=== ripps_ is now known as ripps === seb128_ is now known as seb128 === seb128 is now known as seb128_ === seb128_ is now known as seb128 [08:56] anyone know whether there is a how-to on getting different settings (such as keyboard layout) per input device in X ? [09:08] apw: yes, you can rely on the name and/or the id of the input device (which you can get from the "xinput list" command). [09:08] as regards keyboard layout you can use the device id [09:09] for example setxkbmap -device 2 -print [09:09] or something like that [09:09] tseliot, and is there somewhere you add that to get it to persist, as it gets lost on suspend [09:09] * tseliot hasn't played much with setxkbmap [09:12] apw: in Gnome I *guess* the gnome-settings-daemon would take care of it [09:13] jcristau: ^^ [09:13] tseliot, does xmodmap work per device too? [09:14] that is a poor question ... i want to change the layout, xsetxkbmap is good for that. i also want to swap two keys on the new keyboard, can i do that on a per device basis [09:14] I don't think there's gui support for it yet [09:14] happy to use command line for this for now [09:15] I meant for g-s-d [09:15] ahh [09:15] the tools are lacking [09:17] apw: I don't think xmodmap allows you to have per device settings [16:09] tseliot, if you're around, there is a conf call with AMD today in about an hour [16:11] bryce: thanks but I don't think I'll be around in an hour. Maybe you can forward the report to me? [16:16] I mean the report on the call [16:17] actually I don't really have anything to discuss with them, so was hoping the call would be of more usefulness to you [16:22] bryce: oh, nothing comes to my mind at the moment but maybe I can ask them next time (I'm quite busy with the new house) === mac_v_ is now known as mac_v [17:17] hi, I try to use ppa-purge, but I´m running a Linux Mint 8 aka Helena and got plenty of error messages [17:19] lyhana8, sorry I don't think it is supported on linux mint [17:20] here is my log : http://pastebin.com/d53d34889 [17:20] LM is ubuntu-based, I think it's only the distro name that trigger the problem [17:21] yeah [17:22] is it based on a ubuntu release? [17:22] yep [17:23] http://www.linuxmint.com/about.php [17:23] you can try DIST="whatever" ppa-purge xorg-edgers [17:23] whatever the one you're using is based on [17:23] http://en.wikipedia.org/wiki/Linux_Mint [17:23] ok [17:24] i meant are those packages from the ubuntu archives instead of linux mint archives, no idea how they handle the dist thing [17:24] I added the xorg-edgers repo manually [17:26] well doesn´t work : Warning: Could not find package list for PPA: xorg-edgers ppa [17:26] how can I do it manually ? [17:26] thats cus it removed it last time it failed [17:26] try uncommenting it from your sources then apt-get update then try again [17:28] you want to sudo apt-get install package/karmic package2/karmic to downgrade manually [17:29] or uncomment the edgers ppa in your apt sources and apt-get update to refresh it and try again with DIST="karmic" ppa-purge xorg-edgers [17:30] bryce: I'll merge the xserver so it's (more) ready for upload once the prereqs are in [17:30] Sarvatt: isn´t removing the ppa and reinstalling the xorg-server package enough ? [17:31] nope [17:32] same error, he look for an helena archive for each package [17:33] did the package use lsb_release or something la that to know the context ? [17:33] tjaalton, excellent [17:33] yeah it did [17:34] DIST=$(lsb_release -c -s) [17:35] Sarvatt: so it override my DIST ? [17:35] kees: is libaudit-dev finally in main (lucid) ?-) [17:35] did you use sudo somewhere in there or did you pass DIST="karmic" as root? [17:36] I use the sudo commadn [17:38] I try to remove the ppa version cause I think it make my firefox buggy [17:39] just run this command and bypass ppa-purge, http://pastebin.com/d6f09bb34 :D [17:39] XD [17:40] any idea if your driver can cause issue to firefox (flash) ? [17:41] yeah it sure could [17:41] or this ppa : launchpad-weyland/xserver-nobackfill/ubuntu [17:41] oh~ [17:44] Sarvatt: downgrade ok, i'll logout later. Thanks [17:44] hmm , even with kernel .32-6 ATI KMS performance is slow :( ... [17:47] anyone know how to effectively measure the fps/speed reduction with KMS ? [17:49] * mac_v was told glxgears isnt a good indicator [17:50] try openarena or similar [17:51] hehe , i have to learn to play a new game :) [17:52] has been "on the wagon" for nearly 2yrs... ;) [17:52] or run phoronix-test-suite [17:52] ah , great , ok [17:54] tjaalton: basically i wanted to help in testing , will those results help , or rather are they needed/essential? [17:55] * mac_v realizes he asked the questions first :) [17:57] we know it's slower [17:57] phoronix has an article about it [17:59] what is launchpad-weyland ? Did someone package wayland? [18:00] nah, the author is mathias weyland :) [18:01] mac_v, what's needed is to locate patches upstream that haven't filtered down to us that sound like they would