[Bug default/19024] New: abidw aborts when run against r300_dri.so

[Bug default/19024] New: abidw aborts when run against r300_dri.so

[woodard at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=19024

            Bug ID: 19024
           Summary: abidw aborts when run against r300_dri.so
           Product: libabigail
           Version: unspecified
            Status: NEW
          Severity: normal
          Priority: P2
         Component: default
          Assignee: dodji at redhat dot com
          Reporter: woodard at redhat dot com
                CC: libabigail at sourceware dot org
  Target Milestone: ---

[1]-  Aborted                 (core dumped) ~/bin/abidw --abidiff
/collab/usr/global/tools/order/spack/opt/chaos_5_x86_64_ib/gcc@4.4.7/mesa@8.0.5-22dd4c4b/lib/dri/r300_dri.so

Functions changes summary: 0 Removed, 820 Changed, 0 Added functions
Variables changes summary: 0 Removed, 1 Changed, 0 Added variable

820 functions with some indirect sub-type change:

  [C]'method llvm::APFloat::APFloat(const llvm::APInt&, bool)' has some
indirect sub-type changes:
    parameter 1 of type 'const llvm::APInt&' changed:
      in referenced type 'const llvm::APInt':

  [C]'method llvm::APFloat::opStatus llvm::APFloat::convertFromAPInt(const
llvm::APInt&, bool, llvm::APFloat::roundingMode)' has some indirect sub-type
changes:
    parameter 1 of type 'const llvm::APInt&' changed:
      in referenced type 'const llvm::APInt':

  [C]'method void llvm::APFloat::initFromAPInt(const llvm::APInt&, bool)' has
some indirect sub-type changes:
    parameter 1 of type 'const llvm::APInt&' changed:
      in referenced type 'const llvm::APInt':

  [C]'method void llvm::APFloat::initFromDoubleAPInt(const llvm::APInt&)' has
some indirect sub-type changes:
    parameter 1 of type 'const llvm::APInt&' changed:
      in referenced type 'const llvm::APInt':

  [C]'method void llvm::APFloat::initFromF80LongDoubleAPInt(const
llvm::APInt&)' has some indirect sub-type changes:
    parameter 1 of type 'const llvm::APInt&' changed:
      in referenced type 'const llvm::APInt':

  [C]'method void llvm::APFloat::initFromFloatAPInt(const llvm::APInt&)' has
some indirect sub-type changes:
    parameter 1 of type 'const llvm::APInt&' changed:
      in referenced type 'const llvm::APInt':

  [C]'method void llvm::APFloat::initFromHalfAPInt(const llvm::APInt&)' has
some indirect sub-type changes:
    parameter 1 of type 'const llvm::APInt&' changed:
      in referenced type 'const llvm::APInt':

  [C]'method void llvm::APFloat::initFromPPCDoubleDoubleAPInt(const
llvm::APInt&)' has some indirect sub-type changes:
    parameter 1 of type 'const llvm::APInt&' changed:
      in referenced type 'const llvm::APInt':

  [C]'method void llvm::APFloat::initFromQuadrupleAPInt(const llvm::APInt&)'
has some indirect sub-type changes:
    parameter 1 of type 'const llvm::APInt&' changed:
      in referenced type 'const llvm::APInt':

  [C]'method llvm::APFloat llvm::APFloat::makeNaN(bool, bool, const
llvm::APInt*)' has some indirect sub-type changes:
abidw: ../../src/abg-comparison.cc:11005: virtual void
abigail::comparison::fn_parm_diff::report(std::ostream&, const std::string&)
const: Assertion `get_type_diff() && get_type_diff()->to_be_reported()' failed.

This abort takes quite a while to reproduce. Let it run.

I am not able to attach the file due to size but it can be found at:
http://ssh.bencoyote.net/~ben/r300_dri.so.gz

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug default/19024] abidw aborts when run against r300_dri.so

[woodard at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=19024

--- Comment #2 from Ben Woodard <woodard at redhat dot com> ---
for future reference I found another case of this in case there are two paths
to the same assertion:

hype260@ben:cat
\\_collab\\_usr\\_global\\_tools\\_order\\_spack\\_opt\\_chaos_5_x86_64_ib\\_gcc\@4.4.7\\_gperftools\@2.4\\_lib\\_libtcmalloc_debug.so/stderr
Functions changes summary: 0 Removed, 2 Changed, 0 Added functions
Variables changes summary: 0 Removed, 0 Changed, 0 Added variable

2 functions with some indirect sub-type change:

  [C]'method HeapLeakChecker* HeapLeakChecker::GlobalChecker()' has some
indirect sub-type changes:
    return type changed:
      in pointed to type 'class HeapLeakChecker':
        1 data member change:


