Re: [PATCH 2/2] block/monitor: add drive_insert HMP command

[Stefan Hajnoczi <stefanha@redhat.com>]

[-- Attachment #1: Type: text/plain, Size: 1064 bytes --]

On Wed, Apr 15, 2026 at 02:32:39PM +0200, Matthieu Rolla wrote:
> Thanks Daniel, 
> 
> It makes sense Thanks. 
> 
> Looking at the existing code, blockdev-insert-medium already does the backend/frontend association via blk_insert_bs(), but is restricted to removable devices. 
> A new QMP command like blockdev-attach could reuse the same logic without the removable restriction, paired with blockdev-add for creating the block node.
> 
> Would that be a better approach ? 

Hi Matthieu,
I was wondering whether the blockdev needs to be changed at all. Since
the disk image remains the same, is it sufficient to inject the PCIe SDN
and then recover the NVMe PCI controller?

I don't understand the test scenario well enough, but it seems like
you're testing at the PCIe level here rather than nothin NVMe- or
blockdev-specific. Therefore blockdev commands may not be necessary.

If the testing can be done completely at the PCIe level then that would
also allow other device types to be tested in the same way, which would
be nice.

Stefan

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 484 bytes --]