Skip to contentSkip to conversation

io_uring: remove double poll wait entry for pure poll

PATCH In review Hao Xu 3 participants 8 messages Jul 23, 2021 – Jul 28, 2021

Subsystems (2)
lore
Share & tools
lore

Hao Xu

author
#0 lore raw

PATCH v2 1/1 io_uring: remove double poll wait entry for pure poll

For pure poll requests, we should remove the double poll wait entry. And io_poll_remove_double() is good enough for it compared with io_poll_remove_waitqs().

---

v1-->v2
delete redundant io_poll_remove_double()

 fs/io_uring.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)
fs/io_uring.c +2 -3
index f2fe4eca150b..c5fe8b9e26b4 100644
@@ -4903,7 +4903,6 @@ static bool io_poll_complete(struct io_kiocb *req, __poll_t mask)
 	if (req->poll.events & EPOLLONESHOT)
 		flags = 0;
 	if (!io_cqring_fill_event(ctx, req->user_data, error, flags)) {
-		io_poll_remove_waitqs(req);
 		req->poll.done = true;
 		flags = 0;
 	}
@@ -4926,6 +4925,7 @@ static void io_poll_task_func(struct io_kiocb *req)
 
 		done = io_poll_complete(req, req->result);
 		if (done) {
+			io_poll_remove_double(req);
 			hash_del(&req->hash_node);
 		} else {
 			req->result = 0;
@@ -5113,7 +5113,7 @@ static __poll_t __io_arm_poll_handler(struct io_kiocb *req,
 		ipt->error = -EINVAL;
 
 	spin_lock_irq(&ctx->completion_lock);
-	if (ipt->error)
+	if (ipt->error || (mask && (poll->events & EPOLLONESHOT)))
 		io_poll_remove_double(req);
 	if (likely(poll->head)) {
 		spin_lock(&poll->head->lock);
@@ -5185,7 +5185,6 @@ static int io_arm_poll_handler(struct io_kiocb *req)
 	ret = __io_arm_poll_handler(req, &apoll->poll, &ipt, mask,
 					io_async_wake);
 	if (ret || ipt.error) {
-		io_poll_remove_double(req);
 		spin_unlock_irq(&ctx->completion_lock);
 		if (ret)
 			return IO_APOLL_READY;
Signature · 1 line
2.24.4
#1 lore raw
Quoted 3 lines · Hao Xu

For pure poll requests, we should remove the double poll wait entry. And io_poll_remove_double() is good enough for it compared with io_poll_remove_waitqs().

5.14 in the subject hints me that it's a fix. Is it? Can you add what it fixes or expand on why it's better?

Quoted 47 lines

---

v1-->v2
delete redundant io_poll_remove_double()

 fs/io_uring.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)
fs/io_uring.c +2 -3
index f2fe4eca150b..c5fe8b9e26b4 100644
@@ -4903,7 +4903,6 @@ static bool io_poll_complete(struct io_kiocb *req, __poll_t mask)
 	if (req->poll.events & EPOLLONESHOT)
 		flags = 0;
 	if (!io_cqring_fill_event(ctx, req->user_data, error, flags)) {
-		io_poll_remove_waitqs(req);
 		req->poll.done = true;
 		flags = 0;
 	}
@@ -4926,6 +4925,7 @@ static void io_poll_task_func(struct io_kiocb *req)
 
 		done = io_poll_complete(req, req->result);
 		if (done) {
+			io_poll_remove_double(req);
 			hash_del(&req->hash_node);
 		} else {
 			req->result = 0;
@@ -5113,7 +5113,7 @@ static __poll_t __io_arm_poll_handler(struct io_kiocb *req,
 		ipt->error = -EINVAL;
 
 	spin_lock_irq(&ctx->completion_lock);
-	if (ipt->error)
+	if (ipt->error || (mask && (poll->events & EPOLLONESHOT)))
 		io_poll_remove_double(req);
 	if (likely(poll->head)) {
 		spin_lock(&poll->head->lock);
@@ -5185,7 +5185,6 @@ static int io_arm_poll_handler(struct io_kiocb *req)
 	ret = __io_arm_poll_handler(req, &apoll->poll, &ipt, mask,
 					io_async_wake);
 	if (ret || ipt.error) {
-		io_poll_remove_double(req);
 		spin_unlock_irq(&ctx->completion_lock);
 		if (ret)
 			return IO_APOLL_READY;
Signature · 1 line
Pavel Begunkov
maintainer
#2 lore raw
Quoted 7 lines · Pavel Begunkov
Quoted 3 lines · Hao Xu

For pure poll requests, we should remove the double poll wait entry. And io_poll_remove_double() is good enough for it compared with io_poll_remove_waitqs().

