#milkymist IRC log for Tuesday, 2011-08-02

wolfspraullekernel: yes :-) we will replace all u7/u19/u20 with new parts00:24
wolfsprauland for rc4 we will source more from digikey/mouser00:25
wolfspraulHodapp: hey, welcome00:33
wolfspraulwell developed sounds good, thanks00:33
wolfspraulwhich parts do you like?00:33
wolfspraulindeed we work hard on this, for a number of years already actually :-)00:33
wolfspraullekernel: here's some good news: the first boxes were made, and fit the Fedex Medium Box well! Phew. I will spare you the details...00:38
wolfspraulleaflets are printed, brochures will be done on Friday.00:39
wolfspraulmaybe we make new stickers of the new Milkymist logo on a very nice waterproof synthetic material. we see.00:39
wolfspraulroh's package was sent to Taipei yesterday00:39
xiangfuhttp://fidelio.qi-hardware.com/~xiangfu/compile-log/openwrt-milkymist.minimal-08022011-0333/ now the milkymist openwrt start building. sorry I forget restart the cron jobs.01:38
rohwolfspraul: i forgot the stickers. will add them to the 2nd shipment03:10
rohjust the full logo with the url below on a 10x10 piece of paper03:10
rohits unintrusive advertisement.. i also have lots of logos from other hackspaces or so on the walls in the agency03:11
rohsome do 'buttons' to put on clothes.. they also are easy to fix to the garment tapestry we have03:12
rohbtw.. i do still have one flourescent red and one violet case kit around (completed some month ago)  and ideas? or should i wait and see if they get rare and expensive collectors items? ;)03:15
rohs/and ideas/any ideas/g03:15
wolfsprauldon't know03:15
wolfspraulwe will sell them, no worries03:15
rohyeah. we'll see. i dont have pressure there03:15
rohhm. should add the mic hole maybe.. otherwise.. for rc2 it doesnt matter. but i can add the shield if thats not in there yet. (need to check)03:16
rohthere will be some shielding sheets and buttons as well as 'only the hat missing' button 'pre part' as spares in the second shipment03:17
rohthe second box is still not 'full' by weight btw..03:18
rohits >5kg, but below 10. i guess if you need something other <1-2kg i could add it without extra charges from dhl03:19
HodappGiven the open source nature of this project, are there files I could laser cut myself for the acrylic case?04:05
rohHodapp: there are04:12
rohshould be in the git repo04:12
Hodappalright, cool04:13
rohwe can also sell single cases if you already got a board04:13
Hodappmy local hackerspace (they're on freenode too, #hive13) has a laser cutter so I figured I'd try my hand at making the case myself04:16
rohHodapp: make sure your material is 3.0mm or you will have to make the slots wider04:34
wolfspraulHodapp: do you have an m1 board?04:38
wolfspraulwhat have you done with it so far? any plans?04:38
Hodapproh: noted04:39
Hodappwolfspraul: I don't. I need to look into getting/making one.04:40
wolfspraulif you make anything, make an improvement right away04:41
wolfspraulrealistically you will look at tens of thousands of USD spent though, whether it's in working hours or cash. Try to find something small that you can take out, improve, and then make. that could work...04:42
Hodappwhat's an m1 board cost?04:43
wolfspraul499 USD04:43
wolfspraulthe full product, not just board. including case, accessories, etc.04:43
wolfspraulwhat makes you interested in Milkymist? what's your background?04:44
Hodappdegree was electrical engineering04:47
Hodappbut I have dabbled a lot in pushing pixels around to make interesting graphics, and I realized that a huge amount of what I had coded was basically a half-assed VJ system04:48
wolfspraulah nice04:49
Hodappand I was just poking around github and found this04:51
Hodapp...I need to learn to use git.04:51
Hodappmost source looks like C (for flickernoise at least)... that makes me happy05:03
lekernelwe are not welcome to the "open hardware summit". they said it was technical and geeky. wtf.08:45
wolfspraulwhich summit?08:48
wolfspraullekernel: did you plan to go to New York for that meeting?09:11
wolfspraulthey are probably indeed flooded by stuff and don't understand how Milkymist differs from the boatload of arduino projects09:12
