mm: krealloc: Fix MTE false alarm in __do_krealloc

[PATCH] mm: krealloc: Fix MTE false alarm in __do_krealloc

Posted by Qun-Wei Lin 1 month ago

This patch addresses an issue introduced by commit 1a83a716ec233 ("mm:
krealloc: consider spare memory for __GFP_ZERO") which causes MTE
(Memory Tagging Extension) to falsely report a slab-out-of-bounds error.

The problem occurs when zeroing out spare memory in __do_krealloc. The
original code only considered software-based KASAN and did not account
for MTE. It does not reset the KASAN tag before calling memset, leading
to a mismatch between the pointer tag and the memory tag, resulting
in a false positive.

Example of the error:
==================================================================
swapper/0: BUG: KASAN: slab-out-of-bounds in __memset+0x84/0x188
swapper/0: Write at addr f4ffff8005f0fdf0 by task swapper/0/1
swapper/0: Pointer tag: [f4], memory tag: [fe]
swapper/0:
swapper/0: CPU: 4 UID: 0 PID: 1 Comm: swapper/0 Not tainted 6.12.
swapper/0: Hardware name: MT6991(ENG) (DT)
swapper/0: Call trace:
swapper/0:  dump_backtrace+0xfc/0x17c
swapper/0:  show_stack+0x18/0x28
swapper/0:  dump_stack_lvl+0x40/0xa0
swapper/0:  print_report+0x1b8/0x71c
swapper/0:  kasan_report+0xec/0x14c
swapper/0:  __do_kernel_fault+0x60/0x29c
swapper/0:  do_bad_area+0x30/0xdc
swapper/0:  do_tag_check_fault+0x20/0x34
swapper/0:  do_mem_abort+0x58/0x104
swapper/0:  el1_abort+0x3c/0x5c
swapper/0:  el1h_64_sync_handler+0x80/0xcc
swapper/0:  el1h_64_sync+0x68/0x6c
swapper/0:  __memset+0x84/0x188
swapper/0:  btf_populate_kfunc_set+0x280/0x3d8
swapper/0:  __register_btf_kfunc_id_set+0x43c/0x468
swapper/0:  register_btf_kfunc_id_set+0x48/0x60
swapper/0:  register_nf_nat_bpf+0x1c/0x40
swapper/0:  nf_nat_init+0xc0/0x128
swapper/0:  do_one_initcall+0x184/0x464
swapper/0:  do_initcall_level+0xdc/0x1b0
swapper/0:  do_initcalls+0x70/0xc0
swapper/0:  do_basic_setup+0x1c/0x28
swapper/0:  kernel_init_freeable+0x144/0x1b8
swapper/0:  kernel_init+0x20/0x1a8
swapper/0:  ret_from_fork+0x10/0x20
==================================================================

Fixes: 1a83a716ec233 ("mm: krealloc: consider spare memory for
__GFP_ZERO")
Signed-off-by: Qun-Wei Lin <qun-wei.lin@mediatek.com>
---
 mm/slab_common.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/mm/slab_common.c b/mm/slab_common.c
index 3d26c257ed8b..3445f4500b54 100644
--- a/mm/slab_common.c
+++ b/mm/slab_common.c
@@ -1209,7 +1209,7 @@ __do_krealloc(const void *p, size_t new_size, gfp_t flags)
 		/* Zero out spare memory. */
 		if (want_init_on_alloc(flags)) {
 			kasan_disable_current();
-			memset((void *)p + new_size, 0, ks - new_size);
+			memset(kasan_reset_tag((void *)p + new_size), 0, ks - new_size);
 			kasan_enable_current();
 		}
 
-- 
2.45.2

Re: [PATCH] mm: krealloc: Fix MTE false alarm in __do_krealloc

Posted by David Rientjes 1 month ago

On Fri, 25 Oct 2024, Qun-Wei Lin wrote:

> This patch addresses an issue introduced by commit 1a83a716ec233 ("mm:
> krealloc: consider spare memory for __GFP_ZERO") which causes MTE
> (Memory Tagging Extension) to falsely report a slab-out-of-bounds error.
> 
> The problem occurs when zeroing out spare memory in __do_krealloc. The
> original code only considered software-based KASAN and did not account
> for MTE. It does not reset the KASAN tag before calling memset, leading
> to a mismatch between the pointer tag and the memory tag, resulting
> in a false positive.
> 
> Example of the error:
> ==================================================================
> swapper/0: BUG: KASAN: slab-out-of-bounds in __memset+0x84/0x188
> swapper/0: Write at addr f4ffff8005f0fdf0 by task swapper/0/1
> swapper/0: Pointer tag: [f4], memory tag: [fe]
> swapper/0:
> swapper/0: CPU: 4 UID: 0 PID: 1 Comm: swapper/0 Not tainted 6.12.
> swapper/0: Hardware name: MT6991(ENG) (DT)
> swapper/0: Call trace:
> swapper/0:  dump_backtrace+0xfc/0x17c
> swapper/0:  show_stack+0x18/0x28
> swapper/0:  dump_stack_lvl+0x40/0xa0
> swapper/0:  print_report+0x1b8/0x71c
> swapper/0:  kasan_report+0xec/0x14c
> swapper/0:  __do_kernel_fault+0x60/0x29c
> swapper/0:  do_bad_area+0x30/0xdc
> swapper/0:  do_tag_check_fault+0x20/0x34
> swapper/0:  do_mem_abort+0x58/0x104
> swapper/0:  el1_abort+0x3c/0x5c
> swapper/0:  el1h_64_sync_handler+0x80/0xcc
> swapper/0:  el1h_64_sync+0x68/0x6c
> swapper/0:  __memset+0x84/0x188
> swapper/0:  btf_populate_kfunc_set+0x280/0x3d8
> swapper/0:  __register_btf_kfunc_id_set+0x43c/0x468
> swapper/0:  register_btf_kfunc_id_set+0x48/0x60
> swapper/0:  register_nf_nat_bpf+0x1c/0x40
> swapper/0:  nf_nat_init+0xc0/0x128
> swapper/0:  do_one_initcall+0x184/0x464
> swapper/0:  do_initcall_level+0xdc/0x1b0
> swapper/0:  do_initcalls+0x70/0xc0
> swapper/0:  do_basic_setup+0x1c/0x28
> swapper/0:  kernel_init_freeable+0x144/0x1b8
> swapper/0:  kernel_init+0x20/0x1a8
> swapper/0:  ret_from_fork+0x10/0x20
> ==================================================================
> 
> Fixes: 1a83a716ec233 ("mm: krealloc: consider spare memory for
> __GFP_ZERO")
> Signed-off-by: Qun-Wei Lin <qun-wei.lin@mediatek.com>

