#milkymist IRC log for Friday, 2011-12-02

sh4rm4<wpwrak> hmm, "masala" must mean something naughty in india00:25
sh4rm4it means "mix"00:25
sh4rm4i.e. gharam masala = hot mix (spice mix)00:26
wpwrakmaybe with some second meaning ? like in english, a baffling number of words can be used synonymously to "having sex"00:27
sh4rm4dunno, but maybe the porn site owners had something "interracial" in mind00:28
wpwrake.g., do, have, take, jump, ...00:28
wpwrakah yes, that could be too. "mix" would work well then.00:29
wolfspraulaw_: good morning!01:53
aw_wolfspraul, hi good morning01:54
aw_xiangfu, do we have a quick and easy way to re-build/make if I changed following those two files?02:41
aw_one: https://github.com/milkymist/milkymist/blob/master/software/libhal/vin.c#L12802:42
aw_the other: https://github.com/milkymist/autotest-m1/blob/master/src/tests_videoin.c02:43
aw_and in the end I want a image of test bin?02:43
aw_since i can't build them now. :)02:44
xiangfuaw_, all you need do is type make under 'autotest-m1/' folder.02:44
xiangfuaw_, do you have 'lm32-rtems4.11-gcc' ?02:45
xiangfulet's do this now. :)02:45
aw_but if i modify some content in  vin.c file, so will i still need just make under 'autotest-m1/'?02:46
xiangfuaw_, if you only modify the vin.c file. you needs 'make clean all' under 'autotest-m1/'02:47
xiangfuaw_, like: MMDIR=../../milkymist.git/ make clean all02:48
xiangfuMMDIR is the path of milkymist.git02:48
xiangfubut first you need the compiler (lm-rtems4.11-gcc) in your system.02:49
aw_second...i'm checking my 'milkymist' folder...02:51
kristianpaulwpwrak: had you seen this http://www.circuitsathome.com/mcu/usb-host-shield-library-version-2-0-released ?02:52
aw_xiangfu, alright. now how to check if my system has (lm-rtems4.11-gcc) ?02:53
xiangfuthere should be no lm-rtems4.11-gcc in your system. let's check. just type lm-rtems4.11-gcc. see what it output.02:54
aw_command not found02:54
rohkristianpaul: they use an spi usb controller02:55
roh'cheaters' ;)02:56
xiangfuaw_, yes. download this file : http://fidelio.qi-hardware.com/~xiangfu/build-milkymist/milkymist-firmware-12012011-1513/Flickernoise-lm32-rtems-4.11-SDK-for-Linux-x86_64.tar.bz202:59
aw_xiangfu, wow...big03:03
aw_xiangfu, extracted done.03:11
xiangfuyou have to move the folder (rtems-4.11) to /opt/03:11
xiangfuaw_, then add those lines to the end of ~/.bashrc : http://dpaste.com/664664/03:14
xiangfuhttp://dpaste.com/664664/ line:5 change to your own milkymist.git patch.03:15
aw_xiangfu, where's ~/.bashrc ? under /opt ? or /opt/rtems-4.11 ?03:20
xiangfuaw_, just 'gedit ~/.bashrc'03:21
xiangfu~ means your HOME folder.03:21
aw_like this? adam@adam-laptop:/home$03:23
xiangfulike /home/adam03:23
xiangfu~ == /home/adam03:24
wolfspraulok, so i understand the rc4 reset_ic and usb power switch design verification looks good so far03:24
wolfsprauland the adv7181c is being tested right now03:25
aw_wolfspraul, oah~yes. but I have to learn make on my site firstly then to check registers's setting on how difference between 7181C and 7181B. phew~ long way for me. ha~ :-)03:26
wolfspraulif you believe the soldering is ok, maybe we should just send out one board to Werner?03:30
wolfspraulto speed up...03:30
wolfspraullet's see how it goes today first03:30
wolfspraulXiangfu is going to a milkymist event for the rest of the day, so his availability may be spotty03:31
aw_xiangfu, like this? http://dpaste.com/664683/03:31
wolfspraulmaybe lekernel comes back later and can give you some binaries to speed up, don't know...03:31
xiangfuaw_, yes. correct.03:31
wolfspraulor we just send out the board to werner and take the risk that it has hardware flaws?03:31
wolfspraulso we have multiple options, let's see how it goes today first03:32
wolfspraulthanks for the good work!03:32
xiangfuaw_, add one more line to the end of that file: 'export MMDIR=/home/adam/milkymist/milkymist.git'03:32
aw_wolfspraul, don't know. maybe register settings needs to be changed a lot or few. I don't think that I can quickly digging into this.03:32
aw_but good is that phew~ learn something new for me (always stupid in s/w)... :(03:33
xiangfuaw_, after add that line. open a new terminal.03:34
aw_wolfspraul, seems my first pre-rc4 board is almost soldering well.03:34
wolfspraulaw_: just work on it step by step, concentrated and it will be fine03:34
wolfspraulI'm just preparing you that xiangfu will be mostly offline or reply delayed for the rest of the day03:35
wolfspraulafter 12.30 or so03:35
wolfspraulmaybe lekernel can quickly make some test binaries for you and accelerate the verification03:35
aw_except there's some pins of video decoder we still don't know if it works.03:35
wolfspraulor we send out boards without verification03:35
aw_wolfspraul, oah..yes03:36
wolfspraullots of options03:36
aw_xiangfu, done ... then ?03:39
