Re: [PATCH] mm: gup: fix infinite loop within __get_longterm_locked

[John Hubbard <jhubbard@nvidia.com>]

On 1/20/25 1:26 AM, zhaoyang.huang wrote:
> From: Zhaoyang Huang <zhaoyang.huang@unisoc.com>
> 
> Infinite loop within __get_longterm_locked detected in an unique usage
> of pin_user_pages where the VA's pages are all unpinnable(vm_ops->fault
> function allocate pages via cma_alloc for hardware purpose and leave them
> out of LRU). Fixing this by have 'collected' reflect the actual number
> of pages in movable_folio_list.

The above is rather terse, although perhaps by kernel standards it's OK.

Isn't this missing a Fixes tag?

Fixes: 67e139b02d994 ("mm/gup.c: refactor 
check_and_migrate_movable_pages()")

> 
> Signed-off-by: Zhaoyang Huang <zhaoyang.huang@unisoc.com>
> ---
>   mm/gup.c | 4 ++--
>   1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/mm/gup.c b/mm/gup.c
> index 3b75e631f369..2231ce7221f9 100644
> --- a/mm/gup.c
> +++ b/mm/gup.c
> @@ -2341,8 +2341,6 @@ static unsigned long collect_longterm_unpinnable_folios(
>   		if (folio_is_longterm_pinnable(folio))
>   			continue;
>   
> -		collected++;
> -
>   		if (folio_is_device_coherent(folio))
>   			continue;
>   
> @@ -2359,6 +2357,8 @@ static unsigned long collect_longterm_unpinnable_folios(
>   		if (!folio_isolate_lru(folio))
>   			continue;
>   
> +		collected++;
> +

Well, this seems correct to me. Somehow I talked myself into believing
that it was OK to do collected++ early, even though later on we skip
actually collecting the folio, thus miscounting things.

But now I believe it was just incorrect all along.


Reviewed-by: John Hubbard <jhubbard@nvidia.com>

thanks,
-- 
John Hubbard