[syzbot] KCSAN: data-race in __io_uring_cancel / io_uring_try_cancel_requests

[syzbot] KCSAN: data-race in __io_uring_cancel / io_uring_try_cancel_requests

[syzbot]

Hello,

syzbot found the following issue on:

HEAD commit:    a050a6d2 Merge tag 'perf-tools-fixes-for-v5.13-2021-05-24'..
git tree:       upstream
console output: https://syzkaller.appspot.com/x/log.txt?x=13205087d00000
kernel config:  https://syzkaller.appspot.com/x/.config?x=3bcc8a6b51ef8094
dashboard link: https://syzkaller.appspot.com/bug?extid=73554e2258b7b8bf0bbf
compiler:       Debian clang version 11.0.1-2

Unfortunately, I don't have any reproducer for this issue yet.

IMPORTANT: if you fix the issue, please add the following tag to the commit:
Reported-by: syzbot+73554e2258b7b8bf0bbf@syzkaller.appspotmail.com

==================================================================
BUG: KCSAN: data-race in __io_uring_cancel / io_uring_try_cancel_requests

write to 0xffff88811d8df330 of 8 bytes by task 3709 on cpu 1:
 io_uring_clean_tctx fs/io_uring.c:9042 [inline]
 __io_uring_cancel+0x261/0x3b0 fs/io_uring.c:9136
 io_uring_files_cancel include/linux/io_uring.h:16 [inline]
 do_exit+0x185/0x1560 kernel/exit.c:781
 do_group_exit+0xce/0x1a0 kernel/exit.c:923
 get_signal+0xfc3/0x1610 kernel/signal.c:2835
 arch_do_signal_or_restart+0x2a/0x220 arch/x86/kernel/signal.c:789
 handle_signal_work kernel/entry/common.c:147 [inline]
 exit_to_user_mode_loop kernel/entry/common.c:171 [inline]
 exit_to_user_mode_prepare+0x109/0x190 kernel/entry/common.c:208
 __syscall_exit_to_user_mode_work kernel/entry/common.c:290 [inline]
 syscall_exit_to_user_mode+0x20/0x40 kernel/entry/common.c:301
 do_syscall_64+0x56/0x90 arch/x86/entry/common.c:57
 entry_SYSCALL_64_after_hwframe+0x44/0xae

read to 0xffff88811d8df330 of 8 bytes by task 6412 on cpu 0:
 io_uring_try_cancel_iowq fs/io_uring.c:8911 [inline]
 io_uring_try_cancel_requests+0x1ce/0x8e0 fs/io_uring.c:8933
 io_ring_exit_work+0x7c/0x1110 fs/io_uring.c:8736
 process_one_work+0x3e9/0x8f0 kernel/workqueue.c:2276
 worker_thread+0x636/0xae0 kernel/workqueue.c:2422
 kthread+0x1d0/0x1f0 kernel/kthread.c:313
 ret_from_fork+0x1f/0x30 arch/x86/entry/entry_64.S:294

Reported by Kernel Concurrency Sanitizer on:
CPU: 0 PID: 6412 Comm: kworker/u4:9 Not tainted 5.13.0-rc3-syzkaller #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Workqueue: events_unbound io_ring_exit_work
==================================================================


---
This report is generated by a bot. It may contain errors.
See https://goo.gl/tpsmEJ for more information about syzbot.
syzbot engineers can be reached at syzkaller@googlegroups.com.

syzbot will keep track of this issue. See:
https://goo.gl/tpsmEJ#status for how to communicate with syzbot.

