**** BEGIN LOGGING AT Sat Jun 05 02:59:57 2010 Jun 05 05:24:39 [Rui]: found some info about your alsa-utils issue.. http://thread.gmane.org/gmane.comp.handhelds.openembedded/33269 Jun 05 06:23:51 <[Rui]> JaMa: cool, thanks Jun 05 06:25:37 [Rui]: seen also my reply? Jun 05 06:27:21 [Rui]: the opkg-native problem is similar.. for some reason it's using libldap.so from your host, please pastebing config.log from opkg-native-version/trunk dir Jun 05 06:32:23 <[Rui]> JaMa: will have to do that later, in less than half an hour I'll be leaving for a small conference to talk about Free Software in education. :) Jun 05 06:32:55 <[Rui]> only came here to turn the computer off, really :) Jun 05 06:32:57 <[Rui]> see ya! Jun 05 10:19:55 hi. is it possible to compile c code on the openmoko? i installed gcc via opkg but there is no $gcc.. Jun 05 10:23:08 $CC ... and if it is not there than set it Jun 05 10:23:22 yes very easy to compile c code Jun 05 10:26:15 how can i set it? Jun 05 10:36:12 richy: CC=gcc; export CC Jun 05 10:36:45 simply do: gcc -o abc abc.c Jun 05 10:37:34 hope u have binutils too. Jun 05 10:38:07 richy: at least on debian you can build software just like on your PC Jun 05 10:43:27 lindi- How do i get to manipulate post and pre scripts of packages .... want to change ownership and permission of packages .... if this is correct in the first place Jun 05 10:44:01 THis should affect the population of rootfs ... right? Jun 05 10:44:34 using bitbake and OE Jun 05 10:46:19 shazkhan: debian? Jun 05 10:46:29 shazkhan: bitbake and OE? then I have no idea Jun 05 10:50:17 lindi- How do u do it in debian? The package management is not different. Jun 05 10:50:57 By the way I was trying to get debian for qemuarm and it gave me a very tough time and was not able to get it in working condition. Jun 05 10:51:26 shazkhan: debian is very different from OE Jun 05 10:51:44 shazkhan: bug reports? Jun 05 10:53:29 :) .... aureal web page was followed and the bug was that I was not being able to get filesystem formated :) and had problems with repositories as well .... If I remember correctly it took a lot of time in one task to complete so it took my complete day leaving me in frustration Jun 05 10:54:32 lindi- If debian can get me with non-root user working condition and dbus properly than I would try it again. Jun 05 10:54:36 shazkhan: yep but please report those Jun 05 10:55:08 You want me to try http://lindi.iki.fi/lindi/openmoko/debian-armel-cross.txt Jun 05 10:55:24 Can I use it over my ubuntu host/build Jun 05 10:56:36 As you already know I am interested in selinux inclusion with each package its policy etc Jun 05 10:57:06 If it gets going in one distros then others can be done easily Jun 05 10:57:20 shazkhan: i can't guarantee anything about ubuntu sorry :) Jun 05 10:57:58 shazkhan: why are you doing the selinux stuff on openmoko? Jun 05 10:59:11 shazkhan: gcc/binutils installed but it doesnt works fine.. Jun 05 10:59:15 whereis gcc Jun 05 10:59:26 gcc: /usr/lib/gcc /usr/libexec/gcc Jun 05 10:59:39 hi mickeyl Jun 05 10:59:53 I'm pushing the unpushed changes in the kenrel Jun 05 11:00:06 s/I'm/I pushed Jun 05 11:00:25 *I pushed some unpushed changes that are minor things Jun 05 11:00:31 such as defconfig-> #define Jun 05 11:01:01 so my question is,are you ok with making the refresh thread not killed? Jun 05 11:01:06 now that suspend works Jun 05 11:01:10 maybe it's useless Jun 05 11:01:15 works without FSO Jun 05 11:01:28 because I need to find time to work on the suspend reason Jun 05 11:02:42 lindi- selinux and other hardcore security features is what my objective is. Have you seen security features of Maemo/MeeGo. I am sort of doing the same Jun 05 11:03:00 shazkhan: yes but why not work on those on PC? Jun 05 11:03:14 shazkhan: and only move it to your