#milkymist IRC log for Sunday, 2011-05-15

Fallenouyes00:09
Fallenouthat's why I think /ramdisk is just IMFS00:09
Fallenousince it's a directory in the rootfs00:09
Fallenougotta go to sleep00:09
Fallenousee ya00:09
kristianpaulbye00:09
kristianpaulhe, first crash... lets see what i did wrong00:10
kristianpaulreading seems okay, but for reasong writing crash the system :/00:11
kristianpaulbut simulation went okay..00:19
kristianpaulwell, writing to the gps "buffer" is not a wanted feature, but i still want to know what's wrong..00:20
rejonwhat filesystems can the mm1 read from that microsd?00:32
rejonkristianpaul Fallenou lekernel00:34
kristianpauli had tested fat16 so far rejon00:35
rejonnot vfat00:35
kristianpaulhmm00:37
rejonyes00:39
rejonvfat is default fat1600:39
rejonthat workred kristianpaul ?00:39
kristianpauli'm noob on filesystems, i just got a microsd wolfspraul sent me, pluged to my copmuter and gparted told the fat1600:40
kristianpaulrejon: yeah, but not with my 1Gb kingstone and 512Mb toshiba microsd.. :-/00:40
kristianpaulrejon: are you having problems reading a memory card?00:41
kristianpaulSandDisk is the brand of the card00:43
kristianpaulthe other one seems unbraded (not counting a shiny HBTX sticker on the bottom)00:44
rejonok00:48
kristianpaulI want to mean, memory card reader in milkymist may be not work with all cards.00:50
kristianpaulAlso not write support avaliable00:50
wolfspraulrejon_: memory card support is buggy, you need a little luck that the card you have works00:53
wolfspraulor maybe it's not buggy, but our drivers are not robust/flexible enough to handle the infinite variations on the standard you find in real life cards00:53
wolfspraulthis actually reminds me again of the value of including a 1gb or 2gb card together with the product...00:53
wolfspraulanother thing to look into, maybe00:53
wolfspraulrejon_: I've sold my last rc2 unit!!!00:53
wolfspraulseriously that's big news I think, for the project00:53
wolfspraulthe ability to move ourselves forward with cash income is the #1 thing we need for long-term success00:53
wolfspraulso as of today, I'm officially out of stock, which is bad (but good motivation)00:55
rejon_yes, great00:55
rejon_an african guy00:55
rejon_wanted to buy one on the spot00:55
rejon_yesterday00:55
rejon_and have another offer for something in vietnam00:56
wolfspraulouch00:57
wolfspraulman now I have no stock00:57
kristianpaullekernel: btw, you should  take a look at http://nuttx.sourceforge.net/, seems have lots of missing drivers in rtems like SD/MMC support and USB host stack (with included debug)00:57
wolfspraulwell u know me, now I can hit myself to work faster :-)00:57
Action: kristianpaul will take a look too00:57
wolfspraulrejon_: there may still be a few units at tuxbrain and hackable devices/kd8500:57
wolfspraulwere you able to reflash your m1 with the latest software?00:57
rejon_yes01:09
rejon_the script helps a lot01:09
rejon_taking to this party in an hour01:09
rejon_plugging it in01:09
rejon_man, i have a lot to learn01:10
rejon_http://www.milkymist.org/wiki/index.php?title=Flashing_the_Milkymist_One01:10
rejon_this page needs massive love01:10
rejon_can someone put the link to the reflash script on there01:11
rejon_http://www.milkymist.org/wiki/index.php?title=Flashing_the_Milkymist_One#Easy_JTAG_Method01:11
kristianpaulminute01:11
kristianpaulhe, you are good with mediawiki "tags"01:12
kristianpaulcool, nuttx allow execute separte linked programs too01:19
wolfspraulif you reflash the latest 05/09 snapshot, it should boot right into rendering mode (although boot time will be ca. 35 seconds)01:21
wolfsprauland after that you can click through patches with the left button (the one towards the IR)01:21
Action: kristianpaul boots in 3 seconds with no gui :D01:21
wolfspraulthe reason the boot time increased so much is because we currently compile all patches, there is no precompilation or caching yet01:21
wolfspraulwe'll get it back down01:21
