[Bug regression/50999] New: [4.7 Regression]: g++.dg/lto/20081109 execute -O2 -flto -flto-partition=1to1 -fno-use-linker-plugin

[Bug regression/50999] New: [4.7 Regression]: g++.dg/lto/20081109 execute -O2 -flto -flto-partition=1to1 -fno-use-linker-plugin

[hp at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50999

             Bug #: 50999
           Summary: [4.7 Regression]: g++.dg/lto/20081109 execute -O2
                    -flto -flto-partition=1to1 -fno-use-linker-plugin
    Classification: Unclassified
           Product: gcc
           Version: 4.7.0
            Status: UNCONFIRMED
          Keywords: wrong-code
          Severity: normal
          Priority: P3
         Component: regression
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: hp@gcc.gnu.org
                CC: amodra@gcc.gnu.org, rguenth@gcc.gnu.org
              Host: x86_64-unknown-linux-gnu
            Target: cris-axis-elf


This test previously passed, now it fails.
Revision r180827 introduced this regression.  (Formally, triaged to either
r180826 or r180827 but trivially only the latter could be the cause.  Further,
the regression was introduced while masked by fallout from the libgcc move,
fixed by revision r180854.)
Since then, up till and including at least r181026, it fails as follows:

...
Running /tmp/hpautotest-gcc1/gcc/gcc/testsuite/g++.dg/lto/lto.exp ...
FAIL: g++.dg/lto/20081109 cp_lto_20081109_0.o-cp_lto_20081109_1.o execute -O2
-flto -flto-partition=1to1 -fno-use-linker-plugin 
...

The messages in g++.log are similar:
...
Executing on host:
/tmp/hpautotest-gcc1/cris-elf/gccobj/gcc/testsuite/g++/../../g++
-B/tmp/hpautotest-gcc1/cris-elf/gccobj/gcc/testsuite/g++/../../
cp_lto_20081109_0.o cp_lto_20081109_1.o   -nostdinc++
-I/tmp/hpautotest-gcc1/cris-elf/gccobj/cris-elf/libstdc++-v3/include/cris-elf
-I/tmp/hpautotest-gcc1/cris-elf/gccobj/cris-elf/libstdc++-v3/include
-I/tmp/hpautotest-gcc1/gcc/libstdc++-v3/libsupc++
-I/tmp/hpautotest-gcc1/gcc/libstdc++-v3/include/backward
-I/tmp/hpautotest-gcc1/gcc/libstdc++-v3/testsuite/util -fmessage-length=0 -O2
-flto -flto-partition=1to1 -fno-use-linker-plugin      -isystem
/tmp/hpautotest-gcc1/cris-elf/gccobj/cris-elf/./newlib/targ-include -isystem
/tmp/hpautotest-gcc1/gcc/newlib/libc/include
-B/tmp/hpautotest-gcc1/cris-elf/gccobj/cris-elf/./libgloss/cris/
-L/tmp/hpautotest-gcc1/cris-elf/gccobj/cris-elf/./libgloss/cris
-L/tmp/hpautotest-gcc1/gcc/libgloss/cris 
-B/tmp/hpautotest-gcc1/cris-elf/gccobj/cris-elf/./newlib/
-L/tmp/hpautotest-gcc1/cris-elf/gccobj/cris-elf/./newlib -sim3  
-L/tmp/hpautotest-gcc1/cris-elf/gccobj/cris-elf/./libstdc++-v3/src/.libs 
-B/tmp/hpautotest-gcc1/cris-elf/gccobj/cris-elf/./libstdc++-v3/src/.libs     
-o g++-dg-lto-20081109-31.exe    (timeout = 300)
PASS: g++.dg/lto/20081109 cp_lto_20081109_0.o-cp_lto_20081109_1.o link, -O2
-flto -flto-partition=1to1 -fno-use-linker-plugin 
terminate called after throwing an instance of 'int'

program stopped with signal 6.

FAIL: g++.dg/lto/20081109 cp_lto_20081109_0.o-cp_lto_20081109_1.o execute -O2
-flto -flto-partition=1to1 -fno-use-linker-plugin 
...

Author of suspect patch in revision range CC:ed.

Also CC:ed Alan Modra, as he might have observed the same for powerpc-linux,
mentioned on IRC.  That's supported by
<http://gcc.gnu.org/ml/gcc-testresults/2011-11/msg00620.html> for -m32,
compared to <http://gcc.gnu.org/ml/gcc-testresults/2011-11/msg00153.html>.  A
quick look in g++.log would help verify whether it's an identical (the same)
regression.

[Bug regression/50999] [4.7 Regression]: g++.dg/lto/20081109 execute -O2 -flto -flto-partition=1to1 -fno-use-linker-plugin

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50999

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |ASSIGNED
   Last reconfirmed|                            |2011-11-06
         AssignedTo|unassigned at gcc dot       |rguenth at gcc dot gnu.org
                   |gnu.org                     |
   Target Milestone|---                         |4.7.0
     Ever Confirmed|0                           |1

--- Comment #1 from Richard Guenther <rguenth at gcc dot gnu.org> 2011-11-06 14:38:18 UTC ---
Mine.

