New(?) Radeon 7000 problem |
October, 10, 2003 6:17 AM |
screwj |
Looks like I've got yet another R7000 thread to add. I know these things keep popping up but I don't remember seeing this particular variant anywhere, though I may have missed it due to the "nature" of this forum. Anyway, the problem is on a Beige G3. I finally managed to get the OF output to the monitor attached to the Radeon (a combination of patched firmware, xpostfacto and nuking the nvram), the problem is, that that's only when I boot to the OF prompt. If I then procede to boot in to OS X, the screen goes blank up to the blue screen. Furthermore, logging into console seems to be putting the output onto another (nonexistant) screen (though blindly typing works), and when going to the Display preference pane, the main window collapses to just a titlebar and a second window pops up for the screen (never seen it do that before). So, anybody got any ideas? I'm sort of at my wits end with this one (took me ages jut to sort out the OF issue). BTW, no other display or adapters are connected. BeigeG3, G4 600Mhz, 512MB ram, R7000, ACard ATA66, USB2+FW combo, 3HDs, stock CDRom + LiteOn burner |
. |
RE: New(?) Radeon 7000 problem |
October, 14, 2003 1:04 PM |
screwj |
. |
TV out worked fine for me. I was using it as part of a triple display setup at one time (totally NEAT). Having iTunes visuals pulsating out of a bigass TV is one mesmerizing experience! (funny thing is, at one early stage in the lifespan of OS X I also managed to get a display off the bulit-in AV card on the tele, but it was "removed" from later versions, anyone know anything about this?) Anyway, if you need me to elaborate on anything I've posted just let me know. I'm sure the others would be willing to as well. Just a quick question: have you tried all the R7000 fixes posted in older threads? There were quite a few, and they seemed to work for a lot of people. Maybe this topic needs it's own page or an FAQ on the XPF site |
. |
RE: New(?) Radeon 7000 problem |
October, 14, 2003 12:54 PM |
asarath |
. |
Have the same problem on my 8600. Right now I use an ancient 15" VGA monitor on the on-board controller, but that is a workaround of a strange sort :-). Anybody have any luck with the TV out on the 7000? Each time I try it, it gives me an error message. I wish I had the understanding of software to try the sugested fix Thanks Ashok |
. |
RE: screwj |
October, 13, 2003 1:41 PM |
jseibyl |
. |
Not on my macs. My main video machine is a dual proc windoze with 4 monitors.....don't worry I am saving for the dual g5 to replace it! However, I loaned my external scsi drive to a friend with the backup of my 10.2.6, just to see what would happen, he has an 8500 g4 700 radeon 7000 and onboard vram. I built out that machine. He has three monitors, two on the radeon, one on-board which he added after I brought it up to 9.2.2 and sold it to him..... He tells me it fired right up into jag WITH all three monitors, he had some issues with setting res in X on the other monitors, but he worked it out (he has all vga monitors). The interesting thing is that that install was done on my original 8500(another one) that had the radeon as the main output device and NO VRAM on the board.....clue?? With my beige g3, I had to instal jag to the onboard video, as at the time, I had an old radius vid card not supported. I did the install fine, and when I finally got a Radeon for that machine, it just worked. Not sure if you can gleen anything out of my experiences, I wish I could have a couple of monitors on my mac, just dont have the space.....email and I will send a pic of my confuser rm if u want to see it.... Jim |
. |
RE: New(?) Radeon 7000 problem |
October, 13, 2003 1:25 PM |
lyonsdj88 |
. |
yes, each "f" in ffffffff is a bus, but not all are used. That's 8 bus's. JoeVT asked Ryan to add the feature to XPF so there is a bug report.lyonsdj88@earthlink.net. Thanks , just send it to me when ever you get time ,no rush. If I disable the first bit on my first bus(vci0) 1110 or fffffffe=no boot for me too, sucks that's my built-in video, but I can boot if I probe nothing (pci-probe-list="0") so I'll have to play with it. |
. |
RE: New(?) Radeon 7000 problem |
October, 13, 2003 1:14 PM |
