+++ /dev/null
-/*
- * GPL HEADER START
- *
- * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License version 2 only,
- * as published by the Free Software Foundation.
- *
- * This program is distributed in the hope that it will be useful, but
- * WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
- * General Public License version 2 for more details (a copy is included
- * in the LICENSE file that accompanied this code).
- *
- * You should have received a copy of the GNU General Public License
- * version 2 along with this program; If not, see
- * http://www.gnu.org/licenses/gpl-2.0.html
- *
- * GPL HEADER END
- */
-/*
- * Copyright (c) 2008, 2010, Oracle and/or its affiliates. All rights reserved.
- * Use is subject to license terms.
- *
- * Copyright (c) 2012, 2017, Intel Corporation.
- */
-/*
- * This file is part of Lustre, http://www.lustre.org/
- * Lustre is a trademark of Sun Microsystems, Inc.
- *
- * libcfs/libcfs/watchdog.c
- *
- * Author: Jacob Berkman <jacob@clusterfs.com>
- */
-
-#define DEBUG_SUBSYSTEM S_LNET
-
-#include <linux/kthread.h>
-#include <libcfs/libcfs.h>
-#include "tracefile.h"
-
-#ifndef WITH_WATCHDOG
-#define WITH_WATCHDOG
-#endif
-
-struct lc_watchdog {
- spinlock_t lcw_lock; /* check or change lcw_list */
- int lcw_refcount; /* must hold lcw_pending_timers_lock */
- struct timer_list lcw_timer; /* kernel timer */
- struct list_head lcw_list; /* chain on pending list */
- ktime_t lcw_last_touched;/* last touched stamp */
- struct task_struct *lcw_task; /* owner task */
- void (*lcw_callback)(pid_t, void *);
- void *lcw_data;
-
- pid_t lcw_pid;
-
- enum {
- LC_WATCHDOG_DISABLED,
- LC_WATCHDOG_ENABLED,
- LC_WATCHDOG_EXPIRED
- } lcw_state;
-};
-
-#ifdef WITH_WATCHDOG
-/*
- * The dispatcher will complete lcw_start_completion when it starts,
- * and lcw_stop_completion when it exits.
- * Wake lcw_event_waitq to signal timer callback dispatches.
- */
-static struct completion lcw_start_completion;
-static struct completion lcw_stop_completion;
-static wait_queue_head_t lcw_event_waitq;
-
-/*
- * Set this and wake lcw_event_waitq to stop the dispatcher.
- */
-enum {
- LCW_FLAG_STOP = 0
-};
-static unsigned long lcw_flags = 0;
-
-/*
- * Number of outstanding watchdogs.
- * When it hits 1, we start the dispatcher.
- * When it hits 0, we stop the dispatcher.
- */
-static __u32 lcw_refcount = 0;
-static DEFINE_MUTEX(lcw_refcount_mutex);
-
-/*
- * List of timers that have fired that need their callbacks run by the
- * dispatcher.
- */
-/* BH lock! */
-static DEFINE_SPINLOCK(lcw_pending_timers_lock);
-static struct list_head lcw_pending_timers = LIST_HEAD_INIT(lcw_pending_timers);
-
-/* Last time a watchdog expired */
-static time64_t lcw_last_watchdog_time;
-static int lcw_recent_watchdog_count;
-
-static void
-lcw_dump(struct lc_watchdog *lcw)
-{
- ENTRY;
- rcu_read_lock();
- if (lcw->lcw_task == NULL) {
- LCONSOLE_WARN("Process %d was not found in the task "
- "list; watchdog callback may be incomplete\n",
- (int)lcw->lcw_pid);
- } else {
- libcfs_debug_dumpstack(lcw->lcw_task);
- }
-
- rcu_read_unlock();
- EXIT;
-}
-
-static void lcw_cb(cfs_timer_cb_arg_t data)
-{
- struct lc_watchdog *lcw = cfs_from_timer(lcw, data, lcw_timer);
- ENTRY;
-
- if (lcw->lcw_state != LC_WATCHDOG_ENABLED) {
- EXIT;
- return;
- }
-
- lcw->lcw_state = LC_WATCHDOG_EXPIRED;
-
- spin_lock_bh(&lcw->lcw_lock);
- LASSERT(list_empty(&lcw->lcw_list));
-
- spin_lock_bh(&lcw_pending_timers_lock);
- lcw->lcw_refcount++; /* +1 for pending list */
- list_add(&lcw->lcw_list, &lcw_pending_timers);
- wake_up(&lcw_event_waitq);
-
- spin_unlock_bh(&lcw_pending_timers_lock);
- spin_unlock_bh(&lcw->lcw_lock);
- EXIT;
-}
-
-static int is_watchdog_fired(void)
-{
- int rc;
-
- if (test_bit(LCW_FLAG_STOP, &lcw_flags))
- return 1;
-
- spin_lock_bh(&lcw_pending_timers_lock);
- rc = !list_empty(&lcw_pending_timers);
- spin_unlock_bh(&lcw_pending_timers_lock);
- return rc;
-}
-
-static void lcw_dump_stack(struct lc_watchdog *lcw)
-{
- time64_t current_time = ktime_get_seconds();
- struct timespec64 timediff;
- time64_t delta_time;
-
- timediff = ktime_to_timespec64(ktime_sub(ktime_get(),
- lcw->lcw_last_touched));
-
- /* LU-9235: Don't dump stack if the thread is just touched. */
- if (timediff.tv_sec == 0)
- return;
-
- /*
- * Check to see if we should throttle the watchdog timer to avoid
- * too many dumps going to the console thus triggering an NMI.
- */
- delta_time = current_time - lcw_last_watchdog_time;
- if (delta_time < libcfs_watchdog_ratelimit &&
- lcw_recent_watchdog_count > 3) {
- LCONSOLE_WARN("Service thread pid %u was inactive for %lu.%.02lus. Watchdog stack traces are limited to 3 per %d seconds, skipping this one.\n",
- (int)lcw->lcw_pid,
- timediff.tv_sec,
- timediff.tv_nsec / (NSEC_PER_SEC / 100),
- libcfs_watchdog_ratelimit);
- } else {
- if (delta_time < libcfs_watchdog_ratelimit) {
- lcw_recent_watchdog_count++;
- } else {
- memcpy(&lcw_last_watchdog_time, ¤t_time,
- sizeof(current_time));
- lcw_recent_watchdog_count = 0;
- }
-
- LCONSOLE_WARN("Service thread pid %u was inactive for %lu.%.02lus. The thread might be hung, or it might only be slow and will resume later. Dumping the stack trace for debugging purposes:\n",
- (int)lcw->lcw_pid,
- timediff.tv_sec,
- timediff.tv_nsec / (NSEC_PER_SEC / 100));
- lcw_dump(lcw);
- }
-}
-
-/*
- * Provided watchdog handlers
- */
-
-static void lc_watchdog_dumplog(pid_t pid, void *data)
-{
- libcfs_debug_dumplog_internal((void *)((uintptr_t)pid));
-}
-
-static int lcw_dispatch_main(void *data)
-{
- int rc = 0;
- struct lc_watchdog *lcw;
- struct list_head zombies = LIST_HEAD_INIT(zombies);
-
- ENTRY;
-
- complete(&lcw_start_completion);
-
- while (1) {
- int dumplog = 1;
-
- rc = wait_event_interruptible(lcw_event_waitq,
- is_watchdog_fired());
- CDEBUG(D_INFO, "Watchdog got woken up...\n");
- if (test_bit(LCW_FLAG_STOP, &lcw_flags)) {
- CDEBUG(D_INFO, "LCW_FLAG_STOP set, shutting down...\n");
-
- spin_lock_bh(&lcw_pending_timers_lock);
- rc = !list_empty(&lcw_pending_timers);
- spin_unlock_bh(&lcw_pending_timers_lock);
- if (rc) {
- CERROR("pending timers list was not empty at "
- "time of watchdog dispatch shutdown\n");
- }
- break;
- }
-
- spin_lock_bh(&lcw_pending_timers_lock);
- while (!list_empty(&lcw_pending_timers)) {
- int is_dumplog;
-
- lcw = list_entry(lcw_pending_timers.next,
- struct lc_watchdog, lcw_list);
- /* +1 ref for callback to make sure lwc wouldn't be
- * deleted after releasing lcw_pending_timers_lock */
- lcw->lcw_refcount++;
- spin_unlock_bh(&lcw_pending_timers_lock);
-
- /* lock ordering */
- spin_lock_bh(&lcw->lcw_lock);
- spin_lock_bh(&lcw_pending_timers_lock);
-
- if (list_empty(&lcw->lcw_list)) {
- /* already removed from pending list */
- lcw->lcw_refcount--; /* -1 ref for callback */
- if (lcw->lcw_refcount == 0)
- list_add(&lcw->lcw_list, &zombies);
- spin_unlock_bh(&lcw->lcw_lock);
- /* still hold lcw_pending_timers_lock */
- continue;
- }
-
- list_del_init(&lcw->lcw_list);
- lcw->lcw_refcount--; /* -1 ref for pending list */
-
- spin_unlock_bh(&lcw_pending_timers_lock);
- spin_unlock_bh(&lcw->lcw_lock);
-
- CDEBUG(D_INFO, "found lcw for pid %d\n",
- lcw->lcw_pid);
- lcw_dump_stack(lcw);
-
- is_dumplog = lcw->lcw_callback == lc_watchdog_dumplog;
- if (lcw->lcw_state != LC_WATCHDOG_DISABLED &&
