[Bug debug/55063] New: [4.8 Regression] Thousands of failures in the libstdc++-v3 tests after revision 192739

[Bug debug/55063] New: [4.8 Regression] Thousands of failures in the libstdc++-v3 tests after revision 192739

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55063

             Bug #: 55063
           Summary: [4.8 Regression] Thousands of failures in the
                    libstdc++-v3 tests after revision 192739
    Classification: Unclassified
           Product: gcc
           Version: 4.8.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: debug
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: dominiq@lps.ens.fr
                CC: howarth@nitro.med.uc.edu, iains@gcc.gnu.org,
                    pkoning@gcc.gnu.org
              Host: x86_64-apple-darwin10
            Target: x86_64-apple-darwin10
             Build: x86_64-apple-darwin10


After revision 192739, I got thousands of failures in the libstdc++-v3 tests on
x86_64-apple-darwin10 (Xcode 3.2.6) of the kind:

FAIL: 17_intro/static.cc (test for excess errors)
Excess errors:
warning: invalid DWARF generated by the compiler: DIE 0x0000459e has multiple 
AT_declaration attributes in '/var/tmp//ccYrx3xr.o'.

I don't know if this is an Apple's bug or not (and I don't care! if it is the
case it won't be fixed for darwin 10). I have silenced the failures with the
following hack (i.e., pruning the warning):

--- ../_clean/libstdc++-v3/testsuite/lib/prune.exp    2012-04-11
14:09:59.000000000 +0200
+++ libstdc++-v3/testsuite/lib/prune.exp    2012-10-24 16:55:36.000000000 +0200
@@ -63,6 +63,7 @@ proc libstdc++-dg-prune { system text } 
     regsub -all "(^|\n)\[^\n\]*ld: warning: can't add line info to anonymous
symbol\[^\n\]*" $text "" text
     regsub -all "(^|\n)\[^\n\]*warning:
DWARFDebugInfoEntry::AppendDependants\[^\n\]*AT_\[^\n\]*_bound\[^\n\]*FORM_ref4\[^\n\]*"
$text "" text
     regsub -all "(^|\n)\[^\n\]*warning:\[^\n\]*TAG_variable: 
AT_location\[^\n\]*didn't have valid function low pc\[^\n\]*" $text "" text
+    regsub -all "(^|\n)\[^\n\]*warning: invalid DWARF generated by the
compiler: DIE \[^\n\]* has multiple  AT_declaration attributes in\[^\n\]*"
$text "" text

     # Ignore harmless warnings from Xcode 4.0.
     regsub -all "(^|\n)\[^\n\]*ld: warning: could not create compact unwind
for\[^\n\]*" $text "" text

tested with -m64 only.

[Bug debug/55063] [4.8 Regression] Thousands of failures in the libstdc++-v3 tests after revision 192739

[howarth at nitro dot med.uc.edu]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55063

--- Comment #1 from Jack Howarth <howarth at nitro dot med.uc.edu> 2012-10-24 19:52:07 UTC ---
Also seen on x86_64-apple-darwin12 with Xcode 4.5.1. Note that is actually is
an error from the dsymutil run on the executable.

dsymutil ./static.exe
warning: invalid DWARF generated by the compiler: DIE 0x000045c6 has multiple 
AT_declaration attributes in '/var/tmp//ccnATCH5.o'.

[Bug debug/55063] [4.8 Regression] Thousands of failures in the libstdc++-v3 tests after revision 192739

[howarth at nitro dot med.uc.edu]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55063

--- Comment #2 from Jack Howarth <howarth at nitro dot med.uc.edu> 2012-10-24 20:06:12 UTC ---
A 'dwarfdump -a' of the malformed static.o shows the error is at...

[ 74] TAG_class_type                DW_CHILDREN_yes
       AT_name                      FORM_string
       AT_declaration               FORM_flag
       AT_declaration               FORM_flag
       AT_sibling                   FORM_ref4

[Bug debug/55063] [4.8 Regression] Thousands of failures in the libstdc++-v3 tests after revision 192739

[howarth at nitro dot med.uc.edu]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55063

--- Comment #3 from Jack Howarth <howarth at nitro dot med.uc.edu> 2012-10-24 20:12:03 UTC ---
Also seen as...

0x000045d5:     TAG_class_type [74] *
                 AT_name( "ios_base" )
                 AT_declaration( 0x01 )
                 AT_declaration( 0x01 )
                 AT_sibling( {0x00004663} )

in the output with 'dwarfdump --debug-info' on static.o.

[Bug debug/55063] [4.8 Regression] Thousands of failures in the libstdc++-v3 tests after revision 192739

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55063

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |wrong-debug
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2012-10-25
   Target Milestone|---                         |4.8.0
     Ever Confirmed|0                           |1

--- Comment #4 from Richard Biener <rguenth at gcc dot gnu.org> 2012-10-25 11:24:05 UTC ---
Looks indeed bogus.

[Bug debug/55063] [4.8 Regression] Thousands of failures in the libstdc++-v3 tests after revision 192739

[howarth at nitro dot med.uc.edu]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55063

--- Comment #5 from Jack Howarth <howarth at nitro dot med.uc.edu> 2012-10-25 17:40:26 UTC ---
Interestingly, on x86_64 Fedora 15, I don't see repeated instances of
AT_location but instead...

 <2><2b72>: Abbrev Number: 78 (DW_TAG_class_type)
    <2b73>   DW_AT_name        : (indirect string, offset: 0x27e2): ios_base
    <2b77>   DW_AT_declaration : 1
    <2b77>   DW_AT_declaration : 1
    <2b77>   DW_AT_sibling     : <0x2bdf>

repeated instances of DW_AT_declaration from 'objdump -Wi ./static.o' of the
build of the 17_intro/static.cc test case.

[Bug debug/55063] [4.8 Regression] Thousands of failures in the libstdc++-v3 tests after revision 192739

[paul_koning at dell dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55063

Paul Koning <paul_koning at dell dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |paul_koning at dell dot com

--- Comment #6 from Paul Koning <paul_koning at dell dot com> 2012-10-25 18:15:23 UTC ---
This looks like my misunderstanding of what add_AT_flag does.   I assumed it
turns on a flag in a set of flags, but it actually adds a flag record to a list
of records.  What appears to be needed is a check that the flag is not already
present, and add one only in that case.

I'll work up a patch.

[Bug debug/55063] [4.8 Regression] Thousands of failures in the libstdc++-v3 tests after revision 192739

[ccoutant at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55063

--- Comment #7 from Cary Coutant <ccoutant at gcc dot gnu.org> 2012-10-25 18:30:34 UTC ---
Author: ccoutant
Date: Thu Oct 25 18:30:27 2012
New Revision: 192820

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=192820
Log:
gcc/
    PR debug/55063
    * dwarf2out.c (prune_unused_types_prune): Check whether DIE is
    already a declaration.

Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/dwarf2out.c

[Bug debug/55063] [4.8 Regression] Thousands of failures in the libstdc++-v3 tests after revision 192739

[ccoutant at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55063

Cary Coutant <ccoutant at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
                 CC|                            |ccoutant at gcc dot gnu.org
         Resolution|                            |FIXED
         AssignedTo|unassigned at gcc dot       |ccoutant at gcc dot gnu.org
                   |gnu.org                     |

--- Comment #8 from Cary Coutant <ccoutant at gcc dot gnu.org> 2012-10-25 18:33:24 UTC ---
I've committed a fix at r192820.

http://gcc.gnu.org/ml/gcc-patches/2012-10/msg02316.html