pocket when it works Jun 05 11:03:17 I have already done it on PC Jun 05 11:03:47 not the policy with each package part though ... Jun 05 11:04:33 Why should I put time on PC while the problem is same on pocket? Jun 05 11:04:48 and my objective is pocket Jun 05 11:05:41 shazkhan: well you complained about things being slow under qemu armel Jun 05 11:06:32 I have a GTA02 and HTC Dream at office but qemuarm because it is always available to me :) Jun 05 11:07:06 just have to change MACHINE in auto.conf and thats it Jun 05 11:08:26 lindi- http://shazkhan.wordpress.com/2010/05/27/selinux-and-integrity-measurement-architecture-on-openmoko/ Jun 05 11:09:16 shazkhan: i think you should mention the distro you use in that Jun 05 11:09:33 SHR Jun 05 11:09:40 GNUtoo|laptop: you think the refresh thread is the problem? Jun 05 11:09:58 mickeyl, not at all, it fixes another issue Jun 05 11:10:14 mickeyl, basically sometime at resume you have garbage on the screen Jun 05 11:10:41 keeping the refresh thread on fixes that Jun 05 11:10:53 but I didn't test it extensively Jun 05 11:11:45 i'd rather check that the refresh thread is being reactivated before the display is being reactivated Jun 05 11:12:15 eventually I'd like to get rid of it altogether Jun 05 11:12:20 it's killing performance Jun 05 11:13:13 ah ok Jun 05 11:13:55 so I'll do that later Jun 05 11:14:02 priority is suspend Jun 05 11:14:05 and resume Jun 05 11:14:12 true Jun 05 11:14:18 more precisely get the resume reason Jun 05 11:14:41 btw., with the little delay added, the android lowlevel seems to work ok Jun 05 11:14:59 it's still an inherent race condition, but should work until we have the sysfs entry Jun 05 11:15:18 mickeyl: How do i get to manipulate post and pre scripts of packages .... want to change ownership and permission of packages ... so that the rootfs is populated correctly for the non root user Jun 05 11:16:33 shazkhan: you write pkg_<...>_${PN} scripts Jun 05 11:16:38 take a look at any NTP recipe Jun 05 11:16:48 NTP? Jun 05 11:17:06 net time protocol Jun 05 11:17:14 k Jun 05 11:17:19 mickeyl, ok I'll give a new try Jun 05 11:17:20 like 4.1.2 e.g. Jun 05 11:17:25 there you see Jun 05 11:17:30 altough I need instruction for building SHR Jun 05 11:17:30 that is to say Jun 05 11:17:33 we use autorev Jun 05 11:17:35 pkg_postrm_ntp(){ Jun 05 11:17:35 update-rc.d -f ntp remove Jun 05 11:17:35 } Jun 05 11:17:39 and I use bitbake 1.10 Jun 05 11:17:48 and I don't know well the remote count Jun 05 11:17:52 and SRCPV Jun 05 11:17:55 and similar things Jun 05 11:18:09 mickeyl, I've a new laptop(the old one broke) Jun 05 11:18:11 and it's so fast Jun 05 11:19:19 mickeyl: still using the refresh thread in X then? Jun 05 11:19:34 tmzt: GNUtoo|laptop, yes Jun 05 11:19:36 mickeyl: Thanx ... please give some more example recipes Jun 05 11:19:40 on G1 Jun 05 11:19:49 I'll eat Jun 05 11:20:04 i'm using it on w/ fb, didn't have time to move it to the ioctl yet Jun 05 11:20:11 shazkhan: just grep for it Jun 05 11:20:28 k :) Jun 05 11:20:29 shazkhan: the OE tree is full of it Jun 05 11:20:33 I've looked at porting the 2d parts of the codeaurora driver, the gpu is completely different but the memory mangement should work almost the same Jun 05 11:21:36 http://www.visionmobile.com/blog/2010/06/lead-innovate-or-assemble-how-software-is-changing-the-macro-economics-of-mobile-handsets-2/?utm_source=feedburner&utm_medium=feed&utm_campaign=Feed%3A+Visionmobile+%28VisionMobile+Forum%29&utm_content=Google+Reader Jun 05 11:22:09 ah, a new report from Andreas Jun 05 11:23:22 the most important sentence is that: Jun 05 11:23:23 Only by controlling and integrating hardware, software and services under the same roof can a manufacturer deliver new product experiences that can command top-tier retail prices. Jun 05 11:23:59 