03:00mooch: damn, mwk, you're really working hard!
12:10karolherbst: new crashes :O
12:39karolherbst: pmoreau: meh, you forced me to make the error check cleaner :D
12:40karolherbst: it is a single if clause now
12:40pmoreau: On the other hand, you will (currently) never end up passing a NULL pointer.
12:41karolherbst: but I try to error check every non static function
12:41karolherbst: and we should do that
12:41pmoreau: So… should you do that for the `bios` pointer as well? O:-)
12:42karolherbst: tegras have no bios
12:42karolherbst: but yeah
12:42karolherbst: they shouldn't call the bios subdev in the first place
12:42karolherbst: those error checks aren't there for technical reasons, but to prevent the machine to crash if you code on it
12:42karolherbst: smart compiler will optimize that stuff away anyway
12:43pmoreau: A crash isn’t really great
12:44karolherbst: well currently I find crashes in the glsl part of mesa, there are plenty
14:17karolherbst: pmoreau: what do you do after christmas :D
14:18karolherbst: aka the week before new year
14:19karolherbst: mhh, maybe RSpliet would like to come too to 33c3, and maybe hakzsam_ as well :D
14:19karolherbst: imirkin: I guess it is too far away for you (any maybe even too expensive)?
14:23hakzsam_: I don't plan to attend, christmas is family time :)
14:24karolherbst: don't want to be alone having a nouveau stand there
14:24karolherbst: but yeah, it is a bit silly to be between christmas and new years
14:25pmoreau: karolherbst: The first half of that week, I’ll still be with my family, and the second half I’ll be working as I’m only taking a week of vacation.
14:25karolherbst: mwk: you are my only hope as it seems :D
14:25pmoreau: Wasn’t mupuf planning to go?
14:26karolherbst: he can't
14:26karolherbst: which is really painful, cause he has most of the gpus
14:30karolherbst: pmoreau: ohh actually, I would need your nouveua image pimped
14:30pmoreau: I haven’t added DRI3 to it yet, nor Nouveau staging… --"
14:30karolherbst: the staging thing doesn't matter
14:30karolherbst: dri3 does
14:31karolherbst: most of the people will have laptops
14:31karolherbst: and I was thinking about useing your image for a ready thing to use and re on
14:31pmoreau: Let me finish my rebase of the SPIR-V stuff, and I’ll have a look at the image.
14:31karolherbst: :) awesome
14:32karolherbst: I guess doesn't have any way of persisting changes?
14:33karolherbst: actually it would make sense that somebody with knowledge about the userspace would attend as well, cause my knowledge there is actually low and those are the more visual things
14:34pmoreau: IIRC, it is possible to reserve some space of persistent changes. I was supposed to have a look at it a long time ago: https://github.com/hakzsam/archlive-nouveau/issues/5
14:43pmoreau: Apparently there are some cow options… will investigate later
14:48karolherbst: pmoreau: I would rather use overlay fs: having two partitions on the USB, one with the read only image and the second with all the changes done on /
14:48karolherbst: and the second partitions can be whiped when needed
17:03pmoreau: It looks like it is already using overlayfs
17:23karolherbst: pmoreau: nice!
17:29NanoSector: Lekensteyn: I think your patch broke suspend/resume on my system :x
17:29NanoSector: it suspends fine but when it wakes up it locks up completely
18:32Lekensteyn: NanoSector: does it work with pcie_port_pm=off? The patch results in a fallback to the DSM method, but that will probably not work when the pcie port suspends
19:13madigens: hi! i upgraded fedora 24 to 25 (prerelease) and booting kernel 4.8.6 without nomodeset results in a turned off screen. i have to sysrq-reisub to reboot. here's the log i managed to salvage: https://paste.gnome.org/poveolxsx
19:14imirkin: GTX 970 with 4GB VRAM?
19:14imirkin: boot with nouveau.noaccel=1 nouveau.nofbaccel=1 for now
19:14imirkin: should still get you modesetting.
