| - qbit (QUIT: Ping timeout: 256 seconds) (~qbit@ns2.suah.dev) | 00:07 | |
| sigrid | building | 00:21 |
|---|---|---|
| sigrid | erlehmann: 3068853e8a58ccc7370a5ce977c08223601c497a - issue present | 00:50 |
| sigrid | I was wondering if it could be a problem in irrlicht | 00:50 |
| - mtm (QUIT: Ping timeout: 240 seconds) (~mtm@c-73-27-62-116.hsd1.fl.comcast.net) | 01:03 | |
| + qbit (~qbit@ns2.suah.dev) | 01:09 | |
| erlehmann | mntmn sigrid https://github.com/minetest/minetest/issues/11987#issuecomment-1021719494 | 01:22 |
| sigrid | interesting | 01:26 |
| sigrid | the fix seems to set both front AND back face culling | 01:29 |
| sigrid | now, I am not proficient in mesa, but to me it looks like translate_cull_face in etnaviv logic only translates one of those | 01:30 |
| sigrid | s/one/either/ | 01:30 |
| sigrid | guess it meant to be really used as a bitmask | 01:31 |
| sigrid | lol I think I just fixed it | 01:36 |
| erlehmann | sigrid wait what how | 01:37 |
| sigrid | with a small fix in mesa | 01:37 |
| mntmn | sigrid: woah | 01:38 |
| erlehmann | can it be fixed in minetest too? | 01:38 |
| erlehmann | or is this a mesa thing | 01:38 |
| mntmn | erlehmann: it sounds like a driver bug in mesa | 01:38 |
| sigrid | lemme test again to be sure | 01:38 |
| erlehmann | yeah but workaround this that | 01:38 |
| sigrid | yeah, the fix worked | 01:39 |
| erlehmann | so what now | 01:40 |
| erlehmann | can it be integrated into system image | 01:40 |
| mntmn | erlehmann: we are not patching mesa | 01:41 |
| mntmn | this needs to be upstreamed | 01:41 |
| kfx | mesa's release cadence is prety fast. upstreaming it will get results | 01:41 |
| erlehmann | then i hope sigrid can fire the patch in mesas general direction somehow | 01:41 |
| sigrid | http://okturing.com/src/13029/body | 01:41 |
| erlehmann | thank you sigrid. do you want to write comment here so the issue can be closed? https://github.com/minetest/minetest/issues/11987 | 01:42 |
| sigrid | feel free to upstream it now, I really have to go walk my dog and sleep, going to be most likely very busy to follow through with the process :/ | 01:43 |
| sigrid | yeah, I'll write there | 01:43 |
| erlehmann | i have literally no idea how to mesa upstream | 01:43 |
| sigrid | me neither. guess sending a patch. I'm too tired to figure that out rn | 01:43 |
| kfx | https://docs.mesa3d.org/submittingpatches.html | 01:44 |
| mntmn | erlehmann: you don't need to sort this out | 01:44 |
| mntmn | erlehmann: thanks tho | 01:44 |
| mntmn | sigrid: good night! | 01:44 |
| erlehmann | ok cool | 01:44 |
| erlehmann | :) | 01:44 |
| kfx | looks like someone with a freedesktop gitlab account needs to open a merge request | 01:44 |
| mntmn | maybe i can do it tomorrow | 01:45 |
| erlehmann | sigrid i found your website https://ftrv.se/_about can you tell me how to 9front on reform? | 01:49 |
| sigrid | you can't. it's only through drawterm rn | 01:51 |
| erlehmann | damn | 01:51 |
| sigrid | I'm doing on and off porting but it's very slow, I don't usually have much time in big continuous chunks, unfortunately | 01:52 |
| sigrid | hw/state_3d.xml.h:486:#define VIVS_PA_CONFIG_CULL_FACE_MODE_MASK 0x00000400 | 01:54 |
| sigrid | this seems wrong to me too | 01:54 |
| sigrid | cw is 0x100, ccw is 0x200, mask is 0x400. I expected 0x300 | 01:54 |
| sigrid | maybe it's not "bitmask" but another face culling mode, dunno | 01:55 |
| sigrid | ok, gn. | 01:55 |
| mntmn | n8 sigrid | 01:56 |
| mntmn | huh, i had a foggy memory of this https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3288 | 02:03 |
| mntmn | weird https://gitlab.freedesktop.org/flto/mesa/-/commit/503e9619d269082be620b217edc051ef88f4eb48 | 02:04 |
| mntmn | > HW has no value to cull both faces (setting both CW/CCW bits results in | 02:05 |
| mntmn | only CCW being culled) | 02:05 |
| mntmn | i wonder if that is true for GC7000 | 02:05 |
| erlehmann | i find it mildly amusing that this patch for backface culling is needed for literally culling the back of a face | 02:11 |
| mntmn | hehe | 02:11 |
