james_w | wgrant: bug 293318 | 00:02 |
---|---|---|
ubottu | Launchpad bug 293318 in gnome-settings-daemon "gnome-settings-daemon leaks memory" [Undecided,New] https://launchpad.net/bugs/293318 | 00:02 |
wgrant | james_w: Argh. Let's see... | 00:16 |
james_w | hey wgrant | 00:16 |
james_w | I don't see anything obvious in your change | 00:16 |
wgrant | Neither. | 00:16 |
james_w | I imagine it does call some functions more frequently, and so an existing leak would become an issue | 00:17 |
wgrant | I expected that my previous batch of changes could leak. | 00:17 |
wgrant | Hmm. Possibly. | 00:17 |
wgrant | This shouldn't be calling anything that often, though. | 00:17 |
james_w | ah, ok | 00:19 |
james_w | is it just on device insertion/removal? | 00:19 |
wgrant | Yes. | 00:19 |
wgrant | Well, actually, some of the code will be running on every X event. | 00:20 |
wgrant | Which narrows it down somewhat. | 00:20 |
james_w | shall I ask him if he's inserting 20 devices a second? :-) | 00:20 |
james_w | ah, ok. | 00:20 |
wgrant | The stuff outside the if in devicepresence_filter will be running lots. | 00:20 |
wgrant | But that looks perfectly safe. | 00:22 |
james_w | yeah | 00:24 |
james_w | you don't have to unref the return value of gdk_x11_get_default_xdevice() do you? | 00:24 |
wgrant | xdisplay, you mean? | 00:25 |
james_w | yeah | 00:25 |
wgrant | It doesn't look like it. | 00:27 |
wgrant | It certainly doesn't leak here, on i386. | 00:27 |
wgrant | Even with thousands of hotpluggings. | 00:28 |
james_w | so you think this could be amd64 specific? | 00:29 |
wgrant | I wouldn't think so, but all of the other regressions caused in my changes in Intrepid were my code exposing 64-bit unsafeness in the xserver. | 00:30 |
wgrant | (well, X libs, actually) | 00:32 |
wgrant | I thought I saw it leaking very slowly then, but its usage just dropped, so that can't have been it. | 00:35 |
wgrant | james_w: Do you have amd64 hardware you can test on? | 00:37 |
james_w | sorry, I don't | 00:38 |
ianloic | hey I'm having X freezing up | 02:36 |
ianloic | so I installed the debug version of libgl1-mesa-dri | 02:36 |
ianloic | but it's not being loaded | 02:36 |
ianloic | oh wait | 02:36 |
ianloic | nevermind | 02:36 |
ianloic | wait no | 02:46 |
ianloic | I didn't get it | 02:46 |
ianloic | setting the LD_LIBRARY_PATH doesn't seem to help | 02:46 |
ianloic | how can I actually use these debug packages to get more meaningful stack traces? | 02:46 |
jcristau | gdb | 02:48 |
ianloic | jcristau, how can I convince X to load debug versions of modules & libraries? | 02:53 |
ianloic | jcristau, here's my bug: https://bugs.launchpad.net/ubuntu/+source/mesa/+bug/292234 | 02:54 |
ubottu | Launchpad bug 292234 in mesa "X Freeze in i965_dri.so" [Undecided,New] | 02:54 |
ianloic | jcristau, I feel (perhaps naively) that if I could just get a little more out of that stack trace I'd be able to hack at it myself :) | 02:54 |
tjaalton | omg, now the wayland-crazyness hit brainstorm: http://brainstorm.ubuntu.com/idea/15205/ | 08:01 |
wgrant | tjaalton: Already!? | 08:10 |
tjaalton | wgrant: yep.. | 08:10 |
tjaalton | soo many "facts" are wrong.. | 08:10 |
wgrant | Didn't he specifically say that it wasn't intended to replace X.org? | 08:10 |
tjaalton | right | 08:10 |
tjaalton | and to be useful it needs the xserver running on top of it | 08:11 |
wgrant | Do you have privs to close it and devcomment on it? | 08:11 |
tjaalton | yes | 08:11 |
wgrant | Good. | 08:11 |
tjaalton | done | 08:21 |
Ng | mvo: your HAL fdi thing for scrollwheel emulation - what led you to the properties you ended up setting? | 13:26 |
Ng | (wrt http://mvogt.wordpress.com/2008/08/15/xorg-evdev-and-emulatewheel/) | 13:26 |