screwj |
. |
jseibyl, did you ever use multiple monitors? Maybe it has something to do with OS X not wanting to revert the main display for booting, dunno. lyonsdj88, that explains why it didn't work first time then. I was using 9 with Boot Variables at the time to test the various pci settings and used XPF to get back into X. Could one class this as a bug.? Btw, I used standard pci-probe-list. It's weird though, the first 3 busses of the mask seem completely unused, but setting it to 0 stopped the mac from booting, otherwise I'd have thought it wasn't being used. Btw, where do I get your email addy? |
. |
RE: New(?) Radeon 7000 problem |
October, 13, 2003 11:27 AM |
lyonsdj88 |
. |
Good deal, good deal. If you look at the code I posted you'll see that once you disabled the biult-in video bootx passed the "probe for video" Arg. to Mach_Kernel. now when Ryan comes back he can most likely point you at how to "word" the Arg. Boot_Video=?. BTW XPF sets pci-probe to defalt so if you use it I'll have to auto-boot? false. and set you pci-probe to what ever and reset-all. One ? did your Biege use pci- probe-list or pci-probe-mask? If you ever get a chance make a list of the "ffffffff" hex to disable each device and E-mail it to me, as I'm going to post a data-base of the info for Old-world, at some point. |
. |
RE: New(?) Radeon 7000 problem |
October, 13, 2003 10:40 AM |
screwj |
. |
Oh, one more thing for future reference: On my Beige (rev C desktop) the PCI mask seems to be as follows: FFFXFFFF for built in items, of which 0100 is video, I think 0011 didn't do anything, so I assume 1000 would be mac-io. FFFFXFFF for PCI slots, 0001 unused, 0010 for slot A1 (nearest CPU) to 1000 for slot C1 FFFFF000 seemed to be reduntant, haven't tried the highest three banks(?) Hope this is of some use to someone |
. |
RE: New(?) Radeon 7000 problem |
October, 13, 2003 10:38 AM |
jseibyl |
. |
BTW, I never ran into this trouble with my beige g3 rev B, I yanked the onboard vram as I needed it it for another machine when I put the radeon in, but that shouldn't have made any difference with 2 megs still on the board..... |
. |
RE: New(?) Radeon 7000 problem |
October, 13, 2003 10:34 AM |
screwj |
. |
Just thought I ought to mention several things I did between the first disabling the onboard video and my eventual success... First, I used XPF3a4 to start into X, no success, same problem etc. Another thing I did was to look at WindowManager preferences in /Library, this keeps a list of all the display configs you've used. Funny thing was, despite the Rage Pro not being used since last time I deleted this file, it was still listed, so I deleted it again. Then, tried several boot-args (so restarting several times). At the time I redid the pci-probe-list, boot-args was set to "stdout=devicename" (one of the vars in /chosen OF device), not sure if that helped it find the right output for subsequent launches, but it still worked after disabling the boot-args. Another thing that might have helped: I think some corruption may have caused the caches to be rebuilt on restart, which could also have changed something. |
. |
RE: New(?) Radeon 7000 problem |
October, 13, 2003 10:18 AM |
jseibyl |
. |
cool! I printed this one out, as there is some good info for the future here, and I am limited in experience on the OF side. You guys are good.... Jim |
. |
SUCCESS! |
October, 13, 2003 10:09 AM |
screwj |
. |
Well, looks like I've finally succeeded, though I'm not quite sure what did it. I was messing around with passing different arguments to the kernel, when I noticed in OF that pci-probe-list had been reset to -1, not sure what did it though. resetting it to FFFBFFFF disabled the onboard video and this time it worked! Could XPostFacto or OS X have changed my OF settings? lyonsdj88, it seems the OF variable boot-args get passed to the kernel, so maybe a setting of Boot_Video=devicename would do the trick. Would definitely be the more elegant solution than disabling the video. Anyway, thx to all that helped! I'll keep looking for a solution which works without messing with pci-probe-list, but for now I'm pleased with what I've got! |
. |
RE: New(?) Radeon 7000 problem |
October, 13, 2003 9:59 AM |
lyonsdj88 |
. |
