00:01 mupuf: pmoreau: I guess so, yes
00:01 mupuf: same in sweden?
00:02 pmoreau: It seems there are partying too. :)
00:02 pmoreau: And there will be a cortège leaving Chalmers tomorrow and most likely some more partying :D
00:03 pq: in some university in Finland, "vappu" (May 1st) is actually the two weeks ending around May 1st
00:03 mupuf: pq: right, they have been very active in the past two weeks
00:03 pmoreau: Wut? Two weeks of partying??
00:03 pq: pretty much, yes
00:03 pmoreau: :D
00:04 pq: those who can stand and afford it...
00:04 pq: events every evening here
00:05 pmoreau: Seems crazy!
00:05 mupuf: AFAIK, this is a quiet evening today. Probably gathering strength for the next few days
00:05 pmoreau: But I would be among the ones who can't afford to go
00:12 mupuf: That's what happens when one lives too far from Talinn :)
00:12 pmoreau: Eh?
00:13 mupuf: Well, everyone in Helsinki buys their guns in Talinn
00:13 mupuf: it is *much* cheaper
00:13 mupuf: and the cruise is 2h-long
00:13 mupuf: and costs about 20e
00:14 pmoreau: Oh, ok :)
00:15 RSpliet: mupuf: don't forget to celebrate the 5th of May
00:15 RSpliet: with your Dutch friends ;-)
00:15 mupuf: RSpliet: What's that?
00:15 RSpliet: liberation day
00:22 mupuf: :)
00:25 hakzsam: imirkin, skeggsb_ btw, what is the status of nvif in libdrm?
06:48 hakzsam: mlankhorst, you have a gk20a, right ? :)
08:02 imirkin: hakzsam: no clue, sorry
13:49 imirkin_: fascinating. the mem_out_of_bounds error happens in TCS when reading inputs
13:50 imirkin_: seemingly readning *any* of the inputs causes errors
14:14 nightfuri: hello i get this from wine. fixme:d3d_surface:surface_cpu_blt Filter WINED3D_TEXF_LINEAR not supported in software blit. Has this got anything to do with the driver ?
14:14 imirkin_: no idea... ask the wine guys?
14:15 nightfuri: ok
14:16 glennk: imirkin, any intermediate storage the driver needs to set up?
14:20 imirkin_: glennk: maybe. i'm actually gonna go ahead and backpedal on that comment though
14:20 imirkin_: commenting out most of TES also fixes the problem
14:21 glennk: presumably the parameters get stored to some memory somewhere?
14:21 imirkin_: seems likely
14:22 imirkin_: same thing as for GS
14:22 glennk: afaik its the same as on radeon, tess writes to main memory
14:22 imirkin_: i'm also investigating an unrelated codegen fail
14:22 imirkin_: which is unlikely to be helping matters
14:23 glennk: which incidentally is a problem for tess performance, bouncing data back out to memory
14:23 glennk: might as well just pre-tessellate and read the vertex data directly, less bandwidth needed
14:23 imirkin_: pretty sure that's not an issue on nvidia
14:23 imirkin_: it uses an on-chip buffer for that
14:23 glennk: perhaps, think at least for small enough data it'll stay on chip
14:24 glennk: probably a big cliff to fall off with fine enough tessellation
15:40 pmoreau: Why don't we want to execute nouveau_switcheroo_set_state if it is an Optimus setup and we want to poweroff?
15:44 imirkin_: glennk: looks like even accessing the GENERIC[9] thing is broken
15:44 imirkin_: (in quads.shader_test)
16:07 imirkin_: perhaps the nv hw hates having holes like that, dunno
16:09 imirkin_: i also noticed a giant whoopsie which affects all the code emitters
16:09 imirkin_: (and a minor one)
23:07 effractur: HI
23:08 effractur: with kernel 4.0.1 i am getting on powerpc the following error failed to create 0x00000080, -22
23:08 effractur: for the vbios
23:08 imirkin: effractur: pastebin all the messages from nouveau?
