[PATCH] i386: Don't use frame pointer without stack access

[H.J. Lu]

On Mon, Aug 7, 2017 at 6:32 AM, Michael Matz <matz@suse.de> wrote:
> Hi,
>
> On Mon, 7 Aug 2017, H.J. Lu wrote:
>
>> >> This will break unwinders relying on frame pointers to exist on all
>> >> functions, for which projects conciously forced a frame pointer with this
>> >> option.  I don't think we can simply override user specified explicit
>> >> wishes in this way, presumably they had a reason to use it.
>> >
>> > Hm... yes, you are right.
>> >
>> > HJ, please revert the patch.
>> >
>>
>> Is there a testcae?  I'd like to add it.
>
> Trivial change of your first example, see below.
>
>> [hjl@gnu-tools-1 pr81736]$ clang -S -O2 -fno-omit-frame-pointer x.i
>> [hjl@gnu-tools-1 pr81736]$ cat x.s
> [... no frame ...]
>> [hjl@gnu-tools-1 pr81736]$
>>
>> Does it mean clang is broken?
>
> In my book, yes.

Does GCC do this for all targets or just x86?

>
> Ciao,
> Michael.
>
> Index: gcc/testsuite/gcc.target/i386/force-frame.c
> ===================================================================
> --- gcc/testsuite/gcc.target/i386/force-frame.c (revision 0)
> +++ gcc/testsuite/gcc.target/i386/force-frame.c (working copy)
> @@ -0,0 +1,14 @@
> +/* { dg-do compile } */
> +/* { dg-options "-O2 -fno-omit-frame-pointer" } */
> +
> +int
> +#ifndef __x86_64__
> +__attribute__((regparm(3)))
> +#endif
> +foo (int i)
> +{
> +  return i;
> +}
> +
> +/* The user wants a frame pointer.  */
> +/* { dg-final { scan-assembler "%\[re\]bp" } } */

I am looking for a run-time test which breaks unwinder.

-- 
H.J.