| 00:09:36 | | Quit massiveH (Quit: Leaving) |
| 00:16:25 | | Join massiveH [0] (~massiveH@2600:4040:a982:dc00:c856:873d:bd95:7386) |
| 00:37:49 | | Quit jacobk (Ping timeout: 260 seconds) |
| 00:44:28 | | Join massive_H [0] (~massiveH@2600:4040:a982:dc00:c856:873d:bd95:7386) |
| 00:48:00 | | Quit massiveH (Ping timeout: 276 seconds) |
| 01:00 |
| 01:51:55 | rb-bluebot | Build Server message: New build round started. Revision 188f025f51, 338 builds, 9 clients. |
| 01:51:55 | rb-bluebot | [Bug Fix] Data Abort on Usb Unplug, database browser ran on USB dc by William Wilgus |
| 01:52:21 | _bilgus | This may be the very last of that what a terrible bug to track down |
| 01:54:38 | _bilgus | so we have menus that are called by the root_menu the playlist catalog is just one of them but basically the skin redraw stores a pointer to the current_list but the functions exit and the list is out of scope on USB unplug and it works sometimes |
| 01:54:47 | *** | Saving seen data "./dancer.seen" |
| 01:54:51 | | Quit massive_H (Quit: Leaving) |
| 01:55:46 | _bilgus | we can actually reenter some menus if we store context but I'm kinda thinking returning to root on unplug isn't too bad |
| 02:00 |
| 02:03:43 | rb-bluebot | Build Server message: Build round completed after 708 seconds. |
| 02:03:44 | rb-bluebot | Build Server message: Revision 188f025f51 result: All green |
| 02:58:00 | | Quit othello7 (Ping timeout: 276 seconds) |
| 03:00 |
| 03:31:13 | | Quit braewoods_ (Remote host closed the connection) |
| 03:31:26 | | Join braewoods_ [0] (~braewoods@user/braewoods) |
| 03:32:42 | | Quit Bobathan_ (Quit: ZNC 1.8.2+deb2+b1 - https://znc.in) |
| 03:33:00 | | Join Bobathan_ [0] (~admin@syn-065-029-248-157.res.spectrum.com) |
| 03:54:48 | *** | Saving seen data "./dancer.seen" |
| 05:00 |
| 05:54:52 | *** | No seen item changed, no save performed. |
| 06:00 |
| 06:07:54 | | Join olspookishmagus [0] (~pookie@snf-137798.vm.okeanos.grnet.gr) |
| 06:54:36 | | Join lebellium [0] (~lebellium@2a01cb0405d07f00788fcfdd258d323c.ipv6.abo.wanadoo.fr) |
| 07:00 |
| 07:22:52 | rb-bluebot | Build Server message: New build round started. Revision a983e532fa, 338 builds, 10 clients. |
| 07:22:52 | rb-bluebot | Track info: Hide codec when it differs between tracks by Christian Soffke |
| 07:33:41 | rb-bluebot | Build Server message: Build round completed after 650 seconds. |
| 07:33:42 | rb-bluebot | Build Server message: Revision a983e532fa result: All green |
| 07:54:24 | rb-bluebot | Build Server message: New build round started. Revision a286555578, 338 builds, 10 clients. |
| 07:54:25 | rb-bluebot | plugins: properties: Add Directory Track Info by Christian Soffke |
| 07:54:56 | *** | No seen item changed, no save performed. |
| 08:00 |
| 08:04:27 | rb-bluebot | Build Server message: Build round completed after 603 seconds. |
| 08:04:28 | rb-bluebot | Build Server message: Revision a286555578 result: All green |
| 08:16:17 | | Quit danwellby (Quit: Watch out For sysops carrying carpet and quicklime) |
| 08:25:58 | | Quit Skyrider (Ping timeout: 265 seconds) |
| 08:36:16 | | Join Skyrider [0] (~Skyrider@static.219.32.201.138.clients.your-server.de) |
| 08:38:51 | _bilgus | have another bug I can hit everytime on usb unplug from the shortcut menu causes a div zero on device and sometimes in the sim but I haven't found it yet but seems in a similar vein |
| 09:00 |
| 09:08:24 | _bilgus | Yes similar vein indeed USB screen does Viewport manager init and supplies its own parent viewport, I haven't actually found where it frees it might be here somewhere?? |
| 09:08:40 | _bilgus | making the vps struct static fixes it |
| 09:08:58 | _bilgus | but it'd be better if it cleaned up after itsself |
| 09:30:36 | | Join Moriar [0] (~moriar@107-200-193-159.lightspeed.stlsmo.sbcglobal.net) |
| 09:45:20 | rb-bluebot | Build Server message: New build round started. Revision 06123c3606, 338 builds, 10 clients. |
| 09:45:21 | rb-bluebot | FS #13452: Major update to Korean translation (Hoseok Seo) by Solomon Peachy |
| 09:49:53 | _bilgus | I'm still not sure whats going on with this one |
| 09:51:26 | _bilgus | I get div by zero on the device and at this point nothing in the sim after making the USBs viewport static, but still I see the shortcut screen text still maybe its not doing scroll stop.. |
