Re: debug/5335: ICE when DWARF-1 debug info is generated

Re: debug/5335: ICE when DWARF-1 debug info is generated

[bangerth]

Synopsis: ICE when DWARF-1 debug info is generated

State-Changed-From-To: closed->suspended
State-Changed-By: bangerth
State-Changed-When: Mon Mar 24 15:23:31 2003
State-Changed-Why:
    As Daniel mentioned correcty, dwarf1 is only deprecated,
    not removed in 3.4 (but deprecated already in 3.3). I
    seem to have misread the compiler output, sorry for the
    confusion.
    
    Since chances are low that people will want to fix reports
    for deprecated parts of the compiler, I put this into
    "suspended" state.
    
    W.

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=5335

Re: debug/5335: ICE when DWARF-1 debug info is generated

[Daniel Jacobowitz]

The following reply was made to PR debug/5335; it has been noted by GNATS.

From: Daniel Jacobowitz <drow@mvista.com>
To: bangerth@dealii.org, gcc-bugs@gcc.gnu.org, igusarov@akella.com,
	gcc-gnats@gcc.gnu.org
Cc:  
Subject: Re: debug/5335: ICE when DWARF-1 debug info is generated
Date: Sat, 22 Mar 2003 17:32:28 -0500

 On Sat, Mar 22, 2003 at 07:05:31PM -0000, bangerth@dealii.org wrote:
 > Synopsis: ICE when DWARF-1 debug info is generated
 > 
 > State-Changed-From-To: open->closed
 > State-Changed-By: bangerth
 > State-Changed-When: Sat Mar 22 19:05:30 2003
 > State-Changed-Why:
 >     Confirmed with 3.0 through 3.2.2, but 3.3 deprecates and
 >     3.4 removes support for dward1 debugging
 
 I checked with Zack; this was suggested, but not decided.  Suggest
 leaving the PR open until it is decided.
 
 -- 
 Daniel Jacobowitz
 MontaVista Software                         Debian GNU/Linux Developer

Re: debug/5335: ICE when DWARF-1 debug info is generated

[bangerth]

Synopsis: ICE when DWARF-1 debug info is generated

State-Changed-From-To: open->closed
State-Changed-By: bangerth
State-Changed-When: Sat Mar 22 19:05:30 2003
State-Changed-Why:
    Confirmed with 3.0 through 3.2.2, but 3.3 deprecates and
    3.4 removes support for dward1 debugging

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=5335

Re: debug/5335: ICE when DWARF-1 debug info is generated

[bangerth]

Synopsis: ICE when DWARF-1 debug info is generated

State-Changed-From-To: feedback->open
State-Changed-By: bangerth
State-Changed-When: Sat Mar 22 19:02:03 2003
State-Changed-Why:
    Feedback received

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=5335

Re: debug/5335: ICE when DWARF-1 debug info is generated

[Wolfgang Bangerth]

The following reply was made to PR debug/5335; it has been noted by GNATS.

From: Wolfgang Bangerth <bangerth@ticam.utexas.edu>
To: "Igor A. Goussarov" <igusarov@mail.ru>
Cc: gcc-gnats@gcc.gnu.org
Subject: Re: debug/5335: ICE when DWARF-1 debug info is generated
Date: Fri, 21 Mar 2003 10:06:20 -0600 (CST)

 >     I've found that test case. In fact, it is not a specially written 
 > test case, it is just a piece of preprocessed C++ code, very heavy on 
 > templates. It has been manually reduced, eliminating a lot of unrelated 
 > #includes, types, classes and member functions. What remains is about 
 > 25Kb file which contains several class templates and _one_ function that 
 > takes one parameter of type "pointer to some complicated template class" 
 > and has an empty function body.
 
 Send me this one file in private mail, I will then attach it to the 
 report.
 
 >     How do I send this file to you, then? You've told me not to send 
 > attachments, and 1000 lines of code is too much to send in a mail body. 
 
 But it's ok as an attachment to a mail.
 
 Thanks
  Wolfgang
 
 -------------------------------------------------------------------------
 Wolfgang Bangerth             email:            bangerth@ticam.utexas.edu
                               www: http://www.ticam.utexas.edu/~bangerth/
 
 

Re: debug/5335: ICE when DWARF-1 debug info is generated

[Igor A. Goussarov]

The following reply was made to PR debug/5335; it has been noted by GNATS.