Re: [syzbot] KCSAN: data-race in __io_uring_cancel / io_uring_try_cancel_requests

[Marco Elver]

On Wed, May 26, 2021 at 08:44AM -0700, syzbot wrote:
> Hello,
> 
> syzbot found the following issue on:
> 
> HEAD commit:    a050a6d2 Merge tag 'perf-tools-fixes-for-v5.13-2021-05-24'..
> git tree:       upstream
> console output: https://syzkaller.appspot.com/x/log.txt?x=13205087d00000
> kernel config:  https://syzkaller.appspot.com/x/.config?x=3bcc8a6b51ef8094
> dashboard link: https://syzkaller.appspot.com/bug?extid=73554e2258b7b8bf0bbf
> compiler:       Debian clang version 11.0.1-2
[...]
> write to 0xffff88811d8df330 of 8 bytes by task 3709 on cpu 1:
>  io_uring_clean_tctx fs/io_uring.c:9042 [inline]
>  __io_uring_cancel+0x261/0x3b0 fs/io_uring.c:9136
>  io_uring_files_cancel include/linux/io_uring.h:16 [inline]
>  do_exit+0x185/0x1560 kernel/exit.c:781
>  do_group_exit+0xce/0x1a0 kernel/exit.c:923
>  get_signal+0xfc3/0x1610 kernel/signal.c:2835
>  arch_do_signal_or_restart+0x2a/0x220 arch/x86/kernel/signal.c:789
>  handle_signal_work kernel/entry/common.c:147 [inline]
>  exit_to_user_mode_loop kernel/entry/common.c:171 [inline]
>  exit_to_user_mode_prepare+0x109/0x190 kernel/entry/common.c:208
>  __syscall_exit_to_user_mode_work kernel/entry/common.c:290 [inline]
>  syscall_exit_to_user_mode+0x20/0x40 kernel/entry/common.c:301
>  do_syscall_64+0x56/0x90 arch/x86/entry/common.c:57
>  entry_SYSCALL_64_after_hwframe+0x44/0xae
> 
> read to 0xffff88811d8df330 of 8 bytes by task 6412 on cpu 0:
>  io_uring_try_cancel_iowq fs/io_uring.c:8911 [inline]
>  io_uring_try_cancel_requests+0x1ce/0x8e0 fs/io_uring.c:8933
>  io_ring_exit_work+0x7c/0x1110 fs/io_uring.c:8736
>  process_one_work+0x3e9/0x8f0 kernel/workqueue.c:2276
>  worker_thread+0x636/0xae0 kernel/workqueue.c:2422
>  kthread+0x1d0/0x1f0 kernel/kthread.c:313
>  ret_from_fork+0x1f/0x30 arch/x86/entry/entry_64.S:294

I wasn't entirely sure if io_wq is guaranteed to remain live in this
case in io_uring_try_cancel_iowq(), but the comment there suggests it
does. In that case, I think the below patch would explain the situation
better and also propose a fix.

Thoughts?

Thanks,
-- Marco

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

From: Marco Elver <elver@google.com>
Date: Wed, 26 May 2021 16:56:37 +0200
Subject: [PATCH] io_uring: fix data race to avoid potential NULL-deref

Commit ba5ef6dc8a82 ("io_uring: fortify tctx/io_wq cleanup") introduced
setting tctx->io_wq to NULL a bit earlier. This has caused KCSAN to
detect a data race between between accesses to tctx->io_wq:

  write to 0xffff88811d8df330 of 8 bytes by task 3709 on cpu 1:
   io_uring_clean_tctx                  fs/io_uring.c:9042 [inline]
   __io_uring_cancel                    fs/io_uring.c:9136
   io_uring_files_cancel                include/linux/io_uring.h:16 [inline]
   do_exit                              kernel/exit.c:781
   do_group_exit                        kernel/exit.c:923
   get_signal                           kernel/signal.c:2835
   arch_do_signal_or_restart            arch/x86/kernel/signal.c:789
   handle_signal_work                   kernel/entry/common.c:147 [inline]
   exit_to_user_mode_loop               kernel/entry/common.c:171 [inline]
   ...
  read to 0xffff88811d8df330 of 8 bytes by task 6412 on cpu 0:
   io_uring_try_cancel_iowq             fs/io_uring.c:8911 [inline]
   io_uring_try_cancel_requests         fs/io_uring.c:8933
   io_ring_exit_work                    fs/io_uring.c:8736
   process_one_work                     kernel/workqueue.c:2276
   ...