  [C]'method SpinLockHolder::SpinLockHolder(SpinLock*)' has some indirect
sub-type changes:
abidw: ../../src/abg-comparison.cc:11005: virtual void
abigail::comparison::fn_parm_diff::report(std::ostream&, const std::string&)
const: Assertion `get_type_diff() && get_type_diff()->to_be_reported()' failed.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug default/19024] abidw aborts when run against r300_dri.so

[woodard at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=19024

--- Comment #1 from Ben Woodard <woodard at redhat dot com> ---
For future reference: a similar failure happens against libtcmalloc_debug.so

hype356@ben:~/bin/abidw --abidiff
/collab/usr/global/tools/order/spack/opt/chaos_5_x86_64_ib/gcc@4.4.7/gperftools@2.4/lib/libtcmalloc_debug.so
Functions changes summary: 0 Removed, 2 Changed, 0 Added functions
Variables changes summary: 0 Removed, 0 Changed, 0 Added variable

2 functions with some indirect sub-type change:

  [C]'method HeapLeakChecker* HeapLeakChecker::GlobalChecker()' has some
indirect sub-type changes:
    return type changed:
      in pointed to type 'class HeapLeakChecker':
        1 data member change:


  [C]'method SpinLockHolder::SpinLockHolder(SpinLock*)' has some indirect
sub-type changes:
abidw: ../../src/abg-comparison.cc:11005: virtual void
abigail::comparison::fn_parm_diff::report(std::ostream&, const std::string&)
const: Assertion `get_type_diff() && get_type_diff()->to_be_reported()' failed.
Aborted (core dumped)

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug default/19024] abidw --abidiff fails and aborts when run against r300_dri.so

[dodji at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=19024

dodji at redhat dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|abidw aborts when run       |abidw --abidiff fails and
                   |against r300_dri.so         |aborts when run against
                   |                            |r300_dri.so

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug default/19024] abidw aborts when run against r300_dri.so

[woodard at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=19024

--- Comment #4 from Ben Woodard <woodard at redhat dot com> ---
Yea! The dodji/fixes seems to run to completion.

hype356@ben:time ~/bin/abidw --abidiff
/collab/usr/global/tools/order/spack/opt/chaos_5_x86_64_ib/gcc@4.4.7/mesa@8.0.5-22dd4c4b/lib/dri/r300_dri.so

...


  [C]'llvm::Target llvm::TheX86_64Target' was changed:
    type 'class llvm::Target' changed, as reported earlier



real    11m50.724s
user    11m39.418s
sys     0m8.598s

Now you just have to deal with all the errors.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug default/19024] abidw aborts when run against r300_dri.so

[woodard at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=19024

--- Comment #6 from Ben Woodard <woodard at redhat dot com> ---
Something interesting that I noticed when running abidw on:

 30187 ben       20   0  701m 602m  55m R 100.0  1.9   0:50.32 abidw            
 30188 ben       20   0 4313m 4.1g  97m R 100.0 13.1   0:49.49 abidw            

then later:

 30187 ben       20   0 2110m 2.0g  61m R 100.0  6.3   3:43.84 abidw            
 30188 ben       20   0 4425m 4.2g  97m R 99.7 13.5   3:42.98 abidw         

When started with this command:
~/bin/abidw r300_dri.so.dwz >dwz.abixml & ~/bin/abidw r300_dri.so >plain.abixml
&

So evidently, the compressed DWARF is taking a lot less memory. Of course the
one running on the dwz'd version crashes (see other pr). However, once that one
is resolved it could make debugging easier.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug default/19024] abidw aborts when run against r300_dri.so

[woodard at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=19024

--- Comment #3 from Ben Woodard <woodard at redhat dot com> ---
The reproducer for this bug can be found at
http://ssh.bencoyote.net/~ben/r300_dri.so.gz

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug default/19024] abidw --abidiff fails and aborts when run against r300_dri.so

[dodji at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=19024

dodji at redhat dot com changed:

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

--- Comment #7 from dodji at redhat dot com ---
This issue should now be fixed a series of 13 patches, in master, ending with
commit
https://sourceware.org/git/gitweb.cgi?p=libabigail.git;a=commit;h=6d4472066b11ddba4c0c722cf31aa542f201ceac.

Thank you for filing this issue!

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug default/19024] abidw aborts when run against r300_dri.so

[dodji at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=19024

--- Comment #5 from dodji at redhat dot com ---
Good to know, thank you for that, Ben.

I have pushed the abort fix to master as commit
https://sourceware.org/git/gitweb.cgi?p=libabigail.git;a=commit;h=efeb4e2a8a13bb78a35ed4d11edc16184a337559.
Now I'll look at the type stability issues.

-- 
You are receiving this mail because:
You are on the CC list for the bug.