[05:22] sarvatt, is sna enabled on edgers? i want to get some benchmarks. [05:23] LLStarks: there's ppa:sarvatt/intel-sna [05:23] LLStarks: however, the patch required in X itself is missing. [05:23] so it segfaults nicely [05:24] ah [05:24] it doesn't segfault because the x patch is missing, that just fixes rendering corruption.. :) [05:24] Sarvatt: i got a segfault! [05:25] will sna ship with oneiric and is it comparable to exa, uxa, etc? [05:25] no it wont [05:25] =( [05:25] it's dependent on the xserver 1.12 video abi [05:26] that's alright, we're setting up for PP during the oneiric cycle [05:27] will there be any hiccups similar to the gem/uxa transition? that was harsh. [05:27] pretty much 100% guaranteed there will be :) [05:28] ouch [05:28] sorry I didn't update it today, will update it to the latest in the morning [05:29] so, is the sna ppa (now or tomorrow) ready for vm or live testing in the sense that the desktop is usable? [05:29] *live as in actual install runing on actual hardware [05:29] sure if you want to test it, just avoid chrome/chromium because they dont work [05:30] any benchmarks you'd recommend for showing maximum benefits? [05:30] that ppa hyperair linked needs to be activated at the same time as xorg-edgers [05:32] if at all possible, i'd like to get a hardware table up and running on the wiki. vga string, commit id, benchmark date, results. [05:33] i dont think it's anywhere near ready for something like that, it's going to be very volatile for some time yet.. [05:33] LLStarks: 8086:2a32 -> segfault. [05:34] can you boot to recovery terminal? [05:34] LLStarks: ? [05:35] recovery terminal? [05:36] tty [05:36] well i can switch to ttys yeah [05:36] then that's not so bad. [05:36] in fact, that's how i repaired my X. [05:36] mm yeah it's just a normal X segfault [05:36] it's not a wedge [05:36] ah ok [05:36] i should probably get a proper stack trace [05:36] hyperair: http://dpaste.com/551239/ look like that? [05:36] Sarvatt: not quite [05:36] Sarvatt: it had a few intel_drv.so lines in the middle [05:36] and i lost the log. blargh. [05:37] praying that 8086:27a2 works [05:45] taking the sna plunge. i hope my body can take it. [05:47] the laptop loves it. whenever i test something new it's like "awww, you're still thinking of me. sure, i'll give you some more performance. windows stopped driver updates for me years ago." [05:48] * LLStarks crosses fingers [05:52] holy hell [05:52] http://ie.microsoft.com/testdrive/performance/fishietank/ [05:52] 2 fps to 40 fps [05:53] can i send a cake to intel? [05:53] let's see if it handles 720p gl output [05:55] ...it didn't choke [05:55] this is amazing. [05:57] smooth 1080p. yup. [05:57] it's almost as if the i945gm no longer needs vaapi [06:06] hyperair, no segfaults yet. just the occasional artifact. [06:06] very rare. [06:07] i'm going to be brave and try chromium [06:07] all seems fine [06:09] bryceh: forgot to push xkeyboard-config changes? [06:15] sarvatt, i've noticed that gl subtitle rendering (is broken on edgers. what component should i file against upstream? [06:16] and has been for a while [06:33] RAOF: moo, you haven't pushed the earlier xkeyboard-config merge to git?-) [06:34] Wait, what? [06:34] Let me just get that! [06:34] 2.2.1-1u1 [06:34] maybe we should get 2.3.1 for oneiric? [06:35] debian didn't put it in unstable yet because they still have gnome2 there and there's some conflict [06:35] I don't know of any reason no to grab that. [06:35] but we already have gnome3 for the most part [06:35] in oneiric [06:36] Was the conflict just the naming scheme bits, or something more fundamental? [06:36] don't remember the details [06:37] Pushed like a jimmyhack. [06:37] 2.1 is "the best for gnome2", nothing more than that [06:37] and we already have 2.2.1 :) [06:37] so yes, should be no blockers for 2.3.x [06:38] s/yes/no/ [06:38] Right. I'm pretty sure that refers to the naming stuff. [06:39] the dvorak-stuff? [06:40] No - renaming all the profiles $COUNTRY ($MAP) to $LANGUAGE ($MAP). [06:40] as [06:40] ah [06:40] So the keyboard indicator here says “English (Dvorak)” rather than “USA (Dvorak)”. Etc. [06:40] heh, makes sense [06:42] While that was done for GNOME 3, I'm not sure why it's not regarded