=== bryce_ is now known as bryce2 | ||
ripps | lucid ati drivers still use mesa, right? I thought that gallium3d was ready for ati now. | 02:08 |
---|---|---|
RAOF | radeon gallium is still pretty new; I don't think *any* of the gallium drivers are ready for primetime yet. | 02:10 |
RAOF | Including the intel one, which has been going the longest :). | 02:10 |
RAOF | I could be wrong, though. I don't pay that much attention to the radeon side of things. | 02:11 |
jcristau | the intel one is probably the least ready | 02:12 |
RAOF | Which seems odd, but I presume is due to laribee slipping and slipping and slipping. | 02:14 |
jcristau | or maybe i don't know which one you're talking about. there's probably a poulsbo gallium driver. but meh, poulsbo. | 02:15 |
RAOF | Not one that we have the source to, I presume :) | 02:15 |
jcristau | right | 02:16 |
RAOF | My memory is correct, right? Tungsten Graphics, of Intel-driver fame, started gallium? As what seemed to me to be a project to make Intel's upcoming Larribee super-x86 GPU work? | 02:18 |
NoelJB | As of today, I have a situation where (and I have uninstalled and reinstalled to check), 2.6.32-11 works with nvidia-current, and 2.6.32-12 will start with it, but as soon as I open a terminal window and do nothing other than hit <cr>, the system hangs (requires hard power-off). | 03:33 |
NoelJB | weird thing is that it was working earlier today. | 03:44 |
tjaalton | bdmurray: if it's a serial tablet it won't work yet | 04:31 |
tjaalton | oh, usb id.. | 04:32 |
tjaalton | nevermind then | 04:32 |
tjaalton | * Add klingon language definition. | 04:46 |
tjaalton | in libx11... | 04:46 |
tjaalton | like that's ever going upstream? | 04:46 |
tjaalton | and failed to build | 05:57 |
libv | that'll really help the spread of the linux desktop. | 06:04 |
tjaalton | "look, even trekkies love it" | 06:10 |
libv | the marketing potential of that is enormous. | 06:20 |
libv | trekkies are also a big and popular subculture | 06:20 |
libv | especially known for their low geek count :) | 06:21 |
libv | the uptake of free software with trekkies, since they spend soo much time watching telly and playing trekkie windows/console games, must be very low | 06:22 |
libv | so the potential of growth there is enormous! | 06:22 |
libv | and what young boy doesn't dream of being able to use the ever so popular "openoffice turned my homework into klingon" excuse | 06:23 |
tjaalton | tlhIngan maH! | 06:29 |
libv | is there a definition for vulcan? | 06:31 |
tjaalton | no, that should be fixed | 06:31 |
libv | right, because otherwise this whole brilliant world domination scheme will fail | 06:32 |
libv | s/world/space/ | 06:32 |
Sarvatt | RAOF: need to have lbm-nouveau conflict with all the nvidia-stuff it looks like | 15:20 |
Sarvatt | [ 26.151671] NVRM: The NVIDIA probe routine was not called for 1 device(s). | 15:21 |
Sarvatt | [ 26.151675] NVRM: This can occur when a driver such as rivafb, nvidiafb or | 15:21 |
Sarvatt | [ 26.151677] NVRM: rivatv was loaded and obtained ownership of the NVIDIA | 15:21 |
Sarvatt | [ 26.151678] NVRM: device(s). | 15:21 |
Sarvatt | ah nevermind that was because of my xorg.conf from the nvidia package, thats why it was so late into the boot | 15:26 |
Sarvatt | hmm, why did X start with display :0 on vt1 with nouveau? http://sarvatt.com/downloads/nouveau.1.log | 15:28 |
Sarvatt | also, XAA? RAOF, how did you change the nouveau probe routine to always asume KMS? | 15:29 |
tjaalton | Sarvatt: the log is with -nv btw | 15:31 |
Sarvatt | argh thanks tjaalton | 15:31 |
* Sarvatt needs morning coffee before messing with these things :) | 15:31 | |
Sarvatt | so the xserver patch didnt work there | 15:32 |
Sarvatt | ah my fault of course | 15:32 |
Sarvatt | thats the machine i was playing with 100_rethrow_signals.patch on so i had the same version | 15:32 |
_stink_ | hey folks - i'm playing around with X drivers. little question: i see that x drivers in packages are .so files. when i compile a sort-of-obscure x driver, the ./configure;make ends up with a .la file. i'm vaguely aware that these are different, but can X load the .la if i put it in the module directory along with the other .so files? | 15:35 |
jcristau | no | 15:36 |
