All of lore.kernel.org
 help / color / mirror / Atom feed
From: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
To: linux-kernel@vger.kernel.org
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
	stable@vger.kernel.org, Yang Jihong <yangjihong1@huawei.com>,
	"Peter Zijlstra (Intel)" <peterz@infradead.org>,
	Sasha Levin <sashal@kernel.org>
Subject: [PATCH 4.9 08/26] perf/core: Fix data race between perf_event_set_output() and perf_mmap_close()
Date: Wed, 27 Jul 2022 18:10:37 +0200	[thread overview]
Message-ID: <20220727160959.472445849@linuxfoundation.org> (raw)
In-Reply-To: <20220727160959.122591422@linuxfoundation.org>

From: Peter Zijlstra <peterz@infradead.org>

[ Upstream commit 68e3c69803dada336893640110cb87221bb01dcf ]

Yang Jihing reported a race between perf_event_set_output() and
perf_mmap_close():

	CPU1					CPU2

	perf_mmap_close(e2)
	  if (atomic_dec_and_test(&e2->rb->mmap_count)) // 1 - > 0
	    detach_rest = true

						ioctl(e1, IOC_SET_OUTPUT, e2)
						  perf_event_set_output(e1, e2)

	  ...
	  list_for_each_entry_rcu(e, &e2->rb->event_list, rb_entry)
	    ring_buffer_attach(e, NULL);
	    // e1 isn't yet added and
	    // therefore not detached

						    ring_buffer_attach(e1, e2->rb)
						      list_add_rcu(&e1->rb_entry,
								   &e2->rb->event_list)

After this; e1 is attached to an unmapped rb and a subsequent
perf_mmap() will loop forever more:

	again:
		mutex_lock(&e->mmap_mutex);
		if (event->rb) {
			...
			if (!atomic_inc_not_zero(&e->rb->mmap_count)) {
				...
				mutex_unlock(&e->mmap_mutex);
				goto again;
			}
		}

The loop in perf_mmap_close() holds e2->mmap_mutex, while the attach
in perf_event_set_output() holds e1->mmap_mutex. As such there is no
serialization to avoid this race.

Change perf_event_set_output() to take both e1->mmap_mutex and
e2->mmap_mutex to alleviate that problem. Additionally, have the loop
in perf_mmap() detach the rb directly, this avoids having to wait for
the concurrent perf_mmap_close() to get around to doing it to make
progress.

Fixes: 9bb5d40cd93c ("perf: Fix mmap() accounting hole")
Reported-by: Yang Jihong <yangjihong1@huawei.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Yang Jihong <yangjihong1@huawei.com>
Link: https://lkml.kernel.org/r/YsQ3jm2GR38SW7uD@worktop.programming.kicks-ass.net
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 kernel/events/core.c | 45 ++++++++++++++++++++++++++++++--------------
 1 file changed, 31 insertions(+), 14 deletions(-)

diff --git a/kernel/events/core.c b/kernel/events/core.c
index 2466e2ae54dc..58ef731d52c7 100644
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -5291,10 +5291,10 @@ static int perf_mmap(struct file *file, struct vm_area_struct *vma)
 
 		if (!atomic_inc_not_zero(&event->rb->mmap_count)) {
 			/*
-			 * Raced against perf_mmap_close() through
-			 * perf_event_set_output(). Try again, hope for better
-			 * luck.
+			 * Raced against perf_mmap_close(); remove the
+			 * event and try again.
 			 */
+			ring_buffer_attach(event, NULL);
 			mutex_unlock(&event->mmap_mutex);
 			goto again;
 		}
@@ -9542,14 +9542,25 @@ static int perf_copy_attr(struct perf_event_attr __user *uattr,
 	goto out;
 }
 
