From mboxrd@z Thu Jan 1 00:00:00 1970 From: David Howells Subject: Re: linux-next: Tree for December 18 (fscache) Date: Thu, 18 Dec 2008 17:47:56 +0000 Message-ID: <6448.1229622476@redhat.com> References: <20081218092557.b274e9dd.randy.dunlap@oracle.com> <20081218230637.c879a046.sfr@canb.auug.org.au> Return-path: Received: from mx2.redhat.com ([66.187.237.31]:46325 "EHLO mx2.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751173AbYLRRsD (ORCPT ); Thu, 18 Dec 2008 12:48:03 -0500 In-Reply-To: <20081218092557.b274e9dd.randy.dunlap@oracle.com> Sender: linux-next-owner@vger.kernel.org List-ID: To: Randy Dunlap Cc: dhowells@redhat.com, Stephen Rothwell , linux-next@vger.kernel.org, LKML Randy Dunlap wrote: > Should include/linux/buffer_head.h have an empty stub for fsync_super() > or does fscache even make sense when CONFIG_BLOCK=n? FS-Cache might. CacheFiles probably doesn't. There've been discussions about a separate caching backend to deal with non-rotating media such as large chunks of battery-backed RAM or flash. FS-Cache might make sense in such a situation as these could be accessed in other ways (such as through MTD or even directly). I'll make CONFIG_CACHEFILES dependent on CONFIG_BLOCK. David