#milkymist IRC log for Sunday, 2011-06-19

kristianpaulhe, i almost forgot, now that minimac core dont use DMA i can port it to the avnet, but now i need a PHY.. wich i dont wish to import :-)00:13
wpwraklekernel: (clang) interesting. it's always good to have a second compiler to try things with.00:50
lekernelhttp://okcon.org/2011/programme/open-science-open-data-open-minds "Welcome to the future of recreational drug use."08:40
wpwraklekernel: (recreational drugs) love it ;-)10:27
lekernelwolfspraul, NTSC works by the way13:06
lekernel(from the wiki "Support for NTSC, SECAM, component video, or even multiple multiplexed signals may be possible but is currently unsupported.")13:06
lekernelcomponent video and s-video require an additional little bit of software support, but is definitely possible13:07
lekernelI tested a NTSC source a while ago13:07
wolfspraulI thought you said earlier it's hard?13:07
lekernelI'm talking about the video _input_ right now13:07
wolfspraulah ok13:07
lekerneli'm writing the leaflet atm13:14
lekernelwolfspraul, what cables are we including among those? http://en.qi-hardware.com/wiki/Milkymist_One_accessories#cables13:14
wolfspraulall of those13:16
lekernelthe ones that I see very useful are USB and RCA (or alternatively the little stub that xiangfu sent me)13:16
wolfspraulwe can remove later when we get feedback13:16
wolfspraulI want to remove use obstacles13:16
wolfspraulnot satisfied with the low usage rate of rc2, so this time I want to leave less chance for not doing anything with the box :-)13:16
lekernelrc3 will be a totally different experience... if we get the remaining software items done in time for the flashing (which might need kicking a few asses) we could definitely push it to VJs and artists directly13:18
lekernelbut yeah, good idea to include more cables, especially given how much they cost in europe13:19
wolfspraulthey cost me about 1 usd each :-)13:20
wolfspraulI'm more worried about weight than cost13:20
lekernelok, listing in leaflet is:13:23
lekernel\item a Milkymist One video synthesizer13:23
lekernel\item a CVBS mini-camera13:23
lekernel\item a rubber USB keyboard13:23
lekernel\item a jack-jack and a jack-RCA audio cables13:23
lekernel\item an Ethernet cable13:23
lekernel\item a 5V power supply adapter for the Milkymist One13:23
lekernel\item a 12V power supply adapter for the camera13:23
lekernel\item developer tools (optional):13:23
lekernel\item a JTAG and serial USB adapter13:23
lekernel\item a USB cable13:23
lekernelsounds good?13:23
lekernelah, forgot the camera cable13:24
wolfspraullekernel: ah you're back14:02
wolfspraulso yes the list is correct14:02
wolfspraulthe remote control is up in the air right now, as you know14:02
wolfspraulmaybe instead of 'rubber' you should say silicone keyboard?14:03
wolfspraulis silicone the same as rubber?14:03
wolfspraulchecking :-)14:03
wolfspraullooks like there are many 'silicones' and they are rubber-like14:04
wolfspraulwell, I'm pretty sure the keyboard makers are 'silicone makers' so they use silicone14:05
wolfspraulI could ask for more details what they use chemically, but I think silicone keyboard is better than rubber keyboard14:05
wolfspraulthe developer tools are not optional, they will be included with every rc314:06
wolfspraulI have too little feedback right now to remove anything, it's too early.14:06
wolfspraulhopefully with rc3 we get more real-use feedback and can then remove some stuff in rc414:06
lekernelimo we should be able to make them optional pretty soon14:08
wolfspraulyes but too early to decide now14:08
wolfspraul'silicone keyboard' ok with you?14:08
wolfspraulit's not rubber14:08
wolfspraulremote control is moving, if everything goes well one will be included14:09
wolfspraulbut there are still some moving pieces as you know...14:09
wolfspraulmaybe you add it to the list once we know it's all finalized14:09
