Subject: Re: kern/13232: POOL_DIAGNOSTIC && pool.h
To: Assar Westerlund <assar@netbsd.org>
From: Rafal Boni <rafal@mediaone.net>
List: tech-kern
Date: 06/19/2001 12:24:17
In message <5l66ds4fln.fsf@assaris.sics.se>, you write:
-> Rafal Boni <rafal@mediaone.net> writes:
-> > This makes *everyone* pay the
-> > price (even in totally production-tuned configurations), and as such
-> > seems wrong-headed.
->
-> Um, no.
->
-> It used to be that without POOL_DIAGNOSTIC you directly called
-> pool_get, pool_put, and pool_reclaim. With POOL_DIAGNOSTIC, macros
-> were added that sent in __FILE__ and __LINE__ as separate parameters
-> to the functions (now with an underscore in front). My patch made
-> that always be the case. If POOL_DIAGNOSTIC is not defined, these
-> just get ignored. This will keep the API the same, and hide the fact
-> that diagnostic checks are (or not) performed.
Mmm. Maybe I should drink coffee before I read patches, or read the
sources the patch refers to... 8-)
Ah, I see.. I had read your patch w/out referring to the source (I
thought I recalled what it did 8-) and missed the fact that the
pr_enter/pr_leave/... are all defined to do nothing in the non-
POOL_DIAGNOSTIC case.
Never mind, carry on... Sorry for the hassle.
--rafal
----
Rafal Boni rafal@mediaone.net