Hi Logan, Hmmmm, well at least you updated to much newer packages... but alas, that didn’t help.
If this only seems to happen when lightdm starts (i.e. xorg starts up) and not the nouveau fb console, perhaps sending xorg logs may provide more info... Otherwise, it may be something to discuss with the nouveau devs on the nouveau irc. The dev "imirkin" is quite helpful and i believe he may have access to a G5 machine... I know the nouveau devs would ask you to use the most recent packages available, so its great that you are already on stretch. Nothing else i can think of right now.... Lets see if your xorg logs show anything.. Cheers, Peter On Fri, Jan 22, 2016 at 10:05 AM, Logan Anteau <logan.ant...@gmail.com> wrote: > Peter, > > Sorry for the delay in following up with this. I got a chance to try the > suggestions you provided in your last message. To your first question, yes > this only occurs when lightdm attempts to start up, the nouveau fb console > seems to work fine. I had not previously blacklisted "nv", but I did, no > change. I do not have an xorg.conf. I was on Debian stable before, last > night I upgraded to Debian "stretch" which included newer nouveau, libdrm, > mesa, etc, the issue is not fixed, I get the same error message. > > I appreciate all your help with this and the custom built kernels. I > really want to get this working and for us all to gain the knowledge of > what it takes. This would help any future PMAC 7.2 users with the GeForce > 6800 as well. Do you have any other ideas? > > Thanks, > Logan > > On Thu, Jan 14, 2016 at 7:29 PM, Peter Saisanas <psaisa...@gmail.com> > wrote: > >> Hi Logan, >> On the positive, you do not seem to be having any more EDID issues once >> you have switched to the other LCD screen. >> >> But this line seems a worry: >> >> [ 27.168859] nouveau 0000:f0:10.0: gr: intr 00100000 [ERROR] nsource >> 00000010 [LIMIT_COLOR] nstatus 04000000 [PROTECTION_FAULT] ch 1 [0008e000 >> Xorg[853]] subc 4 class 009f mthd 0308 data 04380780 >> >> Does this only occur when lightdm attempts to start? >> Does the nouveau fb console seem to work fine or does this only occur >> when xorg starts up? >> >> Can you please confirm that the old "nv" driver is blacklisted and not >> conflicting with "nouveau"? >> >> Also, do you happen to have an xorg.conf file as this shouldn't be >> required. >> >> Lastly, what debian repo's are you on, i.e. are you on stable or testing >> (stretch)? as it could be something with the xorg drm driver not playing >> nice with the much newer kernel... Not sure, but something to consider. >> >> The reason why the earlier kernel builds (i.e. < 4.4.0-rc7) didn’t work >> was that they were set up particularly for the PMAC 11.2 machines and not >> the older PMAC 7.2 Powermac G5's. There were drivers missing specifically >> for the AGP Powermac G5's. >> >> OTOH, Aiden was the first one i believe who has a PMAC 7.2 machine that >> actually gave some feedback. So we adjusted the config to also suit those >> machines. >> >> I build these kernels for my own testing purposes and originally just for >> my own G5 Quad (PMAC 11.2). But since it it may help others in the >> community, i can try to adjust the config to suit other G5's as well. >> That's why feedback is great as i don't have other G5 models to test on. >> If it seems to help others in the Debian community to get their Powermac >> G5's running, then why not change the config to suit more machines i say >> It still seems to missing some drivers, but hopefully with feedback we >> can try to sort them out. With my kernel builds, only try the builds > >> 4.4.0-rc7 as these have the kernel config changed for the AGP Powermac G5's. >> >> Lastly, i am on Debian testing repos so it might explain why you have the >> version conflict when installing the headers as it might have newer >> dependancies. >> >> Keep on sending logs as it helps to diagnose what’s wrong with your >> particular config. >> >> Good Luck! >> >> Regards, >> Peter >> >> On Fri, Jan 15, 2016 at 10:12 AM, Logan Anteau <logan.ant...