#qi-hardware IRC log for Monday, 2011-07-04

kristianpaul:D mail replied sooner than i tought !01:37
kristianpaulhttp://paste.debian.net/121798/01:37
kristianpaulwellcome back xiangfu and wpwrak ;-)01:38
xiangfu:)01:39
kristianpaulbtw xiangfu implemnted ramdisk on rtems was not so hard after all https://github.com/kristianpaul/flickernoise/blob/cc201af6a304a880a2523e135c8cc7c1486fdffd/src/main.c01:40
kristianpauljust if you may need it some day01:41
xiangfukristianpaul: thanks01:41
kristianpaulwpwrak: btw i asked osgps developer (Clifford Kelley) about that fdopen and he said it works very well unless  another process uses too much cpu for even a short time it can produce a buffer overrun xD01:45
kristianpauland it can very exiting have real data _but_ i lost bandwitch 01:46
kristianpaulwith namuru seems there is no problem at all, why? because it asumes real data, but as i have complex i just send the I and Q  data from sige to the right mixers i think01:48
kristianpaulanyway, i just read that from the gp2021 datasheet so should not be a problem do same with namuru01:48
cmwkristianpaul: even at boot time when its plugged in still doesnt show up. The driver loads but i dont see it being identified or eth0 :)01:51
kristianpaulwhat dmesg show?01:53
cmwnothing.. just the driver loading (even when i plug it)01:54
kristianpauli encorage you to write mail lists, there are few people having this wifi-sd card, i dont and my 01:54
kristianpaultry rmmod and modprobe it may be?01:54
cmwkristianpaul: will try recompiling the kernel. It's possible its something with the stock01:54
cmwyeah tried that no go01:54
kristianpauli dont and so i really dont know posible issues with it*01:56
cmwthanks anyway01:57
qi-botThe build was successfull, see images here: http://fidelio.qi-hardware.com/~xiangfu/compile-log/openwrt-xburst.full_system-07032011-0915/13:07
kyakthe only guy talking :)13:26
kristianpaulwhy kyak ?14:03
wpwrakkyak: maybe you want to share some experience with the usd wlan thingy with cmw. he has one and it doesn't seem to work (or maybe the driver is busted)14:06
kyakwpwrak: it's also not working for me ;)14:42
wpwrakkyak: do you know someone who has it working ?14:44
zrafawpwrak: you mean wifi usd thing?14:47
wpwrakzrafa: yup14:47
wpwrakzrafa: btw, welcome back on #qi-hardware ! it's been a long while ! :)14:48
zrafawpwrak: I helped somebody but it did not work for him. I did a specific kernel for jlime for him.. but no luck. No idea why14:48
zrafawpwrak: it was a lot of months ago via qi ML14:48
kyakwpwrak: both xiangfu and wolfspraul reported it working14:49
wolfspraulzrafa: yes, happy you are there!14:50
wolfspraulI miss your drinks14:50
zrafawpwrak: welcome: he :-) It is hard for me to follow the irc these days.. that is why I do not get here so much. I can not read the logs either :P14:50
wolfsprauland werner's meat and wine14:50
zrafaa disaster from my side14:50
zrafa:P14:50
wolfspraulzrafa: nah that's ok, just don't forget us entirely :-)14:50
wolfspraulyou get older and wiser, and then you share all the wisdom you gained here :-)14:50
zrafawolfspraul: meat, wine, drinks : haha.. me too! :-). But only if no germany and arg are going to play some match :D14:51
wolfspraulI think I want to leave the Spectec nightmares behind14:52
wolfspraullong live ben-wpan14:52
wpwrakzrafa: ah, i thought your internet at home was working ? btw, does the heating ? ;-)14:52
wpwrakwolfspraul: yeah. more reliable than spectec ;-)14:52
zrafawpwrak: both things work for now.. I will touch wood anyway.. (no sure if "yo toco madera" expresion exists in english)14:53
wpwrakzrafa: yeah. also in german ;-)14:54
zrafawpwrak: great, very known it seems :)14:55
wpwrakstefan_schmidt: ahh, you're back. excellent ! :) how are the atusbs ?17:30
stefan_schmidtwpwrak: yeah, had some weekend dates. Birthday, 10 years "Abitur" meeting etc17:31
stefan_schmidtwpwrak: back to work now.17:31
wpwrakstefan_schmidt: busy week then ;-)17:31
stefan_schmidtwpwrak: reworked it to async usb and still getting scheduling while atomic17:31
stefan_schmidtwpwrak: yup, it was also my 30th birthday. means a lot planing for party etc17:32
stefan_schmidtanyway, alcohol should be gone by now and I should be able to start again17:32
