public-mirrors/linux
1
0
Fork 0
mirror of git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git synced 2025-08-05 16:54:27 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions
linux/fs/afs/vl_rotate.c
David Howells e2c2cb8ef0 afs: Simplify cell record handling 
Simplify afs_cell record handling to avoid very occasional races that cause
module removal to hang (it waits for all cell records to be removed).

There are two things that particularly contribute to the difficulty:
firstly, the code tries to pass a ref on the cell to the cell's maintenance
work item (which gets awkward if the work item is already queued); and,
secondly, there's an overall cell manager that tries to use just one timer
for the entire cell collection (to avoid having loads of timers).  However,
both of these are probably unnecessarily restrictive.

To simplify this, the following changes are made:

 (1) The cell record collection manager is removed.  Each cell record
     manages itself individually.

 (2) Each afs_cell is given a second work item (cell->destroyer) that is
     queued when its refcount reaches zero.  This is not done in the
     context of the putting thread as it might be in an inconvenient place
     to sleep.

 (3) Each afs_cell is given its own timer.  The timer is used to expire the
     cell record after a period of unuse if not otherwise pinned and can
     also be used for other maintenance tasks if necessary (of which there
     are currently none as DNS refresh is triggered by filesystem
     operations).

 (4) The afs_cell manager work item (cell->manager) is no longer given a
     ref on the cell when queued; rather, the manager must be deleted.
     This does away with the need to deal with the consequences of losing a
     race to queue cell->manager.  Clean up of extra queuing is deferred to
     the destroyer.

 (5) The cell destroyer work item makes sure the cell timer is removed and
     that the normal cell work is cancelled before farming the actual
     destruction off to RCU.

 (6) When a network namespace is destroyed or the kafs module is unloaded,
     it's now a simple matter of marking the namespace as dead then just
     waking up all the cell work items.  They will then remove and destroy
     themselves once all remaining activity counts and/or a ref counts are
     dropped.  This makes sure that all server records are dropped first.

 (7) The cell record state set is reduced to just four states: SETTING_UP,
     ACTIVE, REMOVING and DEAD.  The record persists in the active state
     even when it's not being used until the time comes to remove it rather
     than downgrading it to an inactive state from whence it can be
     restored.

     This means that the cell still appears in /proc and /afs when not in
     use until it switches to the REMOVING state - at which point it is
     removed.

     Note that the REMOVING state is included so that someone wanting to
     resurrect the cell record is forced to wait whilst the cell is torn
     down in that state.  Once it's in the DEAD state, it has been removed
     from net->cells tree and is no longer findable and can be replaced.

Signed-off-by: David Howells <dhowells@redhat.com>
cc: Marc Dionne <marc.dionne@auristor.com>
cc: linux-afs@lists.infradead.org
cc: linux-fsdevel@vger.kernel.org
Link: https://lore.kernel.org/r/20250224234154.2014840-16-dhowells@redhat.com/ # v1
Link: https://lore.kernel.org/r/20250310094206.801057-12-dhowells@redhat.com/ # v4
2025-03-10 09:47:15 +00:00

