#milkymist IRC log for Tuesday, 2011-06-21

xiangfumy camera is kind of hot in those two days. maybe because now is summery?02:21
xiangfu~40 degree02:25
wolfspraulit's probably more than 4002:27
wolfspraulyes it gets hot02:27
wolfspraulwe rely on the vendor to make a stable design, right now I have no reason to doubt their work. The camera uses passive cooling, meaning that the entire camera gets warm, and the air around it will dissipate the heat.02:28
wolfspraulfor electronics in general, I think there is no problem with anything that stays under 60 degree celsius02:28
wolfspraulfor the camera, I never measured exact numbers (we should), but you can still easily hold it in your hand, so I don't think it's even 60 degree? I'm not sure.02:29
wolfspraulit's too late now anyway we already sourced it :-)02:29
wolfspraulxiangfu: you can measure the exact temperature, that would be interesting02:29
wolfspraulalso I think it's stable, it will heat up within the first 30 minutes or so, and then it stays stable02:30
wolfspraulit won't heat up uncontrollably and melt down :-)02:30
wolfspraulI think it's all by design... (good or bad design is still up in the air a little)02:30
xiangfuok02:31
wolfspraulthe camera consumes about 93-96 mA at 12V02:31
wolfspraulthat will all be converted into heat :-)02:31
wolfspraulxiangfu: can you still hold it in your hand?02:32
xiangfuwolfspraul: yes. I can hold it in my hand.02:33
wolfspraulso what's your guess on temperature? do you have a thermometer?02:33
wolfspraulAdam has one I'm sure, he can quickly take a measurement...02:33
xiangfu I am measuring. it's about 38, 39. degree.02:33
wolfspraulthat low? I think it's more, feels more to me.02:33
wolfspraulmy guess would be 50 or so02:33
wolfspraul55 even02:34
xiangfuI am using a multimeter. 39~4102:34
wolfspraulI would be worried if it's 70 or more, but even then like I said, we rely on the vendor to make good stuff and we didn't object earlier.02:34
wolfspraulxiangfu: ok, I'm sure it's more, but anyway I think it's OK. We need more precise measurements at some point.02:35
xiangfuok02:35
wolfspraulmy rule of thumb for electronics is < 60 degree is OK02:35
wolfspraul70, 80, maybe you shorten the life expectancy02:35
xiangfuok.02:36
wolfspraulabove that generally no, unless it's specifically designed to handle that kind of heat02:36
wolfspraulof course lower temperature is always good, because there is also stress from warming up and cooling down (different materials)02:37
wolfspraulso if it never even warms up, even better.02:37
xiangfuyes.02:38
xiangfuwolfspraul: I met the error on compile patches in simple mode. two of them.02:38
wolfspraulsaw it02:38
xiangfuafter move those two patches to another folder, simple mode works fine.02:39
xiangfuanother issue, recently I notice when I re-plug the usb cable(while m1 poweron) the m1 freeze sometimes02:44
xiangfuit maybe because when I plug the cable, I move the usb-jtag board.02:45
xiangfumove the usb-jtag board while m1 running will casue m1 freeze sometimes. (I found this when I test memcard in my m1)02:47
wolfspraulxiangfu: ok03:12
wolfspraulI believe Adam has made a few changes to the jtag-serial board so that one can access the memcard without being obstructed by jtag-serial03:12
wolfspraulbut the problem with the usb cable remains, I agree, it's not very stable mechanically so if you unplug the USB cable you may mess with the connectors between jtag-serial and m103:13
wolfspraulthe good news is that jtag-serial is a developer-only feature, and only accessible after opening the case03:13
wolfspraulso I'm not so worried about those freezes, we can just try to make the 2 connectors fit better, which is already something Adam was working on03:13
wpwrakwolfspraul: those debug boards are the soul of murphy, it seems. first, the endless struggle we had with them at openmoko. now the M1 jtag board ...03:15
wolfspraulworks perfectly fine on m103:15
aw_wpwrak, yeah...agreed. btw...well03:16
wolfspraulthe issues we had so far are all cosmetic/polishing03:16