wpwrak(scheduling) hmm, nasty. what's the call stack like ?17:32
stefan_schmidtah, and I'm now using kvm with usb passthrough to avoid all this rebooting17:33
wpwrak(partying) oh yes, i had plenty of that in the last days :) yesterday, i just slept for 12 hours17:33
wpwrakmodule unload still not working then ?17:34
stefan_schmidtwpwrak: ha, 12 hours would have been great. I had 4 hours each night. Only came into bed when the light was back already :)17:34
stefan_schmidtwpwrak: Doing all build-in tests with kvm now.17:35
wpwrakthe nights before the 12 hours were significantly shorter :)17:35
stefan_schmidtWill come back to this topic once the other stuff works17:35
stefan_schmidtheh17:35
wpwrak(modules) okay17:35
stefan_schmidthmm, I can copy'n'paste the call stack from kvm17:35
stefan_schmidts/can/can't/17:36
stefan_schmidtwpwrak: give me a moment to figure this out17:37
wpwrakscreen dump ?17:37
uncloudedis CONFIG_MAC802154 absolutely essential for dirtpan to work?  The 2.6.39-3 Ubuntu kernel has CONFIG_IEEE802154 and CONFIG_IEEE802154_DRIVERS as modules but CONFIG_MAC802154 isn't mentioned.  does this mean I won't be able to use this kernel for dirtpan?17:37
wpwrakhmm, i would think it is needed, yes. haven't tested that hypothesis, though17:38
uncloudeddo you think it's worth me trying without it to find out?17:38
wpwrakah yes, you need it. things like the function that calls xmit are in there ;)17:38
wpwrakoh wait17:39
stefan_schmidtunclouded: you need it. The at86rf230 driver needs it17:39
stefan_schmidtand the driver is not in mainline yet due to this17:39
stefan_schmidtwpwrak: (screen dump) thats lazy17:40
wpwrakno. it's like that. you need it.17:40
uncloudedah ok, that will explain why it isn't in a packaged Ubuntu kernel17:40
wpwrak(lazy) if it works ... :)17:40
stefan_schmidtunclouded: yeah, its only in the linux-zigbee tree right now. The other parts you are seeing are already upstream17:41
stefan_schmidtwpwrak: pfft :)17:41
uncloudedok, will see if I can patch the Ubuntu kernel with it17:41
stefan_schmidtwpwrak: http://pastebin.com/hhnEf3KL17:42
stefan_schmidtwpwrak: full dmesg17:43
stefan_schmidtwpwrak: first call stack is the interesting one17:43
stefan_schmidtwpwrak: What I read out of it so far (could be wrong)17:43
stefan_schmidtwpwrak: __spi_async is the last function form the spi framework calling our .transfer method17:44
stefan_schmidtwpwrak: looking at spi_async it does spin_lock_irqsave() spin_unlock_irqrestore around the __spi_async call17:44
stefan_schmidtwpwrak: and then our driver comes in with USB and it gets scheduled...17:45
wpwrakhmm, seems that you're not quite async yet. can you push your latest version ?17:45
stefan_schmidtwpwrak: yup, moment17:46
qi-bot[commit] Werner Almesberger: install/ben-wpan-config-2.6.38: enabled CONFIG_PROC_PAGE_MONITOR (master) http://qi-hw.com/p/ben-wpan/0c7765817:47
qi-bot[commit] Werner Almesberger: Merge branch 'master' of projects.qi-hardware.com:ben-wpan (master) http://qi-hw.com/p/ben-wpan/848e5af17:47
qi-bot[commit] Stefan Schmidt: spi/atusb: Move read1 to async usb. Broken right now. (ben-wpan-stefan) http://qi-hw.com/p/qi-kernel/5e5d49117:48
stefan_schmidtwpwrak: there you go.17:48
stefan_schmidtwpwrak: read1 in async17:48
stefan_schmidtwpwrak: I'm still doing the wait_for_completion in the same function doing the urb_submit. Did not wrap my head around doing a full dispatcher here17:49
stefan_schmidtwpwrak: that could well be the problem. Never worked on a async usb driver before17:49
wpwrakwait_for_completion is pretty obviously wrong :)17:50
wpwraknow. let's see what you could use instead ...17:51
stefan_schmidtwpwrak: well, thats what I have seen in other usb drivers17:51
wpwraknot in async ones that actually worked :)17:52
stefan_schmidthmm17:53
Action: stefan_schmidt ponders and does a grep17:53
Action: stefan_schmidt ponders more and should read the wait_for_completion implementation...17:53
wpwrakno no .. wait_for_completion sleeps. that's all you need to know about the implementation :)17:55
wpwrakyour async function can't call anything that sleeps17:55
wpwrakin atusb_usb_cb, instead of calling complete, you should call (spi_)msg->complete(msg->context)17:56
