[12:31] <ubotu> New bug: #151426 in xorg (main) "Nvidia flickers since kernel 2.6.22-13 and -14" [Undecided,New]  https://launchpad.net/bugs/151426
[01:21] <ubotu> New bug: #151016 in linux-source-2.6.22 (main) "New in 2.6.22-13: No video after resume from suspend on MacBook (dup-of: 127101)" [Undecided,Invalid]  https://launchpad.net/bugs/151016
[03:42] <bryce> heya pcjc2
[03:42] <pcjc2> hi
[03:42] <pcjc2> where were you adding a usleep?
[03:43] <bryce> --- xserver-xorg-video-intel-2.1.1.orig/src/i830_driver.c
[03:43] <bryce> +++ xserver-xorg-video-intel-2.1.1/src/i830_driver.c
[03:43] <bryce> @@ -1957,6 +1957,7 @@
[03:43] <bryce>     OUTREG(VCLK_POST_DIV, pI830->saveVCLK_POST_DIV);
[03:43] <bryce>     for(i = 0; i < 256; i++) {
[03:43] <bryce> +      usleep(20);
[03:43] <bryce>        OUTREG(PALETTE_A + (i << 2), pI830->savePaletteA[i] );
[03:43] <bryce>     }
[03:43] <pcjc2> ah - see if the register access is too fast for the HW?
[03:44] <pcjc2> the other point in the code where the pallette is programmed made the assertion (in comments) that the PLL / clock had to be running
[03:44] <pcjc2> I'm presuming this is the case after we're restored the clock register (s) above
[03:44] <bryce> mm could be
[03:44] <pcjc2> although I (vaguely) recall delays being used to wait whilst the PLLs stablise
[03:53] <bryce> hrm, running this under gdb isn't working very well
[03:53] <bryce> bbiab (dinner)
[03:54] <pcjc2> how comes?
[03:54] <pcjc2> not crashing with GDB?
[03:55] <pcjc2> or just not easy to trace through
[03:58] <pcjc2> It just occured to me.. the set_cursor_colors function sets the palette regs
[03:59] <pcjc2> and that is another place I've seen it crash (not that function specifically, but after reloading cursors)... I could never make it do it once I'd added a print statement to see what functions were being called... timing does look like a distinct possibility
[04:10] <bryce> when running under gdb when I switch to tty1 it doesn't switch; the background turns purple though, and the system behaves erratically after that, until reboot
[04:10] <pcjc2> gdb presumably affects timing
[04:11] <pcjc2> I have absolutely no idea why... but that ramdac thing "seems" to make a difference here
[04:11] <pcjc2> when I patched it the other way (forced ramdac to load with SWCursor), it said it was already built in
[04:11] <pcjc2> you're supposed to need ramdac to do HW Cursors (and they program the palette)
[04:12] <pcjc2> but I don't see the intel driver using it
 mjg59: adding the usleep() had no effect; still crashed with gray block lockup
 bryce: Ok, so probably not a timing issue. If you skip that block entirely, does it work?
[05:01] <pcjc2> for whatever reason, I'm not getting the crash
[05:01] <pcjc2> but it messes up the video overlay control registers
[05:01] <pcjc2> (and presumably has trampled a load of memory / IO somewhere)
[05:02] <pcjc2> before bug, I have XV frames flipping between buffers at 0x80004000 and 0x80104000
[05:03] <pcjc2> after, I've got 0x80000000 and 0x80100000
[05:03] <pcjc2> (sorry, those are the wrong way around)
[05:04] <bryce> after running the -intel with the aforementioned loop blocks removed, I've been unable to reproduce the bug
[05:04] <bryce> deb, debdiff, etc. at http://people.ubuntu.com/~bryce/Testing/bug127101/
[05:14] <pcjc2> nice
[05:14] <pcjc2> I'll test that here
[05:14] <pcjc2> See if I can get past running XV a 2nd time
[05:29] <pcjc2> no
[05:29] <pcjc2> sorry... something is still not right
[05:29] <pcjc2> just rebooting to make sure the video hardware is really clean
[05:33] <pcjc2> sorry... I've still got a crash here
[05:33] <pcjc2> am running compiz
[05:33] <pcjc2> (or whatever the fancy accelerated desktop is)
[05:33] <pcjc2> mplayer video.ogg
[05:33] <pcjc2> (works)
[05:33] <pcjc2> close mplayer
[05:34] <bryce> like I said on the bug, this fix is not intended to fix the Xv bug
[05:34] <pcjc2> switch to VT1 (nocrash with either your patch or my noramdac patch - goodness knows why), switch back
[05:34] <pcjc2> ok - is that a different known bug?
