Am 10.04.2017 um 17:05 hat Fam Zheng geschrieben: > v3: Respin the unmerged changes from v2 and include one new fix: > > (Yes, it is a big series for the last -rc, and I personally prefer the v2 > approach for the 4-9 part of the problem, which is much more mechanical.) > > - 1, 2 are redoing previous patch 4, using bdrv_parent_drained_begin/end. > [Kevin] > Also fix the ordering against aio_context_release. [Stefan] > - 3 is unchanged from patch 6 in v2. > - 4-9 are reworking of patch 5 following Paolo's suggestion, which allowed > better patch split. > - 10 is finding of a latent bug, which is revealed by patch 9. > > v2: - Drop patch 4 in v1. A second thought made me feel neither it nor Kevin's > suggestion to move the BH process to bdrv_drain_recurse/BDRV_POLL_WHILE > is a complete fix. So leave it for a separate patch. > - Add rev-by to patches 1, 3, 4. > - Split from patch 1 in v1 and add patch 2, for the new assertions. > [Kevin] > - Rewrite patch 5. Fix block job's co when a BDS is moved to a different > aio context. [Kevin] > - Add patch 6. > > Crashes are reported on dataplane devices when doing snapshot and commit under > guest I/O. > > With this series, Ed's test case '176' now passes: > > https://github.com/skyportsystems/qemu-1/commits/eswierk-iotests-2.9
I had only two points for this series. The first is that it adds unused functions, which doesn't hurt (but I might just send a PATCH 11/10 to remove them again). The second one is that some sheepdog code is suspicious, but if anything it just means that this series is incomplete, so not a show stopper either. Reviewed-by: Kevin Wolf <kw...@redhat.com>