| === brainwash_ is now known as brainwash | ||
| Unit193 | bluesabre: So I ran the 'get-pot.sh' script, merged the translations with the msgmerge tool, then diff'd the translation percent: http://paste.openstack.org/show/8OpGNiuGJXRNBIHUuIat/ it's a bit painful. | 03:01 |
|---|---|---|
| === jphilips_ is now known as jphilips | ||
| jphilips | thunar taking up 1 cpu has happened to me twice in the last day :D | 09:00 |
| bluesabre | Unit193: so maybe we don't ship those translations yet, and install call for updates to them and get them out for 20.04.1? | 10:30 |
| bluesabre | s/install/instead | 10:30 |
| -BottyMcBotFace:#xubuntu-devel- ::platform:: Update supported kernel packages @ http://git.launchpad.net/~ubuntu-core-dev/ubuntu-seeds/+git/platform/commit/?id=c0e59184d807f50ec6f89d3800758691fda1b601 (by Dimitri John Ledkov) | 10:33 | |
| -BottyMcBotFace:#xubuntu-devel- Reminder: Next meeting chair is Unit193 | 17:00 | |
| bluesabre | https://gitlab.gnome.org/Community/Ubuntu/gtk-common-themes/-/commit/77468200d2e287ba041aed1e5c475f0958af4a5f :) | 18:49 |
| TJ- | Heads up about an Xorg regression affecting multi-monitors a recent upgrade caused; might not be specific to Xubuntu but affecting 6 new Lenovo E495's here Bug #1873895 | 20:18 |
| ubottu | bug 1873895 in xserver-xorg-video-amdgpu (Ubuntu) "Regression: block staircase display with side-by-side monitors" [Undecided,New] https://launchpad.net/bugs/1873895 | 20:18 |
| bluesabre | Thanks TJ- | 20:19 |
| TJ- | bluesabre: struggling to figure out how to further track it down; any suggestions as to what to attack welcome! | 20:19 |
| bluesabre | TJ-: We updated xfwm4 recently, not sure if that could be related | 20:21 |
| TJ- | bluesabre: let me review the package upgrade list! | 20:22 |
| TJ- | ooo!! xfwm4:amd64 (4.14.0-2ubuntu1, 4.14.1-0ubuntu1) | 20:23 |
| TJ- | so if we downgrade to 4.14.0-2ubuntu1 that would be a worthwhile test | 20:24 |
| bluesabre | Yes. I’d be curious about that. Somebody else mentioned the same issue, but their screenshot showed a completely normal desktop. | 20:27 |
| brainwash | the gitlab issue points to https://bugzilla.xfce.org/show_bug.cgi?id=16716 | 20:28 |
| ubottu | bugzilla.xfce.org bug 16716 in General "don't use xpresent on AMD CARRIZO as it produces garbled screen" [Normal,New] | 20:28 |
| TJ- | that's where we started! in my bug report I linked to the amdgpu 'driver' bug though | 20:29 |
| TJ- | the bit didn't understand is the mention of xpresent vs glx, and "...switched back to glx" | 20:30 |
| brainwash | 4.14.1 uses xpresent when an AMD gpu is detected | 20:31 |
| brainwash | .0 used glx | 20:31 |
| TJ- | aha! | 20:31 |
| TJ- | thanks for the explanation. I'm not at the office but just realised I've a HDMI monitor and cable close to hand so I can test it now | 20:32 |
| brainwash | https://git.xfce.org/xfce/xfwm4/commit/?id=23900123ad8418149897a094d1096d6ecb984d3c | 20:33 |
| TJ- | It would help to test it still fails before downgrading! Didn't have the problem with the older package but just doing a full upgrade to test if it does fail here | 20:42 |
| brainwash | it's a last minute change for sure | 20:44 |
| brainwash | meaning that .1 was not tested much at all | 20:45 |
| TJ- | well grrrr, cannot reproduce! | 20:46 |
| TJ- | Got it!! | 20:48 |
| TJ- | The eDP is 192x1080, the external monitor here (at home) is 1920x1200 and does not suffer | 20:49 |
| TJ- | At the office the external monitor(s) are 1680x1050 and suffer. I just switched the resolution of this HDMI monitor to 1680x105 and reproduced | 20:49 |
