From: Marco Elver <elver@google.com>
To: andrey.konovalov@linux.dev
Cc: Alexander Potapenko <glider@google.com>,
Andrey Konovalov <andreyknvl@gmail.com>,
Dmitry Vyukov <dvyukov@google.com>,
Vlastimil Babka <vbabka@suse.cz>,
kasan-dev@googlegroups.com,
Evgenii Stepanov <eugenis@google.com>,
Oscar Salvador <osalvador@suse.de>,
Andrew Morton <akpm@linux-foundation.org>,
linux-mm@kvack.org, linux-kernel@vger.kernel.org,
Andrey Konovalov <andreyknvl@google.com>
Subject: Re: [PATCH RFC] kasan: use stack_depot_put for Generic mode
Date: Tue, 14 Nov 2023 14:36:15 +0100 [thread overview]
Message-ID: <CANpmjNMJJyiaEWKouY4jho-Qg1+i7eYSxdjn_vEPCbQ0AR9Sew@mail.gmail.com> (raw)
In-Reply-To: <20231103212724.134597-1-andrey.konovalov@linux.dev>
On Fri, 3 Nov 2023 at 22:27, <andrey.konovalov@linux.dev> wrote:
>
> From: Andrey Konovalov <andreyknvl@google.com>
>
> Evict alloc/free stack traces from the stack depot for Generic KASAN
> once they are evicted from the quaratine.
>
> For auxiliary stack traces, evict the oldest stack trace once a new one
> is saved (KASAN only keeps references to the last two).
>
> Also evict all save stack traces on krealloc.
>
> To avoid double-evicting and mis-evicting stack traces (in case KASAN's
> metadata was corrupted), reset KASAN's per-object metadata that stores
> stack depot handles when the object is initialized and when it's evicted
> from the quarantine.
>
> Note that stack_depot_put is no-op of the handle is 0.
>
> Signed-off-by: Andrey Konovalov <andreyknvl@google.com>
Reviewed-by: Marco Elver <elver@google.com>
Maybe mention the space and performance difference from your
experiments. As-is, it's a bit cryptic what the benefit is. I assume
this patch goes along with the other series.
> ---
>
> This goes on top of the "stackdepot: allow evicting stack traces" series.
> I'll mail the patches all together after the merge window.
> ---
> mm/kasan/common.c | 3 ++-
> mm/kasan/generic.c | 22 ++++++++++++++++++----
> mm/kasan/quarantine.c | 26 ++++++++++++++++++++------
> 3 files changed, 40 insertions(+), 11 deletions(-)
>
> diff --git a/mm/kasan/common.c b/mm/kasan/common.c
> index 825a0240ec02..b5d8bd26fced 100644
> --- a/mm/kasan/common.c
> +++ b/mm/kasan/common.c
> @@ -50,7 +50,8 @@ depot_stack_handle_t kasan_save_stack(gfp_t flags, depot_flags_t depot_flags)
> void kasan_set_track(struct kasan_track *track, gfp_t flags)
> {
> track->pid = current->pid;
> - track->stack = kasan_save_stack(flags, STACK_DEPOT_FLAG_CAN_ALLOC);
> + track->stack = kasan_save_stack(flags,
> + STACK_DEPOT_FLAG_CAN_ALLOC | STACK_DEPOT_FLAG_GET);
> }
>
> #if defined(CONFIG_KASAN_GENERIC) || defined(CONFIG_KASAN_SW_TAGS)
> diff --git a/mm/kasan/generic.c b/mm/kasan/generic.c
> index 5d168c9afb32..50cc519e23f4 100644
> --- a/mm/kasan/generic.c
> +++ b/mm/kasan/generic.c
> @@ -449,10 +449,14 @@ struct kasan_free_meta *kasan_get_free_meta(struct kmem_cache *cache,
> void kasan_init_object_meta(struct kmem_cache *cache, const void *object)
> {
> struct kasan_alloc_meta *alloc_meta;
> + struct kasan_free_meta *free_meta;
>
> alloc_meta = kasan_get_alloc_meta(cache, object);
> if (alloc_meta)
> __memset(alloc_meta, 0, sizeof(*alloc_meta));
> + free_meta = kasan_get_free_meta(cache, object);
> + if (free_meta)
> + __memset(free_meta, 0, sizeof(*free_meta));
> }
>
> size_t kasan_metadata_size(struct kmem_cache *cache, bool in_object)
> @@ -489,18 +493,20 @@ static void __kasan_record_aux_stack(void *addr, depot_flags_t depot_flags)
> if (!alloc_meta)
> return;
>
> + stack_depot_put(alloc_meta->aux_stack[1]);
> alloc_meta->aux_stack[1] = alloc_meta->aux_stack[0];
> alloc_meta->aux_stack[0] = kasan_save_stack(0, depot_flags);
> }
>
> void kasan_record_aux_stack(void *addr)
> {
