Re: [PATCH 3/3] x86/build: Specify stack alignment for clang

[Matthias Kaehlcke <mka@chromium.org>]

El Tue, Jun 13, 2017 at 08:38:00AM +0200 Ingo Molnar ha dit:

> 
> * Matthias Kaehlcke <mka@chromium.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 for 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@chromium.org>
> > ---
> >  arch/x86/Makefile | 18 +++++++++++++-----
> >  1 file changed, 13 insertions(+), 5 deletions(-)
> > 
> > diff --git a/arch/x86/Makefile b/arch/x86/Makefile
> > index 86b725d69423..7f6c33f4d428 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)
> > +	stack_align_opt := -mstack-alignment
> > +else
> > +	stack_align_opt := -mpreferred-stack-boundary
> > +endif
> 
> Nit: I'd name it cc_stack_align_opt or so, to make it clear this is a C compiler 
> option.

Sounds good

> > @@ -65,8 +73,8 @@ 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)
> > +        # prevent the compiler from keeping the stack 16 byte aligned
> > +        KBUILD_CFLAGS += $(call cc-option,$(stack_align_opt)=2)
> 
> So the comment appears inaccurate: the point isn't really to 'keep' the compiler 
> from 16-byte alignment (there's nothing wrong with that, functionally), the point 
> is to use a more optimal alignment, right?

I don't know for sure what is the reason for the 4 byte alignment,
since there is no history for the option being added (apparently it
was added in 1999 by Artur Skawina). I suppose it is an optimization
to save stack space and reduce code size by reducing the need for
extra alignment instructions. Please correct me if I'm wrong.

> >          # 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 +106,8 @@ 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)
> > +	# Align the stack to 8 bytes if supported.
> > +	KBUILD_CFLAGS += $(call cc-option,$(stack_align_opt)=3)
> 
> Here too the reason should be outlined: performance, features or correctness?

Probably the same as for 32-bit, plus the stack being aligned in any
case at an 8-byte boundary by the kernel entry. If you have more
wisdom to add please let me know.