as appropriate for GNOME 2. [06:47] Mmm. Looks like this system's borken enough to no longer boot. [06:47] i should work through installing the sandybrigde system.. finally switching to ssd et al [06:48] Maybe I should stop holding out for ivybridge and build a decent desktop system. [06:53] wait for ivy bridge [06:53] :3 [06:54] tr-gate. better than sandy bridge. [06:54] sole purpose is to embarrass amd for the next year or so [06:54] and ~8 months away [06:54] Right. There's *always* something cool ~8 months away. [06:54] sandy bridge + x58 is surefine [06:55] x58? that's old [06:55] wait. did i mess up the chipset? [06:56] z68 [06:57] “After this operation, 258 MB disk space will be freed.” Hm. That seems rather a lot. Oh, wait, it's removing xserver-xorg-core and ubuntu-desktop. [06:57] RAOF: heh, which update? [06:58] Multiarch'd mesa. [06:58] ah, because there's no xserver update to go with it? [06:58] Right. [06:59] Also because llvm's not multiarched, so all the i386 drivers are baroque. [07:26] if it ain't baroque, don't fix it [07:27] Actually, that's not quite true. All the i386 gallium drivers are broke. i965_dri is perfectly happy to run i386 glxinfo and glxgears. Score. [08:01] can sna be done with stable mesa? [08:03] AFAIK sna is largely independent of mesa. You're very much not allowed to make backward-incompatible DRI2 changes :) [08:06] well, i was wondering if i could test it without a full edgers payload [08:07] Probably, but you get to keep all the pieces. [08:07] eh? [08:07] “If it breaks, you get to keep the pieces” [08:08] LLStarks: no fair! [08:08] LLStarks: 32-bit or 64-bit? [08:08] 32-bit [08:08] maybe it hates 64-bit. [08:08] not that i have a choice [08:08] heh [08:09] Oh, atom? [08:09] That was frequently paired with an i945. [08:09] core duo. [08:09] core duo? those support 64-bit, no? [08:09] No, the core duo *2* supports 64-bit :) [08:09] model name : Intel(R) Core(TM)2 Duo CPU T5750 @ 2.00GHz [08:10] core 2 duo. [08:10] t2050 [08:10] still more powerful than an atom [08:10] despite being 5 years old [08:10] * hyperair wants a sandy bridge. [08:10] * RAOF is fairly sure anything 686-derived is faster than an atom. [08:10] lol [08:10] atom is 686 [08:11] p3 architechture [08:11] No, it's not. It's an in-order processor. Which is why performance sucks. [08:11] really? [08:11] * RAOF is pretty sure. [08:11] you're right [08:12] not p6 [09:47] Why must the mesa build system be constructed entirely of hate? [09:59] heh [10:06] Hm, (a) has this worked, and (b) how small are the gallium drivers now? [10:31] Win! 3.6M 2011-06-07 15:12 r300_dri.so → 1.3M 2011-06-07 19:29 r300_dri.so, 3.5M 2011-06-07 15:12 r600_dri.so → 843K 2011-06-07 19:29 r600_dri.so [10:31] woohoo [10:32] so a shared gallium core? [10:32] Yeah. [10:32] nice [10:33] With only the *tiny* problem of 19M 2011-05-27 19:49 /usr/lib/libLLVM-2.9.so.1 [10:36] hehe === jibel_ is now known as jibel [11:57] Has anybody tried Unity on a git master gallium driver? [12:03] Probably some of our intrepid xorg-edgers cadets have? [12:04] probably yeah [12:04] i've only tried swrastg but with 7.10, which lacks tfp support [12:04] I happen to have a nice hot ccache here; want me to try? [12:04] you've pulled a snapshot too? [12:04] (And a radeon 5450ish) [12:05] tjaalton: Mesa build hacking always gets done on master :) [12:05] It's not sufficiently fun to do twice :) [12:05] RAOF: naturally, but is it ending up in oneiric soon?-) [12:06] RAOF: no real need I jast wanted to know if there was a known problem with Unity on gallium, I'm guessing the problem is in our driver. [12:06] Unity works pretty well with vmwgfx except a rendering bug. [12:06] Cool. [12:07] tjaalton: It'll end up in oneiric soon; I just need to finish off the llvm 2.9 mir, which is blocked on actually being able to build openjdk-6b18 on an arm system to check the testsuite. [12:09] What mesa is planned for oneiric? [12:09] 7.11 [12:09] Sigh. [12:09] Problem for KDE? [12:10] Mgraesslin just mentioned on one of the KDE lists he'd done 4.7 to work with 7.10. [12:10] Then mentioned if someone was going to use 7.11 with KDE 4.7 they better test it. [12:11] So not necessarily a problem, but a combination with undefined results at this