**** BEGIN LOGGING AT Tue Jan 29 02:59:58 2013 Jan 29 03:00:57 Interesting for Arm folks: http://www.wired.com/wiredenterprise/2013/01/facebook-arm-chips/ Jan 29 03:03:08 toneeee: pretty everyone has know that for about the last 3 to 5 years Jan 29 03:03:28 toneeee: http://www.calxeda.com/ Jan 29 03:03:33 alan_o: hey bud Jan 29 03:03:46 prpplague: hey man Jan 29 03:04:08 What's new in Tejas? Jan 29 03:04:31 nada, still doing the job hunting thing Jan 29 03:05:01 latest HIDAPI pull request https://github.com/signal11/hidapi/pull/100 Jan 29 03:05:21 I always love when people use words like "incredibly hard" Jan 29 03:05:30 I mean come on Jan 29 03:05:32 it's a text file Jan 29 03:05:40 what's "incredibly hard to read" about it Jan 29 03:06:28 and then, there's a guy who did some work and submitted it to my project, and I appreciate the gesture, but it's work I wasn't really interested in, because I'm more interested in how the file looks in a normal editor, than how it looks on github Jan 29 03:06:40 and I don't like telling people no Jan 29 03:08:58 hehe Jan 29 03:09:13 my 100th pull request though. Do I get a star or something next to my profile now? Jan 29 03:09:17 prplague: I've been out of the hardware side of things for so long that news is news. I mean OE was news to me. Any other pointers are appreciated :D !! Jan 29 03:09:50 alan_o: ever play a flash game called "upgrade complete" ? might be apropos here, heh Jan 29 03:10:03 toneeee: I'll check it out Jan 29 03:10:35 alan_o: it has the best ending ever! hahaha Jan 29 03:11:16 alan_o: you get star tattoos like the russian mofia gangs Jan 29 03:11:31 hidapi? Jan 29 03:11:32 tear drops? Jan 29 03:11:34 or star bellied sneetches Jan 29 03:12:43 oh that Jan 29 03:12:58 not hid over interesting transports Jan 29 03:13:23 ds2: well, unless you find usb or bluetooth interesting, then no Jan 29 03:13:37 still waiting for HID over SATA Jan 29 03:13:52 there are certain2 wire protocols that would be of interest Jan 29 03:21:04 ds2: so I'm digging through my .config and see CONFIG_I2C_HID. Is that what you're referring to? Jan 29 03:21:40 *nod* Jan 29 03:21:51 ds2: if it creates a /dev/hidrawN handle, then my hidapi lib will talk to it. Jan 29 03:22:09 oh I see Jan 29 03:23:00 haven't looked to see if USB HID and I2C HID really do share enough code to do that Jan 29 03:23:48 of course the usefulness of that would be limited probably. hidapi is mostly for shoving arbitrary data over vendor-specific HID reports, which devices do so they can get around having to make drivers for windows (and it's getting easier on Windows with winusb now). Jan 29 03:24:29 it depends... I can see it being useful for newer stuff not necessarily vendor specific Jan 29 03:25:57 the bone sure has a convoluted pinout compared to the classic Jan 29 04:03:34 ds2, depends on where you started from does'nt it? Jan 29 05:13:45 so this is embarrassing.... in my mrf24j40 driver, when I set the long address, I set it backward. Jan 29 05:13:58 never saw it before because I also had promiscuous mode on. Jan 29 05:37:36 crap it 9:30 Jan 29 05:37:38 *is Jan 29 05:44:08 time's fun when you're having flies? Jan 29 05:52:10 er yeah Jan 29 05:52:15 or insane deadlines Jan 29 05:54:21 I want a last supper with one Christ, twelve disciples, no kangaroos by Thursday lunch, or you don't get paid! Jan 29 05:57:04 well legally they have to pay me... well they can fire me too :) Jan 29 05:59:49 Russ: so i was at the range this weekend and some bozo brought a illegal AOW .. *facepalm* Jan 29 06:00:24 which kind of other? Jan 29 06:01:33 shotgun with an overall length under 26" Jan 29 06:01:55 why is that illegal? Jan 29 06:02:04 emeb_mac: federal law Jan 29 06:02:13 yeah. what's the rationale. Jan 29 06:02:23 rationale? :) Jan 29 06:02:31 i know. dumb question. Jan 29 06:02:35 * Russ reads up on aow a bit, interesting, a handgun with a foward vertical grip Jan 29 06:02:56 so if I attach a simple grip to the rail on a glock 19, it's a no-no? Jan 29 06:03:08 oh yeah Jan 29 06:03:11 mranostay, how far under 26"? Jan 29 06:03:49 looking from a slight distance i'd say under the overall length by a lot and the barrel for sure Jan 29 06:04:02 they prompted kicked him off the range Jan 29 06:04:04 geez, not smart Jan 29 06:04:15 at least it wasn't one of the zip guns or something Jan 29 06:07:34 wikipedia article on title II is interesting. Jan 29 06:07:43 lots of non-obvious no-nos Jan 29 06:07:55 emeb_mac: i own a few items.. stored out of state of course :) Jan 29 06:08:14 offshore platform? Jan 29 06:08:28 whatever state you're in, they're out of state. Jan 29 06:08:43 need to transfer them to a friend since i won't be leaving CA anytime soon Jan 29 06:08:46 * emeb_mac is in the state of confusion. Jan 29 06:08:46 emeb_mac: come again? :) Jan 29 06:12:23 emeb_mac: what State are you in again? Jan 29 06:12:45 mranostay: the great state of Arizona. Jan 29 06:12:53 the wild west Jan 29 06:13:54 as opposed to CA which is more "mild west" :) Jan 29 06:13:59 hehe Jan 29 06:14:25 my friends from Ohio think i've turned liberal to give up some gun rights :) Jan 29 06:15:02 wouldn't have thought of OH as a hotbed of 2nd Amdmnt radicalism Jan 29 06:15:32 ah maybe not nutty like Jan 29 06:16:00 i like the political nature of Ohio personally.. not too liberal...not too conservative Jan 29 06:16:22 yep - middle of the road. Jan 29 06:28:18 hmm i shouldn't possibly miss a 2p ham radio test time.. Jan 29 06:29:23 so you ham radio prefix is related to place you received your first license right? Jan 29 06:29:39 not sure Jan 29 07:06:18 hi martin Jan 29 07:08:18 whois ceeper Jan 29 07:08:22 *creeper Jan 29 07:16:41 mranostay, thats mr_creeper to you Jan 29 07:22:11 hi vpopov Jan 29 07:23:08 hi Jan 29 07:24:12 what are the cool kids using for bloging software now? Jan 29 07:25:42 <_av500_> tumblr Jan 29 07:25:59 when did google but Blogspot? Jan 29 07:26:02 *buy Jan 29 07:26:53 <_av500_> decades ago Jan 29 07:26:58 <_av500_> (in internet time) Jan 29 07:30:24 sleepy time almost Jan 29 07:30:37 mranostay: naah, just got to work Jan 29 07:31:21 well there are these timezones Jan 29 07:31:26 mranostay: yeah ;) Jan 29 07:31:39 mranostay: nothing but trouble Jan 29 07:31:56 if could work 12a to 8am to avoid everyone i would :) Jan 29 07:32:32 mranostay: I've lately had to do some work with colleagues 9 time zones away Jan 29 07:32:47 mranostay: just work from home Jan 29 07:33:11 not a option atm Jan 29 07:37:48 sleepy time for sure Jan 29 07:42:29 mranostay, I had a 5am telcon today, working 12a to 8am would solve nothing for me Jan 29 08:03:01 hi trolls Jan 29 08:05:40 hi Jan 29 08:06:22 lolololo panto Jan 29 08:06:27 that's the spirit Jan 29 08:10:04 gm Jan 29 08:10:49 woglinde: moin moin Jan 29 08:18:21 hm no cheap flight to Edinburgh Jan 29 08:18:31 * woglinde plans for elce Jan 29 08:20:22 there's no easyjet hub in berlin? Jan 29 08:21:25 it is Jan 29 08:21:43 you mean than travel with train from stanstead? Jan 29 08:21:49 or what is the nearest? Jan 29 08:22:05 stansted is so close to Edinburgh Jan 29 08:22:08 you can walk Jan 29 08:22:50 av500 good luck than Jan 29 08:24:05 wow, 164€ Jan 29 08:24:08 that is cheap Jan 29 08:26:11 last time I checked easyjet flew directly to edinburgh Jan 29 08:27:17 otherwise it would need a transfer in luton Jan 29 08:27:20 this is LH Jan 29 08:27:25 direct from FRA :) Jan 29 08:27:37 it pays to live next to a real airport Jan 29 08:28:48 av500, btw, if you've even been to cdg, I'm pretty sure the location of the british airway terminal is some kind of snide french joke Jan 29 08:28:53 koen from where? Jan 29 08:29:04 maybee I can travel there with