From: "Igor A. Goussarov" <igusarov@mail.ru>
To: bangerth@dealii.org, gcc-bugs@gcc.gnu.org, gcc-prs@gcc.gnu.org,
	igusarov@akella.com, nobody@gcc.gnu.org, gcc-gnats@gcc.gnu.org
Cc:  
Subject: Re: debug/5335: ICE when DWARF-1 debug info is generated
Date: Fri, 21 Mar 2003 07:14:35 +0300

 Hello Wolfgang,
 
     I've found that test case. In fact, it is not a specially written 
 test case, it is just a piece of preprocessed C++ code, very heavy on 
 templates. It has been manually reduced, eliminating a lot of unrelated 
 #includes, types, classes and member functions. What remains is about 
 25Kb file which contains several class templates and _one_ function that 
 takes one parameter of type "pointer to some complicated template class" 
 and has an empty function body. If I would have commented out this 
 function, the file would compile perfectly producing an empty .o file. 
 (I mean an object file with no one function in it, not a zero-sized 
 file). However, it doesn't look like the problem is in this particular 
 function. This function acts merely like a trigger, causing all of the 
 above templates to be instantiated at once. And it looks asif something 
 goes wrong during template instantiation (or generating dwarf debug info 
 for all these instantiated templates). Well... it's just my speculations...
     How do I send this file to you, then? You've told me not to send 
 attachments, and 1000 lines of code is too much to send in a mail body. 
 I can put its gzipped version somewhere to http, so you can download it. 
 Or upload it to your ftp.
 
 Cheers,
 Igor
 
 
 bangerth@dealii.org wrote:
 > Synopsis: ICE when DWARF-1 debug info is generated
 > 
 > State-Changed-From-To: open->feedback
 > State-Changed-By: bangerth
 > State-Changed-When: Fri Mar 14 02:47:14 2003
 > State-Changed-Why:
 >     Some kind of bitrot seems to have happened to this report
 >     while nobody was staring so hard at it all the time -- the
 >     attachment got corrupted. Do you still have the testcase
 >     you attached back then? I'd like to try to reproduce it,
 >     but without attachment...
 >     
 >     Thanks 
 >       Wolfgang
 > 
 > http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=5335
 

Re: debug/5335: ICE when DWARF-1 debug info is generated

[bangerth]

Synopsis: ICE when DWARF-1 debug info is generated

State-Changed-From-To: open->feedback
State-Changed-By: bangerth
State-Changed-When: Fri Mar 14 02:47:14 2003
State-Changed-Why:
    Some kind of bitrot seems to have happened to this report
    while nobody was staring so hard at it all the time -- the
    attachment got corrupted. Do you still have the testcase
    you attached back then? I'd like to try to reproduce it,
    but without attachment...
    
    Thanks 
      Wolfgang

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=5335

debug/5335: ICE when DWARF-1 debug info is generated

[igusarov]

>Number:         5335
>Category:       debug
>Synopsis:       ICE when DWARF-1 debug info is generated
>Confidential:   no
>Severity:       non-critical
>Priority:       low
>Responsible:    unassigned
>State:          open
>Class:          ice-on-legal-code
>Submitter-Id:   net
>Arrival-Date:   Wed Jan 09 06:26:01 PST 2002
>Closed-Date:
>Last-Modified:
>Originator:     Igor A. Goussarov
>Release:        Reading specs from /usr/lib/gcc-lib/i386-unknown-freebsd4.4/3.0.3/specs   Configured with: ./configure --prefix=/usr   Thread model: posix   gcc version 3.0.3
>Organization:
>Environment:
i386-FreeBSD-4.4-RELEASE
>Description:
The attached file compiles fine if no debug info is 
requested or if DWARF-2 debug info is requested.
However, compilation with -gdwarf+, -gdwarf+1, gdwarf+2 
or -gdwarf+3 switch results in ICE:

../IGusarov/Lists.h:821: Internal compiler error in dwarfout_finish, at
   dwarfout.c:6532
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://www.gnu.org/software/gcc/bugs.html> for instructions.

It looks like generating this debug info requires some 
more templated types (which were a member 'typedefs' in 
another templated type) to be instantiated and these
freshly instantiated types are left in pending_types_list
which violates the final assertion in dwarfout_finish
and thus lead to abort()...
>How-To-Repeat:
unpack an archived "TemplateElements_.i" file and run
g++ -c -gdwarf+3 TemplateElements_.i
>Fix:
Don't use dwarf-1 debug format. In fact, I've
toyed with various debug info formats and found that
4 out of approx. 120 source files in my current project 
fails with the same ICE if compiled with one of 
-gdwarf+[123]
>Release-Note:
>Audit-Trail:
>Unformatted:
----gnatsweb-attachment----
Content-Type: application/octet-stream; name="TemplateElements_.i.gz"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="TemplateElements_.i.gz"