**** BEGIN LOGGING AT Tue Aug 28 03:00:03 2012 Aug 28 07:15:01 heyho Aug 28 07:15:26 hey Aug 28 07:32:17 SHR: 03Martin.Jansa 07shr-chroot * r45cfbf2ae85b 10/ (99 files in 16 dirs): system upgrade Aug 28 07:41:12 SHR: 03Martin.Jansa 07shr-chroot * r16495d5d21ef 10/ (192 files in 18 dirs): system upgrade Aug 28 09:41:49 SHR: 03Martin.Jansa 07shr-makefile * rb17554f3e453 10/Makefile: Makefile: move oebb.sh update after comparing layers and bblayers Aug 28 10:20:50 SHR: 03Martin.Jansa 07meta-smartphone * rf861ffccfb53 10/ (9 files in 9 dirs): README: update meta-oe/oe-core URL to github Aug 28 10:23:52 SHR: 03Martin.Jansa 07meta-smartphone * rbd2fbcd76360 10/meta-android/README: meta-android: update README Aug 28 11:25:57 SHR: 03Martin.Jansa 07meta-smartphone * rb9213bfddfed 10/meta-samsung/recipes-core/systemd/ (4 files in 3 dirs): meta-samsung: systemd: DON'T use MACHINE variable in TUNE_PKGARCH recipes! Aug 28 11:48:13 SHR: 03Martin.Jansa 07shr-chroot * ra2f273334d81 10/OE/bin/ (sstate-cache-sync.sh sstate-diff.sh): bin: improve sstate scripts Aug 28 12:09:54 JaMa, my disablefbcon script was not in the image.... Aug 28 12:09:56 hi btw Aug 28 12:11:42 (the 094) Aug 28 12:11:50 I'll look if there is a 095 Aug 28 12:13:23 hmmm no image in 095 Aug 28 12:13:27 only kenrel Aug 28 12:15:48 hi morphis Aug 28 12:23:26 GNUtoo: heyho Aug 28 12:23:48 I will continue to try to fix crespo but with -b it's really hard Aug 28 12:24:07 else we coould disable framebuffer console Aug 28 12:24:10 GNUtoo: just for you to notice: http://git.shr-project.org/git/?p=meta-smartphone.git;a=commit;h=b9213bfddfedff1315697a6a9d1a599c2a289b23 Aug 28 12:24:11 and reenable it later? Aug 28 12:24:27 ah ok Aug 28 12:24:30 why not Aug 28 12:24:34 it broke sstate Aug 28 12:25:21 I'll pull then Aug 28 12:25:22 ok Aug 28 12:25:32 also we have a problem at boot Aug 28 12:25:37 the screen becomes black Aug 28 12:25:39 apm -s Aug 28 12:25:44 + pressing power button Aug 28 12:25:46 fixes it Aug 28 12:25:54 but you need to ssh inside etc.... Aug 28 12:26:02 to issue apm -s Aug 28 12:28:21 GNUtoo, hello sir Aug 28 12:28:26 hi Aug 28 12:29:01 if you don't off the screen, /sys/class/leds/lcd-backlight/brightness can works Aug 28 12:29:11 ? Aug 28 12:29:23 but about cat /dev/urandom only has effort in android Aug 28 12:29:45 ? Aug 28 12:30:29 if you don't power off the screen, how does /sys/class/leds/lcd-backlight/brightness fails? Aug 28 12:30:44 /sys/class/leds/lcd-backlight/brightness can control the brightness of the screen Aug 28 12:30:58 but you shouldn't off screen before Aug 28 12:31:32 and in android. cat /dev/urandom > /dev/graphica/fb0 works, but not in linux Aug 28 12:31:51 ok Aug 28 12:32:01 did you look at recovery's sources? Aug 28 12:32:14 yes, but it seems it need libc Aug 28 12:33:13 ps, once echo 4 > /sys/class/graphics/fb0/blank, echo > /sys/class/leds/lcd-backlight/brightness doesn't work Aug 28 12:34:06 it needs bionic Aug 28 12:34:10 try to compile it Aug 28 12:34:16 and run the result Aug 28 12:34:24 do you still have the link? Aug 28 12:34:39 look in graphics.c in the repo I pasted you the link to Aug 28 12:36:45 yes, I can see the log, but one problem, how to compiler it, it doesn't have any configure script and makefile Aug 28 12:44:12 where's the link Aug 28 12:44:18 because I know how to compile it Aug 28 12:44:32 you must checkout all the android source with the repo command Aug 28 12:44:54 please send me the link to the github project again(I don't remember where it was) Aug 28 12:45:03 and I'll tell you the right commands Aug 28 12:51:13 GNUtoo, https://github.com/agrabren/android_bootable_recovery/blob/ics/minui/graphics.c Aug 28 12:51:23 GNUtoo: see my last commit in meta-smartphone please Aug 28 12:51:34 GNUtoo: now I'm fixing the same for gpsd Aug 28 12:51:39 JaMa, yes I will study it Aug 28 12:52:44 JaMa, I've some questions about it Aug 28 12:52:51 because I didn't totally understand Aug 28 12:53:00 what is OEBasicHash exactly already? Aug 28 12:53:45 2)TUNE_PKGARCH means armv7? Aug 28 12:54:16 and why using MACHINE_ARCH_crespo is bad? Aug 28 12:54:22 in TUNE_PKGARCH Aug 28 12:54:38 2) yes Aug 28 12:54:41 ok Aug 28 12:54:47 MACHINE_ARCH_crespo is bad, yes Aug 28 12:55:05 it works with OEBasic, but doesn't work very well with OEBasicHash Aug 28 12:55:15 ah ok Aug 28 12:55:20 check sstate-diff.sh script from shr-chroot Aug 28 12:55:24 what's OEBasic and OEBasicHash? Aug 28 12:55:42 basically I know what's sstate Aug 28 12:55:48 but I've no idea of its implementation Aug 28 12:55:55 or rather Aug 28 12:56:01 