easyjet Jan 29 08:29:27 isn't elce not until october? Jan 29 08:29:31 russ sure Jan 29 08:29:42 but why not check the flights now Jan 29 08:29:55 woglinde: 177€ with KLM for you Jan 29 08:30:31 1 stop Jan 29 08:30:37 no direct flights Jan 29 08:30:44 I'm looking at $1137 Jan 29 08:31:15 oh geez, ONT->SEA->AMS->EDI Jan 29 08:31:16 av500 yes and backflight to BER Jan 29 08:31:20 av500 thats funny Jan 29 08:31:25 ? Jan 29 08:31:25 that's only 20h Jan 29 08:31:42 woglinde: TXL Jan 29 08:31:43 return flight only 36h, thanks trip advisor Jan 29 08:31:56 Economy | Snack Jan 29 08:32:29 * Russ changes airport to LAX Jan 29 08:32:43 ams->edi has a direct easyjet flight Jan 29 08:33:01 but it looks like KLM might be cheaper Jan 29 08:33:12 can't we party for a while in london first to make the travel easier/ Jan 29 08:33:42 hmm, 55€ hotel Jan 29 08:33:48 this is all too checp Jan 29 08:33:49 cheap Jan 29 08:33:52 its a trap Jan 29 08:34:38 wow, my shortest return trip is 24hr Jan 29 08:34:55 too bad the concorde is fubar Jan 29 08:34:59 oh, wait, no KLM is more sane, 14hr Jan 29 08:35:14 LAX->AMS->EDI Jan 29 08:35:49 Russ: LAX FRA EDI would work too Jan 29 08:35:52 but I guess $$ Jan 29 08:35:52 WLG-AKL-SFO-ORL-LHR-HKG-AKL-WLG Jan 29 08:36:13 why why why Jan 29 08:36:37 Russ: btw, I have been to CDG, but long ago Jan 29 08:36:46 now I have a high speed train to paris :) Jan 29 08:37:03 the british airways counter was the one at the very end, past every other airline Jan 29 08:38:55 serves them right Jan 29 08:39:01 for being british Jan 29 08:50:07 welcome to #beagle-air Jan 29 08:51:53 "I'm sorry sir, but are not allowed to bring board files on the plane" Jan 29 08:52:18 "remove your shoes and place all device trees in the tray" Jan 29 08:52:31 haha Jan 29 09:02:07 device tray Jan 29 09:02:22 the australian variant Jan 29 09:02:56 * koen stabs of_find_node_by_name Jan 29 09:06:01 hm yes berlin amsterdam and than amsterdam - EDI might possible Jan 29 09:07:49 haha at esyjet you can only book until 20.10 Jan 29 09:07:58 thats the problem Jan 29 09:08:02 hi ant Jan 29 09:41:38 gm woglinde Jan 29 11:52:52 hi there Jan 29 11:53:52 could we run the Tesseract-ocr v3 on Ubuntu on Beagleboard-xm??? Jan 29 11:54:24 koen's rule 3: try and find out? Jan 29 11:57:52 takes too long Jan 29 11:58:32 deadlines are tight, sir Jan 29 11:58:39 money too Jan 29 12:01:54 so elce is in edinburg, neat Jan 29 12:05:20 ynezz yes Jan 29 12:05:51 dm8tbr you need to improve it Jan 29 12:05:53 http://26-26-54.hardwarebug.org/30 Jan 29 12:45:36 lunch Jan 29 12:46:52 doner? Jan 29 12:47:31 gyros Jan 29 12:47:39 where have you been? Jan 29 12:48:00 koth is doener Jan 29 12:48:17 who's koth? Jan 29 12:56:24 -> /whois kotH Jan 29 12:56:42 he's our resident terrorist Jan 29 12:57:21 smuggeld into swiss banks Jan 29 13:03:50 woglinde: did you fix meta-java yet to work with the new bitbake multiline comment thing? Jan 29 13:05:44 \o/ 3800mAh battery for galaxus Jan 29 13:05:51 looks like a hunchback now Jan 29 13:07:33 koen I fixed all stuff Jan 29 13:07:36 10€ with a 51€ voucher to print some business cards Jan 29 13:07:44 I should order hundreds, I will be rich Jan 29 13:07:45 besides shark disabled Jan 29 13:08:04 and icedtea only at 2.1 release Jan 29 13:08:17 tested it yesterday Jan 29 13:08:30 back Jan 29 13:08:38 thanks to bitbake its always a recompile of all packages Jan 29 13:08:45 jo panto Jan 29 13:09:05 av500 lol Jan 29 13:09:20 woglinde: do you need some business cards? Jan 29 13:09:20 but who goves 51€ voucher Jan 29 13:09:24 I can make you a sweet deal Jan 29 13:09:26 gives Jan 29 13:09:41 woglinde: its this stupid vistaprint Jan 29 13:10:03 its not real :) Jan 29 13:10:10 vistaprint sucks Jan 29 13:10:28 sure, they print stuff Jan 29 13:10:35 sure Jan 29 13:10:36 but it all feels a bit dodgy Jan 29 13:10:40 of course Jan 29 13:10:54 I used it once to get the free stamp Jan 29 13:11:09 spent 5€ and 100x unchecking something in their order form Jan 29 13:18:08 * av500 wonders why mranostay is constantly visiting his linkedin profile Jan 29 13:25:06 stalking Jan 29 13:25:37 av500 he likes f5 Jan 29 13:26:03 wow Jan 29 13:26:14 philips is losing all audio/video stuff Jan 29 13:26:23 now its lamps and healthcare only Jan 29 13:26:25 who's finding it? Jan 29 13:26:29 funai Jan 29 13:26:36 *cough* Jan 29 13:27:03 they are on the way to become a zombie brand name Jan 29 13:27:10 see what happens when your royalty abdicates? Jan 29 13:27:15 yeah Jan 29 13:27:26 wont have happened with her still at the helm Jan 29 13:28:20 * emeb_mac wonders who abdicated to zombify all the old US brands Jan 29 14:30:32 could someone recommend a good, recent character device driver to base a new driver on that I wish to write Jan 29 14:30:51 /dev/null Jan 29 14:33:48 so you want a driver with character? Jan 29 14:34:11 he means uart? Jan 29 14:34:37 I'l try explain Jan 29 14:35:10 I current have a driver which exposes an IOCTL which dumps a bunch of registers into userspace using copy_to_user Jan 29 14:35:44 the registers come from an FPGA which fires an interrupt once a second to say it's ready to be read Jan 29 14:36:07 at the moment, I check for this interrupt in userspace on a GPIO Jan 29 14:36:24 then issue an IOCTL call to read the registers from the FPGA through my kernel driver Jan 29 14:37:23 I now want to improve my kernel driver and have that listen for the interrupt, then write to a buffered char device(?) which my userspace can then read at will without having to worry about latency or missing the interrupt while doing data processing Jan 29 14:38:54 therefore, I wondered if anyone knows of any good examples of a character device in the kernel that does something like this Jan 29 14:39:40 the FPGA is connected over SPI, but it'ts the buffering and exposure of the device that is the tricky bit (i think) Jan 29 14:40:00 I guess you would want userspace to select() or poll() on that device Jan 29 14:40:13 yes, that is the end goad Jan 29 14:40:20 s/goad/goal Jan 29 14:40:48 I do that at the moment with the interrupt, but it would be nice to have that done in kernel and not put pressure on userspace to keep the timing correct Jan 29 14:42:09 I was thinking sysfs? isn't using /dev/ frowned upon a bit these days? Jan 29 14:42:35 why? Jan 29 14:42:59 I don't know, I just got that general impression Jan 29 14:44:23 sysfs is just for convenience Jan 29 14:46:01 jackmitchell: from your description, sounds like a job for uio Jan 29 14:46:21 use that standard api for handling your interrupt...registers mmaped via uio api Jan 29 14:46:49 just because your description sounds like you are more-or-less a userspace driver Jan 29 14:47:36 unfortuantly at the moment it is indeed half and half, I would like to push it to kernel and simplify the userspace a little bit Jan 29 14:47:52 I'll have a look in the docs about uio Jan 29 14:49:37 it's a pretty straightforward generic framework for userspace drivers Jan 29 14:49:45 might be a match for you Jan 29 14:50:05 it's the preferred way to write evil binary drivers ;) Jan 29 14:50:24 * av500 laughs demonic laugh Jan 29 14:50:40 mdp: its kinda like a stable binary driver API for linux? Jan 29 14:51:02 av500, yeah, cause the driver is all in userspace ;) Jan 29 14:51:21 av500, it's far less interesting than anybody might think Jan 29 14:51:33 it's just a standardized way to do what you were describing above Jan 29 14:51:43 but he needs to mess with th espi controler Jan 29 14:51:49 write a gplv2 kernel shim to process your h/w's interrupts etc. Jan 29 14:51:50 uio + spidev an option? Jan 29 14:51:57 ahh, I just caught the latter part of it Jan 29 14:52:23 Crofton|work: doubtful, at