Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 9 May 2000 13:04:23 +0930
From:      Greg Lehey <grog@lemis.com>
To:        "Jeffrey J. Mountin" <jeff-ml@mountin.net>
Cc:        freebsd-stable@FreeBSD.ORG, freebsd-hackers@FreeBSD.ORG
Subject:   Re: Debugging Kernel/System Crashes, can anyone help??
Message-ID:  <20000509130423.Q75157@freebie.lemis.com>
In-Reply-To: <4.3.2.20000506175851.00ae9b50@207.227.119.2>
References:  <4.3.2.20000503221040.02051d40@207.227.119.2> <20000503185936.E72341@ethereal.net> <20000504095941.B18453@freebie.lemis.com> <200005040124.VAA55655@account.abs.net> <20000504111136.B22025@freebie.lemis.com> <20000503185936.E72341@ethereal.net> <20000504114013.E22025@freebie.lemis.com> <4.3.2.20000503221040.02051d40@207.227.119.2> <20000506073852.E37507@freebie.lemis.com> <4.3.2.20000506175851.00ae9b50@207.227.119.2>

next in thread | previous in thread | raw e-mail | index | archive | help
On Saturday,  6 May 2000 at 18:15:15 -0500, Jeffrey J. Mountin wrote:
> At 07:38 AM 5/6/00 +0930, Greg Lehey wrote:
>> On Thursday,  4 May 2000 at 17:00:35 -0500, Jeffrey J. Mountin wrote:
>>> At 11:40 AM 5/4/00 +0930, Greg Lehey wrote:
>>>> There's a separate issue about whether to build kernels with debug
>>>> symbols by default.  That takes a lot more space (30 MB as compared to
>>>> about 8).  But if you have a debug kernel, I don't see any reason to
>>>> install a stripped version.
>>>
>>> Running the non-stripped kernel uses more memory
>>
>> No, this isn't correct, not with ELF.
>>
>>> and isn't there also a performance issue.
>>
>> No.
>
> Thanks for clarifying that, but were either memory or performance an
> issue back in the AOUT days?  I seem to recall something to that
> point, but might be wrong.

Yes, if you booted an a.out debug kernel, you ended up loading the
entire kernel, including debug symbols, into memory.

>>> Also what is gained by running it,
>>
>> Disk space.
>
> Then it's an issue of disk layout.  I use a 100 MB root, which should be
> good for number of years to come and /var is not a part of the root
> partition.  IMO space is (or should be) a non-issue or at least should be
> for most.  YMMV and don't care to beat that horse some once more.

Sure, that's one of the issues I mentioned.  But I think 100 MB is
fine for a couple of debug kernels.  My main machine (which I don't
use for debugging) has only 30 MB in the root file system.

>>> as long as you have the complimentary debug kernel in /var/crash (or
>>> wherever) for the stripped boot kernel.
>>
>> It's the one that savecore saves, and there's more opportunity for
>> using the wrong debug kernel for dump analysis.
>
> True, but then one would hope someone doing analysis would know
> better.

It's not a question of knowing better, it's a question of
honest-to-god mistakes.  I make enough of them myself.

> Maybe there should be an "make install debug" that would either
> install the debug version instead of the stripped version *or*
> install it in /var/crash.

Indeed, there *is* a make install.debug (note spelling) which installs
the debug version instead of the stripped version.

> If there are no memory or performance issues with the debug kernel,
> then I wouldn't mind it being installed as the default.  Then those
> that wish to strip it may do so.  As always the point is who gains
> changing the current method.

I tried this a while back.  There was a lot of resistance.

Greg
--
Finger grog@lemis.com for PGP public key
See complete headers for address and phone numbers


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-stable" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20000509130423.Q75157>