+static void mutex_lock_double(struct mutex *a, struct mutex *b)
+{
+	if (b < a)
+		swap(a, b);
+
+	mutex_lock(a);
+	mutex_lock_nested(b, SINGLE_DEPTH_NESTING);
+}
+
 static int
 perf_event_set_output(struct perf_event *event, struct perf_event *output_event)
 {
 	struct ring_buffer *rb = NULL;
 	int ret = -EINVAL;
 
-	if (!output_event)
+	if (!output_event) {
+		mutex_lock(&event->mmap_mutex);
 		goto set;
+	}
 
 	/* don't allow circular references */
 	if (event == output_event)
@@ -9587,8 +9598,15 @@ perf_event_set_output(struct perf_event *event, struct perf_event *output_event)
 	    event->pmu != output_event->pmu)
 		goto out;
 
+	/*
+	 * Hold both mmap_mutex to serialize against perf_mmap_close().  Since
+	 * output_event is already on rb->event_list, and the list iteration
+	 * restarts after every removal, it is guaranteed this new event is
+	 * observed *OR* if output_event is already removed, it's guaranteed we
+	 * observe !rb->mmap_count.
+	 */
+	mutex_lock_double(&event->mmap_mutex, &output_event->mmap_mutex);
 set:
-	mutex_lock(&event->mmap_mutex);
 	/* Can't redirect output if we've got an active mmap() */
 	if (atomic_read(&event->mmap_count))
 		goto unlock;
@@ -9598,6 +9616,12 @@ perf_event_set_output(struct perf_event *event, struct perf_event *output_event)
 		rb = ring_buffer_get(output_event);
 		if (!rb)
 			goto unlock;
+
+		/* did we race against perf_mmap_close() */
+		if (!atomic_read(&rb->mmap_count)) {
+			ring_buffer_put(rb);
+			goto unlock;
+		}
 	}
 
 	ring_buffer_attach(event, rb);
@@ -9605,20 +9629,13 @@ perf_event_set_output(struct perf_event *event, struct perf_event *output_event)
 	ret = 0;
 unlock:
 	mutex_unlock(&event->mmap_mutex);
+	if (output_event)
+		mutex_unlock(&output_event->mmap_mutex);
 
 out:
 	return ret;
 }
 
-static void mutex_lock_double(struct mutex *a, struct mutex *b)
-{
-	if (b < a)
-		swap(a, b);
-
-	mutex_lock(a);
-	mutex_lock_nested(b, SINGLE_DEPTH_NESTING);
-}
-
 static int perf_event_set_clock(struct perf_event *event, clockid_t clk_id)
 {
 	bool nmi_safe = false;
-- 
2.35.1




  parent reply	other threads:[~2022-07-27 16:23 UTC|newest]