least not that I know of Jan 29 14:52:58 I mean uio to wait on interupt that would trigger user space to use spidev Jan 29 14:53:08 he'd need to do the buffering in the fpga though Jan 29 14:53:10 you could do that..yes Jan 29 14:53:47 the latency is just high for processing triggering that spidev transfer Jan 29 14:54:38 you've got to schedule to get the interrupt occurance into userspace..then schedule against to make the spidev xfer request from userspace Jan 29 14:54:44 but why? Jan 29 14:54:53 you would process interrupt and spi in the kernel driver Jan 29 14:54:58 in the model jackmitchell is doing Jan 29 14:55:07 and let user space read it at it's pace Jan 29 14:55:08 av500, yes, I also would ask "why?" Jan 29 14:55:16 av500, that is the proper model, yes Jan 29 14:55:21 ok Jan 29 14:55:53 but the thought experiment is worthwhile to show how inefficient it is to string these things through userspace Jan 29 14:56:00 and uio is only for mmap stuff Jan 29 14:56:02 not for spi Jan 29 14:56:09 o.O Jan 29 14:56:12 mdp: of course Jan 29 14:56:23 we can fill the channel with bad examples easily Jan 29 14:56:24 doesnt her said he wants to be in kernel Jan 29 14:56:30 ups he Jan 29 14:56:30 woglinde: get in line! Jan 29 14:56:49 more output Jan 29 14:56:53 serious uio users are using it with rt-preempt too...no surprise that hans and tglx work on those things together ;) Jan 29 14:57:28 ok, I had a quick read about UIO and it seems like it is on par with what I am currently doing Jan 29 14:57:34 av500 still in train? Jan 29 14:58:03 jackmitchell: just be clear that your latency in processing these interrupts will be quiet high when doing this as a userspace driver Jan 29 14:58:11 s/quiet/quite/ Jan 29 14:59:07 if it meets the reqmts then it's the right approach..that's all that matters Jan 29 14:59:34 * woglinde chuckles Jan 29 14:59:35 mdp: yes indeed, I want to process the interrupts in the kernel, read the registers over SPI in the kernel, then send out the data to userspace Jan 29 14:59:40 hi thurbad Jan 29 14:59:51 heya Jan 29 15:00:14 mdp: but it would be nice if there was some buffering in the kernel for say 3 interrupts worth of data (it is a set amount each time) Jan 29 15:00:15 jackmitchell: the ideal way to implement that is as a kernel driver Jan 29 15:00:42 mdp: yes, and I have a basic kernel driver at the moment, but I want to implement in it basic buffering and interrupt processing Jan 29 15:00:54 let me pastebin it so you can get the gist Jan 29 15:00:56 jackmitchell: should be straightforward Jan 29 15:01:44 mdp: laughing is forbidden :P Jan 29 15:01:47 present the data to userspace via mmap() Jan 29 15:01:47 mdp: http://pastebin.com/k7TbdbQ8 Jan 29 15:01:54 beware, "Buffering" is patented by RealVideo..... Jan 29 15:01:59 royalties might apply Jan 29 15:02:02 lol^H^H^Hok Jan 29 15:02:02 lol Jan 29 15:02:23 mdp: can mmap be buffered? i.e. if I miss an interrupt due to userspace being slow can I still access the previous data? Jan 29 15:02:36 ? Jan 29 15:02:43 you dont mmap, you use spi Jan 29 15:03:00 jackmitchell no prinkts please Jan 29 15:03:06 jackmitchell: you like blank lines? Jan 29 15:03:09 av500, I was talking about the manner in which he could present the fetched data to userspace Jan 29 15:03:21 mdp: read()? Jan 29 15:03:24 av500: it seems so :D Jan 29 15:03:37 jackmitchell: makes code hard to read Jan 29 15:03:41 less fit on a page Jan 29 15:03:56 av500, either way, depending on preference of course Jan 29 15:05:16 av500, just use case dependent depending on how often you are executing it, amount of data getting copied from kernel space..etc..as you know Jan 29 15:08:15 jackmitchell, can a queue of some sort help here? Jan 29 15:08:33 bradfa, +1 for your phrasing Jan 29 15:09:00 so as you can see, the driver is a bit of a mess, and I feel like i'm using outdated kernel paradigms (see registerDriverOnBus()) so I would like a nice, new, clean driver to base my next iteration on Jan 29 15:09:38 jackmitchell, don't worry about "doing it wrong" worry about "getting it to work" (first) Jan 29 15:09:45 then make it pretty Jan 29 15:09:47 bradfa: it works! (tm) Jan 29 15:09:53 jackmitchell, well, now make it pretty :) Jan 29 15:10:17 * bradfa scrolls back Jan 29 15:10:18 bradfa: exactly, but I still want to bring the interrupts into the kernel Jan 29 15:11:03 bradfa: and if I miss an interrupt in userspace, then I still want the old data to be buffered in the kernel, so when my userspace catches up again, it can process it Jan 29 15:11:22 and for this, a char driver would be appropriate.....? Jan 29 15:11:26 jackmitchell, what's wrong with request_irq() and gpio_to_irq()? Jan 29 15:11:55 and struct list_head Jan 29 15:11:59 bradfa: nothing, there is no irq handling in there at the moment Jan 29 15:12:03 ah, k Jan 29 15:12:08 I scrolled through paste quickly Jan 29 15:12:27 I'm thinking it is going to be better to do a fresh proper implementation, rather than hack more things in Jan 29 15:12:33 jackmitchell, I was going to compliemnt your lack of ioctl... but then I noticed... Jan 29 15:12:41 * bradfa has ioctl in his kernel driver for SPI too Jan 29 15:12:49 * jackmitchell hangs head Jan 29 15:12:57 jackmitchell, don't rewrite it, refactor Jan 29 15:13:20 bradfa: yes, ok- I wasn't going to completely chuck it Jan 29 15:13:46 do you know of an example driver which uses a buffered char device, or something similar? Jan 29 15:14:09 I'm looking for something (good) to take inspiration from Jan 29 15:14:53 jackmitchell, why aren't your read() and write() functions doing anything? Jan 29 15:15:04 jackmitchell, the ifx driver isn't a bad place to start Jan 29 15:15:08 one sec, finding name Jan 29 15:15:22 bradfa: I don't use them, I assume the tutorial I followed had them in and I never took them out Jan 29 15:15:40 jackmitchell, aren't read() and write() how you interface to userspace? Jan 29 15:15:56 * bradfa didn't read all the codes Jan 29 15:15:58 I wrote this driver maybe 9 months ago, when I had roughly 3 months of C and linux (developer) experience Jan 29 15:16:06 that's ok Jan 29 15:16:13 I'm no good at C (or any programming) Jan 29 15:16:21 * bradfa isn't even good at the Internets Jan 29 15:16:47 I hope now, with my 12 months of experience I do hero coding, if I have enough good examples.... ;) Jan 29 15:16:47 bradfa how than did you end up here Jan 29 15:17:08 jackmitchell hehe gogo and dont use printk directly Jan 29 15:17:15 woglinde, I dunno, xchat came on debian and it connected to freenode Jan 29 15:17:28 still? Jan 29 15:17:36 woglinde, debian is old Jan 29 15:17:42 I tought they swichted to ofct Jan 29 15:17:47 some ages ago Jan 29 15:17:47 I'm too poor to buy a better Linux Jan 29 15:17:55 oftc Jan 29 15:18:05 sure I am using it too Jan 29 15:18:10 since 1997 Jan 29 15:18:14 jackmitchell, see ./drivers/tty/serial/ifx6x60.c for some ideas but maybe disregard the tty stuff and use the shared workqueue instead of taskelet Jan 29 15:18:21 s/taskelet/tasklet/ Jan 29 15:18:32 thank you, I will go investigate! Jan 29 15:18:49 jackmitchell, and read the changes docs for lddv3 Jan 29 15:19:01 * bradfa forgets what he was googling for Jan 29 15:19:08 bradfa: changes docs? I've only ever seen the actual book Jan 29 15:19:32 jackmitchell, see this, too -> https://lwn.net/Articles/434833/ Jan 29 15:19:51 jackmitchell, changes -> https://lwn.net/Articles/2.6-kernel-api/ Jan 29 15:20:17 some things changed in the, what? 6 years between printing and that changes docs Jan 29 15:20:29 it's still out of date, the changes Jan 29 15:20:33 but better than the book itself Jan 29 15:20:37 for accuracy Jan 29 15:21:25 jackmitchell, there's been some