.nr_samples == 0)
continue;
perf annotate-data: Support --skip-empty option The --skip-empty option is to hide dummy events in a group. Like other output mode in 'perf report' and 'perf annotate', the data-type profiling output should support the option. Committer testing: With dummy: root@number:~# perf annotate --stdio --group --data-type --skip-empty | head -24 Annotate type: 'pthread_mutex_t' in /usr/lib64/libc.so.6 (50 samples): event[0] = cpu_atom/mem-loads,ldlat=30/P event[1] = cpu_atom/mem-stores/P event[2] = dummy:u ============================================================================ Percent offset size field 100.00 100.00 0.00 0 40 pthread_mutex_t { 100.00 100.00 0.00 0 40 struct __pthread_mutex_s __data { 45.21 84.54 0.00 0 4 int __lock; 0.00 0.00 0.00 4 4 unsigned int __count; 0.00 1.83 0.00 8 4 int __owner; 5.19 10.65 0.00 12 4 unsigned int __nusers; 49.61 2.97 0.00 16 4 int __kind; 0.00 0.00 0.00 20 2 short int __spins; 0.00 0.00 0.00 22 2 short int __elision; 0.00 0.00 0.00 24 16 __pthread_list_t __list { 0.00 0.00 0.00 24 8 struct __pthread_internal_list* __prev; 0.00 0.00 0.00 32 8 struct __pthread_internal_list* __next; }; }; 0.00 0.00 0.00 0 0 char[] __size; 45.21 84.54 0.00 0 8 long int __align; }; Skipping it: root@number:~# perf annotate --stdio --group --data-type --skip-empty | head -24 Annotate type: 'pthread_mutex_t' in /usr/lib64/libc.so.6 (50 samples): event[0] = cpu_atom/mem-loads,ldlat=30/P event[1] = cpu_atom/mem-stores/P ============================================================================ Percent offset size field 100.00 100.00 0 40 pthread_mutex_t { 100.00 100.00 0 40 struct __pthread_mutex_s __data { 45.21 84.54 0 4 int __lock; 0.00 0.00 4 4 unsigned int __count; 0.00 1.83 8 4 int __owner; 5.19 10.65 12 4 unsigned int __nusers; 49.61 2.97 16 4 int __kind; 0.00 0.00 20 2 short int __spins; 0.00 0.00 22 2 short int __elision; 0.00 0.00 24 16 __pthread_list_t __list { 0.00 0.00 24 8 struct __pthread_internal_list* __prev; 0.00 0.00 32 8 struct __pthread_internal_list* __next; }; }; 0.00 0.00 0 0 char[] __size; 45.21 84.54 0 8 long int __align; }; Annotate type: 'pthread_mutexattr_t' in /usr/lib64/libc.so.6 (1 samples): root@number:~# Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240807061713.1642924-1-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-06 23:17:13 -07:00
printf(" event[%d] = %s\n", i++, pos->name);
nr_members++;
}
}
if (symbol_conf.show_total_period) {
width = 11;
val_hdr = "Period";
} else if (symbol_conf.show_nr_samples) {
width = 7;
val_hdr = "Samples";
}
printf("============================================================================\n");
printf("%*s %10s %10s %s\n", (width + 1) * nr_members, val_hdr,
"offset", "size", "field");
}
static void print_annotated_data_value(struct type_hist *h, u64 period, int nr_samples)
{
double percent = h->period ? (100.0 * period / h->period) : 0;
const char *color = get_percent_color(percent);
if (symbol_conf.show_total_period)
color_fprintf(stdout, color, " %11" PRIu64, period);
else if (symbol_conf.show_nr_samples)
color_fprintf(stdout, color, " %7d", nr_samples);
else
color_fprintf(stdout, color, " %7.2f", percent);
}
static void print_annotated_data_type(struct annotated_data_type *mem_type,
struct annotated_member *member,
struct evsel *evsel, int indent)
{
struct annotated_member *child;