With the config used, KCSAN only reports data races with value changes:
this implies that in the case here we also know that tctx->io_wq was
non-NULL. Therefore, depending on interleaving, we may end up with:

              [CPU 0]                 |        [CPU 1]
  io_uring_try_cancel_iowq()          | io_uring_clean_tctx()
    if (!tctx->io_wq) // false        |   ...
    ...                               |   tctx->io_wq = NULL
    io_wq_cancel_cb(tctx->io_wq, ...) |   ...
      -> NULL-deref                   |

Note: It is likely that thus far we've gotten lucky and the compiler
optimizes the double-read into a single read into a register -- but this
is never guaranteed, and can easily change with a different config!

Fix the data race by atomically accessing tctx->io_wq. Of course, this
assumes that a valid io_wq remains alive for the duration of
io_uring_try_cancel_iowq(), which should be the case per comment there.

Reported-by: syzbot+bf2b3d0435b9b728946c@syzkaller.appspotmail.com
Signed-off-by: Marco Elver <elver@google.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Pavel Begunkov <asml.silence@gmail.com>
---
 fs/io_uring.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/fs/io_uring.c b/fs/io_uring.c
index 5f82954004f6..c7e27b464cb6 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -8903,12 +8903,16 @@ static bool io_uring_try_cancel_iowq(struct io_ring_ctx *ctx)
 	mutex_lock(&ctx->uring_lock);
 	list_for_each_entry(node, &ctx->tctx_list, ctx_node) {
 		struct io_uring_task *tctx = node->task->io_uring;
+		struct io_wq *io_wq;
 
+		if (!tctx)
+			continue;
 		/*
 		 * io_wq will stay alive while we hold uring_lock, because it's
 		 * killed after ctx nodes, which requires to take the lock.
 		 */
-		if (!tctx || !tctx->io_wq)
+		io_wq = READ_ONCE(tctx->io_wq);
+		if (!io_wq)
 			continue;
 		cret = io_wq_cancel_cb(tctx->io_wq, io_cancel_ctx_cb, ctx, true);
 		ret |= (cret != IO_WQ_CANCEL_NOTFOUND);
@@ -9039,7 +9043,7 @@ static void io_uring_clean_tctx(struct io_uring_task *tctx)
 	struct io_tctx_node *node;
 	unsigned long index;
 
-	tctx->io_wq = NULL;
+	WRITE_ONCE(tctx->io_wq, NULL);
 	xa_for_each(&tctx->xa, index, node)
 		io_uring_del_task_file(index);
 	if (wq)
-- 
2.31.1.818.g46aad6cb9e-goog

Re: [syzbot] KCSAN: data-race in __io_uring_cancel / io_uring_try_cancel_requests

[Marco Elver]

