linux/tools/lib/perf/include/internal/mmap.h

60 lines
1.5 KiB
C
Raw Normal View History

/* SPDX-License-Identifier: GPL-2.0 */
#ifndef __LIBPERF_INTERNAL_MMAP_H
#define __LIBPERF_INTERNAL_MMAP_H
#include <linux/compiler.h>
#include <linux/refcount.h>
#include <linux/types.h>
#include <stdbool.h>
perf cpumap: Give CPUs their own type A common problem is confusing CPU map indices with the CPU, by wrapping the CPU with a struct then this is avoided. This approach is similar to atomic_t. Committer notes: To make it build with BUILD_BPF_SKEL=1 these files needed the conversions to 'struct perf_cpu' usage: tools/perf/util/bpf_counter.c tools/perf/util/bpf_counter_cgroup.c tools/perf/util/bpf_ftrace.c Also perf_env__get_cpu() was removed back in "perf cpumap: Switch cpu_map__build_map to cpu function". Additionally these needed to be fixed for the ARM builds to complete: tools/perf/arch/arm/util/cs-etm.c tools/perf/arch/arm64/util/pmu.c Suggested-by: John Garry <john.garry@huawei.com> Signed-off-by: Ian Rogers <irogers@google.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Andi Kleen <ak@linux.intel.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: James Clark <james.clark@arm.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Kajol Jain <kjain@linux.ibm.com> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Leo Yan <leo.yan@linaro.org> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Mathieu Poirier <mathieu.poirier@linaro.org> Cc: Mike Leach <mike.leach@linaro.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Paul Clarke <pc@us.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Riccardo Mancini <rickyman7@gmail.com> Cc: Stephane Eranian <eranian@google.com> Cc: Suzuki Poulouse <suzuki.poulose@arm.com> Cc: Vineet Singh <vineet.singh@intel.com> Cc: coresight@lists.linaro.org Cc: linux-arm-kernel@lists.infradead.org Cc: zhengjun.xing@intel.com Link: https://lore.kernel.org/r/20220105061351.120843-49-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2022-01-04 22:13:51 -08:00
#include <internal/cpumap.h>
/* perf sample has 16 bits size limit */
#define PERF_SAMPLE_MAX_SIZE (1 << 16)
struct perf_mmap;
libperf: Add support for user space counter access x86 and arm64 can both support direct access of event counters in userspace. The access sequence is less than trivial and currently exists in perf test code (tools/perf/arch/x86/tests/rdpmc.c) with copies in projects such as PAPI and libpfm4. In order to support userspace access, an event must be mmapped first with perf_evsel__mmap(). Then subsequent calls to perf_evsel__read() will use the fast path (assuming the arch supports it). Committer notes: Added a '__maybe_unused' attribute to the read_perf_counter() argument to fix the build on arches other than x86_64 and arm. Committer testing: Building and running the libperf tests in verbose mode (V=1) now shows those "loop = N, count = N" extra lines, testing user space counter access. # make V=1 -C tools/lib/perf tests make: Entering directory '/home/acme/git/perf/tools/lib/perf' make -f /home/acme/git/perf/tools/build/Makefile.build dir=. obj=libperf make -C /home/acme/git/perf/tools/lib/api/ O= libapi.a make -f /home/acme/git/perf/tools/build/Makefile.build dir=./fd obj=libapi make -f /home/acme/git/perf/tools/build/Makefile.build dir=./fs obj=libapi make -C tests gcc -I/home/acme/git/perf/tools/lib/perf/include -I/home/acme/git/perf/tools/include -I/home/acme/git/perf/tools/lib -g -Wall -o test-cpumap-a test-cpumap.c ../libperf.a /home/acme/git/perf/tools/lib/api/libapi.a gcc -I/home/acme/git/perf/tools/lib/perf/include -I/home/acme/git/perf/tools/include -I/home/acme/git/perf/tools/lib -g -Wall -o test-threadmap-a test-threadmap.c ../libperf.a /home/acme/git/perf/tools/lib/api/libapi.a gcc -I/home/acme/git/perf/tools/lib/perf/include -I/home/acme/git/perf/tools/include -I/home/acme/git/perf/tools/lib -g -Wall -o test-evlist-a test-evlist.c ../libperf.a /home/acme/git/perf/tools/lib/api/libapi.a gcc -I/home/acme/git/perf/tools/lib/perf/include -I/home/acme/git/perf/tools/include -I/home/acme/git/perf/tools/lib -g -Wall -o test-evsel-a test-evsel.c ../libperf.a /home/acme/git/perf/tools/lib/api/libapi.a gcc -I/home/acme/git/perf/tools/lib/perf/include -I/home/acme/git/perf/tools/include -I/home/acme/git/perf/tools/lib -g -Wall -L.. -o test-cpumap-so test-cpumap.c /home/acme/git/perf/tools/lib/api/libapi.a -lperf gcc -I/home/acme/git/perf/tools/lib/perf/include -I/home/acme/git/perf/tools/include -I/home/acme/git/perf/tools/lib -g -Wall -L.. -o test-threadmap-so test-threadmap.c /home/acme/git/perf/tools/lib/api/libapi.a -lperf gcc -I/home/acme/git/perf/tools/lib/perf/include -I/home/acme/git/perf/tools/include -I/home/acme/git/perf/tools/lib -g -Wall -L.. -o test-evlist-so test-evlist.c /home/acme/git/perf/tools/lib/api/libapi.a -lperf gcc -I/home/acme/git/perf/tools/lib/perf/include -I/home/acme/git/perf/tools/include -I/home/acme/git/perf/tools/lib -g -Wall -L.. -o test-evsel-so test-evsel.c /home/acme/git/perf/tools/lib/api/libapi.a -lperf make -C tests run running static: - running test-cpumap.c...OK - running test-threadmap.c...OK - running test-evlist.c...OK - running test-evsel.c... loop = 65536, count = 333926 loop = 131072, count = 655781 loop = 262144, count = 1311141 loop = 524288, count = 2630126 loop = 1048576, count = 5256955 loop = 65536, count = 524594 loop = 131072, count = 1058916 loop = 262144, count = 2097458 loop = 524288, count = 4205429 loop = 1048576, count = 8406606 OK running dynamic: - running test-cpumap.c...OK - running test-threadmap.c...OK - running test-evlist.c...OK - running test-evsel.c... loop = 65536, count = 328102 loop = 131072, count = 655782 loop = 262144, count = 1317494 loop = 524288, count = 2627851 loop = 1048576, count = 5255187 loop = 65536, count = 524601 loop = 131072, count = 1048923 loop = 262144, count = 2107917 loop = 524288, count = 4194606 loop = 1048576, count = 8409322 OK make: Leaving directory '/home/acme/git/perf/tools/lib/perf' # Signed-off-by: Rob Herring <robh@kernel.org> Acked-by: Jiri Olsa <jolsa@redhat.com> Acked-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Itaru Kitayama <itaru.kitayama@gmail.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Will Deacon <will@kernel.org> Link: http://lore.kernel.org/lkml/20210414155412.3697605-4-robh@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-04-14 11:07:39 -05:00
struct perf_counts_values;
typedef void (*libperf_unmap_cb_t)(struct perf_mmap *map);
/**
* struct perf_mmap - perf's ring buffer mmap details
*
* @refcnt - e.g. code using PERF_EVENT_IOC_SET_OUTPUT to share this
*/
struct perf_mmap {
void *base;
int mask;
int fd;
perf cpumap: Give CPUs their own type A common problem is confusing CPU map indices with the CPU, by wrapping the CPU with a struct then this is avoided. This approach is similar to atomic_t. Committer notes: To make it build with BUILD_BPF_SKEL=1 these files needed the conversions to 'struct perf_cpu' usage: tools/perf/util/bpf_counter.c tools/perf/util/bpf_counter_cgroup.c tools/perf/util/bpf_ftrace.c Also perf_env__get_cpu() was removed back in "perf cpumap: Switch cpu_map__build_map to cpu function". Additionally these needed to be fixed for the ARM builds to complete: tools/perf/arch/arm/util/cs-etm.c tools/perf/arch/arm64/util/pmu.c Suggested-by: John Garry <john.garry@huawei.com> Signed-off-by: Ian Rogers <irogers@google.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Andi Kleen <ak@linux.intel.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: James Clark <james.clark@arm.