**** BEGIN LOGGING AT Mon Apr 23 02:59:58 2012 Apr 23 05:12:48 <_sundar_> gm Apr 23 05:26:16 <_av500_> gm Apr 23 06:01:35 <_tasslehoff_> good monday Apr 23 06:18:05 no such thing Apr 23 06:23:40 'morning Apr 23 06:24:22 hey, I have a full 36 minutes before I have to face the reality of monday Apr 23 08:14:17 Hi, I'm looking for a manufacture to do a order of Beablebone PCB, anybody know manufactures with experience with this kind of PCB. Apr 23 08:15:04 circuitco :-) Apr 23 08:16:38 and svtronics Apr 23 08:24:11 We want include this product in a proyect but we need delete some componentes of the original PCB, anybody have use beaglebone in a industrial proyect, is it robust plataform? Apr 23 08:25:27 oteros2000: I dont see anything "special" about the beaglebone in terms of PCB Apr 23 08:25:35 also, define "robust" Apr 23 08:26:27 * LetoThe2nd welcomes everybody to join us in proceeding to buzzword land :) Apr 23 08:29:10 is it stable in a environment industrial, with a range of temperature -20 +85ºC ? Apr 23 08:29:58 that must be a pretty cold factory... Apr 23 08:31:21 This is the requeriment in a industrial environment...about temperature Apr 23 08:31:38 so, talk to circuitco Apr 23 08:31:55 humans can't survive at +85, if you're building something to survive inside a volcano you probably should pay an engineer to design a board :p Apr 23 08:32:22 humans don't wear omaps Apr 23 08:32:51 oteros2000: if you're doing your own pcb, probably noones gonna do your ratings for you. check the specs of the compenents, adjust the components if necessary and then do your testing. Apr 23 08:33:24 But human don't live inside of little box, next to power supple IC Apr 23 08:34:22 ok, I'll contact with circuitco and svtronics Apr 23 08:34:42 sounds like an inefficient or badly cooled psu Apr 23 08:35:28 it just sounds like buzzwording. "i need industrial spec, because someone told me. now please tell me it is industrial." Apr 23 08:36:35 i wear an omap every day, my phone uses one ;) Apr 23 08:42:45 85° isn't very hot. according to your specs no car would start if you would leave it in the sun for a while Apr 23 08:44:02 i wouldnt park where it would get that hot, it roasts heaps of parts that are a royal pain to replace and generally shortens its life Apr 23 08:45:27 what do you think what temperature will be inside a car if it's outside 40°C and your car is parked in the sun? Apr 23 08:46:28 http://www.howcast.com/videos/183164-How-To-Bake-Cookies-on-Your-Cars-Dashboard Apr 23 08:46:30 it'd probably reach 60, for sure more than my parked-in-the-shade car! Apr 23 08:46:47 av500, ya, in phoenix, its got to get hotter than an easybake Apr 23 08:47:37 av500: Gerald had an amusing anecdote about people designing for "industrial" Apr 23 08:48:07 pray tell Apr 23 08:48:24 av500: the board would go into the hot part of a cooling system, which would reach 70-80 degrees C Apr 23 08:49:03 av500: the proposed solution to keep be board below that was "make the PCB bigger so it can dissipate the heat better" Apr 23 08:49:16 so a car needs much more than 'industrial' spec if it lives in sunny hot places Apr 23 08:49:19 spot the flaw in that reasoning :) Apr 23 08:49:59 its flawless :) Apr 23 08:50:12 koen, I've look through my datasheets, they all tell me the same thing Apr 23 08:50:18 koen, I don't need you "logic" Apr 23 08:50:32 morning Apr 23 08:50:41 hey XorA Apr 23 08:54:34 * koen stabs openssl Apr 23 08:54:37 don't do git gc --aggressive on linux on a beagleboard ;) Apr 23 08:55:10 it jumped out and latched onto your beard? Apr 23 08:55:21 :) Apr 23 08:56:00 mine tries that since about 12h: Counting objects: 2610217 Apr 23 08:58:15 s/beagleboard/beagleboard classic/ Apr 23 08:58:28 the poor one just has 256mb Apr 23 08:59:18 poor thing Apr 23 09:00:09 at least it isn't unemployed ;) Apr 23 09:00:16 Russ: isn't it like 2 am in your TZ? Apr 23 09:00:53 koen: it's a hot zone Apr 23 09:01:59 ya, I have a call in 6 hr Apr 23 09:02:14 my wife is up studying for a practical Apr 23 09:04:29 I'm still fighting with ktest.pl, I have it behaving to a point Apr 23 09:12:36 does it matter if 3.3 doesn't work without ARM_PATCH_PHYS_VIRT Apr 23 09:13:11 yes, I was hit by that too ;) Apr 23 09:13:24 that's what I'm talking about Apr 23 09:13:34 I know the commit and config option, I just haven't figured out why Apr 23 09:13:52 its mux related, so it probably boots, you just don't get serial output Apr 23 09:13:56 not sure why I had enabled that, I think enabling ARM_PATCH_PHYS_VIRT didn't work before Apr 23 09:14:15 and now it seems to need it. Apr 23 09:15:33 on kirkwood I had to disable it, because enabling it needs a l2-cache disabled by u-boot. Apr 23 09:15:54 it's a bit wired. Apr 23 09:16:45 don't even know what it does. ;) Apr 23 09:17:25 historically, for ARM kernels, your basic PHYS to VIRT mapping and offset was fixed at compile time Apr 23 09:17:35 that limits the number of boards you can run a single kernel on Apr 23 09:17:48 ARM_PATCH_PHYS_VIRT patches in the offsets at boot time Apr 23 09:18:06 ah. understood. thanks. Apr 23 09:19:11 * Russ checks out amazon supply Apr 23 09:25:35 if the option is needed, maybe just enforce it for omaps. Apr 23 09:26:24 DT ftw! Apr 23 09:27:27 I still think DT is a good thing ;) Apr 23 09:30:06 one thing more, vendors can sign ;) Apr 23 09:30:10 ARM DT is like world peace; a good idea, but a few years off :) Apr 23 09:30:38 hmm, is the DT data covered by the GPL? Apr 23 09:31:06 don't think so Apr 23 09:31:26 so vendor kernels will be even more useless than today Apr 23 09:34:32 if you can't start your own kernel, nothing will help. Apr 23 09:34:54 if you can, hidding a DT doesn't make sense Apr 23 09:36:15 I assume something like /proc/dt.gz will appear Apr 23 09:36:45 aholler: that assumes you have a std way to get the device tree Apr 23 09:37:00 I will invent an evil nonstd one :) Apr 23 09:37:29 you will need m$-certificates anyway ;) Apr 23 09:37:35 yes Apr 23 09:38:05 WART will be fun Apr 23 09:40:46 it ensures vendors can't support old hw because the certificate is expired. Apr 23 09:42:21 * Russ thinks dt's would not be difficult to decompile Apr 23 09:45:03 at work we are making good use of java decompilers to find out why Android APKs dont work Apr 23 09:45:23 especially the "google apps" :) Apr 23 09:46:48 av500: I used to have to do that for javacard :-D out customers would even refuse to send us the binary form of the app, but they would send the IDE logs. I wrote a perl script to regenerate the package from the logs Apr 23 09:54:18 meh, now I have to wait until June before I can by a new laptop :( Apr 23 09:54:29 :-( Apr 23 11:17:16 hmm, they should switch in-house power from ~220V to =12V or =5V. Apr 23 11:18:06 w Apr 23 11:18:45 aholler, easy, just live in your car Apr 23 11:18:55 how do you know he doesn't! Apr 23 11:18:56 ;) Apr 23 11:19:39 or a 70s pop-up camper with some quality 12V lighting Apr 23 11:19:43 CanyonMan: :) Apr 23 11:20:17 hm, a trailer park for hardware nerds? Apr 23 11:21:01 * aholler is annoyed by cars Apr 23 11:21:05 I had to spend the last half year to much time inside my car. :/ Apr 23 11:21:52 well we got a nice late april snow storm. There is a giant tree laying across my driveway. So I have to figure out how I am going to get to work. Apr 23 11:40:02 CanyonMan: blast it! Apr 23 11:40:19 LetoThe2nd: nerd trailer tash, indeed Apr 23 11:40:25 trash, even Apr 23 11:40:51 *sing* we're the white trash cicus, don't give a damn.... Apr 23 11:59:59 hmm, my beagle still stops with 3.3.2 after "..., booting the kernel.". enabling patch_phys* and ignoring _HWMOD_SKIP_ENABLE didn't