improve performance, patch your driver, and test *that* [18:01] mac_v, and then if you find something which looks really good, flag it for our attention so we can look into incorporating it [18:02] * mac_v is unfortuantely not good/technical enough to help bryce with patching and testing :( [18:02] * mac_v is more a noob , who can test ppa and whine when things are broken ;p [18:03] heh, we have an oversupply of that at the moment ;-) [18:03] lol ;) [18:05] one other thing we could use is regular performance testing [18:06] do an upgrade each day to latest lucid, run the test suite, plot the data point on a graph [18:06] and then make the graph available to us [18:06] this would enable us to spot performance regressions (or improvements) faster [18:09] i'm surprised how good gallium is right now on i915 at least, simple things like qgears2 and glxgears are 2x faster and openarena is within 10% of the normal driver. the old build i did of it in july is 4-6x slower in the same things (or didnt even run them at all for the most part) [18:11] bryce: that i can do :) ... right now , what i have done is I'm running the Karmic with lucid kernels , in a few days once alpha 1 is released I'll fuly install Lucid on another partition and start doing that... [18:19] wish I knew of a way to make phoronix-test-suite not run some tests 3x so I wouldnt have to babysit it and manually exit the 2nd and 3rd runs, takes out my machine for 3 hours running openarena and smoking guns tests when i just want a rough idea of the speed [18:40] Sarvatt, maybe yank out just that one test and package it to run in isolation? [19:39] bryce: looks like the "remaining changes" -part of the xserver changelog was a mess, partly because of me :) [19:40] it had obsolete information and nothing new, because I haven't been updating it when merging [19:40] the diff, apart from patches, is pretty small [19:43] I thought the random dpms offs were fixed [19:52] virtuald, should be yeah [19:52] tjaalton, ok [19:52] tjaalton, yeah this would be a good opportunity to do some housecleaning on xserver [19:53] btw, I revamped the Xorg.0.log timing patch [19:53] been testing it on my laptop, I can check it in, although I don't think there's much need to enable it [19:55] ok, nice [20:09] bryce: i'm on edgers now, and a while after resume from ram my monitors turned off. i have an rv570. [20:14] virtuald, let me know if you find a patch that fixes it [20:14] although I'm more focused at the moment on what's in lucid than what's in edgers, but I'm open to looking at patches either way [20:19] anyone else have issues loading http://cvs.fedoraproject.org on karmic (firefox), with intel gfx or other? [20:20] it makes ffox take 100% cpu and takes several minutes.. [20:21] umm, http://cvs.fedoraproject.org/viewvc/devel [20:38] i don't know where or how to find such a patch, and i'll try to remember to try to reproduce on lucid [20:39] bryce: should we still autoload poulsbo? [20:40] tjaalton, do we? thought that was defaulted to vesa in our xserver [20:40] bryce: so it seems. anyway, the patch fails now [20:41] hmm, done differently upstream [20:41] } else if (dev->device_id == 0x8108) { [20:41] break; /* "hooray" for poulsbo */ [20:42] that should fall back to vesa then [20:42] so that causes it to fall back to vesa? [20:42] yes [20:42] yes [20:42] should be fine [20:42] echo :) [20:42] I'd defer to alberto, he's more clued in on what the OEM team wants [20:42] but they can patch it themselves for the psb hardware they ship [20:42] they can change it later if needed [20:43] that's the fourth patch I've tried to apply, and fourth to delete :) [20:43] I don't think we have a -psb at present that'll work on lucid with xserver 1.7 anyway. probably needs to go to vesa. [20:44] two of the patches were from fedora, and dropped there. I doubt we should be shipping stuff the "upstream" doesn't [20:44] tjaalton, just please give a mention of the patches being dropped and why. Sometimes in troubleshooting a regression I have to figure out why a given patch was dropped so it would be helpful info. [20:44] bryce: sure [20:45] i haven't looked at what f12 has on top of 1.7.x recently. [20:45] just cvs upped, though :) [20:47] ah there's the glx 1.4 thing [20:48] should probably steal that [21:04] sconklin, btw do you know if the kms pageflip patch is on the radar for the 2.6.32 kernel? http://www.phoronix.com/scan.php?page=news_item&px=NzY5OA [21:05] bryce: I wasn't familar with it, I can see [21:05] if not, would be nice to have it in a ppa for testing purposes [21:07] bryce: according to the page you linked, it didn't make .32 - I'll try dropping it on top of our karmic and see what happens. [21:08] right it's queued for .33 upstream [21:08] bryce: I'm just now uploading a karmic kernel + the moblin speedup test patches to a PPA for Alberto to have a look at. [21:09] sweet [21:09] sconklin, I tested apw's karmic kernel with those patches, but on an SSD didn't spot