patches to ifx driver I think recently, to fix some of its issues Jan 29 15:21:31 maybe worth seeing what those do as well Jan 29 15:21:35 every kernel book is outdated at printing time Jan 29 15:21:46 woglinde, well, so long as you want to run an old kernel it's ok :) Jan 29 15:22:07 2.6.10 worked well for me Jan 29 15:22:14 uh uh Jan 29 15:22:21 I miss 2.4 Jan 29 15:22:23 sure for me too some ages ago Jan 29 15:22:30 :) Jan 29 15:22:43 I do not miss 2.4 Jan 29 15:23:12 newfangled 2.4 stuff Jan 29 15:23:25 av500, newfangled 32 bits Jan 29 15:23:37 wouldn't 16 be enough? Jan 29 15:23:39 bradfa: no Jan 29 15:23:48 my first uC was 32 bit already Jan 29 15:24:03 av500, you had an expensive first uC :) Jan 29 15:24:10 bradfa: exactly Jan 29 15:24:17 and exactly these reactions Jan 29 15:24:34 * bradfa first computer was 32 bits too, but his parents were late to this whole computing thing Jan 29 15:25:36 jackmitchell, also see -> https://lwn.net/Articles/driver-porting/ which has some good articles although is a bit dated Jan 29 15:25:53 my first PC was 486 because we could not decide which 8086, 286 er 386 to buy.... Jan 29 15:26:07 I had a 68040 Jan 29 15:26:14 damn I am older than av500 Jan 29 15:26:36 * bradfa misses m68k Jan 29 15:27:25 bradfa: thanks bradfa Jan 29 15:27:33 jackmitchell, welcome jackmitchell Jan 29 15:27:50 * bradfa waits for big board updates from mdp Jan 29 15:28:19 * av500 feels productivity hitting 0 Jan 29 15:28:23 * av500 blames mdp Jan 29 15:28:32 * bradfa joins av500 in blaming mdp Jan 29 15:28:34 woglinde: doubt that Jan 29 15:28:36 av500 haha I have done my second userstory today Jan 29 15:28:40 * mdp updates the big board with credit for av500, bradfa, and himself Jan 29 15:28:43 woglinde: you look still a but fresh Jan 29 15:28:59 I blame 814x for everything Jan 29 15:29:33 yes Jan 29 15:29:39 mdp: just throw karma signs. devbot should pick it up ;) Jan 29 15:29:52 I wonder what happened to OSD3 and the Centaurus Jan 29 15:30:16 * bradfa is just now confused Jan 29 15:30:31 bradfa, my first 32-bit processor in day-job work was 68020...bare-metal key processor system...mmmm baremetal! Jan 29 15:30:55 SH-1 here Jan 29 15:30:57 av500, lots of stuff on their website...no action Jan 29 15:30:57 barefood Jan 29 15:31:40 mdp: neuros? Jan 29 15:32:16 the thing that was all the rage ~5 years ago? Jan 29 15:32:17 my first computer was an Apple Performa 577 with 4 MB (yes, mega!) RAM and a 320 MB scsi disk Jan 29 15:32:27 to be fair, there was that really nasty i960 in the system too, but I don't like to talk about that...still in therapy from working on that part. Jan 29 15:32:48 that computer was fast once we put 36 MB ram in it! Jan 29 15:33:03 av500, yes...they have an awful lot written on "plans" and community speculation for something that doesn't exist Jan 29 15:33:04 * bradfa remembers 9600 baud modem... Jan 29 15:33:36 bradfa: ah that newfangled stuff Jan 29 15:33:39 bradfa, I would have killed for speed like that Jan 29 15:34:00 mdp, if I throw 19200 baud? two kills? Jan 29 15:34:05 bradfa, just moving to an auto originate modem was a thrill for me Jan 29 15:34:08 mdp, well, I got it in 1994 (I think) Jan 29 15:34:27 bradfa, that was slow by then ;) Jan 29 15:34:29 then mom bought me a 14.4k modem Jan 29 15:34:36 finally I got a 56k in like 1999 Jan 29 15:34:47 mdp, we were cheap Jan 29 15:34:56 http://wiki.neurostechnology.com/index.php/File:NetraCommunityBlockDiagramv2.jpg Jan 29 15:35:00 had free internet thorugh dartmouth college though Jan 29 15:35:23 I recall having a friend with a Gateway get me in on their ultra-cheap 14.4k modem deal in the 92-93ish range while at Uni Jan 29 15:36:01 cheapest unit with a 16550 in that era Jan 29 15:36:15 mdp, mmmm 16550 Jan 29 15:36:23 with a full 1 byte scratch register Jan 29 15:36:24 * bradfa hates fastcgi Jan 29 15:36:30 that changed everything Jan 29 15:36:56 14.4K w/ SL/IP dial-in and linux Jan 29 15:37:05 * av500 wants to expose that byte to userspace with a kernel driver Jan 29 15:37:48 av500, can mmap help here? Jan 29 15:37:58 av500: ioctl READ_SCRATCH_REG Jan 29 15:38:18 byte means bitbang Jan 29 15:38:22 * woglinde runs Jan 29 15:38:28 new nautilus sucks the big one Jan 29 15:38:51 * mdp blames bitbanging for lack of productivity Jan 29 15:39:15 nautilus? Jan 29 15:39:19 like the gnome one? Jan 29 15:39:21 yeah Jan 29 15:39:26 * KotH calls for a JIHAD! Jan 29 15:39:27 av500, so yeah, that's what I'm getting at..for something with block diagrams..description of all it can do...there's little to show Jan 29 15:39:33 use xfce Jan 29 15:39:38 koth you awake nice Jan 29 15:39:40 or fluxbox if on gentoo Jan 29 15:39:41 mdp: its also from 2009 Jan 29 15:39:45 I like Gnome3, most of it Jan 29 15:39:48 everyone on gentoo uses fluxbox Jan 29 15:39:49 woglinde: been awake for quite some time Jan 29 15:39:51 bradfa, the uboot Jan 29 15:39:53 woglinde: just very busy Jan 29 15:39:53 I just don't like the new nautilus Jan 29 15:40:13 ynezz: and it's iskender kebab! Jan 29 15:40:18 mdp, are you suggesting u-boot is a good UI? Jan 29 15:40:19 mdp: when OSD1 came out it was funny, because it was the same DM320 we used Jan 29 15:40:20 av500, right..nowhere did I see a "hey, we've abandoned this!" Jan 29 15:40:30 hm normaly I do not need these file browsers Jan 29 15:40:32 mdp: and our baremetal no-os did things nicely Jan 29 15:40:34 bradfa, talking about nautilus Jan 29 15:40:38 I want to switch to ratpoison but haven't yet gotten up the nerve Jan 29 15:40:38 while their stuff hobbled along Jan 29 15:40:43 bradfa, you guys mean the file thingee?!? Jan 29 15:40:59 * bradfa is now just confused again Jan 29 15:41:01 bradfa: I used i3 for ages, it was pretty good Jan 29 15:41:08 ah, yes, or i3 Jan 29 15:41:13 i7 ftw! Jan 29 15:41:16 * bradfa hates mouse Jan 29 15:41:24 * bradfa got tackball, it runs away less Jan 29 15:41:26 bradfa, The Nautilus....a U-Boot... Jan 29 15:41:34 bradfa: tacky Jan 29 15:41:34 * mdp hands bradfa so coffee Jan 29 15:41:40 s/so/some/ Jan 29 15:41:44 * bradfa puts coffee in face Jan 29 15:41:54 ouch Jan 29 15:42:12 bradfa: you need a big ball: https://plus.google.com/u/0/112266164281670850856/posts/URrzMyAPch5 Jan 29 15:42:17 nsfw? Jan 29 15:42:21 sfw Jan 29 15:42:29 ah yes Jan 29 15:42:31 unless you put it in your pants Jan 29 15:42:33 av500, I have one of those OSD1s that joeborn gave me..never did anything with it since all the TI stuff was closed Jan 29 15:42:36 at that time Jan 29 15:42:39 yes Jan 29 15:42:41 my tackball will fit a billiards ball Jan 29 15:42:44 I think Jan 29 15:42:44 it was ingenient codecs Jan 29 15:43:00 I stole them all from their git Jan 29 15:43:02 er, svn Jan 29 15:43:04 er cvs Jan 29 15:43:06 or rcs Jan 29 15:43:14 tgz Jan 29 15:43:32 for the most part, that insanity continues on any part that gets a DSP/IVA in it Jan 29 15:43:42 yes Jan 29 15:43:57 well, one could live with binary dsp codecs Jan 29 15:44:08 but on dm320, there was closed arm code Jan 29 15:44:16 that talked to closed DSP code Jan 29 15:44:29 it's why none of ti81xx/dm385 is upstream...extremely sick culture Jan 29 15:44:46 dm385 has hw codecs Jan 29 15:44:54 one could expose API and be done Jan 29 15:44:56 mdp: yeah :/ Jan 29 15:45:06 av500, but the culture of "special-ness" continues Jan 29 15:45:09 yes Jan 29 15:45:17 my ssssspeciall Jan 29 15:45:22 lol Jan 29 15:45:35 mdp: ti81xx has what, 9 arm cores + a dsp hiding around somewhere Jan 29 15:45:53 Jacmet: could be 837 cores for all we know ;) Jan 29 15:45:59 mdp: indeed Jan 29 15:46:25 Jacmet: btw, I almost have a base u-boot for 814x to submit upstream Jan 29 15:46:33 couple little cleanups yet Jan 29 15:46:36 mdp: ohh, nice] Jan 29 15:46:54 mdp: I had a quick look at the BSP tree and cried Jan 29 15:46:55 I explicitly ignore 816x..that's an exercise for the reader Jan 29 15:47:07 mdp: anyway, it's a start Jan 29 15:47:30 yeah, I just have no interest in 816x..and as I mentioned, this is just a temporary tool for pcie devel Jan 29 15:47:31 mdp: are you using the SPL stuff? Jan 29 15:47:43 Jacmet: yes Jan 29 15:48:00 mdp: with all the internal sram it almost isn't needed Jan 29 15:48:03 I know one can do a min u-boot due to the huger sram.. Jan 29 15:48:30 right, but I decided to just KISS Jan 29 15:48:38 mdp: ok, makes sense Jan 29 15:49:36 somebody else can do the other approach if they need it Jan 29 15:50:06 it pains me enough to even have to clean this stuff up as is ;) Jan 29 15:50:28 mdp: shouldn't really be needed as spl has the falcon boot stuff nowadays (E.G. direct boot of Linux) Jan 29 15:50:52 Jacmet: learning ti81xx dpll specific programming is pretty much throwaway knowledge :( Jan 29 15:51:07 mdp: :/ Jan 29 15:51:21 in order to rewrite it..you need to understand it Jan 29 15:51:55 it's just a hunk of register banging in the evm.c rather than in a shared code area like omap/am33xx..soo.. Jan 29 15:52:07 anyway, hopefully done later today Jan 29 15:52:28 then add cpsw/pcie on top of that Jan 29 15:53:01 mdp: ok, great. I'll take a look once you have something ready and see how much effort 816x support would be Jan 29 15:53:25 how does TI boot 816x? Jan 29 15:53:28 not with uboot? Jan 29 15:53:31 Jacmet: it shouldn't be bad...Tartarus and I talked a bit about it Jan 29 15:53:37 av500, with uboot Jan 29 15:53:49 av500, there's vendor tree forks Jan 29 15:54:02 av500: with u-boot, boot rom is basically a simplified (earlier?) version of what's in am33xx Jan 29 15:54:31 Jacmet: correct..earlier...the rom for all of these is derived from omap4's Jan 29 15:54:49 that's why they all have xmodem-based uart boot now Jan 29 15:55:03 instead of the old TI-way method Jan 29 15:55:11 Jacmet: And, what mdp said, 816x support shouldn't be too bad, in so far as if you know the platform, it's "easy" to plug in Jan 29 15:55:50 I know mdp has been setting things up so that the code (and headers) are if am335x elif ti814x endif, so you just need to toss in the 816x versions of stuff Jan 29 15:55:58 Jacmet: it'll be easy to see where to add the #elif defined(CONFIG_TI816X) Jan 29 15:56:03 jinx Jan 29 15:56:04 av500: am33xx / ti814x / 816x bootrom afaik even all use the same identifier string in their bootp request - Reuse ;) Jan 29 15:56:21 Jacmet: I just don't have time to go add another platform into the mix ;) Jan 29 15:56:43 Jacmet: you even get the same rom code bugs Jan 29 15:57:12 Tartarus: I only just started working on a 816x based project, but I've done some am33xx work so hopefully it won't be too bad Jan 29 15:57:31 mdp: wee ;) Jan 29 15:57:54 Jacmet: well, just keep in mind that 816x came before 814x which came before am33xx Jan 29 15:58:15 Tartarus: yes, typical TI logic :P Jan 29 15:58:23 Jacmet: the biggest pain is just reorging the am33xx clock setup and ti81xx has different programming and domains Jan 29 15:58:30 Jacmet: so you'll have it easy Jan 29 15:59:02 Jacmet: and as I was telling Tartarus, we also have the little known 813x lurking in the shadows Jan 29 15:59:08 mdp: /me likes it like that ;) Jan 29 15:59:41 mdp: yes, I've seen. Wasn't there also 811x somewhere? Jan 29 15:59:45 Jacmet: biggest peripheral different of the base commodity stuff is that you have EMAC rather than cpsw Jan 29 16:00:15 Jacmet: dunno...I see only 813x and dm385 in the vendor tree sharing the same ti81xx clock codez Jan 29 16:00:58 mdp: there's atleast http://processors.wiki.ti.com/index.php/TI811X_PSP_User_Guide Jan 29 16:01:03 Jacmet: 811x could be x86-based given TI's naming craziness ;) Jan 29 16:01:37 mdp: yes, I've given up trying to find any logic in it Jan 29 16:02:05 I was told you could put in 20 years here and never make any sense of it all Jan 29 16:02:22 * Jacmet is not going to try .. Jan 29 16:02:53 Jacmet: likely they keep 811x in a completely different tree..it might be from a different org Jan 29 16:03:02 anyway, I need to pick up my son - See you Jan 29 16:03:08 they have separate u-boot/kernel trees for different 814x rdk evms even Jan 29 16:03:16 cya Jan 29 16:22:29 till later Jan 29 16:44:58 koen, about to send next version of tilcdc patches (with that fix I sent last week for issue that jason was having, plus DT bindings docs.. and more cowbell).. is it ok if I smash your t-b on it, and if so what parts? Jan 29 16:59:25 more cowbell! Jan 29 17:00:11 robclark: you can put my TB on all parts, I am using your latest patches Jan 29 17:00:24 koen, cool, thx Jan 29 17:00:25 robclark: with a patch from panto to add a generic power gpio Jan 29 17:00:56 koen, fwiw, that you should only need if you are using the lcd panel output driver w/ dvi cape instead of tfp410 output driver Jan 29 17:01:09 koen, I agree with rob Jan 29 17:01:32 I guess panto was doing that to hardcode overscane mode.. I think there should be a way to do that via video="DVI-D-1:..." in bootargs Jan 29 17:01:33 you can drop that (although we might have to add it to the panel drivers) Jan 29 17:01:47 "the patch still applies, I guess I still need it" Jan 29 17:01:56 robclark, it was reduced blanking mode Jan 29 17:02:14 or, yeah, what he said :-P Jan 29 17:02:26 yeah, I haven't tested lcd7 (or anything other than lcd3), so if some other lcds have need gpio then yeah, it should be added to panel output driver Jan 29 17:06:43 morning Jan 29 17:07:13 gm XorA Jan 29 17:19:01 gm Jan 29 17:44:45 re Jan 29 18:01:53 Can I get some open opinions on Sourcery, when developing for Beaglebone? Jan 29 18:02:00 (i.e. "EW!" or "Not bad") Jan 29 18:04:07 not sure what it will gain you over OE/angstrom and using eclipse as an IDE, if you prefer an IDE Jan 29 18:05:27 hello, I need help to being able to wake up the beagle bone with linux and android, is there anybody that could help me with this? Jan 29 18:06:20 I wonder which sense android makes on the bone Jan 29 18:08:41 hi prpplague Jan 29 18:09:05 we are using android 4.1 Billy Jean Jan 29 18:09:11 woglinde: hey Jan 29 18:09:38 any help with the wake up? Jan 29 18:10:01 MZXGiant: (code)sourcery is just gcc Jan 29 18:10:43 these days you're better off with vanilla 4.7 or the linaro releaes Jan 29 18:10:47 +s Jan 29 18:11:14 One more question Jan 29 18:11:17 I want to make sure I'm not nuts. Jan 29 18:11:29 Our "embedded developer", who has "20 years experience" doing this Jan 29 18:11:30 back in the day, codesourcery were contracted to do most of the arm work on gcc Jan 29 18:11:36 that contract has long since expired Jan 29 18:11:51 swears up and down that you have to copy the Sourcery sysroot for ARM to the beaglebone to run the software Jan 29 18:11:56 I... disagree. Jan 29 18:11:57 lol Jan 29 18:12:01 now it's done partially in-house at arm, partially through linaro Jan 29 18:12:26 MZXGiant, it depends Jan 29 18:12:31 I told him "why would you copy an ARM sysroot to an ARM platform?" Jan 29 18:12:42 as long as you have a compatible environment on the target, you don't need to copy anything Jan 29 18:13:03 usually that means kernel+libc no older than what the software was built against Jan 29 18:13:16 We're selecting Cortex-A8 and an glibc-compatible system Jan 29 18:13:19 depends what is that sysroot contains Jan 29 18:13:38 I would normally recommend pointing the compiler at a copy of the target sysroot Jan 29 18:13:45 Ditto. Jan 29 18:13:52 MZXGiant: is there a question in there? Jan 29 18:14:10 thurbad; Basically, "Is copying the sysroot to the target normal