| epilys | hello everyone | 00:05 |
|---|---|---|
| epilys | has rust been tried/run on maemo? | 00:05 |
| dsc_ | most just went to bed before you came in I think :) | 00:21 |
| epilys | my irc bouncer will be here all night | 00:24 |
| missMyN900 | hi everyone - I used to own an N900 and now have a Pinephone (3 GB edition) with Maemo Leste on a microSD and postmarketOS Plasma on the eMMC | 00:59 |
| missMyN900 | I have been playing with ML for some months now and was amazing to be able to see and interact with Hildon again after all these years :) | 01:00 |
| missMyN900 | I have some questions, observations and suggestions | 01:02 |
| missMyN900 | I have been trying a lot of web browsers to try to find something that is somewhat usable for at least a little bit of daily web browsing but I have not found anything that works on ML | 01:03 |
| missMyN900 | on pmOS the best I have found is Falkon (which I also use on my Debian desktop), which has good touch screen support and is quite fast | 01:03 |
| missMyN900 | unfortunately Falkon cannot be installed currently on ML/Devuan because it depends on kf5-walletmanager, which cannot be installed for some reason (systemd dependence?) | 01:04 |
| missMyN900 | I saw on GitHub that there were some plans to hildonize surf IIRC | 01:05 |
| missMyN900 | I would suggest modifying netsurf-gtk instead. 3.10 has a new GTK GUI (both 3 and 2 are supported, by the way) and has been packaged for chimaera, which is quite suitable for small screens | 01:07 |
| missMyN900 | the only thing it needs is support for touch screen scrolling and pinch to zoom. I think it would be a better candidate for the less powerful Maemo Leste devices (especially N900) as well due to its extremely low RAM consumption | 01:08 |
| missMyN900 | furthermore, if desired it can be hildonized more easily because it can still be compiled against GTK2 | 01:09 |
| Wizzup | How modern is netsurf? | 01:21 |
| missMyN900 | @Wizzup in what way? | 01:23 |
| missMyN900 | it is under active development | 01:23 |
| Wizzup | Does it support js? | 01:23 |
| missMyN900 | JS is experimental and disabled by default | 01:24 |
| missMyN900 | still, I thnk it would make for a good default browser. People can always use Falkon (or perhaps Firefox with patches) for websites that do not work in netsurf | 01:25 |
| missMyN900 | the web experience is going to be limited on the N900, Droid 4 etc anyway | 01:25 |
| missMyN900 | when I was using Opera Mini (or was it Mobile?) back in 2013 there were already some issues | 01:25 |
| missMyN900 | same for MicroB, of course | 01:26 |
| missMyN900 | I cannot imagine running a modern WebKit or Blink based browser with 256 MB of RAM | 01:26 |
| missMyN900 | running Chromium on OpenBSD on my 2 GB laptop was already not great | 01:27 |
| missMyN900 | with a few tabs open it would start to swap | 01:27 |
| missMyN900 | @Wizzup you can try netsurf-gtk on your desktop/laptop if you run Debian 11 or Devuan 4.0 at least | 01:28 |
| Wizzup | tmlind: wonder if this is useful - reset on 5.15: https://dpaste.com/2YZ4DJ6JP.txt | 01:33 |
| Wizzup | surf is webkit based afaik | 01:33 |
| missMyN900 | netsurf is great for things like wiki pages, documentation and mobile news websites (e.g. mobile.reuters.com) | 01:36 |
| missMyN900 | sorry my WiFi is working poorly. Maybe switching to connman was not such a good idea after all... | 01:37 |
| * Wizzup zzz | 01:38 | |
| Wizzup | ttyl | 01:38 |
| missMyN900 | there is some glitch on the Pinephone where the time (minutes) keeps flickering back and forth under CPU load (when I web page is loading) | 02:33 |
| missMyN900 | resolved my WiFi issues by getting rid of wpa_supplicant and using iwd with connman instead. Now I can even connect to the 5 GHz networks again, which was broken even when I was still using NetworkManager (on Debian 11) | 03:09 |
| missMyN900 | I wonder whether it is possible to use iwd with Maemo Leste as well | 03:10 |
| sunshavi | iwd is a systemd unit | 04:35 |
| missMyN900 | @sunshavi so it cannot be used with any other init system? | 04:39 |
| missMyN900 | I guess I will have to stick with Debian then on my desktop (I was thinking of installing Devuan eventually) because there is no way I am going back to NetworkManager and wpa_supplicant | 04:40 |
| sunshavi | before iwd I was using wifi-menu. It worked fine. But not autoconnect with different dongles | 04:41 |
| missMyN900 | never heard of it | 04:41 |
| missMyN900 | I loved how WiFi works on OpenBSD | 04:41 |
| missMyN900 | so much simpler | 04:41 |
| sunshavi | mmm. on freebsd they do not support a lot of wifi-cards. is it different on openbsd? | 04:42 |
| missMyN900 | my USB WiFi is supported (Ralink RT5592) | 04:42 |
| missMyN900 | I also bought one of the Alfa Wireless adapters with the AR9271 | 04:43 |
| sunshavi | mine is not supported atheros ar971 or something like that | 04:43 |
| missMyN900 | AR9271 has excellent support | 04:43 |
| missMyN900 | https://man.openbsd.org/athn | 04:44 |
| missMyN900 | https://man.openbsd.org/run | 04:44 |
| missMyN900 | OpenBSD is much better than FreeBSD IMHO | 04:45 |
| missMyN900 | FreeBSD quality control is bad; Ars Technica did an article on it and I have heard the same from other people | 04:45 |
| missMyN900 | OpenBSD is also better for desktop use as FreeBSD devs mainly use macOS | 04:46 |
| missMyN900 | biggest problem with OpenBSD is interoperability | 04:46 |
| missMyN900 | there is no ext4 or UTF16 support, for example | 04:46 |
| missMyN900 | there is no way to run a Linux virtual machine and the Linux compatibility was removed some while back | 04:47 |
| missMyN900 | that is why I had to ditch it | 04:47 |
| sunshavi | r u on irc from ur PP? | 04:48 |
| missMyN900 | I ran NetBSD on a Raspberry Pi B+ for a while that I used for some stuff but I have no need for it anymore and got sick of the limitations of the Raspberry Pi | 04:48 |
| missMyN900 | no | 04:48 |
| missMyN900 | I'm on my desktop | 04:48 |
| missMyN900 | is there a good IRC client? | 04:48 |
| missMyN900 | I am using Quassel on my desktop, which is what I used years ago; I haven't been active on IRC for years | 04:48 |
| sunshavi | opi+2e here. I am on erc now | 04:49 |
| missMyN900 | CLI client? | 04:49 |
| sunshavi | erc is part of emacs | 04:49 |
| missMyN900 | Is that an OrangePi? | 04:49 |
| sunshavi | it is cli if u prefer it | 04:50 |
| sunshavi | Right | 04:50 |
| missMyN900 | I only have a B+ and an old Wandboard. I would love to get a newer SBC but can't justify it right now | 04:50 |
| missMyN900 | especially after I got the Pinephone (I was intending to use it for tethering as my cable internet is unreliable sometimes) | 04:50 |
| sunshavi | country? | 04:51 |
| missMyN900 | US | 04:51 |
| sunshavi | PP is my future. But not this year. Still with n900 | 04:51 |
| missMyN900 | I got rid of my N900 because I thought it was slowing down the WiFi network due to it not supporting N :( | 04:52 |
| missMyN900 | I sold it for very little money | 04:52 |
| missMyN900 | I bought it for $100 in like new condition late 2012 | 04:52 |
| missMyN900 | now I have an iPhone :( | 04:52 |
| sunshavi | But You are almost back with ur PP+mleste | 04:53 |
| missMyN900 | well, it is not quite ready yet in my experience | 04:53 |
| missMyN900 | LTE data is not working, for instance | 04:53 |
| missMyN900 | not on any OS | 04:53 |
| missMyN900 | pmOS + Plasma experience is not great either | 04:54 |
