[02:31] <wgrant> bryce: Any idea what's going on in http://paste.ubuntu.com/59086/? It doesn't look very pleasant...
[02:32] <wgrant> It only happens on that one device, and the property is "Synaptics Edge Motion Speed"
[02:35]  * bryce looks
[02:36] <bryce> ew xcb calls
[02:36] <wgrant> The only atom that it manages to ask the server about is this:
[02:36] <wgrant> 000:<:000d:  8: Request(17): GetAtomName atom=0x10f(unrecognized atom)
[02:36] <wgrant> 000:>:0x000d:60: Reply to GetAtomName:  name='Synaptics Button Scrolling\000\000'
[02:36] <wgrant> But it crashes here:
[02:36] <wgrant> #7  0x0000000000401e6b in print_property (dpy=0x2b0aaac10a00, dev=0x60ef30, property=269) at ../../src/property.c:47
[02:36] <wgrant> 47	    name = XGetAtomName(dpy, property);
[02:37] <wgrant> And atom 269 is Synaptics Edge Motion Speed
[02:37] <bryce> usually if your stack trace shows _XReply, then it indicates there's some sort of error in the client/server interaction
[02:37] <bryce> which could be due to client incorrectly making its server call, or could be an error inside the server
[02:37] <bryce> _typically_ the former more than the latter
[02:38] <wgrant> I'll downgrade the server.
[02:38] <wgrant> As a client stack hasn't changed at all from the working version still fails.
[02:38] <bryce> wgrant: issues with _XReply I've found awfully hard to debug
[02:39] <wgrant> Hmm.
[02:39] <bryce> it's hard to trace into the server and see what's going wrong
[02:39] <wgrant> But no, I'm still running an oldish server.
[02:39] <wgrant> xtrace doesn't show the request ever getting to the server... hmmmm.
[02:40] <bryce> I notice you have null pointers for vector and count... are NULLs valid for those?
[02:40] <wgrant> I've got no idea about xcb.
[02:40] <bryce> wgrant, you could also set up xmond to view the protocol communication between the client and server
[02:41] <wgrant> bryce: I've used xtrace to do that.
[02:41] <bryce> ah good
[02:41] <bryce> well the xcb bits are just the libx11 layer
[02:41] <bryce> whatever the bug is, it's not likely to be in that (possible, but not likely)
[02:43] <bryce> *most* of the issues like this one ended up being the client making an incorrect X protocol call, or passing parameters that weren't valid
[02:44] <wgrant> http://paste.ubuntu.com/59094/
[02:45] <wgrant> Those are the calls that it makes before it dies...
[02:45] <wgrant> 146,36 is XListDeviceProperties.
[02:46] <wgrant> On other devices it then issues a GetDeviceProperty.
[02:47] <bryce> you mentioned you were using an old xserver... what version precisely?
[02:48] <wgrant> 2:1.5.1-1ubuntu3
[02:48] <bryce> so old only in relative terms ;-)
[02:48] <wgrant> I'm upgrading now.
[02:48] <bryce> shouldn't matter
[02:49] <bryce> the 1ubuntu4 change was only for -psb
[02:49] <wgrant> Hmm, maybe not, 400MB is a bit too much of an upgrade for peak.
[02:49] <bryce> er wait
[02:49] <bryce> sorry was thinking 1.5.2-1ubuntu3
[02:49] <wgrant> There's been a merge since then...
[02:49] <wgrant> Yes.
[02:50] <bryce> right.  worth the upgrade.
[02:50] <wgrant> I'm upgrading just X bits.
[02:50] <bryce> glancing at the changelog, nothing that looks likely to be the cause of this
[02:50] <bryce> but who knows
[02:51] <wgrant> The strangest bit is that it worked with this xserver last week...
[02:52] <bryce> hey while you're waiting, there was an email on u-d-d that you'd be best to answer - https://lists.ubuntu.com/archives/ubuntu-devel-discuss/2008-October/005931.html
[02:53] <bryce> I can pass a reply along if you don't want to sub to the list (it's kind of a noisy list)
[02:59] <wgrant> bryce: I'm subscribed, but don't normally keep up to date.
[02:59] <wgrant> I bet he added the options to the Synaptics part of the fdi file rather than the ALPS one.
[02:59] <wgrant> I'll reply shortly.
[03:00] <wgrant> Thanks for pointing it out.
[03:01] <bryce> great thanks
[03:01] <wgrant> Still segfaults.
[03:01] <wgrant> It must be a server-side change, I think.
[03:02] <wgrant> The last updates in the amd64 chroot in which I'm testing were on the 11th, and I've only upgraded libxi6 and xinput since...
[03:03] <wgrant> In fact, the last updates were the day I created the chroot, which means that same set of packages was working fine after that Atom size issue was fixed.
[03:03] <wgrant> So it's not the client.
[03:04] <wgrant> Ah, bug #283921
[03:06] <bryce> ew
[03:06] <bryce> (private bugs ftl)
[03:07] <wgrant> valgrind reports the same sort of stuff and dies similarly on a request that works, however.
[03:07] <bryce> wgrant: I de-privated it
[03:09] <wgrant> Those bad writes aren't healthy, even if they don't kill it in most cases...
[03:11] <wgrant> And those don't happen on i386... so it's something like the same problem.
[03:13] <wgrant> They're always of size 8... very suspicious...
[03:15]  * wgrant will poke it after lunch.
[03:32] <wgrant> You know, I bet it was http://cgit.freedesktop.org/xorg/proto/inputproto/diff/XIproto.h?id=36c8a6f3faf56a8f8ca31455812c9132b379b1b3 (which we don't have)
[03:36] <bryce> mm
[03:42] <wgrant> AHA
[03:43] <wgrant> (intrepid-amd64)root@fisto:/home/fujitsu/libxi-1.1.3# xinput list-props 3
[03:43] <wgrant> Device 'SynPS/2 Synaptics TouchPad': Device Enabled:		1 Synaptics Edges:		1632, 0, 5312, 0 Synaptics Finger:		25, 0, 30
[03:43] <wgrant> It works.
[03:43] <bryce> \o/
[03:43] <wgrant> And the moral of the story is that defines suck.
[03:46] <bryce> the hardest bugs always seem to end up being 1-line fixes
[03:46] <wgrant> Now to work out what needs to be rebuilt...
[03:46] <wgrant> libxi, at least.
[04:35] <wgrant> bryce: It appears that only 5 uploads can have been affected by that bug.
[04:35] <bryce> not bad
[04:36] <wgrant> libxi, xorg-server, gnome-control-center, xf86-input-evtouch and xserver-xorg-input-evdev.
[04:36] <wgrant> Nothing else uploaded since includes XIproto.h or anything else that includes it. 
[06:47] <tjaalton> huh, I thought we had that already
[06:48] <tjaalton> since wasn't that what made properties work on amd64?
[06:48] <tjaalton> or, not fail for some
[06:54] <tjaalton> oh sorry, it's undef, not the #define Atom CARD32 -change
[06:57] <wgrant> tjaalton: Yes, so it's causing the opposite problem now.
[06:58] <wgrant> I'm stupid, basically. I initially fixed it by replacing the Atoms in XIproto.h with CARD32s. I tested that thoroughly. But then upstream fixed it with #define Atom CARD32, so I did too. I then only did basic property getting/setting testing, and didn't try to list them again...
[06:59] <tjaalton> don't worry, it's trivial to fix now
[07:00] <wgrant> It's trivial to fix the header, sure.
[07:00] <wgrant> But I had to script a grep through 300 packages to set what was going to need to be rebuilt.
[07:01] <wgrant> It turns out to be these: https://edge.launchpad.net/~wgrant/+archive
[07:03] <tjaalton> hum, ok
[07:22] <bryce> hey tjaalton
[07:22] <bryce> I've got a patch for 261977, let me post t
[07:24] <bryce> http://launchpadlibrarian.net/18655616/xorg-server_1.5.2-1ubuntu6.debdiff
[07:30] <wgrant> bryce: I like your regression potential mitigation strategy for -synaptics.
[07:30] <wgrant> You left the patch out of debian/patches/series...
[07:30] <bryce> doh!
[07:31] <wgrant> It'll need another upload once x11proto-input is fixed, anyway.
[07:33] <bryce> I'll prep a debdiff to fix, sorry about that
[07:33] <wgrant> Thanks.
[07:35] <bryce> ping me when you want the timing is right to upload it and I'll  put it in
[07:35] <wgrant> I'll need to convince somebody to let x11proto-input in first.
[07:37] <bryce> lp #?
[07:38] <tjaalton> bryce: hey, cool, I'll read it through
[07:38] <wgrant> Bug #283921 is the main symptom, but there's no debdiff attached. I'll generate one.