399 lines
11 KiB
C
Raw Permalink Blame History

// SPDX-License-Identifier: GPL-2.0-or-later
/* Handle vlserver selection and rotation.
*
* Copyright (C) 2018 Red Hat, Inc. All Rights Reserved.
* Written by David Howells (dhowells@redhat.com)
*/
#include <linux/kernel.h>
#include <linux/sched.h>
#include <linux/sched/signal.h>
#include "internal.h"
#include "afs_vl.h"
/*
* Begin an operation on a volume location server.
*/
bool afs_begin_vlserver_operation(struct afs_vl_cursor *vc, struct afs_cell *cell,
struct key *key)
{
static atomic_t debug_ids;
memset(vc, 0, sizeof(*vc));
vc->cell = cell;
vc->key = key;
vc->cumul_error.error = -EDESTADDRREQ;
vc->nr_iterations = -1;
if (signal_pending(current)) {
vc->cumul_error.error = -EINTR;
vc->flags |= AFS_VL_CURSOR_STOP;
return false;
}
vc->debug_id = atomic_inc_return(&debug_ids);
return true;
}
/*
* Begin iteration through a server list, starting with the last used server if
* possible, or the last recorded good server if not.
*/
static bool afs_start_vl_iteration(struct afs_vl_cursor *vc)
{
struct afs_cell *cell = vc->cell;
unsigned int dns_lookup_count;
if (cell->dns_source == DNS_RECORD_UNAVAILABLE ||
cell->dns_expiry <= ktime_get_real_seconds()) {
dns_lookup_count = smp_load_acquire(&cell->dns_lookup_count);
set_bit(AFS_CELL_FL_DO_LOOKUP, &cell->flags);
afs_queue_cell(cell, afs_cell_trace_queue_dns);
if (cell->dns_source == DNS_RECORD_UNAVAILABLE) {
if (wait_var_event_interruptible(
&cell->dns_lookup_count,
smp_load_acquire(&cell->dns_lookup_count)
!= dns_lookup_count) < 0) {
vc->cumul_error.error = -ERESTARTSYS;
return false;
}
}
/* Status load is ordered after lookup counter load */
if (cell->dns_status == DNS_LOOKUP_GOT_NOT_FOUND) {
pr_warn("No record of cell %s\n", cell->name);
vc->cumul_error.error = -ENOENT;
return false;
}
if (cell->dns_source == DNS_RECORD_UNAVAILABLE) {
vc->cumul_error.error = -EDESTADDRREQ;
return false;
}
}
read_lock(&cell->vl_servers_lock);
vc->server_list = afs_get_vlserverlist(
rcu_dereference_protected(cell->vl_servers,
lockdep_is_held(&cell->vl_servers_lock)));
read_unlock(&cell->vl_servers_lock);
if (!vc->server_list->nr_servers)
return false;
vc->untried_servers = (1UL << vc->server_list->nr_servers) - 1;
vc->server_index = -1;
return true;
}
/*
* Select the vlserver to use. May be called multiple times to rotate
* through the vlservers.
*/
bool afs_select_vlserver(struct afs_vl_cursor *vc)
{
struct afs_addr_list *alist = vc->alist;
struct afs_vlserver *vlserver;
unsigned long set, failed;
unsigned int rtt;
s32 abort_code = vc->call_abort_code;
int error = vc->call_error, i;
vc->nr_iterations++;
_enter("VC=%x+%x,%d{%lx},%d{%lx},%d,%d",
vc->debug_id, vc->nr_iterations, vc->server_index, vc->untried_servers,
vc->addr_index, vc->addr_tried,
error, abort_code);
if (vc->flags & AFS_VL_CURSOR_STOP) {
_leave(" = f [stopped]");
return false;
}
if (vc->nr_iterations == 0)
goto start;
WRITE_ONCE(alist->addrs[vc->addr_index].last_error, error);
/* Evaluate the result of the previous operation, if there was one. */
switch (error) {
default:
case 0:
/* Success or local failure. Stop. */
vc->cumul_error.error = error;
vc->flags |= AFS_VL_CURSOR_STOP;
_leave(" = f [okay/local %d]", vc->cumul_error.error);
return false;
case -ECONNABORTED:
/* The far side rejected the operation on some grounds. This
* might involve the server being busy or the volume having been moved.
*/
switch (abort_code) {
case AFSVL_IO:
case AFSVL_BADVOLOPER:
case AFSVL_NOMEM:
/* The server went weird. */
afs_prioritise_error(&vc->cumul_error, -EREMOTEIO, abort_code);
//write_lock(&vc->cell->vl_servers_lock);
//vc->server_list->weird_mask |= 1 << vc->server_index;
//write_unlock(&vc->cell->vl_servers_lock);
goto next_server;
default:
afs_prioritise_error(&vc->cumul_error, error, abort_code);
goto failed;
}
case -ERFKILL:
case -EADDRNOTAVAIL:
case -ENETUNREACH:
case -EHOSTUNREACH:
case -EHOSTDOWN:
case -ECONNREFUSED:
case -ETIMEDOUT:
case -ETIME:
_debug("no conn %d", error);
afs_prioritise_error(&vc->cumul_error, error, 0);
goto iterate_address;
case -ECONNRESET:
_debug("call reset");
afs_prioritise_error(&vc->cumul_error, error, 0);
vc->flags |= AFS_VL_CURSOR_RETRY;
goto next_server;
case -EOPNOTSUPP:
_debug("notsupp");
goto next_server;
}
restart_from_beginning:
_debug("restart");
if (vc->call_responded &&
vc->addr_index != vc->alist->preferred &&
test_bit(alist->preferred, &vc->addr_tried))
WRITE_ONCE(alist->preferred, vc->addr_index);
afs_put_addrlist(alist, afs_alist_trace_put_vlrotate_restart);
alist = vc->alist = NULL;
afs_put_vlserverlist(vc->cell->net, vc->server_list);
vc->server_list = NULL;
if (vc->flags & AFS_VL_CURSOR_RETRIED)
goto failed;
vc->flags |= AFS_VL_CURSOR_RETRIED;
start:
_debug("start");
ASSERTCMP(alist, ==, NULL);
if (!afs_start_vl_iteration(vc))
goto failed;
error = afs_send_vl_probes(vc->cell->net, vc->key, vc->server_list);
if (error < 0) {
afs_prioritise_error(&vc->cumul_error, error, 0);
goto failed;
}
pick_server:
_debug("pick [%lx]", vc->untried_servers);
ASSERTCMP(alist, ==, NULL);
error = afs_wait_for_vl_probes(vc->server_list, vc->untried_servers);
if (error < 0) {
afs_prioritise_error(&vc->cumul_error, error, 0);
goto failed;
}
/* Pick the untried server with the lowest RTT. */
vc->server_index = vc->server_list->preferred;
if (test_bit(vc->server_index, &vc->untried_servers))
goto selected_server;
vc->server_index = -1;
rtt = UINT_MAX;
for (i = 0; i < vc->server_list->nr_servers; i++) {
struct afs_vlserver *s = vc->server_list->servers[i].server;
if (!test_bit(i, &vc->untried_servers) ||
!test_bit(AFS_VLSERVER_FL_RESPONDING, &s->flags))
continue;
if (s->probe.rtt <= rtt) {
vc->server_index = i;
rtt = s->probe.rtt;
}
}
if (vc->server_index == -1)
goto no_more_servers;
selected_server:
_debug("use %d", vc->server_index);
__clear_bit(vc->server_index, &vc->untried_servers);
/* We're starting on a different vlserver from the list. We need to
* check it, find its address list and probe its capabilities before we
* use it.
*/
vlserver = vc->server_list->servers[vc->server_index].server;
vc->server = vlserver;
_debug("USING VLSERVER: %s", vlserver->name);
read_lock(&vlserver->lock);
alist = rcu_dereference_protected(vlserver->addresses,
lockdep_is_held(&vlserver->lock));
vc->alist = afs_get_addrlist(alist, afs_alist_trace_get_vlrotate_set);
read_unlock(&vlserver->lock);
vc->addr_tried = 0;
vc->addr_index = -1;
iterate_address:
/* Iterate over the current server's address list to try and find an
* address on which it will respond to us.
*/
set = READ_ONCE(alist->responded);
failed = READ_ONCE(alist->probe_failed);
vc->addr_index = READ_ONCE(alist->preferred);
_debug("%lx-%lx-%lx,%d", set, failed, vc->addr_tried, vc->addr_index);
set &= ~(failed | vc->addr_tried);
if (!set)
goto next_server;
if (!test_bit(vc->addr_index, &set))
vc->addr_index = __ffs(set);
set_bit(vc->addr_index, &vc->addr_tried);
vc->alist = alist;
_debug("VL address %d/%d", vc->addr_index, alist->nr_addrs);
