09:07 pmoreau: imirkin: I can do that when I get some time. I have been trying to copy nvc0 as much as possible but only when I hit parts that needed fixes or to be changed, so it’s quite possible that there are many other places that could also benefit from copying what nvc0 does.
09:07 pmoreau: I need to bisect that `nouveau 0000:03:00.0: DRM: core notifier timeout` I’m getting whenever I alt-tab between windows since the rewrite.
09:08 pmoreau: So it can be fixed before 5.9 is released.
09:19 pmoreau: karolherbst: How are things looking for the Nouveau CI? I might be able to get a first computer available for it today, once I received the new case fans and extracted an SSD from another computer.
09:21 pmoreau: That one can contain a Tesla, Fermi or Kepler card; the Maxwell, Pascal and Turing cards I have do not fit in that case because of the non-modular drive cage, and I do not have a metal saw to remove it. 🙃
09:38 karolherbst: pmoreau: at this point it's more about getting fundamentals done
09:38 karolherbst: like somebody could figure out how to containeraize lava and declare nodes via config files
09:39 karolherbst: or other things :)
09:39 karolherbst: depending on what you have you could look into various things
09:39 karolherbst: I should write down what needs to be done at some point
09:41 karolherbst: pmoreau: right now I collect stuff here: https://gitlab.freedesktop.org/karolherbst/nouveau_ci
09:41 karolherbst: so the thing I tried to figure out was how to netboot devices via docker container :)
09:41 karolherbst: still needs a bit things figured out like adding iptable rules and stuff
09:42 karolherbst: right now it's all in a script
09:42 karolherbst: but I'd like to have it inside priviliged containers
09:42 karolherbst: and in a way it doesn't suck
09:43 karolherbst: another item would be to work on a gitlab-runner container and configure it against a donstream fork of mesa
09:43 karolherbst: anyway, lot's of things to do :)
09:44 karolherbst: but the first "target" is kind of having a single node lava installation pulling from a daily updated mesa fork running some CTS tests
09:44 karolherbst: shouldn't be toooo much work
09:45 karolherbst: doesn't matter if it had to be installed manually, but it should all be containers in the end within our own registry
09:46 karolherbst: and something like rancheros installed on the "controller" pulling those taged containers and just run things
09:47 karolherbst: in the end I want _all_ machines to be completely stateless
09:48 karolherbst: no ssh to config shit, no storage needed except for storing test results and upload them etc...
09:48 karolherbst: it's a lot more work initially, but once it runs it is very low maintenance
09:49 karolherbst: and interesting aspect later on would be to have it distributed and spawn an VPN network with clients creating private networks and shit... but that's even more work :D
11:31 imirkin: pmoreau: AndrewR reported the same thing with a G92
11:37 pmoreau: I saw that in the logs. We will see whoever gets around to bisecting it first. 🙂
11:41 pmoreau: I’m having the same issue on what I think is the release 5.8 without those patches, but it didn’t trigger as quickly as on the newer version; will need to double-check that. And now I even have the added benefit of no longer seeing my mouse cursor! 🎉 Navigating a GUI without seeing the cursor, is… interesting. 😅
20:09 Lyude: sigh-patch series I sent got held up for moderation approval on the nouveau mailing list, anyone able to fix that?
20:10 imirkin: maybe you can find out from someone who the moderators are, and get some nouveau people added to the list?
20:11 imirkin: i assume it's just like marcheu or something
20:11 Lyude: yeah
20:11 Lyude: probably a good idea
20:11 Lyude: marcheu / skeggsb: do you have any idea who I should poke about thta?
20:11 imirkin: iirc even skeggsb isn't an admin
20:11 imirkin: start with daniels
20:14 airlied: Lyude: skeggsb_ can read dri-devel anyways :-P
22:27 daniels: I think it happens actively because normally opening the moderation page for abandoned lists will completely kill Chrome, but there were only about ten messages
22:28 daniels: pushed it through now anyway
22:44 Lyude: daniels: thanks-feel free to add me to the moderator list if you want btw
22:44 imirkin: me three.
23:34 daniels: see email
23:37 imirkin: daniels: thanks!
23:38 daniels: yw
23:38 imirkin: daniels: authorization failed :(
23:38 imirkin: unless i have no idea what i'm doing
23:38 imirkin: which is a distinct possibility...
23:40 daniels: https://lists.freedesktop.org/mailman/admindb/nouveuaj
23:40 daniels: er, except spelled correctly obv
23:40 daniels: password I sent, no preceding spaces
23:41 imirkin: aha, that works better
23:41 imirkin: the mailman interface itself links to /admin/ and not /admindb/
23:41 airlied: skeggsb: any idea why in nouveau_ttm_io_mem_reserve you stash handle inside offset?
23:42 airlied: I can't see where anything takes it back out again
23:43 airlied: I assume there is some magic code later that figures it out, if I stuck it in base would that still work?
23:46 daniels: imirkin: right, admin is one thing, and moderation is another
23:46 daniels: if you want admin as well then I'm super content with that, but maybe you could follow up with marcheu and ask for it
23:47 imirkin: i didn't realize they were different things
23:48 imirkin: i haven't interacted with mailman admin since ... well, quite some time :)
23:49 daniels: good life choices!