01:43mlankhorst: airlied: it works for me, just abusing the inactive devices
02:42mlankhorst: airlied: you need a syntax identical to 'inactive', but because the nvidia configs abused a bug to make inactive devices be added as a gpu screen I've patched ubuntu to use the inactive syntax for it. :)
03:13airlied: mlankhorst: uun upstreams and abi abuses are someone else problem :)
03:19mlankhorst: heh :p
09:43aaronp: mlankhorst + airlied: on a related note, do you guys want to review the pair of changes in, "Fix GPU screen handling for drivers with no normal screens"?
09:57hakzsam: imirkin, ~4 days later, I finally found the time to answer :)
10:48mlankhorst: aaronp: looks sane to me
11:24pmoreau: bonbons: Ok, thanks. It's most lokely the same issue as my laptop, so that's great: I'll be able to track two bugs at the same time! It will probably be one or two weeks before I get back to doing some Nouveau stuff though… :/
12:40imirkin_: joi: were you going to push your patch out, or do you want me to?
12:43aaronp: mlankhorst, can I count that as reviewed-by for both patches?
12:47mlankhorst: aaronp: ironically I seem to have something similar for ubuntu, just in a different way
12:47mlankhorst: effect's probably the same though
12:47aaronp: Thanks. I'll put together a pull request for Keith.
12:48mlankhorst: I probe in the first loop with GPU_SCREEN if device's set to inactive
12:49aaronp: I'd like to get away from exploiting the inactive trick.
12:49aaronp: I'll update the NVIDIA README file to recommend putting a Load "intel" line in the Module section on sufficiently new X servers once that fix is in.
12:50mlankhorst: won't the effect be the same? :P
12:50aaronp: Yeah, but hopefully it'll be clearer & simpler for everyboyd.
12:58bonbons: pmoreau: there's no hurry at all! When I'm plugging in the nvidia it's more for playing around with system than any "production" use. Whenever there is something to try out, just notify me (on the bug or here assuming i'm connected)
12:59imirkin_: bonbons: seems like you've had rotten luck with nouveau drivers... sorry =/
13:00joi: imirkin_: I wanted to test those ue4 demos first
13:00joi: it seems none of them hit the scratch runout case
13:00bonbons: imirkin_: well I'm probably rather doing non-standard activities or using rather old hardware :)
13:01imirkin_: joi: ok, oh well. i assume piglit went well with your latest version.
13:01imirkin_: bonbons: meh, nouveau _should_ work on all that
13:02bonbons: well, for the really old hw with nvidia igp edid is still missing, because it's unknown where the damn i2c is located
13:03imirkin_: bonbons: your nv1a? heh.
13:03imirkin_: bonbons: you could see if it works with xf86-video-nv
13:03bonbons: imirkin_: yeah!
13:04bonbons: it doesn't
13:04imirkin_: and if it does, try to figure out where it gets its stuff from
13:04imirkin_: or nvidiafb/rivafb perhaps?
13:04imirkin_: they might cheat somehow though
13:04bonbons: in the past I opened a bug exactly for that one in the hop some of the nvidia guys lurking around that code would help get it fixed, but it never happened and the bug was converted to a nouveau one
13:05imirkin_: well, we could send a question to the gpu-doc black hole
13:05imirkin_: and hope for some hawking radiation to come out of it...
13:06bonbons: I did record mmio traces tough with closed driver and did send them in (but no idea if someone reacted to the trace)
13:06bonbons: year, that would be cool
13:07imirkin_: hmmm, well the traces should be in the email account
13:07imirkin_: but that account isn't checked regularly, only when we need to retrieve something specific
13:07bonbons: for secondary gpu on the more recent geforce (PCIe already), yes it should but doing the full bring-up is probably still different from resuming or taking over
13:07bonbons: the trace is certainly already languishing there for years now
13:09bonbons: and I should take the time to dig out my very old PentiumII with riva or so gpu, just for fun :)
13:10imirkin_: well, tnt/tnt2 should work fine
13:10imirkin_: although i'd advise aganist running the latest generation of games on it ;)
13:10bonbons: mostly a matter of building a gentoo image for it!
13:10imirkin_: it should actually even have a more conformant driver than nv1x/nv2x
13:10imirkin_: since it uses swtnl
13:11bonbons: swtnl? not a keyword I recognize...
13:12imirkin_: software tnl
13:12imirkin_: tnl = ... gr
13:12imirkin_: i always forget
13:13imirkin_: texturing and lighting?
13:13imirkin_: tnl = transform & lighting
13:13joi: imirkin_: pushed
13:13imirkin_: joi: cool
13:13bonbons: oh, already heard that in the distant past :)
13:14imirkin_: joi: let me know if you want to do more work on the nouveau mesa bits... i can probably suggest some items
13:25joi: imirkin_: I'm not sure if I have enough time to commit myself to do anything meaningful
13:28imirkin_: joi: ok no worries
14:41aaronp: mlankhorst, thanks for the review! Pull request sent.