From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from gate.crashing.org (gate.crashing.org [63.228.1.57]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (Client did not present a certificate) by ozlabs.org (Postfix) with ESMTP id D0686DDF44 for ; Tue, 30 Jan 2007 20:11:12 +1100 (EST) Subject: Re: [PATCH 5/9] ps3: Preallocate bootmem memory for ps3fb From: Benjamin Herrenschmidt To: Geert Uytterhoeven In-Reply-To: References: <1170124157.19887.116.camel@concordia.ozlabs.ibm.com> <200701300927.52087.arnd@arndb.de> Content-Type: text/plain Date: Tue, 30 Jan 2007 20:08:36 +1100 Message-Id: <1170148116.26655.328.camel@localhost.localdomain> Mime-Version: 1.0 Cc: James Simmons , linuxppc-dev@ozlabs.org, Paul Mackerras , Linux Frame Buffer Device Development , Arnd Bergmann List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , On Tue, 2007-01-30 at 09:36 +0100, Geert Uytterhoeven wrote: > On Tue, 30 Jan 2007, Arnd Bergmann wrote: > > On Tuesday 30 January 2007 09:16, Geert Uytterhoeven wrote: > > > We have a similar problem with the storage driver, which needs a block of 256 > > > KiB of contiguous memory. > > > > I guess at least that one should be solveable. If the 256kb is only > > internal flash access, it is should not be part of the storage driver > > but of whatever else is used to drive this. > > If we convert the FLASH part to a character device driver, the 256 KiB will be > internal to the character device driver. > > > Assuming the module gets loaded shortly after boot, you can still allocate > > 256kb practically always, using get_free_pages(). > > I'm afraid that assumption is usually false, as you only need to access the > FLASH when you want to change boot flags or update the boot loader. At the same time, flash doesn't have to be fast... can't we use a smaller bounce buffer (a page) and divide the transfer ? Ben.