Re: [PATCH] mm: ksize() should silently accept a NULL pointer

[Matthew Wilcox]

On Wed, Jun 17, 2020 at 04:23:04PM +, Christopher Lameter wrote:
> On Tue, 16 Jun 2020, William Kucharski wrote:
> 
> > Other mm routines such as kfree() and kzfree() silently do the right
> > thing if passed a NULL pointer, so ksize() should do the same.
> 
> Ok so the size of an no object pointer is zero? Ignoring the freeing
> of a nonexisting object makes sense. But determining it size?

ksize() is misnamed.  It's not the size of the object, it's the number
of bytes allocated for that object.

A NULL pointer represents a freed object, or one that was never
allocated in the first place.  Clearly that's 0 bytes.  What other
answer would make sense?

Re: [PATCH] mm: ksize() should silently accept a NULL pointer

[Christopher Lameter]

On Tue, 16 Jun 2020, William Kucharski wrote:

> Other mm routines such as kfree() and kzfree() silently do the right
> thing if passed a NULL pointer, so ksize() should do the same.

Ok so the size of an no object pointer is zero? Ignoring the freeing
of a nonexisting object makes sense. But determining it size?

Re: [PATCH] mm: ksize() should silently accept a NULL pointer

[David Hildenbrand]

On 17.06.20 00:54, William Kucharski wrote:
> Other mm routines such as kfree() and kzfree() silently do the right
> thing if passed a NULL pointer, so ksize() should do the same.
> 
> Signed-off-by: William Kucharski 
> ---
>  mm/slab_common.c | 14 +-
>  1 file changed, 5 insertions(+), 9 deletions(-)
> 
> diff --git a/mm/slab_common.c b/mm/slab_common.c
> index 9e72ba224175..2bff01ad94d8 100644
> --- a/mm/slab_common.c
> +++ b/mm/slab_common.c
> @@ -1660,10 +1660,9 @@ static __always_inline void *__do_krealloc(const void 
> *p, size_t new_size,
>  gfp_t flags)
>  {
>   void *ret;
> - size_t ks = 0;
> + size_t ks;
>  
> - if (p)
> - ks = ksize(p);
> + ks = ksize(p);
>  
>   if (ks >= new_size) {
>   p = kasan_krealloc((void *)p, new_size, flags);
> @@ -1723,10 +1722,9 @@ void kzfree(const void *p)
>   size_t ks;
>   void *mem = (void *)p;
>  
> - if (unlikely(ZERO_OR_NULL_PTR(mem)))
> - return;
>   ks = ksize(mem);
> - memset(mem, 0, ks);
> + if (ks)
> + memset(mem, 0, ks);
>   kfree(mem);
>  }
>  EXPORT_SYMBOL(kzfree);
> @@ -1749,8 +1747,6 @@ size_t ksize(const void *objp)
>  {
>   size_t size;
>  
> - if (WARN_ON_ONCE(!objp))
> - return 0;
>   /*
>* We need to check that the pointed to object is valid, and only then
>* unpoison the shadow memory below. We use __kasan_check_read(), to
> @@ -1764,7 +1760,7 @@ size_t ksize(const void *objp)
>* We want to perform the check before __ksize(), to avoid potentially
>* crashing in __ksize() due to accessing invalid metadata.
>*/
> - if (unlikely(objp == ZERO_SIZE_PTR) || !__kasan_check_read(objp, 1))
> + if (unlikely(ZERO_OR_NULL_PTR(objp)) || !__kasan_check_read(objp, 1))
>   return 0;
>  
>   size = __ksize(objp);
> 

Reviewed-by: David Hildenbrand 

-- 
Thanks,

David / dhildenb

Re: [PATCH] mm: ksize() should silently accept a NULL pointer

[Matthew Wilcox]

On Tue, Jun 16, 2020 at 04:54:09PM -0600, William Kucharski wrote:
> Other mm routines such as kfree() and kzfree() silently do the right
> thing if passed a NULL pointer, so ksize() should do the same.
> 
> Signed-off-by: William Kucharski 

Nice simplification.

Reviewed-by: Matthew Wilcox (Oracle) 

[PATCH] mm: ksize() should silently accept a NULL pointer

[William Kucharski]

Other mm routines such as kfree() and kzfree() silently do the right
thing if passed a NULL pointer, so ksize() should do the same.

Signed-off-by: William Kucharski 
---
 mm/slab_common.c | 14 +-
 1 file changed, 5 insertions(+), 9 deletions(-)

diff --git a/mm/slab_common.c b/mm/slab_common.c
index 9e72ba224175..2bff01ad94d8 100644
--- a/mm/slab_common.c
+++ b/mm/slab_common.c
@@ -1660,10 +1660,9 @@ static __always_inline void *__do_krealloc(const void 
*p, size_t new_size,
   gfp_t flags)
 {
void *ret;
-   size_t ks = 0;
+   size_t ks;
 
-   if (p)
-   ks = ksize(p);
+   ks = ksize(p);
 
if (ks >= new_size) {
p = kasan_krealloc((void *)p, new_size, flags);
@@ -1723,10 +1722,9 @@ void kzfree(const void *p)
size_t ks;
void *mem = (void *)p;
 
-   if (unlikely(ZERO_OR_NULL_PTR(mem)))
-   return;
ks = ksize(mem);
-   memset(mem, 0, ks);
+   if (ks)
+   memset(mem, 0, ks);
kfree(mem);
 }
 EXPORT_SYMBOL(kzfree);
@@ -1749,8 +1747,6 @@ size_t ksize(const void *objp)
 {
size_t size;
 
-   if (WARN_ON_ONCE(!objp))
-   return 0;
/*
 * We need to check that the pointed to object is valid, and only then
 * unpoison the shadow memory below. We use __kasan_check_read(), to
@@ -1764,7 +1760,7 @@ size_t ksize(const void *objp)
 * We want to perform the check before __ksize(), to avoid potentially
 * crashing in __ksize() due to accessing invalid metadata.
 */
-   if (unlikely(objp == ZERO_SIZE_PTR) || !__kasan_check_read(objp, 1))
+   if (unlikely(ZERO_OR_NULL_PTR(objp)) || !__kasan_check_read(objp, 1))
return 0;
 
size = __ksize(objp);
-- 
2.26.2