23:08 effractur: i tried with nouveau.config=NvBIOS to specifiy a differnt bios
23:08 effractur: but that does not work
23:08 effractur: imirkin: k
23:09 imirkin: effractur: also, did this work with an earlier kernel?
23:09 effractur: https://p.6core.net/p/c5figpnSRECHnoKIcXG7sPDM
23:09 effractur: imirkin: with 3.18 i got a vbios but after that it did not drive the display correctly
23:09 imirkin: just to confirm -- it this is an old box, right?
23:10 effractur: yes
23:10 imirkin: k. so NV17 seems plausible. is this a mac?
23:10 effractur: yes
23:10 effractur: powerbook g4
23:10 imirkin: i would have expected OpenFirmware
23:10 imirkin: as the vbios source
23:10 effractur: it can't find one
23:10 effractur: with that option
23:11 imirkin: well, i assume you have OF support in your kernel, otherwise it probably wouldn't have booted
23:11 effractur: correct
23:12 effractur: https://p.6core.net/p/f5i5zVfykrUVDhRYtEA4fY4k
23:17 imirkin: effractur: can you boot with nouveau.debug=debug
23:18 effractur: k
23:20 imirkin: [and pastebin the resulting messages]
23:21 effractur: https://p.6core.net/p/5EEwHyCeJxR7inpZiqbO4TQ5
23:21 imirkin: skeggsb_: looks like OF loading got broken :(
23:22 effractur: is there a way to overrule the invalid signature?
23:22 imirkin: effractur: the problem is that data isn't being loaded at all, not some invalid signature.
23:22 effractur: from the PROM?
23:22 imirkin: there's no PROM
23:22 effractur: a
23:22 imirkin: only OF
23:23 effractur: mmk
23:23 imirkin: hmmmmm
23:23 effractur: it never workt with noveau
23:24 effractur: in 3.18 even the bootconsole was broken
23:24 effractur: so there is some progres
23:25 imirkin: effractur: if you don't mind a small patch
23:26 effractur: i am fine
23:26 imirkin: edit drivers/gpu/drm/nouveau/core/subdev/bios/shadowof.c, and remove the __iomem annotation, and replace memcpy_fromio with a plain memcpy
23:27 effractur: were is that in the kernel tree?
23:28 imirkin: i gave you the full path in the kernel tree :)
23:28 effractur: a got it
23:28 effractur: na it is nvkm/subdev/bios/shadowof.c
23:28 imirkin: oh, you're using ben's side tree
23:29 imirkin: yeah
23:30 effractur: k i will recompile the kernel and report back
23:31 imirkin: fwiw i seem to recall an open bug for that mac though, and it's just a pile of fail...
23:31 effractur: yea
23:31 imirkin: i tried to fix it for a while, but eventually gave up
23:32 imirkin: https://bugs.freedesktop.org/show_bug.cgi?id=21273
23:32 effractur: the issue i had with the 3.18 kernel was that noveau did not properly setup the resolution of the lvds panel
23:32 effractur: and also could not fix that
23:32 skeggsb_: imirkin: yeah, i know, it's on my todo list for this week actually
23:33 imirkin: skeggsb_: which... making that NV17 work? or fixing OF?
23:33 skeggsb_: OF
23:33 imirkin: k
23:33 imirkin: that makes more sense :)
23:33 skeggsb_: only barely :P
23:34 imirkin: skeggsb_: any comments on the libdrm fails?
23:34 imirkin: skeggsb_: and also the new supertuxkart (0.9) insta-crashes with a libdrm fail too, although that one's not related to the 2.4.60 situation
23:35 skeggsb_: i haven't noticed either
23:35 effractur: lets see
23:35 imirkin: skeggsb_: well, the "fix races" patch from maarten appears to be destroying the universe
23:36 imirkin: i _think_ it's some sort of fail related to external bo's? not sure though.
23:36 imirkin: the stk fail is different.
23:36 imirkin: somehow the pushbuf list is... corrupted or otherwise very wrong
23:36 effractur: mhe same issue
23:37 imirkin: effractur: doh ok.
23:40 effractur: if it is patched will that be in 4.0.X?
23:41 effractur: or in the nexed kernel release