it's *current* implementation Aug 28 12:56:06 http://git.shr-project.org/git/?p=shr-chroot.git;a=blob;f=OE/bin/sstate-diff.sh;h=a5e2382c1c7d4936e5a876dfa82de251f4294b3e;hb=a2f273334d8108f5435a6dcda450bd74cbe98be5 Aug 28 12:56:39 OEBasicHash is default for current oe-core and will became mandatory soon (because of PR server) Aug 28 12:56:55 OEBasic is used by SHR now, but I'm testing OEBasicHash to be prepared Aug 28 12:57:25 and that commit forced everything depending on systemd (even transitive deps) to rebuild for each machine Aug 28 12:57:55 ok Aug 28 12:58:00 and the difference is that OEBasic creates stamps/*/foo_PV_PR Aug 28 12:58:09 ok Aug 28 12:58:16 but OEBasicHash creates stamps/*/foo_sstate-checksum Aug 28 12:58:21 ok Aug 28 12:58:36 so OEBasic rebuilds stuff only when PE/PV/PR is changed Aug 28 12:58:42 ahhh Aug 28 12:58:45 OEBasicHash on every sstate checksum change Aug 28 12:58:51 ok Aug 28 12:59:09 so with OEBasicHash....you change something in a recipe without bumping PR and it get rebuild? Aug 28 12:59:50 so every MACHINE specific change to some recipe deep in dependency tree forces everything (including all armv7a recipes) to rebuild after MACHINE change Aug 28 12:59:54 GNUtoo: yes Aug 28 13:00:01 ok so it was that thing Aug 28 13:00:08 I just didn't associate it with a name Aug 28 13:00:18 GNUtoo: and PR server will also change PR so it's upgraded on target when sstate checksum is different Aug 28 13:00:31 ok Aug 28 13:00:41 I don't like that at all :) Aug 28 13:00:54 would we be able to build without internet ? Aug 28 13:04:21 JaMa, it fails Aug 28 13:04:23 GNUtoo: not sure, depends on implementation Aug 28 13:04:37 http://pastie.org/private/9twsljkoiec4mcwr3yyuw Aug 28 13:04:45 GNUtoo: my plan is to lock oe-core revision before they make PR server mandatory :/ Aug 28 13:04:53 ok Aug 28 13:06:47 GNUtoo: mmt, will push fix Aug 28 13:06:51 ok Aug 28 13:06:55 thanks a lot Aug 28 13:07:21 repo init -u https://github.com/agrabren/android_bootable_recovery.git -b ics ? Aug 28 13:07:22 SHR: 03Martin.Jansa 07meta-smartphone * r452ec9db72d1 10/meta-samsung/recipes-core/systemd/systemd-machine-units_1.0.bbappend: meta-samsung: systemd-machine-units: fix do_install Aug 28 13:07:25 using git.shr? org github? Aug 28 13:07:28 GNUtoo: ^ Aug 28 13:07:34 ayaka, no Aug 28 13:07:39 ayaka, try to find the manifest Aug 28 13:07:43 and repo init that Aug 28 13:07:52 JaMa, git.shr Aug 28 13:07:56 then pushed Aug 28 13:07:59 ok thanks Aug 28 13:08:05 what is it ? Aug 28 13:09:29 it's a repository which contain an xml file with all the repository to fetch Aug 28 13:11:15 is it a dictionary? Aug 28 13:11:27 no Aug 28 13:11:32 it's an xml file Aug 28 13:13:51 ayaka, do that: Aug 28 13:13:59 repo init -u git://github.com/agrabren/android.git -b ics Aug 28 13:14:08 in a separate dir Aug 28 13:14:08 yes sir Aug 28 13:16:12 done, what to do next ? Aug 28 13:24:16 build it Aug 28 13:24:23 like for instance Aug 28 13:24:36 . build/envsetup.sh Aug 28 13:24:47 lunch cm_shooteru-eng Aug 28 13:24:50 make Aug 28 13:24:54 or something like that Aug 28 13:25:21 oh, I forget to repo sync, but please go on Aug 28 13:26:18 ok Aug 28 13:26:28 then fastboot boot the recovery Aug 28 13:26:34 and look if it works Aug 28 13:26:44 if so we have 2 ways to get the magic stuff: Aug 28 13:26:49 1) strace in android Aug 28 13:26:57 2) look at the source code Aug 28 13:28:06 I see Aug 28 13:29:10 JaMa, is my systemd unit for disablefbcon ok? Aug 28 13:31:10 GNUtoo: looks ok from OE pov Aug 28 13:31:26 I mean is it supposed to run Aug 28 13:31:26 haven't read it from systemd/functionality pov :) Aug 28 13:31:31 ok Aug 28 13:31:47 sorry I have to finish few other things first Aug 28 13:32:09 ok Aug 28 13:36:39 it seems it will sync very long time, after built, what is it ? a initramfs? Aug 28 13:36:50 JaMa, it doesn't seem to run..... Aug 28 13:36:55 I'll wait for you Aug 28 13:37:11 ayaka, you have the recovery and also android itself Aug 28 13:37:51 ? sorry, I can't understand, maybe that time I will know Aug 28 13:38:16 you asked that: Aug 28 13:38:16 after built, what is it ? a initramfs? Aug 28 13:38:46 it builds a recovery image wich is a kernel and an initramfs inside Aug 28 13:39:03 paulk-desktop, hi, do you know how to build only a recovery? Aug 28 13:39:34 hi Aug 28 13:39:36 make recoveryimage Aug 28 13:39:57 ok Aug 28 13:40:16 ayaka, do that when building ^^^ Aug 28 13:40:19 instead of make Aug 28 13:40:20 thanks Aug 28 13:40:44 yes sir and thank you sir Aug 28 13:57:27 SHR: 03Martin.Jansa 07meta-smartphone * r953f9684faad 10/meta-openmoko/recipes-navigation/gpsd/ (8 files in 5 dirs): meta-openmoko: replace gpsd bbappend with gpsd-machine-conf recipe Aug 28 13:57:46 GNUtoo: gpsd fix pushed, please check it when you have time (but opkg upgrade went ok here) Aug 28 13:58:06 disablefbcon.service static Aug 28 13:58:17 in systemctl list-unit-files Aug 28 13:58:25 ok Aug 28 13:58:27 what about dropping shr branches from -contrib repos? to make sure that everybody switched to github? Aug 28 13:58:31 I'll check after crespo is fixed Aug 28 13:58:41 ah? Aug 28 13:58:48 sync doesn't work? Aug 28 13:58:56 because I like having our own repos Aug 28 13:59:18 I'm talking only about -contrib (oe-core/meta-oe) Aug 28 13:59:26 sync for git.shr* Aug 28 13:59:39 yes Aug 28 13:59:49 can't we sync -contrib too? Aug 28 14:02:47 I don't have rights for oe -contrib repos Aug 28 14:03:05 The unit files have no [Install] section. They are not meant to be enabled using systemctl. Aug 28 14:03:07 ah? Aug 28 14:03:08 ok Aug 28 14:03:16 why not switching to git.shr* Aug 28 14:03:37 because shr buildhost is too slow already Aug 28 14:03:49 ah ok Aug 28 14:03:56 then why not gitorious? Aug 28 14:03:59 but we can use that to push if we really want Aug 28 14:04:10 and add hooks there Aug 28 14:04:11 ok Aug 28 14:04:18 yes that would be cool Aug 28 14:04:22 gitorious is usually slower there then github Aug 28 14:04:24 ok Aug 28 14:04:40 (we had already stuff on gitorious such as the kernels) Aug 28 14:04:43 and official OE mirror is already on github Aug 28 14:05:06 yes but now we have "organization" on github (shr-distribution) Aug 28 14:05:17 ok Aug 28 14:05:21 which is better then my private account in case I'm hit by bus or something Aug 28 14:05:26 ok Aug 28 14:05:48 we can even move kernels to github if you want Aug 28 14:05:55 it's just that I don't like very much github because they don't give source code, so migrating away from it is harder Aug 28 14:05:57 or to fso git Aug 28 14:06:07 because there is few kernel trees already :) Aug 28 14:06:08 for fso git we should ask morphis Aug 28 14:06:11 yes Aug 28 14:06:20 morphis, what do you think Aug 28 14:06:59 GNUtoo: I think we should move kernels etc. to github Aug 28 14:07:09 ok Aug 28 14:07:11 they were always in the wrong place on fso.git Aug 28 14:07:19 hmmm Aug 28 14:07:28 but we have a fso-userland mailing list Aug 28 14:07:34 and also an fso-kernel mailing list Aug 28 14:07:39 and? Aug 28 14:07:40 basically SHR is a distro Aug 28 14:07:46 but fso can have kernels Aug 28 14:07:53 how to increase voice mic volume during calls on gta04 ? Aug 28 14:07:53 fso is a middleware Aug 28 14:07:56 so other distros uses fso ekrnels Aug 28 14:07:58 nothing to share with a kernel Aug 28 14:08:15 yes but we often modify the kernel to be able to run fso on it Aug 28 14:08:19 fso doesn't offer any kernel versions Aug 28 14:08:22 like led->backlight Aug 28 14:08:29 yes but we did this for SHR Aug 28 14:08:39 to run fso on SHR Aug 28 14:08:41 paulk-desktop, try some alsamixer controls Aug 28 14:08:49 sure, but which one? Aug 28 14:08:54 morphis, or any other distro Aug 28 14:09:07 paulk-desktop, can I look when I finish discussing + crespo booting? Aug 28 14:09:12 sure Aug 28 14:10:27 GNUtoo: yeah but fso as project doesn't provide any kernels Aug 28 14:10:36 it just hosts them and depend on them Aug 28 14:10:39 yes Aug 28 14:11:03 * GNUtoo prefers gitorious than github Aug 28 14:11:18 but then we have fetch issues with oe Aug 28 14:11:25 hmmm Aug 28 14:13:42 GNUtoo, sorry, sir maybe I cann't complete syncing today, but I will do make recoveryimage after it Aug 28 14:14:05 ok for systemd I'll add that: Aug 28 14:14:07 [Install] Aug 28 14:14:07 WantedBy=multi-user.target Aug 28 14:14:24 ok Aug 28 14:14:27 why not xserver-nodm? Aug 28 14:14:35 ayaka, also the "sir" is not needed Aug 28 14:14:39 because IIRC you have it in Before Aug 28 14:14:52 JaMa, I have already a Before= Aug 28 14:14:56 it just for respecting Aug 28 14:15:14 ayaka, ok, but it's not needed Aug 28 14:15:29 computer science english doesn't uses such formulas Aug 28 14:15:35 GNUtoo: if I recall systemd manuall correctly then Before= is not what you suppose it is Aug 28 14:15:44 I see, and thank you Aug 28 14:15:46 JaMa, ah? Aug 28 14:16:09 ayaka, for instance if you do that in an mail it will sound like SPAM Aug 28 14:16:31 JaMa, what should I do tough Aug 28 14:16:38 s/tough/then/ Aug 28 14:16:38 GNUtoo meant: JaMa, what should I do then Aug 28 14:16:45 GNUtoo: Requires/Wants is independent of After/Before Aug 28 14:16:45 to