[Bug regression/50999] [4.7 Regression]: g++.dg/lto/20081109 execute -O2 -flto -flto-partition=1to1 -fno-use-linker-plugin

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50999

--- Comment #2 from Richard Guenther <rguenth at gcc dot gnu.org> 2011-11-07 12:02:36 UTC ---
Created attachment 25734
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=25734
proposed patch

Hm, on x86_64 the EH table is not empty (look at g++.dg/lto/20081109_1.C,
the usual symtom is that the LTOed unit for that file does not contain
info to unwind from the throw - can you confirm that?).  The key is that
lto_init_eh() has to be called for that unit - previously we saved
flag_exceptions but now we only save explicit -fexceptions (thus, you
can verify if explicitly adding that option to the compile command line
fixes the issue).

Does the attached patch help?  It tries to start to address the (yet unsolved)
issue of preserving lang-dependent options.

[Bug regression/50999] [4.7 Regression]: g++.dg/lto/20081109 execute -O2 -flto -flto-partition=1to1 -fno-use-linker-plugin

[hp at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50999

--- Comment #3 from Hans-Peter Nilsson <hp at gcc dot gnu.org> 2011-11-07 21:24:59 UTC ---
(In reply to comment #2)
> Created attachment 25734 [details]
> proposed patch
> 
> Hm, on x86_64 the EH table is not empty (look at g++.dg/lto/20081109_1.C,
> the usual symtom is that the LTOed unit for that file does not contain
> info to unwind from the throw - can you confirm that?).  The key is that
> lto_init_eh() has to be called for that unit - previously we saved
> flag_exceptions but now we only save explicit -fexceptions (thus, you
> can verify if explicitly adding that option to the compile command line
> fixes the issue).
> 
> Does the attached patch help?  It tries to start to address the (yet unsolved)
> issue of preserving lang-dependent options.

I'll throw this patch to a test session - but I thought you mentioned you have
access to powerpc-linux, wouldn't it be more productive for you to debug there?

[Bug regression/50999] [4.7 Regression]: g++.dg/lto/20081109 execute -O2 -flto -flto-partition=1to1 -fno-use-linker-plugin

[hp at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50999

--- Comment #4 from Hans-Peter Nilsson <hp at gcc dot gnu.org> 2011-11-08 04:01:29 UTC ---
(In reply to comment #2)
> Does the attached patch help?  It tries to start to address the (yet unsolved)
> issue of preserving lang-dependent options.

It does fix this regression, thanks.
Tested at r181063.

[Bug regression/50999] [4.7 Regression]: g++.dg/lto/20081109 execute -O2 -flto -flto-partition=1to1 -fno-use-linker-plugin

[rguenther at suse dot de]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50999

--- Comment #5 from rguenther at suse dot de <rguenther at suse dot de> 2011-11-08 09:29:08 UTC ---
On Mon, 7 Nov 2011, hp at gcc dot gnu.org wrote:

> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50999
> 
> --- Comment #3 from Hans-Peter Nilsson <hp at gcc dot gnu.org> 2011-11-07 21:24:59 UTC ---
> (In reply to comment #2)
> > Created attachment 25734 [details]
> > proposed patch
> > 
> > Hm, on x86_64 the EH table is not empty (look at g++.dg/lto/20081109_1.C,
> > the usual symtom is that the LTOed unit for that file does not contain
> > info to unwind from the throw - can you confirm that?).  The key is that
> > lto_init_eh() has to be called for that unit - previously we saved
> > flag_exceptions but now we only save explicit -fexceptions (thus, you
> > can verify if explicitly adding that option to the compile command line
> > fixes the issue).
> > 
> > Does the attached patch help?  It tries to start to address the (yet unsolved)
> > issue of preserving lang-dependent options.
> 
> I'll throw this patch to a test session - but I thought you mentioned you have
> access to powerpc-linux, wouldn't it be more productive for you to debug there?

;)

Trying to optimize my spare cycles ... when I have to resort to a ppc
machine I have to setup a new devel environment everytime ...

Richard.

[Bug regression/50999] [4.7 Regression]: g++.dg/lto/20081109 execute -O2 -flto -flto-partition=1to1 -fno-use-linker-plugin

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50999

--- Comment #6 from Richard Guenther <rguenth at gcc dot gnu.org> 2011-11-08 13:42:13 UTC ---
Author: rguenth
Date: Tue Nov  8 13:41:57 2011
New Revision: 181160

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=181160
Log:
2011-11-08  Richard Guenther  <rguenther@suse.de>

    PR lto/50999
    * lto-opts.c (append_to_collect_gcc_options): Split out from...
    (lto_write_options): ... here.  Prepend frontend specific flags.

Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/lto-opts.c

[Bug regression/50999] [4.7 Regression]: g++.dg/lto/20081109 execute -O2 -flto -flto-partition=1to1 -fno-use-linker-plugin

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50999

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED

--- Comment #7 from Richard Guenther <rguenth at gcc dot gnu.org> 2011-11-08 13:44:14 UTC ---
Fixed.