[07:39] <bryce> ok
[07:39] <wgrant> LP's package diffing is braindead. Why won't it diff from primary as well as the old version in my PPA!?
[07:41] <tjaalton> bryce: looks good. for completeness sake it could check the system like later in the file (can't check the function now), but since we don't support !*x86 that doesn't matter much
[07:43] <tjaalton> bah, cheese can't find my camera, daughter disappointed
[07:46] <bryce> tjaalton: can do - it'd be helpful though first to just verify it does solve the bug
[07:46] <tjaalton> bryce: sure
[07:51] <wgrant> bryce: See the debdiff on bug #283921. All functionality tested with i386/amd64 clients and i386 server.
[07:57] <tjaalton> hmm, looks like the gstreamer default video input device is wrong
[07:57] <tjaalton> should be v4l2
[08:03] <tjaalton> strange, mouse scroll changes the desktop even when the cursor is on some gtk-based progs
[08:05] <wgrant> tjaalton: I noticed that, and presumed it was a feature.
[08:05] <wgrant> Albeit an annoying one.
[08:05] <wgrant> And it's only on some widgets, AFAICT.
[08:06] <tjaalton> likely so
[08:07] <tjaalton> but anywhere on the cheese window, for instance
[08:21] <bryce> wgrant: with the x11proto-input debdiff, is that a complete fix for 283921 or a partial fix?
[08:22] <wgrant> bryce: It will several no-change rebuilds afterwards.
[08:22] <wgrant> +require
[08:24] <bryce> ok
[08:24] <wgrant> The changes are all in my PPA, but the rest are just rebuilds.
[08:25] <elmargol> any ideas what temperatures are ok for a GeForce Go 7900 GS?
[08:25] <wgrant> None - it's an nvidia card, so it is implicitly unOK.
[08:26] <elmargol> I agree nvidia sucks bigtime
[08:30] <elmargol> I'm nor sure what my problem is a) The fan starts to late or b) the memory gets unstable if the card gets hot
[08:30] <elmargol> or c) the powermixer has a bug an clocks the memory wrong
[08:33] <bryce> wgrant: uploaded x11proto-input
[08:34] <wgrant> bryce: Great, thanks.
[08:34] <wgrant> I really need to get myself some amd64 hardware at some point.
[08:34] <elmargol> uhh there is a dell laptop controll daemon :D
[08:35] <wgrant> For controlling what?
[08:35] <elmargol> the fan
[08:35] <wgrant> Ah.
[09:08] <wgrant> bryce: I suppose you don't want debdiffs for the rebuilds?
[09:10] <bryce> wgrant: probably no need
[09:12] <bryce> tjaalton: philip langdale of VMWare fixed -vmmouse for input-hotplug
[09:13] <bryce> tjaalton: bugs 248521 and 285305
[09:14] <bryce> for 248521, the new upstream version just had one patch so I've packaged it and uploaded it
[09:14] <bryce> but 285305 is to add a hal fdi file.  If you have time could you look at that one?
[09:17] <bryce> (or if we need to bug pitti about it, let me know)
[09:34] <tjaalton> bryce: ok, read the bug. I don't know why it would need a callout script, just setting the driver in the fdi should work
[09:34] <tjaalton> I'll add that to the bug
[09:34] <bryce> ok
[09:35] <tjaalton> ah, running vmmouse_detect is the reason
[09:36] <tjaalton> he also mailed xorg@ about it
[09:39]  * wgrant saw that.
[09:39] <wgrant> Wasn't he saying that it only worked for USB?
[09:40] <tjaalton> with evdev, this is needed for PS/2 emulation
[09:47] <tjaalton> bryce: so, including the fdi file and callout-script to the package would fix both issues
[09:48] <tjaalton> the script could go in /usr/lib/hal like debian-setup-keyboard
[09:52] <bryce> sounds good
[09:58] <bryce> wgrant: your builds should all be in now, assuming I did them right.  Can you doublecheck later on that they all got through okay, and let me know otherwise?
[09:58] <bryce> ok, I'm too tired.  --> bed.  night
[09:59] <wgrant> bryce: Thanks a lot! Night.
[09:59] <tjaalton> night!
[10:24] <elmargol> tseliot: did you see my comment?
