#qi-hardware IRC log for Saturday, 2011-04-23

kristianpaulwolfspraul: hi00:53
kristianpaulwolfspraul: I saw it in Xiangfu blog, also in openwrt xbusrt for qi, support and related information for a hanvon ebook00:53
kristianpaulso i'm curios if were some plans around this technology?00:54
kristianpaulif it was at the same time you founded the color digital dictionary? but then comunity just focus on a general porpuse computer pocket computer instead of a ebook? so thats the end of the history?..00:55
kristianpaulmorning btw :-)00:55
rjeffriesAccording to goo.gl 33 people clicked on my note today about Milkymist (across identi.ca, twitter and Buzz)01:01
wpwrakkristianpaul: the past tense of "find" is "found", the past tense of "found" as in "foundation" is "founded" :)01:02
kristianpaulouch, sorry :/01:03
rjeffrieswpwrak indeed, a combo ofIngenic SOC and Linux with Propeller chip as "helper" chip would give VGA, also a LOT of i/o it's cheap too.01:05
wpwrakkristianpaul: with an english as good as yours, those little mistakes are actually quite amusing ;)01:06
Action: roh made internetz work again.01:07
wpwrakrjeffries: dunno if the prop is worth the trouble. i like the concept, but wouldn't, say, a cpld or a small fpga serve us better ?01:08
rjeffrieswpwrak an OK way to get ethernet connectivity will be ATusb working on a cheap linux router01:08
rohwhat-a-ugly-hack ... use the common wifi from the neighbours till we get our own link (and a wired one to them)01:08
kristianpaulwpwrak: fpga cames in different LUT sizes, so well..  :_)01:08
wpwrakroh: like this ? http://www.dailymotion.com/video/x7vf8c_it-crowd-the-internet_fun01:08
rjeffrieswpwrak the propeller is as you know well a quirky chip... cool but quirky. yes  a smallock.01:09
rjeffrieslekernel uses a big (!) fpga I assume.01:09
kristianpauleven those iglo fpga vendor, that claim low power and small factor should beat a cpld for sure :-)01:09
rohwpwrak: i am guessing now...  not much bandwith01:09
rohvery laggy01:10
kristianpaulha, is funny video :-)01:10
rohhrhr.. similar... our boxes are white and labeled asus wl500g01:10
wpwrakrjeffries: the easiest way for Internet connectivity shall be to just plug an atusb into a linux pc :)01:12
wpwrakrjeffries: admittedly one with the right drivers. but we'll get there as well :)01:13
rohyess... and i just found another antennaposition which made it work much better... well.. back to installing wall-sockets01:13
rohrjeffries: somebody should make sure the neccsary tools and drivers for the atusb get into openwrt01:15
rjeffrieswpwrak yes a Linux PC will work, of course. but something really cheap and low power consumtion  that runs 24x7 would be nice01:15
rohrjeffries: openwrt-able plastic routers with usb start at something like 25E01:15
rjeffriesquestion: once the tools and drivers are in openwrt, how hard to get them into a distro such as oick one, Arch, Debian...01:16
rjeffriesroh that sounds promising about $50 USD in round numbers01:16
rohrjeffries: well.. yes.. but dont forcefully bundle it01:17
wpwrakrjeffries: openwrt isn't really the main path. for kernel stuff, it's via netdev and then vanilla. trickles back down from there.01:17
rohthere are loads of people who already own useable routers01:17
rohrjeffries: openwrt is just 'one' distro.. well.. in that case one optimized mostly for appliances and especially all these plastic routers01:18
wpwrakroh: how could rjeffries forecefully bundle it if tuxbrain is selling the individual components ? :)01:18
rohwpwrak: dunno.. i dont know 'the plans' ;)01:19
rjeffrieswpwrak back to propeller-land for a moment one also gets keyboard, VGA video, mouse[ps2] and keyboard[ps2]01:20
wpwrakroh: the plan is for tuxbrain to get the thingies made and to sell them. whatever else happens is another story. quite simple, actually :)01:21
rjeffriesI think roh meant to not force the softwrae into distros01:21
rjeffriesyup. since it is Easter season, I am hoping for the resurection of Tuxbrain //bad joke, I know01:21
wpwrakrjeffries: _forcing_ software intro distros may be much harder than you'd imagine ;-))01:21
wpwrakrjeffries: (prop) are you thinking of something like ben+ubb+prop or ben++ ?01:22
wolfspraulroh: did you see our latest m1 pictures :-) it's your beautiful case! http://en.qi-hardware.com/wiki/Milkymist_One_pictures01:22
rjeffriesa hood thought experiment is what will stimulate demand for ATusb. that will drive unit vosts lower01:23
wpwrakrjeffries: (atusb) yup. it's another thing with high setup cost but better per unit cost afterwards. less so than UBB, but still01:23
rjeffrieswpwrak since Ben++ is a hazy future, Ben+ubb+simple Propeller boardwith putpose of adding i/o to Ben might work01:24
rjeffriesI winder if we could have an adequate VGA interface that way?01:24
wpwrakrjeffries: (prop) perhaps. don't know how much the prop can do in parallel. e.g., if one peripheral plus communication with the ben already eats up all the cogs, it's not worth the trouble. if it can indeed do a bunch of things, maybe.01:25
wpwrakrjeffries: is there a decent free C compiler for the prop by now ?01:25
rjeffrieswpwrak there is a C compiler ay long last not sure of code quality01:26
wpwrakrjeffries: for vga, consider using a cicuit like the prop uses with ubb. i'm not entirely sure if it could work, but it might01:26
wpwrakrjeffries: (c compiler) is it open source ?01:27
rjeffrieswpwrak I will check on that C compiler just forund out about it  recently01:27
rohlast time i checked propeller didnt even have a compiler running on something else than some dosbox01:27
rohit was supposedly coded in x86 asm01:28
wpwrakroh: who needs compilers anyway ? ;-)01:28
rjeffrieswpwrak http://propeller.wikispaces.com/Programming+in+C+-+Catalina01:29
wpwrakrjeffries: okay, that's a decent enough start. afaik, LCC had an anti-commercial license, though. so that would be a problem.01:30
wpwrakhmm. still no propeller 2. they've been talking about it some 4 years ago. reminds me of those psocs :)01:35
rjeffrieslcc looks ok to me, but I am not a freedom taliban eiether http://en.wikipedia.org/wiki/LCC_(compiler)01:39
wpwrakrjeffries: well, the problem with non-commercial licenses is that they conflict even with loss-making endeavors like ours01:39
rjeffriesyeah {ropeller 2 has been stalled. I think it is still being developed by one genius but oddball engineer and a smart layout person01:40
rjeffriesnot sure that is what that licese means I read it that you can not SELL the LCC software01:40
wpwrakrjeffries: (prop 2) two people ? aw ... is parallax that small or have they basically abandoned the architecture ?01:41
rjeffriesnope the propller was as I remember a one mad design. prety amazing in my opinion01:41
wpwrakrjeffries: "not sell" probably means that you can't ship it with a ben in any form01:42
rjeffriess/mad/man but a funny typo01:42
wpwrakrjeffries: similar to anything mp301:42
rjeffrieswpwrak mayeb I;ll do some email sleuthing on lcc01:42
wpwrakrjeffries: (one mad) indeed ;-)01:42
rjeffriesi meant to write one-man design effort01:43
wpwrakrjeffries: yeah, but some aspects of the thing carry a whiff of madness ;-) and of course, there's genius as well. they sometimes go hand in hand :)01:44
wpwrak(madness) particularly the bit about the own programming language and making the device a self-contained development platform. argh.01:45