| 09:52:36 | _bilgus | that does something like keep a list and compare to the current vp |
| 09:52:50 | _bilgus | so maybe that vp is bunk by then |
| 09:54:58 | *** | Saving seen data "./dancer.seen" |
| 09:56:25 | rb-bluebot | Build Server message: Build round completed after 665 seconds. |
| 09:56:26 | rb-bluebot | Build Server message: Revision 06123c3606 result: All green |
| 09:56:30 | rb-bluebot | Build Server message: New build round started. Revision 3294337a4c, 338 builds, 9 clients. |
| 09:56:30 | rb-bluebot | builds: Based on feedback, switch default Slovak TTS to espeak-ng by Solomon Peachy |
| 09:59:21 | _bilgus | odd since it appears the shortcut screen is supposed to exit on usb plug by virtue of simplelist.. if (simplelist_show_list(&list)) |
| 09:59:21 | _bilgus | break; /* some error happened?! */ |
| 09:59:51 | _bilgus | maybe thats happening and scroo stop doesn't get called till later? |
| 10:00 |
| 10:05:20 | _bilgus | HA |
| 10:06:03 | _bilgus | simplelist_show_list() only cleans up if it doesn't have USB connect |
| 10:06:54 | _bilgus | https://github.com/Rockbox/rockbox/blob/master/apps/gui/list.c#L952 |
| 10:06:56 | rb-bluebot | Build Server message: Build round completed after 627 seconds. |
| 10:06:57 | rb-bluebot | Build Server message: Revision 3294337a4c result: All green |
| 10:07:35 | _bilgus | so it leaves its viewport in the theme stack which gets returned to on USB unplug |
| 10:08:59 | _bilgus | ok so maybe thats the connection to playlist_viewer and plugin_menu not that that they are function_ptr menus i'll try disabing the last patch but I suspect its several sources and hence the reason its been so hard to track down |
| 10:11:22 | _bilgus | nope the previous patch definitely a separate issue |
| 10:12:26 | _bilgus | ODDLY the shortcut text is still scrolling though but atleast no more crash |
| 10:18:02 | _bilgus | No still not quite got it right yet :/ |
| 10:18:46 | user890104 | _bilgus: does it happen on more than one target? i can test on ipod6g |
| 10:19:42 | _bilgus | go for it I'm pretty sure at least one of these are on several targets |
| 10:20:22 | _bilgus | the shortcut one is the easiest to repro and last nights patch takes care of the playlist_viewer and plugin_menu |
| 10:20:43 | _bilgus | feel free to try before you update and see if it fixes it |
| 10:21:38 | _bilgus | so steps for shortcut menu, have one shortcut (I used debug -Dump Rom Contents don't know if it matters) |
| 10:21:59 | _bilgus | open shortcuts plug usb −− unplug crash |
| 10:22:35 | _bilgus | plugin menu, got to top level [Apps, Plugins, Games] plug usb −− unplug, crash |
| 10:22:58 | _bilgus | try in the next level up as well like got to APPs then usb plug unplug |
| 10:23:11 | _bilgus | ditto for the pl viewer |
| 10:24:42 | _bilgus | Uh last nights patch also fixes the 'Returning incorrectly usually) db viewer thing on USB unplug too |
| 10:25:23 | _bilgus | that is what led me down this in the first place |
| 10:34:22 | _bilgus | and these 'were' all reproducible in the sim but I'm still having issues on device with the shortcut one I suspect I'm on the right track due to the scrolling text still going |
| 10:36:08 | _bilgus | the USB screen is kinda asynchronous in whole manner so maybe it should be nuking everything and forcing you back to root |
| 10:42:36 | _bilgus | OK, so wiping out current_lists in the simplelist_show_list() USB handler stops the crash so probably some kind of race there but its odd because the scroller still scrolls the selected menu item and I can see it cleaning up then it comes back |
| 10:43:41 | _bilgus | Ill try making simple list do its own scroll stop on exit |
| 10:59:27 | _bilgus | I also find this odd USB_screen already calls screen->scroll_stop(); |
| 11:00 |
| 11:00:43 | _bilgus | so is it just that the menu continues to run, I'm pretty sure the usb handler basically yields till usb unplug when it hands back execution |
| 11:01:12 | _bilgus | so who is updating the scroller |
| 11:01:55 | _bilgus | IDK user890104 let me know if you can't repro any of these |