xiangfuyes. I will go to a event. there is a big device call 'Media Wall' , when it closed. it's just a big box. when it's open. there are screen. speakers. DMX LED light, DMX laser, etc.03:39
xiangfuaw_, yes. open a new terminal. and goto the autotest-m1.git/ folder. run 'git pull' first. since I just fix the libs DEPENDS.03:39
aw_aha...okay...you go first.03:39
xiangfuaw_, I will leave at ~12:30. so we can finish the compile at lease. :)03:40
xiangfuaw_, after run 'git pull' , goto the autotest-m1.git/src/ folder. just run 'make' you will get a new test bin that include the vin.c's modify. :)03:41
xiangfu'make' ==> will compile boot.bin for you. include all libs modify. like vin.c03:43
xiangfu'make load' ==> will run 'flterm --port /dev/ttyUSB0 --kernel boot.bin' for you. :)03:44
aw_phew~ check where my autotest-m1.git/ folder, it seems that i git cloned before. damn it. :)03:45
aw_does the autotest-m1.git under /milkymist folder?03:48
xiangfuaw_, depends on you. :)03:49
xiangfuaw_, I just put them under milkymist folder.03:49
aw_xiangfu, okay..git cloned it under milkymist folder03:55
xiangfuyes. goto autotest-m1.git/src. type 'make' then there will be 'boot.bin'03:55
aw_adam@adam-laptop:~/milkymist/autotest-m1/src$ make03:56
aw_Makefile:1: /software/include.mak: No such file or directory03:56
aw_make: *** No rule to make target `/software/include.mak'.  Stop.03:56
xiangfuaw_, add another line : export MMDIR=/home/adam/milkymist/milkymist.git to your  ~/.bashrc03:57
xiangfuaw_, then open a new terminal. try again.03:57
aw_xiangfu, i added: http://dpaste.com/664695/04:00
aw_did I miss another sub-folder /milkymist.git/ ?04:01
aw_my cd is 'adam@adam-laptop:~/milkymist/autotest-m1/src$'04:02
xiangfuaw_, can you close the terminal and open a new one. try adain.04:02
kristianpaulxiangfu: you may have already downloaded ISE 13.3 in fidelio?04:02
xiangfukristianpaul, no.04:02
kristianpaulmorning btw !! :-)04:02
xiangfustill 13.204:02
xiangfukristianpaul, good morning.04:03
aw_yes. i reopened:04:03
aw_adam@adam-laptop:~/milkymist/autotest-m1/src$ make04:03
aw_Makefile:1: /home/adam/milkymist/milkymist.git/software/include.mak: No such file or directory04:03
aw_make: *** No rule to make target `/home/adam/milkymist/milkymist.git/software/include.mak'.  Stop.04:03
aw_kristianpaul, good morning. :)04:03
xiangfuaw_, ok. what is '/home/adam/milkymist/milkymist.git/' is that the correct path.04:03
xiangfuaw_, what is the output of 'ls /home/adam/milkymist/' and 'ls /home/adam/milkymist/milkymist.git'?04:04
kristianpaularghhh, 5GB!!! :-|04:05
kristianpaulokay good night !04:05
aw_xiangfu, 'ls /home/adam/milkymist/' : http://dpaste.com/664698/04:05
xiangfuaw_, ok.04:06
xiangfuaw_, try to do this:04:07
aw_xiangfu, 'ls /home/adam/milkymist/milkymist.git'  : http://dpaste.com/664699/04:07
xiangfu1. mv /home/adam/milkymist/autotest-m1    /home/adam04:07
xiangfu2. mv /home/adam/milkymist/  /home/adam/milkymist.git04:07
xiangfu2. mkdir /home/adam/milkymist/04:08
xiangfu4. mv /home/adam/milkymist.git  /home/adam/autotest-m1 /home/adam/milkymist/04:08
xiangfuand cd /home/adam/milkymist/autotest-m1/src. try again. :D04:08
aw_adam@adam-laptop:~/milkymist/autotest-m1/src$ make04:11
aw_lm32-rtems4.11-gcc -I/home/adam/milkymist/milkymist.git/software/include -I/home/adam/milkymist/milkymist.git/software/include/base -I/home/adam/milkymist/milkymist.git/tools -nostdinc   -c -o crt0.o crt0.S04:11
CTCP : from aw_ (aw_!adamwang@host-222.80-43-115.dynamic.totalbb.net.tw) to #milkymist04:11
aw_/opt/rtems-4.11/bin/lm32-rtems4.11-gcc: 1: ELF: not found04:11
aw_/opt/rtems-4.11/bin/lm32-rtems4.11-gcc: 2: Syntax error: ")" unexpected04:11
aw_make: *** [crt0.o] Error 204:11
aw_xiangfu, i just git pull: adam@adam-laptop:~/milkymist/milkymist.git$ git pull04:14
xiangfuaw_, yes.04:15
aw_else i missed?04:15
aw_if you need go 'Media Wall' just go, we check this later. ;-)04:16
wpwrakaw_: seems that you have a 32 bit machine while the pre-compiled SDK is for 64 bit. uname -m is reporting something like "i686", right ? (and not "x86_64")04:51
aw_hi, yes xiangfu also found my system is "i686'. :(04:52
aw_ he just helped me 'remotely' to compile new one. phew~04:53
wpwrakhm,, and i can't see any 32 bit SDK in his directories. i thought he once had made one04:53
wpwrakoh, wow :)04:54
aw_just still got err 2. http://downloads.qi-hardware.com/people/adam/m1/tmp/Error_204:56
aw_he could help me after he's back... well... meanwhile I could do a check on datasheet though. :)04:58
wpwrakapt-get install texinfo  ?05:00
wpwrakthere may be a number of things you'll have to install if you're doing this kind of build for the first time. but let's see.05:01
aw_downloading ...05:02
aw_done.... so keep 'make' under 'adam@adam-laptop:~/milkymist/scripts.git/compile-lm32-rtems$ ' folder ?05:03
aw_he told me this will need 1 hr. :)05:03
