[Bug bootstrap/18972] New: Regression: bootstrap failure of gcc-4.0-20041212 on OpenDarwin 7.2.1/x86 (i686-apple-darwin7.2.1): Bootstrap comparison failure

[Bug bootstrap/18972] New: Regression: bootstrap failure of gcc-4.0-20041212 on OpenDarwin 7.2.1/x86 (i686-apple-darwin7.2.1): Bootstrap comparison failure

[lars dot sonchocky-helldorf at hamburg dot de]

The system:
===========

uname -a
Darwin localhost 7.2.1 Darwin Kernel Version 7.2.1: Wed Jul 14 03:00:02 
PDT 2004; root:tmp/xnu-7.2.1-1-root.obj/RELEASE_I386  x86 i386

gcc -v
Reading specs from /usr/libexec/gcc/darwin/i386/3.3/specs
Thread model: posix
gcc version 3.3 20030304 (Apple Computer, Inc. build 1495)


Prerequisites:
==============

# link libmx to libSystem

# this is necessary because some libraries are missing 
# from OpenDarwin, in this case the -lmx argument results in:
# /usr/bin/ld: can't locate file for: -lmx
# see http://www.opendarwin.org/bugzilla/show_bug.cgi?id=2194
#
# note: this is just enough to bootstrap the compiler, 
# some floating point compiliations might fail

sudo ln -s /usr/lib/libSystem.B.dylib /usr/lib/libmx.dylib

# install new cctools

# according to: http://gcc.gnu.org/ml/gcc-patches/2004-08/msg00104.html
# and: http://gcc.gnu.org/ml/gcc-patches/2004-10/msg01222.html
# newer cctools are needed to build HEAD of gcc
# Since building Apples cctools is somewhat complicated and
# won't work on (Open)Darwin/x86 better use OpenDarwin cctools (odcctools)
# see: http://www.opendarwin.org/projects/odcctools/
# and: http://www.opendarwin.org/projects/odcctools/usingodcctools.html 
# and: 
http://www.opendarwin.org/pipermail/odcctools/2004-October/000033.html

ftp http://www.opendarwin.org/downloads/odcctools-20041018.tar.bz2
tar -jxf odcctools-20041018.tar.bz2
mkdir odcctools-build
cd odcctools-build/
../odcctools-20041018/configure --prefix=/usr
make
sudo make install
cd ..


Bootstrapped like:
==================

# bootstrap the compiler

# note: those STAGE1_CFLAGS are needed to avoid
# http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14780

ftp 
ftp://gcc.gnu.org/pub/gcc/snapshots/4.0-20041212/gcc-4.0-20041212.tar.bz2
openssl md5 gcc-4.0-20041212.tar.bz2
tar -jxf gcc-4.0-20041212.tar.bz2
cd gcc-build/
rm -rf *
../gcc-4.0-20041212/configure --prefix=/tmp/gcc
make bootstrap STAGE1_CFLAGS="-no-cpp-precomp -Wa,-force_cpusubtype_ALL"



Result of attempted bootstrap:
==============================

Comparing stage2 and stage3 of the compiler
rm -f .bad_compare
case "gnucompare" in *compare | *compare-lean ) stage=2 ;; * ) stage=`echo 
gnucompare | sed -e 's,^[a-z]*compare\([0-9][0-9]*\).*,\1,'` ;; esac; \
for dir in . cp java objc build; do \
  if [ "`echo $dir/*.o`" != "$dir/*.o" ] ; then \
    for file in $dir/*.o; do \
      case "gnucompare" in \
        slowcompare* ) \
          tail +16c ./$file > tmp-foo1; \
          tail +16c stage$stage/$file > tmp-foo2 \
            && (cmp tmp-foo1 tmp-foo2 > /dev/null 2>&1 || echo $file 
differs >> .bad_compare) || true; \
          ;; \
        fastcompare* ) \
          cmp $file stage$stage/$file 16 16 > /dev/null 2>&1; \
          test $? -eq 1 && echo $file differs >> .bad_compare || true; \
          ;; \
        gnucompare* ) \
          cmp --ignore-initial=16 $file stage$stage/$file > /dev/null 
2>&1; \
          test $? -eq 1 && echo $file differs >> .bad_compare || true; \
          ;; \
      esac ; \
    done; \
  else true; fi; \
done
rm -f tmp-foo*
case "gnucompare" in *compare | *compare-lean ) stage=2 ;; * ) stage=`echo 
gnucompare | sed -e 's,^[a-z]*compare\([0-9][0-9]*\).*,\1,'` ;; esac; \
if [ -f .bad_compare ]; then \
  echo "Bootstrap comparison failure!"; \
  cat .bad_compare; \
  exit 1; \
else \
  case "gnucompare" in \
    *-lean ) rm -rf stage$stage ;; \
    *) ;; \
  esac; true; \
fi
Bootstrap comparison failure!
./reload.o differs
make[1]: *** [gnucompare] Error 1
make: *** [bootstrap] Error 2