wolfspraulthe "CVBS camera" is typically called "CCD camera" in China, but I guess that's a matter of preference14:12
wolfspraulI wouldn't know what either is :-)14:13
wolfspraulit does have a CCD sendor (Sony 1/3''), not a CMOS sensor14:13
wpwrakwhat are "developer tools" ? a CD with gcc ?14:16
kristianpaulI tought was with fedora FEL14:17
kristianpaulsure is not that anymore14:17
wpwrakbut it's software ? seems odd to make that "optional". copying a CD is rather cheap. i'm sure that's what kinds in china learn even before they're weaned ;-)14:18
wpwrakbesides, including optical media with software is so retro :)14:19
kristianpaulusb stick14:20
wpwrakpricy :)14:21
kristianpaulor the leaflet just poiting to a wiki and xiangfu scripts :-)14:21
lekernelit's jtag/serial adapter + pointing up usb cable14:21
lekernelno we don't include any software (except on the M1 flash)14:22
kristianpaulah, the jtag/serial pod is optional then, i see :-)14:22
lekernelyeah we should make it optional at some point14:22
lekernelmaybe not for the first run3 units14:22
kristianpaulso webupdate is working now?14:22
lekernelno, afaik xiangfu hasn't touched it for... 2 weeks or something?14:23
lekernelit somehow works, but it's not robust and it's incomplete14:24
lekernelI really hope it can be finished before the boards get flashed14:24
wpwrakso "development tools" is a section. \item development tools \begin{itemize} \item jtag board \item usb cable \end{itemize} ?14:24
wpwrakokay, makes sense then :)14:25
kristianpauland later bios also may support PC free like feaures for upgrade and even rescue :-)14:25
kristianpaulmail to patch14:25
lekernelthere's already rescue14:25
lekernelin fact, regarding software, what would be cool is get a frozen "bugfixes only" version before the boards are flashed14:26
wpwrak"usb cable" sounds ambiguous. can you also use it to connect the M1 as a usb device ?14:26
kristianpaulsure i'm just reading apple ios5 bew "features" :)14:26
kristianpaul(M1 as a usb device ) dont think so14:26
kristianpaulis it for debug mainlly14:27
kristianpauland last rescue method14:27
lekernelI hope xiangfu can finish the webupdate fast and fix the GUI fonts and wolfgang does the german translation14:27
lekernelthose are the only remaining blockers on software14:27
lekerneli've done the other tasks, unfortunately it's too often my name which appears in commit logs14:27
wpwraklekernel: you should recruit some minions :)14:28
lekernelkristianpaul, you can definitely use M1 as USB device, see topic14:28
kristianpaulor focus on a SDK, mininios like SDK too14:28
kristianpaulyeah? let see14:29
lekernelthere are even easter egg placeholders on the PCB for you to do that. soldering resistors into them and doing the FPGA/software work is up to you.14:29
kristianpauloh, sure :_)14:30
wpwrakwell, as long as you keep "usb cable" in the "development tools" section and don't "flatten" the list ... :) otherwise, people will think M1 plugs into their PC via usb, using that cable14:31
wpwrakas in "i don't even know what ethernet is, but it also supports USB. great !"14:31
kristianpaullekernel: (to save me some readings) the firmware you used in navre was took/forked from other project or written from scratch?14:39
lekernelwhat other project could I possibly have taken that from?14:40
kristianpaulnot sure, dont know, just asking14:40
kristianpauli'm not very related to the usb world :-)14:40
kristianpaulok thats an answer, thanks ;-)14:40
lekernelopencores only claim they have working USB controllers, but unfortunately they don't ...14:41
wpwraklekernel: (navre) btw, what distressed you there ?14:42
wpwrak"navre" = "distressed" or "heartbroken"14:45
lekernelhaving one more thing to do myself (and a quite pesky one) because existing open source solutions are mostly bullshit14:46
wpwrakah :) thought it was a statement about AVR14:46
wpwrakkinda like saying "a PIG16-compatible core"14:47
lekernelI wouldn't have taken AVR if I thought it was crap14:47
wpwrakhmm, yet that's what the name you picked quite strongy suggests14:48
wpwraksounds a bit like "i had no other choice"14:48
wpwrakas in "my boss made me do it" :)14:48
lekernelyeah, that's basically the situation I was in14:49
wpwrakthe kind yet quite different :)14:49
lekernelin hindsight I should just have stuck a proprietary USB controller in there and let some other company's engineer deal with the pesky idiosyncrasies of the USB protocol14:50
lekernelbut the M1 boards were already out with USB PHYs only14:50
wpwrakanyway, just saying that i don't think anyone reads this as anything else but you expressing negative feelings about AVR/Atmel14:50
lekernelit's just (rightly) negative about other open source USB controllers and AVR clones14:50
wpwrakhaving it all in the FPGA demonstrates the power of your solution. so i think you did the right thing, even if it sucked14:51
wpwraks/it/doing it/14:51
Fallenoulekernel: I think it would be a good idea to have a UVC driver in RTEMS/Milkymist15:13
Fallenouto be able to plug whatever usb webcam and get video in working15:14
lekernelFallenou, see topic15:14
FallenouI can only see a mmlogs link in the topic, what's about it ?15:14
lekernelthere is a gazillion things which would be good to have, but unfortunately so far there isn't a gazillion developers to make it happen15:14
Fallenouwell I could try to do it15:15
FallenouI've been looking for a few weeks at linux uvc driver15:15
Fallenouand am starting to hack into to since 1 week15:15
Fallenouinto it*15:15
lekernelthe driver isn't the only problem, the softusb firmware would need a serious overhaul as well for webcams to work15:16
Fallenouyes I was thinking about it15:16
kristianpaulJFDI? :-)15:16
Fallenoumaybe everything needed is not yet there in the usb controller15:16
Fallenoukristianpaul: JEDI :)15:16
Fallenouthe soft usb needs to be able to send SOF, manage isochronous transfers, and descriptors transfer something like that15:17
Fallenouit could be done in bulk too, but bulk sucks15:17
Fallenouso better do it in isochronou15:17
lekernelalso, the USB ports are full speed only (transceiver limitation) so this quite severely limits the quality of the video that could be transmitted15:18
Fallenouhum yes15:18
lekernelin the end it's going to be a huge amount of work for a mediocre result15:19
lekernelthat being said, it'd be definitely interesting to have an open source USB controller that could handle webcams, even in full speed only15:19
Fallenouhum it's not enough for 640x480 YUV 4:2:2 15fps :/15:19
Fallenouwould need something lik 72 Mbit/sec15:20
Fallenouyes it would be good15:21
Fallenoubut it will be really really small resolution :/15:21
Fallenouwith only 12 Mbps15:22
Fallenouwell maybe with MJPG video stream15:22
Fallenoubut it will take CPU to show the picture15:22
Fallenoudon't know how much cpu it takes to decode a JPG image15:23
GitHub65[extras-m1] sbourdeauducq pushed 1 new commit to master: http://bit.ly/mfBCRn15:32
GitHub65[extras-m1/master] Run 3 leaflet (WIP) - Sebastien Bourdeauducq15:32
lekernelfunny, rejon's "Open clipart library" feature for inkscape has the same bugs as flickernoise web update :-P16:17
lekernelthat is, crash when the network connection fails16:17
wpwraklekernel: btw, speaking of bugs, have you seen this one yet ? https://github.com/MrMEEE/bumblebee/commit/a047be85247755cdbe0acce616:32
lekernelyeah lol16:33
kristianpaulblank spaces are sweet isnt? :)16:37
wpwrakthe comments are precious, too. e.g., https://a248.e.akamai.net/assets.github.com/img/58bb0e2ff4031e355738be85c9fc59aea4eef07c/687474703a2f2f692e696d6775722e636f6d2f496d7831422e706e6716:43
rjeffrieswpwrak something tht could drive atusb sales would be a standalone 802.15 gadget that has various i/o for sensors including i2c, plus gpio and a few analog pins plus the radio17:15
wpwrakrjeffries: #qi-hardware may be more appropriate for this ?17:17
rjeffriessorry wrong channel my error17:17
GitHub138[extras-m1] sbourdeauducq pushed 1 new commit to master: http://bit.ly/kCHTRp17:24
GitHub138[extras-m1/master] leaflet: camera setup diagram - Sebastien Bourdeauducq17:24
lekernelhe, inkscape is cool :)17:28
rejonha lekernel can you fix that inkscape bug too then ;)18:18
GitHub179[extras-m1] sbourdeauducq pushed 1 new commit to master: http://bit.ly/ltxgNo18:30
GitHub179[extras-m1/master] leaflet: almost done - Sebastien Bourdeauducq18:30
mwallelekernel: please consider using status register and interrupt enables in the cores (even if it duplicates the lm32 interrupt pending mechanism), writing proper drivers for linux is a pain in the ass atm :(18:45
mwalleeg i have to poll the cpus pending register within the uart driver18:46
lekernelsounds a lesser evil than duplicating stuff in hw18:47
mwalleso the driver for the core have to be in arch/lm32 and cant be reused for other archs18:48
lekernelseems about right, since the uart hardware itself can't be used with archs that do not have a similar interrupt mechanism in the cpu18:49
lekernel_or_ we can remove the lm32 interrupt pending mechanism from the CPU and put it in the cores18:50
lekernelso it's more "standard"18:51
lekernelit'll be a bit of work though18:51
lekerneland it can impact performance (maybe by a negligible amount) as the CPU will have to issue bus cycles to check interrupt status18:52
mwallewhats so evil with duplicating a resettable bit in the cores? the lm32 interrupt controller are basically 32 ff?18:55
GitHub28[extras-m1] sbourdeauducq pushed 1 new commit to master: http://bit.ly/kDo35o18:56
GitHub28[extras-m1/master] leaflet: add logo - Sebastien Bourdeauducq18:56
mwallebtw im trying to get device trees working with lm3218:56
mwalleso my first OF driver will be the UART one ;)18:56
lekernelmaybe we can simply use level sensitive interrupts, so the LM32 needs no FFs to store the interrupt status18:57
azonenberglekernel: So my hardmask cracked and broke off during the etch step18:57
azonenbergBut i feel like i'm getting very close18:58
lekernelmake the IP register read only (just read the interrupt signals from outside the CPU)18:58
lekerneland keep the interrupt states in the cores only18:58
lekernelactually this would remove the bit of redundancy we have with the minimac2 core atm18:58
lekernelwould that be ok?18:58
lekernelazonenberg, yeah I read that. but very good results still :-)18:59
azonenbergI got the hardmask on there, and in the past i've used the hardmask successfully19:00
azonenbergi think it was either thermal shock (I did a cold rinse then a hot etch right after baking), trying to etch too deep too fast19:00
azonenbergor the hardmask being too thick19:00
azonenbergOne drop wasnt thick enough but two was too much, i think i might need to do several depositions using a dilute solution19:01
azonenbergif it's too thick in a single layer it'll crack more readily19:01
GitHub163[extras-m1] sbourdeauducq pushed 1 new commit to master: http://bit.ly/kvtJDC19:03
GitHub163[extras-m1/master] leaflet: fix Makefiles - Sebastien Bourdeauducq19:03
GitHub139[extras-m1] sbourdeauducq pushed 1 new commit to master: http://bit.ly/jC26Wb19:08
GitHub139[extras-m1/master] leaflet: updated EDK leaflet - Sebastien Bourdeauducq19:08
kristianpaul(leaflet) looks, like a lot of fun and out of the box experince :-)19:17
kristianpauli dont remenber, is rc3 provided with the  case already assembled?19:20
kristianpaulbtw the the super-easy steps dont talk about line in19:23
kristianpauli bet there are lot of people with a portable audio device who want to plug in and feed the effects too19:23
kristianpaulnot saying of course that built-in mic isnt cool for feeding effects related to the locally "sound landspace"19:25
lekernelyes, we _should_ have the case assembled19:46
lekerneleven with a nice cardboard box containing accessories, etc.19:47
wpwraklekernel: hmm, the girl looks like a 2D projection, not like a 3D person standing in front of a screen19:50
lekernelyeah well... I already tried to fix that but failed to get good results with the inkscape "perspective" effect and gave up19:52
wpwrakdraw a little shadow ?19:52
wpwraksome of the padding of lines is excessive. maybe use \raggedright ?19:54
wpwraki actually wonder why latex did them so poorly. maybe it just needs a \par19:55
wpwrakhmm. if you have color printing, the "{\bf green} RCA" could actually be green ;-)19:57
lekernelthe run 3 is already complicated enough without color printing19:59
lekernelso I just put \par at the beginning?19:59
wpwraklemme have a look at the source ...20:00
wpwrakthanks !20:00
wpwrakputting a blank line between "Press the middle pushbutton on the Milkymist One to start." and \includegraphics should improve the formatting20:01
wpwrak(of the paragraph above the picture)20:02
wpwraksimilar for all the other images20:02
lekernel! Class leaflet Error: The text you supplied fills more than six pages20:04
lekernel(leaflet)              and will therefore not fit onto a single flyer.20:04
wpwrakdoes the meta key on the rubber keyboard really not have a label ? and they key next to it neither ?20:04
wpwrakwell, you can worry about the size later :)20:04
lekernelit has a windows label that we probably are going to remove/replace at some point, so ...20:04
wpwrak(window label) hmm, tricky. the key next to ctrl has that "menu" label then ?20:05
GitHub85[extras-m1] sbourdeauducq pushed 1 new commit to master: http://bit.ly/lHEGeu20:08
GitHub85[extras-m1/master] leaflet: fix padding (thanks Werner) - Sebastien Bourdeauducq20:08
wpwrakkeyboard mouse isn't quite intuitive. may be clearer if you show the combinations, e.g., with a "bridge" from Meta to Enter with "Click" written above and a bridge between Meta and a bubble around the cursor keys for "Move Mouse Cursor". not sure if it's worth the effort, though20:08
wpwrakgot the girl picture. maybe if you just make her feet cross the bottom line, that could be enough. real people rarely float in front of screens ;-)20:09
wpwrakhah, s/internet/Internet/ ;-)20:10
wpwrakand s/network do not/network does not/20:10
wpwrak"minijack" is 3.5 mm ?20:14
GitHub40[extras-m1] sbourdeauducq pushed 1 new commit to master: http://bit.ly/iA7UDe20:14
GitHub40[extras-m1/master] leaflet: typos (thanks Werner) - Sebastien Bourdeauducq20:14
lekernelhttp://en.wikipedia.org/wiki/File:Triple_Contact_Plug_1907.png :-)20:16
wpwrakmaybe mention the diameter. never heard of "minijack". also, what is considered "mini" changes with time ;-)20:18
lekernelwell, they'll have the device and cables just in front of them, and it's a very common plug20:19
wpwrakit's a bit unfortunate that de-vel@lists.milkymist.org got hyphenated. maybe use \hbox{devel@lists.milkymist.org}20:19
wpwrakor \mbox{...} :)20:21
lekernelhbox works20:22
GitHub194[extras-m1] sbourdeauducq pushed 1 new commit to master: http://bit.ly/kKGzrW20:29
GitHub194[extras-m1/master] leaflet: hyphenation (thanks Werner) - Sebastien Bourdeauducq20:29
wpwrakthat's all i found. looks very nice !20:30
GitHub90[extras-m1] sbourdeauducq pushed 1 new commit to master: http://bit.ly/iMTk3K20:46
GitHub90[extras-m1/master] leaflet: fix logo - Sebastien Bourdeauducq20:46
mwallelekernel: yeah that should work, but you'll sacrifice compatibility (with other lm32 cores) just to save some FF?21:21
lekernelit's not only about saving some FFs, it's about having a clean design22:42
lekernelwhat's the point of clearing IP when interrupts are level sensitive?22:43
lekernelalso, given the lack of LM32 developments outside of here, there isn't much compatibility to care about ...22:43
lekernelif edge sensitive interrupts are a mess to deal with in linux, just fix that in the processor arch...22:46
kristianpaulmm1 on tv :)23:22
--- Mon Jun 20 201100:00

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