call somebody as sir? Aug 28 14:16:52 GNUtoo: read http://www.freedesktop.org/software/systemd/man/systemd.unit.html Aug 28 14:17:24 ayaka, "dear sir/madam etc..." is not very used..... Aug 28 14:17:53 JaMa, I was looking for a doc and ended up on arch wiki Aug 28 14:18:00 thanks a lot!!!! Aug 28 14:18:36 I see, don't know much about computer science english manners Aug 28 14:19:12 ok Aug 28 14:20:05 ayaka, when sending patches also maintainers can appear to be rude but that's sometimes necessary, because with too much politeness the person would continue in the wrong path and may suicide at the end Aug 28 14:20:54 for instance if the maintainer doesn't clearly tell that the path taken by the contributor is wrong.....the contributor will continue in that path Aug 28 14:21:21 and at the end after years of work if the patch is clearly refused it happens that he suicide Aug 28 14:21:22 could you tell me in a channel or PM, I think it will disturb this channel, I am wonder to learn that Aug 28 14:21:28 it happened with kernel patches Aug 28 14:21:47 like linus torvalds was too shy at the beginning Aug 28 14:23:04 I prefer talking about that here, to get peer review on what I say Aug 28 14:23:14 or to permit other people to learn it Aug 28 14:24:13 JaMa, ok, After is what I want Aug 28 14:24:24 but I'll also read abuot Requires/Wants Aug 28 14:24:48 ok, I am hearing Aug 28 14:24:56 GNUtoo: then I didn't understood when you want that script executed.. Aug 28 14:25:11 JaMa, I need it right before xserver-nodm.service Aug 28 14:25:16 else the phone crashes Aug 28 14:27:55 oh my godness , .repo has been more than 556MB, do I do the correct thing? Aug 28 14:28:53 GNUtoo: then it sounds like WantedBy=xserver-nodm.service, Before=xserver-nodm.service Aug 28 14:29:04 ok Aug 28 14:29:18 thanks a lot Aug 28 14:29:29 I got the WantedBy wrong then Aug 28 14:36:24 paulk-desktop, maybe look at phoneuid.conf Aug 28 14:38:15 in Building Embedded Linux Systems, I know the target vfor arm is ARM, but what is armhf, I am sorry, I don't find out the glibc manual Aug 28 14:38:30 paulk-desktop, which mic do you need? Aug 28 14:40:06 sorry, the target for arm is arm-linux Aug 28 14:46:12 paulk-desktop, hmmmm Aug 28 14:46:25 paulk-desktop, my gta04 doesn't boot, I must find the correct microsd Aug 28 15:01:55 SHR: 03Martin.Jansa 07meta-smartphone * r2bb94e4ad504 10/meta-openmoko/recipes-navigation/gpsd/ (gpsd-machine-conf_1.0.bb gpsd-machine-conf_1.0.bbappend): meta-openmoko: gpsd-machine-conf: change to bbappend Aug 28 15:02:46 Alex[sp3dev]: heyho Aug 28 15:02:53 morphis: yo Aug 28 15:03:05 Alex[sp3dev]: have some questions about libsamsung-ipc and the galaxy nexus Aug 28 15:03:11 sure Aug 28 15:03:18 ok Aug 28 15:03:50 I am issuing IPC_PWR_PHONE_STATE with IPC_TYPE_SET command message to set IPC_PWR_PHONE_STATE_LPM but it isn't accepted Aug 28 15:04:11 morphis: hm, really? Aug 28 15:04:15 as when I am issuing then a IPC_TYPE_GET for IPC_PWR_PHONE_STATE it's still in NORMAL level Aug 28 15:04:52 and IPC_TYPE_SET & IPC_PWR_PHONE_LPM doesn't gives me a response message, is that normal? Aug 28 15:04:56 morphis: in other words, you want me to fire IDA Pro ?) Aug 28 15:05:11 if thats the way how you can help :) Aug 28 15:05:29 I wanted just to ask you if you know something about this Aug 28 15:05:39 sigh..... Aug 28 15:05:39 no, have not noticed Aug 28 15:05:55 the systemd unit with that didn't work Aug 28 15:05:56 I guess for android, I'll just kill rild in flight mode :) Aug 28 15:06:16 SHR: 03Martin.Jansa 07meta-smartphone * rebee875b2177 10/meta-shr/conf/distro/shr.conf: meta-shr: no need for BAD_RECOMMENDATIONS now Aug 28 15:06:28 hm Aug 28 15:06:30 paulk-desktop: hi Aug 28 15:06:39 +[Install] Aug 28 15:06:39 +WantedBy=xserver-nodm.service Aug 28 15:06:45 hi Alex[sp3dev] Aug 28 15:06:49 I'll merge in a couple hours Aug 28 15:06:57 paulk-desktop: xD Aug 28 15:07:00 bbl for now Aug 28 15:19:43 hmmmm Aug 28 15:19:50 JaMa, if I enable it manually it works Aug 28 15:19:51 with: Aug 28 15:19:58 systemctl enable disablefbcon.service Aug 28 15:20:19 JaMa, should I push still? Aug 28 15:22:25 morphis: I've checked the disasm and looks like our ipc code is correct Aug 28 15:22:53 morphis: I still don't get it why you don't want to completely power down the radio in airplane mode Aug 28 15:23:58 GNUtoo: I don't have crespo to test it.. so it's up to you Aug 28 15:24:34 JaMa, ok Aug 28 15:24:42 how to enable the systemd unit Aug 28 15:24:46 I did something bad bts Aug 28 15:24:48 *btw Aug 28 15:24:53 I did that: Aug 28 15:25:34 check