#qi-hardware IRC log for Monday, 2012-09-10

Guest35674I assume I need to compile gmu to have mp3 rubbish support right?00:13
Guest35674what do I need to do to add mp3 yuck support? recompile gmu,sox,mplayer? or for gmu is there a lib I can get from some were? like the devs 7.2 package?01:23
Guest35674right bed time.01:29
kristianpaulyou can lock terminal with byobu ;)01:37
GCW-NowDoes anyone know of an Ingenic 4770 Vivante GC860 OpenGL driver or the ITE IT6610 if so please contact me gcwnow@gmail.com04:54
xiangfuwpwrak, Hi08:33
LunaVoraxHi10:19
alexander_hi10:27
alexander_its me again driff name10:28
wpwrakxiangfu: heya !11:03
paul_boddieHello! I was looking at USB gadget drivers on the NanoNote, and it looks like you can't use the combination drivers because of limitations on the number of endpoints in the jz4740 code. Is this a hardware thing?11:39
paul_boddieI get -ENODEV when the Ethernet-plus-serial gadget tries to bind the serial endpoint, or something like that.11:40
paul_boddieThis being in drivers/usb/gadget/f_acm.c, if I recall correctly. The kernel log reports "can't bind" and a status of -19 (-ENODEV). Anyway, that's my report on the matter for Internet users of the future. :-)11:48
paul_boddieI also found that the jz4740 code won't compile with debugging enabled because it appears to think DEBUG expands to a function call. I may try and make a patch for that, but debugging practices in the kernel seem to be somewhat chaotic.11:52
lindi-paul_boddie: usb host hardware has limits yes11:58
mthlindi-: the gadget driver uses the USB device controller, not the USB host controller12:02
paul_boddieI found that g_ether and g_serial worked fine on their own. There appear to be other drivers that might allow the USB port to work with different "protocols", but I'm not sure what their limitations are.12:03
mthand I think the UDC driver has more limitations than the hardware does12:03
lindi-mth: ok but that shares the endpoint buffers of the host controller?12:03
mthafaik the two controllers are completely separate12:03
lindi-usually they are on the same chip12:03
lindi-really?12:03
mthnot 100% sure, but I've never read anything about shared buffer space12:05
mthlarsc might know more about this12:05
paul_boddieIn any case, I can live with one or the other, although it doesn't appear possible to build multiple gadgets in the same kernel build and dynamically choose between them, so I suspect that there may be some architectural limitations in the kernel, too. Or maybe they don't anticipate anyone doing that.12:05
mthyou could build the gadgets as modules and load/unload them that way, I guess12:08
mthpaul_boddie: if you know of a way to make cdcether work reliably with Windows on the host, I'd like to hear about it12:09
paul_boddieStill seems a bit weird that you can only choose one in the menuconfig, though, and that you have combination gadgets. Maybe changing the "personality" of endpoints is seen as being bad behaviour.12:09
mthit works out of the box with Linux and Mac OS X, but not with Windows12:09
paul_boddieFortunately, I don't have to use Windows. I remember trying to get IRDA networking running between Linux and Windows, but Microsoft had done their usual protocol extension stuff, and although it was documented that Windows might send bad stuff and that you could work around it, the Linux driver just refused to talk to it by default.12:11
xiangfuwpwrak, Hi. 12:19
xiangfu(network problem)12:19
whitequarkpaul_boddie: you're trying to build g_serial and g_ether as modules and switch between them, right?12:40
larscmth: do the new pwm patches look good to you?12:40
whitequarkI've tried that on a different board with the almost same processor (Jz4725B, has exactly same UDC IP core), and it worked for me12:40
paul_boddiewhitequark: I actually didn't compile them as modules, but it's encouraging to know that you can have them both and switch them in and out on the same kernel.12:42