jcristau | you should get a .so with the .la | 15:36 |
_stink_ | ok. any hint on the linking or whatever step that's missing that i need to get the .so? | 15:38 |
jcristau | there isn't one. libtool gives you a .so together with the .la | 15:39 |
_stink_ | ok. thanks for the info! | 15:40 |
Sarvatt | http://sarvatt.com/downloads/nouveau.2.log | 15:41 |
_stink_ | oh, huh, didn't even look at the .la. silly me. | 15:41 |
_stink_ | i thought it was binary. | 15:41 |
Sarvatt | things look good this time | 15:41 |
kklimonda | hey, how can I debug not working xv with nouveau driver in lucid? The one posted on mailing list today. xvinfo says no adaptors present but it worked just a moment ago (maybe a restart ago) | 15:43 |
Sarvatt | the lbm stuff is loaded unconditionally if you have nvidia hardware though, nvidia isn't working when its installed at the same time | 15:43 |
Sarvatt | what does cat /var/log/Xorg.0.log | grep Xv say? | 15:44 |
kklimonda | empty | 15:45 |
kklimonda | grep XV shows: (II) Loading extension XVideo | 15:45 |
kklimonda | (II) Loading extension XVideo-MotionCompensation | 15:45 |
Sarvatt | pastebin the whole log? | 15:45 |
Sarvatt | sarvatt@arcueid:/etc/X11$ cat /var/log/Xorg.0.log | grep Xv | 15:45 |
Sarvatt | (II) NOUVEAU(0): [XvMC] Associated with Nouveau GeForce 8/9 Textured Video. | 15:45 |
Sarvatt | (II) NOUVEAU(0): [XvMC] Extension initialized. | 15:45 |
kklimonda | http://pastebin.com/f4d9e8f42 | 15:46 |
Sarvatt | (EE) NOUVEAU(0): Error creating GPU channel: -16 | 15:47 |
Sarvatt | (EE) NOUVEAU(0): Error initialising acceleration. Falling back to NoAccel | 15:47 |
Sarvatt | do you have the nouveau firmware installed? | 15:47 |
Sarvatt | i had it installed before i installed the ppa packages so i'm not sure about that part, lets see | 15:48 |
kklimonda | yes, I have | 15:48 |
Sarvatt | dmesg | grep firmware show it loading both? | 15:49 |
Sarvatt | [ 2.085401] nouveau 0000:05:00.0: firmware: requesting nouveau/nv86.ctxprog | 15:50 |
Sarvatt | [ 13.683234] nouveau 0000:05:00.0: firmware: requesting nouveau/nv86.ctxvals | 15:50 |
kklimonda | yes - both | 15:50 |
Sarvatt | kklimonda: sorry, had to run to a job there, would you mind putting your dmesg on pastebin too? | 17:16 |
Sarvatt | kklimonda: don't know if you saw my message because of the split right after but would you mind putting your dmesg on pastebin too? | 17:33 |
kklimonda | Sarvatt: http://pastebin.com/f18c60795 | 17:37 |
Sarvatt | hmm, and it happens every boot? [ 17.394177] [lbm-drm] nouveau 0000:01:00.0: GPU lockup - switching to software fbcon is the only thing that stands out compared to mine | 17:48 |
kklimonda | well, not every boot because it worked once - but since then I get this GPU lockup every time | 17:50 |
kklimonda | once as at one boot :) | 17:51 |
Sarvatt | oh just realized i have quiet splash removed from the kernel command line still because nouveau had problems with it in the past, lets see if it works with it there now :) we have basically the same GPU | 17:59 |
Sarvatt | indeed thats the problem | 18:00 |
Sarvatt | (EE) NOUVEAU(0): Error creating GPU channel: -16 | 18:00 |
Sarvatt | (EE) NOUVEAU(0): Error initialising acceleration. Falling back to NoAccel | 18:00 |
Sarvatt | remove quiet splash from your /etc/default/grub GRUB_CMDLINE_LINUX_DEFAULT="quiet splash whatever" line and update-grub after will fix it for now, I know thats no real fix | 18:01 |
Sarvatt | actually, I ran update-initramfs -u between those reboots as well, let me make sure it wasnt that first | 18:03 |
Sarvatt | i had that problem in the past for a long time though and i had to remove quiet splash to work around it | 18:04 |
Sarvatt | i did the update-initramfs because I noticed your ctxvals was getting loaded right after ctxprogs and mine was getting loaded 12 seconds after | 18:05 |
Sarvatt | and now its getting loaded at the same time | 18:05 |
Sarvatt | yep it was quiet splash, removed it again and it works | 18:07 |
Sarvatt | (II) NOUVEAU(0): [XvMC] Associated with Nouveau GeForce 8/9 Textured Video. | 18:07 |
Sarvatt | (II) NOUVEAU(0): [XvMC] Extension initialized. | 18:07 |
Sarvatt | (this kind of defeats the whole purpose of nouveau KMS getting brought in so early for the nice splash across all the major GPU's though) | 18:08 |