This is a regression since bootstrap worked for 4.0.0 20041205. 
See: http://gcc.gnu.org/ml/gcc-testresults/2004-12/msg00276.html


Steps taken to isolate the bug
==============================

the only occurrences of reload.o during bootstrapping were:

gcc -c   -no-cpp-precomp -Wa,-force_cpusubtype_ALL -DENABLE_CHECKING -
DENABLE_ASSERT_CHECKING -DIN_GCC   -W -Wall -Wwrite-strings -Wstrict-prototypes -Wmissing-
prototypes  -fno-common   -DHAVE_CONFIG_H    -I. -I. -I../../gcc-4.0-20041212/gcc -I../../gcc-4.0-
20041212/gcc/. -I../../gcc-4.0-20041212/gcc/../include -I./../intl -I../../gcc-4.0-20041212/gcc/../
libcpp/include  ../../gcc-4.0-20041212/gcc/reload.c -o reload.o

stage1/xgcc -Bstage1/ -B/tmp/gcc/i686-apple-darwin7.2.1/bin/ -c   -O2 -g -fomit-frame-pointer -
DIN_GCC   -W -Wall -Wwrite-strings -Wstrict-prototypes -Wmissing-prototypes -pedantic -Wno-long-
long -Wno-variadic-macros -Wold-style-definition -Werror -fno-common   -DHAVE_CONFIG_H    -I. -
I. -I../../gcc-4.0-20041212/gcc -I../../gcc-4.0-20041212/gcc/. -I../../gcc-4.0-20041212/gcc/../
include -I./../intl -I../../gcc-4.0-20041212/gcc/../libcpp/include  ../../gcc-4.0-20041212/gcc/
reload.c -o reload.o

stage2/xgcc -Bstage2/ -B/tmp/gcc/i686-apple-darwin7.2.1/bin/ -c   -O2 -g -fomit-frame-pointer -
DIN_GCC   -W -Wall -Wwrite-strings -Wstrict-prototypes -Wmissing-prototypes -pedantic -Wno-long-
long -Wno-variadic-macros -Wold-style-definition -Werror -fno-common   -DHAVE_CONFIG_H    -I. -
I. -I../../gcc-4.0-20041212/gcc -I../../gcc-4.0-20041212/gcc/. -I../../gcc-4.0-20041212/gcc/../
include -I./../intl -I../../gcc-4.0-20041212/gcc/../libcpp/include  ../../gcc-4.0-20041212/gcc/
reload.c -o reload.o

whereby the last both commands differ only in the compiler that's used, at first stage1/xgcc and later 
stage2/xgcc. Those commands were the ones which results differed in the comparison, so I reissued 
them (in the gcc subdirectory) with a '-save-temps' option added:

