On Thu, 04/06 16:36, Kevin Wolf wrote: > Am 06.04.2017 um 16:25 hat Fam Zheng geschrieben: > > bdrv_replace_child_noperm tries to hand over the quiesce_counter state > > from old bs to the new one, but if they are not on the same aio context > > this causes unbalance. > > > > Fix this by forcing callers to move the aio context first, and assert > > it. > > > > Reported-by: Ed Swierk <eswi...@skyportsystems.com> > > Signed-off-by: Fam Zheng <f...@redhat.com> > > --- > > block.c | 3 +++ > > blockdev.c | 4 ++-- > > 2 files changed, 5 insertions(+), 2 deletions(-) > > > > diff --git a/block.c b/block.c > > index 927ba89..8893ac1 100644 > > --- a/block.c > > +++ b/block.c > > @@ -1752,6 +1752,9 @@ static void bdrv_replace_child_noperm(BdrvChild > > *child, > > { > > BlockDriverState *old_bs = child->bs; > > > > + if (old_bs && new_bs) { > > + assert(bdrv_get_aio_context(old_bs) == > > bdrv_get_aio_context(new_bs)); > > + } > > if (old_bs) { > > if (old_bs->quiesce_counter && child->role->drained_end) { > > child->role->drained_end(child); > > Can we move this to a separate patch and also add this hunk for > asserting the same thing for newly attached child nodes: > > @@ -1852,6 +1855,7 @@ BdrvChild *bdrv_attach_child(BlockDriverState > *parent_bs, > bdrv_get_cumulative_perm(parent_bs, &perm, &shared_perm); > > assert(parent_bs->drv); > + assert(bdrv_get_aio_context(parent_bs) == > bdrv_get_aio_context(child_bs)); > parent_bs->drv->bdrv_child_perm(parent_bs, NULL, child_role, > perm, shared_perm, &perm, &shared_perm);
OK, will do.