From: Marco Elver <elver@google.com>
To: andrey.konovalov@linux.dev
Cc: Andrew Morton <akpm@linux-foundation.org>,
Juntong Deng <juntong.deng@outlook.com>,
Andrey Konovalov <andreyknvl@gmail.com>,
Alexander Potapenko <glider@google.com>,
Dmitry Vyukov <dvyukov@google.com>,
Andrey Ryabinin <ryabinin.a.a@gmail.com>,
kasan-dev@googlegroups.com, linux-mm@kvack.org,
linux-kernel@vger.kernel.org,
Andrey Konovalov <andreyknvl@google.com>
Subject: Re: [PATCH mm 1/4] kasan: clean up kasan_cache_create
Date: Thu, 21 Dec 2023 21:08:04 +0100 [thread overview]
Message-ID: <CANpmjNMezNWBoH8m38XO2=dP9KQk+_Vb8bo41F7ytQVdbEe-3g@mail.gmail.com> (raw)
In-Reply-To: <20231221183540.168428-1-andrey.konovalov@linux.dev>
On Thu, 21 Dec 2023 at 19:35, <andrey.konovalov@linux.dev> wrote:
>
> From: Andrey Konovalov <andreyknvl@google.com>
>
> Reorganize the code to avoid nested if/else checks to improve the
> readability.
>
> Also drop the confusing comments about KMALLOC_MAX_SIZE checks: they
> are relevant for both SLUB and SLAB (originally, the comments likely
> confused KMALLOC_MAX_SIZE with KMALLOC_MAX_CACHE_SIZE).
>
> Fixes: a5989d4ed40c ("kasan: improve free meta storage in Generic KASAN")
> Signed-off-by: Andrey Konovalov <andreyknvl@google.com>
Reviewed-by: Marco Elver <elver@google.com>
> ---
> mm/kasan/generic.c | 67 +++++++++++++++++++++++++++-------------------
> 1 file changed, 39 insertions(+), 28 deletions(-)
>
> diff --git a/mm/kasan/generic.c b/mm/kasan/generic.c
> index 54e20b2bc3e1..769e43e05d0b 100644
> --- a/mm/kasan/generic.c
> +++ b/mm/kasan/generic.c
> @@ -381,16 +381,11 @@ void kasan_cache_create(struct kmem_cache *cache, unsigned int *size,
>
> ok_size = *size;
>
> - /* Add alloc meta into redzone. */
> + /* Add alloc meta into the redzone. */
> cache->kasan_info.alloc_meta_offset = *size;
> *size += sizeof(struct kasan_alloc_meta);
>
> - /*
> - * If alloc meta doesn't fit, don't add it.
> - * This can only happen with SLAB, as it has KMALLOC_MAX_SIZE equal
> - * to KMALLOC_MAX_CACHE_SIZE and doesn't fall back to page_alloc for
> - * larger sizes.
> - */
> + /* If alloc meta doesn't fit, don't add it. */
> if (*size > KMALLOC_MAX_SIZE) {
> cache->kasan_info.alloc_meta_offset = 0;
> *size = ok_size;
> @@ -401,36 +396,52 @@ void kasan_cache_create(struct kmem_cache *cache, unsigned int *size,
> orig_alloc_meta_offset = cache->kasan_info.alloc_meta_offset;
>
> /*
> - * Add free meta into redzone when it's not possible to store
> + * Store free meta in the redzone when it's not possible to store
> * it in the object. This is the case when:
> * 1. Object is SLAB_TYPESAFE_BY_RCU, which means that it can
> * be touched after it was freed, or
> * 2. Object has a constructor, which means it's expected to
> - * retain its content until the next allocation, or
> - * 3. Object is too small and SLUB DEBUG is enabled. Avoid
> - * free meta that exceeds the object size corrupts the
> - * SLUB DEBUG metadata.
> - * Otherwise cache->kasan_info.free_meta_offset = 0 is implied.
> - * If the object is smaller than the free meta and SLUB DEBUG
> - * is not enabled, it is still possible to store part of the
> - * free meta in the object.
> + * retain its content until the next allocation.
