| Monday, February 3rd, 2020, 03:47 UTC | ||
| [03:47:42] | Steve-Goodey (Steve-Goodey!~steve@2a00:23c5:7d90:bc01:9be6:831c:4ed8:495d) has joined #mythtv | |
| [03:47:42] | Mode for #mythtv by ChanServ!ChanServ@services. : +v Steve-Goodey | |
| [05:00:50] | Steve-Goodey (Steve-Goodey!~steve@2a00:23c5:7d90:bc01:9be6:831c:4ed8:495d) has quit (Quit: Konversation terminated!) | |
| [06:49:00] | peterbennett (peterbennett!~pi@mythtv/developer/peterbennett) has quit (Quit: WeeChat 2.3) | |
| [06:55:49] | Steve-Goodey (Steve-Goodey!~steve@2a00:23c5:7d90:bc01:9be6:831c:4ed8:495d) has joined #mythtv | |
| [06:55:49] | Mode for #mythtv by ChanServ!ChanServ@services. : +v Steve-Goodey | |
| [06:55:55] | markspieth22 (markspieth22!~yaaic@mythtv/developer/markspieth) has joined #mythtv | |
| [06:55:55] | Mode for #mythtv by ChanServ!ChanServ@services. : +v markspieth22 | |
| [06:58:29] | markspieth (markspieth!~yaaic@mythtv/developer/markspieth) has quit (Ping timeout: 264 seconds) | |
| [07:00:46] | peterbennett (peterbennett!~pi@mythtv/developer/peterbennett) has joined #mythtv | |
| [07:00:46] | Mode for #mythtv by ChanServ!ChanServ@services. : +v peterbennett | |
| [08:52:18] | Steve-Goodey (Steve-Goodey!~steve@2a00:23c5:7d90:bc01:9be6:831c:4ed8:495d) has quit (Quit: Konversation terminated!) | |
| [09:02:41] | Merlin83b (Merlin83b!~Daniel@2001:8b0:bbd6:bd9c:3d93:3f4:641:3d06) has joined #mythtv | |
| [10:29:41] | markspieth22 (markspieth22!~yaaic@mythtv/developer/markspieth) has quit (Ping timeout: 264 seconds) | |
| [10:32:53] | markspieth (markspieth!~yaaic@mythtv/developer/markspieth) has joined #mythtv | |
| [10:32:53] | Mode for #mythtv by ChanServ!ChanServ@services. : +v markspieth | |
| [11:01:29] | markspieth (markspieth!~yaaic@mythtv/developer/markspieth) has quit (Ping timeout: 264 seconds) | |
| [11:01:48] | markspieth (markspieth!~yaaic@mythtv/developer/markspieth) has joined #mythtv | |
| [11:01:48] | Mode for #mythtv by ChanServ!ChanServ@services. : +v markspieth | |
| [11:13:01] | markk_ (markk_!~mark@host109-149-186-43.range109-149.btcentralplus.com) has joined #mythtv | |
| [11:25:15] | markk_: | warpme_: nvidia legacy issues should be fixed. re nvidia minimyth libs – do you want cuda stuff as well? |
| [11:30:28] | markk_: | warpme_: the change to gl/gles config has thrown up some interesting issues on pi4. previously it would happily use composed YUV->RGB DMABUF import – now it doesn't work properly (same behaviour I see on desktp with vaapi). presumably because it was pulling in functionality from GL and GLES. so I need to push a fix. |
| [11:31:13] | markk_: | highlights for me that using separate textures for YUV frame import is definitely GLES3.0 functionality |
| [11:32:05] | markspieth (markspieth!~yaaic@mythtv/developer/markspieth) has quit (Ping timeout: 264 seconds) | |
| [11:32:41] | warpme_: | markk_: perfect re. nvidia. Give me sec. Just finished compilation and will do test within next minutes 😉 |
| [11:37:02] | warpme_: | re. separate textures for YUV: indeed I saw EGL texture imports are gles3 thing while lima guys are stating lima offers gles2. I suspect lima offers indeed gles2 level but with some selected gles3 features? |
| [11:39:55] | markk_: | warpme_: do you have the url for the master lima repository? |
| [11:42:02] | warpme_: | sure: here it is: https://gitlab.freedesktop.org/mesa/mesa/commits/master |
| [11:44:54] | markk_: | and now pi4 is working without issue without any changes – but will only boot to 640x480 :( |
| [11:45:53] | warpme_: | markk_: re. ion2 and current master: 1/works perfectly now; 2/do we need libEGL from Nvidia blobs when we have Qt compiled for DesktopGL?; 3/.LiveTV and no 2xHW DI thing: for to 10 livetv starts 7 was with 2x DI OK, but 3 were with progressive scan + 5–10sec stuttering at beginning. |
