[CentOS] build Centos source/rpms from scratch...

Sun Aug 20 21:32:58 UTC 2006
bruce <bedouglas at earthlink.net>

jim...

the linux kernel has the ability to be compiled in a 'debug' mode. the
resulting 'app' is used by various external apps to be able to trap/resolve
dependencies at the kernel level to determine why kernel level (kernel
generated) crashes/hangups occur.

if you're building the kernel, i believe the flag '-g' is used in the CFLAGS
attribute during the complile/build process for the kernel.

it also appears that the kernel-debuginfo app is essentially the same as i
just described. (which would explain the large size of the rpm file. it's
the OS with the debugging/symbolic information)

it also appears that you do not need to 'run' the kernel-debuginfo app,
rather you can point to it with the 'crash' app and other apps to get the
symbols resolved while you're debugging the kernel level errors/isssues...

this is also used with the netdump process.

so what you're doing is either rebuilding the kernel, using the '-g'/debug
flag, or you're using the 'kernel-debuginfo' app to handle the symbolic
resoultion for the debugging process.



-----Original Message-----
From: centos-bounces at centos.org [mailto:centos-bounces at centos.org]On
Behalf Of Jim Perrin
Sent: Sunday, August 20, 2006 2:24 PM
To: CentOS mailing list
Subject: Re: [CentOS] build Centos source/rpms from scratch...


Please don't top post in mailing list replies. It's not considered
"proper" behavior.

> not a school project, a machine is having intermittent crashes, and i want
> to know more detailed information.. so i need to have a debug kernel in
> order to keep testing.

What are you calling a debug kernel? There seems to be a
miscommunication or trouble getting the idea of what you're trying to
do across?

> the issue with the kernel-debuginfo, is that it appears to essentially be
> the OS with the debug/symbolic information. you apparently don't actually
> run the rpm/app after it's installed.. but it gives the 'crash' app the
> ability to match/resolve symbolic information in the event of a crash/hang
> within the OS....

How does this differ from your 'debug kernel'?

> you supplied me with info that i had eventually found after hours of
> searching... i had created debug kernel scenarios years ago, but needed to
> be sure things hadn't changed...

I'm still unclear how this relates to the original post of rebuilding
the CentOS distribution, and to some extent I don't overly follow how
changing to a kernel of your own creation will help you troubleshoot a
crash you think is related to the provided kernel.


--
During times of universal deceit, telling the truth becomes a revolutionary
act.
George Orwell
_______________________________________________
CentOS mailing list
CentOS at centos.org
http://lists.centos.org/mailman/listinfo/centos