That is odd, so I got the bootx source code from Apple. Looking at it now it seems there is a way to pass a video arg. to the Kernal "Boot_Video", but by defalt it should find that radeon if the built-in is disabled.This code:// Video information.. struct Boot_Video { unsigned long v_baseAddr; /* Base address of video memory */ unsigned long v_display; /* Display Code (if Applicable */ unsigned long v_rowBytes; /* Number of bytes per pixel row */ unsigned long v_width; /* Width */ unsigned long v_height; /* Height */ unsigned long v_depth; /* Pixel Depth */ }; typedef struct Boot_Video Boot_Video, *Boot_Video_Ptr; I'll look at it and see if I can tell you how to hand a agr. like "Boot_video ATY,PARENT". |
. |
RE: New(?) Radeon 7000 problem |
October, 13, 2003 7:27 AM |
screwj |
. |
Right, I managed to disable onboard video with pci-probe-list, however, that didn't make any difference (I checked in OF to make sure it was disabled). So, it must be some stored setting that OS X insists on using. I'm not sure whether it's a result of using xpostfacto or whether a clean OS install would fix it, I really am absolutely clueless as to what's going on. I would welcome ANY ideas (sacrificing a chicken ad midnight didn't work either ;-) )! |
. |
RE: New(?) Radeon 7000 problem |
October, 12, 2003 1:20 PM |
screwj |
. |
ok, I've just been and had a look at OF and it seems that screen is set to the internal video, which is the default What was weird though was output-device. In OS X's nvram utility it's shown as "pci/ ATY,RV100Parent@D/ATY,RV100ad_A", however, I just noticed that in OF it's defined as "pci/ATY,RV100Parent@D/ATY,RV100ad_A screen", could it be that this is passing the screen value to the boot process? I tried changing both the screen alias and the output-device variable, but the settings won't stick (output-device immediately adds the screen on to the end, the alias doesn't seem to survive a restart). Btw, I forgot to check the OF version number, but I think it is around 2.0. iirc 2.4 and up appeared with the new-world machines. |
. |
RE: New(?) Radeon 7000 problem |
October, 12, 2003 9:18 AM |
lyonsdj88 |
. |
Maybe, pci-probe-mask? ,also as his Beige is Rev.C could be OF 2.5 .I don't have a Beige so it's just a shot in the dark. |
. |
RE: New(?) Radeon 7000 problem |
October, 12, 2003 4:04 AM |
joevt |
. |
I don't think pci-probe-list is used in the Beige G3. I looked at the Open Firmware part of the Beige G3 ROM (Open Firmware, 2.0f1) and can't find it being used in there. I guess the best way to find out would be to set it to 0 (zero) and see if anything actually changes in the device list when you restart into Open Firmware. Beige G3's have a "screen" devalias. Maybe that has something to do with your problem? Use the "devalias" Open Firmware command to dump the list of device aliases for the G3. What is the screen alias set to? If you do a "printenv" command, what is the output-device set to? Remember, when switching to different versions of XPostFacto, you should use the "Reinstall BootX" command just to make sure that the version of BootX that you want to use is actually going to get used (at least for trouble-shooting purposes - XPostFacto might be smart enough to know that BootX needs to be replaced, but I don't remember). |
. |
RE: New(?) Radeon 7000 problem |
October, 11, 2003 5:47 PM |
screwj |
. |
lyonsdj88, thanks for the long answer, I'll be looking into it tomorrow. As far as I can tell the beige just has a single bus called pci, with the PCI slots and the built-in Video attached. The three slots are A1 to C1, the onboard Video is listed as F1 (this is taken from IORegistryExplorer), there is also an item refered to as mac-io, however, I believe that also belongs to the card in C1, a USB/Firewire card referred to as pci- bridge. Is this possible? This would leave the on-board video to be the fourth digit, so I'd need 0111, yes? joevt, I've tried all three. I used your patched 2.2.5 to get the OF prompt onto the R7k, it seems though that 3.0a4 won't actually write to nvram for some reason, it allows me to change the settings, but nvram doesn't get updated. this is a problem I had with 2.2.5 and a few other OF variable programs (System Disk for example). Clearing the nvram properly helps, maybe I should try it again this time using 3.0a4 straight after initing the nvram? |