wpwrakthat's like building a statue to the knights of NIH (Not Invented Here), resting on a tall tower placed upon a great pyramid (carried by four elephants standing on a tortoise, and so on :)01:47
rjeffriesyes agree wpwrak. I do hope Propeller 2 sees the light of day however02:21
rjeffrieswpwrak yjis Propeller dev board has a loy og available plu-in modules. fairly cheap.02:52
dvdkanybody around here mind to comment about a hacky openwrt recipie?08:33
dvdktrying to package a recent mplayer08:33
dvdkmplayer requires some directories from the ffmpeg source tree.  the part that is is staging_dir/usr/include doesn't suffice08:33
dvdknow I link part from ffmpeg's build directory directory into mplayer's build directory08:34
dvdkthat's ugly.08:34
dvdkis there a way to make such a build recipie work reliably?  i.e. can I write dependencies in a way that ffmpeg build directory is always created before compiling mplayer?08:35
dvdkhow can i find out the version number number of the ffmpeg that is build so i know the name of the build dirs?08:36
kyakdvdk: maybe create a little script for that?08:38
dvdkhi kyak08:38
dvdka script for which part?08:38
dvdkhere is the ugly shell script code i put into build/configure:08:39
dvdkmkdir -p $(PKG_BUILD_DIR)/ffmpeg08:39
dvdkfor i in $(PKG_BUILD_DIR)/../ffmpeg-0.5.*/libavutil; do \08:39
dvdkln -s -f -t $(PKG_BUILD_DIR)/ffmpeg $$$$i; \08:39
kyake.g. when you symlink the ffmpeg dir, first do something like "find $BUILD_DIR -name "ffmpeg*" -depth 108:39
dvdkok, that is similar to the for i in *... loop i use that guarantees i use the latest ffmpeg verison found08:40
kyaki agree it looks ugly08:40
kyakbut should work reliably08:40
dvdkbut as the build system knows which version it builds (and resolves the libffmpeg dependency i gave), can't it give me that knowledge?08:40
dvdkkyak: only as long as shell sorting also matches version sorting :)08:41
dvdkgoing from ffmpeg 5.9 to 5.10 will cause trouble :)08:41
dvdkAh, got it, maybe i can get ffmpeg version from the pkgconfig .pc files installed to the staging dir.08:44
dvdkbut then, still, there is no guarantee that the ffmpeg build_dir actually exists.08:44
dvdkit is only re-generated when ffmpeg package becomes outdated and make rebuilds it, no?08:44
kyakthe build_dir for ffmpeg should always exist, because mplayer depends on ffmpeg08:45
kyaktherefor, ffmpeg always builds before mplayer08:46
dvdkkyak: yeah, but if ffmpeg compilation succeeded, then somebody may cleanup the build dir, and makefile won't regenerate it, since ffmpeg is up to date?08:48
dvdknow build fails with libavutil/sample_fmt.h missing.  well, maybe ffmpeg 0.5.4 is too outdated, and I need to search for a matching mplayer revision...08:50
kyakdvdk: you someone would clean up the build dir via make clean, it would also delete files from staging dir08:51
kyakso it would lead to rebuilding of package08:51
dvdkkyak: rm -rf build_dir :)08:51
kyakif someone had removed the build dir manually, he's on his own08:51
kyakhe knows what he does :)08:51
dvdkok, so then i guess the hacky recepie idea is good :)08:51
dvdkouch, mplayer r32500 _contains_ a private libavutil directory.08:52
kyakdvdk: do you think it might be a good idea to pacakge our own ffmpeg?08:52
kyakand to link mplayer statically to it?08:52
dvdkkyak: not sure.  mplayer is a very special beast.08:52
kyakname it "ffmpeg-for-mplayer"08:53
dvdkkyak: the cleanest solution would be to roll our own mplayer tarball that *includes* ffmpeg.  everything else is just plain hacky08:53
dvdkafter all the mplayer configure script overwrites configure settings of ffmpeg etc.08:53
dvdkso we should treat them as one tarball08:53
dvdkthese two pieces of sw are not really factored.  maybe because developers are mostly the same.08:54
kyakis it problematic to build the ffmpeg that mplayer pulls from ffmpeg git?08:54
dvdkkyak: the ffmpeg mplayer pulls is placed under mplayer's build dir, *because* mplayer build process hacks it in various ways.08:54
kyakyeah, so what?08:54
dvdkinclude files are used that are not even public api includes etc.08:55
dvdkto mplayer ffmpeg  is not a library, it's a part of its source code with deep connections into every part08:55
dvdkwhat i want to say is08:55
kyakdevelopers somehow intend to build mplayer like that.. could we just follow it?08:55
dvdkthat's what i want to say.  building ffmpeg and mplayer separately is a headache08:56
dvdkbut there is no proper tarball that includes both.08:56
dvdkso we'd have to make and host one ourselves08:56
dvdkdoing a non-versioned git pull from ./configure asks for trouble08:56
kyakcould we patch the configure to not ask for anything, and just pull the ffmpeg?08:57
dvdkkyak: we could but i'd advise against it.  this breaks many aspects of openwrt build system08:58
dvdkalso without asking for a special git revision, every compile would yield a different result08:58
kyakbut we can pin the specific git revision08:58
dvdkwe could.  but then we'd pull for every rebuild.08:58
dvdkopenwrt's download rules cache everything in ./dl08:59
dvdkI'm against pulling once per rebuild.08:59
dvdkI'd say let us use the openwrt download/caching system.09:00
dvdkeither customize the download rule to add git checkout once,09:00
dvdkor host our own tarball.09:00
dvdk... or put ffmpeg tarball into the mplayer/files directory and unpack on ./configure :)09:00
kyakactually, i don't think it a problem to pull ffmpeg on every rebuild09:01
dvdknot for your internet connection :)09:01
kyakok, makes sense :)09:01
dvdkok, it's (a) tarball or (b) new download rule or (c) put ffmpeg to files/09:02
dvdkI'd vote for (b) or for (a) if (b) proves  to difficult09:02
dvdkok, going to try (b).  already hacked download rules once, shouldn.t be too bad09:03
dvdkbtw hopefully mplayer allows disabling patented codecs when leaving the ffmpeg config&build part to mplayer's build system09:04
kyakyeah, that's another thing.. the makefile of openwrt's ffmpeg is tricky09:05
dvdklet's look at that once mplayer builds :)09:06
kyakdvdk: btw, that rc4 release of mplayer is too old/lacks features?09:06
kyaki noticed that it doesn't try to pull ffmpeg09:07
dvdkkyak: dunno, but since there are no future releases, going to svn feels more like being future-proof.09:07
dvdkwith vp6 coming out etc. we'll have to upgrade over and over anyways.09:07
kyaki wonder why major distros are still sticking to rc4 and what they will do in future09:08
dvdkkyak: yeah suffering the same problems we have, i guess :)09:08
dvdkstill no decision?09:09
dvdkkyak: how do i find out a revision to use for pinning ffmpeg?09:37
dvdk git checkout $(FFMPEG_REV))09:37
dvdkfatal: reference is not a tree: 9bf81b49cff3945a76ac776f086a1d1adc120e6d09:37
dvdkfor svn i'd use 'svn info', but what do i do for git?09:38
dvdkok, found it:09:41
dvdkcat .git/refs/heads/master09:41
viricdvdk: is there anything in the gpu optimisations you do that allow to play fine videos encoded at higher video size than the nanonote screen?09:48
viricI don't know if mplayer always decodes the blocks totally...09:48
dvdkviric: well, it can do downscaling :)09:48
dvdkif you manage to decode hr video09:49
dvdkyes, everything totally decoded usually09:49
dvdkmplayer has options to skip decoding steps for 3low-res decoding09:49
viriclibjpeg has faster code for downscaled results, for example09:49
viricI don't understand either 'hr video' or '3low-res'09:49
dvdkbut usually you'd just re-encode the video for your nanonote to save space and optimize for decoding with less power.09:50
