Sarvatt | https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-input-synaptics/+bug/944415/comments/3 | 00:08 |
---|---|---|
ubottu | Launchpad bug 944415 in xserver-xorg-input-synaptics (Ubuntu) "Input stops responding" [Undecided,New] | 00:08 |
Sarvatt | dang now it's hung 4 times in the past 5 minutes | 00:11 |
Sarvatt | testing out http://bazaar.launchpad.net/~bregma/utouch-geis/lp-937021/revision/209 | 00:40 |
Sarvatt | err https://code.launchpad.net/~bregma/utouch-geis/lp-937021 | 00:41 |
* Sarvatt shoots in the dark since disabling geis from unity yet again seems to fix it | 00:41 | |
Sarvatt | hmm, xutils-dev could use going into backports | 01:05 |
bryceh | Sarvatt, agreed | 01:06 |
bryceh | Sarvatt, or even just FFe it in? | 01:06 |
* Sarvatt looks into whats needed for that | 01:06 | |
bryceh | I know we have a few things wanting the newer version, does it have any deps itself we can't meet easily? | 01:06 |
Sarvatt | nope | 01:06 |
bryceh | any potential side effects or conflicts? | 01:07 |
Sarvatt | there were one or two releases of xutils-dev that build-dep'ed on xutils-dev that would have caused a headache but that was fixed a year ago | 01:07 |
Sarvatt | can't see how it could conflict or have side effects if its in -backports, people would have to willingly enable that and wouldn't screw with updates, it'd just let people backport newer drivers or X to old releases | 01:10 |
Sarvatt | ah it doesn't look automated, thought it was for some reason | 01:11 |
Sarvatt | hmm failsafeX is busted, no input | 02:10 |
bryceh | yeah I need to figure that out | 02:29 |
Sarvatt | sforshee: so, are you using the macbook air and want it stable, or are you just looking to get it fixed? I can send you unity packages that actually work if you need to be productive on it in the meantime if the latter :) | 03:57 |
sforshee | Sarvatt, I can usually get it to happen if I load up a long page in chromium and scroll with the touchpad, but at times it does seem to stop happening for a while | 04:36 |
sforshee | it also seems to happen sometimes after resume if I suspend by closing the lid | 04:37 |
sforshee | I don't need an immediate workaround, it's not my main machine and I can avoid it pretty well, I just want to see it fixed | 04:38 |
* Sarvatt nods | 04:42 | |
Sarvatt | just wanted to be sure, i use this machine as my primary one and need it working :) | 04:42 |
Sarvatt | i haven't seen it from closing the lid yet, thats strange | 04:43 |
Sarvatt | but i dont suspend on lid close so maybe that explains it | 04:43 |
Sarvatt | that seems like it might not be the same bug though | 04:43 |
sforshee | for a while there was a bug where the machine wouldn't suspend when the lid was closed, and when I opened the lid it definitely looked the same | 04:47 |
sforshee | now it's on the text console rather than the desktop, but it's still similar in that the only thing that updates is the mouse pointer | 04:48 |
sforshee | heh, I just opened the lid and got the freeze. What to I look at to verify it's the same problem? | 04:49 |
Sarvatt | sforshee: i'm 100% sure you're hitting the same problem and it really seems higher up in the stack like in utouch-geis https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-input-synaptics/+bug/944415 | 04:52 |
ubottu | Launchpad bug 944415 in xserver-xorg-input-synaptics (Ubuntu) "Input stops responding" [Undecided,New] | 04:52 |
Sarvatt | from your other bugs compiz backtrack | 04:52 |
Sarvatt | backtrace | 04:53 |
Sarvatt | bah | 04:53 |
Sarvatt | its late, why am i awake :) | 04:53 |
sforshee | Sarvatt, this time the stacktrace is different | 04:56 |
sforshee | it's going through intel_update_renderbuffers() | 04:57 |
Sarvatt | so yeah suspend sounds like a different problem then | 04:57 |
Sarvatt | bugs all around! | 04:57 |
