First of all I think this is related to the same things discussed before over there ->> https://devtalk.nvidia.com/default/topic/754384/recovered-gpu-errors-in-nvidia-settings/?offset=9#4328013
I have recently replaced all of the important guts in my system. Went from AMD/FX 8350 and 2x gtx580s to Intel/i7 5930k and 3x gtx970. Unlike windows… Linux fired right up after the hardware switch. On the old hardware when those errors happened the screen would just hiccup for a sec and then go about its business. With the new hardware when this happens the screen just locks up. I can still ssh in so I was able to grab a bug report, but even if I kill Xorg and I can see that it is killed and no longer running the screen stays the same and I’m forced to reboot via ssh or hardware button. This is making Linux nearly unusable =( which sucks because I spend 90% of my time in it normally.
I saw the new 346.22 beta and it mentions hard lockups on it so I thought I would give it a try so I removed the old drivers rebooted and then installed the new one and then rebooted again. The first time I startx I got a KP. So I reset and removed my xorg.conf and it started up with the one monitor.
I then turned on base mosaic -> saved xorg.conf -> restarted X and I was met with all three but in the wrong order (this is normal at this point). So I rearranged them to the correct order and then saved xorg.conf again and restarted X. This time it simply fails to start then everything locked up. SSH doesn’t accept connections… I can’t switch TTY’s so I can get any kind of bug report.
I then tried what Aaron told me to try with the last 346 beta driver and remove the GPU-* from the metamodes and this time I just get a message about Xorg waiting for the server to accept connections but I think that may be because of the missing GPU-* now that that bug is “fixed”
Please let me know what info you need to try and get this fixed… Im being forced to windows right now because of this and I’m not a happy camper.
EDIT* I ended up getting it to let me switch to tty2 when x failed to start with 346.22 one time… failed every time since then too. I attached it as well.