Re: [PATCH v6 net-next 4/9] vhost-net: allow configuring extended features

["Michael S. Tsirkin" <mst@redhat.com>]

On Fri, Jun 27, 2025 at 04:36:14PM +0200, Paolo Abeni wrote:
> On 6/27/25 3:11 PM, Paolo Abeni wrote:
> > +csky maintainer
> > On 6/27/25 8:41 AM, 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/scripts-kernel_doc-py-properly-handle-VIRTIO_DECLARE_FEATURES/20250624-221751
> >> base:   net-next/main
> >> patch link:    https://lore.kernel.org/r/23e46bff5333015d92bf0876033750d9fbf555a0.1750753211.git.pabeni%40redhat.com
> >> patch subject: [PATCH v6 net-next 4/9] vhost-net: allow configuring extended features
> >> config: csky-randconfig-001-20250627 (https://download.01.org/0day-ci/archive/20250627/202506271443.G9cAx8PS-lkp@intel.com/config)
> >> compiler: csky-linux-gcc (GCC) 15.1.0
> >> reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20250627/202506271443.G9cAx8PS-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/202506271443.G9cAx8PS-lkp@intel.com/
> >>
> >> All warnings (new ones prefixed by >>):
> >>
> >>    In file included from include/linux/uaccess.h:12,
> >>                     from include/linux/sched/task.h:13,
> >>                     from include/linux/sched/signal.h:9,
> >>                     from include/linux/rcuwait.h:6,
> >>                     from include/linux/percpu-rwsem.h:7,
> >>                     from include/linux/fs.h:34,
> >>                     from include/linux/compat.h:17,
> >>                     from drivers/vhost/net.c:8:
> >>    arch/csky/include/asm/uaccess.h: In function '__get_user_fn.constprop':
> >>>> arch/csky/include/asm/uaccess.h:147:9: warning: 'retval' is used uninitialized [-Wuninitialized]
> >>      147 |         __asm__ __volatile__(                           \
> >>          |         ^~~~~~~
> >>    arch/csky/include/asm/uaccess.h:187:17: note: in expansion of macro '__get_user_asm_64'
> >>      187 |                 __get_user_asm_64(x, ptr, retval);
> >>          |                 ^~~~~~~~~~~~~~~~~
> >>    arch/csky/include/asm/uaccess.h:170:13: note: 'retval' was declared here
> >>      170 |         int retval;
> >>          |             ^~~~~~
> >>
> >>
> >> vim +/retval +147 arch/csky/include/asm/uaccess.h
> >>
> >> da551281947cb2c Guo Ren 2018-09-05  141  
> >> e58a41c2226847f Guo Ren 2021-04-21  142  #define __get_user_asm_64(x, ptr, err)			\
> >> da551281947cb2c Guo Ren 2018-09-05  143  do {							\
> >> da551281947cb2c Guo Ren 2018-09-05  144  	int tmp;					\
> >> e58a41c2226847f Guo Ren 2021-04-21  145  	int errcode;					\
> >> e58a41c2226847f Guo Ren 2021-04-21  146  							\
> >> e58a41c2226847f Guo Ren 2021-04-21 @147  	__asm__ __volatile__(				\
> >> e58a41c2226847f Guo Ren 2021-04-21  148  	"1:   ldw     %3, (%2, 0)     \n"		\
> >> da551281947cb2c Guo Ren 2018-09-05  149  	"     stw     %3, (%1, 0)     \n"		\
> >> e58a41c2226847f Guo Ren 2021-04-21  150  	"2:   ldw     %3, (%2, 4)     \n"		\
> >> e58a41c2226847f Guo Ren 2021-04-21  151  	"     stw     %3, (%1, 4)     \n"		\
> >> e58a41c2226847f Guo Ren 2021-04-21  152  	"     br      4f              \n"		\
> >> e58a41c2226847f Guo Ren 2021-04-21  153  	"3:   mov     %0, %4          \n"		\
> >> e58a41c2226847f Guo Ren 2021-04-21  154  	"     br      4f              \n"		\
> >> da551281947cb2c Guo Ren 2018-09-05  155  	".section __ex_table, \"a\"   \n"		\
> >> da551281947cb2c Guo Ren 2018-09-05  156  	".align   2                   \n"		\
> >> e58a41c2226847f Guo Ren 2021-04-21  157  	".long    1b, 3b              \n"		\
> >> e58a41c2226847f Guo Ren 2021-04-21  158  	".long    2b, 3b              \n"		\
> >> da551281947cb2c Guo Ren 2018-09-05  159  	".previous                    \n"		\
> >> e58a41c2226847f Guo Ren 2021-04-21  160  	"4:                           \n"		\
> >> e58a41c2226847f Guo Ren 2021-04-21  161  	: "=r"(err), "=r"(x), "=r"(ptr),		\
> >> e58a41c2226847f Guo Ren 2021-04-21  162  	  "=r"(tmp), "=r"(errcode)			\
> >> e58a41c2226847f Guo Ren 2021-04-21  163  	: "0"(err), "1"(x), "2"(ptr), "3"(0),		\
> >> e58a41c2226847f Guo Ren 2021-04-21  164  	  "4"(-EFAULT)					\
> >> da551281947cb2c Guo Ren 2018-09-05  165  	: "memory");					\
> >> da551281947cb2c Guo Ren 2018-09-05  166  } while (0)
> >> da551281947cb2c Guo Ren 2018-09-05  167  
> > 
> > The intel test report reported the above compile warning on this series:
> > 
> > https://lore.kernel.org/netdev/20250627084609-mutt-send-email-mst@kernel.org/T/#md788de2b3a4e9da23ac93b5f1c773a6070b5b4fb
> > 
> > specifically, in patch 4:
> > 
> > +                       if (get_user(features, featurep + 1 + i))
> > +                               return -EFAULT;
> > 
> > AFAICS such statement is legit, and the bot points to some problem in
> > the arch specific get_user() implementation. Could you please have a look?
> 
> Out of sheer ignorance on my side, I fail to see how the csky get_user()
> could work correctly without something alike the following (which indeed
> fixes the issue here).
> 
> /P
> ---
> diff --git a/arch/csky/include/asm/uaccess.h
> b/arch/csky/include/asm/uaccess.h
> index 2e927c21d8a1..ae0864ad59a3 100644
> --- a/arch/csky/include/asm/uaccess.h
> +++ b/arch/csky/include/asm/uaccess.h
> @@ -167,7 +167,7 @@ do {							\
> 
>  static inline int __get_user_fn(size_t size, const void __user *ptr,
> void *x)
>  {
> -	int retval;
> +	int retval = 0;
>  	u32 tmp;
> 
>  	switch (size) {

Given there's no reaction from the arch maintainers,
I see two options:

- go back to copy_from_user - nothing wrong with it, and the code
in question is off data path.

- go ahead and include this patch in the series, even though
  I'm just as unsure it's right as you are.


Up to you really, but I think it's one of the two.



-- 
MST