From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from rcmenter.iserver.net ([192.41.35.12]) by pentafluge.infradead.org with esmtp (Exim 3.22 #1 (Red Hat Linux)) id 1583Ny-0007l6-00 for ; Thu, 07 Jun 2001 18:15:54 +0100 Subject: Re: HELP From: Russ Dill To: MTD for Linux In-Reply-To: <23825.991907052@redhat.com> References: <20010607114319.A23641@crystal.2d3d.co.za> <20010607110359.A22278@crystal.2d3d.co.za> <19468.991906514@redhat.com> <23825.991907052@redhat.com> Content-Type: text/plain Date: 07 Jun 2001 10:21:06 -0700 Message-Id: <991934470.27485.0.camel@timmy> Mime-Version: 1.0 Sender: linux-mtd-admin@lists.infradead.org Errors-To: linux-mtd-admin@lists.infradead.org List-Help: List-Post: List-Subscribe: , List-Id: Linux MTD discussion mailing list List-Unsubscribe: , List-Archive: On 07 Jun 2001 10:44:12 +0100, David Woodhouse wrote: > > abraham@2d3d.co.za said: > > why is this necessary though? The kernel docs make it pretty clear > > that enabling that causes severe performance penalties which I wanted > > to avoid. > > Then fix the kernel docs. :) I already tried feeding rmk such a patch. The unaligned handly only impacts performance when there is an unaligned read. So for you, its good you now know and are working around this. However, some drivers *cough*jffs2*cough* will make unaligned reads from time to time, but only rarely. In this case, the handler speeds things up by doing the access, instead of producing an oops.