[Bug libfortran/22097] New: libgfortran build failure on mips-sgi-irix6.5

[Bug libfortran/22097] New: libgfortran build failure on mips-sgi-irix6.5

[fxcoudert at gcc dot gnu dot org]

Everything builds fine except for:

gmake[2]: Entering directory
`/tmp/gfortran-20050616/ibin/mips-sgi-irix6.5/libgfortran'
/bin/sh ./libtool --mode=compile (CC_FOR_TARGET) -DHAVE_CONFIG_H -I.
-I../../../gcc/libgfortran -I.  -iquote../../../gcc/libgfortran/io  -std=gnu99
-Wall -Wstrict-prototypes -Wmissing-prototypes -Wold-style-definition -Wextra
-Wwrite-strings (CFLAGS_FOR_TARGET) -c -o environ.lo `test -f
'runtime/environ.c' || echo '../../../gcc/libgfortran/'`runtime/environ.c
/bin/sh: syntax error at line 1 : `(' unexpected
gmake[2]: *** [environ.lo] Error 2

I don't know what to try or how to debug this one... This is a normal build
(languages: c and f95, multilib enabled (default)).

Going into the mips-sgi-irix6.5/libgfortran directory and then doing "gmake"
again works. But then up again in the build directory, it fails:

gmake[3]: *** [multi-do] Error 127
gmake[3]: Leaving directory
`/tmp/gfortran-20050616/ibin/mips-sgi-irix6.5/libgfortran'

I don't understand much about multilib, but it may be such a problem. If that
can help, the mips-sgi-irix6.5 directory now contains:

32/ (which contains libgfortran)
64/ (which contains libgfortran)
libgfortran/

-- 
           Summary: libgfortran build failure on mips-sgi-irix6.5
           Product: gcc
           Version: 4.1.0
            Status: UNCONFIRMED
          Keywords: build
          Severity: normal
          Priority: P2
         Component: libfortran
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: fxcoudert at gcc dot gnu dot org
                CC: gcc-bugs at gcc dot gnu dot org
  GCC host triplet: mips-sgi-irix6.5


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=22097

[Bug libfortran/22097] libgfortran build failure on mips-sgi-irix6.5

[fxcoudert at gcc dot gnu dot org]

------- Additional Comments From fxcoudert at gcc dot gnu dot org  2005-06-16 21:08 -------
Here is a missing info: what makes me think this is related to multilib is that
there are successful builds reported on gcc-testresults on this same host, with
multilib disabled.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=22097

[Bug libfortran/22097] libgfortran build failure on mips-sgi-irix6.5

[fxcoudert at gcc dot gnu dot org]

------- Additional Comments From fxcoudert at gcc dot gnu dot org  2005-06-16 21:32 -------
I tried building with "--disable-multilib" and same thing happens. Ideas, anyone?

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=22097

[Bug libfortran/22097] libgfortran build failure on mips-sgi-irix6.5

[fxcoudert at gcc dot gnu dot org]

------- Additional Comments From fxcoudert at gcc dot gnu dot org  2005-08-04 12:19 -------
This happens again with fresh CVS (20050804). Still no idea?

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=22097

[Bug libfortran/22097] libgfortran build failure on mips-sgi-irix6.5

[fxcoudert at gcc dot gnu dot org]

------- Additional Comments From fxcoudert at gcc dot gnu dot org  2005-09-24 12:56 -------
I don't have access to that mips-sgi-irix6.5 computer anymore, and since nobody
every confirmed this (and there are regular mips-sgi-irix6.5 successful reports
posted to gcc-testresults), this bug is closed.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|                            |INVALID


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=22097

[Bug libfortran/22097] libgfortran build failure on mips-sgi-irix6.5

[fxcoudert at gcc dot gnu dot org]