changes to X's boot time (didn't look at kernel boot time) [21:10] I'll test your kernel too [21:10] and andy is doing research with the sme patches for Lucid. Did you see Andy's email to kernel-list about accounting for the differences in moblin kernel speed? [21:10] He's accounted for all of the differences without the patches we're looking at [21:10] bryce, thats odd as tseliot claimed 2s with some x patches included [21:13] bryce: apw's kernel doesn't have all the patches that mine will, but they still may not make much difference [21:19] sconklin, ok no I'm not on the kernel list [21:19] apw, he thought it might be due to me running on SSD and him not on SSD [21:19] bryce, may be so indeed [21:19] do you use compiz jcristau? i just wrote down the time of every flicker for the past 30 minutes and when i turned on compiz halfway it went from having 80 seconds exactly between flickers alot of the times to 20 seconds [21:19] apw, my X boots in <2 sec on SSD already, so a 2 sec savings would be pretty astounding ;-) [21:20] bryce, yeah it would [21:20] I've a i945 laptop with a much longer X boot, I could try on that [21:20] but since the target hardware is SSD, I'm sort of more interested in that case [21:20] yep. testing here on the same [21:21] Sarvatt: no [21:21] bryce: basically, turning off initramfs and ISA-PNP results in a kernel init time of about 0.8S [21:21] mm [21:22] Sarvatt: so that seems to match what you see [21:22] metacity compositing maybe? i didnt test that [21:22] no compositing for me. plain old fluxbox. [21:22] oh i misread what you said [21:22] sorry [21:22] bryce, you done any testing on lucid yet.. [21:23] thought you said it went 80 -> 20s by turning compiz *off* [21:23] * apw thinks he just had a hang on there on the 10v [21:23] bryce, oh i forgot ... did you note ATI KMS is now on by default [21:24] its weird, if i stared at gnome terminal with nothing else open it'd have chunks of time where it'd go off 80 seconds exactly, if i opened a new program it'd almost always trigger it not long after [21:24] apw, yeah I've got both my laptops on lucid now. [21:24] brave :) [21:24] Sarvatt: jesse suspected the fifo watermarks settings. i haven't had time to play with that yet. [21:25] I ran into an issue where upstart isn't starting gdm, unless I remove some line about kernel tty7 from grub.cfg [21:25] bryce, i think i saw a flicker on my 10v too, like a jump left an inch for a frame [21:25] filed a bug [21:25] bryce, ouch [21:25] apw, not so brave... I still keep my desktop on karmic for now ;-) [21:25] apw: what chip is the 10v? [21:25] but I'm not traveling for the next couple months so if laptop breaks it's not a problem [21:26] jcristau, i965 iirc [21:26] 00:02.0 VGA compatible controller: Intel Corporation Mobile 945GME Express Integrated Graphics Controller (rev 03) [21:26] ah 945 [21:26] could be the same bug Sarvatt and i are seeing [21:26] thats what I have, i get the flickering after a suspend/resume cycle every time on 2.6.32 [21:27] bah ... and we are close to final on the kernel [21:27] * jcristau gets it after boot, no suspend needed [21:27] also i cant suspend with a SD card mounted on 2.6.32 [21:27] and its not a ubuntu patch problem, happens when i compile a vanilla kernel [21:28] Sarvatt, we had a patch for that in karmic and i think its still in lucid [21:28] may not work of course [21:29] hmm wonder if i suspend without a dpms off from pm-utils if it'll still flicker [21:30] or lock up on the solid color [21:30] Sarvatt, on the SD card thing, it hangs hard on suspend yes? [21:30] yeah, the screen wont turn off but the system freezes [21:30] as you can reproduce it, perhaps you could boot with 'no_console_suspend' on the kernel commad line, and try a suspend from VT1 using sudo pm-suspend [21:30] and see if you get a panic [21:31] i'll do that right now [21:38] same issue, /var/log/pm-suspend.log shows a success but it was frozen with the screen still on and unresponsive [21:40] http://paste.ubuntu.com/333422/ [21:40] is it worth sending announcements of test kernels like the moblin patches on Karmic to the ubuntu-x list? [22:04] bryce: preliminary merge done, there are a bunch of patches that need a review and/or a refresh [22:06] we should probably replace patch 140 by disabling acpi altogether [22:07] since it's not used by anything [22:07] http://cvs.fedoraproject.org/viewvc/devel/xorg-x11-server/xserver-1.6.0-less-acpi-brokenness.patch?revision=1.1&view=markup [22:07] one way to do it :) [22:35] hmm, looking like a good excuse to go splurge on an SSD and just shutdown instead of suspending :D [22:35] tjaalton: patch 140 is obsolete anyway [22:36] (commit 7a05c8b1e70680ddd3b3e09ad448788f8d70a428) [22:41] 140 174 178 179 180 fix-dga-removal 181 182 183 185 187 Add-libgcrypt-as-an-option-for-sha1 were all upstream last i looked, but i was looking at master not 1.7 branch [22:42] Sarvatt: the libgcrypt one is only on master === seb128_ is now known as seb128