Re: [PATCH v4 net-next 09/12] testing: net-drv: add basic shaper test

[Simon Horman <horms@kernel.org>]

Cc: Brian Cain, linux-hexagon

On Thu, Aug 22, 2024 at 09:53:22AM +0200, Paolo Abeni wrote:
> 
> 
> On 8/21/24 18:52, kernel test robot wrote:
> > Hi Paolo,
> > 
> > kernel test robot noticed the following build warnings:
> > 
> > [auto build test WARNING on net-next/main]
> > 
> > url:    https://github.com/intel-lab-lkp/linux/commits/Paolo-Abeni/tools-ynl-lift-an-assumption-about-spec-file-name/20240820-231626
> > base:   net-next/main
> > patch link:    https://lore.kernel.org/r/4cf74f285fa5f07be546cb83ef96775f86aa0dbf.1724165948.git.pabeni%40redhat.com
> > patch subject: [PATCH v4 net-next 09/12] testing: net-drv: add basic shaper test
> > config: hexagon-randconfig-r112-20240821 (https://download.01.org/0day-ci/archive/20240822/202408220027.kA3pRF6J-lkp@intel.com/config)
> > compiler: clang version 16.0.6 (https://github.com/llvm/llvm-project 7cbf1a2591520c2491aa35339f227775f4d3adf6)
> > reproduce: (https://download.01.org/0day-ci/archive/20240822/202408220027.kA3pRF6J-lkp@intel.com/reproduce)
> > 
> > If you fix the issue in a separate patch/commit (i.e. not just a new version of
> > the same patch/commit), kindly add following tags
> > | Reported-by: kernel test robot <lkp@intel.com>
> > | Closes: https://lore.kernel.org/oe-kbuild-all/202408220027.kA3pRF6J-lkp@intel.com/
> > 
> > sparse warnings: (new ones prefixed by >>)
> > > > net/shaper/shaper.c:227:24: sparse: sparse: Using plain integer as NULL pointer
> 
> AFAICS this warning comes directly from/is due to the hexgon cmpxchg
> implementation:
> 
> #define arch_cmpxchg(ptr, old, new)                             \
> ({                                                              \
>         __typeof__(ptr) __ptr = (ptr);                          \
>         __typeof__(*(ptr)) __old = (old);                       \
>         __typeof__(*(ptr)) __new = (new);                       \
>         __typeof__(*(ptr)) __oldval = 0;                        \
> 				^^^^^^^ here.

FWIIW, I agree.

It seems that arch_cmpxchg, as implemented above, expects ptr to
be an integer. And indeed it is used in that way from
include/linux/atomic/atomic-arch-fallback.h:raw_atomic_cmpxchg_acquire().

https://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next.git/tree/include/linux/atomic/atomic-arch-fallback.h?id=f8fdda9e4f988c210b1e4519a28ddbf7d29b0038#n2055

As a hack, I allowed the function to handle either int or any type of
pointer. With this in place Sparse no longer flags the problem described
above in shaper.c.

Perhaps someone has a suggestion of how to fix this properly.

diff --git a/arch/hexagon/include/asm/cmpxchg.h b/arch/hexagon/include/asm/cmpxchg.h
index bf6cf5579cf4..d8decb8fb456 100644
--- a/arch/hexagon/include/asm/cmpxchg.h
+++ b/arch/hexagon/include/asm/cmpxchg.h
@@ -51,12 +51,18 @@ __arch_xchg(unsigned long x, volatile void *ptr, int size)
  *  variable casting.
  */
 
+#define arch_cmpxchg_zero(ptr)					\
+	(__typeof__(*(ptr)))					\
+		_Generic(*(ptr),				\
+			 int:		0,			\
+			 default:	NULL)
+
 #define arch_cmpxchg(ptr, old, new)				\
 ({								\
 	__typeof__(ptr) __ptr = (ptr);				\
 	__typeof__(*(ptr)) __old = (old);			\
 	__typeof__(*(ptr)) __new = (new);			\
-	__typeof__(*(ptr)) __oldval = 0;			\
+	__typeof__(*(ptr)) __oldval = arch_cmpxchg_zero(ptr);	\
 								\
 	asm volatile(						\
 		"1:	%0 = memw_locked(%1);\n"		\