[10:34] <tseliot> elmargol: yes, I did
[10:35] <tseliot> but it's not something I can fix
[10:35] <tseliot> only NVIDIA can
[10:39] <elmargol> tseliot: well nvidia does not care
[10:39] <elmargol> tseliot: do you know a tool to control the gpu fan?
[10:43] <tseliot> elmargol: try nvclock
[11:12]  * munckfish rolls up sleeves to work on bug 281610 (and cries)
[11:18] <tjaalton> how do you install a ps3 nowadays?
[11:19] <tjaalton> I've got one to try
[11:25] <tjaalton> I'll ask upstream
[11:29] <Q-FUNK> howdy!
[11:30] <Q-FUNK> how soon can we get around uploading the new geode to hardy-updates and close bug #255991 for good?
[11:31] <tjaalton> is it in -proposed?
[11:32] <Q-FUNK> it
[11:32] <Q-FUNK> it's never been uploaded
[11:32] <Q-FUNK> bryce had submited a debdiff and pitti approved it
[11:32] <Q-FUNK> an actual package just needs to be produced and uploaded
[11:32] <Q-FUNK> based on that diff
[11:36] <tjaalton> sigh
[11:36] <tjaalton> no patch system
[11:36] <Q-FUNK> hm?
[11:36] <tjaalton> should I apply the diff directly??
[11:37] <Q-FUNK> I think that bryce didn't go as far as slicing his diff into a series of debian/patches
[11:37] <tjaalton> no he didn't, but since it's your package you are welcome to do so :)
[11:37] <Q-FUNK> but it would probably work to just throw the whole diff into debian/patches/ and enable the include for simple-patch.mk
[11:38] <Q-FUNK> ok
[11:38] <Q-FUNK> lemme do that now, then we can uplaod
[11:38] <tjaalton> don't forget the changelog, since you know the details
[11:41] <Q-FUNK> yup
[11:41] <Q-FUNK> can I strip the extraneous debian/rules content too?
[11:41] <Q-FUNK> there's duplicated lines that ended up there for no reason
[11:41] <tjaalton> go for it
[11:44] <wgrant> Grumble.
[11:44] <wgrant> There is something still not entirely 64-bit safe there.
[11:45] <wgrant> But it doesn't affect xinput, only syndaemon, so it's probably some header order issue...
[11:50] <munckfish> tjaalton: you can install Ubuntu on PS3 from a daily cd takes about an 2 hours all in
[11:50] <munckfish> http://cdimage.ubuntu.com/ports/daily/current/
[11:50] <munckfish> Yesterdays was the first time for months I've been able to get all the way through an install
[11:51] <munckfish> Unfortunately once installed the nightmare only begins
[11:51] <munckfish> since last week everything changed
[11:51] <munckfish> :D
[11:53] <munckfish> tjaalton: I'm thinking of trying to do a git bisect between 1.5.1-1ubuntu2 and 1.5.1-1ubuntu3, do think that's a good first step ?
[11:54] <munckfish> Or is there other info we could collect first that may help us guess the problem?
[11:54] <munckfish> s/problem/problems/
[11:55] <munckfish> Seems like there's two issues here: graphics and input, not sure if they're separate problems.
[12:17] <tjaalton> I don't understand why it doesn't use evdev for the keyboard even with 1.5.1-1ubuntu2
[12:18] <munckfish> tjaalton: it does use evdev
[12:18] <tjaalton> a lesson for all; don't teach your daughter to play flash-games with an unstable platform
[12:19] <munckfish> I just had to downgrade the xserver-xorg-input-evdev package to match with the 1.5.1-1ubuntu2 of the server
[12:19] <tjaalton> no sound after suspend -> flash-games don't work -> disaster
[12:19] <munckfish> ouch!
[12:19] <munckfish> that's with Flash 10 on Intrepid?
[12:19] <tjaalton> munckfish: well, I'll try to reproduce it
[12:19] <tjaalton> yes
[12:19] <munckfish> nightmare I was hoping flash was more stable in v10
[12:19] <munckfish> :(
[12:19] <munckfish> ho hum
[12:19] <tjaalton> but the sound issue is some alsa trouble
[12:20] <tjaalton> just that the flash game doesn't progress if there's no sound
[12:20] <munckfish> tjaalton: if you decide to install to your PS3, you'll need a couple of workarounds to get to state where you can mess around with it
[12:21] <tjaalton> ok, I'm all ears
[12:21] <Q-FUNK> tjaalton: debdiff and build log attached
[12:21] <munckfish> bug 285231 - cjwatson fixed it last night I'm not sure it's on last nights daily build
[12:21] <munckfish> basically if you can't boot that'll be why
[12:22] <munckfish> next to stop usplash from hanging the system on boot (cries weeps)
[12:22] <munckfish> you'll need to drop to a shell in kboot, remount the target partition and edit the bootloader config
[12:22] <munckfish> best ping me when you get to that stage
[12:23] <munckfish> just need to add in a "Recovery" boot config which boots with kernel parameter "single"
[12:23] <wgrant> AH
[12:23] <munckfish> then you can at least get to a root shell and start messing with X
[12:23] <wgrant> Tracked down the synaptics 64-bit unsafeness issue.