------- Comment #9 from fxcoudert at gcc dot gnu dot org  2005-11-10 15:35 -------
On mips-irix6.5, the libgfortran config.h contains:
#define HAVE_CABSF 1
and
/* #undef HAVE_COMPLEX_H */

and in math.h it does have the following:

struct __cabsl_s { long double a,b; };
extern long double cabsl( struct __cabsl_s );

while the correct definition of cabsl is provided in complex.h. I don't know
why on earth all this is happening, but I guess the root of the problem is
that:

$ cat a.c
#include <math.h>
#include <stdio.h>

int main (void)
{
  long double __complex__ z;

  printf ("%Lg\n", cabsl(z));
  return 0;
}
$ ./bin/gcc -std=c99 a.c
a.c: In function 'main':
a.c:8: error: incompatible type for argument 1 of 'cabsl'

while the cabsl man page clearly states that cabsl is to be found in <math.h>.
I'd like to have the opinion of mips-irix experts here: should the above code
compile fine? what is the right way of doing things here?


-- 

fxcoudert at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |r dot emrich at de dot
                   |                            |tecosim dot com, ro at gcc
                   |                            |dot gnu dot org
   Last reconfirmed|2005-10-25 20:09:15         |2005-11-10 15:35:05
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=22097

[Bug libfortran/22097] libgfortran build failure on mips-sgi-irix6.5

[pinskia at gcc dot gnu dot org]

-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|0                           |1
   Last reconfirmed|2005-10-23 13:15:20         |2005-10-25 20:09:15
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=22097

[Bug libfortran/22097] libgfortran build failure on mips-sgi-irix6.5

[fxcoudert at gcc dot gnu dot org]

------- Comment #8 from fxcoudert at gcc dot gnu dot org  2005-10-24 09:33 -------
(In reply to comment #7)
> I would guess a GNU make bug.  What version of GNU make are you using?  Try a
> newer release.

I had GNU Make version 3.78.1, which seems too old. 3.80 works fine. But...
later I have the following failures:

/tmp/debug/ibin/./gcc/xgcc -B/tmp/debug/ibin/./gcc/
-B/tmp/debug/irun/mips-sgi-irix6.5/bin/ -B/tmp/debug/irun/mips-sgi-irix6.5/lib/
-isystem /tmp/debug/irun/mips-sgi-irix6.5/include -isystem
/tmp/debug/irun/mips-sgi-irix6.5/sys-include -DHAVE_CONFIG_H -I.
-I../../../gcc/libgfortran -I. -iquote../../../gcc/libgfortran/io -std=gnu99
-Wall -Wstrict-prototypes -Wmissing-prototypes -Wold-style-definition -Wextra
-Wwrite-strings -O2 -g -O2 -c
../../../gcc/libgfortran/intrinsics/c99_functions.c   -DPIC -o
.libs/c99_functions.o
../../../gcc/libgfortran/intrinsics/c99_functions.c:582: warning: no previous
prototype for 'clog10f'
../../../gcc/libgfortran/intrinsics/c99_functions.c: In function 'clog10f':
../../../gcc/libgfortran/intrinsics/c99_functions.c:585: warning: implicit
declaration of function 'cabsf'
../../../gcc/libgfortran/intrinsics/c99_functions.c:585: warning: incompatible
implicit declaration of built-in function 'cabsf'
../../../gcc/libgfortran/intrinsics/c99_functions.c:585: warning: implicit
declaration of function 'cargf'
../../../gcc/libgfortran/intrinsics/c99_functions.c:585: warning: incompatible
implicit declaration of built-in function 'cargf'
../../../gcc/libgfortran/intrinsics/c99_functions.c: At top level:
../../../gcc/libgfortran/intrinsics/c99_functions.c:594: warning: no previous
prototype for 'clog10'
../../../gcc/libgfortran/intrinsics/c99_functions.c: In function 'clog10':
../../../gcc/libgfortran/intrinsics/c99_functions.c:597: warning: implicit
declaration of function 'cabs'
../../../gcc/libgfortran/intrinsics/c99_functions.c:597: warning: incompatible
implicit declaration of built-in function 'cabs'
../../../gcc/libgfortran/intrinsics/c99_functions.c:597: warning: implicit
declaration of function 'carg'
../../../gcc/libgfortran/intrinsics/c99_functions.c:597: warning: incompatible
implicit declaration of built-in function 'carg'
../../../gcc/libgfortran/intrinsics/c99_functions.c: At top level:
../../../gcc/libgfortran/intrinsics/c99_functions.c:606: warning: no previous
prototype for 'clog10l'
../../../gcc/libgfortran/intrinsics/c99_functions.c: In function 'clog10l':
../../../gcc/libgfortran/intrinsics/c99_functions.c:609: error: incompatible
type for argument 1 of 'cabsl'
../../../gcc/libgfortran/intrinsics/c99_functions.c:609: warning: implicit
declaration of function 'cargl'
../../../gcc/libgfortran/intrinsics/c99_functions.c:609: warning: incompatible
implicit declaration of built-in function 'cargl'

