Re: [PATCH] fork: zero vmap stack using clear_pages() instead of memset()

["Christophe Leroy (CS GROUP)" <chleroy@kernel.org>]

Le 06/03/2026 à 14:16, Christophe Leroy (CS GROUP) a écrit :
> 
> 
> Le 24/02/2026 à 11:26, Linus Walleij a écrit :
>> After the introduction of clear_pages() we exploit the
>> fact that the process vm_area is allocated in contiguous
>> pages to just clear them all in one swift operation.
>>
>> Suggested-by: Mateusz Guzik <mjguzik@gmail.com>
>> Link: https://eur01.safelinks.protection.outlook.com/? 
>> url=https%3A%2F%2Flore.kernel.org%2Flinux- 
>> mm%2Fdpnwsp7dl4535rd7qmszanw6u5an2p74uxfex4dh53frpb7pu3%402bnjjavjrepe%2F&data=05%7C02%7Cchristophe.leroy%40csgroup.eu%7C5eca287345d5408b59ce08de738f41a2%7C8b87af7d86474dc78df45f69a2011bb5%7C0%7C0%7C639075256284265137%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=PrUINapvx0%2BJH9Dl9%2F2bRfUJmrvPCxwk%2FxGll6rgle8%3D&reserved=0
>> Suggested-by: Pasha Tatashin <pasha.tatashin@soleen.com>
>> Link: https://eur01.safelinks.protection.outlook.com/? 
>> url=https%3A%2F%2Flore.kernel.org%2F20240311164638.2015063-7- 
>> pasha.tatashin%40soleen.com&data=05%7C02%7Cchristophe.leroy%40csgroup.eu%7C5eca287345d5408b59ce08de738f41a2%7C8b87af7d86474dc78df45f69a2011bb5%7C0%7C0%7C639075256284308057%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=J2HT3AP4aZTX0Va0XE0taOtnA9GsETKFY1mWhrHBQbg%3D&reserved=0
>> Signed-off-by: Linus Walleij <linusw@kernel.org>
>> ---
>>   kernel/fork.c | 2 +-
>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/kernel/fork.c b/kernel/fork.c
>> index e832da9d15a4..88d78ccef245 100644
>> --- a/kernel/fork.c
>> +++ b/kernel/fork.c
>> @@ -345,7 +345,7 @@ static int alloc_thread_stack_node(struct 
>> task_struct *tsk, int node)
>>           stack = kasan_reset_tag(vm_area->addr);
>>           /* Clear stale pointers from reused stack. */
>> -        memset(stack, 0, THREAD_SIZE);
>> +        clear_pages(vm_area->addr, vm_area->nr_pages);
> 
> THREAD_SIZE can be smaller that PAGE_SIZE, how do you deal with that ?
> 
> For instance on powerpc/64 when PAGE_SIZE is 64k, while the default 
> THREAD_SHIFT is 14, leading to THREAD_SIZE being 16k
> 
> Or on powerpc/32 when PAGE_SIZE is 16k, while the default THREAD_SHIFT 
> is 13, leading to THREAD_SIZE being 8k.
> 

Oops sorry, I missed it was only for VMAP_STACK, in that case stack size 
is at least one page.

Christophe

> Christophe
> 
> 
> 
>>           tsk->stack_vm_area = vm_area;
>>           tsk->stack = stack;
>>
>> ---
>> base-commit: 6de23f81a5e08be8fbf5e8d7e9febc72a5b5f27f
>> change-id: 20260224-mm-fork-clear-pages-ae7c52d12a93
>>
>> Best regards,
>