[05:34] <bryce> yes
[05:34] <pcjc2> doh
[05:35] <bryce> they both result in the gray blinking blocks
[05:35] <pcjc2> I've probably been seeing a slightly different crash
[05:35] <bryce> but one is 127101, the other is 141063
[05:35] <bryce> I don't know if yours is 141063, but it's not 127101
[05:35] <pcjc2> mine only occasionally resulted in blinking blocks - more usually a pallette corruption, followed by crash
[05:35] <pcjc2> did you try the ramdac thing?
[05:35] <bryce> it could be you're seeing multiple issues
[05:35] <pcjc2> I'm certain it must just be some fluke
[05:36] <bryce> no I stopped after my change seemed to resolve it
[05:36] <pcjc2> as it seemed to work.. but having chased the code paths which I commented out, shouldn't actually have any effect!
[05:36] <pcjc2> We need the Intel docs to resolve this _properly_
[05:36] <bryce> plus it's getting late and I'm going to be out the rest of the week at a customer site, so need to do some packing and spend a little time with the gf ;-)
[05:36] <pcjc2> Presumably there is a reason to restore the pallette regs
[05:37] <pcjc2> That sounds like a good point
[05:37] <pcjc2> its 4:37am here
[05:37] <bryce> yikes!
[05:37] <bryce> yeah I was up late last night myself on this
[05:37] <bryce> you'll probably have better ideas once you've had time to sleep on it
[05:37] <pcjc2> I've caught myself in a nasty insomniac cycle which will give me real problems for tomorrow's 9am lab demonstration!
[05:38] <pcjc2> (todays)
[05:38] <pcjc2> I'll chase the report on the Xv bug, see what I can learn
[05:38] <pcjc2> Goodnight!
[05:38] <bryce> night!
[08:05] <ubotu> New bug: #151484 in scim (main) "SCIM Disallows renaming of files (dup-of: 66104)" [Undecided,New]  https://launchpad.net/bugs/151484
[08:09] <tepsipakki> bryce: seems that the blackscreen-ati
[08:09] <tepsipakki> uh
[08:09] <tepsipakki> let me try again
[08:12] <bryce> hi tepsipakki
[08:12] <tepsipakki> seems that bug 132716 can be narrowed down pretty well
[08:12] <ubotu> Launchpad bug 132716 in xserver-xorg-video-ati "ATI Driver Gets Black Screen on Radeon 7500 Mobile (Regression)" [High,Confirmed]  https://launchpad.net/bugs/132716
[08:14] <tepsipakki> it should be safe to add a workaround at least for the inspiron
[08:18] <bryce> ok
[08:18] <bryce> I think we've got a potential fix for bug 127101 too
[08:18] <ubotu> Launchpad bug 127101 in xserver-xorg-video-intel "laptop hangs when switching video mode" [High,In progress]  https://launchpad.net/bugs/127101
[08:20] <bryce> tepsipakki: I've posted a debdiff to fix up a couple xresprobe issues, but I'm going to be out of town the rest of the week
[08:21] <bryce> it includes the vesa portion of tormod's patch
[08:21] <bryce> tepsipakki: I'd appreciate it if you could speak with slangasek about getting it in for final.
[08:22] <bryce> it's not the end of the world if it doesn't make it, though.
[08:23] <tepsipakki> bryce: I already uploaded the vesa-fix :)
[08:24] <bryce> ah ok
[08:24] <tepsipakki> slangasek asked for it yesterday
[08:24] <bryce> I saw it in the backlog but didn't notice the result
[08:26] <tepsipakki> did soren test the xresprobe fix?
