Re: [PATCH 02/16] ehca: module infrastructure
From: Jörn Engel <hidden>
Date: 2006-04-27 12:09:55
Also in:
lkml
From: Jörn Engel <hidden>
Date: 2006-04-27 12:09:55
Also in:
lkml
On Thu, 27 April 2006 12:48:05 +0200, Heiko J Schick wrote:
+ if (ehca_module->cache_pd == NULL) {Hmm.
+ ret = kmem_cache_destroy(ehca_module->cache_pd); + if (ret != 0)
The " != 0" is completely superfluous. Above NULL check is a matter of taste, this one demonstates lack of boolean algebra understanding.
+ rblock = kzalloc(H_CB_ALIGNMENT, GFP_KERNEL);
+ if (!rblock) {Hmm. And your taste seems to change. :)
+ if (ehca_hw_level == 0) {And since we're on the subject. Ignoring the recent discussion involving akpm, viro and others, the kernel historically used int both for integer and boolean, plus return values as a special kind of "boolean with error indication attached". For boolean, it is nicer to do things like "if (!error)", for integers, a comparison as above is nicer. Return codes fall into the boolean category. Pointers after kmalloc() and similar are viewed as rich boolean by some people, but not by all. Jörn -- Geld macht nicht glücklich. Glück macht nicht satt.