Commit 82d94856 authored by Peter Zijlstra's avatar Peter Zijlstra Committed by Ingo Molnar

perf/core: Fix lock inversion between perf,trace,cpuhp

Lockdep gifted us with noticing the following 4-way lockup scenario:

        perf_trace_init()
 #0       mutex_lock(&event_mutex)
          perf_trace_event_init()
            perf_trace_event_reg()
              tp_event->class->reg() := tracepoint_probe_register
 #1             mutex_lock(&tracepoints_mutex)
                  trace_point_add_func()
 #2                 static_key_enable()

 #2     do_cpu_up()
          perf_event_init_cpu()
 #3         mutex_lock(&pmus_lock)
 #4         mutex_lock(&ctx->mutex)

        perf_event_task_disable()
          mutex_lock(&current->perf_event_mutex)
 #4       ctx = perf_event_ctx_lock()
 #5       perf_event_for_each_child()

        do_exit()
          task_work_run()
            __fput()
              perf_release()
                perf_event_release_kernel()
 #4               mutex_lock(&ctx->mutex)
 #5               mutex_lock(&event->child_mutex)
                  free_event()
                    _free_event()
                      event->destroy() := perf_trace_destroy
 #0                     mutex_lock(&event_mutex);

Fix that by moving the free_event() out from under the locks.
Signed-off-by: default avatarPeter Zijlstra (Intel) <peterz@infradead.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Steven Rostedt (VMware) <rostedt@goodmis.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vince Weaver <vincent.weaver@maine.edu>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: default avatarIngo Molnar <mingo@kernel.org>
parent 1f07476e
...@@ -1231,6 +1231,10 @@ static void put_ctx(struct perf_event_context *ctx) ...@@ -1231,6 +1231,10 @@ static void put_ctx(struct perf_event_context *ctx)
* perf_event_context::lock * perf_event_context::lock
* perf_event::mmap_mutex * perf_event::mmap_mutex
* mmap_sem * mmap_sem
*
* cpu_hotplug_lock
* pmus_lock
* cpuctx->mutex / perf_event_context::mutex
*/ */
static struct perf_event_context * static struct perf_event_context *
perf_event_ctx_lock_nested(struct perf_event *event, int nesting) perf_event_ctx_lock_nested(struct perf_event *event, int nesting)
...@@ -4196,6 +4200,7 @@ int perf_event_release_kernel(struct perf_event *event) ...@@ -4196,6 +4200,7 @@ int perf_event_release_kernel(struct perf_event *event)
{ {
struct perf_event_context *ctx = event->ctx; struct perf_event_context *ctx = event->ctx;
struct perf_event *child, *tmp; struct perf_event *child, *tmp;
LIST_HEAD(free_list);
/* /*
* If we got here through err_file: fput(event_file); we will not have * If we got here through err_file: fput(event_file); we will not have
...@@ -4268,8 +4273,7 @@ int perf_event_release_kernel(struct perf_event *event) ...@@ -4268,8 +4273,7 @@ int perf_event_release_kernel(struct perf_event *event)
struct perf_event, child_list); struct perf_event, child_list);
if (tmp == child) { if (tmp == child) {
perf_remove_from_context(child, DETACH_GROUP); perf_remove_from_context(child, DETACH_GROUP);
list_del(&child->child_list); list_move(&child->child_list, &free_list);
free_event(child);
/* /*
* This matches the refcount bump in inherit_event(); * This matches the refcount bump in inherit_event();
* this can't be the last reference. * this can't be the last reference.
...@@ -4284,6 +4288,11 @@ int perf_event_release_kernel(struct perf_event *event) ...@@ -4284,6 +4288,11 @@ int perf_event_release_kernel(struct perf_event *event)
} }
mutex_unlock(&event->child_mutex); mutex_unlock(&event->child_mutex);
list_for_each_entry_safe(child, tmp, &free_list, child_list) {
list_del(&child->child_list);
free_event(child);
}
no_ctx: no_ctx:
put_event(event); /* Must be the 'last' reference */ put_event(event); /* Must be the 'last' reference */
return 0; return 0;
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment