pespin has uploaded this change for review.
io_uring: RECVMSG_SENDMSG: Reset fd to blocking after osmo_fd_register() workaround
When IOFD_FLAG_NOTIFY_CONNECTED is requested by the user through
osmo_iofd_notify_connected(), a write_cb(0, NULL) callback is done
towards the user to notify the socket is connected.
In mode RECVMSG_SENDMSG, at least for SCTP sockets, according to comment
in iofd_uring_register() (see also OS#5751) we cannot do the write(0)
trick to get notifications, so instead we need to workaround by using a
temporary osmo_fd to poll() for write status.
To do so, we call osmo_fd_register(), which internally sets the fd as
non-blocking (O_NONBLOCK). However, this is becomes an undesired
behavior later on when connect happens and io_uring is used to
recvmsg/sendmsg, since that could cause sqes to be answered with -EAGAIN
cqes at the kernel isntead of keeping (blocking) them internally until
the transaction can be resolved. This seems was a "desired" or
"expected" behavior in older kernels according to public
discussions/tickets, but it seeems it changed over time (see eg. GH#364
below).
In summary, the conclusion seems to be: try to avoid as much as possible mixing
O_NONBLOCK with io_uring, as you may get totally unexpected/changing
behavior.
Hence, avoid keeping O_NONBLOCK for those sockets when moving back from
osmo_fd to io_uring.
See regarding related discussion on the expected behavior with io_uring and O_NONBLOCK:
https://www.spinics.net/lists/io-uring/msg04058.html
https://github.com/axboe/liburing/issues/364
Change-Id: Idba623730230bc049b827e51b058cd64d23b730f
---
M src/core/osmo_io_uring.c
1 file changed, 21 insertions(+), 1 deletion(-)
git pull ssh://gerrit.osmocom.org:29418/libosmocore refs/changes/71/41671/1
diff --git a/src/core/osmo_io_uring.c b/src/core/osmo_io_uring.c
index cac6272..278b686 100644
--- a/src/core/osmo_io_uring.c
+++ b/src/core/osmo_io_uring.c
@@ -534,6 +534,23 @@
static void iofd_uring_write_enable(struct osmo_io_fd *iofd);
static void iofd_uring_read_enable(struct osmo_io_fd *iofd);
+/* make an FD blockig:
+ * osmo_fd_register(ofd) did set fd flag O_NONBLOCK previously. We don't
+ * want to keep the fd as O_NONBLOCK once we start using io_uring,
+ * otherwise we'd end up getting cqes with -EAGAIN; better let the kernel
+ * wait internally for the sqe to complete. */
+static int iofd_reset_fd_blocking(int fd)
+{
+ int flags;
+
+ /* make FD nonblocking */
+ flags = fcntl(fd, F_GETFL);
+ if (flags < 0)
+ return flags;
+ flags &= ~O_NONBLOCK;
+ flags = fcntl(fd, F_SETFL, flags);
+ return flags;
+}
/* called via osmocom poll/select main handling once outbound non-blocking connect() completes */
static int iofd_uring_connected_cb(struct osmo_fd *ofd, unsigned int what)
@@ -547,6 +564,7 @@
/* Unregister from poll/select handling. */
osmo_fd_unregister(ofd);
+ iofd_reset_fd_blocking(ofd->fd);
IOFD_FLAG_UNSET(iofd, IOFD_FLAG_NOTIFY_CONNECTED);
/* Notify the application about this via a zero-length write completion call-back. */
@@ -591,7 +609,8 @@
* Use a temporary osmo_fd which we can use to notify us once the connection is established
* or failed (indicated by FD becoming writable). This is needed as (at least for SCTP sockets)
* one cannot submit a zero-length writev/sendmsg in order to get notification when the socket
- * is writable.*/
+ * is writable.
+ * NOITE: osmo_fd_setup() sets iofd->fd as O_NONBLOCK. */
if (IOFD_FLAG_ISSET(iofd, IOFD_FLAG_NOTIFY_CONNECTED)) {
osmo_fd_setup(&iofd->u.uring.connect_ofd, iofd->fd, OSMO_FD_WRITE,
iofd_uring_connected_cb, iofd, 0);
@@ -664,6 +683,7 @@
if (IOFD_FLAG_ISSET(iofd, IOFD_FLAG_NOTIFY_CONNECTED)) {
osmo_fd_unregister(&iofd->u.uring.connect_ofd);
+ iofd_reset_fd_blocking(iofd->u.uring.connect_ofd.fd);
IOFD_FLAG_UNSET(iofd, IOFD_FLAG_NOTIFY_CONNECTED);
}
To view, visit change 41671. To unsubscribe, or for help writing mail filters, visit settings.