Sarvatt | works fine with just quiet, just splash screwing with it | 18:12 |
Sarvatt | else if (strcmp (driver_name, "nouveau") == 0) | 18:17 |
Sarvatt | yeah plymouth needs a little adapting for lbm-nouveau I think | 18:17 |
Sarvatt | http://cgit.freedesktop.org/plymouth/tree/src/plugins/renderers/drm/plugin.c | 18:19 |
Sarvatt | good news, the plymouth change fixed it | 18:29 |
Sarvatt | line 444 of http://cgit.freedesktop.org/plymouth/tree/src/plugins/renderers/drm/plugin.c | 18:29 |
Sarvatt | changing it to else if (strcmp (driver_name, "lbm-nouveau") == 0) | 18:30 |
Sarvatt | probably better to just add a new elseif section for lbm-nouveau but I just wanted to see if it would work | 18:31 |
kklimonda | Sarvatt: ok, thanks | 18:31 |
Sarvatt | i'll upload a fixed plymouth to that PPA | 18:31 |
Sarvatt | if i can figure out a versioning change that wont screw with the shlibs.. | 18:32 |
Sarvatt | http://sarvatt.com/downloads/patches/0001-src-plugins-renderers-drm-plugin.c-Add-check-for-lbm.patch | 18:49 |
Sarvatt | ^ non-intrusive plymouth patch to work with lbm-nouveau without wiping out normal nouveau support, works here | 18:50 |
Sarvatt | heyo bryce, if you were using the nouveau on edgers/nouveau ya have to remove splash to get acceleration it looks like, plymouth has to be adapted to the lbm-nouveau module name to use the drm renderer otherwise its screwing with the framebuffer and not initializing properly in X | 18:56 |
bryce2 | Sarvatt, ok thanks | 18:57 |
Sarvatt | i'm not sure what version i should call the new plymouth in the PPA though, its really odd | 18:57 |
Sarvatt | plymouth (0.8.0~-8) | 18:57 |
Sarvatt | i get shlibs warnings with anything i try | 18:58 |
bryce2 | why not 0.8.0~8 ? | 18:58 |
bryce2 | or 0.8.0~0ubuntu1~sarvat~1 or similar | 18:58 |
Sarvatt | 0.8.0~-8ubuntu1 doesn't work at least | 18:59 |
Sarvatt | well i guess i'll just do it and ignore the shlibs warnings :D | 19:00 |
Sarvatt | oh weird dpkg-gensymbols: warning: new libraries appeared in the symbols file: libplybootclient.so.2 | 19:00 |
Sarvatt | http://sarvatt.com/downloads/patches/0001-src-plugins-renderers-drm-plugin.c-Add-check-for-lbm.patch is the patch to fix it, i linked it on #ubuntu-devel since its not really upstream material since its specific to our nouveau rename just for one release | 19:01 |
bjsnider | refresh my memory: how much has to be removed for the nvidia installer to work? jockey and all of the packaged drivers isn't it? | 19:03 |
Sarvatt | i dont think it'll work at all regardless on lucid with the way things are done now, will it? | 19:04 |
bryce2 | Sarvatt, sounds like you need to update the symbols file? | 19:04 |
bjsnider | well, it may not work afterwards, but i mean for the installer to run | 19:05 |
Sarvatt | maybe we should just disable glx-tls in mesa so we dont have to do the whole /usr/lib/mesa or /usr/lib/nvidia-current stuff and can just use an alternative for /usr/lib/libGL directly | 19:05 |
bryce2 | Sarvatt, uhh | 19:06 |
Sarvatt | err so mesa's libGL doesn't have to get put in /usr/lib/mesa/libGL.so.1 | 19:06 |
bryce2 | Sarvatt, run this issue by tseliot first, he may have a better idea for solving that | 19:07 |
tseliot | Sarvatt: ?? | 19:11 |
tseliot | why? What happened? | 19:11 |
tseliot | tjaalton: do you know if calibration works for wacom touchscreens? It doesn't seem to work here | 19:16 |
tseliot | it says the calibration property has no items | 19:16 |
tseliot | is it still WIP? | 19:17 |
Sarvatt | tseliot: I was just saying that because we couldn't leave libGL.so.1 in /usr/lib because it wouldn't take the alternatives due to the tags because of tls and I might be missing something but noone can install nvidia drivers directly from nvidia with how our libgl is an alternative always now | 19:25 |
Sarvatt | plus I have no idea about this - https://bugs.edge.launchpad.net/ubuntu/+source/mesa/+bug/259219 | 19:26 |
ubottu | Ubuntu bug 259219 in mesa "Broken TLS support in libGL.so" [Undecided,Confirmed] | 19:26 |
tseliot | Sarvatt: right, I haven't worked on the installer yet | 19:27 |
tseliot | Sarvatt: that bug has been there since intrepid | 19:31 |
