On Thu, Nov 03, 2022 at 11:16:53PM +0900, Hyeonggon Yoo wrote: > On Tue, Nov 01, 2022 at 03:33:11PM -0700, Kees Cook wrote: > > Since GCC cannot apply the __alloc_size attributes to inlines[1], all > > allocator inlines need to explicitly call into extern functions that > > contain a size argument. Provide these wrappers that end up just > > ignoring the size argument for the actual allocation. > > > > This allows CONFIG_FORTIFY_SOURCE=y to see all various dynamic allocation > > sizes under GCC 12+ and all supported Clang versions. > > > > [1] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96503 > > > > Cc: Vlastimil Babka <vbabka@xxxxxxx> > > Cc: Christoph Lameter <cl@xxxxxxxxx> > > Cc: Pekka Enberg <penberg@xxxxxxxxxx> > > Cc: David Rientjes <rientjes@xxxxxxxxxx> > > Cc: Joonsoo Kim <iamjoonsoo.kim@xxxxxxx> > > Cc: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> > > Cc: Roman Gushchin <roman.gushchin@xxxxxxxxx> > > Cc: Hyeonggon Yoo <42.hyeyoo@xxxxxxxxx> > > Cc: linux-mm@xxxxxxxxx > > Signed-off-by: Kees Cook <keescook@xxxxxxxxxxxx> > > --- > > include/linux/slab.h | 8 ++++++-- > > mm/slab_common.c | 14 ++++++++++++++ > > 2 files changed, 20 insertions(+), 2 deletions(-) > > > > diff --git a/include/linux/slab.h b/include/linux/slab.h > > index 970e9504949e..051d86ca31a8 100644 > > --- a/include/linux/slab.h > > +++ b/include/linux/slab.h > > @@ -442,6 +442,8 @@ static_assert(PAGE_SHIFT <= 20); > > > > void *__kmalloc(size_t size, gfp_t flags) __assume_kmalloc_alignment __alloc_size(1); > > void *kmem_cache_alloc(struct kmem_cache *s, gfp_t flags) __assume_slab_alignment __malloc; > > +void *kmem_cache_alloc_sized(struct kmem_cache *s, gfp_t flags, size_t size) > > + __assume_slab_alignment __alloc_size(3); > > void *kmem_cache_alloc_lru(struct kmem_cache *s, struct list_lru *lru, > > gfp_t gfpflags) __assume_slab_alignment __malloc; > > void kmem_cache_free(struct kmem_cache *s, void *objp); > > @@ -469,6 +471,8 @@ void *__kmalloc_node(size_t size, gfp_t flags, int node) __assume_kmalloc_alignm > > __alloc_size(1); > > void *kmem_cache_alloc_node(struct kmem_cache *s, gfp_t flags, int node) __assume_slab_alignment > > __malloc; > > +void *kmem_cache_alloc_node_sized(struct kmem_cache *s, gfp_t flags, int node, size_t size) > > + __assume_slab_alignment __alloc_size(4); > > > > #ifdef CONFIG_TRACING > > void *kmalloc_trace(struct kmem_cache *s, gfp_t flags, size_t size) > > @@ -482,7 +486,7 @@ void *kmalloc_node_trace(struct kmem_cache *s, gfp_t gfpflags, > > static __always_inline __alloc_size(3) > > void *kmalloc_trace(struct kmem_cache *s, gfp_t flags, size_t size) > > { > > - void *ret = kmem_cache_alloc(s, flags); > > + void *ret = kmem_cache_alloc_sized(s, flags, size); > > > > ret = kasan_kmalloc(s, ret, size, flags); > > return ret; > > @@ -492,7 +496,7 @@ static __always_inline __alloc_size(4) > > void *kmalloc_node_trace(struct kmem_cache *s, gfp_t gfpflags, > > int node, size_t size) > > { > > - void *ret = kmem_cache_alloc_node(s, gfpflags, node); > > + void *ret = kmem_cache_alloc_node_sized(s, gfpflags, node, size); > > > > ret = kasan_kmalloc(s, ret, size, gfpflags); > > return ret; > > diff --git a/mm/slab_common.c b/mm/slab_common.c > > index 33b1886b06eb..5fa547539a6a 100644 > > --- a/mm/slab_common.c > > +++ b/mm/slab_common.c > > @@ -1457,6 +1457,20 @@ size_t ksize(const void *objp) > > } > > EXPORT_SYMBOL(ksize); > > > > +/* Wrapper so __alloc_size() can see the actual allocation size. */ > > +void *kmem_cache_alloc_sized(struct kmem_cache *s, gfp_t flags, size_t size) > > +{ > > + return kmem_cache_alloc(s, flags); > > +} > > +EXPORT_SYMBOL(kmem_cache_alloc_sized); > > + > > +/* Wrapper so __alloc_size() can see the actual allocation size. */ > > +void *kmem_cache_alloc_node_sized(struct kmem_cache *s, gfp_t flags, int node, size_t size) > > +{ > > + return kmem_cache_alloc_node(s, flags, node); > > +} > > +EXPORT_SYMBOL(kmem_cache_alloc_node_sized); > > The reason that we have two implementations of kmalloc_trace* > depending on CONFIG_TRACING is to save additional function call when > CONFIG_TRACING is not set. > > With this patch there is no reason to keep both. > So let's drop #ifdefs and use single implementation in mm/slab_common.c. Okay, I'll respin... -- Kees Cook
