Re: [PATCH] xen-block: Fix removal of backend instance via xenstore

[Paolo Bonzini <pbonzini@xxxxxxxxxx>]

On 08/03/21 18:29, Anthony PERARD wrote:
> > If nothing else works then I guess it's okay, but why can't you do the
> > xen_block_drive_destroy from e.g. an unrealize callback?
> I'm not sure if that's possible.
>
> xen_block_device_create/xen_block_device_destroy() is supposed to be
> equivalent to do those qmp commands:
>      blockdev-add node-name=xvdz-qcow2 driver=qcow2 
> file={"driver":"file","filename":"disk.qcow2","locking":"off"}
>      device_add id=xvdz driver=xen-disk vdev=xvdz drive=xvdz-qcow2
>
> But I tried to add a call xen_block_drive_destroy from
> xen_block_unrealize, but that still is called too early, it's called
> before object_property_del_all() which would delete "drive" and call
> release_drive() which would free the node.
Can you use blockdev_mark_auto_del?  Then you don't have to call 
xen_block_drive_destroy at all.
Paolo

> So, no, I don't think we can use an unrealized callback.
>
> I though of trying to delete the "drive" property ahead of calling
> object_unparent() but I didn't figure out how to do so and it's maybe
> not possible.
>
> So either drain_call_rcu or adding call_rcu(xen_block_drive_destroy)
> seems to be the way, but since xen_block_drive_destroy uses
> qmp_blockdev_del, it seems better to drain_call_rcu.
>
> Cheers,