**** BEGIN LOGGING AT Sun Oct 16 02:59:58 2016 Oct 16 03:21:53 lol @ li Oct 16 03:44:58 lil Oct 16 08:08:42 Wizzup: tried both pali's latest (v4.6) and mainline linux (v4.8) Oct 16 08:08:57 still unlucky with booting. maybe it's just my device Oct 16 08:09:14 i used rx51_defconfig and omap2plus_defconfig (just changed the cmdline to yours) Oct 16 08:10:08 uh Oct 16 08:10:09 hm Oct 16 08:10:18 did you diff the generated configs? Oct 16 08:10:26 I'd do make oldconfig from my prev config Oct 16 08:10:29 and then diff it Oct 16 08:11:34 hmm ok, i'll try Oct 16 08:16:08 I'm going to do it somewhere today as well Oct 16 08:16:20 also, do you create a uImage as well? Oct 16 08:17:32 yep Oct 16 08:17:46 i'm setting up your config to build for arm-sdk so i'll see what happens Oct 16 08:31:34 I compile on my laptop :p Oct 16 08:32:01 I'm also considering getting another chromebook at some point - the one with nvidia tetra k1 Oct 16 08:34:18 nice Oct 16 08:34:26 i only tried the acer chromebook Oct 16 08:34:31 quite decent Oct 16 08:38:54 I think that is the acer one Oct 16 08:46:02 Wizzup: yes i think so too Oct 16 08:46:07 it's this one https://www.chromium.org/chromium-os/developer-information-for-chrome-os-devices Oct 16 08:46:14 oh, wrong paste Oct 16 08:46:25 https://www.chromium.org/chromium-os/developer-information-for-chrome-os-devices/acer-cb5-311-chromebook-13 Oct 16 08:46:27 this Oct 16 08:47:28 that one, yeah. Oct 16 08:47:45 it's good but the battery went to hell Oct 16 08:47:58 now it doesn't last for 5 minutes without being plugged in Oct 16 08:48:06 a friend of mine had exactly the same Oct 16 08:48:15 he returned it within the warranty Oct 16 09:24:03 Wizzup: no luck. i've still the same issue. comes to "freeing unused kernel memory" and just turns off after 10 seconds Oct 16 09:24:19 used your config and accepted defaults for (NEW) Oct 16 09:28:21 parazyd: ha! Oct 16 09:28:26 parazyd: you need the watchdog module Oct 16 09:28:39 at least, that is what it sounds like Oct 16 09:29:42 they all seem to be set Oct 16 09:29:57 right, but have you installed it/them? Oct 16 09:30:05 If they are build as module... you need the modules Oct 16 09:30:16 And they should be loadable at the start - which may be tricky if you have an encrypted root Oct 16 09:30:35 (in which case it should be either not build as module, or modules should be provided in the initramfs) Oct 16 09:30:56 actually, hm Oct 16 09:31:09 what you describe seems like the kernel not booting properly, and then the watchdog will kick in, so maybe that is not an issue Oct 16 09:31:16 ok, give me some time to try a newer kernel today Oct 16 09:31:34 p.s. Pali's latest seems to be https://github.com/pali/linux-n900/tree/v4.7-rc4-n900 Oct 16 09:31:40 http://sprunge.us/UjWO Oct 16 09:31:56 i was trying mainline 4.8 Oct 16 09:32:05 will try pali's now Oct 16 09:33:45 bbiab Oct 16 09:34:00 do you need CONFIG_WATCHDOG_NOWAYOUT? Oct 16 09:34:21 in that case make sure that some watchdog kicker daemon is running Oct 16 09:34:42 otherwise you will get device reboot after 30s Oct 16 09:37:27 Pali: not sure, using Wizzup's config. i'll try without Oct 16 09:37:49 it depends on your userspace configuration Oct 16 09:37:56 init daemons, etc... Oct 16 09:39:12 e.g. Maemo has dsme which with module for watchdogs autokicking Oct 16 09:39:45 on Debian is package watchdog with watchdog daemon which do same think... Oct 16 09:54:52 Pali: I think the problem he is experiencing is not actually related to the watchdog Oct 16 09:55:15 parazyd: do you have my kernel still? can you try it with my kernel? Oct 16 09:55:49 Pali: That is, I had a feeling that the kernels he compiled didn't get very far booting, so it's not like it fails randomly during agetty being visible Oct 16 09:56:21 Wizzup: yeah fails very early Oct 16 09:56:47 i have your 4.5.0rc5 Oct 16 09:57:05 i'll try Oct 16 09:57:22 just want to try 4.6 now, as it finished compiling Oct 16 09:57:30 If that one fails, then maybe your rootfs is not OK Oct 16 09:58:14 You can always disable the watchdog by putting the device in R&D mode. Oct 16 09:58:18 it's tough to debug it Oct 16 09:58:48 well, and setting the flags that say the watchdogs are disabled in that mode. Oct 16 10:00:04 I don't think that the watchdog is the problem. Oct 16 10:00:48 It might be related. Oct 16 10:01:12 My root filesystem takes far longer to mount that it would take fo the watchdog to kick in. Oct 16 10:01:22 It's very obvious when the watchdog kicks in - I've had it happen during openrc booting. It takes at least 30s. Oct 16 10:01:28 (on 2.6.28) Oct 16 10:01:36 and since my kernel has the watchdog compiled in the kernel, it should be fine Oct 16 10:01:45 (I have it working fine, and I don't have extra userspace tools for it afaik) Oct 16 10:02:06 but I do have NOWAYOUT=y Oct 16 10:03:11 I vaguely seem to remember thinking that the kernel didn't ping the watchdog during mounts under some version. Oct 16 10:03:22 Maybe in 2.6.28 Oct 16 10:03:38 We're talking about 4.5 Oct 16 10:03:38 (unless you