postinst script if it looks right Aug 28 15:25:51 +SYSTEMD_SERVICE_crespo = "rndissetup.service disablefbcon.service" Aug 28 15:25:53 ok Aug 28 15:26:00 with ar x? Aug 28 15:26:03 on the ipk Aug 28 15:27:14 /var/lib/opkg/info/foo/ is easier Aug 28 15:27:19 if you have crespo running Aug 28 15:28:11 http://pastie.org/private/ybsxtrkcqelfvwmicgolq Aug 28 15:29:43 and does it work if you execute it? Aug 28 15:30:17 I'll try Aug 28 15:30:37 I'm re-pushing an fso2-console-image.ext2 Aug 28 15:30:45 because else the phone reboots Aug 28 15:30:55 I'll push what I have now Aug 28 15:31:00 and fix right after Aug 28 15:31:12 anyway one fix at a time in one commit Aug 28 15:31:18 not multi-fix-catch-all commits Aug 28 15:31:36 *multi-fix-catch-all in one single commit are bad Aug 28 15:31:46 GNUtoo, if you power down the radio it reboots the device? Aug 28 15:31:55 no idea Aug 28 15:32:03 from sstate-diff.sh it looks like we have sane sstate checksums again Aug 28 15:32:04 under what Image? Aug 28 15:32:11 uff 3 hours to fix that Aug 28 15:32:16 ok sorry Aug 28 15:32:28 ka6sox-away, under what Image Aug 28 15:32:41 sorry. I misread.... Aug 28 15:32:48 was mixing up what was being said. Aug 28 15:32:51 ok Aug 28 15:33:02 GNUtoo: not your fault.. it's really easy to break them and tooling is not very helpfull when trying to fix them Aug 28 15:33:09 ok Aug 28 15:33:18 trying to understand why radio was not completely off during airplane mode. Aug 28 15:34:06 ka6sox-away: on crespo or gta04 or what? Aug 28 15:34:29 http://pastie.org/private/ek8b1lt3ow752lszrj3w Aug 28 15:34:30 hmmm Aug 28 15:34:32 strange Aug 28 15:34:39 I'll look if the link is already there Aug 28 15:34:45 Alex[sp3dev], I thought it was referring to GN Aug 28 15:36:25 ka6sox-away: it applies to all samsung devices. basically, someone (either paulk-desktop or someone else) implemented it so that FW is loaded only on init. and now everyone's lazy to do a proper reinit in airplane mode Aug 28 15:36:48 ka6sox-away: my point of view is that this sucks hard because I don't trust the proprietary fw Aug 28 15:38:52 blobs are always tough...but we can't fix that... Aug 28 15:39:07 JaMa, ls: /etc/systemd/system/xserver-nodm.service.wants/disablefbcon.service: No such file or directory Aug 28 15:39:10 well, shutting off the modem would be a good fix Aug 28 15:39:24 JaMa, I wonder if we need to make separate packages for each systemd unit Aug 28 15:39:27 I tried to avoid that Aug 28 15:39:28 I know that some devices I've used before do power control thru the modem firmware. Aug 28 15:39:32 and cheated using that: Aug 28 15:39:39 SYSTEMD_SERVICE_crespo = "rndissetup.service disablefbcon.service" Aug 28 15:39:51 ka6sox-away: qualcomm is a curse word here :) Aug 28 15:39:53 GNUtoo: check systemd.bbclass IIRC it should support multiple services per package Aug 28 15:40:01 GNUtoo: but maybe there is a but Aug 28 15:40:03 g Aug 28 15:40:29 yes I already checked and I think SYSTEMD_SERVICE_crespo = "rndissetup.service disablefbcon.service" can't work but I'm not sure Aug 28 15:40:38 Alex[sp3dev], I wasn't say'n any names... Aug 28 15:41:42 ka6sox-away: most other SoCs (nvidia, TI, samsung) don't have the modem cpu control everything and don't distrust the application cpu Aug 28 15:42:56 Alex[sp3dev]: somehow the ril from samsung does the power state change correctly (as the log shows) Aug 28 15:43:23 Alex[sp3dev]: I want to do it the right way rather than going a big way around which makes it hard to map everything in the right place Aug 28 15:43:33 JaMa, maybe I push what I have now Aug 28 15:43:39 and we look right after how to fix it Aug 28 15:43:40 morphis: well, you could trace the IPC on the proprietary ril then Aug 28 15:43:50 thats what I am doing right now Aug 28 15:44:21 GNUtoo: multiple services should work in rootfs time (see 9d72ff10c51963513f33ce8dbf42822abf42f6b7 in meta-oe) Aug 28 15:44:31 ok Aug 28 15:44:39 morphis: what's wrong with closing the ipc channel on airplane request and reiniting on exit? The only problem I see is that libsamsung-ipc is so fragile that any change can break everything :) Aug 28 15:44:49 GNUtoo: you can report it as issue in postinst generation in systemd.bbclass if that postinst cannot work like this Aug 28 15:44:59 Alex[sp3dev]: one reason, yes :) Aug 28 15:45:51 Alex[sp3dev]: what you mentioned is rather well from the security point of view Aug 28 15:46:08 but I am trying to get this done inside ofono and will get this upstream later Aug 28 15:46:24 and they will not accept such expriments if there is another