On Wed, May 26, 2021 at 05:48PM +0200, Marco Elver wrote:
> On Wed, May 26, 2021 at 08:44AM -0700, syzbot wrote:
> > Hello,
> > 
> > syzbot found the following issue on:
> > 
> > HEAD commit:    a050a6d2 Merge tag 'perf-tools-fixes-for-v5.13-2021-05-24'..
> > git tree:       upstream
> > console output: https://syzkaller.appspot.com/x/log.txt?x=13205087d00000
> > kernel config:  https://syzkaller.appspot.com/x/.config?x=3bcc8a6b51ef8094
> > dashboard link: https://syzkaller.appspot.com/bug?extid=73554e2258b7b8bf0bbf
> > compiler:       Debian clang version 11.0.1-2
> [...]
> > write to 0xffff88811d8df330 of 8 bytes by task 3709 on cpu 1:
> >  io_uring_clean_tctx fs/io_uring.c:9042 [inline]
> >  __io_uring_cancel+0x261/0x3b0 fs/io_uring.c:9136
> >  io_uring_files_cancel include/linux/io_uring.h:16 [inline]
> >  do_exit+0x185/0x1560 kernel/exit.c:781
> >  do_group_exit+0xce/0x1a0 kernel/exit.c:923
> >  get_signal+0xfc3/0x1610 kernel/signal.c:2835
> >  arch_do_signal_or_restart+0x2a/0x220 arch/x86/kernel/signal.c:789
> >  handle_signal_work kernel/entry/common.c:147 [inline]
> >  exit_to_user_mode_loop kernel/entry/common.c:171 [inline]
> >  exit_to_user_mode_prepare+0x109/0x190 kernel/entry/common.c:208
> >  __syscall_exit_to_user_mode_work kernel/entry/common.c:290 [inline]
> >  syscall_exit_to_user_mode+0x20/0x40 kernel/entry/common.c:301
> >  do_syscall_64+0x56/0x90 arch/x86/entry/common.c:57
> >  entry_SYSCALL_64_after_hwframe+0x44/0xae
> > 
> > read to 0xffff88811d8df330 of 8 bytes by task 6412 on cpu 0:
> >  io_uring_try_cancel_iowq fs/io_uring.c:8911 [inline]
> >  io_uring_try_cancel_requests+0x1ce/0x8e0 fs/io_uring.c:8933
> >  io_ring_exit_work+0x7c/0x1110 fs/io_uring.c:8736
> >  process_one_work+0x3e9/0x8f0 kernel/workqueue.c:2276
> >  worker_thread+0x636/0xae0 kernel/workqueue.c:2422
> >  kthread+0x1d0/0x1f0 kernel/kthread.c:313
> >  ret_from_fork+0x1f/0x30 arch/x86/entry/entry_64.S:294
> 
> I wasn't entirely sure if io_wq is guaranteed to remain live in this
> case in io_uring_try_cancel_iowq(), but the comment there suggests it
> does. In that case, I think the below patch would explain the situation
> better and also propose a fix.
> 
> Thoughts?

Due to some moving around of code, the patch lost the actual fix (using
atomically read io_wq) -- so here it is again ... hopefully as intended.
:-)

Thanks,
-- Marco

From: Marco Elver <elver@google.com>
Date: Wed, 26 May 2021 16:56:37 +0200
Subject: [PATCH] io_uring: fix data race to avoid potential NULL-deref

Commit ba5ef6dc8a82 ("io_uring: fortify tctx/io_wq cleanup") introduced
setting tctx->io_wq to NULL a bit earlier. This has caused KCSAN to
detect a data race between between accesses to tctx->io_wq:

  write to 0xffff88811d8df330 of 8 bytes by task 3709 on cpu 1:
   io_uring_clean_tctx                  fs/io_uring.c:9042 [inline]
   __io_uring_cancel                    fs/io_uring.c:9136
   io_uring_files_cancel                include/linux/io_uring.h:16 [inline]
   do_exit                              kernel/exit.c:781
   do_group_exit                        kernel/exit.c:923
   get_signal                           kernel/signal.c:2835
   arch_do_signal_or_restart            arch/x86/kernel/signal.c:789
   handle_signal_work                   kernel/entry/common.c:147 [inline]
   exit_to_user_mode_loop               kernel/entry/common.c:171 [inline]
   ...
  read to 0xffff88811d8df330 of 8 bytes by task 6412 on cpu 0:
   io_uring_try_cancel_iowq             fs/io_uring.c:8911 [inline]
   io_uring_try_cancel_requests         fs/io_uring.c:8933
   io_ring_exit_work                    fs/io_uring.c:8736
   process_one_work                     kernel/workqueue.c:2276
   ...

