From mboxrd@z Thu Jan 1 00:00:00 1970 Received: with ECARTIS (v1.0.0; list linux-mips); Thu, 07 Jun 2007 19:11:06 +0100 (BST) Received: from wf1.mips-uk.com ([194.74.144.154]:29905 "EHLO dl5rb.ham-radio-op.net") by ftp.linux-mips.org with ESMTP id S20027233AbXFGSLD (ORCPT ); Thu, 7 Jun 2007 19:11:03 +0100 Received: from denk.linux-mips.net (denk.linux-mips.net [127.0.0.1]) by dl5rb.ham-radio-op.net (8.14.1/8.13.8) with ESMTP id l57I44vN005341; Thu, 7 Jun 2007 19:04:04 +0100 Received: (from ralf@localhost) by denk.linux-mips.net (8.14.1/8.14.1/Submit) id l57I44eA005340; Thu, 7 Jun 2007 19:04:04 +0100 Date: Thu, 7 Jun 2007 19:04:04 +0100 From: Ralf Baechle To: "Maciej W. Rozycki" Cc: linux-mips@linux-mips.org Subject: Re: [PATCH] DECstation: Optimised early printk() Message-ID: <20070607180404.GA3285@linux-mips.org> References: MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.5.14 (2007-02-12) Return-Path: X-Envelope-To: <"|/home/ecartis/ecartis -s linux-mips"> (uid 0) X-Orcpt: rfc822;linux-mips@linux-mips.org Original-Recipient: rfc822;linux-mips@linux-mips.org X-archive-position: 15347 X-ecartis-version: Ecartis v1.0.0 Sender: linux-mips-bounce@linux-mips.org Errors-to: linux-mips-bounce@linux-mips.org X-original-sender: ralf@linux-mips.org Precedence: bulk X-list: linux-mips On Tue, Jun 05, 2007 at 11:45:07AM +0100, Maciej W. Rozycki wrote: > This is an optimised implementation of early printk() for the DECstation. > After the recent conversion to a MIPS-specific generic routine using a > character-by-character output the performance dropped significantly. > This change reverts to the previous speed -- even at 9600 bps of the > serial console the difference is visible with a naked eye; I presume for a > framebuffer it is even worse (it may depend on exactly which one is used > though). > > Additionally the change includes a fix for a problem that the old > implementation had -- the format used would not actually limit the length > of the string output. This new implementation uses a local buffer to deal > with it -- even with this additional copying it is much faster than the > generic function. > > Plus this driver is registered much earlier than the generic one, > allowing one to see critical messages, such as one about an incorrect CPU > setting used, that are produced beforehand. :-) Queued up for 2.6.23, too. > As a side note, the SYS_HAS_EARLY_PRINTK option could probably be called > SYS_HAS_GENERIC_EARLY_PRINTK or something... I take a patch to rename this :-) Ralf