localhost:~/GCC/FSF/gcc-build/gcc lars$ stage1/xgcc -Bstage1/ -B/tmp/gcc/i686-apple-
darwin7.2.1/bin/ -c   -O2 -g -fomit-frame-pointer -DIN_GCC   -W -Wall -Wwrite-strings -Wstrict-
prototypes -Wmissing-prototypes -pedantic -Wno-long-long -Wno-variadic-macros -Wold-style-
definition -Werror -fno-common   -DHAVE_CONFIG_H    -I. -I. -I../../gcc-4.0-20041212/gcc -I../../
gcc-4.0-20041212/gcc/. -I../../gcc-4.0-20041212/gcc/../include -I./../intl -I../../gcc-4.0-
20041212/gcc/../libcpp/include -save-temps ../../gcc-4.0-20041212/gcc/reload.c -o reload.o
localhost:~/GCC/FSF/gcc-build/gcc lars$ mv reload.i stage2-reload.i
localhost:~/GCC/FSF/gcc-build/gcc lars$ mv reload.s stage2-reload.s
localhost:~/GCC/FSF/gcc-build/gcc lars$ stage2/xgcc -Bstage2/ -B/tmp/gcc/i686-apple-
darwin7.2.1/bin/ -c   -O2 -g -fomit-frame-pointer -DIN_GCC   -W -Wall -Wwrite-strings -Wstrict-
prototypes -Wmissing-prototypes -pedantic -Wno-long-long -Wno-variadic-macros -Wold-style-
definition -Werror -fno-common   -DHAVE_CONFIG_H    -I. -I. -I../../gcc-4.0-20041212/gcc -I../../
gcc-4.0-20041212/gcc/. -I../../gcc-4.0-20041212/gcc/../include -I./../intl -I../../gcc-4.0-
20041212/gcc/../libcpp/include -save-temps ../../gcc-4.0-20041212/gcc/reload.c -o reload.o
localhost:~/GCC/FSF/gcc-build/gcc lars$ mv reload.i stage3-reload.i
localhost:~/GCC/FSF/gcc-build/gcc lars$ mv reload.s stage3-reload.s

I found out that the .i files and .s files don't differ, so I attached them only once

regards, Lars

-- 
           Summary: Regression: bootstrap failure of gcc-4.0-20041212 on
                    OpenDarwin 7.2.1/x86 (i686-apple-darwin7.2.1): Bootstrap
                    comparison failure
           Product: gcc
           Version: 4.0.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: bootstrap
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: lars dot sonchocky-helldorf at hamburg dot de
                CC: gcc-bugs at gcc dot gnu dot org
 GCC build triplet: i686-apple-darwin7.2.1
  GCC host triplet: i686-apple-darwin7.2.1
GCC target triplet: i686-apple-darwin7.2.1


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18972

[Bug bootstrap/18972] Regression: bootstrap failure of gcc-4.0-20041212 on OpenDarwin 7.2.1/x86 (i686-apple-darwin7.2.1): Bootstrap comparison failure

[lars dot sonchocky-helldorf at hamburg dot de]

------- Additional Comments From lars dot sonchocky-helldorf at hamburg dot de  2004-12-14 00:50 -------
Created an attachment (id=7733)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=7733&action=view)
result of stage 2 compilation derived with '-save-temps'


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18972

[Bug bootstrap/18972] Regression: bootstrap failure of gcc-4.0-20041212 on OpenDarwin 7.2.1/x86 (i686-apple-darwin7.2.1): Bootstrap comparison failure

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-12-14 00:54 -------
"I found out that the .i files and .s files don't differ, so I attached them only once"
If the .s don't differ, then this is not a gcc bug or rather just a makefile problem or something is really 
wrong somewhere.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18972

[Bug bootstrap/18972] Regression: bootstrap failure of gcc-4.0-20041212 on OpenDarwin 7.2.1/x86 (i686-apple-darwin7.2.1): Bootstrap comparison failure

[lars dot sonchocky-helldorf at hamburg dot de]

------- Additional Comments From lars dot sonchocky-helldorf at hamburg dot de  2004-12-14 01:27 -------
maybe my attempt to "simulate" stage2 and stage3 were not alright, I am doing another clean bootstrap 
now:

../gcc-4.0-20041212/configure --prefix=/tmp/gcc
make bootstrap STAGE1_CFLAGS="-no-cpp-precomp -Wa,-force_cpusubtype_ALL" BOOTCFLAGS="-O2 
-g -save-temps"

after the bubblestrap attempt (touch ../../gcc-4.0-20041212/gcc/reload.c followed by: make 
bubblestrap BOOTCFLAGS="-O2 -g -save-temps") didn't work (-save-temps was ignored)


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18972

[Bug bootstrap/18972] Regression: bootstrap failure of gcc-4.0-20041212 on OpenDarwin 7.2.1/x86 (i686-apple-darwin7.2.1): Bootstrap comparison failure

[lars dot sonchocky-helldorf at hamburg dot de]

------- Additional Comments From lars dot sonchocky-helldorf at hamburg dot de  2004-12-14 13:46 -------
Well next time I better read http://gcc.gnu.org/install/build.html thoroughly
instead of just banking on the "word of irc" instead:
It had to be 'BOOT_CFLAGS' instead of 'BOOTCFLAGS' of course.

