Subject: Re: PR 31468 -- distributed LKMs are not useful because of incompatible config
To: None <firstname.lastname@example.org>
From: Michael van Elst <email@example.com>
Date: 10/31/2005 12:06:34
firstname.lastname@example.org (Johan Danielsson) writes:
>Why do we have both DIAGNOSTIC and DEBUG if the former isn't as
>low-impact as its description suggests?
low-impact doesn't mean no impact.
DIAGNOSTIC tests are usually fast but still cost performance. You
will notice this in critical code paths, e.g. disk I/O and network I/O,
especially on slower systems.
DEBUG tests can be very costly.
Both options _do_ change data size and layout.
Both options even _do_ change behaviour of the kernel (and I am not
talking about panics...).
Michael van Elst
"A potential Snark may lurk in every tree."