Virtualization - Dec 2017 - Question about eliminating virtio-scsi 30s timeout and hot-unplug

Hi Paolo,
I had a question about this patch that eliminates virtio-scsi timeout on
IOs - https://patchwork.kernel.org/patch/9802047/.  How does this work when
we have IOs outstanding to a device that is then hot-unplugged.  I'm under
the impression that those IOs will never get returned with any status
(please correct me if I'm wrong), and we will then end up waiting on them
forever with this patch.

Thanks,
Jason
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.linuxfoundation.org/pipermail/virtualization/attachments/20171221/1994b38c/attachment-0001.html>

On 12/21/2017 08:29 PM, Jason Behmer via Virtualization
wrote:
> Hi Paolo,
> I had a question about this patch that eliminates virtio-scsi timeout on
> IOs - https://patchwork.kernel.org/patch/9802047/.? How does this work
> when we have IOs outstanding to a device that is then hot-unplugged.?
> I'm under the impression that those IOs will never get returned with
any
> status (please correct me if I'm wrong), and we will then end up
waiting
> on them forever with this patch.
> 
It's no different from the virtio-block path; the I/O is pending in the
qemu process. When a device is hot-unplugged the I/O will be returned to
the qemu process, which then will return the final status to the guest.

Cheers,

Hannes
-- 
Dr. Hannes Reinecke		   Teamlead Storage & Networking
hare at suse.de			               +49 911 74053 688
SUSE LINUX GmbH, Maxfeldstr. 5, 90409 N?rnberg
GF: F. Imend?rffer, J. Smithard, J. Guild, D. Upmanyu, G. Norton
HRB 21284 (AG N?rnberg)