paul_boddieI just wondered why they'd bother to have combination modules, but I suppose it has something to do with simultaneously advertising both kinds of service.12:43
whitequarkpaul_boddie: ah, simple then12:44
whitequarkwhen you compile both in the kernel, you're stating that you want to use them simultaneously12:44
whitequarkand the jz4740 UDC doesn't have enough endpoints for that12:44
paul_boddieI thought so. Thanks for confirming that!12:45
whitequarkOTOH you might have some success with CDC-ECM (as opposed to CDC Ethernet), as the former is less complex12:45
whitequarkit was developed for exactly this application12:45
whitequarknote that CDC-ECM doesn't work on Windows AFAIK12:45
paul_boddieI was just using the usual CDC Ethernet driver, which is g_ether, I think.12:45
whitequarkyeah12:45
whitequarkit's more complex than ECM one12:45
paul_boddieI'm talking to a Linux host, so I guess I don't have the same interoperability problems as others have.12:48
paul_boddieI'm actually looking to play a bit with the USBIP driver, just for fun.12:49
whitequarkUSBIP? what's that?12:49
paul_boddieUSB over TCP/IP. It's in the staging drivers directory.12:51
paul_boddieThe TCP/IP part seems overkill to me. I may look into using plain file descriptors rather than sockets.12:52
mthlarsc: I didn't do a thorough review, but reading the patches I think he changed everything as we discussed12:53
whitequarkpaul_boddie: erm, so what do you want to do exactly? usb over tcpip over usb?12:53
mthwhitequark: cdcether doesn't work with Windows either, in practice12:54
paul_boddieUSB over something, perhaps not TCP/IP, which would use USB as the underlying transport. Just for fun, as I said. :-)12:54
paul_boddieUSBIP provides a host interface. That's the crucial distinction.12:55
whitequarkmth: ah, maybe. I didn't really work on Windows for 5 years or so12:55
mthI had two friends with Windows netbooks try to connect to the A320 and neither of them succeeded even after trying various things for over an hour12:56
alexander_I am working on a decent cl countdown script. it plays sounds using "aplay /somesound 2>/dev/null &". If a sound is still playing when the next one starts. the next one does not play. any advice? this works on my desktop13:29
larscmake sure the dmix plugin is enabled and used13:30
alexander_were is that set?13:31
larscaplay -D, but in theory it should be the default if it is enabled13:31
alexander_thanks13:32
larscwhat does aplay -L list?13:32
alexander_nope don't work. 13:41
alexander_aplay -L lists one card can post the full info13:42
alexander_default:CARD=LB6013:45
alexander_QI LB6013:45
alexander_I have had this problem before on puppy. I think it was solved by telling it to use alsa not oss13:47
larscaplay will use alsa, I'm not even sure there is oss support on the nanonote13:47
larschave you tried with aplay -D plug:dmix?13:48
alexander_oh13:48
alexander_that did it. thanks and apologizes.13:51
larscIf you want to you can make that default in your .asoundrc. See http://alsa.opensrc.org/Dmix and http://en.qi-hardware.com/wiki/Mpd the later also adds a software volume13:54
alexander_thanks14:06
alexander_one more thing and then I will post it here.14:06
alexander_any recommended fonts for a countdown?14:20
xiangfuHi wpwrak 14:25
wpwrakhi xiangfu !14:30
wpwraknetwork better ?14:30
xiangfuwpwrak, yes. 14:30
xiangfu:)14:30
xiangfuwpwrak, I have problem when I run 'make -C pgm/fw on HOST=root@ben'14:30
xiangfuit make ben crash and reboot.14:30
wpwrakmaybe you're having trouble with inrush current14:31
wpwrakfirst, did you run the "ub" script before the "make ... on" ?14:31
xiangfuwpwrak, I have to manually insmod /lib/modules/3.3.8/at86rf230.ko14:31
xiangfuthen the lsmod output is like: http://pastebin.com/Ms6SjesZ14:31
wpwrakyou don't need at86rf230.ko14:32
