#milkymist IRC log for Monday, 2012-03-19

Fallenouthank you for your help wpwrak & lekernel00:00
Fallenougn8 !00:00
wpwrakit was fun :-) sweet dreams !00:00
Fallenouyep :'00:01
Fallenouthanks see you00:01
wolfspraulcladamw: hi good morning! :-)00:50
wolfspraulanother new week, wow00:50
cladamwhi good morning !00:51
wolfspraulwhat's the plan this week?00:54
wolfspraulschematics are finished now? onto layout?00:54
wpwrakwe still have a few open issues from the review. and two more things need looking at, SDRAM and the scary big FPGA00:56
wpwrakis would actually be good if sebastien could check the FPGA. i don't know a tenth of the quirks and restrictions it may have on its I/Os00:57
wpwraks/is/it/00:57
wolfspra1lhmm, I see01:03
cladamwwpwrak, power supply sheet replies from Werner I'll look at it, then we check fpga late tonight, I would combine all replied sheets then send a final M1R4, is it okay ?01:04
wpwraksounds good01:13
wpwrakpity that we had so little public response to the review. based on the number of quite insightful suggestions we got on previous hardware issues, i had hoped for a bit more interest01:14
wpwrakwell, maybe people just find reviews boring :)01:14
wolfspra1lnot enough people that feel this is 'their' platform01:19
wpwrakperhaps, yes. they're not used to co-ownership yet01:20
wolfspra1lI think review can be understood as satisfying and fun, but you need to feel how this connects to your own use in the future01:20
cladamwboring may just not meet one's won designs. :( One may reply when seeing similar sch sheet alike 'their' platform.01:20
cladamws/won/own01:20
wpwrakyes, there's always that chance01:21
kristianpaulfpga review looks time demading plus *i think* it should have done just once, or at least defiend over a baseline to follow01:43
kristianpauli dunno that last part in the milkymist one history..01:43
kristianpaulas wpwrak said, there are tenth quirks and restrictions about asigments on all over the banks about I/O, clocks, buffers and wonder what else i miss right now..01:45
kristianpaulkind stuff imho should do just once :-)01:47
wpwrakthat's basically my plan: look at the things we've changed ... and hope for the best for the rest ;-)01:51
kristianpaul:)01:52
kristianpaulwpwrak: that rule will apply for bitstream as well?02:01
wpwrakmh ? we don't review the bitstream :)02:04
kristianpaulsure, just kidding02:16
wpwrakbesides, it's software. it can ripen in transit ;-)02:21
kristianpaulthats the worst part indeed02:22
wpwrakthe ripening ?02:23
wpwrakwhy ? it ensures it arrives fresh02:23
kristianpaulhmm?02:23
wpwrakwould you rather have it full of dessicated bugs ? ;-)02:23
kristianpaullol, you are very good with words :)02:25
wpwrakmaybe i should consider a political career02:26
kristianpaulin argentina? lol02:26
wpwraki think, with a little effort, i can be very corrupt02:26
wpwrakand i've read Machiavelli. what else do you need ?02:27
kristianpauljust you ;)02:28
kristianpaulergh, i corrupted the standy bitstream.. and forgot we're still "using" it02:29
kristianpaulfor recovery right? no way get rid of it  :-)02:29
wpwrakyou can skip it via jtag :)02:30
kristianpaulhmm truth, where is the scriping fu when you needed02:38
kristianpaulbut solved, i just reflash from my otherm milkymist upstream folder02:38
stekernFallenou: you could add a watchdog to the wishbone accesses and generate an error on timeout03:32
stekerninstead of just having the cpu hanging03:32
GitHub58[flickernoise] xiangfu pushed 1 new commit to master: http://git.io/Cpk7zA03:32
GitHub58[flickernoise/master] Ctrl+H: shortcuts help: combine some help messages - Xiangfu03:32