ran something in userspace) Oct 16 10:04:29 I don't know what versions I'm talking about. Oct 16 10:04:50 It's just something I think I've experienced. Presumably there are reasons for those flags. Oct 16 10:05:00 afk Oct 16 10:05:19 some day I want to load grub for fun, from the u-boot Oct 16 10:05:52 luke-jr: I think you asked about the mainline blog post, getting closer now. doing a 'start by flashing n900 to mainline on n900' post now Oct 16 10:06:27 Pali: would you recommend using your 4.6-rc4 or your 4.7-rc4? Oct 16 10:06:47 sorry do not know now Oct 16 10:06:52 will try 4.7 first Oct 16 10:07:15 It would be good to know eventually how much your kernel is still different from the kernel.org mainline one Oct 16 10:08:14 Wizzup: you have 4.8 from before? Oct 16 10:08:33 I haven't tried/compiled 4.8 yet Oct 16 10:08:39 Wizzup: freemangordon mentioned almost all pali's stuff was upstreamed Oct 16 10:16:33 Wizzup: i think something's not right with my device. your kernel doesn't manage to boot either. it goes further in the boot process though, then rgb stripes all over the screen and dies Oct 16 10:17:27 last message is sshd urandom read Oct 16 10:18:58 not everything yet Oct 16 10:19:06 see: http://elinux.org/N900 Oct 16 10:39:33 parazyd: I have a few spare N900 devices, and I have a sd card that works. Easy to verify if your device is messed up. Oct 16 10:39:51 we can see at TI wednesday i guess Oct 16 10:40:17 yeah, I was thinking over going over today to play with openocd and the cortex-m debuggers, but not sure if I have the time Oct 16 10:56:02 Pali: is there something missing besides BT, DSP, cameras and SGX? Oct 16 10:56:46 Pali: iirc everything else is in 4.8 Oct 16 10:56:50 freemangordon: support for disabling input device (touchscreen) Oct 16 10:57:11 well, ok, but devices support is there Oct 16 10:57:19 the migt be some glitches, but still Oct 16 10:57:22 fm radio receiver Oct 16 10:57:23 *might Oct 16 10:57:32 but that depends on bt Oct 16 10:57:34 a prt of BT Oct 16 10:57:36 *part Oct 16 10:57:39 yes Oct 16 10:57:54 I can't remember anything else Oct 16 10:57:55 and maybe some alsa input/switches Oct 16 10:57:59 but do not know now Oct 16 10:58:18 ok, but all those are details, which should not matter much in xase of devuan Oct 16 10:58:40 the fuck, what's with my fingers today Oct 16 10:59:09 * freemangordon is going to try upstream 4.8 Oct 16 10:59:40 freemangordon: wifi bt coex mode Oct 16 10:59:52 and wifi permanent mac address Oct 16 11:00:09 musb host mode support Oct 16 11:00:31 hmm, yeah. what was the last status for wifi/bt coexist mode? Oct 16 11:00:41 do you remember> Oct 16 11:00:42 ? Oct 16 11:01:02 maintainers are not happy with sysfs attribute Oct 16 11:01:14 even if exactly same is for wl127x driver Oct 16 11:01:19 and they want it implemented how? Oct 16 11:01:26 netlink Oct 16 11:01:42 hmm, this is above my level I guess Oct 16 11:01:48 = totally incompatible with Maemo + totally incompatible with all other drivers Oct 16 11:02:13 every other driver has either sysfs attribute or module parameter Oct 16 11:02:36 and they are not happy with module parameter either? Oct 16 11:02:57 iirc not Oct 16 11:03:02 I see Oct 16 11:03:21 and I do not see reason to implement third different method for bt coex Oct 16 11:03:30 which will be used only by one driver Oct 16 11:03:44 so we will keep that patch out-of-tree? Oct 16 11:03:53 probably.. Oct 16 11:03:57 makes sense Oct 16 11:04:48 maybe we can start discussion again how to solve both those problems (how to set permanent mac address and confiure bt coex from userspace) Oct 16 11:05:05 I guess first we need BT driver in some form Oct 16 11:05:25 it does not depend on bt driver Oct 16 11:05:29 ok Oct 16 11:05:40 it is feature of wifi firmware Oct 16 11:06:03 well, lets start the discussion again then. But please, be polite on the ML :p Oct 16 11:06:27 then maybe you can start it :-) Oct 16 11:06:33 BTW, what has happened with your USB friend? how was his name... Oct 16 11:06:45 Pali: no, I lack the details for that Oct 16 11:07:25 btcoex is easy, to support both bt & wifi at same time you need to activate btcoex feature of wl1251 Oct 16 11:07:33 and code is already here Oct 16 11:08:12 permanent mac address: it is stored in CAL and currently we have userspace CAL NAND parser which can read it and do something with it Oct 16 11:08:38 which means we need interface to set permanent mac address from userspace for wlan0 Oct 16 11:09:10 can't we read CAL from the kernel? Oct 16 11:09:25 and it is stupid to implement proprietary Nokia's CAL format in kernel Oct 16 11:09:35 I know, but still Oct 16 11:09:37 and have it in kernel just for one device Oct 16 11:09:48 we have GPLv3 code Oct 16 11:09:56 which is incompatible with kernel's GPLv2 Oct 16 11:10:04 do clean room reimplementation would be needed Oct 16 11:10:12 libcal code is mine, I ca relicense it as much as I like Oct 16 11:10:12 plus lot of other stuff... Oct 16 11:10:22 *can Oct 16 11:10:47 I doubt that kernel's dev accept such horrible