Re: [PATCH net] rds: annotate data-race around rs_seen_congestion
From: Allison Henderson <achender@kernel.org>
Date: 2026-05-22 04:49:20
Also in:
linux-rdma, lkml
On Fri, 2026-05-22 at 09:16 +0800, Jiayuan Chen wrote:
rs_seen_congestion is read in rds_poll() and written in rds_sendmsg()
and rds_poll() without any lock. Use READ_ONCE()/WRITE_ONCE() to
annotate these lockless accesses and silence KCSAN.
Fixes: b98ba52f96e7 ("RDS: only put sockets that have seen congestion on the poll_waitq")
Reported-by: syzbot+fbf3648ae7f5bdb05c59@syzkaller.appspotmail.com
Closes: https://lore.kernel.org/netdev/6a0f8d94.050a0220.6b33c.0000.GAE@google.com/ (local)
Signed-off-by: Jiayuan Chen <jiayuan.chen@linux.dev>Hi Jiayuan, I was about to send the same patch, so no complaints here. Thanks for catching this. Reviewed-by: Allison Henderson <achender@kernel.org> Tested-by: Allison Henderson <achender@kernel.org> Allison
quoted hunk ↗ jump to hunk
--- net/rds/af_rds.c | 4 ++-- net/rds/send.c | 2 +- 2 files changed, 3 insertions(+), 3 deletions(-)diff --git a/net/rds/af_rds.c b/net/rds/af_rds.c index 76f625986a7f..93b2da63ed42 100644 --- a/net/rds/af_rds.c +++ b/net/rds/af_rds.c@@ -219,7 +219,7 @@ static __poll_t rds_poll(struct file *file, struct socket *sock, poll_wait(file, sk_sleep(sk), wait); - if (rs->rs_seen_congestion) + if (READ_ONCE(rs->rs_seen_congestion)) poll_wait(file, &rds_poll_waitq, wait); read_lock_irqsave(&rs->rs_recv_lock, flags);@@ -247,7 +247,7 @@ static __poll_t rds_poll(struct file *file, struct socket *sock, /* clear state any time we wake a seen-congested socket */ if (mask) - rs->rs_seen_congestion = 0; + WRITE_ONCE(rs->rs_seen_congestion, 0); return mask; }diff --git a/net/rds/send.c b/net/rds/send.c index d8b14ff9d366..e5d58c29aabe 100644 --- a/net/rds/send.c +++ b/net/rds/send.c@@ -1388,7 +1388,7 @@ int rds_sendmsg(struct socket *sock, struct msghdr *msg, size_t payload_len) ret = rds_cong_wait(conn->c_fcong, dport, nonblock, rs); if (ret) { - rs->rs_seen_congestion = 1; + WRITE_ONCE(rs->rs_seen_congestion, 1); goto out; } while (!rds_send_queue_rm(rs, conn, cpath, rm, rs->rs_bound_port,