#qi-hardware IRC log for Friday, 2010-09-24

wpwrakroh: they also say something about writing to the USB gadget "if it permits this", but it's not quite clear to me what this means00:00
qi-bot[commit] Xiangfu Liu: add new packages from qipackage http://qi-hw.com/p/openwrt-xburst/f0e56a301:33
SiENcEhey04:14
SiENcEgot my nanonote today and i want to flash the latest dist. where can i get it?04:15
SiENcEis there any other than openwrt? debian?04:15
wpwrakjlime04:30
wpwrakto flash openwrt: http://en.qi-hardware.com/wiki/Updating_Ben_with_usbboot04:33
wpwrakimages: http://downloads.qi-hardware.com/software/images/Ben_NanoNote_2GB_NAND/latest/04:33
wpwraktools: http://projects.qi-hardware.com/index.php/p/xburst-tools/downloads/04:34
wpwrakthat's all :)04:34
qi-bot[commit] Werner Almesberger: Use two faces and show how they overlap (in progress) http://qi-hw.com/p/ben-scans/525e15505:39
qi-bot[commit] Werner Almesberger: The top face can now be shifted. http://qi-hw.com/p/ben-scans/d49f34905:39
qi-bot[commit] Werner Almesberger: When leveling, the center circle is now shown when the pointer approaches it. http://qi-hw.com/p/ben-scans/edb12b605:39
qi-bot[commit] Werner Almesberger: Overlapping now has an OSD, too. http://qi-hw.com/p/ben-scans/85855ec05:39
qi-bot[commit] Werner Almesberger: Various coordinate transform corrections in overlap.c http://qi-hw.com/p/ben-scans/f3e7c1e05:39
qi-bot[commit] Werner Almesberger: Infrastructure for generating POV-Ray output (not useful yet) http://qi-hw.com/p/ben-scans/37e840f05:39
tuxbrain_awaywpwrak: I'm interpreting right the commits if I suppose you are  in the way of generate the full 3d mesh of NN?05:41
wpwraktuxbrain_away: well, of the parts. since there doesn't seem to be anything that's useful for doing it, i'm hacking my own part combiner.05:42
wpwraktuxbrain_away: it's easy to try. just check it out, make, then mkdir .cache; make run05:43
wpwraktuxbrain_away: you'll get the top face and on the side three buttons. you can orient the top face with the scroll wheel. the goal is to have as much blue as possible.05:44
Action: tuxbrain wants to be like you when I grow up05:45
wpwraktuxbrain_away: then click on B and do the same with the bottom face. finally, click on A+B and make them overlap by shifting (scroll wheel inside the circle) or rotating (scroll wheel outside the circle)05:45
wpwraktuxbrain: then press Q to exit. after that, make pov to render the scene in POV-Ray. lots of things don't quite work the way they should. e.g., the coordinate transforms in A+B aren't completely correct (the Z correction is missing) and the way user input gets applied is a bit counter-intuitive.05:47
wpwraktuxbrain: the pov-ray stuff is worse. right now, i just dump the data without actually aligning, scaling, etc., anything. so the result just tells me that i got the height field right, but the rest is a big no-op05:48
tuxbraincloning git05:48
wpwraktuxbrain: the trick is to never grow up ;-)05:48
wpwrakhmm, if there's a way to clone only a subset of a git, that may be useful. else, you'll also clone all the 3D scans. they're biggish.05:49
tuxbrainabout 250Mb of data... wow all is 3d info?05:49
wpwrakyup05:50
wpwraklots of data and a tiny bit of code :)05:50
tuxbrainin which dir is the makefile?05:50
wpwrakin solidify/05:51
tuxbrainmaking05:51
wpwrakyou have a very quick connection :)05:52
tuxbrainheheehh05:52
tuxbrainjust downloaded at 1'6Mb :P05:53
tuxbrainhey I have an colorized thing in the screen cool!!!05:53
wpwrakwhen you move the mouse around, you'll see a yellow circle05:54
tuxbrainyes05:54
wpwrakmouse wheel inside the circle raises or lowers the z0 plane. where the plane intersects with the scan, you have blue color. anything below z0 is red. anything above green.05:54
wpwrakthe steps at z0+1 and z0-1 are darker green and darker red, to make it easier to see where the plane goes.05:55
wpwrakmouse wheel outside the circle tilts the z0 plane.05:56
