#qi-hardware IRC log for Sunday, 2010-06-27

rafaurandom_: you there?00:10
rafacan you tell me which boot loader you use?00:11
rafaurandom_: it is clear for me that you did not use the bootloader from jlime right?00:11
urandom_i updated the bootloader when i flashed the latest openwrt image so i use that one00:12
rafaah.. okey, well, perhaps you could have realized that it did not boot from fat because you did not boot from the bootloder from jlime wiki :)00:13
rafaurandom_: have you booted finally?00:14
urandom_the kernel at least, (seen jlime logo) but i think it did not find the rootfs which was on second partition00:15
rafatoo weird.. perhaps that bootloader just want to mount the first partition?00:16
rafaI do not like too many different binary bootloaders around00:16
rafaperhaps I write a short mail explaining that it should work with bootloader in jlime. That thing would annoy users surely.. it is not nice to use specific files for different stuff :(00:17
urandom_why not just use the "official" qi one?00:17
rafaurandom_: the bootloader from jlime wiki was a tmp qi one, and the official one at that moment did not work for microsd00:18
urandom_also debian is booting fine from one big ext2 partitions so no need for 3 different partitions at all00:19
rafaurandom_: so if it changes going to the good way okey, but now you can not boot jlime because the "current" official one does not mount second partition00:19
rafaurandom_: try to install in debian a package with 5 or 6 dependences00:19
rafanevermind.. I think that it is just a temporary stuff.. all the stuff would mature with the time.. and if somebody just need debian like it is then he would use the proper bootloader, if somebody needs jlime he will have to use the proper files, and so on..00:21
urandom_i am not really interested in debian, jlime seems much better for my needs00:21
urandom_just got the idea using ext2 from debian00:22
rafaurandom_: I would like to use initramfs and kexec, so we can have a general second stage bootloader for any needs..00:22
rafaso you could boot from any partition in NAND or sd, or just to do maintanence tasks from that initramfs00:23
rafaand also, it would let users to update theirs system without usbboot and all that no nice stuff00:24
urandom_well the good thing of this story is that i got epic fdisk skills now ;)00:24
rafaurandom_: haha great man00:24
urandom_yeah i would like to easy flash everything from sdcard (but thats now partly possible, at least for kernel und rootfs)00:25
arctanxrafa: your uImage is tagged -qi... are you using the same kernel config as the openwrt-xburst kerneL?00:34
urandom_now booting jlime with rootfs00:35
urandom_it takes a bit long00:36
arctanxmine's doing a fsck and generating ssh keys00:36
urandom_omg x and no crapy gmenu2x, awesome!00:37
arctanxI think it wins the long boot award00:37
urandom_rafa thanks for making jlime, it is awesome!00:38
urandom_openwrt currently looks like shit against it (but boots damn fast)00:40
urandom_wow looking movies on the ben00:44
urandom_second boot isnt that bad00:47
urandom_omg it comes without lua preinstalled! unforgiveable00:49
urandom_and no python, no ruby, no guile00:50
kyakcalamarz: do you have some documention/howto about the keymouse/scummvm?01:44
kyakrafa: "Kernel panic - not syncing: No init found. Try passing init= option to kernel" when booting fresh install of Jlime...02:18
kyakbtwm i had to put the uImage to /boot/ on the first ext2 partition02:24
kyakotherwise it won't boot at all02:24
kyakrafa: ok, it seems that it is no longer required to have two partitions for Jlime03:35
kyak"uboot will load "/boot/uImage" in the sd card first ext2 partition.03:36
kyakthen kernel will also boot to rootfs at sd card first ext2 partition."03:36
kyakyep, got it running from a first ext2 partition03:47
arctanxHumm has anyone else run into errors compiling qt4 for the openwrt build?04:05
kyakxiangfu: you there?07:50
xiangfukyak: hi07:50
kyakhi man07:50
kyakwanted to piont out to something.. don't know how to fix this correctly07:50
kyakin your latest commit, you removed the zImage build07:51
xiangfukyak: yes.07:51
kyakbut, have a look:07:51
kyakcp: cannot stat `/home/bas/build/openwrt-xburst/build_dir/linux-xburst_qi_lb60/linux-': No such file or directory07:51
kyakmake[5]: *** [install] Error 107:51
kyakit is still trying to install it07:51
kyaktherefore, build fails07:52
xiangfukyak: thanks for report. there is another makefile need modify.07:53
kyakno problem :) if i knew how to fix it, i would do it. But i definitely lack knowledge about the internals of build process07:53
xiangfukyak: in this file: http://projects.qi-hardware.com/index.php/p/openwrt-xburst/source/tree/xburst/target/linux/xburst/image/Makefile07:54
kyakok, i see it :)07:55
xiangfukyak: just for you information :). I will commit it now , thanks again.07:55
kyakthank you!07:55
qi-commitsXiangfu Liu: remove cp the zImage file to bin/xburst folder http://qi-hw.com/p/openwrt-xburst/191f22d07:58
