From 10369080454d87ee5b2db211ce947cb3118f0e13 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Tue, 7 Mar 2023 14:59:59 +0000 Subject: net: reclaim skb->scm_io_uring bit Commit 0091bfc81741 ("io_uring/af_unix: defer registered files gc to io_uring release") added one bit to struct sk_buff. This structure is critical for networking, and we try very hard to not add bloat on it, unless absolutely required. For instance, we can use a specific destructor as a wrapper around unix_destruct_scm(), to identify skbs that unix_gc() has to special case. Signed-off-by: Eric Dumazet Cc: Pavel Begunkov Cc: Thadeu Lima de Souza Cascardo Cc: Jens Axboe Reviewed-by: Jens Axboe Reviewed-by: Pavel Begunkov Signed-off-by: David S. Miller --- net/unix/garbage.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'net/unix/garbage.c') diff --git a/net/unix/garbage.c b/net/unix/garbage.c index dc2763540393..2405f0f9af31 100644 --- a/net/unix/garbage.c +++ b/net/unix/garbage.c @@ -305,7 +305,7 @@ void unix_gc(void) * release.path eventually putting registered files. */ skb_queue_walk_safe(&hitlist, skb, next_skb) { - if (skb->scm_io_uring) { + if (skb->destructor == io_uring_destruct_scm) { __skb_unlink(skb, &hitlist); skb_queue_tail(&skb->sk->sk_receive_queue, skb); } -- cgit v1.2.3