[tip:perfcounters/core] perf_counter: Fix race in counter initialization

From: tip-bot for Peter Zijlstra
Date: Wed Jun 03 2009 - 09:08:27 EST


Commit-ID: a96bbc16418bc691317f265d6bf98ba941ca9c1a
Gitweb: http://git.kernel.org/tip/a96bbc16418bc691317f265d6bf98ba941ca9c1a
Author: Peter Zijlstra <a.p.zijlstra@xxxxxxxxx>
AuthorDate: Wed, 3 Jun 2009 14:01:36 +0200
Committer: Ingo Molnar <mingo@xxxxxxx>
CommitDate: Wed, 3 Jun 2009 14:57:03 +0200

perf_counter: Fix race in counter initialization

We need the PID namespace and counter ID available when the
counter overflows and we need to generate a sample event.

[ Impact: fix kernel crash with high-frequency sampling ]

Reported-by: Ingo Molnar <mingo@xxxxxxx>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@xxxxxxxxx>
Cc: Mike Galbraith <efault@xxxxxx>
Cc: Paul Mackerras <paulus@xxxxxxxxx>
Cc: Corey Ashford <cjashfor@xxxxxxxxxxxxxxxxxx>
Cc: Marcelo Tosatti <mtosatti@xxxxxxxxxx>
Cc: Arnaldo Carvalho de Melo <acme@xxxxxxxxxx>
Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
Cc: John Kacur <jkacur@xxxxxxxxxx>
LKML-Reference: <new-submission>
[ fixed a further crash and cleaned up the initialization a bit ]
Signed-off-by: Ingo Molnar <mingo@xxxxxxx>


---
kernel/perf_counter.c | 25 ++++++++++++++-----------
1 files changed, 14 insertions(+), 11 deletions(-)

diff --git a/kernel/perf_counter.c b/kernel/perf_counter.c
index 317cef7..ab44554 100644
--- a/kernel/perf_counter.c
+++ b/kernel/perf_counter.c
@@ -48,6 +48,8 @@ int sysctl_perf_counter_priv __read_mostly; /* do we need to be privileged */
int sysctl_perf_counter_mlock __read_mostly = 512; /* 'free' kb per user */
int sysctl_perf_counter_limit __read_mostly = 100000; /* max NMIs per second */

+static atomic64_t perf_counter_id;
+
/*
* Lock for (sysadmin-configurable) counter reservations:
*/
@@ -3351,14 +3353,18 @@ perf_counter_alloc(struct perf_counter_attr *attr,

mutex_init(&counter->mmap_mutex);

- counter->cpu = cpu;
+ counter->cpu = cpu;
counter->attr = *attr;
- counter->group_leader = group_leader;
- counter->pmu = NULL;
- counter->ctx = ctx;
- counter->oncpu = -1;
+ counter->group_leader = group_leader;
+ counter->pmu = NULL;
+ counter->ctx = ctx;
+ counter->oncpu = -1;
+
+ counter->ns = get_pid_ns(current->nsproxy->pid_ns);
+ counter->id = atomic64_inc_return(&perf_counter_id);
+
+ counter->state = PERF_COUNTER_STATE_INACTIVE;

- counter->state = PERF_COUNTER_STATE_INACTIVE;
if (attr->disabled)
counter->state = PERF_COUNTER_STATE_OFF;

@@ -3402,6 +3408,8 @@ done:
err = PTR_ERR(pmu);

if (err) {
+ if (counter->ns)
+ put_pid_ns(counter->ns);
kfree(counter);
return ERR_PTR(err);
}
@@ -3419,8 +3427,6 @@ done:
return counter;
}

-static atomic64_t perf_counter_id;
-
/**
* sys_perf_counter_open - open a performance counter, associate it to a task/cpu
*
@@ -3515,9 +3521,6 @@ SYSCALL_DEFINE5(perf_counter_open,
list_add_tail(&counter->owner_entry, &current->perf_counter_list);
mutex_unlock(&current->perf_counter_mutex);

- counter->ns = get_pid_ns(current->nsproxy->pid_ns);
- counter->id = atomic64_inc_return(&perf_counter_id);
-
fput_light(counter_file, fput_needed2);

out_fput:
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/