PCI Card Incompatibility Feature Request |
October, 03, 2003 11:39 PM |
joevt |
I put a Radeon 7000 and 9000 in a Power Mac 8600 which also has a Sonnet Trio card. It boots fine into OS 9 but when I try to boot into OS X, I get a CLAIM FAILED Open Firmware error message. I guess the same error would occur if you used two Radeon 7000's instead. I think it would be helpful if XPostFacto could disable selected PCI cards by setting the pci-probe-list mask so that incompatible PCI cards don't need to be physically removed from the computer. The same option would be useful for other combinations of PCI cards (it won't help if the problem is electrical instead of Open Firmware related which might be the case for me). ------ If "CLAIM FAILED" means that there is not enough Open Firmware memory then couldn't XPostFacto be made to increase that somehow (maybe partly by setting "real-size" and/or "virt-size")? |
. |
JoeVT? |
October, 11, 2003 11:16 PM |
lyonsdj88 |
. |
Did you try "fcode-debug? true" long shot but mabe this would let you use 2 pci cards that use compression. fcode-debug? true should free up some ram right? |
. |
RE: PCI Card Incompatibility Feature Request |
October, 09, 2003 6:39 AM |
lyonsdj88 |
. |
No reason to go to any troublr over.;-) |
. |
RE: PCI Card Incompatibility Feature Request |
October, 09, 2003 1:31 AM |
joevt |
. |
You're right. The 8600 doesn't seem to want to boot into either OS X or 9 without the control@B built-in monitor out device being probed. The solution would probably require another OF patch added to the nvram script. Is there any reason why you'd want to disable that device? |
. |
RE: PCI Card Incompatibility Feature Request |
October, 08, 2003 10:10 PM |
lyonsdj88 |
. |
Got it ,thanks, but the bad news is of you set vci0's first bit to 0 then no boot(-2). I'll have to check but I think it has something to do with sixty6(vidoeout) and the fact that it is on grand central. Mybe there is a work around. I did disable planb and that worked fine(-3). Also tried (-8)1000 to disable all the dev's on vci but still no boot. |
. |
RE: PCI Card Incompatibility Feature Request |
October, 08, 2003 2:30 AM |
joevt |
. |
The value used for pci-probe-list is a 32 bit number. Each bit represents a device. Devices that have their bit set to 0 are skipped by Open Firmware. Bit 0 is the right-most bit and represents vci0/ control@B in Open Firmware. Bit 31 is the left-most bit. fffffbf converted to binary is 1111 1111 1111 1111 1111 1111 1011 1111 Each hex digit or group of four binary digits is a different pci bus. The bit in that number (fffffbf) that is set to 0 is bit 6 which represents the device in slot "B1" which, in Open Firmware, would be pci1/@E. As for telnet, I suppose you could write a telnet Open Firmware package like the one in the G4 but you would have to figure out how to get it loaded into Open Firmware. I guess you could put whatever Open Firmware code you like in the rom of a video card... If you have a Mac with USB then you can get a Keyspan serial adapter and use that in OS X. |
. |
RE: PCI Card Incompatibility Feature Request |
October, 07, 2003 1:20 PM |
lyonsdj88 |
. |
Ok, I don't understand how to get the hex value of a device. Joe if you get time could you post the math to get from bit "0: vci0/control@B = monitor out" for the hex value of not probing it. how do you get "fffffbf" to disable slot b? |
. |
RE: PCI Card Incompatibility Feature Request |
October, 07, 2003 1:01 PM |
lyonsdj88 |
. |
Ok got it :sudo nvram pci-probe-list="-65". "0x" is assumed or default now :nvram - p returns pci-probe-list? 0xffffffbf . on reboot the FW card in pci slot be is not probed and not "seen". So, thanks Joe really appriciate your help. I'll see if I can make a list of all the comands to shutdown dev's on an 8600, that way anyone trouble shooting conflicts won't have to open there mac to remove pci cards and the built-in device's can be turned "off" as well. |
. |
RE: PCI Card Incompatibility Feature Request |
October, 07, 2003 8:09 AM |
lyonsdj88 |
. |
I know only new world mac's have the Telnet Package in OF, but can't I load Telnet in to OF after boot? I think sudo nvram pci-probe-list="YOURTEXT" is the propper format, but it may want "YOURTEXT" in dec. or it could be a bug in the terminal. I killed my ADB kbd so only way I can control OF is zterm from another Mac in 9, as I only have one mac(8600) that can run OS X and has the localtalk printer and modem ports. |
. |
RE: PCI Card Incompatibility Feature Request |
October, 07, 2003 2:51 AM |
joevt |
. |