| mntmn | there appear to be two different patches and none of them got picked up | 02:11 |
| mntmn | i commented in marex' patch to get it some attention again | 02:12 |
| erlehmann | i bet there is a funny commit message to be written for the patch | 02:12 |
| erlehmann | sigrid do you get more fps with your fix in minetest when looking at a wall of glasslike nodes in devtest? i am asking bc not doing backface culling could have affected performance maybe | 02:24 |
| erlehmann | the backfaces of glasslike nodes are not drawn when they are up against another node i think | 02:24 |
| erlehmann | comparet the allfaces node, where each face is drawn every time | 02:24 |
| - erlehmann (QUIT: Quit: Just say no, then the virus can not enter your body without your consent.) (~erle@ip5f5bd566.dynamic.kabel-deutschland.de) | 02:29 | |
| - GNUmoon (QUIT: Remote host closed the connection) (~GNUmoon@gateway/tor-sasl/gnumoon) | 03:01 | |
| + bibliocar (~EricShmar@195.82.99.14) | 03:08 | |
| + mtm (~mtm@c-73-27-62-116.hsd1.fl.comcast.net) | 03:09 | |
| - nsc (QUIT: Ping timeout: 268 seconds) (~nicolas@i5C74452E.versanet.de) | 03:57 | |
| - bibliocar (QUIT: Ping timeout: 256 seconds) (~EricShmar@195.82.99.14) | 03:58 | |
| + nsc (~nicolas@i5C7447EC.versanet.de) | 03:59 | |
| * nsc -> Guest1530 | 04:00 | |
| - Christoph_ (QUIT: Remote host closed the connection) (~Christoph@p54bf60eb.dip0.t-ipconnect.de) | 04:21 | |
| - sts-q (QUIT: Ping timeout: 240 seconds) (~sts-q@212.53.219.208) | 04:43 | |
| + sts-q (~sts-q@212.53.219.144) | 04:50 | |
| - Ar|stote|is (QUIT: Quit: https://quassel-irc.org - Chat comfortably. Anywhere.) (~linx@149-210-32-252.mobile.nym.cosmote.net) | 05:19 | |
| + Ar|stote|is (~linx@149-210-32-252.mobile.nym.cosmote.net) | 05:20 | |
| - Ar|stote|is (QUIT: Client Quit) (~linx@149-210-32-252.mobile.nym.cosmote.net) | 05:21 | |
| + Ar|stote|is (~linx@149-210-32-252.mobile.nym.cosmote.net) | 05:26 | |
| - doppler (QUIT: Quit: doppler) (~doppler@user/doppler) | 05:49 | |
| - lexik_ (QUIT: Remote host closed the connection) (~lexik@171.25.222.230) | 06:01 | |
| + lexik (~lexik@171.25.222.230) | 06:02 | |
| + GNUmoon (~GNUmoon@gateway/tor-sasl/gnumoon) | 07:22 | |
| - vagrantc (QUIT: Quit: leaving) (~vagrant@2600:3c01:e000:21:21:21:0:100e) | 07:25 | |
| sigrid | mntmn: hah, now there are three (if you count the wrong one I did) | 08:05 |
| + Major_Biscuit (~MajorBisc@86-88-79-148.fixed.kpn.net) | 09:08 | |
| josch | mntmn: would you like me to build mesa in reform-debian-packages with http://okturing.com/src/13029/body on top? | 09:16 |
| - qbit (QUIT: Quit: WeeChat 3.3) (~qbit@ns2.suah.dev) | 09:16 | |
| - Major_Biscuit (QUIT: Ping timeout: 268 seconds) (~MajorBisc@86-88-79-148.fixed.kpn.net) | 09:38 | |
| Boostisbetter | Do we know if Reform's default polkit is vulnerable to the pwnkit exploit the media is raving about now? | 09:44 |
| + qbit (~qbit@ns2.suah.dev) | 09:55 | |
| + Major_Biscuit (~MajorBisc@c-001-019-027.client.tudelft.eduvpn.nl) | 10:01 | |
| sigrid | void already had polkit updated yday | 10:15 |
| mntmn | josch: no, lets stick with debian mesa for now, but thanks | 10:21 |
| + erlehmann (~erle@ip5f5bd566.dynamic.kabel-deutschland.de) | 11:01 | |
| josch | Boostisbetter: what is your version of policykit-1? | 11:06 |
| josch | Boostisbetter: you can check whether your version is still affected by comparing it to the list here: https://security-tracker.debian.org/tracker/CVE-2021-4034 | 11:10 |
| sigrid | in case anyone gets bored and wants to try a minimal void linux image on their reform: https://github.com/ftrvxmtrx/void-mklive/releases/tag/mntreform2-20220126 | 11:56 |
| Boostisbetter | the default Reform distro uses lxpolkit. Is it vulernable? When can we apt update? mntmn: not in a rush or anything, but just curious. | 12:03 |
| sknebel | ... the reform runs debian. update your debian, all good | 12:07 |
| Boostisbetter | i was told initially that updating shouldn't be done as things can break... | 12:08 |
| + mjw (~mjw_@2001:1c06:2488:1400:9e5c:8eff:fe8f:a440) | 12:08 | |
| Boostisbetter | and we were all kind of waiting for the ok | 12:09 |