way Aug 28 15:46:39 morphis: just don't tell them there's LPM IPC :) Aug 28 15:47:02 :) Aug 28 15:47:17 will mean less features Aug 28 15:47:21 ok, I'll read ofono code and say what I think should be done Aug 28 15:47:42 morphis: btw, does ofono work yet? Aug 28 15:47:44 SHR: 03GNUtoo 07meta-smartphone * rdbccd987bd98 10/meta-samsung/recipes-core/systemd/ (2 files in 2 dirs): meta-samsung: start fixing the disablefbcon systemd unit for crespo. Aug 28 15:47:48 Alex[sp3dev]: yes Aug 28 15:47:57 hmmm Aug 28 15:48:00 now I wonder what to do Aug 28 15:48:06 JaMa, what should we do now? Aug 28 15:48:16 with what? Aug 28 15:48:16 Alex[sp3dev]: just power up/down yet Aug 28 15:48:29 I'd like to workarround to finsh that bug Aug 28 15:48:42 because I've spent too much time already on it Aug 28 15:48:48 check that postinst Aug 28 15:48:59 as I said multiple services should work Aug 28 15:49:03 yes Aug 28 15:49:09 but I wonder what to check Aug 28 15:49:16 because when I run it manually it works Aug 28 15:49:21 but I need it at do_rootfs Aug 28 15:49:28 because else the image will never start Aug 28 15:49:38 ah wait a second Aug 28 15:49:41 I'll try shr-image Aug 28 15:49:49 I'm on fso2-console-image Aug 28 15:49:59 which lacks the xserver-nodm-init Aug 28 15:50:27 I have no idea where you got 17:39:27 < GNUtoo> JaMa, ls: /etc/systemd/system/xserver-nodm.service.wants/disablefbcon.service: No such file or directory Aug 28 15:50:39 GNUtoo: so check if that ipk file has both .service files Aug 28 15:50:44 ok Aug 28 15:50:51 and if postinst works then it should work from opkg upgrade too Aug 28 15:50:56 I got it before postinst Aug 28 15:51:00 I'll check shr-image Aug 28 15:51:14 and if it does then only difference is that 1st time postinst after flashing is a bit different Aug 28 15:51:15 like in shr-image.ext2 mounted on laptop Aug 28 15:51:30 yes "$D" is not there on target Aug 28 15:52:00 and some postinst were failing when executed before /tmp mount Aug 28 15:52:24 that was fixed in http://git.openembedded.org/meta-openembedded/commit/?id=402a0814947b310835c194b15d3b844e710dd394 Aug 28 15:52:56 so maybe your disablefbcon.service is executed too soon Aug 28 15:53:07 and fails for some reason Aug 28 15:53:32 ok Aug 28 15:53:38 I'll look at do_rootfs Aug 28 15:54:13 you're talking of which postinst? Aug 28 15:54:17 the one at do_rootfs? Aug 28 15:54:26 Alex[sp3dev]: maybe acceptance of LPM level belongs to RFS handling Aug 28 15:56:16 morphis: dunno Aug 28 15:56:36 it's the same postinst Aug 28 15:56:41 but using different systemctl Aug 28 15:56:55 http://pastie.org/private/ujaurwdlhn3uyxb3loxxqq Aug 28 15:57:17 just checked that disable/enable works for multiple services on target Aug 28 15:57:44 and? Aug 28 15:57:49 it works Aug 28 15:57:51 morphis: btw, I may be wrong, but from ril disasm, it looks that RFS IPC group is 0x15 (21), not 0x42 Aug 28 15:57:52 ah ok Aug 28 15:57:54 strange Aug 28 15:57:59 morphis: any good way to test rfs? Aug 28 15:58:04 but it's clearly not enabled at boot on crespo Aug 28 15:58:04 sorry I don't see anything interesting if you grep it like this Aug 28 15:58:07 samsung-modem-mgr Aug 28 15:58:21 Alex[sp3dev]: it just needs to be a little bit more verbose Aug 28 15:58:41 morphis: if it works with 0x42, I'd assume it is correct, but you may check with 0x15 Aug 28 15:59:13 will do Aug 28 16:00:58 Alex[sp3dev], actually there is no rfs group Aug 28 16:01:36 paulk-desktop: but there's rfs handling in samsung ril. just like other groups. what about that? Aug 28 16:01:49 yes that's a workaround I've set Aug 28 16:01:53 I have chosen 0x42 Aug 28 16:02:02 because basically, there is no IPC header for RFS Aug 28 16:02:12 the RFS command is set in the global header Aug 28 16:02:18 Operation failed: No such file or directory Aug 28 16:02:20 paulk-desktop: no, I mean, libsec-ril.so has RFS handling in the IPC loop Aug 28 16:02:23 in a chroot Aug 28 16:02:29 let me strace it Aug 28 16:02:38 Alex[sp3dev], so they have chosen 0x15 Aug 28 16:02:46 paulk-desktop: I see now Aug 28 16:02:54 everybody lies Aug 28 16:03:18 what I mean is that this doesn't reflect anything coming from the modem: there will never be messages with any RFS group Aug 28 16:03:46 but it makes requests handling easier to have one Aug 28 16:05:32 JaMa, ok Aug 28 16:05:44 JaMa, what should I do then, paste more? Aug 28 16:07:35 at least the section where the postinst is executed Aug 28 16:07:41 http://paste.debian.net/hidden/8f2d592a/ better ? Aug 28 16:08:46 yes WantedBy=basic.target found in rndissetup.service Aug 28 