With the config used, KCSAN only reports data races with value changes:
this implies that in the case here we also know that tctx->io_wq was
non-NULL. Therefore, depending on interleaving, we may end up with:

              [CPU 0]                 |        [CPU 1]
  io_uring_try_cancel_iowq()          | io_uring_clean_tctx()
    if (!tctx->io_wq) // false        |   ...
    ...                               |   tctx->io_wq = NULL
    io_wq_cancel_cb(tctx->io_wq, ...) |   ...
      -> NULL-deref                   |

Note: It is likely that thus far we've gotten lucky and the compiler
optimizes the double-read into a single read into a register -- but this
is never guaranteed, and can easily change with a different config!

Fix the data race by atomically accessing tctx->io_wq. Of course, this
assumes that a valid io_wq remains alive for the duration of
io_uring_try_cancel_iowq(), which should be the case per comment there.

Reported-by: syzbot+bf2b3d0435b9b728946c@syzkaller.appspotmail.com
Signed-off-by: Marco Elver <elver@google.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Pavel Begunkov <asml.silence@gmail.com>
---
 fs/io_uring.c | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)

diff --git a/fs/io_uring.c b/fs/io_uring.c
index 5f82954004f6..e681ece1bbca 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -8903,14 +8903,18 @@ static bool io_uring_try_cancel_iowq(struct io_ring_ctx *ctx)
 	mutex_lock(&ctx->uring_lock);
 	list_for_each_entry(node, &ctx->tctx_list, ctx_node) {
 		struct io_uring_task *tctx = node->task->io_uring;
+		struct io_wq *io_wq;
 
+		if (!tctx)
+			continue;
 		/*
 		 * io_wq will stay alive while we hold uring_lock, because it's
 		 * killed after ctx nodes, which requires to take the lock.
 		 */
-		if (!tctx || !tctx->io_wq)
+		io_wq = READ_ONCE(tctx->io_wq);
+		if (!io_wq)
 			continue;
-		cret = io_wq_cancel_cb(tctx->io_wq, io_cancel_ctx_cb, ctx, true);
+		cret = io_wq_cancel_cb(io_wq, io_cancel_ctx_cb, ctx, true);
 		ret |= (cret != IO_WQ_CANCEL_NOTFOUND);
 	}
 	mutex_unlock(&ctx->uring_lock);
@@ -9039,7 +9043,7 @@ static void io_uring_clean_tctx(struct io_uring_task *tctx)
 	struct io_tctx_node *node;
 	unsigned long index;
 
-	tctx->io_wq = NULL;
+	WRITE_ONCE(tctx->io_wq, NULL);
 	xa_for_each(&tctx->xa, index, node)
 		io_uring_del_task_file(index);
 	if (wq)
-- 
2.31.1.818.g46aad6cb9e-goog

Re: [syzbot] KCSAN: data-race in __io_uring_cancel / io_uring_try_cancel_requests

[Pavel Begunkov]

On 5/26/21 4:52 PM, Marco Elver wrote:
> Due to some moving around of code, the patch lost the actual fix (using
> atomically read io_wq) -- so here it is again ... hopefully as intended.
> :-)

"fortify" damn it... It was synchronised with &ctx->uring_lock
before, see io_uring_try_cancel_iowq() and io_uring_del_tctx_node(),
so should not clear before *del_tctx_node()

The fix should just move it after this sync point. Will you send
it out as a patch?

diff --git a/fs/io_uring.c b/fs/io_uring.c
index 7db6aaf31080..b76ba26b4c6c 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -9075,11 +9075,12 @@ static void io_uring_clean_tctx(struct io_uring_task *tctx)
 	struct io_tctx_node *node;
 	unsigned long index;
 
-	tctx->io_wq = NULL;
 	xa_for_each(&tctx->xa, index, node)
 		io_uring_del_tctx_node(index);
-	if (wq)
+	if (wq) {
+		tctx->io_wq = NULL;
 		io_wq_put_and_exit(wq);
+	}
 }
 
 static s64 tctx_inflight(struct io_uring_task *tctx, bool tracked)

 
-- 
Pavel Begunkov

Re: [syzbot] KCSAN: data-race in __io_uring_cancel / io_uring_try_cancel_requests