Sarvatt | intel_update_renderbuffers(), can you pastebin it? | 04:58 |
sforshee | yeah, just a minute | 04:58 |
Sarvatt | think i can point you at that bug | 04:58 |
Sarvatt | sounds like the other big bug we're hitting in mesa at the moment | 04:59 |
sforshee | http://pastebin.ubuntu.com/864587/ | 05:00 |
Sarvatt | ok no thats different, its just trying to draw, the mesa problem was an actual crash | 05:01 |
Sarvatt | nothing in ~/.xsession-errors that time? | 05:01 |
sforshee | there was, but I forgot to save it. D'oh. | 05:02 |
sforshee | i should be able to produce it again with a little effort | 05:02 |
Sarvatt | me too, same machine and all | 05:03 |
Sarvatt | but cant seem to hit it.. 10 s3's so far | 05:04 |
sforshee | yeah, I don't see it that often anymore | 05:04 |
Sarvatt | maybe something in /var/log/lightdm/x-0.log? | 05:05 |
Sarvatt | just got http://paste.ubuntu.com/864594/ but it resumed fine that time | 05:06 |
sforshee | I'm not finding what I saw in .xsession-errors anywhere in /var/log, although x-0.log does show the 'unable to find touch point' messages | 05:07 |
Sarvatt | oh duh, i'm using a unity with geis support ripped out that actually works | 05:08 |
Sarvatt | thats probably why i can't reproduce it | 05:08 |
Sarvatt | i'll mess with it tomorrow, its too late here, thanks for the pointers :) | 05:09 |
sforshee | yeah, I need to go to bed too | 05:09 |
sforshee | thanks for looking! | 05:09 |
sforshee | Sarvatt, a little something for tomorrow morning: http://pastebin.ubuntu.com/864597/ | 05:14 |
sforshee | that's .xsession-errors from the freeze after s3 | 05:15 |
Sarvatt | wth is up with all the g-s-d xrandr spam | 05:15 |
Sarvatt | i dont get that | 05:16 |
* Sarvatt wonders what he's changed locally now | 05:16 | |
sforshee | the xrandr spam isn't there before s3 but is afterwards, whether or not it freezes | 05:19 |
=== yofel_ is now known as yofel | ||
sforshee | Sarvatt, cnd: just got the desktop freeze coming out of s3 again. I posted some information I collected to bug 938770. | 15:30 |
ubottu | Launchpad bug 938770 in xserver-xorg-input-synaptics (Ubuntu) "Desktop becomes completely unresponsive" [High,Confirmed] https://launchpad.net/bugs/938770 | 15:30 |
cnd | sforshee, if you do "unity --replace" does it fix things? | 16:09 |
sforshee | cnd, yes, "unity --replace" does fix it | 17:41 |
FernandoMiguel | alias compizB='DISPLAY=:0 compiz --replace &' | 17:41 |
FernandoMiguel | alias compizC='DISPLAY=:0 compiz --reload &' | 17:41 |
FernandoMiguel | :D | 17:41 |
cnd | sforshee, then it's probable that a touch is not getting rejected or accepted properly | 17:42 |
cnd | sforshee, please let me know if you figure out a reliable way to reproduce it | 17:42 |
sforshee | cnd, I can reproduce it pretty reliably on this machine, but not on any others | 17:43 |
sforshee | in fact, it just happened when I opened the lid | 17:43 |
cnd | sforshee, ok, see if you can figure out a series of steps that always or usually makes it happen | 17:44 |
sforshee | cnd, always is tough. I can usually get it by scrolling in chromium using gestures, although it might take a while to trigger it | 17:45 |
sforshee | but once in a while it becomes very hard to reproduce | 17:45 |
cnd | sforshee, I wonder if it occurs when you accidentally lightly brush a third finger while scrolling? | 17:46 |
sforshee | hmm, you might be right | 17:47 |
sforshee | two for two so far | 17:48 |
cnd | heh | 17:48 |
cnd | sforshee, are you momentarily touching a third finger? | 17:49 |
cnd | or dragging it along too | 17:49 |
cnd | i.e. would it be seen as a tap | 17:49 |
sforshee | i can trigger it with a 3-finger drag | 17:50 |
cnd | sforshee, any 3-finger drag? | 17:51 |
cnd | I mean to say, it triggers always when you do a 3-finger drag? | 17:51 |