| Boostisbetter | also while I wouldn't consider myself a Linux noob, I am a little nervous when I check for updates and see that over a thousands thing can be updated. | 12:13 |
| Boostisbetter | I'm really worried about stuff breaking. | 12:13 |
| Boostisbetter | as I have my Reform perfectly setup and good to go | 12:14 |
| erlehmann | hey sigrid, i saw you did some tga stuff in the past. would you like to help me get minetest interpret A1R5G5B5 and 16 bit grayscale+alpha pictures correctly? context: https://github.com/minetest/minetest/pull/11978 | 12:18 |
| erlehmann | as you may be able to see from the picture, currently A1R5G5B5 textures have no alpha channel and 16 bit grayscale+alpha is erroneously interpreted as A1R5G5B5 | 12:19 |
| sigrid | does that involve writing c++? | 12:31 |
| sigrid | tbh almost nothing supports all types of tga because it's a mess | 12:32 |
| sigrid | 9front does. | 12:32 |
| Boostisbetter | well, I just said f-it and set the thing to update | 12:39 |
| Boostisbetter | guess we'll see if I had any reason to worry | 12:39 |
| Boostisbetter | I did see the note about llvm so I shouldn't have any problems there. | 12:39 |
| Boostisbetter | If I accidentally overwrite the Reform keymap, how can I put back in the needed one? Is it something i can just pull off the SD card that came with the Reform? (I am using an SSD and not the SD) | 12:41 |
| mntmn | Boostisbetter: keymap for console or sway (or which DE are you using)? | 12:41 |
| mntmn | plomlompom: also ich hab hier ein image, vmtl irgendwie leicht verändert, wo es das delay nicht gibt, wenn ich 1 neuen user make | 12:43 |
| mntmn | sorry, wrong channel (still relevant tho lol) | 12:44 |
| Boostisbetter | mntmn: I use sway | 12:47 |
| mntmn | Boostisbetter: then you configured your keymap in .config/sway/config (look for "input" lines) | 12:47 |
| Boostisbetter | ok, then update isn't touching that. I am running an apt upgrade right now, and it was asking me about a keymap, and while I wanted to select keep it accidentally went to manually select and then continued updating. | 12:49 |
| Boostisbetter | which is why I was asking, just to be prepared to need to find it. | 12:49 |
| mntmn | Boostisbetter: yeah, it might update the console keymap | 12:52 |
| Boostisbetter | mntmn: if it does that, is it easy to get it back in place? | 12:52 |
| mntmn | Boostisbetter: https://wiki.debian.org/Keyboard#Keyboard_configuration | 12:53 |
| Boostisbetter | mntmn: vielen Dank! | 12:54 |
| Boostisbetter | I am about finished updating and will be restarting, fingers crossed. | 12:55 |
| mntmn | Boostisbetter: did you move away from our /usr/local stuff already? | 12:56 |
| Boostisbetter | no I did not | 12:57 |
| mntmn | Boostisbetter: ok i can explain it later, i am really stressed atm trying to finish sysimage v3 | 12:57 |
| Boostisbetter | mntmn: no worries. I don't think it will prevent me booting or anything like that so it should be fine. | 12:58 |
| mntmn | well, there was some breakage but i don't remember if sway will still come up | 12:58 |
| mntmn | Boostisbetter: anyway, if you get stuck: https://source.mnt.re/reform/reform-system-image/-/issues/15#note_1601 | 12:59 |
| mntmn | Boostisbetter: and then install sway via apt | 12:59 |
| mntmn | Boostisbetter: then remove "deb https://mntre.com/reform-debian sid/" from /etc/apt/sources.list | 13:01 |
| mntmn | Boostisbetter: then instead add "deb [trusted=yes] https://mntre.com/reform-debian-repo reform main" | 13:01 |
| mntmn | Boostisbetter: then apt update, then apt install xwayland (should pull in patched xwayland from our new repo) | 13:01 |
| - mtm (QUIT: Ping timeout: 256 seconds) (~mtm@c-73-27-62-116.hsd1.fl.comcast.net) | 13:03 | |
| Boostisbetter | mntmn: yeah it is weird, it rebooted and I got back into sway but the waybar doesn't work. I'll look through all that you've said here when I get home. Thanks! | 13:06 |
| mntmn | Boostisbetter: also in your .config/sway/config, find instances of /usr/local and remove the local | 13:06 |
| + Christoph_ (~Christoph@p4fe73f85.dip0.t-ipconnect.de) | 13:55 | |