GitHub177[autotest-m1] xiangfu pushed 2 new commits to master: https://github.com/milkymist/autotest-m1/compare/4e53273...bede88809:41
GitHub177[autotest-m1/master] add script file for append image's crc and length to boot.bin - Xiangfu Liu09:41
GitHub177[autotest-m1/master] Makefile: add create boot.crc.bin - Xiangfu Liu09:41
xiangfuhttp://dpaste.com/584319/  I can only get three values at the end of the file.09:45
xiangfuI can sure I append the correct value to the end of boot.bin09:45
lekernelxiangfu, you do not have to copy the image to SDRAM before CRC09:50
lekernelyou can crc directly a pointer in flash09:51
lekerneland either way, there's the compiler for you to handle memory allocation; no need for hacks like unsigned int membase = SDRAM_BASE + 2000000;09:51
lekernelgrmbl. cvs $id$ tag + git merge = time wastage.10:27
lekernelwhen will people stop using cvs ...10:27
kristianpaul(open-hardware-summit) yeah, not for milkymist. at least not the geek side ;) at least your presententation will show current status of Vjing with open hardware, art related work, what people doint wiht it11:32
kristianpaulsome kind simlar to what rejon showed at fisl..11:32
kristianpaulto get an idea of course :)11:33
kristianpaulor something like the history behind milkmist manufacturing? seems the like that work too, isnt?11:33
lekernel[lekernel@localhost rtems]$ git merge -s theirs master11:39
lekernelCould not find merge strategy 'theirs'.11:39
lekernelAvailable strategies are: octopus ours recursive resolve subtree.11:39
lekernelok, for some weird reason that particular one needs -X and not -s11:41
GitHub91[rtems] sbourdeauducq pushed 1 new commit to mmstaging: https://github.com/milkymist/rtems/commit/14b56ec2a07fe587009b49f3186946628e96938211:59
GitHub91[rtems/mmstaging] Merge branch 'master' into mmstaging - Sebastien Bourdeauducq11:59
GitHub187[rtems] sbourdeauducq pushed 1 new commit to mmstaging: https://github.com/milkymist/rtems/commit/0304c8c0ae7cdce0274aaf10a6c7095e9bd9d71412:01
GitHub187[rtems/mmstaging] Remove unmerged ChangeLog entry - Sebastien Bourdeauducq12:01
lekernel(gdb) info all-registers12:37
lekernelIE             0x40146ec0107508089612:37
lekernelah, the interrupt regs are not initialized on the stack by the gdbstub crt ...12:47
GitHub80[extras-m1] sbourdeauducq pushed 1 new commit to master: https://github.com/milkymist/extras-m1/commit/751a2ff97c41f4150612963bb23c21a0dea36b9e13:35
GitHub80[extras-m1/master] Remove oshw logo from brochures - Sebastien Bourdeauducq13:35
xiangfulekernel, http://pastebin.com/NkTrRwwR here is the new tests_image.c14:16
xiangfumy problem is when I add some "printf" at line 105, I got different result.14:17
xiangfufor example: I add "printf("  %08x:%08x\t", file_end-1, images[i][1]);" to line 105, I can got 5 correct values.14:17
xiangfuwithout this "printf", all values I got from "*file_end++" is 0,14:18
lekernelxiangfu, good14:18
lekernelexcept this problem of course14:18
lekernelfile_end-1 is a pointer14:19
lekernelyou meant *(file_end-1) ?14:20
xiangfuI just print the address for debug.14:20
xiangfuif I add printf("hello world\n") at line 105, it give me 6 correct value.14:23
xiangfulike random printf give random correct values. but I am sure the value at the end of boot.bin is correct.14:24
xiangfupushing last append_crc_len.sh to github14:26
GitHub28[autotest-m1] xiangfu pushed 2 new commits to master: https://github.com/milkymist/autotest-m1/compare/bede888...a0c226714:26
GitHub28[autotest-m1/master] clean boot.crc.bin and tmp files - Xiangfu Liu14:26
GitHub28[autotest-m1/master] we need bios-rescue-without-CRC.bin, not bios-rescue.bin for factory flash - Xiangfu Liu14:26
GitHub46[rtems] sbourdeauducq pushed 1 new commit to mmstaging: https://github.com/milkymist/rtems/commit/212e17a77652f785760375bdbdd4a865eadb041b14:33
GitHub46[rtems/mmstaging] simplify isr fix - Sebastien Bourdeauducq14:33
lekernelxiangfu, but printf("  %s\t", images_name[i]); doesn't "fix" the problem?14:45
lekernelyou can try to have a look at the disassembler listing14:47
lekernelbut it's pretty weird14:48
xiangfuseems different boot.bin length give different correct values.14:52
