Re: Still hard, was: g77 Build Problems

Re: Still hard, was: g77 Build Problems

[Toon Moene]

Henry Hotz wrote:

> setenv CFLAGS -O0
> setenv BOOT_CFLAGS -g
> ../gcc/configure --enable-languages="c f77"

> The problem I have now is that there is no f77 compiler there.  
> There are no files in the object tree with either a '7' or a '2c' 
> in their names at all.

I ran into this problem recently, too.

The culprit is you configure line.  It should read:

../gcc/configure --enable-languages="c,f77"

Note the comma !

Hope this helps,

--
Toon Moene, KNMI, PO Box 201, 3730 AE De Bilt, The Netherlands.
Tel. +31302206443, Fax +31302210407,  e-mail moene@knmi.nl
URL: http://www.knmi.nl/hirlam

Re: Still hard, was: g77 Build Problems

[Henry B. Hotz]

Success!!!  Or at least partial success.  Thanks to everyone for their help.

cvs today's version of gcc source 5-9-01
mkdir&cd obj
setenv CFLAGS -O0		# Maybe not needed
setenv BOOT_CFLAGS "-g -O0"	# Should be fixed in a couple of months
../gcc/configure --enable-languages="c,f77"
limit stack 800
make all-libiberty
make bootstrap
su
make install
exit

I can now compile a "Hello world." program in Fortran!!!!

I cannot do a make check however so I don't have a good feel for how 
robust the g77 compiler is yet.  Separate posting to 
darwin-development on that subject.  We'll be trying it soon.

Signature held pending an ISO 9000 compliant
signature design and approval process.
h.b.hotz@jpl.nasa.gov, or hbhotz@oxy.edu

Re: Still hard, was: g77 Build Problems

[Andrew W. Reynolds]

Henry,

If you did not specify a new "prefix" then the files wre installed in 
/usr/local/bin and /usr/local/lib. To over-ride the version /usr/bin/cc 
you should ensure /usr/local/bin precedes /usr/bin in your PATH 
variable. Also, there is not gcc installed already so using gcc instead 
of cc will help.

Hope this helps.

Drew

On Tuesday, May 8, 2001, at 08:21 PM, Henry B. Hotz wrote:

Ok, I did
setenv CFLAGS -O0
setenv BOOT_CFLAGS -g
../gcc/configure --enable-languages="c f77"
limit stack 800
make all-libiberty
make bootstrap
make check

The bootstrap target appeared to pass the stage2/3 compare properly. The 
check target appeared to do nothing useful?  I did download/install 
dejagnu 1.4.  Tcl is already there.

The problem I have now is that there is no f77 compiler there.  There 
are no files in the object tree with either a '7' or a '2c' in their 
names at all.

I also did a make install and no f77 or g77 executable is installed. A 
gcc of a fortran hello world program said
[localhost:~] hotz% gcc test.f
gcc: test.f: Fortran compiler not installed on this system
gcc: installation problem, cannot exec `#Fortrancollect2': No such file 
or directory


I guess that's pretty conclusive.  Now is there some quoting convention 
for the enable-languages configure option that I need to figure out?  
Reinspecting the configure docs on the web site I see they are using 
commas instead of spaces to separate languages.  Not what I remember 
seeing, but worth a try if nobody tells me differently.

Re: Still hard, was: g77 Build Problems

[Henry B. Hotz]

>At 8:23 PM +0200 5/7/01, Martin Costabel wrote:
>>Here is the next one after the one you mentioned:
>
>In other words I don't have a prayer unless I play games with 
>optimization levels.  OK, I'm trying -O0 as I write.  Thanks.

Ok, I did
setenv CFLAGS -O0
setenv BOOT_CFLAGS -g
../gcc/configure --enable-languages="c f77"
limit stack 800
make all-libiberty
make bootstrap
make check

The bootstrap target appeared to pass the stage2/3 compare properly. 
The check target appeared to do nothing useful?  I did 
download/install dejagnu 1.4.  Tcl is already there.

The problem I have now is that there is no f77 compiler there.  There 
are no files in the object tree with either a '7' or a '2c' in their 
names at all.