sforshee | specifically over a chromium window, it doesn't do it with a 3-finger drag over the desktop | 17:52 |
sforshee | except now I can't :( | 17:52 |
cnd | sforshee, what about over other windows? | 17:53 |
sforshee | cnd, my system got completely hosed, I'm rebooting now | 17:54 |
cnd | ok | 17:54 |
sforshee | oddly, if i do a 3-finger drag over a blank desktop it acts like it's dragging a window | 17:55 |
sforshee | it's not happening with libreoffice calc, nor with firefox | 17:57 |
sforshee | but it happens immediately with chromium | 17:57 |
cnd | sforshee, chromium installed from the precise archives, correct? | 18:04 |
cnd | sforshee, do you have ginn running? | 18:05 |
sforshee | cnd, yes | 18:05 |
cnd | ps aux | grep ginn | 18:05 |
sforshee | cnd, sometimes it gets in a state where I can't trigger it even with chromium however | 18:05 |
cnd | sforshee, as soon as I get a new version of X synaptics uploaded to precise, I'll be switching to work on this bug | 18:06 |
cnd | hopefully I'll be able to reproduce it too | 18:06 |
Sarvatt | sforshee: did you have more macs than just the air? | 18:07 |
Sarvatt | wondering if its happening on other bcm5974's too | 18:07 |
sforshee | Sarvatt, I also have a macbook pro with the bcm5974 | 18:07 |
sforshee | I just tried a minute ago and reproduced it | 18:07 |
cnd | Sarvatt, sforshee: when the bug occurs, does the whole desktop lock up? | 21:10 |
cnd | as though unity is hung? | 21:10 |
Sarvatt | cnd: yeah, it looks fine but you cant interact with anything, the cursor does change between a pointer and a text entry line when you move it over things though | 21:11 |
sforshee | cnd, pretty much, the only thing that works at all is that the mouse pointer moves | 21:11 |
Sarvatt | vt switch then switch back to vt7 and its black with just a cursor | 21:11 |
cnd | yeah, so I think that may be bug 942625 | 21:11 |
ubottu | Launchpad bug 942625 in unity (Ubuntu) "Unity hangs when touching my touchpad/trackpad" [Undecided,Confirmed] https://launchpad.net/bugs/942625 | 21:11 |
cnd | we have a wallpaper fix | 21:11 |
cnd | we're trying to figure out the underlying root cause | 21:11 |
cnd | the wallpaper fix is really adding proper error handling, so it's needed anyways | 21:12 |
Sarvatt | awesome, i'll check that out and see if it is | 21:13 |
Sarvatt | hopefully it doesn't need new nux and all the other stuff | 21:14 |
Sarvatt | oh cool r2047 is in https://launchpad.net/~unity-team/+archive/staging | 21:15 |
Sarvatt | going through and finding all of the dupes will be fun, they're splattered across all of the drivers and unity and compiz and the kernel | 21:17 |
Sarvatt | got it running now, lets see how this goes | 21:29 |
Sarvatt | wow 175 manual tests in checkbox-unity, seriously? | 21:31 |
Sarvatt | should randomize the test questions so everything gets some coverage because thats nuts | 21:32 |
Sarvatt | cnd: that does indeed look like the fix | 21:40 |
Sarvatt | select 11 | 21:40 |
Sarvatt | print *data | 21:40 |
Sarvatt | $1 = {id = 4, device_id = 64176, window = 172, touches = 3, timestamp = 4764291, focus_x = 1129054208, focus_y = 1139212288, bound_x1 = -2594, bound_y1 = 5313, bound_x2 = 1465, | 21:40 |
Sarvatt | bound_y2 = 6572} | 21:40 |
cnd | Sarvatt, what are you printing? | 21:41 |
Sarvatt | #11 GeisAdapter::GestureStart (cookie=<optimized out>, gesture_type=<optimized out>, gesture_id=<optimized out>, count=<optimized out>, attrs=<optimized out>) | 21:41 |
Sarvatt | at /build/buildd/unity-5.4.0/plugins/unityshell/src/GeisAdapter.cpp:136 | 21:41 |
Sarvatt | data = 0x4b763d0 | 21:41 |
Sarvatt | self = <optimized out> | 21:41 |
Sarvatt | that fix did - if (data->touches == 3) | 21:42 |
Sarvatt | + if (data->touches == 3 && data->window != 0) | 21:42 |
cnd | yeah | 21:42 |
Sarvatt | 3 touches and window = 172 when its stuck | 21:42 |