GitHub124[rtems] sbourdeauducq pushed 1 new commit to mmstaging: https://github.com/milkymist/rtems/commit/f4e74941eb89629058742e9b076d0e6d548d0c7014:53
GitHub124[rtems/mmstaging] merge - Sebastien Bourdeauducq14:53
lekernelah, maybe it's an alignment problem?14:58
lekernelthe unsigned int * pointer should be pointing to an address which is a multiple of 414:58
lekernelcheck that14:59
xiangfuit is multiple of 4.15:04
xiangfutime to sleep. work on that tomorrow.15:05
wolfspraullekernel: do you remember the video decoder rework picture? as seen on page 32 of this presentation http://lekernel.net/presentations/Milkymist_THSF2011/mm_thsf.pdf16:12
wolfsprauldo you have a link to the original somewhere?16:12
wolfspraulit's a nice picture...16:13
mwallelekernel:  whats your gdbs version?21:01
mwallelekernel: posted a patch on the mailinglist, just compile tested :)21:19
mwallelarsc: theres quite a large discussion about that gcc bug ;)21:24
lekernelmwalle, thanks :)21:27
mwallelekernel: mh, maybe we should fix IE21:29
mwalleeg store IE.EIE in IE.IE in case of an exception on loading and the other way around on storing21:30
mwalleand IE.BIE rsp21:30
lekernelfrom my understanding IE is not affected by debug exceptions... is it?21:31
mwalleIE is disabled21:32
mwallepc <- DEBA21:32
lekernelbut setting IE.IE to 0 doesn't disable debug exceptions21:32
mwalleno but a debug exception stores IE.IE in IE.BIE and disables interrupts21:32
mwalleso if you do a info registers, IE.IE will never be 121:33
lekernelah, ok21:34
mwallebut not now, im too tired to do sth right ;)21:34
lekernelso debug exceptions are just pulsed, they don't stay asserted like interrupts21:35
mwallethe cpu interrupt won't stay asserted either21:36
mwallewell, in case no interrupt is pending ;)21:37
lekernelyes :)21:37
lekernelbut what prevents the PC from getting stuck at the interrupt vector is that IE.IE is set to 0 upon entering the ISR21:37
lekernelit's a different mechanism for debug exceptions21:37
mwallefor debug exceptions IE.IE is set to 0, too21:38
mwallenon-debug: BA <- PC, IE.BIE <- IE.IE, IE.IE = 0, PC <- DEBA + offset21:39
mwallethats of course debug21:39
mwallenon-debug: EA <- PC, IE.EIE <- IE.IE, IE.IE = 0, PC <- EBA (or DEBA) + offset21:40
mwalleand bret does PC <- BA, IE.IE <- IE.BIE21:41
mwallewhile eret does PC <- EA, IE.IE <- IE.EIE21:41
mwalleso yes there is no ack or pending exception, but thats true for all exceptions including interrupts21:45
GitHub167[milkymist] sbourdeauducq pushed 1 new commit to master: https://github.com/milkymist/milkymist/commit/c48437b9ccbafc49dbfcc22cc695e0081ae6f6f021:45
GitHub167[milkymist/master] lm32: gdbstub save and restore IE/IM/IP - Michael Walle21:45
GitHub99[milkymist] sbourdeauducq pushed 1 new commit to master: https://github.com/milkymist/milkymist/commit/e6000e3d468f5c4d53c4cf450a87d19894e142e321:47
GitHub99[milkymist/master] Update debug ROM - Sebastien Bourdeauducq21:47
mwallebtw no one answers my pull/commit requests for the 0.15 qemu branch :(21:47
mwalleedgar, the one who committed all my patches, seems to be on vacation, no one else casres21:48
lekernelah, the joys of open source :)21:49
lekernelI can write a mail to Fabrice Bellard, but he's not very active on QEMU anymore21:49
lekernelthat's all?21:51
lekernelthere was something with the framebuffer as well iirc21:51
lekernelmwalle, do we need only the USB "full speed" patch?21:54
mwallelekernel: i dont think fabrice has commit access anymore21:55
mwalleframebuffer? dunno21:56
lekernelwell, does everything work for you with this one patch?21:57
mwallelet me test with a current flickernoise bianry22:01
mwallelekernel: whats the size of the fbi header again?22:05
lekernel8 bytes22:05
lekernelthere's also the elf online, but with 7MB of extra debug info22:06
mwallelekernel: visuals, network, audio is working22:10
lekernelok :)22:10
lekerneli have sent the email22:11
mwalleand of course mouse and keyboard ;)22:11
mwallethere was the framebuffer resize bug, but the fix for that is already upstream22:11
mwallelol, even webupdate works22:13
--- Wed Aug 3 201100:00

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