- (dumplog || !is_dumplog)) {
- lcw->lcw_callback(lcw->lcw_pid, lcw->lcw_data);
- if (dumplog && is_dumplog)
- dumplog = 0;
- }
-
- spin_lock_bh(&lcw_pending_timers_lock);
- lcw->lcw_refcount--; /* -1 ref for callback */
- if (lcw->lcw_refcount == 0)
- list_add(&lcw->lcw_list, &zombies);
- }
- spin_unlock_bh(&lcw_pending_timers_lock);
-
- while (!list_empty(&zombies)) {
- lcw = list_entry(zombies.next,
- struct lc_watchdog, lcw_list);
- list_del_init(&lcw->lcw_list);
- LIBCFS_FREE(lcw, sizeof(*lcw));
- }
- }
-
- complete(&lcw_stop_completion);
-
- RETURN(rc);
-}
-
-static void lcw_dispatch_start(void)
-{
- struct task_struct *task;
-
- ENTRY;
- LASSERT(lcw_refcount == 1);
-
- init_completion(&lcw_stop_completion);
- init_completion(&lcw_start_completion);
- init_waitqueue_head(&lcw_event_waitq);
-
- CDEBUG(D_INFO, "starting dispatch thread\n");
- task = kthread_run(lcw_dispatch_main, NULL, "lc_watchdogd");
- if (IS_ERR(task)) {
- CERROR("error spawning watchdog dispatch thread: %ld\n",
- PTR_ERR(task));
- EXIT;
- return;
- }
- wait_for_completion(&lcw_start_completion);
- CDEBUG(D_INFO, "watchdog dispatcher initialization complete.\n");
-
- EXIT;
-}
-
-static void lcw_dispatch_stop(void)
-{
- ENTRY;
- LASSERT(lcw_refcount == 0);
-
- CDEBUG(D_INFO, "trying to stop watchdog dispatcher.\n");
-
- set_bit(LCW_FLAG_STOP, &lcw_flags);
- wake_up(&lcw_event_waitq);
-
- wait_for_completion(&lcw_stop_completion);
- clear_bit(LCW_FLAG_STOP, &lcw_flags);
-
- CDEBUG(D_INFO, "watchdog dispatcher has shut down.\n");
-
- EXIT;
-}
-
-struct lc_watchdog *lc_watchdog_add(int timeout,
- void (*callback)(pid_t, void *),
- void *data)
-{
- struct lc_watchdog *lcw = NULL;
- ENTRY;
-
- LIBCFS_ALLOC(lcw, sizeof(*lcw));
- if (lcw == NULL) {
- CDEBUG(D_INFO, "Could not allocate new lc_watchdog\n");
- RETURN(ERR_PTR(-ENOMEM));
- }
-
- spin_lock_init(&lcw->lcw_lock);
- lcw->lcw_refcount = 1; /* refcount for owner */
- lcw->lcw_task = current;
- lcw->lcw_pid = current_pid();
- lcw->lcw_callback = (callback != NULL) ? callback : lc_watchdog_dumplog;
- lcw->lcw_data = data;
- lcw->lcw_state = LC_WATCHDOG_DISABLED;
-
- INIT_LIST_HEAD(&lcw->lcw_list);
- cfs_timer_setup(&lcw->lcw_timer, lcw_cb, (unsigned long)lcw, 0);
-
- mutex_lock(&lcw_refcount_mutex);
- if (++lcw_refcount == 1)
- lcw_dispatch_start();
- mutex_unlock(&lcw_refcount_mutex);
-
- /* Keep this working in case we enable them by default */
- if (lcw->lcw_state == LC_WATCHDOG_ENABLED) {
- lcw->lcw_last_touched = ktime_get();
- mod_timer(&lcw->lcw_timer, cfs_time_seconds(timeout) +
- jiffies);
- }
-
- RETURN(lcw);
-}
-EXPORT_SYMBOL(lc_watchdog_add);
-
-static void lcw_update_time(struct lc_watchdog *lcw, const char *message)
-{
- ktime_t newtime = ktime_get();
-
- if (lcw->lcw_state == LC_WATCHDOG_EXPIRED) {
- ktime_t lapse = ktime_sub(newtime, lcw->lcw_last_touched);
- struct timespec64 timediff;
-
- timediff = ktime_to_timespec64(lapse);
