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]

Re: [libstdc++] Un-precious the CC-related variables


 > Andreas Jaeger <aj@suse.de> writes:
 > > Phil Edwards <phil@jaj.com> writes:
 > > > Folks are finding that we have to remove the "precious variable"
 > > > protection from CC, like I did with CXX et al.  This should help
 > > > with the multilibs problems until multilibs get moved to the top
 > > > level, although I don't have x86_64 or irix systems to verify
 > > > that.
 > > 
 > > Thanks, this seems to fix it for me on x86_64,
 > 
 > And for me on irix, thanks.
 > Richard

Hmm, the patch went in, right?  I still get an error on irix6.5 as of
last night:

	creating reloadable object files...
	creating a temporary reloadable object file: .libs/libg2c.la-2.o
	/usr/bin/ld -n32 -64 -r -o .libs/libg2c.la-1.o libE77/.libs/Labort.o
	libE77/.libs/Lderf.o [...]
	ld64: ERROR   4  : Conflicting flag setting: -32-set-32bit-path
	ld64: ERROR   4  : Conflicting flag setting: -64
	        -n32 may conflict with -o32
	ld64: FATAL   12 : Expecting n32 objects: libE77/.libs/Labort.o is n64.
	make[5]: *** [libg2c.la] Error 4
	make[5]: Leaving directory
	`/caip/u58/ghazi/gcc-testing/build/mips-sgi-irix6.5/mabi=64/libf2c'

I use CC='cc -n32' for stage1 if that is relevant (I bet it is).  I
have to do that because the system compiler generates o32 by default.

Strangely, it only fails when I do a serial make.  Parallel make -j2
or higher seems to work. (?)

		--Kaveh
--
Kaveh R. Ghazi			ghazi@caip.rutgers.edu


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