unfortunately it's also somewhat FOSS undfriendly Jun 05 11:24:02 I thoguht this was interesting Jun 05 11:24:04 Network operators/carriers saw Android as an opportunity to reduce their dependency on two players, Apple and RIM whose stellar sales were depriving operators from any negotiating power. Operators have always tried to divide and conquer amongst their suppliers, for example working in 2002 with HTC and Windows Mobile to reduce their dependency on Nokia, or in 2007 using a three-pronged OS strategy (WinMo, Symbian, Linux Jun 05 11:24:06 but tight integration has always been FOSS weakness Jun 05 11:25:11 the less of FR was in part that it needed a full software stack to sell right? and, as stated on the gta03-list, was targeted as a mass market device rather than a 2nd gen developer device Jun 05 11:25:19 (lesson) Jun 05 11:26:10 .. have launched or preparing to introduce out-of-the-box Android designs that reduce the time to market for Android handsets to 6-9 months (or circa 3 months once Mediatek’s design hits the market). Jun 05 11:26:46 this is why I like qualcomm, their firmware is closed but they have the best integrated chipset and readily available reference designs (for the ODM of course) Jun 05 11:26:58 (FR) yes. given the device weaknesses, it was an incredible mistake to even try to compete with the iPhone UI experience Jun 05 11:27:51 having a great of the box UX is crucial, having openness to replace that fully if you prefer is also desirable Jun 05 11:28:00 have you seen linaro yet? Jun 05 11:28:22 heh, no Jun 05 11:28:27 yet another consortium, fun Jun 05 11:28:28 a big part of that is to get a single linux kernel tree with support for all the major SoCs Jun 05 11:28:48 very unlike LiMO and even OHA (which exists on paper from what I can tell) Jun 05 11:28:56 could be interesting next year, only for new devices though Jun 05 11:29:32 i'm somewhat tired these days. thinking of letting all that rest for 12 months and then coming back Jun 05 11:29:35 they are using devicetree to standarize kernel development, and have comments from linus rejecting the idea of each SoC vendor submitting a huge new NIH tree ot the kernel Jun 05 11:30:03 ya, good plan Jun 05 11:30:25 yeah, personally I'm waiting on an awesome qwerty snapdragon device Jun 05 11:30:35 probably shipping with android so we have working kernel source Jun 05 11:30:42 *nod* Jun 05 11:30:46 and then building the open, living system I want Jun 05 11:30:52 no more firmware flashing, I hate that :) Jun 05 11:30:55 hehe Jun 05 11:30:59 yea Jun 05 11:31:45 best things, snapdragon support is almost upstream and there's a real open X driver for accelerated 2d Jun 05 11:32:13 and it seems z430 is more similar to radeon then I ever would have thought, so it's possible we can get an open 3d driver for it Jun 05 11:32:26 nice Jun 05 11:32:27 which solves pretty much all the issues (still the optimist :)) Jun 05 11:32:34 well Jun 05 11:32:37 it even has a command processor Jun 05 11:32:40 we still don't have a compelling UI at all Jun 05 11:32:50 all the solutions are half-baken at best Jun 05 11:32:52 Qt/QtML? Jun 05 11:33:12 I might prefer clutter/mutter/gnome3/gjs personnally, but I could live with Qt if I have to Jun 05 11:33:36 yes, QML might be a good contender, but it will take a couple of years until all the building blocks have been done Jun 05 11:34:18 too bad WebOS is closed Jun 05 11:34:40 i would have enjoyed putting that on top of FSO Jun 05 11:34:50 mickeyl, about palm pre...I think they sell some in france Jun 05 11:35:19 also I saw the status Jun 05 11:35:25 it has progressed a lot Jun 05 11:35:56 btw is alain2210 in irc? Jun 05 11:36:01 sometimes Jun 05 11:36:02 GNUtoo|laptop: do you know if the snapdragon video driver can be ported to 7x0x? Jun 05 11:36:08 only