I also did a make install and no f77 or g77 executable is installed. 
A gcc of a fortran hello world program said
[localhost:~] hotz% gcc test.f
gcc: test.f: Fortran compiler not installed on this system
gcc: installation problem, cannot exec `#Fortrancollect2': No such 
file or directory


I guess that's pretty conclusive.  Now is there some quoting 
convention for the enable-languages configure option that I need to 
figure out?  Reinspecting the configure docs on the web site I see 
they are using commas instead of spaces to separate languages.  Not 
what I remember seeing, but worth a try if nobody tells me 
differently.


Signature held pending an ISO 9000 compliant
signature design and approval process.
h.b.hotz@jpl.nasa.gov, or hbhotz@oxy.edu

Re: Still hard, was: g77 Build Problems

[Henry B. Hotz]

At 8:23 PM +0200 5/7/01, Martin Costabel wrote:
>Here is the next one after the one you mentioned:

In other words I don't have a prayer unless I play games with 
optimization levels.  OK, I'm trying -O0 as I write.  Thanks.

For the record there are some notes about building with 2.95 at 
< http://www-jlc.kek.jp/~fujiik/macosx/memo/HEPonX.html > which I may 
try.  That's a pretty comprehensive page.  I wish it covered MacOS X 
release, instead of beta, but it references most of the general 
system stuff I'm interested in and MPI as well.  No Postgres or AFS, 
but I've already gotten them working.


Signature held pending an ISO 9000 compliant
signature design and approval process.
h.b.hotz@jpl.nasa.gov, or hbhotz@oxy.edu

Re: Still hard, was: g77 Build Problems

[Martin Costabel]

Here is the next one after the one you mentioned:

../../Sources/gnu-cvs-gcc/gcc/reload1.c: In function `emit_reload_insns':
../../Sources/gnu-cvs-gcc/gnu-cvs-gcc/gcc/reload1.c:7310: Insn does not
satisfy its constraints:
(insn 4696 4695 4697 (set:SI (reg/u:SI 0 r0)
        (plus:SI (reg:SI 31 r31)
            (high:SI (const:SI (minus:SI (symbol_ref:SI ("&L_rld$non_lazy_ptr"))
                        (symbol_ref:SI ("*L56$pb"))))))) 37
{addsi3_high} (nil)
    (nil))
../../Sources/gnu-cvs-gcc/gcc/reload1.c:7310: Internal compiler error in
reload_cse_simplify_operands, at reload1.c:8343

You will come across this or similar ones either during "make bootstrap"
or, depending on the age of your sources, after you succeeded to make a
g77 executable, when you try to compile FORTRAN programs. Sometimes it
happens only with "-O0" and sometimes with "-O2", depending on the sources.

BTW, all these errors have been reported already on the
darwin-development list 3 weeks ago. And as long as there is no darwin
backport to the gcc-3.0 branch, we will probably be able to play this
game for another couple of months :-) Beats Bugdom hands down ;-)

