Re: Build regressions/improvements in v6.5-rc3

On Mon, Jul 24, 2023 at 02:43:02PM +0200, Geert Uytterhoeven wrote:
> On Mon, 24 Jul 2023, Geert Uytterhoeven wrote:
> > JFYI, when comparing v6.5-rc3[1] to v6.5-rc2[3], the summaries are:
> >  - build errors: +5/-0
> 
>   + /kisskb/src/include/linux/fortify-string.h: error: call to '__write_overflow_field' declared with attribute warning: detected write beyond size of field (1st parameter); maybe use struct_group()? [-Werror=attribute-warning]:  => 583:25, 493:25
> 
> mips-gcc13/mips-allmodconfig
> 
> Full context:
> 
>     In function 'fortify_memset_chk',
> 	inlined from 'memset_io' at /kisskb/src/arch/mips/include/asm/io.h:486:2,
> 	inlined from 'build_auth_frame' at /kisskb/src/drivers/net/wireless/legacy/ray_cs.c:2697:2:
>     /kisskb/src/include/linux/fortify-string.h:493:25: error: call to '__write_overflow_field' declared with attribute warning: detected write beyond size of field (1st parameter); maybe use struct_group()? [-Werror=attribute-warning]
>       493 |                         __write_overflow_field(p_size_field, size);
> 	  |                         ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>     In function 'fortify_memcpy_chk',
> 	inlined from 'memcpy_toio' at /kisskb/src/arch/mips/include/asm/io.h:494:2,
> 	inlined from 'translate_frame' at /kisskb/src/drivers/net/wireless/legacy/ray_cs.c:955:3,
> 	inlined from 'ray_hw_xmit.constprop' at /kisskb/src/drivers/net/wireless/legacy/ray_cs.c:912:12:
>     /kisskb/src/include/linux/fortify-string.h:583:25: error: call to '__write_overflow_field' declared with attribute warning: detected write beyond size of field (1st parameter); maybe use struct_group()? [-Werror=attribute-warning]
>       583 |                         __write_overflow_field(p_size_field, size);
> 	  |                         ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> 
> Single-element flexible array abuse in drivers/net/wireless/legacy/rayctl.h:tx_msg.var

^^^^
is this line from you or is there a tool outputting this? Because, yes,
very true:

struct tx_msg {
    struct tib_structure tib;
    struct phy_header phy;
    struct mac_header mac;
    UCHAR  var[1];
};

I'll send a patch.

-- 
Kees Cook