02:20pmoreau: I didn't expected to see the same EVO calls without acceleration and with: I thought EVO would be considered as some kind of acceleration.
02:25pmoreau: So I'll have to find out what in the acceleration stuff makes EVO unhappy.
07:22Dan39: hi. nouveau is throwing tons of errors. mostly when i try to start firefox... http://sprunge.us/ehaS
07:23Dan39: it is doing it at other times too. just restarted wm, and ran glxgears, it displayed fine, but when i closed it, then X froze
07:23Dan39: all except for the mouse cursor freezes :/
07:23Dan39: sometimes it will start responding again, sometimes it wont
07:24imirkin: Dan39: do you have a G92?
07:24imirkin: remove libvdpau_nouveau.so
07:25Dan39: ah yes i had added that
07:25Dan39: i guess it isnt completely stable? :P
07:25imirkin: it's known to be broken on G92's
07:25imirkin: i haven't had time to look into it
07:25Dan39: i read that mine card supported it, so gave it a try haha
07:25imirkin: should probably kill it in the kernel so that it doesn't cause harm
07:26Dan39: hmmm wait something seems a lil fishy here with packages i have installed
07:26Dan39: minor paste...
07:26Dan39: extra/libvdpau 1.1-1 [installed]
07:26Dan39: Nvidia VDPAU library
07:26Dan39: extra/mesa-vdpau 10.5.6-1 [installed]
07:26Dan39: Mesa VDPAU drivers
07:26Dan39: should i have both installed? <_<
07:26imirkin: but remove mesa-vdpau, that should avoid the plugin-container fail
07:26Dan39: o, ok. thought maybe i shouldnt and some package maintainer missed that
07:26imirkin: (which i assume is making use of vdpau)
07:27imirkin: you should probably reboot the box and i suspect your issues will go away
07:27imirkin: anyways, i gtg. good luck!
07:34Dan39: well lets see :D
07:35aaronp: Dan39, the description on libvdpau in the Arch database is wrong. It's not an NVIDIA-specific library.
07:36Dan39: i see
07:36Dan39: one is lib other is drivers
07:37Dan39: can anyone recommend a program, possibly with sliders, to adjust brightness and gamma?
07:37aaronp: Right. libvdpau is a wrapper that loads the appropriate back-end driver depending on your graphics drivers.
07:37Dan39: i usually do it by xrandr commands, but it is horribly tedious typing in the different values over and over
07:46Karlton: Dan39: not sure about gamma, but I know gnome and kde have sliders for brightness
07:47Dan39: i wonder which package provides that, or if its part of the whole big thing
07:47Dan39: id rather not install one of those
07:48Karlton: you could create key bindings for it as well using xbindkeys
07:51Karlton: it is possible that xfce4 has one also that wouldn't be dependency heavy, but I am not sure
07:52Dan39: omg what did i just do setting brightness -1 hahaha
07:52Dan39: inverted the colors
07:52Dan39: kind of...
07:53Dan39: set some like sepcial dark mode
07:54Dan39: ok so setting -1 +1 etc. just sets it to 0 or 1 <_<
07:55Dan39: wow it does really weird stuff in the negatives
08:22imirkin_: aaronp: it was made by nvidia, and it *is* a vdpau library ;)
08:23imirkin_: it's accurate but incredibly misleading
08:23imirkin_: Dan39: btw, you may be interested in ^R
08:23imirkin_: (history search in any readline-enabled program, including bash)
12:25moochmcgee: imirkin_: I found an old reference for nvidia registers. It calls my elusive CRTC register 38h RMA
12:26moochmcgee: Which is confusing because of there's an RMA BAR on NV3.
12:27moochmcgee: any idea what's going on?
12:27moochmcgee: also, windows 9x drivers write 5, 3, 5, 3, 7, 0 in that order over and over again
12:27imirkin_: haha no
12:28moochmcgee: i got this from the rivatv website
12:28imirkin_: i believe you
12:29imirkin_: i just haven't the faintest clue
12:29imirkin_: i'd look at other writes that those drivers do
12:29moochmcgee: There's nothing useful
12:29imirkin_: you might also find vbetrace instructive
12:29moochmcgee: just stuff for vesa
12:29moochmcgee: though vesa modes do this too
12:30moochmcgee: but they actually work
12:30imirkin_: vbetool post? i forget exactly
12:30imirkin_: you can get a trace of what all vbe does
12:30moochmcgee: well, i don't have a real card
12:30moochmcgee: i have a later nvidia card...
12:31imirkin_: oh. that's unfortunate.
