Re: [Qemu-devel] [PATCH v2 0/6] Acquire the AioContext during _realize()

[Alberto Garcia]

On Wed 16 Jan 2019 02:54:44 PM CET, Stefan Hajnoczi wrote:
> The x-blockdev-set-iothread command is for low-level tests so I don't
> expect users to invoke it.

As I said in a different e-mail maybe this is not necessary then.

> This patch series makes virtio-blk and virtio-scsi more robust,
> although I haven't checked what happens if the drive is attached to a
> different IOThread than the device - will the switchover work?

You mean if you use x-blockdev-set-iothread with a drive and then attach
it to a device using a different thread? That seems to work (with my
patch).

> What I'm unclear about is why fdc, ide, usb-msd, and nvme should worry
> about IOThreads/AioContext in .realize() since they don't support it
> in their data path.  What happens when you submit I/O requests to
> devices configured like this?  I guess they would crash or hit
> assertion failures since they invoke blk_aio_*() APIs from outside the
> appropriate AioContext.
>
> Maybe fdc and friends should forbid this scenario:

Yeah, that seems like a simple solution.

Berto