cnd | that's the root window case | 21:42 |
cnd | which is fixed elsewhere | 21:43 |
cnd | I'm currently trying to figure out why data->window would ever be 0 | 21:43 |
Sarvatt | cnd: awesome, ya added a synclient property for clickpad? \o/ | 22:00 |
cnd | Sarvatt, yep | 22:00 |
cnd | synclient ClickPad=1 | 22:01 |
Sarvatt | so much easier to tell people that instead of explaining how to use xinput properties | 22:01 |
cnd | Sarvatt, yeah, though I'm not sure how well it works when there are more than two trackpads | 22:02 |
jcristau | badly. | 22:03 |
cnd | for people who use a magic trackpad with a laptop | 22:03 |
Sarvatt | this unity is looking good so far, it does indeed look worked around in 5.6.0 | 22:06 |
Sarvatt | cnd: is it normal none of the 3 finger gestures work? | 22:11 |
cnd | Sarvatt, I'm working on that right now :) | 22:11 |
cnd | there's a bug somewhere in the stack | 22:12 |
Sarvatt | oh cool | 22:12 |
cnd | where a rejected touch isn't being rejected fully | 22:12 |
cnd | or something like that | 22:12 |
cnd | and this causes future touch event handling to be wrong | 22:12 |
Sarvatt | 4 finger is kinda iffy too, sometimes it does a right click when i'm not pressing and have tap to click disabled | 22:13 |
cnd | yeah | 22:13 |
Sarvatt | first time using any of this | 22:14 |
cnd | we have the stack in place, just need to debug it now :) | 22:14 |
cnd | it's close, I think we're about 2 or 3 bug fixes from it working properly | 22:14 |
cnd | I need to figure out if unity/utouch is not accepting/rejecting touches properly | 22:15 |
cnd | or if the server isn't not handling accept/reject properly | 22:15 |
cnd | that's the next step | 22:15 |
cnd | Sarvatt, if you're interested in watching the recognizer, run: GRAIL_DEBUG=1 unity --replace | 22:16 |
Sarvatt | heh | 22:18 |
Sarvatt | Warning: failed to get previous touch value | 22:18 |
Sarvatt | but it doesn't die now | 22:18 |
Sarvatt | i get that from every single gesture | 22:18 |
Sarvatt | Warning: failed to get previous touch value | 22:19 |
Sarvatt | Warning: failed to get previous touch value | 22:19 |
Sarvatt | GRAIL ERROR (../../src/v3/recognizer.cpp:ProcessEvent:691): touch 305 failed to be rejected | 22:19 |
Sarvatt | GRAIL ERROR (../../src/v3/recognizer.cpp:ProcessEvent:691): touch 306 failed to be rejected | 22:19 |
Sarvatt | GRAIL ERROR (../../src/v3/recognizer.cpp:ProcessEvent:691): touch 307 failed to be rejected | 22:19 |
Sarvatt | GRAIL ERROR (../../src/v3/recognizer.cpp:ProcessEvent:691): touch 308 failed to be rejected | 22:19 |
Sarvatt | only grail debug i've seen so far | 22:19 |
cnd | hmm... that's not good :) | 22:19 |
cnd | that sounds like you have an old utouch-geis installed | 22:20 |
cnd | actually, that doesn't make sense | 22:20 |
Sarvatt | ah need a daily? | 22:20 |
cnd | nm | 22:20 |
cnd | I don't know why it would fail to be rejected | 22:20 |
cnd | but let me try to resolve the bug I can reproduce here first :) | 22:20 |
Sarvatt | http://paste.ubuntu.com/865832/ all i've seen so far | 22:21 |
Sarvatt | no worries sorry to distract you | 22:21 |
cnd | nah, no distraction :) | 22:21 |
cnd | just letting you know the order in which I'm tackling stuff | 22:22 |
Sarvatt | only doing 4 finger swipes up and down in the log, every gesture that does actually go through gives 4 errors | 22:23 |
cnd | ok, I found the problem in the X server | 22:28 |
cnd | but I know this area of code | 22:28 |
cnd | and there have been lots of issues with the logic here | 22:28 |
cnd | I hope I don't break anything else with a fix :) | 22:28 |
cnd | I'm going to make this fix, and propose an xorg-gtest testcase to go with it this time | 22:29 |
FernandoMiguel | my touchpad is terribly jumpy today :\ | 22:32 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!