08:30 karolherbst: mupuf: the nve6 is perfect
08:30 karolherbst: don't even need to enable a boost level and it hangs
09:25 karolherbst: nice
09:25 karolherbst: it hangs when going from 07 -> 0f, but not for 07 -> 0a -> 0f
09:26 karolherbst: I already assumed that the issue is somewhere in the pll configuration here, seems likely
11:23 mupuf: karolherbst: wonderful :)
12:41 karolherbst: mupuf: after taking a second look, it looks like a memory clock issue on your nve6
12:48 karolherbst: yeah, memory corruption and something messed up
12:50 karolherbst: but still, another reclocking issue
13:32 doctorpepper: hi guys
13:36 doctorpepper: can anyone help me please , i am trying to offloading from intel to nvidia gtx860M but when i try to set the provider offloadsink i get the message Could not find provider with name nouveau (and the same for Intel)
13:39 doctorpepper: and the result of xrandr --listproviders doesnt seems to be right see: https://paste.fedoraproject.org/paste/tnquTskIFbTIuCUKSHiEKg
14:59 Sophira: I had another freeze just now; interestingly, OBS was not involved last time like it was last time, and the error was different. It also never froze completely, so I was able to reboot safely. http://matrix.theblob.org/2017-09-13.dmesg-freeze-2-cropped.txt (as a reminder, the dmesg from last time it happened a few days ago is at http://matrix.theblob.org/2017-09-08.dmesg-freeze-cropped.txt )
15:02 Sophira: I'm noticing that the "nouveau 0000:01:00.0: gr: TRAP ch 2 [103fb84000 X[2375]]" line is the same in both (except for X's PID, obviously).
15:02 Sophira: But the "103fb84000" figure is the same.
15:04 wjx: What does 'mc' in nvkm_mc mean?
15:13 wjx: Multi Controller?
18:53 karolherbst: seems like we have an issue within our SEQ parser: 000000: Invalid param 00010200 for op 0x14
18:53 karolherbst: RSpliet: did you encounter this at some point?
18:55 karolherbst: currently this gets printed out: WAIT STATUS !(unknown) , 45478000 ns
18:55 RSpliet: yeah, that's the best it can do. This status is new in whatever blob/GPU combo you're currently trying to trace
18:56 karolherbst: I just read an old trace
18:56 karolherbst: no idea which version that was
18:56 karolherbst: okay, hopefully it ain't important
18:57 RSpliet: I did document what I could back then http://envytools.readthedocs.io/en/latest/nvrm/pmu/seq.html#wait-status
18:57 RSpliet: I bet it has something to do with waiting for VBLANK on head 2 or 3, which didn't exist when I RE'd SEQ
19:00 karolherbst: might be
19:30 karolherbst: okay nice, this indeed looks like a 10 hour task I could do on the flight :)
20:30 jushur: lets say i have a laptop with intel i7 and intel igp+nvidia dedicated gfx (using nouveau) i have a hdmi output wich i belive could be used solely by the nvidia card.
20:31 jushur: would it be possible to do some magic and run one X on the intel igp, and one on the nvidia dedicated?. or possibly even passtrough.
20:35 karolherbst: jushur: search for reverse prime
20:35 jushur: karolherbst: ok, thx
21:25 jushur: That worked better/easier then I expected =)