point. [12:11] Ah, right. Well, xorg-edgers contains 7.11 snapshots, but we won't be switching until 7.11 gets branched. [12:13] Oh! At what point did libGL start to fallback on swrastg_dri? [12:20] aiui Sarvatt right, it would be 7.11? [12:20] uh [12:20] "iirc what Sarvatt said,..." [12:22] RAOF: We didn't get 4.7 beta 1 packaged yet either. [12:23] Fun multiarch fact: dpkg --remove mesa-utils won't remove the i386 mesa-utils package; you need dpkg --remove mesa-utils:i386 [12:25] Wow. glxgears is curiously messed up. [13:24] Hrm, Im having an issue where I have the laptop setup with the external screen, and only the external screen in use. However, when powermanagement turns the screen off after a period of inactivity, it doesnt come up again when you become active. the machine seems fine and reponsive, but no screen. [13:35] jussi: best bet is to try the mainline .39 kernel [13:36] tjaalton: how would I go about getting that ? is it relatively stable? [13:37] http://kernel.ubuntu.com/~kernel-ppa/mainline/v2.6.39.1-oneiric/ [13:37] certainly stable enough to test that [13:40] tjaalton: right, however, this machine is on natty [13:41] sure [13:41] just install the deb :) [13:41] ok :) [13:41] it doesn't matter if it's built on oneiric tools [13:41] *with [13:42] at least, it shouldn't matter much, and hasn't for the usecases I've had [13:42] ok, was just unsure. Ill let you know how it goes :) [13:42] good [13:42] which system did you have? [13:43] is there a chance it will fix my bug also? (bug 793487) [13:43] Launchpad bug 793487 in xserver-xorg-video-intel (Ubuntu) "Kubuntu does not shut down when external monitor attached through displayport (affects: 1) (heat: 6)" [Undecided,New] https://launchpad.net/bugs/793487 [13:43] Its a HP probook 5320m [13:44] also, do I need to install the headers? [13:44] no need for the headers, unless you have dkms packages [13:45] ah so it's arrandale, yeah .39 might fix that bug [13:45] dkms packages? [13:45] .39-rc2 has been tested to fix some bugs concerning external monitors [13:46] dynamic kernel module source, probably don't have them if it's something you don't know :) [13:46] used for binary blobs [14:04] tjaalton: Unfortunately the kernel did not boot. I installed hte headers also, but still no boot. I pressed esc to see the progress, and it seems to stop after apparmor [14:05] jussi: dang [14:06] tjaalton: Im more than happy to try anything else you ahve to offer though. :) (or try troubleshoot the kernel) [14:06] you could try this one, in case the point release broke something http://kernel.ubuntu.com/~kernel-ppa/mainline/v2.6.39-oneiric/ [14:07] at least one person booted that succesfully on natty [14:09] * jussi tries [14:22] tjaalton: unfortunately no go on that either. it gets a little further, but then dies [14:22] meh [14:23] you could try rc4 from the parent dir next, it's also built on natty if it makes a difference after all.. [14:28] tjaalton: the blobs won't work with the new kernel as we prevent the module from building against unsupported kernels [14:29] tseliot: right, but is it already in natty too? [14:29] tjaalton: yep [14:29] nice [14:29] :) [14:29] hrm, theres also this in the install proceedure: W: Possible missing firmware /lib/firmware/rtl_nic/rtl8105e-1.fw for module r8169 [14:29] well, i was just pointing out when the headers would be useful [14:29] jussi: that shouldn't matter [14:30] ok, just thought to mention it [14:30] some new nic module i guess [14:32] right, reboot time. hope this works. [14:34] right then. (it boots :D ) now let me shut down and see if it fixes that bug, Ill try the initial bug I mentioned in a min. [14:35] phew, great [14:40] right, so that bug I mentioned before (793487) is fixed with this kernel. [14:40] niice [14:41] kde seems a little slow, not sure whats going on. [14:41] just seems. [14:42] anywya, Ill go away for a bit, and see if the other bug still persists [14:42] ok [14:42] (and if someone wants to build the newer .1 kernel for natty, I would be happy to try that.) [14:44] the hwe team has a set of patches that'll hopefully get added to the natty kernel, even though there will be no upstream point-releases anymore for .38 [14:44] Im also getting a few artifacts left