wolfspraullike the usb high-speed fix, or the mechanical adjustment to better reach the memcard, or a very stable connection between jtag-serial and m103:16
aw_the jtag rc2 i believe it will fit m1 connect better.03:16
wolfspraulin all this, the actual boards always worked, 100% of the time03:16
wolfspraulwe almost have too much focus on the jtag-serial board, I think03:17
wolfspraulit's a developer feature03:17
wolfspraulI rather polish the end-user experience from outside the box, not when the case is disassembled and people are playing with jtag :-)03:17
wpwrakwolfspraul: the devil's usually in the electromechanical bits :) but yes, over-focus is a possibility03:17
wolfsprauljtag-serial on m1 is a success story03:17
wolfspraulthe issues we have are closer to (bad) perfectionism than anything else03:17
wolfspraulI mean the boards always work, really always. never once had a problem...03:18
wolfspraulI think I have seen some cases where I had to run the reflash_m1.sh script twice, because the first time it would fail with some error03:18
wpwrakwolfspraul: but it causes freezes ?03:19
wolfspraulthat looks like a jtag or software issue though03:19
wolfspraulnah03:19
wolfspraulwpwrak: xiangfu is unplugging the usb cable03:19
wolfspraulthe usb cable is rather big, and the connector quite strong03:19
wpwrakah :) that can easily cause trouble03:19
wolfspraulnow, on the other side the jtag-serial is held to the m1 only with simple pin headers03:19
wolfspraulso the force needed to unplug the USB is sometimes more than it takes to unplug the pin headers on the other side03:20
wpwrakit may even be just the FTDI spewing garbage at JTAG. there should be two reset lines in there. much fun can be had with these.03:20
wolfspraulbut keep in mind: this is all happening with an open case03:20
wolfspraulit's a developer-only situation03:20
wolfspraulthe connector between jtag-serial and m1 is not designed for live plugging03:20
wolfspraulthe sequence in which the wires connect can even cause damage03:21
wolfspraulso yes, it is a (very small) problem03:21
wolfspraulbut once someone opens his case, and starts working with jtag, that's a very small issue to take care of03:21
wolfspraulsame as, say, to not spill liquids onto the open pcb :-)03:21
wpwrakno cleaning with alcohol allowed :)03:22
wolfspraulto be safe, you may want to power off your m1 before unplugging the usb-jtag cable03:22
wolfspraulwhy would you even want to unplug it while running the unit?03:22
wolfspraulremember that we are in a developer lab situation now, someone has jtag wired up for some reason03:22
wolfspraulthis is not a user at a party/event trying to use the product03:22
wpwrakyou never know ;-)03:23
wpwrakhacker party ;-)03:23
wolfsprauland actually, with the case open, and the usb connector being metallic, you have to be careful to not accidentally short something even before you connect to jtag-serial03:23
wolfspraulin the bigger context the mechanical issue xiangfu pointed out is not more problematic than many other things you expose yourself to once you open the case and start working with the board03:23
wolfspraulbut then, like with the other jtag-serial bugs, we shouldn't take it out of proportion03:24
wolfspraulit works just fine...03:24
wolfsprauland as Adam said we already try to make the connectors to m1 fit tighter/be stronger03:24
wpwrakyeah, often you just need to know where the traps are03:25
wolfspraulif you reach with a metallic connector directly onto an open pcb, you should already be at some alert mental state03:25
wolfspraulotherwise you better not even open the case...03:25
xiangfu:)03:26
wolfspraulxiangfu: I think I always first powered off my m1 before unplugging the jtag-serial usb cable03:27
wolfspraulnot even out of running into a freeze or so like you, just general precaution03:27
xiangfuwolfspraul: yes. that is better.03:28
wolfspraulhah, I have a horrible idea03:30
wolfspraulwpwrak: Adam can put a huge blob of hot glue around all this (jtag-serial sitting on m1)03:30
wolfspraulthat would be niiiiice03:30
wolfspraulthat way the connection between jtag-serial and m1 will sure be stronger than the one between usb cable and jtag-serial03:30
