Christoph Egger wrote: > Steven Chamberlain <steven@pyro.eu.org> writes: > > It's the same thing we saw in sid recently until they added a workaround > > (https://bugs.debian.org/796798). > > Oh great we know the actual problem? :-) Yeah, pretty sure that's what it is. > > It began after the kernel was upgraded from 9.0 to 10.0. It may also be > > related to softupdates if we have that enabled now; or even a bug in > > some other filesystem layer if schroot uses e.g. nullfs > > I should see that in `mount` right? It doesn't seem to have any > softupdate there: I don't think that tells you. You can't change it via mount options either, but only through tunefs? While it's unmounted. Although, softupdates are rather important for performance if the media type is slow (not SSD). This is some really odd edge case / race that some Perl test is sensitive to. > As DSA has been asking about that issue at least twice I'm inclined to > have a fix uploaded. probably jessie-kfreebsd-proposed-updates and I'd > do it tomorrow "morning" I guess? It's really just a 1-line diff to stat.t, whereas getting a kernel bug diagnosed+solved+patched could take weeks. So the version number would be something like +deb8u2+kbsd8u1? That sounds good. The next perl bugfix would be +deb8u3 having higher precedence, so once the buildd is fixed we won't need to upload patched versions to jessie-kfreebsd-p-u any more. Regards, -- Steven Chamberlain steven@pyro.eu.org
Attachment:
signature.asc
Description: Digital signature