wpwrakin fact, the "ub" script tries to turn it off in case it's there :)14:32
wpwrak"ub" tries to turn off at86rf230 and the mmc driver, so that nothing in the kernel uses the 8:10 card slot14:33
xiangfuwhen I run 'echo spi*.0 >unbind' . I got no such device. is that normal?14:33
wpwrakyes, that's okay14:34
xiangfudon't want broken anything... :-)14:34
wpwrakactually .. the "no device" is a bit surprising. it suggests a partial initialization. but okay, if there's no spi under the at86rf230, then that's fine.14:36
wpwrakvery important: when you do the "make ... on", the programmer must be disconnected. only connect it after the "make ... on"14:36
xiangfuyes.14:37
wpwrakthe ben is extremely fragile when it comes to inrush current on the 8:10 card port. even a small external capacitative load can crash it.14:37
xiangfuI do that exactly follow readme.  when 'poke 0x10010318 4' it crash nanonote. 14:37
wpwrakhmm. do you have another ben you could try ? maybe this one is even more sensitive than they usually are14:38
xiangfutrying another nanonote 14:39
wpwraksolved the ATtiny167.conf mystery. the next time you pull, ATtiny167.conf will be deleted. run  make -C common  to get it back14:43
xiangfuanother ben success on 'make -C pgm/fw prog HOST=root@ben'14:45
xiangfuwpwrak, (ATtiny167.conf) got it.14:45
wpwrak(success) excellent ! :)14:46
xiangfuwpwrak, for reflash the bootload. I have to reboot and setup them again right?14:46
xiangfucan I reflash bootload right after reflash programmer?14:46
wpwrakyes, you can do everything in the same session14:47
wpwrakthe ben should be able to stay up very long. the main reasons why you may have to restart it are some short-circuit when inserting a board in the programmer (shouldn't happen very often since the fixture already guides the board) or when power is turned off by accident and you aren't quick enough to turn it back on14:49
xiangfuwpwrak, after reflash. the led blinking very fast. after led walk. is that normal?14:58
xiangfuI cannot see if it display image correct. :-)14:58
xiangfuI cannot figure out if it display image correct. 14:58
wpwrakhmm. after the led walk, the application should be in standby mode and there should be no led activity14:59
wpwrakyou can get some weird behaviour if the power source is weak. try this:15:00
wpwrakconnect usb, make -C fw/app flash15:01
wpwrakleave usb still connected. afther the "make ... flash", the boot loader jumps to the application. pressing the button should then wake things up.15:01
xiangfuwhen I plug usb. 15:03
xiangfuthe 1~2 led on.15:03
xiangfuthen I run the make -C fw/app flash15:03
xiangfuoutput looks normal.15:03
xiangfuthen the device start to walk led. 15:03
xiangfuafter walk led. 15:03
xiangfuhalf of the leds blinking very fast.15:04
wpwrakhmm, that's strange15:04
xiangfuhalf leds I mean:  1, 3, 5, 7, 9 .. is blinking.15:04
wpwrakthat sounds like a power supply issue15:04
xiangfu2, 4, 6, 8,... are off.15:04
wpwrakalso, when the "make ... flash" finishes, the device should go dark15:07
wpwrakdid you pull the latest versions of everything ?15:07
xiangfuyes. I think so.15:10
xiangfulet me try again.15:10
wpwrakalso please make sure the avrdude run really succeeded. avrdude prints a lot of stuff and it's easy to overlook an error. if it says that it verified the efuse ("avrdude: 1 bytes of efuse verified"), then everything is okay15:12
xiangfuthis time seems right.15:13
xiangfulast time I plug in the 'programmer'. none of the leds is light.15:13
xiangfunot. the yellow led is light.15:13
xiangfus/not/now15:14
wpwrakno lights in the programmer would be odd indeed :)15:14
xiangfureflashing..15:15
wpwrakperhaps avrdude didn't even detect the device and never changed the boot loader. that would explain weird behaviour since the boot loader i had flashed originally is not compatible with the current application. (they share some interrupt handling, and that has changed)15:15
