Re: [Qemu-devel] qemu_futex_wait() lockups in ARM64: 2 possible issues

[Rafael David Tinoco <rafaeldtinoco@ubuntu.com>]

> Note that the RCU thread is expected to sit most of the time doing 
> nothing, so I don't think this matters.

Agreed.

> Zhengui's theory that notify_me doesn't work properly on ARM is more
> promising, but he couldn't provide a clear explanation of why he thought
> notify_me is involved.  In particular, I would have expected notify_me to
> be wrong if the qemu_poll_ns call came from aio_ctx_dispatch, for example:
> 
> 
>     glib_pollfds_fill
>       g_main_context_prepare
>         aio_ctx_prepare
>           atomic_or(&ctx->notify_me, 1)
>     qemu_poll_ns
>     glib_pollfds_poll
>       g_main_context_check
>         aio_ctx_check
>           atomic_and(&ctx->notify_me, ~1)
>       g_main_context_dispatch
>         aio_ctx_dispatch
>           /* do something for event */
>             qemu_poll_ns 
> 

Yep, will focus there.

> 
> Can you place somewhere your util/async.o object file for me to look at it?

Sure!

https://send.firefox.com/download/45c26bbe1075eea1/#ZD_e_96imPG2QuDqaX-jhg

Note: this async.o has value as int, EV_BUSY as 3, aborts if any errno
in qemu_futex() and uses &ev->value as 1st argument to wake/wait (as in
https://pastebin.ubuntu.com/p/xk8D6H6kgM/).

> 
> You could change it to 3, but it has to have all the bits in EV_FREE 
> (see atomic_or(&ev->value, EV_FREE) in qemu_event_reset).
> 
> You could also change it to -1u, but I don't see a particular need to do so.
> 

Yep, it was a dead end on my side.

>> - Should qemu_event_set() check return code from
>> qemu_futex_wake()->qemu_futex()->syscall() in order to know if ANY
>> waiter was ever woken up ? Maybe even loop until at least 1 is awaken ?
> 
> Why would it need to do so?
> 

No need, just realized after I saw no tasks waking that thread up. Like
you said, ctx->notify_me seems more promising, will give it a try.