Re: [PATCH -next] mm/slub: fix a slab missed to be freed problem

[Feng Tang <feng.tang@intel.com>]

On Fri, Sep 30, 2022 at 07:25:54PM +0800, Vlastimil Babka wrote:
> 
> On 9/30/22 12:07, Feng Tang wrote:
> > When enable kasan and kfence's in-kernel kunit test with slub_debug on,
> > it caught a problem (in linux-next tree):
> > 
> >  ------------[ cut here ]------------
> >  kmem_cache_destroy test: Slab cache still has objects when called from test_exit+0x1a/0x30
> >  WARNING: CPU: 3 PID: 240 at mm/slab_common.c:492 kmem_cache_destroy+0x16c/0x170
> 
> Assuming the warning was preceded by some kunit test failures?
> I don't see how leaving more empty slabs on free list than needed would
> cause this warning, the shutdown should just drop the empty slab.

The previous code only call remove_partial() to dequeue the slab from
partial list, and miss to call discard_slab() for it.

From the debug dump, the n->nr_partils stays at 5, while n->nr_slabs
keeps increasing. And during shutdown, the free_partial() only free
the 5 slabs on partial list, and n->nr_slabs still has a big numbers
of orphan slabs

> >  Modules linked in:
> >  CPU: 3 PID: 240 Comm: kunit_try_catch Tainted: G    B            N 6.0.0-rc7-next-20220929 #52
> >  Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014
> >  RIP: 0010:kmem_cache_destroy+0x16c/0x170
> >  Code: 41 5c 41 5d e9 a5 04 0b 00 c3 cc cc cc cc 48 8b 55 60 48 8b 4c 24 20 48 c7 c6 40 37 d2 82 48 c7 c7 e8 a0 33 83 e8 4e d7 14 01 <0f> 0b eb a7 41 56 41 89 d6 41 55 49 89 f5 41 54 49 89 fc 55 48 89
> >  RSP: 0000:ffff88800775fea0 EFLAGS: 00010282
> >  RAX: 0000000000000000 RBX: ffffffff83bdec48 RCX: 0000000000000000
> >  RDX: 0000000000000001 RSI: 1ffff11000eebf9e RDI: ffffed1000eebfc6
> >  RBP: ffff88804362fa00 R08: ffffffff81182e58 R09: ffff88800775fbdf
> >  R10: ffffed1000eebf7b R11: 0000000000000001 R12: 000000008c800d00
> >  R13: ffff888005e78040 R14: 0000000000000000 R15: ffff888005cdfad0
> >  FS:  0000000000000000(0000) GS:ffff88807ed00000(0000) knlGS:0000000000000000
> >  CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> >  CR2: 0000000000000000 CR3: 000000000360e001 CR4: 0000000000370ee0
> >  DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
> >  DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
> >  Call Trace:
> >   <TASK>
> >   test_exit+0x1a/0x30
> >   kunit_try_run_case+0xad/0xc0
> >   kunit_generic_run_threadfn_adapter+0x26/0x50
> >   kthread+0x17b/0x1b0
> > 
> > It was biscted to commit c7323a5ad078 ("mm/slub: restrict sysfs
> > validation to debug caches and make it safe")
> > 
> > The problem is inside free_debug_processing(), in one path, the slab
> > on partial list is missed to be freed when partial list is full.
> > 
> > Signed-off-by: Feng Tang <feng.tang@intel.com>
> > ---
> > 
> > Hi reviewers,
> > 
> > Sorry for the late reporting, but it's curious that this problem didn't
> > show up in my earlier test (which caught some other problems).
> 
> I think we can reuse the slab_free and don't need a new bool?

Yes, much simpler!

Thanks,
Feng

> diff --git a/mm/slub.c b/mm/slub.c
> index 5c3c31a154ba..a63953f649ed 100644
> --- a/mm/slub.c
> +++ b/mm/slub.c
> @@ -2886,22 +2886,25 @@ static noinline void free_debug_processing(
>                 set_freepointer(s, tail, prior);
>                 slab->freelist = head;
>  
> -               /* Do we need to remove the slab from full or partial list? */
> +               /*
> +                * If the slab is empty, and node's partial list is full,
> +                * it should be discarded anyway no matter it's on full or
> +                * partial list.
> +                */
> +               if (slab->inuse == 0 && n->nr_partial >= s->min_partial)
> +                       slab_free = slab;
> +
>                 if (!prior) {
> +                       /* was on full list */
>                         remove_full(s, n, slab);
> -               } else if (slab->inuse == 0 &&
> -                          n->nr_partial >= s->min_partial) {
> +                       if (!slab_free) {
> +                               add_partial(n, slab, DEACTIVATE_TO_TAIL);
> +                               stat(s, FREE_ADD_PARTIAL);
> +                       }
> +               } else if (slab_free) {
>                         remove_partial(n, slab);
>                         stat(s, FREE_REMOVE_PARTIAL);
>                 }
> -
> -               /* Do we need to discard the slab or add to partial list? */
> -               if (slab->inuse == 0 && n->nr_partial >= s->min_partial) {
> -                       slab_free = slab;
> -               } else if (!prior) {
> -                       add_partial(n, slab, DEACTIVATE_TO_TAIL);
> -                       stat(s, FREE_ADD_PARTIAL);
> -               }
>         }
>  
>         if (slab_free) {