On 1/23/25 20:49, Salvatore Bonaccorso wrote:
Additionally please try with 6.1.120 and revert this commit
3ab9326f93ec ("io_uring: wake up optimisations")
(which landed in 6.1.120).
If that solves the problem maybe we miss some prequisites in the 6.1.y
series here?
I hope I did all this right. I found this:
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id=3181e22fb79910c7071e84a43af93ac89e8a7106
and attempted to undo that change in the vanilla 6.1.124 source by
making the following change to io_uring/io_uring.c:
585,594d584
< static inline void __io_cq_unlock_post_flush(struct io_ring_ctx *ctx)
< __releases(ctx->completion_lock)
< {
< io_commit_cqring(ctx);
< spin_unlock(&ctx->completion_lock);
< io_commit_cqring_flush(ctx);
< if (!(ctx->flags & IORING_SETUP_DEFER_TASKRUN))
< __io_cqring_wake(ctx);
< }
<
1352c1342
< __io_cq_unlock_post_flush(ctx);
---
> __io_cq_unlock_post(ctx);
I rebooted into the resulting kernel and am happy to report that the
problem did NOT occur!