. |
RE: New(?) Radeon 7000 problem |
October, 11, 2003 3:39 PM |
joevt |
. |
What version of XPostFacto are you using, 2.2.5, 3.0a4, or my patched version of 2.2.5? You should try 3.0a4 and if that doesn't work, you can try my patched version. If you don't have that then I can send it to you if you e-mail me. |
. |
RE: New(?) Radeon 7000 problem |
October, 11, 2003 3:16 PM |
lyonsdj88 |
. |
OF comands: dev pci1(return) ls(return) dev vci0(return) ls(return) or at boot dev / ls to see all device's in your system. Also to get a list of pci1 then dev pci1 / ls is short hand.Remember the (space). Sorry the format sucks here. |
. |
RE: New(?) Radeon 7000 problem |
October, 11, 2003 3:06 PM |
lyonsdj88 |
. |
Terminal in OS X :sudo nvram pci-probe-list="-2" That will disable the first bit(device) on your first pci bus(bus 0).Now on my 8600 the first pci bus is vci0 the next pci bus is pci1.There are 6 more pci bus's but they are not used. Now -2 is Decimal for fffffffe Each letter is a bus from right to left not left to right.So,"e" in fffffffe is bus 0(vci0 on my 8600) Now each letter has four bits and each bit is a spot for a device,not all are used. So, now you need to know that a bit is a 1 or a 0 so four bits is 1111 or any four didgats of 1 or 0 like 0011 or 1010 or 1110 or 1011 and so on. A 0 tells Open Frimware Not to probe that bit(or device) and 1 means Do probe that device.So now How do you get "e" from 1110. Please see this chart:http:// www.tpub.com/neets/book13/53m.htm ......Remember right to left so on that bus,bus 0 or "e" that 0 is the first bit(or device) If you look back at the chart you will see that "f" is 1111 and 1 means do probe.So seven "f"'s is probe all 28 bits of the Last 7 bus's(most do not have a device but it won't matter because OF just won't find anything and move to the next.) So if your built in video card in the first device on bus 0 then fffffffe(1111 1111 1111 1111 1111 1111 1111 1110) will probe every thing but your built-in video. Now if your built-in video is the 2nd device on bus 0 then 1111 1111 1111 1111 1111 1111 1111 1101(fffffffd) Now my pci slots are on bus 1 so to disable them I turn off 3 bits on bus 1(I have 3 slots) Some thing like this ffffff1f(1111 1111 1111 1111 1111 1111 0001 1111) How do I know what bit a device is on see joeVT's list for the 8600 http://www.opendarwin.org/bugzilla/ show_bug.cgi?id=986 ...Now your Biege could be different so you'll have to check. In Open FirmWare :dev pci1 ls Now you should see any cards in your 3 pci slots and mybe a 4th device(I have pci106b,1 and my 3 slots) So if you see the names of your pci cards then pci1 is your pci slot bus too. Most likely you built-in video is on bus 0(vci0 in my system) :dev vci0 ls if you see a list and on that list is the OF name of your built-in video then vci0 is your 1st bus if you get can't find dev after :dev vci0 then vci0 is not the name of bus 0 in your system so try :dev pci0 ls So now How did I get the Decimal(-2) from fffffffe check here: http://www.eclectic-web.co.uk/awr/ hex2dec.php?num=ffffffbf ....Well good luck ,hope I was clear and if you did not understand part or I forgot something post your ?. |
. |
RE: New(?) Radeon 7000 problem |
October, 11, 2003 12:29 PM |
jseibyl |
. |
Crap forgot, it's OF time! |
. |
RE: New(?) Radeon 7000 problem |
October, 10, 2003 6:48 PM |
screwj |
. |
jseibyl, now I come to think of it, yanking the VRAM could be a problem, as the original 2MB are soldered to the MB IIRC. Nice thought though gregoryy, I used the nvram reset earlier to get the OF prompt onto the radeon, the problem is the OS X kernel seems to insist on ignoring it. Btw, after zapping the PRAM, the Display Preferences has stopped behaving strangely (ie, only the main window for the one monitor), the other problems persist though :-( |
. |
OpenFirmware reset (Re: Radeon 7k) |
October, 10, 2003 5:08 PM |
gregoryy |
. |
FWIW: OpenFirmware reset for Beige: > init-nvram > set-defaults > reset-all |
. |
RE: New(?) Radeon 7000 problem |
October, 10, 2003 4:19 PM |
jseibyl |
. |
good luck, at least there are a couple of options, hope something works, let us know... |