=== lilstevi is now known as lilstevie [00:26] have there been any reports of spurious mouse clicks? [00:28] mdeslaur, synaptics or evdev? [00:28] evdev...it's a usb mouse [00:29] I've not seen reports of spurious mouse clicks recently [00:29] there seem to be none open on my natty list - http://www.bryceharrington.org/Arsenal/Reports/ubuntu-x-swat/workqueue-natty.html [00:30] ok, am just wondering if it's my hardware...I'll try another mouse [00:30] thanks bryce_ [00:30] there were a couple weird bugs with erratic pointers but those were synaptics-specific [00:34] And now would be the time for crazy pointer-based bugs. [00:34] oh yeah, new input stack [00:34] mdeslaur, just updated today and noticed it for the first time? [00:35] bryce_: I noticed it maybe a day or two ago, after an update [00:35] but, as I said, it maybe hardware [00:35] I just plugged in another mouse, we'll see [00:36] mdeslaur, ok, then that gets cnd and raof off the hook ;-) [00:36] hehe [00:36] although if it is hardware, I'll be sure to bring the broken mouse to UDS for a practical joke :P === cnd` is now known as cnd [01:06] RAOF, bryce_: changing the tls type to local-dynamic didn't help any... [01:06] so I don't really know what a good resolution is [01:08] and it seems to have just made things worse all around [01:09] RAOF, unless these variables are used in other libs? [01:11] Does it interact with the split-glapi stuff? You're testing against natty's mesa, right? [01:12] RAOF, I took natty's mesa package and replaced any occurrences of the initial-exec model with local-dynamic [01:12] rebuilt [01:12] and installed [01:14] it doesn't look like the two headers with the occurrences are installed in any packages [01:14] so they are confined to the library [01:14] Yeah, that'd be right. [01:14] I'm going to try to make them generic [01:14] just to see if that works [01:14] Is there any raw asm that's getting used? [01:16] yeah [01:17] do you think that means it might require rebuilding other packages, or that the variables need to be generic since they may be referenced by other libraries? [01:17] I can't imagine the latter, since if you can't find them in a header file you can't access them [01:18] I'm not sure. [01:18] I should just make a test.c to try to load the libs in different orders [01:18] and see what happens [01:19] * RAOF goes to find drepper's tls paper [01:31] RAOF, well, I can't seem to trigger the bug with a simple test case [01:31] and if I can't fix this kind of bug quickly, it moves down my priority list :) [01:31] I gave the libavg devs a two line python workaround [01:31] Heh. [01:31] Ok. [01:31] it just loads libstdc++ before anything else :) [01:31] I'll look into it. [01:31] cheeky [01:32] so it's not a high priority [01:36] Hey! Now that I think of it, doesn't libGL *link* to libstdc++? [01:36] But the dri drivers do. [01:41] So libstdc++ is *already* being loaded by dlopen after libgl. [01:43] RAOF, does it? [01:43] that seems odd [01:43] to use opengl you have to load all the c++ stuff? [01:43] Yup. [01:44] Because the glsl compiler is written in C++, so needs libstdc++ [01:44] hmm, then I have no clue :) [01:44] And the glsl compiler is built into the dri drivers, which are dlopened by the non-C++-using libgl [01:45] What's the actual problem again? :) [01:45] RAOF, then you're beyond my comprehension now :) [01:46] I just know that libavg dlopens libGL [01:46] and if you don't preload libstdc++, you get a segfault [01:46] and it's claimed that this is due to tls stuff [01:46] Maybe it's passing incorrect flags to dlopen? [01:46] RTLD_NOW? [01:47] that's all it's passing [01:47] Let's ask someone who understands the dynamic linker! [01:47] * RAOF nominates Colin Watson. [01:47] :) [01:47] good luck finding him at this hour [01:49] I don't. I plan to read about TLS circa 2001 and eat toast. === Solarion_ is now known as Solarion [06:05] humber:~/ubuntu/linux/linux-2.6$ sudo lsinput [06:05] [sudo] password for bryce: [06:05] /dev/input/event0 [06:05] protocol version mismatch (expected 65536, got 