- LCONSOLE_WARN("Service thread pid %u %s after %lu.%.02lus. This indicates the system was overloaded (too many service threads, or there were not enough hardware resources).\n",
- lcw->lcw_pid, message,
- timediff.tv_sec,
- timediff.tv_nsec / (NSEC_PER_SEC / 100));
- }
- lcw->lcw_last_touched = newtime;
-}
-
-static void lc_watchdog_del_pending(struct lc_watchdog *lcw)
-{
- spin_lock_bh(&lcw->lcw_lock);
- if (unlikely(!list_empty(&lcw->lcw_list))) {
- spin_lock_bh(&lcw_pending_timers_lock);
- list_del_init(&lcw->lcw_list);
- lcw->lcw_refcount--; /* -1 ref for pending list */
- spin_unlock_bh(&lcw_pending_timers_lock);
- }
-
- spin_unlock_bh(&lcw->lcw_lock);
-}
-
-void lc_watchdog_touch(struct lc_watchdog *lcw, int timeout)
-{
- ENTRY;
- LASSERT(lcw != NULL);
-
- lc_watchdog_del_pending(lcw);
-
- lcw_update_time(lcw, "resumed");
-
- mod_timer(&lcw->lcw_timer, jiffies + cfs_time_seconds(timeout));
- lcw->lcw_state = LC_WATCHDOG_ENABLED;
-
- EXIT;
-}
-EXPORT_SYMBOL(lc_watchdog_touch);
-
-void lc_watchdog_disable(struct lc_watchdog *lcw)
-{
- ENTRY;
- LASSERT(lcw != NULL);
-
- lc_watchdog_del_pending(lcw);
-
- lcw_update_time(lcw, "completed");
- lcw->lcw_state = LC_WATCHDOG_DISABLED;
-
- EXIT;
-}
-EXPORT_SYMBOL(lc_watchdog_disable);
-
-void lc_watchdog_delete(struct lc_watchdog *lcw)
-{
- int dead;
-
- ENTRY;
- LASSERT(lcw != NULL);
-
- del_timer(&lcw->lcw_timer);
-
- lcw_update_time(lcw, "stopped");
-
- spin_lock_bh(&lcw->lcw_lock);
- spin_lock_bh(&lcw_pending_timers_lock);
- if (unlikely(!list_empty(&lcw->lcw_list))) {
- list_del_init(&lcw->lcw_list);
- lcw->lcw_refcount--; /* -1 ref for pending list */
- }
-
- lcw->lcw_refcount--; /* -1 ref for owner */
- dead = lcw->lcw_refcount == 0;
- spin_unlock_bh(&lcw_pending_timers_lock);
- spin_unlock_bh(&lcw->lcw_lock);
-
- if (dead)
- LIBCFS_FREE(lcw, sizeof(*lcw));
-
- mutex_lock(&lcw_refcount_mutex);
- if (--lcw_refcount == 0)
- lcw_dispatch_stop();
- mutex_unlock(&lcw_refcount_mutex);
-
- EXIT;
-}
-EXPORT_SYMBOL(lc_watchdog_delete);
-
-#else /* !defined(WITH_WATCHDOG) */
-
-struct lc_watchdog *lc_watchdog_add(int timeout,
- void (*callback)(pid_t pid, void *),
- void *data)
-{
- static struct lc_watchdog watchdog;
- return &watchdog;
-}
-EXPORT_SYMBOL(lc_watchdog_add);
-
-void lc_watchdog_touch(struct lc_watchdog *lcw, int timeout)
-{
-}
-EXPORT_SYMBOL(lc_watchdog_touch);
-
-void lc_watchdog_disable(struct lc_watchdog *lcw)
-{
-}
-EXPORT_SYMBOL(lc_watchdog_disable);
-
-void lc_watchdog_delete(struct lc_watchdog *lcw)
-{
-}
-EXPORT_SYMBOL(lc_watchdog_delete);
-
-#endif
#define DEBUG_SUBSYSTEM S_RPC
#include <linux/kthread.h>
+#include <linux/ratelimit.h>
+
#include <obd_support.h>
#include <obd_class.h>
#include <lustre_net.h>
return svcpt->scp_at_check;
}
+/*
+ * If a thread runs too long or spends to much time on a single request,
+ * we want to know about it, so we set up a delayed work item as a watchdog.