it seems "sudo nvram pci-probe-list" doesn't like "ffffffbf" because it is turning that into 0 which of course will stop all pci slots from being probed. Try a different format, maybe without the quotes, and maybe with a 0x or $ prefix. Maybe the number has to be decimal instead of hex? I'm not running OS X right now so I don't know what the correct format is. Keep trying different formats until "nvram -p" does not say 0 for pci-probe-list. You can run Z-Term in OS X. You can set pci-probe-list to -1 in Open Firmware without zapping the rest of PRAM. Only Open Firmware for newer Macs like the G4 can be controlled over Ethernet. |
. |
RE: PCI Card Incompatibility Feature Request |
October, 05, 2003 11:14 PM |
lyonsdj88 |
. |
no dev's on chaos or bandit in the device tree in X. I'll have to go to 2 mechine mode to check dev / ls in OF as I get no video without VCI or my PCI slots.GC is still there but not control,planb or 106,1 , A1 ,B1 , C1.So if you need me to I can go two mech mode ,but it's a pain to boot 9 on my rig and pull an old mac to run zterm. I wish knew a way to telnet in old world or ztem over ethernet. Do you know if there is a way? To set my pci-probe-list I use terminal in X so I don't have to zap the NVRAM to get it back to "-1" this is the command I use.sudo nvram pci-probe-list="ffffffbf" ,but when I use nvram -p I get pci-probe-list? 0. I'm I doing it correct? |
. |
RE: PCI Card Incompatibility Feature Request |
October, 05, 2003 10:21 PM |
tomquinlan |
. |
Top Slot: Tempo ATA 66 Card Second Slot: USB 1 Third Slot: Firewire 400 Fourth Slot USB 2 Fifth Slot: 64mb Radeon 7000 PC Convert Bottom Slot: 32mb Radeon 7000 Mac Edition Tom |
. |
RE: PCI Card Incompatibility Feature Request |
October, 05, 2003 6:49 PM |
joevt |
. |
Thanks, Tom. Maybe my problem is mostly with the Tempo Trio card (or it could be the Radeon 9000; I should try again with either one removed). Do you have both Radeon 7000's on the same bus? What slots are your 6 PCI cards in? lyonsdj88, what difference do you see when you do dev / ls after you startup into Open Firmware between when pci-probe-list is set to -1 and when it is set to FFFFFFBF (-0x65)? pci106b,1 is associated with pci1 and pci2 (the two bandit devices), not with vci0. 106b is the vendor-id of Apple. I found this URL that lists all the device-ids: http://www.yourvote.com/pci/pciread.asp?venid=0x106b It says that device-id 1 is for the PowerPC Host-PCI Bridge (you were right about it being a bridge). Also the class-code of 60000 means "PCI to Host Bridge". I thought bandit was the host bridge but bandit has no vendor/device id or class-code. |
. |
RE: PCI Card Incompatibility Feature Request |
October, 05, 2003 1:05 PM |
tomquinlan |
. |
Just for the record, I have two Radeon 7000's installed in my 9500 w/Sonnet G4/700. All six PCI slots are filled and working. Entrega 4port USB 1 card; IOGear 2port USB 2 card with 3port connector to front; ADS Tech 2port Firewire card; Sonnet Tempo ATA 66 Controller running 2 IDE drives; 1.09 GB RAM from various vendors. The two Radeons are a Mac Radeon 7000 running two 15inch monitors and a PC 64mb 7000 convert running a 19inch. The only playing around I did had to do with getting the Apple to appear on the main 19inch monitor when booting up. Otherwise, everything (besides sound in) has just worked. Tom |
. |
RE: PCI Card Incompatibility Feature Request |
October, 05, 2003 9:29 AM |
lyonsdj88 |
. |
"To disable slot B1 of pci1 use this Open Firmware command: setenv pci-probe-list FFFFFFBF" This does not seem to work for me on my 8600. Nothing seems to get probed on the pci bus. |
. |
RE: PCI Card Incompatibility Feature Request |
October, 05, 2003 8:55 AM |
lyonsdj88 |
. |
Could be that pci106b,1 is some sort of bridge for vci0 as control@b is pci106b,3 and planb@D is pci106b,4. Not sure what is dev 2 on pci106b. |
. |
RE: PCI Card Incompatibility Feature Request |
October, 05, 2003 4:09 AM |
joevt |
. |
I posted the details (slightly re-worded) in BugZilla. It formats stuff much better. I know that the Sonnet Trio uses some kind of compression (it creates a property, then does something with that like uncompress it I guess, and then calls byte-load on the result). The ATI cards don't do anything like that. But one thing the dual head ATI cards do is encode lots of bytes for the ndrv and then use the same encoded bytes when creating the "driver,AAPL,MacOS,PowerPC" for both video outputs (the two child nodes _A and _B of the video card). The driver is more than 64k is size. One thing that both the Sonnet Trio and ATI dual head cards have in common is that they create child devices but I don't think that would cause the problem. I would try to figure out what exactly is going on but my 8600 usually doesn't want to start up with all three cards installed. |
. |
RE: PCI Card Incompatibility Feature Request |
October, 04, 2003 8:36 PM |
chibi_delenn |
. |
joevt, Your best bet for long tutorials that need to be monospaced properly is to either email a text file as such, or link to a page that can display it properly. You're right, this forum's current software royally blows when it comes to white space and truncation. BG. - Chibi Delennâ„¢ |