| - Major_Biscuit (QUIT: Ping timeout: 240 seconds) (~MajorBisc@c-001-019-027.client.tudelft.eduvpn.nl) | 14:18 | |
| + Major_Biscuit (~MajorBisc@86-88-79-148.fixed.kpn.net) | 14:25 | |
| mntmn | josch: do you have a suggestion of how to split out reform-tools from the huge reform-debian-packages build? it's too unweildly for me to rebuild all those packages if i just want to make small updates to the package | 14:31 |
| mntmn | josch: and i'm very time constrained atm so i need a quick solution to push a new reform-tools update | 14:31 |
| mntmn | josch: also, that repo is not currently building because of some problem building linux https://source.mnt.re/reform/reform-debian-packages/-/jobs/547 | 14:35 |
| mntmn | josch: can we remove the linux kernel from the build for now as it is not synced with the linux build in reform-system-image? | 14:35 |
| josch | mntmn: the reform-tools package from reform-debian-packages is built the same way as it is in reform-system-image, so just by calling dpkg-deb. Have a look into build.sh for the exact call. | 15:01 |
| mntmn | josch: yeah, i did a local workaround now by building it in reform-system-image | 15:03 |
| josch | mntmn: yes, the linux kernel isn't building right now because linux in Debian unstable fails to build. This is not because of our patches to it but because dwarves was renamed to pahole and the empty transitional package is m-a:foreign, so the :native dependency cannot be resolved. I already filed a merge request to fix this: https://salsa.debian.org/kernel-team/linux/-/merge_requests/445 | 15:03 |
| mntmn | josch: oof! | 15:03 |
| josch | yes, since you want to build kernel 5.17 we have to remove the kernel from reform-debian-packages | 15:04 |
| mntmn | josch: yeah, as a temporary solution that would be great | 15:04 |
| mntmn | josch: there's already changes/fixes in 5.17 that we would like to have | 15:05 |
| josch | okay | 15:05 |
| josch | mntmn: can I add linux back once it builds 5.17 with your patches from reform-system-image? | 15:07 |
| mntmn | josch: yes, sure! | 15:07 |
| mntmn | https://source.mnt.re/reform/reform-debian-packages/-/merge_requests/1 | 15:07 |
| josch | mntmn: Cool! But as far as I'm concerned you can just push to main directly. :) | 15:09 |
| - erlehmann (QUIT: Ping timeout: 256 seconds) (~erle@ip5f5bd566.dynamic.kabel-deutschland.de) | 15:09 | |
| + mtm (~mtm@c-73-27-62-116.hsd1.fl.comcast.net) | 15:10 | |
| mntmn | josch: haha ok, will merge | 15:10 |
| josch | I also just removed building linux so the pipeline should pass again. | 15:10 |
| mntmn | cool | 15:34 |
| mntmn | thanks | 15:34 |
| mntmn | interesting https://www.cnx-software.com/2022/01/26/directfb2-project-brings-back-directfb-graphics-library-for-linux-embedded-systems/ | 15:47 |
| mntmn | now with GLES support | 15:47 |
| Boostisbetter | So the only issue I am having after following the instructions is that I can get waybar to start. It complains that there is an error loading libfmt.so.7 | 15:55 |
| sigrid | oh wow, directfb is back from the dead? | 15:59 |
| mntmn | yeah | 16:01 |
| mntmn | Boostisbetter: how is waybar started in your sway config? | 16:01 |
| mntmn | Boostisbetter: maybe you are still starting /usr/local/bin/waybar? you should also delete it | 16:02 |
| Boostisbetter | mntmn: | 16:03 |
| Boostisbetter | bar { | 16:03 |
| Boostisbetter | swaybar_command waybar | 16:03 |
| Boostisbetter | } | 16:03 |
| mntmn | Boostisbetter: ok that doesn't specify a path, so i assume it calls /usr/local/bin/waybar, delete it | 16:04 |
| Boostisbetter | mntmn: the line which calls exec waybar pointing to local was already commented out with # | 16:04 |
| mntmn | Boostisbetter: fine, but it doesn't matter | 16:05 |
| Boostisbetter | Mntmn: meaning I should still delete it? | 16:06 |
| mntmn | Boostisbetter: yes | 16:06 |
| mntmn | Boostisbetter: and apt install waybar if you haven't | 16:06 |
| + erlehmann (~erle@ip5f5bd566.dynamic.kabel-deutschland.de) | 16:10 | |
| Boostisbetter | mntmn: Yeah I installed it because I saw that in the scripts notes | 16:10 |
