glibc w/egcs problems

[Dan Jacobowitz]

On Fri, Oct 09, 1998 at 12:55:27PM -0700, Ulrich Drepper wrote:
> I thought I start trying to compile glibc with the current egcs but I
> didn't get far.  At
> 
> 	http://www.cygnus.com/~drepper/strtof.i.gz
> 
> you can find the preprocessed source of the first file crashing gcc.
> Running egcs as of 1998-10-09T12:00-0700 started with
> 
> 	cc1 strtof.i -fPIC -O3
> 
> prints
> 
>   gcc/reload.c:4401: Internal compiler error in function find_reloads_toplev
> 
> before crashing.

I didn't get this as early in the build process as Ulrich did, but on
powerpc and using the 1005 snapshot I see this:

egcc test-float.i -c -O

In file included from ../include/bits/mathcalls.h:1,
                 from ../math/math.h:97,
                 from ../include/math.h:1,
                 from libm-test.c:123,
                 from test-float.c:33:
../math/bits/mathcalls.h:63: warning: conflicting types for built-in function `cosl'
../math/bits/mathcalls.h:65: warning: conflicting types for built-in function `sinl'
../math/bits/mathcalls.h:146: warning: conflicting types for built-in function `sqrtl'
../math/bits/mathcalls.h:165: warning: conflicting types for built-in function `fabsl'
../../gcc/reload.c:4421: Internal compiler error in function find_reloads_toplev


Interestingly enough, if I leave out the -O, I see this:

libm-test.c: In function `random_value':
In file included from test-float.c:33:
libm-test.c:212: inconsistent operand constraints in an `asm'
libm-test.c: In function `test_not_exception':
libm-test.c:324: inconsistent operand constraints in an `asm'

(a total of ten times).

Dan
-------------- next part --------------
A non-text attachment was scrubbed...
Name: test-float.i.bz2
Type: application/x-bzip2
Size: 32697 bytes
Desc: not available
URL: <http://gcc.gnu.org/pipermail/gcc-bugs/attachments/19981010/77c61bff/attachment.bz2>