[3.3 branch] bootstrap failures on {arm,mips,mipsel,m68k}-linux

[3.3 branch] bootstrap failures on {arm,mips,mipsel,m68k}-linux

[Matthias Klose]

snapshots built form HEAD and then the 3.3 branch fail for the
following architectures (on which 3.2.x builds fine).
all builds failed from the beginning when the snapshots were built
from the HEAD/3.3 branch.

arm: http://buildd.debian.org/build.php?arch=arm&pkg=gcc-snapshot

- fails with ICE for CVS 20020714
- starting with CVS 20020809 the make compare fails.
- (the results prior to 20020701 are from the gcc-3_1-branch).

mips: http://buildd.debian.org/build.php?arch=mips&pkg=gcc-snapshot

- latest try to build from CVS 20020907
gcc-3.2  -c   -g -DIN_GCC   -W -Wall -Wwrite-strings
-Wstrict-prototypes -Wmissing-prototypes -Wtraditional -pedantic
-Wno-long-long -fno-common  -DHAVE_CONFIG_H    -I. -I. -I../../src/gcc
-I../../src/gcc/. -I../../src/gcc/config -I../../src/gcc/../include
../../src/gcc/expr.c -o expr.o
/tmp/ccs4dU7E.s: Assembler messages:
/tmp/ccs4dU7E.s:28502: Error: Branch out of range
[...]
/tmp/ccs4dU7E.s:33314: Error: Branch out of range
make[4]: *** [expr.o] Error 1

mipsel: http://buildd.debian.org/build.php?arch=mipsel&pkg=gcc-snapshot

- latest try to build from CVS 20021201
  same error as mips


m68k: http://buildd.debian.org/build.php?arch=m68k&pkg=gcc-snapshot

- build from 20021201:
stage1/xgcc -Bstage1/ -B/usr/lib/gcc-snapshot/m68k-linux/bin/ -c   -O2
 -DIN_GCC   -W -Wall -Wwrite-strings -Wstrict-prototypes
-Wmissing-prototypes -Wtraditional -pedantic -Wno-long-long
-fno-common  -DHAVE_CONFIG_H -DGENERATOR_FILE
-I. -I. -I../../src/gcc -I../../src/gcc/. -I../../src/gcc/config
-I../../src/gcc/../include \
  ../../src/gcc/gengtype-yacc.c -o gengtype-yacc.o
gengtype-yacc.c: In function `yydestruct':
gengtype-yacc.c:725: warning: traditional C rejects ISO C style
function definitions
stage1/xgcc -Bstage1/ -B/usr/lib/gcc-snapshot/m68k-linux/bin/   -O2
-DIN_GCC   -W -Wall -Wwrite-strings -Wstrict-prototypes
-Wmissing-prototypes -Wtraditional -pedantic -Wno-long-long
-fno-common  -DHAVE_CONFIG_H -DGENERATOR_FILE  -o gengtype \
 gengtype.o gengtype-lex.o gengtype-yacc.o ../libiberty/libiberty.a
./gengtype
make[4]: *** [s-gtype] Segmentation fault


- build from 20020809:
./xgcc -B./ -B/usr/lib/gcc-snapshot/m68k-linux/bin/ -isystem
/usr/lib/gcc-snapshot/m68k-linux/include -isystem
/usr/lib/gcc-snapshot/m68k-linux/sys-include -O2  -DIN_GCC     -W
-Wall -Wwrite-strings -Wstrict-prototypes -Wmissing-prototypes
-isystem ./include  -fPIC -g -DHAVE_GTHR_DEFAULT -DIN_LIBGCC2
-D__GCC_FLOAT_NOT_NEEDED  -I. -I. -I../../src/gcc
-I../../src/gcc/. -I../../src/gcc/config -I../../src/gcc/../include
-fexceptions -c ../../src/gcc/unwind-dw2-fde-glibc.c -o
libgcc/./unwind-dw2-fde-glibc.o
../../src/gcc/unwind-dw2-fde.c: In function `init_object':
../../src/gcc/unwind-dw2-fde.c:924: internal compiler error: Internal
compiler error in reg_overlap_mentioned_p, at rtlanal.c:1543
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://www.gnu.org/software/gcc/bugs.html> for instructions.
make[5]: *** [libgcc/./unwind-dw2-fde-glibc.o] Error 1

Re: [3.3 branch] bootstrap failures on {arm,mips,mipsel,m68k}-linux

[Alexandre Oliva]

On Dec 28, 2002, Matthias Klose <doko@cs.tu-berlin.de> wrote:

> - latest try to build from CVS 20020907

> /tmp/ccs4dU7E.s: Assembler messages:
> /tmp/ccs4dU7E.s:28502: Error: Branch out of range
> [...]
> /tmp/ccs4dU7E.s:33314: Error: Branch out of range

This is a bug in the bootstrap compiler.  There's nothing GCC 3.3 can
do about it.  The bug is won't occur if you use GCC 3.3 as the
bootstrap compiler, since it's already fixed.  You can build it
without bootstrapping and then use that for a bootstrap, or build
stage1 with optimization (STAGE1_CFLAGS='-O -g'), or use a CVS version
of binutils that supports branch relaxation in the assembler and
enable that by adding -Wa,--relax-branch to STAGE1_CFLAGS.

-- 
Alexandre Oliva   Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer                 aoliva@{redhat.com, gcc.gnu.org}
CS PhD student at IC-Unicamp        oliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist                Professional serial bug killer

Re: [3.3 branch] bootstrap failures on {arm,mips,mipsel,m68k}-linux

[Matthias Klose]

Alexandre Oliva writes:
> On Dec 28, 2002, Matthias Klose <doko@cs.tu-berlin.de> wrote:
> 
> > - latest try to build from CVS 20020907
> 
> > /tmp/ccs4dU7E.s: Assembler messages:
> > /tmp/ccs4dU7E.s:28502: Error: Branch out of range
> > [...]
> > /tmp/ccs4dU7E.s:33314: Error: Branch out of range
> 
> This is a bug in the bootstrap compiler.  There's nothing GCC 3.3 can
> do about it.  The bug is won't occur if you use GCC 3.3 as the
> bootstrap compiler, since it's already fixed.  You can build it
> without bootstrapping and then use that for a bootstrap, or build
> stage1 with optimization (STAGE1_CFLAGS='-O -g'), or use a CVS version
> of binutils that supports branch relaxation in the assembler and
> enable that by adding -Wa,--relax-branch to STAGE1_CFLAGS.

thanks. now it fails building gnat.

what should be done with bootstap failures on architectures, where
gcc-3.2.x does bootstrap? File bug reports with priority high?
same for Ada bootstrap failures, where gnat-3.2 succeeds?

	Matthias

Re: [3.3 branch] bootstrap failures on {arm,mips,mipsel,m68k}-linux

[Alexandre Oliva]

On Dec 29, 2002, Matthias Klose <doko@cs.tu-berlin.de> wrote:

> what should be done with bootstap failures on architectures, where
> gcc-3.2.x does bootstrap? File bug reports with priority high?
> same for Ada bootstrap failures, where gnat-3.2 succeeds?

That's a regression too, right?  Filing gnats bug reports with high
priority is the right way to get them accounted for, yes.

-- 
Alexandre Oliva   Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer                 aoliva@{redhat.com, gcc.gnu.org}
CS PhD student at IC-Unicamp        oliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist                Professional serial bug killer