wpwrakthe goal is to have as much blue as possible05:56
wpwrak"A" is difficult, because the piece seems to be a bit uneven. "B" is easier.05:57
wpwrakthen, at A+B, you can rotate (mouse wheel outside the circle) and shift (mouse wheel inside the circle, in one of the four quadrants shown)05:58
tuxbrainman this is impresive05:59
wpwraki'll have to add some more visual aids in A and B. it's too easy to get lost there.06:00
tuxbrainI barelly understand a fuck, but is like having a magnetic resonance  on NN06:00
wpwrakyeah, looks like x-ray images ;-)06:00
tuxbrainoh it doesn't scale full screen :P06:01
wpwraknaw, it uses the size of the scan :) each pixel is a scan point06:02
tuxbrainnow I can read wpwrak thoughts from here... (damn lusers...)06:02
wpwrak;-))06:02
tuxbrainwpwrak: make sense :)06:02
kristianpaulSiENcE: jlime is great i use it too :)06:47
qwebirc70375i am trying to compile a simple hello world C code06:47
qwebirc70375plz help me with Makefile for it06:48
SiENcEok. i'm going to test jlime. where to get the latest one?06:48
kristianpaulqwebirc70375: http://developer.ubicom.com/wiki/index.php/Add_a_new_application_to_openWRT06:49
kristianpaulSiENcE: http://www.jlime.com/wiki/documentation/user/nanonote/installation06:49
kristianpaulSiENcE: http://www.jlime.com/wiki/downloads06:50
qwebirc70375kristianpaul : if we go with this procedure then we will have to flush the kernel again and agin if some minor change is there in the application06:52
qwebirc70375cant we compile it seperately and port the bin file via storage card and then execute it06:52
kristianpaulqwebirc70375: hmm not flush the kernel i think, you can just build the ipk look here http://en.qi-hardware.com/wiki/Building_Software_Image#A_short_guide06:54
SiENcEkristianpaul, thanks06:55
kristianpaulwpwrak: even when you ran the make all you got packages in the bin folder i think, so no need to reflash thre whole thing either06:55
kristianpaulwpwrak: sorry06:55
kristianpaulqwebirc70375: ^06:55
qwebirc70375yes06:56
kristianpaulis kay for you that way?06:56
kristianpaulkay/okay06:56
kristianpaulqwebirc70375: http://nbd.name/openwrt.html#x1-460002.1.206:57
kristianpaulmore info06:57
qwebirc70375have you tried this at ur end06:57
kristianpaulnope :p07:00
kristianpauli dint get in to openwrt packing07:00
kristianpauli used jlime last time i wanted compile a simple app07:00
qwebirc70375i have tried the Makefile lying on Nanonote official site but getting error : cannot find -lSDL07:02
qwebirc70375what does it mean07:02
kristianpaulthat  cannot find -lSDL/lib SDL ?07:02
kristianpaulwpwrak: ah yes id just listing ID (i dint check the las table) i tought it was related with usb host or client stack..07:03
kristianpauls/id/it07:04
wpwrakkristianpaul: from the sound of it, it acts as a host, no ? then it need a fair bit of the stack to be able to get at the IDs.07:05
wpwrakkristianpaul: from the description, it isn't quite clear to me if they've realized yet just how much work it is :)07:06
kristianpaulwpwrak: yes host it is, but as you said not quite clear indeed :)07:07
kristianpaullot work indeed07:08
qwebirc91287I am able to execute hello world C program on nanonote successfully07:17
qwebirc91287from where I can have usbbot command07:46
qwebirc91287I want to reflash the kernel07:46
xiangfuqwebirc91287: http://en.qi-hardware.com/wiki/Xburst-tools#usbboot08:13
qwebirc91287i am getting error while reflashing kernel : cant read bulk data from Ingenic device08:16
xiangfuqwebirc91287: make sure there is no USB-HUB. and use the usb-cable that send with NanoNOte.08:24
SiENcExiangfu, why the usb cable send with nanonote?08:39
wpwrakthat seems to be another myth that ought to die. i don't think the nanonote cable is *that* superior to the average usb cable bought within the last years.08:57
xiangfuI should say use better usb cable :)09:00
wpwrakxiangfu: or "don't use the usb cable grandma knitted herself" ;-)09:01