65537) [06:05] wth? [06:08] Oh, does it need a rebuild again? [06:08] possibly I just haven't rebooted since updating [06:09] (although I'm not being prompted to reboot) [06:09] I think it needs to be rebuilt against newer kernel headers. [06:10] I'll do a upgrade and reboot and see if it still repros [06:11] Looks like this might be bug #723944 [06:11] Launchpad bug 723944 in xorg (Ubuntu) "Synaptics touchpad no more working after disribution upgrade (affects: 2) (heat: 12)" [Undecided,New] https://launchpad.net/bugs/723944 [06:38] bryce_: A quick rebuild fixes input-utils. [06:38] cnd: You wouldn't happen to be awake, would you? [06:41] we should phone him (tee hee) [06:42] So, those logs on the bugs don't seem to be particularly informative. [06:42] Except that the X driver is loading, and thinks it's driving the devices. [06:44] yeah [06:45] nothing remarkable in dmesg either [06:46] RAOF, I'm dist-upgrading 3 laptops now, will try reproducing on each [06:46] RAOF, have you had luck reproducing it on your hw? [06:47] Not so far. [06:47] But I'll upgrade my other laptop and see. [06:50] the protocol mismatch thingee seems to pre-date the xserver update today [06:50] kernel bug maybe? [06:51] the synaptics-thing? [06:52] * bryce_ points tjaalton at #canonical [06:52] bryce_: That's a regular occurrence - input-utils needs to be kept in sync with the kernel evdev header, and I guess it hasn't had a rebuild recently enough. [06:52] ok, so that's item #1. guessing we need a kernel guy to twiddle something? [06:52] those spurious mouseclicks.. think I'm seeing those sometimes [06:52] can we get the same info from proc or sysfs? [06:52] with thunderbird. clicking on emails open in new tabs [06:53] but not always [06:53] bryce_: No; we just need a no-change rebuild of input-utils. [06:53] RAOF, ok [06:53] And I don't think we can get that info anywhere else; it reads the raw event stream out of the evdev nozzle. [06:53] RAOF, looks like input device info can be gotten at /proc/bus/input/devices [06:54] Yeah, that gets some. [06:55] Hm. My laptop says “failed to open grail, no gesture support”, and I didn't notice that in any of the other xorg.0.logs [07:01] Yeah. All the affected systems seem to have utouch support. [07:02] interesting [07:03] yeah I have two laptops using synaptics, neither seems affected [07:03] What are the buttons reported in Xorg.0.log? [07:04] My unaffected laptop has “left right”, my other laptop (which I predict will be affected once it's finished upgrading) has “left right double triple” [07:04] http://paste.ubuntu.com/571563/ [07:05] Ok. You *also* fail to open grail. [07:05] http://paste.ubuntu.com/571565/ [07:06] I installed utouch on that second one, still no repro [07:07] Oh, the driver links to libutouch-grail1; the “failed to open grail” is from the failure of grail_open(device) [07:08] ok, guess we can call that issue #2. I take it it's innocuous? [07:09] I presume it'll happen when the touchpad doesn't support enough for grail to be all gesturiffic. [07:09] But given that second Xorg.0.log of yours is for a *non* reproducer, it's clearly not the (sole) issue. [07:10] Although we do apply some quirks to your second Xorg.0.log. [07:11] second box is a dell mini, I think if that had busted mouse we'd get LOTS of bug reports today [07:11] third box: http://paste.ubuntu.com/571566/ [07:12] also verified using synaptics but no repro [07:14] That one also has a bonus PS/2 mouse :) [07:14] fwiw that one's in a docking station [07:15] it's got a touchscreen lvds and a touchscreen external monitor and touchpad [07:15] huh, we still have -input-fpit in the archive, provides -input-4 :P [07:15] but no ps/2 mouse [07:16] dev/input/event6 would care to differ :) [07:17] it can think what it wants [07:18] actually the external touchscreen functionality isn't hooked up [07:18] but anyway [07:18] well I'm running out of ideas [07:18] it seems not to be abi breakage else it would be universally broken [07:19] Right. [07:19] I'd like to see the raw event stream from “xinput test”, or at least know that such a