On 29.10.25 18:10, Kevin Wolf wrote:
Am 28.10.2025 um 17:33 hat Hanna Czenczek geschrieben:
The request coroutine may not run in the BDS AioContext. We should wake
it in its own context, not move it.
With that, we can remove GlusterAIOCB.aio_context.
Also add a comment why aio_co_schedule() is safe to use in this way.
**Note:** Due to a lack of a gluster set-up, I have not tested this
commit. It seemed safe enough to send anyway, just maybe not to
qemu-stable. To be clear, I don’t know of any user-visible bugs that
would arise from the state without this patch; the request coroutine is
moved into the main BDS AioContext, which feels wrong, but I’m not sure
whether it can actually produce hard bugs.
Doesn't moving into a different AioContext mean that everything down to
the AIO callback in the guest device runs in a different thread? That's
definitely a big no-no for virtio-scsi, and I think also for virtio-blk.
It does, yes.
I’ll leave it to others’ opinions whether to keep or drop this patch
under these circumstances.
Signed-off-by: Hanna Czenczek <[email protected]>
It looks trivially correct (famous last words...) and the bugs not
unlikely to be hit, so I'd say keep it.
I have no idea if the gluster library is actually thread safe, but if it
isn't, that breaks before gluster_finish_aiocb(). After reentering the
coroutine, the library isn't called any more.
OK.
Hanna