| missMyN900 | I barely use the PP | 04:54 |
| missMyN900 | just hoping for LTE to start working and for Maemo Leste to mature | 04:54 |
| sunshavi | be patient. It is still evolving | 04:55 |
| missMyN900 | @sunshavi for sure. I have installed some applications today to try to make the Pinephone a bit more useful right now | 05:09 |
| missMyN900 | @sunshavi I am leaving the ML card inserted for now as I will not be using pmOS anyway as long as LTE data does not work | 05:10 |
| missMyN900 | I think smplayer, xournal and pluma are useful | 05:11 |
| missMyN900 | I hope that battery status support is implemented soon for the Pinephone. There are also graphical glitches that need fixing. | 05:12 |
| sicelo | sunshavi: iwd doesn't necessarily need systemd. at least the pmOS people use it (and they definitely have no systemd) | 07:33 |
| sicelo | missMyN900: LTE data not working on PP in any OS? That's interesting. I thought it works on at least pmOS and Mobian. Anyway, not a PP owner, so I might have misread | 07:34 |
| freemangordon | uvos: seems all other drivers have &dri2ConfigQueryExtension.base in their screen extenstions, no matter that they *do not* announce __DRI2_CONFIG_QUERY in version info | 07:57 |
| missMyN900 | @sicelo: yes, I tried the Manjaro Plasma image my PP came with, the latest pmOS Plasma stable, the latest pmOS Phosh edge and now pmOS Plasma edge and still have not gotten LTE data to work once | 08:02 |
| missMyN900 | text messages and calls worked at least on pmOS Plasma stable | 08:03 |
| missMyN900 | nothing cellular related worked at all on Mobian Phosh | 08:03 |
| missMyN900 | this is with T-Mobile US | 08:03 |
| freemangordon | Wizzup: where is the mesa source code? | 08:03 |
| missMyN900 | @Wizzup TMO works well in netsurf as well :) just checking out your post in the ML alpha thread | 08:04 |
| freemangordon | oh, my remote is wrong | 08:05 |
| missMyN900 | @sicelo oh nice, I did not know pmOS was using iwd as well | 08:07 |
| missMyN900 | I quite like Alpine/pmOS | 08:07 |
| missMyN900 | will be installing Alpine on my old Wandboard over the holidays; I used to run it on my Raspberry Pi some years ago | 08:07 |
| missMyN900 | I don't like Plasma Mobile though and don't get me started on Phosh... | 08:08 |
| missMyN900 | just reading about the Qt5 port: that is awesome. I had no idea that Maemo widgets had been ported. I was wondering what was being done about GTK2 having been deprecated a long time ago (does not feel like it though ;) ) | 08:11 |
| missMyN900 | this reminds me, there is some Qt demo web browser for mobile devices that is available on pmOS (scaling is broken though so it is currently unusable); it even has its own virtual keyboard | 08:12 |
| missMyN900 | packaging and fixing that could be quite a win for Maemo Leste IMHO | 08:13 |
| missMyN900 | found its homepage: https://doc.qt.io/QtWebBrowser/ | 08:14 |
| missMyN900 | it uses QtWebEngine (Blink/Chromium) | 08:15 |
| freemangordon | uvos: going to fix that, lets see if it will affect swaps as well | 08:15 |
| missMyN900 | if that were to be packaged and tested and touch screen support was added to netsurf-gtk, there would be two excellent browsers for Maemo Leste | 08:15 |
| missMyN900 | one traditional mobile browser with full support for the modern web and an ultralightweight alternative for low RAM devices | 08:16 |
| freemangordon | Wizzup: hmm, what is this 0002-pvr-build-with-new-mesa.patch about? I see code changes there. | 08:17 |
| sicelo | missMyN900: maybe ask about LTE on either the mobian or pmos rooms | 08:22 |
| missMyN900 | @sicelo: I will try that some time, thanks | 08:23 |
| sicelo | including other cellular things ... for sure i know there are people who daily drive the pp | 08:23 |
| missMyN900 | yes, I am aware. Unfortunately there are also multiple reports on the forums of cellular related issues :( | 08:24 |
| freemangordon | Wizzup: oh, Isee | 08:25 |
| missMyN900 | I downloaded a PDF today with Firefox on the Pp and moved it to MyDocs with Spacefm and then viewed it with the PDF reader | 08:26 |
| missMyN900 | the PDF reader is working very nicely on Maemo Leste | 08:26 |
| freemangordon | Wizzup: though, I think we shall incorporate those changes in the code, not as patches | 08:27 |
| missMyN900 | based on my brief testing I would recommend Pluma, Xournal and Spacefm for use on Maemo Leste | 08:27 |
| sicelo | we did have xournal on fremantle. i still use it there :-) | 08:27 |
| missMyN900 | it seems to work well on the Pp with Leste | 08:28 |
| missMyN900 | I tried to find some non-Gnome Gtk apps with big buttons | 08:28 |
| missMyN900 | Smplayer works okay as well though | 08:29 |
| missMyN900 | the stylus of the N900 would have been great to have with the Pp... | 08:29 |
| missMyN900 | the hardware keyboard will be a massive improvement though | 08:30 |
| sicelo_ | PING 1637739169 293661 | 08:32 |
| sicelo_ | PING 1637739174 545487 | 08:32 |
| missMyN900 | netsurf really shines on "traditionalist" sites like TMO | 08:35 |
| missMyN900 | Alright, I've got to go | 08:35 |
| sicelo | i think netsurf has been tried on ML, yes | 08:36 |
| uvos | freemangordon: huh strange | 09:39 |
| uvos | missMyN900: on d4 firefox works really well, if you activate all the touch gestures and hide the tab and menu bar, see http://uvos.xyz/maserati/screenshots/2020-06-17-195309_960x540_scrot.png im also hoping to modify http://uvos.xyz/maserati/screenshots/qtwebbrowser.png a bit to make it maybe the default leste browser in the future, at least for devices with 512mb+ ram | 09:44 |
| uvos | on n900 its a bit harder, both ff and qtwebbrowser just get oom killed on startup for me | 09:44 |
| uvos | so something like surf etc will have to do | 09:45 |
| uvos | problem i have with surf is its abysmal scroling performance | 09:45 |
| uvos | freemangordon: btw qwebengine renders black on ddk1.17 | 09:48 |
| uvos | not that this is your fault most likley | 09:48 |
| uvos | [12977:13001:1124/094752.436859:ERROR:gles2_cmd_decoder.cc(2603)] [.RenderWorker-0x4f8fa8]GL ERROR :GL_INVALID_OPERATION : ScopedTextureBinder::dtor: <- error from previous GL command | 09:48 |
| parazyd | uvos: For Firefox we could also provide a user.js that modifies it to be more mobile friendly. | 09:49 |
| uvos | parazyd: yeah but long term i dread the next -esr | 09:50 |
| parazyd | True, it's getting worse | 09:50 |
| uvos | parazyd: ff changed a lot behind the scenes with the replacement of gecko | 09:50 |
| parazyd | There's some alternatives called palemoon or something | 09:51 |
| parazyd | Perhaps worth a look | 09:51 |
| uvos | yeah | 09:51 |
| Wizzup | freemangordon: hi | 11:17 |
| Wizzup | freemangordon: I think you figured out the patches I made to make it build with newer mesa? | 11:18 |
| freemangordon | yes | 11:32 |
| freemangordon | the point is that we shall incorporate them in code, rather than ceeping them n debian/patches | 11:33 |
| freemangordon | *keeping | 11:33 |
| Wizzup | freemangordon: yes, my mesa patches are "wip" in that sense since I didn't expect the rebase to surface these problems | 11:34 |
| Wizzup | so I just stuck them in there since I didn't want to change the git log otherwise (yet) | 11:34 |
| Wizzup | please feel free to take that and apply it | 11:34 |
| freemangordon | ok | 11:34 |
| freemangordon | will do | 11:34 |
| Wizzup | the other one I would leave (the fix for freedreno) | 11:34 |
| Wizzup | mesa master hasn't fixed it yet | 11:34 |