Anyway, surprisingly 'make bootstrap STAGE1_CFLAGS="-no-cpp-precomp
-Wa,-force_cpusubtype_ALL" BOOT_CFLAGS="-O2 -g -save-temps"' worked flawlessly -
no Bootstrap comparison failure any more.

I also came to know that '-save-temps' doesn't save the .i and .s files in the
stageX directories but just in the gcc directory (so the stage2 .i and .s files
would get erased by a bootstrap or bubblestrap)

So I started over like this:

cd GCC/FSF/gcc-build/
rm -rf *
../gcc-4.0-20041212/configure --prefix=/tmp/gcc
make bootstrap STAGE1_CFLAGS="-no-cpp-precomp -Wa,-force_cpusubtype_ALL"
cd gcc

redoing stage 2 (note the additional '-fomit-frame-pointer' (which I omitted in
the first place too ;-))):

touch ../../gcc-4.0-20041212/gcc/reload.c
make restage2 BOOT_CFLAGS="-O2 -g -fomit-frame-pointer -save-temps"

now I moved reload.i and reload.s into a safe place and then redoing stage 3:

touch ../../gcc-4.0-20041212/gcc/reload.c
make restage3 BOOT_CFLAGS="-O2 -g -fomit-frame-pointer -save-temps"

then I saved the "other" reload.i and reload.s too

The comparison of all stage2 vs. stage3 files yieled absolutelly no difference
(even the .o files are similar)


Heisenbug?

Next thing I'll do is a clean 'make bootstrap STAGE1_CFLAGS="-no-cpp-precomp
-Wa,-force_cpusubtype_ALL" BOOT_CFLAGS="-O2 -g -fomit-frame-pointer"' and then
again the same with '-save-temps' added to see which (if?) one of those fails.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18972

[Bug bootstrap/18972] Regression: bootstrap failure of gcc-4.0-20041212 on OpenDarwin 7.2.1/x86 (i686-apple-darwin7.2.1): Bootstrap comparison failure

[lars dot sonchocky-helldorf at hamburg dot de]

------- Additional Comments From lars dot sonchocky-helldorf at hamburg dot de  2004-12-14 16:18 -------
Indeed a Heisenbug:

reload.o from stage2 and stage3 differ when produced like this:

touch ../../gcc-4.0-20041212/gcc/reload.c
make restage2 BOOT_CFLAGS="-O2 -g -fomit-frame-pointer"
make restage3 BOOT_CFLAGS="-O2 -g -fomit-frame-pointer"

the resultingg reload.o differ just in one byte at offset 0000B577h:

for stage2 it is: D0
for stage3 it is: C4

now in context:

stage2: 0000 B570h: 7E B9 31 C0 8B 8C 24 D0 00 00 00 89 84 24 6C 02
stage3: 0000 B570h: 7E B9 31 C0 8B 8C 24 C4 00 00 00 89 84 24 6C 02

Does anyone know how I could disassemble those .o files to get something more
human readable?

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18972

[Bug bootstrap/18972] Regression: bootstrap failure of gcc-4.0-20041212 on OpenDarwin 7.2.1/x86 (i686-apple-darwin7.2.1): Bootstrap comparison failure

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-12-14 16:21 -------
(In reply to comment #5)
> Indeed a Heisenbug:
> 
> Does anyone know how I could disassemble those .o files to get something more
> human readable?

Use otool.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18972

[Bug bootstrap/18972] Regression: bootstrap failure of gcc-4.0-20041212 on OpenDarwin 7.2.1/x86 (i686-apple-darwin7.2.1): Bootstrap comparison failure

[lars dot sonchocky-helldorf at hamburg dot de]

------- Additional Comments From lars dot sonchocky-helldorf at hamburg dot de  2004-12-18 05:18 -------
(In reply to comment #6)
> (In reply to comment #5)
> > Indeed a Heisenbug:
> > 
> > Does anyone know how I could disassemble those .o files to get something more
> > human readable?
> 
> Use otool.