[12:23] <munckfish> all good fun!
[12:24] <tjaalton> munckfish: heh, sounds like a lot of fun
[12:24] <munckfish> Yeah and to think it was all working so nicely just a week ago
[12:25] <munckfish> tjaalton: hmmm git bisect isn't going well
[12:25] <munckfish> :(
[12:25] <tjaalton> Q-FUNK: ok, I'll look at it later
[12:25] <Q-FUNK> tjaalton: thanks!
[12:25] <munckfish> it's landed me in commit 4e6cbd323854709ae00c44108c93ab6596151de2
[12:26] <munckfish> with no debian/ dir, so I grabbed the debian dir from an earlier commit
[12:26] <munckfish> but build is failing :(
[12:31] <munckfish> dix/devices.c:99 error: 'XI_PROP_ENABLED' undeclared (first use ...
[12:32] <tjaalton> you must have the latest properties-patch for the server
[12:32] <jcristau> i'd just try to revert the 3 xkb commits between 1.5.1 and 1.5.2..
[12:32] <tjaalton> so git-bisecting the ubuntu-branch probably doesn't work
[12:32] <tjaalton> yeah, that could be it
[12:41] <munckfish> right yes this is a nightmare with or without the debian dir, other deps are causing a pain the arse now
[12:41] <munckfish> ok so I'll try to selectively revert different patches
[12:42] <munckfish> jcristau: so that will deal with the keyboard issue, but I think need to deal with the no graphics at all issue first and that gets introduced in 1.5.3 as well I believe
[12:42] <munckfish> duh
[12:42] <munckfish> not 1.5.3
[12:43] <munckfish> 1.5.1-1ububuntu3
[12:43]  * munckfish slaps forehead
[13:27] <wgrant> jcristau: That conditional redefinition of Atom depending on _XSERVER64's definedness just bit me. It's rather inconvenient!
[13:27] <wgrant> There are some really strange things lurking in the depths of the X headers...
[13:43] <tjaalton> Q-FUNK: uploaded
[13:51] <jcristau> wgrant: yeah, it's a mess
[13:53] <wgrant> jcristau: -synaptics' config.h is included in both its driver and clients, and it includes xorg-server.h. Chaos ensues.
[13:54] <jcristau> wgrant: hah. should probably have a separate config header for the clients, or something
[13:56] <wgrant> For now I've worked around it in my PPA by wrapping the #include xorg-server.h in #ifndef SYNAPTICS_CLIENT_ONLY and defining those in the clients... but that's because I'm no good with autotools and can't work out how to do it properly.
[13:56] <jcristau> yeah, that should work too
[14:08] <Q-FUNK> tjaalton: thanks!   can you post on the bug to show the upload has been done and close it?
[14:10] <tjaalton> sure
[14:11] <tjaalton> but it's only in proposed, so it's fix committed
[14:11] <Q-FUNK> ah yes
[14:19] <tjaalton> oops, bryce uploaded -evtouch when he meant to upload evdev
[14:20] <tjaalton> and a wrong version at that
[14:20] <tjaalton> -4 when we had -3ubuntu12 :)
[14:20] <tjaalton> (-4build1 but still)
[14:20] <tjaalton> well, ogra will fix it :)
[14:21] <jcristau> shouldn't all of them have been XubuntuY instead of Xbuild1?
[14:21] <wgrant> tjaalton: -evtouch was needed as well as -evdev.
[14:21] <tjaalton> wgrant: how so?
[14:21] <tjaalton> anyway, I think all the changes were dropped
[14:22] <tjaalton> hmm, xorg-server 1.5.2-2build1
[14:22] <wgrant> tjaalton: It built against the broken XIproto.h.
[14:22] <tjaalton> I think xorg-server is busted as well
[14:23] <wgrant> libxi too.