struct type_hist *h = mem_type->histograms[evsel->core.idx];
perf annotate-data: Support --skip-empty option The --skip-empty option is to hide dummy events in a group. Like other output mode in 'perf report' and 'perf annotate', the data-type profiling output should support the option. Committer testing: With dummy: root@number:~# perf annotate --stdio --group --data-type --skip-empty | head -24 Annotate type: 'pthread_mutex_t' in /usr/lib64/libc.so.6 (50 samples): event[0] = cpu_atom/mem-loads,ldlat=30/P event[1] = cpu_atom/mem-stores/P event[2] = dummy:u ============================================================================ Percent offset size field 100.00 100.00 0.00 0 40 pthread_mutex_t { 100.00 100.00 0.00 0 40 struct __pthread_mutex_s __data { 45.21 84.54 0.00 0 4 int __lock; 0.00 0.00 0.00 4 4 unsigned int __count; 0.00 1.83 0.00 8 4 int __owner; 5.19 10.65 0.00 12 4 unsigned int __nusers; 49.61 2.97 0.00 16 4 int __kind; 0.00 0.00 0.00 20 2 short int __spins; 0.00 0.00 0.00 22 2 short int __elision; 0.00 0.00 0.00 24 16 __pthread_list_t __list { 0.00 0.00 0.00 24 8 struct __pthread_internal_list* __prev; 0.00 0.00 0.00 32 8 struct __pthread_internal_list* __next; }; }; 0.00 0.00 0.00 0 0 char[] __size; 45.21 84.54 0.00 0 8 long int __align; }; Skipping it: root@number:~# perf annotate --stdio --group --data-type --skip-empty | head -24 Annotate type: 'pthread_mutex_t' in /usr/lib64/libc.so.6 (50 samples): event[0] = cpu_atom/mem-loads,ldlat=30/P event[1] = cpu_atom/mem-stores/P ============================================================================ Percent offset size field 100.00 100.00 0 40 pthread_mutex_t { 100.00 100.00 0 40 struct __pthread_mutex_s __data { 45.21 84.54 0 4 int __lock; 0.00 0.00 4 4 unsigned int __count; 0.00 1.83 8 4 int __owner; 5.19 10.65 12 4 unsigned int __nusers; 49.61 2.97 16 4 int __kind; 0.00 0.00 20 2 short int __spins; 0.00 0.00 22 2 short int __elision; 0.00 0.00 24 16 __pthread_list_t __list { 0.00 0.00 24 8 struct __pthread_internal_list* __prev; 0.00 0.00 32 8 struct __pthread_internal_list* __next; }; }; 0.00 0.00 0 0 char[] __size; 45.21 84.54 0 8 long int __align; }; Annotate type: 'pthread_mutexattr_t' in /usr/lib64/libc.so.6 (1 samples): root@number:~# Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240807061713.1642924-1-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-06 23:17:13 -07:00
int i, nr_events = 0, samples = 0;
u64 period = 0;
int width = symbol_conf.show_total_period ? 11 : 7;
perf annotate-data: Support --skip-empty option The --skip-empty option is to hide dummy events in a group. Like other output mode in 'perf report' and 'perf annotate', the data-type profiling output should support the option. Committer testing: With dummy: root@number:~# perf annotate --stdio --group --data-type --skip-empty | head -24 Annotate type: 'pthread_mutex_t' in /usr/lib64/libc.so.6 (50 samples): event[0] = cpu_atom/mem-loads,ldlat=30/P event[1] = cpu_atom/mem-stores/P event[2] = dummy:u ============================================================================ Percent offset size field 100.00 100.00 0.00 0 40 pthread_mutex_t { 100.00 100.00 0.00 0 40 struct __pthread_mutex_s __data { 45.21 84.54 0.00 0 4 int __lock; 0.00 0.00 0.00 4 4 unsigned int __count; 0.00 1.83 0.00 8 4 int __owner; 5.19 10.65 0.00 12 4 unsigned int __nusers; 49.61 2.97 0.00 16 4 int __kind; 0.00 