| [11:49:04] | warpme_: | markk_: re. lima: this might be good place to ask for details: https://gitlab.freedesktop.org/lima/mesa/issues/130 Vasily/Qiang are very helpful regarding questions! |
| [11:51:25] | warpme_: | ah pi4.... getting 2 video decoders: state-full for <HVEC and state-less for HVEC is "nice" HW feature (2 separate APIs for this in ffmpeg) isn't? |
| [11:56:35] | warpme_: | i got almost perfect hw.video decode on amlogic vdec (s912) – except it randomly enters jumpy video dendering for 5–30sec periods and next recovers itself again to nice playback. Still don't know where root cause is (myth or 3D driver or vdec+ffmpeg cooperation). If anybody interested for log from such playback: |
| [11:56:39] | warpme_: | https://pastebin.com/ezGvPPtw |
| [12:09:43] | markspieth (markspieth!~yaaic@mythtv/developer/markspieth) has joined #mythtv | |
| [12:09:43] | Mode for #mythtv by ChanServ!ChanServ@services. : +v markspieth | |
| [12:16:38] | markk_ (markk_!~mark@host109-149-186-43.range109-149.btcentralplus.com) has quit (Ping timeout: 260 seconds) | |
| [12:29:37] | warpme_: | markk_: just give try for current master on aarch64 and a bit surpliced that on GLES build Qt5 mythtv configure reports: OpenGLES: no see:https://pastebin.com/BUTH4ywc |
| [13:00:46] | markk_ (markk_!~mark@host109-149-186-43.range109-149.btcentralplus.com) has joined #mythtv | |
| [13:03:30] | markk_: | warpme_: re aarch64 – are you building qt5 after mythtv configure? or is that just a cut and paste log? |
| [13:05:21] | warpme_: | markk_: oh no. qt5 was build sometime ago. I paste build log from time when it was build. mythtv (current master) build today. what interesting: it works ok! |
| [13:07:27] | warpme_: | what killing me (and stated to be seen by mm2 testers) is "white rectriangle" issue. cf. msg 1/02, 3:14PM. |
| [13:07:50] | warpme_: | it started manifest week or two ago... |
| [13:07:58] | stuarta: | that's the window for mythbrowser isn't it? |
| [13:08:55] | markk_: | warpme_: as long as it finds opengl – it should be fine. Qt will link to the version it wants. it's just certain platforms where both gl and gles cause problems. would however be best to fix the detection. |
| [13:09:10] | warpme_: | nope. in my case mythwebbrowser is fs + top bar |
| [13:10:15] | markk_: | warpme_: maybe we need to check for QT_OPENGL_ES_3 as well |
| [13:10:52] | warpme_: | and it seems to appear when myth draws main window: after return from external app launched in mythtv (i.e. chrome browser) or sometimes after resume from suspend. |
| [13:11:52] | markk_: | warpme_: white rectangle or white screen? |
| [13:12:18] | warpme_: | or i.e. after launching rxvt terminal from myth.... |
| [13:13:28] | warpme_: | pls see picture https://paste.pics/d14441406e25c912f3aee6e8c88635b9 or https://www.tapatalk.com/groups/minimyth2/vie . . . =4&t=153 |
| [13:15:03] | warpme_: | it is on nvidia 340 blob and on panfrost GL and on lima GL.... |
| [13:15:37] | warpme_: | for me seems like thing from last touches to myth GUI/windowing code |
| [13:38:12] | markspieth: | 32 bit android build now works |
| [13:39:01] | peterbennett: | markspieth: Great thanks I will try it later today :) |
| [13:42:07] | peterbennett: | I am having ongoing problems with #13479 |
| [13:42:07] | ** MythLogBot https://code.mythtv.org/trac/ticket/13479 ** | |
| [13:42:26] | markk_ (markk_!~mark@host109-149-186-43.range109-149.btcentralplus.com) has quit (Ping timeout: 240 seconds) | |
| [13:49:54] | ** stuarta fires up shield to apply outstanding updates ** | |
