Re: [BUG] allmodconfig build error in next-20240108

[Stephen Rothwell <sfr@rothwell.id.au>]

[-- Attachment #1: Type: text/plain, Size: 5080 bytes --]

Hi Lucas,

On Tue, 9 Jan 2024 10:58:40 -0600 Lucas De Marchi <lucas.demarchi@intel.com> wrote:
>
> On Mon, Jan 08, 2024 at 03:15:23PM -0800, Paul E. McKenney wrote:
> >On Tue, Jan 09, 2024 at 09:57:57AM +1100, Stephen Rothwell wrote:  
> >> Hi Paul,
> >>
> >> On Mon, 8 Jan 2024 13:33:36 -0800 "Paul E. McKenney" <paulmck@kernel.org> wrote:  
> >> >
> >> > Recent -next trees get the following build error for allmodconfig builds:
> >> >
> >> > ------------------------------------------------------------------------
> >> >
> >> > drivers/gpu/drm/xe/xe_gt_pagefault.c: In function ‘xe_guc_pagefault_handler’:
> >> > ./include/linux/fortify-string.h:57:33: error: writing 16 bytes into a region of  size 0 [-Werror=stringop-overflow=]
> >> >    57 | #define __underlying_memcpy     __builtin_memcpy
> >> >       |                                 ^
> >> > ./include/linux/fortify-string.h:644:9: note: in expansion of macro ‘__underlying_memcpy’
> >> >   644 |         __underlying_##op(p, q, __fortify_size); \
> >> >       |         ^~~~~~~~~~~~~
> >> > ./include/linux/fortify-string.h:689:26: note: in expansion of macro ‘__fortify_memcpy_chk’
> >> >   689 | #define memcpy(p, q, s)  __fortify_memcpy_chk(p, q, s, \
> >> >       |                          ^~~~~~~~~~~~~~~~~~~~
> >> > drivers/gpu/drm/xe/xe_gt_pagefault.c:340:17: note: in expansion of macro ‘memcpy’
> >> >   340 |                 memcpy(pf_queue->data + pf_queue->tail, msg, len * sizeof(u32));
> >> >       |                 ^~~~~~
> >> > In file included from drivers/gpu/drm/xe/xe_device_types.h:17,
> >> >                  from drivers/gpu/drm/xe/xe_vm_types.h:16,
> >> >                  from drivers/gpu/drm/xe/xe_bo.h:13,
> >> >                  from drivers/gpu/drm/xe/xe_gt_pagefault.c:16:
> >> > drivers/gpu/drm/xe/xe_gt_types.h:102:25: note: at offset [1144, 265324] into destination object ‘tile’ of size 8
> >> >   102 |         struct xe_tile *tile;
> >> >       |  
> >>
> >> Which architecture?  What compiler and version?  Anything special in your build
> >> setup?  I do x86_64 allmodconfig builds all day with gcc v13.2 and I don't see
> >> this failure.  
> >
> >Good point!
> >
> >I am using gcc version 11.3.1 20230605 (Red Hat 11.4.1-2) on x86_64.
> >I see the same behavior on gcc version 8.5.0, which for all I know might
> >be too old.  
> 
> I could reproduce it with allmodconfig and gcc 11.4.1 from rockylinux,
> but not with gcc 9.3 or 12.3. Also it's not reproduced with gcc 11.4.1
> when using defconfig + CONFIG_DRM_XE  (even if  -Wstringop-overflow is
> still added).
> 
> I don't see a bug in the code, even if it inverts the head/tail
> convention.
> 
> Searching around showed this which may be relevant: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101854
> At least I can reproduce the same issue as in the snippet provided
> (https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101854#c7) with the buggy
> compiler.
> 
> So, maybe the best thing to do for now is to disable -Wstringop-overflow
> for gcc < 12?
> 
> 
> ------8<-----
> diff --git a/drivers/gpu/drm/xe/Makefile b/drivers/gpu/drm/xe/Makefile
> index 6952da8979ea..0433a3c6cbfd 100644
> --- a/drivers/gpu/drm/xe/Makefile
> +++ b/drivers/gpu/drm/xe/Makefile
> @@ -17,7 +17,7 @@ subdir-ccflags-y += $(call cc-option, -Wunused-const-variable)
>   subdir-ccflags-y += $(call cc-option, -Wpacked-not-aligned)
>   subdir-ccflags-y += $(call cc-option, -Wformat-overflow)
>   subdir-ccflags-y += $(call cc-option, -Wformat-truncation)
> -subdir-ccflags-y += $(call cc-option, -Wstringop-overflow)
> +subdir-ccflags-$(call gcc-min-version, 120000) += $(call cc-option, -Wstringop-overflow)
>   subdir-ccflags-y += $(call cc-option, -Wstringop-truncation)
>   # The following turn off the warnings enabled by -Wextra
>   ifeq ($(findstring 2, $(KBUILD_EXTRA_WARN)),)
> ------8<-----
> 
> and if we are tweaking the warnings, then do similarly in scripts/Makefile.extrawarn
> so it doesn't show up again with W=1 builds. Thoughts?

The top level Makefile (in linux-next) has:

#Currently, disable -Wstringop-overflow for GCC 11, globally.
KBUILD_CFLAGS-$(CONFIG_CC_NO_STRINGOP_OVERFLOW) += $(call cc-option, -Wno-stringop-overflow)
KBUILD_CFLAGS-$(CONFIG_CC_STRINGOP_OVERFLOW) += $(call cc-option, -Wstringop-overflow)

and init/Kconfig has:

# Currently, disable -Wstringop-overflow for GCC 11, globally.
config GCC11_NO_STRINGOP_OVERFLOW
        def_bool y

config CC_NO_STRINGOP_OVERFLOW
        bool
        default y if CC_IS_GCC && GCC_VERSION >= 110000 && GCC_VERSION < 120000 && GCC11_NO_STRINGOP_OVERFLOW

config CC_STRINGOP_OVERFLOW
        bool
        default y if CC_IS_GCC && !CC_NO_STRINGOP_OVERFLOW

So, what does "grep -E '(STRINGOP_OVERFLOW|GCC_VERSION)' .config" show for your
breaking build(s)?

-- 
Cheers,
Stephen Rothwell

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]