io_uring_cancel_task_requests() doesn't imply that the ring is going
away, it may continue to work well after that. The problem is that it
sets ->cq_overflow_flushed effectively disabling the CQ overflow feature
Split setting cq_overflow_flushed from flush, and do the first one only
on exit. It's ok in terms of cancellations because there is a
io_uring->in_idle check in __io_cqring_fill_event().
It also fixes a race with setting ->cq_overflow_flushed in
io_uring_cancel_task_requests, whuch's is not atomic and a part of a
bitmask with other flags. Though, the only other flag that's not set
during init is drain_next, so it's not as bad for sane architectures.
Signed-off-by: Pavel Begunkov <asml.silence@xxxxxxxxx>
---
fs/io_uring.c | 6 ++----
1 file changed, 2 insertions(+), 4 deletions(-)
diff --git a/fs/io_uring.c b/fs/io_uring.c
index f53356ced5ab..7115147a25a8 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -1734,10 +1734,6 @@ static bool io_cqring_overflow_flush(struct io_ring_ctx *ctx, bool force,
spin_lock_irqsave(&ctx->completion_lock, flags);
- /* if force is set, the ring is going away. always drop after that */
- if (force)
- ctx->cq_overflow_flushed = 1;
-
cqe = NULL;
list_for_each_entry_safe(req, tmp, &ctx->cq_overflow_list, compl.list) {
if (!io_match_task(req, tsk, files))
@@ -8663,6 +8659,8 @@ static void io_ring_ctx_wait_and_kill(struct io_ring_ctx *ctx)
{
mutex_lock(&ctx->uring_lock);
percpu_ref_kill(&ctx->refs);
+ /* if force is set, the ring is going away. always drop after that */
+ ctx->cq_overflow_flushed = 1;
if (ctx->rings)
io_cqring_overflow_flush(ctx, true, NULL, NULL);
mutex_unlock(&ctx->uring_lock);
--
2.24.0
