[PATCH 4/5] kasan, mm: integrate slab init_on_alloc with HW_TAGS
Marco Elver
elver at google.com
Mon Mar 8 11:36:59 GMT 2021
On Sat, Mar 06, 2021 at 01:15AM +0100, Andrey Konovalov wrote:
> This change uses the previously added memory initialization feature
> of HW_TAGS KASAN routines for slab memory when init_on_alloc is enabled.
>
> With this change, memory initialization memset() is no longer called
> when both HW_TAGS KASAN and init_on_alloc are enabled. Instead, memory
> is initialized in KASAN runtime.
>
> The memory initialization memset() is moved into slab_post_alloc_hook()
> that currently directly follows the initialization loop. A new argument
> is added to slab_post_alloc_hook() that indicates whether to initialize
> the memory or not.
>
> To avoid discrepancies with which memory gets initialized that can be
> caused by future changes, both KASAN hook and initialization memset()
> are put together and a warning comment is added.
>
> Combining setting allocation tags with memory initialization improves
> HW_TAGS KASAN performance when init_on_alloc is enabled.
>
> Signed-off-by: Andrey Konovalov <andreyknvl at google.com>
> ---
> include/linux/kasan.h | 8 ++++----
> mm/kasan/common.c | 4 ++--
> mm/slab.c | 28 +++++++++++++---------------
> mm/slab.h | 17 +++++++++++++----
> mm/slub.c | 27 +++++++++++----------------
> 5 files changed, 43 insertions(+), 41 deletions(-)
>
> diff --git a/include/linux/kasan.h b/include/linux/kasan.h
> index 4c0f414a893b..bb756f6c73b5 100644
> --- a/include/linux/kasan.h
> +++ b/include/linux/kasan.h
> @@ -216,12 +216,12 @@ static __always_inline void kasan_slab_free_mempool(void *ptr)
> }
>
> void * __must_check __kasan_slab_alloc(struct kmem_cache *s,
> - void *object, gfp_t flags);
> + void *object, gfp_t flags, bool init);
> static __always_inline void * __must_check kasan_slab_alloc(
> - struct kmem_cache *s, void *object, gfp_t flags)
> + struct kmem_cache *s, void *object, gfp_t flags, bool init)
> {
> if (kasan_enabled())
> - return __kasan_slab_alloc(s, object, flags);
> + return __kasan_slab_alloc(s, object, flags, init);
> return object;
> }
>
> @@ -306,7 +306,7 @@ static inline bool kasan_slab_free(struct kmem_cache *s, void *object)
> static inline void kasan_kfree_large(void *ptr) {}
> static inline void kasan_slab_free_mempool(void *ptr) {}
> static inline void *kasan_slab_alloc(struct kmem_cache *s, void *object,
> - gfp_t flags)
> + gfp_t flags, bool init)
> {
> return object;
> }
> diff --git a/mm/kasan/common.c b/mm/kasan/common.c
> index 6107c795611f..7ea747b18c26 100644
> --- a/mm/kasan/common.c
> +++ b/mm/kasan/common.c
> @@ -428,7 +428,7 @@ static void set_alloc_info(struct kmem_cache *cache, void *object,
> }
>
> void * __must_check __kasan_slab_alloc(struct kmem_cache *cache,
> - void *object, gfp_t flags)
> + void *object, gfp_t flags, bool init)
> {
> u8 tag;
> void *tagged_object;
> @@ -453,7 +453,7 @@ void * __must_check __kasan_slab_alloc(struct kmem_cache *cache,
> * Unpoison the whole object.
> * For kmalloc() allocations, kasan_kmalloc() will do precise poisoning.
