Re: [PATCH v2] powerpc/smp: Add check for kcalloc() failure in parse_thread_groups()

[Christophe Leroy <christophe.leroy@csgroup.eu>]

Le 22/09/2025 à 17:38, Christophe JAILLET a écrit :
> Le 22/09/2025 à 17:10, Guangshuo Li a écrit :
>> As kcalloc() may fail, check its return value to avoid a NULL pointer
>> dereference when passing it to of_property_read_u32_array().
>>
>> Fixes: 790a1662d3a26 ("powerpc/smp: Parse ibm,thread-groups with 
>> multiple properties")
>> Cc: stable@vger.kernel.org
> 
> Signed-off-by that was part of v1, is missing in v2.

I see it below the ---

> 
>> ---
>> changelog:
>> v2:
>> - Return -ENOMEM directly on allocation failure.
> 
> Except for a newline that is removed, v2 is the same as v1, or I miss 
> something?

v1 was:

+       if (!thread_group_array) {
+               ret = -ENOMEM;
+               goto out_free;
+       }

Which was wrong.

Well maybe there was several v1, I'm talking about 
https://lore.kernel.org/all/20250918131513.3557422-1-lgs201920130244@gmail.com/

> 
> CJ
> 
>>
>> Signed-off-by: Guangshuo Li <lgs201920130244@gmail.com>
>> ---
>>   arch/powerpc/kernel/smp.c | 2 ++
>>   1 file changed, 2 insertions(+)
>>
>> diff --git a/arch/powerpc/kernel/smp.c b/arch/powerpc/kernel/smp.c
>> index 5ac7084eebc0..cfccb9389760 100644
>> --- a/arch/powerpc/kernel/smp.c
>> +++ b/arch/powerpc/kernel/smp.c
>> @@ -822,6 +822,8 @@ static int parse_thread_groups(struct device_node 
>> *dn,
>>       count = of_property_count_u32_elems(dn, "ibm,thread-groups");
>>       thread_group_array = kcalloc(count, sizeof(u32), GFP_KERNEL);
>> +    if (!thread_group_array)
>> +        return -ENOMEM;
>>       ret = of_property_read_u32_array(dn, "ibm,thread-groups",
>>                        thread_group_array, count);
>>       if (ret)
>