0.00 0.00 20 2 short int __spins; 0.00 0.00 0.00 22 2 short int __elision; 0.00 0.00 0.00 24 16 __pthread_list_t __list { 0.00 0.00 0.00 24 8 struct __pthread_internal_list* __prev; 0.00 0.00 0.00 32 8 struct __pthread_internal_list* __next; }; }; 0.00 0.00 0.00 0 0 char[] __size; 45.21 84.54 0.00 0 8 long int __align; }; Skipping it: root@number:~# perf annotate --stdio --group --data-type --skip-empty | head -24 Annotate type: 'pthread_mutex_t' in /usr/lib64/libc.so.6 (50 samples): event[0] = cpu_atom/mem-loads,ldlat=30/P event[1] = cpu_atom/mem-stores/P ============================================================================ Percent offset size field 100.00 100.00 0 40 pthread_mutex_t { 100.00 100.00 0 40 struct __pthread_mutex_s __data { 45.21 84.54 0 4 int __lock; 0.00 0.00 4 4 unsigned int __count; 0.00 1.83 8 4 int __owner; 5.19 10.65 12 4 unsigned int __nusers; 49.61 2.97 16 4 int __kind; 0.00 0.00 20 2 short int __spins; 0.00 0.00 22 2 short int __elision; 0.00 0.00 24 16 __pthread_list_t __list { 0.00 0.00 24 8 struct __pthread_internal_list* __prev; 0.00 0.00 32 8 struct __pthread_internal_list* __next; }; }; 0.00 0.00 0 0 char[] __size; 45.21 84.54 0 8 long int __align; }; Annotate type: 'pthread_mutexattr_t' in /usr/lib64/libc.so.6 (1 samples): root@number:~# Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240807061713.1642924-1-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-06 23:17:13 -07:00
struct evsel *pos;
perf annotate-data: Support --skip-empty option The --skip-empty option is to hide dummy events in a group. Like other output mode in 'perf report' and 'perf annotate', the data-type profiling output should support the option. Committer testing: With dummy: root@number:~# perf annotate --stdio --group --data-type --skip-empty | head -24 Annotate type: 'pthread_mutex_t' in /usr/lib64/libc.so.6 (50 samples): event[0] = cpu_atom/mem-loads,ldlat=30/P event[1] = cpu_atom/mem-stores/P event[2] = dummy:u ============================================================================ Percent offset size field 100.00 100.00 0.00 0 40 pthread_mutex_t { 100.00 100.00 0.00 0 40 struct __pthread_mutex_s __data { 45.21 84.54 0.00 0 4 int __lock; 0.00 0.00 0.00 4 4 unsigned int __count; 0.00 1.83 0.00 8 4 int __owner; 5.19 10.65 0.00 12 4 unsigned int __nusers; 49.61 2.97 0.00 16 4 int __kind; 0.00 0.00 0.00 20 2 short int __spins; 0.00 0.00 0.00 22 2 short int __elision; 0.00 0.00 0.00 24 16 __pthread_list_t __list { 0.00 0.00 0.00 24 8 struct __pthread_internal_list* __prev; 0.00 0.00 0.00 32 8 struct __pthread_internal_list* __next; }; }; 0.00 0.00 0.00 0 0 char[] __size; 45.21 84.54 0.00 0 8 long int __align; }; Skipping it: root@number:~# perf annotate --stdio --group --data-type --skip-empty | head -24 Annotate type: 'pthread_mutex_t' in /usr/lib64/libc.so.6 (50 samples): event[0] = cpu_atom/mem-loads,ldlat=30/P event[1] = cpu_atom/mem-stores/P ============================================================================ Percent offset size field 100.00 100.00 0 40 pthread_mutex_t { 100.00 100.00 0 40 struct __pthread_mutex_s __data { 45.21 84.54 0 4 int __lock; 0.00 0.00 4 4 unsigned int __count; 0.00 1.83 8 4 int __owner; 5.19 10.65 12 4 unsigned int __nusers; 49.61 2.97 16 4 int __kind; 0.00 0.00 20 2 short int __spins; 0.00 0.00 22 2 short int __elision; 0.00 0.00 24 16 __pthread_list_t __list { 0.00 0.00 24 8 struct __pthread_internal_list* __prev; 0.00 0.00 32 8 struct __pthread_internal_list* __next; }; }; 0.00 0.00 0 0 char[] __size; 45.21 84.54 0 8 long int __align; }; Annotate type: 'pthread_mutexattr_t' in /usr/lib64/libc.so.6 (1 samples): root@number:~# Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240807061713.1642924-1-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-06 23:17:13 -07:00
