[Bug bootstrap/51787] New: [4.7.0 Regression] internal compiler error: in inline_small_functions, at ipa-inline.c:1410

[Bug bootstrap/51787] New: [4.7.0 Regression] internal compiler error: in inline_small_functions, at ipa-inline.c:1410

[dougmencken at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51787

             Bug #: 51787
           Summary: [4.7.0 Regression] internal compiler error: in
                    inline_small_functions, at ipa-inline.c:1410
    Classification: Unclassified
           Product: gcc
           Version: 4.7.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: bootstrap
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: dougmencken@gmail.com


I'm trying to get GCC 4.7. Latest weekly snapshot done Dec 31, 2011
(gcc-v4.7-20111231) fails to bootstrap on stage2:

/root/build-farm/_gcc_bootstrap/./prev-gcc/g++
-B/root/build-farm/_gcc_bootstrap/./prev-gcc/
-B/usr/powerpc-gnu-linux-uclibc/bin/ -nostdinc++
-B/root/build-farm/_gcc_bootstrap/prev-powerpc-gnu-linux-uclibc/libstdc++-v3/src/.libs
-B/root/build-farm/_gcc_bootstrap/prev-powerpc-gnu-linux-uclibc/libstdc++-v3/libsupc++/.libs
-I/root/build-farm/_gcc_bootstrap/prev-powerpc-gnu-linux-uclibc/libstdc++-v3/include/powerpc-gnu-linux-uclibc
-I/root/build-farm/_gcc_bootstrap/prev-powerpc-gnu-linux-uclibc/libstdc++-v3/include
-I/root/build-farm/gcc-v4.7-20111231.sourcedir/libstdc++-v3/libsupc++
-L/root/build-farm/_gcc_bootstrap/prev-powerpc-gnu-linux-uclibc/libstdc++-v3/src/.libs
-L/root/build-farm/_gcc_bootstrap/prev-powerpc-gnu-linux-uclibc/libstdc++-v3/libsupc++/.libs
-c   -O3 -gtoggle -DIN_GCC   -fno-exceptions -fno-rtti -W -Wall -Wno-narrowing
-Wwrite-strings -Wcast-qual -Wmissing-format-attribute -pedantic -Wno-long-long
-Wno-variadic-macros -Wno-overlength-strings   -DHAVE_CONFIG_H -I. -I.
-I../../gcc-v4.7-20111231.sourcedir/gcc
-I../../gcc-v4.7-20111231.sourcedir/gcc/.
-I../../gcc-v4.7-20111231.sourcedir/gcc/../include
-I../../gcc-v4.7-20111231.sourcedir/gcc/../libcpp/include 
-I../../gcc-v4.7-20111231.sourcedir/gcc/../libdecnumber
-I../../gcc-v4.7-20111231.sourcedir/gcc/../libdecnumber/dpd -I../libdecnumber  
 ../../gcc-v4.7-20111231.sourcedir/gcc/cgraph.c -o cgraph.o
In file included from ../../gcc-v4.7-20111231.sourcedir/gcc/cgraph.c:3060:0:
./gt-cgraph.h:66:2: internal compiler error: in inline_small_functions, at
ipa-inline.c:1410
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.
make[3]: *** [cgraph.o] Error 1
make[3]: Leaving directory `/root/build-farm/_gcc_bootstrap/gcc'
make[2]: *** [all-stage2-gcc] Error 2
make[2]: Leaving directory `/root/build-farm/_gcc_bootstrap'
make[1]: *** [stage2-bubble] Error 2
make[1]: Leaving directory `/root/build-farm/_gcc_bootstrap'
make: *** [all] Error 2
ERROR: stage2 bootstrap error. Abort.

I'm building everything with -O3.

Configure line:

../gcc-v4.7-20111231.sourcedir/configure --prefix=/usr --sysconfdir=/etc
--mandir=/usr/share/man --build=powerpc-gnu-linux-uclibc
--host=powerpc-gnu-linux-uclibc --target=powerpc-gnu-linux-uclibc
--enable-languages=c,c++,fortran,java --enable-stage1-languages=c,c++,fortran
--with-cpu=default32 --enable-threads=posix --with-system-zlib
--with-libiconv-prefix=/usr --without-included-gettext --disable-nls
--disable-werror --disable-multilib --disable-libssp --disable-__cxa_atexit
--enable-shared --without-gjdoc --enable-libgcj --enable-libgcj-multifile
--enable-interpreter --disable-libjava-multilib --disable-browser-plugin
--enable-checking=release

stage0's gcc -v:

Using built-in specs.
COLLECT_GCC=gcc
COLLECT_LTO_WRAPPER=/usr/libexec/gcc/powerpc-gnu-linux-uclibc/4.6.1/lto-wrapper
Target: powerpc-gnu-linux-uclibc
Configured with: ../gcc-v4.6.1.sourcedir/configure --prefix=/usr
--sysconfdir=/etc --mandir=/usr/share/man --build=powerpc-gnu-linux-uclibc
--host=powerpc-gnu-linux-uclibc --target=powerpc-gnu-linux-uclibc
--enable-languages=c,c++,fortran,java --with-cpu=default32
--enable-threads=posix --with-system-zlib --with-libiconv-prefix=/usr
--disable-nls --disable-werror --disable-multilib --disable-libssp
--disable-__cxa_atexit --disable-checking --enable-shared --without-gjdoc
--enable-libgcj --enable-libgcj-multifile --enable-interpreter
--disable-libjava-multilib --disable-browser-plugin
Thread model: posix
gcc version 4.6.1 (GCC) 

cgraph.ii is attached.

