Wizzup | freemangordon: I suppose I could make some fake pkg that provides all these translations, just for a small image | 01:34 |
---|---|---|
Wizzup | right now /usr/share/locale is like 150MB mostly because of these files I think | 01:34 |
Wizzup | but without it, I have a ~700MB image which should fit on the mz617 | 01:34 |
Wizzup | tmlind: ok, have my mz617 set up again, will look at what you mentioned re: cable being a problem | 02:37 |
uvos | Wizzup: sure, ill try and remember for next time | 09:15 |
uvos | i gues having a bunch of stuff built into the kernel and not as kernel modules might bite us eventually | 09:17 |
uvos | it would be good if it fits onto the n900s boot partition for instance | 09:17 |
uvos | and n900 uboot did have some small kernel size limit that was a problem in the past | 09:18 |
uvos | not sure if that was lifted | 09:18 |
uvos | i imagine btrfs is pretty large, given its featureset, but dont know how large in absolute terms | 09:19 |
sicelo | on n900 the u-boot sits on the kernel area. kernel itself can sit anywhere. i believe it's the same thing on D4. IOW, the size of actual kernel isn't really an issue for N900 | 09:24 |
uvos__ | sicelo: it is if you want to append the kenrel | 11:02 |
uvos__ | D4 can have any size kernel yeah | 11:02 |
Wizzup | uvos__: well as a module we can't use it as root | 11:36 |
Wizzup | unless we start making initramfses | 11:36 |
sicelo | where are we planning to use btrfs | 11:43 |
uvos__ | sicelo: mz6xx | 11:45 |
uvos__ | to span the partitions | 11:45 |
Wizzup | well to span it we would still need a initramfs | 11:50 |
Wizzup | in any case I did build a ~700MB ext4 leste image with the image builder that boots fine | 11:51 |
Wizzup | (stripped down ofc) | 11:51 |
Wizzup | so that is enough for mz617 to install to userdata | 11:51 |
Wizzup | (which is like 13GB at least) | 11:51 |
uvos__ | you sure we need initramfs to span? | 11:56 |
uvos__ | i think just root=/one/device/of/the/span should work | 11:57 |
uvos__ | or could work at least | 11:57 |
Wizzup | uvos__: doesn't work, tried that | 11:59 |
uvos__ | :( | 11:59 |
Wizzup | btrfs doesn't scan for devices in kernel | 11:59 |
Wizzup | it needs userspace to scan somehow | 11:59 |
Wizzup | I agree it's dumb | 12:00 |
uvos__ | well now i dont know what we need btrfs to be built in for | 12:01 |
uvos__ | its not really a good filesystem for small flash volumes | 12:01 |
Wizzup | compression is still useful | 12:01 |
uvos__ | neither is ext4 but ok | 12:01 |
Wizzup | root@devuan-mz617:~# df -h | 12:03 |
Wizzup | Filesystem Size Used Avail Use% Mounted on | 12:03 |
Wizzup | /dev/mmcblk1p20 13G 2.1G 11G 18% / | 12:03 |
Wizzup | this is default image with ext4 | 12:03 |
Wizzup | er btrfs | 12:03 |
Wizzup | but no compression | 12:03 |
uvos__ | yeah sure | 12:03 |
Wizzup | so it's not too bad space wise atm | 12:03 |
uvos__ | but the problem is user_data cant be flashed directly or | 12:03 |
uvos__ | that was one that could only be erased? | 12:04 |
Wizzup | neither cache nor preinstall nor user data can be flashed directly | 12:04 |
uvos__ | cache can be flashed directly | 12:04 |
Wizzup | but from android you can write an image to cache | 12:04 |
Wizzup | well my cache can't I seem to remember | 12:05 |
Wizzup | in any case, | 12:05 |
Wizzup | I flashed user data from my tiny leste on cache, which I dd't from android to cache | 12:05 |
Wizzup | if we can flash directly to cache with fastboot that would be nice, and I'll probably try it some time soon | 12:05 |
uvos__ | ok but cache was where we originally placed fastboot because that was flashable | 12:06 |
