Hi,

On Fri, 24 May 2024 at 16:23, Yong Huang <yong.hu...@smartx.com> wrote:
> I'm not testing the latest QEMU version while theoretically it is
> reproducible, I'll check it and give a conclusion.

* Yes, that'll help. Thank you.

> I'm not sure this usage is common but in our production environment, it is 
> used.

* I see. If it's being used in a production environment and the crash
occurs there, then it's a reasonable change.

 > IMHO, resetting the CDROM device may be a work around because
> the request SHOULD not be lost. No requests are found may be
> caused by other reasons, resetting the CD ROM seems crude.
> The path that executes the scsi_get_buf() is in a USB mass storage
> device,  and it called by the UHCI controller originally, which just
> handles the Frame List blindly, reset solution is kind of complicated
> in implementation
>
> Migrating the requests may be a graceful solution.

* Yes, true. Migration should migrate guest's devices along with their
state and data. Resetting was suggested for the case if CDROM is not
used in production and so the migration was not required. But you are
using it in a production environment so migrating SCSI requests makes
sense.

Thank you.
---
  - Prasad


Reply via email to