Re: c++/2274: loading elf generates error msg

Re: c++/2274: loading elf generates error msg

[drow]

Synopsis: loading elf generates error msg

State-Changed-From-To: open->closed
State-Changed-By: drow
State-Changed-When: Mon Jun 18 18:00:08 2007
State-Changed-Why:
    We established that this was a linker or object file format problem.

http://sourceware.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gdb&pr=2274

Re: c++/2274: loading elf generates error msg

[Daniel Jacobowitz]

The following reply was made to PR c++/2274; it has been noted by GNATS.

From: Daniel Jacobowitz <drow@false.org>
To: "C.Gers" <c.gers@Jenoptik-si.de>
Cc: gdb-gnats@sourceware.org
Subject: Re: c++/2274: loading elf generates error msg
Date: Mon, 18 Jun 2007 13:19:14 -0400

 On Mon, Jun 18, 2007 at 06:59:13PM +0200, C.Gers wrote:
 > I think the linker/debugger combination was verry smart because the same stuff worked with
 > gcc 3.4.6 / gdb 6.3. I was able to debug mixed modul code without
 > problems. Now gcc 4.1.1 gdb 6.5 combination is not able to handle
 > the same thing ? I could not remove debug info because i need it to
 > debug my prog !
 
 Then you will have to figure out what is going on.  There is
 absolutely no way that GDB can handle a DWARF section containing both
 big and little endian data; the underlying format is endian-sensitive.
 
 -- 
 Daniel Jacobowitz
 CodeSourcery

Re: c++/2274: loading elf generates error msg

[Daniel Jacobowitz]

The following reply was made to PR c++/2274; it has been noted by GNATS.

From: Daniel Jacobowitz <drow@false.org>
To: c.gers@jenoptik-si.de
Cc: gdb-gnats@sources.redhat.com
Subject: Re: c++/2274: loading elf generates error msg
Date: Mon, 18 Jun 2007 12:08:00 -0400

 On Mon, Jun 18, 2007 at 03:54:55PM -0000, c.gers@jenoptik-si.de wrote:
 > i think this is a problem of mixed little and big endian modules in
 my app. The boot module for the ppc has to be in big endian. After
 init i switch endianess and jum into other modules now compiled as
 little endian.
 
 This is not a GDB bug; I suggest you remove all debug info from the
 little endian files, since your final file is presumably big-endian.
 Otherwise you will need a very smart linker.
 
 -- 
 Daniel Jacobowitz
 CodeSourcery

c++/2274: loading elf generates error msg

[c.gers]

>Number:         2274
>Category:       c++
>Synopsis:       loading elf generates error msg
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          change-request
>Submitter-Id:   net
>Arrival-Date:   Mon Jun 18 15:58:01 UTC 2007
>Closed-Date:
>Last-Modified:
>Originator:     c.gers@jenoptik-si.de
>Release:        unknown-1.0
>Organization:
>Environment:
gcc 4.1.1, binutils 2.16.0 cross development pc as host ppc as target (xilinx environment + tools)
>Description:
loading an elf file causes the message :
Error: Dwarf Error: wrong version in compilation unit header (is 512, should be 2) [in module /cygdrive/x/Projekte/VCI_2_Project/FPGAs/EDK/_BL/BL_PPC_0/executable.elf] 

i think this is a problem of mixed little and big endian modules in my app. The boot module for the ppc has to be in big endian. After init i switch endianess and jum into other modules now compiled as little endian. 

I tried readelf -wi and saw :
The section .debug_info contains:
Compilation Unit @ offset 0x0:
Length 20300443136
Version 512
:
readelf -a brings up
.debug_info offset = 021913
:
Now i tried to repair the fault :
Hexeditor offset 021913
00 00 00 79 00 02 -> changed to 79 00 00 00 02 00
and heureca things work again.

i dont know if this is a problem of gdb or of the compiler/linker so please forward this information if you think other people shoud handle this.

tks
>How-To-Repeat:
Use readelf -wi executable.elf
>Fix:
Hexeditor offset 021913
00 00 00 79 00 02 -> changed to 79 00 00 00 02 00
>Release-Note:
>Audit-Trail:
>Unformatted: