#qi-hardware IRC log for Friday, 2010-06-18

xiangfueharrington: let me test again in my nanonote.00:00
eharringtonJust ssh'd  in to check files again. I had to mount mmcblk0p1 /sd - then ls -l on /sd/boot shows uImage is there.00:01
xiangfueharrington: I got a "ext2fs_devread()" error in another 8GB sdcard. I will try to format the sdcard to 2 partitions and try again.00:17
xiangfueharrington: maybe something wrong with the mmc driver. you can try some sd card < 8GB, I will work on this recently.00:35
eharringtonOK, thank-you. I do recall reading somewhere that there might be a > 4gb, not sure if it is related, sorry don't remember where I saw it. I will try some things also.00:38
pitangaHello folks! In http://en.qi-hardware.com/wiki/Updating_Ben_NanoNote_software#Alternative_to_using_the_reflash_ben.sh_script -> Notes it is stated "Usbboot is not very robust. Neither on the side of the code that is running on the NanoNote, nor how it implements the USB protocol, nor on how it handles NAND. It's much better to let the Linux kernel do this. Unless there is a real new feature in u-boot, leave u-boot unchanged on your NanoNote now". 00:58
kyakpitanga: so?01:13
eharringtonI find it stable, albeit slow, just scp'd 135 mb to nano via usb, took 8 min's.01:15
eharringtonor is that my sd card that is slow? hmmm, will have to experiment01:16
kyakusbboot as far as i know is using other driver than usb gadget01:16
eharringtonTried re-partition 8 gb sd to only 1 3 gb part, still not able to boot debian at /boot/uImage. Will find some 2 or 4 gb sd's tomorrow to test.01:34
eharringtonAnyone used: downloads.qi-hardware.com/people/xiangfu/tmp/openwrt-xburst-qi_lb60-u-boot.bin with smaller sd?01:35
eharringtonpitanga: Are you having errors with:  "nprog 2048 openwrt-xburst-qi_lb60-root.ubi 0 0 -n" ? If so, you should try "nprog 2048 openwrt-xburst-qi_lb60-root.ubi 0 0  -n"  (extra space between last 0 & -n)01:45
pitangaeharrington: do I need to "nerase" before "nprog"?01:47
eharringtonI did that step the first attempt, but skipped it on later tries (because all steps other than #11 appeared to work fine)01:49
xiangfupitanga: yes. to refresh rootfs. we need "nerase"01:52
pitangaxiangfu: what if I don't want to erase u-boot, as recommended on the note I mentioned above? Or should I just forget about that note and reflash everything?01:53
xiangfupitanga: [nerase 16 512 0 0] this command will only erase the rootfs partition01:54
pitangaxiangfu: thanks!01:54
xiangfupitanga: for more info about usbboot you can search in en.qi-hardware.com :-)01:55
pitangaxiangfu: thks!01:55
nllpntri have a minor problem with the nanonote i just received, if anyone cares to pipe up...03:01
nllpntrtried to flash debian, something failed, although without any errors, and it won't boot at all.03:03
jluisnllpntr: Have you tried to press alt+f2? did yow see any message when powering up?03:06
nllpntrno message on power up, nothing at all. i assume alt+f2 would be after power up?03:07
jluisdebian don't present info on tty1 so you have to change to tty2 to loggin but the flasshing process sometimes fail without error03:11
nllpntrso assuming flashing failed, how might I resurrect this thing?03:12
nllpntroh, followed instructions from here: http://pyneo.org/howto/debian/nano.html03:12
jluisto make you use the carbonized ruber ;P03:12
nllpntrI have a feeling I missed some major detail and should not have used said instructions...03:13
jluisthe instructions worked for me but  just to be sure connect your neo to usb anb see if it is recognized as a ethernet03:15
nllpntrit's behaving as a brick.03:16
freespaceif you short the usb boot pin, and then plug it in, does a new usb device show up?03:16
nllpntrno, can't usb boot03:17
freespacethat doesn't sound good03:17
nllpntrwhoah, wait a tic. it just booted.03:18
nllpntrwell, connected via usb03:18
freespacei am that fear some :P03:18
jluisor low battery03:22
freespaceyou would need to take the battry out for short the usb boot pads03:23
nllpntrshould I reinsert the batter while it flashes?03:23
nllpntrdoes that matter?03:24
nllpntrperhaps it died last time during the flash? doesn't seem likely with usb power...03:24
freespacewhen i reflash03:24
freespacei only short it until the flash script finds it03:24
freespacethen i leave it alone03:24
freespacew/o battery until it finishes03:25
freespacethen i unplug03:25
freespaceput battery in03:25
freespaceand it boots03:25
freespacei noticed some strange behaviour if i put the battery in while still plugged in03:25
nllpntrwell, that's no different from my first attempt. it's still flashing now, we shall see03:25
jluisben recives suficient power from usb, is not needed to reinsert the battery03:25
nllpntrthought so03:26
freespacenod, the important thing is it isn't bricked :D03:26
nllpntrI guess i'm just finding it difficult to short those pins reliably03:26
nllpntrdon't have proper tools using <ahem> steel wool03:26
freespacedid you get the little rubber dome thingy?03:27
nllpntrno, is that supposed to be included?03:27
freespaceoh i got one03:27
freespacecheck carefully03:27
nllpntrhey, the wiki does say "_WHATEVER TECHNIQUE YOU CAN COME UP WITH_"03:27
freespacei almost threw it out03:27
freespacemine came in a small palstic bag03:27
freespaceit is wha you find if you take a kb apart03:28
freespacethe carbonised rubber in it is what i use03:28
freespacei position it over the switch, then press down03:28
freespaceit should be with in the instructions03:28
freespaceand the screen cleaning cloth03:28
nllpntrodd. I was careful. didn't see a smaller bag.03:29
freespaceok, maybe it is a new thing03:29
freespacei ordered last week03:29
nllpntrI'll figure something out03:29
freespacelick it03:30
nllpntralmost did try that actually03:30
nllpntrok, flash is done. I unplugged, inserted the battery... waiting for something to happen03:31
freespacewhich firmware?03:31
nllpntrhmm. embarassingly, I don't know...03:33
freespaceflash the latest offical from qi03:37
freespaceand once you know that works, go from there?03:37
nllpntrgot a quick link? I'm finding their pages a bit spaghettified03:38
freespacethose were the instructions03:38
freespaceand then just use reflash_ben.sh03:39
freespaceit will automagic download03:39
freespacethe latest images03:39
nllpntrah right03:39
freespaceso just download the .deb or whathave you03:39
freespacethen reflash_ben03:39
freespacethat page should be restrucutred03:39
freespaceonce i can pass the captcha i will edit it03:40
nllpntrsweet. working. perhaps I will do some more homework before operating on this thing.03:43
freespaceit is quite addictive03:48
nllpntrseriously. this is all i've been playing with since I got home from work.03:49
nllpntryay! all is right with the world03:54
nllpntrthanks for all the help folks03:54
nllpntrI need to sleep before I get assimilated03:54
rafalarsc: or any one knows if there is a git for the current kernel in the official openwrt distro QI uses?08:00
rafaxiangfu: you know perhaps right?08:01
rafaxiangfu: I mean, without to use the whole openwrt building process.08:03
xiangfurafa: no. the current git not sync with openwrt. so it's old08:04
rafaxiangfu: but I would like to build just the 2.6.32 kernel, should I use the  git://projects.qi-hardware.com/openwrt-xburst.git ?08:06
rafaI only want to work a bit with kernel, no with the whole openwrt. Perhaps I do not understand well that git, but it looks like it manages the whole official distro08:07
xiangfurafa: yes. the openwrt-xburst.git manages the whole official distro.08:10
xiangfurafa: and there is another kernel.git in projects.qi-hardware.com08:10
mthdo you need 2.6.32 specifically? otherwise you could compile the 2.6.34 kernel from the qi-kernel git08:10
xiangfurafa: which only have kernel source code.08:10
rafamth: yes, we use 2.6.34 kernel in jlime. But we need 2.6.32 to test zimage initramfs and kexec.08:19
rafaxiangfu: the kernel.git in projects.qi-hardware.com is just for 2.6.34 it seems. Well, thanks anyway, I will try to take 2.6.32 from openwrt-xburst.git ;)08:20
xiangfurafa: the master bransh is 2.6.32. but it's old. I will try to update it now08:41
xiangfurafa: I want make a automatic sync with openwrt-xburst.git but always have problem with patch the openwrt-xburst.git's kernel patches.08:42
rafaxiangfu: I can imagine.. a sync between two tools-repositories should be hard to keep :(08:45
xiangfuHi rafa,  here the network is not good. so I will go home then upload the kernel to qi-kernel.git. will be three hours later.09:11
rafaxiangfu: thanks a lot !!!! I will be too happy if you can do that .. Again thanks09:21
mthlarsc, zear: in the gmenu2x git, there is the autotools build but also src/Makefile-zear09:41
mthit would be easier to maintain only one build system09:41
mthI'm not a fan of autotools, so I actually wouldn't mind building with a hand-coded Makefile, but I don't know how the other maintainers feel about that09:42
zearmth, Makefile-zear is what you guys renamed the Makefile when you adopted my nanonote port to owrt09:43
zearoh right, you weren't maintaining it, well, it's what them guys did ;)09:43
mthI only got commit rigths yesterday09:44
zearyeah, my bad ;)09:44
zearmth, for some reason i confuse you with qi-guys when you talk in this channel :D09:44
mthanother question is how to do support for different devices09:46
zearmth, well, once it's adopted to owrt i don't touch it, because owrt is black magic for me :D09:46
larscvirtual Device class and an implementation for each device09:47
larscwith setBacklight, getPower and such ops09:47
mthsounds good09:47
zearthat's a good idea09:47
mthselect device at compile time or run time? (since you said 'virtual')09:48
zearalso, it needs setKeymapping09:48
zearbecause i hate how we use virtual GP2X buttons now09:48
larscboth would be poissble09:48
mthI think compile time would be better, to keep the footprint low in case more devices are added later09:49
mthalso it would allow a particular device implementation to use libraries that are not present on other devices09:49
mthzear: it's not only in the openwrt build, but also in the gmenu2x git tree09:50
mthI'm building it from git for the Dingoo09:50
zearmth, if you want to build it for the dingoo, you need to revert the root path and battery status device back to dingux settings09:51
mthswitching from sparsehash 1.7 installed in the toolchain to sparsehash 1.6 in the gmenu2x source tree fixed a crash, although I have no idea why09:51
mthzear: that's why I was asking how to handle different devices :)09:51
zearso download joyrider port's code and compare it with the one from the git09:51
mthbattery path is the same between NN and OpenDingux, but backlight probably not09:52
zearmth, well, this git build is based on a very very dirty hack of mine09:52
larscwell, backlight is /sys/class/backlight/*/09:52
mthok, so you won't be offended if I remove it then and add the needed flags to the autotools build?09:53
mthlarsc: but "*" is not the same, right?09:53
larscmth: as a starting point http://metafoo.de/gmenu2x-device.patch09:57
mththanks, I'll use that10:02
rafamth: do you maintain the gmenu2x git code?10:34
mthwell, I can commit to it, but I don't want to be the sole maintainer10:35
rafaso who is the main maintainer? or it is just who has commit rights will do whatever he wants?10:36
mthI don't know if there is a maintainer10:36
mthif I am in doubt, I check with larsc10:37
mth(whether or not to commit something)10:37
rafamth:  cool10:38
rafazear: we know now where to send our gmenu2x complains :D10:38
zearrafa, i made it clear a couple of times that the flickering of the nn's screen is my fault because i left the jz4740 overclocking code in gmenu2x and it messes with LCD timings ;)10:39
mthI don't like GUI programming, I prefer looking at the system parts and the build10:39