[14:23] <wgrant> Just the versioning is wrong, but it's still a bit messy...
[14:23] <Kano> hi, will u get new nv legacy drivers or not for release?
[14:24] <tjaalton> Kano: there are no new legacy versions
[14:25] <Kano> and what do you tell your users then ;)
[14:25] <tjaalton> force them to nv
[14:26] <Kano> nv does not work correctly, at least not for my nvidia 8800 gts 512
[14:26] <Kano> when will the vesa override work?
[14:26] <tjaalton> patch exists
[14:26] <tjaalton> sorry, doesn't
[14:29] <tjaalton> Kano: is there a bug about it, do you know?
[14:30] <Kano> well not in lauchpad i guess, i told aaronp,but he said he has got no time for it
[14:30] <tjaalton> I mean the xforcevesa not working
[14:31] <tjaalton> wgrant: yep, the diff reveals that only the changelogs were modified
[14:31] <Kano> tjaalton: if you did not add one, i only told it here
[14:32] <wgrant> tjaalton: I checked them on LP and came to that same conclusion. Good.
[14:32] <tjaalton> Kano: it might be filed already
[14:33] <tjaalton> wgrant: now the question is, how to fix them :)
[14:33] <wgrant> tjaalton: We can't until we merge from Debian.
[14:34] <tjaalton> doesn't look like anyone has filed a bug about xforcevesa b0rkedness
[14:34] <Kano> there are 20 bugs with xforcevesa...
[14:34] <tjaalton> please tell me
[14:35] <tjaalton> searching for it gave no results
[14:35]  * wgrant wins.
[14:35] <wgrant> I now have confirmation that syndaemon works without SHMConfig, even on amd64. Yay!
[14:36] <tjaalton> hehe, bug 283979
[14:36] <wgrant> What a summary.
[14:37] <tjaalton> yep
[15:28] <mnemo> if I have a BLAH.patch creating using git-format-patch, how can I apply that to a "apt-get source" tree? git-am won't work but I want to try this patch on the ubuntu version of this app
[15:28] <jcristau> patch -p1
[15:32] <tjaalton> Kano: so you have those 20 bug reports, please share them
[15:33] <Kano> well i used the first search option
[15:33] <Kano> but there is no specific bug against xserver-xorg with xforcevesa
[15:35] <tjaalton> I get two bugs with 'xforcevesa'
[15:35] <Kano> you can search on first page too
[15:36] <tjaalton> what. first. page?
[15:36] <Kano> https://launchpad.net/+search?field.text=xforcevesa
[15:37] <tjaalton> right, so not searching the _bugs_
[15:37] <tjaalton> and it also lists closed bugs
[15:37] <tjaalton> but yes, no current ones filed
[15:38] <tjaalton> now floorball ->
[15:48] <mnemo> jcristau: thanks
[18:05] <bryce_> bleah, I need to not do packaging until 2am
[18:05] <bryce_> tjaalton: anything we can do to fix the version numbers?  I clearly wasn't thinking
[18:07] <tjaalton> bryce_: hehe :)
[18:12] <tjaalton> bryce_: well, maybe ask jcristau to release the versions, and we could merge them then. libxi has only received some packaging fixes which should be trivial, and xorg-server has one commit by lool (included in our package too)
[18:19] <mnemo> bryce_: my G45 (desktop) machine current freezes directly after login and I now finally have some clues as to why
[18:19] <mnemo> in fact I know a specific patch that makes EXA work for me
[18:19] <mnemo> I've tried to apply this patch to ubuntu's intel-agp.ko and ti worked
[18:19] <mnemo> the specific change that allows G45 to boot is this one:
[18:19] <mnemo> http://git.kernel.org/?p=linux/kernel/git/anholt/drm-intel.git;a=blobdiff;f=drivers/char/agp/intel-agp.c;h=f1fe7490131468b40c5817e232442026d9f8ff0d;hp=016fdf0623a4ce7b5f849efc8ab099a2c5026497;hb=2052746fc8397130c120f0194a89938b0b62b6cb;hpb=b34c87315b1a2822111fc8ef744ef504f9be2f85
[18:19] <tjaalton> so it's a kernel bug?
[18:19] <tjaalton> right
[18:19] <tjaalton> file a bug on the kernel then
[18:20] <bryce_> mnemo: cool, good work tracking that down
[18:20] <mnemo> that change is VERY small and it's also "if its G45 then do blah" so maybe we can get it in?
