public inbox for [email protected]
 help / color / mirror / Atom feed
* [PATCH v2] io_uring: fix io_kiocb.flags modification race in IOPOLL mode
@ 2020-06-11 15:39 Xiaoguang Wang
  2020-06-11 15:46 ` Jens Axboe
  0 siblings, 1 reply; 2+ messages in thread
From: Xiaoguang Wang @ 2020-06-11 15:39 UTC (permalink / raw)
  To: io-uring; +Cc: axboe, asml.silence, joseph.qi, Xiaoguang Wang

While testing io_uring in arm, we found sometimes io_sq_thread() keeps
polling io requests even though there are not inflight io requests in
block layer. After some investigations, found a possible race about
io_kiocb.flags, see below race codes:
  1) in the end of io_write() or io_read()
    req->flags &= ~REQ_F_NEED_CLEANUP;
    kfree(iovec);
    return ret;

  2) in io_complete_rw_iopoll()
    if (res != -EAGAIN)
        req->flags |= REQ_F_IOPOLL_COMPLETED;

In IOPOLL mode, io requests still maybe completed by interrupt, then
above codes are not safe, concurrent modifications to req->flags, which
is not protected by lock or is not atomic modifications. I also had
disassemble io_complete_rw_iopoll() in arm:
   req->flags |= REQ_F_IOPOLL_COMPLETED;
   0xffff000008387b18 <+76>:    ldr     w0, [x19,#104]
   0xffff000008387b1c <+80>:    orr     w0, w0, #0x1000
   0xffff000008387b20 <+84>:    str     w0, [x19,#104]

Seems that the "req->flags |= REQ_F_IOPOLL_COMPLETED;" is  load and
modification, two instructions, which obviously is not atomic.

To fix this issue, add a new iopoll_completed in io_kiocb to indicate
whether io request is completed.

Signed-off-by: Xiaoguang Wang <[email protected]>

---
V2:
  Use READ_ONCE/WRITE_ONCE while reading/setting iopoll_completed.
---
 fs/io_uring.c | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/fs/io_uring.c b/fs/io_uring.c
index 5b0249140ff5..61fca5afaac8 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -529,7 +529,6 @@ enum {
 	REQ_F_INFLIGHT_BIT,
 	REQ_F_CUR_POS_BIT,
 	REQ_F_NOWAIT_BIT,
-	REQ_F_IOPOLL_COMPLETED_BIT,
 	REQ_F_LINK_TIMEOUT_BIT,
 	REQ_F_TIMEOUT_BIT,
 	REQ_F_ISREG_BIT,
@@ -574,8 +573,6 @@ enum {
 	REQ_F_CUR_POS		= BIT(REQ_F_CUR_POS_BIT),
 	/* must not punt to workers */
 	REQ_F_NOWAIT		= BIT(REQ_F_NOWAIT_BIT),
-	/* polled IO has completed */
-	REQ_F_IOPOLL_COMPLETED	= BIT(REQ_F_IOPOLL_COMPLETED_BIT),
 	/* has linked timeout */
 	REQ_F_LINK_TIMEOUT	= BIT(REQ_F_LINK_TIMEOUT_BIT),
 	/* timeout request */
@@ -640,6 +637,8 @@ struct io_kiocb {
 	struct io_async_ctx		*io;
 	int				cflags;
 	u8				opcode;
+	/* polled IO has completed */
+	u8				iopoll_completed;
 
 	u16				buf_index;
 
@@ -1798,7 +1797,7 @@ static int io_do_iopoll(struct io_ring_ctx *ctx, unsigned int *nr_events,
 		 * If we find a request that requires polling, break out
 		 * and complete those lists first, if we have entries there.
 		 */
-		if (req->flags & REQ_F_IOPOLL_COMPLETED) {
+		if (READ_ONCE(req->iopoll_completed)) {
 			list_move_tail(&req->list, &done);
 			continue;
 		}
@@ -1979,7 +1978,7 @@ static void io_complete_rw_iopoll(struct kiocb *kiocb, long res, long res2)
 		req_set_fail_links(req);
 	req->result = res;
 	if (res != -EAGAIN)
-		req->flags |= REQ_F_IOPOLL_COMPLETED;
+		WRITE_ONCE(req->iopoll_completed, 1);
 }
 
 /*
@@ -2012,7 +2011,7 @@ static void io_iopoll_req_issued(struct io_kiocb *req)
 	 * For fast devices, IO may have already completed. If it has, add
 	 * it to the front so we find it first.
 	 */
-	if (req->flags & REQ_F_IOPOLL_COMPLETED)
+	if (READ_ONCE(req->iopoll_completed))
 		list_add(&req->list, &ctx->poll_list);
 	else
 		list_add_tail(&req->list, &ctx->poll_list);
@@ -2140,6 +2139,7 @@ static int io_prep_rw(struct io_kiocb *req, const struct io_uring_sqe *sqe,
 		kiocb->ki_flags |= IOCB_HIPRI;
 		kiocb->ki_complete = io_complete_rw_iopoll;
 		req->result = 0;
+		req->iopoll_completed = 0;
 	} else {
 		if (kiocb->ki_flags & IOCB_HIPRI)
 			return -EINVAL;
-- 
2.17.2


^ permalink raw reply related	[flat|nested] 2+ messages in thread

* Re: [PATCH v2] io_uring: fix io_kiocb.flags modification race in IOPOLL mode
  2020-06-11 15:39 [PATCH v2] io_uring: fix io_kiocb.flags modification race in IOPOLL mode Xiaoguang Wang
@ 2020-06-11 15:46 ` Jens Axboe
  0 siblings, 0 replies; 2+ messages in thread
From: Jens Axboe @ 2020-06-11 15:46 UTC (permalink / raw)
  To: Xiaoguang Wang, io-uring; +Cc: asml.silence, joseph.qi

On 6/11/20 9:39 AM, Xiaoguang Wang wrote:
> While testing io_uring in arm, we found sometimes io_sq_thread() keeps
> polling io requests even though there are not inflight io requests in
> block layer. After some investigations, found a possible race about
> io_kiocb.flags, see below race codes:
>   1) in the end of io_write() or io_read()
>     req->flags &= ~REQ_F_NEED_CLEANUP;
>     kfree(iovec);
>     return ret;
> 
>   2) in io_complete_rw_iopoll()
>     if (res != -EAGAIN)
>         req->flags |= REQ_F_IOPOLL_COMPLETED;
> 
> In IOPOLL mode, io requests still maybe completed by interrupt, then
> above codes are not safe, concurrent modifications to req->flags, which
> is not protected by lock or is not atomic modifications. I also had
> disassemble io_complete_rw_iopoll() in arm:
>    req->flags |= REQ_F_IOPOLL_COMPLETED;
>    0xffff000008387b18 <+76>:    ldr     w0, [x19,#104]
>    0xffff000008387b1c <+80>:    orr     w0, w0, #0x1000
>    0xffff000008387b20 <+84>:    str     w0, [x19,#104]
> 
> Seems that the "req->flags |= REQ_F_IOPOLL_COMPLETED;" is  load and
> modification, two instructions, which obviously is not atomic.
> 
> To fix this issue, add a new iopoll_completed in io_kiocb to indicate
> whether io request is completed.

Looks good, applied, thanks.

-- 
Jens Axboe


^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2020-06-11 15:46 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2020-06-11 15:39 [PATCH v2] io_uring: fix io_kiocb.flags modification race in IOPOLL mode Xiaoguang Wang
2020-06-11 15:46 ` Jens Axboe

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox