linux-kernel - Re: [PATCH] perf: Fix missing SIGTRAPs

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]

Date:   Sat, 8 Oct 2022 10:41:28 +0200
From:   Marco Elver <elver@...gle.com>
To:     Peter Zijlstra <peterz@...radead.org>
Cc:     Ingo Molnar <mingo@...hat.com>,
        Arnaldo Carvalho de Melo <acme@...nel.org>,
        Mark Rutland <mark.rutland@....com>,
        Alexander Shishkin <alexander.shishkin@...ux.intel.com>,
        Jiri Olsa <jolsa@...nel.org>,
        Namhyung Kim <namhyung@...nel.org>,
        linux-perf-users@...r.kernel.org, linux-kernel@...r.kernel.org,
        kasan-dev@...glegroups.com, Dmitry Vyukov <dvyukov@...gle.com>
Subject: Re: [PATCH] perf: Fix missing SIGTRAPs

On Fri, Oct 07, 2022 at 06:14PM +0200, Marco Elver wrote:
> On Fri, Oct 07, 2022 at 03:58PM +0200, Marco Elver wrote:
> > On Fri, Oct 07, 2022 at 03:09PM +0200, Peter Zijlstra wrote:
> > > On Fri, Oct 07, 2022 at 11:37:34AM +0200, Marco Elver wrote:
> > > 
> > > > That worked. In addition I had to disable the ctx->task != current check
> > > > if we're in task_work, because presumably the event might have already
> > > > been disabled/moved??
> > > 
> > > Uhmmm... uhhh... damn. (wall-time was significantly longer)
> > > 
> > > Does this help?
> > 
> > No unfortunately - still see:
> > 
> > [   82.300827] ------------[ cut here ]------------
> > [   82.301680] WARNING: CPU: 0 PID: 976 at kernel/events/core.c:6466 perf_sigtrap+0x60/0x70
> 
> Whenever the warning fires, I see that event->state is OFF.

The below patch to the sigtrap_threads test can repro the issue (when
run lots of them concurrently again). It also illustrates the original
problem we're trying to solve, where the event never gets rearmed again
and the test times out (doesn't happen with the almost-working fix).

Thanks,
-- Marco

------ >8 ------

>From 98d225bda6d94dd793a1d0c77ae4b301c364166e Mon Sep 17 00:00:00 2001
From: Marco Elver <elver@...gle.com>
Date: Sat, 8 Oct 2022 10:26:58 +0200
Subject: [PATCH] selftests/perf_events: Add a SIGTRAP stress test with
 disables

Add a SIGTRAP stress test that exercises repeatedly enabling/disabling
an event while it concurrently keeps firing.

Signed-off-by: Marco Elver <elver@...gle.com>
---
 .../selftests/perf_events/sigtrap_threads.c   | 35 +++++++++++++++++--
 1 file changed, 32 insertions(+), 3 deletions(-)

diff --git a/tools/testing/selftests/perf_events/sigtrap_threads.c b/tools/testing/selftests/perf_events/sigtrap_threads.c
index 6d849dc2bee0..d1d8483ac628 100644
--- a/tools/testing/selftests/perf_events/sigtrap_threads.c
+++ b/tools/testing/selftests/perf_events/sigtrap_threads.c
@@ -62,6 +62,8 @@ static struct perf_event_attr make_event_attr(bool enabled, volatile void *addr,
 		.remove_on_exec = 1, /* Required by sigtrap. */
 		.sigtrap	= 1, /* Request synchronous SIGTRAP on event. */
 		.sig_data	= TEST_SIG_DATA(addr, id),
+		.exclude_kernel = 1, /* To allow */
+		.exclude_hv     = 1, /* running as !root */
 	};
 	return attr;
 }
@@ -93,9 +95,13 @@ static void *test_thread(void *arg)
 
 	__atomic_fetch_add(&ctx.tids_want_signal, tid, __ATOMIC_RELAXED);
 	iter = ctx.iterate_on; /* read */
-	for (i = 0; i < iter - 1; i++) {
-		__atomic_fetch_add(&ctx.tids_want_signal, tid, __ATOMIC_RELAXED);
-		ctx.iterate_on = iter; /* idempotent write */
+	if (iter >= 0) {
+		for (i = 0; i < iter - 1; i++) {
+			__atomic_fetch_add(&ctx.tids_want_signal, tid, __ATOMIC_RELAXED);
+			ctx.iterate_on = iter; /* idempotent write */
+		}
+	} else {
+		while (ctx.iterate_on);
 	}
 
 	return NULL;
@@ -208,4 +214,27 @@ TEST_F(sigtrap_threads, signal_stress)
 	EXPECT_EQ(ctx.first_siginfo.si_perf_data, TEST_SIG_DATA(&ctx.iterate_on, 0));
 }
 
+TEST_F(sigtrap_threads, signal_stress_with_disable)
+{
+	const int target_count = NUM_THREADS * 3000;
+	int i;
+
+	ctx.iterate_on = -1;
+
+	EXPECT_EQ(ioctl(self->fd, PERF_EVENT_IOC_ENABLE, 0), 0);
+	pthread_barrier_wait(&self->barrier);
+	while (__atomic_load_n(&ctx.signal_count, __ATOMIC_RELAXED) < target_count) {
+		EXPECT_EQ(ioctl(self->fd, PERF_EVENT_IOC_DISABLE, 0), 0);
+		EXPECT_EQ(ioctl(self->fd, PERF_EVENT_IOC_ENABLE, 0), 0);
+	}
+	ctx.iterate_on = 0;
+	for (i = 0; i < NUM_THREADS; i++)
+		ASSERT_EQ(pthread_join(self->threads[i], NULL), 0);
+	EXPECT_EQ(ioctl(self->fd, PERF_EVENT_IOC_DISABLE, 0), 0);
+
+	EXPECT_EQ(ctx.first_siginfo.si_addr, &ctx.iterate_on);
+	EXPECT_EQ(ctx.first_siginfo.si_perf_type, PERF_TYPE_BREAKPOINT);
+	EXPECT_EQ(ctx.first_siginfo.si_perf_data, TEST_SIG_DATA(&ctx.iterate_on, 0));
+}
+
 TEST_HARNESS_MAIN
-- 
2.38.0.rc1.362.ged0d419d3c-goog