Date: Fri, 23 Jan 2004 16:13:26 -0500 From: John Baldwin <jhb@FreeBSD.org> To: Peter Jeremy <PeterJeremy@optushome.com.au>, Dag-Erling =?iso-8859-1?q?Sm=F8rgrav?= <des@des.no> Cc: re@freebsd.org Subject: Re: Fix make release for 4-STABLE Message-ID: <200401231613.26241.jhb@FreeBSD.org> In-Reply-To: <20040123205005.GB4759@cirb503493.alcatel.com.au> References: <20040118235148M.matusita@jp.FreeBSD.org> <xzpk73loiip.fsf@dwp.des.no> <20040123205005.GB4759@cirb503493.alcatel.com.au>
next in thread | previous in thread | raw e-mail | index | archive | help
On Friday 23 January 2004 03:50 pm, Peter Jeremy wrote: > On Wed, Jan 21, 2004 at 08:41:50AM +0100, Dag-Erling Sm=F8rgrav wrote: > >John Baldwin <jhb@FreeBSD.org> writes: > >> [...] I have th= is > >> bad feeling that there is some kind of memory corruption bug in the > >> loader and that the problem goes away if you use a loader that has for= th > >> in it. > > I presume this was the trying-to-DMA-across-a-64k-boundary problem > that was fixed recently. Yes. > >why use a loader w/o forth? a complete loader with all 4th and conf > >files takes about 250k, which will definitely fit on the boot floppy, > >and with splitfs we don't need to worry about there being enough room > >for the kernel. > > "goes away" doesn't mean that the problem isn't there. It just means > that there's no obvious problem in John's particular splitfs test case. > A proper fix is far preferable to masking the problem and finding the > bug is far easier when it clearly manifests itself. > > FORTH in the install loader is an interesting question. Do the > benefits of FORTH in the initial install justify its size? (I don't > know the answer to this). I agree that we can do it (with the aid of > splitfs) but we don't want to unnecessarily bloat the install image. > > Firstly, writing and reading floppies is painfully slow, though not > slow enough that you can usefully do something else whilst waiting. > > Secondly, install image bloat translates to a higher minimum RAM > requirement and the floppy install is likely to be mostly used in > older systems which are likely to have less RAM. The last figure I > can remember is 16MB RAM (it doesn't seem to be documented in the > -RELEASE Hardware notes). The current 16MB limit means that it is > non-trivial to install FreeBSD on any normal 386-based system and > increasing it further will start biting 486-based systems. I agree > that these systems are obsolete but they are still useful as SOHO- > grade routers, simple firewalls, printservers etc. The size of the loader doesn't affect how much memory sysinstall needs for = the=20 install, so adding forth support back in won't affect the minimum memory=20 requirement. =2D-=20 John Baldwin <jhb@FreeBSD.org> <>< http://www.FreeBSD.org/~jhb/ "Power Users Use the Power to Serve" =3D http://www.FreeBSD.org
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200401231613.26241.jhb>