in #htc-linux i guess Jun 05 11:36:47 tmzt, it's msm_fb.c ? Jun 05 11:37:08 I mean the X driver and the use of drm/dri2 over pmem Jun 05 11:37:16 ah no it can't Jun 05 11:37:22 because it depends on dri Jun 05 11:37:32 and we don't have that for dream Jun 05 11:37:41 right, but the dri ioctl interface is very similar to pmem Jun 05 11:37:45 which we already have Jun 05 11:37:47 The msm dri driver only support higher phones Jun 05 11:38:03 ah cool Jun 05 11:38:17 maybe add a request to the status then Jun 05 11:38:20 in the TODO Jun 05 11:38:27 is G1 in a state were thinking about this is worth it? Jun 05 11:38:42 TODO means to do later Jun 05 11:38:44 maybe should focus on the Slide or something like that with 7x27 Jun 05 11:38:48 I'd say GPS, BT is more important Jun 05 11:38:55 yeah, where's this TODO? Jun 05 11:39:03 http://htc-linux.org/wiki/index.php?title=Dream#Kernel_Status Jun 05 11:39:11 let alone accelerometers and compass Jun 05 11:39:15 so we don't forget later Jun 05 11:39:20 yes Jun 05 11:39:25 but suspend is top priority Jun 05 11:39:27 gps still crashes but I thought the userspace was working Jun 05 11:39:47 tmzt, talk to phh for GPS Jun 05 11:39:52 when i build the code phh posted, it crashes immediately Jun 05 11:39:59 s/build/build and run/ Jun 05 11:39:59 mickeyl meant: when i build and run the code phh posted, it crashes immediately Jun 05 11:40:29 could be a miscompilation of librpc though, it's been converted to GNU Jun 05 11:40:49 mickeyl, I've an android build system Jun 05 11:41:02 on the hdd of my old laptop which is broken but I've chroot Jun 05 11:41:17 ls Jun 05 11:41:18 oops Jun 05 11:41:36 ya, might be worth trying out whether the same code built under android works Jun 05 11:41:43 if so, we need to find out the differneces Jun 05 11:41:43 ok Jun 05 11:41:52 pastebin me the code then Jun 05 11:42:01 http://git.freesmartphone.org/?p=utilities.git;a=tree;f=android/rpc;h=c409c68931acd9c4708458b1982b5b970f622a0a;hb=a6839f3d22675c2b59f6f05172429f635439ff5e Jun 05 11:42:06 it's your tarball Jun 05 11:42:07 :) Jun 05 11:42:12 (converted to GNU) Jun 05 11:42:15 I mean the code phh pasted Jun 05 11:42:22 not converted to GNU Jun 05 11:42:37 that's the tarball i got from you Jun 05 11:42:50 so it works sometimes and crashes other times Jun 05 11:42:51 ok Jun 05 11:42:57 mickeyl, didn't you talk to phh? Jun 05 11:43:06 i did Jun 05 11:43:08 wasn't helpful though Jun 05 11:43:12 ah ok Jun 05 11:43:15 he said the wince RPC should work as well Jun 05 11:43:39 so roughly he said: Jun 05 11:43:45 and since i don't understand a single line of this fricking code, i have no idea where it can go wrong Jun 05 11:43:48 SHOULDWORK WONTFIX? Jun 05 11:44:07 well, apparantly he doesn't bother much with GNU Jun 05 11:44:09 I think he doesn't understand it fully either Jun 05 11:44:20 so he could care less until it works in Android Jun 05 11:44:22 GNU or not GNU the android code can crash Jun 05 11:44:25 s/until/as long/ Jun 05 11:44:25 mickeyl meant: so he could care less as long it works in Android Jun 05 11:44:48 and your code doesn't crash every time Jun 05 11:45:10 well Jun 05 11:45:12 "every time" Jun 05 11:45:14 i tried twice Jun 05 11:45:17 and then i lost interest Jun 05 11:45:26 it sucks having take out the battery and do a full reboot Jun 05 11:45:33 so i gave up after two attempts Jun 05 11:46:34 the code is great though Jun 05 11:46:44 it's a prime example for all those who come to me saying android is open source Jun 05 11:47:03 it crashes the radio Jun 05 11:48:01 is it the same on nexus? Jun 05 11:48:23 or does it have a sane peripheral control abstraction? Jun 05 11:48:26 I don't know Jun 05 11:48:30 he said it works on g1? Jun 05 11:48:32 but also on wince? Jun 05 11:48:36 ok Jun 05 11:48:39 I'll retry Jun 05 11:48:39 works