#milkymist IRC log for Tuesday, 2011-12-20

Action: kristianpaul reads01:07
kristianpaul_Operator is a bit crazy, guess python fault here01:12
kristianpaulah, yes cluttered same as with _Assign, but it have method, very easy to read01:14
kristianpaulof course _Operator should not be used, i see why ;)01:15
kristianpaul"breaking down a Migen system into multiple sub-systems, possibly                                                                                     using different clock domains."01:22
cladamw( M1rc3 yield ) 85 sets until now.03:57
xiangfuabout the /ssd mode. after rebase rtems-yaffs2.git to last upstream. now it's 0777 again.05:51
xiangfuoh. I have send the rtems-yaffs2.git 20 patches to yaffs2 upstream.05:51
xiangfuthis patch fix the ftpd root problem. http://www.rtems.org/pipermail/rtems-devel/2011-December/000087.html07:02
GitHub115[flickernoise] xiangfu pushed 1 new commit to master: http://git.io/cUHP_g07:04
GitHub115[flickernoise/master] assign FTPD_ROOT to /ssd/ - Xiangfu Liu07:04
xiangfuafter fix the ftpd_root. now only 'nautilus' give me some errors. filezilla works fine now.07:05
cladamwxiangfu, have you met that mouse works well on both usb port in GUI but got 'RX timeout error' in 'boot.crc.be5b0e7.bin' ?08:33
xiangfucladamw, oh. then the 'RX timeout error' should be also under GUI. have to check the serial output when boot to GUI?08:34
cladamwxiangfu, and VID & PID are all detected well.08:34
xiangfucladamw, this is a regression on new USB software core.08:34
xiangfucladamw, yes. all works fine only give RX timeout error.08:34
cladamwthere's no show 'RX timeout error' serial output when boot to GUI.08:35
cladamwxiangfu, hmm.... so I'd have to use GUI to double confirm the test results now. :008:36
xiangfucladamw, wait. maybe because the recently commit in milkymist.git? I will build a boot.crc.bin now. by using the milkymist.git 1.1 tag08:37
xiangfucladamw, try again with this boot.crc:  http://downloads.qi-hardware.com/people/xiangfu/tmp/boot.crc.be5b0e7.bin08:40
cladamwxiangfu, before you compile a new one. could you check that 2011-11-29 version with 'boot.crc.be5b0e7.bin' to see if the same as mine?08:40
xiangfucladamw, see if there still RX timeout error.08:40
cladamwxiangfu, okay.08:40
xiangfucladamw, run the 'md5sum boot.crc.be4b0e7.bin' in your local.08:41
xiangfuI will upload the md5sum to server08:41
cladamwxiangfu, where's the folder md5sum supposed to be in my local folder?08:45
xiangfucladamw, the boot.crc....bin md5 have add to : http://milkymist.org/updates/current/for-rc3/md5sums08:49
cladamwxiangfu, great ! there's no 'RX timeout error' msg in the new 'boot.crc.be5b0e7.bin'.08:49
xiangfucladamw, ok. then there is some regression in recently commit.  :)08:50
xiangfucladamw, will replace new one in server08:50
cladamwxiangfu, but wait a moment. let me test all items to see if else. :-)08:51
cladamwxiangfu, seems good but only 0045 and midi.09:00
xiangfuold 'boot.crc...bin' don't have midi error?09:01
cladamwthe phy id 0045, we knew it already. and the midi test result will show up 'PASS' at first round/time, if you retest midi 'i' item, then it will show 'failes', check it out if you are the same as mine. :-)09:01
cladamwuse this:  http://downloads.qi-hardware.com/people/xiangfu/tmp/boot.crc.be5b0e7.bin09:02
cladamwmaybe I didn't notice well when i used previous bin. but you can check to retest 'i' for midi then if show failed. :-)09:03
xiangfucladamw, sorry one moment.09:14
cladamwxiangfu, yes, i can surely confirm that 'retest' midi by selecting 'i' will always show 'failed' now. :-)09:18
cladamw( M1rc3 yield ) 86 sets until now.09:55
wpwrakthe RX timeout is a regression in the USB firmware. i'll fix it, just didn't have time to look into it.10:06
wpwrakflickernoise doesn't report thnigs like "RX timeout" but they still happen. you just don't see them :)10:07
wpwrakalso, "RX timeout" means that one transfer failed. not that the whole communication broke down. USB will just retry.10:08
cladamwwpwrak, ha ... thanks !10:10
wpwrakso if your mouse works, there's nothing to worry about. and if it doesn't work, try another one ;-)10:11
cladamwoah ~yes. since I was testing a new rc3 board, just found. :) Thanks told me this. ;-)10:13
lekernel86/90? that's pretty cool10:41
cladamwlekernel, these days I'm still fixing remaining boards. :-)10:42
lekernelwhat are the most common problems?10:43
cladamwvictims( clean ) after several times of reworks and unqualified usb transceiver.10:45
wpwrakalmost 96% yield. not bad at all. i think we started with something like 50% :)10:48
lekernelit was less than that, thanks to the counterfeit buffers10:49
lekernelmaybe 20% or something10:49
wpwrakah yes, the very early beginnings :) adam found those quickly, though10:51
wpwrakazonenberg: yeah. the dark side of china :)15:11
azonenbergThey have many dark sides, thats just one of 'em15:11
azonenbergI wouldnt think anyone would counterfeit a buffer though lol15:11
azonenbergusually i'd expect something cheap like a buffer to be sold as something more expensive like a MCU15:11
azonenbergbut what do you pass off as a buffer that's worth even less than a buffer?15:12
lars_a package with pins and nothing else ;)15:13
azonenberglars_: I suppose thats possible, but where do you get one?15:15
azonenbergMore likely possibility is factory scrap15:15
azonenbergwith a bad die inside15:15
wpwraklars_: naw, it does a little bit more. what's funny is that the package marking does differ. so they're not just rejects15:15
azonenbergIf you have any of them left15:16
azonenbergmy microscope is hungry15:16
wpwrakazonenberg: someone actually made an effort of faking them, believe it or not ;-)15:16
azonenbergi'd love to see one of them naked15:16
wpwrakyou'd have to ask adam15:16
azonenbergwhats his name on here? or is he AFK atm15:16
wpwraklately, cladamw15:18
wpwrakhere yuo can see how the package markings change: http://en.qi-hardware.com/wiki/File:M1_rc3_0x37_u20s_mark_is_faked.png15:19
wpwrakand it's a very fancy chip: a SN74LVC1G17DBVR :)15:20
azonenbergLooks like worse quality engraving and the lines arent straight15:20
wpwrakdetails are here: http://en.qi-hardware.com/wiki/Milkymist_One_run_3_schedule#Classifications_of_Failure15:21
azonenbergProbably IRL its a vreg or something15:21
azonenbergBecause at 11 cents per 500 it'd be hard to fake economically unless the source material was REALLY cheap15:21
azonenbergWhere did you get the bad parts?15:22
wpwrakwolfgang told a story that sounded like it's some chinese (taiwanese ?) mom and pop store15:23
wpwraksome things are just scary and don't make sense :)15:31
lars_the marking almost looks like it was drawn by hand15:34
lars_or using a cheap stencil15:34
azonenbergLol, probably a cheap stencil15:36
azonenbergwas it painted on or laser engraved?15:36
azonenbergNot that lasers are very expensive15:36
wpwrakhand-drawn .. original contemporary chinese folk art ;-)15:37
wolfspraula happy kristian paul?17:05
wpwrakwolfspraul: did you brnig the chinese firewall to germany ? or is it now the bundestrojaner that disconnects you ? :-)17:08
wolfspraulperfecting my configuration ;-)17:09
lekernelwolfspraul: where are you?17:12
lekernelbbl... going to the semi-local hamradio meeting... with a camera this time :)17:14
kristianpaullekernel: radio cool, ask then for satelite bolitha or similar18:56
rohlekernel: do i understand right that arbitrary shapes and or alpha masks (your last mail) means that we could do mapping with the m1?21:43
--- Wed Dec 21 201100:00

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