16:08:47 Enabled rndissetup.service for basic.target. Aug 28 16:08:55 is missing for disablefbcon.service Aug 28 16:09:08 ok Aug 28 16:09:09 no try to execute systemctl wrapper in rootfs dirr Aug 28 16:10:03 hmm it want's something with .target Aug 28 16:10:18 wanted_by=$(grep WantedBy $ROOT/$service_file \ Aug 28 16:10:18 | sed 's,WantedBy=,,g' \ Aug 28 16:10:18 | tr ',' '\n' \ Aug 28 16:10:18 | grep '\.target$') Aug 28 16:11:23 oh Aug 28 16:11:32 that was my fear at some point Aug 28 16:11:35 but not within oe Aug 28 16:11:38 but within the rootfs Aug 28 16:11:44 like on target running **** ENDING LOGGING AT Tue Aug 28 16:11:53 2012 **** BEGIN LOGGING AT Tue Aug 28 16:12:23 2012 Aug 28 16:12:25 if systemctl on target supports WantedBy=foo.service then systemctl wrapper is too strict and should be fixed Aug 28 16:13:42 GNUtoo: or try Alias which should be also supported by systemctl wrapper Aug 28 16:13:48 ok Aug 28 16:13:53 Alias=foo.service.wants/bar.service Aug 28 16:13:56 so Aug 28 16:14:10 Alias=xserver-nodm.service.wants/disablefbcon.service Aug 28 16:14:18 ah? Aug 28 16:14:22 I'll read about Alias Aug 28 16:15:11 ok Aug 28 16:15:19 I'll try alias right after trying that: Aug 28 16:15:43 from that manuall: WantedBy=foo.service in a service bar.service is mostly equivalent to Alias=foo.service.wants/bar.service in the same file. Aug 28 16:15:51 WantedBy=basic.target Aug 28 16:16:05 ok Aug 28 16:16:16 but systemctl wrapper implementation doesn't ensure that foo.service.wants directory exists (like WantedBy= does) Aug 28 16:16:50 so probably right fix would be to change that grep to allow not only .target but also .service (if you can confirm it works on target) Aug 28 16:17:59 I can test Aug 28 16:18:15 but I don't know how to grep 2 things at once Aug 28 16:18:23 so I'd prefer if you do it Aug 28 16:19:06 you can test it by echo 'WantedBy=foo.service' | grep '\.target$' :) Aug 28 16:19:08 GNUtoo: grep "\(foo\|bar\)" ? Aug 28 16:19:19 yes like Alex[sp3dev] says Aug 28 16:19:20 yes there was something like that Aug 28 16:19:31 I just didn't remember the syntax by heart Aug 28 16:20:43 thanks a lot!!!! Aug 28 16:20:44 echo 'WantedBy=foo.service' | grep '\.\(target\|service\)$' Aug 28 16:21:22 | grep '\(\.target$\)\|\(\.service$\)' Aug 28 16:21:26 is a bit better Aug 28 16:22:34 but yours should work too Aug 28 16:23:07 ok Aug 28 16:24:41 if yours is better I use yours Aug 28 16:30:22 hmmm Aug 28 16:30:30 only ./etc/systemd/system/basic.target.wants/disablefbcon.service is in the rootfs Aug 28 16:31:10 - | grep '\.target$') Aug 28 16:31:10 + | grep '\(\.target$\)\|\(\.service$\)') Aug 28 16:32:36 JaMa, where did I made a mistake? Aug 28 16:32:57 show that section of rootfs.log again please Aug 28 16:33:03 ok Aug 28 16:34:15 http://pastie.org/private/utjq6b6m1ohpxppe40qw Aug 28 16:34:33 WantedBy=basic.target found in disablefbcon.service Aug 28 16:34:36 did you change that? Aug 28 16:34:51 or do you have old .ipk with newer PV with this somewhere? Aug 28 16:35:53 no I kept it Aug 28 16:35:56 I'll remove that Aug 28 16:36:13 ? Aug 28 16:36:26 I thought you have changed it to WantedBy=xserver-nodm.service Aug 28 16:36:41 that's why you had to fix systemctl wrapper Aug 28 16:37:06 I had both and forgott to remove one of them(the basic.target) Aug 28 16:37:20 I had that: Aug 28 16:37:21 [Install] Aug 28 16:37:22 WantedBy=xserver-nodm.service Aug 28 16:37:22 +WantedBy=basic.target Aug 28 16:37:30 I git reset --hard Aug 28 16:37:37 and I'm retrying Aug 28 16:37:53 after the cleansstate of machine units Aug 28 16:39:28 then something is wrong Aug 28 16:39:35 you should get 2 links Aug 28 16:41:20 http://pastie.org/private/5xxtkn9uvlhbf1h48zalmw Aug 28 16:42:29 GNUtoo: did you rebuild systemd-systemctl-native ? Aug 28 16:42:37 no Aug 28 16:42:38 sorry Aug 28 16:43:33 I'll go right after fixing and pushing(wifi cafee is closing) Aug 28 16:46:20 it seem that I won't be able to push, so if you can push after testing on do_rootfs it would be so nice Aug 28 16:46:34 JaMa, ^^^ Aug 28 16:48:54 bye Aug 28 16:55:59 I guess it can wait for tomorrow Aug 28 19:32:06 Alex[sp3dev], ping Aug 28 19:32:18 paulk-desktop: pong Aug 28 19:32:28 Alex[sp3dev], I'm integrating your commits now Aug 28 19:32:36 paulk-desktop: unbelievable Aug 28 19:32:36 though I'd like to get rid of compat.h Aug 28 19:32:39 :D Aug 28 19:32:51 paulk-desktop: that's... blasphemy Aug 28 19:33:41 I would just prefer to make in-code ajustements Aug 28 19:34:04 anyways that LOGD/ALOGD thing is really a pain Aug 28 19:34:13 why in hell did they need