on g1 on android Jun 05 11:48:58 mickeyl: the rpc goes directly to the radio through the rpc router Jun 05 11:49:07 when radio crashes you have to reboot the phone Jun 05 11:49:18 yes Jun 05 11:49:35 and it's of incredible inconvenience that there is no sysfs abstraction Jun 05 11:49:39 hell i would have even applaused an ioctl Jun 05 11:49:57 but no, stupid engineers need to expose this via RPC Jun 05 11:50:01 :/ Jun 05 11:50:07 you want the rpc client in kernel space? Jun 05 11:50:12 no Jun 05 11:50:16 i want peripheral control Jun 05 11:50:27 abstracted over simple interfaces Jun 05 11:50:28 rpc is the protocol used to take to the radio, it's the native protocol Jun 05 11:50:33 i know Jun 05 11:50:37 everything is going rpc, proc_comm is too simple Jun 05 11:50:42 for more than really basic stuff Jun 05 11:50:52 nmea from the radio would be better though Jun 05 11:51:38 phh: we're trying to understand briefly what's going on with gps on g1 Jun 05 11:52:12 all controls are done through rpc methods between A11 and A9 (can be implemented in kernel or userland) Jun 05 11:52:36 the modem then reports back NMEA on /dev/smd27, and also the same infos on rpc callbacks Jun 05 11:52:55 (enough ?) Jun 05 11:53:19 oh, what's working as far as the rpc/librpc code Jun 05 11:53:31 is it crashing on android on g1 or just on oe Jun 05 11:54:30 ah that, i'm not so sure, the code I did for G1 was before I understand stuff a bit better, so there are weird stuff in it Jun 05 11:55:49 so is this supposed to work or rather not? http://git.freesmartphone.org/?p=utilities.git;a=blob;f=android/rpc/gps/main.c;h=d8d4f14790ac37dcc080f72f20edd07a8dde30c4;hb=a6839f3d22675c2b59f6f05172429f635439ff5e Jun 05 11:55:53 and I know that the code that worked, shouldn't have worked. Jun 05 11:56:07 mickeyl, http://pastebin.com/Ld2kxr5h Jun 05 11:56:23 isn't position working xdandroid now though? Jun 05 11:56:29 tmzt: it is Jun 05 11:56:29 it just doesn't let you shut it off or something Jun 05 11:56:36 altough it sometime crashes Jun 05 11:56:40 even in android Jun 05 11:56:53 oh, but that's on wince devices Jun 05 11:56:53 GNUtoo|laptop: is this the code from the FSO repo? Jun 05 11:56:57 yes Jun 05 11:56:59 mickeyl: the code that worked is the "android libgps" stuff Jun 05 11:57:10 GNUtoo|laptop: ok, so it somewhat works for you Jun 05 11:57:15 what is xdr? Jun 05 11:57:23 this is sunrpc or dcerpc or someting? Jun 05 11:57:24 mickeyl, maybe different radio firmware? Jun 05 11:57:29 tmzt: sunrpc Jun 05 11:57:34 GNUtoo|laptop: yes, i have 1.x Jun 05 11:57:40 you have 2.x Jun 05 11:57:43 mickeyl, ok Jun 05 11:57:53 mickeyl: the program segfault for you ? Jun 05 11:58:02 phh: no, hard lock Jun 05 11:58:03 leading to instant A9 crash Jun 05 11:58:07 hum Jun 05 11:58:25 mickeyl: any logs from the program ? Jun 05 11:58:43 not now Jun 05 11:58:45 would need to recreate Jun 05 11:59:14 phh: would a basic gps only client work better? Jun 05 11:59:16 mickeyl: anyway I had such a problem when I tried to build my test program as static to run it on ubuntu Jun 05 11:59:21 than using this full sun stack Jun 05 11:59:44 there is no reason for it to work better Jun 05 11:59:55 sun stack does nothing clever Jun 05 11:59:58 phh: so you mean it probably won't work in a GNU environment? Jun 05 12:00:06 maybe something written in vala that just talks to kernel's rpc device Jun 05 12:00:20 mickeyl: I mean that I've got such problems, that's all. Jun 05 12:00:21 mickeyl, for me it works Jun 05 12:00:24 tmzt: oO Jun 05 12:00:29 s/works/somehow works Jun 05 12:00:32 tmzt: I'll let you have fun. Jun 05 12:00:49 hey, I don't even have this hardware Jun 05 12:01:06 but it's at the top of the TODO for the g1 fso stuff apparently Jun 05 12:01:10 well you have TP2, it's the same thing, but the values Jun 05 