ACK: [PATCH 0/1] Partial fix for mm/slub regression (LP #1895109)

[Andrea Righi]

On Thu, Sep 10, 2020 at 12:20:52PM +0200, Kleber Souza wrote:
> On 10.09.20 12:13, Andrea Righi wrote:
> > On Thu, Sep 10, 2020 at 12:04:05PM +0200, Kleber Sacilotto de Souza wrote:
> >> [ Impact ]
> >> As reported on bug 1894780 (Oops and hang when starting LVM snapshots
> >> on 5.4.0-47), some changes applied to focal/linux 5.4.0-45 introduced a
> >> regression likely on the mm/slub code. A fix hasn't been identified yet,
> >> however, reverting commit "mm/slub: fix a memory leak in
> >> sysfs_slab_add()", upstream dde3c6b72a16c2db826f54b2d49bdea26c3534a2,
> >> applied to focal/linux as part of v5.4.47 upstream stable release
> >> (bug 1884089), seems to improve the situation.
> >>
> >> [ Fix ]
> >> I propose to revert commit "mm/slub: fix a memory leak in
> >> sysfs_slab_add()" while we don't find a better fix for it.
> >>
> >> [ Test Case ]
> >> Please see bug 1894780.
> > 
> > Did the bug reporter (or anyone else able to reproduce the oops) test a
> > kernel with this revert? I don't see any feedback about this in
> > LP: #1894780.
> 
> Hi Andrea,
> 
> It was not commented on the bug but the reporter provided the feedback that
> the test kernel made the situation better, at least his systems now boot.
> But still regressing from -42, so more investigation is needed.
> 
> We are still seeing the following errors on dmes:
> 
> [   23.553649] kobject_add_internal failed for :a-0000152 with -EEXIST,
> don't try to register things with the same name in the same directory.
> [   23.567674] device-mapper: table: 253:6: snapshot: Failed to read
> snapshot metadata

OK, I've also looked at this bug and reverting the commit makes sense to
me, therefore:

Acked-by: Andrea Righi <andrea.righi at canonical.com>