#qi-hardware IRC log for Saturday, 2012-03-17

Action: pabs3 wonders if all the happy Debian users are using Debian on their nanonote00:48
zrafalinux kernel + busybox counts as a distribution? :P00:59
wpwrakzrafa: that would solve the "messy boot process" issue :)01:27
wpwrakalright. time to reinstall ... the old system was shot too badly to recover01:33
wpwrakoh joy, a read error on the ubuntu install CD01:45
kristianpaulzrafa: i wish :)01:55
kristianpaulpabs3: you trying?01:56
kristianpauli dont as i said,perhaps i dont need all tha debian offer in a embeded device... as rafa said linux and busybox userland are a good start02:06
kristianpaulplus the custom ports and builds :-)02:06
zrafawhat what what?02:11
zrafafor text console only applications debian should be one of the best for nn02:15
zrafayou have thousands of ready packages to test.. Of course a lot of them useless in nn.. But many console applications will be there as well02:16
kristianpauluseless, afaik..02:17
kristianpaulzrafa: i will try OW first, at least more apps than in owrt02:19
kristianpauland cool guis (X11) :-)02:19
zrafagreat stuff : X Factor by Guy Gerber...02:21
kristianpauldont remenber that :)02:21
zrafakristianpaul: what is OW?02:21
zrafaah :)02:22
zrafayes.. OE is the best thing to have a repository for embedded devices02:22
kristianpaulalso dunno how optmized mips from debian is for nn.. i guess similar history to mipsel for lemote yeeloong02:22
kristianpaulalso i can run owrt for milkysmist :)02:24
kristianpaulso i'll stick to it for a while too 02:24
vanoQuite busy here! 08:33
vanoI am trying to get debian booting on a uSD card on the nano note, but when booting into the SD card, in the boot screen I get the line "ousted. Opts: (null)fs (mmcblk0p1): re-" (all that I can see) and I cannot get a login console screen with the alt+ctrl+F2.08:35
vanoWHen I boot back into the NAND and check the logs I don't get to see anything special...08:35
vanoI have already tried this several times, but no go. Any tips?08:36
vanoI can boot into NAND with the uSD inside, so there doesn't seem to be anything wrong with fsck blocking the boot...08:37
wolfsprauldon't know, but that looks like your Debian boot process doesn't get very far08:59
AylaThe list of accepted organisations for GSoC 2012 is online: http://www.google-melange.com/gsoc/accepted_orgs/google/gsoc201211:05
pabs3Ayla: can you translate from JavaScript for us?11:17
Aylayou don't have a JS-enabled browser?11:18
Aylapabs3: here's a CSV: http://pastebin.com/C0KzKwQW11:19
pabs3I have one but I'm not willing to turn on JavaScript11:19
qi-botThe build was successful: http://fidelio.qi-hardware.com/~xiangfu/build-nanonote/openwrt-xburst.full_system-20120316-1257 13:55
infobotDocScrutinizer51 meant: Hi15:39
DocScrutinizer51who's admin of qibot?15:39
lindi-who needs such bots anyway...15:41
DocScrutinizer51me, when somebody asks me what's the meaning of NFC15:42
DocScrutinizer51others obviously to get notifications about build process15:42
DocScrutinizer51qi-bot: status15:43
DocScrutinizer51qi-bot: help15:44
DocScrutinizer51damn ansty bot15:44
wpwraklindi-: we should get DocScrutinizer a pet rock. then he won't feel so lonely and doesn't need to surround himself with robots :)15:44
DocScrutinizer51~seen wolfspraul 15:45
infobotwolfspraul is currently on #qi-hardware (16m 43s), last said: 'don't know, but that looks like your Debian boot process doesn't get very far'.15:45
wpwrakDocScrutinizer: by the way, any comments on those ethernet transformers ?15:52
wpwrakDocScrutinizer: i have to admit that all this stuff is black magic to me :)15:52
DocScrutinizer51so what kind of comment are you interested in?15:53
DocScrutinizer51the transformer(s) just do some glavanic separation15:55
DocScrutinizer51wpwrak: so waht's the black magic about all that?15:58
DocScrutinizer51wpwrak: or are you talking bout sth else than what's commonly known as ethernet PHY?15:59
wpwrakno, it's that PHY16:02
wpwraki don't know if the various coil arrangements are really equivalent16:02
wpwrakalso, does polarity matter ?16:02
DocScrutinizer51think of it as a number of unrelated baluns16:03
DocScrutinizer51about polarity I'm not sure, but I'd guess it should be taken care of though driver chip will recover from twistedvswapped pairs I think16:04
DocScrutinizer51tbh I never looked into 'modulation' on nBt16:05
DocScrutinizer51you can loo it up in wikipedia for sure16:06
DocScrutinizer51I'd expect a datasheet for ethernet PHY component usually not elaborating on ploarities and coil arrangements, but rather name pins in a defacto standardized way, so you can do the "chinese EE" (connect pins with similar names)16:09
wpwrakdid you real my mail about the ethernet PHY ?16:10
wpwrakthe problem is that the thing we have is just not a 1:1 copy of the recommended PHYs16:11
DocScrutinizer51hmm, if you think I should, I might look into it16:13
wpwrakthat would be great, thanks ! :)16:13
wpwrakadam also seems to be mainly based on the assurance of the manufacturer that it's a valid replacement16:13
DocScrutinizer51I could do now if you post the URL to PHY DS16:13
wpwrakdid you install dsv ? that would give you instant access to all those goodies16:14
wpwraki set you a detailed description of how to do it16:14
DocScrutinizer51nah. nfc what that is, for sure not available on N90016:14
wpwrakah no, needs a real computer :)16:15
DocScrutinizer51my 'real computer' is in maintenance mode16:15
larscdsv = datasheet viewer?16:19
DocScrutinizer51NRZ(I) is basically not polarity aware16:21
wpwraklarsc: yup. you give it a list of all the datasheets your project uses with a (set of) name(s) for each. it downloads and caches them. then you look them up with   dsv <name>16:21
larscwpwrak: were do i get it?16:21
wpwrakDocScrutinizer: excellent. so the winding direction won't matter ?16:21
DocScrutinizer51prolly not much16:22
wpwrakgood. the chip doens't seem to have anything that lets you swap + and -. although i have to admit that i understand only a fraction of the items discussed there16:23
DocScrutinizer51in 100BT iirc only two pairs of the 4 are used, each for one direction16:24
DocScrutinizer51and if there's only one pair available, some chips do halfdumplex16:25
DocScrutinizer51which then is slow as hell16:26
wpwrakmaybe it does that, too16:26
wpwrakit even has a cable fault analysis function :)16:26
DocScrutinizer51as I see the chip block schem, it has TX and TX'16:28
DocScrutinizer51I don't see a function block to swap TX and RX for that auto thing 16:31
DocScrutinizer51so you need a special '0modem' patchcable to connect two identical devices with taht PHY16:32
DocScrutinizer51and I guess that also means it can't do halfduplex16:32
wpwrakit has TX/RX swapping16:36
DocScrutinizer51ooh it's no complete PHY as it needs a transformer16:36
DocScrutinizer51what's the black magic in that?16:37
wpwrakyes. that transformer is what gives me some headaches :)16:37
wpwrakfor tx/rx swap, see for example page 2116:37
DocScrutinizer51you need one 1:1 transformer for TX and one for RX16:37
wpwrakdo you have my mail with the ethernet review at hand ?16:38
wpwrak(subject something like "M1r4 review: ethernet")16:38
DocScrutinizer51no mail on N90016:38
DocScrutinizer51I've read it though16:39
wpwrakhere are some of the transformers they recommend:16:39
DocScrutinizer51can't recall the whole 16:39
DocScrutinizer51so what?16:39
wpwrakjust look at their structure16:39
DocScrutinizer51I guess they have 4 (or 2*4) pins16:39
wpwrakthis is the critter we have: http://downloads.qi-hardware.com/hardware/milkymist_one/datasheet/Ethernet/RTF-114B8A1A(023-00).pdf16:40
wpwrakcheck your copy & paste :)16:42
DocScrutinizer51no c+p, opens directly from xchat16:43
DocScrutinizer51the ( spoils it16:43
wpwrakxchat gets the parenthesis wrong16:43
DocScrutinizer51anyway, afk for some shopping and a proper argentinian beef ass16:44
wpwrakyou eat the ass of beef ? :)16:45
DocScrutinizer51what else is steak?16:46
DocScrutinizer51if only I could find a proper visor to watch the 'screen' while walking16:48
wpwrakshoulder and back16:48
wpwrakthe car drivers in your city will be thankful if you don't find such a visor too soon :)16:48
DocScrutinizer51how so? they are more scared by the guy watching his phone rrather than street DAMN almost ran into a biker16:49
DocScrutinizer51I'm not talking about opaque visor16:50
wpwrakthe best is the tenderloin (lomo in spanish)16:50
DocScrutinizer51whatever that means16:51
DocScrutinizer51there's also sir*16:51
wpwrakin german, ":ende"16:52
DocScrutinizer51k cya16:54
larsclende gut alles gut?16:56
rjeffrieswpwrak what is status of atben and atusb? I think hardwrae can be ordered from sharism? what about the code for each?16:59
wpwrakrjeffries: i think wolfgang has all the stock from tuxbrain now, yes16:59
rjeffrieswpwrak for example, on atusb, is have drivers benn pushed upstream?16:59
rjeffriesI am out of date sorry16:59
rjeffriess/is have/have17:00
wpwrakrjeffries: code is still the same. works quite well for some simple things, but you wouldn't want to use for, say, bulk transfers with tcp17:00
rjeffrieswpwrak so is there a 6LoWpan layer under Dirtpan?17:00
wpwrakno, there's no 6lowpan when there is dirtpan17:00
wpwrakdirtpan is directly on ieee 802.15.417:01
rjeffriesactually that could be useful in some cases17:01
rjeffriesis it a subset of 802.15.4 I assume, or..17:01
rjeffriess/it it/is it/17:02
wpwrakthe ieee 802.15.4 implementation on linux was a subset of the standard, yes. and probably still is. i haven't tracked it for a while.17:03
rjeffrieswpwrak I sit fair to say (no judgement implied) that your level of interest in the atben/atusb has faded, and you are now on Milkymist primarily (where you are making major contributions IMO)?17:22
rjeffriess/I sit/Is it/17:45
infobotrjeffries meant: wpwrak Is it fair to say (no judgement implied) that your level of interest in the atben/atusb has faded, and you are now on Milkymist primarily (where you are making major contributions IMO)?17:45
larscwpwrak: where can i get that vds thingy?17:50
wpwraklarsc: it's in project eda-tools. here's how to set it up for use with M1r4: http://lists.milkymist.org/pipermail/devel-milkymist.org/2012-March/002894.html19:43
wpwraklarsc: or a more manual use would be with ben-wpan. there you'd "make dsv" in the project's base directory and could then "dsv txrx" (transceiver) etc.19:44
larscok thanks19:45
wpwrakhmm, interesting. there doesn't seem to be an easy way to resync a git repository whose database has become corrupted20:31
wpwrak(still cleaning up things after the file system mishap ...)20:32
wpwrakhm. just cloning a new. there's very little unpushed data anyway. (or so i hope ;-)20:45
viricAyla: I see there is a "NMI Watchdog" in linux...20:59
viricthat could detect lockups no?21:00
viricx86 only...21:01
derRichardhi, long time no see :)21:26
viricAyla: http://www.mjmwired.net/kernel/Documentation/nmi_watchdog.txt21:26
derRichardcan i use a mainline kernel on ben?21:28
larscyes.it will lack some features though21:29
larscmainly usb gadget support21:30
derRichardis the gadget driver too ugly for mainline? ;)21:30
derRicharddoes it support dma? AFAIK it was broken when i tried it last time21:31
larscdma works. sg dma not21:31
larscscatter gather21:32
derRichardlarsc: so, using ben as usb disk will not be damn slow? :D21:32
larscah, you mean the gadget driver21:32
derRichardthis issue i mean21:33
derRichardi've reported it some time ago21:33
larscactually i have no idea if dma works for the gadget driver21:33
derRichard1 year 9 months ago it did not work :D21:34
derRichardlarsc: see "static unsigned int use_dma = 0;" in jz4740_udc.c21:54
viricAyla: got the debug with kexec + crashdump in qemu x86_64! first steps... gdb loads it fine, and the gdbmacros work well too22:03
viricI've just seen "perf kvm top". incredible.22:27
vanoHi! Any links showing how to make the debian .uni image to include the remaining 1.5GB (or at least .5GB more) of the NAND?22:34
vanoor to make the .ubi image from the debian.tgz? 22:34
viricvano: it's partitioned so you get 0,5GB of openwrt, 1,5GB of /home22:34
vanoI know. But I would like it to be 1GB - 2GB of / 22:35
viricvano: and both debian and openwrt kernels have that partitioning hardcoded.22:35
vanoah I see.22:35
viricvano: changing one of them, could make users destroy filesystems22:35
vanounderstand& so basically I would have to build my own kernel for this...22:36
vanoI am trying to install too many things (debian) and run out of space...22:36
viricvano: or change its commandline. Many years ago, there was a tool to change the kernel commandline in an image22:36
viricI wonder if it still exists22:37
viricvano: ah you have debian in the 0,5GB of / ?22:37
vanoMaybe a better idea would be to move something like /usr/bin to the SD or something like that ?22:37
vanoI absolutely cannot get debian boot from the uSD22:37
vanotried everything...22:38
viricmkdir /home/usr22:38
viriccp -a /usr/* /home/usr22:38
viricmount -o bind /home/usr /usr22:38
viricmount --bind22:38
vanotrue :)22:38
vanothat should do it&. 22:38
vanoAny ideas as to debian / on the uSD? 22:39
vanoI cannot see the error message& I was going to try with an older boot without the fb screen& but didn't get to find any.22:39
vanoI have ruled out any fsck issues (or so I think)& cause I can get to boot into NAND with the uSD inserted...22:40
viricvano: you have to press a key on boot22:41
viricvano: the uboot source has a check, if a key is pressed, it picks the kernel from the uSD, assuming it's in a known filesystem22:41
viricand a particular partition.22:41
vanowhich partition?22:42
viric#define CONFIG_BOOTCOMMANDFROMSD    "mmc init; ext2load mmc 0 0x8060000022:42
viric/boot/uImage; bootm"22:42
vanothat is the correct partition I have& the first partition (and only one in my case)22:43
vanothe thing it is that, it seems to boot 22:43
vanobut I cannot get to the consoles via ctrl+alt+F222:44
vanothe message logs (that I check later when booting into NAND) don''t show anything wrong either..22:44
vanohowever of course I cannot get to see the console messages properly.22:44
viricdo you want to do some kernel debugging? :)22:44
viricyou need the serial wire: http://viric.name/cgi-bin/nanonixos/doc/trunk/doc/serial.wiki22:46
vanoam not a hw guy& but will look at that :)22:48
vanobtw just in case.. the lat message I get in the console when booting into the uSD is: "ounted. Opts: (null)fs (mmcblk0p1): re-m"    .. just in case you could guess what it could be :)22:49
viriconly that? it looks like with broken words22:49
viricI've no idea on the debian kernel, I never used it. I simply imagine they prepared it right.22:50
vanoyou are not aware of any problems with the size of the uSD? this is 8GB.22:52
viricI used at most 2GiB22:52
viricbut never booted from sd22:52
vanowell the uSD seems to work ok mounted when booted into NAND.22:53
viricuboot may be more stupid22:53
viricthan linux22:53
virictake a picture of your nanonote screen with debian booted22:54
viricanyway I go to bed22:54
vanok c ya.22:55
derRichardwhere can i get the sources of xburst-tools?22:58
wpwrakroh: do you know an algorithm/library/etc. that produces a toolpath for milling a 2D area delimited by a polygon ?23:34
--- Sun Mar 18 201200:00

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