[Pavel Begunkov]

On 5/26/21 5:29 PM, Pavel Begunkov wrote:
> On 5/26/21 4:52 PM, Marco Elver wrote:
>> Due to some moving around of code, the patch lost the actual fix (using
>> atomically read io_wq) -- so here it is again ... hopefully as intended.
>> :-)
> 
> "fortify" damn it...

fwiw, it's a reference to my own commit that came after -rc

-- 
Pavel Begunkov

Re: [syzbot] KCSAN: data-race in __io_uring_cancel / io_uring_try_cancel_requests

[Marco Elver]

On Wed, 26 May 2021 at 18:29, Pavel Begunkov <asml.silence@gmail.com> wrote:
> On 5/26/21 4:52 PM, Marco Elver wrote:
> > Due to some moving around of code, the patch lost the actual fix (using
> > atomically read io_wq) -- so here it is again ... hopefully as intended.
> > :-)
>
> "fortify" damn it... It was synchronised with &ctx->uring_lock
> before, see io_uring_try_cancel_iowq() and io_uring_del_tctx_node(),
> so should not clear before *del_tctx_node()

Ah, so if I understand right, the property stated by the comment in
io_uring_try_cancel_iowq() was broken, and your patch below would fix
that, right?

> The fix should just move it after this sync point. Will you send
> it out as a patch?

Do you mean your move of write to io_wq goes on top of the patch I
proposed? (If so, please also leave your Signed-of-by so I can squash
it.)

So if I understand right, we do in fact have 2 problems:
1. the data race as I noted in my patch, and
2. the fact that io_wq does not live long enough.

Did I get it right?

Thanks,
-- Marco

> diff --git a/fs/io_uring.c b/fs/io_uring.c
> index 7db6aaf31080..b76ba26b4c6c 100644
> --- a/fs/io_uring.c
> +++ b/fs/io_uring.c
> @@ -9075,11 +9075,12 @@ static void io_uring_clean_tctx(struct io_uring_task *tctx)
>         struct io_tctx_node *node;
>         unsigned long index;
>
> -       tctx->io_wq = NULL;
>         xa_for_each(&tctx->xa, index, node)
>                 io_uring_del_tctx_node(index);
> -       if (wq)
> +       if (wq) {
> +               tctx->io_wq = NULL;
>                 io_wq_put_and_exit(wq);
> +       }
>  }
>
>  static s64 tctx_inflight(struct io_uring_task *tctx, bool tracked)
>
>
> --
> Pavel Begunkov

Re: [syzbot] KCSAN: data-race in __io_uring_cancel / io_uring_try_cancel_requests

[Pavel Begunkov]

On 5/26/21 5:36 PM, Marco Elver wrote:
> On Wed, 26 May 2021 at 18:29, Pavel Begunkov <asml.silence@gmail.com> wrote:
>> On 5/26/21 4:52 PM, Marco Elver wrote:
>>> Due to some moving around of code, the patch lost the actual fix (using
>>> atomically read io_wq) -- so here it is again ... hopefully as intended.
>>> :-)
>>
>> "fortify" damn it... It was synchronised with &ctx->uring_lock
>> before, see io_uring_try_cancel_iowq() and io_uring_del_tctx_node(),
>> so should not clear before *del_tctx_node()
> 
> Ah, so if I understand right, the property stated by the comment in
> io_uring_try_cancel_iowq() was broken, and your patch below would fix
> that, right?

"io_uring: fortify tctx/io_wq cleanup" broke it and the diff
should fix it.

>> The fix should just move it after this sync point. Will you send
>> it out as a patch?
> 
> Do you mean your move of write to io_wq goes on top of the patch I
> proposed? (If so, please also leave your Signed-of-by so I can squash
> it.)

No, only my diff, but you hinted on what has happened, so I would
prefer you to take care of patching. If you want of course.

