Challenging things to do: SIGSEGV catcher and backtrace extractor
A few moons ago we have discussed on the dev list a tool for automatic segfault detection (including multiple segfaults during 'make test') and core backtrace generation. I'm quite frankly tired of explaining again and again that we need a core file backtrace, how to get the core file dumped and generating trace (no offense, just facts). I'm sure there is a better way to do. If we can automate this process, like many other apps do (e.g. gnome/kde apps) this will save us a lot of traffic on the list and frustration on users' behalf and make us all happier. I've planned on working on this idea for a long time, but other things seem to get on my way. (like helping getting the core dumped :) Any takers? If you are interested, here is some more info: There are two situations when we want this. 1. during 'make test'. which is sort of very easy, because the new test suite already scans for core files. And there is a perl module on CPAN that automatically extracts the backtrace from a given core file (which almost works). The catch is that there can be more than one core dumped and they all be overwritten by the last core dumped. Therefore it's better to install a SIGSEGV to do the work. You will probably need to do it in XS (could write a prototype with Inline). I think I've tried in Perl and it didn't work (since SIGSEGV is caught on the C level by httpd). If this works, you will be able to generate as many backtraces as the number of segfaults 2. during normal use. Those users who get segfaults will load something like Apache::SEGV2BT and it'll write a backtrace every time segfault happens. Therefore it's most likely case 2 which can be used as case 1. For generating segfaults for testing, see my Bad::Segv module, updated by Kevin A. McGrail at http://www.peregrinehw.com/downloads/apache/mod_perl and discussed here: http://perl.apache.org/preview/modperl-docs/dst_html/docs/2.0/devel/debug_c/debug_c.html#Getting_the_core_File_Dumped p.s. I'm talking about mod_perl 2.0 mainly, but it probably should be easy to make it work for 1.x as well once it works for 2.0. __ Stas BekmanJAm_pH -- Just Another mod_perl Hacker http://stason.org/ mod_perl Guide --- http://perl.apache.org mailto:[EMAIL PROTECTED] http://use.perl.org http://apacheweek.com http://modperlbook.org http://apache.org http://ticketmaster.com
Re: Challenging things to do: SIGSEGV catcher and backtrace extractor
On Fri, 12 Apr 2002, Stas Bekman wrote: If you read the rest of the post I mention it (without telling the name :). The problem with this module is that it's useful only after you have the core file. which is not good, because (as I've already explained): it's important to mention Devel::CoreStack, as it is a good starting point. 1. Many users have problems getting the core file dumped then there'd be no way to automate generating a stacktrace anyhow. 2. There can be multiply segfaults with different causes which will overwrite each other, so we want to catch SEGVs as they happen. that's ok, we'll deal with one at time. Not talking about the fact that this module is not slick, e.g. you need manual interaction to help it get to the trace. (it shows the gdb's *more* pager for long output of loading symbols). don't have to use the module as-is, but there is plenty of logic in there that can be borrowed, rather than figuring out everything from scratch.
Re: Challenging things to do: SIGSEGV catcher and backtrace extractor
Doug MacEachern wrote: On Fri, 12 Apr 2002, Stas Bekman wrote: If you read the rest of the post I mention it (without telling the name :). The problem with this module is that it's useful only after you have the core file. which is not good, because (as I've already explained): it's important to mention Devel::CoreStack, as it is a good starting point. true, as it has the config for a few known debuggers, but otherwise it just invokes the debugger and pipes the 'bt' or equivalent command to it and grabs the output. 1. Many users have problems getting the core file dumped then there'd be no way to automate generating a stacktrace anyhow. You can get a backtrace if you run the process under debugger without dumping a core file. No special setup required. I was thinking to attach the debugger on SIGSEGV event. Is it too late? I see certain gnome apps failing and they ask you if you want to get the stack, without me doing anything at all. That's what I want for modperl. You say it's not possible? __ Stas BekmanJAm_pH -- Just Another mod_perl Hacker http://stason.org/ mod_perl Guide --- http://perl.apache.org mailto:[EMAIL PROTECTED] http://use.perl.org http://apacheweek.com http://modperlbook.org http://apache.org http://ticketmaster.com
Re: Challenging things to do: SIGSEGV catcher and backtrace extractor
On Fri, 12 Apr 2002, Stas Bekman wrote: You can get a backtrace if you run the process under debugger without dumping a core file. No special setup required. I was thinking to attach the debugger on SIGSEGV event. Is it too late? I see certain gnome apps failing and they ask you if you want to get the stack, without me doing anything at all. That's what I want for modperl. You say it's not possible? anything is possible of course. but then you have to run the tests with httpd running under gdb, not something that should be done by default. maybe you don't need gdb either, i dunno, if gnome has a trick up its sleeve, might be worth looking at.
Challenging things to do: SIGSEGV catcher and backtrace extractor
Date: Fri, 12 Apr 2002 00:24:11 +0800 To: [EMAIL PROTECTED] From: Stas Bekman [EMAIL PROTECTED] Subject: Challenging things to do: SIGSEGV catcher and backtrace extractor Message-ID: [EMAIL PROTECTED] A few moons ago we have discussed on the dev list a tool for automatic segfault detection (including multiple segfaults during 'make test') and core backtrace generation. I'm quite frankly tired of explaining again and again that we need a core file backtrace, how to get the core file dumped and generating trace (no offense, just facts). I'm sure there is a better way to do. If we can automate this process, like many other apps do (e.g. gnome/kde apps) this will save us a lot of traffic on the list and frustration on users' behalf and make us all happier. While I can't step up to the plate to help on development because I am still working on GETTING a core file, I can second the motion. The frustration in getting a good working core file to trace has been enormous and I am neither a newbie nor someone who gives up easily. Regards, KAM
Re: Challenging things to do: SIGSEGV catcher and backtrace extractor
On Fri, Apr 12, 2002 at 02:09:44AM +0800, Stas Bekman wrote: Doug MacEachern wrote: On Fri, 12 Apr 2002, Stas Bekman wrote: If you read the rest of the post I mention it (without telling the name :). The problem with this module is that it's useful only after you have the core file. which is not good, because (as I've already explained): it's important to mention Devel::CoreStack, as it is a good starting point. true, as it has the config for a few known debuggers, but otherwise it just invokes the debugger and pipes the 'bt' or equivalent command to it and grabs the output. 1. Many users have problems getting the core file dumped then there'd be no way to automate generating a stacktrace anyhow. You can get a backtrace if you run the process under debugger without dumping a core file. No special setup required. I was thinking to attach the debugger on SIGSEGV event. Is it too late? I see certain gnome apps failing and they ask you if you want to get the stack, without me doing anything at all. That's what I want for modperl. You say it's not possible? Sure, that's possible from the SEGV handler. -- Daniel Jacobowitz Carnegie Mellon University MontaVista Software Debian GNU/Linux Developer