stream exists or doesn't. [07:19] the mouse0 / mouse1 bit seems not particularly unusual (I have that same bit my xorg log) [07:20] Yeah. We don't match anything against mouse* [07:20] It's a symlink to somethnig more interesting, so we'll already have covered it earlier. [07:20] RAOF, maybe you could forward one or both of those bug reports to peter (and cnd when he comes on)? [07:20] I shall. [07:20] Well, peter? [07:21] hutterer [07:21] this isn't upstream, if it's due to the multitouch patches [07:21] Yeah, I know who you meant, but it's going to be multitouch releated. [07:21] ok, just thought if he and cnd were in communication about this [07:22] daniels and cnd are [07:22] ahh [08:36] hmm pbuilder didn't fail on x11-utils without the old changes, so it should be sync'able [09:05] Hello, anyone an idea why my mouse scrollwheel doesn't work anymore in my natty installation? [09:06] geser: does xev list any events for it? or evtest? [09:16] RAOF: heh, so keithp is asking if removing randr-1.4 from 1.10 would be something people want [09:17] Well, I guess he could; no one's using it yet. [09:17] At least, I don't *think* anyone's using it yet :) [09:17] it'd revert the abi as well aiui [09:17] Yeah, it would. [09:18] But that wasn't an ABI that drivers *needed* to care about IIUC [09:18] right, it's just another round of rebuilds :) [09:18] tjaalton: no, xev doesn't show any events when I scroll (but pressing the scroll wheel is shown as button 2 as expected) [09:18] tjaalton: I'm not sure it'd even require a round of rebuilds. [09:18] geser: then check if you get events from evtest on the console [09:19] lunch-> [09:25] tjaalton: evtest doesn't list any events either :( [09:30] Looks like something's broken with the Synaptics touchpad driver [09:32] trying out my other mouse (a Logitech Wireless M305) everything works as expected [09:34] hmm, re-pluging my mouse seems to work too [09:38] geser: so if evtest doesn't show anything it's because of the kernel [09:39] cdbs: Aha! You've got a piece of affected hardware? [09:40] RAOF: yes, its an old Synaptics touchpad that doesn't support multitouch [09:40] RAOF: Should I post evtest output to bug? [09:40] * cdbs got the bug [09:41] It wouldn't hurt. Also, working out whether “xinput test” shows events when you do things with the touchpad would be nice. [09:44] RAOF: My thing doesn't have /dev/input/event7 . It has only upto 5 [09:44] Which one should I use? [09:44] xinput list [09:44] pick the id for the device [09:53] RAOF: xinput test doesn't help [09:54] RAOF: no output, except when I clicked the buttons (which are working fine) [10:00] okay, I downgraded using the downgrade scrip [10:00] t [10:14] Hi, I did downgrade the packages, but after that I got a far less usable system [10:15] Every application I would open, would crash [10:15] EVERY [10:15] from nautilus, to even gnome-terminal [10:17] that's not X's fault [10:19] well, not [10:19] but that opens up a possibility: [10:19] Why not just upload a 'revert' package of xserver-xorg-input-synaptic? [10:20] It appears that these applications and GTK bindings are built against the new ABI, while the downgrade script as given by RAOF takes us back to the old one [10:20] lets give cnd a chance to fix it [10:25] seb128: there's a new(ish) libgnomekbd (2.30.2), are there plans to update? [10:26] tjaalton, we are on 2.32 which is a newer serie [10:26] tjaalton, or do you mean a sru for some stable version? [10:26] tjaalton, 2.30 is used in lucid [10:26] seb128: there's 2.30.2 in debian [10:26] but it has 2.30.2? [10:26] tjaalton, well lucid has that [10:26] grah, sorry [10:26] tjaalton, natty has 2.32 [10:26] mixed up the versions [10:27] debian has 2.30.2, we have 2.32.0 :) [10:27] right [10:27] just curious how did you run into that? [10:27] is there any bug in natty? [10:27] no, it's just listed on the versions_current.html list [10:27] dunno why [10:27] since it's not maintained by the x team [11:07] hm, -fpit is still maintained upstream it seems.. [11:07] for abi changes at least [11:15] and it seems to still