dvdkalso, for NN we won't ship patented codecs, so that mostly leaves Theora for modern video09:50
viricor vp809:50
dvdkyeah, migrht be to cpu intensive for NN.  also theora performs quite well for low-res.09:51
viricvp8 can be compiled specifically for mips, iirc.09:51
dvdkstill, thechnology-wise theora is much easier on the cpu09:51
dvdkvp8 you mean webm?09:51
viricwebm is matroska + vp8 + ogg, isn't it?09:52
dvdkdid you see the bbb.ogv demo movie i encoded with theora ptalarbvorm?09:53
dvdkquality-wise that looks great, especially on the small nn screen09:53
dvdkwhat's that:09:54
dvdkffmpeg/libavcodec/libavcodec.a(aacsbr.o): In function `sbr_dequant':09:54
dvdk[..] MPlayer-r33304/ffmpeg/libavcodec/aacsbr.c:1100: undefined reference to `exp2f'09:54
viricI'm not sure vp8 decoding is cpu intensive.09:54
viricdvdk: -lm09:54
dvdkthere is -lm.09:54
dvdkmaybe wrong order?09:54
dvdk -ltheora -logg    -ldl -rdynamic  -lm  -ldirectfb -lggi -laa -lvga -lSDL09:54
viricmaybe the uclibc in openwrt does not have exp2f09:54
viricI personally use glibc in the nanonote09:55
viricdvdk: remember that the NN does not have a fpu. Are you configuring libavcodec for without-fpu?09:58
viric(maybe it ends up calling floating point operations in any case, I don't know)09:58
dvdkeverything right msoft-float.09:58
dvdkhowever, uclibc lack exp2f09:58
viricno no09:58
dvdkit has expf() and exp2() though09:58
dvdk-Dexp2f=expf :)09:58
viricexp2f looks for c9909:58
dvdkno exp2f=exp209:59
viriclooks like a c9909:59
viricOk I leave. Sant Jordi.09:59
dvdkok, mplayer from svn compiles.  let's see how it works.10:37
dvdkok, it now uses fftheora, i.e. ffmpeg's theora reimplementation10:40
dvdkdemuxing works, too.10:41
dvdkhmm, the vidix driver fails to work.10:42
dvdkwow, -vo cvidix -geometry is working, cpu load down to 50% :)10:44
wolfsprauldvdk: in your mail you mentioned that the scaler can scale anything up to 320x240, can it also scale down?10:48
wolfspraulsay from a 640x480 theora video down to 320x240 full-screen on Ben10:49
dvdkyes it can, although image quality might not bee too good10:49
wolfspraulthere are only a few hundred videos on Wikimedia Commons now, but it still gives sort of an idea over resolutions... http://commons.wikimedia.org/wiki/Category:Videos_by_display_resolution10:50
wolfspraulso there is a lot actually right in 320x240 (60 videos), then another cluster at 640x480 (178 videos), 1280x720 (49 videos), and even full HD I guess 1920x1080 (12 videos)10:51
dvdkwell, the best idea would be to transcode for nanonote resolution anyways.10:51
dvdkthe scale can scale from any size to (almost) any size.10:51
wolfspraulI am hoping one day we can have a client that can browse, download and display from a repository like Wikimedia Commons10:51
wolfspraultranscoding would probably be very painful, if someone watches a video only once10:52
wolfspraul(Wikimedia Commons is not a serious video repo now - simply too few videos. Just to get the idea across)10:52
dvdkwell, i could easily write a script that transcode just everything and puts it on qi-hardware.com :)10:53
dvdknanonote cpu-power isn't currently strong enough for 640x480, so downscaling on the NN wouldn't be very useful.10:54
dvdkvideo aborted with OOM after a few minutes.  are we leaking memory?10:54
dvdktrying with -vc theora (i.e. official theora decoder) instead10:55
dvdkwolfspraul: then there is also vodo.net10:57
wolfspraulyou mean it's not realistic that a 640x480 theora video is scaled down and displayed at 320x240, in real-time?10:57
dvdkmaybe after some tweaking.  currently not.  not without simd optimizations etc.10:57
dvdkscaling is not the problem.  it's mostly for free10:58
dvdkbut decoding is costly, and we only have a tiny cache, no L2 cache etc.10:58
wolfspraulnice link about vodo.net - thanks! didn't know about it11:00
dvdkyou should see pioneer one :)11:00
wolfspraulunfortunately non-commercial, but nice project nonetheless11:01
dvdkwas just going to complain about the license :)11:01
wolfspraulwell it's ok. it seems they are trying to create ways to build a distribution and monetization network/channel for the creators. more power to them.11:01
wolfspraulI believe this should be done on 'free first', then 'pay', but it's good that many people try different things in parallel.11:02
wolfspraulnothing better than paying for free content, imo11:02
wolfspraulvoluntarily of course, based on the degree of appreciation for the work :-)11:03
dvdkok, we have a demuxer problem.  the ffmpeg ogg demuxer leaks memory (or just needs lots of it)11:16
dvdkrunning mplayer -demuxer ogg seems to fixes the leak11:16
dvdkMaybe we should get the Nanonote listed under Theora Hardware at http://wiki.xiph.org/Theora_Hardware (once everything works stably / is part of official firmware)12:49
DocScrutinizerwolfspraul: h-e-n hostmode for N900 earned some 60EUR donations from 2010-11 til now12:51
wolfsprauldvdk: definitely, we should list it12:52
DocScrutinizer60..100¬ total12:53
wolfspraulmaybe one day we have Theora encoding in Milkymist too, who knows... I see Elphel 333 in that list :-)12:53
wolfspraulwhat is h-e-n hostmode?12:53
dvdkwolfspraul: looks like a wiki, so only needs  signing up for an account :)12:53
wolfspraullet's get it into a OpenWrt or Jlime image first though12:54
dvdkmaybe wait a week or two, until everything stabalized.12:54
DocScrutinizerhostmode-easy-now, my project to put USB hostmode back into N90012:54
wolfspraulotherwise it's a bit of vaporware/wishful thinking12:54
dvdkwolfspraul: sure.12:54
dvdkwolfspraul: what i coded so far is tested with and packaged for our openwrt image.12:55
wolfspraulDocScrutinizer: well that's not too bad, at least some bits... how did you get that (which payment method)?12:55
kyakdvdk: i think '-demuxer ogg' option can be added to nanonote-files/script-files/root/.mplayer/config13:00
dvdkkyak: but can we then still playback mpg, avi etc.?13:01
kyakdvdk: you are seeking for disabling patented codecs at all, or leave it depending on PATENTED flag?13:01
kyakdvdk: i was referring to "Maybe we can patch mplayer to use that ogg demuxer by default?13:02
dvdkkyak: maybe depending on flag, if possible.  else rip out as much as needed.13:03
dvdkogg demuxer by default meens the mplayer 'ogg'  demuxer instead of ffmpeg's ogg demuxer13:03
qi-bot[commit] David Kühling: mplayer_jz47xx: upgrade to latest version w/ scaler support and some fixes http://qi-hw.com/p/openwrt-packages/de2140413:07
kristianpaulDocScrutinizer: your lucky paypal allow you receive donations, sadl not my case :/13:09
kristianpaul(vodo) looks nice, okay i just installed the firefox extention now..13:10
DocScrutinizerkristianpaul: believe me, I feel like it's not been worth the effort - and I actually *hate* paypal13:13
kristianpaulhe :-)13:15
DocScrutinizerwell, it came handy for paying some weird stuff for N900 I ordered in HK13:16
DocScrutinizerspare speakers for example13:16
kristianpaulyes paying is what all people is allowing to :-)13:17
kristianpaulI got my MM1 that way13:17
DocScrutinizerthe N900 tends to fry speakers due to the very good class-D amp13:17
viricMaybe I should try to get mplayer working in the nanonote too...13:17
kristianpaulviric: then and you dvdk and kyak can do a hackatoon ;)13:18
DocScrutinizer0..30kHz +/-0dB ;-D13:18
kristianpaul0? 0_o13:18
DocScrutinizersure, class-D13:18
DocScrutinizerthat's the problem13:19