| freemangordon | Wizzup: how to remove your patches? I mean - shall I revert or drop (and overwrite history)? | 11:35 |
| freemangordon | or, make another branch? | 11:36 |
| Wizzup | patch -p1 < debian/patches/foo.patch ; git add -p ; git commit -m "foo" ; git rm debian/patches/foo vi debian/patches/series; git add -p debian/patches/series maybe ? | 11:36 |
| Wizzup | the thing is if you add it to git history you will need to re-tag | 11:36 |
| Wizzup | but that should not be a problem I think :) | 11:37 |
| Wizzup | I don't think you need to keep my work, you can also drop the commits if you want | 11:37 |
| Wizzup | do whatever really | 11:37 |
| freemangordon | Wizzup: I know how to apply the patches by hand, the point is that we have commit messages for them in the history I would like to drom | 11:37 |
| freemangordon | but ok, I'll force-push with patches dropped | 11:38 |
| freemangordon | hmm, mesa failed to build for me :( | 11:38 |
| freemangordon | Wizzup: https://pastebin.com/YX8vqAGd any idea? | 11:39 |
| Wizzup | freemangordon: out of memory? | 11:43 |
| Wizzup | check dmesg for oom killer | 11:43 |
| freemangordon | trying now with a single job only | 11:54 |
| freemangordon | yeah | 11:54 |
| freemangordon | Out of memory: Killed process 14933 (cc1plus) total-vm:600988kB, anon-rss:421664kB, file-rss:304kB, shmem-rss:0kB, UID:0 pgtables:590kB oom_score_adj:0 | 11:54 |
| freemangordon | added 1GB swap, lets see | 11:56 |
| uvos | i see we are still compileing mesa on omap4 :) | 11:58 |
| Wizzup | uvos: he is, I am not, I use bencoh's lxc setup and the CI | 11:58 |
| Wizzup | but yeah the CI does it on raspi4's :p | 11:59 |
| Wizzup | (for now) | 11:59 |
| uvos | do we plan on using lxc on ci? | 11:59 |
| Wizzup | no, I ordered an ~1000 usd powerful arm server | 12:01 |
| Wizzup | this one: https://shop.solid-run.com/product/SRLX216S00D00GE064H09CH/ | 12:01 |
| Wizzup | (well with the ram its 1000 usd) | 12:01 |
| Wizzup | I'll put it in a chassic at my place so it shares the LAN with th rest of the builders | 12:02 |
| uvos | ok great | 12:02 |
| dreamer | fancy | 12:02 |
| Wizzup | mid december or so I'll set it up | 12:02 |
| dreamer | oh only 16 core :P | 12:02 |
| Wizzup | well we got some funding to get the hw, so might as well since it's slowing us down | 12:02 |
| dreamer | ja, good idea | 12:02 |
| uvos | any idea what LX2160A corrisponds to roughly in terms of x86 processors? | 12:02 |
| Wizzup | not really, others might know, we'll find out I supose | 12:02 |
| Wizzup | the CI doesn't run on super fast x86 processors either | 12:03 |
| bencoh | oh, solid-run <3 | 12:03 |
| Wizzup | It runs on AMD Opteron(tm) Processor 6278 | 12:03 |
| Wizzup | bencoh: they're basically the only feasible option atm (from what I could see) | 12:03 |
| uvos | i mean operatorn 62xx is really old | 12:04 |
| Wizzup | yes | 12:04 |
| uvos | bulldozer - outch :P | 12:04 |
| Wizzup | but it fits in my coreboot kdpe-d16 :p | 12:04 |
| uvos | heh | 12:04 |
| Wizzup | kgpe-d16* | 12:04 |
| sunshavi | 64 Gb emmc. I have the same but with a uSD adapter on my SBC | 12:04 |
| Wizzup | sunshavi: I think it has pci or something, so we'll go for with ssd on sata or m.2 | 12:05 |
| bencoh | :) | 12:05 |
| sunshavi | Yes. I have compiled mesa in the past 2 Gb memory was enough for doing an in-memory compilation | 12:05 |
| sunshavi | Wizzup: Do You remember when I was trying to put ML on my SBC. And I failed. It was because of an error with EmmmModule+uSDAdapter. | 12:07 |
| sunshavi | After a couple of lines within mmc-sunxi.c I have been able of booting archlinux on it. So I think I could try soon ML again | 12:08 |
| Wizzup | freemangordon: I'd like to package osso-abook for -devel some time, that ok? | 12:09 |
| Wizzup | I know it's not finished, but it's finished enough to build some pkgs | 12:09 |
| freemangordon | ok | 12:10 |
| freemangordon | lemme see if I have pushed everything so far | 12:13 |
| freemangordon | Wizzup: yep, everything is pushed | 12:17 |
| freemangordon | please, have a look if you can invent a better versioning scheme :) | 12:17 |
| Wizzup | freemangordon: this is about abook? | 12:22 |
| Wizzup | freemangordon: uvos: do you use a powervr.ini on d4 or n900? I moved mine on n900 and I don't think there is a difference | 12:37 |
| Wizzup | sicelo: # sleep 10 ; /etc/init.d/n900-powermanagement status | 12:49 |
| Wizzup | d=2021-11-24|t=13:48:57|i=OFF:0,RET:0|p=408|c=NA|b=ST_SDRC,ST_OMAPCTRL,ST_I2C1,ST_MCSPI4 | 12:49 |
| Wizzup | I think this is useful :-) | 12:49 |
| Wizzup | tmlind: looks like spi4 is wifi, and it does idle some time, i2c1 is twl, but a lot is twl, so that might be hard to narrow | 12:55 |
| Wizzup | I don't know about SDRC and OMAPCTRL really | 12:55 |
| Wizzup | I'm going to figure out how we can add to the bootmenu from our partition(s) | 12:55 |
| Wizzup | and then add emergency mode and such | 12:55 |
| Wizzup | hm: [ 30.164733] rx51-audio n900-audio: snd_soc_register_card failed (-517) | 12:57 |
| Wizzup | https://dpaste.com/8JHZ8A5YF | 12:58 |
| sicelo | mmm, yes, our N900 needs some loving. its dts is getting outdated, it seems | 12:59 |
| Wizzup | sicelo: so you've seen this before I reckon? | 12:59 |
| Wizzup | this might be one for pavel maybe | 12:59 |
| sicelo | i don't think i've seen it. but i've seen similar stuff. most of the time it's because our dts needs updating in line with latest binding specifications, etc. | 13:01 |
| Wizzup | ok | 13:02 |
| Wizzup | so I am thinking that there are some random boot failures which are the real reason the device would reset, and I'm seeing the same with the droid on 5.15, so that's probably some of the problems we're seeing (might be able to use console on the d4 to find that) | 13:03 |
| lel | MerlijnWajer opened an issue: https://github.com/maemo-leste/bugtracker/issues/587 (N900 Linux 5.15) | 13:04 |
| Wizzup | sicelo: if you have time, please add/comment ^ | 13:04 |
| uvos | random boot failures on d4 are "normal" | 13:04 |
| uvos | if you mean right after kexec | 13:04 |
| uvos | thats the 3.0.8 kernels fault | 13:05 |
| Wizzup | it happens more frequently now I'm sure | 13:05 |
| Wizzup | I'll attach serial some time, but there's so much to do :) | 13:05 |
| uvos | dose it vibrate when it happens? | 13:06 |
| uvos | ie is it kexecboot menu->select entry->vibrate->motorola logo | 13:06 |
| uvos | or kexecboot menu->select entry->motorola logo | 13:06 |
| Wizzup | I don't remember, but I am pretty sure after kexecboot menu | 13:07 |
| uvos | sure both of those are after the menu | 13:07 |
| sicelo | i'm surprised nokia modem isn't enabled by default. i am quite sure omap2plus has it enabled ootb | 13:52 |
| freemangordon | Wizzup: I don;t think we shall user powervr.ini on any of the devices | 13:53 |
| freemangordon | it was used because of glamor | 13:53 |
| Wizzup | freemangordon: ok, it is not purged by default | 13:56 |
| Wizzup | sicelo: no @ enabled ootb | 13:57 |
| freemangordon | Wizzup: it should not be | 14:04 |
| Wizzup | freemangordon: right | 14:05 |
| sicelo | ah i see. i was enabling it in my own config | 14:07 |
| Wizzup | yes | 14:08 |
| Wizzup | uvos: parazyd: I want to tweak the omap2plus_defconfig for our purposes a bit like spinal did, making some more essential things built in | 14:08 |
| Wizzup | any suggestion on how to best approach it? | 14:08 |
| Wizzup | in particular keeping in mind upstream omap2plus_defconfig will change with time | 14:08 |
