[Bug ada/14060] New: Crash regression for GCC HEAD

[Bug ada/14060] New: Crash regression for GCC HEAD

[jan.kratochvil at redhat dot com]

http://sourceware.org/bugzilla/show_bug.cgi?id=14060

             Bug #: 14060
           Summary: Crash regression for GCC HEAD
           Product: gdb
           Version: HEAD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: ada
        AssignedTo: unassigned@sourceware.org
        ReportedBy: jan.kratochvil@redhat.com
                CC: brobecker@adacore.com
    Classification: Unclassified
            Target: x86_64-unknown-linux-gnu


rm -rf gdb.ada/arrayidx/;git reset --hard # workaround ada testcases rebuilding

PATH="/your/path/to/gcc-head/bin:$PATH" runtest GNATMAKE_FOR_TARGET='gnatmake
--GCC=gcc\ -gdwarf-4\ -fdebug-types-section\ -g0' gdb.ada/arrayidx.exp 

also reproducible with:
../gdb -nx gdb.ada/arrayidx/p -ex 'break p.adb:44' -ex r -ex 'print
e_one_two_three'

ada-lang.c:10546: internal-error: to_fixed_range_type: Assertion `TYPE_NAME
(raw_type) != NULL' failed.

10545      gdb_assert (raw_type != NULL);
10546      gdb_assert (TYPE_NAME (raw_type) != NULL);
(gdb) p *raw_type.main_type
$3 = 
{name = 0x0,
 tag_name = 0x7ffff7e12282 "p__index",
 code = TYPE_CODE_ENUM,
 flags = [objfile_owned],
 owner = 0x216d0f0 (objfile),
 target_type = 0x0,
 vptr_basetype = 0x0,
 type_specific_field = TYPE_SPECIFIC_NONE}

GDB definitely should not internal error on GCC output.
Whether the GCC HEAD change is valid or not I do not know.

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

[Bug ada/14060] Crash regression for GCC HEAD

[jan.kratochvil at redhat dot com]

http://sourceware.org/bugzilla/show_bug.cgi?id=14060

Jan Kratochvil <jan.kratochvil at redhat dot com> changed:

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

--- Comment #1 from Jan Kratochvil <jan.kratochvil at redhat dot com> 2012-05-04 18:06:56 UTC ---
This is some general Ada non-specific GCC regression incl. C++.

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

[Bug ada/14060] Crash regression for GCC HEAD

[brobecker at adacore dot com]

http://sourceware.org/bugzilla/show_bug.cgi?id=14060

--- Comment #2 from brobecker at adacore dot com 2012-05-04 18:09:15 UTC ---
> ada-lang.c:10546: internal-error: to_fixed_range_type: Assertion `TYPE_NAME
> (raw_type) != NULL' failed.
[...]
> GDB definitely should not internal error on GCC output.
> Whether the GCC HEAD change is valid or not I do not know.

I agree that internal-error, even on bad GCC output, is bad.
Can you send me the binary? I'll take a look.

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

[Bug ada/14060] Crash regression for GCC HEAD

[jan.kratochvil at redhat dot com]

http://sourceware.org/bugzilla/show_bug.cgi?id=14060

--- Comment #3 from Jan Kratochvil <jan.kratochvil at redhat dot com> 2012-05-04 18:32:56 UTC ---
Created attachment 6391
  --> http://sourceware.org/bugzilla/attachment.cgi?id=6391
gdb.ada/arrayidx/p after xz

It seems to be too serious debug regression by GCC that GDB does not have to
handle it IMO.
Filed as:
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53235

Attaching the build only if you are really interested.

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.