wpwrak(make) yeah, just retry the last command05:04
aw_make ... again now.05:04
wpwrak(1 hour) depends a lot on your system ;-) what does   cat /proc/cpuinfo   say ?05:05
wpwrakdual-core, good. if you're using "make" without options now, you could try  make -j2  to use both cores instead05:08
aw_wow... do you next time i can always use 'make -j2' ?05:11
aw_do you mean ... :)05:11
wpwrakyes :)05:11
aw_great info.05:12
wpwraknote that, if a build fails, parallel make (make -j2, ...) may produce more confusing error messages, because the command that fails may not be the last command05:12
wpwrakso if this happens, you can just do a  make  (without -j), and you'll get the regular error. that "make" may either come back quickly or take a while. depends on how the thing is structured.05:14
wpwrakof course, if you just re-make things that are known to work or that have only small changes, there's no problem05:14
aw_alright... my goal today is try to learn this firstly then see how difference between 7181B & C video decoder, if i see there's obvious/clear register settings then i modify files myself and make under 'autotest-m1.git' folder. phew~05:18
aw_at least hope to see capture from video-in source. :-) don't know ...05:19
aw_hmm...thinking(idea) and doing(s/w) are super different thing to me. :)05:22
wpwraktakes a bit of time to become a master. but you're on a good path :)05:27
aw_thank you sir. :)05:29
wpwrakhmm, you're upgrading from 7181B to 7181C or from 7125 to 7181C ?05:34
aw_from 7181B to 7181C05:36
aw_make is done: "touch .install.gcc.ok" ...05:37
wpwrakhmm, my m1rc3 has a 712505:39
aw_no...the 7125 is a encoder from digital r,g,b into analog r/g/b :)05:39
wpwrakaah, wrong end. i see :-)05:40
aw_7125 is U18, i replaced 7181C instead on U21 footprint for video con (g/b/r)05:40
wpwrakthey seems to have changed the multiplexer05:41
aw_i got err 2: http://dpaste.com/664764/ when typed 'make' under '~/milkymist/autotest-m1.git/src$'05:42
aw_how do think that err 2 ?05:43
wpwrak*hmm* :)05:43
aw_'avr-gcc: Command not found' ?05:44
aw_what's that 'avr-gcc' ?05:44
wpwraklet's see how demanding that build process is ...05:44
aw_you mean you check your site ?05:45
wpwrakyou could try  apt-get install gcc-avr05:46
aw_do you mean that to see "Makefile" always to see its build process ?05:46
wpwrakif that doesn't help (you may get now, weird errors indead), then you'd have to build avr-gcc from sources05:46
wpwrakthe instructions for building a somewhat recent version of avr-gcc would be here: http://projects.qi-hardware.com/index.php/p/ben-wpan/source/tree/master/atusb/fw/README05:49
aw_installed .. gcc-avr05:49
wpwrakmaybe doa   avr-gcc --version ? (just the first line)05:50
aw_solved ... then another:05:50
wpwrakvictory ! :)05:50
aw_lm32-rtems4.11-ld: cannot find -lmath05:51
aw_make: *** [boot.elf] Error 105:51
wpwrakwow. that's weird05:51
wpwraki dont't even know that that library is. i can see a libmath for C++, one for Perl, ... and then libmathlib2-dev - CERNLIB data analysis suite05:55
wpwrakbut none of this sound quite right05:55
wpwrakwhat were you trying to build when you got the -lmath error ?05:57
aw_phew ~ just googled ... don't know hehe ...05:57
aw_after installed gcc-avr then go to '~/milkymist/autotest-m1.git/src$' folder and 'make'05:58
wpwrakworks here ;-)06:00
wpwrakaha. libmath is a milkymist.git-ishm06:00
wpwrakmake, in milkymist.git,  make -C software/libmath06:01
wpwrakthen try finishing the autotest build again06:02
wpwrakwhile you're at it, also06:02
wpwrakmake -C software/libhal06:02
wpwrakmake -C software/libfpvm06:04
wpwrak(i hope the last one is right. there are two ways of building libfpvm and i'm not sure if they conflict or not)06:04
aw_followed them above. second ...06:05
aw_chmod -x boot.elf06:05
aw_lm32-rtems4.11-objcopy -j .text -j .data -j .rodata -O binary boot.elf boot.bin06:05
aw_chmod -x boot.bin06:05
aw_make -C /home/adam/milkymist/milkymist.git/tools06:05
aw_make[1]: Entering directory `/home/adam/milkymist/milkymist.git/tools'06:05
aw_make[1]: Nothing to be done for `all'.06:05
aw_make[1]: Leaving directory `/home/adam/milkymist/milkymist.git/tools'06:05
aw_/home/adam/milkymist/milkymist.git/tools/mkmmimg boot.bin06:05
aw_CRC32 (boot.bin): 8f26268406:05
aw_does this mean it's done on making (boot.bin) ?06:05
wpwraki have no clue what yuo just did ;-)06:06
wpwrakdid you run   make -C software/libmath    ?06:06
wpwrakor just "make" ?06:06
aw_oah..yes i did your instructions.06:07
aw_make -C software/libmath under milkymist.git06:07
aw_then finished 'make' under '~/milkymist/autotest-m1.git/src$' then shows up 'CRC32 (boot.bin): ....06:08
wpwrakvery strange. that should just build libmath. there's nothing for it to do in tools/06:08
wpwrakaah !!!06:09
wpwraknow i get it :)06:09
aw_you get what?06:09