| 11:13:47 | user890104 | ok, i'll try on a normal clean build first, then on debug |
| 11:19:45 | | Quit Galois (Remote host closed the connection) |
| 11:25:23 | rb-bluebot | Build Server message: New build round started. Revision 9f366b1b8a, 338 builds, 10 clients. |
| 11:25:24 | rb-bluebot | voice: Flag ':' in voiced strings. by Solomon Peachy |
| 11:26:36 | | Join othello7 [0] (~Thunderbi@pool-100-36-176-164.washdc.fios.verizon.net) |
| 11:34:48 | rb-bluebot | Build Server message: Build round completed after 565 seconds. |
| 11:34:49 | rb-bluebot | Build Server message: Revision 9f366b1b8a result: All green |
| 11:35:20 | | Join Galois [0] (djao@efnet.math.uwaterloo.ca) |
| 11:35:24 | | Join danwellby [0] (~danwellby@cynthiajndevey.plus.com) |
| 11:51:04 | rb-bluebot | Build Server message: New build round started. Revision fe5375a6cb, 338 builds, 10 clients. |
| 11:51:04 | rb-bluebot | Improve Crossfade handling in Single Mode by Roman Artiukhin |
| 11:52:24 | _bilgus | IDK I'll keep looking into it I'll put up a patch but I think if I can figure out why the scroll engine isn't stopping it may just fix the issue entirely and hopefully maybe the previous bugfix will be unnecessary too |
| 11:55:01 | *** | Saving seen data "./dancer.seen" |
| 11:56:11 | _bilgus | I feel like these are all just manifestation of the same underlying issue but none seem to be directly connected I think because there is still a root issue hiding in there |
| 11:56:26 | | Join Demosthe1ex [0] (~Demosthen@user/demosthenex) |
| 12:00 |
| 12:00:00 | | Quit Demosthenex (Ping timeout: 272 seconds) |
| 12:02:02 | rb-bluebot | Build Server message: Build round completed after 658 seconds. |
| 12:02:03 | rb-bluebot | Build Server message: Revision fe5375a6cb result: All green |
| 12:06:40 | | Join Demosthenex [0] (~Demosthen@user/demosthenex) |
| 12:09:58 | | Quit Demosthe1ex (Ping timeout: 252 seconds) |
| 12:15:10 | _bilgus | I think what might be happening is that the scroll engine holds a pointer to a vp and when it goes to check bounds its checking a pointer to data that no longer exists in the stack frome hence the static viewport fixing one of the issues |
| 12:15:32 | _bilgus | but there is probably somewhere else that I haven't found yet |
| 12:33:48 | user890104 | _bilgus: i got a different crash on the most recent commit |
| 12:33:55 | user890104 | https://i.imgur.com/bLx6X7t.jpeg |
| 12:38:41 | user890104 | i'm unable to reproduce the usb plug/unplug crash on 2b7b4c24, rockbox just goes to the main menu if i plug/unplug usb while in the shortcuts menu |
| 12:41:27 | user890104 | so plugging/unplugging usb in plugins-apps just makes it go back one level i.e. if i go to apps, plug/unplug it goes to plugins. if i do the same in plugins, it goes to the main menu |
| 12:57:41 | _bilgus | user890104, returning up a level is what mine does after 188f025f, though the sim with ASAN enabled would probably still show it could be that your memory hasn't been overwritten yet |
| 12:58:34 | _bilgus | the imgur one sounds like more of a failure to communicate with the host but memory corruption is funny |
| 12:59:01 | _bilgus | esp when you start reading and writing others stack |
| 13:00 |
| 13:01:06 | _bilgus | I do notice the scroll engine has a provision for custom scroll functions and synclist appears to have its own but it looks like its all handled properly so I'm not sure this is even the root cause maybe something else is corruption the pointers I'm holding on to |
| 13:05:46 | _bilgus | like it tries to stop the scrolling but it compares the viewport it has and it doesn't match the current ones anymore IDK its weird and its reached by send_event so pretty much could be a race at the root |
| 13:10:13 | _bilgus | user890104, I suppose it depends on what 'TRUE' actually maps to on your device and also what GO_TO directive is mapped to that number but prior to 188f025f mine returns incorrectly to the db browser from the file browser and I think plugin catalog on usb unplug |
| 13:11:55 | _bilgus | which is what originally caught my eye since it kept trying to build the db |
| 13:36:11 | | Join Demosthe1ex [0] (~Demosthen@user/demosthenex) |
| 13:39:25 | | Quit Demosthenex (Ping timeout: 252 seconds) |