thing in kernel just for one device... Oct 16 11:11:06 iirc there was such an idea from the maintainers Oct 16 11:11:54 Pali: but yes, you are right, lets restart the discussion ane what the winds are Oct 16 11:12:17 /s/ane/and see/ Oct 16 11:12:25 ~ping Oct 16 11:12:25 1 packet transmitted, 1 packet received, 0.0% packet loss Oct 16 11:12:42 if every one device starts implementing its own format in kernel, then there will be 100 different formats... Oct 16 11:12:53 s/... without leading slash Oct 16 11:12:56 Pali: sure, I agree Oct 16 11:13:07 yeah, it seems today is my typo day Oct 16 11:50:58 Pali: regarding musb host mode, do you have that working in your kernel? Oct 16 11:51:31 Wizzup: no Oct 16 11:51:39 okay Oct 16 11:51:45 but I think it should be already supported Oct 16 11:51:57 ok, last time I tried it did not manage to get it to work Oct 16 11:52:00 we just needs to export musb registers via /sys/kernel/debug Oct 16 11:52:03 I'll try to do it again in some time in the future Oct 16 11:52:05 I see Oct 16 11:52:20 in my tree there is patch for it Oct 16 11:57:30 ok, I have a 4.5.0-rc5 kernel from your repo. I can try it again Oct 16 11:58:30 ~tabletsdev Oct 16 11:58:30 hmm... tabletsdev is http://tablets-dev.nokia.com/ http://wiki.maemo.org/Tabletsdev , http://tabletsdev.maemo.org (all defunct, thanks Nokia) or the nice site http://www.fladnag.net/downloads/telephone/n900/tools/, or http://www.mmnt.net/db/0/0/93.81.63.203/repositories/skeiron.org/skeiron.org/tablets-dev/maemo_dev_env_downloads, or http://maemo.muarf.org/tablets-dev/maemo_dev_env_downloads/ Oct 16 11:58:48 ~flasher Oct 16 11:58:49 i heard flasher is at http://www.jedge.com/n810/flasher/maemo_flasher-3.5_2.5.2.2.tar.gz (also .exe!), or http://www.chakra-project.org/ccr/packages.php?ID=5027 or generally http://www.google.com/search?q=maemo_flasher-3.5_2.5.2.2.tar.gz. HARMattan(N9): https://aur.archlinux.org/packages/fl/flasher-harmattan/flasher-harmattan.tar.gz, or -- list of filenames/md5sums: http://pastebin.com/sYKdNJSH, or http://galif.eu/nokia/ Oct 16 12:01:02 Wizzup: use newer, 4.5 has a nasty onenand bug Oct 16 12:02:08 alright, will go for a newer one Oct 16 12:03:34 Wizzup: what distro are you going to use? Oct 16 12:04:37 well, I have a phone right now with gentoo on it (for several months), I decided to turn it on again and just let it be on my wifi on a public v4, run some containers on it for fun Oct 16 12:04:45 I picked gentoo specifically for easy driver support/testing Oct 16 12:05:03 I've given parazyd that setup, and I think he's working on getting devuan to work on that Oct 16 12:05:29 I am asking becuse I just build vanilla 4.8 wit rx51_defconfig and booted devuan to login Oct 16 12:05:39 That's sweet! Oct 16 12:06:21 now I only need for somebody to tell me how to bring wifi up :) Oct 16 12:06:47 assuming I can;t enter digits from the kbd Oct 16 12:06:59 well, presuming the interface is there and the firmware is loaded Oct 16 12:07:08 parazyd: poke Oct 16 12:07:31 hmm, no wlan0, I guess modules are not loaded Oct 16 12:07:41 lets see for the usb0 Oct 16 12:07:48 freemangordon: if you can autocomplete, with ctrl+i you can autocomplete Oct 16 12:07:57 lemme try Oct 16 12:08:21 that sentence is weird, but I wanted to say that, if you are in a situation where you're typing something from the shell, and it can be autocompleted, then you can use ctrl+i as replacement for tab Oct 16 12:08:35 yeah, got it Oct 16 12:09:15 freemangordon: bastard! :D Oct 16 12:09:24 could my toolchain be borked in some way? Oct 16 12:09:38 could be Oct 16 12:09:43 parazyd: if you need some test devices, I can give you a few, would like to get them back at some point though Oct 16 12:09:50 I have several spares Oct 16 12:09:55 I am using gcc4.7.2 from maemo cssu-thumb scratchbox Oct 16 12:10:19 well, if he can't boot my 4.5.x kernel, something else may be up, since that used to work for him, afaik Oct 16 12:10:24 Wizzup: not necessary, we can only try one or two at TI Oct 16 12:10:27 (that was made from a native atm system using gcc 4.7) Oct 16 12:10:36 s/atm/arm/ Oct 16 12:10:36 Wizzup meant: (that was made from a native arm system using gcc 4.7) Oct 16 12:10:38 Wizzup: yes it did use to work, now it doesn't boot Oct 16 12:10:49 now, how to bring that wifi up dammit Oct 16 12:11:08 freemangordon: ifconfig wlan0 up? Oct 16 12:11:19 dmesg | grep wl1251 # to see if firmware is loaded? Oct 16 12:11:22 wpa_supplicant should be installed Oct 16 12:11:50 guys, I don't have any special symbol on the kbd Oct 16 12:11:51 ;) Oct 16 12:12:04 freemangordon: do you have access to the rootfs Oct 16 12:12:07 (sd card) Oct 16 12:12:11 Wizzup: sure Oct 16 12:12:19 what I did was, take it out, touch /123456789|<> Oct 16 12:12:29 and then use autocomplete + arrows + backspace to get the right symbols Oct 16 12:12:35 *shrug* it's painful Oct 16 12:12:46 yep, I'd prefer to load n900 keymap Oct 16 12:12:50 well, touch /aaa123456789|<> Oct 16 12:12:59 let me see if I already did that somewhere... Oct 16 12:13:05 lemme