Re: [STABLE BACKPORT 4.4.y, 4.9.y and 4.14.y] compiler.h: Raise minimum version of GCC to 5.1 for arm64

[Nick Desaulniers]

On Mon, Jan 18, 2021 at 5:54 AM Will Deacon <will@xxxxxxxxxx> wrote:
>
> commit dca5244d2f5b94f1809f0c02a549edf41ccd5493 upstream.
>
> GCC versions >= 4.9 and < 5.1 have been shown to emit memory references
> beyond the stack pointer, resulting in memory corruption if an interrupt
> is taken after the stack pointer has been adjusted but before the
> reference has been executed. This leads to subtle, infrequent data
> corruption such as the EXT4 problems reported by Russell King at the
> link below.
>
> Life is too short for buggy compilers, so raise the minimum GCC version
> required by arm64 to 5.1.
>
> Reported-by: Russell King <linux@xxxxxxxxxxxxxxx>
> Suggested-by: Arnd Bergmann <arnd@xxxxxxxxxx>
> Signed-off-by: Will Deacon <will@xxxxxxxxxx>
> Tested-by: Nathan Chancellor <natechancellor@xxxxxxxxx>
> Reviewed-by: Nick Desaulniers <ndesaulniers@xxxxxxxxxx>
> Reviewed-by: Nathan Chancellor <natechancellor@xxxxxxxxx>
> Acked-by: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx>
> Cc: <stable@xxxxxxxxxxxxxxx> # 4.4.y, 4.9.y and 4.14.y only
> Cc: Theodore Ts'o <tytso@xxxxxxx>
> Cc: Florian Weimer <fweimer@xxxxxxxxxx>
> Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
> Cc: Nick Desaulniers <ndesaulniers@xxxxxxxxxx>
> Link: https://lore.kernel.org/r/20210105154726.GD1551@xxxxxxxxxxxxxxxxxxxxx
> Link: https://lore.kernel.org/r/20210112224832.10980-1-will@xxxxxxxxxx
> Signed-off-by: Catalin Marinas <catalin.marinas@xxxxxxx>
> [will: backport to 4.4.y/4.9.y/4.14.y]

Merging this from stable into "Android Common Kernel" trees that were
built with AOSP GCC 4.9, I expect this to break some builds.  Arnd,
IIRC did you mention that AOSP GCC had picked up a fix?  If so, did
you verify that via disassembly, or gerrit patch file?

(AOSP GCC 4.9: https://android.googlesource.com/platform/prebuilts/gcc/linux-x86/aarch64/aarch64-linux-android-4.9/
master branch, roll back a few commits).
It looks like AOSP GCC `#defines __android__ 1`.

I'm not arguing against a backport, just trying to think through how
we'll need to sort this out downstream.  (Revert or patch on top that
checks for __android__, if AOSP GCC does in fact have a fix.)

> Signed-off-by: Will Deacon <will@xxxxxxxxxx>
> ---
>  include/linux/compiler-gcc.h | 6 ++++++
>  1 file changed, 6 insertions(+)
>
> diff --git a/include/linux/compiler-gcc.h b/include/linux/compiler-gcc.h
> index af8b4a879934..3cc8adede67b 100644
> --- a/include/linux/compiler-gcc.h
> +++ b/include/linux/compiler-gcc.h
> @@ -145,6 +145,12 @@
>
>  #if GCC_VERSION < 30200
>  # error Sorry, your compiler is too old - please upgrade it.
> +#elif defined(CONFIG_ARM64) && GCC_VERSION < 50100
> +/*
> + * https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63293
> + * https://lore.kernel.org/r/20210107111841.GN1551@xxxxxxxxxxxxxxxxxxxxx
> + */
> +# error Sorry, your version of GCC is too old - please use 5.1 or newer.
>  #endif
>
>  #if GCC_VERSION < 30300
> --
> 2.30.0.284.gd98b1dd5eaa7-goog
>


-- 
Thanks,
~Nick Desaulniers