practice?" Jan 29 18:14:19 Because my thought is "No" Jan 29 18:14:19 MZXGiant lol fire him Jan 29 18:14:44 wakeup help??? Please Jan 29 18:14:46 but when I told him that, he went off and told me that I am just "dicking up" to make him look foolish Jan 29 18:14:47 I build the sysroot on a pc, then copy it to the target Jan 29 18:14:58 or for dev work, nfsroot it Jan 29 18:15:14 nfsroot is no copy Jan 29 18:15:18 so fire Jan 29 18:15:25 hi trolls, friends, and mutually inclusive people Jan 29 18:15:35 jo mranostay Jan 29 18:15:38 either way, I'm not using the minimal "sysroot" that's usually part of the compiler bundle Jan 29 18:15:44 Yeah, that's what he wants Jan 29 18:15:54 hi unreasonably rebellious dude Jan 29 18:15:56 to move that minimal sysroot to /arm-sysroot on the bone Jan 29 18:15:59 MZXGiant what did you embedded guy the 20 years Jan 29 18:16:05 qnx? Jan 29 18:16:10 and point the interpreter to /arm-sysroot/ld-linux.so.3 Jan 29 18:16:15 and let it run wild Jan 29 18:16:15 good evening panto Jan 29 18:16:22 * panto bows Jan 29 18:16:26 MZXGiant: what, and have _two_ copies of libc on the target? Jan 29 18:16:28 woglinde; I'm assuming something not embedded :P Jan 29 18:16:36 mru; Correct! Jan 29 18:16:45 and where does the first on come from? Jan 29 18:17:08 The base OS I already have running on the bone Jan 29 18:17:10 Javier_ is it commercial ask your ti sales contact Jan 29 18:17:17 I have a stripped down copy of angstrom running X and twm Jan 29 18:17:32 what is a stripped down copy? Jan 29 18:17:58 gcc --stripped-down-copy Jan 29 18:18:17 panto you are genius Jan 29 18:18:21 woglinde: objcopy --strip ? Jan 29 18:18:27 see, it's simple Jan 29 18:18:32 mru we already do it Jan 29 18:18:39 and have debug symbols in -dbg package Jan 29 18:18:44 iKnow Jan 29 18:18:49 he proably means opkg remove Jan 29 18:18:58 for packages he do not need Jan 29 18:19:01 strip can mean many things Jan 29 18:19:10 right Jan 29 18:19:14 some of it with one dollar bills Jan 29 18:19:23 thats why I am asking Jan 29 18:19:23 heh Jan 29 18:20:04 hello woglinde, this question is a technical question about the beagleboard not commercial question Jan 29 18:20:06 panto dont talk about money Jan 29 18:20:27 Javier_ your question is not precise Jan 29 18:21:04 and do not expect everybody here in the mood to help you Jan 29 18:21:16 ok, my question is. Since there are not any button available or connection with the P8 or P9 connector possible to sleep the board and wake it up, we can sleep it by software, but how could we wake it up? Jan 29 18:22:00 examine the irc logs we had this discussion some days ago Jan 29 18:22:31 solutions was something with cutting lines are external circuits with battery logic Jan 29 18:22:37 ups s/are/or Jan 29 18:23:54 ol.. to have all those dollars back spent back in the late 70's in the girly bars ... Jan 29 18:24:49 noice i didnt use the t word Jan 29 18:24:57 where can I find that discussion... looking with what text? Jan 29 18:25:08 cutting line Jan 29 18:25:15 3 or 4 days ago Jan 29 18:25:31 djlewis wow Jan 29 18:52:26 ... Jan 29 18:53:04 djlewis: .... Jan 29 18:53:16 oops.. Jan 29 19:08:54 ...___... Jan 29 19:09:46 * KotH shoots mranostay Jan 29 19:10:21 * KotH then takes his soul and puts it into a bottle Jan 29 19:10:29 <_av500_> again... Jan 29 19:10:47 hey! he asked for his soul to be safed... i just did that Jan 29 19:11:01 when i did i ever ask for that? Jan 29 19:11:20 * KotH points up Jan 29 19:12:00 mranostay: you know what SOS means, do you? Jan 29 19:12:51 and here I thought he was just doing an emoticon Jan 29 19:12:53 save our ship Jan 29 19:13:17 save ohio strays Jan 29 19:13:39 <_av500_> send oreos safely Jan 29 19:13:59 [doublestuff] Jan 29 19:14:07 quadstuff Jan 29 19:15:11 where stuff==beer Jan 29 19:15:24 .equ beer Jan 29 19:18:24 5 ale pumps here... Jan 29 19:21:05 * KotH was just reading something about the "ale high frequency network" Jan 29 19:22:15 KotH: some kind of rapid pub crawl? Jan 29 19:22:39 mru: "crash kurs in wirtschaftskunde" :) Jan 29 19:23:02 in american please Jan 29 19:23:20 crash course in economics Jan 29 19:23:40 module the joke on words Jan 29 19:23:42 modulo* Jan 29 19:24:01 KotH: your company related to banking? Jan 29 19:24:14 mranostay: we use banks, yes Jan 29 19:25:17 <_av500_> banks use us Jan 29 19:25:26 abuse Jan 29 19:26:16 not the best place to launder money anymore Jan 29 19:26:32 Belize is so much better... and warmer Jan 29 19:28:10 KotH: what a coincidence, my company uses banks too Jan 29 19:28:21 small world Jan 29 19:28:45 mdp: you don't get paid in sheep? Jan 29 19:28:56 re Jan 29 19:29:00 kids in bed Jan 29 19:29:07 we trade in fatted calves in the heartland Jan 29 19:38:30 Mmmmm fatted calves Jan 29 19:38:33 roasted over a fire Jan 29 19:41:59 and the fire burns burns burns Jan 29 19:42:09 in a ring of fire Jan 29 19:44:03 heh Jan 29 19:45:35 ;) Jan 29 19:54:34 what would be the best place to buy us power socket to european schuko plug adapters? Jan 29 19:54:51 friend of mine is sitting in cambridge and has a slight power problem Jan 29 19:57:21 lol Jan 29 19:57:29 frist trip? Jan 29 19:58:00 woglinde: postdoc Jan 29 19:58:17 woglinde: mathematician... you can imagine how well prepared he is Jan 29 20:00:49 I would ask on #macgyver Jan 29 20:00:58 any electrics retailer should have such things Jan 29 20:01:29 but if he's in cambridge, what does he need schuko plugs for? Jan 29 20:01:53 mru: to power his old electronics Jan 29 20:02:57 since when mathematician has electronics? Jan 29 20:03:04 they have paper and pen Jan 29 20:04:22 and a waste basket Jan 29 20:04:36 philosophy is simpler, it doesn't need the waste basket Jan 29 20:04:45 lol Jan 29 20:06:25 philosophy do not need paper and pen too Jan 29 20:06:50 they have to write the ramblings down somehow Jan 29 20:06:56 some said philosophy? I'm an expert at failing at it Jan 29 20:07:01 *someone Jan 29 20:07:17 panto what you came from the land of philosophy Jan 29 20:07:35 another mark to be ashamed Jan 29 20:07:37 *g* Jan 29 20:07:44 ah, but have you ever met one philosopher on the flesh? :) Jan 29 20:07:46 money philosophy Jan 29 20:07:55 you have to be ashamed :P Jan 29 20:07:56 what is the third mark Jan 29 20:08:02 panto: unemployed people all the time :) Jan 29 20:08:06 garlic? Jan 29 20:08:22 mranostay, yes, and they seem to pick up the best looking chicks Jan 29 20:08:24 damn them Jan 29 20:08:49 haha Jan 29 20:12:42 <_av500_> since when does the UK have schuko? Jan 29 20:12:54 <_av500_> thtey have these battle ships Jan 29 20:16:51 panto: so no money or a lot money gets the girls? :) Jan 29 20:18:46 pretty much Jan 29 20:19:01 extra points if you're an alcoholic drug-abusing biker Jan 29 20:30:20 is the google group not sending emails? Jan 29 20:31:38 bradfa: you turn it off? Jan 29 20:32:01 mranostay, no Jan 29 20:32:12 but someone may have... Jan 29 20:32:24 * mranostay hands bradfa a tin foil hat Jan 29 20:32:27 * bradfa notices people have been active on the beagle group in the past 2 days... Jan 29 20:32:38 * bradfa hasn't gotten any emails except 1 yesterday Jan 29 20:33:25 mranostay, tin foil hat fits fine, google groups settings says I get 1 email per each message Jan 29 20:33:30 however, I do not Jan 29 20:33:34 hmmm Jan 29 20:35:04 yeah, I definitely am missing a good 2 days worth of messages Jan 29 20:35:53 with the usual quality of messages, that's not saying much Jan 29 20:36:15 mru: that is why i don't think i'm even subscribed Jan 29 20:36:27 had to turn off the digest for the LinkedIn group as well Jan 29 20:37:04 jkridner__, are there any issues with the google groups? Jan 29 21:04:17 Anyone has