try to find it in rescueos image Oct 16 12:13:28 you can do loadkeys on rx51_us.map Oct 16 12:13:31 I have that file here... Oct 16 12:13:45 let me know if you can't find it, then I can upload it Oct 16 12:14:27 ]I have n900 keymap from rescueos Oct 16 12:14:38 freemangordon: could you share your kernel and modules for me to try? Oct 16 12:14:48 parazyd: sure, just a second Oct 16 12:16:31 parazyd: http://46.249.74.23/devuan/kernel-n900/ Oct 16 12:17:41 thanks! Oct 16 12:17:50 np Oct 16 12:18:03 parazyd: how to autoload n900 keymap? Oct 16 12:18:32 /etc/profile ? Oct 16 12:18:45 parazyd: ok :) Oct 16 12:19:03 parazyd: also, what is the preferred way to contribute to the sdk? Oct 16 12:19:49 parazyd: shall I create a user on git.devuan.org and fork and...? Oct 16 12:20:00 keymap not included :) Oct 16 12:20:08 you can if you wish Oct 16 12:20:21 there is also on https://github.org/dyne/arm-sdk Oct 16 12:20:32 ah, github Oct 16 12:20:32 if you prefer Oct 16 12:20:35 for me it's the same Oct 16 12:20:38 yeah, I prefer :) Oct 16 12:21:22 parazyd: could you elaborate on "(15,19,59) parazyd: keymap not included :)"? was that pointed to me? Oct 16 12:21:24 (awesome to see the devuan work - going afk for a bit) Oct 16 12:22:00 parazyd: you want n900 keymap as well? Oct 16 12:22:18 freemangordon: yes, and yes Oct 16 12:22:27 oh, sorry, ok Oct 16 12:22:46 freemangordon: and i'm still experiencing the same issue with your kernel :/ Oct 16 12:23:11 what is the issue? Oct 16 12:23:29 parazyd: http://46.249.74.23/devuan/ Oct 16 12:23:32 freeing unused kernel memory... 10 seconds ... poweroff Oct 16 12:23:49 parazyd: switch the device to RnD mode Oct 16 12:23:55 thanks for the keymap Oct 16 12:24:19 with the flasher, change rd flags for no watchdogs etc Oct 16 12:25:11 yay, even reboot seems to work under devuan :) Oct 16 12:25:59 freemangordon: how do i change the flags? i don't know how to do it with 0xffff Oct 16 12:26:17 ah Oct 16 12:26:19 found it :) Oct 16 12:26:29 use flasher-3.5 Oct 16 12:26:36 no idea about 0xffff Oct 16 12:30:17 parazyd: is there a way to change the default keymap during image building? keyboard-configuration or somesuch? Oct 16 12:30:39 yes it's possible to script it Oct 16 12:30:48 arm-sdk/boards/nokia-n900.sh Oct 16 12:31:00 can be added either in prebuild() or postbuild() Oct 16 12:32:13 ok, lets first make a fork on github to see if image would build here Oct 16 12:34:43 query me if you need help using Oct 16 12:34:56 ok Oct 16 12:36:26 disabling RnD on mine doesn't help Oct 16 12:37:27 does it boot maemo? Oct 16 12:37:32 yep Oct 16 12:37:41 kernel power and vanilla Oct 16 12:37:44 also, what is u-boot version you use? Oct 16 12:38:32 2013.04 Oct 16 12:38:42 parazyd: also, is you battery charged and in good condition? Oct 16 12:39:12 yeah Oct 16 12:39:23 weird Oct 16 12:39:23 i can charge another and try though Oct 16 12:39:40 lemme get the flags on my device to compare with yours Oct 16 12:41:25 parazyd: http://pastebin.com/WZX59jvm Oct 16 12:43:58 nope, with those fails as well Oct 16 12:44:01 * parazyd sighs Oct 16 12:45:13 doing power off is weird Oct 16 12:45:52 do you have anything in mtdlog partition? Oct 16 12:45:53 i'm starting over Oct 16 12:45:58 flashing vanilla Oct 16 12:46:05 should not matter Oct 16 12:46:16 devuan and kernel does not touch onenand Oct 16 12:47:30 ooh, I have dmesg log in /var/log/, sweet, now I can check what is loaded Oct 16 12:47:37 yeah i realize Oct 16 12:47:52 i got xorg working on devuan even Oct 16 12:47:56 (before) Oct 16 12:48:05 was only missing 3d accel though Oct 16 12:48:12 i'd kill for those blobs :D Oct 16 12:48:27 wait, there should be driver from TI Oct 16 12:48:39 yeah i had no real luck with it Oct 16 12:48:45 which is supposed to work Oct 16 12:48:48 really? Oct 16 12:48:52 but perhaps someone more experienced can make it work Oct 16 12:48:53 hmm Oct 16 12:49:07 well, I'll take a look when we reach that point Oct 16 12:49:23 parazyd: what is your device HW revision? Oct 16 12:49:52 2204 Oct 16 12:50:14 same here, so it MUST work :) Oct 16 12:50:54 i've really no idea why this is happening Oct 16 12:51:31 parazyd: http://pastebin.com/p7kFje8U Oct 16 12:52:09 :) Oct 16 12:52:26 Pali: isn't wl1251 driver supposed to be auto-loaded? Oct 16 12:52:38 it is Oct 16 12:52:52 not in my devuan image yet Oct 16 12:52:57 you sould add it to /etc/modules Oct 16 12:53:05 echo wl1251 >> /etc/modules Oct 16 12:53:10 you need also wl1251_spi Oct 16 12:53:11 IIRC Oct 16 12:53:39 it should have DT modinfo Oct 16 12:53:48 so based on DT alias it is loaded Oct 16 12:54:26 hmm, maybe I should do depmod -a? Oct 16 12:54:44 from within booted devuan when I change the kernel? Oct 16 12:54:59 Pali: ^^^? Oct 16 12:55:12 cant hurt Oct 16 12:55:15 depmod is needed after installation Oct 16 12:55:26 but someting like modules install calls it Oct 16 12:55:44 there is make modules_install Oct 16 12:55:46 yes Oct 16 12:56:23 when building image yes, but I just copied 4.8 modules on the sdcard :) Oct 16 13:06:28 whoa linux 4.8 