Sarvatt | ugh, I miss my htc dream, this new phone has horrible reception | 19:31 |
Sarvatt | bryce2: libplymouth2.symbols is entirely 0.8.0~ and I didnt touch any of these things getting shlibs warnings, i think the current package probably had the shlibs warnings.. i'll just version it 0.8.0~-9~nouveau I guess | 19:36 |
Sarvatt | spinfinity seems to be the best plymouth plugin IMO | 19:36 |
Sarvatt | solar has way too much cpu usage, spinfinity doesnt hardly use more than the default one and the progress bar looks muuch nicer | 19:37 |
Sarvatt | the bar takes up the whole bottom of the screen instead of being offcenter under the ubuntu logo | 19:38 |
tjaalton | tseliot: what tool are you using? | 19:39 |
tseliot | tjaalton: xinput | 19:39 |
tseliot | Sarvatt: we'll use our own plymouth theme | 19:40 |
tseliot | (not the one which is currently in Lucid though) | 19:40 |
Sarvatt | tried xsetwacom tseliot? | 19:41 |
tseliot | Sarvatt: with the new wacom driver? Would it work? | 19:41 |
tjaalton | tseliot: there are at least two calibration tools, recently discussed on xorg@. apparently they work and so should xinput | 19:41 |
jcristau | Sarvatt: did 259219 ever get forwarded upstream? | 19:41 |
tjaalton | tseliot: if you are using evdev.. | 19:42 |
tseliot | tjaalton: I'm getting 3 wacom devices | 19:42 |
tjaalton | tseliot: so use the wacom tools | 19:42 |
Sarvatt | i dont think so, tjaalton brought it up on dri-devel a few years ago it looks like but no responses | 19:42 |
Sarvatt | ok uploaded the fixed plymouth to xorg-edgers/nouveau, should be fixed when you get that update kklimonda | 19:43 |
kklimonda | Sarvatt: thanks | 19:43 |
tjaalton | bbl -> | 19:44 |
Sarvatt | kklimonda: no thank you for testing it, wouldn't have seen that bug otherwise :D | 19:57 |
Sarvatt | what do we want to do about nouveau coexistance with nvidia? | 19:58 |
Sarvatt | we need a conflicts somewhere, nvidia doesnt work with lbm-nouveau installed since that takes over the device before nvidia can, and there are a bunch of libgl errors with the nvidia libgl still hanging around while using nouveau | 19:59 |
tseliot | tjaalton: I'm using xsetwacom but things such as Rotation don't seem to work | 19:59 |
tseliot | Sarvatt: I've just uploaded a fix for that | 19:59 |
tseliot | I'm about to do the same for -173 and -96 | 19:59 |
tseliot | all we need to do is put the blacklist file in the initramfs | 20:00 |
tseliot | so that nouveau is blacklisted early | 20:00 |
tseliot | is this what you were referring to? | 20:01 |
Sarvatt | kind of | 20:01 |
Sarvatt | that'll fix nvidia to work but theres still the problem getting nouveau to work with nvidia installed | 20:02 |
tseliot | Sarvatt: why? If you switch to nouveau, do ldconfig and reboot it should work | 20:03 |
Sarvatt | are you blacklisting lbm-nouveau or just nouveau? | 20:04 |
tseliot | just nouveau | 20:04 |
tseliot | what's lbm-nouveau? | 20:04 |
Sarvatt | its called lbm-nouveau in what we're going with | 20:04 |
Sarvatt | linux-backports-modules-nouveau | 20:04 |
tseliot | aah, so you're backporting the whole thing | 20:04 |
Sarvatt | apw renamed all the relevant modules lbm-* to not interfere with the normal ones | 20:05 |
Sarvatt | thats why I had to do that change in plymouth since it was just looking for nouveau to use the drm renderer | 20:05 |
tseliot | let me correct myself: you should switch to nouveau, do ldconfig, update the initramfs and reboot | 20:05 |
Sarvatt | plus remove the xorg.conf.. | 20:06 |
tseliot | Sarvatt: ah, do you mean it was trying to use plymouth+nouveau drm with nvidia? | 20:06 |
Sarvatt | right now i have nvidia-current and nouveau installed, if nvidia-current is installed ldconf is still going to prefer the nvidia-current libgl isnt it? | 20:07 |
tseliot | Sarvatt: it depends on what you set the alternative to | 20:07 |
tseliot | update-alternatives --display gl_conf should tell you what is set to | 20:08 |
Sarvatt | nope plymouth's nouveau drm renderer plugin was hardcoded to look for the nouveau module and i had to add another match for lbm-nouveau for it to use it -- http://sarvatt.com/downloads/patches/plymouth-lbm-nouveau.patch | 20:09 |