wpwraki understood what you're doing. i was missing the last step06:09
wpwrakit's kinda weird that autotest-m1 would go over to milkymist.git and build things there. that confused me06:10
aw_so can i say that a 'newer' (boot.bin) is built already and now every time that after I modify files under '/autotest-m1.git/src/' folder then i just type 'make' ?06:10
wpwrakokay, seems that you're good now. now you try to run the critter :)06:11
wpwrakyes, i think that should work06:11
aw_hehe...critter... okay..thanks.. now go back to see this file : https://github.com/milkymist/milkymist/blob/master/software/libhal/vin.c#L12806:12
aw_phew ~ i need to see each of them, but seems lekernel told me those three:06:13
aw_static const char vreg_addr[] = {06:13
aw_0x1d, 0xc3, 0xc406:13
aw_<lekernel> };06:13
aw_hehe :)06:13
wpwrakthat would be crystal and multiplexer06:14
wpwraklovely. there's no complete data sheet06:19
wpwrakthe 7181B still had one06:20
aw_second... in another place. :)06:21
wpwrakbah. why simple if it can be awkward ? ;-)06:23
wpwrakthanks ! that looks better. actually there's too much information now ;-))06:23
wpwraki think you want to change the 0x05 in vin.c to 0x06 (variable vreg_dat)06:30
wpwrakthen rebuild libhal, then rebuild the autotest06:30
aw_I put its link recorded here: http://en.qi-hardware.com/wiki/Milkymist_One_RC3_Known_Issues#switch_to_ADV7181C06:30
wpwraki think i should teach dsv a few more tricks ... :)06:31
aw_will back to see 0x06 ...06:31
aw_firstly i need to run that new boot.bin to see if it can detect 0x19 ID(7181C) :-)06:32
wpwrakoh, instead of changing vin.c you may also be able to get the video signal simply by plugging the camera into the middle connector06:38
aw_the new boot.bin can't show menu items that i can pick. strange ! http://dpaste.com/664775/06:38
aw_let's see if xiangfu did something in this repository ?06:38
wpwrakmaybe you now have a SOC incompatibility06:39
aw_oah ~ he committed two hours ago: https://github.com/milkymist/autotest-m1/commit/d41c1c9a5fd41735fbb6062ad31461666828fbed06:39
aw_i have to compile MMDIR/tools before run flterm. phew ~06:40
aw_what does this commit mean ?06:45
wpwrak(commit) perhaps some minor dependency. nothing really important06:45
aw_aha ... you did them for me. ?06:46
wpwrakyuo need the new soc, the new bios, and then load boot.bin via flterm06:46
wpwrakyes. these are the files i currently use06:46
aw_just copy your files into my current folder.06:46
aw_so your boot.bin is for adv7181c and should detect id: 0x19 ?06:47
wpwrakm1nor soc.fpg && m1nor bios.bin &&  flterm --port /dev/ttyUSB0 --kernel boot.bin06:47
wpwrakit should accept the 0x1906:47
aw_wow..wow...super script shows up now ...06:48
wpwrakit doesn't set the mux right, though. but let's try my middle connector hypothesis first06:48
wpwrakm1nor is all about keeping things simple ;-)06:48
aw_let's see where your m1nor is .06:49
aw_good i found it06:50
wpwrakyou're quicker than me :)06:50
wpwrakyup, that's the one06:51
wpwraksince yuo're getting the latest version, you can abbreviate the above command sequence to06:51
wpwrak m1nor soc.fpg bios.bin &&  flterm --port /dev/ttyUSB0 --kernel boot.bin06:52
wpwrakwhen things are running, connect the camera to the middle connector, then do the VGA test (d), then the video input test (h)06:58
wpwrakand with a little luck, you'll see an image :)06:58
aw_block 67 unlocked07:07
aw_erasing block 67: 007:07
aw_addr: 0x0087246207:07
aw_verify skipped07:07
aw_Flashed soc.fpg at offset 0x6e0000 using /home/adam/.qi/milkymist/latest/fjmem.bit07:07
aw_Flashed bios.bin at offset 0x86000007:07
aw_[FLTERM] Starting...07:07
aw_after above msg, i pressed middle btn but there's no items I can choose. :(07:08
wpwrakyou mean the bios doesn't start ?07:09
aw_yes... it must show somethings log out ...07:10
aw_or maybe I use the old method to load your boot.bin07:10
wpwrakwhat's "the old method" ?07:11
aw_just ' flterm --port /dev/ttyUSB0 --kernel boot.bin' but after power-cycle07:12
wpwraklet's go step by step. after you press the middle button, does the BIOS print anything on the serial console ? "libHPDMC SDRAM initialization runtime" or such ?07:12
wpwrakpower cycle is always a good idea :)07:12
wpwrakand yes, flterm ... is still how you'd load07:12
wpwrakthe m1nor was just for updating the NOR. you don't need to do this each time07:13
aw_no show up "libHP....." just blank ...07:13
wpwrakoh wait .. there's somethign wrong with soc.fpg07:14
wpwrakupdating ...07:15
wpwraknow it should he okay. sorry about that. md5sum soc.fpg: 04a486a928d4f737763521ddd64203f3  soc.fpg07:16
aw_GREAT !!! we DID.07:21
aw_I2C pass. and video captured !!!07:21
wpwrakcongratulatinos ! :)07:22
aw_in middle connector (BLUE)07:22
wpwrakhow's the image quality ?07:22
wolfspraulphew, great07:22
aw_hehe.... so far i think it's good. you know the image is enlarged. :-)07:22