+ * If it fires, we display a stack trace of the delayed thread,
+ * providing we aren't rate-limited
+ *
+ * Watchdog stack traces are limited to 3 per 'libcfs_watchdog_ratelimit'
+ * seconds
+ */
+static struct ratelimit_state watchdog_limit;
+
+static void ptlrpc_watchdog_fire(struct work_struct *w)
+{
+ struct ptlrpc_thread *thread = container_of(w, struct ptlrpc_thread,
+ t_watchdog.work);
+ u64 ms_lapse = ktime_ms_delta(ktime_get(), thread->t_touched);
+ u32 ms_frac = do_div(ms_lapse, MSEC_PER_SEC);
+
+ if (!__ratelimit(&watchdog_limit)) {
+ LCONSOLE_WARN("%s: service thread pid %u was inactive for %llu.%03u seconds. The thread might be hung, or it might only be slow and will resume later. Dumping the stack trace for debugging purposes:\n",
+ thread->t_task->comm, thread->t_task->pid,
+ ms_lapse, ms_frac);
+
+ libcfs_debug_dumpstack(thread->t_task);
+ } else {
+ LCONSOLE_WARN("%s: service thread pid %u was inactive for %llu.%03u seconds. Watchdog stack traces are limited to 3 per %u seconds, skipping this one.\n",
+ thread->t_task->comm, thread->t_task->pid,
+ ms_lapse, ms_frac, libcfs_watchdog_ratelimit);
+ }
+}
+
+static void ptlrpc_watchdog_init(struct delayed_work *work, time_t time)
+{
+ INIT_DELAYED_WORK(work, ptlrpc_watchdog_fire);
+ schedule_delayed_work(work, cfs_time_seconds(time));
+}
+
+static void ptlrpc_watchdog_disable(struct delayed_work *work)
+{
+ cancel_delayed_work_sync(work);
+}
+
+static void ptlrpc_watchdog_touch(struct delayed_work *work, time_t time)
+{
+ struct ptlrpc_thread *thread = container_of(&work->work,
+ struct ptlrpc_thread,
+ t_watchdog.work);
+ thread->t_touched = ktime_get();
+ mod_delayed_work(system_wq, work, cfs_time_seconds(time));
+}
+
/**
* requests wait on preprocessing
* user can call it w/o any lock but need to hold
struct l_wait_info lwi = LWI_TIMEOUT(svcpt->scp_rqbd_timeout,
ptlrpc_retry_rqbds, svcpt);
- lc_watchdog_disable(thread->t_watchdog);
+ ptlrpc_watchdog_disable(&thread->t_watchdog);
cond_resched();
if (ptlrpc_thread_stopping(thread))
return -EINTR;
- lc_watchdog_touch(thread->t_watchdog,
- ptlrpc_server_get_timeout(svcpt));
+ ptlrpc_watchdog_touch(&thread->t_watchdog,
+ ptlrpc_server_get_timeout(svcpt));
return 0;
}
int counter = 0, rc = 0;
ENTRY;
+ thread->t_task = current;
thread->t_pid = current_pid();
unshare_fs_struct();
/* wake up our creator in case he's still waiting. */
wake_up(&thread->t_ctl_waitq);
- thread->t_watchdog = lc_watchdog_add(ptlrpc_server_get_timeout(svcpt),
- NULL, NULL);
+ thread->t_touched = ktime_get();
+ ptlrpc_watchdog_init(&thread->t_watchdog,
+ ptlrpc_server_get_timeout(svcpt));
spin_lock(&svcpt->scp_rep_lock);
list_add(&rs->rs_list, &svcpt->scp_rep_idle);
}
}
- lc_watchdog_delete(thread->t_watchdog);
- thread->t_watchdog = NULL;
+ ptlrpc_watchdog_disable(&thread->t_watchdog);
out_srv_fini:
/*
if (ptlrpc_hr.hr_partitions == NULL)
RETURN(-ENOMEM);
+ ratelimit_state_init(&watchdog_limit,
+ cfs_time_seconds(libcfs_watchdog_ratelimit), 3);
+
init_waitqueue_head(&ptlrpc_hr.hr_waitq);
weight = cpumask_weight(topology_sibling_cpumask(smp_processor_id()));
git://git.whamcloud.com / fs / lustre-release.git / commitdiff