wolfspraul(no worries, I am joking, we will not do that :-))03:31
wpwraksounds like a solution that's worse than the problem ;-)03:31
aw_wpwrak, dis you know how david's smt vendor to go through your front.pos and back.pos files while mounting atben?03:47
aw_wpwrak, since i made http://lists.en.qi-hardware.com/pipermail/discussion/2010-December/006638.html03:49
wpwrakaw_: i converted them to csv. they didn't complain :)03:49
wpwrakaw_: e.g., here's the SMT fab package i made: http://downloads.qi-hardware.com/people/werner/wpan/fab/atusb-smt-110330.tar.gz03:50
aw_wpwrak, hmm...some sort of same results, good. :-) csv also they can transfer to xls in windows.03:50
wpwrakaw_: there was one thing the smt fab didn't like: i had not included the fiducials in the positions. the latest version of my scripts now does that too03:51
wpwrakaw_: pcb and smt fab package generation is just a matter of "make fab". the only thing not automated is the README. still haven't found a good artificial intelligence to do the documenting for me ;-)03:54
aw_wpwrak, already pretty good tool though.:-)03:55
aw_wpwrak, with regards to if adding fiducials in each single gerber can be determined by their skills. but it's good that i saw you added it into single gerber.03:56
aw_wpwrak, this time i asked smt technician first then decided not to add fiducial. :-)03:58
wpwrakaw_: the smt fab in spain gave us a detailed specification of what fiducials they wanted. but in the end, it seems they didn't really need them :)04:00
wpwrakaw_: (or maybe they just picked up their positions from somewhere else. there's enough redundancy for that.)04:00
wpwrakaw_: in any case, the next time, the fiducials will be perfect ;-)04:01
aw_wpwrak, exactly heard eventually here too, the super smart optical system on their smt machine seem that can cover this.04:01
aw_wpwrak, but how it did? didn't know. :-)04:02
aw_wpwrak, just liked you said; me too here, they picked up one pad or through hole then done. ;-)04:03
wpwrakaw_: i was kinda surprised they asked for fiducials at all, with so many nice pads to choose from :)04:05
aw_wpwrak, :)04:07
lekernelxiangfu, do not plug the usb-jtag board when the m1 is running, it's not made for this and there is risk of short circuits07:08
lekernelif it freezes, it's probably because it sent a break on the serial port and entered debugger mode07:08
lekernelalso, some distros run crap like gpsd at every new serial port which is connected; the garbage data such things send might just well be interpreted as a break07:10
xiangfulekernel: thanks for the info.07:14
lekernellet's not give too much importance to the remote control...07:17
lekernelany news on the libcurl timeout issues? or fixing the GUI fonts?07:17
lekernelhardoded t003 timing is fine now07:17
lekernelhardcoded07:17
lekernellibcurl related freezes and bogus characters aren't07:18
xiangfulekernel:(remote) yes. I just ask another people to help, (libcurl) not much progress.  :(07:57
xiangfuI am take a look the check version today. if the expat can decode the html?07:58
lekernelhtml?07:58
lekerneloh, don't do anything complicated please07:58
xiangfulekernel: I create a temp URL: http://downloads.qi-hardware.com/people/xiangfu/milkymist-one/msd/latest/07:58
xiangfumy idea is get file name, compare the version by filename.07:59
lekerneljust download files named "soc-version" or "flickernoise-version" which contain the version numbers in plain text07:59
xiangfulekernel: or we just push VERSIONS?07:59
xiangfulekernel: oh. yes. that is simple and fast.07:59
xiangfulekernel: I think one file is enought.08:00
lekernelyes, if you like manipulating strings in C (I don't)08:00
lekernelboth solutions are acceptable08:00
lekernelbut no xml please :)08:01
lekernelfor the patch pool, download a list of files08:02
lekernelone patch name per line08:02
xiangfu(no xml) yes. sure.08:02
lekernelthen compare with those in /ssd/patchpool and download the missing ones08:02
xiangfu(patch pool) ok. got it08:02
xiangfuok.08:02
xiangfubtw what is the meaning of 'msd'?08:03
lekernelthe "?" in the GUI should be replaced with the number of patches08:03
