Re: [RFC PATCH] mm/vmalloc: fix vmalloc which may return null if called with __GFP_NOFAIL

[Hailong Liu <hailong.liu@xxxxxxxx>]

On Thu, 09. May 20:57, Barry Song wrote:
>
> Upon further examination, it's not a bug, but we can still utilize 'nofail'.
> The current code is very hard to read about gfp and "nofail" :-)
>
> maybe:
>
> diff --git a/mm/vmalloc.c b/mm/vmalloc.c
> index 6641be0ca80b..7c66fe16c2ad 100644
> --- a/mm/vmalloc.c
> +++ b/mm/vmalloc.c
> @@ -3498,7 +3498,7 @@ vm_area_alloc_pages(gfp_t gfp, int nid,
>  {
>         unsigned int nr_allocated = 0;
>         gfp_t alloc_gfp = gfp;
> -       bool nofail = false;
> +       bool nofail = !!(gfp & __GFP_NOFAIL);
>         struct page *page;
>         int i;
>
> @@ -3555,7 +3555,6 @@ vm_area_alloc_pages(gfp_t gfp, int nid,
>                  * and compaction etc.
>                  */
>                 alloc_gfp &= ~__GFP_NOFAIL;
> -               nofail = true;
>         }

Thanks for suggestion. I think that makes more clearly. Will
try it in next version.

--

Best Regards,
Hailong.