From mboxrd@z Thu Jan 1 00:00:00 1970 From: erik quanstrom Date: Sun, 5 Sep 2010 01:22:22 -0400 To: bruce.ellis@gmail.com, 9fans@9fans.net Message-ID: <0329becc4ed03493a2467d0c9b02851f@plug.quanstro.net> In-Reply-To: References: <621112A569DAE948AD25CCDCF1C0753329998E@dolly.ntdom.cupdx> <621112A569DAE948AD25CCDCF1C0753329998F@dolly.ntdom.cupdx> <3a082ce7e33442198025b18230b8bba3@plug.quanstro.net> MIME-Version: 1.0 Content-Type: text/plain; charset="US-ASCII" Content-Transfer-Encoding: 7bit Subject: Re: [9fans] latest plan9.iso Topicbox-Message-UUID: 4f8ea368-ead6-11e9-9d60-3106f5b1d025 On Sat Sep 4 22:56:22 EDT 2010, bruce.ellis@gmail.com wrote: > i came across this the other day - i thought i'd finally found a > machine for which P9 supports nothing. > > in fact i beat you - no PCI devices at all. > > any help appreciated. it boots any flavour of windows. new HP Pav quadcore. 9atom has this change to 9load. this fixed up a funky motherboard with ipmi. pci.c:528,535 - /n/sources/plan9/sys/src/boot/pc/pci.c:514,520 * according to the spec. */ n = inl(PciADDR); - - if(!(n & 0x7f000003)){ + if(!(n & 0x7FF00000)){ outl(PciADDR, 0x80000000); outb(PciADDR+3, 0); if(inl(PciADDR) & 0x80000000){ perhaps this will help. if not the value of n would be interesting. in the same vein, i've got several machines in the lab that have pci busses >= 64. this change was all that was required (and setting *pcimaxbno in the .ini file for the kernel) pci.c:37,43 - /n/sources/plan9/sys/src/boot/pc//pci.c:36,42 static Lock pcicfglock; static Lock pcicfginitlock; static int pcicfgmode = -1; - static int pcimaxbno = 255; + static int pcimaxbno = 7; static int pcimaxdno; static Pcidev* pciroot; static Pcidev* pcilist; i'm tempted to set maxbno to 255 in the kernel as well, but i don't think i understand why it was set to less than 255 in the first place. if this is just a speedup, i'd be tempted. anyone with first-hand knowledge? - erik