@gmail.com> >> wrote: >> >>> Peter, >>> >>> I have installed your 4.4.0 kernel. With the same monitor I was using >>> before I get this dmesg log (seems to be about the same issue). Same EDID >>> issue with the other DVI port too: >>> http://pastebin.com/NJDvK56K >>> >>> >>> I tried another monitor and I get a black screen that eventually has >>> white,purple,black, and green bars. Here's a dmesg log from that monitor: >>> http://pastebin.com/ZuZWFCXy >>> >>> >>> Here's an output of /proc/interrupts from your 4.4.0 kernel: >>> http://pastebin.com/jUqmhUSi >>> >>> Also btw, I assume you are aware, but if not, your linux-headers-4.4.0 >>> is uninstallable because it depends on a version of libncurses5 >= 6 (5.9 >>> is installed) and libtinfo5 >= 6 (5.9 is installed). Hopefully this new >>> information helps. Let me know what I can try next... >>> >>> Logan >>> >>> >>> >>> On Thu, Jan 14, 2016 at 3:04 PM, Peter Saisanas <psaisa...@gmail.com> >>> wrote: >>> >>>> Hi Logan, >>>> Not sure if you have intentionally disabled MSI interrupts for nouveau, >>>> but i believe the MSI interrupts only need to be disabled for PCIe based >>>> nVidia gpu's running the nouveau driver on Powermac G5's only. I have my >>>> doubts that AGP based nVidia GPU's ever have MSI interrupts enabled in the >>>> first place. >>>> >>>> Either way, MSI interrupts are disabled on your machine. >>>> >>>> Only try the 4.4.0 kernel out. Originally these kernels were for the >>>> Powermac G5's (PMAC 11.2) but only recently i have had feedback from an AGP >>>> based machine (PMAC 7.2) and tweaked the kernel configs to accommodate >>>> those machines as well. >>>> >>>> Anyway, if you can run 4.4.0 and give feedback along with full dmesg >>>> logs and cat /proc/interrupts logs, this would be most helpful. >>>> >>>> You probably are aware of this, but make sure you create a new entry in >>>> yaboot.conf and run "ybin -v" for the new test kernel. >>>> >>>> Regards, >>>> Peter >>>> >>>> >>>> >>>> >>>> >>>> On 15/01/16 06:01, Logan Anteau wrote: >>>> >>>> Peter, >>>> >>>> Thanks for your suggestions. Here's a pastebin of my /proc/interrupts: >>>> >>>> http://pastebin.com/2ZG4qESD >>>> >>>> I'll try one of your newer kernels either tonight or tomorrow when I >>>> get time. I believe I tried your 3.8.16 kernel and had other issues with >>>> it. Networking didn't work out of the box, the system didn't shut down >>>> (only halted) and it hung when trying to start lightdm (I couldn't even ssh >>>> into it or get to the ttys through Ctrl+Alt+F1,F2,etc). >>>> >>>> I'll give you 4.4/4.3 kernels a try though and report back. Unless you >>>> have any intermediate recommendations first. >>>> >>>> Logan >>>> >>>> >>>> On Thu, Jan 14, 2016 at 3:42 AM, Peter Saisanas <psaisa...@gmail.com> >>>> wrote: >>>> >>>>> Hi Logan, >>>>> The kernel version is quite old (3.16.7). Perhaps you can try >>>>> something newer. Nouveau continually has improvements practically with >>>>> each >>>>> newer kernel version. >>>>> >>>>> Perhaps something to consider, try the easy way out and use a >>>>> pre-compiled kernel deb package that i have available. I have kernels as >>>>> new as 4.4.0 available suitable for Powermac G5's. >>>>> At least these could perhaps be a better baseline as these kernels are >>>>> known to be working for many Powermac G5's (AGP & PCIe GPU's along with a >>>>> few generations of nVidia GPU's (some even older than yours). Makes it >>>>> easier to have a side by side comparison. >>>>> Your call, link is below if interested. >>>>> >>>>> >>>>> https://drive.google.com/folderview?id=0B8pqd5Ots1vfT2puX09CYjEwcFk&usp=drive_web >>>>> >>>>> Many have used them successfully in the past and have got their G5 >>>>> Powermacs running with nouveau. >>>>> >>>>> I'm not so sure that MSI interrupts are ever enabled on AGP GPU's in >>>>> the first place (PCIe based GPU's are a different story). >>>>> It would be interesting to see your cat /proc/interrupts log as well. >>>>> >>>>> Regarding your EDID issue, can you try the other DVI connector of your >>>>> video card or cable or alternatively try another display out if the issue >>>>> still persists. >>>>> >>>>> Most Apple OEM ATI Radeon GPU's have their own issues as well. But >>>>> that's another story. >>>>> >>>>> Regards, >>>>> Peter >>>>> >>>> >>>> >>>> >>> >> >
