| T42 | <A_T_R> sdk-foreach-su -ly ssu re 4.4.0.72", then "sdk-foreach-su -ly zypper ref" and finally "sdk-foreach-su -ly zypper dup" | 04:05 |
|---|---|---|
| T42 | <A_T_R> upgraded from 4.4.0.68 to 4.4.0.72 ,is any method to check sdk version? | 04:07 |
| T42 | <elros34> 'ssu re' but I guess if you already have 4.4.0.68 you do not need to upgrade to newer version | 08:38 |
| T42 | <adampigg> damn, we ran out of build minutes on the CI! | 10:13 |
| T42 | <adampigg> i wonder if we can register as an open source project to get more ci minutes | 10:18 |
| piggz | seems like something we should discuss as a community | 10:19 |
| piggz | probably dont qualify https://about.gitlab.com/solutions/open-source/join/ | 10:20 |
| *** amccarthy is now known as Guest4510 | 11:03 | |
| *** amccarthy_ is now known as amccarthy | 11:03 | |
| T42 | <b100dian> So this wasnt just a month worth of CI minutes, is it actually over? | 15:11 |
| T42 | <adampigg> no, it will restart on 1-nov (re @b100dian: So this wasnt just a...) | 16:53 |
| T42 | <adampigg> we got 400min/mo | 16:53 |
| T42 | <RealDanct12> 400 min is basically nothing (re @adampigg: we got 400min/mo) | 17:09 |
| T42 | <adampigg> well, yeah, its not ideal (re @RealDanct12: 400 min is basically...) | 17:09 |
| T42 | <adampigg> esp with all the project we have, and me triggering builds like theyre free! | 17:10 |
| T42 | <A_T_R> my mistake ,upgrade from 4.4.0.58 to 4.4.0.72 (re @elros34: 'ssu re' but I guess...) | 17:24 |
| T42 | <NotKit> there is also GitHub Actions, but migration would require changes to pipeline | 17:27 |
| T42 | <adampigg> sure, im considering this (re @NotKit: there is also GitHub...) | 17:46 |
| T42 | <elros34> @A_T_R still minor update so I wouldn't bother about sdk upgrade. Maybe even upgrading sdk target is not needed | 18:36 |
| mal | that sdk-foreach-su method updates sdk, tooling and all targets, it's quite easy way and I always use it | 18:55 |
| poetaster | mal, elaborate? | 18:56 |
| mal | poetaster: this way "sdk-foreach-su -ly ssu re 4.4.0.72", then "sdk-foreach-su -ly zypper ref" and finally "sdk-foreach-su -ly zypper dup" | 18:56 |
| poetaster | neato! | 18:57 |
| poetaster | not quite as much fun as: https://github.com/JohnEarnest/ok/blob/gh-pages/docs/Manual.md | 19:00 |
| poetaster | but very useful. | 19:00 |
| poetaster | mal, I traced the godo 3.5 crash on volla/gs290/gs5 to glEnable which leaves me in a quandry. | 19:13 |
| poetaster | dies here: https://github.com/savegame/godot/blob/dec12506b930a3f4a66c9483be5dc80f9e13b3e2/drivers/gles2/rasterizer_gles2.cpp#L256 | 19:13 |
| poetaster | How best do proceed to debug. Don't want to lean on piggz. | 19:13 |
| poetaster | well, or you for that matter :) | 19:13 |
| piggz | can you compare against the working godot verison? | 19:15 |
| poetaster | it's a fair bit of time. almost 2 years of foreign changes and more recent changes from shashi. | 19:16 |
| poetaster | I could walk back sashikknox work from about: https://github.com/savegame/godot/commit/6ed4b10baf4dd1841b41d0036018f7028b9d6557 | 19:20 |
| poetaster | but frankly, I have a lot of irons in a bunch of fires. like. every. one. else. here. beer. | 19:20 |
| mal | poetaster: did some older godot version work? | 19:25 |
| mal | that code has been in godot since 3.1 version | 19:26 |
| mal | is there some difference in working and that 3.5 version related to how much debugging is enabled? | 19:27 |
| T42 | <b100dian> Thaodan: ngfd PR 6 docs should be merged I think? | 19:31 |
| poetaster | mal, the 3.2 version works fine everywhere. the 3.5 works fine on xperia 10ii (and fxprotech thingy from piggz). just not on volla and gs5 | 19:32 |
| poetaster | the last savegame commit is last 3 months and certainly not in godot, otherwise ;) | 19:33 |
| poetaster | I looked at the entire commit history from april 2020 up to shashi's changes which is the diff between 3.2 and 3.5 ++. | 19:34 |
| poetaster | I'd be inclined to think it was shashi's changes, but work on at least 2 ports. | 19:35 |
| piggz | id be inclined to say its mediatek weirdness | 19:36 |
| poetaster | i'd be inclined to agree. shit. | 19:36 |
| piggz | but id love to be shown it was soemthing else! | 19:38 |
| poetaster | I really really tried to blame pulse audio, but it get's all the way to glEnable which is external to godot. At least as far as I can step. How to proceed?! | 19:40 |