mththe overclocking code should be moved out of the menu and into the kernel10:40
mthwell, the menu should ask the kernel for a certain frequency and the kernel should set that freq10:40
rafamth: is not the menu a sdl application? why it should ask something to the kernel?10:42
rafazear: is it your fault that gmenu2x has that problem? coool! :) at least you will be famous between openwrt nn users10:44
zearrafa, :D10:45
zearmore like infamous :D10:45
rafaa virus110:45
mthrafa: because that's better than mmap-ing /dev/mem and accessing the SoC registers directly10:45
rafamth: well, for sdl I would say that the application just should do the proper sdl calls, and IIRC sdl does not have nothing to do with kernel if you use just the sdl API. But things could change from time to time10:48
rafasome day Vi will ask openoffice to ask kernel to let Vi to use the keyboard :P10:48
kristofferdont say that10:49
kristofferI would be lost without vim and mutt10:49
mthrafa: there is no SDL API for changing the CPU freq10:49
rafakristoffer: haha.. me too10:50
rafamth: that is why gmenu2x should not ask kernel nothing10:50
mthyou want to remove the feature then?10:50
rafamth: of course, if that application is a GUI application written using sdl I do not understand how it would like to do kernel questions. But I am not so interested in gmenu2x anyway. If we use gmenu2x someday then it should be just a GUI app I think.10:52
rafazear: it is your fault I know10:53
mththings like battery status also communicate with the kernel, through the /sys pseudo filesystem10:54
rafasurely because there is not a multi platform middle ware to use10:55
rafaor just a simple lib10:56
mthwe want to introduce a Device class in gmenu2x, which has one implementation per device10:57
mthso not a separate lib, but a separate unit within gmenu2x10:57
rafamth: anyway, cool if you are worning on gmenu2x. I will be happy if it improves ;)10:58
calamarzrafa: is there a way to enable acpi support in the kernel? (sorry if newbie question)11:53
mthisn't ACPI x86 specific?11:56
calamarzoh, I had no idea of that :/11:57
mthI'm not sure, it's an actual question, not a rhetoric one :)11:58
rafacalamarz: no idea, I have set Power Management support in 2.6.34. If something under that works either it is APM or ACPI. I would think that APM could be possible, but no sure.12:05
rafacalamarz: larsc could know12:05
rafaor krisstofer, who is not here :P12:05
calamarzmmm looking at the debian package for acpi-modules, I only see it for x86 and amd64, if that means something12:08
rafacalamarz: do you acpid (daemon)12:08
rafado you see*12:08
calamarzmmm yes, acpid is for mipsel too12:14
qi-commitsXiangfu Liu: add-config-file. SYNC AT: Fri Jun 18 21:01:10 CST 2010 http://qi-hw.com/p/qi-kernel/df6af7e12:51
qi-commitsLars-Peter Clausen: From 9a4567a733b689d5dc803cd27dfaa01bc03dc374 Mon Sep 17 00:00:00 2001 http://qi-hw.com/p/qi-kernel/05256e012:51
qi-commitsLars-Peter Clausen: From 806ead1e454a8a5876b777b22ca67187c4749f32 Mon Sep 17 00:00:00 2001 http://qi-hw.com/p/qi-kernel/80d9e4212:51
qi-commitsXiangfu Liu: From 4939c63a61175f87d93ac0164c7adb40e8410a47 Mon Sep 17 00:00:00 2001 http://qi-hw.com/p/qi-kernel/7c47ff212:51
qi-commitsLars-Peter Clausen: From d76e6b85f28891eecded962793fb8a02cdf26f39 Mon Sep 17 00:00:00 2001 http://qi-hw.com/p/qi-kernel/6373d8b12:51
qi-commitsLars-Peter Clausen: From 2d00c901d3a438c6f750f8b13b329845775ec3b5 Mon Sep 17 00:00:00 2001 http://qi-hw.com/p/qi-kernel/03e316d12:51
qi-commitsLars-Peter Clausen: From df07ed6a52d9f6027ff1753c00b3128fa18dde31 Mon Sep 17 00:00:00 2001 http://qi-hw.com/p/qi-kernel/811022012:51
qi-commitsLars-Peter Clausen: From d3699249d687dc0b4d8d4e0e5ac3f9405d31b1ac Mon Sep 17 00:00:00 2001 http://qi-hw.com/p/qi-kernel/3cbbb1312:51
qi-commitsLars-Peter Clausen: From f6bc212:51
calamarznice date on the qi-commits :p13:13
rafawejp: hey man, could you tell me a link for the gmu source code?14:05