Lines 1399-1416 from ipa-inline.c:

1399:      /* Be sure that caches are maintained consistent.  
1400:         We can not make this ENABLE_CHECKING only because it cause
differnt
1401:         updates of the fibheap queue.  */
1402:      cached_badness = edge_badness (edge, false);
1403:      reset_edge_growth_cache (edge);
1404:      reset_node_growth_cache (edge->callee);
1405:
1406:      /* When updating the edge costs, we only decrease badness in the
keys.
1407:         Increases of badness are handled lazilly; when we see key with
out
1408:         of date value on it, we re-insert it now.  */
1409:      current_badness = edge_badness (edge, false);
1410:>>>>  gcc_assert (cached_badness == current_badness);
1411:      gcc_assert (current_badness >= badness);
1412:      if (current_badness != badness)
1413:        {
1414:          edge->aux = fibheap_insert (heap, current_badness, edge);
1415:          continue;
1416:        }

[Bug bootstrap/51787] [4.7.0 Regression] internal compiler error: in inline_small_functions, at ipa-inline.c:1410

[dougmencken at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51787

--- Comment #1 from Douglas Mencken <dougmencken at gmail dot com> 2012-01-07 20:48:15 UTC ---
Created attachment 26268
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=26268
pre-processed output of last failed operation (compressed)

[Bug bootstrap/51787] [4.7.0 Regression] internal compiler error: in inline_small_functions, at ipa-inline.c:1410

[burnus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51787

Tobias Burnus <burnus at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |burnus at gcc dot gnu.org

--- Comment #2 from Tobias Burnus <burnus at gcc dot gnu.org> 2012-01-07 21:07:39 UTC ---
Works for me with GCC 4.7.0 20120107 (experimental) [trunk revision 182980] on
x86-64-linux (with and without -m32; without one gets the error "cast from
'void*' to 'size_t' loses precision").

I used the following flags but also some variants:
g++ -O3 -gtoggle -m32 -fno-exceptions -fno-rtti -W -Wall -Wno-narrowing -c
-Wwrite-strings -Wcast-qual -Wmissing-format-attribute -pedantic -Wno-long-long
-Wno-variadic-macros -Wno-overlength-strings cgraph.ii

[Bug bootstrap/51787] [4.7 Regression] internal compiler error: in inline_small_functions, at ipa-inline.c:1410

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51787

Richard Guenther <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.7.0
            Summary|[4.7.0 Regression] internal |[4.7 Regression] internal
                   |compiler error: in          |compiler error: in
                   |inline_small_functions, at  |inline_small_functions, at
                   |ipa-inline.c:1410           |ipa-inline.c:1410

[Bug bootstrap/51787] [4.7 Regression] internal compiler error: in inline_small_functions, at ipa-inline.c:1410

[dougmencken at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51787

--- Comment #3 from Douglas Mencken <dougmencken at gmail dot com> 2012-01-09 17:43:20 UTC ---
I can re-confirm that snapshots 4.7-20111231, 4.7-20120107 do fail (can't enter
this into "Known to fail" field, because despite that
http://gcc.gnu.org/bugzilla/page.cgi?id=fields.html#cf_known_to_fail says that
it is `A custom Free Text field', I'm facing the error: `The value "4.7" used
in the "Known to fail" field is not a valid version for the "gcc" product').

Bootstrapped 4.6.2 successfully (with the same config I'm using for
4.7-prereleases, see initial post), and tried to bootstrap 4.7-20111231 and
4.7-20120107 with that 4.6.2 as stage0 compiler. No success. Same error.

--------------------------------

(In reply to comment #2)

> Works for me with GCC 4.7.0 20120107 (experimental) [trunk revision 182980] on
> x86-64-linux

Well, maybe it only fails on 32-bit linux@powerpc systems. (Also I doubt
there's something related to uClibc.)

[Bug bootstrap/51787] [4.7 Regression] internal compiler error: in inline_small_functions, at ipa-inline.c:1410

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51787

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jakub at gcc dot gnu.org

--- Comment #4 from Jakub Jelinek <jakub at gcc dot gnu.org> 2012-01-09 23:29:26 UTC ---
ICEs in the same spot as PR51801.

[Bug bootstrap/51787] [4.7 Regression] internal compiler error: in inline_small_functions, at ipa-inline.c:1410

[dougmencken at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51787

--- Comment #5 from Douglas Mencken <dougmencken at gmail dot com> 2012-01-11 03:33:40 UTC ---
Fixed PR51801 fixes this issue too. Current checkout bootstraps successfully.

[Bug bootstrap/51787] [4.7 Regression] internal compiler error: in inline_small_functions, at ipa-inline.c:1410

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51787

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

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

--- Comment #6 from Jakub Jelinek <jakub at gcc dot gnu.org> 2012-01-11 07:30:07 UTC ---
Fixed.

*** This bug has been marked as a duplicate of bug 51801 ***