for N900? Oct 16 13:06:46 mhm Oct 16 13:07:35 is N900 is mainline? Oct 16 13:07:38 in* Oct 16 13:07:48 sure Oct 16 13:08:04 nice Oct 16 13:08:05 I suppose everything except modem/telephony Oct 16 13:08:14 nobody needs those :P Oct 16 13:08:38 well, not exactly - cameras, BT, DSP and SGX are missing, while modem is supported :) Oct 16 13:08:58 rip phone Oct 16 13:10:22 &1& Oct 16 13:10:27 parazyd: there is no loadkma in devuan, as it is part of busybox, any idea how to loda that keymap? Oct 16 13:10:35 *loadkmap Oct 16 13:11:13 loadkeys Oct 16 13:11:37 not found, I guess I need package installed Oct 16 13:11:39 /usr/bin/loadkeys Oct 16 13:12:10 apt-get install kbd Oct 16 13:12:20 try that Oct 16 13:12:39 no way - there is no '-' symbol and network ;) Oct 16 13:12:51 I need to include that in the image Oct 16 13:13:13 add it to the extra_packages array in the script Oct 16 13:13:39 ok Oct 16 13:13:48 also add the wifi modules to custmodules then Oct 16 13:13:56 space separated Oct 16 13:14:05 no need to do it, depmod fixed it Oct 16 13:14:10 ah ok Oct 16 13:14:25 any common place in the sdk for device specific files? Oct 16 13:14:37 where to put n900 keymap in Oct 16 13:15:08 make a n900 dir in extra/ Oct 16 13:15:28 then call it as $R/extra/n900/n900.kmap from the script Oct 16 13:15:47 ok, thanks Oct 16 13:16:19 oh my! vanilla maemo. haven't seen this in a while Oct 16 13:16:34 gonna put a fresh uboot and see where it goes Oct 16 13:16:41 what are you guys doing anyway? putting devuan on a N900? Oct 16 13:17:16 yeah Oct 16 13:19:49 nice Oct 16 13:20:04 get it ready, i'll have a N900 soonish ;) Oct 16 13:20:43 same I hope Oct 16 13:21:04 but I'd rather something tried and tested Oct 16 13:21:24 I'll stick to Maemo Oct 16 13:21:52 should port as many maemo parts away from maemo and into n900-devuan ;) Oct 16 13:21:58 as possible Oct 16 13:22:23 also, from what I remember, the N900/maemo don't have proper OpenGL, yes? Oct 16 13:22:34 neither does maemo ;) Oct 16 13:22:41 only OpenGLES , i would assum Oct 16 13:22:46 e Oct 16 13:22:49 yep, gles2 Oct 16 13:22:52 I was thinking of SDL2 Oct 16 13:23:06 that might be obstacle in porting it, I think Oct 16 13:23:17 libsdl2 is in devuan :) Oct 16 13:23:32 ok, you have my interest Oct 16 13:23:37 hahahahaha Oct 16 13:23:40 doubt you can find any linux distro without that really Oct 16 13:24:12 but i guess many maemo parts lack source Oct 16 13:51:59 parazyd: how to add rx51_defconfig to the sdk, besides copying it in kernel-configs dir? Oct 16 13:52:22 I can't figure out what is the "sdk" way of copying it to kernel source directory Oct 16 13:57:51 oh, got it Oct 16 13:58:19 it should be called n900.config Oct 16 14:00:50 yeah Oct 16 14:00:55 see lib/helpers Oct 16 14:01:01 copy-kernel-config Oct 16 14:01:04 yeah, saw it Oct 16 14:01:12 cool Oct 16 14:01:34 trying to build it now, for 4.8.1 stable Oct 16 14:02:14 oh, I forgot to reload the build script after modifying it :( Oct 16 14:12:33 freemangordon: going well? Oct 16 14:13:37 still rough on the edges, will needs some more time :) Oct 16 14:14:23 it turned out copy-kernel-config needs to be called explicitly :) Oct 16 14:15:11 also, console-config wants me to select keyboard :) Oct 16 14:15:24 well, keyboard-configuration that is Oct 16 14:15:51 is it possible for me to connect in a tmux and see? Oct 16 14:15:59 i can show you what to do exactly Oct 16 14:16:12 I already figured it out Oct 16 14:16:53 ok Oct 16 14:17:14 add -E to any sudo calls that chroot or do something in the rootfs Oct 16 14:17:22 it will export necessary env vars Oct 16 14:17:46 but now it seems kernel make doesn't like my rx51_defconfig, it seems it is not the .config it expects Oct 16 14:18:34 you should use a complete config Oct 16 14:18:35 or Oct 16 14:18:42 copy n900.config to arch/arm/configs/rx51_defconfig and call it that way Oct 16 14:21:02 well, I copied it, did make rx51_defconfig and copied the resulted .config back to kernel-configs directory as n900.config. That should do the job, no? Oct 16 14:21:21 yep Oct 16 14:21:59 trying it Oct 16 14:23:40 parazyd: ok, kompiles compiles Oct 16 14:23:46 *kernel compiles Oct 16 14:25:31 nice Oct 16 14:36:18 parazyd: hmm, I don;t see uImage copied anywhere, weird Oct 16 14:36:42 it should be either in /root or /boot Oct 16 14:36:46 there is "sudo cp $CPVERBOSE uImage $strapdir/root/" Oct 16 14:36:52 no, it is not there Oct 16 14:36:53 if not, i think you're using the master branc Oct 16 14:37:08 i have 2-3 branches Oct 16 14:37:13 yes, which branch shall I use? Oct 16 14:37:14 wait, letmesee Oct 16 14:37:19 next Oct 16 14:38:03 here's my current version on my machine Oct 16 14:38:04 http://sprunge.us/YDUQ Oct 16 14:38:45 hmm, no such branch on github Oct 16 14:39:11 ok let me commit what i have then i'll push Oct 16 14:39:22 ok, thanks Oct 16 14:39:47 parazyd: where is upstream git repo? Oct 16 14:39:54 (I lost the address) Oct 16 14:40:48 my real upstream is git://git.parazyd.cf/arm-sdk.git Oct 16 14:40:58 but anyway Oct 16 14:41:03 branch is