qi-bot[commit] Xiangfu Liu: change the INITRAMFS to LZMA compression http://qi-hw.com/p/openwrt-xburst/49182fa09:07
xiangfuwpwrak: :)09:08
kyakhm10:58
kyakbad news is that even "Hello World" of ncursesw is giving bad output10:59
kyaktherefore, some problem with ncursesw10:59
kyak"Bad output" means it output crap instead of utf-810:59
qi-bot[commit] Carlos Camargo: Adding sram_gpio_QT example http://qi-hw.com/p/nn-usb-fpga/83f727f11:29
qi-bot[commit] Carlos Camargo: Adding qmake.conf sample file http://qi-hw.com/p/nn-usb-fpga/545953211:57
larscAyla: have you managed to get rtc wakeup working?14:16
Aylanope14:18
AylaI'll work on that tonight14:19
Aylait seems that the interrupt is never called, so I guess it is masked somewhere14:19
larschm, ok.14:23
larscyou should check /proc/interrupts to be sure that it really has not fired14:24
Ayla 23:          0            INTC  jz4740-rtc14:27
Aylano interrupt at all14:27
kristianpaulxbust have a sleep instruction or it go to low power mode?14:29
larscsleep14:38
larsckristianpaul: well it has both. there are two different sleep modes14:43
larscso you can change the semantics of the sleep intstruction by writing to a mmio register14:43
larscone is the normal sleep behaviour known from mips. the cpu will stop executing instructions until the next interrupt14:44
larscthe other is going to low power state until the next interrupt happens14:44
larscthe first one is used when the cpu is idle, the second one is used when we send the system to suspend14:48
Aylalarsc: the driver currently does not implement irq_set_state15:16
Aylais that a problem?15:16
Ayla(ah no, it seems it's only for the periodic alarm)15:18
kristianpaullarsc: so interrupt can be trigger by a counter or a internal clock?15:40
kristianpauljust thinking how setup an alarm clock15:41
wpwrakkristianpaul: counters often aren't powered in deep sleep states (not sure without looking how it is on xburst, though)15:47
larscwell at least the rtc clock is of course running15:48
kristianpaulhmm15:48
kristianpaulok15:48
kristianpaulso how the nanonote will wakeup?15:49
larscand i think you can also keep the other SoC timers running15:49
kristianpaulkeyboard pressing i guess15:49
kristianpaulbut.15:49
kristianpaulan alarm clock is so nice :)15:49
Aylalarsc: hmm15:49
Aylajz4740_rtc_ctrl_set_bits: mask=0x4, set=1.15:50
Aylaare you sure that this function actually works? :)15:50
larscno15:50
Aylais it possible that the interrupt is activated on the RTC subsystem, but not on the AIC?15:54
larscdon't think so15:58
Aylathat's funny, if I type "jz4740 RTC" on google I get your name everywhere :)16:05
Aylalarsc: could you give me a hint?16:25
larscAyla: do you have the jz4740 ds?16:26
AylaJZ_RTC_CTRL_AE bit is enabled in rtc_set_alarm,16:26
AylaI have a PDF document, but not really useful16:26
AylaJZ_RTC_CTRL_AF_IRQ is enabled in rtc_alarm_irq_enable,16:27
Aylabut JZ_RTC_CTRL_AF is not enabled at all16:27
Aylaeven if it is disabled in the IRQ handler16:28
Aylaor maybe I just can't see it... x_x16:28
larscAF is set by the hardware16:31
Aylaok16:34
Aylalarsc: no it's not16:38
larscAyla: but it should be16:39
Aylabits 3,4,5 are not set on the first call of rtc_ctrl_set_bits16:39
Ayla(before applying flags)16:39
Aylabut maybe they're cleared before that16:39
Aylabut even if JZ_RTC_CTRL_AF is set by the hardware,16:43
Aylathe bit is cleared on the interrupt16:43
Aylaso there can be only one interrupt16:43
larscyes. there should only be one interrupt16:48
larscotherwise the handler will be called in a constant loop16:49
Aylashouldn't that flag be set on rtc_alarm_irq_enable ?16:49
Aylawhen I say there can be only one interrupt, I mean that it's not possible to reprogram the alarm for a new interrupt16:49
larscwhy not?16:52
Aylabecause the interrupt handler desactivates the AF bit16:53
larscthe AF flag indicates whether a interrupt is pending16:53
Aylayes, I just read that....16:54
Aylalarsc: another idea,17:27
Aylayou do not call wait_write_ready on rtc_reg_read17:28
