Il 24/01/2013 08:42, Dietmar Maurer ha scritto:
>>> But when I use a thread it triggers the bug in bdrv_drain_all(). So
>>> how can I fix bdrv_drain_all() if I use a separate thread to write data?
>>
>> The bug is, in all likelihood, in your own code. Sorry. :)
>
> yes. I still not fully understand that aio code.
>
> If I detect a incomplete write, I register an aio handler like this:
>
> ret = write(fd, buf, count)
> if (ret < 0 && (errno == EAGAIN || errno == EWOULDBLOCK) ) {
> ...
> qemu_aio_set_fd_handler(fd, NULL, vma_co_continue_write, NULL,
> mydata);
> qemu_coroutine_yield();
>
> But seems that qemu_aio_wait() returns false after that.
>
> I thought that should return true instead?
You need a flush handler (fourth argument to qemu_aio_set_fd_handler)
that returns true.
Paolo
