On 19.03.19 14:16, Kevin Wolf wrote: > Am 13.02.2019 um 23:53 hat Max Reitz geschrieben: >> The commit and mirror block nodes are filters, so they should be marked >> as such. >> >> Signed-off-by: Max Reitz <mre...@redhat.com> >> Reviewed-by: Alberto Garcia <be...@igalia.com> > > Then we need to update the definition of a filter: > > /* set to true if the BlockDriver is a block filter. Block filters pass > * certain callbacks that refer to data (see block.c) to their bs->file if > * the driver doesn't implement them. Drivers that do not wish to forward > * must implement them and return -ENOTSUP. > */ > bool is_filter; > > This talks about bs->file specifically, and commit and mirror nodes use > bs->backing instead. Most places that check bs->in_filter do so to check > whether they can forward stuff to bs->file, and they don't consider > bs->backing. > > I think the latter will be fixed after patch 3, but the definition in > the comment is still wrong at the end of the series.
Right, I forgot this comment. (Interesting how Quorum has already ignored this definition for a long time.) O:-) Max
signature.asc
Description: OpenPGP digital signature