#qi-hardware IRC log for Wednesday, 2011-09-21

wpwrakcongratulations ! :) now that they "know" you, it should be much easier00:00
wpwrakthe passport scan is a bit of a routine thing. a lot of places want that for credit card payments from abroad. and of course, even more so if from "strange" countries :)00:01
wolfspraulgood morning qiots01:33
kristianpaulevening :)01:51
qi-botThe build has FAILED, see log here: http://fidelio.qi-hardware.com/~xiangfu/compile-log/openwrt-xburst.full_system-09202011-0001/02:56
rjeffrieslekernel thanks for the Aleksander Zawada report. so cool. I shared your blog link on Google+03:26
DocScrutinizeraw, hey still here :-D06:51
awDocScrutinizer, hi yes, strange.06:52
awwhy total members are such few? strange!06:53
DocScrutinizermega netsplit07:00
DocScrutinizer[2011-09-21 08:53:02] [Benutzer] There are 41 users and 6379 invisible on 1 servers07:00
DocScrutinizerI can tell you must be on kornbluth.freenode.net07:01
DocScrutinizerother servers are all the same, docscrutinizer51 not here but on another server with similar symptoms07:02
DocScrutinizerso it's not like "our" server got orphaned, it's like whole freenode fell apart07:02
Action: DocScrutinizer grrr mega netsplit07:09
Action: DocScrutinizer this is kornbluth, but on barjavel situation is all the same. Seems all servers got disconnected from each other07:10
Action: DocScrutinizer [2011-09-21 09:13:30] <KindOne|kornblut> You can connect to one of the US servers, or just wait it out - http://freenode.net/irc_servers.shtml07:14
Action: DocScrutinizer [2011-09-21 09:33:11] <KindOne|kornblut> looks like the EU hub went offline, this was not planned07:34
Action: DocScrutinizer WB everybody07:52
viricerror building uboot's mkimage08:02
viric/nix/store/9b3q62czfh5zld9zg2p8sbk65hkbvpzi-binutils-2.21/bin/ld: errno: TLS definition in /nix/store/vxycd107wjbhcj720hzkw2px7s7kr724-glibc-2.12.2/lib/libc.so.6 section .tbss mismatches non-TLS reference in fit_image.o08:02
viricuboot people wrote fit_image thinking of a very specific libc?08:02
viricmaybe thinking 'errno' is not in TLS?08:04
virichm it's a matter of order in -I / -idirafter08:07
kyakviric: you can have a look in the git repo for who makes the nanomap08:18
viricok ok. I thought maybe she would be here08:18
DocScrutinizer51wb everybody08:22
tuxbrainHi all :)08:54
wolfspra1ltuxbrain: tuxbrain!09:03
virichej tuxbrain 09:15
kristianpaulwb tuxbrain 11:20
wpwraktuxbrain: btw, we can discuss this here :) other people may like spi, too12:05
tuxbrainsure, summarizing I have the goal to communicate NN and Arduino though SPI (using UBB) for the OSHWCON 2011 this weekend on Madrid. I ask for help to wpwrak to develop the NN part an console app (master) that can read/write strings trough spi and that can be piped. I will focus on the arduino part (slave) to do the same. we will start with fixed chars strings to avoid wasting time in defining ends of transmision etc protocols. but will be implemented12:10
tuxbrain in the future to be able to transmit whatever size strings or chunk of bytes.12:10
tuxbrainThe main goal is to mount a NN here http://www.youtube.com/watch?v=NAiXv0Nvt0E and make the NN a brain/HMI enchancement, of course due this can be applied to any arduino board alot of other uses can florish from that.12:12
tuxbrainbut to focus the objective before saturday is to have test at least the send/reiceve of for example a 4char string form MM to an Arduino board, and make it documented to make it reproducible with no need of in deep knowledge of the whole thing, just download the console app, c&p the arduino code, connect UBB-ISCP cable and work. Any help will be really really really appreciated12:15
qi-bot[commit] Werner Almesberger: m1/perf/sched.c: added asserts and list poisoning (master) http://qi-hw.com/p/wernermisc/60cf6f512:20