To be entirely fair, assuming that aligned ptr
reads can't be torn, I don't see any _real_ problem. But surely
the report is very helpful and the current state is too wonky, so
should be patched.

TL;DR;
The synchronisation goes as this: it's usually used by the owner
task, and the owner task deletes it, so is mostly naturally
synchronised. An exception is a worker (not only) that accesses
it for cancellation purpose, but it uses it only under ->uring_lock,
so if removal is also taking the lock it should be fine. see
io_uring_del_tctx_node() locking.

> 
> So if I understand right, we do in fact have 2 problems:
> 1. the data race as I noted in my patch, and

Yes, and it deals with it

> 2. the fact that io_wq does not live long enough.

Nope, io_wq outlives them fine. 

> Did I get it right?
> 
>> diff --git a/fs/io_uring.c b/fs/io_uring.c
>> index 7db6aaf31080..b76ba26b4c6c 100644
>> --- a/fs/io_uring.c
>> +++ b/fs/io_uring.c
>> @@ -9075,11 +9075,12 @@ static void io_uring_clean_tctx(struct io_uring_task *tctx)
>>         struct io_tctx_node *node;
>>         unsigned long index;
>>
>> -       tctx->io_wq = NULL;
>>         xa_for_each(&tctx->xa, index, node)
>>                 io_uring_del_tctx_node(index);
>> -       if (wq)
>> +       if (wq) {
>> +               tctx->io_wq = NULL;
>>                 io_wq_put_and_exit(wq);
>> +       }
>>  }
>>
>>  static s64 tctx_inflight(struct io_uring_task *tctx, bool tracked)

-- 
Pavel Begunkov

Re: [syzbot] KCSAN: data-race in __io_uring_cancel / io_uring_try_cancel_requests

[Marco Elver]

On Wed, May 26, 2021 at 09:31PM +0100, Pavel Begunkov wrote:
> On 5/26/21 5:36 PM, Marco Elver wrote:
> > On Wed, 26 May 2021 at 18:29, Pavel Begunkov <asml.silence@gmail.com> wrote:
> >> On 5/26/21 4:52 PM, Marco Elver wrote:
> >>> Due to some moving around of code, the patch lost the actual fix (using
> >>> atomically read io_wq) -- so here it is again ... hopefully as intended.
> >>> :-)
> >>
> >> "fortify" damn it... It was synchronised with &ctx->uring_lock
> >> before, see io_uring_try_cancel_iowq() and io_uring_del_tctx_node(),
> >> so should not clear before *del_tctx_node()
> > 
> > Ah, so if I understand right, the property stated by the comment in
> > io_uring_try_cancel_iowq() was broken, and your patch below would fix
> > that, right?
> 
> "io_uring: fortify tctx/io_wq cleanup" broke it and the diff
> should fix it.
> 
> >> The fix should just move it after this sync point. Will you send
> >> it out as a patch?
> > 
> > Do you mean your move of write to io_wq goes on top of the patch I
> > proposed? (If so, please also leave your Signed-of-by so I can squash
> > it.)
> 
> No, only my diff, but you hinted on what has happened, so I would
> prefer you to take care of patching. If you want of course.
> 
> To be entirely fair, assuming that aligned ptr
> reads can't be torn, I don't see any _real_ problem. But surely
> the report is very helpful and the current state is too wonky, so
> should be patched.

In the current version, it is a problem if we end up with a double-read,
as it is in the current C code. The compiler might of course optimize
it into 1 read into a register.

Tangent: I avoid reasoning in terms of compiler optimizations where
I can. :-) It's is a slippery slope if the code in question isn't
tolerant to data races by design (examples are stats counting, or other
heuristics -- in the case here that's certainly not the case).
Therefore, my wish is that we really ought to resolve as many data races
as we can (+ mark intentional ones appropriately). Also, so that we're
left with only the interesting cases like in the case here.  (More
background if you're interested: https://lwn.net/Articles/816850/)

The problem here, however, has a nicer resolution as you suggested.