had success insmodding this on a beaglebone? http://processors.wiki.ti.com/index.php/EDMA_sample_test_application Jan 29 21:04:30 It hangs here Jan 29 21:13:48 this is frustrating... I have no idea why I've stopped getting google group emails. Jan 29 21:15:44 jsabeaudry: change numTCs to 3 Jan 29 21:16:05 jsabeaudry: that sample is not as generic as it first appears Jan 29 21:19:17 Is there a 'official' BBone HCL anywhere (or more likely Ångström HCL on BBone: interested in 802.11n/ac USB 'nubs' and WebCams)? Jan 29 21:19:31 Ok, maybe Bluetoothaches too Jan 29 21:21:08 mdp, Thanks, it works, too bad it does not include an AB-Sync transfer, I'll give it a spin see if I can crash the bone like with my driver Jan 29 21:21:50 jsabeaudry: crashing is but a milestone in the exact steps on your way to "working" Jan 29 21:22:23 heh Jan 29 21:22:28 you can see an example of an AB-sync transfer in several kernel drivers, if that helps Jan 29 21:24:40 mdp, hmmm, I'll grep SYNCDIM see what I can find, thanks Jan 29 21:24:50 old kernel Jan 29 21:25:43 on 3.2 at the moment but I was getting the same issue on 3.8 only more often Jan 29 21:25:48 jsabeaudry: actually, you can look at mainline in drivers/dma/edma.c Jan 29 21:25:58 jsabeaudry: or any 3.6+ beagleboard/kernel Jan 29 21:28:37 isn't it all supposed to use dmaengine nowadays? Jan 29 21:28:48 jsabeaudry: edma_prep_slave_sg() .. if you ignore all the dmaengine vchan stuff then it's a concise example Jan 29 21:28:55 and don't care about {e,s,o}dma? Jan 29 21:29:14 koen, the edma dmaengine driver is a wrapper around the private edma api.. Jan 29 21:29:25 I know Jan 29 21:29:28 as a wrapper driver, it's not yet ready to support MEMCPY transfers Jan 29 21:29:33 ah Jan 29 21:29:40 so those can still be done using direct calls for the time being Jan 29 21:29:55 we must have stern words with the TIer who wrote that wrapper! Jan 29 21:29:57 unlike everything else in the kernel, I've been trying not to break working use cases ;) Jan 29 21:30:22 that's how the audio stuff works without being converted to dmaengine as well Jan 29 21:30:54 I have some memcpy support already implemented, btw...just waiting to follow all this in Jan 29 21:31:20 and it's the least useful feature anyway ;) Jan 29 21:33:28 mdp, oh wow, I need to create a param slot for each sg list element and link them? Jan 29 21:33:49 if you are doing SG lists, yes Jan 29 21:34:23 after much much much much much much (am I clear?) studying, that's the only known way to make use of EDMA to handle SG lists Jan 29 21:34:36 it's a feature of the EDMA design Jan 29 21:34:40 mdp, OMG I've been doing things very wrong! Jan 29 21:35:17 jsabeaudry: one of the points of dmaengine is that you shouldn't have to waste your time having to know EDMA-isms for the common cases Jan 29 21:35:39 dmaengine is only 3.6+ ? Jan 29 21:36:02 jsabeaudry: if you want them to be handled efficiently you do them as a linked param set like this Jan 29 21:37:23 jsabeaudry: that's the first beagleboard kernel that applied the patches series with edma dmaengine support Jan 29 21:38:19 Is copying bytes from a gpmc-bound address to a kfifo a common case? Jan 29 21:38:24 jsabeaudry: the driver itself showed up in the 3.6 release, but only supporting davinci platforms Jan 29 21:42:20 jsabeaudry, yes, but your target of the transfer is some buffer that is passed by reference to the kfifo api Jan 29 21:42:37 so kfifo is irrelevant to the dma operation Jan 29 21:43:24 on completion of the dma transfer I suppose in your driver architecture you may do a kfifo_put to pass the received data along Jan 29 21:44:01 so what you want to do is look at a SLAVE_SG dmaengine example, what you describe is a common case, yes Jan 29 21:44:37 your doing a a transfer with a src of the gpmc to a dst buffer is all..same thing done in the mmc and spi drivers Jan 29 21:45:14 from a dma pov only the magic src address (and whatever address width and depth of burst you allow) change Jan 29 21:46:32 Ok, I'll guess I'll move back to 3.8 and go through the dmaengine api Jan 29 21:46:50 mdp, thank you so much, I've been searching for this bug for a long time Jan 29 21:46:56 jsabeaudry, drivers/mmc/host/davinci_mmc.c:mmc_davinci_start_dma_transfer() has an example of usage Jan 29 21:46:58 np Jan 29 21:48:19 Documentation/DMA-API.txt for streaming mappings and Documentation/dmaengine.txt for the slave dmaengine api are the two relevant kernel docs Jan 29 21:48:47 that above example, coupled with what I do in drivers/dma/edma.c on the backend is reasonably straightforward Jan 29 21:49:17 if you just look at the map/unmap_sg() and dma_*() usage and ignore mmc-isms ;) Jan 29 23:16:33 anyone doing a talk? Want to have bad dreams? https://www.youtube.com/watch?v=ZTdUmlGxVo0&t=12m Jan 29 23:36:31 Russ, if that guy was a reference that's pretty dang close to my real fears of talking Jan 29 23:37:19 it goes on for the whole talk Jan 29 23:37:35 (no argument if the guy is right or wrong, just the situation) Jan 29 23:42:31 Russ: probably don't want to watch that then Jan 29 23:43:30 that's a pretty long talk for that to keep going on like that Jan 29 23:44:19 the internet never forgets Jan 29 23:44:38 or forgives Jan 29 23:50:32 if lennart turned up in my audience I'd ask him to leave Jan 29 23:52:20 heh Jan 29 23:52:45 anyone like that kid? Jan 29 23:53:47 apparently he has some anti-fans Jan 29 23:57:27 he tends to stay away from my kernel, so I'm ok Jan 29 23:58:39 systemd.ko :-D Jan 30 00:02:08 XorA: had to google to be sure you were kidding Jan 30 00:04:01 I do wonder if people would have a more positive view of lennart and co projects if he had a different attitude, or if it weren't for the attitude, his software would be ignored completely Jan 30 00:04:49 mranostay: there was a dbus.ko in the past Jan 30 00:28:48 XorA: did you see we have found the install task? Jan 30 00:51:53 toneeee: nope Jan 30 00:55:20 XorA: http://hipstercircuits.com/?p=245 Jan 30 00:55:57 XorA: I found the openembedded-core manual and see that though there is a deploy task listed in the recipe, it looks like the install task is actually working Jan 30 00:56:16 toneeee: yes, that is normal Jan 30 00:57:12 XorA: fantastisch Jan 30 00:57:15 toneeee: bitbake -C compile should work, but it seems Angstrom bitbake version is too old Jan 30 00:57:23 ohhh Jan 30 00:59:27 might have to wait until koen bumps bitbake version Jan 30 01:00:05 I will be at FOSDEM soon anyway and no coding shall be done :-D Jan 30 01:24:59 sorry if this is asked too often, but are there any plans for a C compiler for the PRU? to not have to hand-code assembly? Jan 30 01:26:49 why on earth (or anywhere else) would anyone want a c compiler for pru? Jan 30 01:29:57 me me !! I want one :D Jan 30 01:30:40 but: I don't mind skipping it because the PRU might be obsoleted by the Cortex-Ms on the newer Sitara SoCs... Jan 30 01:31:00 they serve somewhat different purposes Jan 30 01:31:29 the pru has the advantage that you can easily count exactly how many cycles a piece of code will use Jan 30 01:31:52 but of course only if you write assembly by hand Jan 30 01:31:52 also the instuction space you have is really small Jan 30 01:32:07 well it's no smaller than Arduino Jan 30 01:32:44 it's bigger in many regards because you can swap memory into it from the rest of the SoC Jan 30 01:32:50 if you involve a compiler, you lose that determinism Jan 30 01:33:11 which was probably the reason for using it in the first place Jan 30 01:33:27 if you don't need that accurate timing, a cortex-m is much easier to use Jan 30 01:33:47 okay. nice advice. Jan 30 01:34:03 also you aren't running complex stuff on a PRU Jan 30 01:34:10 