On 4/16/08, crash-utility-request@redhat.com <crash-utility-request@redhat.com> wrote:

Send Crash-utility mailing list submissions to
       crash-utility@redhat.com

To subscribe or unsubscribe via the World Wide Web, visit
       https://www.redhat.com/mailman/listinfo/crash-utility
or, via email, send a message with subject or body 'help' to
       crash-utility-request@redhat.com

You can reach the person managing the list at
       crash-utility-owner@redhat.com

When replying, please edit your Subject line so it is more specific
than "Re: Contents of Crash-utility digest..."

Today's Topics:

  1. x86_64 crash (cross-platform) (Cliff Wickman)
  2. Re: x86_64 crash (cross-platform) (Chip Coldwell)

----------------------------------------------------------------------

Message: 1
Date: Tue, 15 Apr 2008 15:34:38 -0500
From: Cliff Wickman <cpw@sgi.com>
Subject: [Crash-utility] x86_64 crash (cross-platform)
To: anderson@redhat.com
Cc: crash-utility@redhat.com
Message-ID: <E1JlrrS-0002i7-Oe@eag09.americas.sgi.com>

Hi Dave,

We've recently seen lcrash fail with a floating point exception
very early in initialization.
Seems to be a binary compiled under one distribution and then
executed on another.

And the solution seems to be the -Wl,--hash-style=both option.

Have others reported this?
There must be a cleaner solution than in my notes below.   That is,
an easier way to link with this option.

Thanks.
-Cliff
--------------------------------------------------------------------------

To make crash so that it runs under SLES or RedHat:
cd /home/estes02/cpw/crash-4.0-6.2
make

Compile it on a SLES system:
good:
cpw@spandau:/tmp/cpw/crash-4.0-6.2> readelf -S crash | grep -A1 hash
[ 4] .hash             HASH             0000000000400290  00000290
      0000000000009490  0000000000000004   A       5     0     8
on a RHEL:
bad:
cpw@alcatraz crash-4.0-6.2 $ readelf -S crash | grep -A1 hash
[ 3] .gnu.hash         GNU_HASH         0000000000400240  00000240
      000000000000a040  0000000000000000   A       4     0     8
OR

vi gdb-6.1/gdb/configure
---
if test "$GCC" = yes; then
   CFLAGS="-g -O2 -Wl,--hash-style=both"
else
   CFLAGS="-g -Wl,--hash-style=both"
fi
else
if test "$GCC" = yes; then
   CFLAGS="-O2 -Wl,--hash-style=both"
else
   CFLAGS="-Wl,--hash-style=both"
---
make
readelf -S crash | grep -A1 hash   (should show GNU form)

cpw@alcatraz crash-4.0-6.2 $ readelf -S crash | grep -A1 hash
[ 3] .hash             HASH             0000000000400240  00000240
      0000000000009490  0000000000000004   A       5     0     8
[ 4] .gnu.hash         GNU_HASH         00000000004096d0  000096d0
      000000000000a040  0000000000000000   A       5     0     8
------------------------------------------------------------------------

------------------------------

Message: 2
Date: Tue, 15 Apr 2008 16:43:30 -0400 (EDT)
From: Chip Coldwell <coldwell@redhat.com>
Subject: Re: [Crash-utility] x86_64 crash (cross-platform)
To: "Discussion list for crash utility usage,   maintenance and
       development" <crash-utility@redhat.com>
Message-ID:
       <alpine.LRH.1.00.0804151642150.32524@bogart.boston.redhat.com>
Content-Type: TEXT/PLAIN; charset=US-ASCII

On Tue, 15 Apr 2008, Cliff Wickman wrote:

>
> Hi Dave,
>
> We've recently seen lcrash fail with a floating point exception
> very early in initialization.
> Seems to be a binary compiled under one distribution and then
> executed on another.
>
> And the solution seems to be the -Wl,--hash-style=both option.
>
> Have others reported this?

This isn't specific to crash; it's the result of a change in the
Linux/glibc ABI.  Specifically, the new ABI has a .hash.gnu section
whereas the old one has .hash.  Your linker option above causes both
sections to be included, so that the old dynamic loader can load a
binary built with a new toolchain.

Chip

--
Charles M. "Chip" Coldwell
Senior Software Engineer
Red Hat, Inc
978-392-2426

GPG ID:  852E052F
GPG FPR: 77E5 2B51 4907 F08A 7E92  DE80 AFA9 9A8F 852E 052F

------------------------------

--
Crash-utility mailing list
Crash-utility@redhat.com
https://www.redhat.com/mailman/listinfo/crash-utility

End of Crash-utility Digest, Vol 31, Issue 10
*********************************************