12:31moochmcgee: well, i do have an AGP nv3 but no agp computers
12:31imirkin_: so close
12:31moochmcgee: writing 1 to bit 0 of that register cleared the screen on someone's card
12:32imirkin_: you may be interested in http://envytools.readthedocs.org/en/latest/hw/bus/prma.html though
12:32moochmcgee: i already looked over that
12:32moochmcgee: it doesn't try to access that bar
12:33moochmcgee: but i have the feeling that this is the key to getting linear framebuffers to work
12:33moochmcgee: so far, i at least have banking registers implemented
12:33imirkin_: you really should try to get ahold of mwk
12:34imirkin_: i think he's done a lot of RE of nv3
12:34imirkin_: he's in this chan
12:34moochmcgee: yeah, i saw
12:34imirkin_: but i don't think i've seen him say anything in 6+ months
12:34moochmcgee: ping mwk i need help with my riva 128 emulator
12:35moochmcgee: once i'm done getting riva 128 to work in pcem, i'm going to go after the intel 740
12:35imirkin_: preserving old gpu history?
12:35moochmcgee: pretty much
12:35imirkin_: there are hwtests in envytools for the nv3 accel stuff
12:36moochmcgee: where in the repo is this? i'm too lazy to look
12:36imirkin_: iirc they had something super-funky
12:36imirkin_: and non-triangular
12:36imirkin_: nv3_pgraph iirc
12:36moochmcgee: i think you want nv01_pgraph.c
12:37moochmcgee: but okay
12:37imirkin_: oh my bad.
12:37imirkin_: yeah. nv01_pgraph indeed. fail.
12:37imirkin_: your gpu's too recent ;)
12:38moochmcgee: i've got an nvd9
12:38imirkin_: i meant the nv03
12:38imirkin_: but the nvd9 is also too recent for nv01_pgraph
12:38moochmcgee: although this doesn't test the NV_CIO registers
12:39imirkin_: well the hwtest is only for pgraph
12:39moochmcgee: but that's what i need to look at
12:39imirkin_: i think the nv3 pgraph was similar to nv4, but used a diff fifo submission mechanism?
12:40imirkin_: shouldn't be too hard to add nouveau support for, should someone be motivated
12:40imirkin_: the framework is pretty generic in that regard now
12:40moochmcgee: yeah, but i'm not good with graphics drivers. :)
12:41imirkin_: it's not graphics
12:41imirkin_: just the kernel bits
12:41imirkin_: to be able to feed commands in
12:41imirkin_: the drivers to make the commands happen already exist
12:41moochmcgee: eh, true
12:41moochmcgee: still, emulators are my true calling
12:41imirkin_: anyways, this is all predicated on having an accurate pgraph impl
12:42imirkin_: you could cheat and farm it out to GL :)
12:42moochmcgee: that's not the way this emulator does it
12:42moochmcgee: it already emulates the s3 virge and voodoo 1
12:42moochmcgee: both accurately
12:42moochmcgee: except the voodoo filter isn't quite done yet
12:43imirkin_: does it have a virtio interface?
12:43imirkin_: how do i plug it in? :)
12:43moochmcgee: oh, that
12:43moochmcgee: it's kinda complicated
12:43moochmcgee: lots of things have to be hooked up
12:43moochmcgee: you could look at my code for an example.
12:43imirkin_: you might want to talk to the linux kvm people, they might have suggestions on integrating it nicely
12:44moochmcgee: well, it's not my emulator
12:44moochmcgee: it's someone else's
13:02mwk: moochmcgee: sounds like a fun project
13:02mwk: unfortunately I'm in the middle of crunch time here
13:03mwk: please talk to me on saturday
13:24Dan39: imirkin_: lol yes i use ctrl-r all the time. in this case just hitting the up arrow is even faster you know ;)
17:17oldornew: What do you think is happening when I'm browsing the internet through chrome and in random places square blocks of the screen have random colors, sometimes some text that is obviously from some other texture? Chrome's error, Driver's error, Hardware error?
17:17oldornew: I'm just looking for gpu experts to ask this, the driver I'm using is native nvidia.
17:17tobijk: libdrm 2.4.60?
17:17oldornew: This is happening for the first time now, just as I'm using gpugrid@home that's probably using 100% of the gpu
17:18tobijk: then you are wrong herem this is not a channel about the nvidia driver
17:18tobijk: go ask their support
23:36pmoreau: How should I proceed to get EVO stuff happening when loading Nouveau? Because mmt on insmod/modprobe doesn't work
23:39pq: wasn't mmt a thing to trace accesses from user space only?
23:40pq: but the kernel module is obviously running in kernel space, so no dice
23:40pmoreau: Oh maybe
23:40pq: there's mmiotrace to trace kernel accesses, but it doesn't catch accesses from user space
23:42pmoreau: I know about mmiotrace, but it would like submitted EVO methods and data
23:42pmoreau: I'll go through the code then
23:52pmoreau: I could use debug=spam + enable EVO prints to get everything in dmesg.