have users.. [13:08] RAOF, if you're still up, I'm now up too :) [13:08] I read the backscroll here and in ubuntu-devel [13:08] taking a look at the bug now [13:24] if I had to guess, it looks like grail is opening on non-multitouch hardware and this is throwing things off [14:53] Hey guys I have a reproduceable crash I want to get a bt from and I install debug packages but gdb tells that /usr/bin/X has no debugging symbols [14:54] What is the problem? [14:56] soreau: the real binary is /usr/bin/Xorg [14:56] Well &%* [15:00] soreau, it's not a big problem [15:00] do you have a core dump? [15:01] or are you wanting to attach to X? [15:01] Yes I got the dump now [15:01] thanks jcristau [15:01] ok [15:01] soreau, by any chance are you using multitouch hardware? [15:02] :) [15:02] nope [15:02] It happens when trying to vncviewer into the machine [15:02] It worked until a couple weeks ago [15:02] ok, cool [15:02] not my problem :) [15:03] fdo#30032 is a crash when using x11vnc, could be the same thing? [15:09] tjaalton: got a few mins to upload a fix for the synaptics issue? [15:11] or bryce_, if you happen to be up already [15:33] cnd: i'm at the hw store, back home in 30min [15:33] comp hw :) [15:33] tjaalton: ok, thanks! [16:00] any idea what's going on https://bugs.launchpad.net/bugs/714280 (my last messages) [16:00] Launchpad bug 714280 in xorg-server (Ubuntu) (and 3 other projects) "The error was 'BadLength (poly request too large or internal Xlib length erro'. (affects: 5) (dups: 1) (heat: 233)" [High,Confirmed] === bigon_ is now known as bigon [16:03] what's going on is you're mixing different bugs. [16:05] my 2 last messages are still refering to BadLength (poly request too large or internal Xlib length error) [16:05] for a different request [16:05] changedrawableattributes is fixed in 4324d6fdfbba17e66b476cf008713d26cac83ad1 [16:06] should I open a different bug for this one? [16:06] dunno [16:06] you could just upgrade your X server [16:06] or your mesa. [16:07] both machine are uptodate (with natty) [16:07] then natty doesn't yet have the fixes [16:07] indeed [16:08] so upgrade your X server or mesa, or live with the bug until natty bumps either of them. [16:10] actually, natty's X server has that fixed. [16:12] cnd: ok, finally home [16:13] tjaalton, great [16:13] tjaalton, I suppose you'll probably want a source package [16:13] didn't think to create one... [16:13] cnd: yep [16:13] heh [16:13] I'll do that right now [16:21] cnd: I'll have dinner in the meantime [16:21] won't take long [16:21] tjaalton: I just uploaded [16:21] ok [16:21] oh [16:21] where? [16:21] it's at people.canonical.com/~cndougla/utouch/ [16:22] Checksum doesn't match for /tmp/people.canonical.com/~cndougla/utouch/utouch-frame_1.1.1-0ubuntu1.debian.tar.gz [16:22] hmm... [16:23] that's odd [16:23] I'll try to reupload it [16:24] tjaalton, try once mre [16:24] if that doesn't work, I'll try things locally to figure out what's wrong [16:24] still the same [16:24] hmmm [16:24] ok [16:24] 4cf85322570b8ad7ce800212c582e655 [16:25] is the md5sum [16:25] when source.changes says it should be e113aa341f8e3b4e784fdf374f4d1c02 [16:25] ummm... my version locally doesn't match either! [16:25] https://wiki.ubuntu.com/ChaseDouglas/DeveloperApplication-uTouch [16:25] oops [16:25] dc9a4f63e84871a89984b1fd839d9236 [16:26] that's the tarball [16:26] oh, I was looking at the wrong file [16:27] I'll recreate the source package and see what's wrong [16:27] I did a test build after the source package build [16:27] maybe the test build overwrote it [16:27] could be [16:28] i'll eat now :) [16:29] ok, it's reuploaded when you are ready [16:34] cnd: yep, didn't complain this time [16:35] great [16:44] jcristau: I will open an other bug I've 1.10 rc2 installed and still have the issue X_GLXChangeDrawableAttributes request [16:56] installed or running? [17:27] bryce_, RAOF, tjaalton: I'm applying for package upload rights for the utouch packages [17:27] if you'd like to sponsor me, please do so at https://wiki.ubuntu.com/ChaseDouglas/DeveloperApplication-uTouch [17:27] thanks! [17:32] 17:56 < jcristau> installed or running? << running I've rebooted to be sure [18:06] morning [18:06] synaptics issue get sorted out? [18:11] bryce_, we think so [18:11] new utouch-frame should fix it [18:12] the package should be synced to the mirrors by now I believe [18:13] it is for amd64 at least [18:13] it affected multi-finger, non-mt trackpads :) [18:15] cnd, excellent [18:16] bryce_, in case you didn't see above, I'm trolling for package upload rights sponsors [18:16] so we can fix these issues faster :) [18:16] we waited around for about an hour before tjaalton became available [18:16] and didrocks is our normal uploader, but he's swamped [18:17] cnd, right! hitting the page now [18:17] bryce_, thanks! [18:18] i'll add my comments after bryce :) [18:18] cool [18:22] hmmm, so I can reproduce this on my netbook by just attempting to "install ubuntu" https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/714829 [18:22] Launchpad bug 714829 in xserver-xorg-video-intel (Ubuntu) "Xorg segfaults during LiveCD installation using preseed file (affects: 1) (heat: 201)" [High,Incomplete] [18:22] only, X never crashes [18:23] nice [18:23] i get tons of the [ 194.542] (WW) intel(0): intel_uxa_prepare_access: bo map failed: Cannot allocate memory though and X is still alive and kicking but unresponsive [18:23] can you get to the machine somehow? [18:24] yeah i'm using it now [18:24] gem objects look fine http://sarvatt.com/downloads/i915_gem_objects.txt [18:24] hmm I bought an ssd, could pyt it in my x61 [18:24] put even [18:25] dont have to actually wipe anything even, just booting the livecd, going to install ubuntu then hitting next kills things [18:25] hm ok [18:27] so something is eating the memory, and it's not x/intel [18:27] its right when it'd be starting up partman to do all the disk partitioning that things die [18:28] hm, have you checked it's logs? [18:28] i cant even see the panel at the top in unity on a 1024x600 screen [18:30] now i'm getting some aufs error spam, aufs au_new_inode:412:dbus-launch[2578]: Warning: Un-notified UDBA or repeatedly renamed dir, b0, tmpfs, ubuntu, hi1965, i273. [18:31] oh got a crash that time, went to VT immediately after hitting install ubuntu [18:33] try ubuntu is fine, looks like installing it from inside the live session works fine too.. [18:34] better let it finish before I say that I guess :) [18:43] I have an Xorg crash http://pastebin.com/Etjjzue8 when trying to vncviewer into it from another machine :) [18:54] bryce_, btw, I'm going to ask at the DMB meeting for motu rights too, so we'll see [18:54] thanks for the endorsement :) [19:03] cnd, yep, good luck! [19:08] bryce_, because I'm not sure the best route for me, I asked cody on the dmb [19:08] he said he'd be happy to review an application of mine before going to the dmb for core dev rights [19:08] I just don't want to waste the dmb time [19:08] their meetings get long and slow enough as it is [19:09] so I'm going to create a new application page for core dev [19:09] and ask for some endorsements [19:12] cnd, sounds good [19:13] cnd, yeah it would not be a bad idea to shoot for utouch rights in the coming meeting, then apply for motu the meeting after that, and core dev the one after that, in sequence [19:13] something like that [19:13] cnd, however you do it, you got my endorsement :-) [19:14] heh [19:14] Sarvatt, aha you have a lead on our elusive memory leak bug [19:15] bryce_: i'm drawing up blanks on why this is happening only when "install ubuntu" is picked, installing from inside "try ubuntu" just hangs before the disk partitioning tool comes up but X is still here and i'm still able to quit the install fine even [19:15] Sarvatt, yeah we've suspected it was something like the installer. I've seen the bug reports only against use in the livecd environment [19:15] Sarvatt, have you found a way to view memory usage that shows what process is gobbling mem? [19:16] nothing is using an abnormally large amount of memory [19:16] Sarvatt, i.e. does top, cat /proc/meminfo, xrestop, or