tracing, perf_events: Protect the buffer from recursion in perf

While tracing using events with perf, if one enables the
lockdep:lock_acquire event, it will infect every other perf
trace events.

Basically, you can enable whatever set of trace events through
perf but if this event is part of the set, the only result we
can get is a long list of lock_acquire events of rcu read lock,
and only that.

This is because of a recursion inside perf.

1) When a trace event is triggered, it will fill a per cpu
   buffer and submit it to perf.

2) Perf will commit this event but will also protect some data
   using rcu_read_lock

3) A recursion appears: rcu_read_lock triggers a lock_acquire
   event that will fill the per cpu event and then submit the
   buffer to perf.

4) Perf detects a recursion and ignores it

5) Perf continues its work on the previous event, but its buffer
   has been overwritten by the lock_acquire event, it has then
   been turned into a lock_acquire event of rcu read lock

Such scenario also happens with lock_release with
rcu_read_unlock().

We could turn the rcu_read_lock() into __rcu_read_lock() to drop
the lock debugging from perf fast path, but that would make us
lose the rcu debugging and that doesn't prevent from other
possible kind of recursion from perf in the future.

This patch adds a recursion protection based on a counter on the
perf trace per cpu buffers to solve the problem.

-v2: Fixed lost whitespace, added reviewed-by tag

Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Reviewed-by: Masami Hiramatsu <mhiramat@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Li Zefan <lizf@cn.fujitsu.com>
Cc: Jason Baron <jbaron@redhat.com>
LKML-Reference: <1257477185-7838-1-git-send-email-fweisbec@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

1 files changed, 18 insertions, 23 deletions

diff --git a/kernel/trace/trace_event_profile.c b/kernel/trace/trace_event_profile.c
index c9f687ab0d4f..e0d351b01f5a 100644
--- a/kernel/trace/trace_event_profile.c
+++ b/kernel/trace/trace_event_profile.c
@@ -8,41 +8,36 @@
 #include <linux/module.h>
 #include "trace.h"
 
-/*
- * We can't use a size but a type in alloc_percpu()
- * So let's create a dummy type that matches the desired size
- */
-typedef struct {char buf[FTRACE_MAX_PROFILE_SIZE];} profile_buf_t;
 
-char            *trace_profile_buf;
-EXPORT_SYMBOL_GPL(trace_profile_buf);
+struct perf_trace_buf *perf_trace_buf;
+EXPORT_SYMBOL_GPL(perf_trace_buf);
 
-char            *trace_profile_buf_nmi;
-EXPORT_SYMBOL_GPL(trace_profile_buf_nmi);
+struct perf_trace_buf *perf_trace_buf_nmi;
+EXPORT_SYMBOL_GPL(perf_trace_buf_nmi);
 
 /* Count the events in use (per event id, not per instance) */
 static int      total_profile_count;
 
 static int ftrace_profile_enable_event(struct ftrace_event_call *event)
 {
-        char *buf;
+        struct perf_trace_buf *buf;
         int ret = -ENOMEM;
 
         if (atomic_inc_return(&event->profile_count))
                 return 0;
 
         if (!total_profile_count) {
-                buf = (char *)alloc_percpu(profile_buf_t);
+                buf = alloc_percpu(struct perf_trace_buf);
                 if (!buf)
                         goto fail_buf;
 
-                rcu_assign_pointer(trace_profile_buf, buf);
+                rcu_assign_pointer(perf_trace_buf, buf);
 
-                buf = (char *)alloc_percpu(profile_buf_t);
+                buf = alloc_percpu(struct perf_trace_buf);
                 if (!buf)
                         goto fail_buf_nmi;
 
-                rcu_assign_pointer(trace_profile_buf_nmi, buf);
+                rcu_assign_pointer(perf_trace_buf_nmi, buf);
         }
 
         ret = event->profile_enable(event);
@@ -53,10 +48,10 @@ static int ftrace_profile_enable_event(struct ftrace_event_call *event)
 
 fail_buf_nmi:
         if (!total_profile_count) {
-                free_percpu(trace_profile_buf_nmi);
-                free_percpu(trace_profile_buf);
-                trace_profile_buf_nmi = NULL;
-                trace_profile_buf = NULL;
+                free_percpu(perf_trace_buf_nmi);
+                free_percpu(perf_trace_buf);
+                perf_trace_buf_nmi = NULL;
+                perf_trace_buf = NULL;
         }
 fail_buf:
         atomic_dec(&event->profile_count);
@@ -84,7 +79,7 @@ int ftrace_profile_enable(int event_id)
 
 static void ftrace_profile_disable_event(struct ftrace_event_call *event)
 {
-        char *buf, *nmi_buf;
+        struct perf_trace_buf *buf, *nmi_buf;
 
         if (!atomic_add_negative(-1, &event->profile_count))
                 return;
@@ -92,11 +87,11 @@ static void ftrace_profile_disable_event(struct ftrace_event_call *event)
         event->profile_disable(event);
 
         if (!--total_profile_count) {
-                buf = trace_profile_buf;
-                rcu_assign_pointer(trace_profile_buf, NULL);
+                buf = perf_trace_buf;
+                rcu_assign_pointer(perf_trace_buf, NULL);
 
-                nmi_buf = trace_profile_buf_nmi;
-                rcu_assign_pointer(trace_profile_buf_nmi, NULL);
+                nmi_buf = perf_trace_buf_nmi;
+                rcu_assign_pointer(perf_trace_buf_nmi, NULL);
 
                 /*
                  * Ensure every events in profiling have finished before