io_uring/net: disable partial retries for recvmsg with cmsg
authorJens Axboe <axboe@kernel.dk>
Mon, 19 Jun 2023 15:41:05 +0000 (09:41 -0600)
committerJens Axboe <axboe@kernel.dk>
Wed, 21 Jun 2023 13:34:07 +0000 (07:34 -0600)
We cannot sanely handle partial retries for recvmsg if we have cmsg
attached. If we don't, then we'd just be overwriting the initial cmsg
header on retries. Alternatively we could increment and handle this
appropriately, but it doesn't seem worth the complication.

Move the MSG_WAITALL check into the non-multishot case while at it,
since MSG_WAITALL is explicitly disabled for multishot anyway.

Link: https://lore.kernel.org/io-uring/0b0d4411-c8fd-4272-770b-e030af6919a0@kernel.dk/
Cc: stable@vger.kernel.org # 5.10+
Reported-by: Stefan Metzmacher <metze@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
io_uring/net.c

index c0924ab1ea11224bb6d8563d003fd6dadca42b10..2bc2cb2f4d6c2dbf2b686ec1891b9fdc34336a89 100644 (file)
@@ -789,16 +789,19 @@ retry_multishot:
        flags = sr->msg_flags;
        if (force_nonblock)
                flags |= MSG_DONTWAIT;
-       if (flags & MSG_WAITALL)
-               min_ret = iov_iter_count(&kmsg->msg.msg_iter);
 
        kmsg->msg.msg_get_inq = 1;
-       if (req->flags & REQ_F_APOLL_MULTISHOT)
+       if (req->flags & REQ_F_APOLL_MULTISHOT) {
                ret = io_recvmsg_multishot(sock, sr, kmsg, flags,
                                           &mshot_finished);
-       else
+       } else {
+               /* disable partial retry for recvmsg with cmsg attached */
+               if (flags & MSG_WAITALL && !kmsg->msg.msg_controllen)
+                       min_ret = iov_iter_count(&kmsg->msg.msg_iter);
+
                ret = __sys_recvmsg_sock(sock, &kmsg->msg, sr->umsg,
                                         kmsg->uaddr, flags);
+       }
 
        if (ret < min_ret) {
                if (ret == -EAGAIN && force_nonblock) {