| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <1588766821.7534.3.camel@mtksdccf07>
Date: Wed, 6 May 2020 20:07:01 +0800
From: Walter Wu <walter-zh.wu@...iatek.com>
To: Dmitry Vyukov <dvyukov@...gle.com>
CC: Andrey Ryabinin <aryabinin@...tuozzo.com>,
Alexander Potapenko <glider@...gle.com>,
Matthias Brugger <matthias.bgg@...il.com>,
kasan-dev <kasan-dev@...glegroups.com>,
Linux-MM <linux-mm@...ck.org>,
LKML <linux-kernel@...r.kernel.org>,
Linux ARM <linux-arm-kernel@...ts.infradead.org>,
wsd_upstream <wsd_upstream@...iatek.com>,
<linux-mediatek@...ts.infradead.org>
Subject: Re: [PATCH 2/3] kasan: record and print the free track
On Wed, 2020-05-06 at 13:59 +0200, Dmitry Vyukov wrote:
> On Wed, May 6, 2020 at 1:56 PM Walter Wu <walter-zh.wu@...iatek.com> wrote:
> >
> > On Wed, 2020-05-06 at 11:50 +0200, Dmitry Vyukov wrote:
> > > On Wed, May 6, 2020 at 7:22 AM Walter Wu <walter-zh.wu@...iatek.com> wrote:
> > > >
> > > > We add new KASAN_RCU_STACK_RECORD configuration option. It will move
> > > > free track from slub meta-data (struct kasan_alloc_meta) into freed object.
> > > > Because we hope this options doesn't enlarge slub meta-data size.
> > > >
> > > > This option doesn't enlarge struct kasan_alloc_meta size.
> > > > - add two call_rcu() call stack into kasan_alloc_meta, size is 8 bytes.
> > > > - remove free track from kasan_alloc_meta, size is 8 bytes.
> > > >
> > > > This option is only suitable for generic KASAN, because we move free track
> > > > into the freed object, so free track is valid information only when it
> > > > exists in quarantine. If the object is in-use state, then the KASAN report
> > > > doesn't print call_rcu() free track information.
> > > >
> > > > [1]https://bugzilla.kernel.org/show_bug.cgi?id=198437
> > > >
> > > > Signed-off-by: Walter Wu <walter-zh.wu@...iatek.com>
> > > > Cc: Andrey Ryabinin <aryabinin@...tuozzo.com>
> > > > Cc: Dmitry Vyukov <dvyukov@...gle.com>
> > > > Cc: Alexander Potapenko <glider@...gle.com>
> > > > ---
> > > > mm/kasan/common.c | 10 +++++++++-
> > > > mm/kasan/report.c | 24 +++++++++++++++++++++---
> > > > 2 files changed, 30 insertions(+), 4 deletions(-)
> > > >
> > > > diff --git a/mm/kasan/common.c b/mm/kasan/common.c
> > > > index 32d422bdf127..13ec03e225a7 100644
> > > > --- a/mm/kasan/common.c
> > > > +++ b/mm/kasan/common.c
> > > > @@ -321,8 +321,15 @@ void kasan_record_callrcu(void *addr)
> > > > /* record last call_rcu() call stack */
> > > > alloc_info->rcu_free_stack[1] = save_stack(GFP_NOWAIT);
> > > > }
> > > > -#endif
> > > >
> > > > +static void kasan_set_free_info(struct kmem_cache *cache,
> > > > + void *object, u8 tag)
> > > > +{
> > > > + /* store free track into freed object */
> > > > + set_track((struct kasan_track *)(object + BYTES_PER_WORD), GFP_NOWAIT);
> > > > +}
> > > > +
> > > > +#else
> > > > static void kasan_set_free_info(struct kmem_cache *cache,
> > > > void *object, u8 tag)
> > > > {
> > > > @@ -339,6 +346,7 @@ static void kasan_set_free_info(struct kmem_cache *cache,
> > > >
> > > > set_track(&alloc_meta->free_track[idx], GFP_NOWAIT);
> > > > }
> > > > +#endif
> > > >
> > > > void kasan_poison_slab(struct page *page)
> > > > {
> > > > diff --git a/mm/kasan/report.c b/mm/kasan/report.c
> > > > index 7aaccc70b65b..f2b0c6b9dffa 100644
> > > > --- a/mm/kasan/report.c
> > > > +++ b/mm/kasan/report.c
> > > > @@ -175,8 +175,23 @@ static void kasan_print_rcu_free_stack(struct kasan_alloc_meta *alloc_info)
> > > > print_track(&free_track, "Last call_rcu() call stack", true);
> > > > pr_err("\n");
> > > > }
> > > > -#endif
> > > >
> > > > +static struct kasan_track *kasan_get_free_track(struct kmem_cache *cache,
> > > > + void *object, u8 tag, const void *addr)
> > > > +{
> > > > + u8 *shadow_addr = (u8 *)kasan_mem_to_shadow(addr);
> > > > +
> > > > + /*
> > > > + * Only the freed object can get free track,
> > > > + * because free track information is stored to freed object.
