Re: [PATCH] xfs: use kmem_cache_free() for kmem_cache objects

From: Rustam Kovhaev
Date: Thu Sep 30 2021 - 20:32:29 EST


On Thu, Sep 30, 2021 at 11:10:10PM +0200, Vlastimil Babka wrote:
> On 9/30/21 8:48 PM, Rustam Kovhaev wrote:
> > On Thu, Sep 30, 2021 at 10:13:40AM +0200, Vlastimil Babka wrote:
> >>
> >> I think it's fair if something like XFS (not meant for tiny systems AFAIK?)
> >> excludes SLOB (meant for tiny systems). Clearly nobody tried to use these
> >> two together last 5 years anyway.
> >
> > +1 for adding Kconfig option, it seems like some things are not meant to
> > be together.
>
> But if we patch SLOB, we won't need it.

OK, so we consider XFS on SLOB a supported configuration that might be
used and should be tested.
I'll look into maybe adding a config with CONFIG_SLOB and CONFIG_XFS_FS
to syzbot.

It seems that we need to patch SLOB anyway, because any other code can
hit the very same issue.

> >> Maybe we could also just add the 4 bytes to all SLOB objects, declare
> >> kfree() is always fine and be done with it. Yes, it will make SLOB footprint
> >> somewhat less tiny, but even whan we added kmalloc power of two alignment
> >> guarantees, the impact on SLOB was negligible.
> >
> > I'll send a patch to add a 4-byte header for kmem_cache_alloc()
> > allocations.
>
> Thanks. Please report in the changelog slab usage from /proc/meminfo
> before and after patch (at least a snapshot after a full boot).

OK.