Return to BSD News archive
Path: euryale.cc.adfa.oz.au!newshost.carno.net.au!harbinger.cc.monash.edu.au!munnari.OZ.AU!news.ecn.uoknor.edu!feed1.news.erols.com!howland.erols.net!bloom-beacon.mit.edu!senator-bedfellow.mit.edu!usenet From: ghudson@mit.edu (Greg Hudson) Newsgroups: comp.unix.bsd.netbsd.misc Subject: Re: NetBSD/i386 1.2, 128MB, "mp_map full", help? Date: 13 Jan 1997 04:41:22 -0500 Organization: Massachusetts Institute of Technology Lines: 23 Sender: ghudson@the-light-fantastic.MIT.EDU Message-ID: <x7d3ew552v1.fsf@the-light-fantastic.MIT.EDU> References: <VIXIE.97Jan12223511@wisdom.vix.com> NNTP-Posting-Host: the-light-fantastic.mit.edu In-reply-to: vixie@vix.com's message of 13 Jan 1997 06:35:11 GMT X-Newsreader: Gnus v5.1 Xref: euryale.cc.adfa.oz.au comp.unix.bsd.netbsd.misc:5111 > panic: ptdi 1ba063 > panic: malloc: out of space in kmem_map So, I'm not totally confident of my answer here, but on a 128MB machine, it's fairly likely that you need to adjust the value of NKPDE in arch/i386/include/pmap.h. You can't use a kernel config option. In 1.2, NKPDE is set to a fixed value of 12, which is generally sufficient for up to about 32MB of RAM (given the MAXBSIZE value of 16K in 1.2; it's gone up to MAXPHYS or 64K in -current). In -current, NKPDE is computed automatically using the formula: min(NKPDE_MAX, NKPDE_BASE + (biosextmem >> 10) * NKPDE_SCALE) where NKPDE_MAX is 31, NKPDE_BASE is 4, and NKPDE_SCALE is 1. There's nothing magic about this formula; it's just something Jason and Chris thought would work in the large majority of situations. So on a 128MB machine, I think the calculated value of nkpde comes out to min(31, 4 + 128), or 31. So I would try setting NKPDE to 31 for starters. We would be interested in knowing if you need to set the value higher in order to avoid crashes. I believe each increment of NKPDE will cost you 4K of wired memory.