eharrington_greetings, anyone know what default un & pw are for debian in sd card image (it's working!!!) Success with 2 gb sd card with exact same procedur/files which failed on 8 gb card & 8gb with 3 gb part. Progress.14:16
eharrington_xiangfu: got it booted on 2 gb card, unable to login yet14:19
eharrington_am stuck at (none) login: prompt, doesn't like my openwrt un/pw14:21
eharrington_so, dual booting is now running14:24
eharrington_never mind, editted passwd from openwrt, I'm in!14:53
calamarzrafa: http://wejp.k.vu/wp-content/uploads/2010/03/gmu-0.7.0.tar.gz ? or you were asking for a git or st?15:26
wejpi have no idea what he is asking about, i mean the source is on my website of course15:27
calamarzeharrington_: that's good news, yeah15:27
rafacalamarz: thanks15:41
rafawejp: I did not know that you had a web site.. then, google gave me it, so I was guessing that wejp.k.vu would be the proper site to look.15:43
wejpoh ok15:43
rafawejp: perhaps the website is very famous, just that it is the first time that I visit it :)15:43
wejpi thought this would be well known as the site is even mentioned in gmu itself ;)15:43
rafawejp: I have not used gmu yet ;)15:44
wejpok, i see :)15:44
calamarzwejp: i'm trying to finish the debian package this weekend... cannot tell about its quality, it'll be my first debian package :p15:46
wejpnice :)15:46
wejpoh, btw, there will be a new gmu release in the near future, only few things left to do :)15:47
tuxbrainrafa: I want gmu in Jlime :P15:50
rafatuxbrain: yeah, we will have it soon, zear already convinced us  ;)16:06
kyakok, i really think to have to write to mailing list about this, but..16:33
kyaki'm doing some tests with SDIO WiFi card, and it's not working very well16:34
kyaknow i've lowered the configuration to a simple ad-hoc between my laptop and Ben16:34
kyakand there is a 50 % packet loss16:35
kyaknot 49 and not 5116:35
kyakcan be seen so clear with ping16:35
kyakand a weird messages about sparse IRQs and sdop interrupts in dmesg16:36
kyakso i just posted16:54
tuxbrainkiak, you think is the card or the sdio of BNN?17:10
kyaki suppose it's the ks7010 driver17:12
kyakbut i can't be sure17:12
kyakbut the only reason why i think so is because the message coming from ks7010 driver17:15
kyaki thought at first it might be a conflict with mmc card driver, so i tryed to disable it in various ways17:16
kyakbut ks7010 seems to rely on some functions of mcc17:16
kyakdriver, so it doesn't work at all without mmc driver17:17
kyaktuxbrain: when you have some free time, would you mind to test it with the latest software?17:18
kyakad-hoc or OPEN configuration.. to see if this problem is existing for you17:18
kyaki'm praying for it not to be a hardware issue17:19
kyakbecause i must admit that the Wi-Fi card is not fitting so well as regular sd memory card17:19
kyakfitting in the slot i mean17:19
kyakit's not ejected17:20
kyakot not always ejected17:20
tuxbraincross your fingers to I found the sd... I have not see it arround for long time, I will do some archoelogical research later on this weekend17:21
kyaklol, ok :)17:22
kyakit's easy to loose, yes17:22
kyakthat's why i try to keep it in a case17:22
calamarztuxbrain: did you get any progress with the keymouse configs?17:27
tuxbrainsorry calamarz, too many fronts, I still having to recompile the kernel due input-core has to be a module as well,... and the worst of all,  the gps module arrives, the elphel cam is working like a charm, so body is calling for some hot tin, wires , video tutorial and true metal. so I think keymouse will wait until I get a gps fix :P , pics on the preliminars http://www.tuxbrain.org/downloads/nanonote/gps/18:35
kyaki see this problem hasn't been solved since then18:40
calamarzyeah that gps looks promising ! :)18:56
Textmodemorning all19:05