Did so.

I copied the reload.o files that were created by:

make restage2 BOOT_CFLAGS="-O2 -g -fomit-frame-pointer"

and:

make restage3 BOOT_CFLAGS="-O2 -g -fomit-frame-pointer"

into:

GCC/FSF/bootstrap-failure_gcc-4.0-20041212/stage2

and:

GCC/FSF/bootstrap-failure_gcc-4.0-20041212/stage3 

and then did the following with them:

otool -tV GCC/FSF/bootstrap-failure_gcc-4.0-20041212/stage3/reload.o &> stage3.reload.o.otool.s
otool -tV GCC/FSF/bootstrap-failure_gcc-4.0-20041212/stage2/reload.o &> stage2.reload.o.otool.s

which yielded two nice assembler files that I'll attach here (plus a diff for quick overview)

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18972

[Bug bootstrap/18972] Regression: bootstrap failure of gcc-4.0-20041212 on OpenDarwin 7.2.1/x86 (i686-apple-darwin7.2.1): Bootstrap comparison failure

[lars dot sonchocky-helldorf at hamburg dot de]

------- Additional Comments From lars dot sonchocky-helldorf at hamburg dot de  2004-12-18 05:20 -------
Created an attachment (id=7774)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=7774&action=view)
reload.o of stage2 mangled through otool


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18972

[Bug bootstrap/18972] Regression: bootstrap failure of gcc-4.0-20041212 on OpenDarwin 7.2.1/x86 (i686-apple-darwin7.2.1): Bootstrap comparison failure

[lars dot sonchocky-helldorf at hamburg dot de]

------- Additional Comments From lars dot sonchocky-helldorf at hamburg dot de  2004-12-18 05:21 -------
Created an attachment (id=7775)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=7775&action=view)
reload.o of stage3 mangled through otool


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18972

[Bug bootstrap/18972] Regression: bootstrap failure of gcc-4.0-20041212 on OpenDarwin 7.2.1/x86 (i686-apple-darwin7.2.1): Bootstrap comparison failure

[lars dot sonchocky-helldorf at hamburg dot de]

------- Additional Comments From lars dot sonchocky-helldorf at hamburg dot de  2004-12-18 05:22 -------
Created an attachment (id=7776)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=7776&action=view)
diff between stage2 and stage3 reload.o otool -tV output


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18972

[Bug bootstrap/18972] Regression: bootstrap failure of gcc-4.0-20041212 on OpenDarwin 7.2.1/x86 (i686-apple-darwin7.2.1): Bootstrap comparison failure

[lars dot sonchocky-helldorf at hamburg dot de]

------- Additional Comments From lars dot sonchocky-helldorf at hamburg dot de  2004-12-18 05:38 -------
please note (I might not have stated this clearly enough in comment #4):

the bootstrap doesn't fail if:

 '-fomit-frame-pointer' is omitted (by specifying BOOT_CFLAGS="-O2 -g" instead of BOOT_CFLAGS="-
O2 -g -fomit-frame-pointer" (which is the default and doesn't need to be specified)

or if:
'-save-temps' is specified (for instance: BOOT_CFLAGS="-O2 -g -fomit-frame-pointer -save-temps" 
works)

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18972

[Bug bootstrap/18972] Regression: bootstrap failure of gcc-4.0-20041212 on OpenDarwin 7.2.1/x86 (i686-apple-darwin7.2.1): Bootstrap comparison failure

[lars dot sonchocky-helldorf at hamburg dot de]

------- Additional Comments From lars dot sonchocky-helldorf at hamburg dot de  2005-01-04 00:47 -------
I could bootstrap (but not install) the gcc-4.0-20050102 snapshot. Should this bug be closed?

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18972

[Bug bootstrap/18972] Regression: bootstrap failure of gcc-4.0-20041212 on OpenDarwin 7.2.1/x86 (i686-apple-darwin7.2.1): Bootstrap comparison failure

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-01-04 00:52 -------
(In reply to comment #12)
> I could bootstrap (but not install) the gcc-4.0-20050102 snapshot. Should this bug be closed?

Yes which I am doing right now (for the install problem if there was one, please file it).

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|                            |FIXED
   Target Milestone|---                         |4.0.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18972