[10:13] apw: posted email on kms to ubuntu-devel@ [11:00] bryce_: has jbarnes left UDS already? [11:01] * hyperair thinks KMS is the best thing that ever happened to the intel driver =D [11:03] tjaalton: no, he leaves tomorrow [11:03] tjaalton: however I've not seen him so far this morning. Maybe he's sleeping in. [11:03] bryce_: heh, could be [11:03] there was just a question I had [11:06] lp 377875 is weird. would be nice to have the guy's dmesg [11:06] I'll direct him here if/when I see him [11:06] Launchpad bug 377875 in xserver-xorg-video-intel "[945G/GZ] Kubuntu: no direct rendering (KMS bug)" [High,Incomplete] https://launchpad.net/bugs/377875 [11:06] we have a session after lunch for doing more bug triage [11:08] jcristau: yeah; it's the one bug so far reported against KMS, however sounds like maybe just a futzed up config [11:08] bryce_: yeah i was following the links from your kms mail :) [11:10] i need to get a .30-rc kernel.. [11:13] yep [11:13] fwiw, .30 seems to be a kickass improvement over .28 all around [11:13] well, for -intel I mean [11:18] .30 got the bugfixes from f11 testing :) [11:21] there he is [11:21] jbarnes_BCN: tjaalton had a question for you [11:31] tjaalton: hi [11:37] jbarnes_BCN: hi, I'll finish this ice cream first ;) [11:38] tjaalton: hurry, almost lunch time :-) [11:40] i have a feeling that guy is just not adding intel_agp and drm before i915 in his /etc/initramfs-tools/modules like you need to do when you go that route.. [11:41] bryce_: just had mine :) [11:41] jbarnes_BCN: we have a bunch of new fujitsu desktops with built-in G33/35, and some of them blank every now and then for a second or two [11:41] running 9.04 [11:42] and I've had that same on my laptop a couple of times with karmic [11:42] Sarvatt: update-initramfs should do that for him though [11:43] jbarnes_BCN: on the fujitsus it happens even on the gdm login screen, and sometimes very frequently. Is it due to failing hardware or something in the driver? going back 2.6.3 -> 2.4 didn't help [11:44] tjaalton: probably driver... [11:44] tjaalton: lemme dig up a bug # and patch for you to test [11:44] it doesn't for me on ubuntu, maybe a change in initramfs-tools in debian? [11:44] 0.92bubuntu30 [11:44] jcristau: cool, thanks [11:44] Sarvatt: let me check [11:44] tjaalton: itym jbarnes ;) [11:45] uh [11:45] lazy tab-complete [11:46] Sarvatt: looks like 0.93.2 added agp modules to the initramfs. and the next upload will add i915 (and thus drm) [11:46] https://bugs.freedesktop.org/show_bug.cgi?id=19304 has a test patch to address some fifo underruns [11:46] Freedesktop bug 19304 in Driver/intel "[i945 FBC] spontaneous black screen (major pipe-A underrun)" [Major,Assigned] [11:46] it'll also print some output to the log if an underrun occurs [11:47] hmm, AIUI it doesn't log anything [11:48] only the old driver would log anything [11:48] if 2.4 doesn't show pipe underruns, then it could be something else [11:48] ah, ok [11:49] right, I'll add the patch and try [11:49] Sarvatt: also i think the driver was fixed for some kms failure when i915 wasn't loaded before X startup, so that may help too [12:24] yep I'm 100% convinced that is his problem in lp 377875\ [12:24] Launchpad bug 377875 in xserver-xorg-video-intel "[945G/GZ] Kubuntu: no direct rendering (KMS bug)" [High,Incomplete] https://launchpad.net/bugs/377875 [12:25] I just added only i915 to my /etc/initramfs-tools/modules and I get the exact same error [12:25] Sarvatt: good :) [12:25] http://sarvatt.com/downloads/Xorg.txt [12:26] cant open drm, falls back to swrast [12:27] so that should be fixed with either new initramfs-tools or new driver [12:27] we're both using the latest driver, need initramfs-tools for sure [12:36] [ 4.469436] [drm:drm_fill_in_dev] *ERROR* Cannot initialize the agpgart module. [12:36] then it loads intel_agp after that [12:45] using kms on .30-rc7 with ddx 2.7.99.1 now. looks pretty good. [12:47] there's a update drm-intel that fixes the font corruption/swap problem now too [12:47] yeah i know === crevette__ is now known as crevette [14:20] bryce_: kms is tempting me to upgrade my laptop to karmic, really quite horribly much ;) [14:21] Ng: do it :-) [14:21] I wonder if I can do it from archive.conference, but still use update-manager [14:24] I think I'll start with the jaunty test option [14:35] mmm, high res consoles [14:36] there's still some flickering when starting X though! [14:37] weirdly, the i915.modeset=1 option on the kernel cmdline whined about it being unknown. I thought ordering there didn't matter? (I just put it on the end after quiet splash) [14:48] bryce: just a note for the wiki that i've edited a few times, you never need to manually enable UXA in xorg.conf on any driver, KMS always only uses UXA [14:49] * Ng tries a suspend/resume cycle [14:49] even if you have EXA picked in xorg.conf it'll ignore it and use UXA instead, no need for the extra step [14:50] I was promised faster resume! It's the same ;) [14:51] Ng: the kernel complains about the option (because at that time i915 isn't loaded yet), but it works anyway [14:51] ah [14:52] will we want to move i915 really early in the initramfs so the kernel modes are there for the majority of the boot? [14:53] probably not [14:53] AIUI [14:53] doesn't that mean the usplash screen will resize? [14:53] because the boot will be fast enough anyway [14:53] don't know [14:53] breaks the usplash progress bar doing that or building it in the kernel right now [14:53] can usplash die kthx [14:56] could be that even plymouth is "too heavy" [14:56] ok I take it back, usplash appears to be unaffected [14:56] tjaalton: our target boot time on a Dell mini9 for 10.04 is 10 seconds. Barely seems worth showing more than a spinny thing ;) [14:57] so chvt X->Console is instant, but going Console->X has a couple of flickers [14:57] (on jaunty with a karmic KMS kernel) [14:57] thats fixed in xorg-server 1.6.1.901 [14:57] \o/ [14:57] its in xorg-edgers [14:57] or karmic :D [14:58] I'm off work next week and I was going to re-install my laptop with 64bit anyway, so I can do karmic then [14:58] seems a bit silly to risk nuking my OS on the penultimate day of UDS ;) [15:01] sounds like a plan to me ;) [15:02] hehe [15:02] Sarvatt: or sid ;) [15:03] sid schmid! ;) [15:03] * Ng wanders off to his next session [15:28] ping bryce [15:29] ping again bryce [15:38] bryce_: WELCOME BACK [15:38] https://bugs.edge.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bugs?field.searchtext=&orderby=-importance&field.status%3Alist=CONFIRMED&field.status%3Alist=TRIAGED&assignee_option=any&field.assignee=&field.bug_reporter=&field.bug_supervisor=&field.bug_commenter=&field.subscriber=&field.status_upstream=hide_upstream&field.status_upstream-empty-marker=1&field.omit_dupes.used=&field.omit_dupes=on&field.has_pa [15:38] tch.used=&field.has_cve.used=&field.tag=&field.tags_combinator=ANY&search=Search [15:38] wow. that's like a gitweb url, except worse [15:39] indeed :-) [15:42] http://launchpadlibrarian.net/22029875/Screenshot.png [15:43] http://launchpadlibrarian.net/26349077/Evince-garbled.png [15:43] bug #326008 [15:43] Launchpad bug 326008 in xserver-xorg-video-intel "screen does not redraw properly" [Undecided,Confirmed] https://launchpad.net/bugs/326008 [15:43] 1st one [15:48] jbarnes_BCN: the patch didn't help :/ [15:48] jbarnes_BCN: next we'll try another display card to rule out a broken monitor or such [15:51] heya tjaalton, we're doing X bug triaging for a couple sessions [15:52] if anyone wants a kernel to test out the swap related font/glyph corruption fix for intel/UXA -- http://sarvatt.com/downloads/2.6.30-rc7-sarvatt2/ [15:53] slangasek: https://wiki.ubuntu.com/X/Triaging [15:53] need xorg-edgers to go with it [15:57] bryce_: have fun :) [15:57] karmic didn't provide audio for my laptop, so I can't listen to you ;) [15:58] Sarvatt: is it the "funny characters in firefox" -bug? [15:58] yup! [15:58] ok.. [15:58] affects only lwn.net and linuxtoday.com here [15:58] wonder why [15:59] tjaalton: don't think we have mikes anyway [16:08] https://launchpad.net/~ubuntu-x-swat/+archive/x-freeze-test [16:10] Question for later when people aren't busy: do you guys have any plans regarding xserver-xorg-video-intel 2.7.99.1 in karmic? I would just like to suggest not just syncing debian-experimental because there have been so many fixes past that release from when they first dropped EXA/XAA. the fixes from 2.7.0 to 2.7.1 aren't even in it. [16:12] Sarvatt: yeah I was planning on merging it in after UDS, next week [16:12] Sarvatt: I was thinking using a git snapshot - maybe what you stuck in edgers, if you think that's sane [16:20] for sure, I would muuuuuch rather use 2.7.1 than the tagged 2.7.99.1 release thats in debian-experimental right now, the desktop didn't stabilize in UXA for me until around may 2nd (about a week after 2.7.99.1 tag) and there have been so many fixes since then like the firefox hangs that 100% of people using UXA hit on specific websites and all the "random crash" stability fixes that went into 2.7.1. of course libdrm needs to be updated [16:20] too, part of the firefox hang fix is post 2.4.11 so a git version of that would be needed too [16:20] oh, a fix so that nouveau compiles on 2.6.30 just went into mesa/drm too [16:22] http://www.gossamer-threads.com/lists/linux/kernel/1081109?page=last [16:22] PAE patch ^^ [16:25] jbarnes: gotten any feedback on that cursor flicker patch? [16:30] talked to a few people using xorg-edgers since i started adding to it right after you posted it, on my 945GME and 3 people on 965 it does get rid of the constant flicker that was there without it, but it makes the animated cursor corruption that's always happened more prominant. every few seconds randomly it'll break the animation and the cursor disappears, sometimes theres green colored garbage in a box around the cursor. it's there w [16:30] ithout the patch too only less noticable because its constantly flickering on and off without anyway [16:35] don't think theres any 32 bit PAE kernels shipped by ubuntu anymore since 2.6.28 [16:44] http://git.kernel.org/?p=linux/kernel/git/anholt/drm-intel.git;a=commit;h=07f4f3e8a24138ca2f3650723d670df25687cd05 [16:45] apw: can we get a kernel with that patch (actually a drm-intel based kernel would be even better) [16:45] Sarvatt: huh [16:46] Sarvatt: so sounds like a newer git of 2.7.99.1 would make sense to pull in? Any plans to update the snapshot in xorg-edgers? [16:48] do you know who's handling the 2.8 release process? [16:48] i update it almost every commit, so depends on what gets pushed there :) you'd probably want to add the quirk patches from the current one and drop the 101_kms_cursor_flicker.patch I add to it though [16:49] jcristau: cworth is doing it again [16:50] jbarnes_BCN: ok [16:51] bryce: I think it'd make *alot* of sense, would have to update libdrm too though newer than the tagged 2.4.11 because part of the firefox crash fix is in there and that affected everyone using UXA as far as I know. didn't talk to a single person that could go to the websites in the bug report without it hanging them (happens in 2.7.1 UXA too) [16:55] Sarvatt: yeah that sounds like something we'd want to pull [17:01] jbarnes: it's not 100% to the ubuntu config, but if you need a deb to test real quick I've got one here http://sarvatt.com/downloads/2.6.30-rc7-sarvatt2 [17:07] changes from the config were enabling pcie aspm and memory stick, anticipatory default, core2 arch generic tune, KMS built in and default instead of a module, git log of what i pulled on top of linux-2.6 last night here - http://sarvatt.com/git/cgit.cgi/linux/log/?h=2.6.30-rc7-sarvatt2 [17:15] l8r [17:17] so with KMS can the resume scripts not have to do the chvt dancing? [17:17] yes [17:19] I guess that's why my resume isn't any quicker, it doesn't know that :) [17:19] does here. [17:19] (I'm running jaunty still, just with the newer kernel and driver) [17:22] it's quicker in karmic [17:36] Ng, does your /usr/lib/pm-utils/sleep.d/98smart-kernel-video have the have_kms() section around line 40? [17:37] Sarvatt: that was added recently i'd be surprised if jaunty had that [17:37] line 40 here is in smart_kernel_fglrx [17:37] my intel section has nothing about kms [17:38] yeah, karmic has that [17:38] ah yeah that checks for kms and does add_parameters --quirk-no-chvt on karmic [18:06] Ng, put an updated pm-utils up here since i didnt have any other jaunty packages in the PPA. https://edge.launchpad.net/~sarvatt/+archive/xorg-testing working fine with no flicker on resume and its back up in 2 seconds in KMS on a jaunty machine I just tested it on just incase you want it :) [18:06] its just karmic's renumbered 0ubuntu0 so it'll get overwritten when you upgrade [18:08] Sarvatt: aha :) [18:08] I'm reading terminfo man pages, but when I get done with that, I'll go back to playing ;) [18:29] ahh yeah, about that 2.6.30-rc7-sarvatt2 kernel, i also enabled mtrr cleanup with 1 spare reg because its required for my aspire one to have an open mtrr for video.. these darn things have all 8 full with junk by default [22:08] bryce_: coming for vino tinto? [22:09] I've got a bunch of free wine in 1205, invitation is open [22:26] jbarnes_BCN: no answer at the door; everyone passed out? :-) [22:27] jbarnes_BCN: team dinner went late; the place we had reservations at, after seating us, said their kitchen had a problem and there was no food. (??) So we ended up wandering around and finding another place. [22:35] bryce_: we were downstairs [22:35] you walked right past me [22:35] even after I called your name :) [22:35] oh! [22:35] heh [22:35] we're back in the room now [22:35] heh, I was intent on getting upstairs and to room 1205 :-) [22:36] apw: room 1205 [22:36] apw: gotta catch up on your drinking