From mboxrd@z Thu Jan 1 00:00:00 1970 Content-Type: text/plain; charset="iso-8859-1" From: Trevor Woerner Reply-To: ppc339@vtnet.ca To: linuxppc-dev@lists.linuxppc.org Subject: Re: [PATCH] educate users about x86 specific kludges instead of hiding thier ignorance Date: Thu, 15 May 2003 23:10:50 -0400 References: <20030515074214.GA787@plato.local.lan> <20030515152245.A23204@forte.austin.ibm.com> In-Reply-To: <20030515152245.A23204@forte.austin.ibm.com> MIME-Version: 1.0 Message-Id: <200305152310.50885.ppc339@vtnet.ca> Sender: owner-linuxppc-dev@lists.linuxppc.org List-Id: On May 15, 2003 04:22 pm, linas@austin.ibm.com wrote: > On Thu, May 15, 2003 at 02:11:42PM -0500, Tony 'Nicoya' Mantler wrote: > > Speaking of which, is it even possible these days to build a kernel > > small enough for (non-b) zImage on x86? I think it's been a while > > since I've even made one that can fit on a floppy. > > If you put most things into modules, then yes, its still very > possible. This might be hard to believe, but when I had to make small x86 kernels (everything including compressed filesystem had to fit in 2MB) for a little embedded device I was working with, I could usually make smaller kernels by *not* turning on module support and simply including the couple of modules I needed directly in the kernel. Being an embedded device I didn't need very many extras. Enabling module support inflates the kernel quite a bit, not to mention all the extra space for the modules themselves and the modutils. But yes, if you are trying to create a general kernel for a range of devices (instead of one specific device) that have a lot of extras, then building modules does keep the kernel size down. I'm just trying to point out that there is a point where it makes more sense to compile-in, rather than create modules, when trying to create small kernels. Trevor ** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/