This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

[committed] Check for -march in config/mips/iris6.h


Richard Sandiford <rsandifo@redhat.com> writes:
> As things stand, the patch will cause a few regressions
> for mips-sgi-irix6.5/-mabi=32 because of the following iris6.h definition:
>
>     /* Force the default ABI onto the command line in order to make the specs
>        easier to write.  Default to the mips2 ISA for the O32 ABI.  */
>     #define DRIVER_SELF_SPECS \
>       "%{!mabi=*: -mabi=n32}", \
>       "%{mabi=32: %{!mips*: -mips2}}"
>
> We really ought to be checking -march as well as -mips here.
> I'm testing that patch now, and if everything goes OK, I'll commit
> it before I commit anything like the changes below.

Here's the IRIX specs patch.  Bootstrapped & regression tested
on mips-sgi-irix6.5.  Applied to mainline.

Richard


	* config/mips/iris6.h (DRIVER_SELF_SPECS): Check -march as well as
	-mipsN before forcing a default of -mips2.

Index: config/mips/iris6.h
===================================================================
RCS file: /cvs/gcc/gcc/gcc/config/mips/iris6.h,v
retrieving revision 1.83
diff -u -p -F^\([(a-zA-Z0-9_]\|#define\) -r1.83 iris6.h
--- config/mips/iris6.h	7 Feb 2005 09:27:46 -0000	1.83
+++ config/mips/iris6.h	17 Apr 2005 07:48:14 -0000
@@ -32,7 +32,7 @@ #define MULTILIB_DEFAULTS { "mabi=n32" }
    easier to write.  Default to the mips2 ISA for the O32 ABI.  */
 #define DRIVER_SELF_SPECS \
   "%{!mabi=*: -mabi=n32}", \
-  "%{mabi=32: %{!mips*: -mips2}}"
+  "%{mabi=32: %{!mips*: %{!march*: -mips2}}}"
 
 /* Force the generation of dwarf .debug_frame sections even if not
    compiling -g.  This guarantees that we can unwind the stack.  */


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]