that too Jan 30 01:34:40 i'm loving that PRU Jan 30 01:34:41 :-) Jan 30 01:34:47 what a godsend of a device Jan 30 01:34:49 and it's a rather unusual architecture that it would be hard to target with gcc Jan 30 01:35:00 gcc _hates_ special-purpose registers Jan 30 01:36:11 ... " he pru has the advantage that you can easily count exactly how many cycles a piece of code will use " Jan 30 01:36:12 not to mention you'd still have to constantly look over your shoulder to make sure it didn't mess up with memory layout and such Jan 30 01:36:25 exactly what makes it appealing Jan 30 01:36:46 unless you touch the L3 interconnect of course Jan 30 01:39:19 So the first time I saw C, I was an assembly type guy. My friend showed me how C saved my time by dumping the assembly output of the version 6 unix compiler, primitive but useful: cc -s helloworld.c and there was glory in that simple compiler. I'd like something to avoid the assembly paradigm but maybe with predictable outputs for each statement so you are still in control of the timing like mrpackethead suggests Jan 30 01:40:26 heh Jan 30 01:40:27 so you want an assembler with macros Jan 30 01:40:31 they all have 'em Jan 30 01:40:32 sometimes Jan 30 01:40:39 of course I know they have them Jan 30 01:40:44 thanks guys. so which are the newer chips with the cortex-m's? they're on the same IC? Jan 30 01:41:02 and I don't want an assembler with macros, I want a higher-level language. I can get by with a macro assembler Jan 30 01:41:23 a more advanced compiler is by definition unpredictable Jan 30 01:41:39 well cc -s will show the output and it can't be that huge. Jan 30 01:42:00 hmmm missing the point Jan 30 01:42:10 you save time and effort by putting the compiler in charge of some decisions Jan 30 01:42:21 by doing that you no longer know the precise outcome Jan 30 01:42:27 yes. and then I tweak the output. cc -s stupid.c > stupid.asm Jan 30 01:42:36 you can of course check what it did in any specific case Jan 30 01:42:50 yes, if I care that much that I'd spin on asm, sure I do. Jan 30 01:42:50 but a different version of the compiler will produce different output Jan 30 01:42:57 otherwise it wouldn't be a different version Jan 30 01:43:16 don't be surprised if you see a non-gcc pru C compiler in the future Jan 30 01:43:35 I'll be surprised if it works Jan 30 01:43:50 oh you are suggesting that I'd end up rebuilding my stupid.c code in another situation and my compiler's output would need to be retweaked. Accepted/good point Jan 30 01:43:55 but I'm never surprised, so it won't Jan 30 01:43:57 mru, we did have that discussion about other TI C compilers didn't we? ;) Jan 30 01:44:13 exactly Jan 30 01:44:28 hehe Jan 30 01:44:52 I wasn't thinking that far into the future. This is throwaway tech! :D Jan 30 01:45:11 mru, I was thinking about that while um, looking at something related recently ;) Jan 30 01:46:03 toneeee: you are holding your breath waiting for a general purpose cortex-M in a Sitara processor? Jan 30 01:46:19 the omaps have them... Jan 30 01:46:25 yep Jan 30 01:46:29 yes Jan 30 01:46:40 even Sitara now has cortex-M..just not general purpose ;) Jan 30 01:46:56 the power sequencer? Jan 30 01:46:58 dra74x has two M4s Jan 30 01:47:23 mru, yeah, you don't get the m3 to yourself because of that Jan 30 01:47:28 but I am not on the TI roadmap mailing list. I'm a beagle hobbiest who has been looking at the various TI SoC designs to see the different peripherals they are packing into them Jan 30 01:47:50 my crystal ball says more PRU Jan 30 01:47:55 yay!! Jan 30 01:47:59 thanks crystal ball! Jan 30 01:48:24 (I like Harvard architectures, they are fun. Weitek had some 'back in the day' ) Jan 30 01:48:28 the crystal ball has more details at conferences with drinks Jan 30 01:49:49 where there are no logs? :P Jan 30 01:50:16 lololol Jan 30 01:50:19 i get it, the PRU is most useful for very precise machine/motor timing. but still, right now i'm probably going to be adding a second cheap m0 or something core to do some not-too-complex-but-too-complex-for-PRU stuff Jan 30 01:50:19 everything is logged Jan 30 01:50:53 nemik_: I used it for precise bus cycles, myself Jan 30 01:51:10 now I'm using it as a highspeed spi slave Jan 30 01:51:14 The PRU-feeds-the-6502-bus is pretty keen. maybe we could PRU-feeds-arduino. lol Jan 30 01:51:17 nemik_: what is too complex? Jan 30 01:51:50 mranostay, bitbanged spi slave...bang bang bang bang Jan 30 01:52:34 mranostay: routine that does some control of channels (some h-bridges) and reads wiegand data Jan 30 01:53:18 plus some processing inside there to SHA256 some of that wiegand data but that could be done on beaglebone then by just writing it to some section of the shared memory and issuing some interrupt then Jan 30 01:53:37 mranostay, it really is the best bitbanger around at 200mhz Jan 30 01:54:05 still...assembly :/ but my hesitation is probably because i'd never really done it before Jan 30 01:54:52 nemik: it's really "easy" on the small scale … but tedious on the larger scale. Jan 30 01:54:57 nemik_: I don't even have a neckbeard and even I can do assembly ;) not sure why it's a big deal Jan 30 01:55:11 it's not like you are writing a spreadsheet app in assembly here Jan 30 01:56:17 it's a tiny investment for having determinism..ask mranostay how it quickly fixed his nixie tube problem Jan 30 01:56:31 yea you're right. i should spend a couple days with it and try it. the real nice thing there is the bin file is then upgradeable as part of the OS. no need to deal with the bootloaders on separate chips and their firmware Jan 30 01:56:56 I can't grow a neckbeard, freakin' follicles! who needs 'em! Jan 30 01:57:06 nah, this isn't really about determinism for me but i get what you mean Jan 30 01:57:25 yeah it is a solid conversation Jan 30 01:59:05 speaking of which, it appears the PRU can access the GPIO banks 0 and 1? and can then make any of those pins into interrupts? that would be real nice... Jan 30 02:02:15 nemik_: well if you mean poll their state then yes :) Jan 30 02:03:00 and it isn't really GPIO there is mux setting to have the PRU access it Jan 30 02:03:08 yeah that mux is nifty Jan 30 02:03:25 that's how I found those DMTimers Jan 30 02:03:43 and ended up here…. Jan 30 02:06:26 mranostay: hmmm the pulses i'm looking for are ~50 uS and i can't miss one. Jan 30 02:08:22 at 200 mhz i suppose that could work depending on how many i need and what else it might be doing Jan 30 02:08:35 nemik_: two cores :) Jan 30 02:09:16 are you just counting pulses? Jan 30 02:09:46 toneeee: yea, just that. and good point mranostay Jan 30 02:20:03 i got four outputs of DMX up and running on teh PRU Jan 30 02:20:19 its wonderful for that kind of thing. Jan 30 02:20:46 DMX has got some whacky breaks and pauses in it, that makes using a Standard UART possible, but not allways easy Jan 30 02:20:58 being able to bit bang it out, is great Jan 30 02:20:59 you are Mr DMX, mph Jan 30 02:21:25 nice! Jan 30 02:21:39 it's cool; I run a recording studio for funzies but I do not do much live work and thus only know peripherally of DMX (pun intended) Jan 30 02:21:40 The PRU also can do some pretty complex task Jan 30 02:22:07 like run ethercat or profinet Jan 30 02:22:18 where you need very high determinitic behavior Jan 30 02:22:26 for industrial control. Jan 30 02:26:42 mrpackethead_: thanks for reminding me to look up Ethercat. I'd never heard of it (not in controls) until I looked through the SPRUH73G document and was curious about it. Jan 30 02:27:03 TI have a board called the ICE Jan 30 02:27:17 AM3559 Jan 30 02:27:22 k Jan 30 02:27:26 plus the relevent PHY **** ENDING LOGGING AT Wed Jan 30 02:59:59 2013