mvo | Ng: you mean how I figured out the exact names? | 13:28 |
Ng | mvo: yeah | 13:28 |
Ng | I'm curious why it worked at all | 13:28 |
mvo | Ng: I think I asked around here | 13:28 |
mvo | it seems to be not plagued by the bug that it does not work after supsend | 13:28 |
mvo | or rather, affected differently, it seems like every second resume its fine, but about half/25% of the time its not. but cures itself on the next resume again | 13:29 |
Ng | mvo: that is kinda weird. It would be interesting to know if the patch fixes that | 13:30 |
Ng | but I'd also like to construct a HAL fdi for it. Not for my own use, but to prove that it works with those settings as well | 13:30 |
Ng | the thing is, you don't specify the wheel emulation Y axis, so I'm wondering if it defaulted to 4 5. It can't be relating to the ZAxisMapping you set, because you actually spelt Axis wrong, so unless X corrects for that, that setting was probably ignored ;) | 13:31 |
Ng | (I built a deb of the current intrepid i386 package with the patch included, if you fancy giving it a test run, it's attached to the bug) | 13:31 |
Ng | I'd also like to figure out if it's possible to get horizontal scrolling going, but that's way less important | 13:36 |
Ng | oh, I just did that :) | 13:37 |
tjaalton | Ng: got an X300 to play with.. should it work with hardy?-) | 13:58 |
Ng | tjaalton: all the important stuff other than sound will work | 13:59 |
Ng | that just needs a newer alsa to get the driver | 14:00 |
Ng | should all work with intrepid, apart from the scrollwheel emulation ;D | 14:00 |
tjaalton | Ng: heh, ok.. I'll try with hardy first | 14:00 |
mvo | Ng: haha, I spelt it wrong? oh well | 15:27 |
Ng | ZAxsis | 15:27 |
Ng | tjaalton: are you borrowing it? why not chuck intrepid straight on? :) | 15:27 |
tjaalton | Ng: I'm installing it for a colleague, and yes intrepid it is since hardy refused to netboot | 15:28 |
tjaalton | or, failed to install the kernel | 15:28 |
Ng | ah | 15:29 |
Ng | steal it ;) | 15:29 |
tjaalton | it's tempting ;) | 15:30 |
crevette | X300 ... hmmm | 15:30 |
tjaalton | could swap my X61 for it | 15:30 |
crevette | X61 is not bad | 15:30 |
tjaalton | "look, it's smaller" | 15:30 |
crevette | :) | 15:30 |
tjaalton | no it's not | 15:30 |
crevette | I'm happy with y T6A but I should have wait some week to have a X400 with a powerful cpu | 15:30 |
crevette | T61 | 15:31 |
crevette | damn technical evolution | 15:31 |
tjaalton | got my X61 last December, and a month after they released X300.. | 15:32 |
Ng | crevette: the X61 and the X300 seem to have very similar internals | 15:37 |
Ng | the 301 has a much newer chipset | 15:37 |
crevette | I didn't know there is a 301 | 15:37 |
Ng | hmm, or not, I can't remember exactly. i think I remember being surprised that it wasn't ICH10, which would make it ICH9 (where the 300 is ICH8) | 15:37 |
Ng | crevette: physically the same apart from DisplayPort, then a newer chipset and a faster CPU, with the option of a 128GB SSD instead of just 64GB | 15:38 |
crevette | Ng, DisplayPort is the external display port so is it DVI now ? | 16:16 |
Ng | displayport is supposed to replace DVI, I believe | 16:17 |
crevette | ah DisplayPort is a type of port | 16:18 |
Ng | yeah | 16:18 |
* crevette is not into hardware :) | 16:18 | |
Ng | it's not compatible with DVI, although DVI signals can be passed over it, hardware support willing | 16:18 |
Ng | (the X300 just has VGA) | 16:19 |
* crevette reads wikipedia article | 16:19 | |
crevette | yeah my T61 too :/ | 16:19 |
Ng | having VGA is no bad thing, every projector in the world can talk to it, vs almost none talking displayport :) | 16:23 |
bryce | morning | 16:52 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!