| [14:21:31] | stuarta: | hmm, the upcoming "2019" shield remote looks far more useful than the original |
| [14:30:03] | markk_ (markk_!~mark@host109-149-186-43.range109-149.btcentralplus.com) has joined #mythtv | |
| [14:56:38] | markk_ (markk_!~mark@host109-149-186-43.range109-149.btcentralplus.com) has quit (Quit: Konversation terminated!) | |
| [14:56:56] | markk_ (markk_!~mark@host109-149-186-43.range109-149.btcentralplus.com) has joined #mythtv | |
| [14:57:12] | markk_: | warpme_: can you paste the conttents of QtGui/qtgui-config.h for the aarch64 build – from wherever your qt5 includes are installed |
| [15:27:17] | Steve-Goodey (Steve-Goodey!~steve@2a00:23c5:7d90:bc01:9be6:831c:4ed8:495d) has joined #mythtv | |
| [15:27:17] | Mode for #mythtv by ChanServ!ChanServ@services. : +v Steve-Goodey | |
| [15:49:21] | Merlin83b (Merlin83b!~Daniel@2001:8b0:bbd6:bd9c:3d93:3f4:641:3d06) has quit (Quit: Leaving) | |
| [15:52:43] | markk_ (markk_!~mark@host109-149-186-43.range109-149.btcentralplus.com) has quit (Ping timeout: 265 seconds) | |
| [15:53:41] | markspieth (markspieth!~yaaic@mythtv/developer/markspieth) has quit (Ping timeout: 264 seconds) | |
| [15:54:20] | markspieth (markspieth!~yaaic@mythtv/developer/markspieth) has joined #mythtv | |
| [15:54:20] | Mode for #mythtv by ChanServ!ChanServ@services. : +v markspieth | |
| [16:02:36] | SteveGoodey (SteveGoodey!~steve@host86-150-60-252.range86-150.btcentralplus.com) has joined #mythtv | |
| [16:02:36] | Mode for #mythtv by ChanServ!ChanServ@services. : +v SteveGoodey | |
| [16:30:17] | markspieth (markspieth!~yaaic@mythtv/developer/markspieth) has quit (Ping timeout: 264 seconds) | |
| [17:11:32] | gregl (gregl!~greg@cpe-24-194-253-7.nycap.res.rr.com) has joined #mythtv | |
| [17:14:24] | markspieth (markspieth!~yaaic@mythtv/developer/markspieth) has joined #mythtv | |
| [17:14:24] | Mode for #mythtv by ChanServ!ChanServ@services. : +v markspieth | |
| [17:18:53] | markspieth (markspieth!~yaaic@mythtv/developer/markspieth) has quit (Ping timeout: 264 seconds) | |
| [17:24:36] | markspieth (markspieth!~yaaic@mythtv/developer/markspieth) has joined #mythtv | |
| [17:24:36] | Mode for #mythtv by ChanServ!ChanServ@services. : +v markspieth | |
| [17:47:20] | gregl (gregl!~greg@cpe-24-194-253-7.nycap.res.rr.com) has quit (Quit: Leaving) | |
| [18:59:12] | markk_ (markk_!~mark@host109-149-186-43.range109-149.btcentralplus.com) has joined #mythtv | |
| [19:58:08] | markk_ (markk_!~mark@host109-149-186-43.range109-149.btcentralplus.com) has quit (Ping timeout: 265 seconds) | |
| [20:55:26] | peterbennett: | I have analyzed #13479 and proposed a possible solution – if anybody has any input or other solution please let me know. |
| [20:55:26] | ** MythLogBot https://code.mythtv.org/trac/ticket/13479 ** | |
| [20:57:01] | peterbennett: | markspieth: I successfully built android 32-bit – testing it now.. Thanks :) |
| [21:04:37] | markspieth2: | peterbennett: I even checked that \I didnt break 64bit build too. It now takes 10 min less for a complete libs+myth build. 26 min for me vs 35 before. Half of qt is no longer built. |
| [21:28:47] | markk_ (markk_!~mark@host109-149-186-43.range109-149.btcentralplus.com) has joined #mythtv | |
| [21:30:53] | markspieth (markspieth!~yaaic@mythtv/developer/markspieth) has quit (Ping timeout: 264 seconds) | |
| [21:37:23] | markspieth (markspieth!~yaaic@mythtv/developer/markspieth) has joined #mythtv | |
| [21:37:23] | Mode for #mythtv by ChanServ!ChanServ@services. : +v markspieth | |
| [21:37:59] | stuarta: | markspieth: android build improvements sound good |