kristianpaulwell 5 seconds..01:21
wolfspraulbut you can click through all patches, which is nice01:21
kristianpauloh01:21
wolfspraulkristianpaul: I'm talking about the 05/09 image, which will get you into a mode where you can click through all patches, and they all need to be compiled first (currently)01:21
kristianpauli dint knew it that (compile all patches)01:22
kristianpaulyes sure wolfspraul01:22
kristianpaulha, you mising ack wire!!02:13
kristianpaulokay, now other hours to test this ix02:13
kristianpauls/ix/fix02:13
kristianpaulhour*02:13
kristianpaulhum i think build-rescue is not been delete with make clean at flash/ dir and also byt clen_all13:12
kristianpaulorr..13:12
kristianpaulis the third time i haved to delete the content of that folder my self..13:13
wolfspraulkristianpaul: don't understand. what do you want?13:30
wolfspraulyou are proposing a fix to a build target?13:30
kristianpaulmay be, yes13:33
kristianpauli need to confirm first..13:33
wolfspraulah ok, I misread what you said13:33
kristianpaulno no, i wast to clear on word, i admin13:33
wolfspraulI thought you said "this is the third time I propose to ..."13:33
wolfspraulbut you said "this is the third time I had to delete..."13:34
wolfspraulgot it now :-)13:34
kristianpauli just noticed this, because a non-ending error was solved by deleting the content of build-rescue13:34
kristianpaul;-)13:34
wolfspraulsure - inadequate error handling in build scripts can lead to many unpleasant and hard to track down surprises13:34
CIA-51flickernoise: Sebastien Bourdeauducq master * r464f056 / patches/VideoIn/Jeremie - conduirebourre.com.fnp : New patch - http://bit.ly/iZzbJG14:26
kristianpaularghh, timing was not met :S14:28
kristianpaulhere we go again..14:36
kristianpaulwolfspraul: i'll install Xilinx ISE on fidelio, is okay with you?14:43
kristianpaulwee 11.2M/s14:44
kristianpaul:-=14:44
kristianpaulcloud computing is not bad at all ;-)14:44
wolfspraulsure of course if that helps you14:45
wolfspraulnot sure the specs of the server are enough for it, but if you need anything, let me know14:45
wolfsprauldownload speed should be good, yes :-)14:46
kristianpaulseems as fast as my pc, well it is Athlon mine is sempron14:47
kristianpauland NO i will not even try to run two P&R instances.. on my pc, i dont want end burning out something14:47
kristianpaulargh, what is wrong... with 0xf0000000 ..20:07
kristianpaulhum, the system freeze even from the upstream bitstream version, so it should be newby mistake related with pointers..22:47
lekernelwhat are you trying to do?22:51
kristianpauli cloned the minimac core, in order the re-used the dual ported ram memory, plus add the shift core that sync data coming from the L1 SiGE front end22:52
kristianpaulso I decied wrote a small commad to write and read that memory22:53
kristianpaulreading is okay. but when i write from this address (0xf0000000) the system just crash22:53
kristianpauli tried same procudere from soft usb and ethernet cores, and no crash22:55
kristianpauli tried with my hacked milkymist core and the upstream one22:55
kristianpaulsame problem22:55
kristianpaulthe comand is this http://paste.debian.net/117067/22:55
kristianpauls/procudere/procedure22:56
kristianpauls/cores/address22:57
lekernelyou can use mr/mw in BIOS instead of writing your own code23:01
lekernelthat's what those commands are for23:01
kristianpaulyeah but i wanted to do it from rtems as a custom command23:06
lekernelget it to work from the bios first ... then rtems23:07
kristianpaulwell i could hack the bios..23:07
kristianpauli'll do23:07
kristianpaul:-)23:07
lekernelnah, use mr/mw first23:07
kristianpaulok23:07
kristianpaulhmm bios dint crashed, BUT now i realize i got some garbage back in the console if i write to 0xf000000023:09
kristianpaulalso reading the supposed written value just give 0..23:09
kristianpaulokay i need to check whats wrong here23:09
--- Mon May 16 201100:00

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