diff mbox series

[1/2] VMCI: Remove handle_arr_calc_size()

Message ID adf0c48a57d911be5509688614d2e00694ff57eb.1702068153.git.christophe.jaillet@wanadoo.fr
State New
Headers
Series [1/2] VMCI: Remove handle_arr_calc_size() |

Commit Message

Christophe JAILLET Dec. 8, 2023, 8:46 p.m. UTC 
  Use struct_size() instead of handle_arr_calc_size().
This is much more conventionnal.

Suggested-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
---
 drivers/misc/vmw_vmci/vmci_handle_array.c | 12 +++---------
 1 file changed, 3 insertions(+), 9 deletions(-)

Comments

Kees Cook Dec. 8, 2023, 8:59 p.m. UTC | #1 
On Fri, Dec 08, 2023 at 09:46:09PM +0100, Christophe JAILLET wrote:
> Use struct_size() instead of handle_arr_calc_size().
> This is much more conventionnal.
> 
> Suggested-by: Kees Cook <keescook@chromium.org>
> Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>

Looks good. And since capacity in u32, there's no need for size_add().

Reviewed-by: Kees Cook <keescook@chromium.org>
Christophe JAILLET Dec. 8, 2023, 9:14 p.m. UTC | #2 
Le 08/12/2023 à 21:59, Kees Cook a écrit :
> On Fri, Dec 08, 2023 at 09:46:09PM +0100, Christophe JAILLET wrote:
>> Use struct_size() instead of handle_arr_calc_size().
>> This is much more conventionnal.
>>
>> Suggested-by: Kees Cook <keescook@chromium.org>
>> Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
> 
> Looks good. And since capacity in u32, there's no need for size_add().

Hmm,

isn't u32 + u32 --> u32, so can overflow?
If I understand correctly, the type promotion to size_t will occur after 
the addition.

So size_add() looks needed, or I missed something?

CJ

> 
> Reviewed-by: Kees Cook <keescook@chromium.org>
>
Kees Cook Dec. 8, 2023, 9:27 p.m. UTC | #3 
On Fri, Dec 08, 2023 at 10:14:35PM +0100, Christophe JAILLET wrote:
> Le 08/12/2023 à 21:59, Kees Cook a écrit :
> > On Fri, Dec 08, 2023 at 09:46:09PM +0100, Christophe JAILLET wrote:
> > > Use struct_size() instead of handle_arr_calc_size().
> > > This is much more conventionnal.
> > > 
> > > Suggested-by: Kees Cook <keescook@chromium.org>
> > > Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
> > 
> > Looks good. And since capacity in u32, there's no need for size_add().
> 
> Hmm,
> 
> isn't u32 + u32 --> u32, so can overflow?
> If I understand correctly, the type promotion to size_t will occur after the
> addition.

Oh lovely, I thought the promotion was first. Ugh.

> So size_add() looks needed, or I missed something?

Yeah, and I'm also to stuck in pretending 32-bit systems don't exist.
So, yes, please include size_add()...

-Kees

> 
> CJ
> 
> > 
> > Reviewed-by: Kees Cook <keescook@chromium.org>
> > 
>
diff mbox series

Patch

diff --git a/drivers/misc/vmw_vmci/vmci_handle_array.c b/drivers/misc/vmw_vmci/vmci_handle_array.c
index de7fee7ead1b..56d48d42736b 100644
--- a/drivers/misc/vmw_vmci/vmci_handle_array.c
+++ b/drivers/misc/vmw_vmci/vmci_handle_array.c
@@ -8,12 +8,6 @@ 
 #include <linux/slab.h>
 #include "vmci_handle_array.h"
 
-static size_t handle_arr_calc_size(u32 capacity)
-{
-	return VMCI_HANDLE_ARRAY_HEADER_SIZE +
-	    capacity * sizeof(struct vmci_handle);
-}
-
 struct vmci_handle_arr *vmci_handle_arr_create(u32 capacity, u32 max_capacity)
 {
 	struct vmci_handle_arr *array;
@@ -25,7 +19,7 @@  struct vmci_handle_arr *vmci_handle_arr_create(u32 capacity, u32 max_capacity)
 		capacity = min((u32)VMCI_HANDLE_ARRAY_DEFAULT_CAPACITY,
 			       max_capacity);
 
-	array = kmalloc(handle_arr_calc_size(capacity), GFP_ATOMIC);
+	array = kmalloc(struct_size(array, entries, capacity), GFP_ATOMIC);
 	if (!array)
 		return NULL;
 
@@ -51,8 +45,8 @@  int vmci_handle_arr_append_entry(struct vmci_handle_arr **array_ptr,
 		struct vmci_handle_arr *new_array;
 		u32 capacity_bump = min(array->max_capacity - array->capacity,
 					array->capacity);
-		size_t new_size = handle_arr_calc_size(array->capacity +
-						       capacity_bump);
+		size_t new_size = struct_size(array, entries,
+					      array->capacity + capacity_bump);
 
 		if (array->size >= array->max_capacity)
 			return VMCI_ERROR_NO_MEM;