On Tue, Feb 16, 2016 at 11:20 AM, Geert Uytterhoeven
<[email protected]> wrote:
> Hi Kees,
>
> On Tue, Feb 16, 2016 at 7:38 PM, Kees Cook <[email protected]> wrote:
>> This fixes the build for both allnoconfig (!CONFIG_MMU) and with
>> XIP_KERNEL (due to missing pieces in the linker script), as seen after
>> both "ARM: 8502/1: mm: mark section-aligned portion of rodata NX" and
>> "ARM: 8513/1: xip: Move XIP linking to a separate file".
>>
>> Reported-by: Russell King <[email protected]>
>> Reported-by: Arnd Bergmann <[email protected]>
>> Signed-off-by: Kees Cook <[email protected]>
>
>> diff --git a/arch/arm/kernel/vmlinux-xip.lds.S
>> b/arch/arm/kernel/vmlinux-xip.lds.S
>> index 40bc4cadb959..2da60cd1b295 100644
>> --- a/arch/arm/kernel/vmlinux-xip.lds.S
>> +++ b/arch/arm/kernel/vmlinux-xip.lds.S
>
>> @@ -310,6 +311,17 @@ SECTIONS
>> }
>>
>> /*
>> + * Without CONFIG_DEBUG_ALIGN_RODATA, __start_rodata_section_aligned will
>> + * be the first section-aligned location after __start_rodata. Otherwise,
>> + * it will be equal to __start_rodata.
>> + */
>> +#ifdef CONFIG_MMU
>
> CONFIG_DEBUG_ALIGN_RODATA or CONFIG_MMU?
>
>> +__start_rodata_section_aligned = ALIGN(__start_rodata, 1 << SECTION_SHIFT);
CONFIG_MMU controls the availability of SECTION_SHIFT, so I used that.
Either will work, since without CONFIG_DEBUG_ALIGN_RODATA, the values
will end up being identical, as in the #else.
Though it looks like XIP_KERNEL should kill all of the
CONFIG_DEBUG_*RODATA defines anyway. I think we need a different
version of this combined with the patch from Ard...
-Kees
>> +#else
>> +__start_rodata_section_aligned = __start_rodata;
>> +#endif
>> +
>> +/*
>> * These must never be empty
>> * If you have to comment these two assert statements out, your
>> * binutils is too old (for other reasons as well)
>> diff --git a/arch/arm/kernel/vmlinux.lds.S b/arch/arm/kernel/vmlinux.lds.S
>> index 1707b9431761..f13960e44955 100644
>> --- a/arch/arm/kernel/vmlinux.lds.S
>> +++ b/arch/arm/kernel/vmlinux.lds.S
>> @@ -328,7 +328,11 @@ SECTIONS
>> * be the first section-aligned location after __start_rodata. Otherwise,
>> * it will be equal to __start_rodata.
>> */
>> +#ifdef CONFIG_MMU
>
> Likewise (the out-of-scope comment says CONFIG_DEBUG_ALIGN_RODATA)?
>
>> __start_rodata_section_aligned = ALIGN(__start_rodata, 1 << SECTION_SHIFT);
>> +#else
>> +__start_rodata_section_aligned = __start_rodata;
>> +#endif
>
> Gr{oetje,eeting}s,
>
> Geert
>
> --
> Geert Uytterhoeven -- There's lots of Linux beyond ia32 --
> [email protected]
>
> In personal conversations with technical people, I call myself a hacker. But
> when I'm talking to journalists I just say "programmer" or something like
> that.
> -- Linus Torvalds
--
Kees Cook
Chrome OS & Brillo Security