| sicelo | it hardly changes (upstream) | 14:09 |
| uvos | i think its perfectly sane to have a leste_omap_defconfig or whatever | 14:11 |
| uvos | and have just the modules we need for the devices the omap kernel is built for + whatever you want built in | 14:11 |
| uvos | i would not go around patching omap2plus forever | 14:11 |
| uvos | this is quite usual behavior for a distrobution in general | 14:13 |
| uvos | (having an own defconfig) | 14:13 |
| Wizzup | yeah n9xx-linux has n900_defconfig | 14:14 |
| Wizzup | Pali: do you create your bootmenu entries from fremantle still, or do you manage them externally? I'm wondering what the best way for leste to manage this would be | 14:58 |
| Wizzup | I think we probably want some bootmenu.scr file on the sd card | 14:58 |
| Wizzup | I was looking at u-boot/include/configs/nokia_rx51.h but it wasn't quite clear to me | 14:58 |
| Wizzup | reading u-boot doc... | 14:59 |
| Pali | IIRC upstream u-boot read bootmenu.scr also from sd card | 15:01 |
| Wizzup | any clue what has preference? | 15:01 |
| Wizzup | or are they combined somehow? | 15:01 |
| Wizzup | I saw the preboot section | 15:01 |
| Pali | now I checked it | 15:02 |
| Pali | yes, it first loads it from sd card and then fallbacks to emmc | 15:02 |
| Pali | include/configs/nokia_rx51.h | 15:02 |
| Pali | preboot= | 15:02 |
| Wizzup | ok, so if we have it on sd card we better make sure we have fremantle entry to allow booting fremantle with sd card in | 15:02 |
| Pali | yes! | 15:02 |
| Wizzup | parazyd: fyi ^ | 15:03 |
| Wizzup | well, whatever I did, it no longer boots 5.15 kernels anymore, meh | 15:07 |
| Wizzup | interesting, now 5.1 also gives panic trying to find root device | 15:09 |
| sicelo | :-) n900 is always 'fun' when playing with new kernels, haha | 15:11 |
| Wizzup | *not* :) | 15:14 |
| * uvos grumbels over uart ... no one hears him | 15:19 | |
| Wizzup | uvos: ? | 15:19 |
| uvos | its a joke, thats whats happening to n900 its complaining over uart, but no one listens | 15:20 |
| Wizzup | ah yeah | 15:21 |
| Wizzup | well we just need to find a way to easily make a few of these serial adapters | 15:21 |
| sicelo | we are listening actually. just only Wizzup has serial for now. | 15:21 |
| sicelo | i'm still going to go ahead and permanently solder some thin wires onto my old broken on | 15:22 |
| Wizzup | uvos: I don't have any cnc equipment available | 15:23 |
| Wizzup | I guess we could ask some commercial place to make some | 15:23 |
| uvos | Wizzup: im fine with using my equitment, but it dont want to invest the time to design something. | 15:25 |
| uvos | Wizzup: ie give me sane and compleat cad files (ie sres isent either) and i can make you some parts. | 15:26 |
| uvos | we have milling, turning, fdm printing and laser cutting available | 15:26 |
| uvos | with 3,2,3,2 axies | 15:27 |
| uvos | respectively | 15:27 |
| Wizzup | with drm built in at least I quickly see a penguin, but seems console is not on that tty - weird | 15:29 |
| Wizzup | maybe something with run sdboot | 15:29 |
| Wizzup | uvos: hm, ok... | 15:29 |
| uvos | theres no better time to learn freecad than now :P | 15:54 |
| uvos | Wizzup: is power duplicated on the pogo landing pads? | 15:54 |
| uvos | Wizzup: if so it should be realy easy, just mill out a battery shaped object with some holes in the right places to glue pogo pins | 15:55 |
| Wizzup | the serial module I have I tihnk uses the battery pads | 15:58 |
| Wizzup | s/pads/connector/ | 15:58 |
| uvos | probubly not then | 15:58 |
| uvos | well designing something to fit the battery connector will be a bit more dificult | 15:59 |
| uvos | unless the n900 can be powerd via only usb via some trick like moto a85x | 15:59 |
| uvos | any how im happy to help with any other infomation you might need, eg desin rules etc. | 16:00 |
| uvos | or hints on tools etc | 16:01 |
| Wizzup | yeah, I'm not going to volunteer to do it since it will take me weeks probably | 16:06 |
| uvos | Wizzup: ok yeah makes sense | 16:07 |
| freemangordon | Wizzup: I think I am ready with mesa fixes | 16:08 |
| freemangordon | doing one last test and will force-push | 16:08 |
| uvos | what are we fixing? | 16:08 |
| freemangordon | may I leave tagin/versioning to you? | 16:08 |
| freemangordon | segfault | 16:08 |
| uvos | great :) | 16:08 |
| freemangordon | I have glx enabled in omap.conf and no segfault anymore | 16:09 |
| freemangordon | enabled like "not disabled" :) | 16:09 |
| Wizzup | freemangordon: do you mean to me? | 16:09 |
| freemangordon | yes | 16:09 |
| uvos | it falls back to llvm/swrast or dose it fail to create a surface? | 16:09 |
| freemangordon | uvos: no idea, how to check? | 16:09 |
| Wizzup | freemangordon: ok, I'll make a note and try to remember but I am fighting a losing fight with my n900 and kernrel | 16:09 |
| uvos | glxgears? | 16:09 |
| freemangordon | Wizzup: ok | 16:09 |
| freemangordon | uvos: sec | 16:09 |
| freemangordon | uvos: do you know which pacjage is that? | 16:10 |
| freemangordon | *package | 16:10 |
| uvos | mesa-demos | 16:10 |
| uvos | iirc | 16:10 |
| freemangordon | hmm, maybe mesa-tools | 16:10 |
| freemangordon | or mesa-utils :) | 16:10 |
| uvos | dpkg -S /usr/bin/glxgears | 16:11 |
| uvos | mesa-utils: /usr/bin/glxgears | 16:11 |
| uvos | yeah | 16:11 |
| freemangordon | mhm | 16:11 |
| freemangordon | Error: glXCreateContext failed | 16:12 |
| Wizzup | great | 16:12 |
| uvos | freemangordon: ok thats fine | 16:12 |
| freemangordon | cool | 16:12 |
| uvos | freemangordon: but ddk1.9 fell back to llvm | 16:13 |
| uvos | witch is slightly bettermaybe | 16:13 |
| Wizzup | I don't think we want that really | 16:13 |
| Wizzup | if we do anything here at all we should use gl4es | 16:13 |
| uvos | idk sometimes soemthing working at all is better than it being slow | 16:13 |
| Wizzup | things falling back to llvmpipe for "acceleration" is just an outright nightmare | 16:13 |
| uvos | you can still block glx if you want in xorg.conf | 16:13 |
| uvos | but yeah its fine for now | 16:13 |
| uvos | it might be because of the pvr override acctually | 16:15 |
| Wizzup | weird, modules_install does not install pvrsrvkm and related modules to the path for me | 16:24 |
| Wizzup | that explains why my n900 resets after trying to start X (without anything in dmegs) | 16:24 |
| Wizzup | hmm | 16:25 |
| uvos | lifeguard strikes again? :P | 16:26 |
| Wizzup | no | 16:26 |
| Wizzup | lf is not happening | 16:26 |
| uvos | then why would it reboot? | 16:26 |
| Wizzup | some internal kernel oops? | 16:26 |
| Wizzup | I had SGX_OMAP=y instead of SGX_OMAP=m, I think this is why it did not install the other modules | 16:27 |
| uvos | ah ok | 16:27 |
| uvos | merly missing pvrsrvkm should not cause a reset ofc | 16:27 |
| Wizzup | looks like it does though | 16:27 |
| Wizzup | in any case this is not important atm | 16:27 |
| Wizzup | this really illustrates why computers are so painful, it still doesn't install it with modules_install | 16:28 |
| Wizzup | so maybe DRM=y and DRM_OMAP=y causes it just to skip installing the modules | 16:29 |
| * bencoh headscratches | 16:29 | |
| uvos | that would be very wierd | 16:29 |
| uvos | but idk | 16:29 |
| Wizzup | I'll do make clean and new defconfig.. | 16:29 |
| bencoh | do make savedefconfig first | 16:30 |
| Wizzup | no need | 16:30 |
| bencoh | unless you don't want to debug it | 16:30 |
| Wizzup | not really, I just want my modified defconfig to work | 16:30 |
