linux-kernel - [PATCH 1/3 v2] perf_event: fix race in perf_swevent_get_recursion

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <4B556F66.3070402@cn.fujitsu.com>
Date:	Tue, 19 Jan 2010 16:37:58 +0800
From:	Xiao Guangrong <xiaoguangrong@...fujitsu.com>
To:	Ingo Molnar <mingo@...e.hu>
CC:	Peter Zijlstra <peterz@...radead.org>,
	Frederic Weisbecker <fweisbec@...il.com>,
	Paul Mackerras <paulus@...ba.org>,
	Masami Hiramatsu <mhiramat@...hat.com>,
	Jason Baron <jbaron@...hat.com>,
	LKML <linux-kernel@...r.kernel.org>
Subject: [PATCH 1/3 v2] perf_event: fix race in perf_swevent_get_recursion_context()

It only disable preemption in perf_swevent_get_recursion_context(),
it can't avoid race of hard-irq and NMI since they are nested that
will re-enter this path and make the recursion value chaotic

In this patch, we use atomic operation to avoid it and reduce
cpu_ctx->recursion size, it also make this patch no need disable
preemption

Signed-off-by: Xiao Guangrong <xiaoguangrong@...fujitsu.com>
---
 include/linux/perf_event.h |    2 +-
 kernel/perf_event.c        |   14 ++++----------
 2 files changed, 5 insertions(+), 11 deletions(-)

diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h
index c6f812e..8474ab0 100644
--- a/include/linux/perf_event.h
+++ b/include/linux/perf_event.h
@@ -730,7 +730,7 @@ struct perf_cpu_context {
 	 *
 	 * task, softirq, irq, nmi context
 	 */
-	int				recursion[4];
+	unsigned long			recursion;
 };
 
 struct perf_output_handle {
diff --git a/kernel/perf_event.c b/kernel/perf_event.c
index eae6ff6..77ef16e 100644
--- a/kernel/perf_event.c
+++ b/kernel/perf_event.c
@@ -3921,7 +3921,7 @@ static void perf_swevent_ctx_event(struct perf_event_context *ctx,
 
 int perf_swevent_get_recursion_context(void)
 {
-	struct perf_cpu_context *cpuctx = &get_cpu_var(perf_cpu_context);
+	struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context);
 	int rctx;
 
 	if (in_nmi())
@@ -3933,13 +3933,8 @@ int perf_swevent_get_recursion_context(void)
 	else
 		rctx = 0;
 
-	if (cpuctx->recursion[rctx]) {
-		put_cpu_var(perf_cpu_context);
+	if (test_and_set_bit(rctx, &cpuctx->recursion))
 		return -1;
-	}
-
-	cpuctx->recursion[rctx]++;
-	barrier();
 
 	return rctx;
 }
@@ -3948,9 +3943,8 @@ EXPORT_SYMBOL_GPL(perf_swevent_get_recursion_context);
 void perf_swevent_put_recursion_context(int rctx)
 {
 	struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context);
-	barrier();
-	cpuctx->recursion[rctx]--;
-	put_cpu_var(perf_cpu_context);
+
+	clear_bit(rctx, &cpuctx->recursion);
 }
 EXPORT_SYMBOL_GPL(perf_swevent_put_recursion_context);
 
-- 
1.6.1.2


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