the like show something? [19:16] I've also wondered if it might be just video memory getting exhausted [19:17] but I don't know if we have a tool for measuring that... is it shown in one of the gem files in sysfs? [19:17] was playing around in the try ubuntu option there, rebooting into install ubuntu to see [19:17] it looks fine http://sarvatt.com/downloads/i915_gem_objects.txt [19:17] (thats from right before an X crash while it was spamming the bo map failed errors) [19:18] I wonder if the Install Ubuntu functionality can be run independently of the livecd [19:18] i can't see the menu bar on the selection screen where you pick try or install either, its above the screen so the window will fit [19:19] yeah trying to figure out how the heck to do just that at the moment [19:19] Sarvatt, have you talked with evand or cjwatson yet? [19:20] if not, maybe we should spin them up, they might have some suggestions that'd save us some time [19:24] I'll poke through their bug tracker and then see if I can grab evan [19:26] fwiw, on radeon since updating to latest stuff yesterday I've been noticing with popup menus that sometimes in the fraction of a second before the menu is displayed where what I suppose is supposed to be a black rectangle, I catch a momentary glimpse of what looks like uninitialized video memory [19:26] bryce_, I have copied my utouch endorsement to my coredev application: https://wiki.ubuntu.com/ChaseDouglas/DeveloperApplication-CoreDev [19:26] it has some utouch rights specific phrasing [19:27] would you like to edit it, or do you want me to fix it up for you? [19:27] but it only displays for a few milliseconds so can't see it properly, and seems hard to reproduce on demand [19:27] cnd, I can edit it up, thanks [19:28] ok [19:33] done [19:48] Sarvatt, ok I pinged ev on #ubuntu-devel. guessing he's EOD'd [19:48] Sarvatt, gave him a rundown maybe he can help [19:53] Sarvatt, hrm, finding nothing interesting amongst the ubiquity bug reports [19:55] Sarvatt, reviewing our own bug report history with this issue, the first report (from mario) was Jan 19th, so I would guess the issue first entered the archive around Jan 17/18 [19:55] ubiquity 2.5.6 is dated Jan 17 [20:07] bryce_, RAOF, btw, I forgot to mention that I had been planning to fix up lsinput [20:07] in fact, I think I'll sit down right now to do so :) [20:14] bryce_: apparently just sitting on the welcome screen is enough to trigger the bo map failed errors [20:16] * Sarvatt is going to try reverting this force background redrawing commit from jan 17th [20:17] Sarvatt, quite unexpectedly just following random hunches I found some leaky C code [20:18] Sarvatt, going on the assumption that it was ubiquity, and that it must have happened shortly before feb 19th, I looked at the diff [20:18] ubiquity 2.5.6 is a huge diff but 99.9% is just python and translations stuff [20:18] however there is one change to one C file [20:18] and wouldn't you know it, that change involves adding a pointer, strduping to it, and there is no matching free [20:19] it's in d-i/source/netcfg/netcfg-common.c routine netcfg_write_common [20:22] Sarvatt, anyway I can do up a patch at this point. need food first [20:22] I'm 70% certain this could be our culprit. [20:23] no idea how we'd test it though [20:26] bryce_: awesome [20:54] is it just me, or are all lp related functions really slow today? [21:10] I've had a few timeouts but nothing out of the ordinary [21:11] bryce_: so uh.. http://sarvatt.com/downloads/patches/wtf.diff has lasted 10 minutes on the welcome screen with no bo map errors [21:12] bryce_: 204 seconds was my record before [21:12] Sarvatt, aha, where'd you find that? [21:12] http://bazaar.launchpad.net/~ubuntu-branches/ubuntu/natty/ubiquity/natty/revision/380#src/panel/panel.c [21:14] something tells me it might have been queuing a redraw way more often than they wanted? :D going to attempt an install now and see what happens [21:14] ok [21:15] of course it wont go past the preparing to install screen because partman wont load up but still not crashing [21:17]