Aylabut on the doc it's written "The read17:28
Aylavalue from the target register is also undefined.17:28
Ayla"17:28
Ayla(when the bit is 0)17:28
qi-bot[commit] Werner Almesberger: Instead of performing the tranformations stepwise for each point, pre-calculate http://qi-hw.com/p/ben-scans/898970b17:29
qi-bot[commit] Werner Almesberger: More corrections to handling of the "user" matrix. Made controls more intuitive. http://qi-hw.com/p/ben-scans/7c2478317:29
qi-bot[commit] Werner Almesberger: solidify/main.pov: don't use weird "x is up" perspective http://qi-hw.com/p/ben-scans/018695917:29
qi-bot[commit] Werner Almesberger: Increase dynamic range of rotation and follow "far from center = faster" http://qi-hw.com/p/ben-scans/276415517:29
qi-bot[commit] Werner Almesberger: Some simple optimizations increase drawing speed in overlap by 33%. http://qi-hw.com/p/ben-scans/d4a557517:29
qi-bot[commit] Werner Almesberger: Bottom face was flipped in overlap. http://qi-hw.com/p/ben-scans/9c614db17:29
qi-bot[commit] Werner Almesberger: Like rotations, shifts can now be accelerated by changing the mouse position. http://qi-hw.com/p/ben-scans/a7105ad17:29
Aylawhen I write to the /sys/class/rtc/rtc0/wakealarm file, it does not seem to set the IRQ flag17:31
Aylathe alarm_irq_enable function is never called...17:38
Aylaoh!17:39
Aylartc-dev.c: "RTC_WKALM_SET eliminates the need for a separate RTC_AIE_ON call"17:40
Aylawhich enable the alarm interrupt17:40
Aylawhich means that RTC_WKALM_SET should actually set the interrupt17:41
Aylawhat it does, is call the rtc_set_alarm() function17:41
Aylaunfortunately,17:41
Aylajz4740's version of that function only modify the RTC_SEC_ALARM and the RTC_CTRL_AE flags17:42
Aylait does not set the RTC_CTRL_AF_IRQ flag17:42
larscok.17:51
larscwill you fix it?17:51
Aylayes17:52
larscgood :)17:52
AylaI just hope that's the reason why it does not work :)17:52
wpwrakAyla: if not, you get to enjoy the adventure a bit longer. a win-win situation ;-)17:56
Ayla 23:         21            INTC  jz4740-rtc17:56
Aylawhohooo17:56
AylaI get interrupts :p17:56
qi-bot[commit] kyak: mupdf: added libjpeg dependency as noted by Xiangu http://qi-hw.com/p/openwrt-packages/0e186f017:56
Aylaone every second17:56
Aylaah, yeah17:58
AylaI took the wrong interrupt xD17:58
Aylatook the 1Hz one17:59
qi-bot[commit] kyak: stupid workaround to make links compile http://qi-hw.com/p/openwrt-packages/7e5961a17:59
Aylaokay, I get alarm interrupts...18:09
Aylabut that does not wake the dingoo18:10
mthit is posisble to suspend and resume individual drivers instead of the entire system?18:14
wpwrakPh'nglui mglw'nafh dingoo Ayla wgah'nagl fhtagn, in Ayla's house, dead dingoo waits dreaming18:14
tuxbrainwpwrak: LOL18:15
Aylahere's a patch: http://pastebin.com/UhBLc2LB18:16
mthassuming you have no serial connection, there is not really much of a difference between "not waking up" and "crashes on wakeup"18:16
Ayladoes it crash on wakeup?18:17
Aylamaybe it crash on suspend18:17
mthalso possible18:19
mthmaybe this is useful: Documentation/power/basic-pm-debugging.txt18:21
mthvia /sys/power/pm_test you can specify how deep the kernel is allowed to sleep18:23
mthinstead of doing a full suspend18:23
Aylaokay18:28
AylaI'll check that tomorrow18:28
qi-bot[commit] Werner Almesberger: The #endif at end of header file must have a comment saying /* !name */ http://qi-hw.com/p/ben-scans/0c4940620:36
qi-bot[commit] Werner Almesberger: Step sizes are now auto-determined, allowing use of files of any resolution. http://qi-hw.com/p/ben-scans/2cc9a9e20:36
qi-bot[commit] Werner Almesberger: solidify now stores the context of sessions in project description files. http://qi-hw.com/p/ben-scans/f6ed3bf20:36
qi-bot[commit] Werner Almesberger: Use the project name to disambiguate names in POV-Ray output. http://qi-hw.com/p/ben-scans/01d8e4120:36
--- Sat Sep 25 201000:00

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