linux-mips
[Top] [All Lists]

Re: Backtrace capability

To: Sudeep Kottilingal <skottilingal@kinetowireless.com>
Subject: Re: Backtrace capability
From: David Daney <ddaney@avtrex.com>
Date: Tue, 30 Sep 2003 12:14:55 -0700
Cc: linux-mips@linux-mips.org, Joe Baranowski <jbaranowski@kinetowireless.com>
In-reply-to: <2EEAE99C5AA0B14BB6A0BBB4ABF8D1E117960E@blunote.bluzona.com>
Original-recipient: rfc822;linux-mips@linux-mips.org
References: <2EEAE99C5AA0B14BB6A0BBB4ABF8D1E117960E@blunote.bluzona.com>
Sender: linux-mips-bounce@linux-mips.org
User-agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.2.1) Gecko/20021130
Sudeep Kottilingal wrote:

Hi all,

I am trying to debug some SIGSEGV when I leave my system ON for a few days. At the time of the

Crash I am not able to get a nice back-trace/stack trace yet. (I do not see the execinfo.h file on my uclibc toolchain

Which implies its not supported yet on my toolchain release).

*Has anyone had success in incorporating backtrace capability into a uclibc binary yet …?*

*If your answer is yes, could u give me some pointers on how to do it and set it up ..?*

*If your answer is no is there any other alternative set up to get more information.*

I do not run an ICD or ICE on my board. Neither do I have a serial console. My console is redirected via Telnet.

Currently I am running the application which crashes on the linux PC through totalview debugger.

Waiting for it to crash.

sudeep

I hacked together a backtrace that you may be able to adapt. It was originally for gcj, but we use several variations for other things as well. It kind of relies on an o32 ABI calling convention.

The code is in this mail message:

http://gcc.gnu.org/ml/java-patches/2003-q3/msg00584.html

Look at:

libjava/sysdep/mips/mipsel-backtracer.c
libjava/sysdep/mips/mipsel-backtracer.h
libjava/sysdep/mips/mipsel-bthelper.S

David Daney





<Prev in Thread] Current Thread [Next in Thread>