| Boostisbetter | but still getting the libfmt.so.7 error | 16:10 |
| - erlehmann (QUIT: Read error: Connection reset by peer) (~erle@ip5f5bd566.dynamic.kabel-deutschland.de) | 16:11 | |
| mntmn | Boostisbetter: `which waybar` resolves to /usr/bin/waybar, yes? | 16:15 |
| - GNUmoon (QUIT: Remote host closed the connection) (~GNUmoon@gateway/tor-sasl/gnumoon) | 16:15 | |
| mntmn | Boostisbetter: any results for `ldd /usr/bin/waybar | grep local` ? | 16:15 |
| Boostisbetter | mntmn: no it resolves to local | 16:16 |
| mntmn | Boostisbetter: ok but... didn't you just delete that? can you restart your shell? | 16:16 |
| Boostisbetter | I deleted it from the config file | 16:16 |
| mntmn | argh | 16:16 |
| mntmn | i meant `rm /usr/local/bin/waybar` | 16:16 |
| mntmn | it is shadowing the installation from debian | 16:17 |
| Boostisbetter | mntmn: ahh | 16:17 |
| Boostisbetter | one second | 16:17 |
| Boostisbetter | ok restarting now | 16:18 |
| Boostisbetter | mntmn: that did the trick. Now the only other thing I see amiss is that rofi no longer shows the app your selected as being highlighted. | 16:20 |
| mntmn | Boostisbetter: good that you mentioned that, i forgot to include that fix with the system image v3 | 16:21 |
| mntmn | Boostisbetter: the fix is to remove the 2 occurences of "element" in .local/share/rofi/themes/mnt-reform.rasi | 16:21 |
| mntmn | ah i did not forget it | 16:22 |
| mntmn | Boostisbetter: the file has to look like this https://source.mnt.re/reform/reform-system-image/-/blob/sysimage-v3/reform2-imx8mq/template-skel/.local/share/rofi/themes/mnt-reform.rasi | 16:22 |
| Boostisbetter | yep was element, and that fixed it! | 16:24 |
| Boostisbetter | mntmn: so basically we are all set for just apt update apt upgrade in the future? | 16:24 |
| Boostisbetter | provided various components updates don't break something else of course. | 16:24 |
| Boostisbetter | mntmn: honestly thank you for your help!!! I can't believe have attached to the Reform I have become. I love this thing!!! | 16:26 |
| mntmn | Boostisbetter: yes, i believe apt upgrade should be fine now | 16:27 |
| mntmn | Boostisbetter: you're welcome! | 16:27 |
| + erlehmann (~erle@ip5f5bd566.dynamic.kabel-deutschland.de) | 16:27 | |
| mntmn | Boostisbetter: now i need to package these instructions into a guide or better a script, so people can upgrade... | 16:27 |
| Boostisbetter | Ruff: the upgrading had broken the suspend script. | 16:28 |
| - mtm (QUIT: Ping timeout: 240 seconds) (~mtm@c-73-27-62-116.hsd1.fl.comcast.net) | 16:28 | |
| mntmn | Boostisbetter: how so? | 16:29 |
| mntmn | i am shipping the new suspend script/method with the new image btw | 16:29 |
| Boostisbetter | I think it has to do with local. | 16:29 |
| mntmn | Boostisbetter: how so? | 16:29 |
| Boostisbetter | mntmn: well I believe I set mine up with /usr/local | 16:31 |
| mntmn | ok | 16:31 |
| Boostisbetter | mntmn: I was thinking it might have to do with that since we are not really using that | 16:31 |
| + GNUmoon (~GNUmoon@gateway/tor-sasl/gnumoon) | 16:32 | |
| Boostisbetter | Ruff: any tips on getting that working again? | 16:35 |
| mntmn | Boostisbetter: well, what exactly is the problem? | 16:36 |
| Boostisbetter | mntmn: it will suspend like it should but does not resume. It hangs everytime | 16:37 |
| mntmn | Boostisbetter: how are you triggering the suspend? directly with the script? | 16:38 |
| mntmn | Boostisbetter: or with systemctl suspend? | 16:38 |
| mntmn | new system-image build (beta) https://source.mnt.re/reform/reform-system-image/-/jobs/554/artifacts/browse/reform2-imx8mq/ | 16:49 |
| Boostisbetter | mntmn: with systemctl suspend set to a keyboard shortcut via sway config | 16:59 |
| Boostisbetter | mntmn: sorry for the delay was getting some spaghetti started | 17:00 |
| mntmn | Boostisbetter: so what's in /etc/systemd/system/reform-sleep.service? does it look right? | 17:00 |
| mntmn | esp. ExecStart and ExecStop | 17:01 |
| Boostisbetter | mntmn: | 17:06 |
| Boostisbetter | ExecStart=/usr/local/bin/standby suspend | 17:06 |
| Boostisbetter | ExecStop=/usr/local/bin/standby resume | 17:06 |
| mntmn | Boostisbetter: and does /usr/local/bin/standby exist? | 17:06 |
| Boostisbetter | mntmn, | 17:06 |
| Boostisbetter | mntmn: yes it does | 17:06 |
| Boostisbetter | mntmn: when i push the keyboard combo it does suspend | 17:06 |
| mntmn | Boostisbetter: well, it would suspend also without that script, i think | 17:07 |
| mntmn | but without doing any of the preparations therein | 17:07 |
| Boostisbetter | mntmn: well I enabled the terminal so I can see what is happening, and it seemed like it was resuming correctly. The lockscreen even came up, but keyboard and trackball were unresponsive then. | 17:07 |
| mntmn | huh | 17:08 |
| mntmn | hmm, that's new/weird | 17:08 |
| Boostisbetter | mntmn: indeed. I'll keep futzing around with it and see if I can make any progress. | 17:11 |
| mntmn | interesting https://github.com/nwg-piotr/nwg-shell | 17:24 |
| sknebel | "It's aimed at those, who do not believe that the rationale of the modern computer is to use less than 200 MiB RAM and 0.5% CPU." lol | 17:25 |
| sigrid | suspend/resume seems to work fine for me besides one thing: | 17:25 |
| sigrid | [ 371.709693] ath: phy1: Unable to reset channel, reset status -22 | 17:25 |
| sigrid | [ 371.747298] ath: phy1: DMA failed to stop in 10 ms AR_CR=0xffffffff AR_DIAG_SW=0xffffffff DMADBG_7=0xffffffff | 17:25 |
| sigrid | [ 372.211247] ath: phy1: Chip reset failed | 17:25 |
| sigrid | rmmod helps but that's not great | 17:26 |
| mntmn | sigrid: ah, we usually remove the ath from pci and then rescan pci after resume | 17:26 |
| mntmn | sigrid: to work around this problem | 17:26 |
| sigrid | ok, so it's known? | 17:26 |
| sigrid | thought I was doing something wrong :) | 17:26 |
| mntmn | sigrid: https://source.mnt.re/reform/reform-system-image/-/blob/sysimage-v3/reform2-imx8mq/reform-tools_1.0-8/usr/sbin/reform-standby#L22 | 17:27 |
| mntmn | sigrid: yes, known problem | 17:27 |
| sigrid | ah, ok | 17:27 |
| Boostisbetter | mntmn: well I just resumed from suspend and it worked as before, so who knows. Maybe I just needed a few reboots. | 17:31 |
| Boostisbetter | mntmn: nwg-shell is intersting, but I think with the 4gb that the Reform has, my WM using as little memory as possible is a good thing. Still love the idea. | 17:33 |
| + mtm (~mtm@c-73-27-62-116.hsd1.fl.comcast.net) | 18:23 | |
| + vagrantc (~vagrant@2600:3c01:e000:21:21:21:0:100e) | 18:35 | |
| - mjw (QUIT: Quit: Leaving) (~mjw_@2001:1c06:2488:1400:9e5c:8eff:fe8f:a440) | 18:35 | |
| - Major_Biscuit (QUIT: Ping timeout: 256 seconds) (~MajorBisc@86-88-79-148.fixed.kpn.net) | 19:50 | |
| + mjw (~mark@gnu.wildebeest.org) | 19:55 | |
| - Ar|stote|is (QUIT: Read error: Connection reset by peer) (~linx@149-210-32-252.mobile.nym.cosmote.net) | 20:01 | |
| + Ar|stote|is (~linx@149-210-32-252.mobile.nym.cosmote.net) | 20:10 | |
| mntmn | new beta image: https://source.mnt.re/reform/reform-system-image/-/jobs/555/artifacts/file/reform2-imx8mq/reform-system.img.gz | 20:10 |
| + Major_Biscuit (~MajorBisc@2a02:a461:129d:1:193d:75d8:745d:e91e) | 20:26 | |
| Boostisbetter | and just accomplished the 2nd resume from standby after upgrading. I think everything is back to 100% normal. | 20:28 |
| - Major_Biscuit (QUIT: Ping timeout: 250 seconds) (~MajorBisc@2a02:a461:129d:1:193d:75d8:745d:e91e) | 20:30 | |
| + Major_Biscuit (~MajorBisc@2a02:a461:129d:1:193d:75d8:745d:e91e) | 20:32 | |
| - Major_Biscuit (QUIT: Ping timeout: 252 seconds) (~MajorBisc@2a02:a461:129d:1:193d:75d8:745d:e91e) | 20:37 | |
| mntmn | with the new system image, chromium has no problem handling 1080p fullscreen youtube playback | 20:45 |
| mntmn | cool | 20:45 |
| dj-death | nice | 20:45 |
| dj-death | is it just updated kernel? | 20:45 |
| technomancy | what about firefox? | 20:55 |
| mntmn | firefox is slow | 21:05 |
| mntmn | dj-death: not sure which parts of the stack make it fast | 21:05 |
| mntmn | i mean, firefox is kinda ok with layer acceleration force-enabled, but too slow for me to be fun | 21:06 |
| Boostisbetter | mntmn: does doing an apt upgrade get us that kernel after pointing towards the current mntre repo? | 21:09 |
