kcm_write_msgs() calls unreserve_psock() to release its hold on the
underlying TCP socket if it has run out of things to transmit, but if we
have nothing in the write queue on entry (e.g. because someone did a
zero-length sendmsg), we don't actually go into the transmission loop and
as a consequence don't call reserve_psock().

Fix this by skipping the call to unreserve_psock() if we didn't reserve a
psock.

Fixes: c31a25e1db48 ("kcm: Send multiple frags in one sendmsg()")
Reported-by: syzbot+dd1339599f1840e4cc65@syzkaller.appspotmail.com
Link: https://lore.kernel.org/r/000000000000a61ffe05fe0c3d08@google.com/
Signed-off-by: David Howells <dhowells@redhat.com>
Tested-by: syzbot+dd1339599f1840e4cc65@syzkaller.appspotmail.com
cc: Tom Herbert <tom@herbertland.com>
cc: Tom Herbert <tom@quantonium.net>
cc: "David S. Miller" <davem@davemloft.net>
cc: Eric Dumazet <edumazet@google.com>
cc: Jakub Kicinski <kuba@kernel.org>
cc: Paolo Abeni <pabeni@redhat.com>
cc: Jens Axboe <axboe@kernel.dk>
cc: Matthew Wilcox <willy@infradead.org>
cc: netdev@vger.kernel.org
---
 net/kcm/kcmsock.c |    4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/net/kcm/kcmsock.c b/net/kcm/kcmsock.c
index d75d775e9462..d0537c1c8cd7 100644
--- a/net/kcm/kcmsock.c
+++ b/net/kcm/kcmsock.c
@@ -661,6 +661,7 @@ static int kcm_write_msgs(struct kcm_sock *kcm)
 				kcm_abort_tx_psock(psock, ret ? -ret : EPIPE,
 						   true);
 				unreserve_psock(kcm);
+				psock = NULL;
 
 				txm->started_tx = false;
 				kcm_report_tx_retry(kcm);
@@ -696,7 +697,8 @@ static int kcm_write_msgs(struct kcm_sock *kcm)
 	if (!head) {
 		/* Done with all queued messages. */
 		WARN_ON(!skb_queue_empty(&sk->sk_write_queue));
-		unreserve_psock(kcm);
+		if (psock)
+			unreserve_psock(kcm);
 	}
 
 	/* Check if write space is available */

Hello:

This patch was applied to netdev/net-next.git (main)
by Jakub Kicinski <kuba@kernel.org>:

On Thu, 15 Jun 2023 12:32:02 +0100 you wrote:
> kcm_write_msgs() calls unreserve_psock() to release its hold on the
> underlying TCP socket if it has run out of things to transmit, but if we
> have nothing in the write queue on entry (e.g. because someone did a
> zero-length sendmsg), we don't actually go into the transmission loop and
> as a consequence don't call reserve_psock().
> 
> Fix this by skipping the call to unreserve_psock() if we didn't reserve a
> psock.
> 
> [...]

Here is the summary with links:
  - [net-next] kcm: Fix unnecessary psock unreservation.
    https://git.kernel.org/netdev/net-next/c/9f8d0dc0ec4a

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html