[CentOS] How do I debug the kernel?

Tue Mar 6 21:48:53 UTC 2007
Rodrigo Barbosa <rodrigob at darkover.org>

Hash: SHA1

On Tue, Mar 06, 2007 at 04:38:21PM -0500, James Olin Oden wrote:
> On 3/6/07, Rodrigo Barbosa <rodrigob at darkover.org> wrote:
> >> >Humm, not that I know of. And considering we are talking about PC 
> >hardware
> >> >here (are we?), I can't imagine how it can be possible.
> >> I'm definately no expert here, but PC hardware has many hardware level
> >> debuging hooks that apps such as oprofile take advantage of.
> >
> >Sure does, but how does one recover from a hardware lockup ?
> >
> Well say the PCI buses timing gets hosed and the whole system seizes
> up what do you intened to do at that point (that is a hardware
> lockup)?  And what does that thing you do have to do with debuging a
> kernel?  Just curious.

A hardware lockup can be induced by software (quite easily, actually).
On a 2 computers debug environment, you can see it happening and get
enough data to fix it.

> >Even if it were possible to have that kind of debugging, it would be
> >fairly limited.
> When you say "hardware" I start thinking jtag interfaces, or even more
> interesting was this thing we would place every pin of the processor
> through and then connect to the socket on the motherboard, but I think
> I missing something.

Oh, I remember that one. SoftICE it was called, I think. I'm not sure.
That was many years ago. Yes, that is one option but, if I remember correctly,
that debugging hardware had a small processor of its own, completely

And yes, jtag is also quite nice, but I never saw anything like that on a 
PC. Used them on some MIPS boards.


- -- 
Rodrigo Barbosa
"Quid quid Latine dictum sit, altum viditur"
"Be excellent to each other ..." - Bill & Ted (Wyld Stallyns)

Version: GnuPG v1.4.5 (GNU/Linux)