lekerneljust to give an imprecise indication of how many new ones will be downloaded08:03
xiangfuok08:03
lekerneluse \e before to disable translation08:04
xiangfugot it. thanks.08:04
lekerneli'll have a look at this gethostbyname thing08:07
lekernelapparently this gets you stuck, so focus on the more "end user" web update code08:07
xiangfuthanks. the problem may from the cpukit/libnetworking/libc/res_*.c files.08:12
xiangfuthe last few files I look into is res_*.c files.08:12
lekernelthat sounds like the right track08:14
lekernelalso it'd be useful to know how libcurl sets the timeout08:14
lekernelis that with alarm() ?08:14
xiangfudon't know that.08:22
lekernelthat's one thing that needs to be figured out in order to be able to debug08:30
xiangfuhttp://downloads.qi-hardware.com/people/xiangfu/milkymist-one/msd/latest/08:32
xiangfuhave you update the08:32
xiangfu(sorry)08:32
xiangfuhttp://downloads.qi-hardware.com/people/xiangfu/milkymist-one/patchpool/08:32
xiangfulekernel:  I create a test url, is this ok for you? there is file named 'versions'08:32
lekernelyes, looks good08:38
lekernelfor the patches, "versions" looks inappropriate, use "list" or something like that08:39
lekerneloh, and we should not include the DMX patches in the patchpool, since in "out of the box" mode the user probably won't have a DMX controller attached08:40
wolfspraullekernel: it would be cool if patches were somewhat 'smart', maybe with a flag or so, and a patch could indicate if it should be skipped if there is no camera signal present, or no DMX controller08:41
wolfspraullike a command in the patch "needs camera"08:41
lekernelyeah I brought up this idea already, it's in the "other ideas" software roadmap08:42
wolfspraulnot sure whether this is a good idea, but in simple mode it may help to skip patches that just don't look good, without camera, or without dmx08:42
wolfspraulok08:42
lekernelbut that's one small detail08:42
wolfspraulnot very high priority08:42
lekernelif people want to get a "professional" experience, they won't use simple mode08:42
xiangfu(list) ok. I was thinking. maybe in future we do have a version in patchpool, anyway, let's use 'list'.08:56
lekernelthen it's going to be a different file format, and it makes sense to use a different filename so software and people do not get confused08:57
xiangfujust checked libcurl using alarm() and sigsetjmp()09:40
lekernelok09:42
lekernelmy guess at the bug is that gethostbyname() calls alarm() again to set its 2 minute timeout09:42
lekernelhttp://lists.uclibc.org/pipermail/uclibc/2002-April/023978.html09:44
lekernelthe problem is, there's no select() in rtems...09:44
lekernelah: http://www.rtems.com/ml/rtems-users/2004/september/msg00122.html09:45
lekernelSO_RCVTIMEO09:45
lekernelgethostbyname should use SO_RCVTIMEO09:45
lekernelcheck it does09:46
lekernelby the way, maybe this bug is already fixed in rtems head... https://github.com/milkymist/rtems09:46
lekernelmaybe worth looking at that code09:47
Fallenoulekernel: ahah I just watched the youtube video about CVS you put in the description of the git mirror of rtems :)09:51
Fallenoulekernel: strange it says in github "forked from kristianpaul/rtems-milkymist" about the milkymist/rtems-milkymist repo10:07
Fallenoudidn't you fork from mine ?10:07
lekernelthere has been a lot of renaming and repository deletion on github when I set up the organization account10:09
lekernelso the "forked from" is messed up sometimes, but this is the correct code inside10:09
lekernelthis repository will be deleted as well anyway10:09
lekerneli'm merging the patches upstream10:09
Fallenouok, I was fearing you lost some commits in the battle10:10
xiangfuthe problem maybe here: https://github.com/milkymist/rtems/blob/master/cpukit/libnetworking/libc/res_send.c#L66610:12
xiangfuit's will try 4 times. totally 75 seconds. I don't know how to make the alarm work on res_send.c10:13
lekernelxiangfu, the code you are looking at is the rtems head10:14
lekernelthe code which is running atm is https://github.com/milkymist/rtems-milkymist/blob/master/cpukit/libnetworking/libc/res_send.c#L666 ... it seems to be the same here, but just so you know.10:14
xiangfuyes.10:14
xiangfu(my network is very slow, I am under the GFW :) so I just using the one I opened. I am re-compile the rtems for make sure.10:17
lekerneland res_send.c should NOT use alarm()10:23
lekernelimo even when SO_RCVTIMEO is used, alarm() should work as well10:24
xiangfumaybe we set the _res.retry to 2. fix most of the problem.10:39
xiangfuthen the totally DNS search time is 15 seconds.10:40
lekernelacceptable for run310:40
lekernelhow does one set _res.retry? I hope it's not buried deep in the RTEMS code...10:41
xiangfuI guess some ENV can apply on _res.10:42
xiangfuthe code is here: https://github.com/milkymist/rtems-milkymist/blob/master/cpukit/libnetworking/libc/res_init.c#L18110:42
xiangfuok.10:43
xiangfuthe res_setoptions only work on _res.options. not _res.retry10:43
lekernelstruct __res_state _res10:44
lekernelthis symbol is exported10:44
lekerneljust set in flickernoise main.c:10:44
lekernelextern struct __res_state _res;10:44
lekernelthen _res.retry = 2; at the beginning of the Init task10:45
lekernelit's a hack, but it'll get the thing done10:45
lekernelmark it as a hack with a "FIXME" comment in the code10:46
lekernelalarm() should work anyway in gethostbyname()... maybe the rtems people will fix that bug someday10:47
lekernelin the meantime we have a workaround10:48
lekernelif it's not fixed in RTEMS HEAD (which i'll patch shortly with the milkymist drivers) i'll file a PR at RTEMS10:49
xiangfulekernel: ok. got it11:13
awxiangfu, has newest rc5.v  been included for t003 under http://www.milkymist.org/snapshots/latest/ ?12:08
lekernelah, the rc3 boards get immersion gold too.12:41
lekerneland Expected PCB Return Date 06/2312:43
GitHub106[flickernoise] sbourdeauducq pushed 2 new commits to master: http://bit.ly/jmHceU14:09
GitHub106[flickernoise/master] Simplify/cleanup patches to avoid register allocation problems - Sebastien Bourdeauducq14:09
GitHub106[flickernoise/master] Sort interactive (DMX) patches - Sebastien Bourdeauducq14:09
GitHub182[flickernoise] sbourdeauducq pushed 1 new commit to master: http://bit.ly/iKBY1314:12
GitHub182[flickernoise/master] Added comet wallpapers - Sebastien Bourdeauducq14:12
Action: xiangfu also just want commit :)14:15
Action: xiangfu now git pull -r :)14:15
GitHub172[flickernoise] xiangfu pushed 1 new commit to master: http://bit.ly/jSFG3514:19
GitHub172[flickernoise/master] let the gethostbyname return in  15 seconds - Xiangfu Liu14:19
lekernelgood :)14:26
GitHub67[flickernoise] sbourdeauducq pushed 1 new commit to master: http://bit.ly/jVv1td14:46
GitHub67[flickernoise/master] Work around black screen bug in MTK - Sebastien Bourdeauducq14:46
Fallenouwow, lattice has a funny way of "obfuscating" their IP core code15:46
carlobarhi, im trying to run qemu but i got some errors. Can some help me? This is the output:15:56
carlobar$ qemu-system-lm32 -M milkymist -nographic -kernel software/bios/bios.elf15:56
carlobarBad ram pointer (nil)15:56
carlobarAbortado15:56
Fallenoudon't run the elf15:57
carlobarbios.bin?15:58
kristianpaulyes yes15:59
lekernelthe .elf should work as well15:59
carlobarwith bios.bin the output is: pflash_write: Unimplemented flash cmd sequence (offset 00002000, wcycle 0x0 cmd 0x0 value 0x1)16:00
lekernelworks for me16:00
lekernel$ qemu-system-lm32 -M milkymist -nographic -kernel software/bios/bios.elf16:00
lekernellibHPDMC SDRAM initialization runtime16:00
lekernel(c) Copyright 2010 Sebastien Bourdeauducq, released under GNU LGPL version 3.16:00
lekernelVersion 1.016:00
lekernelInitialization sequence completed.16:00
lekernelhave you modified the bios?16:01
lekernelor qemu?16:01
kristianpaulbtw what is the mask for in the addressing for the mm core?16:03
kristianpaulis kay if have to RTFM, is just not my topic..16:03
lekernelhuh?16:03
lekernelwtf is a "mask for in the addressing for the mm core"?16:04
kristianpaulxD16:04
kristianpaulsorry16:04
kristianpaulbit mask*16:04
lekernelstill doesn't make sense ....16:04
kristianpauls/core/soc16:04
carlobari had modified the bios... memory settings on linker.ld are:16:04
carlobarMEMORY {16:04
carlobarflash : ORIGIN = 0x00000000, LENGTH = 0x1000000   /* 16M */16:04
carlobarsdram : ORIGIN = 0xc1000000, LENGTH = 0x1000000  /* use the upper 16M of the SDRAM */16:04
carlobarbram  : ORIGIN = 0x20000000, LENGTH = 0x2000/* 16k */16:04
carlobar}16:04
kristianpauldevice addressing*16:04
lekernelcarlobar, get it to work first without modifications16:05
kristianpaulanyway i'll ask later with more sane arguments :-)16:05
lekernelcarlobar, and apparently you have changed the SDRAM address; don't expect QEMU to magically guess what the new address is16:05
carlobarhehe, ok, SDRAM address is 0x40000000 or 0xc0000000?16:06
lekernel416:07
lekernelhm, c is for uncached access16:08
kristianpauls/mask/shadow!!16:08
carlobardoes HPDMC have into account cached and uncached access?16:12
lekernelwhy should it?16:13
kristianpauli think thats task for the fmlbrg16:13
lekernelbtw all the upper bit does is bypass the L1 cache (inside LM32°16:14
kristianpaulwith upper you mean MSB?16:15
lekernelyes16:16
Fallenouis it possible to run Lattice fpga dev software on linux x86_64 ?16:18
Fallenouor is it only runnable on x86_3216:18
lekernelFallenou, you should have got a M1 instead of a lattice board; the xilinx tools work on 6416:18
kristianpaul(lattice board) :o16:19
Fallenoukristianpaul: yes I just bought a $99 lattice board16:20
kristianpaulah, cool cheap stuff ;-)16:20
Fallenouhttp://www.latticesemi.com/products/developmenthardware/developmentkits/ecp3versadevelopmentkit.cfm16:20
Fallenouthis one :)16:20
FallenouI will try to play with LM3216:20
kristianpaulwow,1Gb DDR3 :D16:21
FallenouI received it yesterday16:23
Fallenoudidn't have time to plug it in atm16:23
Fallenoudid some home made sushi/maki instead ;)16:23
kristianpaul:)16:23
kristianpaullooks very cool for networking16:24
Fallenouyet, 2 NICs16:24
Fallenouyep*16:24
kristianpaullike running a snort implementation in verilog or something16:25
kristianpaulor Fallenou implementation for perl regex libray in verilog ;-)16:25
Fallenouyesssss definetely !16:25
carlobardo i have to install the avr toolchain in order to build the bios?16:39
kristianpaulnope that i remenber16:40
GitHub190[milkymist] sbourdeauducq pushed 1 new commit to master: http://bit.ly/j0CSGD16:40
GitHub190[milkymist/master] bt656cap: send IRQ at end of frame, not at start... - Sebastien Bourdeauducq16:40
kristianpaulwhy?16:40
lekernelah, now the video-in is very smooth now16:40
lekernel:)16:40
lekernelno longer this little intermittent "pause/tearing" bug16:41
carlobarkristianpaul: i got errors building softusb-input:16:41
carlobarmake[1]: avr-gcc: No se encontró el programa16:41
carlobarmake[1]: *** [softusb-input.elf] Error 12716:41
kristianpaulah, thats because the navre core, yes you need avr-gcc indeed16:43
carlobarnavre core?16:46
Fallenouthere is an AVR compatible core inside the SoC16:47
Fallenouit is used as a USB host controller16:47
Fallenouthe USB host controller is a software running on an AVR compatible core16:47
Fallenouso you need to compile some C files corresponding to the USB host controller, the target needs to be avr, so you need avr-gcc16:48
Fallenounavre is the name of the AVR core used16:48
kristianpauland also a french word i think.. :-)16:49
carlobarbut im not using the usb host, can i disable libraries related to softusb-input to build the bios and run it with qemu?16:49
Fallenouyes navré is a french word16:50
Fallenoudon't know is bios is doing some usb testing at boot time or not16:50
Fallenouif bios*16:51
kristianpaulit does something i remenber16:51
Fallenougotta go++16:52
carlobarmaybe its better to install the toolchain...16:52
Fallenouthat's one command to install it under ubuntu/debian16:53
carlobarwhere is it available? is it in repos?16:53
kristianpaulas always ;)16:53
Fallenouyes16:53
Fallenousudo aptitude install gcc-avr avr-libc16:54
Fallenouor apt-get if you prefer16:54
Fallenou*gone*16:54
kristianpaulor with aptitude }:-)16:55
carlobarthank you, im installing it16:55
kristianpaulmay be binutils-avr too, is on my system dont remenber why right now..16:56
carlobaryes, binutils-avr  is installed too17:06
carlobari was getting this error compiling softusb-input.elf:17:09
carlobar/usr/lib/gcc/avr/4.3.2/../../../avr/bin/ld: section .bss [00000000 -> 0000000f] overlaps section .text [00000000 -> 00000f0f]17:09
carlobarso i modified memory settings in navre.ld17:09
GitHub106[llvm-lm32] jpbonn pushed 20 new commits to master: https://github.com/milkymist/llvm-lm32/compare/ea7d9dc...979f70518:42
GitHub106[llvm-lm32/master] Completely short-circuit out ARC optimization if the ARC runtime - Dan Gohman18:42
GitHub106[llvm-lm32/master] Don't mark the eh.dispatch.setup with a memory access marker. We want this to - Bill Wendling18:42
GitHub106[llvm-lm32/master] Re-apply 132758 and 132768 which were speculatively reverted in 132777.  - Akira Hatanaka18:42
kristianpaula bit OT herebut18:57
kristianpaullekernel: http://ballonsolaire.pagesperso-orange.fr/en-historique5.htm18:57
kristianpaulit was in enlgish actually18:57
kasbahhi22:37
kasbahI stumbled across and irc log where jackgassett and lekernel discussed porting milkymist to papilio one22:39
kasbahdid you make any progress with that jack?22:40
kristianpaulhey, kasbah22:40
kristianpaulwellcome :-)22:40
kasbahthanks :)22:40
kasbahdon't use irc much22:41
kasbahguess jack and lekernel are actually afk22:42
kasbahi am working on this project: alphasphere.com22:43
kasbahwanted to evaluate moving the processor to fpga22:44
kasbaham interested in the processor core used on the milkymist22:45
kasbahand wether i could run that on one of jack's papilio boards22:45
kristianpaulwell, move procesors to fpga have some  pros and conrs as always, but it depends on your needs22:46
kasbahanyway... no one is in it seems22:46
kasbahyeah22:46
kasbahi kinda want to just have a low cost way to try it out22:47
kristianpaulcan you tell us what are you currently doint with the processor22:47
kristianpaul?22:47
kasbahwe are using XMOS processor at the moment22:47
kasbahit is is similar to a paralax propeller22:47
kristianpauland what you expect from the fpga ?22:47
kristianpaulyeah, multire core no mmu22:47
kristianpaulmutlicore*22:47
kasbahyeah but the multicore is not really helping us at the moment22:48
kasbahi could be using something much less powerful22:48
kasbahbut for what we want to do it is not powerful enough22:48
kasbahbasically i was hoping with RTEMS it would be easier to do sound synthesis and playback on board22:49
kristianpaulwell, actually neither of tha is implement, or that i'm aware, so development go first22:50
kristianpaulabout lowcost, i guess you are aware fpga are not the costless solustion of the market right now..22:51
kristianpaulbut of course,always is interesting do coll stuff on it, like VJing :)22:51
kasbahi was hoping we could port some music languages like Puredata or Supercollider to uClinux22:53
kristianpaulis the alphasphere currently produced?22:53
kasbahwe are working on prototype 0.2 right now22:53
kristianpaulwow, Puredata, i had heard some people screaming even making it run on a PC ;)22:54
kasbahi have seen it run on an old ipod :)22:54
kristianpaulwow22:54
kristianpauli dint knew it22:54
kristianpaulthats fantantism you have the link to that information?22:54
kristianpaulfantastic*22:55
kristianpaulso still hope, really i never tought it could run on old iphones :-)22:55
kristianpaulipods*22:55
kristianpaulkasbah: please hang on here, more often, you are from UK right? may be n some hours lekernel could also reply to you22:56
kasbahyeah i will do22:56
kristianpaulbut you're wellcome to ask more about milkymist in the mail list22:56
kristianpauland feel free to ask,22:56
kasbahpdpod:http://www.youtube.com/watch?v=kukNp4uwcKc22:57
kristianpauli'm hurry know i;ll leave and back in some hours, bye!22:57
kristianpaulyour project looks very interesting22:57
kasbahok cool22:57
kristianpaulmusical instruments are always wellcome i think..22:58
kasbah:)22:58
Action: kristianpaul away22:58
kasbahnow the 100th follower of @milkymistvj on twitter :)23:58
--- Wed Jun 22 201100:00

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