now pushed to GH Oct 16 14:41:12 ok Oct 16 14:50:21 parazyd: hmm, uImage is supposed to be copied in /, not in /root Oct 16 14:50:35 unless i am missing something Oct 16 14:50:38 yes i remember, this was before we talked about it :) Oct 16 14:50:45 change it to /boot Oct 16 14:50:45 ah, ok Oct 16 14:51:00 boot is vfat partition? Oct 16 14:51:08 /boot is the root of mmcblk0p1 Oct 16 14:51:12 yes Oct 16 14:51:15 ok Oct 16 15:00:40 parazyd: it boots :), now I need some help on how to set default keymap Oct 16 15:02:12 add "loadkeys /path/to/keymap" to /etc/profile Oct 16 15:02:27 loadkeys does not like the keymap file :( Oct 16 15:02:39 hmm why? Oct 16 15:02:54 "syntax error" Oct 16 15:03:34 parazyd: it is binary keymap :) Oct 16 15:04:03 so I guess I need loadkeys -b Oct 16 15:04:47 probably Oct 16 15:05:30 hmm, no, it *kreates* a binary keymap Oct 16 15:05:33 *creates Oct 16 15:06:08 so what about loadkmap? Oct 16 15:06:27 it is part of busybox iiuc Oct 16 15:06:37 and there is no bb in devuan installed Oct 16 15:08:57 try installing busybox, loading the keymap and then try to dump it with dumpkeys maybe? Oct 16 15:09:39 see dumpkeys(1) Oct 16 15:10:10 hmm, right Oct 16 15:11:42 or I can do it directly from maemo Oct 16 15:12:37 no dumpkeys in maemo Oct 16 15:37:29 freemangordon: sorry was having dinner Oct 16 15:38:06 here's dumpkeys from maemo: http://sprunge.us/RICi Oct 16 15:38:27 now you can have < > instead of pound and euro sign :) Oct 16 15:38:40 also add the pipe char somewere Oct 16 15:41:56 parazyd: I have that already, sorry for not telling you :) Oct 16 15:42:07 I was trying to bring wlan1 up Oct 16 15:42:36 mmm yeah i need to put in the udev hack Oct 16 15:42:49 because wlanX will keep ++ Oct 16 15:49:58 what is the problem with that ++? Oct 16 15:51:27 wifi changes mac address on every boot Oct 16 15:51:45 so udev increments the number because it looks like a new device Oct 16 15:52:03 this is problem with the driver, we discussed it with pali ^^^ Oct 16 15:54:44 echo "#" > /etc/udev/rules.d/75-persistent-net-generator.rules Oct 16 15:54:57 this is the solution Oct 16 15:55:12 yes the driver generates a new address every boot, as mentioned Oct 16 16:00:33 parazyd: where to put keymap file? Oct 16 16:01:13 in /usr/share/keymaps/dir-of-choice Oct 16 16:01:50 no such dir exists, also it is for text keymaps, not binary Oct 16 16:02:18 as busybox seems to be installed anyway, I'd better skip kbd package Oct 16 16:02:36 ok, in any case i can play with it later Oct 16 16:02:44 and use "busybox loadkmap < $kmap" from a script in /etc/profile.d Oct 16 16:03:02 sounds good Oct 16 16:03:05 I will do it, the only thing I am missing is where to place that kmap file Oct 16 16:04:03 maybe simply in /etc Oct 16 16:04:17 or /root/ Oct 16 16:04:42 "/etc" seems better. will place it there Oct 16 16:07:37 ack Oct 16 16:26:00 parazyd: https://github.com/dyne/arm-sdk/pull/1 Oct 16 16:31:06 freemangordon: fantastic! merged :) big thanks! Oct 16 16:32:29 parazyd: does it refuse to boot on your device still? Oct 16 16:33:03 i will build yours now and see Oct 16 16:33:38 parazyd: also, which packages do I need in order to try to bring X up? Oct 16 16:33:47 only lxde working? Oct 16 16:35:33 xserver-xorg xserver-xorg-video-omap xserver-xorg-input-evdev xinit should do the trick Oct 16 16:36:23 xserver-xorg-video-fbdev also won't hurt Oct 16 16:38:35 ok, lemme first try to connect to my home wifi :) Oct 16 16:42:29 parazyd: no wl1251 firmware? Oct 16 16:43:50 hmm? it's there in the config Oct 16 16:44:03 the driver. you need separate firmware? Oct 16 16:44:19 yep, firmware-ti-connectivity deb it seems :) Oct 16 16:45:30 that exists only in testing Oct 16 16:45:37 rather include it manuall Oct 16 16:45:43 how? Oct 16 16:46:20 put it in extra? Oct 16 16:46:28 yeah Oct 16 16:46:31 get it from fremantle Oct 16 16:46:58 no such package there iirc Oct 16 16:47:09 I can get firmware files though Oct 16 16:47:23 Pali: is wl1251 nvs fw device specific? Oct 16 16:47:34 wl1251-firmware Oct 16 16:47:35 no, device specific Oct 16 16:47:43 Pali: hmm? Oct 16 16:47:51 not firmware, but device specific Oct 16 16:48:01 the meego package should be used if possible Oct 16 16:48:07 their license permits redist Oct 16 16:48:13 Pali: every device has different nvs? Oct 16 16:48:19 yes Oct 16 16:48:21 OR Oct 16 16:48:26 nvs is stored in CAL Oct 16 16:48:36 ah Oct 16 16:48:45 i'll just include the debian .deb Oct 16 16:48:48 wl1251-fw.bin -> firmware Oct 16 16:48:56 wl1251-nvs.bin -> device calibration data Oct 16 16:49:19 wl1251-fw.bin in linux-firmware repository exists and contains "example" file Oct 16 16:49:27 s/wl1251-fw.bin/wl1251-nvs.bin/ Oct 16 16:49:27 Pali meant: wl1251-nvs.bin in linux-firmware repository exists and contains "example" file Oct 16 16:49:47 Pali: are you sure? as what I see on my device is that nvs file has the same date as fw file Oct 16 16:49:54 it is not used Oct 16 16:50:03 wl1251-nvs.bin is "example" file Oct 16 16:50:08 aah, I see Oct 16 16:50:24 wl1251_cal reads REAL