qi-bot[commit] Werner Almesberger: m1/perf/sched.c: fixed compilation with DEBUG enabled (master) http://qi-hw.com/p/wernermisc/e34e8eb12:20
qi-bot[commit] Werner Almesberger: m1/perf/sched.c: slight cleanup (no functional changes) (master) http://qi-hw.com/p/wernermisc/1ad7db712:20
qi-bot[commit] Werner Almesberger: spiio/: simple SPI I/O example (master) http://qi-hw.com/p/wernermisc/05bbe9112:20
wpwraktuxbrain: this should do the trick: http://projects.qi-hardware.com/index.php/p/wernermisc/source/tree/master/spiio12:21
wpwrakyou need to edit the signal assignment at the beginning. you may also want to check the timing. finally, you must decide if setting the I/O's to input at the end is really what your device likes to happen.12:23
wpwrak(should be fine as long as nSS isn't on MMC_CLK. all the other MMC signals have a pull-up, which means that nSS is deasserted)12:24
wpwraki shuold add that this is completely untested ;-)12:24
tuxbrainI will try along this week. I will report my success/fails12:51
wpwrakgood luck ! :)13:02
wpwraki'll be afk for a bit now13:02
RandomUser4Hello. Not sure this is the adequate channel but here goes. I've recently assembled a new computer. I have a problem: I can run each of both 2GB RAM sticks on their own but not both at the same time. Win 7 Home Premium x64. I've been reading a lot on the matter, but no clear solution so far13:36
RandomUser4The computer gets a BSOD some minutes after Windows logs in when both sticks are in, although yesterday I managed to run it for hours, until I turned it off, without any issue.13:38
ArtlavOn my nanonote the screen suddenly started flickering, starting from u-boot and on. What could be wrong? 15:34
larscwhat kind of flickering? low or high frequency? blacklight or screen contents?15:36
ArtlavContent. In uboot the text lines jump up and down, later every second line moves a pixel up and back. High frequency.15:37
ArtlavIn mc black lines run on the right side of the screen, and the impression is of shaking up and down.15:39
larscdoes it change if you push the case above the keyboard down?15:39
larscnext to the power button15:40
ArtlavIt changes a little. Too much and all stops and screen starts to go white.15:41
larschm, looks like a problem with the screen connector15:42
larscyou could try to remove the front cover and try to check whether the cable is loose and reconnect it if necessary15:43
ArtlavAnything unusual to take care of in disassembly?15:43
larscnope. there are 4 screws which have to be removed15:45
larsctwo at the outedges at the front two under the battery cover15:45
ArtlavDoes not seem to help. It stopped getting worse but didn't got better.15:53
ArtlavIs the screen side disassembleable?15:58
larscArtlav: http://en.qi-hardware.com/wiki/Disassemble16:03
wpwraktuxbrain: back. how is it going ?16:08
tuxbrainI'm on family guy mode now, I will switch to PC addon mode in about 3-4 hours16:09
Artlavlarsc: Thanks. Is that even the right place to ask problem questions?16:10
wpwrakArtlav: probably the very best place :)16:11
wpwrakArtlav: regarding the cable, did you remove it from the connector and put it back in ? are you familiar with how these connectors work ?16:12
Artlavwpwrak: Yes. I removed it, cleaned with alcohol and reinserted. It is no longer sensitive to pressure, but the flicker remains.16:14
ArtlavHere is what it looks like: http://www.youtube.com/watch?v=ChNsagwf_9A16:14
ArtlavThe problem, that is.16:15
wpwrakheh, bug reports with a youtube video. that's new :)16:15
ArtlavBetter than "screen flickers", no?16:15
wpwrakhmm, odd symptoms. may be a failing hsync line. let's see where it is located ...16:17
wpwraknot every exposed between ben and lcm. and on the lcm itself ...16:18