12:01:19 the remoko version shr ships is completly broken and cannot work - there are newer revisions in svn which could work Jun 05 12:01:34 what's with all this threading? Jun 05 12:01:42 i would check out the code on the neo, but there seems to be no svn package Jun 05 12:01:47 bumbl, you mean they ported remoko to bluez4??? Jun 05 12:01:49 bumbl: afaik remoko has never been updated to bluez4 Jun 05 12:01:53 anyways, need to run Jun 05 12:01:56 tmzt: receiving messages is asynchronous Jun 05 12:02:42 this was a port from bionic to glibc? Jun 05 12:02:54 so different pthread implementation right? Jun 05 12:03:52 no Jun 05 12:03:56 I was using android's libs Jun 05 12:03:59 just statically linked Jun 05 12:04:05 okay, practically how many things need user rpc on g1 simultaneously? Jun 05 12:04:11 just audio and gps? Jun 05 12:04:18 clock too? Jun 05 12:04:21 user=userland ? Jun 05 12:04:26 in userspace? Jun 05 12:04:27 yeah Jun 05 12:04:29 clock is kernel Jun 05 12:04:33 ah ok Jun 05 12:04:34 in userspace, only gps Jun 05 12:04:38 audio is in kernel Jun 05 12:04:40 so why the CLIENT list? Jun 05 12:05:00 ah sorry I didn't saw the userspace thing Jun 05 12:05:04 gps needs two endpoints Jun 05 12:05:05 I thought plain rpc Jun 05 12:05:22 sure, but that's just muxing Jun 05 12:05:28 couldn't you use a switch? Jun 05 12:05:30 clock is obviouly kenrel but is it over rpc? Jun 05 12:05:47 GNUtoo|laptop: it's proc comm Jun 05 12:05:53 ok Jun 05 12:05:56 mickey|away: according to the svn it says so: "finished initial bluez4 port - bogus atm" and then ~10 commits with concerning refactoring Jun 05 12:06:02 tmzt: a switch ... ? Jun 05 12:06:07 bumbl, wow nice Jun 05 12:06:18 like a switch statement for the endpoint address Jun 05 12:06:32 maybe I'll make bluetooth work on htcdream then,but after fso suspend Jun 05 12:06:33 unless you expect to have to handle (queue?) multiple callbacks Jun 05 12:06:38 GNUtoo|laptop: the question is whether it works Jun 05 12:06:47 bumbl, try it then Jun 05 12:06:56 GNUtoo|laptop: no svn package in shr Jun 05 12:07:08 bumbl, cross-compile it then: Jun 05 12:07:29 build an image with openembedded and use the devshell to cross-compile it Jun 05 12:07:46 http://docs.sun.com/app/docs/doc/816-1435/xdrnts-21693?a=view Jun 05 12:07:50 or mail the author Jun 05 12:08:06 or ask shr people for a recipe on their mailing list Jun 05 12:08:12 hi Jun 05 12:08:13 s/recipe/package Jun 05 12:08:16 hi Kensan Jun 05 12:08:23 GNUtoo|laptop: i am sitting on an eee at the moment - so no cross compiling ;) Jun 05 12:08:31 ok Jun 05 12:08:34 eee701? Jun 05 12:08:43 tmzt: are you porting msm_fb from codeaurora's chromium to some other tree? Jun 05 12:09:51 Kensan: is it different? Jun 05 12:10:42 tmzt: when I looked through the history it seems there's many fixes and improvements compared to the one in android's htc-msm-2.6.32 tree Jun 05 12:11:10 tmzt: plus some(most?) of the scheduling issues seem to have been flushed out. Jun 05 12:11:27 that's good Jun 05 12:12:06 I'm not porting that though, I'm looking at the memory management functions of kgsl to see if the can work with the more basic gpu/mdp Jun 05 12:12:14 so we can get rid of the refresh thread Jun 05 12:12:38 there's some other stuff I'm doing with X as well, trying to get more into that codebase Jun 05 12:12:40 too bad we don't have kgsl Jun 05 12:12:45 on the dream Jun 05 12:12:49 GNUtoo|laptop: bless god no - a 1005pe - quite decent for most things but compiling takes ages + i should get a bigger hdd, those 250GB are nothing today; i will install fso ms4 and an old remoko version and be happy Jun 05 12:12:55 we have pmem which is similar Jun 05 12:13:12 bumbl, ok Jun 05 12:13:29 tmzt, the problem is that we removed pmem from the kenrel Jun 05 12:13:53 yeah Jun 05 12:13:58 what kernel are