From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: From: Arnd Bergmann To: cbe-oss-dev@ozlabs.org Subject: Re: [Cbe-oss-dev] [PATCH 4/5] powerpc: export symbols for page size selection Date: Fri, 7 Apr 2006 18:03:01 +0200 References: <20060407150124.422051000@dyn-9-152-242-103.boeblingen.de.ibm.com> <20060407152813.GA382@lst.de> <20060407154247.GJ952@pb15.lixom.net> In-Reply-To: <20060407154247.GJ952@pb15.lixom.net> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Message-Id: <200604071803.01836.arnd.bergmann@de.ibm.com> Cc: linuxppc-dev@ozlabs.org List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , On Friday 07 April 2006 17:42, Olof Johansson wrote: > Yeah, what's the need for that, really? It needs to know so much of > kernel internals that it's getting silly to allow it to be a module. > It's still quite handy to have the actual file system as a module, because a) the module is rather bloated atm, it's about 200kb including the embedded spu code that it runs for context switch. I don't want to burden the generic distros with that b) It simplifies development a bit to not have to reload the whole kernel when making changes. Booting with all the HW debugger stuff enabled takes quite a bit of time, especially when you have to tunnel the binaries through our internal network ;-). We currently have two modules, and the base module has most of the stuff that has dependencies on exported symbols. The binary is 25k on my system (with some debugging options enabled). How about if I do a patch that always includes the base but not the actual file system? Arnd <><