Re: [PATCH 1/1] x86/speculation, objtool: Annotate indirect calls/jumps for objtool on 32bit

[Peter Zijlstra <peterz@infradead.org>]

On Wed, Mar 14, 2018 at 11:24:27AM +0000, Andy Whitcroft wrote:
> In 9e0e3c5130e9 ("x86/speculation, objtool: Annotate indirect calls/jumps
> for objtool") we added annotations for CALL_NOSPEC/JMP_NOSPEC on x86 64bit.
> We did not annotate the 32bit path.  Annotate it similarly.
> 
> Signed-off-by: Andy Whitcroft <apw@canonical.com>
> ---
>  arch/x86/include/asm/nospec-branch.h | 5 ++++-
>  1 file changed, 4 insertions(+), 1 deletion(-)
> 
>     While reviewing indirect calls in our builds I noted that the
>     i386 retpoline CALL_NOSPEC is not annotated safe even though
>     its amd64 equivalent is.  I cannot see any reason this is not
>     also inherantly safe.  Peter was there a reason that you did
>     not annotate this one too?  Anyhow, on the assumption this was
>     just missed, this patch annotates it.

Yeah, just an oversight aided by the fact that I (obviously) never build
32bit kernels.

> diff --git a/arch/x86/include/asm/nospec-branch.h b/arch/x86/include/asm/nospec-branch.h
> index d0dabeae0505..07886162bdf8 100644
> --- a/arch/x86/include/asm/nospec-branch.h
> +++ b/arch/x86/include/asm/nospec-branch.h
> @@ -183,7 +183,10 @@
>   * otherwise we'll run out of registers. We don't care about CET
>   * here, anyway.
>   */
> -# define CALL_NOSPEC ALTERNATIVE("call *%[thunk_target]\n",	\
> +# define CALL_NOSPEC 						\
> +	ALTERNATIVE(						\
> +	ANNOTATE_RETPOLINE_SAFE					\
> +	"call *%[thunk_target]\n",				\
>  	"       jmp    904f;\n"					\
>  	"       .align 16\n"					\
>  	"901:	call   903f;\n"					\

Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>