Re: [PATCH] mm: slub: Really fix slab walking for init_on_free

[Alexander Potapenko <glider@google.com>]

On Wed, Nov 6, 2019 at 11:22 PM Laura Abbott <labbott@redhat.com> wrote:
>
> Commit 1b7e816fc80e ("mm: slub: Fix slab walking for init_on_free")
> fixed one problem with the slab walking but missed a key detail:
> When walking the list, the head and tail pointers need to be updated
> since we end up reversing the list as a result. Without doing this,
> bulk free is broken. One way this is exposed is a NULL pointer with
> slub_debug=F:
Thanks for the fix!
Is it possible to reproduce the problem on a small test case that
could be added to lib/test_meminit.c?

> =============================================================================
> BUG skbuff_head_cache (Tainted: G                T): Object already free
> -----------------------------------------------------------------------------
>
> INFO: Slab 0x000000000d2d2f8f objects=16 used=3 fp=0x0000000064309071 flags=0x3fff00000000201
> BUG: kernel NULL pointer dereference, address: 0000000000000000
> PGD 0 P4D 0
> Oops: 0000 [#1] PREEMPT SMP PTI
> CPU: 0 PID: 0 Comm: swapper/0 Tainted: G    B           T 5.3.8 #1
> Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 0.0.0 02/06/2015
> RIP: 0010:print_trailer+0x70/0x1d5
> Code: 28 4d 8b 4d 00 4d 8b 45 20 81 e2 ff 7f 00 00 e8 86 ce ef ff 8b 4b 20 48 89 ea 48 89 ee 4c 29 e2 48 c7 c7 90 6f d4 89 48 01 e9 <48> 33 09 48 33 8b 70 01 00 00 e8 61 ce ef ff f6 43 09 04 74 35 8b
> RSP: 0018:ffffbf7680003d58 EFLAGS: 00010046
> RAX: 000000000000005d RBX: ffffa3d2bb08e540 RCX: 0000000000000000
> RDX: 00005c2d8fdc2000 RSI: 0000000000000000 RDI: ffffffff89d46f90
> RBP: 0000000000000000 R08: 0000000000000242 R09: 000000000000006c
> R10: 0000000000000000 R11: 0000000000000030 R12: ffffa3d27023e000
> R13: fffff11080c08f80 R14: ffffa3d2bb047a80 R15: 0000000000000002
> FS:  0000000000000000(0000) GS:ffffa3d2be400000(0000) knlGS:0000000000000000
> CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> CR2: 0000000000000000 CR3: 000000007a6c4000 CR4: 00000000000006f0
> Call Trace:
>  <IRQ>
>  free_debug_processing.cold.37+0xc9/0x149
>  ? __kfree_skb_flush+0x30/0x40
>  ? __kfree_skb_flush+0x30/0x40
>  __slab_free+0x22a/0x3d0
>  ? tcp_wfree+0x2a/0x140
>  ? __sock_wfree+0x1b/0x30
>  kmem_cache_free_bulk+0x415/0x420
>  ? __kfree_skb_flush+0x30/0x40
>  __kfree_skb_flush+0x30/0x40
>  net_rx_action+0x2dd/0x480
>  __do_softirq+0xf0/0x246
>  irq_exit+0x93/0xb0
>  do_IRQ+0xa0/0x110
>  common_interrupt+0xf/0xf
>  </IRQ>
>
> Given we're now almost identical to the existing debugging
> code which correctly walks the list, combine with that.
>
> Link: https://lkml.kernel.org/r/20191104170303.GA50361@gandi.net
> Reported-by: Thibaut Sautereau <thibaut.sautereau@clip-os.org>
> Fixes: 1b7e816fc80e ("mm: slub: Fix slab walking for init_on_free")
> Signed-off-by: Laura Abbott <labbott@redhat.com>
Acked-by: Alexander Potapenko <glider@google.com>
> ---
>  mm/slub.c | 39 +++++++++------------------------------
>  1 file changed, 9 insertions(+), 30 deletions(-)
>
> diff --git a/mm/slub.c b/mm/slub.c
> index dac41cf0b94a..d2445dd1c7ed 100644
> --- a/mm/slub.c
> +++ b/mm/slub.c
> @@ -1432,12 +1432,15 @@ static inline bool slab_free_freelist_hook(struct kmem_cache *s,
>         void *old_tail = *tail ? *tail : *head;
>         int rsize;
>
> -       if (slab_want_init_on_free(s)) {
> -               void *p = NULL;
> +       /* Head and tail of the reconstructed freelist */
> +       *head = NULL;
> +       *tail = NULL;
>
> -               do {
> -                       object = next;
> -                       next = get_freepointer(s, object);
> +       do {
> +               object = next;
> +               next = get_freepointer(s, object);
> +
> +               if (slab_want_init_on_free(s)) {
>                         /*
>                          * Clear the object and the metadata, but don't touch
>                          * the redzone.
> @@ -1447,29 +1450,8 @@ static inline bool slab_free_freelist_hook(struct kmem_cache *s,
>                                                            : 0;
>                         memset((char *)object + s->inuse, 0,
>                                s->size - s->inuse - rsize);
> -                       set_freepointer(s, object, p);
> -                       p = object;
> -               } while (object != old_tail);
> -       }
> -
> -/*
> - * Compiler cannot detect this function can be removed if slab_free_hook()
> - * evaluates to nothing.  Thus, catch all relevant config debug options here.
> - */
> -#if defined(CONFIG_LOCKDEP)    ||              \
> -       defined(CONFIG_DEBUG_KMEMLEAK) ||       \
> -       defined(CONFIG_DEBUG_OBJECTS_FREE) ||   \
> -       defined(CONFIG_KASAN)
>
> -       next = *head;
> -
> -       /* Head and tail of the reconstructed freelist */
> -       *head = NULL;
> -       *tail = NULL;
> -
> -       do {
> -               object = next;
> -               next = get_freepointer(s, object);
> +               }
>                 /* If object's reuse doesn't have to be delayed */
>                 if (!slab_free_hook(s, object)) {
>                         /* Move object to the new freelist */
> @@ -1484,9 +1466,6 @@ static inline bool slab_free_freelist_hook(struct kmem_cache *s,
>                 *tail = NULL;
>
>         return *head != NULL;
> -#else
> -       return true;
> -#endif
>  }
>
>  static void *setup_object(struct kmem_cache *s, struct page *page,
> --
> 2.21.0
>


-- 
Alexander Potapenko
Software Engineer

Google Germany GmbH
Erika-Mann-Straße, 33
80636 München

Geschäftsführer: Paul Manicle, Halimah DeLaine Prado
Registergericht und -nummer: Hamburg, HRB 86891
Sitz der Gesellschaft: Hamburg