On Sun 16-01-22 13:49:22, Christophe JAILLET wrote:
> kstrdup_quotable_cmdline() is given a gfp flag that is passed and used for
> memory allocation in kstrdup_quotable() just a few lines below.
>
> It looks reasonable to use this gfp value for the buffer allocated and
> freed in kstrdup_quotable_cmdline() as well.
>
> Fixes: 0ee931c4e31a ("mm: treewide: remove GFP_TEMPORARY allocation flag")
I do not think this commit is changing much here. It just replaces
GFP_TEMPORARY with GFP_KERNEL so the code has ignored the gfp mask even
before that change.
All existing callers of kstrdup_quotable_cmdline use GFP_KERNEL so would
it make more sense to simply drop the gfp argument altogether and use
GFP_KERNEL internally?
Normally it is better to have a full control of the allocation mask but
if we have any non-GFP_KERNEL caller then I would rather have the
argument added and the function checked whether all internal paths are
gfp mask aware.
> Signed-off-by: Christophe JAILLET <christophe.jaillet@xxxxxxxxxx>
> ---
> According to what I've found in 5.16, all callers use GFP_KERNEL, so this
> patch should be a no-op.
> But who knows how it will be used in the future. Better safe than sorry.
> ---
> lib/string_helpers.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/lib/string_helpers.c b/lib/string_helpers.c
> index 90f9f1b7afec..7aceeb40dfd7 100644
> --- a/lib/string_helpers.c
> +++ b/lib/string_helpers.c
> @@ -624,7 +624,7 @@ char *kstrdup_quotable_cmdline(struct task_struct *task, gfp_t gfp)
> char *buffer, *quoted;
> int i, res;
>
> - buffer = kmalloc(PAGE_SIZE, GFP_KERNEL);
> + buffer = kmalloc(PAGE_SIZE, gfp);
> if (!buffer)
> return NULL;
>
> --
> 2.32.0
--
Michal Hocko
SUSE Labs