for_each_group_evsel(pos, evsel) {
h = mem_type->histograms[pos->core.idx];
perf annotate-data: Support --skip-empty option The --skip-empty option is to hide dummy events in a group. Like other output mode in 'perf report' and 'perf annotate', the data-type profiling output should support the option. Committer testing: With dummy: root@number:~# perf annotate --stdio --group --data-type --skip-empty | head -24 Annotate type: 'pthread_mutex_t' in /usr/lib64/libc.so.6 (50 samples): event[0] = cpu_atom/mem-loads,ldlat=30/P event[1] = cpu_atom/mem-stores/P event[2] = dummy:u ============================================================================ Percent offset size field 100.00 100.00 0.00 0 40 pthread_mutex_t { 100.00 100.00 0.00 0 40 struct __pthread_mutex_s __data { 45.21 84.54 0.00 0 4 int __lock; 0.00 0.00 0.00 4 4 unsigned int __count; 0.00 1.83 0.00 8 4 int __owner; 5.19 10.65 0.00 12 4 unsigned int __nusers; 49.61 2.97 0.00 16 4 int __kind; 0.00 0.00 0.00 20 2 short int __spins; 0.00 0.00 0.00 22 2 short int __elision; 0.00 0.00 0.00 24 16 __pthread_list_t __list { 0.00 0.00 0.00 24 8 struct __pthread_internal_list* __prev; 0.00 0.00 0.00 32 8 struct __pthread_internal_list* __next; }; }; 0.00 0.00 0.00 0 0 char[] __size; 45.21 84.54 0.00 0 8 long int __align; }; Skipping it: root@number:~# perf annotate --stdio --group --data-type --skip-empty | head -24 Annotate type: 'pthread_mutex_t' in /usr/lib64/libc.so.6 (50 samples): event[0] = cpu_atom/mem-loads,ldlat=30/P event[1] = cpu_atom/mem-stores/P ============================================================================ Percent offset size field 100.00 100.00 0 40 pthread_mutex_t { 100.00 100.00 0 40 struct __pthread_mutex_s __data { 45.21 84.54 0 4 int __lock; 0.00 0.00 4 4 unsigned int __count; 0.00 1.83 8 4 int __owner; 5.19 10.65 12 4 unsigned int __nusers; 49.61 2.97 16 4 int __kind; 0.00 0.00 20 2 short int __spins; 0.00 0.00 22 2 short int __elision; 0.00 0.00 24 16 __pthread_list_t __list { 0.00 0.00 24 8 struct __pthread_internal_list* __prev; 0.00 0.00 32 8 struct __pthread_internal_list* __next; }; }; 0.00 0.00 0 0 char[] __size; 45.21 84.54 0 8 long int __align; }; Annotate type: 'pthread_mutexattr_t' in /usr/lib64/libc.so.6 (1 samples): root@number:~# Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240807061713.1642924-1-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-06 23:17:13 -07:00
if (symbol_conf.skip_empty &&
evsel__hists(pos)->stats.nr_samples == 0)
continue;