I'm investigating. For now, all I can say is I *hate* C99.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=22097

[Bug libfortran/22097] libgfortran build failure on mips-sgi-irix6.5

[wilson at gcc dot gnu dot org]

------- Comment #7 from wilson at gcc dot gnu dot org  2005-10-23 21:27 -------
I would guess a GNU make bug.  What version of GNU make are you using?  Try a
newer release.

Alternatively, this could be a shell bug.  What shell and version are you
using?  Try using a different shell and/or a more recent shell version. 
Preferably a recent version of bash.

There should be a "gmake all-am" line right before the failure.  The
libgfortran all rule does a recursive make, and it appears that the value of
CC, which is $(CC_FOR_TARGET) is getting clobbered somehow by this recursive
make invocation.  It was OK before this recursive make, because there are some
rules invoked before the recursive make, and they didn't fail for you.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=22097

[Bug libfortran/22097] libgfortran build failure on mips-sgi-irix6.5

[fxcoudert at gcc dot gnu dot org]

------- Comment #6 from fxcoudert at gcc dot gnu dot org  2005-10-23 13:18 -------
Hum... confirming this bug doesn't reopen it, so here it is.


-- 

fxcoudert at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |UNCONFIRMED
  GCC build triplet|                            |mips-sgi-irix6.5
 GCC target triplet|                            |mips-sgi-irix6.5
         Resolution|INVALID                     |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=22097

[Bug libfortran/22097] libgfortran build failure on mips-sgi-irix6.5

[fxcoudert at gcc dot gnu dot org]

------- Comment #5 from fxcoudert at gcc dot gnu dot org  2005-10-23 13:15 -------
Well, looks like I'm going to continue this long monologue. I got my access to
a mips-irix6.5 back, and it's still failling:

$ ../gcc/configure --prefix=/tmp/debug/irun --enable-languages=c,fortran
--with-gmp=$HOME/gfortran/gfortran_libs/mips-irix
--with-mpfr=$HOME/gfortran/gfortran_libs/mips-irix
--with-as=$HOME/gfortran/gfortran_libs/mips-irix/bin/as
--with-ld=$HOME/gfortran/gfortran_libs/mips-irix/bin/ld --disable-libssp
--disable-libmudflap

[configure works fine]

$ gmake
[... works fine until ...]
gmake[3]: Entering directory `/tmp/debug/ibin/mips-sgi-irix6.5/libgfortran'
/bin/sh ./libtool --mode=compile (CC_FOR_TARGET) -DHAVE_CONFIG_H -I.
-I../../../gcc/libgfortran -I.  -iquote../../../gcc/libgfortran/io  -std=gnu99
-Wall -Wstrict-prototypes -Wmissing-prototypes -Wold-style-definition -Wextra
-Wwrite-strings (CFLAGS_FOR_TARGET) -c -o compile_options.lo `test -f
'runtime/compile_options.c' || echo
'../../../gcc/libgfortran/'`runtime/compile_options.c
/bin/sh: syntax error at line 1 : `(' unexpected

I have no idea what is going on here.


-- 

fxcoudert at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|0000-00-00 00:00:00         |2005-10-23 13:15:20
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=22097