to change that? Aug 28 19:34:24 paulk-desktop: because they are morons and fuckers Aug 28 19:34:32 good point Aug 28 19:34:47 paulk-desktop: my rule is to try and concentrate all ifdeffery in one header (the same goes for debugging macros). because when someone edits code in an ifdef and doesn't compile the other ifdef case, bad things happen Aug 28 19:40:07 well I'll probably keep the header then Aug 28 19:40:46 I'm not in the mood of spending hours of ifdefing everything in the code, perhaps introducing errors, to do nothing more that what the header already does Aug 28 19:43:36 or maybe I will, but another day Aug 28 19:47:36 Alex[sp3dev], also, your commits often introduce new lines with only tabs Aug 28 19:47:53 paulk-desktop: I was lazy to rebase Aug 28 19:48:09 paulk-desktop: but generally I hate spaces and want to kill them with fire Aug 28 19:52:12 ok I'll push your commits Aug 28 19:52:51 thanks a lot for your great work, Alex[sp3dev] ! Aug 28 19:53:34 paulk-desktop: your work is greater. If it was not for libsamsung-ipc, I guess I wouldn't even start hacking samsung :) Aug 28 20:06:02 paulk-desktop: does ABSTRACT socket namespace not work on GB? why not keep it for both GB and ICS? Aug 28 20:06:26 Mhh I didn't really check but I thought it was new from ICS Aug 28 20:06:42 I'll test tomorrow and maybe revert my commit if it works Aug 28 20:06:49 paulk-desktop: GNUToo told me he edited samsung-ril-client and it worked so you may want to test Aug 28 20:07:02 ok Aug 28 20:53:22 paulk-desktop: so, now it's time to blackmail cyanogenmod developers into using our ril? Aug 28 20:53:35 lol Aug 28 20:54:07 one day maybe, but for the moment, we need full implementation of the protocol Aug 28 20:54:13 paulk-desktop: https://gitorious.org/~ast/replicant/asts-hardware_ril_samsung-ril-client you may want to push my "tuna_final" branch to a separate branch on replicant. so that we have everything upstream Aug 28 20:54:35 paulk-desktop: btw, have you seen the progress over there? https://github.com/grindars/android_hardware_samsung_freeril Aug 28 20:54:38 ah right, there is the client too Aug 28 20:55:05 no, I didn't see, but what he's doing is crazy Aug 28 20:55:29 crazy in good sense Aug 28 20:55:41 sure Aug 28 20:55:56 like he's done a crazy amount of work rewriting it all Aug 28 20:56:09 maybe the result will be better Aug 28 20:56:13 but anyways I don't like CPP Aug 28 20:56:58 paulk-desktop: C code in userland looks scary Aug 28 20:57:45 at least in C++ you have stl with vectors, priority queues and hash tables. I still don't like c++ but I'll write a blog post once with my thoughts on programming languages Aug 28 20:58:24 okay Aug 28 20:58:29 I'll gladly read it then Aug 28 20:58:38 As I'm not a very experienced dev Aug 28 20:59:00 I would benefit learning the real differences between languages Aug 28 20:59:10 but I still wouldn't like OOP Aug 28 21:01:14 paulk-desktop: I think that practice makes perfect. I would advise to try ruby (or python), ocaml and haskell, common lisp, erlang (to see *different* languages). Also, I advise learning CGI and perl (to see how web works) and opengl (to improve math skills). Aug 28 21:01:50 generally, you should 1) try reading a bit about everything and 2) enjoy your projects Aug 28 21:02:25 yeah great advices. I should also try to learn with books Aug 28 21:03:43 Alex[sp3dev]: wouldn't it be faster to explain what you do not recommend :) ? Aug 28 21:03:47 sometimes I don't have patience to read books and a small tutorial is good to bring fun and get interested with the subject. well, the general idea is that you should always try to answer "why", no "how". Aug 28 21:03:57 misc: I do not recommend java and php Aug 28 21:04:46 well, i would not recommend php, but on the other hand, this is a easy stuff for doing fast web page Aug 28 21:05:06 php is probably the first language I started with Aug 28 21:05:15 back at the time when I was doing web pages Aug 28 21:05:25 it made C learning easier Aug 28 21:06:19 misc: I think that virtually all areas of science are interesting, and even in CS it is impossible to master everything, but stuff like machine learning, DSP, CV are amazing. Math is also amazing. Maybe not quite amazing at high school, but amazing from later on Aug 28 22:01:19 is there a way to toggle arbitrary gpios from userspace on gta04? Mine is out of the case and I wanted to play a little with irda Aug 28 22:14:50 nm, I found it: https://groups.google.com/forum/?fromgroups=#!topic/pandaboard/rVO41Ig2lhc **** ENDING LOGGING AT Wed Aug 29 02:59:58 2012