5.14 in the subject hints me that it's a fix. Is it? Can you add what it fixes or expand on why it's better?

Ditto that, the commit message explains what is being done, it should explain _why_ it's being done. For the 'what' part you can read the code. So while the patch doesn't look wrong, I also can't quite tell why the change is necessary.

Signature · 1 line
Jens Axboe
author
#3 lore raw
In reply to Pavel Begunkov
Quoted 7 lines · Pavel Begunkov
Quoted 3 lines · Hao Xu

For pure poll requests, we should remove the double poll wait entry. And io_poll_remove_double() is good enough for it compared with io_poll_remove_waitqs().

5.14 in the subject hints me that it's a fix. Is it? Can you add what it fixes or expand on why it's better?

Hi Pavel, I found that for poll_add() requests, it doesn't remove the double poll wait entry when it's done, neither after vfs_poll() or in the poll completion handler. The patch is mainly to fix it.

Quoted 11 lines
Quoted 11 lines

---

v1-->v2
delete redundant io_poll_remove_double()

  fs/io_uring.c | 5 ++---
  1 file changed, 2 insertions(+), 3 deletions(-)
Quoted 9 lines

fs/io_uring.c static bool io_poll_complete(struct io_kiocb *req, __poll_t mask) · line 4906

Quoted 9 lines
fs/io_uring.c +0 -1
index f2fe4eca150b..c5fe8b9e26b4 100644
@@ -4903,7 +4903,6 @@ static bool io_poll_complete(struct io_kiocb *req, __poll_t mask)
  	if (req->poll.events & EPOLLONESHOT)
  		flags = 0;
  	if (!io_cqring_fill_event(ctx, req->user_data, error, flags)) {
-		io_poll_remove_waitqs(req);

Currently I only see it does that with io_poll_remove_waitqs() when cqring overflow and then ocqe allocation failed. Using io_poll_remove_waitqs() here is not very suitable since (1) it calls __io_poll_remove_one() which set poll->cancelled = true, why do we set poll->cancelled and poll->done to true at the same time though I think that doesn't cause any problem. (2) it does list_del_init(&poll->wait.entry) and hash_del(&req->hash_node) which has been already done.
Correct me if I'm wrong since I may misunderstand the code.

Regards,
Hao

Quoted 30 lines
Quoted 29 lines
fs/io_uring.c +2 -2
  		req->poll.done = true;
  		flags = 0;
  	}
@@ -4926,6 +4925,7 @@ static void io_poll_task_func(struct io_kiocb *req)
 
  		done = io_poll_complete(req, req->result);
  		if (done) {
+			io_poll_remove_double(req);
  			hash_del(&req->hash_node);
  		} else {
  			req->result = 0;
@@ -5113,7 +5113,7 @@ static __poll_t __io_arm_poll_handler(struct io_kiocb *req,
  		ipt->error = -EINVAL;
 
  	spin_lock_irq(&ctx->completion_lock);
-	if (ipt->error)
+	if (ipt->error || (mask && (poll->events & EPOLLONESHOT)))
  		io_poll_remove_double(req);
  	if (likely(poll->head)) {
  		spin_lock(&poll->head->lock);
@@ -5185,7 +5185,6 @@ static int io_arm_poll_handler(struct io_kiocb *req)
  	ret = __io_arm_poll_handler(req, &apoll->poll, &ipt, mask,
  					io_async_wake);
  	if (ret || ipt.error) {
-		io_poll_remove_double(req);
  		spin_unlock_irq(&ctx->completion_lock);
  		if (ret)
  			return IO_APOLL_READY;
#4 lore raw
Quoted 11 lines · Hao Xu
Quoted 7 lines · Pavel Begunkov
Quoted 3 lines · Hao Xu

For pure poll requests, we should remove the double poll wait entry. And io_poll_remove_double() is good enough for it compared with io_poll_remove_waitqs().

5.14 in the subject hints me that it's a fix. Is it? Can you add what it fixes or expand on why it's better?

Hi Pavel, I found that for poll_add() requests, it doesn't remove the double poll wait entry when it's done, neither after vfs_poll() or in the poll completion handler. The patch is mainly to fix it.

Ok, sounds good. Please resend with updated description, and let's add some tags.

Fixes 88e41cf928a6 ("io_uring: add multishot mode for IORING_OP_POLL_ADD")

Also, I'd prefer the commit title to make more clear that it's a fix. E.g. "io_uring: fix poll requests leaking second poll entries".