you working from? Jun 05 12:14:00 for oe Jun 05 12:14:21 leviathan-incomming Jun 05 12:14:31 htc-msm-2-6-32/leviathan-incoming.git Jun 05 12:14:35 msm gnu conform branch Jun 05 12:14:57 I see pmem-free Jun 05 12:15:30 tmzt: which tree are you working on? Jun 05 12:16:27 yes it's somewhat free of pmem Jun 05 12:16:36 can blanking be done from sysfs? Jun 05 12:16:48 tmzt: the codeaurora chromium repos also has a xf86-video-msm tree Jun 05 12:17:42 what's mach/pmem.h Jun 05 12:17:45 yeah, I know Jun 05 12:17:55 freesmartphone.org: 03tom 07framework * r8f618948c4c7 10/framework/subsystems/opimd/pimd_messages.py: Added following name changes of ogsmd (i.e ogsmd restarts). #557 Thanks to Paul Fertser. Jun 05 12:17:59 chaning the msmfb to report as mdp2.2 won't exaclt work though Jun 05 12:18:00 PaulFertser, too me a long time, sorry :P Jun 05 12:18:02 the hardware is different Jun 05 12:18:14 PaulFertser, I'm mostly not here lately :( Jun 05 12:18:36 what's the st1.5 device that is mentioned in most chromium commits? Jun 05 12:21:37 no idea Jun 05 12:21:46 except it seems to be made by quanta Jun 05 12:29:10 how different is this to pmem? Jun 05 12:29:12 https://www.codeaurora.org/gitweb/quic/chrome/?p=kernel.git;a=blob;f=drivers/char/msm_kgsl/kgsl_sharedmem.c;h=be720ca72e3811315aa30c24075d27c2c5503081;hb=refs/tags/Q8650DTBCANLXA1070 Jun 05 12:30:09 as far as i can tell, it's the same thing Jun 05 12:30:37 the mdp on msm7x0x doesn't have multiple level page tables, so it has to be changed Jun 05 12:30:59 but porting the implementation from pmem.c and changing to shmem from filp should work shoudln't it? Jun 05 12:31:50 we should see what Palm did, the mmu parts of 7x27 are the same as 7x0x I think Jun 05 12:32:00 I don't think Jun 05 12:32:05 at least I remember seeing that mentioned in a few commits Jun 05 12:32:06 no? Jun 05 12:32:09 I think 7x27 is more like 8x50 than 7x00 Jun 05 12:32:21 the gpu is, but not mmu, I'll see if I can find it Jun 05 12:32:41 tmzt: kgsl sharedmem has a "no mmu" option somewhere Jun 05 12:32:44 well, it's mmu the option Jun 05 12:32:49 I don't know how releavant is that Jun 05 12:33:40 http://android.git.kernel.org/?p=kernel/msm.git;a=log;h=refs/heads/android-msm-2.6.32 Jun 05 12:33:44 sweet, stacktrace Jun 05 12:41:26 phh: https://www.codeaurora.org/gitweb/quic/la/?p=kernel/msm.git;a=commitdiff;h=2b27b56d4f97fe697dc9348b5184f51bd914f923 Jun 05 12:42:50 phh: also, rpc, https://www.codeaurora.org/gitweb/quic/la/?p=kernel/msm.git;a=commitdiff;h=de57c7022a80e88895cbcfba8dbbfb1cf20b24b0 Jun 05 12:44:53 "trivial merge" Jun 05 12:44:55 thanks. Jun 05 12:47:20 ha JaMa is not there Jun 05 13:25:15 hm, Qi has a really poor support of u-boot images :/ Jun 05 13:25:57 ThibG: yes. make u-boot not write to glamo and we have no reason to use it :) Jun 05 13:26:09 ThibG: also, it would be nice if u-boot was part of debian Jun 05 13:27:12 aptitude show u-boot \o/ Jun 05 13:27:26 woot! Jun 05 13:28:34 that's very recent Jun 05 13:28:39 ThibG: no openmoko support yet though Jun 05 13:28:51 yeah :( Jun 05 13:28:58 and it doesn't build from source: 'make[1]: Leaving directory `/home/lindi/debian/debian-u-boot/u-boot-2010.06~rc1' No targets defined for your architecture. exit 1 make: *** [build] Error 1 dpkg-buildpackage: error: debian/rules build gave error exit status 2' Jun 05 13:29:02 i'll report a bug Jun 05 13:30:30 ThibG: ah, it's just not available for amd64 Jun 05 13:30:50 hm ok Jun 05 13:31:05 I just know it's in Debian, I haven't had a look yet :) Jun 05 13:31:14 odd that apt-get --build source does not warn about that Jun 05 13:33:54 phh: for trivial merge, click the second sha hash at the top, that links to the commit Jun 05 13:33:57 sorry Jun 05 13:56:23 hi, how do I build SHR with autorev for SHR+FSO with bitbake master? Jun 05 