> */
> - kasan_unpoison(tagged_object, cache->object_size, false);
> + kasan_unpoison(tagged_object, cache->object_size, init);
>
> /* Save alloc info (if possible) for non-kmalloc() allocations. */
> if (kasan_stack_collection_enabled())
> diff --git a/mm/slab.c b/mm/slab.c
> index 51fd424e0d6d..936dd686dec9 100644
> --- a/mm/slab.c
> +++ b/mm/slab.c
> @@ -3216,6 +3216,7 @@ slab_alloc_node(struct kmem_cache *cachep, gfp_t flags, int nodeid, size_t orig_
> void *ptr;
> int slab_node = numa_mem_id();
> struct obj_cgroup *objcg = NULL;
> + bool init = false;
>
> flags &= gfp_allowed_mask;
> cachep = slab_pre_alloc_hook(cachep, &objcg, 1, flags);
> @@ -3254,12 +3255,10 @@ slab_alloc_node(struct kmem_cache *cachep, gfp_t flags, int nodeid, size_t orig_
> out:
> local_irq_restore(save_flags);
> ptr = cache_alloc_debugcheck_after(cachep, flags, ptr, caller);
> -
> - if (unlikely(slab_want_init_on_alloc(flags, cachep)) && ptr)
> - memset(ptr, 0, cachep->object_size);
> + init = slab_want_init_on_alloc(flags, cachep);
>
> out_hooks:
> - slab_post_alloc_hook(cachep, objcg, flags, 1, &ptr);
> + slab_post_alloc_hook(cachep, objcg, flags, 1, &ptr, init);
> return ptr;
> }
>
> @@ -3301,6 +3300,7 @@ slab_alloc(struct kmem_cache *cachep, gfp_t flags, size_t orig_size, unsigned lo
> unsigned long save_flags;
> void *objp;
> struct obj_cgroup *objcg = NULL;
> + bool init = false;
>
> flags &= gfp_allowed_mask;
> cachep = slab_pre_alloc_hook(cachep, &objcg, 1, flags);
> @@ -3317,12 +3317,10 @@ slab_alloc(struct kmem_cache *cachep, gfp_t flags, size_t orig_size, unsigned lo
> local_irq_restore(save_flags);
> objp = cache_alloc_debugcheck_after(cachep, flags, objp, caller);
> prefetchw(objp);
> -
> - if (unlikely(slab_want_init_on_alloc(flags, cachep)) && objp)
> - memset(objp, 0, cachep->object_size);
> + init = slab_want_init_on_alloc(flags, cachep);
>
> out:
> - slab_post_alloc_hook(cachep, objcg, flags, 1, &objp);
> + slab_post_alloc_hook(cachep, objcg, flags, 1, &objp, init);
> return objp;
> }
>
> @@ -3542,18 +3540,18 @@ int kmem_cache_alloc_bulk(struct kmem_cache *s, gfp_t flags, size_t size,
>
> cache_alloc_debugcheck_after_bulk(s, flags, size, p, _RET_IP_);
>
> - /* Clear memory outside IRQ disabled section */
> - if (unlikely(slab_want_init_on_alloc(flags, s)))
> - for (i = 0; i < size; i++)
> - memset(p[i], 0, s->object_size);
> -
> - slab_post_alloc_hook(s, objcg, flags, size, p);
> + /*
> + * memcg and kmem_cache debug support and memory initialization.
> + * Done outside of the IRQ disabled section.
> + */
> + slab_post_alloc_hook(s, objcg, flags, size, p,
> + slab_want_init_on_alloc(flags, s));
> /* FIXME: Trace call missing. Christoph would like a bulk variant */
> return size;
> error:
> local_irq_enable();
> cache_alloc_debugcheck_after_bulk(s, flags, i, p, _RET_IP_);
> - slab_post_alloc_hook(s, objcg, flags, i, p);
> + slab_post_alloc_hook(s, objcg, flags, i, p, false);
> __kmem_cache_free_bulk(s, i, p);
> return 0;
> }
> diff --git a/mm/slab.h b/mm/slab.h
> index 076582f58f68..0116a314cd21 100644
> --- a/mm/slab.h
> +++ b/mm/slab.h
> @@ -506,15 +506,24 @@ static inline struct kmem_cache *slab_pre_alloc_hook(struct kmem_cache *s,
> }
>
> static inline void slab_post_alloc_hook(struct kmem_cache *s,
> - struct obj_cgroup *objcg,
> - gfp_t flags, size_t size, void **p)
> + struct obj_cgroup *objcg, gfp_t flags,
> + size_t size, void **p, bool init)
> {
> size_t i;
>
> flags &= gfp_allowed_mask;
> +
> + /*
> + * As memory initialization is integrated with hardware tag-based
> + * KASAN, kasan_slab_alloc and initialization memset must be
> + * kept together to avoid discrepancies in behavior.
