01:09 gnurou: secure boot accomplished on GM206 \o/
01:10 karolherbst: yay
01:10 karolherbst: gnurou: and how _ugly_ is it to do? :D
01:11 gnurou: I will just have Ben close his eyes when doing the git merge
01:12 karolherbst: gnurou: does stuff like opengl works then already or are there still missing mesa bits?
01:13 gnurou: karolherbst: a few missing Mesa bits, but I got kmscube and parts of glmark2 working
01:13 karolherbst: and performance compared to the binary driver in that cases?
01:13 gnurou: the parts non working are arguably because I quickly hacked Maxwell texture support (without my hack they don't crash, but display black)
01:14 karolherbst: ohh I see
01:14 gnurou: performance will be bad, mainly because the shaders have no scheduling information
01:15 gnurou: but this should be fixable once someone takes a look at maxas
01:15 karolherbst: is it really _that_ important?
01:15 gnurou: textures should also not take more than a couple hours for imirkin_ if I can release the format documentation
01:15 gnurou: it can make quite a difference yes
01:15 gnurou: then of course you have the beloved reclocking...
01:15 pmoreau: gnurou: \o/
01:16 karolherbst: I am asking because on kepler nouveau stucks at around 60-70% perf
01:16 karolherbst: pmoreau: though furmark from gputest runs at over 90% binary performance
01:16 karolherbst: ...
01:16 karolherbst: gnurou: ^
01:16 gnurou: IIUC the kepler compiler adds scheduling information, doesn't it?
01:16 karolherbst: yeah
01:16 karolherbst: but no clue if its right
01:16 karolherbst: it isn't bad though
01:16 pmoreau: gnurou: Double sushi treat for you tonight!
01:17 karolherbst: gnurou: I found some differences, but they didn't change the performance
01:17 karolherbst: gnurou: like the seventh instruction is marked as dual issued, but I don't think that makes sense
01:17 karolherbst: even the binary drivers doesn't do this
01:17 pmoreau: I could try that on my 960!
01:17 karolherbst: pmoreau: and which firmware do you use
01:17 karolherbst: :D
01:17 pmoreau: …
01:18 pmoreau: Well, I would also need a computer to plug it into.
01:18 pmoreau: But, my workstation could do at worse.
01:19 karolherbst: gnurou: but I think the main issue is binary optimizazions on kepler, because the pixmark_piano (a gpu core only benchmark) runs at around 50% performance, and that's the biggest difference I found so far
01:19 pmoreau: gnurou: Any idea when we could see a public version of uploading code + firmware (not necessarily upstreaming-ready)?
01:25 gnurou: pmoreau: high secure firmware is still under code review, then we need to sign it, get permission to release, and we should be good
01:25 gnurou: before Christmas?
01:25 gnurou: I hate giving ETAs though
01:27 pmoreau: Ok, great! Do you know if it also works on other chipsets, or is some more modifications needed?
01:35 karolherbst: gnurou: will the firmware be different from what the binary driver uses?
01:37 gnurou: the part that performs secure boot, yes - at least at the moment. FECS and GPCCS will be the only secure firmwares released in a first time, and they will be identical to resman. I will think about the other falcons once GR is merged.
01:39 gnurou: so this means that in a first time, PMU will only be used to perform secure boot of GR - but I will try to fix that ASAP.
01:39 gnurou: PMU fw release is the next battle...
01:41 gnurou:should write some doc that explains how secure boot works since the above is certainly confusing without some context
01:42 karolherbst: what is FE?
01:42 gnurou: pmoreau: the plan is to release fw for all GM20X. Kernel will be able to load all of them, both dGPU and Tegra.
01:43 gnurou: Front End I believe (of GR)
01:43 pmoreau: Great to hear!
01:43 hakzsam: cool
01:44 pmoreau: So no talk about it at FOSDEM? :'-(
01:44 gnurou: well feel free to talk about it :) I don't think I can make the trip sadly
01:45 gnurou: will be at XDC though
01:45 gnurou: but mupuf may have other talk plans for me there :P
01:46 pmoreau: Ah ah ah! And you're going to give me all the necessary documentation so that I can understand what's going on? :-D
01:46 gnurou: the source code is heavily commented ;)
01:46 gnurou: I also don't mind answering questions ;)
01:46 pmoreau: I have to take care of SPIR-V though, and prepare the related talk so…
01:47 gnurou: hehe, everyone is busy, that's good
01:47 pmoreau: ;-)
01:48 pmoreau: I'll be happy to look at compute on my 960 once the firmwares land.
01:49 gnurou: that would be great
01:50 pmoreau: I guess there are already some NVIDIA people looking at it, since some CUDA related patches landed in chrome-os (IIRC)
01:50 hakzsam: pmoreau, gm20X compute seems to be quite similar to kepler...
01:50 gnurou: indeed
01:50 gnurou: Andrew already made a few public posts iirc
01:51 gnurou: you guys definitely want to work together
01:51 pmoreau: He didn't replied on upstreaming those to Nouveau though… :-(
01:51 marcheu: it's hard to upstream if we don't have an open source user space consumer
01:52 marcheu: which, we don't
01:53 pmoreau: Might take some time before we reach that point
01:53 hakzsam: pmoreau, btw, I fixed demmt for compute on gm20X yesterday, it can now decode the pushbuf ;)
01:53 pmoreau: Hans will probably reach it first
01:53 pmoreau: hakzsam: Yep, saw that, thanks! :-)
01:56 marcheu: but either way, you can find the change in the chrome os 3.18 kernel tree when you need them :):
01:56 marcheu: changes even
01:57 pmoreau: True enough :-) I had a look at them
01:58 marcheu: FWIW this is all working and shipping code, so you can probably mostly trust it
01:59 pmoreau: And it was written by someone having access to the documentation :-D
02:33 gnurou: too bad a lot of that code is hastily written to ship the product in time ;)
02:58 karolherbst: gnurou: like always :D
03:01 RSpliet: the world would be such a nicer place if people didn't have expectations
03:02 karolherbst: yeah and boring :p
03:27 mupuf: gnurou: very good news!
03:27 mupuf: so, good time for me to have bought the GM206 yesterday :p
03:28 mupuf: by the time the code and fw gets published, envytools should work quite well on it
03:29 mupuf: pmoreau: didn't know you had a gm206! Where is the trace and bios :p?
03:29 mupuf: anyway, something like wtrpm but easier to set up: https://www.kickstarter.com/projects/658903329/remoteboot-wifi-remote-management-module-for-pcs
03:30 mupuf:pre-ordered some C.H.I.P. to do the same thing ($8 computers with wifi on the board)
03:30 RSpliet: why not a RPi Zero? ($5) :-P
03:30 mupuf: because it has no wifi?
03:30 RSpliet: fair enough
03:30 mupuf: and the chip has populated headers
03:31 mupuf: which means no soldering
03:31 RSpliet: and the RPI is sold out, and probably a "limited edition"
03:31 mupuf: ack
03:31 RSpliet: anyway, yeah
03:31 mupuf: I will have the CHIP in january
03:31 RSpliet: I don't have much fate in CHIPs reliability, knowing the NAND controller on these AllWinner devices
03:32 mupuf: :s
03:32 mupuf: well, if it boots linux, then it is enough
03:32 mupuf: there will be no writes there
03:32 RSpliet: mmm, my experience is that there's little resilience to power cuts
03:33 mupuf: ah, not cool
03:33 mupuf: because it is kind of something that will happen
03:33 mupuf: I guess I can mount the system in ro
03:33 mupuf: that should work, right?
03:33 RSpliet: when MLC memory chips are used; whereas SLC cannot be booted by the Allwinner boot rom
03:33 RSpliet: if you want reliability, use microSD :-P
03:34 RSpliet: (or eMMC)
03:34 mupuf: that would double the price :D
03:34 mupuf: anyway, that's talk for later :p
03:34 mupuf: I am just fed up with having the ethernet cables running around
03:34 RSpliet: yeah, I don't know whether the NAND framework will mature enough before CHIP is released
03:35 mupuf: and a ad-hoc wifi network would do everything I need for discovery
03:36 RSpliet: Boris Brezillon is the man working on AllWinner Nand currently btw :-P
03:45 mupuf: no idea who he is
03:48 RSpliet: I think he gave a presentation on NAND a few years ago at FOSDEM
03:48 RSpliet: and did a GSoC on it... good chap :-P
03:51 pmoreau: mupuf: I would have uploaded them (and played with the card) if I had managed to fit that card in my desktop computer. :-/
05:28 karlmag: pmoreau: sounds like you need a new (and bigger) computer case..
05:32 RSpliet: or no case at all
05:32 karlmag: RSpliet: or that I suppose.
05:32 karlmag: Personally I would prefer a case in general.
05:33 RSpliet: hmm, I agree, the unwrapping is half the fun
05:33 karlmag: hehe
05:33 pmoreau: karlmag: Exactly! I left my old computer and its case to my sister, and I'll build two new ones pretty soon it seems
05:36 karlmag: I did keep a couple of chieftech tower cases. Though I've found that new cases are somewhat different layed out.
05:39 RSpliet: interesting btw... intel now has compute shaders, but no tesselation shaders yet? :-P
05:40 RSpliet: on a more serious note, glad to hear they did a lot of infrastructure work there, that seemed like the biggest challenge for GL 4.3
07:35 Edu-J: hi! I have and old computer with a Nvidia FX5600XT (nvidia 173 drivers needed). I've tried several linux distros but in "additional drivers" just find "Processor microcode firmware for intel CUPS".
07:36 Edu-J: "Nouveau" is not in the list
07:36 Edu-J: any advice?
07:36 imirkin: Edu-J: nouveau should work fine
07:37 imirkin: Edu-J: if you're having trouble operating your distro, you should ask in a distro-specific support channel
07:37 Edu-J: all distros work fine
07:38 imirkin: so... what's the problem?
07:38 Edu-J: but when I scroll in firefox
07:38 Edu-J: the response is awful
07:38 imirkin: are you using blob or nouveau?
07:39 Edu-J: I don't really know. As I said just "intel-microcode" is in the "additional drivers"
07:39 imirkin: Edu-J: pastebin dmesg
07:39 Edu-J: 1 sec
07:39 Tom^: intel microcode is CPU specific things.
07:41 Edu-J: [ 0.000000] Initializing cgroup subsys cpuset
07:41 Edu-J: [ 0.000000] Initializing cgroup subsys cpu
07:41 Edu-J: [ 0.000000] Initializing cgroup subsys cpuacct
07:41 Edu-J: [ 0.000000] Linux version 4.2.0-19-generic (buildd@lgw01-31) (gcc version 5.2.1 20151010 (Ubuntu 5.2.1-22ubuntu2) ) #23-Ubuntu SMP Wed Nov 11 11:38:40 UTC 2015 (Ubuntu 4.2.0-19.23-generic 4.2.6)
07:42 Edu-J: [ 0.000000] KERNEL supported cpus:
07:42 Edu-J: [ 0.000000] Intel GenuineIntel
07:42 Edu-J: [ 0.000000] AMD AuthenticAMD
07:42 Edu-J: [ 0.000000] NSC Geode by NSC
07:42 Edu-J: [ 0.000000] Cyrix CyrixInstead
07:42 Edu-J: [ 0.000000] Centaur CentaurHauls
07:42 Edu-J: [ 0.000000] Transmeta GenuineTMx86
07:42 Edu-J: [ 0.000000] Transmeta TransmetaCPU
07:42 pmoreau: :-/
07:42 Edu-J: [ 0.000000] UMC UMC UMC UMC
07:42 Edu-J: [ 0.000000] x86/fpu: Legacy x87 FPU detected.
07:42 Edu-J: [ 0.000000] x86/fpu: Using 'lazy' FPU context switches.
07:42 Edu-J: [ 0.000000] e820: BIOS-provided physical RAM map:
07:42 Edu-J: [ 0.000000] BIOS-e820: [mem 0x0000000000000000-0x000000000009fbff] usable
07:42 Edu-J: [ 0.000000] BIOS-e820: [mem 0x000000000009fc00-0x000000000009ffff] reserved
07:42 Edu-J: [ 0.000000] BIOS-e820: [mem 0x00000000000f0000-0x00000000000fffff] reserved
07:42 Edu-J: [ 0.000000] BIOS-e820: [mem 0x0000000000100000-0x000000007ffeffff] usable
07:42 Edu-J: [ 0.000000] BIOS-e820: [mem 0x000000007fff0000-0x000000007fff7fff] ACPI data
07:42 Edu-J: [ 0.000000] BIOS-e820: [mem 0x000000007fff8000-0x000000007fffffff] ACPI NVS
07:42 Edu-J: [ 0.000000] BIOS-e820: [mem 0x00000000fec00000-0x00000000fec00fff] reserved
07:42 Edu-J: [ 0.000000] BIOS-e820: [mem 0x00000000fee00000-0x00000000fee00fff] reserved
07:42 Edu-J: [ 0.000000] BIOS-e820: [mem 0x00000000fff00000-0x00000000ffffffff] reserved
07:42 Edu-J: [ 0.000000] Notice: NX (Execute Disable) protection missing in CPU!
07:42 Edu-J: [ 0.000000] Legacy DMI 2.3 present.
07:42 Edu-J: [ 0.000000] DMI: MICRO-STAR INC. MS-6758/MS-6758, BIOS V3.6 10/08/2004
07:42 Edu-J: [ 0.000000] e820: update [mem 0x00000000-0x00000fff] usable ==> reserved
07:42 imirkin: ouch
07:42 Edu-J: [ 0.000000] e820: remove [mem 0x000a0000-0x000fffff] usable
07:42 Edu-J: [ 0.000000] e820: last_pfn = 0x7fff0 max_arch_pfn = 0x1000000
07:42 Edu-J: [ 0.000000] MTRR default type: uncachable
07:42 Edu-J: [ 0.000000] MTRR fixed ranges enabled:
07:42 Edu-J: [ 0.000000] 00000-9FFFF write-back
07:43 pmoreau: imirkin: Any way to stop it?
07:43 Edu-J: [ 0.000000] A0000-BFFFF uncachable
07:43 Edu-J: [ 0.000000] C0000-C7FFF write-protect
07:43 Edu-J: [ 0.000000] C8000-EFFFF uncachable
07:43 Edu-J: [ 0.000000] F0000-FFFFF write-back
07:43 Tom^: this escelated fast.
07:43 Edu-J: [ 0.000000] MTRR variable ranges enabled:
07:43 Edu-J: [ 0.000000] 0 base 000000000 mask F80000000 write-back
07:43 Edu-J: [ 0.000000] 1 disabled
07:43 Edu-J: [ 0.000000] 2 disabled
07:43 Edu-J: [ 0.000000] 3 disabled
07:43 Edu-J: [ 0.000000] 4 disabled
07:43 imirkin: that fixed it
07:43 pmoreau: Yup
07:43 Tom^: atleast we know now he is on ubuntu with 4.2.0 :P
07:43 pmoreau: :-D
07:43 karlmag: I *think* (s)he might not know about the pastebin concept.
07:44 pmoreau: It's great that each line took some time to be pasted
07:44 imirkin: yeah, anti-flooding measures
07:45 Tom^: isnt that a freenode thing?
07:45 imirkin: it's a pretty common thing on all irc servers i've seen since the 90's
07:45 Tom^: hm perhaps, i guess when i spammed ascii arts in my youths was in the 90s :P
07:48 imirkin: yeah, coz of people like you ;)
07:48 karolherbst: but seriously, how can someone think it is a good idea to paste the entire dmesg into irc :D
07:48 imirkin: pretty easily
07:48 imirkin: apparently :)
07:49 karolherbst: :D
07:51 karlmag: karolherbst: I think you're at the core of something... To think or not to think.. is that even a question? :-P
07:52 karolherbst: or maybe that person missunderstood the order and followed it blindly ;)
07:53 karlmag: yeah, with lack of knowledge about the concept "to pastebin"
07:54 Tom^: and thus we concluded it was imirkin misleading helping that caused this misdeeds.
07:54 Tom^: shame on you
07:54 imirkin: aye
07:55 imirkin: glad we got to the bottom of that, i was getting worried!
07:56 karlmag: hehe
07:57 karlmag: freaking awesome... managed to tip my water bottle into my lap... around 2-3dl of water in my lap, on the chair and on the floor... *sigh*
07:57 karlmag: :-P
07:58 karlmag: I wear black pants, it's dark outside and I have a fairly long jacket/coat...
07:58 karlmag: so... oh well
07:59 karlmag: I think I'll take that as a cue to go home..
08:08 imirkin: ... or to wear rubber pants next time
09:09 paneidos: hi, can anyone help me get this bug solved? https://bugs.freedesktop.org/show_bug.cgi?id=92178
09:13 imirkin: i seem to recall you and i debugged this for a bit
09:13 imirkin: after which time i told you to create a bugzilla entry... do you remember if i concluded anything else?
09:14 paneidos: imirkin: you didn't conclude anything else according to the IRC logs
09:15 paneidos: imirkin: apart from concluding the data for lvds is useless, iirc
09:16 imirkin: sad
09:16 imirkin: i'll bbiab...
09:16 paneidos: imirkin: there is this line however: 17:17 < skeggsb_> conclusion: we need to scare up some other way of detecting the panel
09:16 imirkin: RSpliet: btw, i just added short-imad code locally. might try to get the immediate inlining going as well, although it's trickier since there's merges in the way.
09:31 paneidos: imirkin: I gotta go eat, dinner is ready. I did try a boot with nouveau.debug=1, output can be found here: https://paneidosdesu.stackstorage.com/index.php/s/THwzTiFAYcyyEsl
09:45 imirkin_: bleh, i have to download it to see it? not worth it :p probably nothing of note in there anyways
10:26 paneidos: imirkin: got you a paste here: http://paste.pound-python.org/show/GG8wD5pz3dtNYhPYPo2p/
10:26 paneidos: imirkin: but not much beyond the one in the bug report
10:28 imirkin_: oh right. coz nouveau.debug=1 doesn't actually do anything
10:29 imirkin_: you can do nouveau.debug=debug drm.debug=0xe
10:30 paneidos: imirkin: I'll do that, give me a sec
10:33 john_cephalopoda: Hi
10:34 paneidos: imirkin: booted with those options, dmesg here: http://paste.pound-python.org/show/gT91O8FkNBvFVBeRN8Uj/
11:01 imirkin_: airlied_: i suspect that the commit i just pushed out (79bff488bc) will fix that CTS crash with the null vertex buffer thing
11:01 imirkin_: airlied_: dunno how hard it is for you to run it, but if you get a chance, might be interesting to see what it does.
11:02 imirkin_: paneidos: right. so the issue was that we need the panel info but we don't have it right? no amount of debug options is going to fix that :)
11:02 imirkin_: paneidos: i guess we need to go spelunking in the vbios or... something
11:52 john_cephalopoda: When I activate Atmospheric Light Scattering in flightgear, everything becomes yellow and green: http://i.imgur.com/9xJCvly.jpg
11:53 imirkin_: john_cephalopoda: yeah, this has been reported before
11:53 imirkin_: iirc there might even be a bug about it
11:53 john_cephalopoda: I think it's an other bug.
11:53 imirkin_: oh whoa, yeah that looks different
11:53 imirkin_: is this a recent thing?
11:53 john_cephalopoda: This one is only on the interior while the other reported bug was only on the exterior.
11:53 john_cephalopoda: Not sure, if it is recent.
11:53 imirkin_: can you make an apitrace?
11:53 john_cephalopoda: How do I do that?
11:54 john_cephalopoda: Ah, a program.
11:54 imirkin_: https://github.com/apitrace/apitrace
12:08 john_cephalopoda: https://bpaste.net/show/1e26ffe07d74
12:09 john_cephalopoda: imirkin_: ^
12:10 imirkin_: sad :(
12:12 john_cephalopoda: imirkin_: Hmm?
12:22 imirkin_: john_cephalopoda: what version of mesa are you using and what gpu?
12:26 john_cephalopoda: GL_RENDERER = Gallium 0.4 on NVE6 || GL_VERSION = 3.0 Mesa 11.0.7 || GL_VENDOR = nouveau
12:26 john_cephalopoda: 01:00.0 VGA compatible controller: NVIDIA Corporation GK106 [GeForce GTX 645 OEM] (rev a1)
12:28 john_cephalopoda: Also happened on 11.0.6 and some 10.x
12:28 john_cephalopoda: Kernel-nouveau from Linux 4.3.0
12:29 john_cephalopoda: imirkin_: Did you time out? Did you get my messages?
12:29 imirkin_: john_cephalopoda: nope, still here..
12:30 imirkin_: and even if i had, there are logs
12:30 john_cephalopoda: Yeah, just remembered that.
12:30 Tom^: being online in irc doesnt necessery mean you are infront of the keyboard
12:31 imirkin_: either way... there's a particular source of kepler fail that i'm aware of which COULD end up in something like that, but a bit unlikely. if you feel like applying a patch, you could try this: https://github.com/imirkin/mesa/commit/9fc207b9394c7b62c80b01994d97d3cfe9ce0971
12:33 imirkin_: john_cephalopoda: you could also try git master and see if one my random changes has fixed it
12:34 Tom^: im dling flightgear, gonna see if i get same issue
12:34 john_cephalopoda: I'm trying to apply the patch
12:35 imirkin_: there are a handful of issues that didn't make it into 11.0.7 that i fixed, but i find it unlikely they'd cause those particular visual artifacts
12:39 john_cephalopoda: I added that patch and I'm compiling now.
12:39 john_cephalopoda: Using a source-based distro is a bit of a hassle sometimes but it makes it easier to fiddle with code.
12:40 Tom^: john_cephalopoda: wait, how do i run flightgear? O_o
12:41 john_cephalopoda: Tom^: The executable is called "fgfs"
12:41 Tom^: ah
12:41 john_cephalopoda: View -> Rendering options -> Atmospheric Light Scattering
12:45 Tom^: john_cephalopoda: nop works fine here, tho im on 11.2.0_devel.74897.0ef5c8a-1
12:45 Tom^: so either its specific to your gpu or its already fixed without imirkin_ patch he just showed. :p
12:45 Tom^: imirkin_: or did that make it into mesa?
12:46 imirkin_: Tom^: GK110 has more registers than GK104, so it could still be that texture thing... just how the elemental thing only showed up on GK110
12:46 imirkin_: i.e. purely due to luck and register assignments
12:46 Tom^: ah
12:47 Tom^: http://i.imgur.com/LW7DIys.jpg 28 fps with everything maxed under rendering options :p
12:48 imirkin_: helps to have a monster gpu ;)
12:48 Tom^: seems i broke my uh horizontal gadget/panel
12:48 Tom^: it thinks im rolled 45 degrees :D
12:49 imirkin_: not enough yaw!
12:50 Tom^: LOL im not meant to pilot planes http://i.imgur.com/VdqntwH.png , now im leaving this at that.
12:51 imirkin_: er, that looks all messed up...
12:51 Tom^: imirkin_: game bug, i went underground. :p
12:51 imirkin_: oh ok
12:51 Tom^: john_cephalopoda: do you play this with a controller or the keyboard?
12:55 john_cephalopoda: Tom^: Keyboard.
12:56 john_cephalopoda: Using mouse for controlling movement.
12:57 john_cephalopoda: imirkin_: Your patch fixed it.
12:58 imirkin_: john_cephalopoda: ok cool. ugh
12:58 john_cephalopoda: It also fixed an other really ugly bug. Everything looks great now. Thank you very much.
12:58 imirkin_: i need to think about how to fix it properly
13:13 paneidos: imirkin_: what do you mean by spelunking in the vbios?
13:13 imirkin_: looking for info in the vbios
13:14 paneidos: imirkin_: any way I can help?
13:14 imirkin_: not without doing the work :)
13:14 imirkin_: you could help by figuring out the info we're missing
13:14 imirkin_: you could also make a mmiotrace of the blob, that could provide useful info
13:25 paneidos: ok, I read about mmiotrace, but it seems to me I need userspace stuff to start it
13:26 imirkin_: you just need to run X with the blob drivers
13:26 imirkin_: paneidos: alternatively, if xf86-video-nv works for you
13:26 imirkin_: that could be a place to look
13:26 imirkin_: [i probably suggested all that last time you inquired]
13:28 paneidos: by blob you mean nvidia-drivers? probably need to dig some old ones up then
13:28 imirkin_: yes, nvidia-drivers. you probably need the 96 series
13:28 paneidos: and no, you did not tell me this last time
13:29 imirkin_: but first try xf86-video-nv
13:30 imirkin_: and the nvidia framebuffer in the kernel, which is a totally separate driver
13:30 imirkin_: if either of those work
13:30 imirkin_: then we can look at what they do
13:32 paneidos: ok, i'll go work on that
14:00 imirkin_: hakzsam: actually user_buffer is the opposite of UBO :)
14:01 imirkin_: for maximal confusion
14:01 imirkin_: (it makes sense once you understand it, but def confusing until you do)
14:01 imirkin_: client_buffer might have been a better name
14:02 imirkin_: basically it's "not a buffer object, but random application memory"
14:02 hakzsam: oh, I didn't know :)
14:02 hakzsam: weird names
14:03 imirkin_: well... it makes sense... UBO is a GL name, the DX name is something else
14:03 imirkin_: a user buffer is a user-supplied memory area
14:03 imirkin_: but yeah, def at odds with GL naming
14:03 imirkin_: like a lot of gallium stuff
14:04 hakzsam: okay, makes sense now
14:06 hakzsam: imirkin_, well, I thought that the U of UBO was User but it's actually Uniform... :)
14:06 imirkin_: hehe
14:14 paneidos: imirkin_: xf86-video-nv works
14:14 imirkin_: paneidos: awesome!
14:14 imirkin_: so now to figure out where it gets info from
14:16 paneidos: ok, I'll probably need to enable some stuff in the kernel
14:17 imirkin_: paneidos: can you include the output of 'lspci -nnv -d 10de::300' in the bug?
14:18 imirkin_: er actually nm. dmesg has the relevant info.
14:19 imirkin_: hm, looks like it sticks your pci id into a flatpanel = 1 list
14:20 paneidos: imirkin_: I think I need CONFIG_MMIOTRACE and CONFIG_DEBUG_FS, anything else?
14:20 imirkin_: no that's it
14:21 imirkin_: get a mmiotrace of xf86-video-nv and nouveau and try to compare them i guess
14:23 imirkin_: paneidos: can you also include your xorg log for xf86-video-nv?
14:23 imirkin_: there are a few points of interest in there
14:23 paneidos: ok
14:24 hakzsam: imirkin, btw, there are no regressions with piglit
14:24 imirkin_: hakzsam: awesome. presumably with the new version -- old one should have had some crashes...
14:25 hakzsam: some? 95% of piglit :D
14:25 hakzsam: the v2 was a crap...
14:25 hakzsam: the new one is much better
14:30 paneidos: imirkin_: uhm
14:31 paneidos: mmiotrace of nv is empty
14:31 imirkin_: paneidos: ohhhh... um. hm. mmiotrace might not work with xf86-video-nv
14:31 paneidos: probably because there is no kernel driver
14:31 imirkin_: can you attach the xorg log from it though?
14:32 paneidos: I did capture that one
14:36 paneidos: imirkin_: https://bugs.freedesktop.org/attachment.cgi?id=120457
14:51 paneidos: imirkin_: the old nvidia-drivers require some work to install, xorg must be downgraded and I need an older kernel. the former is still in the package tree, the latter is not, but I might be able to get one from an older system
14:52 imirkin_: paneidos: i wouldn't worry about that
14:52 imirkin_: paneidos: i think the info should be enough to make progress
14:52 imirkin_: for an interested party
15:16 paneidos: imirkin_: thanks for all the help so far, I'm off to go sleep now.
16:55 imirkin_: Tom^: fyi this is a better fix for that texbar issue: http://patchwork.freedesktop.org/patch/67752/
16:57 imirkin_: it also paradoxically helps a bunch of shaders... i think in ways that don't actually matter though.