On 8/14/24 23:12, Patrick Leis wrote:
I would go with your aio_wait_bh_oneshot suggestion. Please consider
adding it to QEMU as I pointed above. I can send a patch.
Hey - this is an important race condition to resolve, can we get some
attention back on this patchset please. It's pressing.
Sorry about the delay.
The patch is good, and there is no reason why it should break. But
without a user (and as far as I understand there is no occurrence of
this race condition in upstream QEMU code) there is no reason to include
it in QEMU. QEMU is not a library, and therefore there should be no
unused functions in the code base.
You can keep it as a solution for the race condition in your code;
depending on how your fork is organized, it may be kept in qemu-timer.c
or in a file that is part of your additional device models.
Thanks,
Paolo