> TL;DR;
> The synchronisation goes as this: it's usually used by the owner
> task, and the owner task deletes it, so is mostly naturally
> synchronised. An exception is a worker (not only) that accesses
> it for cancellation purpose, but it uses it only under ->uring_lock,
> so if removal is also taking the lock it should be fine. see
> io_uring_del_tctx_node() locking.

Did you mean io_uring_del_task_file()? There is no
io_uring_del_tctx_node().

> > So if I understand right, we do in fact have 2 problems:
> > 1. the data race as I noted in my patch, and
> 
> Yes, and it deals with it
> 
> > 2. the fact that io_wq does not live long enough.
> 
> Nope, io_wq outlives them fine. 

I've sent:
https://lkml.kernel.org/r/20210527092547.2656514-1-elver@google.com

Thanks,
-- Marco

Re: [syzbot] KCSAN: data-race in __io_uring_cancel / io_uring_try_cancel_requests

[Pavel Begunkov]

On 5/27/21 10:32 AM, Marco Elver wrote:
> On Wed, May 26, 2021 at 09:31PM +0100, Pavel Begunkov wrote:
>> On 5/26/21 5:36 PM, Marco Elver wrote:
>>> On Wed, 26 May 2021 at 18:29, Pavel Begunkov <asml.silence@gmail.com> wrote:
>>>> On 5/26/21 4:52 PM, Marco Elver wrote:
>>>>> Due to some moving around of code, the patch lost the actual fix (using
>>>>> atomically read io_wq) -- so here it is again ... hopefully as intended.
>>>>> :-)
>>>>
>>>> "fortify" damn it... It was synchronised with &ctx->uring_lock
>>>> before, see io_uring_try_cancel_iowq() and io_uring_del_tctx_node(),
>>>> so should not clear before *del_tctx_node()
>>>
>>> Ah, so if I understand right, the property stated by the comment in
>>> io_uring_try_cancel_iowq() was broken, and your patch below would fix
>>> that, right?
>>
>> "io_uring: fortify tctx/io_wq cleanup" broke it and the diff
>> should fix it.
>>
>>>> The fix should just move it after this sync point. Will you send
>>>> it out as a patch?
>>>
>>> Do you mean your move of write to io_wq goes on top of the patch I
>>> proposed? (If so, please also leave your Signed-of-by so I can squash
>>> it.)
>>
>> No, only my diff, but you hinted on what has happened, so I would
>> prefer you to take care of patching. If you want of course.
>>
>> To be entirely fair, assuming that aligned ptr
>> reads can't be torn, I don't see any _real_ problem. But surely
>> the report is very helpful and the current state is too wonky, so
>> should be patched.
> 
> In the current version, it is a problem if we end up with a double-read,
> as it is in the current C code. The compiler might of course optimize
> it into 1 read into a register.

Absolutely agree on that

> Tangent: I avoid reasoning in terms of compiler optimizations where
> I can. :-) It's is a slippery slope if the code in question isn't
> tolerant to data races by design (examples are stats counting, or other
> heuristics -- in the case here that's certainly not the case).
> Therefore, my wish is that we really ought to resolve as many data races
> as we can (+ mark intentional ones appropriately). Also, so that we're
> left with only the interesting cases like in the case here.  (More
> background if you're interested: https://lwn.net/Articles/816850/)
> 
> The problem here, however, has a nicer resolution as you suggested.
> 
>> TL;DR;
>> The synchronisation goes as this: it's usually used by the owner
>> task, and the owner task deletes it, so is mostly naturally
>> synchronised. An exception is a worker (not only) that accesses
>> it for cancellation purpose, but it uses it only under ->uring_lock,
>> so if removal is also taking the lock it should be fine. see
>> io_uring_del_tctx_node() locking.
> 
> Did you mean io_uring_del_task_file()? There is no
> io_uring_del_tctx_node().

Ah, yes, that's from patches I sent for next.

-- 
Pavel Begunkov