DocScrutinizerthat's why Nokia delayed rollout of N900 - they had to invent xprot speaker protection in PulseAudio13:20
virickristianpaul: Maybe you mean '-ton', and not '-toon'? :)13:21
DocScrutinizerwell, actually INput is decoupled by 100nF/?10kR?13:23
viricOr even '-thon'13:23
kristianpaulviric: sure sure  :-)13:23
wolfspraulDocScrutinizer: how do you feel about Nokia/Meego/WinMobile and the future of phones at Nokia?13:34
DocScrutinizerdark sad story13:34
viricwhat is the state of GSM stacks usability in linux?13:37
viricIsn't there any 'arduino gsm module' around? :)13:38
wolfspraulDocScrutinizer: he, come on. That's not the final word :-) The future begins now.13:40
wolfsprauleven inside Nokia there will be some signs of life once it's becoming more clear that they hit the self-destruct button13:41
wolfspraulalthough I wouldn't count on that or wait for that13:41
DocScrutinizermeego dudes are rather agnostic about special system architecture needs for mobile battery-powered devices...13:41
DocScrutinizerNokia officially put meego on "strictly educational" status. No further plans for a product line based on it13:42
DocScrutinizermaemo been killed in favour of meego long ago13:43
wolfspraulyes ok but what's next...13:44
viricwhat's the next to fall?13:44
wolfspraulno what's the next best mobile device/smartphone to run free software on13:44
wolfspraulDocScrutinizer: any particular mode or line in mind at Samsung?13:45
kristianpaulyes plese tell!!, my 10 year old phone will die soon, i'll need buy a mobilephone soor or later :/13:46
DocScrutinizernope, nothing that's matching my taste13:46
DocScrutinizerI got 2 pcs N900, considering to get a 3rd one13:46
kristianpaulhmm, what GSM chips uses that N900?13:47
DocScrutinizermaemo's still alive13:47
wolfspraulkristianpaul: buy the cheapest 20 USD phone and let's continue with NanoNote, Milkymist, etc. if you want to go pro join the osmocom project13:48
wolfspraul:-) (I had to give that feedback, no? :-))13:48
viricwolfspraul: that's what I planned to do :)13:48
kristianpaulyeah, i tought that alredy :-)13:48
viricCan someone with arduino pieces make something close to a gsm phone?13:49
kristianpauljust somethimes, i miss mobile connectivity, may be i just need a gsm-3g modem dongle..13:49
virickristianpaul: sometimes I miss lack of connectivity13:50
wolfspraulwe'll get to the connectivity13:51
wolfspraulat least once we do, there will be something new there13:51
wolfspraulfor sure13:51
Action: kristianpaul is at gome13:51
kristianpaulviric: surely you can, but the point is not just buy something thats works byt it self (gms side), in the other hand better get something you can build/hack on :-)13:51
kristianpaularduino and his majority shield market is not that way to go i think..13:52
Action: kristianpaul will buy a 20usd phone with lighting :-)13:52
DocScrutinizerkristianpaul: http://www.gsmrapid.com/showthread.php?t=495713:52
wolfspraulI can't believe how the industry botched the whole Wi-Fi mesh story and potential. greed killed the mesh star.13:52
wolfspraulbut we need to do better, not complain...13:53
viricSome university teacher here should make their students do a project like "let's make a mobile phone"13:54
wolfspraulviric: no we are much further than that. specifically, they can join osmocombb or the related bunch of projects there.13:56
wolfspraulthey can advance the state of KiCad, they can work on essential tools like the initial boom sourcing system Werner did13:56
viricfine. that way then13:56
wolfspraulwait, I have some more13:56
wolfspraulthey can buy some Milkymist One and hack wireless basebands and DSP into it13:57
wolfspraulheck they can buy some N900 and help Joerg with usb hostmode, at least it's a device that works today :-)13:57
viricwolfspraul: how to get a wireless baseband? with gnu radio pieces?13:57
wolfspraulah, good point. another project.13:57
kristianpaulviric: i think thats is what wolfspraul means, http://bb.osmocom.org/trac/wiki/GsmDevelBoard (even to remennber my self)13:57
wolfspraulI think the key is to join existing projects, and incrementally improve those.13:57
wolfspraulrather than starting yet another big meta-project that will produce nothing but bloat and distraction after a while.13:58
DocScrutinizerwolfspraul: USB hostmode is pretty much final13:58
viricI've a friend, university teacher, that gives as homework to students to improve wikipedia articles13:58
viricsomeone on electronics should do something similar :)13:58
wolfspraulactually all the pieces are there, they are just combined in horrible ways (well, most pieces are there, some black spots arguably)13:58
kristianpaulDocScrutinizer: nice link, no usefull datasheets afaiks ;)13:58
wolfspraulviric: yes, perfect!13:58
DocScrutinizerkristianpaul: there are no datasheets for BB513:59
wolfspraulthat's exactly what I mean, indeed13:59
kristianpaul(combined in horrible ways ) oh yes13:59
viricIs GSM here to stay many years still?13:59
DocScrutinizerhmm, yup13:59
kristianpaulworld is slow,hardware costs money..13:59
wolfspraulhard to predict13:59
wolfspraulbut that kind of thinking may mislead you - gsm is here today, join osmocom, n900, etc. and do something today14:00
viricIf someone made an easyly hackable GSM device, it would be potentially forbidden to be sold to the public, isn't it?14:00
wolfspraulotherwise you think too strategically, always looking for the perfect entry spot, factor in inertia, new development, etc. etc. you will never stop thinking :-)14:00
wolfspraulthat's total nonsense, really14:01
viricah ok14:01
viricI remember hearing stories like this.14:01
wolfspraulsome kids believe in santa claus too, which is not a bad thing necessarily14:01
wolfspraulyou bet14:01
wolfspraulof course14:01
wolfspraulthe Internet is full of rubbish nowadays, isn't it :-)14:01
wolfspraulamazing to see the comment section on many serious news publications nowadays14:01
wolfspraulany insanity clinic certainly would have no problem looking for patients there14:02
viricwolfspraul: the standards for 'insanity' have changed along time ;)14:02
wolfspraulto keep it real: there is absolutely no worry in my mind that a 100% free phone would not meet regulatory requirements in all important markets14:03
wolfspraulin a licensed spectrum, let alone an unlicensed spectrum14:04
viricas you know chinese production...14:04
wolfspraulof course it's all work, many markets, many local jurisdictions, many powerful players who may create this or that entry barrier14:04
wolfspraulwelcome to the real world14:04
viricI've seen a thread about getting a Loongson3 computer out of china14:04
viricA chinese said: it's available, but not to be sold abroad, due to lack of certifications14:05
Action: kristianpaul have a loongsoon computer14:05
viricand people saying "I want one anyway"14:05
wolfspraulwhat is your question?14:05
viricwolfspraul: what is the chance that, nowadays, a product meant for the chinese market, meets european certifications, let's say?14:05
wolfspraulif you take a random 'china only' product, and submit it to CE and especially RoHS testing (long story), chances that it will not meet those are 90% at least14:06
wolfspraul_but_ the whole story is much longer14:06
kristianpaulyou should ask insted how by pass that regulations and get that item shippend to your home easilly ;)14:06
wolfsprauloften it may be only some small blip somewhere in a measurement, or difference in measurement, that will make a device pass CCC test (the Chinese one), but fail the equivalent CE test14:07