Btw, seems it should fix hangs in ./poll-mshot-update

Quoted 60 lines
Quoted 10 lines
Quoted 10 lines

---

v1-->v2
   delete redundant io_poll_remove_double()

  fs/io_uring.c | 5 ++---

  1 file changed, 2 insertions(+), 3 deletions(-)
Quoted 9 lines

fs/io_uring.c static bool io_poll_complete(struct io_kiocb *req, __poll_t mask) · line 4906

Quoted 9 lines
fs/io_uring.c +0 -1
index f2fe4eca150b..c5fe8b9e26b4 100644
@@ -4903,7 +4903,6 @@ static bool io_poll_complete(struct io_kiocb *req, __poll_t mask)
       if (req->poll.events & EPOLLONESHOT)
           flags = 0;
       if (!io_cqring_fill_event(ctx, req->user_data, error, flags)) {
-        io_poll_remove_waitqs(req);

Currently I only see it does that with io_poll_remove_waitqs() when cqring overflow and then ocqe allocation failed. Using io_poll_remove_waitqs() here is not very suitable since (1) it calls __io_poll_remove_one() which set poll->cancelled = true, why do we set poll->cancelled and poll->done to true at the same time though I think that doesn't cause any problem. (2) it does list_del_init(&poll->wait.entry) and hash_del(&req->hash_node) which has been already done.
Correct me if I'm wrong since I may misunderstand the code.

Regards,
Hao

Quoted 28 lines
Quoted 27 lines
fs/io_uring.c +2 -2
           req->poll.done = true;
           flags = 0;
       }
@@ -4926,6 +4925,7 @@ static void io_poll_task_func(struct io_kiocb *req)
             done = io_poll_complete(req, req->result);
           if (done) {
+            io_poll_remove_double(req);
               hash_del(&req->hash_node);
           } else {
               req->result = 0;
@@ -5113,7 +5113,7 @@ static __poll_t __io_arm_poll_handler(struct io_kiocb *req,
           ipt->error = -EINVAL;
         spin_lock_irq(&ctx->completion_lock);
-    if (ipt->error)
+    if (ipt->error || (mask && (poll->events & EPOLLONESHOT)))
           io_poll_remove_double(req);
       if (likely(poll->head)) {
           spin_lock(&poll->head->lock);
@@ -5185,7 +5185,6 @@ static int io_arm_poll_handler(struct io_kiocb *req)
       ret = __io_arm_poll_handler(req, &apoll->poll, &ipt, mask,
                       io_async_wake);
       if (ret || ipt.error) {
-        io_poll_remove_double(req);
           spin_unlock_irq(&ctx->completion_lock);
           if (ret)
               return IO_APOLL_READY;
Signature · 1 line
Pavel Begunkov
author
#5 lore raw
Quoted 16 lines · Pavel Begunkov
Quoted 11 lines · Hao Xu
Quoted 7 lines · Pavel Begunkov
Quoted 3 lines · Hao Xu

For pure poll requests, we should remove the double poll wait entry. And io_poll_remove_double() is good enough for it compared with io_poll_remove_waitqs().

5.14 in the subject hints me that it's a fix. Is it? Can you add what it fixes or expand on why it's better?

Hi Pavel, I found that for poll_add() requests, it doesn't remove the double poll wait entry when it's done, neither after vfs_poll() or in the poll completion handler. The patch is mainly to fix it.

Ok, sounds good. Please resend with updated description, and let's add some tags.

Quoted 7 lines
Fixes 88e41cf928a6 ("io_uring: add multishot mode for IORING_OP_POLL_ADD")

Also, I'd prefer the commit title to make more clear that it's a fix. E.g. "io_uring: fix poll requests leaking second poll entries".

Btw, seems it should fix hangs in ./poll-mshot-update

Sure,I'll send v3 soon, sorry for my unprofessionalism..

Quoted 63 lines
Quoted 62 lines
Quoted 12 lines
Quoted 12 lines

---

v1-->v2
   delete redundant io_poll_remove_double()

   fs/io_uring.c | 5 ++---
   1 file changed, 2 insertions(+), 3 deletions(-)
Quoted 9 lines

fs/io_uring.c static bool io_poll_complete(struct io_kiocb *req, __poll_t mask) · line 4906

Quoted 9 lines
fs/io_uring.c +0 -1
index f2fe4eca150b..c5fe8b9e26b4 100644
@@ -4903,7 +4903,6 @@ static bool io_poll_complete(struct io_kiocb *req, __poll_t mask)
       if (req->poll.events & EPOLLONESHOT)
           flags = 0;
       if (!io_cqring_fill_event(ctx, req->user_data, error, flags)) {
-        io_poll_remove_waitqs(req);

Currently I only see it does that with io_poll_remove_waitqs() when cqring overflow and then ocqe allocation failed. Using io_poll_remove_waitqs() here is not very suitable since (1) it calls __io_poll_remove_one() which set poll->cancelled = true, why do we set poll->cancelled and poll->done to true at the same time though I think that doesn't cause any problem. (2) it does list_del_init(&poll->wait.entry) and hash_del(&req->hash_node) which has been already done.
Correct me if I'm wrong since I may misunderstand the code.