on the screen. [14:44] looks like it would fix a bunch of issues [14:44] ah [14:45] well, it has around 90 commits, of which maybe 8 are needed [14:45] rc4 that is, to the drm/i915 code [14:45] not artefacts per se, but parts of highlighting etc [15:12] RAOF: pushed xserver merge with the fedora pointer-barrier patch added [15:13] now xorg merge finalized so we can start pushing these to oneiric.. [17:35] anyone seen X server dumping ? on intel ? [17:35] [ 25658.288] Segmentation fault at address 0x7f039ec32010 [17:35] [ 25658.288] [17:35] Caught signal 11 (Segmentation fault). Server aborting [17:36] this is natty updated as of this morning [18:16] apw: full stack trace? [18:17] hyperair, at the bottom of: http://paste.ubuntu.com/621075/ [18:17] apw: one sec, let me get you the bug number [18:18] i attached a fix to it, its x-x-i-synaptics [18:18] https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/774978 comment #25 [18:18] Launchpad bug 774978 in xserver-xorg-input-synaptics (Ubuntu) "xserver crashes in RecordAReply when XRecord is enabled in syndaemon (affects: 91) (dups: 22) (heat: 444)" [High,Triaged] [18:19] it just started now though? it happened at random before but amazing you went from release till now and just started hitting it [18:20] yeah first time i've ever seen it [18:20] i right clicked somewhere and boom [18:20] oh ... of course i don't use my touchpad most of the time on this machine as i have external keybaord an mouse [18:21] and i was using the touchpad unusually ... so [18:35] weird, i've been using my touchpad though [18:35] why would a touchpad cause intel_drv.so to segfault? [18:35] i see that in the stack trace somewhere [18:57] hyperair, i've seen it happen in israel due to voltage differences [18:57] Sarvatt: IT WORKS. [18:57] my xserver choked whenever i used the touchpad [18:58] LLStarks: ah, right voltage differences. [18:58] well my xserver never choked due to the touchpad [18:58] but the touchpad itself choked. [18:59] btw, all of my observed improvements that i attributed to SNA yesterday were all from the new ddx/mesa/drm WITHOUT SNA [18:59] i don't get it [19:00] what's crippling the stable stack so much? [19:00] Sarvatt: time to sru -synaptics? [19:01] synaptics needs tapbutton2=2 [19:01] for two-finger middle click [19:01] right now, it's tapbutton2=3 [19:02] that and out-of-the box two-finger scrolling [19:17] * hyperair doesn't observe any improvements to SNA though.. [19:22] intel ddx is overrated [19:23] i have no idea what's going on between oneiric's 2.15 and the non-SNA git, but something is slowing me down [19:24] it can't be mesa. this is 2d stuff. [19:26] i wish i could source this slowdown, but xorg log is clean [19:37] can i run -intel master on a stable oneiric stack? [19:40] hmm i wonder why 3d activity causes my entire computer to freeze for short periods of time [19:40] enough to miss a number of interrupts for psmouse [19:40] and enuogh to screw over eth0 transmission [19:41] and enough to make iwl4965 trigger a hardware reset [19:41] x is just weird [19:42] heh [19:42] i think it's more like i915 is doing too much processing in the irq [19:45] i945gm slowdown with the 2.15 ddx confirmed. [19:45] 2.15git is fine. [19:46] there's like zero acceleration in the former [19:47] and this is without SNA [19:47] time for some bisecting [19:47] should be fun [19:51] wait wtf... 2.15-release git is fine. did somebody screw up packaging? [20:03] nvm. i might be wrong. [21:07] tjaalton: i dont suppose there's any chance you already did any of the mesa lucid backport? :P [21:07] Sarvatt: not sure, probably not [21:07] silly sentence [21:08] yeah it was a long shot, gonna start throwing it together now since thats the last part [21:23] ricotz, is it possible to change the sound theme in gnome 3? it's not an option with the tweak tool === yofel_ is now known as yofel [21:44] bjsnider, this can be done with dconf-editor [21:45] bjsnider, org/gnome/desktop/sound [22:06] got it close, http://sarvatt.com/downloads/mesa/ [22:11] Sarvatt, should I go ahead and upload/sru the fix you posted on #774978? Or is there more we need to do first? [22:14] that'd be much appreciated [22:38] Sarvatt: would adding -lrt to LDFLAGS help there?