xiangfuwpwrak, ok. 15:21
xiangfuyou update the avrdude to 5.11.115:21
xiangfuI am still with 5.10.15:22
xiangfuupdating..15:22
wpwrakyou need the latest versions of all the patches from ben-blinkenlights. there are things the regular avrdude or avrdude 5.10 with older patches can't do. e.g., multiple configuration files15:23
xiangfuwpwrak, README 123. there is a typo.15:28
xiangfufw/app/flash --> fw/app flash15:28
wpwrakfixed. thanks !15:29
alexander_How the heck do you temporary change the console so ones text fills the screen? fonts don't look like the way. what about changing the resolution?15:34
alexander_The aim is full screen countdown.15:34
wpwrakxiangfu: i just noticed that the cap sensor may be unreliable when on battery power. on usb power, it should be fine, though. in case of persistent trouble, you can try leaving off the top of the case and touching the sensor directly.15:37
alexander_https://filetea.me/t1sd06b2 expires on my internet disconnection15:37
alexander_https://filetea.me/t1s9389c15:39
alexander_new url15:39
alexander_its the beginis of a Decent CL countdown bash script15:39
alexander_with sound15:40
qi-bot[commit] Xiangfu: cgminer: update to 2.7.5 (master) http://qi-hw.com/p/openwrt-packages/8e9f39215:43
qi-bot[commit] Xiangfu: avrdude: update to 5.11.1, update ben-blinkenlights patches (master) http://qi-hw.com/p/openwrt-packages/bde99a815:43
qi-bot[commit] Xiangfu: avrdude: add depends libftdi (master) http://qi-hw.com/p/openwrt-packages/bfd847215:43
wpwrakthanks :)15:45
xiangfuwpwrak, great. I can see the smile by waving the device.15:47
wpwrakyeah ! :)15:48
wpwrakimage selection (the 2nd menu item) should also work. so you can pick the vampire or the ufo.15:48
alexander_oh to use  the script # decentclcountdown 00:01:0015:50
alexander_# decentclcountdown 00:01:0515:50
paul_boddiealexander_: What are you using to show the numbers? The console?15:56
alexander_yep15:56
alexander_bash15:56
alexander_printf15:56
paul_boddieIf you want more control, you could use something that accesses the framebuffer. SDL seems to work quite well if the framebuffer is available.15:57
alexander_I know but thats a lot to lean about. I didn't think it would be so hard to have a huge font size. somehow. Its the last hurdle. the rest is easier improvements.15:59
xiangfuwpwrak, it like I tap once. the image change once. 16:00
xiangfuwpwrak, the image selection under 'menu' not working here.16:00
alexander_setfont /user/share/kbd/consolefonts/ter-v32n.psf16:00
paul_boddieYou could use pygame if you don't want to write stuff in a systems programming language. :-)16:00
alexander_gets close16:00
wpwrakxiangfu: if you tap once and the image changes, then you're in the image selection16:01
xiangfuwpwrak, ok. great. 16:01
wpwrakxiangfu: you have, in this order: run, image selection, image speed / width, and run endlessly16:01
wpwrakxiangfu: image speed isn't implemented yet. all the rest should work.16:02
xiangfuwpwrak, ok. got it. trying endless.16:02
xiangfuwpwrak, I can adjust the width by waving slow or fast. :-D 16:04
wpwrakyes, manual speed control ;-)16:05
alexander_I quite like bash maybe on year I'll  dabble in python  but that would be need more days I don't have currently in less I need to dabble for my biz16:05
alexander_on =one16:05
xiangfuwpwrak, how long is the battery life? if I run under endless mode?16:06
wpwraki haven't tried it yet. i would estimate about 5-10 hours. also depends on the image.16:08
paul_boddiealexander_: You can get a lot done in shell script, certainly.16:08
xiangfuwpwrak, got it.16:08
alexander_I know. just go on the puppy linux forums16:09
alexander_pail_boddie: Ding! I should ask on the puppy forum16:11
alexander_posted onto puppy forums17:12
--- Tue Sep 11 201200:00

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