calamarzkyak: http://en.qi-hardware.com/wiki/Virtual_mouse08:14
nebajothyes hello o/08:27
calamarznebajoth: hi08:30
calamarzsomebody knows which is the option in the kernel to output info when booting?08:30
nebajothwhat are you working on, calamarz?08:31
calamarznebajoth: jumping from one thing to another :p08:38
calamarznow playing with bloating the kernel08:38
calamarzyesterday I finally got the keymouse working (it was only enabling mouse/uinput in the kernel)08:39
calamarzand now I'm trying to enable acpi for getting power info and suspend stuff working08:40
calamarznebajoth: are you behind the nanohacks wordpress?08:46
nebajothcalamarz: yep08:47
nebajothooooh, acpi would be super nice08:48
urandom_whats that "unable to open display" error on jlime about? first time i got it i just reflashed since i thought something might have got corupted but now i got it a second time08:51
calamarznebajoth: I thought the other day we could use an elgg there... nothing against the wordpress, in fact the theme is really nice... but I know some people that are working hard on elgg for community work (integration with wiki, task management, integration with mailin list etc...)08:52
calamarznebajoth: --> http://lorea.cc08:53
nebajothI have no religious attachment to buddypress09:04
nebajothin fact09:04
nebajothI've deployed elgg before09:04
nebajothand I used this opportunity to see how buddypress was09:04
nebajothI like it pretty well09:05
nebajothbut I'll look into this09:05
nebajothlorea.cc looks pretty cool09:05
nebajothboth sitewise and missionwise09:05
bartbesurandom_: when do you get that error?09:22
urandom_while booting09:23
bartbeswhat program?09:23
bartbes(does it originate from)09:23
urandom_couldnt read it, they use a very small font, also i have reflashed now and everything works now09:24
bartbesehm.. okay09:25
urandom_there seems to be no working way to shutdown the ben with jlime09:26
bartbesnot even 'halt'?09:27
urandom_well it is doing something but i still have to take out battery for the screen to go blank09:28
rafaurandom_: shutdown, no, it does not work yet, as muffinman wiki says :)09:29
kyakcalamarz: cool, thanks!09:30
wolfspraullekernel: nice! do you have a download link for this video somewhere so I can convert to Ogg Theora and upload to the Qi wiki?09:50
wolfsprauldo you have a higher resolution than 320x240?09:52
urandom_if somone else has problem with "unable to open display" in jlime and is booting it form one big ext2 then editing etc/muffinman-extras/00fsck.sh is the solution (just for the archive)09:59
calamarzhow can the kernel boot parameters be changed? (the ones in /proc/cmdline)10:04
larsccalamarz: in uboot10:29
lekernelno higher res unfortunately... as I said on the MM list I have a rather crappy camera11:48
lekernelmaybe Adam or Michael could make a better video...11:49
bartbeswow, irc really is instant :P11:50
bartbesnext time try email :D11:51
xdpiratei think my nanonote is faulty12:24
xdpiratei was using it just now, and suddenly the screen faded into a complete white screen12:25
xdpiratetried taking out the battery, leaving it out, then reinserting it, it just lights up the lcd with the same completely white screen12:25
wolfspraulxdpirate: were you on battery or powered over the USB cable?12:25
wolfspraultry to take the battery our and power it over USB cable12:25
wolfspraulthere are two options: 1) the battery is empty 2) the batery connector does not sit very tight mechanically with the battery itself12:26
xdpiratejust tried reinserting the battery, it remained off and wouldn't turn on, i'll try connecting it via usb now12:26
wolfspraulwell maybe the battery is just empty?12:26
rafaurandom_: about "editing etc/muffinman-extras/00fsck.sh".. that should be done from a fresh install12:26
xdpiratei find that very weird, i just charged it a couple of days ago and haven't used the nanonote at all since12:26
rafaif somebody sees "unable to open display"12:27
rafaurandom_: then he surely already has a broken ext2, well it perhaps works, but it already has errors and he should install from scrasch12:27
wolfspraulxdpirate: let's first find the root cause12:28
wolfspraulscreen turning white is typically because of loss of power12:28
xdpirateit booted via usb+usb to power adapter12:28
wolfspraulwhich can be low/empty battery, or sometimes the battery connector12:28
wolfspraulok, leave it plugged in, now insert the battery and see whether the red charging LED comes on12:28
xdpirateyay, great success12:29
xdpirateprobably just a low capacity on the battery12:29
wolfspraulnext thing would be to find out why it went down even though you said it was full12:30
wolfspraulbattery should discharge itself only about 1% per day I think12:30
xdpirateyeah haven't used it much since last charge12:30
wolfspraulok, first leave it connected until fully charged now12:31
xdpiratebtw can i get the battery % remaining from anywhere?12:31
wolfspraulI think you can but I don't know where :-)12:31
xdpiratehehe :P12:31
wolfspraulit's not very precise because we have no coulomb counter or precise current measurements12:31
wolfspraulonly voltage12:31
wolfspraulmaybe someone else knows...12:31
urandom_xdpirate http://en.qi-hardware.com/wiki/Ben_NanoNote_New_User_Guide#Power_state_transitions last one12:32
wolfspraulxdpirate: hey I saw you added a translation to the wiki, nice and thanks a lot!12:33
xdpirateurandom_, thanks12:33
xdpiratewolfspraul, hey no probs, will probably do more12:34
wolfspraulgreat, that would be fantastic!12:34
xdpiratewolfspraul, i just noticed that the battery compartement is abit bigger in length than the battery12:37
xdpirateso there's a bad connection there12:37
xdpirateneed to find something to put in there so it sits right12:37
wolfspraulyes that is possible too12:38
wolfspraulI am not really happy with the battery connector12:38
wolfspraulyou can try to pull it out a bit with a small screwdriver, bending the metal into the battery compartment. but be careful to not break anything (so maybe not :-))12:39
xdpirate850 mAh isn't all that great tho12:39
xdpiratei got a nokia battery here that's also 3.7V but it doesn't say anything about the mAh12:39
wolfspraulor you put some paper into the other side so the battery presses more firmly against the connector12:39
xdpirateyeah that was my idea aswell12:39
wolfspraulthey should also work, but the nokia is even lightly narrower so you definitely need something on the other side12:40
xdpirateit's a BL-4C battery, according to the great interwebz it's only 780 mAh12:40
xdpirateanother source says 72012:41
xdpirateboth smaller tho12:41
urandom_it just would be nice to get some "pls load battery" message before it fades to white, would be more usefull than the new image printing lot of shit everytime i insert an sdcard12:42
xdpiratetrue, i never know how much battery charge is left12:43
wolfspraulyeah definitely we could be smarter about the battery running low12:43
xdpiratethe LED powers off when it's fully charged right?12:46
xdpiratemkay i'll leave it off and charging now then12:46
xdpiratei put a piece of cardboard from a pizza box to press the battery up against the connector12:47
xdpiratethanks for the help wolfspraul12:47
wolfspraulno problem12:47
wolfspraulthanks for your patient with our little patient12:47
wolfspraulpatience :-)0000000000012:48
xdpirateyeah urandom_ i tried that script12:48
wolfspraulit's a long way until software is rock stable and polished in all cases...12:48
wolfspraulyears I think12:48
xdpiratedoesn't work, when fully charged it told me that i had a minus 35% charge12:48
urandom_wrote it in the issue list http://projects.qi-hardware.com/index.php/p/ben-nanonote/issues/84/12:50
urandom_i really would prefer fixing bugs instead of reporting bugs :/12:51
emebhmmm - using the 6/15 openwrt image and SD card is acting odd.12:55
emebdoesn't automount at boot12:55
emeband if I eject/insert then kernel spews errors if trying to mount later.12:56
emebonly way to use it is to boot with card in and then issue 'mount /dev/mmcblk0p1 <mount point>' manually.12:56
xdpirateadd that line to /etc/profile12:57
xdpirateand it'll mount at boot12:57
kyaknot a good solution. it will try to mount after each login12:57
kyakbetter /etc/rc.local12:58
urandom_emeb i dont think they are errors, just messages12:58
kyakor /etc/fstab, if there is such :)12:58
emeburandom_: but it doesn't actually mount when those occur. Complains about not finding valid FAT12:59
urandom_it can mount ext2 and fat16 partitions without problems for me13:00
emebkyak: agree. rc is prolly best sinc fstab has some stuff in it about being auto generated13:01
emeburandom_: hmm will try some more.13:02
urandom_but i gues the sd stuff on openwrt really needs some more work13:03
emeburandom_: true, but I seem to recall the previous image worked better...13:04
emebanother question: gmu blanks the display, even if you've switched to a text term via alt+ctl+Fx. One solution is to disable blanking in gmu config13:05
emebis there a way to poke something in /sys to turn display back on?13:05
emeblarsc: thx - I figured there was something like that.13:10
calamarzsdschulze: do you mind if I put your notes about compiling debian kernel in the qi wiki?15:00
lunavoraxEvening everyone !15:23
lunavorax51 people and nobody's talking ?15:25
calamarzlunavorax: we're bots :p15:27
lunavoraxThat's what I was expecting15:28
lunavoraxI have a few questions about  the os of the ben nanonote15:29
lunavoraxI can't remember who told me that the flickering screen bug will be fixed soon15:29
lunavoraxBut again I can't remember when I read that15:29
lunavoraxSo I don't know if it's soon soon or already made15:29
larsclunavorax: the bug has been fixed quite some time ago. Unfortunately it hasn't made it into the current images yet :/15:43
lunavoraxOh ok15:43
lunavoraxno pb15:43
lunavoraxWhat happens when it says "Host key verification failed." when I try to connect to the ben nanonote15:44
lunavoraxWhat have I done wrong ?15:44
larscyou have a different key for the nanonotes ip in your ~/.ssh/known_hosts15:45
lunavoraxlarsc, that means ? I have to delete this file and try again ?15:55
bartbesremove the entry for that ip/hostname15:56
lunavoraxhum ok bartbes15:56
lunavoraxThis files doesn't exist on the nanonote so i believe it's on the computer then15:57
larscssh should tell you which line it is when you try to connect15:57
bartbesyes, on the client15:57
larscyes, on your host15:57
bartbeslarsc: client15:57
lunavoraxclient != host, right ?15:57
bartbesthe ssh client15:57
bartbesmeaning, your computer15:58
larscbartbes: depends on your point of view ;)15:59
lunavoraxhaha ok15:59
bartbeslarsc: no, that's the client15:59
lunavoraxI'll just try15:59
bartbesit's definitely the client15:59
bartbesbecause it's the ssh *client* which error16:00
larscbartbes: it's the ssh client, but the usb host16:00
bartbesbut we're not talking usb now are we?16:00
lunavoraxOk hum that doesn't solve the problem16:00
bartbeswe're talking about a network connection16:00
bartbesbut yeah, it's always point-of-view stuff16:01
lunavoraxhum gentlemens16:02
bartbesI'm looking through the man page16:02
lunavoraxWell I should be doing the same even if i'm a noob16:03
bartbesyou could try: ssh -o StrictHostKeyChecking=false <nanonote>16:04
bartbes(a temporary solution but that should work)16:05
lunavoraxok i'm trying it16:07
ezdagorAnyone know how to get the mouse to work under Debian? There is no kernel module for it.. if there is one..16:09
ezdagorI used the Debian-lenny.ubi16:12
lunavoraxYes i tried to boot but it didn't worked16:12
lunavoraxI much more used to debian that will be easier for me when I'll succeed to boot it16:13
lunavoraxnod ?16:14
ezdagorNodding my head.16:15
lunavoraxah ok16:15
lunavoraxI was thinking "Nintendo optical disc"16:15
ezdagorTwo things I haven't figured out yet... sound and the mouse..16:16
lunavoraxon the nanonote ?16:16
ezdagorMatchbox works ok with the keyboard, but would be easier with a mouse.16:17
ezdagorYeah. The NanoNote.16:17
urandom_ezdagor have you seen http://en.qi-hardware.com/wiki/Virtual_mouse ?16:17
lunavoraxbartbes, your trick doesn't work16:19
lunavoraxIt ask for lunavorax's password on the nanonote16:19
lunavoraxBut the nanonote only have one user which is root16:20
lunavoraxAnd useradd doesn't seems to be here16:20
lunavoraxor maybe16:20
lunavoraxah ?16:20
lunavoraxOk no16:20
lunavoraxworking sorry16:20
lunavoraxI apologise bartbes16:21
lunavoraxi wrote instead of root@
lunavoraxFirst time ever i finally connect to my ben w/ ssh, thank you very very much16:21
lunavoraxIs it normal that I can't ping google with my ben ?16:29
bartbesif you haven't set your comp up to route, then it is16:29
lunavoraxOk just can't communicate w/ anything else that the pc16:29
lunavoraxoh ok16:29
lunavoraxIf instructions are on the wiki i'm going to read them nom16:30
urandom_lunavorax http://en.qi-hardware.com/wiki/Ethernet_over_USB#NAT_config16:30
lunavoraxthanks urandom_ , if found the article at the same time :P16:31
bartbeslunavorax: you're going to nom the instructions, aren't you? :P16:31
lunavoraxOh weird16:32
urandom_good you found it same time, some people really have problems finding the right wiki pages16:32
lunavoraxI can't echo "1" > /proc/sys/net/ipv4/ip_forward even if I'm sudo16:32
lunavoraxI'll google that16:32
mthare you sure the redirect is done inside the sudo as opposed to outside?16:33
lunavoraxI did not really understand what you've said mth16:33
lunavoraxOr what you meant16:34
mthif you write "sudo echo 1 > somewhere", it first redirects as a normal user and then switches to root for the echo16:34
lunavoraxHum i didn't knew that16:34
lunavoraxSo i'll have to launch su first16:34
lunavoraxAnd then the command line ?16:34
mththe redirection is done by the shell, not by the command you execute in it16:34
bartbessudo -s16:34
bartbesthen the echo command16:35
lunavoraxIndeed, thanks bartbes16:35
lunavoraxI didn't knew that16:35
lunavoraxSo I only input thoses 3lines and  it's done ?16:37
lunavoraxNo extra work ?16:37
bartbesthere shouldn't be16:37
bartbes*unless* you want extras16:37
lunavoraxBut I have to reboot my ben then ?16:38
lunavoraxBut it's still ain't working16:38
bartbesis the dns server set up correctly?16:38
bartbesmaybe you can try pinging an ip first16:38
lunavoraxI don't even know how to set up a dns16:39
bartbesbtw, urandom_, since you linked it, do you use that snippet yourself?16:39
bartbeslunavorax: actually on that same wiki page it tells you how to set up the nanonote16:40
lunavoraxbartbes, no ping doesn't work w/ ip address of computers inside or outside the network16:40
lunavoraxYes and ?16:40
lunavoraxI don't see the connection16:41
bartbesroute add default gw <ip of computer>16:41
urandom_bartbes no never did the ethernet over usb thing, just translated it once to german :P16:41
bartbesokay, because it looks like you have to do that every time you want to use it16:41
bartbesthough of course it *can* be automated16:41
bartbesand I happen to semi-know how16:41
lunavoraxI have to admit this part of the article is unclear for me16:41
bartbeskristianpaul: ok?16:42
kristianpaulirssi is atricky client16:43
urandom_i dont think it has to be done every time or at least there is a way to not have to16:45
bartbesdoes anyone want to know how to automate it?16:45
bartbeswell you sure have to modprobe every time?16:45
bartbescome on, finally I know something!16:46
urandom_just say it already16:47
bartbesokay, first you add iptable_nat to /etc/modules.conf (or /etc/conf.modules)16:49
bartbesthen you write a script with the other 2 commands16:49
bartbessave it somewhere, make it executable16:49
bartbesedit /etc/udev/rules.d/10-local.rules16:49
bartbesadd the following line16:50
bartbesKERNEL=="usb0", RUN+="/path/to/your/script"16:50
bartbes(modify usb0 to the /dev/ device where the nanonote ends up)16:50
bartbesand you're done16:50
bartbesit automagically runs the script when the device is inserted16:50
max_posedonI suggest disctro-specific solution16:51
max_posedonmost distros detect and allow pre-up post-up's for ethernets16:51
bartbesudev powns!16:51
max_posedonyes, but solution to low-level imho16:52
urandom_bartbes sounds like this should be wikified, feel free to add it :)16:52
bartbesurandom_: well somebody should test it first16:53
lunavoraxOk well I just give up16:54
lunavoraxIt's not a problem, but it's better than messing around16:54
urandom_so someone test bartbes solution!16:55
lunavoraxwhat solution ?16:55
urandom_how to not have to modprobe every time16:57
bartbesand not have to do the other stuff16:57
bartbesthe modprobe is relatively easy16:57
lunavoraxI think I've missed that16:57
urandom_lunavorax yeah you have, it will show off in the archives soon16:58
lunavoraxhum ok i'll try this asap then16:58
lunavoraxIs it really complex or is it just me being really incompetent ?16:59
urandom_well if it did not work for you per hand then bartbes method wll not work too16:59
urandom_its a bit complex17:00
lunavoraxhum :/17:00
bartbesunless of course you failed to follow the instructions..17:01
lunavoraxWell there wasn't that much instruction (!?)17:01
lunavoraxWell maybe17:01
lunavoraxI can't telle17:01
lunavoraxI'll retry another day, it's late now I'm a bit tired (and tired of it)17:02
tuxbraincalamarz, rafa: bravo, bravissimo!!!!17:30
tuxbrainI hope both openwrt and jlime has the virtual mouse thing in his respectives future releases :)17:33
tuxbrainrafa: what is the resolution+codec+bitrate recomended to play the video so smoth :)17:42
kristianpaultuxbrain: mouse why?17:42
kristianpaularrow akeys are perfect for small screens17:43
tuxbrainbut there are a lot of soft out there that can be run with minimum porting if they have the mouse thing enabled, (I agree is better a good key driven app, but the example  scummvm17:45
tuxbrainis just to add one feature more to the nanonote17:45
kristianpaullets wait touch screen with  Ya17:45
rafatuxbrain: check the muffinman wiki17:45
tuxbrainno touch screen in ya17:46
kristianpaulyou can sell then tuxbrain17:46
rafatuxbrain: ffmpeg -i your_video_file -ac 1 -ar 22050 -s 320x240 output.mpg17:46
tuxbrainrafa: sorry I was so impatient :) I ask on first link, I have just arrive to it , thanks17:47
rafatuxbrain: and BTW, I think that it is not a good idea to try to have mouse pointer.. X is nice, and X works okey for 320x240.. but our idea is to have just applications useful for qwerty keyboard. But well, you can show us the mouse options :)17:48
tuxbrainkristianpaul: and where do you will connect it ? suggestions?17:48
rafatuxbrain: you there?17:51
tuxbrainagain, I agree than main apps, and news apps created with nn in mind should be qwerty driven, but have such feature enabled to test "already made soft" that was though with mouse in mind, should not harm isn't it?17:51
tuxbrainyes I am17:51
rafasome guys were able to use jlime using the current qi official bootloader, but...17:51
rafacurrent jlime does some little tasks in second ext2 partition17:52
rafaso they were able to boot jlime and use it17:52
rafabut after few reeboots17:52
rafathey broke their filesystems17:52
tuxbrainsounds bad17:52
rafabut they did not note that17:52
rafatuxbrain: yes, of course17:52
rafaI just want to tell you that until we can fix that jlime works nice if you use jlime instructionts17:53
rafaI mean, people do not follow the jlime instructions, and they use their skills to boot it.. but then ... after some reboots the distributions has very weird behaviours...17:54
rafaand they complain that it works bad :)17:54
kristianpaultuxbrain: uart, gpio17:55
kristianpauli think there are two free pins17:55
kristianpaulnear to TP4/TP517:55
tuxbrainkristianpaul: sure on this? I was searching for free gpio on ben to do some experiments17:56
kristianpauli have a backlog about it17:56
kristianpaulfrom some weeks ago17:56
kristianpaullet me check17:56
kristianpaulwhy lattice release in an open source way its lm32 core17:57
urandom_nah i wasnt complaining rafa, also it works good now (even after some reboots)17:58
rafaurandom_: well, you did the proper steps... you realized that it did now work okey and you fixed it.. but other no18:01
tuxbrainI have to find some time to collect all the news , and post them in some visible place18:03
tuxbrainyou are making this device shine even brighter guys18:04
tuxbrainalso the milkimist video is awesome18:05
kristianpaultuxbrain: you sell fpga stuff?18:10
kristianpaultuxbrain: http://paste.debian.net/78939/18:11
kristianpaulthe driver will be as the mmc in openwrt with the wrt54gl i think18:11
tuxbrainkristianpaul: I plan to sell sack and milkymist, but I have no news from sack part, I was waiting for one of the prototypes18:20
tuxbrainkristianpaul lars, so tp35 tp36 are free to use?18:21
tuxbrainsorry larsc18:21
larscthose are connected to a flash socket which is not used18:21
larscso you can use them, yes18:23
tuxbrainhumm? so is not a quimera think to bitbang them to use as 9600 uart?18:23
mthlarsc: I was thinking of reconfiguring the PLL only when scaling_max_freq changes and handle changes to scaling_cur_freq by changing the dividers18:23
mthsince changing the dividers has a much lower latency18:24
mthdoes that approach make sense?18:24
qi-commitsLars-Peter Clausen: MIPS: JZ4740: Drop eccpos array from ecclayout. http://qi-hw.com/p/qi-kernel/4826e1418:26
qi-commitsLars-Peter Clausen: MTD: NAND: jz4740: Add custom nand read/write callbacks http://qi-hw.com/p/qi-kernel/87577b618:26
qi-commitsLars-Peter Clausen: MTD: NAND: jz4740: Minor cleanups http://qi-hw.com/p/qi-kernel/1e2d11f18:26
qi-commitsLars-Peter Clausen: mmc: jz4740: request mmc irq with IRQF_DISABLED http://qi-hw.com/p/qi-kernel/24a28ac18:26
qi-commitsLars-Peter Clausen: mmc: jz4740: Rework irq handler http://qi-hw.com/p/qi-kernel/d55f3fe18:26
qi-commitsLars-Peter Clausen: mmc: jz4740: Only start timeout timeer when sending inital command http://qi-hw.com/p/qi-kernel/117a44c18:26
qi-commitsLars-Peter Clausen: mmc: jz4740: Drop unnecesassry locks http://qi-hw.com/p/qi-kernel/bb3d06c18:26
qi-commitsLars-Peter Clausen: mmc: jz4740: Merge jz_mmc_cmd_done into jz_mmc_irq_worker http://qi-hw.com/p/qi-kernel/128b6b318:26
qi-commitsLars-Peter Clausen: mmc: jz4740: Use bitops to modify waiting flag instead of using a spinlock http://qi-hw.com/p/qi-kernel/0483cdc18:26
qi-commitsLars-Peter Clausen: mmc: jz4740: Minor cleanups http://q18:26
larscmth: sorry can follow you. what are scaling_max_freq and scaling_cur_freq?18:27
mththey're files in /sys/devices/system/cpu/cpu0/cpufreq18:27
tuxbrainmmm then maybe I can attach the gps module without loosing the hwd uart tp4/5, well step by step first I have to test the module in the hwd uart, then I will try the bitbanging thing, also , there is any other gpio free to also add on/off gps functionality?18:27
mthactually scaling_cur_freq is read-only, how it gets its value depends on the governor that is in use18:27
larsctuxbrain: there should be at least one more18:28
larscmth ok18:29
mthI reworked the PLL rate setting code and now I can set any frequency between 25 MHz and 500 MHz with at least 1 MHz precision18:29
tuxbrainlarsc: yay!18:29
mthbut every time the PLL needs a delay to restabilize18:29
mthand I found no info on how long it needs exactly18:29
mthso I'm just sticking to the value that Uli copied from Ingenic (I assume)18:29
mthwhich seems to work but might be on the conservative side18:30
mth32 RTC ticks, so 1 ms18:30
tuxbrainrafa: you know if there is avrdude and avr-gcc in OE repos?18:31
mthwith that kind of latency, it's not possible to switch frequencies often18:31
larscmth: there should be a bit which is set when the pll is stable18:31
tuxbrainrafa: if so I would like to change my actual debian -arduino platform to jlime-arduino :)18:31
mthyes, but I think it just indicates whether the counter reached zero, not whether it's actually stable18:32
mthotherwise, why even have a counter?18:32
larscno idea18:33
mthalso, changing dividers means we could scale just the CPU without changing the clocks of all other subsystems18:33
mthwhen done properly, the overhead of changing all clocks derived from the PLL could also add up to something significant18:34
tuxbrainlarsc: with no rush , but when you know another free gpio(s) to be able to use please let me(us) know, with out hurry, first I have to deal with gps just connected to serial. kristianpaul larsc thank you both :)18:36
tuxbraingood night channel, time to some sleep18:36
mththe only drawback I see is that the resolution is low18:37
mthif the PLL is set to 400 MHz, the CPU can be set to 400 or 200, but nothing inbetween18:38
mthfor max CPU frequencies below 250 MHz, we could use a divider of 2 for the max speed and then 2/3 of the max speed would also be a possible setting18:39
mthbut I don't think many users will cap their CPU speed at 250 MHz18:39
mthsince if the frequency scaling works well, there will not be much gain over having a high max speed with a low average speed18:40
mth"bool enabled" -> there is a bool type?18:44
mthis it a Linux thing or a newer version of the C spec?18:45
larsci think it is in c99 or so. but it's also a linux thing18:46
mthbut it's a typedef then, so things like "!!" are still needed?18:48
larsci don't think so18:49
larscit's a typedef for _Bool18:50
larscand _Bool is from c9918:50
mth"When a value is demoted to a _Bool, the result is 0 if the value equals 0, and 1 otherwise." - that's good18:59
larschm, does 2.6.34 still boot for you with my recent changes?19:02
mthI haven't merged today's changes yet19:02
mtheverything before today seems to work fine19:02
mthshould I try now?19:03
larscif you have time.19:04
mthyes, there is no way I can sleep in this heat anyway19:04
larscit does not boot on the nanonote. hard locks somewhere. although they were only backports from my for-upstream branch19:04
larscah, i might have an idea whats wrong19:05
larscand why it works for you19:06
larscyup... it's the framebuffer driver19:06
qi-commitsLars-Peter Clausen: fbdev: jz4740: Minor cleanups http://qi-hw.com/p/qi-kernel/bd9ee3a19:11
larscrafa: jz-2.6.34 should work again19:11
mthI can still load openMSX on the A320, so I think the mmc changes are OK: openMSX is big enough to break on most mmc bugs19:13
mththe other changes are not used on the A32019:13
bartbesa320, isn't that the dingoo?19:14
mthI'm porting the qi-kernel to it, under the name OpenDingux19:15
bartbesbecause dingux isn't open?19:17
mththe source is open, but the development process was not because it was all done by one man19:17
mthwho stopped updating it over half a year ago19:18
bartbesthat sucks19:18
mthit's the risk of depending on a single hero19:20
mthso far there isn't a big kernel team for opendingux either though19:20
mthbut by reusing the works of larsc and some contributions from Ayla and Uli, it's at least not just me that's working on it19:21
bartbesso was the old developer not willing to let people work on it, or was there a lack of volunteers?19:21
mthboth, I think19:21
mthI'm not sure if he was really unwilling, but he didn't setup up the infrastructure for people to contribute19:22
bartbesI guess that's the first thing you did19:22
mthone change I made is to use git instead of svn, so it's easier to use people's patches without having to give them commit rights from the start19:23
mthand I'm more active on IRC, so more approachable for people who run into problems19:23
mthwe're not there yet though: there is no new boot loader and root fs set up yet19:25
mthwe can still use the ones booboo (original dev) made, but at some point we'll have to make our own to move forwards19:26
mthfor example, to use uImage instead of zImage so the boot loader can pass info about the hardware19:27
qi-commitsLars-Peter Clausen: RTC: jz4740: Propergate write errors in jz4740_rtc_reg_write up to its callers http://qi-hw.com/p/qi-kernel/3deecc119:43
qi-commitsLars-Peter Clausen: RTC: jz4740: Use kzalloc instead of kmalloc http://qi-hw.com/p/qi-kernel/a4c9bcb19:43
qi-commitsLars-Peter Clausen: RTC: JZ4740: Remove deadlocks which could occur if the HW was broken http://qi-hw.com/p/qi-kernel/f934e4b19:43
qi-commitsLars-Peter Clausen: RTC: JZ4740: Check whether rtc structure could be allocated http://qi-hw.com/p/qi-kernel/034b40e19:43
qi-commitsLars-Peter Clausen: RTC: JZ4740: Whitespace cleanups http://qi-hw.com/p/qi-kernel/bdb548b19:43
qi-commitsLars-Peter Clausen: RTC: JZ4740: Add common implementation for jz4740_rtc_update_irq_enable and jz4740_rtc_alarm_irq_enable http://qi-hw.com/p/qi-kernel/15c909719:43
qi-commitsLars-Peter Clausen: RTC: JZ4740: Use dev_get_drvdata directly instead of wrapping it in dev_to_rtc http://qi-hw.com/p/qi-kernel/9e07a1319:43
qi-commitsLars-Peter Clausen: RTC: JZ4740: Whitespace cleanup http://qi-hw.com/p/qi-kernel/cf1058719:43
rafatuxbrain_away: sorry, I did "find... | grep -i avr" and I did not get any output20:46
larscrafa: did you saw my comment?20:53
mthlarsc: I got the "maximum CPU freq determines sets PLL freq" part working now21:07
mthfor the "desired CPU freq determines CCLK freq" part, I think it's good to use the cpufreq table support21:08
mththis allows gathering of statistics, for example (how long the CPU runs in which frequency)21:08
mthto build that table, I have to take the PLL freq and apply all possible dividers21:09
mthI could do that in cpufreq.c, but that would be duplicating knowledge already available in clock.c21:09
mthhowever, there is no way yet to get all supported rates from a clock21:10
mthare you OK with me adding such a method?21:10
mthin the sense that I'd like it to get merged into qi-kernel at some point in the future21:11
larscsure. go ahead. you know better than me whats needed for cpu freq support21:18
larscand it can be changed later on if it turns out to be a problem anyway ;)21:19
qi-commitsLars-Peter Clausen: MIPS: jz4740: PWM: Fix handling if the pwm clock is not available http://qi-hw.com/p/qi-kernel/3b8a52a21:35
rafalarsc: I think that I should pull and try again right ? ;)21:40
larscit should work now21:40
rafagreat, let me try..21:45
larscI was able to drop the first patches from my for-upstream branch :)21:45
rafalarsc: my big problem is to power off.. sorry to be soo insistent about.. I have not idea what is the problem21:46
larsci don't know either21:46
rafalarsc: now I am trying to build current git, current branch 2.6.34 with config from arch/mips/configs/qi*21:46
rafalarsc: but if that git+branch with that config at arch/mips/config/ works for you to power offf I am sure that it is a problem from my side21:47
larscrafa: I'm compiling a 2.6.34 kernel now to be sure21:53
larscworks fine22:01
rafalarsc: great, I have tried as well, I can boot with rootfs on sd22:02
rafalarsc: could you check power off / reboot from a rootfs in sd?22:02
rafafor me it still does not work :(22:02
larscnope, i can not22:02
larscbut there shouldn't be any difference22:03
larscanyway... the sun starts rising again, time to get some sleep22:05
rafalarsc: cool, thank anyway..22:06
rafalarsc: if I can realize which is the problem I will let you know22:06
rafahi.. somebody can tell me the output of "cat /proc/cmdline" in his/her nn?22:59
rafaI am debuggin poweroff in 2.6.34 and I would like to know the arguments in 2.6.3222:59
rafaxiangfu ? :)22:59
xiangfurafa: wait one second.23:00
rafathanks a lot23:00
xiangfurafa: mem=32M console=tty0 console=ttyS0,57600n8 ubi.mtd=2 rootfstype=ubifs root=ubi0:rootfs rw rootwait23:01
rafaxiangfu: I am building 2.6.34 from qi git, using arch/mips/configs/qi*.. but I can not poweroff the machine from userland23:01
rafaxiangfu: thanks a lot!23:01
rafano luck..23:01
xiangfurafa: do you use command: "poweroff" "halt" ?23:02
rafayes, I have tried those23:02
rafaxiangfu: I have sysvinit installed and busybox.. I have tried poweroff and halt using both.. sysvinit and busybox versions23:02
rafano luck,23:02
rafathose shutdown the machine, but no power off.. it just halt/freezes after if ends the shutdown process23:03
rafawolfspraul: hey, are you coming to colombia it seems right?23:03
rafaif so, cool ! :) I would like to be there to listen you ;)23:04
rafaxiangfu: btw, have you done the qi git for 2.6.32 kernel?23:04
xiangfurafa: yes.23:06
rafaxiangfu: cool, perhaps I need to try it..23:06
wolfspraulrafa: I am already in Colombia since Friday I think23:06
rafaha!.. great :) .. drink coffe.. they have the best on this world ;)23:07
rafaxiangfu: which is the git link?23:08
rafaI do not see it under http://projects.qi-hardware.com/23:08
xiangfurafa: http://projects.qi-hardware.com/index.php/p/qi-kernel/source/changes/master/ the master branch23:08
rafathe master branch is 2.6.32?..23:08
rafaah okey23:09
rafaxiangfu: and which .config file should I use? sorry for the many questions23:11
rafaI do not find it under arch/mips/configs23:11
xiangfurafa: there is ".config"  file in master branch23:12
rafaah.. I just use the current, thanks xiangfu23:12
xiangfurafa: you are welcome23:13
rafaxiangfu: I got an error.. let me paste you here.. perhaps you know which is the probem23:14
rafagen_initramfs_list.sh: Cannot open '/home/xiangfu/workspace/PanGu/openwrt-xburst/build_dir/target-mipsel_uClibc-'23:14
rafamake[1]: *** [usr/initramfs_data.cpio.gz] Error 123:14
rafamake: *** [usr] Error 223:14
rafaIt wants to open some file in your home it seems.23:14
xiangfurafa: oh. I always built the kernel with initramfs. so let me upload a new .config. it's one options of menuconfig23:16
xiangfurafa: sorry23:16
rafaxiangfu: no problem, let me disable initramfs here23:16
qi-commitsXiangfu Liu: remove the build INITRAMFS http://qi-hw.com/p/qi-kernel/4a5edae23:20
rafawolfspraul: so try to use timezone from colombia, no chat NOW!23:20
rafawolfspraul: and again, drink their coffe, this is the best :)23:21
wolfspraulno chat now?23:23
wolfspraulI am working :-)23:23
rafait is time to sleep in colombia now23:23
rafawell, also in argentina.. perhaps I should go to sleep .. too tired these days23:24
rafabut first, I will try xiangfu kernel ;)23:24
rafaxiangfu: building.. it also has rootfs as /dev/ram.. so I disabled kernel command line arguments default as well23:26
xiangfurafa: yes. I just committed it. :)23:32
rafaxiangfu: no luck with my rootfs on sd.. kernel says that it mounts rootfs, but no init found. It is there.. well, I will check tomorrow. thanks for your help23:43
xiangfurafa: this kernel will run "/etc/preinit" as the init.23:48
--- Mon Jun 28 201000:00

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