From: Zilin Guan <zi...@seu.edu.cn> The function netfs_unbuffered_write_iter_locked() in fs/netfs/direct_write.c contains an unnecessary smp_rmb() call after wait_on_bit(). Since wait_on_bit() already incorporates a memory barrier that ensures the flag update is visible before the function returns, the smp_rmb() provides no additional benefit and incurs unnecessary overhead.
This patch removes the redundant barrier to simplify and optimize the code. Signed-off-by: Zilin Guan <zi...@seu.edu.cn> Signed-off-by: David Howells <dhowe...@redhat.com> cc: Akira Yokosawa <aki...@gmail.com> cc: Jeff Layton <jlay...@kernel.org> cc: ne...@lists.linux.dev cc: linux-fsde...@vger.kernel.org Link: https://lore.kernel.org/r/20241207021952.2978530-1-zi...@seu.edu.cn/ --- fs/netfs/direct_write.c | 1 - 1 file changed, 1 deletion(-) diff --git a/fs/netfs/direct_write.c b/fs/netfs/direct_write.c index 88f2adfab75e..173e8b5e6a93 100644 --- a/fs/netfs/direct_write.c +++ b/fs/netfs/direct_write.c @@ -104,7 +104,6 @@ ssize_t netfs_unbuffered_write_iter_locked(struct kiocb *iocb, struct iov_iter * trace_netfs_rreq(wreq, netfs_rreq_trace_wait_ip); wait_on_bit(&wreq->flags, NETFS_RREQ_IN_PROGRESS, TASK_UNINTERRUPTIBLE); - smp_rmb(); /* Read error/transferred after RIP flag */ ret = wreq->error; if (ret == 0) { ret = wreq->transferred;