Action: xiangfu try to use wpwrak's usb command for write full parse HID report :-)03:32
wpwrakit's a messy one, isn't it ? :)03:33
Action: xiangfu have more mouse that didn't working. the X/Y goes to 1024 etc..03:33
xiangfuwpwrak, trying ...03:33
xiangfuwpwrak, which file I should load ? softusb-input.bin?03:36
xiangfuno needs reboot any more and netboot. that same a lot of time03:41
xiangfuwpwrak, great feature for me. thanks03:42
wpwrakah. now i understand which "usb command" you meant :)03:44
xiangfuwpwrak, I am try to add another one "usb restore" :-)03:48
wpwrak"return to factory settings" ? :)03:53
wpwrakyou could of course just keep the original image on the M1, in ssd03:54
xiangfuwpwrak, yes. that one option.04:02
xiangfuI have a mouse. the hid report that have 10 buttons. but in fact it is just normal mouse.04:03
xiangfulike:04:03
xiangfu        Usage Page (Button),            ; Button (09h)04:03
xiangfu        Usage Minimum (01h),04:03
xiangfu        Usage Maximum (10h),04:03
xiangfu        Logical Minimum (0),04:03
xiangfu        Logical Maximum (1),04:03
xiangfu        Report Count (16),04:04
wpwrakyou mean 16 buttons ?04:15
wpwrakwell, maybe that little mouse has great ambitions for the future ;-)04:15
xiangfu:-)04:15
wpwrakin any case, you should skip two bytes for its buttons, no matter how many there really are04:15
xiangfuI will only take care mouse hid descriptor on mouse buttons and x/y04:16
wpwraka wise decision :)04:16
cladamwwpwrak, just measured U12 for actual temperature. :-) Your concern is right, but I don't think we need to change. :) A 478 mW(i.e. 39 C ) would be there for 149 mA out for video-in codec. The next problem is to put M1 in ambient 70 C is common ? or even 45 C ?06:56
cladamwif over 39 C, U120 almost starts to keep on 150 mA as constant.06:58
cladamws/U120/U1206:58
wpwrakwhat is your room temperature ?07:00
cladamw(C145/C234) wpwrak do you think it's okay for you if we change to 1uF/10uF or both 10uF ?07:00
cladamwmy room temperature is now 23 C.07:00
wpwraki'm just answering your mail :) (C145 et al. yes, perfect)07:01
wpwrakso add 16 C for a hot day :)07:01
cladamwwpwrak, yeah..sure, if put M1 in a hot day. :-)07:02
wpwrakwell, mine survived such days. though i usually don't have the camera running. maybe i should :)07:04
cladamwit's good you can confirm and have a try, then we know the answer will get closer. :-) Happy to solve such mistakes though. :-)07:06
cladamwalso about R62's 0.1W/0.05W in TI's ds. phew~07:06
wpwraksubtle, eh ? ;-)07:10
wpwrakbut it seems that we can also keep the 1/16 W. but better safe than sorry07:10
wpwrakparticularly since the price is the same :)07:11
cladamwyes, let's change for 0603 package.07:11
cladamwsince we don't know which one is correct is in TI's ds, so 0603 package( 0.1W ) is no 'sorry' .:-)07:12
lekernelFallenou: CSRs (the bus ones), ethernet memory, and USB memory are non cacheable09:23
GitHub76[flickernoise] xiangfu pushed 1 new commit to master: http://git.io/Pq1E1g09:27
GitHub76[flickernoise/master] shellext: add usb restore for restore to default firmware - Xiangfu09:27
Fallenouok09:31
Fallenouit's fixed now, the weird too high address is not accessed any more09:32
Fallenougcc was missing some piece of information in the inline assembly to generate correct code (register cloberring)09:32
Fallenou04:37 < stekern> Fallenou: you could add a watchdog to the wishbone accesses and generate an error on timeout < that would be cleaner indeed but what will the cpu do if wishbone says "ERR" ? it will retry ? and fail again ? thus hanging the whole system again09:34
cladamwwolfspraul, rii 2.4G keyboard received. it contains its own usb cable.09:48