| mntmn | Boostisbetter: no | 21:09 |
| mntmn | i will provide a migration script/guide in the next days | 21:10 |
| * Guest1530 -> nsc | 21:43 | |
| * nsc -> Guest2923 | 21:44 | |
| - GNUmoon (QUIT: Ping timeout: 276 seconds) (~GNUmoon@gateway/tor-sasl/gnumoon) | 22:10 | |
| Boostisbetter | mntmn: most excellent | 22:23 |
| Boostisbetter | ruff: I would like to pick your brain about possible reasons why the suspend script is not nearly as reliable after updating. | 22:23 |
| erlehmann | it is indeed interesting why ff is so slow on reform, given that it is fast on old thinkpads still | 22:45 |
| technomancy | huh, I never installed chromium; I just assumed it was roughly as slow as firefox | 22:45 |
| technomancy | gotta say not having chromium installed feels *really* good | 22:46 |
| technomancy | but having a fast browser might feel better =( | 22:47 |
| mntmn | technomancy: give it a try :3 | 22:53 |
| technomancy | but I don't like google =( | 22:54 |
| technomancy | j/k I'll see how it feels | 22:54 |
| kfx | I don't have a problem with firefox's speed, but I do all kinds of stuff like run ublock origin and localcdn, and disable serviceworkers entirely | 22:54 |
| kfx | sure I don't get push notifications but then I have fre cores | 22:54 |
| kfx | s/fre/free/ | 22:55 |
| technomancy | wasn't chromium going to nerf ublock origin or something? | 22:55 |
| erlehmann | yeah | 22:56 |
| erlehmann | ad company nerfs adblocking | 22:56 |
| erlehmann | not sus at all | 22:56 |
| technomancy | you'd hope that's something they could break in chrome and un-break in chromium but I don't suppose we are that lucky | 22:57 |
| erlehmann | well | 22:57 |
| erlehmann | chromium is basically throwing source out there | 22:57 |
| vkoskiv | I wish Apple bit the bullet and went back to distributing Safari for other platforms. | 22:58 |
| + GNUmoon (~GNUmoon@gateway/tor-sasl/gnumoon) | 22:58 | |
| vkoskiv | They have the money, and they know just as well as we do that browser diversity is vitally important to a healthy web. | 22:58 |
| erlehmann | apple, the company known for caring about other platforms so much? ^^ | 22:58 |
| vkoskiv | They used to do it. You could get Safari for Windows. | 22:58 |
| technomancy | I mean, that's basically why I'm still on firefox; mozilla is inept but at least they're not outright hostile. | 22:58 |
| vkoskiv | Yeah. I did use ungoogled-chromium, but didn't like to use software compiled and distributed by *some dude* | 22:59 |
| vkoskiv | So Firefox it is. | 22:59 |
| vkoskiv | I find it mostly inoffensive, so I don't mind using it. But I'd prefer to have one or more viable alternatives as well. | 22:59 |
| - mtm (QUIT: Ping timeout: 240 seconds) (~mtm@c-73-27-62-116.hsd1.fl.comcast.net) | 23:00 | |
| + mtm (~mtm@c-73-27-62-116.hsd1.fl.comcast.net) | 23:00 | |
| technomancy | hm; ok, yeah this is legitimately a lot faster. | 23:01 |
| technomancy | not sure whether to be happy or sad about that | 23:01 |
| mntmn | technomancy: same | 23:03 |
| technomancy | thanks for the tip! | 23:03 |
| mntmn | somebody should tell mozilla | 23:03 |
| technomancy | I bet the freescale people all use chrome and optimize for it | 23:07 |
| mntmn | that's kinda true, but also mozilla decided to ignore embedded systems | 23:08 |
| mntmn | while chromium devs did not | 23:08 |
| mntmn | and there's a lot of open source work around chromium embedded stuff | 23:09 |
| mntmn | or, webkit | 23:09 |
| mntmn | for example https://webkit.org/wpe/ | 23:09 |
| mntmn | mozilla just doesn't do things like that | 23:09 |
| mntmn | (except maybe when they had the firefox phone) | 23:09 |
| technomancy | huh, interesting | 23:10 |
| mntmn | https://wpewebkit.org/ | 23:10 |
| mntmn | > WPE powers hundreds of millions of embedded devices | 23:10 |
| mntmn | another strategic blunder by mozilla | 23:10 |
| mntmn | https://wpewebkit.org/about/supported-hardware.html | 23:11 |
| technomancy | I knew my days avoiding googleware were numbered; I just thought I had maybe another year or two before FF really crashed and burned | 23:11 |