wpwraknope, not very exposed there either16:19
wpwrakone possibility would be a hairline crack in the FPC. that would be fairly fatal problem :-(16:20
wpwrakyou should be able to still get the ben to work for a while by pressing the cable at just the right spot and in the right way. but the crask is likely to get worse with time.16:21
ArtlavNo pressing or bending of the cable itself seems to affect anything. Can there be something wrong on the screen side of it?16:23
larscit has the same type of connector on the screen side, so in theory yes16:25
wpwrakof course :) it seems a lot less likely, though. there is also a FPC in the screen. it's ben mainboard --FPC--> PCB in the display--FPC-->LCD module16:25
larscbut you'd have to pull the cable pretty hard to get it loose16:26
wpwraklarsc: you mean the FPC between ben main pcb and the LCD's PCB ? that one looks very much as if it was soldered to the PCB. of course, with enough force, it will come off ...16:27
larscif you had a scope you could check the lcd signal for irregularities16:27
ArtlavNothing suspicious inside the screen part. No scope.16:34
ArtlavCan there be some other reason? If i loosen the cable a little, all kinds of artefacts start to appear, but the flicker is always there. As if it was from somewhere else.16:34
wpwrakthat would suggest the problem is at a different place. could be a bad solder joint or a hairline crack in a trace.16:36
wpwrakor some chip's driver failing. not sure how like that is, though. it must happen occasionally :)16:37
larscit looks a bit as if there is jitter on the lcd clock and some edges are missed16:38
larsceg. blue truns red16:38
ArtlavIs there any documentation on prior nanonote hardware failure modes? Sounds like i have a unique case on my hands.16:40
larscone thing you could try is to see if there is any change if you downclock the lcd clock16:42
larscbut that would require a custom kernel16:42
wpwraklcd failure has occasionally been observed before. but i don't think we have a case where it's clear that it looked like that. (maybe it did, maybe it didn't)16:43
ArtlavCustom kernel is not a big problem, what to do?16:43
wpwrakso the video is indeed quite useful :)16:43
wpwrakyou could try to press on the contacts of the FPC connector on the ben's PCB with a non-conducting object16:44
larscArtlav: in arch/mips/jz4740/qi_lb60.c reduce qi_lb60_video_modes.rate16:44
wpwrakthat would reveal bad solder joints, cracks near the connector, and anything fishy inside the connector16:44
wpwrakben schematics are here: http://en.qi-hardware.com/w/images/9/9c/Lb60_schematic.pdf16:45
wpwrakyou'd be looking for hsync or maybe the pixel clock16:45
wpwrakyeah, slowing things down could also be useful. you never know ...16:46
Artlavboard-qi_lb60.c ? There is refresh in it, but no rate. refresh=30.16:48
larscyes. refresh sorry16:48
ArtlavReduced to 20. Compiling...16:50
ArtlavFunniest thing is Murphy's law - just as i seriously started combing the software and streamlining the stuff up, the hardware suddenly failed!16:51
ArtlavWhat do you know, half of the problem is gone.16:55
ArtlavIt no longer goes up and down, but the lines are still here.16:55
larschalf of the problem?16:55
ArtlavThe cord is a part too - now there is a clear relation between pressures on it and the amount of lines. But no pressure to remove them completely.16:56
larsctry to only apply pressure to the pins that go to the pcb16:59
wpwrakthat's a very strange one. looking closer at the video, the background color varies but the forground text looks pretty stable16:59
wpwrakcould there be conductive contamination ? metal dust or such ?16:59
larscwpwrak: if you have a series of white pixels only the first and the last will change color17:00