cladamwxiangfu, new image supports Rii mini keyboard ?09:52
xiangfucladamw, not the release image.09:54
xiangfucladamw, if you want test the Rii mini keyboard. reflash it with --qi option.09:54
cladamwokay09:55
cladamwxiangfu, thanks. :)10:00
wolfspraulcladamw: what do you mean with 'its own usb cable'?10:02
wolfspraulyou should have received a version that works with a wireless dongle10:02
wolfspraulyou plug the wireless dongle into the m1, and use the riitek keyboard and mouse remotely10:02
wolfspraulis that what you got?10:03
cladamwwolfspraul, Rii mini equipped with its own usb cable to recharge battery.10:05
wolfspraulah yes, true10:05
wolfspraulbut you have the wireless version, right?10:05
cladamwright10:06
wolfspraulgood10:08
cladamwxiangfu, which version is for Rii ? I mean "--qi [VERSION] [--data]"10:11
xiangfujust reflash_m1.sh --qi10:12
xiangfuis ok.10:12
cladamwokay10:12
cladamwha~ nice ! control remotely very well. :-)10:31
wolfsprauland mouse included too10:32
wolfspraulwe should have shipped like this from day 1, oh well10:32
wolfspraulbut we can only learn as fast as we can, and improve, and keep moving10:32
cladamwyes, well.. but this tested by my own applied usb power switch patch, need to reopen new packed m1 to test again. :-)10:33
cladamwxiangfu, if I've never reflashed by '--qi --data', then directly '--qi', there's failed during compiling patch, so i go for ''--qi --data' again and without failure. right ?10:59
cladamwxiangfu, i seems like that from me, but I'd like to know as usal shipment steps of reflash. :-) Now i have to do:11:01
cladamw1) reflash_m1.sh --qi --data 2) reflash_m1.sh --bios-mac XX XX11:01
cladamwcorrect me if i am wrong.11:02
xiangfucladamw, (failed during compiling patch) it's normal. because there are some update on patches. if11:04
xiangfucladamw, --qi --data will overwirte all patches. update them to latest.11:04
cladamwokay, so each m1 from me, must those two steps above. :-)11:04
wpwraksome day, we should think about what to do about patch and FN compatibility12:01
mwallelekernel: wpwrak: why is sync 0x80 ? and not 0xff or 0x00 ?18:59
wpwrakerrr, context please ?19:00
mwalleoh *g* :)19:00
mwalleUSB19:00
mwallesoftusb19:01
mwallebut i think i already found it19:01
wpwrakyeah. it's just how the pattern is defined. blame intel ;)19:05
mwallewpwrak: yeah spec says, max. change of bits, so with nrzi that would be 0x00, i guessed19:08
wpwrakdoes it say that ? i see i section 7.1.10 "he SYNC pattern used for low-/full-speed transmission is required to be 3 KJ pairs followed by 2 K's for a total of eight symbols."19:22
wpwrakso that's 7 changes (0) followed by one non-change (1)19:22
lekernelmwalle: that's to provide an oscillating pattern to make the clock synchronization easier on the receiver19:32
lekerneland the last bit (MSB) to 1 disables the last transition, marking the end of the sync19:34
mwallewpwrak: yeah you're right, dunno where i had that info from19:34
mwallelekernel: so that marking is just an extra indication which can be checked19:39
mwalleor is there any technical reason for this?19:40
larscyou need to know where the packet ends19:41
wpwrakit's basically the start bit19:41
wpwrake.g., the receiver may not see the entire SYNC pattern19:42
qi-botThe firmware build was successful, see images here: http://fidelio.qi-hardware.com/~xiangfu/build-milkymist/milkymist-firmware-20120319-2040/21:19
qi-botThe firmware (using branch) build was successful, checkout the VERSIONS for detail, see images here: http://fidelio.qi-hardware.com/~xiangfu/build-milkymist/milkymist-firmware-20120319-2219/22:59
--- Tue Mar 20 201200:00

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