| mntmn | WPE is very performant, but there is no full blown shell/UI for it afaik, you would need to roll it yourself | 23:11 |
| technomancy | firefox on android is just as good as chromium tho | 23:12 |
| mntmn | there is cog though | 23:12 |
| mntmn | technomancy: depends on the phone i think? i use ff on my fairphone and it seems slower than chrome | 23:12 |
| mntmn | (it's fast enough though) | 23:12 |
| erlehmann | oh wow https://bugzilla.mozilla.org/show_bug.cgi?id=1552132 | 23:14 |
| erlehmann | > This is super fast on my M1 MBP, so I will assume this is fixed by the new wasm optimizing backend for arm64. | 23:14 |
| erlehmann | mntmn, they do not even have arm64 CI apparently for linux? https://bugzilla.mozilla.org/show_bug.cgi?id=1677963 | 23:15 |
| mntmn | erlehmann: > 3 years ago | 23:15 |
| erlehmann | > Arm64 builds for Linux have been available on CI since bug 1532952. They're not tested, and not shippable builds, but they exist. | 23:16 |
| kfx | "this is super fast on my desktop class processor known for being fast" | 23:26 |
| kfx | now THAT'S the mozilla bugfixing ethos I know so well! | 23:26 |
| kfx | (I think I'm allowed to snark because I'm in about:credits, right?) | 23:27 |
| mntmn | haha | 23:27 |
| mntmn | granted | 23:27 |
| mntmn | i wonder if ff webrender works with the vivante blob | 23:27 |
| mntmn | but then, probably a nightmare to set it all up | 23:27 |
| mntmn | (the vivante blob can do GLES 3) | 23:28 |
| mntmn | technomancy: btw, you use "layers.acceleration.force-enabled = true" in FF, yes? | 23:31 |
| technomancy | it's not on; lemme try that | 23:33 |
| mntmn | that will make things a bit faster | 23:36 |
| sigrid | if I run iwd instead of wpa_supplicant and do "echo 1 > /sys/bus/pci/devices/0000:01:00.0/reset 2>/dev/null" on resume from suspend, I get just a few ath9k errors in dmesg, the bonus is that the network connection is still intact | 23:41 |
| sigrid | so far I got no hangs at all. just once keyboard and mouse didn't work, luckily I could ssh and go back into suspend/resume and it worked again | 23:42 |
| mntmn | sigrid: neat | 23:45 |
| mntmn | sigrid: if keyboard and mouse don't work, maybe the usb hub can be reset as a workaround | 23:45 |
| mntmn | technomancy: i just tested 1080p fullscreen youtube playback in ff96 with accelerated layers, it works ok, the JS ui stuff is not as fast as chromium but ok | 23:47 |
| sigrid | for me firefox full hd video eats 100% cpu time | 23:49 |
| mntmn | yeah | 23:49 |
| sigrid | the same video with mpv or ffplay is much better | 23:49 |
| mntmn | software decöde | 23:49 |
| sigrid | I am sure I have no hw decoding though | 23:50 |
| mntmn | yeah if you have the patched ffmpeg those will then use hw decode | 23:50 |
| mntmn | ah ok | 23:50 |
| mntmn | mpv is also better at sw decoding for sure | 23:50 |
| mntmn | or, at getting those pixels to the display without much blitting | 23:50 |
| sigrid | I just play youtube urls with mpv directly at this point | 23:53 |
| sigrid | I wonder how firefox decodes the video, whether it does it in separate threads with ffmpeg, and how exactly do decoded frames end up in firefox window | 23:54 |
| sigrid | maybe there is too much data shoved around | 23:54 |
| kfx | the secret to fast video in firefox is in libva | 23:57 |
| kfx | I can no longer remember which versions support it by default and which versions turn it off by default and under which gui system (x or wayland) | 23:57 |
| sigrid | yeah, libva except it seems all broken for me here | 23:57 |
| sigrid | I have this: | 23:58 |
| sigrid | [ 19.289573] hantro_vpu: module is from the staging directory, the quality is unknown, you have been warned. | 23:58 |
| sigrid | [ 19.307253] hantro-vpu 38300000.video-codec: registered nxp,imx8mq-vpu-dec as /dev/video0 | 23:58 |
| sigrid | but none of the libva forks have been able to do anything with it | 23:58 |
| sigrid | partially because they haven't been updated to new kernels (which had a lot of v4l2 changes) | 23:59 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!