6.10-stable review patch. If anyone has any objections, please let me know.
------------------
From: Martin Karsten <mkarsten@xxxxxxxxxxxx>
commit b9ca079dd6b09e08863aa998edf5c47597806c05 upstream.
A struct eventpoll's busy_poll_usecs field can be modified via a user
ioctl at any time. All reads of this field should be annotated with
READ_ONCE.
Fixes: 85455c795c07 ("eventpoll: support busy poll per epoll instance")
Cc: stable@xxxxxxxxxxxxxxx
Signed-off-by: Martin Karsten <mkarsten@xxxxxxxxxxxx>
Link: https://lore.kernel.org/r/20240806123301.167557-1-jdamato@xxxxxxxxxx
Reviewed-by: Joe Damato <jdamato@xxxxxxxxxx>
Signed-off-by: Christian Brauner <brauner@xxxxxxxxxx>
Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
---
fs/eventpoll.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/fs/eventpoll.c b/fs/eventpoll.c
index 6c0a1e9715ea..145f5349c612 100644
--- a/fs/eventpoll.c
+++ b/fs/eventpoll.c
@@ -420,7 +420,7 @@ static bool busy_loop_ep_timeout(unsigned long start_time,
static bool ep_busy_loop_on(struct eventpoll *ep)
{
- return !!ep->busy_poll_usecs || net_busy_loop_on();
+ return !!READ_ONCE(ep->busy_poll_usecs) || net_busy_loop_on();
}
static bool ep_busy_loop_end(void *p, unsigned long start_time)
--
2.46.2
