Using -fpic rather than -fPIC for m68k

[Richard Sandiford]

Richard Sandiford <richard at codesourcery.com> writes:
> The attached patch makes m68k targets use -fpic instead of -fPIC.
> Tested on a coldfire m68k-uclinux configuration.  Please install if OK.

Sorry, please ignore the second hunk of that patch, which was a local hack
to get around the fact that linux's binfmt_flat.c doesn't initialize %a1.
I've re-attached the proper hunk below.

Richard


-------------- next part --------------
Index: Rules.mak
===================================================================
--- Rules.mak	(revision 14626)
+++ Rules.mak	(working copy)
@@ -250,6 +250,13 @@ ifeq ($(TARGET_ARCH),cris)
 	PIEFLAG_NAME:=-fpie
 endif
 
+ifeq ($(TARGET_ARCH),m68k)
+	# -fPIC is only supported for 68020 and above.  It is not supported
+	# for 68000, 68010, or Coldfire.
+	PICFLAG:=-fpic
+	PIEFLAG_NAME:=-fpie
+endif
+
 ifeq ($(TARGET_ARCH),powerpc)
 # PowerPC can hold 8192 entries in its GOT with -fpic which is more than
 # enough. Therefore use -fpic which will reduce code size and generates