data from CAL and send it to kernel driver Oct 16 16:50:30 ok Oct 16 16:50:42 but... only on maemo kernel Oct 16 16:50:43 I guess we'll need to build that for devuan Oct 16 16:50:49 yeah Oct 16 16:50:56 upstream kernel does not have support for loading via maemo specific interface Oct 16 16:50:58 missing interface in the upstream? Oct 16 16:51:03 right Oct 16 16:51:14 but upstream kernel can load wl1251-nvs.bin via firmware_request function Oct 16 16:51:35 ok, so we can generate it on-the-fly? Oct 16 16:51:37 and so you can hook up udev to supply own script which provide those data Oct 16 16:52:01 instead reading it from disk Oct 16 16:52:18 can;t we just write the file with the data from CAL? Oct 16 16:52:49 it is stupid to oeverwrite files in /lib/.... to something device specific Oct 16 16:53:27 I guess we can ln it to a file in /var and create that file if it doesn;t exist Oct 16 16:53:51 or just create it in /lib on the very first boot Oct 16 16:54:32 correct way is to write that udev script Oct 16 16:54:52 which read nvs data from cal and then send them it to kernel Oct 16 16:54:58 ok, ok :) Oct 16 16:55:02 parser is already written: https://github.com/community-ssu/wl1251-cal Oct 16 16:56:01 parazyd: what is the "politically correct" way to include device specific packages in devuan? Oct 16 16:56:40 you mean custom debs? Oct 16 16:56:40 parazyd: also, is building packages a part of the sdk? Oct 16 16:56:45 parazyd: yes Oct 16 16:57:04 see ^^^ Oct 16 16:57:27 i had it implemented in legacy branch, but not yet in master/next where i'm now using libdevuansdk (which i'm also writing) Oct 16 16:57:33 give me a few minues and i'll implement it Oct 16 16:57:38 package building, not yet Oct 16 16:57:43 but soom(tm) Oct 16 16:57:51 ok, but here it is supposed to be build? Oct 16 16:57:57 *where Oct 16 16:58:05 on the device itself? Oct 16 16:58:13 you do it yourself i guess Oct 16 16:58:24 hmm, ok Oct 16 16:58:31 the debian directory is already there Oct 16 16:59:01 parazyd: can't parse, could you elaborate on that debian dir? Oct 16 16:59:10 https://github.com/community-ssu/wl1251-cal Oct 16 16:59:14 oh, sure Oct 16 16:59:16 debian dir is the files needed Oct 16 16:59:25 yeah, got it now Oct 16 16:59:41 I know I can build that in maemo SDK, but what about devuan SDK? Oct 16 16:59:58 not yet Oct 16 17:00:02 ok Oct 16 17:00:19 i'm the only one writing the stuff, so time management isn't quite good :D Oct 16 17:00:30 I know ;) Oct 16 17:08:32 parazyd: how to install a deb? Oct 16 17:08:44 dpkg -i package.deb Oct 16 17:08:57 I mean - during image creaton :) Oct 16 17:09:07 i'm implementing it right now Oct 16 17:09:20 oh, ok, sorry Oct 16 17:16:51 ok, wireless is up :) Oct 16 17:17:48 nice Oct 16 17:34:51 parazyd: also, X seems to work, unfortunately there is no enough free space on /. How to make root bigger? Oct 16 17:35:21 you should do it from maemo/your computer Oct 16 17:35:28 unmount the root partition Oct 16 17:35:52 repartition it with parted/cfdisk, resize2fs on the partition and finally e2fsck on the partition Oct 16 17:36:24 parazyd: I was asking about the image Oct 16 17:36:38 how to create an image with bigger ext4 partition Oct 16 17:36:48 ah you want to include it when building? Oct 16 17:36:52 sure Oct 16 17:36:54 just change the size variable Oct 16 17:36:57 in the script Oct 16 17:37:01 now is 666 Oct 16 17:37:05 ;D Oct 16 17:37:10 hmm, good number :) Oct 16 17:37:36 i pushed deb support btw Oct 16 17:37:43 great Oct 16 17:37:46 just drop .deb files in extra/$device_name Oct 16 17:37:58 thanks, will do Oct 16 17:38:06 update libdevuansdk also Oct 16 17:38:13 git submodule update or whatever it's called Oct 16 17:53:46 freemangordon: the new image still doesn't boot for me. i'll see what's up when i meet Wizzup Oct 16 17:53:56 :( Oct 16 17:54:00 weird Oct 16 17:54:07 yeah Oct 16 17:54:18 dumb reason, could the sd card be too big? Oct 16 17:54:21 it's 32G Oct 16 17:54:40 should not matter, though, who knows Oct 16 17:54:59 yeah... i'll see wednesday and report back Oct 16 17:55:11 Wizzup: do you want me to upload devuan image for you to test if it boots on your device? Oct 16 17:55:30 hello Oct 16 17:56:20 parazyd: hmm, dropping the .deb in extra/n900 kade no change Oct 16 17:56:23 *made Oct 16 17:58:32 did you pull the latest commit? Oct 16 17:58:37 yes Oct 16 17:58:47 you should see install-custom-packages function Oct 16 17:58:49 at least I think so Oct 16 17:58:49 yeah Oct 16 17:58:52 hmm... Oct 16 17:58:59 lemme check Oct 16 17:59:16 hmm, no such function Oct 16 17:59:31 pull again Oct 16 17:59:37 from github Oct 16 17:59:56 git pull && git submodule update did the trick for me on my server Oct 16 18:00:22 ah, I see, it is not "just drop .deb files in extra/$device_name", it seems I have to list them, right? Oct 16 18:01:19 or not Oct 16 18:02:19 list? no Oct 16 18:02:37 it's supposed to automagically find them Oct 16 18:02:41 unless i missed something Oct 16 18:02:42 ok, I pulled again, seems it was still not pushed when I pull it early Oct 16 