| [21:38:17] | stuarta: | i will have to try building it again and uploading it to the shield |
| [21:40:14] | SteveGoodey (SteveGoodey!~steve@host86-150-60-252.range86-150.btcentralplus.com) has quit (Quit: Konversation terminated!) | |
| [21:42:19] | Tobbe5178 (Tobbe5178!~asdf@81-235-210-78-no62.tbcn.telia.com) has quit (Ping timeout: 272 seconds) | |
| [21:49:26] | peterbennett: | markspieth2: Unfortunately it crashes :( See my email. |
| [21:49:58] | peterbennett: | markspieth2: Cannot find iconv |
| [22:06:47] | Tobbe5178 (Tobbe5178!~asdf@2001:2002:51eb:d24e:a40b:3a10:3792:4b69) has joined #mythtv | |
| [22:12:39] | amessina (amessina!~amessina@unaffiliated/amessina) has joined #mythtv | |
| [22:15:18] | Steve-Goodey (Steve-Goodey!~steve@2a00:23c5:7d90:bc01:9be6:831c:4ed8:495d) has quit (Quit: Konversation terminated!) | |
| [22:20:07] | gigem: | peterbennett: I try to very hard to not touch the slave wake/shutdown code. It's always seemed like fragile house of cards to me. If you think m_schedLock needs to be unlocked, then follow the existing "if (m_recListChanged) return/continue" paradigm to short-circuit back to the top of the scheduler loop. |
| [22:20:46] | gigem: | markspieth: Android 32-bit fails to build for me. I haven't had time to look into why? |
| [22:21:49] | peterbennett: | gigem: I looked through the various calls and I think I only need to unlock when calling m_mainServer->isClientConnected() |
| [22:23:18] | peterbennett: | gigem: I am not very familiar with scheduler – what is m_recListChanged – I don't see it |
| [22:25:27] | peterbennett: | gigem: I am looking at V30 – is that something newer? |
| [22:38:55] | peterbennett: | gigem: There are already 9 places in scheduler where calls are surrounded by schedLock.unlock and schedLock.lock – I would be adding two more – so maybe this is the same solution used before for tother s |
| [22:42:23] | stuarta: | sounds painful |
| [22:43:55] | peterbennett: | The deadlock is painful. requires kill -KILL of mythbackend to get running again. Regular kill is ignored. |
| [22:45:04] | stuarta: | feel free to mark the relevant ticket as a blocker |
| [22:45:22] | markk_ (markk_!~mark@host109-149-186-43.range109-149.btcentralplus.com) has quit (Ping timeout: 265 seconds) | |
| [22:46:43] | peterbennett: | It has been there since v30, I think it requires some unusual timing issues although I can create it quite easily. There have been 2 or 3 others that reported it. |
| [22:55:36] | gigem: | peterbennett: m_recListChanged is called reclist_changed in v30. hampton renamed it in his cleanups. It's a boolean that gets set whenever anything in the scheduler, recording list changes while the lock is not held. If it's set when the lock is relocked, the list is potentially no longer safe to traverse and the call chain should return back to the top of the main loop in Scheduler::run(). |
| [22:57:12] | peterbennett: | gigem: I see okay thanks I will take a look at it. |
| [23:02:53] | gigem: | I finally had an instance of bad, frame time (i.e. skip forward actually goes back) last night with the change markk_ and I have discussed. I think this is a case of GIGO, though. The recording was 1080i mpeg2 ts that I'd transcoded to 1080i hevc mkv. The actual recording was 1:01 in length after transcoding and running mythcommflag --rebuild to rebuild the seek table, mythtv only reported it as |
| [23:02:55] | gigem: | about 0:55 in length. I've seen that before and think we have a bug somewhere in our hevc frame counting. |
| [23:20:49] | amessina (amessina!~amessina@unaffiliated/amessina) has quit (Quit: Konversation terminated!) | |
IRC Logs collected by
BeirdoBot.
Please use the above link to report any bugs.