wpwraklarsc: what i mean is there there's strong horizontal distortion (brightness and sometiomes color)  in the background but none in the foreground. pretty weird.17:04
ArtlavNo longer sure about "clear" relationship. The lines seem to up and ebb by themselves whether i touch the cord or not.17:05
ArtlavA sub-pixel shift. Starting from some line everything is shifted right by a subpixel - that's what these lines are.17:10
ArtlavThe text and background changes colour and shifts right a little.17:11
ArtlavStarting from approximately 100 by x.17:11
ArtlavClean cut-off.17:11
ArtlavThere is a tiny black dot at the left side of each line.17:12
ArtlavAlmost as if a clock is skipped and a pixel missed, with the rest send in regular order.17:13
ArtlavAnd it's not limited to video - the sound now have tiny chirps in it, the higher the pitch the worse.17:14
ArtlavMaybe the problem is somewhere else entirely?17:15
wpwrakinteresting .. what on earth did you do to your ben ? :)17:15
wpwrakdoes USB still work ?17:15
larscit's probably interference surveilance van outside17:16
larscit's probably interference from surveilance van outside17:16
ArtlavUSB - yes, kernel flashed OK.17:17
ArtlavAnd to use the oldest phrase - i didn't do anything.17:17
wpwrakor that large flying saucer hovering over the house. they're probably adjusting the abduction beam.17:17
wpwrak(USB) that would suggest the system clock is okay17:18
ArtlavNot to hardware anyway. I tried all kinds of software experiments, like 64Kb pages and dynamic ticks.17:18
ArtlavAnd last UFO around here was a hot air balloon in the night.17:19
ArtlavThere seems to be a relation of line intensity to the CPU usage.17:20
Action: wpwrak notes that suspicious flying objects are a frequent occurrence in Artlav's vicinity17:20
wpwrakmaybe it's a power instability17:20
wpwrakhard to debug without a scope, though17:21
wpwrakwell, you could put some resistor, say 33 Ohm, between 3.3 V and ground and see what happens.17:21
kristianpauli havent seems anything like that, even when i dealed witht that lcm cable semi-loose for a month 17:22
ArtlavI tried it on USB cord and no battery, no changes.17:22
kristianpaulremenbering lcm have its own firmware right?17:23
wpwrakcould be a problem on the secondary side of the regulator17:26
ArtlavSound problems cleared - software issue in gmu.17:28
wejwhat issue?17:29
ArtlavNo idea. chrooted into the old system - good sound. back - chirps in sound.17:30
wejdoesn't sound like an issue with Gmu17:31
ArtlavBut does sound like a software issue?17:31
wejprobably yes17:32
Artlavmplayer gives clear sound, gmu gives chirps. 17:32
ArtlavIn same system. So, not likely to be a hardware thing.17:33
ArtlavOk, 22 Om resistor between GND and + of battery slot did nothing.17:38
ArtlavThe lines seem to get more and more intense as the time go.17:39
wpwrakis the downstairs neighbour experimenting with strong magnets or such ? we had that once at the university ... our sysadmin, sun support, etc., were getting gray hair. i suggested that it was time to seek the help of an exorcist, ... then, a few days later, our sysadmin had lunch with the folks from the lab below ours and they told him about their cool experiments ...17:43
kristianpaultrying to be free gravity? ;)17:47
ArtlavUnfortunately, the issue occurs on both ends of quite a large city, so surroundings are hardly the problem.17:49
ArtlavFurther downclocking do nothing.17:54
ArtlavThere is a definite CPU load relation - with video playback the whole right 2 thirds of the screen is in distorted flicker, with a picture viewer it's a bit of lines, with just the terminal its a few lines.17:55
ArtlavThe blanking go into all-white every odd time.17:56
ArtlavOr not...17:57