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Kajol Jain <kjain@linux.ibm.com> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Leo Yan <leo.yan@linaro.org> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Mathieu Poirier <mathieu.poirier@linaro.org> Cc: Mike Leach <mike.leach@linaro.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Paul Clarke <pc@us.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Riccardo Mancini <rickyman7@gmail.com> Cc: Stephane Eranian <eranian@google.com> Cc: Suzuki Poulouse <suzuki.poulose@arm.com> Cc: Vineet Singh <vineet.singh@intel.com> Cc: coresight@lists.linaro.org Cc: linux-arm-kernel@lists.infradead.org Cc: zhengjun.xing@intel.com Link: https://lore.kernel.org/r/20220105061351.120843-49-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2022-01-04 22:13:51 -08:00
struct perf_cpu cpu;
refcount_t refcnt;
u64 prev;
u64 start;
u64 end;
bool overwrite;
u64 flush;
libperf_unmap_cb_t unmap_cb;
char event_copy[PERF_SAMPLE_MAX_SIZE] __aligned(8);
struct perf_mmap *next;
};
struct perf_mmap_param {
int prot;
int mask;
};
size_t perf_mmap__mmap_len(struct perf_mmap *map);
void perf_mmap__init(struct perf_mmap *map, struct perf_mmap *prev,
bool overwrite, libperf_unmap_cb_t unmap_cb);
int perf_mmap__mmap(struct perf_mmap *map, struct perf_mmap_param *mp,
perf cpumap: Give CPUs their own type A common problem is confusing CPU map indices with the CPU, by wrapping the CPU with a struct then this is avoided. This approach is similar to atomic_t. Committer notes: To make it build with BUILD_BPF_SKEL=1 these files needed the conversions to 'struct perf_cpu' usage: tools/perf/util/bpf_counter.c tools/perf/util/bpf_counter_cgroup.c tools/perf/util/bpf_ftrace.c Also perf_env__get_cpu() was removed back in "perf cpumap: Switch cpu_map__build_map to cpu function". Additionally these needed to be fixed for the ARM builds to complete: tools/perf/arch/arm/util/cs-etm.c tools/perf/arch/arm64/util/pmu.c Suggested-by: John Garry <john.garry@huawei.com> Signed-off-by: Ian Rogers <irogers@google.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Andi Kleen <ak@linux.intel.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: James Clark <james.clark@arm.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Kajol Jain <kjain@linux.ibm.com> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Leo Yan <leo.yan@linaro.org> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Mathieu Poirier <mathieu.poirier@linaro.org> Cc: Mike Leach <mike.leach@linaro.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Paul Clarke <pc@us.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Riccardo Mancini <rickyman7@gmail.com> Cc: Stephane Eranian <eranian@google.com> Cc: Suzuki Poulouse <suzuki.poulose@arm.com> Cc: Vineet Singh <vineet.singh@intel.com> Cc: coresight@lists.linaro.org Cc: linux-arm-kernel@lists.infradead.org Cc: zhengjun.xing@intel.com Link: https://lore.kernel.org/r/20220105061351.120843-49-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2022-01-04 22:13:51 -08:00
int fd, struct perf_cpu cpu);
void perf_mmap__munmap(struct perf_mmap *map);
void perf_mmap__get(struct perf_mmap *map);
void perf_mmap__put(struct perf_mmap *map);
u64 perf_mmap__read_head(struct perf_mmap *map);
libperf: Add support for user space counter access x86 and arm64 can both support direct access of event counters in userspace. The access sequence is less than trivial and currently exists in perf test code (tools/perf/arch/x86/tests/rdpmc.c) with copies in projects such as PAPI and libpfm4. In order to support userspace access, an event must be mmapped first with perf_evsel__mmap(). Then subsequent calls to perf_evsel__read() will use the fast path (assuming the arch supports it). Committer notes: Added a '__maybe_unused' attribute to the read_perf_counter() argument to fix the build on arches other than x86_64 and arm. Committer testing: Building and running the libperf tests in verbose mode (V=1) now shows those "loop = N, count = N" extra lines, testing user space counter access. # make V=1 -C tools/lib/perf tests make: Entering directory '/home/acme/git/perf/tools/lib/perf' make -f /home/acme/git/perf/tools/build/Makefile.build dir=. obj=libperf make -C /home/acme/git/perf/tools/lib/api/ O= libapi.a make -f /home/acme/git/perf/tools/build/Makefile.build dir=./fd obj=libapi make -f /home/acme/git/perf/tools/build/Makefile.build dir=./fs obj=libapi make -C tests gcc -I/home/acme/git/perf/tools/lib/perf/include -I/home/acme/git/perf/tools/include -I/home/acme/git/perf/tools/lib -g -Wall -o test-cpumap-a test-cpumap.c ../libperf.a /home/acme/git/perf/tools/lib/api/libapi.a gcc -I/home/acme/git/perf/tools/lib/perf/include -I/home/acme/git/perf/tools/include -I/home/acme/git/perf/tools/lib -g -Wall -o test-threadmap-a test-threadmap.c ../libperf.a /home/acme/git/perf/tools/lib/api/libapi.a gcc -I/home/acme/git/perf/tools/lib/perf/include -I/home/acme/git/perf/tools/include -I/home/acme/git/perf/tools/lib -g -Wall -o test-evlist-a test-evlist.c ../libperf.a /home/acme/git/perf/tools/lib/api/libapi.a gcc -I/home/acme/git/perf/tools/lib/perf/include -I/home/acme/git/perf/tools/include -I/home/acme/git/perf/tools/lib -g -Wall -o test-evsel-a test-evsel.c ../libperf.a /home/acme/git/perf/tools/lib/api/libapi.a gcc -I/home/acme/git/perf/tools/lib/perf/include -I/home/acme/git/perf/tools/include -I/home/acme/git/perf/tools/lib -g -Wall -L.. -o test-cpumap-so test-cpumap.c /home/acme/git/perf/tools/lib/api/libapi.a -lperf gcc -I/home/acme/git/perf/tools/lib/perf/include -I/home/acme/git/perf/tools/include -I/home/acme/git/perf/tools/lib -g -Wall -L.. -o test-threadmap-so test-threadmap.c /home/acme/git/perf/tools/lib/api/libapi.a -lperf gcc -I/home/acme/git/perf/tools/lib/perf/include -I/home/acme/git/perf/tools/include -I/home/acme/git/perf/tools/lib -g -Wall -L.. -o test-evlist-so test-evlist.c /home/acme/git/perf/tools/lib/api/libapi.a -lperf gcc -I/home/acme/git/perf/tools/lib/perf/include -I/home/acme/git/perf/tools/include -I/home/acme/git/perf/tools/lib -g -Wall -L.. -o test-evsel-so test-evsel.c /home/acme/git/perf/tools/lib/api/libapi.a -lperf make -C tests run running static: - running test-cpumap.c...OK - running test-threadmap.c...OK - running test-evlist.c...OK - running test-evsel.c... loop = 65536, count = 333926 loop = 131072, count = 655781 loop = 262144, count = 1311141 loop = 524288, count = 2630126 loop = 1048576, count = 5256955 loop = 65536, count = 524594 loop = 131072, count = 1058916 loop = 262144, count = 2097458 loop = 524288, count = 4205429 loop = 1048576, count = 8406606 OK running dynamic: - running test-cpumap.c...OK - running test-threadmap.c...OK - running test-evlist.c...OK - running test-evsel.c... loop = 65536, count = 328102 loop = 131072, count = 655782 loop = 262144, count = 1317494 loop = 524288, count = 2627851 loop = 1048576, count = 5255187 loop = 65536, count = 524601 loop = 131072, count = 1048923 loop = 262144, count = 2107917 loop = 524288, count = 4194606 loop = 1048576, count = 8409322 OK make: Leaving directory '/home/acme/git/perf/tools/lib/perf' # Signed-off-by: Rob Herring <robh@kernel.org> Acked-by: Jiri Olsa <jolsa@redhat.com> Acked-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Itaru Kitayama <itaru.kitayama@gmail.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Will Deacon <will@kernel.org> Link: http://lore.kernel.org/lkml/20210414155412.3697605-4-robh@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-04-14 11:07:39 -05:00
int perf_mmap__read_self(struct perf_mmap *map, struct perf_counts_values *count);
#endif /* __LIBPERF_INTERNAL_MMAP_H */