| poetaster | ah, linking to debug symbol libhybris-libGLESv2? | 19:43 |
| mal | piggz: I wouldn't rule out something in libhybris needing mediatek specific hacks | 20:16 |
| piggz | yeah, probably | 20:17 |
| mal | poetaster: something I was thinking, if it crashes in android side, do you use the env var to get symbols for android side via libhybris | 20:18 |
| mal | HYBRIS_ENABLE_LINKER_DEBUG_MAP=1 gdb something | 20:19 |
| mal | poetaster: have you checked whether that part of the code is used in 3.2 build? | 20:24 |
| piggz | poetaster: mal: out of interest, openlara works fine on the v22 | 20:39 |
| mal | ok | 20:46 |
| piggz | i assume that code also calls glEnable at some point? | 20:56 |
| mal | piggz: it's probably not about glEnable itself but the parameter to it | 20:57 |
| mal | is it certain it's the glEnable and not the callback call after it | 20:58 |
| piggz | poetaster: what glEnable call is it? | 20:58 |
| mal | piggz: https://github.com/savegame/godot/blob/dec12506b930a3f4a66c9483be5dc80f9e13b3e2/drivers/gles2/rasterizer_gles2.cpp#L256 the was mentioned earlier | 20:59 |
| mal | I remember having issues with gl debug stuff during vulkan porting | 20:59 |
| piggz | is that a vital call? tried commenting it out/ | 21:01 |
| piggz | ? | 21:01 |
| mal | no it's not, it's just about some extra debug | 21:01 |
| mal | you can see that the whole part of the code is ifdef'd with CAN_DEBUG | 21:02 |
| mal | that is why I asked if it's used in 3.2 build, in case that had less debugging enabled | 21:04 |
| poetaster | mal, the code in 3.5 has diverged considerably from 3.2 if only you consider the matrix transforms shash does but more. | 21:17 |
| poetaster | mal, env var to get symbols sounds like a tip! | 21:18 |
| poetaster | jeez, where do I find openlara!? | 21:19 |
| mal | poetaster: I meant that is CAN_DEBUG defined in both 3.2 and 3.5 build you have made | 21:19 |
| mal | poetaster: sfos build of openlara is not publicly available yet but I can give you a link in private if you want | 21:20 |
| poetaster | mal, I'm not sure about CAN_DEBUG? directly after glEnable is a 'callback(_gl_debug_print, NULL);' followed by another 'glEnable(_EXT_DEBUG_OUTPUT);' which irritates me. | 21:23 |
| mal | have you added debug prints between each of those | 21:24 |
| poetaster | Nope. I haven't even got scons to build it aarch64 yet. sigh. | 21:24 |
| piggz | poetaster: perhaps try a build with can_debug undefined? | 21:24 |
| poetaster | piggz, ah, now I'm curious and confused. | 21:26 |
| poetaster | piggz, your suggesting the debugging itself is the issue? I'm still trying to figure out what _EXT_DEBUG_OUTPUT_SYNCHRONOUS_ARB means :) | 21:27 |
| piggz | poetaster: yes, thats my suggestion | 21:27 |
| piggz | i googled that, andit only appears in godot afaict! | 21:28 |
| poetaster | hah, go figure. I attract one hit wonders with google. Vis. Spectre/Meltdown and linux vserver kernels ;) | 21:29 |
| mal | piggz: https://github.com/libhybris/libhybris/blob/master/hybris/include/GLES2/gl2ext.h#L72 it's also named this | 21:31 |
| mal | you can see the value is the same | 21:32 |
| piggz | in any case, it hardly seems an important call | 21:32 |
| poetaster | yeah, it's odd | 21:33 |
| mal | anyway, we don't actually implement any of those ext functions in libhybris but since those are used with eglGetProcAddress it should not matter, maybe | 21:34 |
| poetaster | mal, GL_DEBUG_OUTPUT_SYNCHRONOUS_KHR != GL_DEBUG_OUTPUT_SYNCHRONOUS_ARB? | 21:36 |
| mal | https://github.com/savegame/godot/blob/dec12506b930a3f4a66c9483be5dc80f9e13b3e2/drivers/gles2/rasterizer_gles2.cpp#L36 vs https://github.com/libhybris/libhybris/blob/master/hybris/include/GLES2/gl2ext.h#L72 same value is used | 21:37 |
| poetaster | ah. got it. | 21:38 |
| poetaster | piggz, mal I've pinged shashikknox ... I'll take another kick at building it (yet another build system, yabs). tomorrow I have 12 10 year olds and a halloween birthday party so I'm off to bed. Thanks for all the input! | 21:44 |
| poetaster | mal, the #ifdef CAN_DEBUG stuff seems the same 3.2/3.5 ok. bed. | 21:48 |
| piggz | https://github.com/savegame/godot/blob/3.2.1-stable-sailfish/drivers/gles2/rasterizer_gles2.cpp#L223 | 21:54 |
| piggz | requires glad_enabled | 21:54 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!