From: Jens Axboe <axboe@kernel.dk>
Date: Sat, 21 Jan 2023 17:39:22 +0000 (-0700)
Subject: io_uring: allow re-poll if we made progress
X-Git-Url: http://git.maquefel.me/?a=commitdiff_plain;h=a79b13f24967b68b49aa23eda5fa9eb14d07b3a4;p=linux.git

io_uring: allow re-poll if we made progress

commit 10c873334febaeea9aa0c25c10b5ac0951b77a5f upstream.

We currently check REQ_F_POLLED before arming async poll for a
notification to retry. If it's set, then we don't allow poll and will
punt to io-wq instead. This is done to prevent a situation where a buggy
driver will repeatedly return that there's space/data available yet we
get -EAGAIN.

However, if we already transferred data, then it should be safe to rely
on poll again. Gate the check on whether or not REQ_F_PARTIAL_IO is
also set.

Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---

diff --git a/io_uring/io_uring.c b/io_uring/io_uring.c
index 3fb76863fed4d..997a7264e1d4b 100644
--- a/io_uring/io_uring.c
+++ b/io_uring/io_uring.c
@@ -5853,7 +5853,7 @@ static int io_arm_poll_handler(struct io_kiocb *req)
 
 	if (!req->file || !file_can_poll(req->file))
 		return IO_APOLL_ABORTED;
-	if (req->flags & REQ_F_POLLED)
+	if ((req->flags & (REQ_F_POLLED|REQ_F_PARTIAL_IO)) == REQ_F_POLLED)
 		return IO_APOLL_ABORTED;
 	if (!def->pollin && !def->pollout)
 		return IO_APOLL_ABORTED;
@@ -5869,7 +5869,10 @@ static int io_arm_poll_handler(struct io_kiocb *req)
 		mask |= POLLOUT | POLLWRNORM;
 	}
 
-	apoll = kmalloc(sizeof(*apoll), GFP_ATOMIC);
+	if (req->flags & REQ_F_POLLED)
+		apoll = req->apoll;
+	else
+		apoll = kmalloc(sizeof(*apoll), GFP_ATOMIC);
 	if (unlikely(!apoll))
 		return IO_APOLL_ABORTED;
 	apoll->double_poll = NULL;