vc->call_responded = false;
_leave(" = t %pISpc", rxrpc_kernel_remote_addr(alist->addrs[vc->addr_index].peer));
return true;
next_server:
_debug("next");
ASSERT(alist);
if (vc->call_responded &&
vc->addr_index != alist->preferred &&
test_bit(alist->preferred, &vc->addr_tried))
WRITE_ONCE(alist->preferred, vc->addr_index);
afs_put_addrlist(alist, afs_alist_trace_put_vlrotate_next);
alist = vc->alist = NULL;
goto pick_server;
no_more_servers:
/* That's all the servers poked to no good effect. Try again if some
* of them were busy.
*/
if (vc->flags & AFS_VL_CURSOR_RETRY)
goto restart_from_beginning;
for (i = 0; i < vc->server_list->nr_servers; i++) {
struct afs_vlserver *s = vc->server_list->servers[i].server;
if (test_bit(AFS_VLSERVER_FL_RESPONDING, &s->flags))
vc->cumul_error.responded = true;
afs_prioritise_error(&vc->cumul_error, READ_ONCE(s->probe.error),
s->probe.abort_code);
}
failed:
if (alist) {
if (vc->call_responded &&
vc->addr_index != alist->preferred &&
test_bit(alist->preferred, &vc->addr_tried))
WRITE_ONCE(alist->preferred, vc->addr_index);
afs_put_addrlist(alist, afs_alist_trace_put_vlrotate_fail);
alist = vc->alist = NULL;
}
vc->flags |= AFS_VL_CURSOR_STOP;
_leave(" = f [failed %d]", vc->cumul_error.error);
return false;
}
/*
* Dump cursor state in the case of the error being EDESTADDRREQ.
*/
static void afs_vl_dump_edestaddrreq(const struct afs_vl_cursor *vc)
{
struct afs_cell *cell = vc->cell;
static int count;
int i;
if (!IS_ENABLED(CONFIG_AFS_DEBUG_CURSOR) || count > 3)
return;
count++;
rcu_read_lock();
pr_notice("EDESTADDR occurred\n");
pr_notice("CELL: %s err=%d\n", cell->name, cell->error);
pr_notice("DNS: src=%u st=%u lc=%x\n",
cell->dns_source, cell->dns_status, cell->dns_lookup_count);
pr_notice("VC: ut=%lx ix=%u ni=%hu fl=%hx err=%hd\n",
vc->untried_servers, vc->server_index, vc->nr_iterations,
vc->flags, vc->cumul_error.error);
pr_notice("VC: call er=%d ac=%d r=%u\n",
vc->call_error, vc->call_abort_code, vc->call_responded);
if (vc->server_list) {
const struct afs_vlserver_list *sl = vc->server_list;
pr_notice("VC: SL nr=%u ix=%u\n",
sl->nr_servers, sl->index);
for (i = 0; i < sl->nr_servers; i++) {
const struct afs_vlserver *s = sl->servers[i].server;
pr_notice("VC: server %s+%hu fl=%lx E=%hd\n",
s->name, s->port, s->flags, s->probe.error);
if (s->addresses) {
const struct afs_addr_list *a =
rcu_dereference(s->addresses);
pr_notice("VC: - nr=%u/%u/%u pf=%u\n",
a->nr_ipv4, a->nr_addrs, a->max_addrs,
a->preferred);
pr_notice("VC: - R=%lx F=%lx\n",
a->responded, a->probe_failed);
if (a == vc->alist)
pr_notice("VC: - current\n");
}
}
}
pr_notice("AC: t=%lx ax=%u\n", vc->addr_tried, vc->addr_index);
rcu_read_unlock();
}
/*
* Tidy up a volume location server cursor and unlock the vnode.
*/
int afs_end_vlserver_operation(struct afs_vl_cursor *vc)
{
struct afs_net *net = vc->cell->net;
_enter("VC=%x+%x", vc->debug_id, vc->nr_iterations);
switch (vc->cumul_error.error) {
case -EDESTADDRREQ:
case -EADDRNOTAVAIL:
case -ENETUNREACH:
case -EHOSTUNREACH:
afs_vl_dump_edestaddrreq(vc);
break;
}
if (vc->alist) {
if (vc->call_responded &&
vc->addr_index != vc->alist->preferred &&
test_bit(vc->alist->preferred, &vc->addr_tried))
WRITE_ONCE(vc->alist->preferred, vc->addr_index);
afs_put_addrlist(vc->alist, afs_alist_trace_put_vlrotate_end);
vc->alist = NULL;
}
afs_put_vlserverlist(net, vc->server_list);
return vc->cumul_error.error;
}
Reference in a new issue View git blame Copy permalink