wolfspraulI need to email rafa about the ben for miri, but I think it's still on07:23
wpwrakusing the current code, composite will simply be on the middle connector07:23
wpwrak(can be changed, of course)07:23
wolfspraulaw_: wpwrak : what should we send to Werner?07:23
wolfspraulaw_: please include one retail new Ben NanoNote07:23
aw_yeah...just change register of multiplexer07:23
wolfspraulthen one of the two reworked boards07:23
wolfspraulthen what?07:23
wolfspraulcamera? power supply? any accessories?07:24
wpwrakNO POWER SUPPLY !!07:24
wolfspraulah :-)07:24
aw_he... wpwrak you're saying you don't have POWER supply or don't want it. ;-)07:25
wolfspraulhe doesn't want it07:25
wpwrakif you really really want me to have one, you may send it in separate mail. just the power supply, nothing else. using air mail, not a courier07:26
wolfspraullet's find out what you want07:26
aw_wolfspraul, i just rework one board only. surely we ship this one first then I back to rework another. :-)07:26
wpwrakaw_:  in argentina, they often don't let things into the country that connect to mains and dno't have the argentine certification07:26
wpwrakaw_: and when they find such a thing, they block the entire shipment. so you can't just remove the power supply and get the rest. very messy.07:27
wpwrakaw_: of course, with a sufficient bribe, also this can be solved ...07:27
wpwrakaw_: may be better to rework #2 first. if you run into problems, you'll have one to compare to07:28
wolfspraulwpwrak: let's make a quick list07:28
wolfspraulnothing else?07:28
wolfspraulspare parts?07:28
wpwrakwolfspraul: i have a camera. case parts would be nice. IR remote, too (i don't have that one yet)07:29
aw_parts: AP2142ASG * 2pcs, APX803-40SAG * 2pcs, few 100uF capacitors, 7181C * 1 or 2pcs ?07:29
wolfspraul2 IR remote07:29
wolfspraulcase parts, we still had some old violet color07:29
wolfsprauland the new light-blue (with logo)07:29
wpwrakif you don't mind parting with a light-blue one, that would be better. that way, i can see how the colors look (e.g., when trying leds)07:30
wpwrakif you're short on them, purple will do07:30
wolfsprauloh no07:30
wolfspraulalright, so 1 light-blue case07:30
wolfspraulwhat else?07:32
wpwrakaw_: (parts) 2  APX803-40SAG isthe most important.  AP2142ASG can't hurt. i have 100 uF caps. and i'm too chicken to try to replace the 7181C anyway, so i'm not sure this one will be useful ;-)07:33
wpwrakmaybe a rubber keyboard, for completeness07:35
aw_wpwrak, for 7181C, it's okay for 1pcs. maybe later you'd like your previous one built with C version. :-)07:36
wolfspraulI doubt Werner will risk that07:36
wolfspraulI'm not against including one at all, of course07:36
wolfsprauleven if the chance is only 10%07:36
wolfspraulbut maybe it's more like 0% :-)07:36
wolfspraulit's also good to be able to compare the two07:37
wpwrakyeah, i'd only think of making such a change if we have an emergency with the C and my pre-rc4 board suffers some fatal accident07:38
aw_the AP2142ASG, since I saw some emails about 4 usb channels. not sure if we will use. so ship 2 pcs to you, then I still have two.07:38
wolfspraulI think no need now for the adv7181c07:38
aw_i bought 5 sets of parts though.07:39
wpwrakthinking of it .. if you have space power/ether/video-on sidewalls, 2-3 of them would be nice to have07:40
wpwrakthat way, i could cut holes for JTAG ... and close the box :)07:40
wolfspraulaw_: how many spares did roh send you?07:40
wpwrakor maybe i should just hack off a bit of cardboard. thinking of it, that may actually be substantially more efficient :)07:42
wolfspraulno wait, roh sent us spares07:42
wolfspraulI just don't know how many exactly, so this may be a very good opportunity to get some actually used07:42
aw_i have three full sets of cases(matched) available. and now two rc3 boards available to assemble. So a light-blue case is okay.07:44
aw_there's two cases with wrong 'M' log. if you don't mind I ship that one to you. so I can still reserve those three full sets of cases.07:46
wpwraksure, no problem07:47
rohwolfspraul: :) holler if you need more cases07:48
wpwrakaw_: btw, here's a flickernoise to go with the rest: http://downloads.qi-hardware.com/people/werner/tmp/adv7181c/flickernoise.fbi07:49
aw_wpwrak, aha..tks.07:50
wpwrakyou;ll need that for the new soc. the older ones will have incompatibilities with uart and/or usb07:50
wpwrakwhen the smoke settles a bit, we should have a great upgrade day. so that everyone gets synchronized again :)07:50
wolfspraulaw_: how about *spare* parts?07:51
wolfspraulI think roh included some extra parts07:52
wolfsprauldo you have any?07:52
wolfspraulroh: do you remember how many spares you sent?07:52
rohnot sure if there were acryllic spares07:53
rohi know i had spare buttons/shields07:53
wolfspraulah ok07:53
aw_wolfspraul, second... needs to count them again from my site.07:53
