2022-11-05 07:05:08

by Kees Cook

Subject: [PATCH] mm/slab_common: Restore passing "caller" for tracing

The "caller" argument was accidentally being ignored in a few places
that were recently refactored. Restore these "caller" arguments, instead
of _RET_IP_.

Cc: Hyeonggon Yoo <[email protected]>
Cc: Vlastimil Babka <[email protected]>
Cc: Christoph Lameter <[email protected]>
Cc: Pekka Enberg <[email protected]>
Cc: David Rientjes <[email protected]>
Cc: Joonsoo Kim <[email protected]>
Cc: Andrew Morton <[email protected]>
Cc: Roman Gushchin <[email protected]>
Cc: [email protected]
Fixes: 11e9734bcb6a ("mm/slab_common: unify NUMA and UMA version of tracepoints")
Cc: [email protected]
Signed-off-by: Kees Cook <[email protected]>
---
mm/slab_common.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/mm/slab_common.c b/mm/slab_common.c
index 33b1886b06eb..0e614f9e7ed7 100644
--- a/mm/slab_common.c
+++ b/mm/slab_common.c
@@ -941,7 +941,7 @@ void *__do_kmalloc_node(size_t size, gfp_t flags, int node, unsigned long caller

if (unlikely(size > KMALLOC_MAX_CACHE_SIZE)) {
ret = __kmalloc_large_node(size, flags, node);
- trace_kmalloc(_RET_IP_, ret, size,
+ trace_kmalloc(caller, ret, size,
PAGE_SIZE << get_order(size), flags, node);
return ret;
}
@@ -953,7 +953,7 @@ void *__do_kmalloc_node(size_t size, gfp_t flags, int node, unsigned long caller

ret = __kmem_cache_alloc_node(s, flags, node, size, caller);
ret = kasan_kmalloc(s, ret, size, flags);
- trace_kmalloc(_RET_IP_, ret, size, s->size, flags, node);
+ trace_kmalloc(caller, ret, size, s->size, flags, node);
return ret;
}

--
2.34.1

2022-11-06 11:15:28

by Hyeonggon Yoo

Subject: Re: [PATCH] mm/slab_common: Restore passing "caller" for tracing

On Fri, Nov 04, 2022 at 11:35:34PM -0700, Kees Cook wrote:
> The "caller" argument was accidentally being ignored in a few places
> that were recently refactored. Restore these "caller" arguments, instead
> of _RET_IP_.
>
> Cc: Hyeonggon Yoo <[email protected]>

Acked-by: Hyeonggon Yoo <[email protected]>

> Cc: Vlastimil Babka <[email protected]>
> Cc: Christoph Lameter <[email protected]>
> Cc: Pekka Enberg <[email protected]>
> Cc: David Rientjes <[email protected]>
> Cc: Joonsoo Kim <[email protected]>
> Cc: Andrew Morton <[email protected]>
> Cc: Roman Gushchin <[email protected]>
> Cc: [email protected]
> Fixes: 11e9734bcb6a ("mm/slab_common: unify NUMA and UMA version of tracepoints")
> Cc: [email protected]


BTW I think it can be just sent to next release candidate.
The referred commit was merged in this development cycle.

> Signed-off-by: Kees Cook <[email protected]>
> ---
> mm/slab_common.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/mm/slab_common.c b/mm/slab_common.c
> index 33b1886b06eb..0e614f9e7ed7 100644
> --- a/mm/slab_common.c
> +++ b/mm/slab_common.c
> @@ -941,7 +941,7 @@ void *__do_kmalloc_node(size_t size, gfp_t flags, int node, unsigned long caller
>
> if (unlikely(size > KMALLOC_MAX_CACHE_SIZE)) {
> ret = __kmalloc_large_node(size, flags, node);
> - trace_kmalloc(_RET_IP_, ret, size,
> + trace_kmalloc(caller, ret, size,
> PAGE_SIZE << get_order(size), flags, node);
> return ret;
> }
> @@ -953,7 +953,7 @@ void *__do_kmalloc_node(size_t size, gfp_t flags, int node, unsigned long caller
>
> ret = __kmem_cache_alloc_node(s, flags, node, size, caller);
> ret = kasan_kmalloc(s, ret, size, flags);
> - trace_kmalloc(_RET_IP_, ret, size, s->size, flags, node);
> + trace_kmalloc(caller, ret, size, s->size, flags, node);
> return ret;
> }
>
> --
> 2.34.1


Thanks for catching this!

--
Thanks,
Hyeonggon

2022-11-06 20:37:54

by Vlastimil Babka

Subject: Re: [PATCH] mm/slab_common: Restore passing "caller" for tracing

On 11/6/22 11:46, Hyeonggon Yoo wrote:
> On Fri, Nov 04, 2022 at 11:35:34PM -0700, Kees Cook wrote:
>> The "caller" argument was accidentally being ignored in a few places
>> that were recently refactored. Restore these "caller" arguments, instead
>> of _RET_IP_.
>>
>> Cc: Hyeonggon Yoo <[email protected]>
>
> Acked-by: Hyeonggon Yoo <[email protected]>
>
>> Cc: Vlastimil Babka <[email protected]>
>> Cc: Christoph Lameter <[email protected]>
>> Cc: Pekka Enberg <[email protected]>
>> Cc: David Rientjes <[email protected]>
>> Cc: Joonsoo Kim <[email protected]>
>> Cc: Andrew Morton <[email protected]>
>> Cc: Roman Gushchin <[email protected]>
>> Cc: [email protected]
>> Fixes: 11e9734bcb6a ("mm/slab_common: unify NUMA and UMA version of tracepoints")
>> Cc: [email protected]
>
>
> BTW I think it can be just sent to next release candidate.
> The referred commit was merged in this development cycle.

Yep, dropped the Cc stable for that reason. Pushed to
slab/for-6.1-rc4/fixes and will send a PR to Linus in few days.
Thanks!

>> Signed-off-by: Kees Cook <[email protected]>
>> ---
>> mm/slab_common.c | 4 ++--
>> 1 file changed, 2 insertions(+), 2 deletions(-)
>>
>> diff --git a/mm/slab_common.c b/mm/slab_common.c
>> index 33b1886b06eb..0e614f9e7ed7 100644
>> --- a/mm/slab_common.c
>> +++ b/mm/slab_common.c
>> @@ -941,7 +941,7 @@ void *__do_kmalloc_node(size_t size, gfp_t flags, int node, unsigned long caller
>>
>> if (unlikely(size > KMALLOC_MAX_CACHE_SIZE)) {
>> ret = __kmalloc_large_node(size, flags, node);
>> - trace_kmalloc(_RET_IP_, ret, size,
>> + trace_kmalloc(caller, ret, size,
>> PAGE_SIZE << get_order(size), flags, node);
>> return ret;
>> }
>> @@ -953,7 +953,7 @@ void *__do_kmalloc_node(size_t size, gfp_t flags, int node, unsigned long caller
>>
>> ret = __kmem_cache_alloc_node(s, flags, node, size, caller);
>> ret = kasan_kmalloc(s, ret, size, flags);
>> - trace_kmalloc(_RET_IP_, ret, size, s->size, flags, node);
>> + trace_kmalloc(caller, ret, size, s->size, flags, node);
>> return ret;
>> }
>>
>> --
>> 2.34.1
>
>
> Thanks for catching this!
>