[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAGXu5j+wXoAaAFzmnU=VuYeHrGTyzBsEnCdeALYb=jXUaeeQ6g@mail.gmail.com>
Date: Wed, 16 May 2018 06:59:02 -0700
From: Kees Cook <keescook@...omium.org>
To: Nadav Amit <namit@...are.com>
Cc: LKML <linux-kernel@...r.kernel.org>,
Nadav Amit <nadav.amit@...il.com>,
Thomas Gleixner <tglx@...utronix.de>,
Ingo Molnar <mingo@...hat.com>,
"H. Peter Anvin" <hpa@...or.com>, X86 ML <x86@...nel.org>,
Jan Beulich <JBeulich@...e.com>,
Josh Poimboeuf <jpoimboe@...hat.com>
Subject: Re: [RFC 5/8] x86: refcount: prevent gcc distortions
On Tue, May 15, 2018 at 7:11 AM, Nadav Amit <namit@...are.com> wrote:
> GCC considers the number of statements in inlined assembly blocks,
> according to new-lines and semicolons, as an indication to the cost of
> the block in time and space. This data is distorted by the kernel code,
> which puts information in alternative sections. As a result, the
> compiler may perform incorrect inlining and branch optimizations.
>
> The solution is to set an assembly macro and call it from the inlined
> assembly block. As a result GCC considers the inline assembly block as
> a single instruction.
>
> This patch allows to inline functions such as __get_seccomp_filter().
> The effect of the patch is as follows on the kernel size:
>
> text data bss dec hex filename
> 18146418 10064100 2936832 31147350 1db4556 ./vmlinux before
> 18148228 10063968 2936832 31149028 1db4be4 ./vmlinux after (+1678)
>
> Static text symbols:
> Before: 39673
> After: 39649 (-24)
>
> Cc: Thomas Gleixner <tglx@...utronix.de>
> Cc: Ingo Molnar <mingo@...hat.com>
> Cc: "H. Peter Anvin" <hpa@...or.com>
> Cc: x86@...nel.org
> Cc: Kees Cook <keescook@...omium.org>
> Cc: Jan Beulich <JBeulich@...e.com>
> Cc: Josh Poimboeuf <jpoimboe@...hat.com>
>
> Signed-off-by: Nadav Amit <namit@...are.com>
> ---
> arch/x86/include/asm/refcount.h | 55 ++++++++++++++++++++-------------
> 1 file changed, 33 insertions(+), 22 deletions(-)
>
> diff --git a/arch/x86/include/asm/refcount.h b/arch/x86/include/asm/refcount.h
> index 4cf11d88d3b3..a668c534206d 100644
> --- a/arch/x86/include/asm/refcount.h
> +++ b/arch/x86/include/asm/refcount.h
> @@ -14,34 +14,43 @@
> * central refcount exception. The fixup address for the exception points
> * back to the regular execution flow in .text.
> */
> -#define _REFCOUNT_EXCEPTION \
> - ".pushsection .text..refcount\n" \
> - "111:\tlea %[counter], %%" _ASM_CX "\n" \
> - "112:\t" ASM_UD2 "\n" \
> - ASM_UNREACHABLE \
> - ".popsection\n" \
> - "113:\n" \
> +
> +asm ("\n"
> + ".macro __REFCOUNT_EXCEPTION counter:vararg\n\t"
Why are these vararg?
Also, I think for the whole series, these #define-a-macro cases need a
comment in the code. It's not obvious from looking at the code why
they've defined a macro instead of just leaving the asm as it was.
Beyond that, as long as there is no behavioral changes, I'm fine with
the changes.
-Kees
--
Kees Cook
Pixel Security
Powered by blists - more mailing lists