wolfspraulaw_: we mean individual parts, for example the sides07:53
wolfspraulbut it seems no such thing exists...07:53
aw_since most of them are assembled already.07:53
wolfspraulyes, so either you have *extra* parts somewhere, or not07:54
wolfspraulbut it seems no07:54
aw_moments...i have to clean files firstly07:54
rohi could laser some sidewalls for the video/eth side with another hole for the jtag07:54
aw_no. i guess with reverse/wrong side cases, i can match hopefully if Werner needs 2pcs.07:55
rohwpwrak: which color is your case?07:56
wpwrakroh: the one i have now is purple08:00
wpwrakbut i guess the easiest approach is to use a bit of cardboard and a sharp knife. don't have to be pretty anyway08:01
aw_wpwrak, type 'm1nor soc.fpg bios.bin flickernoise.fbi &&  flterm --port /dev/ttyUSB0 --kernel boot.bin' or 'm1nor soc.fpg && m1nor bios.bin && m1nor flickernoise.fbi && flterm --port /dev/ttyUSB0 --kernel boot.bin' ?08:06
wpwraknow, you just need  m1nor flickernoise.fbi08:07
wpwrakyou already have all the rest08:07
wpwrakfor a new machine,  m1nor soc.fpg bios.bin flickernoise.fbi08:07
wpwrakand then flterm08:07
aw_i see. i mean I need to use '&&' to link them. right ?08:07
wpwrakyou don't need the &&. i only used that in the first example because i didn't know if you already had an older version of m1nor, which would accept only one file at a time. the current version accepts multiple files, so it's all in one command08:09
aw_good the rendering run well with video source (middle connector) :-) but with your new flickernoise.fbi my usb mouser doesn't work.08:17
aw_I count 'extra' cases part now ...08:17
wpwrak(mouse) urk :-(08:18
wpwrakif you go to the BIOS, does it report anything about the mouse ?08:19
wpwrakmaybe stop booting with Esc, then connect the mouse and watch the serial console for USB debug output08:20
wolfspraulgood morning!08:32
wpwraklekernel_: good news: the new codec seems to work. just needs different multiplexer settings08:34
wpwrakaw_: mouser, n.: a cat proficient at mousing   :-)08:41
rohwpwrak: i could laser you a new sidewall in purple. i still have some sheets of that color08:43
rohwpwrak: just no clue how to get it to you afterwards08:43
wpwrakyeah, shipping overhead (also including customs processing on my end) probably vastly exceeds the actual work08:45
rohwpwrak: know some laser shops around? ;) i could also send you a file *g*08:46
wpwrakduh. find shop. send and explain file. go there to pick things up. that's as much overhead as global shipping :)08:47
lekernel_cool :)08:47
lekernel_btw, it seems that codec might support a "raw ADC" mode :)08:47
wpwraknaw, i'll try the cardboard plus knife approach :)08:47
wpwrakoh, and i could make it cardboard plus mill :) that sounds like even more fun08:48
rohwpwrak: you could just mill it08:48
rohsome fr4 also should look geeky08:49
wpwrakyeah. just checked the size and it would fit. most M1 case parts are too large for my mill, but the short sidewalls are okay08:49
rohoh. its that small?08:49
wpwrakpretty small, yes: only 10 x 15 cm work area08:49
aw_(acrylic cases) there's five full sets just after matched rest of them. One with reverse 'M' log, the other I think the surface of acrylic is not good. so I keep remaining 3 sets.08:50
aw_how do you think? two sets or just one set?08:50
wpwraklekernel_: (raw adc) so we can use M1 to control the sensors in our secret nuclear research facilities ?08:50
wpwrakaw_: one set will be okay. i'll just do the cardboard approach. that's easier ;)08:51
wpwrakroh: hmm, my uncoated FR4 boards are a little bit too short08:54
wpwrakroh: i have a ton of aluminium sheets, though. but then, that may be risky :)08:54
rohmill too weak?08:57
Fallenou16:07 < lekernel> I'm setting up http://milkymist.org/blog/ - who wants an  account? < I want one :)08:58
lekernel_hmm, what is the SOY pin?09:00
lekernel_see fig. 30 p. 85 of the 7181C manual09:00
wpwrakroh: the mill is weak, yes. but it can do aluminium, very slowly. but it's be worried to have so much bare metal around. not sure if there isn't anything on the sides of the M1 pcb that's not supposed to make contact09:02
wpwraklekernel_: i think the technical term may be "a surprise" ? :)09:06
aw_wpwrak, with yours, my usb mouse. http://dpaste.com/664826/09:08
aw_but well at least it works well these 2 days without your flickernoise.fbi :-)09:09
rohwpwrak: well.. if you have any advice on shipping services from germany...  i think simply for finding out how well it works we could send a small letter sized acryllic part *g*09:14
lekernel_seems to be useful only in power save mode09:17
wpwrakhmm, that's an unhappy mouse indeed. darn. what kind of mouse is it ?09:20
wpwrak(2 days) so you bought it recently ?09:20
rohwpwrak: hm. dhl says 'brief international' should work09:23