| bluesabre | No AMD gpu to test things with. Will be my next hardware purchase. | 20:55 |
| bluesabre | If you can confirm, we can probably patch that back out after reporting upstream | 20:58 |
| TJ- | Well, now we've got it where should I dig? I've updated the bug report with my observations about the resolution and my hypothesis as to WHY it occurs as it does; I need sleep but tomorrow we can have a crack at fixing it if I know which code to attack | 20:58 |
| TJ- | I don't think xfwm4 is at fault here so instead of reverting there, let's attack the driver itself | 20:59 |
| TJ- | bluesabre: which country are you in? | 20:59 |
| TJ- | basically, looks like the driver is trying to render 1920 pixels to 1680 display - we can try to measure the block width tomorrow to confirm | 21:04 |
| TJ- | I can work on it for now but if I get out of my depth I'm prepared to ship a loan E495 to one of you if you think you can get further | 21:08 |
| bluesabre | TJ-: US. I wouldn’t recommend shipping me anything currently, not sure if make reasonable progress. | 21:34 |
| TJ- | bluesabre: OK ... I'm looking at the driver... a bisect might get us somewhere based on this: $ gitlog | grep present: | wc -l => 20 | 21:35 |
| TJ- | I suspect a636f42b496b0604ca00a144690ece61d1a88a27 present: Check that flip and screen pixmap pitches match | 21:37 |
| TJ- | looks like there's a series of incremental patches around flip/pitch in present that are suspect | 21:50 |
| xu-core-beta0tw | Core installer crashes in both VMWare and Virtual Box. Had no problem installing the Desktop ISO. | 22:00 |
| Unit193 | Running the installer from the boot menu or the live desktop session? | 22:03 |
| xu-core-beta0tw | Both. | 22:06 |
| xu-core-beta0tw | Tried unchecking downloads and using safe graphics option, neither helped. Crash is always near the end of the install. | 22:07 |
| Unit193 | OK, the boot menu option is unfortunately common, but live session isn't. Did the installer complete, just crash at the end? Do you have logs of what happened? | 22:08 |
| xu-core-beta0tw | No, never completed so I don't know if there's an install log. I'll try again and look. | 22:15 |
| Unit193 | That'd be the logs in /var/log/installer/ and /var/log/syslog (ubiquity pushes everything to there..) | 22:16 |
| TJ- | I've noticed an apparent xf86-video-amdgpu 'pitch' problem in Xorg log. When using single output eDP 1920x1080 the log shows "=> pitch 7680 bytes" and 7680/4=1920. with ext monitor at 1920x1200 I see "=> pitch 15360 bytes" and 15360/4-1920=1920 BUT when ext monitor is at 1680x1050 I see " => pitch 14848 bytes" however 14848/4-1920=1792 NOT 1680 as would be expected | 22:22 |
| Unit193 | TJ-: I may have missed it, did you end up trying downgrading xfwm4? | 22:25 |
| TJ- | Unit193: yes but that wasn't conclusive since I was testing with a 1920x1200 ext HDMI - didn't spot until later that the issue occurs with the 1680x1050 output | 22:26 |
| TJ- | Unit193: with the downgrade I didn't observe the issue BUT didn't test with the HDMI at 1680 | 22:26 |
| xu-core-beta0tw | Couldn't find anything useful in the logs so tried the installer from the desktop again in Virtual Box and this time it installed. This was a fresh image, last time I installed from the desktop after the crash took me there. If I can replicate the crash again I'll let you know. Thanks for your help! | 22:46 |
| Unit193 | xu-core-beta0tw: Hey, glad it seems to be working! | 22:47 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!