IORING_OP_RECVMSG fails when msg_control is set

[romange]

struct msghdr has msg_control and msg_controllen fields that could be set before calling recvmsg().
recvmsg returns successfully with those fields set.

IORING_OP_RECVMSG, on the other hand fails with 22/Invalid argument.

To reproduce, see https://github.com/romange/liburing/commit/fd9c810c6cf6c73c7f3b27aacf144a229ee56313

[axboe]

This is known, cmsg is currently disabled through io_uring. This will be relaxed or 5.10.

[isilence]

Do we really need this? Is there a user for whom SCM*
are performance critical?

I guess they would require one by one handling in io_uring
with a swtich or something + extra resources. ->files for
SCM_RIGHTS who knows what else for others.

[axboe]

Taking a closer look, I don't think we do.

[isilence]

#128, but don't know their use case and how critical to do that via io_uring.

[axboe]

I don't think we need it for SCM_RIGHTS, I don't view that as performance critical. But other uses cases of cmsg are useful and we do need, like MSG_ERRQUEUE. I'd be quite content to leave SCM_RIGHTS disabled forever.

[axboe]

https://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next.git/commit/net/socket.c?id=583bbf0624dfd8fc45f1049be1d4980be59451ff

[axboe]

Above is queued for 5.10 in net-next

[romange]

I wanted to use cmsg in order to timestamp RX packets and to see how much my application layer/io_uring contributes to the latency tail.

[axboe]

Closing this one