wpwrakroh: the thing is that mail shipments often get held up at customs. and that's basically half a day wasted.09:25
rohah. so you need to go there in person for every letter?09:26
wpwrakroh: fedex and such pass as long as you avoid the traps, and you just pay customs fees and such on delivery09:26
aw_wpwrak, no. the mouse I used to test all rc3 boards. i have run this mouse well two days after I added usb switch AP2142ASG. It's FUJITSU Model : MP117-P09:26
aw_wpwrak, but after your flickernoise.fbi then doesn't work. :)09:27
wpwraknot for letters. but parcels have a high probability. and they x-ray things, so they see if the content of a letter is really paper or something more interesting.09:27
wpwrak(mouse) ah, i see. does unplugging and re-plugging help ? some mice need a few tries before they come to life09:28
wpwraklekernel_: (soy) not sure if it's just for power-saving. seems that this is a new feature in the C and the B didn't even support sync-on-green (never heard of sync-on-luma)09:30
lekernel_sync on luma is what S-Video and YPbPr do09:31
wpwrakah, i see09:31
lekernel_shall we connect this SOY pin? it's just one capacitor ...09:32
lekernel_we can just DNP it if we have problems ...09:32
wpwraki know the old component video cables, before vga. they had a BNC for red, green, blue, and at least one sync if not two09:32
wpwrakyes, i think it's a good idea to connect it09:33
lekernel_from my understanding it is not needed though09:33
lekernel_as aw_'s test has shown, the chip is able to acquire sync without it09:33
lekernel_CVBS sync is the same sync as s-video and ypbpr09:33
wpwrakmaybe only with somewhat peculiar equipment09:33
wpwrakaw_: if that mouse doesn't work, do you have another mouse that does ?09:34
aw_wpwrak, I'll test other three models on hands ... second ...09:36
wpwrakah, you have a ton of them :) very good. some ought to work09:36
aw_wpwrak, four mouses are tested. :-)10:04
aw_1. Fujitsu MP117-P, never work.10:04
aw_2. ASUS M-U005, can plug/replug and work well10:05
aw_3. msi SM02, work only power-cycle but cant be detected when replug10:06
aw_4. Logitech M-UAS144, can plug /replug and work well.10:06
aw_i think i won't cross my hands on them now until next pre-rc4 rework is done. :)10:08
wolfspraullekernel_: what do you think about 4 USB connectors?10:51
lekernel_I don't really know where they would fit10:51
wolfsprauland putting that aside?10:52
lekernel_well, if you want. but 3 are enough imo...10:58
wolfspraulI don't feel strongly about it10:59
wolfspraulmeaning that we don't even fully use the 2 yet, our usb support can still improve a lot it sounds11:00
wolfspraulit's just that werner brought it up so I try to investigate a little11:00
wolfspraulfor example you could have said "oh no, horrible idea, will cause lots of trouble in the soc"11:00
wolfspraulbut you didn't :-)11:00
lekernel_the 3 port scenario I have is keyboard + mouse + LV311:00
lekernel_no, adding more ports should be relatively painless with the current design11:01
lekernel_unless more unexpected bugs show up11:01
wolfspraulwell there are still logs of potential fixes/improvements I'm sure, throughout the stack11:02
lekernel_there's just one multiplexer, and the USB core services the ports one by one11:02
wolfsprauljust look at Adam's mouse list above :-)11:02
wolfspraulwill 3 fit?11:02
lekernel_maybe, if we cramp them a little11:03
lekernel_can we say "version" instead of "rc"?12:02
GitHub111[milkymist] sbourdeauducq pushed 1 new commit to master: http://git.io/OFTuJA12:53
GitHub111[milkymist/master] bios: make the LIBS as .PHONY - Xiangfu Liu12:53
wpwraki think the main question about adding USB ports would be whether the power supply can keep up. at least the path through M1 shuold be able to handle N x 500 mA plus a bit of tolerance13:09
wpwrakregarding uses, i'm sure they will be found. you can never have too many USB ports ;-)13:10
wolfsprauloh good point, the current power supply we include is only rated 2A, but we can source a bigger one. then would need to check the circuit on the board.13:19
lekernel_http://sourceforge.net/projects/opendf/develop (...and for the 1st time, svn didn't crash on a > 1GB checkout)13:22
wolfspraulseems we are in class13:25
wolfspraulcan you summarize what these links are about?13:25
lekernel_building hardware accelerators more efficiently13:26
wpwrakfor 4 ports, you may want to consider one of those stacked USB connectors. like this: http://search.digikey.com/us/en/products/AU-Y1008-R/AE9926-ND/82168013:26
wpwrakor aybe this one. looks tidier and it's "USB 2.0", so more future-proof: http://search.digikey.com/us/en/products/690-008-621-013/151-1085-ND/80618413:28
wpwraknaw, the bot here doesn't do it yet :)13:30
wolfspraulok those are pairs of 2, so we can only do 2 or 413:38
wolfspraulwith the current vertical ones we could do 313:38
