linux-kernel - Re: [PATCH v4 3/3] x86/build: Specify stack alignment for clang

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20170619204704.GP141096@google.com>
Date:   Mon, 19 Jun 2017 13:47:04 -0700
From:   Matthias Kaehlcke <mka@...omium.org>
To:     hpa@...or.com
Cc:     Thomas Gleixner <tglx@...utronix.de>,
        Ingo Molnar <mingo@...hat.com>,
        "H . J . Lu" <hjl.tools@...il.com>,
        David Woodhouse <dwmw2@...radead.org>,
        Masahiro Yamada <yamada.masahiro@...ionext.com>,
        Michal Marek <mmarek@...e.com>, x86@...nel.org,
        linux-kbuild@...r.kernel.org, linux-kernel@...r.kernel.org,
        Michael Davidson <md@...gle.com>,
        Greg Hackmann <ghackmann@...gle.com>,
        Nick Desaulniers <ndesaulniers@...gle.com>,
        Stephen Hines <srhines@...gle.com>,
        Kees Cook <keescook@...omium.org>,
        Arnd Bergmann <arnd@...db.de>,
        Bernhard.Rosenkranzer@...aro.org,
        Peter Foley <pefoley2@...oley.com>,
        Behan Webster <behanw@...verseincode.com>,
        Douglas Anderson <dianders@...omium.org>
Subject: Re: [PATCH v4 3/3] x86/build: Specify stack alignment for clang

El Mon, Jun 19, 2017 at 01:17:03PM -0700 hpa@...or.com ha dit:

> On June 19, 2017 11:37:57 AM PDT, Matthias Kaehlcke <mka@...omium.org> wrote:
> >For gcc stack alignment is configured with
> >-mpreferred-stack-boundary=N,
> >clang has the option -mstack-alignment=N for that purpose. Use the same
> >alignment as with gcc.
> >
> >If the alignment is not specified clang assumes an alignment of
> >16 bytes, as required by the standard ABI. However as mentioned in
> >d9b0cde91c60 ("x86-64, gcc: Use -mpreferred-stack-boundary=3 if
> >supported") the standard kernel entry on x86-64 leaves the stack
> >on an 8-byte boundary, as a consequence clang will keep the stack
> >misaligned.
> >
> >Signed-off-by: Matthias Kaehlcke <mka@...omium.org>
> >---
> > arch/x86/Makefile | 26 +++++++++++++++++++++-----
> > 1 file changed, 21 insertions(+), 5 deletions(-)
> >
> >diff --git a/arch/x86/Makefile b/arch/x86/Makefile
> >index b2dae639f778..9406d3670452 100644
> >--- a/arch/x86/Makefile
> >+++ b/arch/x86/Makefile
> >@@ -11,6 +11,14 @@ else
> >         KBUILD_DEFCONFIG := $(ARCH)_defconfig
> > endif
> > 
> >+# Handle different option names for specifying stack alignment with
> >gcc and
> >+# clang.
> >+ifeq ($(cc-name),clang)
> >+	cc_stack_align_opt := -mstack-alignment
> >+else
> >+	cc_stack_align_opt := -mpreferred-stack-boundary
> >+endif
> >+
> ># How to compile the 16-bit code.  Note we always compile for
> >-march=i386;
> > # that way we can complain to the user if the CPU is insufficient.
> > #
> >@@ -28,7 +36,7 @@ REALMODE_CFLAGS	:= $(M16_CFLAGS) -g -Os -D__KERNEL__
> >\
> > 
> >REALMODE_CFLAGS += $(call __cc-option, $(CC), $(REALMODE_CFLAGS),
> >-ffreestanding)
> >REALMODE_CFLAGS += $(call __cc-option, $(CC), $(REALMODE_CFLAGS),
> >-fno-stack-protector)
> >-REALMODE_CFLAGS += $(call __cc-option, $(CC), $(REALMODE_CFLAGS),
> >-mpreferred-stack-boundary=2)
> >+REALMODE_CFLAGS += $(call __cc-option, $(CC), $(REALMODE_CFLAGS),
> >$(cc_stack_align_opt)=2)
> > export REALMODE_CFLAGS
> > 
> > # BITS is used as extension for files which are available in a 32 bit
> >@@ -65,8 +73,10 @@ ifeq ($(CONFIG_X86_32),y)
> >         # with nonstandard options
> >         KBUILD_CFLAGS += -fno-pic
> > 
> >-        # prevent gcc from keeping the stack 16 byte aligned
> >-        KBUILD_CFLAGS += $(call
> >cc-option,-mpreferred-stack-boundary=2)
> >+        # Align the stack to the register width instead of using the
> >default
> >+        # alignment of 16 bytes. This reduces stack usage and the
> >number of
> >+        # alignment instructions.
> >+        KBUILD_CFLAGS += $(call cc-option,$(cc_stack_align_opt)=2)
> > 
> ># Disable unit-at-a-time mode on pre-gcc-4.0 compilers, it makes gcc
> >use
> >         # a lot more stack due to the lack of sharing of stacklots:
> >@@ -98,8 +108,14 @@ else
> >         KBUILD_CFLAGS += $(call cc-option,-mno-80387)
> >         KBUILD_CFLAGS += $(call cc-option,-mno-fp-ret-in-387)
> > 
> >-	# Use -mpreferred-stack-boundary=3 if supported.
> >-	KBUILD_CFLAGS += $(call cc-option,-mpreferred-stack-boundary=3)
> >+        # By default gcc and clang use a stack alignment of 16 bytes
> >for x86.
> >+        # However the standard kernel entry on x86-64 leaves the stack
> >on an
> >+        # 8-byte boundary. If the compiler isn't informed about the
> >actual
> >+        # alignment it will generate extra alignment instructions for
> >the
> >+        # default alignment which keep the stack *mis*aligned.
> >+        # Furthermore an alignment to the register width reduces stack
> >usage
> >+        # and the number of alignment instructions.
> >+        KBUILD_CFLAGS += $(call cc-option,$(cc_stack_align_opt)=3)
> > 
> > 	# Use -mskip-rax-setup if supported.
> > 	KBUILD_CFLAGS += $(call cc-option,-mskip-rax-setup)
> 
> Goddammit.
> 
> How many times do I have to say NAK to
> 
> >+ifeq ($(cc-name),clang)
> 
> ... before it sinks in?

The initial version of this patch doesn't have this condition and just
uses cc-option to select the appropriate option. Ingo didn't like the
duplication and suggested the use of a variable, which kinda implies a
check for the compiler name. I also think this is a cleaner
solution. but I'm happy to respin the patch if you have another
suggestion that is ok for both of you.