tjh.dev / kernel
Linux kernel mirror (for testing) git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel os linux
1
fork

Configure Feed

Select the types of activity you want to include in your feed.

io_uring/net: don't use io_net_kbuf_recyle() for non-provided cases

A previous commit used io_net_kbuf_recyle() for any network helper that
did IO and needed partial retry. However, that's only needed if the
opcode does buffer selection, which isnt support for sendzc, sendmsg_zc,
or sendmsg. Just remove them - they don't do any harm, but it is a bit
confusing when reading the code.

Link: https://lore.kernel.org/r/20250821020750.598432-4-axboe@kernel.dk
Signed-off-by: Jens Axboe <axboe@kernel.dk>

Jens Axboe 15ba5e51 5e73b402

+3 -3
+3 -3
io_uring/net.c
··· 570 570 kmsg->msg.msg_controllen = 0; 571 571 kmsg->msg.msg_control = NULL; 572 572 sr->done_io += ret; 573 - return io_net_kbuf_recyle(req, kmsg, ret); 573 + return -EAGAIN; 574 574 } 575 575 if (ret == -ERESTARTSYS) 576 576 ret = -EINTR; ··· 1503 1503 zc->len -= ret; 1504 1504 zc->buf += ret; 1505 1505 zc->done_io += ret; 1506 - return io_net_kbuf_recyle(req, kmsg, ret); 1506 + return -EAGAIN; 1507 1507 } 1508 1508 if (ret == -ERESTARTSYS) 1509 1509 ret = -EINTR; ··· 1573 1573 1574 1574 if (ret > 0 && io_net_retry(sock, flags)) { 1575 1575 sr->done_io += ret; 1576 - return io_net_kbuf_recyle(req, kmsg, ret); 1576 + return -EAGAIN; 1577 1577 } 1578 1578 if (ret == -ERESTARTSYS) 1579 1579 ret = -EINTR;