> > > > + */
> > > > + if (*shadow_addr == KASAN_KMALLOC_FREE)
> > > > + return (struct kasan_track *)(object + BYTES_PER_WORD);
> > >
> > > Humm... the other patch defines BYTES_PER_WORD as 4... I would assume
> > > seeing 8 (or sizeof(long)) here. Why 4?
> > It should be a pointer size, maybe sizeof(long) makes more sense.
> >
> > > Have you tested all 4 modes (RCU/no-RCU x SLAB/SLUB)? As far as I
> > > remember one of the allocators stored something in the object.
> > Good question, I only tested in RCU x SLUB, would you tell mew how do
> > no-RCU? I will test them in v2 pathset.
>
> I meant with CONFIG_KASAN_RCU_STACK_RECORD=y and with
> CONFIG_KASAN_RCU_STACK_RECORD not set.
> But if we drop CONFIG_KASAN_RCU_STACK_RECORD config, then it halves
> the number of configurations to test ;)
>
Ok, I have be tested by RCU xSLUB and no_RCUxSLUB, It is workable. So I
only miss this SLAB combination.
>
> > >
> > > Also, does this work with objects with ctors and slabs destroyed by
> > > rcu? kasan_track may smash other things in these cases.
> > > Have you looked at the KASAN implementation when free_track was
> > > removed? That may have useful details :)
> > Set free_track before put into quarantine, free_track should not have to
> > be removed, it only have to overwirte itself.
> >
> > >
> > >
> > > > + else
> > > > + return NULL;
> > > > +}
> > > > +
> > > > +#else
> > > > static struct kasan_track *kasan_get_free_track(struct kmem_cache *cache,
> > > > void *object, u8 tag, const void *addr)
> > > > {
> > > > @@ -196,6 +211,7 @@ static struct kasan_track *kasan_get_free_track(struct kmem_cache *cache,
> > > >
> > > > return &alloc_meta->free_track[i];
> > > > }
> > > > +#endif
> > > >
> > > > static void describe_object(struct kmem_cache *cache, void *object,
> > > > const void *addr, u8 tag)
> > > > @@ -208,8 +224,10 @@ static void describe_object(struct kmem_cache *cache, void *object,
> > > > print_track(&alloc_info->alloc_track, "Allocated", false);
> > > > pr_err("\n");
> > > > free_track = kasan_get_free_track(cache, object, tag, addr);
> > > > - print_track(free_track, "Freed", false);
> > > > - pr_err("\n");
> > > > + if (free_track) {
> > > > + print_track(free_track, "Freed", false);
> > > > + pr_err("\n");
> > > > + }
> > > > #ifdef CONFIG_KASAN_RCU_STACK_RECORD
> > > > kasan_print_rcu_free_stack(alloc_info);
> > > > #endif
> > > > --
> > > > 2.18.0
Powered by blists - more mailing lists