Sarvatt | do we want to allow nvidia and nouveau to coexist on the system? wont it always prefer the nvidia libGL.so.1 if nvidia is installed becaue of the priorities? | 20:09 |
Sarvatt | plus we dont want the nvidia xorg.conf to be there if someones using nouveau | 20:10 |
Sarvatt | I know they can change the alternative manually its just doesn't seem like something we should expect a normal user to do to me | 20:14 |
Sarvatt | are you using an intuos3 tseliot? spotted a kernel rotation fix for intuos3 tablets that might not be in 2.6.32 | 20:17 |
Sarvatt | ah nope its in 2.6.32 | 20:18 |
kklimonda | Sarvatt: still no luck - I've even removed quiet and splash.. It worked at least once but after reboot I'm still getting "GPU lockup" in my dmesg. | 20:19 |
Sarvatt | hmmm, odd because i could reproduce the exact errors and it was fixed here | 20:20 |
Sarvatt | might have screwed up the packaging | 20:20 |
Sarvatt | built it locally on another machine where i had the problem and packaged it on this one, wouldn't put it past me to have applied the patch in the wrong terminal :D | 20:22 |
kklimonda | Sarvatt: I've checked source package already and your patch is there | 20:24 |
Sarvatt | yeah indeed it is.. | 20:25 |
Sarvatt | so its still giving the gpu channel -16 error in the Xorg.0.log? | 20:35 |
kklimonda | yes | 20:36 |
Sarvatt | unrelated but i just noticed a little problem with the nouveau detection patch for xserver RAOF is using | 20:39 |
Sarvatt | Screen "Builtin Default fbdev Screen 0" | 20:39 |
Sarvatt | Screen "Builtin Default vesa Screen 0" | 20:39 |
Sarvatt | Screen "Builtin Default fbdev Screen 0" | 20:39 |
Sarvatt | (loading fbdev twice) | 20:39 |
Sarvatt | its odd its setting the resolution so late for you | 20:44 |
tseliot | Sarvatt: nvidia and nouveau can coexist and if you switch to the alternative provided by nouveau you will get mesa's libGL.so.1 | 20:44 |
Sarvatt | yeah but do you really expect users to do that manually? | 20:44 |
jcristau | they can't really coexist because only one of the kernel drivers can own the device though | 20:48 |
kklimonda | Sarvatt: the whole boot is odd and random - after I press enter in grub I get either clean screen or a bit of color characters or the whole screen is set to the weird color (even before resolution is changed). Then either resolution changes and screen stays black or screen shuts down completely and It goes back on only a second before X kicks in. | 20:48 |
tseliot | Sarvatt: no, Jockey should deal with that | 20:48 |
tseliot | jcristau: a reboot will be required. And | 20:49 |
tseliot | it works here | 20:49 |
jcristau | jockey adds the blacklist as needed when you tell it to switch? | 20:49 |
tseliot | yes, it will, I'm working on it | 20:50 |
Sarvatt | does it remove the xorg.conf too? | 20:50 |
jcristau | ok | 20:50 |
tseliot | Sarvatt: it switches back to whatever was there before. I can make sure that xorg.conf is removed when switching to nouveau | 20:51 |
Sarvatt | i dont use jockey at all so all of this change just adds a ton more manual configuring to me but I guess I'm in the minority there | 20:51 |
bjsnider | is xorg.conf ever necessary when using nouveau? | 20:51 |
bjsnider | is there an option that needs to be passed to nouveau through xorg.conf? | 20:51 |
Sarvatt | not required with the xserver detection priority changes but plenty of people do use xorg.conf for other options and there is alot of stuff you can tweak in there for it | 20:53 |
bjsnider | maybe it should just be changed from nvidia to nouveau then | 20:53 |
Sarvatt | if you have an xorg.conf and dont specify the driver to use it'll only try to use nouveau and give up with the detection change, at least we dont ship the generic xorg.conf with one device section anymore | 20:54 |
Sarvatt | kklimonda: can you pastebin your dmesg with the newer plymouth? just curious at the changes since the other one | 20:56 |
kklimonda | Sarvatt: http://pastebin.com/f70276761 | 20:57 |
Sarvatt | anyone know how to tell what plymouth theme is currently in use? | 20:57 |
Sarvatt | oh yeah both of us are using nouveau.modeset=1, that doesnt do anything so we can drop it :D | 20:58 |
Sarvatt | (the modules lbm-nouveau now and it only has KMS) | 20:59 |
Sarvatt | http://sarvatt.com/downloads/nouveau.3.log | 21:01 |