ArtlavIf i look at a picture of a circle, then the center of the screen is in lines, while the up and bottom is not. More image - more lines.17:58
ArtlavYes, related to amount of image on the screen, and possibly related to CPU load.17:59
ArtlavI guess the real question is - is it worth buying a new one now, or would there be a better version soon?18:01
ArtlavWell, thanks all for helping and good night.18:16
wpwrakArtlav: nothing new coming anytime soon. so another ben seems to be best choice.18:20
ignatius-Can anyone recommend a good MP3 player for the Ben?19:07
AwAylaignatius-: GMU19:15
ignatius-I'ved tried GMU. It doesn't play MP3s fast enough. It "skips"19:18
wejno, it plays mp3 just fine. if you have compiled the mp3 plugin that is19:19
ignatius-Yes, I have.19:19
wejwith any bitrate up to 320 kbit19:19
wejthen you did something wrong. my guess is that you have built the mp3 decoder with floating point math19:20
ignatius-Ah. How do I circumvent that?19:20
wejby telling mpg123 to build with the fixed point decoder19:20
ignatius-How do I do that?19:21
wejyou need to tell the configure script to do that, otherwise it will compile with float, which does not work on a cpu without an fpu19:21
ignatius-No configure script. Is there a CFLAG entry for that?19:26
wejthe decoder comes with a configure script, i am 100 % sure of that19:28
wejif yours doesn't. redownload it19:28
ignatius-http://wejp.k.vu/projects/files/gmu-0.8.0BETA1.tar.gz does not have a configure script.19:31
wejit doesn't need one19:32
ignatius-So, where do I find a version with a configure script? Any idea?19:32
ignatius-Thanks, BTW.19:33
wejthere is no version of gmu with a configure script19:33
weji am talking about the mpg123 decoder19:33
ignatius-Ah. Ok. But, I thought GMU uses it's own mpg decoder? At least there is in the source tree... or appears to be..19:33
wejGmu uses mpg123 for decoding mp3, if it is configured with mp3 support19:34
wejof course it has to implement support for that library19:34
ignatius-Ok. So just recompile a version of mpg123, than?19:34
wejwhich is why there are files  for that in the Gmu source19:35
ignatius-Ok. Thanks, man. I appreciate it.19:35
wpwrakhmm, fun spam: "New LED Candle bulb here:" "Advantages: 1.Patented product!"  somehow their marketing doesn't seem to be very targeted :)19:48
mthignatius-: buildroot passes this option to mpg123 configure: --with-cpu=generic_nofpu20:45
ignatius-Ok. Yeah, the package version of "GMU" fails to compile. As do every other package for some reason.20:54
ignatius-On my desktop, I mean. I'm compiling it on the Ben.20:54
qi-bot[commit] Werner Almesberger: m1/perf/sched.c: revamped to handle static registers correctly as well (master) http://qi-hw.com/p/wernermisc/8f82a0e21:08
qi-bot[commit] Werner Almesberger: m1/perf/main.c: exit immediately on compilation error (master) http://qi-hw.com/p/wernermisc/a19a19621:08
qi-bot[commit] Werner Almesberger: m1/perf/: "spotless" target that also removes the "main" executable (master) http://qi-hw.com/p/wernermisc/23c7bd621:08
qi-bot[commit] Werner Almesberger: m1/perf/sched.c (foreach): improved type safety (master) http://qi-hw.com/p/wernermisc/b177acd21:08
qi-bot[commit] Werner Almesberger: m1/perf/sched.c: light code cleanup (master) http://qi-hw.com/p/wernermisc/e98f08e21:08
qi-bot[commit] Werner Almesberger: m1/perf/sched.c (schedule): reduced conditionally compiled code size for optimizer (master) http://qi-hw.com/p/wernermisc/2bc4c3321:08
qi-bot[commit] Werner Almesberger: m1/perf/try: disable profiling (master) http://qi-hw.com/p/wernermisc/403917a21:08
qi-bot[commit] Werner Almesberger: m1rc3/norruption/LOG: push results of on-going testing (master) http://qi-hw.com/p/wernermisc/7567c4f21:08
--- Thu Sep 22 201100:00

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