--
Martin

"Henry B. Hotz" wrote:
> 
> With respect and thanks to the suggestions regarding optimization,
> the response from Stan Shebs seemed to be most relevant:
> 
> At 5:35 PM -0700 5/4/01, Stan Shebs wrote:
> >Anyway, this is a known failure point - needs more than the default
> >512K of stack space.  Do "limit stack 800" before building (in .cshrc
> >if you like).
> 
> Now I still have a problem:
> 
> stage1/xgcc -Bstage1/ -B/usr/local/powerpc-apple-darwin1.3.2/bin/ -c
> -DIN_GCC    -g -O2 -W -Wall -Wwrite-strings -Wstrict-prototypes
> -Wmissing-prototypes -Wtraditional -pedantic -Wno-long-long
> -DHAVE_CONFIG_H -DGENERATOR_FILE    -I. -I. -I../../gcc/gcc
> -I../../gcc/gcc/. -I../../gcc/gcc/config -I../../gcc/gcc/../include
> ../../gcc/gcc/hash.c -o hash.o
> ../../gcc/gcc/hash.c: In function `hash_lookup':
> ../../gcc/gcc/hash.c:93: warning: promoted argument `create' doesn't
> match prototype
> ../../gcc/gcc/hash.h:100: warning: prototype declaration
> stage1/xgcc -Bstage1/ -B/usr/local/powerpc-apple-darwin1.3.2/bin/ -c
> -DIN_GCC    -g -O2 -W -Wall -Wwrite-strings -Wstrict-prototypes
> -Wmissing-prototypes -Wtraditional -pedantic -Wno-long-long
> -DHAVE_CONFIG_H    -I. -I. -I../../gcc/gcc -I../../gcc/gcc/.
> -I../../gcc/gcc/config -I../../gcc/gcc/../include
> ../../gcc/gcc/ifcvt.c -o ifcvt.o
> stage1/xgcc -Bstage1/ -B/usr/local/powerpc-apple-darwin1.3.2/bin/ -c
> -DIN_GCC    -g -O2 -W -Wall -Wwrite-strings -Wstrict-prototypes
> -Wmissing-prototypes -Wtraditional -pedantic -Wno-long-long
> -DHAVE_CONFIG_H -DGENERATOR_FILE    -I. -I. -I../../gcc/gcc
> -I../../gcc/gcc/. -I../../gcc/gcc/config -I../../gcc/gcc/../include
> ../../gcc/gcc/genattrtab.c
> stage1/xgcc -Bstage1/ -B/usr/local/powerpc-apple-darwin1.3.2/bin/
> -DIN_GCC    -g -O2 -W -Wall -Wwrite-strings -Wstrict-prototypes
> -Wmissing-prototypes -Wtraditional -pedantic -Wno-long-long
> -DHAVE_CONFIG_H -DGENERATOR_FILE  -o genattrtab \
>   genattrtab.o rtl.o bitmap.o ggc-none.o gensupport.o hashtab.o
> safe-ctype.o print-rtl.o errors.o ` case "obstack.o" in ?*) echo
> obstack.o ;; esac ` ` case "" in ?*) echo  ;; esac ` ` case "" in ?*)
> echo  ;; esac ` ` case "" in ?*) echo  ;; esac `
> ./genattrtab ../../gcc/gcc/config/rs6000/rs6000.md > tmp-attrtab.c
> make[1]: *** [s-attrtab] Segmentation fault
> make: *** [stage2_build] Error 2

Re: Still hard, was: g77 Build Problems

[Henry B. Hotz]

With respect and thanks to the suggestions regarding optimization, 
the response from Stan Shebs seemed to be most relevant:

At 5:35 PM -0700 5/4/01, Stan Shebs wrote:
>Anyway, this is a known failure point - needs more than the default
>512K of stack space.  Do "limit stack 800" before building (in .cshrc
>if you like).

Now I still have a problem:

stage1/xgcc -Bstage1/ -B/usr/local/powerpc-apple-darwin1.3.2/bin/ -c 
-DIN_GCC    -g -O2 -W -Wall -Wwrite-strings -Wstrict-prototypes 
-Wmissing-prototypes -Wtraditional -pedantic -Wno-long-long 
-DHAVE_CONFIG_H -DGENERATOR_FILE    -I. -I. -I../../gcc/gcc 
-I../../gcc/gcc/. -I../../gcc/gcc/config -I../../gcc/gcc/../include 
../../gcc/gcc/hash.c -o hash.o
../../gcc/gcc/hash.c: In function `hash_lookup':
../../gcc/gcc/hash.c:93: warning: promoted argument `create' doesn't 
match prototype
../../gcc/gcc/hash.h:100: warning: prototype declaration
stage1/xgcc -Bstage1/ -B/usr/local/powerpc-apple-darwin1.3.2/bin/ -c 
-DIN_GCC    -g -O2 -W -Wall -Wwrite-strings -Wstrict-prototypes 
-Wmissing-prototypes -Wtraditional -pedantic -Wno-long-long 
-DHAVE_CONFIG_H    -I. -I. -I../../gcc/gcc -I../../gcc/gcc/. 
-I../../gcc/gcc/config -I../../gcc/gcc/../include 
../../gcc/gcc/ifcvt.c -o ifcvt.o
stage1/xgcc -Bstage1/ -B/usr/local/powerpc-apple-darwin1.3.2/bin/ -c 
-DIN_GCC    -g -O2 -W -Wall -Wwrite-strings -Wstrict-prototypes 
-Wmissing-prototypes -Wtraditional -pedantic -Wno-long-long 
-DHAVE_CONFIG_H -DGENERATOR_FILE    -I. -I. -I../../gcc/gcc 
-I../../gcc/gcc/. -I../../gcc/gcc/config -I../../gcc/gcc/../include 
../../gcc/gcc/genattrtab.c
stage1/xgcc -Bstage1/ -B/usr/local/powerpc-apple-darwin1.3.2/bin/ 
-DIN_GCC    -g -O2 -W -Wall -Wwrite-strings -Wstrict-prototypes 
-Wmissing-prototypes -Wtraditional -pedantic -Wno-long-long 
-DHAVE_CONFIG_H -DGENERATOR_FILE  -o genattrtab \
  genattrtab.o rtl.o bitmap.o ggc-none.o gensupport.o hashtab.o 
safe-ctype.o print-rtl.o errors.o ` case "obstack.o" in ?*) echo 
obstack.o ;; esac ` ` case "" in ?*) echo  ;; esac ` ` case "" in ?*) 
echo  ;; esac ` ` case "" in ?*) echo  ;; esac `
./genattrtab ../../gcc/gcc/config/rs6000/rs6000.md > tmp-attrtab.c
make[1]: *** [s-attrtab] Segmentation fault
make: *** [stage2_build] Error 2

Unless someone has a better suggestion, I think I'll try cvs'ing a 
newer copy of the source (mine is a week old now, must be suffering 
from bit-rot by now, mustent it ;-?) and maybe try The -O0 thing too.


Signature held pending an ISO 9000 compliant
signature design and approval process.
h.b.hotz@jpl.nasa.gov, or hbhotz@oxy.edu

Re: Still hard, was: g77 Build Problems

[Phil Edwards]

On Fri, May 04, 2001 at 09:09:24PM -0700, David A. Gatwood wrote:
> > Did I goof something up?  Next suggestion?
> 
> Try turning off -O2 in the CFLAGS, or try -O0.  Worth a shot.  Used to
> help in the early ppc days, anyway.

There was a patch posted by someone last month on gcc-patches ("RFA/Patch:
gcc.c retry") which got no responses.  It introduces a flag for gcc.c
(intended mostly for bootstrapping, I think, but could be wrong) which
says, "if we used -On, n>0, and the compiler failed, try again with -O0
automatically."

-- 
pedwards at disaster dot jaj dot com  |  pme at sources dot redhat dot com
devphil at several other less interesting addresses in various dot domains
The gods do not protect fools.  Fools are protected by more capable fools.

Re: Still hard, was: g77 Build Problems

[David A. Gatwood]

On Fri, 4 May 2001, Henry B. Hotz wrote:

> stage1/xgcc -Bstage1/ -B/usr/local/powerpc-apple-darwin1.3.2/bin/ -c 
> -DIN_GCC    -g -O2 -W -Wall -Wwrite-strings -Wstrict-prototypes 
> -Wmissing-prototypes -Wtraditional -pedantic -Wno-long-long 
> -DHAVE_CONFIG_H    -I. -I. -I../../gcc/gcc -I../../gcc/gcc/. 
> -I../../gcc/gcc/config -I../../gcc/gcc/../include 
> ../../gcc/gcc/expr.c -o expr.o
> ../../gcc/gcc/expr.c: In function `move_by_pieces_1':
> ../../gcc/gcc/expr.c:1564: warning: `to1' might be used uninitialized 
> in this function
> ../../gcc/gcc/expr.c: In function `emit_single_push_insn':
> ../../gcc/gcc/expr.c:3129: warning: unused parameter `mode'
> ../../gcc/gcc/expr.c:3128: warning: unused parameter `x'
> ../../gcc/gcc/expr.c:3130: warning: unused parameter `type'
> ../../gcc/gcc/expr.c: In function `store_constructor':
> ../../gcc/gcc/expr.c:4685: warning: `minelt' might be used 
> uninitialized in this function
> ../../gcc/gcc/expr.c:4686: warning: `maxelt' might be used 
> uninitialized in this function
> ../../gcc/gcc/expr.c: In function `store_field':
> ../../gcc/gcc/expr.c:5215: warning: comparison between signed and unsigned
> ../../gcc/gcc/expr.c: In function `expand_expr':
> ../../gcc/gcc/expr.c:7168: warning: comparison between signed and unsigned
> xgcc: Internal error: Illegal instruction (program cc1)
> Please submit a full bug report.
> See <URL: http://www.gnu.org/software/gcc/bugs.html > for instructions.
> make[1]: *** [expr.o] Error 1
> make: *** [stage2_build] Error 2
> [localhost:gcc-cvs/obj/gcc] hotz%
> 
> Did I goof something up?  Next suggestion?