perf annotate-data: Support --skip-empty option The --skip-empty option is to hide dummy events in a group. Like other output mode in 'perf report' and 'perf annotate', the data-type profiling output should support the option. Committer testing: With dummy: root@number:~# perf annotate --stdio --group --data-type --skip-empty | head -24 Annotate type: 'pthread_mutex_t' in /usr/lib64/libc.so.6 (50 samples): event[0] = cpu_atom/mem-loads,ldlat=30/P event[1] = cpu_atom/mem-stores/P event[2] = dummy:u ============================================================================ Percent offset size field 100.00 100.00 0.00 0 40 pthread_mutex_t { 100.00 100.00 0.00 0 40 struct __pthread_mutex_s __data { 45.21 84.54 0.00 0 4 int __lock; 0.00 0.00 0.00 4 4 unsigned int __count; 0.00 1.83 0.00 8 4 int __owner; 5.19 10.65 0.00 12 4 unsigned int __nusers; 49.61 2.97 0.00 16 4 int __kind; 0.00 0.00 0.00 20 2 short int __spins; 0.00 0.00 0.00 22 2 short int __elision; 0.00 0.00 0.00 24 16 __pthread_list_t __list { 0.00 0.00 0.00 24 8 struct __pthread_internal_list* __prev; 0.00 0.00 0.00 32 8 struct __pthread_internal_list* __next; }; }; 0.00 0.00 0.00 0 0 char[] __size; 45.21 84.54 0.00 0 8 long int __align; }; Skipping it: root@number:~# perf annotate --stdio --group --data-type --skip-empty | head -24 Annotate type: 'pthread_mutex_t' in /usr/lib64/libc.so.6 (50 samples): event[0] = cpu_atom/mem-loads,ldlat=30/P event[1] = cpu_atom/mem-stores/P ============================================================================ Percent offset size field 100.00 100.00 0 40 pthread_mutex_t { 100.00 100.00 0 40 struct __pthread_mutex_s __data { 45.21 84.54 0 4 int __lock; 0.00 0.00 4 4 unsigned int __count; 0.00 1.83 8 4 int __owner; 5.19 10.65 12 4 unsigned int __nusers; 49.61 2.97 16 4 int __kind; 0.00 0.00 20 2 short int __spins; 0.00 0.00 22 2 short int __elision; 0.00 0.00 24 16 __pthread_list_t __list { 0.00 0.00 24 8 struct __pthread_internal_list* __prev; 0.00 0.00 32 8 struct __pthread_internal_list* __next; }; }; 0.00 0.00 0 0 char[] __size; 45.21 84.54 0 8 long int __align; }; Annotate type: 'pthread_mutexattr_t' in /usr/lib64/libc.so.6 (1 samples): root@number:~# Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240807061713.1642924-1-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-06 23:17:13 -07:00
samples = 0;
period = 0;
for (i = 0; i < member->size; i++) {
samples += h->addr[member->offset + i].nr_samples;
period += h->addr[member->offset + i].period;
}
perf annotate-data: Support --skip-empty option The --skip-empty option is to hide dummy events in a group. Like other output mode in 'perf report' and 'perf annotate', the data-type profiling output should support the option. Committer testing: With dummy: root@number:~# perf annotate --stdio --group --data-type --skip-empty | head -24 Annotate type: 'pthread_mutex_t' in /usr/lib64/libc.so.6 (50 samples): event[0] = cpu_atom/mem-loads,ldlat=30/P event[1] = cpu_atom/mem-stores/P event[2] = dummy:u ============================================================================ Percent offset size field 100.00 100.00 0.00 0 40 pthread_mutex_t { 100.00 100.00 0.00 0 40 struct __pthread_mutex_s __data { 45.21 84.54 0.00 0 4 int __lock; 0.00 0.00 0.00 4 4 unsigned int __count; 0.00 1.83 0.00 8 4 int __owner; 5.19 10.65 0.00 12 4 unsigned int __nusers; 49.61 2.97 0.00 16 4 int __kind; 0.00 0.00 0.00 20 2 short int __spins; 0.00 0.00 0.00 22 2 short int __elision; 0.00 0.00 0.00 24 16 __pthread_list_t __list { 0.00 0.00 0.00 24 8 struct __pthread_internal_list* __prev; 0.00 0.00 0.00 32 8 struct __pthread_internal_list* __next; }; }; 0.00 0.00 0.00 0 0 char[] __size; 45.21 84.54 0.00 0 8 long int __align; }; Skipping it: root@number:~# perf annotate --stdio --group --data-type --skip-empty | head -24 Annotate type: 'pthread_mutex_t' in /usr/lib64/libc.so.6 (50 samples): event[0] = cpu_atom/mem-loads,ldlat=30/P event[1] = cpu_atom/mem-stores/P ============================================================================ Percent offset size field 100.00 100.00 0 40 pthread_mutex_t { 100.00 100.00 0 40 struct __pthread_mutex_s __data { 45.21 84.54 0 4 int __lock; 0.00 0.00 4 4 unsigned int __count; 0.00 1.83 8 4 int __owner; 5.19 10.65 12 4 unsigned int __nusers; 49.61 2.97 16 4 int __kind; 0.00 0.00 20 2 short int __spins; 0.00 0.00 22 2 short int __elision; 0.00 0.00 24 16 __pthread_list_t __list { 0.00 0.00 24 8 struct __pthread_internal_list* __prev; 0.00 0.00 32 8 struct __pthread_internal_list* __next; }; }; 0.00 0.00 0 0 char[] __size; 45.21 84.54 0 8 long int __align; }; Annotate type: 'pthread_mutexattr_t' in /usr/lib64/libc.so.6 (1 samples): root@number:~# Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240807061713.1642924-1-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-08-06 23:17:13 -07:00
print_annotated_data_value(h, period, samples);
nr_events++;
}
printf(" %#10x %#10x %*s%s\t%s",
member->offset, member->size, indent, "", member->type_name,
member->var_name ?: "");
if (!list_empty(&member->children))
printf(" {\n");
list_for_each_entry(child, &member->children, node)
print_annotated_data_type(mem_type, child, evsel, indent + 4);
if (!list_empty(&member->children))
printf("%*s}", (width + 1) * nr_events + 24 + indent, "");
printf(";\n");
}
int hist_entry__annotate_data_tty(struct hist_entry *he, struct evsel *evsel)
{
print_annotated_data_header(he, evsel);
print_annotated_data_type(he->mem_type, &he->mem_type->self, evsel, 0);
printf("\n");
perf annotate-data: Add hist_entry__annotate_data_tui() Support data type profiling output on TUI. Testing from Arnaldo: First make sure that the debug information for your workload binaries in embedded in them by building it with '-g' or install the debuginfo packages, since our workload is 'find': root@number:~# type find find is hashed (/usr/bin/find) root@number:~# rpm -qf /usr/bin/find findutils-4.9.0-5.fc39.x86_64 root@number:~# dnf debuginfo-install findutils <SNIP> root@number:~# Then collect some data: root@number:~# echo 1 > /proc/sys/vm/drop_caches root@number:~# perf mem record find / > /dev/null [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.331 MB perf.data (3982 samples) ] root@number:~# Finally do data-type annotation with the following command, that will default, as 'perf report' to the --tui mode, with lines colored to highlight the hotspots, etc. root@number:~# perf annotate --data-type Annotate type: 'struct predicate' (58 samples) Percent Offset Size Field 100.00 0 312 struct predicate { 0.00 0 8 PRED_FUNC pred_func; 0.00 8 8 char* p_name; 0.00 16 4 enum predicate_type p_type; 0.00 20 4 enum predicate_precedence p_prec; 0.00 24 1 _Bool side_effects; 0.00 25 1 _Bool no_default_print; 0.00 26 1 _Bool need_stat; 0.00 27 1 _Bool need_type; 0.00 28 1 _Bool need_inum; 0.00 32 4 enum EvaluationCost p_cost; 0.00 36 4 float est_success_rate; 0.00 40 1 _Bool literal_control_chars; 0.00 41 1 _Bool artificial; 0.00 48 8 char* arg_text; <SNIP> Reviewed-by: Ian Rogers <irogers@google.com> Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240411033256.2099646-5-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2024-04-10 20:32:53 -07:00
/* move to the next entry */
return '>';
}