Thread overview: 33+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-07-27 16:10 [PATCH 4.9 00/26] 4.9.325-rc1 review Greg Kroah-Hartman
2022-07-27 16:10 ` [PATCH 4.9 01/26] security,selinux,smack: kill security_task_wait hook Greg Kroah-Hartman
2022-07-27 16:10 ` [PATCH 4.9 02/26] xen/gntdev: Ignore failure to unmap INVALID_GRANT_HANDLE Greg Kroah-Hartman
2022-07-27 16:10 ` [PATCH 4.9 03/26] misc: rtsx_usb: fix use of dma mapped buffer for usb bulk transfer Greg Kroah-Hartman
2022-07-27 16:10 ` [PATCH 4.9 04/26] misc: rtsx_usb: use separate command and response buffers Greg Kroah-Hartman
2022-07-27 16:10 ` [PATCH 4.9 05/26] misc: rtsx_usb: set return value in rsp_buf alloc err path Greg Kroah-Hartman
2022-07-27 16:10 ` [PATCH 4.9 06/26] xfrm: xfrm_policy: fix a possible double xfrm_pols_put() in xfrm_bundle_lookup() Greg Kroah-Hartman
2022-07-27 16:10 ` [PATCH 4.9 07/26] power/reset: arm-versatile: Fix refcount leak in versatile_reboot_probe Greg Kroah-Hartman
2022-07-27 16:10 ` Greg Kroah-Hartman [this message]
2022-07-27 16:10 ` [PATCH 4.9 09/26] ip: Fix a data-race around sysctl_fwmark_reflect Greg Kroah-Hartman
2022-07-27 16:10 ` [PATCH 4.9 10/26] tcp/dccp: Fix a data-race around sysctl_tcp_fwmark_accept Greg Kroah-Hartman
2022-07-27 16:10 ` [PATCH 4.9 11/26] tcp: Fix a data-race around sysctl_tcp_probe_threshold Greg Kroah-Hartman
2022-07-27 16:10 ` [PATCH 4.9 12/26] i2c: cadence: Change large transfer count reset logic to be unconditional Greg Kroah-Hartman
2022-07-27 16:10 ` [PATCH 4.9 13/26] igmp: Fix data-races around sysctl_igmp_llm_reports Greg Kroah-Hartman
2022-07-27 16:10 ` [PATCH 4.9 14/26] igmp: Fix a data-race around sysctl_igmp_max_memberships Greg Kroah-Hartman
2022-07-27 16:10 ` [PATCH 4.9 15/26] tcp: Fix a data-race around sysctl_tcp_notsent_lowat Greg Kroah-Hartman
2022-07-27 16:10 ` [PATCH 4.9 16/26] be2net: Fix buffer overflow in be_get_module_eeprom Greg Kroah-Hartman
2022-07-27 16:10 ` [PATCH 4.9 17/26] Revert "Revert "char/random: silence a lockdep splat with printk()"" Greg Kroah-Hartman
2022-07-27 16:10 ` [PATCH 4.9 18/26] mm/mempolicy: fix uninit-value in mpol_rebind_policy() Greg Kroah-Hartman
2022-07-27 16:10 ` [PATCH 4.9 19/26] bpf: Make sure mac_header was set before using it Greg Kroah-Hartman
2022-07-27 16:10 ` [PATCH 4.9 20/26] ALSA: memalloc: Align buffer allocations in page size Greg Kroah-Hartman
2022-07-27 16:10 ` [PATCH 4.9 21/26] tty: drivers/tty/, stop using tty_schedule_flip() Greg Kroah-Hartman
2022-07-27 16:10 ` [PATCH 4.9 22/26] tty: the rest, " Greg Kroah-Hartman
2022-07-27 16:10 ` [PATCH 4.9 23/26] tty: drop tty_schedule_flip() Greg Kroah-Hartman
2022-07-27 16:10 ` [PATCH 4.9 24/26] tty: extract tty_flip_buffer_commit() from tty_flip_buffer_push() Greg Kroah-Hartman
2022-07-27 16:10 ` [PATCH 4.9 25/26] tty: use new tty_insert_flip_string_and_push_buffer() in pty_write() Greg Kroah-Hartman
2022-07-27 16:10 ` [PATCH 4.9 26/26] net: usb: ax88179_178a needs FLAG_SEND_ZLP Greg Kroah-Hartman
2022-07-27 22:18 ` [PATCH 4.9 00/26] 4.9.325-rc1 review Florian Fainelli
2022-07-28  9:42 ` Pavel Machek
2022-07-28 10:08 ` Naresh Kamboju
2022-07-28 14:32 ` Jon Hunter
2022-07-28 14:52 ` Shuah Khan
2022-07-28 22:57 ` Guenter Roeck

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20220727160959.472445849@linuxfoundation.org \
    --to=gregkh@linuxfoundation.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=peterz@infradead.org \
    --cc=sashal@kernel.org \
    --cc=stable@vger.kernel.org \
    --cc=yangjihong1@huawei.com \
    --subject='Re: [PATCH 4.9 08/26] perf/core: Fix data race between perf_event_set_output() and perf_mmap_close()' \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.