> + *
> + * As p[i] might get tagged, memset and kmemleak hook come after KASAN.
> + */
> for (i = 0; i < size; i++) {
> - p[i] = kasan_slab_alloc(s, p[i], flags);
> - /* As p[i] might get tagged, call kmemleak hook after KASAN. */
> + p[i] = kasan_slab_alloc(s, p[i], flags, init);
> + if (p[i] && init && !IS_ENABLED(CONFIG_KASAN_HW_TAGS))
Same as the other patch, it seems awkward to have
!IS_ENABLED(CONFIG_KASAN_HW_TAGS) rather than a kasan_*() helper that
returns this information.
> + memset(p[i], 0, s->object_size);
> kmemleak_alloc_recursive(p[i], s->object_size, 1,
> s->flags, flags);
> }
> diff --git a/mm/slub.c b/mm/slub.c
> index e26c274b4657..f53df23760e3 100644
> --- a/mm/slub.c
> +++ b/mm/slub.c
> @@ -2822,6 +2822,7 @@ static __always_inline void *slab_alloc_node(struct kmem_cache *s,
> struct page *page;
> unsigned long tid;
> struct obj_cgroup *objcg = NULL;
> + bool init = false;
>
> s = slab_pre_alloc_hook(s, &objcg, 1, gfpflags);
> if (!s)
> @@ -2899,12 +2900,10 @@ static __always_inline void *slab_alloc_node(struct kmem_cache *s,
> }
>
> maybe_wipe_obj_freeptr(s, object);
> -
> - if (unlikely(slab_want_init_on_alloc(gfpflags, s)) && object)
> - memset(kasan_reset_tag(object), 0, s->object_size);
> + init = slab_want_init_on_alloc(gfpflags, s);
>
> out:
> - slab_post_alloc_hook(s, objcg, gfpflags, 1, &object);
> + slab_post_alloc_hook(s, objcg, gfpflags, 1, &object, init);
>
> return object;
> }
> @@ -3356,20 +3355,16 @@ int kmem_cache_alloc_bulk(struct kmem_cache *s, gfp_t flags, size_t size,
> c->tid = next_tid(c->tid);
> local_irq_enable();
>
> - /* Clear memory outside IRQ disabled fastpath loop */
> - if (unlikely(slab_want_init_on_alloc(flags, s))) {
> - int j;
> -
> - for (j = 0; j < i; j++)
> - memset(kasan_reset_tag(p[j]), 0, s->object_size);
> - }
> -
> - /* memcg and kmem_cache debug support */
> - slab_post_alloc_hook(s, objcg, flags, size, p);
> + /*
> + * memcg and kmem_cache debug support and memory initialization.
> + * Done outside of the IRQ disabled fastpath loop.
> + */
> + slab_post_alloc_hook(s, objcg, flags, size, p,
> + slab_want_init_on_alloc(flags, s));
> return i;
> error:
> local_irq_enable();
> - slab_post_alloc_hook(s, objcg, flags, i, p);
> + slab_post_alloc_hook(s, objcg, flags, i, p, false);
> __kmem_cache_free_bulk(s, i, p);
> return 0;
> }
> @@ -3579,7 +3574,7 @@ static void early_kmem_cache_node_alloc(int node)
> init_object(kmem_cache_node, n, SLUB_RED_ACTIVE);
> init_tracking(kmem_cache_node, n);
> #endif
> - n = kasan_slab_alloc(kmem_cache_node, n, GFP_KERNEL);
> + n = kasan_slab_alloc(kmem_cache_node, n, GFP_KERNEL, false);
> page->freelist = get_freepointer(kmem_cache_node, n);
> page->inuse = 1;
> page->frozen = 0;
> --
> 2.30.1.766.gb4fecdf3b7-goog
>
More information about the linux-arm-kernel
mailing list