uvos__ | er kexecboot not fastboot | 12:06 |
Wizzup | ok | 12:06 |
Wizzup | cdrom is my kexecboot | 12:06 |
uvos__ | im not sure of the others | 12:06 |
Wizzup | well I can try this now that my userdata works | 12:06 |
uvos__ | preinstall for sure can be flashed too but that could be singed | 12:06 |
uvos__ | user_data cant be flashed at all, not evenby a signed image | 12:06 |
Wizzup | I made some image-builder changes with the hildon-meta-core split, and we can get pretty tiny images with just keyboard,wifi,h-d,cpa,ofono btw | 12:07 |
uvos__ | great | 12:07 |
Wizzup | locales are still 150MB, but without that it's like 750MB | 12:07 |
uvos__ | that seams pretty large still | 12:07 |
Wizzup | (that is the debian translations) | 12:07 |
Wizzup | yup, mesa is *huge* | 12:07 |
Wizzup | libllvm is like 70MB or so | 12:07 |
uvos__ | yeah gpu drivers can get absurldy huge | 12:08 |
Wizzup | There was some other low hanging fruit that I tackled, like not adding iphbd or dkms, which pulls in gcc, and removing some pkgs devuan image builder installs that we don't need (htop, zsh, etc) | 12:08 |
uvos__ | (cires in 4GB AMD libdevice_opertions.so) | 12:08 |
Wizzup | but <800MB is enough for the mz6xx purposes I think | 12:08 |
Wizzup | yeah that's probably some deeplearning stuff | 12:08 |
uvos__ | i think htop can go everywhere | 12:10 |
uvos__ | i was allways puzzeled why this was preinstalled | 12:10 |
Wizzup | devuan imagebuild does it | 12:10 |
Wizzup | and yes it can go away | 12:10 |
Wizzup | I think having vim or htop preinstalled is nice for users of the regular image | 12:11 |
Wizzup | it shows something in the debian submenu | 12:11 |
uvos__ | yeah | 12:11 |
uvos__ | i gues vim is needed for strict posix compat | 12:11 |
uvos__ | htop is just useless | 12:11 |
Wizzup | it has a certain hackery feeling to it :P | 12:11 |
Wizzup | btw, I'm going to add razr and mzxxx images | 12:11 |
Wizzup | I was thinking of what we go through every time | 12:12 |
Wizzup | leste-config can do in the xyboard "Section" I suppose | 12:12 |
Wizzup | for mzxxx | 12:12 |
Wizzup | (since we made different ones for all the other devices) | 12:12 |
uvos__ | yes | 12:12 |
Wizzup | I wonder if we need more specific meta packages or leste-config's for the different mz61x | 12:12 |
uvos__ | well we dont know yet | 12:12 |
uvos__ | the accels could be oriented differently | 12:13 |
Wizzup | and what do we call mz607,mz608,mz609 and what do we call mz615,mz616,mz617 - mz60x, mz61x ? | 12:13 |
uvos__ | for one thing | 12:13 |
uvos__ | mz60x and mz61x? :P | 12:13 |
uvos__ | oh yeah | 12:13 |
Wizzup | ok | 12:13 |
Wizzup | I would like to prevent buildimg images for all of the different submodels if possible (because the image building is inefficient) | 12:13 |
uvos__ | yeah | 12:14 |
Wizzup | so I could name it either specifically, like mz617 or mz616, or mz61x) | 12:14 |
uvos__ | hopefully this wont bite us | 12:14 |
Wizzup | I suppose they will need a different inittab | 12:14 |
Wizzup | (/boot/ on mz617 is really android /system) | 12:14 |
Wizzup | if you keep the standard inittab it will error on boot and ask you to type in the root pw or ^D | 12:14 |
uvos__ | yeah | 12:15 |
Wizzup | (since it wants to check /dev/mmcblk1p1 but can't) | 12:15 |
uvos__ | also what about the variants with sdcards | 12:15 |
uvos__ | i think sdcard images for those would be usefull | 12:15 |
Wizzup | yes | 12:15 |
uvos__ | i know this gets messy | 12:15 |
Wizzup | so maybe we can manage /etc/inittab with leste-config | 12:15 |
uvos__ | you also need to change boot.cfg | 12:15 |