Acked-by: David Rientjes <rientjes@google.com>

Re: [PATCH] mm: krealloc: Fix MTE false alarm in __do_krealloc

Posted by Vlastimil Babka 1 month ago

On 10/25/24 10:58, Qun-Wei Lin wrote:
> This patch addresses an issue introduced by commit 1a83a716ec233 ("mm:
> krealloc: consider spare memory for __GFP_ZERO") which causes MTE
> (Memory Tagging Extension) to falsely report a slab-out-of-bounds error.
> 
> The problem occurs when zeroing out spare memory in __do_krealloc. The
> original code only considered software-based KASAN and did not account
> for MTE. It does not reset the KASAN tag before calling memset, leading
> to a mismatch between the pointer tag and the memory tag, resulting
> in a false positive.
> 
> Example of the error:
> ==================================================================
> swapper/0: BUG: KASAN: slab-out-of-bounds in __memset+0x84/0x188
> swapper/0: Write at addr f4ffff8005f0fdf0 by task swapper/0/1
> swapper/0: Pointer tag: [f4], memory tag: [fe]
> swapper/0:
> swapper/0: CPU: 4 UID: 0 PID: 1 Comm: swapper/0 Not tainted 6.12.
> swapper/0: Hardware name: MT6991(ENG) (DT)
> swapper/0: Call trace:
> swapper/0:  dump_backtrace+0xfc/0x17c
> swapper/0:  show_stack+0x18/0x28
> swapper/0:  dump_stack_lvl+0x40/0xa0
> swapper/0:  print_report+0x1b8/0x71c
> swapper/0:  kasan_report+0xec/0x14c
> swapper/0:  __do_kernel_fault+0x60/0x29c
> swapper/0:  do_bad_area+0x30/0xdc
> swapper/0:  do_tag_check_fault+0x20/0x34
> swapper/0:  do_mem_abort+0x58/0x104
> swapper/0:  el1_abort+0x3c/0x5c
> swapper/0:  el1h_64_sync_handler+0x80/0xcc
> swapper/0:  el1h_64_sync+0x68/0x6c
> swapper/0:  __memset+0x84/0x188
> swapper/0:  btf_populate_kfunc_set+0x280/0x3d8
> swapper/0:  __register_btf_kfunc_id_set+0x43c/0x468
> swapper/0:  register_btf_kfunc_id_set+0x48/0x60
> swapper/0:  register_nf_nat_bpf+0x1c/0x40
> swapper/0:  nf_nat_init+0xc0/0x128
> swapper/0:  do_one_initcall+0x184/0x464
> swapper/0:  do_initcall_level+0xdc/0x1b0
> swapper/0:  do_initcalls+0x70/0xc0
> swapper/0:  do_basic_setup+0x1c/0x28
> swapper/0:  kernel_init_freeable+0x144/0x1b8
> swapper/0:  kernel_init+0x20/0x1a8
> swapper/0:  ret_from_fork+0x10/0x20
> ==================================================================
> 
> Fixes: 1a83a716ec233 ("mm: krealloc: consider spare memory for
> __GFP_ZERO")
> Signed-off-by: Qun-Wei Lin <qun-wei.lin@mediatek.com>

Thanks, added to slab hotfixes for 6.12

> ---
>  mm/slab_common.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/mm/slab_common.c b/mm/slab_common.c
> index 3d26c257ed8b..3445f4500b54 100644
> --- a/mm/slab_common.c
> +++ b/mm/slab_common.c
> @@ -1209,7 +1209,7 @@ __do_krealloc(const void *p, size_t new_size, gfp_t flags)
>  		/* Zero out spare memory. */
>  		if (want_init_on_alloc(flags)) {
>  			kasan_disable_current();
> -			memset((void *)p + new_size, 0, ks - new_size);
> +			memset(kasan_reset_tag((void *)p + new_size), 0, ks - new_size);
>  			kasan_enable_current();
>  		}
>