19:15madigens: ah, okay. is this a known bug?
19:15imirkin: hmmm... maybe it'll do secboot either way
19:15imirkin: madigens: https://bugs.freedesktop.org/show_bug.cgi?id=94990
19:15imirkin: i believe there are some hackpatches in there you can play with
19:16madigens: oh okay, good thing it's known. saves me the trouble of reporting it ;)
19:17madigens: hm... my current goal is to get the damn proprietary blob going, will do that if i fail.
19:19imirkin: if you're looking for good linux support, i highly recommend looking at AMD's offering
19:23madigens: i'm actually waiting for vega to come out :)
19:23madigens: and zen
20:20Lekensteyn: NanoSector: btw, does this affect just systemd sleep/resume or also the runtime PM case?
20:22Lekensteyn: oh.. I am stupid
20:36mwk:is fresh out of simple methods to reverse
20:38karolherbst: mwk: well, if you got time, how does the 33c3 situation looks like for you?
20:39NanoSector: [19:32] <Lekensteyn> NanoSector: does it work with pcie_port_pm=off?
20:39NanoSector: I haven't actually checked
20:59Manoa: I have a problem, my fermi card is not reclockable, it get message function not implemented, so you guys suggested I try modify the bios and load it at the boot loader with nouveau.NvBios=/boot/nvbios.rom,nvforcepost=1
20:59karolherbst: Manoa: well, fermis aren't reclockable yet
20:59Manoa: when I do that, the system freezes the moment nouveau is loading xD
20:59karolherbst: who said that you should try this?
20:59Manoa: sory I don't remember, but the story is not over yet XD
20:59imirkin: Manoa: first off, i doubt /boot/nvbios.rom is a valid path. check how firmware loading works.
21:00imirkin: Manoa: secondly, what makes you think your modified vbios is correct for your board?
21:00Manoa: I had to install windows and use nibitor/fermibioseditor
21:00Manoa: and it was successful, anyway since the bootloader option didn't worked, I had the bios flashed into the card
21:00imirkin: and do those tools change the *init* sequence?
21:01Manoa: and it worked, all 3 performance levels were correctly detected in /sys/kernel/debug/...
21:01imirkin: i very much doubt they do - they probably just adjust the perf tables that the blob reads
21:01Manoa: and I think now you hit on the problem
21:01Manoa: is there anything that can be done about that ?
21:02imirkin: not really. i don't think there ever was.
21:02karolherbst: Manoa: memory reclocking still has to be implemented
21:02karolherbst: Manoa: though there is something: if you can do some reverse engineering you could actually help out
21:04Manoa: yhe sutch a shame this fermi is sutch a card xD
21:05Manoa: yhe reverse engineering good idea, I never done anything like this before
21:05Manoa: I would have to load the nvidia driver for linux and intercept the instructions it sends to the hardware ?
21:05Manoa: the MMIO things ?
21:34Lekensteyn: NanoSector: actually based on some tests I think that even pcie_port_pm=off won't work for you. Setting acpi_osi="!Windows 2013" should work though
21:35Lekensteyn: you are affected by https://bugs.acpica.org/show_bug.cgi?id=1333
23:47fmiz: Hello, I need some guidance on a bug, I don't even know if it's nouveau related. I have an old geforce 9300 and hdmi stopped working with fedora 25 beta, I've reported my issue here: https://bugzilla.redhat.com/show_bug.cgi?id=1385394
23:52imirkin: most likely a nouveau issue.
23:52imirkin: there's been some reports of that on NVAC boards
23:53fmiz: Is there anything I can do (aside from plugging in a VGA display)?
23:53imirkin: search for 'NVAC "No Signal"' on the nouveau mailing list
23:53imirkin: basically you just need to comment out a line in drivers/gpu/drm/nouveau/nvkm/engine/disp/nv50.c
23:54imirkin: skeggsb: --^ looks like it's not just poma