Sarvatt | theres my logs | 21:01 |
kklimonda | you don't have the huge gap after "[lbm-drm] Initialized nouveau" | 21:02 |
Sarvatt | reinstalling plymouth and libplymouth2 from the PPA just incase | 21:06 |
Sarvatt | had my locally built one installed still | 21:07 |
Sarvatt | because I got the exact same GPU lockup in dmesg and noaccel in X when I had the old plymouth installed | 21:08 |
Sarvatt | ok it works fine with the PPA plymouth here - (II) NOUVEAU(0): [XvMC] Associated with Nouveau GeForce 8/9 Textured Video. | 21:12 |
Sarvatt | kklimonda: sudo update-initramfs -u -k 'all' and reboot | 21:16 |
Sarvatt | i just noticed you booted generic-pae that time and -generic before, plymouth might not have rebuilt all your initrds | 21:16 |
Sarvatt | more like, it DIDNT update all your initrd's, it only does 1 by default | 21:16 |
Sarvatt | so you had the old plymouth in the other kernel's initrd | 21:17 |
kklimonda | Sarvatt: I don't even have plymouth in initrd :) | 21:20 |
Sarvatt | parts of it go in there always afaik, it just doesnt stash the whole theme into it | 21:21 |
kklimonda | Sarvatt: I've done as you asked - no change - but it was just a slip on my part. I have both -generic and -pae installed and I've been testing both after you have uploaded new plymouth to see if there is something different between them. here is the last boot with -generic: http://pastebin.com/f174369d | 21:22 |
Sarvatt | darn, of course it wouldn't be that easy :D | 21:27 |
RAOF | Ah, good. We're debugging the -pae weirdness? | 21:30 |
kklimonda | RAOF: no - I can't get it to work at all :/ | 21:30 |
kklimonda | RAOF: and it worked twice just fine - I've even send you an email ;) | 21:31 |
RAOF | Why didn't I see the plymouth issue on my laptop, I wonder? | 21:36 |
kklimonda | hardware specific? or my gpu is dying again and that's some bizarre symptom.. | 21:36 |
Sarvatt | i'm wondering if it has something to do with the mtrr allocation problem | 21:36 |
Sarvatt | you had splash enabled before RAOF? | 21:37 |
Sarvatt | we both had to remove splash from the kernel command line messing with nouveau before and it didnt happen with splash removed | 21:38 |
Sarvatt | thats why I didnt notice it at least | 21:38 |
RAOF | I had splash enabled! I even switched the plymouth screen to spinfinity. | 21:38 |
RAOF | I *did* see some GPU lockups, but it seemed only when nv loaded, not nouveau. | 21:39 |
Sarvatt | wish we had an updated nouveau, it would display NV_WARN(dev, "failed to reserve VGA memory\n"); if its the MTRR problem | 21:40 |
RAOF | Which part is that? The kernel module? | 21:41 |
Sarvatt | yeah was looking at this from just after rc4 http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=ac8fb975e8c88d312a376b035494be17548d01c6 | 21:42 |
RAOF | There seem to be quite a lot of fixes streaming into the nouveau kernel module recently. I could see how easy it is to update linux-backport-modules, if you like. | 21:42 |
Sarvatt | doesn't look too bad but i'm not sure, the scripts are just hardcoded for apw's machines paths | 21:45 |
Sarvatt | wonder why he picked rc4 for nouveau | 21:45 |
RAOF | Probably because that was about the time we were talking about it. | 21:46 |
Sarvatt | shoot, hopefully I can continue booting with the lid closed after drm/nouveau: report LVDS as disconnected if lid closed since I use that machine headless, my intel doesnt start gdm until I open the lid if i boot closed | 21:49 |
RAOF | :) | 21:49 |
Sarvatt | oh nevermind it has a module option to ignore it too | 21:49 |
bryce2 | I was talking with apw about it yesterday | 21:51 |
bryce2 | he picked rc4 just because that's what he had at the time. He hasn't updated it just because he was unsure if anyone was needing it for testing | 21:51 |
bryce2 | I told him it'd be a good idea to update it since we're testing on it | 21:52 |
bryce2 | he didn't seem to think it would be much work to do it, but I don't know if he's had a chance to look at updating it yet | 21:52 |
RAOF | I might have a look; it can't hurt to learn a little more about it :) | 21:54 |
RAOF | That seems pretty painless. | 22:09 |
Sarvatt | oh yeah looks really easy, updates/UPDATE-NOUVEAU then updates/MUNGE-NOUVEAU.. maybe we should stuff the firmware in there too? | 22:10 |
