"BFD 2.13.2.1 assertion fail" on Solaris

[Dimitri Papadopoulos-Orfanos]

Hi,

I get GNU ld assertion failures when linking with libGLU from Sun OpenGL 
1.3 on Solaris 8.


To reproduce the problem on a machine with Sun OpenGL 1.3 installed:

$ cat > foo.c
#include <GL/glu.h>
void foo()
{
   gluCheckExtension(0, 0);
}
$
$ gcc -shared foo.c \
	-I/usr/openwin/include -L/usr/openwin/lib -lGLU
/usr/local/binutils/bin/ld: BFD 2.13.2.1 assertion fail elf32-sparc.c:1490
/usr/local/binutils/bin/ld: BFD 2.13.2.1 assertion fail elf32-sparc.c:1764
$


Linking with Sun OpenGL 1.2.3 or earlier works just fine. The problem 
happens only with Sun OpenGL 1.3.


What are the differences between libGLU from Sun OpenGL 1.2.3 and libGLU 
from Sun OpenGL 1.3?


On a machine with Sun OpenGL 1.2.3 installed:

$ /usr/ccs/bin/nm /usr/openwin/lib/libGLU.so | fgrep '|1 '
[2]     |       148|       0|SECT |LOCL |0    |1      |
[102]   |         0|       0|OBJT |LOCL |0    |1      |_START_
$


On a machine with Sun OpenGL 1.3 installed:

$ /usr/ccs/bin/nm /usr/openwin/lib/libGLU.so | fgrep '|1 '
[2]     |       148|       0|SECT |LOCL |0    |1      |
[1833]  |    673804|       0|OBJT |GLOB |1    |14     |_GLOBAL_OFFSET_TABLE_
[102]   |         0|       0|OBJT |LOCL |0    |1      |_START_
[1939]  |    602172|     192|FUNC |GLOB |1    |11     |gluCheckExtension
[1885]  |    589708|    4728|FUNC |GLOB |1    |11     |gluPartialDisk
[1893]  |    400764|     788|FUNC |GLOB |1    |11     |gluTessEndPolygon
$


As you can see the "Other" field is set to 1 for 3 functions of the Sun 
OpenGL 1.3 libGLU library. The 'nm' manual page reads:
	A field reserved for future use, currently containing 0.
See:
	http://docs.sun.com/db/doc/816-0210/6m6nb7mg9?a=view#OUTPUT


So there is a problem with Sun OpenGL 1.3 here. The "Other" field is set 
to 1 although it should be set to 0 according to the Sun manuals.


That said, the Sun linker is able to link libGLU, and GNU ld should fail 
in such cases. There are libraries with the "Other" field set on 
Solaris, whether we want it or not, we have to live with it?


Has anyone any more clues about it? Will this bug report be taken into 
account, or should I rather send it to the bug-binutils list?


Regards,
--
Dimitri Papadopoulos