This is the mail archive of the
newlib@sources.redhat.com
mailing list for the newlib project.
Re: gcc 3.2 CVS + newlib CVS status report, i960-elf failure
- From: Thomas Fitzsimmons <fitzsim at redhat dot com>
- To: joel dot sherrill at OARcorp dot com
- Cc: newlib at sources dot redhat dot com
- Date: 06 May 2002 18:55:52 -0400
- Subject: Re: gcc 3.2 CVS + newlib CVS status report, i960-elf failure
- References: <3CBDAF49.9D7889B0@OARcorp.com>
On Wed, 2002-04-17 at 13:22, Joel Sherrill wrote:
>
> Hi,
>
> I built gcc 3.2 CVS and newlib CVS overnight for a bunch of
> targets. Ignoring the mem_err/debug.h sparc problem I have
> locally worked around, it should match CVS. I only built
> C since that was enough to exercise newlib and find similar
> multiline asm problems in RTEMS. The targets listed as OK
> that have simulators have testresults reported to gcc
> already. Here is what I have (ignoring warnings):
>
> a29k-coff - gcc internal error in unwind-dw2-fde.c
> arc-elf - gcc problem with crtinit .. gcc PR3587
> arm-elf - OK
> arm-rtems - OK
> avr-elf - OK
> d30v-elf - OK
> fr30-elf - GCC ICE at emit-rtl.c:1874
> h8300-coff - OK
> h8300-rtems - OK
> hppa1.1-proelf - ICE compiling findfp.c
> i386-elf - OK
> i386-rtems - OK
> i960-coff - OK
> i960-elf - not ok, libgloss fails to build
Hello,
I looked into this failure. There are two issues. The simpler issue is
that the i960-elf gcc port doesn't support the "section" attribute that
appears in the ELF-specific section of libgloss/libnosys/warning.h.
This is easy to workaround. However, the larger issue is that in
newlib, "as" fails with an internal error while assembling vfprintf.s
(generated by compiling newlib/libc/stdio/vfprintf.c). I've submitted a
bug report to bug-binutils@gnu.org.
I'll hold off committing a workaround for the libgloss problem until I
can properly test it with newlib.
Tom
--
Thomas Fitzsimmons
Red Hat Canada Limited e-mail: fitzsim@redhat.com
2323 Yonge Street, Suite 300
Toronto, ON M4P2C9