Re: [PATCH] crypto: using size_add() for kmalloc()

[Su Hui <suhui@nfschina.com>]

On 2025/4/21 16:46, Su Hui wrote:
> On 2025/4/21 16:32, Christophe JAILLET wrote:
>> Le 21/04/2025 à 09:43, Su Hui a écrit :
>>> On 2025/4/21 15:10, Christophe JAILLET wrote:
>>>> Le 21/04/2025 à 07:51, Su Hui a écrit :
>>>>> It's safer to use size_add() to replace open-coded aithmetic in 
>>>>> allocator
>>>>> arguments, because size_add() can prevent possible overflow problem.
>>>>>
>>>>> Signed-off-by: Su Hui <suhui@nfschina.com>
>>>>> ---
>>>>>   include/crypto/aead.h     | 3 ++-
>>>>>   include/crypto/akcipher.h | 4 +++-
>>>>>   include/crypto/kpp.h      | 3 ++-
>>>>>   3 files changed, 7 insertions(+), 3 deletions(-)
>>>>>
>>>>> diff --git a/include/crypto/aead.h b/include/crypto/aead.h
>>>>> index 0e8a41638678..cf212d28fe18 100644
>>>>> --- a/include/crypto/aead.h
>>>>> +++ b/include/crypto/aead.h
>>>>> @@ -10,6 +10,7 @@
>>>>>     #include <linux/atomic.h>
>>>>>   #include <linux/container_of.h>
>>>>> +#include <linux/overflow.h>
>>>>
>>>> You could move this 1 line below, to keep alphabetical order.
>>>> And why do you say that it is redundant in your follow-up mail?
>>> Thanks for your suggestion, I didn't notice this alphabetical order 
>>> at first :( .
>>> Because I found that  <linux/crypto.h> includes <linux/slab.h>, and
>>> <linux/slab.h> includes <linux/overflow.h>, so this overflow.h is 
>>> redundant.
>>
>> It is usually considered best practice to include what is used, and 
>> not relying on indirect includes.
>>
>> Should these others includes change one day, then some apparently 
>> unrelated files will fails to built.
>>
> I already send a v2 patch, too fast for this v2 sending :(.
> v2: 
> https://lore.kernel.org/all/20250421083116.1161805-1-suhui@nfschina.com/
>
> I agreed with 'include what is used'.  So I guess v1 is enough and v2 
> maybe a wrong patchset.
> Sorry for the noise.

Oh, I forget to keep alphabetical order, so v3 is needed if there is no 
other suggestions.