From mboxrd@z Thu Jan 1 00:00:00 1970 From: Linus Torvalds Subject: Re: matroxfb: fix regression with uninitalized fb_info->mm_lock mutex (second head) Date: Tue, 7 Jul 2009 13:30:54 -0700 (PDT) Message-ID: References: <20090707223413.d56c6a7f.krzysztof.h1@poczta.fm> Mime-Version: 1.0 Return-path: In-Reply-To: <20090707223413.d56c6a7f.krzysztof.h1@poczta.fm> Sender: linux-kernel-owner@vger.kernel.org List-ID: Content-Type: TEXT/PLAIN; charset="us-ascii" Content-Transfer-Encoding: 7bit To: Krzysztof Helt Cc: Linux-fbdev-devel , linux-kernel@vger.kernel.org, akpm@linux-foundation.org, a.p.zijlstra@chello.nl, rjw@sisk.pl, stable@kernel.org On Tue, 7 Jul 2009, Krzysztof Helt wrote: > > Remove redundant locking by the mm_lock mutex before a second head of matrox > framebuffer is registered. Why do you write misleading commentary like this. > +/* > + * This function is called before the register_framebuffer so > + * no locking is needed. > + */ Or this? It's not about "needed". The locking is not only not needed, it would be BUGGY. And it's not "redundant". That implies that it's done somewhere else. It's more than "not needed" - it would be actively buggy to lock things there. I really don't like how you're approaching this. You're ignoring the real issues I ask you, you're writing misleading comments and commit messages, and the end result is fragile code. I still don't understand why you insist on initializing those things late, which is the primary problem here. Linus