RAOF | I think the separate firmware package is a better idea for now; that makes it easier to strip out later should we want/need to. | 22:11 |
bryce2 | agreed | 22:11 |
* RAOF test-builds the updated linux-backport-modules-nouveau. | 22:13 | |
Sarvatt | already done it? i was doing the same | 22:16 |
RAOF | Hm. Looks like I didn't do it quite right. | 22:17 |
Sarvatt | installing the -server headers to try building it here, you updated the paths in UPDATE-NOUVEAU, ran it then ran MUNGE-NOUVEAU and it didnt work? | 22:19 |
RAOF | When checking the diff against the current package it didn't appear to contain the changes I expected. | 22:20 |
Sarvatt | looks right here | 22:24 |
RAOF | Oh, no; they are there. There just seemed to be lots of *-> lbm_* changes I wasn't expecting hiding them. | 22:24 |
Sarvatt | http://pastebin.com/f3a17357f | 22:25 |
Sarvatt | oh maybe it runs MUNGE-NOUVEAU in the build process | 22:25 |
RAOF | I think it might, yes. | 22:25 |
RAOF | My build failed looking for double-munged objects. | 22:25 |
Sarvatt | yep it does make[3]: *** No rule to make target `/home/sarvatt/source/linux-backports-modules-2.6.32-2.6.32/debian/build/build-generic/nouveau/lbm_lbm_drm_kms_helper.c', needed by `/home/sarvatt/source/linux-backports-modules-2.6.32-2.6.32/debian/build/build-generic/nouveau/lbm_lbm_drm_kms_helper.o'. Stop. | 22:26 |
RAOF | Yup. There it is in debian/rules.d/2-binary-arch.mk | 22:26 |
Sarvatt | makes sense why update wasnt executable and that one was then | 22:26 |
RAOF | Yeah. | 22:26 |
RAOF | Oh, yeah. That's why it's building compat-wireless again. Three different kernel flavours! | 22:35 |
RAOF | Delicious! | 22:35 |
tjaalton | tseliot: rotation isn't related to calibration :) and if it doesn't work it's a bug | 22:37 |
tseliot | tjaalton: yes, I know, they are separate issues | 22:37 |
Sarvatt | theres a bunch of rotation changes in wacom git since our snapshot, no idea if theres any fixes there though | 22:39 |
Sarvatt | http://linuxwacom.git.sourceforge.net/git/gitweb.cgi?p=linuxwacom/xf86-input-wacom;a=summary | 22:40 |
Sarvatt | hmm might just affect intuos4 mice | 22:43 |
Sarvatt | eww | 22:57 |
Sarvatt | it no likey the updated module | 22:58 |
RAOF | In what way? I'm updating mine now. | 22:58 |
Sarvatt | http://sarvatt.com/downloads/nouveau.4.log | 22:59 |
Sarvatt | maybe because the lid is closed, trying with the module parameter now | 23:00 |
Sarvatt | ignorelid:Ignore ACPI lid status (int) | 23:01 |
* Sarvatt ponders making that default on if thats it | 23:01 | |
RAOF | Also, FIFODEATH! | 23:01 |
RAOF | I'm not sure why X adds the vesa & fbdev sections to the default config; possibly it *always* adds them. In which case we need to patch vesa to not blow up horribly in the presence of kms. | 23:03 |
Sarvatt | yeah it does always add those on pc arches | 23:04 |
kklimonda | Sarvatt, RAOF: well, rejoice - my problem disappeared on a clean install. At least for now. | 23:04 |
Sarvatt | oh? | 23:04 |
kklimonda | Sarvatt, my new dmesg looks more or less like yours | 23:04 |
Sarvatt | still would like to know what was hanging around to cause it | 23:04 |
Sarvatt | did you have nouveau-generic or nouveau-generic-pae installed? | 23:05 |
Sarvatt | do you still get the mtrr error? | 23:05 |
kklimonda | yes, I still get it | 23:05 |
RAOF | Does that mean suspend/resume & Xv works for both -generic and -generic-pae for you now? | 23:06 |
kklimonda | RAOF, both seem to work fine with -pae, I'll test -generic in a moment | 23:07 |
Sarvatt | lbm-nouveau.ignorelid=1 fixed my problem with the newer module | 23:07 |
Sarvatt | glad i looked at the changes beforehand | 23:08 |
Sarvatt | RAOF: how did you get in a situation where the servers autoconfiguration actually tried to use vesa or fbdev and caused a problem? you'd have to have nouveau and nv not installed | 23:13 |
Sarvatt | unless you mean in failsafe, thats a different script in /etc/gdm/ forcing vesa there, i have my /etc/X11/xorg.failsafe.conf usinf fbdev to make failsafe work | 23:14 |
RAOF | Sarvatt: For some reason in some cases I found vesa was binding before nv, producing a pretty bloom pattern on my lvds. | 23:14 |
