Return-path: Received: from mail-vc0-f174.google.com ([209.85.220.174]:48897 "EHLO mail-vc0-f174.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751114AbaBCUSy (ORCPT ); Mon, 3 Feb 2014 15:18:54 -0500 MIME-Version: 1.0 In-Reply-To: <20140203201228.GE26722@mwanda> References: <52ee2ee736e00_2c3211fc86c5851f@209.249.196.67.mail> <20140202160512.GA4946@redhat.com> <201402021807.37772.s.L-H@gmx.de> <52EFCC66.9020304@lwfinger.net> <20140203201228.GE26722@mwanda> Date: Mon, 3 Feb 2014 12:18:53 -0800 Message-ID: (sfid-20140203_211940_455911_333EF5E9) Subject: Re: rtl8821ae. From: Linus Torvalds To: Dan Carpenter Cc: Larry Finger , Stefan Lippers-Hollmann , Dave Jones , Andrea Merello , Greg Kroah-Hartman , Linux Wireless List , Linux Kernel , Linux Driver Project Content-Type: text/plain; charset=UTF-8 Sender: linux-wireless-owner@vger.kernel.org List-ID: On Mon, Feb 3, 2014 at 12:12 PM, Dan Carpenter wrote: > On Mon, Feb 03, 2014 at 11:05:42AM -0600, Larry Finger wrote: >> A combined driver would require very many branches based on chip >> number and would certainly execute much more slowly. > > I seriously doubt there are performance issues with merging the drivers. Quite frankly, merging drivers can be a f*cking pain. I would seriously suggest avoiding it *unless* the hardware is literally almost identical, or the drivers have been written with multi-chip support from the ground up by people who actually understood the hardware. But the reason isn't performance - it's subtle breakage. Merged drivers have a tendency to break support for chip A when you fix something for chip B. And the end result is a driver that nobody understands, and nobody can sanely test. Sometimes it's simply better to leave old drivers alone. Linus