viricAnd a question related... I imagine it's much safer to wait for a product to pass the certifications, than to bypass them. At least, from my ignorance, this is the path I prefer now.14:07
wolfsprauland then the problem in a high-volume production is that any change is risky14:07
wolfspraulso if the product is already made in high-volume, and now someone wants to make 'a few' changes to pass CE, no matter how irrelevant that failure is, who will take the risk?14:08
wolfspraulso unless there is a strong sales partner in the EU, it will not happen14:08
wolfspraulRohS is another story, it is typically not checked, and the way RohS is enforced is such that EU agencies 'co-work' with importers and manufacturers 'towards' RoHS14:08
wolfspraulbecause in the end it is just impossible to be sure about every last connector, every cable, the print color on the box, etc.14:09
wolfspraulso if the Chinese manufacturer is willing to sign a RohS self-compliance document, good for you14:09
viricthe whole free market is optimised to economic benefit, and not to safety... so no wonder.14:09
wolfspraulif not, you have to sign it, and be ready to dig through the materials if something comes up one day14:09
wolfspraulnah, that's too cynical.14:10
wolfspraulLoongson is very political, it's the Chinese (government) answer to Intel.14:10
viricthis is as I understood it too14:10
wolfspraulso those are all big players, I have zero insights into the strategies that are going on around Loongson right now.14:10
wolfspraulbut you cannot see it so simple as in 'that's all crap and they will never pass the great EU standards'14:11
wolfspraul1) not everything in China is crap14:11
viricwell, I have two loongson2f and a NN, so I imagine they passed all the certifications :)14:11
wolfspraul2) the EU standards may by far not be as great as some people think they are, and they may not be enforced as diligently as some people think they should be14:11
viricsure sure.14:11
viricin some products we are using Lin Engineering motors, chinese, and I think they are around the best stepper motors around14:12
wolfspraulfor everything I manufacture, I definitely try to meet all CE, FCC, RohS, WEEE requirements14:12
wolfspraulbut it's ongoing work, that kind of stuff never stops14:12
DocScrutinizerkristianpaul: there's a 'leaked' schematic of N900, with full detail of BB5, and there's the ISI API specs, and ofono, to learn talk to BB5 chip14:12
wolfspraulalso actually those standards are moving targets, they are being edited, tightened, etc.14:12
DocScrutinizerkristianpaul: still insufficient to build own equipment based on BB5 chips, not to mention you probably can't buy any14:13
viricThere was a time where some software companies started to give out the source code of their games (ID software, Apogee, ...)14:14
DocScrutinizerwolfspraul: you know about zoom-II witch never came to market in the UMTS enabled variant, due to RoHS issues? :-D14:14
viricNo phone company is releasing all about their old phones? :)14:14
Jay7viric: it's very expensieve because of patents law14:14
DocScrutinizerthe 3G modules they planned to use was not compliant14:15
Jay7lawyers should be sure that they open no patented tech's14:15
viricJay7: how so? Patents are public designs, isn't it?14:15
Jay7main problem is cross-licensing14:16
viricI mean, the *already patented tech* should be that more easily publicable14:16
viricJay7: but I'm not understanding anything under 'cross-licensing problem'14:17
viricdvdk: what is the performance comparision between mplayer with your patches or without?14:17
Jay7viric: company A licensing some technology from company B to make some product14:17
viricJay7: not-patent-related you mean, rigth?14:18
Jay7then company A can't publish some things that affected by tech from company B w/o permit from B14:18
viricunless they agreed to publish that.14:18
Jay7and because of total patent's mess nobody can publish anything14:19
viricthat could be a good proof about the lack of public benefit from patents14:19
viric(if it is about patents)14:19
dvdkviric: you have a nanonote: download the package and test yourself :)14:23
dvdkwhat do you mean with 'patches'? the hw acceleration part?14:23
viricdvdk: yes14:24
viricdvdk: I don't use openwrt in it...14:24
viricdvdk: the same bbb.ogv video could be played, before your patches, in the nanonote?14:25
dvdkviric: that get the source and compile it for whatever you use :)14:25
dvdkviric: yes it could, however software-scaling to fullscreen was too slow.14:25
viricbut for a video meant for the nanonote screen size... thinking of no later scaling...14:25
dvdknot so sure about bbb320.ogv (saw my last mail?), here no scaling is required, only yuv->rgb conversion14:25
dvdkhad problems with 320 width video at 30 fps, that's why the scaler.14:26
dvdkthe updated mplayer uses ffmpeg's fftheora for decoding which should be fastre.14:26
viricfaster than what?14:26
dvdkviric: faster than using official libtheora14:28
dvdkffmpeg integrates better with mplayer (less copying data around etc.), also i guess that ffmpeg developers produced faster code.14:28
dvdkviric: which os are you using?14:29
dvdkon nanonote14:29
viricdvdk: "nanonixos" (cough)14:29
dvdkhu?  you mean your nn is a paperweight :) ?14:29
viricit works, boots, has mplayer, rss reader, browser, network, ... :)14:30
dvdkwell, if it has mplayer, it should be trivial to package the accelerated vidix  driver for it.14:31
viricI'll try to.14:31
dvdktell me if you need any help with that.14:31
viricwhat about that 'mplayer2' btw?14:31
dvdkso the vidix stuff currently consists of (1) the jz47xx_vid.so library (sourceforge), and a tiny patch to mplayer that makes it load the lib.14:31
dvdkviric: mplayer2 dunno.14:32
dvdkaway for a few minutes14:32
dvdkviric: interesting, this nanonixos.  didn't even here of nixos until now.14:39
viricBe welcome in #nixos :)14:40
dvdkviric: maybe once openwrt works :)14:40
viricyou will wait for the distribution to work, before switching to another? That's altruism ;)14:41
dvdkviric: stubbornness14:41
viricdvdk: I think only me uses nanonixos14:42
dvdkviric: it looks like you are the inventor of nanonixos :)14:43
viricright :)14:43
dvdkwrt mplayer acceleration: you need this tiny patch: http://projects.qi-hardware.com/index.php/p/openwrt-packages/source/tree/master/mplayer/patches/010-jz47xx-vidix.patch14:43
viricbut it's an easy thing, using nix.14:43
dvdkwhich should apply against any mplayer version from the last years14:43
viricah any?14:43
viricI did not expect something that easy.14:43
dvdkwait, now comes the second part:14:44
dvdkyou need to compile the jz47xx_vid driver https://sourceforge.net/projects/mplayer-jz47xx/14:44
dvdk(latest tarball 0.1.1)14:44
dvdkstandard autoconf/automake, cross-compiles without problems.  install the jz47xx_vid.so to /usr/lib14:44
dvdkthat's all14:44
dvdkcommand line for use is -vo cvidix -sreenw 320 -screenh 240 -fs (put it into the config file)14:45
dvdkjz47xx_vid is a library, no kernel patches required14:46
dvdki'm going to try to get jz47xx_vid into mplayer upstream, once it works stably.  until then i'm quite happy about that thing being a separate package, so i don't have to apply monster-patches whenever i rebuild mplayer14:47
viricdvdk: what do you think about http://www.mplayer2.org/comparison.html14:48
dvdkhey, they finally started to rewrite the mess that is named mplayer.  cool.14:52
viricand getting vlc on the nanonote? no chance?14:58
viricdvdk: the jz47xx thing builds just fine by nix14:58
dvdkviric: that.s the power of autoconf/automake :)14:59
dvdkviric: vlc is a little too bloated? lots of thread use etc. not good for a tiny cpu.14:59
viricah ok14:59
viricI did not know14:59
dvdkjust guessing.15:00
viricdvdk: I'll be using mplayer-snapshot-20101227. May it work?15:08
dvdkif the patch applies, you should be fine.  same patch worked for 1.0rc2 and for svn head15:09
viricI don't disable any codecs... :)15:09
dvdkviric: btw nix is not cross-compiled as far as i see.  so you compile on NN or using qemu-mips?15:12
viricnanonixos is cross compiled15:12
viricby nix15:13
viricnixpkgs allow cross compilation15:13
viricnix-build -A MPlayer   => builds locally15:13
viricnix-build -A MPlayer.hostDrv => cross builds15:13
dvdkok, so not a port of standard nixos, but kind of a new flavour15:13
viricnixos is not cross-copmiled.15:13
dvdkusing functional language for build specification is extremely nice15:13
viricbut nixpkgs, that nixos and nanoixos use, work not only cross-compiled, but they work locally compiled also in bsd, darwin, cygwin, ...15:13
viricso you can ask for 'vim' on darwin, cygwin, cross-build  to a nanonote, to a sheevaplug...15:14
viricwith a single specification, customisable: nixpkgs.15:14
viricnixos is a GNU/Linux distribution not cross build.15:15
dvdkread about it.  fixes many, many problems that come with openwrt and debian.  (especialy with autoconf doing things it wasn't configured to do, resulting in bad dependency infos etc.)15:15
viricsure sure. I know. :)15:15
viricI don't have to read more about it :)15:15
viricdvdk: http://vicerveza.homeunix.net/~viric/cgi-bin/nanonixos/ci/4f46b1528515:18
viricthere are some additional changes... but it should not be that relevant.15:18
dvdknot sure i can understand that :)  so jz47xx_id is part fo the mplayer package?15:19
dvdkdid you apply the patch15:20
viricjz4xxpatch is about the patch15:20
viricjz47xx_vid is the lib15:20
viricI've them apart by now. I'll use LD_LIBRARY_LOAD15:20
viricLD_LIBRARY_PATH I mean15:20
dvdkthat's better.  because the _vid.so can be upgraded often without relation to the mplayer pkg15:21
viricI'll write a mplayer sh wrapper that sets the LD_LIBRARY_PATH.15:21
viricbut now my computer will be building for a long time still :) I'm not building a nanonixos since nixpkgs dated months ago15:22
viricdvdk: If the _vid.so has no relationship with mplayer, it would be overcomplicated to make it build inside the mplayer nix expression :)15:25
viricdvdk: for a nixos user, additionally, it is very hard to run an openwrt toolchain, I think.15:28
Action: dvdk does not understand15:29
viricopenwrt supposes some /usr and things like that15:29
dvdkthe patch i saw is that multiple nix expressions, or just one containing all expressions for th efull build?15:29
viricdvdk: the patch also adds offrss, my RSS feeds reader15:30
dvdkviric: now i understand even less.  mplayer? offrss? huh?15:32
viricwell, I had some things in the checkout I committed alltogether.15:32
viricI thought you could understand what is related to your _vid and what not :)15:32
dvdkno way to browse the full nanonixos thing at ~viric/cgi-bin15:33
viricyes, click on Files15:33
viric(you need to login anonymously)15:33
dvdkah, only seoing 'Home Timeline Branches Tags Tickets Wiki Login '15:34
dvdkyou mean no login, no access to source?  going to register you with bugmenot.com :)15:34
viricthe login is to allow full access only to humans, and not to crawlers15:34
viricClick the login. Then "Fill automatically captcha", then Login again.15:34
dvdkah, ok.15:34
viricdvdk: the root file is default.nix, and the others come cited.15:35
dvdk5 lines for jz47xx_vid package?  where is the rest?15:35
dvdkalso mplayer package is sooo compact.  is that all?15:36
dvdkwhat are the options (          x264Support = false; etc.) referring to?15:36
viricdvdk: it requires the upstream nixpkgs.15:36
dvdkyou convinced me, going to migrate to nixos :)15:36
virichttp://vicerveza.homeunix.net/~viric/cgi-bin/nanonixos/doc/trunk/doc/usage.wiki :)15:36
dvdkyou mean RTFM?  not my strength :)15:37
viricYou can use nix+nixpkgs in any GNU/Linux distribution15:37
viricyou don't need to switch to start using it15:37
dvdkok, very interesting.  thanks.  people have been using build server extensions on debian to track dependencies.  Nix is much cleaner.15:37
dvdkyeah, don't worry, i cannot migrate yet, as openwrt is not yet working :)15:38
dvdkdon't have enough time on my hands.  still, good to know that makefile and per scripts are not the only options left.15:38
viricBut I mean... it should be trivial to install nix, get a copy of nixpkgs, and "nix-env -i MPlayer" for example :)15:38
viricSo don't imagine that you need to do a lot of work to start using ig15:39
viricAnd it works orthogonal to your linux distribution (all stored in /nix/... paths)15:39
dvdkok, it's just starting to climb upward on my todo list.15:39
dvdkwell, keep me posted how the mplayer thing works.  have to leave for some matters in meatspace.15:40
wpwrakhmm. ingenic specify the 4720 up to 240 MHz17:06
wpwrakif i toggle GPIOs as fast as i can, i get a square wave with a frequency of 6.588 Hz. so that's 13.176 Mchanges/sec17:10
wpwrak316/13.176 = 23.98, so the GPIO update rate seems to be CCLK/2417:12
wpwrakthe GPIOs are in the address range 0x10000000-0x10ffffff, so according to 4720pm section 24, they're on APB. APB is supplied by PCLK.17:16
kyakdvdk: trying to follow your instructions in "Nanonote video playback update", have no luck -\17:18
dvdkkyak: where's the problem?17:18
dvdkchecking the mail, maybe i gave wrong urls?17:18
kyakdvdk: i start the "mplayer bbb320.ogv -vo cvidix -screenw 320 -screenh 240 -fs -demuxer ogg", and then there is black screen17:18
wpwrakaccording to CPCCR (PDIV=2), PCLK = X1/3. X1 is not defined anywhere. we do know though that CCLK=X1. however, this still leaves the possibility that "X1" only has local significance, such as "input of the respective divider"17:18
kyakthough i hear some noise from speaker17:18
kyakdvdk: the urls all worked fine17:19
dvdkkyak: strange.17:19
dvdkcan you record the stdout/stderr using mplayer ... &> log17:20
wpwrakit we assume that X1 is global, that would make PCLK = 105.333 MHz. the GPIO update rate would thus be PCLK/8. seems sluggish.17:21
wpwraklet's see what else we know about PCLK ...17:21
kyakdvdk: http://dpaste.com/535032/17:21
kyakdvdk: i interuppt it myself17:21
dvdkworks here.17:22
dvdk md5sum /usr/bin/mplayer17:22
dvdk645ffaa1c871d339d3a2d6d62a9dbb11  /usr/bin/mplayer17:22
kyak# md5sum /usr/lib/jz47xx_vid.so17:23
kyak0deeaad9d69193776fdff3b731778e3f  /usr/lib/jz47xx_vid.so17:23
kyakif it matters17:23
dvdkfind / -iname "jz47xx_vid.so"17:24
dvdksame md5suum.  you notice it sead  0dead ? :)17:24
kyakyeah, noticed, it;s funy :)17:24
dvdkruning a shell-script with17:25
dvdkexec mplayer /data/video/bbb320.ogv -vo cvidix -screenw 320 -screenh 240 -fs -demuxer ogg17:25
kyakdo you have /root/.mplayer/config?17:25
kyaki.e. have you changed the default one?17:25
kyakdvdk: oh! it plays now!17:26
dvdklooks like default.17:26
dvdkkyak: hah.17:26
kyaki don't know what happened17:26
dvdkmaybe some race?17:26
kyaki left it for several tens of seconds17:26
dvdklet's see how far it gets :)17:26
kyaknow if i interrupt and start playing, it starts immediately17:26
dvdki guess it's the flash reading being slow or something (nand or sd-card?)17:27
kyakdvdk: btw, you should be able to change volume level with Volume keys on ben :)17:27
kyakit's in nand, yes17:27
dvdkin nand for me, too17:27
dvdkbtw the log says video codec 'theora'17:27
dvdkstrange, thought it wouuld use fftheora (i.e. ffmpeg) nowadays.17:27
kyakdvdk: when i rewind, there is that squeaky  noise17:30
dvdkyeah, the demuxer is maybe not too good :)17:30
kristianpaul(you should be able to change volume level with Volume keys on ben) oh, realy? since when.. ?17:30
dvdkok, I guess the old demuxer keeps the newer ffmpeg 'fftheora' driver from working.  that's a pitty.17:30
kristianpaulI need that code for a WIP i have with the ben to control a TV Tunner17:30
kyakkristianpaul: since always :)17:30
kyak# cat /usr/share/mplayer/input.conf17:30
kyakF12 volume -117:30
kyakF11 volume 117:30
dvdkwithout -demuxer ogg it should run faster, but memory overflows probably at some point17:30
wpwrakthe UART is also on APB. but is doesn't use PCLK, if the manual can be believed :-(17:30
kristianpaulkyak: you mena i can use that keys even on a tty, or is mplayer specific?17:30
kyakkristianpaul: i'm talking about mplayer17:30
kristianpaulah ok17:30
dvdkkyak: how do you like the video qualtiy?17:30
kristianpaulsorry  i misudenrstood ;)17:30
kyakdvdk: it's superb :)17:31
kyakdvdk: works very fast17:31
dvdkwell, theora-alpha encoder is incredible (used it for dvd ripping recently)17:31
kristianpauli love this map :_) http://www.nhm.ac.uk/research-curation/research/projects/worldmap/diversity/c1.htm17:31
dvdkok, comparing fftheora and libtheora, difference in performance is not too string.17:31
kyakdvdk: how to get rid of that blinking cursor?17:32
dvdkkyak: there should not be a cursor.  have you still &> logfile?17:32
dvdkit sends an ioctl() to /dev/tty to hide the cursor17:32
dvdkworks here17:32
dvdk(from ash, launched from gmenu2x)17:32
kyakah ok17:32
kyaki launch via ssh17:32
dvdkno, you're right there is a cursor at the bottom .  hmm.17:32
dvdkahh, same for me, launched from ssh for testing :)17:33
wpwraklet's see what happens if i lower HCLK ...17:33
kyakdvdk: btw, have no problems with alsamixer17:33
dvdkkyak: hmm.17:33
dvdkmaybe the newer mplayer is fixed and unmutes at startup?17:33
kyakactually, i don't remember having such problem at all17:33
dvdkkyak: no mplayer does not unmute.17:34
wpwraksystem no longer responds. sigh. not unexpected, though.17:34
dvdkmaybe problem with my gmenu2x config.  gmenu2x config once crashed on me, resetting screen brightness etc.17:34
kristianpaulwpwrak: playing with UART-SPI bridge with the atmega? :)17:34
wpwrakkristianpaul: naw, video out17:35
kyakdvdk: you are right, there is not cursor when i launch mplayer from tty17:35
wpwrakkristianpaul: we're about 5% too fast for CGA (pixel clock)17:35
kyakdvdk: the videos are pretty much unsrollable -\17:36
kristianpaulvide out? in the atmega??17:36
wpwrakkristianpaul: no, ben+ubb17:36
kristianpauloh !17:36
kyakdvdk: i mean, can't rewind or fast forward17:36
kristianpaulwpwrak: why you mentioned UART then ? i dont get it yet or was not related?17:37
dvdkthe ogg demuxer is flakey.17:38
dvdktry to rewind in larger steps (up/down or pgup/pgdown)17:38
dvdkwithout -demuxer ogg you have better rewinding, but memory is going to run out17:39
wpwrakkristianpaul: i'm trying to figure out how the clocks in the ben are really related. the documentation is not very helpful.17:39
wpwrakkristianpaul: in particular, i'm trying to find out if i could make the gpio update rate a little faster.17:40
wpwrakkristianpaul: not sure yet how i'll deal with the clock difference. one possibility would be to tweak the ben's system clock. another would be to insert delay cycles in the video timing. both approaches have their ups and downs.17:43
wpwraksigh. doesn't like it if i change PCLK either. neither up nor down.17:45
wpwrakof course, i could change it within my evil loop. that one shuts down most of the rest of the system anyway.17:46
kristianpaulevil loop :D17:47
viricdvdk: trouble building mplayer. grrr17:49
dvdkkyak: you're right seeking is pretty broken.  hmm.17:49
wpwrakkristianpaul: well, it disables the lcd controller and interrupts ... :)17:49
dvdkkyak: worked with earlier mplayer versions i tried, though.17:49
viricmipsel-unknown-linux-gcc -O -DCODECS2HTML -I. -Iffmpeg -o codec-cfg codec-cfg.cc17:50
viric./codec-cfg etc/codecs.conf > codecs.conf.hh17:50
viric/bin/sh: ./codec-cfg: cannot execute binary file17:50
viricMaybe a newer version fixes that.17:50
dvdkviric: configure --enable-cross-compile ?17:51
wpwraknope. no luck. even changing PCLK inside my loop kills the system.17:51
viriclet me test17:51
viricI was not using it17:51
kyakdvdk: never worked with any mplayer version. btw, it's same broken in jlime17:52
dvdkworks on my desktop, newer ogg skeleton helps with that.17:53
dvdkmaybe we should use mkv container?17:54
kyakis it better?17:54
dvdkmkv is much better than ogg17:55
dvdkseeking worked when i tested it yesterday with mplayer 1.0rc2 which had just this other ogg demuxing bug.17:59
kyakthis seeking bug is not always triggered18:00
kyakyou have to seek several times in a row18:00
kyaki even started to get used to it back then, need to give it 5-6 seconds to settle after i seek18:01
kyakthen i can seek again18:01
kyaki noticed that mplayer now ignores Volume keys -\18:01
dvdkdoes -nocache help?18:01
Action: viric never got softvol working....18:01
viricdvdk: Checking for VIDIX ... no18:02
viricis this good or bad?18:02
dvdkah, ok, you need to enable some otpions..18:02
dvdkcan you lookup the mplayer/Makefile in openwrt-packages.git on qi-hw?18:03
viricgrmbl. It still makes the 'cannot execute binary file'18:03
viriclet me see.18:03
dvdk --enable-vidix \18:03
dvdk  --disable-vidix-pcidb \18:03
dvdk  --with-vidix-drivers="no"18:03
kyakdvdk: it hits oom with -nocache18:03
viricMay I ask... what is 'vidix'?18:04
dvdkno i mean -nocache -demuxer ogg18:04
viricdvdk: why disable that pcidb?18:05
dvdkno pci bus18:05
viricah, it's nanonote specific.18:05
kyakdvdk: it's better.. kind of.. still this sqeaky noise, but at least it's not stuck at it and i don't have to reboot Ben18:05
dvdkkyak: as soon as you test, everything stops working.  stop that :)18:06
dvdkviric: some more config opts:  --target=mips \18:07
dvdk  --disable-mencoder \18:07
dvdk  --disable-pthreads \18:07
dvdk --disable-x11 \ disable-xv \ disable-vm \ disable-vdpau \ disable-gl \ disable-xf86keysym18:07
kyakdvdk: :)18:08
wpwrakah, cool. if i add memory accesses, i get exactly the right frequency very convenient ;-))18:09
viricdvdk: codec-cfg$(EXESUF): codec-cfg.c codec-cfg.h help_mp.h $(HOST_CC) -O -DCODECS2HTML -I. -Iffmpeg -o $@ $<18:10
viricit's clearly broken18:10
viricI'll try a newer snapshot.18:10
dvdkwhy it works w openwrt?18:11
viricI've no idea.18:11
viricmaybe they fixed it. let's see.18:11
dvdkviric: i thought you'd start with a pre-existing nix package for nanonote, only adding the patch.18:13
dvdkof course porting from scratch for cross-compiling might become a PITA18:14
viricI'm seeing maybe I never built mplayer before :)18:14
viricbut I'll fix it ;)18:14
kyakdvdk: will you commit the Makefile?18:18
dvdkcan i?18:22
kyakwhy can't you? :)18:26
viriccan ffmpeg have both fftheora and libtheora at once?18:37
kristianpaulwpwrak: you can use the rtc clock as timer or something no?18:37
wpwrakkristianpaul: what use do you have in mind ?18:39
kristianpaulwpwrak: you need a clock divider right? you can use the..hmm rtc clock have a timer? :P18:40
wpwrakkristianpaul: oh, but i need a fast clock. VGA pixel clock is 25.175 MHz18:41
wpwrakgrmbl. having to wait for memory sucks. maybe i can squeeze a prefetch into hsync ...18:45
dvdkkyak: thought i shouldn't commit until patent codec problems are checked18:53
dvdkok, have to go for now.18:53
dvdkkyak: btw the patch i send on the mailinglist is up-to-date and corresponds to my Makefile.18:54
wpwrakblargh. 12 us and i have only 3.77 us :-(18:54
viric_wpwrak: what are you doing?18:55
wpwrakviric_: adding VGA output to the ben18:57
viric_aah ok18:58
viric_through UBB18:58
wpwrakyeah. 31.6 us a line (without hsync) and i have a budget of 34.35 us19:03
wpwrakerr no, wait19:03
wpwrakdarn. i don't :-(19:03
viric_I remember something like 14KHz19:03
wpwrakonly 31.77 us19:03
wpwrakand without hsync, only 28 us. grmbl. too slow.19:04
viric_it's like raw VGA video at 60fps through SDIO, almost19:05
wpwrakviric_: (14 kHz) sounds like TV ?19:05
viric_I think 14kHz is the line frequency19:06
wpwrakway too slow. well, CGA perhaps.19:06
wpwrakVGA has 528 lines (including the invisible ones)19:07
wpwraknot sure if modern monitors accept CGA vertical timing. in any case, the difficult one is horizontal timing. vertical is just more memory ;-)19:08
wpwrakgood. the memcpy to cache is done in 2.82 us. i have a budget of 3.77 us.19:11
wpwrakah .. wait ... let's do this a little smarter ...19:15
wpwrakintersting. a mere read takes longer than a memcpy. hmm. probably something not too good with the loop check ...19:18
wpwrakoopsie, killed the system ...19:19
wpwrakstill slow :-(19:21
wpwrakah, we can of course go a little faster ...19:22
viricwpwrak: some fun for you19:25
wpwrak2.8 us. good. and i could probably make it even faster. the 34.6 us in a 31.77 us time slot are still there, though. hmm.19:30
Action: roh doesnt get what all that complexity should be for19:38
rohi mean.. vga out? seriously? what a waste of time19:38
wpwrakhmm ... lbu $5,... followed by andi $5,$5,0xff  that's kinda stupid19:38
wpwrakgrmbl. but does't get faster if i hand-optimize it.19:44
wpwrakand if i pre-compute the whole word to store, memory bandwidth makes the prefetching too slow19:57
wpwraksigh. need to interpolate the pixels then.20:01
viricfinally I built mplayer...20:53
virickyak: were you running mplayer properly before? I don't have much luck...21:19
viricit runs super-slow here.21:23
viric30% of CPU decoding an OGG audio track22:19
wpwrakgrmbl. my monitors don't like my beautiful vga signal. bastards.22:23
andy753421wow, there's more people here than i expected22:30
andy753421what the difference between usbboot, zbboot, and jzboot?22:32
viricI only use usbboot22:35
viricbut I'm a bit outdated on that22:35
andy753421none of them work with my hardware.. so i need to know which one to try to fix :)22:35
viricwhat is your hardware?22:38
andy753421it's a cheap ebook, the cpu is a jz475522:39
andy753421i need to take it apart again because i forgot to check what the flash chip was :(22:39
viricI don't think it's generic code.....22:39
andy753421the windows version of usbboot that ingenics provides can boot it and access the flash, but there doesn't seem to be a firmware for the jz475x in the linux port22:40
viriclarsc knows well some jz47xx chips22:46
kristianpaulandy753421: hanvon ebook?23:00
andy753421kristianpaul: no, it's a crazy generic one with no identifiable markings, i found a link on some bulk-purchase site for something similar though, hold on23:02
andy753421(it was given to me for free)23:02
andy753421$60 :) http://www.alibaba.com/product-gs/439080383/Private_model_7_ebook_with_WIFI.html23:03
kristianpaulbut i see you take apart in order to indentify cpu was jz475523:03
kristianpaulah, CPU::  JZ4755 400HZ23:04
viricit's not with eink?23:04
kristianpaulHZ.. ;)23:04
viric400Hz has to be annoying in the ears, if it couples to the internal spekaer :)23:05
andy753421it's a generic LCD, really not that much use as an e-book, but it should make a good portable movie player23:05
andy753421(assuming i can update the software to not be horrible ;)23:05
viricIt can't play movies now?23:06
viricVa, bona nit a tothom.23:07
andy753421it can, but the a/v tracking slowly gets skewed somehow, so after about ~10 minutes it starts to get pretty noticeable23:07
andy753421it can't connect to network shares either at the moment, which i would really like as well23:07
kristianpaulandy753421: got indentified serial port already?23:08
andy753421there's a usb port, which i can connect to with the recover boot-mode23:09
andy753421(i have to press the `back arrow' and hit reset, then it goes to recover mode,etc)23:09
kristianpaulhmm but is that actually give a uboot promtp?23:09
kristianpaulusb port emulates a serial port?23:10
andy753421kristianpaul: oh no, i don't think there's any way to do that, but i can access the flash chip that way23:10
andy753421i don't think there are any actual serial ports brought out on it, unless there's some trick to doing it with the usb port23:11
kristianpauli cant find 4755 pinout, but if is same as 4725 i think there is23:11
andy753421kristianpaul: the processor supports it, but i don't think the board is set up for it23:12
kristianpauli dount that but i let the question open23:12
kristianpaulat least you find the pads for selecting the boot mode?23:13
kristianpaulpin 28 and 33 is UART123:13
andy753421i haven't tried to do anything directly, here's the data sheet for the cpu though: http://andy753421.ath.cx/temp/Jz4755_ds.pdf23:15
kristianpaulhmm that lined sqaure in the right down corber could be the seria port pad..or.. i cant see very well in the pic23:18
kristianpaulin any case you could try solder some wires directly to the soc been carefull not take too long or internal gold wires may die23:19
kristianpaulhave fun !!!23:19
kristianpaulopeninkpot project will be gratefull at least if you publish some full res pic for the whole PCB and Display23:21
kristianpauli think23:21
kristianpaulalso your findings etc..23:21
kristianpaulandy753421: i think dingux uses same chip you have but not sure at all..23:29
andy753421kristianpaul: i can try to take some more pictures, but i don't have a very good camera23:32
andy753421(those are `full res' ;)23:32
andy753421oh, there's 0, 1, and 2, pictures too if you hadn't noticed: http://andy753421.ath.cx/temp/ebook-2.jpg23:33
wpwrakheh, mystery solved. hsync must keep on running also when there's no line to display.23:33
kristianpaulandy753421: yes i asaw it, but no too much res as i want23:34
andy753421kristianpaul: ah, that's the best resolution i can do, but i'll see if I can get some better light, and i could take some close-ups if you want23:35
kristianpaulclose up nea the right down corner around the soc is calling my atention :-)23:36
andy753421i'll see what i can do, i've got to head in to work for a couple hours right now though23:38
wpwrakgood. now i get my test pattern on all my three monitors (xenon, samsung, lg). they don't keep sync, though. wish i had an analog monitor :)23:40
Action: kristianpaul still having his trusty compaq MV500 monitor23:44
wpwraktrusty or rusty ? ;-)23:48
kristianpaulwell, it run mm1 patches very well, of course it is not white as in 1995 but it looks resonable good :D23:50
--- Sun Apr 24 201100:00

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