Wizzup | I think for mz616 we can safely assume people will probably just use sdcards and not te internal emmc | 12:15 |
uvos__ | sure and if not mz617 is presuambly compatable | 12:16 |
Wizzup | the one thing lacking for mz61x images -now- is the kernel | 12:16 |
Wizzup | yeah I think we'll find it all to be very compatible | 12:16 |
Wizzup | I haven't found any differences at all really | 12:16 |
Wizzup | (apart from modem presumably) | 12:16 |
uvos__ | well mz60x is for sure different in varous ways | 12:17 |
uvos__ | but yeah the sub variants are presumably very samey | 12:17 |
Wizzup | yeah that is what I meant | 12:17 |
Wizzup | btw, please ack that you saw my sphone changes | 12:17 |
uvos__ | i did | 12:17 |
Wizzup | I pushed to master only a minor debian/ change but then the build failed so I just decided to push a build fix to master | 12:17 |
uvos__ | dont worry, i hired the hitman already | 12:18 |
Wizzup | :D | 12:18 |
uvos__ | so otherwise you know what to do right? | 12:20 |
uvos__ | for sphone | 12:20 |
uvos__ | add a config to leste-config for it | 12:20 |
uvos__ | and tell me, i can then drop the maemo stuff from the default config | 12:21 |
Wizzup | oh, so you mean the divert the default sphone.ini ? | 12:21 |
uvos__ | no, add a 20-maemo.ini or so | 12:21 |
Wizzup | ok, I thought we couldn't change the module load line there, did you change that? | 12:21 |
uvos__ | you can change anything | 12:22 |
uvos__ | do tell me if it dosent work | 12:22 |
uvos__ | but anything in sphone.init should be changeable | 12:22 |
Wizzup | ok, so if I add the comm-voicecallmanager via leste-config as a 20-maemo.ini it should work? | 12:22 |
uvos__ | yeah i mean you need to copy the whole modules= line | 12:22 |
Wizzup | same for qtloop? | 12:22 |
uvos__ | and overwrite it | 12:22 |
uvos__ | yes | 12:23 |
Wizzup | ok, I'll try that | 12:23 |
Wizzup | thanks! | 12:23 |
uvos__ | and you can disable the sms feature to | 12:23 |
uvos__ | theres a value for that | 12:23 |
uvos__ | (and dont load the sms ui modules to save some ram) | 12:23 |
uvos__ | for recents to be the entry point (as you would like) | 12:24 |
uvos__ | you need to change the .desktop file atm | 12:24 |
uvos__ | need to figure out how to deal with this | 12:24 |
Wizzup | yeah that's tricky | 12:25 |
Wizzup | btw, for new kernel, I think tony pushed our some more stuff recently, we still have the cpufreq issue with it not showing up right? | 12:25 |
uvos__ | have not tried it recently, sorry | 12:25 |
uvos__ | could be fixed yeah, no idea | 12:25 |
Wizzup | okay, for the mz61x images themselves we'll need a new kernel build I think | 12:26 |
uvos__ | yes | 12:26 |
Wizzup | if you could re-add the droid3 dts in there too, that would be good | 12:26 |
Wizzup | I have a local change for it on maemo-6.6-buildeb that I could push somewhere | 12:26 |
uvos__ | it needs to rewirtten because the layout of the dts files changed alot | 12:26 |
uvos__ | but yeah | 12:26 |
Wizzup | since 6.6? | 12:27 |
uvos__ | no to 6.6 | 12:27 |
Wizzup | I did this | 12:27 |
uvos__ | lots of stuff moved around | 12:27 |
uvos__ | great | 12:27 |
Wizzup | let me push this somewhere | 12:27 |
Wizzup | https://github.com/maemo-leste/droid4-linux/tree/maemo-6.6-buildeb | 12:28 |
Wizzup | it also updates this scripts/package/builddeb thing, but maybe you did have that already | 12:28 |
uvos__ | yeah | 12:28 |
uvos__ | btw mesa | 12:34 |
uvos__ | we build or own mesa for pvr anyhow | 12:34 |
uvos__ | we could disable all the drivers besides prv to make it pretty tiny presumably | 12:35 |
uvos__ | why have a drivers for all the desktop dgpus in there, its not like leste devices have pcie slots | 12:35 |