[08:27] <bryce> not that I know of
[08:29] <bryce> to be honest I'm still not sure if it is a critical enough fix to need going in
[08:29] <bryce> it only is needed for intel laptops where ddcprobe fails, which I think would be fairly rare
[08:29] <bryce> otoh, chances for regression are pretty small
[08:29] <tepsipakki> and the installer does work after that, it's just that you need to hit enter a couple of times :)
[08:30] <bryce> but the bug is marked as milestoned for -rc
[08:36] <bryce> night!
[08:42] <tepsipakki> night :)
[08:43] <tepsipakki> I wouldn't mind going to bed again.. instead I cycled 9km to work when it was -3C
[09:29] <soren> tepsipakki: Which xresprobe fix is that?
[09:30] <tepsipakki> the intel corruption with alternative installer
[09:31] <soren> No, that's the bug, not the fix :)
[09:31] <tepsipakki> heh, right
[09:32] <tepsipakki> check the bug, Bryce asked you to test the fix (links there) .)
[09:32] <tepsipakki> :)
[09:33] <soren> Yes, just saw it.
[09:33] <soren> Um.. Sure, I'll try it when I'm in my office.
[09:34] <tepsipakki> great
[10:05] <ubotu> New bug: #141401 in linux-restricted-modules-2.6.22 (restricted) "xen amd64 nvidia module missing or broken" [Undecided,New]  https://launchpad.net/bugs/141401
[11:27] <ubotu> New bug: #148709 in xrandr (main) "Xrandr rotate with metacity artifacts" [Undecided,New]  https://launchpad.net/bugs/148709
[11:29] <soren> tepsipakki: I just tested the fix from Bryce. It doesn't break the display, but asks me for my resolution, but I suppose that's to be extected?
[11:30] <soren> ...since it neither gets the info from ddcprobe nor xprobe..
[11:36] <tepsipakki> soren: right..
[11:41] <soren> tepsipakki: Ok, so I guess the fix works as intended. It's not optimal, but better than before.
[11:43] <tepsipakki> yeah
[11:46] <ubotu> New bug: #146954 in linux-restricted-modules-2.6.22 (restricted) "gnome freezes after proprietary nvidia driver install" [Undecided,Confirmed]  https://launchpad.net/bugs/146954
[01:40] <ubotu> New bug: #151558 in linux-restricted-modules-2.6.22 (restricted) "nvidia driver crashes sauerbraten" [Undecided,New]  https://launchpad.net/bugs/151558
[01:50] <ubotu> New bug: #151557 in compiz (main) "compiz dies immediately in kde/kubuntu gutsy latest after dist upgrade (dup-of: 144241)" [Medium,New]  https://launchpad.net/bugs/151557
[02:31] <ubotu> New bug: #66481 in linux-source-2.6.17 (main) "Atheros WLAN (MadWifi) poor signal strength  (dup-of: 62106)" [Medium,Confirmed]  https://launchpad.net/bugs/66481
[02:41] <ubotu> New bug: #147361 in xserver-xorg-video-ati (main) "[gutsy]  kdm crashes when randr is enabled" [Undecided,Incomplete]  https://launchpad.net/bugs/147361
[03:29] <slowhand> hi
[04:40] <pcjc2> Hi Bryce
[04:40] <bryce> heya pcjc2
[04:41] <pcjc2> am playing with the Xv issue
[04:41] <pcjc2> I'm just trying to make sure it forces a reset of the Xv regs / setup after its mode-switched, see if that helps
[04:42] <pcjc2> I wonder if the mode-setting is clearing the Xv regs, and as we don't seem to be reprogramming them afterwards - it might be an issue
[04:42] <pcjc2> To be honest though, I don't know what bits of the driver Compiz invokes which would make it only appear with compiz
[04:43] <pcjc2> yep
[04:43] <pcjc2> fixes the issue for me!
[04:43] <bryce> ooh, awesome
[04:43] <pcjc2> Also, ubuntu patch 01 has a bug 
[04:44] <pcjc2> kindof... it introduces an extra condition test checking to see whether to return BadAlloc or not
[04:44] <bryce> can you report that to the ubuntu-x@ list?  I'm about to hop on a plane and will be gone the next couple days
[04:44] <pcjc2> but doesn't add () around it, the test is like (A && B || C)
[04:44] <bryce> ah
[04:44] <pcjc2> where it should be (for clarity at least), (A && ( B || C))
[04:45] <pcjc2> I will disect to make sure it wasn't just this tiny little change (as well as my "hack" of a  forced reset) which helps
[04:45] <pcjc2> It shouldn't be the if statement, as if that were to be the problem, it would either return BadAlloc un-necessarily
[04:46] <pcjc2> or NOT return BadAlloc when it should.