13:56:33 mickey|away, any idea? Jun 05 13:56:49 because bitbake 1.10 implement something like remote count Jun 05 13:57:14 so I don't know the impact on SRCPV and AUTOREV Jun 05 20:33:39 anyone uses 2.6.34? or at least tried? Jun 05 20:34:01 radekp: there's still at least the touchscreen bug Jun 05 20:34:44 lindi-: which one? Jun 05 20:35:02 lindi-: i get "selected device is not a touchscreen I understand" Jun 05 20:35:07 lindi-: is this the bug? Jun 05 20:35:09 radekp: http://lindi.iki.fi/lindi/cgi-bin/openmoko-bugs.py?q=touchscreen Jun 05 20:36:14 lindi-: but there is fix for the #2328 Jun 05 20:36:26 radekp: yes very recently Jun 05 20:36:47 radekp: the bug is still open :) Jun 05 20:36:50 lindi-: i wish to test it but, getting the message above ^^^ Jun 05 20:37:05 radekp: what prints that? Jun 05 20:37:19 libts Jun 05 20:37:30 when i run qtopia Jun 05 20:37:43 hmm let me try X Jun 05 20:39:36 X gives the same error Jun 05 20:39:48 radekp: hexdump? Jun 05 20:40:09 i tried input-events 1 Jun 05 20:40:15 and it gives events Jun 05 20:40:39 lindi-: tried geny's patch? Jun 05 20:43:38 JaMa: not yet Jun 05 20:43:44 JaMa: hi, in SHR 2.6.34 touchscreen works? Jun 05 20:43:57 radekp: y Jun 05 20:44:27 hmm maybe you have more recent tslib version... Jun 05 20:44:31 radekp: but it's not filtered in kernel driver Jun 05 20:44:35 radekp: evdev actually Jun 05 20:44:48 and ie in enlightenment it's quite problematic Jun 05 20:44:57 so you dont use tslib in SHR? Jun 05 20:45:13 single click is too often detected as 1px drag Jun 05 20:45:25 no, we use evdev Jun 05 20:45:42 ahh oki, that explains it... Jun 05 21:44:27 TAsn: good evening, i have a small question about using window mangager Jun 05 21:47:52 hello, I have a question: after starting an application and pressing on the home button (on top) howto switch bach to that application ? Jun 05 21:50:10 nschle85: illume2? Jun 05 21:50:15 TAsn: cookie for ya in patchwork Jun 05 21:50:32 JaMa: yes Illume2 Jun 05 21:50:37 nschle85, if you press somewhere without icons in the bar, a shelf should appear under the bar with the buttons you are looking for :) Jun 05 21:50:59 if not, run: killall shr_elm_softkey && shr_elm_softkey & Jun 05 21:51:04 and it'll appear Jun 05 21:51:17 dos1: the problem is with first wizard when it will take you more then 10 sec.. Jun 05 21:51:23 dos1: as I said on ML.. Jun 05 21:52:12 JaMa: well, then that's probably illume which should be fixed; but it's still better with this workaround than without it Jun 05 21:53:50 pespin: no shelf appears Jun 05 21:54:05 dos1: TAsn proposed proper fix.. just nobody implemented it.. Jun 05 21:55:25 JaMa, TAsn: what was that? Jun 05 21:55:47 pespin: ok ill try the killall ...... Jun 05 21:55:54 dos1: http://www.mail-archive.com/community@lists.openmoko.org/msg59535.html Jun 05 21:58:43 pespin: wow ! your killall ... worked Jun 05 21:59:12 but does it now wor after reboot ? Jun 05 22:00:29 nschle85: also read http://www.mail-archive.com/community@lists.openmoko.org/msg59535.html Jun 05 22:00:48 nschle85: and no, won't work after reboot if you don't follow that e-mail Jun 05 22:04:55 JaMa: thank you Jun 05 22:14:34 JaMa: i have a second question Jun 05 22:15:29 JaMa: month ago the system clock was automatically updated after gsm login Jun 05 22:16:18 JaMa: but since several month this does not work, what i am doing wrong ? Jun 05 22:24:54 nschle85: check /etc/localtime /etc/timezone, sometimes fsotdld had some problems with those Jun 05 23:24:07 freesmartphone.org: 03daniele.athome 07libframeworkd-glib * r95a266d3ffa1 10/src/odeviced/ (4 files in 2 dirs): Jun 05 23:24:07 freesmartphone.org: odevice: implemented Audio.GetVolume, Audio.SetVolume Jun 05 23:24:07 freesmartphone.org: Signed-off-by: Daniele Ricci **** ENDING LOGGING AT Sun Jun 06 02:59:56 2010