| uvos | do the sgx modules have any depends in Kconfig? | 16:30 |
| uvos | maybe they depend on omapdrm being a module | 16:30 |
| uvos | you can do that | 16:31 |
| Wizzup | well they get build... | 16:31 |
| Wizzup | they just don't get installed | 16:31 |
| uvos | ¯\_(ツ)_/¯ | 16:32 |
| * freemangordon hates meson/ninja | 16:35 | |
| freemangordon | why the hell a simple touch of some config file would cause 2000 files to be recompiled is beyond my capabilities to grok | 16:36 |
| Wizzup | make clean made it work.. | 16:37 |
| freemangordon | use ninja next time, it cleans everything even if you don't want it to :D | 16:39 |
| bencoh | :] | 16:39 |
| uvos | the user side experiance of meson is not so great yeah | 16:40 |
| uvos | but developer side is quite good | 16:40 |
| uvos | not that i recommend using it | 16:40 |
| bencoh | well, many projects are slowly moving to it | 16:41 |
| bencoh | so I feel like we won't have much choice in some close future | 16:41 |
| uvos | yeah because its really easy to maintin dev side, using autotools is a mess really | 16:41 |
| freemangordon | uvos: hmm, so I am user? | 16:41 |
| uvos | freemangordon: yeah | 16:41 |
| bencoh | (maybe not so near, but ....) | 16:41 |
| freemangordon | good to know :D | 16:41 |
| uvos | freemangordon: of the build system | 16:41 |
| uvos | freemangordon: your not writing mesas buld system | 16:42 |
| freemangordon | Wizzup: ok, what you want me to do wiht mesa tag? | 16:42 |
| freemangordon | shall I increase version or what? | 16:42 |
| Wizzup | freemangordon: do you want it built soon? | 16:42 |
| Wizzup | I'm a phone call atm | 16:42 |
| freemangordon | ah, sorry | 16:42 |
| Wizzup | I think I tagged it | 16:42 |
| Wizzup | if I did, then feel free to delete tag and re-tag | 16:42 |
| freemangordon | yes, you did | 16:42 |
| freemangordon | ok | 16:42 |
| freemangordon | will do | 16:42 |
| freemangordon | and will spin a build in CI | 16:43 |
| Wizzup | cool | 16:46 |
| freemangordon | uvos: do you have an idea how to fix one and the same 12MiB .so being copied 30 times? | 16:47 |
| freemangordon | xxx_dri.so files I mean | 16:48 |
| uvos | freemangordon: where? | 16:49 |
| freemangordon | /usr/lib/armhf-xxxx/dri | 16:49 |
| uvos | no i mean what builds it? mesa? | 16:49 |
| freemangordon | yes | 16:50 |
| uvos | what so are we talking about here | 16:55 |
| uvos | pvr? | 16:55 |
| freemangordon | uvos: do 'ls -al /usr/lib/arm-linux-gnueabihf/dri/' | 16:56 |
| freemangordon | no matter pvr or not | 16:56 |
| freemangordon | it copies one and the same .so under different names | 16:57 |
| uvos | is it that way in your debian direcory? | 16:57 |
| freemangordon | using 200-300 MiB instead of ~15 | 16:57 |
| uvos | ie before install runs | 16:57 |
| freemangordon | I guess, but I have stopped the build | 16:58 |
| freemangordon | lemme restart it | 16:58 |
| freemangordon | but it will take 2 hours | 16:58 |
| uvos | heh | 16:58 |
| freemangordon | bencoh: do you have mesa build around? | 16:58 |
| uvos | maybe ill build it on vm later | 16:58 |
| uvos | *i will build it on vm later | 16:58 |
| uvos | and see | 16:58 |
| freemangordon | lemme check what is installed in debian | 16:59 |
| uvos | essentally check if its mesa buildsystem screwing up the build the install or if its the debian packaging | 16:59 |
| freemangordon | yes, I know | 16:59 |
| freemangordon | I mean - I understand your point | 16:59 |
| bencoh | freemangordon: do you want me to build it? | 16:59 |
| freemangordon | if possible | 16:59 |
| bencoh | sure, which commit/branch? | 16:59 |
| freemangordon | https://github.com/maemo-leste-upstream-forks/mesa/tree/maemo/beowulf-experimental | 17:00 |
| freemangordon | bencoh: wait | 17:00 |
| freemangordon | I did 'ninja -j2' all the time | 17:01 |
| freemangordon | so my debian directory is intact | 17:01 |
| bencoh | hm? | 17:01 |
| freemangordon | debian build was not cleaned | 17:02 |
| bencoh | ah | 17:02 |
| freemangordon | uvos: https://pastebin.com/Nbnb792w | 17:03 |
| freemangordon | as expected | 17:03 |
| bencoh | I guess you fetched libdrm/llvm8/libclang8 from backports btw? | 17:04 |
| freemangordon | yeah | 17:04 |
| bencoh | tell me if you still want me to build it :) | 17:04 |
| freemangordon | no need, thanks | 17:04 |
| uvos | freemangordon: its the same on arch linux | 17:35 |
| uvos | freemangordon: looks like there really is only 2 different .so files | 17:35 |
| uvos | mesa classic and gallium3d | 17:35 |
| uvos | wth | 17:35 |
| uvos | aperantly mesa people dont know what a symbolic link is | 17:35 |
| freemangordon | yeah | 17:43 |
| freemangordon | looks like | 17:43 |
| freemangordon | maybe this is because of dx12 support :D | 17:43 |
| uvos | freemangordon: they link everyting into one so | 17:44 |
| uvos | on purpose it seams | 17:44 |
| uvos | and then call ./bin/install_megadrivers.py on it | 17:44 |
| uvos | i dont get wth they are trying to achive here | 17:45 |
| freemangordon | yes, linking everyting into once might have sense | 17:45 |
| freemangordon | but copying this 30 times is insane | 17:45 |
| freemangordon | this will easily fill onenand on n900 when we decide to move there :) | 17:46 |
| Wizzup | maybe they all use btrfs with CoW | 17:46 |
| bencoh | huhu | 17:46 |
| freemangordon | no matter, this needs to be fixed | 17:47 |
| bencoh | so I wasn't entirely wrong when I said that there were duplicate files :] | 17:47 |
| Wizzup | :P | 17:47 |
| Wizzup | maybe we make an issue and fix it later? | 17:47 |
| uvos | bencoh: yeah | 17:47 |
| freemangordon | Wizzup: well, better fix it while we are on it | 17:47 |
| bencoh | I think they expect package managers to split it | 17:47 |
| freemangordon | but, no me :D | 17:47 |
| uvos | i dont like them linking everything into one so either | 17:47 |
| uvos | huge waste of ram | 17:47 |
| freemangordon | uvos: yeah, but his is easy to be maintained | 17:48 |
| uvos | what do i need the driver for r300 | 17:48 |
| freemangordon | not really | 17:48 |
| freemangordon | (waste of RAM) | 17:48 |
| uvos | how not? | 17:48 |
| freemangordon | you dont read pages into RAM unless needed | 17:48 |
| bencoh | yeah | 17:48 |
| uvos | ok | 17:48 |
| bencoh | in that regard it's not *that* bad | 17:48 |
| freemangordon | *that bad* being the key here :) | 17:49 |
| bencoh | ;) | 17:49 |
| freemangordon | but otherwise you have to maintain several .so files | 17:49 |
| freemangordon | and keep iterfaces comparibel, etc,etc | 17:49 |
| freemangordon | *compatible | 17:50 |
| freemangordon | one big blob is easier to maintain | 17:50 |
| freemangordon | "(17,40,32) uvos: but developer side is quite good" | 17:51 |
| freemangordon | :p | 17:51 |
| bencoh | :] | 17:51 |
| freemangordon | uvos: did you have time to further look into buffer swap issue? | 17:51 |
| uvos | well there writing custom python scripts here | 17:52 |
| uvos | its hardly just meson anymore | 17:52 |
| uvos | but sure | 17:52 |
| uvos | freemangordon: not yet | 17:52 |
| freemangordon | ok | 17:52 |
| uvos | but i will get to it | 17:52 |
| freemangordon | thanks | 17:52 |
| freemangordon | appreciated | 17:53 |
| freemangordon | I want to start looking into vrfb | 17:53 |
| freemangordon | hmm, maybe I shall power-up my n950 soon | 17:53 |
| bencoh | wow, n950 | 17:53 |
| freemangordon | mhm | 17:53 |
| freemangordon | I was awarded back then | 17:53 |
| bencoh | yeah I know, I was just surprised to read that now :) | 17:54 |