| 13:41:20 | | Join Demosthenex [0] (~Demosthen@user/demosthenex) |
| 13:44:55 | | Quit Demosthe1ex (Ping timeout: 260 seconds) |
| 13:46:25 | | Join Demosthe1ex [0] (~Demosthen@user/demosthenex) |
| 13:49:35 | | Quit Demosthenex (Ping timeout: 260 seconds) |
| 13:55:04 | *** | Saving seen data "./dancer.seen" |
| 14:00 |
| 14:13:09 | | Quit danwellby (Quit: Watch out For sysops carrying carpet and quicklime) |
| 14:28:15 | | Join dys [0] (~dys@user/dys) |
| 14:29:44 | _bilgus | well not really any closer but I do see that any screen with scrolling text under the USB screen continues to scroll its lines |
| 14:33:03 | | Join jacobk [0] (~quassel@2601:380:837f:3520:b413:1241:5e3:eebd) |
| 14:37:15 | speachy | huh, that's potentially a major issue regardless |
| 14:38:40 | | Join danwellby [0] (~danwellby@cynthiajndevey.plus.com) |
| 14:42:16 | _bilgus | I do notice a comment about the line buffer possibly needing a mutex so perhaps thats it but I think on USB plug when it calls scroll_stop those lines should be cleared and they appear to be but something reloads them |
| 14:42:57 | _bilgus | so maybe somebody didn't get the memo, IDK i'll keep investigating |
| 14:44:15 | speachy | looks like something motivated translators. We're up to 10 with >90% coverage. would be higher but I've made things a lot stricter in recent months |
| 14:45:35 | speachy | I wonder if I'm being overly nitpicky, but it'll never get better without raising standards |
| 14:55:52 | speachy | cue the infighting in ..-1, -2, -3, -999 |
| 15:00 |
| 15:55:08 | *** | Saving seen data "./dancer.seen" |
| 16:00 |
| 16:30:04 | rb-bluebot | Build Server message: New build round started. Revision f3a5fe2738, 338 builds, 10 clients. |
| 16:30:04 | rb-bluebot | FS #13453: Updated Polish translation (Adam Rak) by Solomon Peachy |
| 16:39:59 | rb-bluebot | Build Server message: Build round completed after 595 seconds. |
| 16:40:00 | rb-bluebot | Build Server message: Revision f3a5fe2738 result: All green |
| 16:42:38 | | Nick Demosthe1ex is now known as Demosthenex (~Demosthen@user/demosthenex) |
| 17:00 |
| 17:06:49 | | Join Demosthe1ex [0] (~Demosthen@user/demosthenex) |
| 17:10:15 | | Quit Demosthenex (Ping timeout: 260 seconds) |
| 17:42:39 | | Quit jacobk (Ping timeout: 276 seconds) |
| 17:50:53 | | Quit lebellium (Quit: Leaving) |
| 17:51:08 | | Join jacobk [0] (~quassel@2601:380:837f:3520:b413:1241:5e3:eebd) |
| 17:55:10 | *** | Saving seen data "./dancer.seen" |
| 18:00 |
| 18:07:46 | | Quit olspookishmagus (Quit: All for nothing) |
| 18:12:49 | rb-bluebot | Build Server message: New build round started. Revision ca332b5473, 338 builds, 9 clients. |
| 18:12:49 | rb-bluebot | Fixes for the SysCfg viewer by Vencislav Atanasov |
| 18:22:48 | rb-bluebot | Build Server message: Build round completed after 601 seconds. |
| 18:22:49 | rb-bluebot | Build Server message: Revision ca332b5473 result: All green |
| 19:00 |
| 19:38:21 | | Join massiveH [0] (~massiveH@2600:4040:a982:dc00:c856:873d:bd95:7386) |
| 19:55:11 | *** | Saving seen data "./dancer.seen" |
| 21:00 |
| 21:55:15 | *** | No seen item changed, no save performed. |
| 22:00 |
| 22:08:41 | | Join __builtin [0] (~quassel@rockbox/developer/builtin) |
| 22:16:45 | __builtin | good to see so much development happening recently :) |
| 22:28:39 | | Quit jacobk (Ping timeout: 276 seconds) |
| 23:00 |
| 23:00:05 | | Quit Moriar (Ping timeout: 248 seconds) |
| 23:32:49 | | Join jacobk [0] (~quassel@2601:380:837f:3520:1113:1c75:8d3b:2220) |
| 23:51:30 | _bilgus | After way too much looking at code I think I finally understand whats going on here, so the list view function uses put_line which stylizes our lines according to the theme/settings, line bar inverted etc. |
| 23:54:44 | _bilgus | so listview has its own scroller function and when scroll_stop is called it putsxy the string to the screen instead and thats what shows in the USB screen |
| 23:55:19 | *** | Saving seen data "./dancer.seen" |
| 23:56:16 | _bilgus | So AFAICT this isn't a deadly thing unlike the others |
| 23:57:36 | _bilgus | we may be able to just suspend that if usb == connected or spin up a new empty list IDK yet |
icon identifies that the person is a core developer (has commit access).