This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: builtin_return_addr vs frame_pointer_needed vs -O3
- From: Ian Dall <ian at beware dot dropbear dot id dot au>
- To: rth at redhat dot com
- Cc: ian at beware dot dropbear dot id dot au, schwab at suse dot de, dj at redhat dot com, gcc at gcc dot gnu dot org, gcc-patches at gcc dot gnu dot org
- Date: Mon, 1 Jul 2002 23:19:11 +0930
- Subject: Re: builtin_return_addr vs frame_pointer_needed vs -O3
- References: <200101312226.RAA29943@greed.delorie.com><20010131170452.A13854@redhat.com><15641.30554.720720.529367@gargle.gargle.HOWL><20020626074538.C25683@redhat.com><jebs9y6pq3.fsf@sykes.suse.de><15642.18479.449591.859902@sibyl.beware.dropbear.id.au><20020628184211.A27214@redhat.com>
Richard Henderson writes:
> On Thu, Jun 27, 2002 at 08:33:11AM +0930, Ian Dall wrote:
> > Maybe that particular test should just not be run with -fomit-frame-pointer?
>
> That's probably easiest. You should be able to detect
> that in a .x file.
I'm no TCL guru, but the following patch seems to work. It is a bit ugly
because it assumes the existance of "option" and also that it is
evaluated in a loop so that "continue" means something. It might be
better to have a "torture_skip_pattern" variable which can optionally
be set. It would need some change to c-torture.exp though.
Also you might not want to do this for all targets, but I'd suggest
the default to be not to do the -fomit-frame-pointer case, and opt in
for certain targets if thats what people want to do.
Ian
*** testsuite/gcc.c-torture/execute/20010122-1.x 30 Apr 2002 23:39:33 -0000 1.1.1.1
--- testsuite/gcc.c-torture/execute/20010122-1.x 1 Jul 2002 13:37:46 -0000
***************
*** 5,8 ****
--- 5,10 ----
set torture_execute_xfail "mips*-sgi-irix6*"
}
+ set torture_eval_before_compile {if {[string match {*-fomit-frame-pointer*} "$option"]} {continue}}
+
return 0