| bencoh | it might be easier to tinker with than n900 though :) | 17:54 |
| freemangordon | maybe | 17:54 |
| bencoh | what about serial on it? | 17:54 |
| freemangordon | no idea | 17:54 |
| bencoh | is it simpler? | 17:54 |
| bencoh | ah | 17:54 |
| freemangordon | and don;'t wan't to risk it | 17:55 |
| bencoh | right | 17:55 |
| freemangordon | I plan this to be my next device | 17:55 |
| bencoh | oh | 17:55 |
| freemangordon | yeah | 17:55 |
| uvos | daily carring a unreplaceable device | 17:55 |
| uvos | ok.. | 17:55 |
| uvos | whatever floats your boat | 17:55 |
| bencoh | well ... | 17:55 |
| uvos | :P | 17:55 |
| freemangordon | why non-replaceable? | 17:56 |
| Wizzup | ok, I have a custom boot menu on the n900 and it prints sensible stuff on the console now, but something causing resets isn't fun surely | 17:56 |
| bencoh | assuming leste goes strong and works on other devices, he could still migrate to droid4 if needed | 17:56 |
| freemangordon | I mean - if it breaks, I have d4 ;) | 17:56 |
| freemangordon | bencoh: yeah | 17:56 |
| bencoh | ;) | 17:56 |
| freemangordon | Wizzup: if you have high-fps camera around, you may try to capture a video | 17:57 |
| Wizzup | no, the screen is black, it happens after X is being started | 17:57 |
| freemangordon | Wizzup: or, set mtdoops partition | 17:57 |
| uvos | Wizzup: boot to emergency shell | 17:57 |
| uvos | start xorg by hand via ssh | 17:57 |
| freemangordon | or move dsme out of /etc/init.d | 17:58 |
| uvos | that way you have some output while it starts | 17:58 |
| freemangordon | you'll get login shell | 17:58 |
| Wizzup | there are many things to try, first I will just try building from within the lxc container and get something that works again | 17:58 |
| freemangordon | ok | 17:58 |
| Wizzup | freemangordon: doing that (mv dsme) means I have to manually type on the keyboard to get usb net up | 17:58 |
| Wizzup | and it's a pain | 17:58 |
| Wizzup | :D | 17:58 |
| freemangordon | I use wpa_supplicant with conf file | 17:59 |
| Wizzup | mhm | 17:59 |
| freemangordon | I typed the command once | 17:59 |
| freemangordon | and from then on I use ctrl-r ;) | 17:59 |
| uvos | the n900 having so little special keys in console can get pretty frustrating | 18:00 |
| uvos | yeah | 18:00 |
| Wizzup | I will do that if it turns out to be a problem that keeps recurring | 18:01 |
| Wizzup | having drm built in is real nice though since you get display in seconds rather than after 30 seconds | 18:04 |
| Wizzup | damn, now even the kernels that used to work reset in X, ok.... I guess I'll have to debug it now :( | 18:25 |
| Wizzup | maybe it's the powervr.ini that I moved away somehow | 18:31 |
| freemangordon | I have powervr.ini on neither d4 or n900 | 18:32 |
| Wizzup | ok | 18:32 |
| freemangordon | *nor | 18:32 |
| Wizzup | weird, X and h-d are fine, so wth is causing the resets | 18:43 |
| freemangordon | who resets? dsme? | 18:44 |
| Wizzup | no, kernel just resets | 18:44 |
| freemangordon | hmm | 18:44 |
| Wizzup | leaving all log files corrupted | 18:44 |
| freemangordon | bad battery? | 18:44 |
| Wizzup | seems unlikely | 18:44 |
| freemangordon | ok | 18:44 |
| freemangordon | uvos__: hmm, after thinking about that a bit, DRI2 actually uses doulble-buffering, but it works like that: | 18:50 |
| freemangordon | 1. a frame is rendered and is send to be displayed | 18:50 |
| freemangordon | 2. driver waits for vsync and only then it completes the swap | 18:51 |
| freemangordon | 3. client renders back buffer again and requests swap | 18:52 |
| freemangordon | 4. after vsync back buffer becomes front and only the swap is completed | 18:52 |
| freemangordon | *then | 18:52 |
| freemangordon | so, we need to implement tripple-buffering in DDX | 18:53 |
| freemangordon | maybe we can do it better if the first swap is returned immediately | 18:57 |
| bencoh | is triple-buffering really mandatory here? | 18:58 |
| freemangordon | no | 18:58 |
| bencoh | it's usually not that good perf-wise | 18:58 |
| freemangordon | but it seems DDX is done in such a way that client waits in SwapBuffers until vsync | 18:59 |
| freemangordon | I am not sure I explain this correctly | 18:59 |
| bencoh | (hmm, actually the screen is small enough that it wouldn't hurt *that* much the memory bus) | 18:59 |
| freemangordon | actually I think we shall block only if we have more than one buffer waiting for vsync | 19:00 |
| freemangordon | yes, that should be it | 19:00 |
| bencoh | sounds reasonable | 19:01 |
| bencoh | that would mean the driver always have another buffer ready to push to display | 19:01 |
| freemangordon | lemme see if I can implement that | 19:01 |
| bencoh | (assuming I got that right) | 19:01 |
| bencoh | s/have/has/ | 19:01 |
| dsc_ | epilys: https://i.imgur.com/LtWQclB.png | 19:07 |
| dsc_ | ship it(tm) | 19:07 |
| Wizzup | freemangordon: looks like icd2 starting causes the kernel reset | 19:11 |
| bencoh | uh | 19:11 |
| freemangordon | weird | 19:11 |
| Wizzup | maybe wifi driver | 19:11 |
| freemangordon | yeah | 19:11 |
| Wizzup | it was working before though | 19:11 |
| Wizzup | yeah it does... wow | 19:12 |
| uvos__ | freemangordon: ok yeah id avoid tripple buffering if at all possible | 19:43 |
| freemangordon | uvos__: yeah, but I think there is something broken in mesa with double-buffering | 19:43 |
| freemangordon | I think SwapBuffers shall not block | 19:44 |
| freemangordon | not that I debugged it, but I think it does | 19:44 |
| freemangordon | some random thoughts - 'vsync' on d4 is 80 Hz | 19:44 |
| freemangordon | it seems to be the max rate display can be updated to | 19:45 |
| freemangordon | max FPS reported by clutter on d4 for h-d scrolling is 40 | 19:45 |
| freemangordon | on n900, vsync is 60, max fps reported by clutter - 30 | 19:45 |
| freemangordon | so, we are not vsync limited, but vsinc/2 limited | 19:46 |
| freemangordon | which is single-buffering IIUC :) | 19:46 |
| freemangordon | hmm, wait | 19:46 |
| freemangordon | no, this is double-buffering | 19:46 |
| freemangordon | because we cannot render on buffer A until buffer B is not flipped | 19:47 |
| freemangordon | uvos: DRI3 does tripple-buffering by default | 19:47 |
| freemangordon | so I don;t see a reason why dri2 shall not | 19:48 |
| freemangordon | we have 1/60th of a second delay, so what? | 19:48 |
| freemangordon | worse-case | 19:48 |
| uvos | i mean 1/60 of a second is quite a bit | 19:48 |
| freemangordon | *worst | 19:48 |
| uvos | you can feal that for sure | 19:49 |
| uvos | so dose vblank_mode on hildon-desktop change this any | 19:49 |
| uvos | ? | 19:49 |
| freemangordon | no | 19:49 |
| uvos | ok | 19:49 |
| uvos | so its not h-d | 19:49 |
| freemangordon | vblank_mode, like clutter_vblank_mode? | 19:49 |
| freemangordon | no, it is not h-d | 19:49 |
| uvos | no the mesa override | 19:49 |
| freemangordon | hmm, what is this? | 19:49 |
| uvos | https://dri.freedesktop.org/wiki/ConfigurationOptions/ | 19:50 |
| uvos | looks light it might just be implemented for some drivers | 19:50 |
| uvos | (that list is wrong for sure tho) | 19:50 |
| uvos | same effect as clutter_vblank_mode | 19:51 |
| uvos | most likely | 19:51 |
| uvos | just mesa forcing things | 19:51 |
| freemangordon | anyway, with modesetting/dri3 we have tripple-buffering | 19:52 |
| uvos | only if you set it in xorg.conf no? | 19:53 |
| freemangordon | no | 19:53 |