[04:46] <pcjc2> We should probably send the info via Intel at least.. I'm so new to the driver - and they have the docs / knowledge of _how_ it should work
[04:47] <pcjc2> Being dumb... where is the ubuntu-x@ list subscription?
[04:50] <pcjc2> darn.... still managed hang the GPU mode-switching with Xrandr whilst its playing video
[04:51] <tepsipakki> pcjc2: lists.ubuntu.com
[04:52] <pcjc2> tepsipakki: thanks
[05:14] <ubotu> New bug: #151351 in xserver-xorg-video-intel (main) "Corrupted screen on G33 with -14 kernel; regression from -13" [High,Confirmed]  https://launchpad.net/bugs/151351
[06:15] <ubotu> New bug: #151621 in xorg (main) "Cursor Disappears - Nvidia Card" [Undecided,New]  https://launchpad.net/bugs/151621
[06:55] <pcjc2> I've sent a possible fix for #141063 to ubuntu-x@
[06:56] <ubotu> New bug: #151635 in xserver-xorg-video-intel (main) "Intel G33 X broken again by -14 kernel" [Undecided,New]  https://launchpad.net/bugs/151635
[06:57] <tepsipakki> pcjc2: that's already been fixed by dropping a patch
[06:58] <tepsipakki> which needed a patch for the kernel, which has also been dropped :)
[07:00] <pcjc2> ok, can you point me at the upload?
[07:01] <pcjc2> ok, I see it
[07:04] <pcjc2> that fixed it?
[07:04] <pcjc2> I didn't think I had "G33" class hardware!
[07:07] <pcjc2> was it the kernel patch which was causing the corruption?
[07:07] <pcjc2> because the patch dropped from the xorg driver looks like a NOP on my hardware
[07:15] <pcjc2> No, I can confirm the 0ubuntu7 release which drops the G33 patch does not fix #151621
[07:15] <pcjc2> (sorry... wrong number... that was unrelated)
[07:15] <pcjc2> I can confirm the 0ubuntu7 release which drops the G33 patch does not fix #141063
[07:33] <tepsipakki> oops, sorry.. I was confused
[07:39] <tepsipakki> right after your message ubotu mentioned about a bug, which I thought was the one you mentioned :)
[07:40] <pcjc2> Unfortunately Bryce's workaround for the grey blocks crash breaks palette restore when switching back to console
[09:51] <ubotu> New bug: #151674 in linux-restricted-modules-2.6.22 (restricted) "compiz+nvidia: tearing in xvideo (totem/mplayer/VLC)" [Undecided,New]  https://launchpad.net/bugs/151674
[10:26] <ubotu> New bug: #147837 in xserver-xorg-video-intel (main) "[gutsy]  Text Installation failed - 2 reasons: encrypted+lvm fs, screen font hang" [Undecided,New]  https://launchpad.net/bugs/147837
[10:31] <ubotu> New bug: #132739 in linux-restricted-modules-2.6.22 (restricted) "init.d/linux-restricted-modules-common script issues "Read only file system" error" [Undecided,Fix released]  https://launchpad.net/bugs/132739
[10:55] <ubotu> New bug: #151689 in xserver-xorg-driver-ati (main) "gdm or kdm has a huge font" [Undecided,New]  https://launchpad.net/bugs/151689
[11:30] <ubotu> New bug: #147081 in xserver-xorg-video-intel (main) "movie player distorts image horizontally" [Low,New]  https://launchpad.net/bugs/147081
[11:30] <ubotu> New bug: #151704 in xorg (main) "gutsy, xorg cpu usage during video playback" [Undecided,New]  https://launchpad.net/bugs/151704
[11:58] <pcjc2> well... its not pretty (still got a lot of debug code which stops me setting the palette anywhere), but I may have got a fix for #127101
[11:59] <pcjc2> I took the block Bryce commented out (stops it crashing, but messes up the console palette) and moved it to before other registers are restored