wpwrakyou can also find monsters with 2 x USB + RJ45 if you want :)13:40
wpwrakbut i don't trust those. tall = big lever13:41
wolfspraullekernel_: ok so it's about the "CAL Actor Language"? http://en.wikipedia.org/wiki/CAL_Actor_Language14:13
wolfspraulhow does this relate to Milkymist?14:14
lekernel_well, dataflow languages like CAL (there are others, and I can also brew my own) can be used to design things like graphics pipelines more efficiently14:15
lekernel_the problem I see with CAL is there is no support for things like automatic generation of DMA masters and their SoC interconnect14:16
wolfspraulok. do you plan on using CAL for the Milkymist SoC? do you want to encourage others to study/use CAL? for what features in Milkymist specifically do you think it could be suited well?14:20
lekernelfor the design of a more powerful and flexible graphics pipeline14:21
lekernellike, right now, switching to 5:6:5 to 10:10:10 means a lot of manual coding14:22
lekernelfrom 5:6:514:22
lekernelideally, with a high level synthesis language, we could easily parametrize such things14:22
lekernelautomatic memory interfaces also look interesting14:23
lekernelright now there's a lot of copy and paste and "manual compilation"14:23
lekernelthis is daunting, inefficient, and not very flexible... with high level synthesis, it'd be much easier to explore the design space (e.g. enable/disable prefetching, caches, etc.)14:24
lekernelso... yes, I plan to rewrite the TMU using dataflow (not necessarily CAL)14:25
lekernelI'm just looking at what's already existing14:25
wolfspraulah ok, nice14:31
wolfspraulthat's very clear now - thanks!14:31
wpwrak(manual coding) no macros in verilog ?15:54
lekernelthey're not very powerful15:56
wpwrakmmh, pity15:57
wpwraksuch layered languages are often more trouble that they're worth. they have to be really really good, or you'll just end up having to work on/debug your actual code, plus the gobbledygook the precompiler has generated15:58
wpwrakhmm. thinking of it, we have no word on the camera image quality yet. what exactly is the ADV7181C supposed to improve with respect to the B ? anything specific that should be observable ?17:46
lekernelbetter colors17:52
mwallemh, the ft2232 samples with 120mhz and 10 bit oversampling giving us a max baudtick of 12mhz18:19
mwalleunfortunately the error above 4mbps is rather high with our 20mhz (4bit oversampling) and that 12mhz18:20
wpwraklekernel: "better" in what sense ? dynamic range ? pixel-to-pixel crosstalk ? linearity ? ...18:27
lekernelwhat can be expected from the effects of a better antialiasing filter on a CVBS signal18:33
wpwrakhmm. so that would mean that signals with HF contamination will be better but those what stay within the design bandwidth will be unaffected ?18:34
wpwrakvery nice. without MIDI traffic, the "tunnel" patch renders at 24 +/- 0.03 FPS. if i hammer traditional MIDI at 1000 control changes per second, the frame rate drops to 15.44 +/- 0.12 FPS.18:40
wpwraknow, let's make this faster ...18:40
wpwraknice. 23.99 +/- 0.1 FPS18:55
wpwrakperformance is good. now let's see about correctness ...18:55
wpwrakhmm. something ain't quite right ... "optimized" midi seems to trigger false mouse action. not the improvement i was looking for. so where did i go astray ....19:02
wpwrakseems this was just my rf mouse acting up at the point of losing battery power. ah well.20:00
kristianpaulroh: send me the file :-)20:17
rohkristianpaul: will do after i made one20:49
kristianpaullol i tought you alredy did it :)20:50
rohnah. wasted the afternoon making new screw-locks for the fusebox20:51
rohthe box is as big as 2 wardrobes and has plastic thingies which keep the frontpanels in.. 4 per panel, i think 10 panels in total20:52
rohthere were only 2 on average per panel left in working order, the rest broke since 198420:52
rohand i couldnt find the manuf. so i used calipers and openscad and printed new ones. replicator(tm) for real20:53
rohwpwrak: check if the messages contain a 'empty bat' packet20:53
rohmost devices send such prewarns... maybe your stack just misinterpreted that20:53
wpwrakoh. that would be an interesting possibility. i'll have a look the next time this happens. thanks !21:19
wpwraklekernel: btw, did you get a reaction yet from xilinx regarding that xdl comma bug ?21:30
wpwrakjust machined a few more buttons. now i'm ready to receive the pre-rc4 :)23:11
kristianpauldynamical code loading in rtems GREAT !!23:48
kristianpaulhum but suppoty got lm32..23:48
wpwrakdynamic code loading ?23:52
kristianpaulwpwrak: http://www.rtems.org/pipermail/rtems-users/2011-December/009216.html23:53
wpwrakah :) that should be as easy as  if (read(fd, buffer, sizeof(buffer)) > 0) ((void (*)(void)) buffer)();  no ? needs -fPIC, of course23:56
--- Sat Dec 3 201100:00

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