| freemangordon | this is out of x control | 19:53 |
| freemangordon | in dri3 clients control buffers | 19:53 |
| uvos | ok iirc x has a option for this | 19:53 |
| uvos | but might just apply to dri2/be broken | 19:54 |
| freemangordon | no, this is totally out of server's control | 19:54 |
| uvos | well xorg also dose its own rendering | 19:54 |
| uvos | might be refering to that | 19:54 |
| freemangordon | could be | 19:54 |
| freemangordon | but in terms of page flipping, server has no sayng | 19:54 |
| freemangordon | *saying | 19:55 |
| uvos | ok | 19:55 |
| uvos | i hear you | 19:55 |
| freemangordon | sh, I think we shall implement tripple-buffering in ddx | 19:55 |
| freemangordon | 40fps is too low | 19:55 |
| uvos | so why is h-d special? | 19:56 |
| uvos | btw | 19:56 |
| freemangordon | what do you mean? | 19:56 |
| uvos | well other egl clients in non-composing mode arnt locked to 40fps (at least they dont think they are) | 19:57 |
| freemangordon | it is not special, it is just that for h-d ddx does page-flipping, glmark does blits | 19:57 |
| uvos | ah ok | 19:57 |
| uvos | yeah | 19:57 |
| freemangordon | uvos: https://gitlab.freedesktop.org/xorg/driver/xf86-video-omap/-/commit/ce46bb1f32e07d1be6c55114b6f92d97d0636676 | 20:13 |
| uvos | freemangordon: but its not working? | 20:21 |
| uvos | otherwise why is it wating | 20:21 |
| freemangordon | well, I don't understand what is "client explicitly request the third buffer." supposed to mean | 20:22 |
| freemangordon | somehow, someone shall request DRI2BufferThirdLeft | 20:22 |
| freemangordon | hmm seems armsoc ddx has this 3buffering implemented | 20:24 |
| freemangordon | I can use that as an example | 20:24 |
| Wizzup | interesting... even with wifi module unloaded icd still causes reset - maybe it's me adding nokia modem... | 20:39 |
| Wizzup | (which you wouldn't see since you probably don't have ofono installed) | 20:39 |
| uvos | well at least your having fun. | 20:40 |
| uvos | maybe try touching the modem interfaces | 20:41 |
| Wizzup | it's only fun because I'm not stuck anymore :) | 20:41 |
| Wizzup | I don't think I will debug the modem problems now if that turn out to be the problem | 20:41 |
| Wizzup | I want to get a more usable config first (drm built in, etc) | 20:42 |
| uvos | ah yes the endless cycle: i hate this, oh neat, i hate this, oh neat, i hate this.... | 20:42 |
| uvos | Wizzup: ok | 20:42 |
| uvos | Wizzup: makes sense | 20:42 |
| Wizzup | yes, removing nokia modem and friends made the freeze go away | 20:43 |
| uvos | might be usefull to know from sicelo if he used the modem on some newer version than 5.1 to have something to bisect from | 20:43 |
| uvos | sicelo: ^^^^ | 20:43 |
| Wizzup | It's probably a unbalanced pm thing I bet | 20:43 |
| Wizzup | most other things don't cause resets like these | 20:44 |
| uvos | anything that hangs the kernel can cause a watchdog reset like this or? | 20:44 |
| uvos | sure a unbalanced pm sleep/wake would do it | 20:45 |
| Wizzup | uvos: not sure how it works exactly (when it hangs and resets) | 21:03 |
| Wizzup | blacklisting nokia_model makes the kernel stuck trying to load it anyway it seems | 21:07 |
| Wizzup | modem* | 21:07 |
| Wizzup | easier to not build the module yet, hehe | 21:07 |
| uvos | hmm that def should not happen.. | 21:08 |
| Wizzup | # cat /mnt/extern/etc/modprobe.d/nokia-modem.conf | 21:10 |
| Wizzup | options nokia-modem pm=1 | 21:10 |
| Wizzup | maybe I should have written nokia-modem and not nokia_modem | 21:10 |
| Wizzup | (lsmod shows nokia_modem though) | 21:10 |
| Wizzup | anyway we have that pm=1 thing in there specifically on leste | 21:10 |
| sicelo | freemangordon: "I plan this to be my next device" ... meaning, daily drive N950? | 21:26 |
| freemangordon | why not? | 21:29 |
| Wizzup | uvos: yeah looks like the nokia-modem.conf caused it to be loaded | 21:33 |
| sicelo | uvos, Wizzup - yes i was building nokia modem on 5.9-5.12. i don't remember if i actually used it | 21:37 |
| sicelo | re - N950, just interesting. it definitely seems like a fun device | 21:38 |
| sicelo | nokia-modem - even if i might not have used it, i distinctly remember it was probing just fine | 21:38 |
| Wizzup | it probes fine | 21:41 |
| Wizzup | the problem occurs when icd2 talks to ofono which talks to modem | 21:41 |
| sicelo | ok. i really can't remember if i put it to any use. maybe i didn't | 21:44 |
| Wizzup | it works in 5.1 for sure | 21:44 |
| Wizzup | it might be the pm= option | 21:44 |
| Wizzup | at least I got the following done today: working bootmenu on sd card, drm built in for smoother boot, zImage booting works | 21:45 |
| sicelo | it'd be really sad if more stuff is 'broken' for N900 by now, although i hope fixing all of them should not be too hard | 21:45 |
| Wizzup | lol as I wrote this xorg ddx crashed | 21:45 |
| Wizzup | sicelo: this was the case every time we try newer kernel after a while | 21:45 |
| sicelo | oh great @zImage! | 21:45 |
| Wizzup | nobody but us uses it | 21:45 |
| uvos | well pmos | 21:45 |
| uvos | they use 5.6 i think | 21:45 |
| sicelo | 5.7 | 21:46 |
| Wizzup | they are also looking for a maintainer, and I am not sure if they are finding volunteers | 21:46 |
| uvos | so that pobubly works | 21:46 |
| Wizzup | (for n900) | 21:46 |
| Wizzup | sicelo: yeah I need to test booting without dtb appended but I think it will work too | 21:46 |
| Wizzup | that is necessary to make our single kernel pkg | 21:46 |
| sicelo | i had started to 'track' the kernel and test for n900 each time (from 5.9 until there was 5.13-rc1, i think). when it became fun, then life got extra busy ... hopefully mid-dec things normalize | 21:48 |
| Wizzup | freemangordon: I apt upgraded to latest mesa and I think my n900 is unhappy now, let me reboot and try again - anything you changed that might affect it? | 21:52 |
| Wizzup | (can't get a break hehe) | 21:53 |
| Wizzup | freemangordon: nevermind for now, let me debug further... | 21:55 |
| sicelo | cannice to know built in drm works. i once tried it, and my n900 wouldn't boot. but it was booting fine with drm as module | 21:56 |
| sicelo | s/cannice/nice/ | 21:56 |
| Wizzup | yeah but something broke since I upgraded mesa | 21:57 |
| Wizzup | now the device is unusable again :D | 21:57 |
| Wizzup | but I think the problem is that powervr doesn't init now, and I am pretty sure it did | 21:57 |
| uvos | well what dose pvrsrvinit say? (in logs dont run it again) | 21:57 |
| Wizzup | I can't get into the device because usbnet also doesn't work anymore, and the display doesn't respond, so I cna't tell you for sure | 21:58 |
| uvos | mhh ok | 21:58 |
| uvos | fun | 21:58 |
| Wizzup | powervr |PVR:(Error): OpenServices: PVRDRMOpenRender failed [0, ] | 21:58 |
| Wizzup | powervr |PVR:(Error): PVRSRVInitSrvConnect: PVRSRVConnect failed [0, ] | 21:58 |
| Wizzup | powervr |PVR:(Error): SrvInit: PVRSRVInitSrvConnect failed (4) [0, ] | 21:58 |
| Wizzup | powervr |failed to initialize server | 21:58 |
| Wizzup | powervr | * ERROR: powervr failed to start | 21:58 |
| Wizzup | prety generic | 21:58 |
| uvos | im pretty sure thats the same as when you dont have the module probed | 21:59 |
| uvos | but yeah its not very detailed | 21:59 |
| Wizzup | maybe it's also racy somehow | 22:00 |
| Wizzup | oh, I see | 22:01 |
| Wizzup | 2981 Nov 24 19:22:15 localhost kernel: [ 2.190307] [drm] Initialized omapdrm 1.0.0 20110917 for omapdrm.0 on minor 0 | 22:01 |
| Wizzup | vs | 22:01 |