Wizzup | right we need to rebase on mesa-amber right? | 12:42 |
Wizzup | well we do need mesa with other drivers for other devices | 12:42 |
Wizzup | pinephone for example | 12:42 |
Wizzup | or the librem soon (looks at sicelo :D) | 12:42 |
Wizzup | and yes, we could make a special pvr mesa, but it actually costs a lot of time | 12:43 |
Wizzup | even to just upgrade mesa and deal with the packaging mess | 12:43 |
uvos__ | Wizzup: we need to split pvr mesa anyhow | 12:57 |
uvos__ | pvr mesa needs to be amber | 12:57 |
uvos__ | while the gallium3d drivers (like for pp) must not be amber | 12:57 |
uvos__ | so special mesa pvr is inevitable | 12:57 |
Wizzup | right, ok, well, shall we do that for daedalus, or is there a need to do it sooner? | 13:00 |
Wizzup | because I'm not kidding if I say it takes me 1-2 full days to switch us to new mesa, even with no code backporting :p | 13:00 |
Wizzup | btw, I think we took a bunch of notes on atrix2 to make some dts changes, I think you hosted those somewhere, right? | 13:00 |
uvos__ | i dont think i have ever hosted any notes | 13:12 |
uvos__ | i only have the related files hosted | 13:12 |
uvos__ | https://uvos.xyz/maserati/stockinfo/ | 13:12 |
Wizzup | yeah that counts I guess | 13:18 |
inky | gsmarena has this info about mz617: it has gsm connectivity, but it cannot do neither voice nor sms. what does that mean? (i personally won't put a sim card in it anyway). | 13:40 |
inky | also, i think it doesn't list the ram size. | 13:40 |
Wizzup | it means that you can use it for data | 13:42 |
tmlind | cool tablet install images sounds great :) | 13:45 |
inky | Wizzup, oh thank you. but why sms won't work? just curious. or will (should) it work under linux? | 13:48 |
inky | also if you work on it now, what's the ram size? | 13:48 |
inky | and the graphics card is powervr? | 13:48 |
sicelo | inky: there are data only modems ... e.g. what's inside those 'mifi' thingies. | 13:58 |
sicelo | of course it may also just be that there's no software for it :-) | 13:58 |
Wizzup | tony said the modem or tablet hung when issuing sms or call commands iirc | 14:01 |
Wizzup | uvos: I pushed one more commit to the 6.6 builddeb | 14:04 |
Wizzup | there is also 6.6 in experimental https://maedevu.maemo.org/pkgweb/chimaera-experimental/main/binary-armhf/linux-image-omap.html | 14:05 |
Wizzup | but I think this is an older version, without droid3 stuff, not that it matters | 14:05 |
buZz | Wizzup: searched for those carmounts yet? | 14:16 |
Wizzup | buZz: can only find droid3 ones | 14:18 |
buZz | aw :( | 14:19 |
Wizzup | but I am going to the US in ~4-5 days, so ... | 14:19 |
Wizzup | freemangordon: ping | 14:48 |
sicelo | < Wizzup> or the librem soon (looks at sicelo :D) | 15:01 |
* sicelo looks outside the window :p | 15:01 | |
Wizzup | :p | 15:08 |
Wizzup | uvos: experimental has the latest 6.6 based on your work with the d3 stuff mixed in, no btrfs yet | 20:00 |
uvos | Wizzup: check | 20:00 |
Wizzup | it's not the latest 6.6 code or anything | 20:01 |
Wizzup | sicelo: today on my mz617 I also did not find the usb0 interface on the tablet to be up | 21:25 |
Wizzup | it did expose the rndis interface, but did not bring it up nor assign an ip | 21:25 |
Wizzup | I guess I should check if pcsuite-enable.sh fails or something | 21:26 |
Wizzup | tmlind: thinking about it some more, I think it's pretty unlikely that both my tablets have damaged cables wrt ts not always working | 21:26 |
Wizzup | tmlind: I could bring in a third one that I haven't opened up to see how that works | 21:26 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!