[18:20] <tjaalton> ask the kernel people
[18:20] <mnemo> I will open a bug and nominate for intrepid and we'll see what happens
[18:20] <bryce_> yeah, and also make sure to grab a kernel guy on irc, because I don't think there's much time left to get it in
[18:20] <mnemo> true
[18:20] <bryce_> nominating probably won't be sufficient at this point
[18:20] <mnemo> ok
[18:21] <bryce_> also (esp if you can't get a kernel guy), let the release managers know about it
[18:21] <munckfish> hi still trying to get my head around bug 281610. Could someone gives me some tips as to where input device config actually gets into the xserver if not specified on the cli or xorg.conf?
[18:21] <bryce_> like cjwatson, pitti, slangasek, or hobsee
[18:22] <tjaalton> filed bug 285562 about the xforcevesa issue
[18:22] <munckfish> I've been running gdb on xf86Config.c in checkCoreInputDevices but it just doesn't find any anywhere :(
[18:23] <tjaalton> munckfish: have you tried reverting the three xkb related commits?
[18:24] <munckfish> tjaalton: not yet, my build failed so I decided to grab the -dbg package and see if I could get any hints from that. I can revert those next.
[18:24] <tjaalton> munckfish: please do, probably easier to get results that way
[18:25] <munckfish> tjaalton: ok thx
[21:31] <wgrant> tseliot: It seems that the dist-upgrader is commenting out the inputdevice sections, but not commenting out the lines in serverlayout. That causes problems.
[21:33] <tseliot> wgrant: mvo wrote that script. From what I remember it should remove the entries in the serverlayout section too
[21:33] <tseliot> I can talk to mvo again
[21:34] <tseliot> wgrant: he didn't use my script that uses xkit because xkit removes comments (from sections) and doesn't comment things out
[21:36] <wgrant> tseliot: Ah. I've seen at least two fairly sane reports of it on ubuntuforums.
[21:38] <tseliot> wgrant: having the links to such reports would help
[21:38] <wgrant> I'll see if I can find them again.
[21:39] <tseliot> thanks
[21:42] <wgrant> tseliot: Here's one: http://ubuntuforums.org/showthread.php?t=951659
[21:43] <tseliot> ok
[21:44] <wgrant> tjaalton: Shouldn't bug #285562 be High with an Intrepid task? Critical without an Intrepid task doesn't imply release-criticality.
[22:05] <tjaalton> wgrant: yep, milestone alone isn't enough I guess
[22:05] <wgrant> tjaalton: Yep. There was an email about it a while ago.
[22:05] <wgrant> Any idea how I make my joystick not be a mouse?
[22:05] <wgrant> Even if I add it to the fdi file, it still breaks things.
[22:06] <tjaalton> hrm
[22:06] <tjaalton> how exactly does it break things? I keep hearing that a lot but I'm missing something
[22:07] <wgrant> It it seems to scroll up. And up. And up.
[22:07] <wgrant> I would xinput query-state it, but it's not easy!
[22:07] <tjaalton> and there is no way to not let evdev grab them, at least no practical way (since they tend to have a input.keyboard capability)
[22:09] <wgrant> What are other distros doing?
[22:09] <tjaalton> launchpad is failing on me
[22:09] <tjaalton> they don't care, or use input-hotplug?-)
[22:10] <tjaalton> I'll check fedora
[22:10] <wgrant> They don't use input-hotplug, you mean?
[22:10] <tjaalton> haha
[22:10] <tjaalton> they don't have -joystick
[22:10] <tjaalton> I saw that earlier but forgot
[22:10] <tjaalton> wgrant: right, not many distros do yet. fedora is one but I'm not aware of others
[22:10] <tjaalton> besides us
[22:12] <wgrant> Hmmm.
[22:12] <tjaalton> duh, forgot that evdev refuses to use my joysticks
[22:12] <wgrant> How!?
[22:13] <tjaalton> (WW) Logitech Logitech RumblePad 2 USB: Don't know how to use device
[22:13] <tjaalton> (II) UnloadModule: "evdev"
[22:13] <tjaalton> (EE) PreInit returned NULL for "Logitech Logitech RumblePad 2 USB"
[22:13] <tjaalton> (EE) config/hal: NewInputDeviceRequest failed
[22:13] <wgrant> I wonder if we should try to harvest joystick names from xorg.conf.
