diff mbox series

[1/1] arm64: hibernate: add __force attribute to gfp_t casting

Message ID 20210201150306.54099-2-pasha.tatashin@soleen.com (mailing list archive)
State New, archived
Headers show
Series arm64/hibernate: sparse warnings fix | expand

Commit Message

Pasha Tatashin Feb. 1, 2021, 3:03 p.m. UTC 
Two new warnings are reported by sparse:

"sparse warnings: (new ones prefixed by >>)"
>> arch/arm64/kernel/hibernate.c:181:39: sparse: sparse: cast to
   restricted gfp_t
>> arch/arm64/kernel/hibernate.c:202:44: sparse: sparse: cast from
   restricted gfp_t

gfp_t has __bitwise type attribute and requires __force added to casting
in order to avoid these warnings.

Fixes: 50f53fb72181 ("arm64: trans_pgd: make trans_pgd_map_page generic")
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Pavel Tatashin <pasha.tatashin@soleen.com>
---
 arch/arm64/kernel/hibernate.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

Comments

Christoph Hellwig Feb. 2, 2021, 8:42 a.m. UTC | #1 
On Mon, Feb 01, 2021 at 10:03:06AM -0500, Pavel Tatashin wrote:
> Two new warnings are reported by sparse:
> 
> "sparse warnings: (new ones prefixed by >>)"
> >> arch/arm64/kernel/hibernate.c:181:39: sparse: sparse: cast to
>    restricted gfp_t
> >> arch/arm64/kernel/hibernate.c:202:44: sparse: sparse: cast from
>    restricted gfp_t
> 
> gfp_t has __bitwise type attribute and requires __force added to casting
> in order to avoid these warnings.
> 
> Fixes: 50f53fb72181 ("arm64: trans_pgd: make trans_pgd_map_page generic")
> Reported-by: kernel test robot <lkp@intel.com>
> Signed-off-by: Pavel Tatashin <pasha.tatashin@soleen.com>

What about just passing the gfp_t value by reference which would be much
cleaner?
Pasha Tatashin Feb. 2, 2021, 3:05 p.m. UTC | #2 
On Tue, Feb 2, 2021 at 3:42 AM Christoph Hellwig <hch@infradead.org> wrote:
>
> On Mon, Feb 01, 2021 at 10:03:06AM -0500, Pavel Tatashin wrote:
> > Two new warnings are reported by sparse:
> >
> > "sparse warnings: (new ones prefixed by >>)"
> > >> arch/arm64/kernel/hibernate.c:181:39: sparse: sparse: cast to
> >    restricted gfp_t
> > >> arch/arm64/kernel/hibernate.c:202:44: sparse: sparse: cast from
> >    restricted gfp_t
> >
> > gfp_t has __bitwise type attribute and requires __force added to casting
> > in order to avoid these warnings.
> >
> > Fixes: 50f53fb72181 ("arm64: trans_pgd: make trans_pgd_map_page generic")
> > Reported-by: kernel test robot <lkp@intel.com>
> > Signed-off-by: Pavel Tatashin <pasha.tatashin@soleen.com>
>
> What about just passing the gfp_t value by reference which would be much
> cleaner?


Hi Christoph,

Thank you for your suggestions. Passing by reference is OK. I am not
sure it would be much cleaner because for GFP_ATOMIC we would need to
declare another variable on stack before trans_info.

Pasha
diff mbox series

Patch

diff --git a/arch/arm64/kernel/hibernate.c b/arch/arm64/kernel/hibernate.c
index 9df32ba0d574..b1cef371df2b 100644
--- a/arch/arm64/kernel/hibernate.c
+++ b/arch/arm64/kernel/hibernate.c
@@ -178,7 +178,7 @@  EXPORT_SYMBOL(arch_hibernation_header_restore);
 
 static void *hibernate_page_alloc(void *arg)
 {
-	return (void *)get_safe_page((gfp_t)(unsigned long)arg);
+	return (void *)get_safe_page((__force gfp_t)(unsigned long)arg);
 }
 
 /*
@@ -198,7 +198,7 @@  static int create_safe_exec_page(void *src_start, size_t length,
 {
 	struct trans_pgd_info trans_info = {
 		.trans_alloc_page	= hibernate_page_alloc,
-		.trans_alloc_arg	= (void *)GFP_ATOMIC,
+		.trans_alloc_arg	= (__force void *)GFP_ATOMIC,
 	};
 
 	void *page = (void *)get_safe_page(GFP_ATOMIC);