From mboxrd@z Thu Jan 1 00:00:00 1970 From: Christopher Li Subject: Re: How do I get the latest sparse? Date: Tue, 16 Jan 2007 22:31:47 -0800 Message-ID: <20070117063147.GA2077@chrisli.org> References: <20070116001735.GA12002@chrisli.org> <45ACC3DD.6050702@freedesktop.org> <20070117014923.GA641@chrisli.org> <45ADB71D.3080502@freedesktop.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Received: from alnrmhc12.comcast.net ([206.18.177.52]:36085 "EHLO alnrmhc12.comcast.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750976AbXAQGyr (ORCPT ); Wed, 17 Jan 2007 01:54:47 -0500 Content-Disposition: inline In-Reply-To: <45ADB71D.3080502@freedesktop.org> Sender: linux-sparse-owner@vger.kernel.org List-Id: linux-sparse@vger.kernel.org To: Josh Triplett Cc: Linux-Sparse On Tue, Jan 16, 2007 at 09:41:49PM -0800, Josh Triplett wrote: > > Without cross function/file checking, the interrupt checking generates way > > too many false positives. > > Sounds a lot like the context checking. :) That is because the rules of the checking is too restrictive. > The use of strcmp on asm strings for x86 instruction mnemonics seems like a > problem, though. Yes, I agree. The interrupt checking is an example of how the checker can be written. It is not very usable as it is. I have a few idea to address it, but it will not help the usability issue. Cross functions checking is the way to go. Chris