18:05:00 it works Oct 16 18:05:02 just tested Oct 16 18:05:38 yeah, seems your latest changes were not pulled 10 minutes ago, now everything seems in -place Oct 16 18:05:43 cool Oct 16 18:06:02 will try to make an image with lxde pre-installed Oct 16 18:06:46 BTW for some reason TS y axis is inverted :) Oct 16 18:10:07 freemangordon: here's Wizzup's advice http://sprunge.us/GJRh Oct 16 18:10:56 yeah, makes sense Oct 16 18:11:33 anything you wish, just drop in extra Oct 16 18:11:40 yep Oct 16 18:11:49 i keep the dist images minimal, but everything can stay commented in the script Oct 16 18:11:56 for ppl to have :) Oct 16 18:12:19 sidenote, i tried a class 4 4GB microsd, still the same Oct 16 18:12:28 it's some hardware issue now Oct 16 18:12:28 well, I think a tool to bring wireless up on n900 is a must Oct 16 18:12:30 has to be Oct 16 18:14:16 add this to prebuild() to stop wlanX incrementing on boot btw: print "#" | sudo tee $strapdir/etc/udev/rules.d/75-persistent-net-generator.rules Oct 16 18:14:37 ok Oct 16 18:16:20 do you like this concept of the sdk in general? Oct 16 18:16:28 i call it "simple distro kit" :) Oct 16 18:17:21 yes, the fact it took me less than an hour to build my first image (with upgraded kernel) means it is usefull Oct 16 18:17:48 it only lacks custom debs build functionality aiui Oct 16 18:18:01 that will come Oct 16 18:18:14 i'm still learning how to make debian packages Oct 16 18:18:22 * parazyd is not really a debian user Oct 16 18:18:39 I can try to make devuan based maemo distro Oct 16 18:19:09 lxde install takes a while :D Oct 16 18:23:04 parazyd: BTW, who provides those funny notes while image is being build? Oct 16 18:24:51 i got them from mailpile, then just reimplemented in shell :) Oct 16 18:24:56 sometimes they're called Oct 16 18:25:06 see bottom of lib/helpers Oct 16 18:25:40 ok Oct 16 18:27:03 actually libdevuansdk/zlibs/helpers Oct 16 18:27:04 sorry :D Oct 16 18:28:11 no problem, it is not important, I was just curious. Actually it is a very good idea to have some fun while waiting Oct 16 18:28:43 heheh Oct 16 18:31:30 freemangordon: sure, I can test that Oct 16 18:31:48 lots of backlog to catch up with, too. might not manage to do it tonight Oct 16 18:32:24 ok, I will put it on my web server and will query you to test it when you have time Oct 16 18:32:57 parazyd: "install-custom-packages:14: command not found: install-custdebs" Oct 16 18:33:28 you didn't do git submodule update Oct 16 18:34:03 I did Oct 16 18:34:12 helpers file is updated Oct 16 18:34:19 O.o Oct 16 18:34:28 what about libdevuansdk Oct 16 18:34:34 did it checkout to origin? Oct 16 18:34:43 hmm, what? Oct 16 18:34:45 git submodule update --recursive --remote Oct 16 18:34:45 Skipping unmerged submodule lib/libdevuansdk Oct 16 18:34:56 what is that supposed to mean? Oct 16 18:34:59 the commit should be d28e0a93f4561099adc6c270027ed3331ce44489 Oct 16 18:35:19 i dunno Oct 16 18:35:28 --recursive --remote isn't really necessary Oct 16 18:40:56 freemangordon: did you figure out how to load keys? Oct 16 18:41:19 Wizzup: yeah, through busybox Oct 16 18:41:38 though it doesn;t work in X Oct 16 18:42:20 parazyd: somehow I managed to update libdevuansdk to latest version Oct 16 18:42:26 yes, X is different Oct 16 18:42:28 that's setxkbmap Oct 16 18:42:29 X needs its own mapping Oct 16 18:42:33 yeah Oct 16 18:42:38 freemangordon: what was the reason? Oct 16 18:42:40 I didn't do that yet Oct 16 18:42:45 s/reason/solution/ Oct 16 18:42:46 parazyd meant: freemangordon: what was the solution? Oct 16 18:42:48 no idea, some git weirdness Oct 16 18:42:53 aah Oct 16 18:43:05 git reset --hard on the submodule Oct 16 18:43:20 /o\ Oct 16 18:43:21 git reset HEAD that is Oct 16 18:43:24 ok :) Oct 16 18:43:38 and then git submodule update Oct 16 18:43:43 anyway, it will do the thing if you `git pull && git submodule update` Oct 16 18:43:59 why it was thinking the submodule was modified is beyond my understanding :) Oct 16 18:44:20 beats me Oct 16 18:44:56 Wizzup: btw, if you ever feel like baking gentoo images, feel free to try out arm-sdk. should be a lot simpler to do gentoo than devuan actually Oct 16 18:45:13 there is 12 devices already Oct 16 18:46:14 ok, custom .deb seems to be installed now Oct 16 18:46:24 great Oct 16 18:47:05 parazyd: hm - what would I use it for? Oct 16 18:48:47 don't know really. just a thought since you like to play with ARM boards Oct 16 19:58:08 parazyd: lxde works :) Oct 16 19:58:41 awesome :) Oct 16 20:19:42 freemangordon: will you submit the wifi firmware blobs to the repo or should i? Oct 16 20:20:34 parazyd: I just installed firmware-ti-connectivity_0.43_all.deb Oct 16 20:20:46 the one from debian? Oct 16 20:20:51 yes Oct 16 20:20:57 all right. and modules autoloaded? Oct 16 20:21:07 yes Oct 16 20:21:19 cool :) many thanks for your work today! Oct 16 20:21:54 well, modules are auto-loaded without firmware as well, but when you try to connect with wpa_supplicant it complains because of the missing FW **** ENDING LOGGING AT Mon Oct 17 02:59:58 2016