Try turning off -O2 in the CFLAGS, or try -O0.  Worth a shot.  Used to
help in the early ppc days, anyway.


david

---------------------------------------------------------------------
                    Check out my weekly web comic:
                     http://www.techmagazine.org

Still hard, was: g77 Build Problems

[Henry B. Hotz]

Got the as at http://dryden.biol.yorku.ca/darwin.html and put it in 
/usr/libexec...

Already had gcc from a few days ago.

../gcc/configure --enable-languages="c f77"

make all-libiberty
cd gcc
make bootstrap

...

stage1/xgcc -Bstage1/ -B/usr/local/powerpc-apple-darwin1.3.2/bin/ -c 
-DIN_GCC    -g -O2 -W -Wall -Wwrite-strings -Wstrict-prototypes 
-Wmissing-prototypes -Wtraditional -pedantic -Wno-long-long 
-DHAVE_CONFIG_H    -I. -I. -I../../gcc/gcc -I../../gcc/gcc/. 
-I../../gcc/gcc/config -I../../gcc/gcc/../include 
../../gcc/gcc/expmed.c -o expmed.o
../../gcc/gcc/expmed.c: In function `expand_shift':
../../gcc/gcc/expmed.c:2007: warning: comparison between signed and unsigned
stage1/xgcc -Bstage1/ -B/usr/local/powerpc-apple-darwin1.3.2/bin/ -c 
-DIN_GCC    -g -O2 -W -Wall -Wwrite-strings -Wstrict-prototypes 
-Wmissing-prototypes -Wtraditional -pedantic -Wno-long-long 
-DHAVE_CONFIG_H    -I. -I. -I../../gcc/gcc -I../../gcc/gcc/. 
-I../../gcc/gcc/config -I../../gcc/gcc/../include 
../../gcc/gcc/expr.c -o expr.o
../../gcc/gcc/expr.c: In function `move_by_pieces_1':
../../gcc/gcc/expr.c:1564: warning: `to1' might be used uninitialized 
in this function
../../gcc/gcc/expr.c: In function `emit_single_push_insn':
../../gcc/gcc/expr.c:3129: warning: unused parameter `mode'
../../gcc/gcc/expr.c:3128: warning: unused parameter `x'
../../gcc/gcc/expr.c:3130: warning: unused parameter `type'
../../gcc/gcc/expr.c: In function `store_constructor':
../../gcc/gcc/expr.c:4685: warning: `minelt' might be used 
uninitialized in this function
../../gcc/gcc/expr.c:4686: warning: `maxelt' might be used 
uninitialized in this function
../../gcc/gcc/expr.c: In function `store_field':
../../gcc/gcc/expr.c:5215: warning: comparison between signed and unsigned
../../gcc/gcc/expr.c: In function `expand_expr':
../../gcc/gcc/expr.c:7168: warning: comparison between signed and unsigned
xgcc: Internal error: Illegal instruction (program cc1)
Please submit a full bug report.
See <URL: http://www.gnu.org/software/gcc/bugs.html > for instructions.
make[1]: *** [expr.o] Error 1
make: *** [stage2_build] Error 2
[localhost:gcc-cvs/obj/gcc] hotz%

Did I goof something up?  Next suggestion?

Signature held pending an ISO 9000 compliant
signature design and approval process.
h.b.hotz@jpl.nasa.gov, or hbhotz@oxy.edu