> */
> if ((cache->flags & SLAB_TYPESAFE_BY_RCU) || cache->ctor) {
> cache->kasan_info.free_meta_offset = *size;
> *size += sizeof(struct kasan_free_meta);
> - } else if (cache->object_size < sizeof(struct kasan_free_meta)) {
> - if (__slub_debug_enabled()) {
> - cache->kasan_info.free_meta_offset = *size;
> - *size += sizeof(struct kasan_free_meta);
> - } else {
> - rem_free_meta_size = sizeof(struct kasan_free_meta) -
> - cache->object_size;
> - *size += rem_free_meta_size;
> - if (cache->kasan_info.alloc_meta_offset != 0)
> - cache->kasan_info.alloc_meta_offset += rem_free_meta_size;
> - }
> + goto free_meta_added;
> + }
> +
> + /*
> + * Otherwise, if the object is large enough to contain free meta,
> + * store it within the object.
> + */
> + if (sizeof(struct kasan_free_meta) <= cache->object_size) {
> + /* cache->kasan_info.free_meta_offset = 0 is implied. */
> + goto free_meta_added;
> }
>
> + /*
> + * For smaller objects, store the beginning of free meta within the
> + * object and the end in the redzone. And thus shift the location of
> + * alloc meta to free up space for free meta.
> + * This is only possible when slub_debug is disabled, as otherwise
> + * the end of free meta will overlap with slub_debug metadata.
> + */
> + if (!__slub_debug_enabled()) {
> + rem_free_meta_size = sizeof(struct kasan_free_meta) -
> + cache->object_size;
> + *size += rem_free_meta_size;
> + if (cache->kasan_info.alloc_meta_offset != 0)
> + cache->kasan_info.alloc_meta_offset += rem_free_meta_size;
> + goto free_meta_added;
> + }
> +
> + /*
> + * If the object is small and slub_debug is enabled, store free meta
> + * in the redzone after alloc meta.
> + */
> + cache->kasan_info.free_meta_offset = *size;
> + *size += sizeof(struct kasan_free_meta);
> +
> +free_meta_added:
> /* If free meta doesn't fit, don't add it. */
> if (*size > KMALLOC_MAX_SIZE) {
> cache->kasan_info.free_meta_offset = KASAN_NO_FREE_META;
> @@ -440,7 +451,7 @@ void kasan_cache_create(struct kmem_cache *cache, unsigned int *size,
>
> /* Calculate size with optimal redzone. */
> optimal_size = cache->object_size + optimal_redzone(cache->object_size);
> - /* Limit it with KMALLOC_MAX_SIZE (relevant for SLAB only). */
> + /* Limit it with KMALLOC_MAX_SIZE. */
> if (optimal_size > KMALLOC_MAX_SIZE)
> optimal_size = KMALLOC_MAX_SIZE;
> /* Use optimal size if the size with added metas is not large enough. */
> --
> 2.25.1
>
prev parent reply other threads:[~2023-12-21 20:08 UTC|newest]
Thread overview: 11+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-12-21 18:35 andrey.konovalov
2023-12-21 18:35 ` [PATCH mm 2/4] kasan: reuse kasan_track in kasan_stack_ring_entry andrey.konovalov
2023-12-21 20:10 ` Marco Elver
2023-12-21 20:19 ` Andrey Konovalov
2023-12-21 18:35 ` [PATCH mm 3/4] kasan: simplify saving extra info into tracks andrey.konovalov
2023-12-21 20:11 ` Marco Elver
2023-12-21 20:21 ` Andrey Konovalov
2023-12-21 18:35 ` [PATCH mm 4/4] kasan: simplify kasan_complete_mode_report_info for tag-based modes andrey.konovalov
2023-12-21 20:13 ` Marco Elver
2023-12-21 20:22 ` Andrey Konovalov
2023-12-21 20:08 ` 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='CANpmjNMezNWBoH8m38XO2=dP9KQk+_Vb8bo41F7ytQVdbEe-3g@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=glider@google.com \
--cc=juntong.deng@outlook.com \
--cc=kasan-dev@googlegroups.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=ryabinin.a.a@gmail.com \
/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