wpwrakbasically move everything after calling read1 into the callback. now there's one more difficulty: some messages have two transfers17:57
stefan_schmidtwpwrak: drivers/usb/usb-skeleton.c was the one I peaked into17:58
stefan_schmidtwpwrak: And yes, sleeping and async does not make sense together17:58
wpwrakso you'd need to chain them in the callback. maybe move the urb submission simply into a suitable function that takes a struct spi_transfer and that calls a callback when done17:58
wpwrak(usb-skeleton) skel_read isn't async17:59
stefan_schmidtwpwrak: grr, why don' they use the sync usb api then. Fooling me all over :)18:00
wpwrakand you need to set master->bus_lock_flag. else, you may get new requests while still working on an older one18:01
wpwraksee also drivers/spi/spi.c:spi_async18:02
stefan_schmidtwpwrak: so in summary: Kick off the completion stuff, move msg->complete and msg->status into the callback and set the bus_lock_flag18:04
stefan_schmidtwpwrak: and as second step take care of messages with two transfers18:04
wpwrakyes. you can do bus_lock_flag later. you're probably nicely synchronized from the callers anyway18:06
stefan_schmidtok, lets see how this works out18:06
wpwrak(2nd step) or do them both at once. what you want is a function that submits the "next" transfer.18:06
qi-bot[commit] Werner Almesberger: ubb-vga.c (usage): correct synopsis (-r is now called -m) (master) http://qi-hw.com/p/ben-blinkenlights/92ce05318:10
wpwrakbtw, since spi_atben.c is now in the ben-wpan branch, i just deleted the ben-wpan-atben branch18:15
wpwrakstefan_schmidt: btw2, i'll remove the interrupt redirection from spi_atben soon, since i can just pass the real interrupt. so if you want a copy of that code, you'll make it now or dig a bit into the history18:17
stefan_schmidtwpwrak: just mention interrupt redirection in the commit message and I get it from there18:19
stefan_schmidtwpwrak: already merging the plain ben-wpan branch nowadays18:19
wpwrakperfect18:19
qi-bot[commit] Werner Almesberger: spi_atben: removed interrupt redirection (ben-wpan) http://qi-hw.com/p/qi-kernel/9e362c819:17
wpwrak83 lines shorter :)19:17
qi-bot[commit] Werner Almesberger: switch linux-zigbee tools source from (old) tarball to git repository (master) http://qi-hw.com/p/ben-wpan/ae348fb23:36
qi-bot[commit] Werner Almesberger: tools/Makefile (BEN_DIRS): add dirtpan iff its lowpan tools dependency is met (master) http://qi-hw.com/p/ben-wpan/71f095d23:36
qi-bot[commit] Werner Almesberger: xxx (master) http://qi-hw.com/p/ben-wpan/6d1198c23:36
qi-bot[commit] Werner Almesberger: tools/: moved get_key from atrf-path/gui.c to libatrf, for sharing (master) http://qi-hw.com/p/ben-wpan/56f8b2d23:36
qi-bot[commit] Werner Almesberger: tools/atrf-rssi/gui.c (gui): accept keyboard input both from stdin and SDL (master) http://qi-hw.com/p/ben-wpan/10f618a23:36
qi-bot[commit] Werner Almesberger: Merge branch 'master' of projects.qi-hardware.com:ben-wpan (master) http://qi-hw.com/p/ben-wpan/f27990723:36
wolfspraulwpwrak: good morning :-)23:50
wolfspraulso you are back in Buenos Aires now?23:51
wolfspraulwhat's your take in FISL?23:51
wpwrakwolfspraul: yup, back home and well rested :)23:54
wpwrakwolfspraul: it was pretty good. people liked what rejon and i said and had to show. we may get a few customers for ben, atben/atusb, and M1 :)23:55
wolfspraulany wisdom you learnt that I should now learn from you?23:55
wolfspraulhow many people were there? how many young people (say < 25, or < 20)?23:57
wpwrakwolfspraul: one guy was also starting to think of producing things in brazil (to escape the horrendous import taxes). he has done some small things already. he's also thinking of making a wpan+arduino blend.23:57
wolfspraulnice23:57
wpwrakat the talks ? hmm, mine had about 50. rejon got 30-40. define "young" :)23:58
wolfspraulyoung < 25 years old, or better yet < 20 years old23:58
wpwrakpeople were very excited about the M1 demos23:58
wolfspraulI hope a download link for your talk will emerge23:58
wolfspraulI mean for the video23:58
Action: kristianpaul hope the same23:59
--- Tue Jul 5 201100:00

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