Re: Bug#1093243: Upgrade to 6.1.123 kernel causes mariadb hangs

[Pavel Begunkov <asml.silence@gmail.com>]

On 1/24/25 16:30, Xan Charbonnet wrote:
> On 1/24/25 04:33, Pavel Begunkov wrote:
>> Thanks for narrowing it down. Xan, can you try this change please?
>> Waiters can miss wake ups without it, seems to match the description.
>>
>> diff --git a/io_uring/io_uring.c b/io_uring/io_uring.c
>> index 9b58ba4616d40..e5a8ee944ef59 100644
>> --- a/io_uring/io_uring.c
>> +++ b/io_uring/io_uring.c
>> @@ -592,8 +592,10 @@ static inline void __io_cq_unlock_post_flush(struct io_ring_ctx *ctx)
>>        io_commit_cqring(ctx);
>>        spin_unlock(&ctx->completion_lock);
>>        io_commit_cqring_flush(ctx);
>> -    if (!(ctx->flags & IORING_SETUP_DEFER_TASKRUN))
>> +    if (!(ctx->flags & IORING_SETUP_DEFER_TASKRUN)) {
>> +        smp_mb();
>>            __io_cqring_wake(ctx);
>> +    }
>>    }
>>    void io_cq_unlock_post(struct io_ring_ctx *ctx)
>>
> 
> 
> Thanks Pavel!  Early results look very good for this change.  I'm now running 6.1.120 with your added smp_mb() call.  The backup process which had been quickly triggering the issue has been running longer than it ever did when it would ultimately fail.  So that's great!
> 
> One sour note: overnight, replication hung on this machine, which is another failure that started happening with the jump from 6.1.119 to 6.1.123.  The machine was running 6.1.124 with the __io_cq_unlock_post_flush function removed completely.  That's the kernel we had celebrated yesterday for running the backup process successfully.
> 
> So, we might have two separate issues to deal with, unfortunately.

Possible, but it could also be a side effect of reverting the patch.
As usual, in most cases patches are ported either because they're
fixing sth or other fixes depend on it, and it's not yet apparent
to me what happened with this one.

> This morning, I found that replication had hung and was behind by some 35,000 seconds.  I attached gdb and then detached it, which got things moving again (which goes the extra mile to prove that this is a very closely related issue).  Then it hung up again at about 25,000 seconds behind.  At that point I rebooted into the new kernel, the 6.1.120 kernel with the added smp_mb() call.  The lag is now all the way down to 5,000 seconds without hanging again.
> 
> It looks like there are 5 io_uring-related patches in 6.1.122 and another 1 in 6.1.123.  My guess is the replication is hitting a problem with one of those.
> 
> Unfortunately, a replication hang is much harder for me to reproduce than the issue with the backup procedure, which always failed within 15 minutes.  It certainly looks to me like the patched 6.1.120 does not have the hang (but it's hard to be 100% certain).  Perhaps the next step is to apply the extra smp_mb() call to 6.1.123 and see if I can get replication to hang.

Sounds like it works as expected with mb(), at least for now. I agree,
it makes sense to continue testing with the patch, and I'll send it to
stable in the meantime. Thanks for testing!

-- 
Pavel Begunkov