2011-10-05 01:08 http://moxielogic.org/blog/?p=537 2011-10-05 01:27 linuxperv: :D 2011-10-05 01:27 :) 2011-10-05 01:31 the milkymist does have usb 2011-10-05 01:31 well there you go 2011-10-05 01:44 paroneayea: hey, you come here :-) welcome! 2011-10-05 01:45 kristianpaul: what is the moxie processor? 2011-10-05 01:53 "The Moxie processor is a general purpose load-store processor." 2011-10-05 01:53 I jsut had been following, looks itneresing project, they also have a WIP SoC 2011-10-05 01:54 i think you can ask antgreen at #milkymist for details 2011-10-05 01:54 There is a wiki http://moxielogic.org/wiki 2011-10-05 01:58 not much in the wiki at first glance 2011-10-05 01:58 is it all freely licensed? 2011-10-05 01:58 where are the sources? 2011-10-05 01:58 should we add the blog to the qi planet? 2011-10-05 02:01 blog yes, why not 2011-10-05 02:01 sources https://github.com/atgreen 2011-10-05 02:03 license seems gplv2 for HDL 2011-10-05 03:14 The build was successfull, see images here: http://fidelio.qi-hardware.com/~xiangfu/compile-log/openwrt-xburst.full_system-10032011-2306/ 2011-10-05 03:18 kristianpaul: ok interesting! 2011-10-05 03:19 how could this project relate with Milkymist? 2011-10-05 03:36 well, i just see another SoC coming,so is an alternative, but in early state still it seems 2011-10-05 06:34 i don't understand why moxie.  doesn't opencores.org have lots of those? 2011-10-05 06:36 wpwrak: ping 2011-10-05 06:38 blogic: phew. back home from dinner with quite a lot of beer. let's see how useful i'm now :) 2011-10-05 06:51 haha 2011-10-05 06:57 wpwrak: the story goes like this ... i was contatced by some old friends that are playing with 802.15.4. the aim is to integrate it in owrt and build some demo apps 2011-10-05 06:57 blogic: do you have any specific hardware in mind ? 2011-10-05 06:57 i saw that there are patches sent upstream to support the 802.15.4 inside the kernel and then went on to find hw 2011-10-05 06:57 yes, that is where you come into the play :) 2011-10-05 06:58 hehe :) we have the finest hardware :) 2011-10-05 06:58 i browsed around and saw the nanonote has the wpan dongle 2011-10-05 06:58 so i ma told 2011-10-05 06:58 anyhow i know mirco and lars quite well and the pinted me your way 2011-10-05 06:58 *pointed 2011-10-05 06:58 (well, actually we don't. but that's a good marketing message nevertheless. and the other may be harder to put to use than ours anyway :) 2011-10-05 06:58 hehe 2011-10-05 06:58 is the dongle already on sale ? 2011-10-05 06:59 i saw some pics i think on your web site of a large batch of pcbs 2011-10-05 06:59 yup. both - the one for the ben and the one for usb 2011-10-05 06:59 got a link ? 2011-10-05 06:59 not a lot have been made, but i think tuxbrain would be more than happy to tell the fab to produce some more 2011-10-05 07:00 ok 2011-10-05 07:00 do you know whom at tuxbrain i oculd contact ? 2011-10-05 07:00 the project overview page is here: http://downloads.qi-hardware.com/people/werner/wpan/web/ 2011-10-05 07:01 the tuxbrain shop page is here: https://www.tuxbrain.net/shop/index.php?cPath=347 2011-10-05 07:01 404 2011-10-05 07:01 looks like the server is down 2011-10-05 07:01 which one ? 2011-10-05 07:02 tuxbrain is almost one man: David Reyes Samblas Martinez 2011-10-05 07:02 tuxbrain.net 2011-10-05 07:02 works from here 2011-10-05 07:02 hmmm 2011-10-05 07:03 maybe your government is at war with spain ? :) 2011-10-05 07:03 i am based in germany 2011-10-05 07:04 there are no spanish troops here and if german troops were in spain the tv would probably tell me 2011-10-05 07:04 :D 2011-10-05 07:04 hmm. well, spain is part of the PIGS/PIIGS. so in a way, yes. 2011-10-05 07:04 dont mention the war ! :) 2011-10-05 07:04 (-:C 2011-10-05 07:05 No match for "WWW.TUXBRAIN.NET". 