| Wizzup | 2580 Nov 24 19:07:56 localhost kernel: [ 28.333038] [drm] Initialized pvr 1.17.4948957 20110701 for 50000000.gpu on minor 1 | 22:01 |
| Wizzup | the one with built-in drm doesn't seem to init pvr | 22:02 |
| Wizzup | so yeah, probably it is not being loaded somehow | 22:02 |
| Wizzup | maybe because the mode is not available drm is being loaded | 22:02 |
| Wizzup | 3267 Nov 24 19:22:16 localhost kernel: [ 2.283264] [drm] Initialized omapdrm 1.0.0 20110917 for omapdrm.0 on minor 0 | 22:02 |
| Wizzup | (2.2s after boot) | 22:02 |
| uvos | inserting the pvr moule later works fine | 22:03 |
| uvos | like haveing it blacklisted and inserting it minutes after boot | 22:03 |
| uvos | i have done that plenty of times | 22:03 |
| Wizzup | but nothing loads it | 22:04 |
| Wizzup | I guess pvrsrvctl would ;) | 22:04 |
| Wizzup | looks like keyboard also doesn't work when drm loads early :( | 22:05 |
| Wizzup | and usb otg | 22:05 |
| uvos | it should be loaded by kernel normaly | 22:06 |
| Wizzup | they are not :( | 22:06 |
| uvos | omapdrm being built in has no bearing on it | 22:06 |
| Wizzup | right, except that when omapdrm is not built in, the keyboard works | 22:06 |
| Wizzup | so there's some loading magic going on | 22:06 |
| uvos | i mean this suggests some terrible issue | 22:07 |
| uvos | maybe compile the kernel with all the debuging stuff enabled | 22:07 |
| Wizzup | maybe another day, I just wanted to do the packaging for now | 22:09 |
| Wizzup | but looks like I won't finish that today either ;) | 22:09 |
| Wizzup | I wouldn't be surprised if some parts not probing blocks other parts from probing, I've seen this a lot | 22:09 |
| Wizzup | this is the current score https://github.com/maemo-leste/bugtracker/issues/587 | 22:10 |
| sicelo | i do recall having issues with built in drm, but i was too noob to debug back then | 22:10 |
| Wizzup | I am too | 22:10 |
| uvos | i mean omapdrm built in has no issues on mapphones w/o sgx | 22:10 |
| uvos | thats how bionic boots | 22:11 |
| * freemangordon is zzzz | 22:11 | |
| freemangordon | night | 22:11 |
| Wizzup | uvos: then what loads sgx? | 22:11 |
| Wizzup | gn | 22:11 |
| uvos | w/o sgx | 22:11 |
| uvos | nothing | 22:11 |
| uvos | its a untainted kernel | 22:11 |
| Wizzup | I'm going to reboot my machine in a bit, maybe all my usb ports are dead and that's why I am not seeing the device now | 22:11 |
| sicelo | i was doing drm fror sgx purposes | 22:11 |
| Wizzup | the droid4 with leste has a habit of killing the usb controllers | 22:12 |
| Wizzup | lol | 22:12 |
| uvos | laptops dont like it if you pull high power usb class current but dont ask first | 22:12 |
| Wizzup | it's more about the switching between charging and discharging | 22:13 |
| Wizzup | that happens every second or so for an hour, and it'll die | 22:13 |
| uvos | inrush current maybe | 22:14 |
| uvos | either way | 22:14 |
| uvos | yeah cpcap-charger is terribly behaved wrt usb spec | 22:14 |
| missMyN900 | uvos: I would suggest netsurf(-gtk) instead of surf, particularly for the N900. It has very low RAM consumption and can be compiled against GTK2 still. They just introduced a new (small-screen-friendly) GTK UI in 3.10. It is available in Devuan Chimaera (and Debian 11, also pmOS edge). | 22:46 |
| missMyN900 | uvos: awesome that you are looking into qtwebbrowser. I think it has a lot of potential. | 22:46 |
| missMyN900 | uvos: according to Phoronix benchmarks the LX2160A is faster than an i7-4770K for video encoding, which is something that x86 excels at | 23:02 |
| missMyN900 | @Wizzup: M.2 can be SATA as well. I have a SATA M.2 SSD in my desktop (the X470 motherboard accepts both NVMe and SATA SSDs in the M.2 slot) | 23:03 |
| missMyN900 | @Wizzup: and my laptop (older HP EliteBook) only is compatible with M.2 SATA SSDs (I have an M.2 860 Evo installed) | 23:04 |
| * sicelo wants one of those for his T440p | 23:05 | |
| uvos | missMyN900: sure netsurf is useful on n900 maybe | 23:09 |
| uvos | missMyN900: otherwise it think its way to janky for general usage/ recommendation | 23:09 |
| missMyN900 | @uvos: all it needs is touch screen support | 23:09 |
| missMyN900 | @uvos: it works well on TMO, wiki pages, documentation and some mobile news sites (e.g. mobile.reuters.com) | 23:09 |
| uvos | it also needs a layout engine that works on popular websites | 23:10 |
| missMyN900 | Pine64 forums as well probably | 23:10 |
| uvos | i mean yeah | 23:10 |
| uvos | if thats what you want thats fine | 23:10 |
| missMyN900 | @uvos: well, but there is a limit to what you can do with low resource devices | 23:10 |
| uvos | but the deafult browser must apeal to a bit more casual audiance | 23:10 |
| missMyN900 | even on devices with a bit more RAM it could be useful as a second browser | 23:10 |
| uvos | only the n900 is low resource in this sense | 23:10 |
| missMyN900 | @uvos: Chromium struggles with 2 GB RAM | 23:11 |
| uvos | but yes for n900 its likely the best you can do | 23:11 |
| missMyN900 | even the Pinephone non-Convergence edition only has 2 GB RAM | 23:11 |
| uvos | i use d4 daily with ff and 2-3 pages open | 23:11 |
| uvos | i have never had problems with ram ever | 23:11 |
| ashley | Chrom(e|ium) struggles with 8GB of RAM on desktop/laptop ;-) | 23:11 |
| uvos | this is true | 23:12 |
| missMyN900 | netsurf is useful even on a Pinephone or some 1 GB RAM device when you are multitasking | 23:12 |
| uvos | sure im not saying: dont make it instalable | 23:13 |
| uvos | just its not suitable as a default browser | 23:13 |
| missMyN900 | I think it should be the default browser with qtwebbrowser as an alternative browser | 23:13 |
| missMyN900 | the default browser should work well on every ML device IMHO | 23:13 |
| missMyN900 | it is also useful to have a browser with JS disabled by the default, whether for privacy, security or battery life reasons | 23:14 |
| missMyN900 | @uvos: I think that if Maemo Leste ever becomes a real alternative to Android and iOS, then that will be the time to make something other than netsurf the default browser | 23:38 |
| missMyN900 | @uvos: but as long as it is an enthusiast OS run on old and cheap hardware by people who are used to the limitations of that, netsurf is an adequate default browser IMHO | 23:39 |
| missMyN900 | maybe a whole new Maemo web browser can be developed from scratch (well, except for the rendering/JS engine obviously) eventually with special privacy and security features, strong integration with the OS etc, but until then netsurf-gtk is something that is relatively easy modify into something useful and that works well on all current popular ML devices | 23:41 |
| missMyN900 | @sunshavi: iwd is available in the Devuan Chimaera repositories so it definitely can work without systemd and could be used in Maemo Leste as well | 23:42 |
| sicelo | i switched to iwd on Debian sid after reading a pmOS discussion on it ... it works. haven't really seen how 'better' it is. yes, has a nicer interactive shell (iwctl) than wpa_supplicant's (wpa_cli). anyway, i guess it's the future | 23:47 |
| missMyN900 | sicelo: switching to connman+iwd from networkmanager+wpa_supplicant fixed my 5 GHz WiFi | 23:52 |
| missMyN900 | before I was not able to connect to my 5 GHz network with Debian 11. No 5 GHz networks would show up at all | 23:52 |
| missMyN900 | connman+wpa_supplicant worked extremely poorly though | 23:52 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!