kklimonda | RAOF, works fine now on both -generic and -pae - only issue I have is that before kms kicks in and sets the right resolution I see garbage on my screen | 23:15 |
RAOF | Sarvatt: Where's your newer plymouth? | 23:16 |
Sarvatt | xorg-edgers/nouveau | 23:16 |
kklimonda | and plymouth kicks in way too late | 23:16 |
Sarvatt | yeah nouveau is switching to high rez way too late here too | 23:17 |
RAOF | Hm. Must be my apt proxy kicking in. *Now* I can see that. | 23:17 |
RAOF | Plymouth kicks in pretty early for me. That might just be my nv4x not needing firmware loaded, I guess. | 23:20 |
Sarvatt | RAOF: you said you were using spinfinity? did you plymouth-set-default-theme with --rebuild-initrd to do that? the --rebuild-initrd option seems to be making plymouth work for other people | 23:20 |
Sarvatt | [ 14.962137] Console: switching to colour frame buffer device 160x50 14 seconds in :( | 23:20 |
RAOF | Faouwch. | 23:20 |
Sarvatt | trying to sudo plymouth-set-default-theme spinfinity --rebuild-initrd and see if it comes up earlier | 23:22 |
jcastro | hi jbarnes, I ran into this today: https://bugs.freedesktop.org/show_bug.cgi?id=26418 | 23:22 |
ubottu | Freedesktop bug 26418 in Driver/intel "[i965] GPU hang on -intel 2.9.1 on Lenovo (triggered by GnomeDo?)" [Critical,New] | 23:22 |
jcastro | jbarnes: I am sprinting with bryce so I am able to do whatever you need | 23:23 |
RAOF | Sarvatt: Takes my laptop 2sec to switch to the colour framebuffer. | 23:23 |
Sarvatt | oh boy, think its hung in plymouth like my intel is doing sometimes, cant ssh in after that reboot :) | 23:25 |
Sarvatt | hopefully its just an extra long fsck | 23:26 |
RAOF | First reboot after rebuilding the initramfs again & plymouth hung before properly starting the spinfinity animation. | 23:27 |
Sarvatt | yeah the first boot after upgrading hung for me as well on 0.8.0~-8 | 23:28 |
Sarvatt | looks like it hung on my nvidia laptop too after changing themes and rebuilding | 23:29 |
RAOF | But works the second time around. | 23:30 |
RAOF | Sarvatt: Are you going to uploaad the updated l-b-m-nouveau to xorg-edgers/nouveau? | 23:31 |
Sarvatt | second boot worked fine here too, had to dig that laptop out :D | 23:32 |
RAOF | Oh, man. It takes so long to load the firmware for your card. | 23:32 |
Sarvatt | its a little screwy, i got the 80x25 text console for most of the boot, then it showed the spinfinity theme for about a second and went to X | 23:33 |
Sarvatt | no progress bar or anything | 23:33 |
Sarvatt | guess it is the firmware loading that slows it down so much :( | 23:33 |
RAOF | Well, that does take a whopping 6 seconds out of your boot time, yes. | 23:34 |
Sarvatt | think i'll try with the firmware-less patch | 23:34 |
RAOF | What's the firmware-less patch? | 23:34 |
Sarvatt | the blob takes 10 seconds to load too | 23:35 |
RAOF | You could try just removing nouveau-firmware for testing. | 23:35 |
RAOF | Everything should still work, as long as you don't mind not having accel. | 23:35 |
Sarvatt | argh didnt bookmark it | 23:40 |
Sarvatt | oh yeah good point | 23:40 |
Sarvatt | it was the patch so nv50 didnt need firmware | 23:40 |
RAOF | Oh? The voodoo generator patch? | 23:42 |
Sarvatt | yeah | 23:43 |
RAOF | Sarvatt: You're after http://0x04.net/~mwk/gen.diff | 23:45 |
Sarvatt | thats it | 23:45 |
Sarvatt | thanks | 23:45 |
Sarvatt | btw theres a doc in his directory there where it looks like they are working on CUDA support and it says to expect major api breakage soon :P | 23:45 |
Sarvatt | no luck removing firmware | 23:46 |
Sarvatt | http://pastebin.com/f3ebbb01b | 23:47 |
RAOF | So they're working on the final set of ABI changes, then? Woot. Excellent timing. | 23:47 |
Sarvatt | in a few months after the other things in the TODO :) | 23:48 |
Sarvatt | its got timetables | 23:48 |
RAOF | It looks like it takes ages to timeout on fetching the firmware. Maybe the voodoo generator will work better. | 23:49 |
Sarvatt | yeah going to try that now. did you want me to upload lbm or did you want to do it? | 23:49 |
Sarvatt | i'll just upload it and see if it clashes :D | 23:51 |
Sarvatt | just did ~pre2 | 23:51 |
RAOF | If you're doing so now, I won't :) | 23:52 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!