Re: [PATCH 5/6] x86-64: Omit frame pointers on vread_tsc

[Ingo Molnar <mingo@elte.hu>]

* Andy Lutomirski <luto@MIT.EDU> wrote:

> vread_tsc is short and hot, and it's userspace code so the usual
> reasons to keep frame pointers around don't apply.
> 
> Signed-off-by: Andy Lutomirski <luto@mit.edu>
> ---
> 
> It might be nicer to move vread_tsc into a separate file and change
> the options directly.
> 
>  arch/x86/kernel/tsc.c |    3 +++
>  1 files changed, 3 insertions(+), 0 deletions(-)
> 
> diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c
> index a159fba..1aea4a6 100644
> --- a/arch/x86/kernel/tsc.c
> +++ b/arch/x86/kernel/tsc.c
> @@ -764,6 +764,8 @@ static cycle_t read_tsc(struct clocksource *cs)
>  }
>  
>  #ifdef CONFIG_X86_64
> +#pragma GCC push_options
> +#pragma GCC optimize ("omit-frame-pointer")
>  static cycle_t __vsyscall_fn vread_tsc(void)
>  {
>  	cycle_t ret;
> @@ -807,6 +809,7 @@ static cycle_t __vsyscall_fn vread_tsc(void)
>  	asm volatile ("");
>  	return last;
>  }
> +#pragma GCC pop_options
>  #endif

The cleaner option would be to put the __vsyscall functions into a new .c file 
and build it with framepointers off in the Makefile, which can be done via:

CFLAGS_vtsc.o := -fno-omit-frame-pointer

or so.

We could also consider putting this function into a .S too. Not sure it works 
out in a maintainable fashion though.

Thanks,

	Ingo