[22:13] <wgrant> tjaalton: I fixed that by adding it to the -joystick fdi file.
[22:13] <tjaalton> yes, I don't have it installed atm
[22:13] <tjaalton> rumblepad should be there already since I added it
[22:13] <tjaalton> let's try with it
[22:14] <wgrant>       <match key="info.product" contains="RumblePad">
[22:14] <wgrant> Yep
[22:15] <tjaalton> upstream master supports properties, but I sort of didn't care about it ;)
[22:15] <wgrant> What do the properties allow?
[22:15] <wgrant> This whole stack should be fairly nice for Jaunty, I must say.
[22:15] <tjaalton> "Set type_name to XI_JOYSTICK. Let's not claim we are XI_MOUSE."
[22:15] <tjaalton> I wonder if that would help
[22:16] <tjaalton> should be yes
[22:16] <wgrant> I tried that, but it's not actually defined anywhere.
[22:16] <wgrant> And I don't think much uses the type at the moment.
[22:17] <wgrant> synaptics only just started reporting itself as XI_TOUCHPAD within the last few months.
[22:17] <tjaalton> ok, so with joystick it moves the mouse now
[22:18] <wgrant> Hmmm.
[22:18] <wgrant> Must be evdev being stupid.
[22:18] <tjaalton> it's stupid only with evdev?
[22:19] <wgrant> joystick hasn't traditionally performed mouse actions, so it must be evdev.
[22:19] <tjaalton> not according to the log
[22:19] <tjaalton> but:
[22:19] <tjaalton> (II) XINPUT: Adding extended input device "Logitech Logitech RumblePad 2 USB" (type: MOUSE)
[22:21] <wgrant> I think that's fine.
[22:21] <tjaalton> I mean that there's no evidence that evdev is handling it
[22:22] <wgrant> Oh, indeed. There was for the failure when it wasn't in the fdi file, but you're right...
[22:22] <wgrant> (**) Microsoft SideWinder Precision 2 Joystick: always reports core events
[22:22] <wgrant> Shouldn't it not be doing that?
[22:23] <tjaalton> there was a time when I was able to crash the server with this thing, and then evdev was handling it
[22:23] <tjaalton> (**) Logitech Logitech RumblePad 2 USB: always reports core events
[22:24] <wgrant> What is the internal difference between a joystick and mouse?
[22:26] <tjaalton> that's a good question, I don't know
[22:26] <tjaalton> but there was a discussion about the patches a while ago on xorg@
[22:35] <wgrant> tjaalton: I'm eating now, but try setting StartMouseEnabled to false in the fdi file.
[22:35] <wgrant> It's exposed through properties now.
[22:39] <tjaalton> wgrant: yep, that did the trick
[22:39] <wgrant> tjaalton: I wonder why it's changed.
[22:41] <wgrant> The option was added in January, but the default was TRUE before AFAICT.
[22:44] <tjaalton> hm
[22:46] <wgrant> I wish I'd thought to try a joystick earlier
[22:53] <wgrant> I wonder which 64-bit-unsafeness bugs I'll find today.
[23:00] <tjaalton> heh
[23:02] <wgrant> One of the three was my fault, but still...
[23:17] <wgrant> This joystick's axes are behaving decidedly mouselike
[23:17] <wgrant> ie. only valuators 0 and 1 have values, only button 1, 2, 3 work.
[23:17] <wgrant> The working valuators are also relative.
[23:18] <wgrant> And affected by more than one real axis.
[23:18] <wgrant> How odd.
[23:25] <wgrant> Damn.
[23:26] <wgrant> X didn't like me dropping a copy of -joystick master over the top of the old driver...
[23:26] <wgrant> It crashed and hardlocked before I released enter!
[23:26] <tjaalton> that's to be expected
[23:28] <tjaalton> don't remember anymore how to prevent that from happening
[23:28] <tjaalton> some option for cp or so
[23:33] <wgrant> It is behaving exactly like a mouse now.
[23:33] <wgrant> I can perform appropriate clicks with the first three buttons, drag, etc.
[23:34] <wgrant> Oh
[23:34] <wgrant> The throttle scrolls, too.
[23:44] <jcristau> tjaalton: --remove-destination
[23:46] <wgrant> jcristau: Thanks.
[23:47] <jcristau> otherwise you're truncating and modifying the file the server has mapped, which doesn't work too well :)
[23:48] <wgrant> Indeed.