tuxbrainoh is a pitty I can't stream outside on line the video  on nanonote with the elphel at 1920x1088 at 25fps  is streaming to my PC,  , once I install the new image on BNN I will do a quick tour with this...., man I can see my every line of my fingerprints, I'm still amazed by the definition on the close ups.19:07
tuxbrainwell let's stop playing with the toys and let rest until moring, good night Textmode :)19:07
Textmoderest well, tuxxy19:09
mthlarsc: which branch is this commit attached to? jz4740: fb: Small cleanups http://qi-hw.com/p/qi-kernel/f6cda2a19:39
mthah never mind19:39
mthI was confused because the commit is 17 days old, but was not pushed until yesterday19:40
larscyeah, i forgot to push some cleanups19:47
larscpreparing v2 of the jz4740 patches right now, which will probably go into ralfs queue19:48
mth-   mmc_suspend_host(host->mmc, PMSG_SUSPEND);19:48
mth+ mmc_suspend_host(host->mmc, 0);19:48
mththis change breaks compilation in my tree19:48
mthRalf being the MIPS maintainer?19:49
larschow can that break compilation?19:52
mthdrivers/mmc/host/jz4740_mmc.c:946: error: incompatible type for argument 2 of 'mmc_suspend_host'19:52
larschm, k19:54
mthpm_message_t is a struct, so 0 seems to be invalid indeed19:54
larsci think it happend because the second parameter is gone in .35 and i backported a change from there19:56
larscwill fix it later19:56
mthwith that change reverted, it compiles fine and boots on the A32020:00
urandom_someone already flashed the latest openwrt image?21:11
qi-commitsMaarten ter Huurne: Fixed most warnings. http://qi-hw.com/p/gmenu2x/9f05aaf21:15
qi-commitsMaarten ter Huurne: Remove manually created Makefile, since the autotools build is now usable. http://qi-hw.com/p/gmenu2x/d8dd09021:15
qi-commitsMaarten ter Huurne: Enabled code specific for embedded devices and enable more warnings. http://qi-hw.com/p/gmenu2x/29b827721:15
qi-commitsXiangfu Liu: add jz4760 support to stage2, make it can reflash jz4760 device http://qi-hw.com/p/xburst-tools/4cdac8821:21
qi-commitsXiangfu Liu: add jz4760 support to xburst_stage1 http://qi-hw.com/p/xburst-tools/bcc23ae21:21
qi-commitsXiangfu Liu: make the usbboot detect the target cpu type by usb id http://qi-hw.com/p/xburst-tools/046a7e721:21
qi-commitsXiangfu Liu: cleanup head files http://qi-hw.com/p/xburst-tools/d0db32721:21
qi-commitsXiangfu Liu: add jz4760 support to stage2 http://qi-hw.com/p/xburst-tools/7a5a09021:24
qi-commitsMaarten ter Huurne: Fixed order of includes in "cpu.cpp". http://qi-hw.com/p/gmenu2x/7a5d09921:35
qi-commitsMaarten ter Huurne: Removed ListView and ListViewItem class, since they were not used anywhere. http://qi-hw.com/p/gmenu2x/1d6ad0321:35
emebhey - 6-15-2010 image. That's new.21:43
qi-commitsMaarten ter Huurne: Do not run the main loop inside the GMenu2X constructor, but in a method. http://qi-hw.com/p/gmenu2x/3efa60221:51
Textmodeew...was it really doing that?21:54
mthsometimes I'm amazed it runs at all21:58
urandom_ahh no! i bricked my ben once again :(22:04
urandom_so nobody tryed the latest image yet? just want to know if it works at all22:08
qi-commitsMaarten ter Huurne: Cleaned up BrowseDialog class: removed unused methods and fields, restricted access where possible. http://qi-hw.com/p/gmenu2x/ad22a9022:09
urandom_yeah!!! sucess!22:24
qi-commitsMaarten ter Huurne: Removed SDL_ttf and FreeType from the libraries to link with, since there is no code that uses these libs. http://qi-hw.com/p/gmenu2x/6ec615522:32
qi-commitsMaarten ter Huurne: Add explicit destructor to BrowseDialog. http://qi-hw.com/p/gmenu2x/9c5799c22:56
qi-commitsMaarten ter Huurne: In "utilities.h", only include those headers that are needed by the definitions in "utilities.h". http://qi-hw.com/p/gmenu2x/660b4f022:56
qi-commitsMaarten ter Huurne: Fixed intendation. http://qi-hw.com/p/gmenu2x/c52b23923:10
qi-commitsMaarten ter Huurne: Removed SelectorDetector class, since it is not used. http://qi-hw.com/p/gmenu2x/5db8ac823:10
--- Sat Jun 19 201000:00

Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!