Re: PATCH: Support VER_FLG_INFO in readelf

[Rainer Orth <ro at CeBiTec dot Uni-Bielefeld dot DE>]

Ian Lance Taylor <iant@google.com> writes:

> Fixed like so.

Links now, thanks.

> How does gold do on its testsuite on Solaris?

Pretty terrible, please see the attached output of make -k check.

The errors are of a few types only, though:

gcctestdir/ld: error: /usr/lib/libm.so: duplicate definition for version 0

gcctestdir/ld: fatal error: basic_test.o: readv failed: Invalid argument

gcctestdir/ld: error: cannot find -lm
gcctestdir/ld: error: cannot find -lc

gcctestdir/ld: error: /vol/gcc-4.4/lib/gcc/sparc-sun-solaris2.10/4.4.2/../../../libstdc++.a(class_type_info.o): unsupported reloc 18 against local symbol

plus a couple of Sun as errors:

/usr/ccs/bin/as: "/var/tmp//ccBsZ09e.s", line 10: error: unknown opcode ".symver
"

Sun as doesn't know about .symver.

/usr/ccs/bin/as: "/var/tmp//ccBsZ09e.s", line 10: error: invalid character (0x40
)

I haven't checked this one.

/usr/ccs/bin/as: "/var/tmp//ccazMM5u.s", line 12: error: unknown opcode ".protec
ted"

This is a gcc bug: Sun as uses .symbolic for STV_PROTECTED.

/usr/ccs/bin/as: "/var/tmp//ccQGkh7a.s", line 37: error: unknown opcode ".intern
al"

Another gcc bug: STV_INTERNAL is marked as `currently reserved'.  Both
ld and Sun cc emit .hidden instead.

I'm testing a patch for the last two.

Unfortunately, make -j<N> check output is completely unusable, so
running the testsuite on a T5220 is pretty slow and it took me some time
to get back with this.

        Rainer

-- 
-----------------------------------------------------------------------------
Rainer Orth, Center for Biotechnology, Bielefeld University

Attachment: gold-check.out.bz2
Description: gold make -k check output on Solaris 10/SPARC