Re: [PATCH] Compile debug/stack_chk_fail_local.c with stack protector

[Carlos O'Donell <carlos at redhat dot com>]

On 07/05/2018 07:28 AM, Florian Weimer wrote:
> The resulting object file is statically linked into applications, so
> it is desirable to have (formal) stack protector coverage there.

The broader context is this:

 60 # Building the stack-protector failure routines with stack protection
 61 # makes no sense.
 62 
 63 CFLAGS-stack_chk_fail.c += $(no-stack-protector)
 64 CFLAGS-stack_chk_fail_local.c += $(no-stack-protector)
 65 

Should we expand the comment to say something like this?

# Building the stack-protector failure routines with stack protection
# is not required since we have already failed a stack check and are
# exiting the process. However, the local aliases which jump to the real
# routines should still be compiled with stack protection
# e.g. stack_chk_fail_local.c

Otherwise the comment is out-of-sync with what we're doing here.

OK with that change.

Reviewed-by: Carlos O'Donell <carlos@redhat.com>

> 2018-07-05  Florian Weimer  <fweimer@redhat.com>
> 
> 	* debug/Makefile (CFLAGS-stack_chk_fail_local.c): Remove
> 	$(no-stack-protector).  stack_chk_fail_local.c can be compiled
> 	with stack protector enabled because there is no risk of infinite
> 	recursion.
> 
> diff --git a/debug/Makefile b/debug/Makefile
> index c6f6feb0ec..d9683a8c8c 100644
> --- a/debug/Makefile
> +++ b/debug/Makefile
> @@ -61,7 +61,6 @@ elide-routines.o := stack_chk_fail_local
>  # makes no sense.
>  
>  CFLAGS-stack_chk_fail.c += $(no-stack-protector)
> -CFLAGS-stack_chk_fail_local.c += $(no-stack-protector)
>  
>  CFLAGS-backtrace.c += -fno-omit-frame-pointer -funwind-tables
>  CFLAGS-sprintf_chk.c += $(libio-mtsafe)
> 


-- 
Cheers,
Carlos.