Re: [PATCH RFC 00/11] migration/block: disk activation rewrite

[Peter Xu]

On Tue, Dec 03, 2024 at 07:51:27PM -0500, Peter Xu wrote:
>   migration/block: Merge block reactivations for fail/cancel
>   migration/block: Extend the migration_block_* API to dest side
>   migration/block: Apply the migration_block_* API to postcopy

I just noticed these three patches cannot be separate, because right after
we introduce a flag to cache disk activation status, it will need to apply
to all the code or inconsistency can happen if someone applies one of the
patches, for example.

I also overlooked that we must take qmp_cont() into the picture too, so
that will also need to use the API so the flag will be consistent.

So in the next version I'll squash the three patches into one, and also use
the new API in qmp_cont(), so the status flag should always be consistent.

Again, to block layer developers: please help if any of you know how to
make bdrv_inactivate_all() safe to be called on top of inactivated disks,
or any reasoning on why it mustn't.  It could be very helpful.

-- 
Peter Xu