Comment 13 for bug 1837869

New candidates:
5fbfabd3 block: Formats don't need CONSISTENT_READ with NO_IO

cc954f01 block: Open backing image in force share mode for size probe

0152bf40 block: Don't notify parents in drain call chain
d736f119 block: Allow graph changes in subtree drained section
1a63a907 block: Keep nodes drained between reopen_queue/multiple

There also are some commits around switching to ".bdrv_co_block_status" and ".bdrv_co_create" but I'd like to avoid those as it seems a bigger overhaul.

We do implicit image creation via qmp in the testcase here:
(QEMU) blockdev-snapshot-sync device=disk0 snapshot-file=tmp.qcow2 format=qcow2
Formatting 'tmp.qcow2', fmt=qcow2 size=10737418240 backing_file=./top-vm01.img backing_fmt=qcow2 cluster_size=65536 lazy_refcounts=off refcount_bits=16
And there is a commit saying:
"This adds the .bdrv_co_create driver callback to file, which enables image creation over QMP."
But this seems to be an explicit create, the former implicit create might be ok and our issues are on the base file not the tmp.qcow2 anyway.

None of the above seem to be the perfect candidate, the one that looks somewhat closer to the issue is:
1a529736 block: Fix flags in reopen queue

Of these we already have on top of 2.11:
cc954f01 block: Open backing image in force share mode for size probe

I'm trying a build of Ubuntu's 2.11 with the smaller patches identified (no drain queue and no new ops used):
5fbfabd3 block: Formats don't need CONSISTENT_READ with NO_IO
1a529736 block: Fix flags in reopen queue

This is only a first stabbing in the dark via PPA [1] to be eventually more fine grained and selective.

[1]: https://launchpad.net/~paelzer/+archive/ubuntu/bug-1837869-locking-of-snapshots