> - return __kasan_record_aux_stack(addr, STACK_DEPOT_FLAG_CAN_ALLOC);
> + return __kasan_record_aux_stack(addr,
> + STACK_DEPOT_FLAG_CAN_ALLOC | STACK_DEPOT_FLAG_GET);
> }
>
> void kasan_record_aux_stack_noalloc(void *addr)
> {
> - return __kasan_record_aux_stack(addr, 0);
> + return __kasan_record_aux_stack(addr, STACK_DEPOT_FLAG_GET);
> }
>
> void kasan_save_alloc_info(struct kmem_cache *cache, void *object, gfp_t flags)
> @@ -508,8 +514,16 @@ void kasan_save_alloc_info(struct kmem_cache *cache, void *object, gfp_t flags)
> struct kasan_alloc_meta *alloc_meta;
>
> alloc_meta = kasan_get_alloc_meta(cache, object);
> - if (alloc_meta)
> - kasan_set_track(&alloc_meta->alloc_track, flags);
> + if (!alloc_meta)
> + return;
> +
> + /* Evict previous stack traces (might exist for krealloc). */
> + stack_depot_put(alloc_meta->alloc_track.stack);
> + stack_depot_put(alloc_meta->aux_stack[0]);
> + stack_depot_put(alloc_meta->aux_stack[1]);
> + __memset(alloc_meta, 0, sizeof(*alloc_meta));
> +
> + kasan_set_track(&alloc_meta->alloc_track, flags);
> }
>
> void kasan_save_free_info(struct kmem_cache *cache, void *object)
> diff --git a/mm/kasan/quarantine.c b/mm/kasan/quarantine.c
> index 152dca73f398..37fb0e3f5876 100644
> --- a/mm/kasan/quarantine.c
> +++ b/mm/kasan/quarantine.c
> @@ -141,11 +141,22 @@ static void *qlink_to_object(struct qlist_node *qlink, struct kmem_cache *cache)
> static void qlink_free(struct qlist_node *qlink, struct kmem_cache *cache)
> {
> void *object = qlink_to_object(qlink, cache);
> - struct kasan_free_meta *meta = kasan_get_free_meta(cache, object);
> + struct kasan_alloc_meta *alloc_meta = kasan_get_alloc_meta(cache, object);
> + struct kasan_free_meta *free_meta = kasan_get_free_meta(cache, object);
> unsigned long flags;
>
> - if (IS_ENABLED(CONFIG_SLAB))
> - local_irq_save(flags);
> + if (alloc_meta) {
> + stack_depot_put(alloc_meta->alloc_track.stack);
> + stack_depot_put(alloc_meta->aux_stack[0]);
> + stack_depot_put(alloc_meta->aux_stack[1]);
> + __memset(alloc_meta, 0, sizeof(*alloc_meta));
> + }
> +
> + if (free_meta &&
> + *(u8 *)kasan_mem_to_shadow(object) == KASAN_SLAB_FREETRACK) {
> + stack_depot_put(free_meta->free_track.stack);
> + free_meta->free_track.stack = 0;
> + }
>
> /*
> * If init_on_free is enabled and KASAN's free metadata is stored in
> @@ -155,14 +166,17 @@ static void qlink_free(struct qlist_node *qlink, struct kmem_cache *cache)
> */
> if (slab_want_init_on_free(cache) &&
> cache->kasan_info.free_meta_offset == 0)
> - memzero_explicit(meta, sizeof(*meta));
> + memzero_explicit(free_meta, sizeof(*free_meta));
>
> /*
> - * As the object now gets freed from the quarantine, assume that its
> - * free track is no longer valid.
> + * As the object now gets freed from the quarantine,
> + * take note that its free track is no longer exists.
> */
> *(u8 *)kasan_mem_to_shadow(object) = KASAN_SLAB_FREE;
>
> + if (IS_ENABLED(CONFIG_SLAB))
> + local_irq_save(flags);
> +
> ___cache_free(cache, object, _THIS_IP_);
>
> if (IS_ENABLED(CONFIG_SLAB))
> --
> 2.25.1
>
prev parent reply other threads:[~2023-11-14 13:36 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-11-03 21:27 andrey.konovalov
2023-11-14 13:36 ` Marco Elver [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=CANpmjNMJJyiaEWKouY4jho-Qg1+i7eYSxdjn_vEPCbQ0AR9Sew@mail.gmail.com \
--to=elver@google.com \
--cc=akpm@linux-foundation.org \
--cc=andrey.konovalov@linux.dev \
--cc=andreyknvl@gmail.com \
--cc=andreyknvl@google.com \
--cc=dvyukov@google.com \
--cc=eugenis@google.com \
--cc=glider@google.com \
--cc=kasan-dev@googlegroups.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=osalvador@suse.de \
--cc=vbabka@suse.cz \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox