#milkymist IRC log for Wednesday, 2011-01-19

renzohello :)06:50
renzoi am new there, so hello everybody06:50
renzoi would like to speak about an idea06:52
renzoadd dmx player features to Milkymist06:53
renzoat the moment i'm using Qlc software on linux : http://sourceforge.net/projects/qlc/06:54
rohrenzo: you mean use the hw as a lighting-controller?06:55
renzokind of06:55
rohthat should be possible (only sw needed) .. it can already send and recieve dmx with flickrnoise06:55
renzofor example, be able to play a dmx show06:55
renzoyes, but it need an engine to play on a timeline, isn't it?06:56
rohengine?06:58
renzodoes flickrnoise make fades between two values?06:58
rohdunno. i guess it can06:58
rohflickrnoise plays 'patches' like the old winamp did06:59
lekernelit can, but it doesn't have the relatively advanced timeline support you probably need06:59
rohi guess it would make sense to write a specialized app for the mm1 for your usecase06:59
renzoanother thing: should it be possible to add real time clock on the hardware?06:59
lekernelyup06:59
rohhm. realtime clock. there is none now. and without soldering you could only count crystal cycles...07:00
lekernel(rtc) you could add a rtc chip on the gpio expansion header, but if you're not into electronics/fpga/low-level it's probably easier to use NTP07:00
rohi guess one could add one via expansion header.. an i2c thingie with backup battery07:00
lekernelyup07:00
rohdoes one need a rtc 'on stage' ?07:00
rohi thought clocks are important.. but not the absolute time07:01
renzoeven if i could use ntp, i am not always connected, but i can try to.07:01
rohas in midi or so masterclock07:01
rohrenzo: counting crystal cycles is 'stable enough' for normal stuff. but a real clock would drift less (could be seconds a day)07:01
renzoor maybe i could use separate clock module, which start power when needed.07:01
renzoroh: i see07:02
renzofirst, i will try in qemu to understand current dmx features07:04
renzo(rtc) there is a rtc in every pc, is it a simple thing?07:06
rohyes and no.07:06
rohyes its simple07:06
rohno, building one yourself doesnt make much sense right now07:06
rohmostly because one needs a low-current asic backed by a battery/accumulator07:06
rohand a extra 32khz crystal07:07
rohso i guess its not really making sense e.g. designging one on vhdl  or verilog besides if you want to build chips from that07:08
renzook07:08
rohof course it would be nice to have a foss rtc design ;)07:09
renzo:)07:09
rohjust not that real-world useable with an fpga till made an asic07:09
nightlybuildhey roh. Did you finally receive cases order confirmation from CÚcile?07:12
rohi did07:14
rohwell.. no extra confirmation.. but a mail explaining my confusion ;)07:15
rohanyhow.. the acryllic is ordered and will be sent out to me today i was told on the phone earlier today07:15
rohso i hope it will all work out in time07:15
nightlybuildroh: great07:15
rohany idea how to get the stuff to fosdem? or where does it need to go to?07:16
nightlybuildroh: depends on shedule actually. May be lekernel may bring them with him. If not, you'll have to ship them out to Paris07:17
rohi see07:18
rohabout dates.. when will you leave paris for brussels?07:18
lekernelroh: actually for simple and slow stuff like rtc's, actel fpga or cpld might be useful07:22
lekernelor even just make your own IC. shouldn't be very hard.07:23
lekernelI can't bring more stuff to fosdem. i'm not in Berlin anymore and will be back on Feb 7th07:25
rohi see07:26
rohthen i'll send it07:26
nightlybuildroh: february 4th in the morning to Paris, then afternoon train to Bruxelles07:27
nightlybuildroh: so, postal delivery to Paris still the last solution07:27
nightlybuildtime will tell07:27
rohwe'll see how fast i get it done07:29
nightlybuildroh: yep. No worries07:32
nightlybuildwe are also familiar with hacking time ;)07:33
rohhehe.. thats not my worries.. rather logistics and all the stuff i dont do07:34
lekernelroh: how much do those "triple head resolution expanders" cost?07:56
lekernelmh, $280. good! they're expensive :)07:57
rohah.. didnt know08:02
renzohello again10:34
renzoi've found this : http://www.sparkfun.com/products/9910:35
renzodo you think this kind of module would be easy to connect on mm1 ?10:36
larscis the minimac driver supposed to work?14:45
mwallehi17:08
draskohi17:21
mwalledrasko: i thought you would submit some code or patches..17:23
draskofinaly it turned out to be very specific17:23
draskoand I guess useles17:23
draskofor the rest of the world17:23
draskowhot can be usefull is the breakpoint manipulation17:24
draskowhich was not present in your code17:24
draskoSo, what I commited was AMR946e target that was missing :http://openocd.git.sourceforge.net/git/gitweb.cgi?p=openocd/openocd;a=commit;h=9e3d43cfe75df7c4f6797d630576f1a02428b21817:27
draskoPoint is that for LM32 all the commands processed by VHDL code are completely specific for the core that I have (that was customised hevily), so I did not want to commit these.17:28
draskomwalle, did you progress further on OpenOCD port for Milkymist core ?17:31
mwalledrasko: not really17:36
draskoOK, then I will take up from the file you sent me, and try to update it wit all the general/non-specific stuff that can be relevant for Milkymist and send it by the end of the wee.17:37
drasko*week17:37
draskoI will not be able to test anything17:38
draskoI corrected several things, I thing cashed registers handling on break mode entry...17:39
mwalledrasko: hm there was some rewrite some months ago, you can find the updated source at my git repository17:39
mwallehttp://git.serverraum.org/?p=mw/openocd-lm32.git;a=summary17:40
mwallebut thats wip :)17:40
draskoOK, thanks17:43
draskomost of your code was very correct17:43
draskoI had most of the trouble on commands part17:43
draskobecause I did not know the format of the commands (documentation lost), so I had to look though VHDL code. Also Lattce's monitor comes as a closed source, so I had to disassemble it and reverse engineer the commands to send to it.17:45
draskoI basically added single stepping, and breakpoint handling17:46
draskoand just corrected few minor things in your code17:47
draskoand it works now correctly17:47
larscyeay!17:48
draskolarsc, not for Milkymist, though :)17:49
larscis the jtag interface different ?17:50
draskocompletely17:51
rohoh?17:52
draskobut if Michael got Milkymist jtag iface working with UrJTAG, I guess it will not be a problem, because he knows all the commands needed to pass to the TAP controller17:53
draskoI'll integrate the changes of OpenOCD code, and send some instructions how I made my stuff working17:54
mwalledrasko: that would be cool :)17:54
draskoI can not test anything for Milkymist because I do not have platform, but it should be very close17:54
draskomwalle, just for my information, what is the cirrent status on your OPenOCD port in the sense of functionality ? I guess you can send BREAK, and communicate with Monitor ROM ?18:02
draskoI.e. what is working for you and what is not ?18:02
mwalleiirc breakpoints and reset were working the last time i tested18:04
mwallesinglestep would be cool :)18:05
mwallelarsc: apropos gdb, we should redefine the registers (of course breaking lattice/theobromas/jons toolchain)18:06
mwalleas you noticed there are some missing :)18:06
draskoMost of the time I spend with toolchain18:09
draskoLattice's toolchain sucks, GDB is complaining of Dwarf header error18:10
draskoso I used gcc-4.5.1, although Lattices toolchain whorks also when recompiled with new binutils (and preferebly new GDB also).18:11
mwalledrasko: for elf binaries?18:12
draskoyes18:13
mwallegn818:35
draskobye18:37
--- Thu Jan 20 201100:00

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!