. |
RE: PCI Card Incompatibility Feature Request |
October, 04, 2003 8:25 PM |
lyonsdj88 |
. |
You lost me in here some where. That new forum software can't come too soon. I've always wanted to disable my built-in video on my 8600 to see if I could get more throughput across to the Radeon. You and Ryan are also right that this could help XFP and mybe some conflicts with CPU upgrades. So, some thing like Setenv pci-probe- list FFFFFFEF would disable vci0? What other built-in device's could be disabled in this way? BTW You can set pci-probe-list frrom the terminal in OS X. "nvram -p" to see your env. sudo nvram pci-probe-list="-1" replace -1 with your text.JoeVT and Ryan , am sure know this, but others who read this may not. |
. |
RE: PCI Card Incompatibility Feature Request |
October, 04, 2003 7:22 PM |
OSXGuru |
. |
Sorry about the forum software mangling your posts. That's a great idea, and I'd be happy to implement the feature. I've opened a bug-- here's the link--if you could repost some of the technical details there I would be most grateful. Use pci-probe-list to let user disable PCI cards As for the source of the CLAIM FAILED error, I believe the problem relates to OF drivers that use compression. Apparently, there is some kind of bug on the older machines that causes trouble if more than one card does that. My guess is that the machine is not actually running out of memory--it's just a bug. I don't have more details, though, on exactly what the bug is. |
. |
RE: PCI Card Incompatibility Feature Request |
October, 04, 2003 4:57 PM |
joevt |
. |
This is frustrating. I'll skip part 2: For pci1 and pci2, slot B is this weird made-up device (with different reg properties for pci2): /bandit@F2000000/pci106b,1@B PROPERTIES: name pci106b,1 vendor-id 0000106B device-id 00000001 revision-id 00000003 class-code 00060000 min-grant 00000000 max-latency 00000000 devsel-speed 00000001 fast-back-to-back reg 00005800 00000000 00000000 00000000 00000000 PRE> For vci0, the slots are for the following built-in devices: /control@B // monitor out /planb@D // video in
Slot E is I don't know what. Examples: To disable slot E of pci1 use setenv pci-probe-list FFFFFFBF |
. |
RE: PCI Card Incompatibility Feature Request |
October, 04, 2003 4:48 PM |
joevt |
. |
Part 2: The bus number is the three bits 27,26,25 of the bus device unit address: devalias vci0 /chaos@F0000000 // 0xF0000000 BITSHIFTRIGHT 25 BITAND |
. |
RE: PCI Card Incompatibility Feature Request |
October, 04, 2003 4:46 PM |
joevt |
. |
Nope. I hate this forum. Is there a manual for this? In IEEE-1275, pci-probe-list is defined as a string listing the slots in the order they should be probed. On the 9600 and similar Macs, pci-probe-list is a 32 bit number that has the following format: bus number: 7 6 5 4 3 pci2 pci1 vci0 slot unit address: - - - - - - - - - - - - - - - - - - - - F E D B F E D B - E D B Slots that have their bit cleared are not probed. The default value of -1 = FFFFFFFF means everything gets probed. The format does not allow changing the order in which slots are probed. The bus number is the three bits 27,26,25 of the bus device unit address: devalias vci0 /chaos@F0000000 // 0xF0000000 BITSHIFTRIGHT 25 BITAND |
. |
RE: PCI Card Incompatibility Feature Request |
October, 04, 2003 4:42 PM |
joevt |
. |
It keeps truncating in the same spot... Maybe this will work: In IEEE-1275, pci-probe-list is defined as a string listing the slots in the order they should be probed. On the 9600 and similar Macs, pci-probe-list is a 32 bit number that has the following format: bus number: 7 6 5 4 3 pci2 pci1 vci0 slot unit address: - - - - - - - - - - - - - - - - - - - - F E D B F E D B - E D B Slots that have their bit cleared are not probed. The default value of -1 = FFFFFFFF means everything gets probed. The format does not allow changing the order in which slots are probed. The bus number is the three bits 27,26,25 of the bus device unit address: devalias vci0 /chaos@F0000000 // 0xF0000000 SHIFTRIGHT 25 BIT AND |
. |
RE: PCI Card Incompatibility Feature Request |
October, 04, 2003 4:29 PM |
joevt |
. |
One more try: In IEEE-1275, pci-probe-list is defined as a string listing the slots in the order they should be probed. On the 9600 and similar Macs, pci-probe-list is a 32 bit number that has the following format: bus number: 7 6 5 4 3 pci2 pci1 vci0 slot unit address: - - - - - - - - - - - - - - - - - - - - F E D B F E D B - E D B Slots that have their bit cleared are not probed. The default value of -1 = FFFFFFFF means everything gets probed. The format does not allow changing the order in which slots are probed. The bus number is the three bits 27,26,25 of the bus device unit address: devalias vci0 /chaos@F0000000 // 0xF0000000 >> 25 AND |