Regards,
Hao

Quoted 28 lines
Quoted 27 lines
fs/io_uring.c +2 -2
           req->poll.done = true;
           flags = 0;
       }
@@ -4926,6 +4925,7 @@ static void io_poll_task_func(struct io_kiocb *req)
             done = io_poll_complete(req, req->result);
           if (done) {
+            io_poll_remove_double(req);
               hash_del(&req->hash_node);
           } else {
               req->result = 0;
@@ -5113,7 +5113,7 @@ static __poll_t __io_arm_poll_handler(struct io_kiocb *req,
           ipt->error = -EINVAL;
         spin_lock_irq(&ctx->completion_lock);
-    if (ipt->error)
+    if (ipt->error || (mask && (poll->events & EPOLLONESHOT)))
           io_poll_remove_double(req);
       if (likely(poll->head)) {
           spin_lock(&poll->head->lock);
@@ -5185,7 +5185,6 @@ static int io_arm_poll_handler(struct io_kiocb *req)
       ret = __io_arm_poll_handler(req, &apoll->poll, &ipt, mask,
                       io_async_wake);
       if (ret || ipt.error) {
-        io_poll_remove_double(req);
           spin_unlock_irq(&ctx->completion_lock);
           if (ret)
               return IO_APOLL_READY;
maintainer
#6 lore raw
Quoted 17 lines · Hao Xu
Quoted 16 lines · Pavel Begunkov
Quoted 11 lines · Hao Xu
Quoted 7 lines · Pavel Begunkov
Quoted 3 lines · Hao Xu

For pure poll requests, we should remove the double poll wait entry. And io_poll_remove_double() is good enough for it compared with io_poll_remove_waitqs().

5.14 in the subject hints me that it's a fix. Is it? Can you add what it fixes or expand on why it's better?

Hi Pavel, I found that for poll_add() requests, it doesn't remove the double poll wait entry when it's done, neither after vfs_poll() or in the poll completion handler. The patch is mainly to fix it.

Ok, sounds good. Please resend with updated description, and let's add some tags.

Quoted 8 lines
Quoted 7 lines
Fixes 88e41cf928a6 ("io_uring: add multishot mode for IORING_OP_POLL_ADD")

Also, I'd prefer the commit title to make more clear that it's a fix. E.g. "io_uring: fix poll requests leaking second poll entries".

Btw, seems it should fix hangs in ./poll-mshot-update

Sure,I'll send v3 soon, sorry for my unprofessionalism..

Are you going to send out v3?

Signature · 1 line
Jens Axboe
author
#7 lore raw
Quoted 21 lines · Jens Axboe
Quoted 20 lines · Hao Xu
Quoted 19 lines · Pavel Begunkov
Quoted 11 lines · Hao Xu
Quoted 7 lines · Pavel Begunkov
Quoted 3 lines · Hao Xu

For pure poll requests, we should remove the double poll wait entry. And io_poll_remove_double() is good enough for it compared with io_poll_remove_waitqs().

5.14 in the subject hints me that it's a fix. Is it? Can you add what it fixes or expand on why it's better?

Hi Pavel, I found that for poll_add() requests, it doesn't remove the double poll wait entry when it's done, neither after vfs_poll() or in the poll completion handler. The patch is mainly to fix it.

Ok, sounds good. Please resend with updated description, and let's add some tags.

Fixes 88e41cf928a6 ("io_uring: add multishot mode for IORING_OP_POLL_ADD")
Quoted 8 lines
Quoted 5 lines
Quoted 4 lines

Also, I'd prefer the commit title to make more clear that it's a fix. E.g. "io_uring: fix poll requests leaking second poll entries".

Btw, seems it should fix hangs in ./poll-mshot-update

Sure,I'll send v3 soon, sorry for my unprofessionalism..

Are you going to send out v3?

v3 sent. Btw I'm working on letting fast poll support multishot, I believe that will benefit non-persistent programming, let's see if it helps accept().