2011-10-05 07:05 >>> Last update of whois database: Wed, 05 Oct 2011 07:02:45 UTC <<< 2011-10-05 07:05 that is what whois says 2011-10-05 07:05 ok started working again 2011-10-05 07:05 weird 2011-10-05 07:05 hi 2011-10-05 07:05 that's normal for whois. but it should know tuxbrain.net 2011-10-05 07:06 whee ! :) 2011-10-05 07:06 hi 2011-10-05 07:06 you summoned the master himself :) 2011-10-05 07:06 good 2011-10-05 07:06 tuxbrain: aloha 2011-10-05 07:06 what must i do to get some wpan dongles ? :D 2011-10-05 07:06 buy them? :P 2011-10-05 07:06 how many do you need? 2011-10-05 07:07 not sure yet 2011-10-05 07:07 probbaly 5 or so 2011-10-05 07:07 i just need some hw to integrate this stuf finto openwrt 2011-10-05 07:08 so i just need 2 2011-10-05 07:08 :) 2011-10-05 07:08 not sure what the other guys need 2011-10-05 07:08 but now that i know the link and price and so on i can move along with the planning 2011-10-05 07:08 is the firmware used int he atmel also open ? 2011-10-05 07:09 sure :) almost all wpwrak work is on his repo :) 2011-10-05 07:09 good 2011-10-05 07:09 i also saw some imx based hw, but it looked like some closed source stuff 2011-10-05 07:11 so the atben is not a sdio card i assume 2011-10-05 07:11 and i can hook it up to some other hw that has enough spare gpios i assume 2011-10-05 07:12 i don't know about imx. but there's a somewhat smarter chip. pretty nice actually. but in an evil LGA package (LGA = BGA with a twist or two) 2011-10-05 07:12 atben needs GPIO bit-banging 2011-10-05 07:12 ok 2011-10-05 07:12 lga is not that bad 2011-10-05 07:12 :) 2011-10-05 07:12 let me dig it out ... 2011-10-05 07:13 it's the MC1322x 2011-10-05 07:14 yes 2011-10-05 07:14 not imx 2011-10-05 07:14 got confused there 2011-10-05 07:14 needs to dig out his eu vatid so he can order 2011-10-05 07:16 the MC13224 data sheet is here: http://www.freescale.com/files/rf_if/doc/data_sheet/MC1322x.pdf 2011-10-05 07:16 the good stuff is on page 52 ;-) 2011-10-05 07:16 ok 2011-10-05 07:17 it does have a pretty little ARM7 though 2011-10-05 07:17 but anyway, what we here have it a bit simpler, with the idea that the host does most of the work 2011-10-05 07:18 yes 2011-10-05 07:18 i think i'll just start with ordering 2 usb dongles 2011-10-05 07:21 ok thx for the info ! 2011-10-05 09:11 [commit] Maarten ter Huurne: media: radio: RDA5807: Configure inputs and outputs using platform data. (jz-3.0) http://qi-hw.com/p/qi-kernel/9034541 2011-10-05 09:11 [commit] Maarten ter Huurne: media: radio: RDA5807: Print fatal problems using dev_err instead of dev_warn. (jz-3.0) http://qi-hw.com/p/qi-kernel/26e4716 2011-10-05 09:11 [commit] Maarten ter Huurne: MIPS: A320: Specify platform data for RDA5807. (jz-3.0) http://qi-hw.com/p/qi-kernel/40201cd 2011-10-05 09:26 wpwrak: still around ? 2011-10-05 09:26 wpwrak: do you use kicad ? 2011-10-05 09:32 you bet he does, he augmented it severely to make it fit for that purpose 2011-10-05 09:33 actually i asked rather then bet 2011-10-05 09:33 but thanks for th einfo 2011-10-05 11:23 blogic: DocScrutinizer would have won the bet for you :) yes, i use kicad 2011-10-05 11:24 ok 2011-10-05 11:24 and yes, there's quite a bit of scripts and things that make it work better for out purposes :) http://projects.qi-hardware.com/index.php/p/eda-tools/ 2011-10-05 11:24 and components/footprints for kicad: http://projects.qi-hardware.com/index.php/p/kicad-libs/ 2011-10-05 11:24 ok 2011-10-05 11:24 moo wpwrak :-D 2011-10-05 11:25 whois DocScrutinizer 2011-10-05 11:25 DocScrutinizer: oink :) 2011-10-05 11:25 lol 2011-10-05 15:12 wolfspraul: I missed your welcome, but thanks :) 2011-10-05 15:26 no problem, of course 2011-10-05 15:26 happy to see you here 2011-10-05 15:27 linuxperv is also a pretty great person who does a bit of stuff in the maker / electronics space and was getting interested in the milky mist 2011-10-05 15:27 her joining was a reminder that I should probably join this channel also :) 2011-10-05 15:30 nice - any links to linuxperv's projects or areas of interest? 2011-10-05 15:35 "unixsluts.com" nice ;-) 2011-10-05 15:41 oh hi 2011-10-05 15:41 no i don't have much on the web and don't do anything too exciting ;p 2011-10-05 16:55 hi all 2011-10-05 16:55 wpwrak, did you take a look to this http://www.openscad.org/ ? 2011-10-05 16:56 i remember you did a sw for the 3D scan of the Ben case, isn`t it? 2011-10-05 17:02 (openscad) yes, i experimented a bit with it. i like the approach, the speed at which it works, and also the "IDE" 2011-10-05 17:03 what i don't like about it is the still very limited set of functions, the rendering artefacts in the "quick preview", and the poor handling of ambiguities in the model 2011-10-05 17:04 i didn't make software for the actual scanning, but i made a tool for bringing a top and a bottom scan into alignment 2011-10-05 20:54 Damn I'm angry at Apple 2011-10-05 20:54 iOS 5 looks so cool with it's voice recognition that understand what you mean 2011-10-05 20:54 Why don't I have the same thing under Linux ? 2011-10-05 20:55 It would be like the future is now :P 2011-10-05 20:55 because at Apple, they're paid 2011-10-05 20:55 Aren't Linux developers paid also ? 2011-10-05 20:57 5% of them, maybe 2011-10-05 20:59 Hum 2011-10-05 20:59 Just a moment 2011-10-05 20:59 I've read an article about that some time ago 2011-10-05 20:59 Here you go Ayla http://apcmag.com/linux-now-75-corporate.htm 2011-10-05 21:00 that's kernel though 2011-10-05 21:00 I think desktop is still mostly volunteers 2011-10-05 21:00 which also explains why the desktop is in a relatively poor shape 2011-10-05 21:01 Yep I made a misconseption 2011-10-05 21:01 It really is a shame imho 2011-10-05 21:01 and that also explains why I feel Android is so great :D 2011-10-05 21:01 Android sucks 2011-10-05 21:02 (imo) 2011-10-05 21:02 at least with Linux you can run apps that are not signed; iOS won't let you do that 2011-10-05 21:02 Yeah... But Java, seriously ? 2011-10-05 21:02 On a phone ? 2011-10-05 21:03 Java has been running on phones for ages: J2ME 2011-10-05 21:03 the main problem with J2ME was that all phones had different bugs 2011-10-05 21:03 what's wrong with Java? 2011-10-05 21:03 Yeah and therefore been criticized for ages 2011-10-05 21:03 Virtual machines 2011-10-05 21:03 That's the whole problem 2011-10-05 21:04 I have yet to experience a non-sluggy Java application 2011-10-05 21:04 I'm not joking 2011-10-05 21:04 the problem is not Java or virtual machines 2011-10-05 21:04 the problem is how those Java apps are coded 2011-10-05 21:05 Java have his part in the process 2011-10-05 21:05 As well as some other languages have other problems 2011-10-05 21:06 I just happen to hate Java more than the others 2011-10-05 21:06 (maybe not as much as C# though) 2011-10-05 21:22 Sorry, I disagree. 2011-10-05 21:22 and Android's virtual machine seems to be quite good 2011-10-05 22:03 LunaVorax: you anger is good. the more you hate a problem, the more likely you are to solve it :-) 2011-10-05 22:04 Hahaha 2011-10-05 22:04 Probably not, I don't use to be a smart person. 2011-10-05 22:05 wpwrak, but I would just head down on similar software for the Nanonote :P 2011-10-05 22:05 no, really. the motivation for almost anything that is done well is some deep discomfort with the status quo ante 2011-10-05 22:06 Oh I see 2011-10-05 22:06 wpwrak, an unrealted serious question 2011-10-05 22:07 Is wanting to work at Qi a good goal in like ? 2011-10-05 22:07 what you need is what is called "an itch to scratch". something that fills you with an irresistible desire to solve the problem 2011-10-05 22:07 there is not much of "working at qi". qi is just some 3-4 people, not a big company 2011-10-05 22:08 That's what I wanted to know in fact 2011-10-05 22:08 If in some future there could be some place 2011-10-05 22:08 but doing the kind of work that's happening here, be it for qi or in some other way, seems very reasonable goal to me 2011-10-05 22:09 Ok so I'm on the right road already 2011-10-05 22:09 A long one but still the good one 2011-10-05 22:09 i don't know if qi will ever be a big employer. but the movement is not limited to qi. 2011-10-05 22:12 Maybe it needs a big income boom to think about it 2011-10-05 22:14 do you guys use Dosbox on your nanonote? 2011-10-05 22:14 of course, having more money will help to move things forward 2011-10-05 22:14 and if so, are you using Slaaneesh's statically linked and closed source port? 2011-10-05 22:14 I tried to and I failed Ayla it seems to have some troubles with the screen res 2011-10-05 22:14 yes, same here, on dingoo ... 2011-10-05 22:15 Ayla: me, no. gave up on DOS some 20 years ago :) 2011-10-05 22:15 Ayla, indeed :) 2011-10-05 22:15 it wants to use a 640x400 screen resolution 2011-10-05 22:15 wpwrak: what about your "msdos" driver then? :p 2011-10-05 22:16 1992 ;-) 2011-10-05 22:16 fair enough :) 2011-10-05 22:16 and i didn't need DOS for it ;-) 2011-10-05 22:16 wpwrak, I think Qi needs better advertisment but it's kind of hard since Qi's product aren't really for "everyone" for now (exemple, I don't really see my brother buying a Nanonote) 2011-10-05 22:17 you must be old 2011-10-05 22:17 I bet you have a big beard 2011-10-05 22:17 LunaVorax: oh yeah, advertizing is weak. 2011-10-05 22:17 Beware, MS-DOS fanboys still exist 2011-10-05 22:17 like every linux guru 2011-10-05 22:17 Ayla: naw, i shave almost each time i crawl out of my cave 2011-10-05 22:17 Linus doesn't have a beard 2011-10-05 22:18 and I get laid everytime I reboot my linux computer 2011-10-05 22:18 considering that linux needs rebooting almost never, i kinda feel sorry for you :) 2011-10-05 22:19 and do you actually crawl out of your cave from time to time? :p 2011-10-05 22:20 wpwrak, don't even reboot when you recompile the kernel ? :o 2011-10-05 22:20 maybe he uses kexec 2011-10-05 22:20 Oh, I don't know that 2011-10-05 22:20 ...yet 2011-10-05 22:20 Ayla: oh yes, often enough to make the unbearding experience not too dreadful :) 2011-10-05 22:21 ok :) 2011-10-05 22:21 Ayla: (kexec) i actually like clean reboots, at least during development. but yes, being one of the kexec lobbyists, that would indeed be a choice :) 2011-10-05 22:23 about kexec 2011-10-05 22:23 I've been working on a bootloader for the JZ4740 2011-10-05 22:24 the goal is to have a small (< 8k) bootloader which can then load a linux kernel from a UBI drive 2011-10-05 22:24 or better, to launch kexecboot 2011-10-05 22:26 I thought you'd like to hear about that 2011-10-05 22:55 does anyone possibly know a place where one could talk with an experienced USB developer? 2011-10-05 22:55 I'm trying to de-proprietarize an h.264 HD webcam 2011-10-05 22:57 whitequark: you could pm 'CareBear\' (Peter Stuge, one of the libusb/coreboot devs) 2011-10-05 22:58 steve|m: thanks! 2011-10-05 22:58 but sniffing what the windows driver does and then trying to replay it with a small libusb-application and see what happens is a good start 2011-10-05 22:58 steve|m: O 2011-10-05 22:59 oops 2011-10-05 22:59 I'm currently doing that, but it does several hundreds of control transfers at init 2011-10-05 22:59 (I'm sniffing WinXP machine under virtualbox, by the way) 2011-10-05 23:00 I am able to capture a valid H.264 stream from the webcam, but it has one weird property 2011-10-05 23:00 each frame is compressed to ~12K (yes, kilobytes) of data 2011-10-05 23:00 they contain a very distorted, but actual image 2011-10-05 23:01 hmm.. some sort of RLE? 2011-10-05 23:01 steve|m: nope 2011-10-05 23:01 it's h.264 2011-10-05 23:01 but compressed with a ratio of maybe 99% 2011-10-05 23:02 there are 7 extension units defined in this camera, of which 4 are also present on non-h264-capable ones (checked by googling up lsusb -v's) 2011-10-05 23:02 so, other three (which are only defined on my exact webcam) somehow tune the encoder 2011-10-05 23:03 they contain a total of 19 vendor-specific controls. 2011-10-05 23:04 does the h264 data you captured on windows have the same properties? 2011-10-05 23:05 steve|m: well. the only app known to capture h264 from this camera is Skype (it's skypecam, moreover, it is originally intended for use with skype TVs), and it crashes on my VM by some reason 2011-10-05 23:05 but it is known to work on a real machine 2011-10-05 23:05 I cannot capture h264, however, as only Skype knows how to handle this webcam 2011-10-05 23:06 ah okay.. so it can't be used as a WDM camera device on windows? 2011-10-05 23:06 it's a valid UVC cam 2011-10-05 23:06 it supports MJPEG, YUYV and H264 2011-10-05 23:06 mjpeg and yuyv work perfectly on both windows & linux 2011-10-05 23:06 I see, so only h264 doesn 2011-10-05 23:07 Windows itself does not even try to initialize h264 stream, as that is an extension or something like that 2011-10-05 23:07 't work.. 2011-10-05 23:07 well. 2011-10-05 23:07 there is a standard on H264, but it is broken 2011-10-05 23:07 so they just took the way MPEG streams are presented, changed first four bytes (they're fourcc by convention) in GUID to H264 2011-10-05 23:07 maybe the ffmpeg/vlc guys could help you 2011-10-05 23:07 and that's it -- now they have an "h264" webcam 2011-10-05 23:07 the problem isn't with the stream, actually 2011-10-05 23:08 it's with the way to tell the camera to compress the stream with a given ratio 2011-10-05 23:08 stream is fine, it can be played by mplayer e.g. 2011-10-05 23:08 I just concatenate all the frames and that's it 2011-10-05 23:09 I should somehow snoop the URBs which are setting XU parameters on Windows, but to date I'm unable to filter them from all the requests stream 2011-10-05 23:10 which tool are you using? device monitoring studio? 2011-10-05 23:10 is it possible the H.264 encoder has a configurable maximum bitrate and it's set extremely low by default? 2011-10-05 23:10 currently I start WinXP in virtualbox, load usbmon module on Linux host and launch wireshark there 2011-10-05 23:10 mth: that's it 2011-10-05 23:10 I think it is exactly the problem 2011-10-05 23:11 ah yes, wireshark 2011-10-05 23:11 now I "just" need to determine which of the 19 (at least) vendor-specific parameters with an unknown type control that. 2011-10-05 23:12 can you selectively block them between the VM and the hardware? 2011-10-05 23:12 mth: I think no. EHCI controller in VirtualBox is a blob 2011-10-05 23:13 and between VB and Linux? is there only monitoring there or also filtering? 2011-10-05 23:13 only monitoring, VB uses a kernel module to forward requests 2011-10-05 23:14 ah, that's a pain 2011-10-05 23:14 steve|m: so, Device Monitoring Studio is an analogue of usbmon+wireshark for Windows? 2011-10-05 23:14 mth: even if there were some filtering capabilities, I am still unable to determine which requests go to XUs 2011-10-05 23:15 there are over 800 of them in the init sequence. I'll get mad knocking them off one by one 2011-10-05 23:15 not one by one, ideally you should be able to filter out about half 2011-10-05 23:15 whitequark: yes 2011-10-05 23:16 is there a client setting for image quality maybe that changes the max bitrate? 2011-10-05 23:17 I mean, if they go through the trouble of making it configurable, I'd expect there could be a way to change the value at runtime 2011-10-05 23:17 mth: it is adjusted by Skype automatically according to network conditions 2011-10-05 23:17 that may be exploitable, yes 2011-10-05 23:18 even funnier, Skype actually has a nice spec for encoding webcams on their site 2011-10-05 23:18 of course, my one is not even nearly conformant 2011-10-05 23:21 can you do record and playback of the full init sequence? 2011-10-05 23:22 if so, you could also stop halfway and check if the bitrate config was already applied 2011-10-05 23:24 mth: I don't think there is an easy way (i.e. not manual) of replaying a recorded usbmon dump 2011-10-05 23:25 steve|m: that looks affordable. I think my employer may get it for me. 2011-10-05 23:33 okay, thanks for the suggestions. I'll try these tomorrow