help. Apr 23 12:00:35 aholler: 3.3.2 is obsolete Apr 23 12:00:50 I know, but 3.3.3 isn't announced ;) Apr 23 12:01:17 I have preordered it.... Apr 23 12:01:27 Linux eiche 3.3.3-dockstar-00019-g1a6ce14 #43 PREEMPT Mon Apr 23 12:04:39 CEST 2012 armv5tel Feroceon 88FR131 rev 1 (v5l) Seagate FreeAgent DockStar GNU/Linux Apr 23 12:01:38 ;) Apr 23 12:01:50 what, no 3.3.4RC? Apr 23 12:01:55 I am dissapoint! Apr 23 12:02:05 wow, hail storm Apr 23 12:02:25 I'm just a half-backed kernel friggler and avoid rc-stuff ;) Apr 23 12:02:29 free crushed ice and no mojito :( Apr 23 12:03:00 the sun will come afterwards, had that storm 30minutes before ;) Apr 23 12:03:11 you mean the oracle Apr 23 12:07:24 allhail the oracle. Apr 23 12:08:20 aholler: 3.3.3 was announced tonight and is on kernel.org Apr 23 12:08:41 yeah! halfway to hell! Apr 23 12:09:08 jannau: lkml (and lwn) doesn't know about 3.3.3 Apr 23 12:09:48 aholler: lkml is for old people Apr 23 12:10:04 aholler: announcements seems to be moved to g+ Apr 23 12:10:08 there is ph0rum? Apr 23 12:10:21 aholler: sorcial notwork Apr 23 12:11:06 gregkh was probably too shocked by the raspi vendor kernel and forgot the announcement mails Apr 23 12:11:17 lol Apr 23 12:11:29 looks like I've got unsocial by not being a member of social-networks Apr 23 12:11:48 aholler: yes Apr 23 12:12:55 aholler: the nice thing about g+ is that all the people from school that you dont care about are not there - coz they never heard of it... Apr 23 12:13:45 how do you know I was at school? ;) Apr 23 12:14:00 s/school/work camp/ Apr 23 12:14:09 s/work/boot/ Apr 23 12:14:15 yeah Apr 23 12:15:14 aholler: I guess I assume everybody got a free high school diploma like the kids in HE :) Apr 23 12:22:49 the third try to boot that hacked 3.3.2 kernel revealed a bit more: http://fpaste.org/0fdp/ But befor digging further, I will switch to 3.3.3 ;) Apr 23 12:27:40 jannau: I assume greg fall asleep as it was 00:39 ;) Apr 23 12:30:27 my bets are still on 'stunned by the raspi kernel': https://plus.google.com/111049168280159033135/posts/1jusHFDG1v8 Apr 23 12:32:20 didn't know nxp is associated with the raspi? Apr 23 12:39:21 err, right. marvell and not broadcom. I just assumed that gregkh was still looking at the raspberry pi kernel Apr 23 12:43:51 hmm, gregkh uses lego too ;) Apr 23 13:13:12 I've installed the latest Ångström distro on my BeagleBone. I had the impression that it included some source/demos on how to use the expansion ports. Could someone point me in the right direction? Apr 23 13:25:23 thomas_fogh: i think it has some examples for the cloud9 ide, yes. Apr 23 13:25:43 nice, our cable got upgraded to 128/10 Apr 23 13:25:54 all we called for was a fixed IP Apr 23 13:26:05 thomas_fogh: should be accessible thorugh the webinterface if you have the cloud9 image. the small systemd image does not have exmples, i think. Apr 23 13:31:28 LetoThe2nd: some files are missing compared to https://github.com/jadonk/bonescript but I'll look into it. Thanks! Apr 23 13:41:17 * mdp wonders if Boris read the entire precise PWM article he +1ed Apr 23 14:01:11 mdp: I would +1 it, but that's because I played with their RC car at ELC-E cambridge :) Apr 23 14:01:47 I just found it interesting that it was an epic fail Apr 23 14:02:37 though that's an important part of experimentation Apr 23 14:03:46 mdp: in fact, after like 15 people here asking to get xenomai to even compile, this one is a welcome "success" Apr 23 14:04:23 still, a simple FIQ IRQ on a timer would be a much simpler PWM solution.... Apr 23 14:04:58 doesn't the omap have pwm hardware? Apr 23 14:05:34 also their conclusion at the end is wrong Apr 23 14:05:38 yes, but they need more channels Apr 23 14:05:48 they compare an external signal with the kernel generated one Apr 23 14:05:56 with unsynced clocks Apr 23 14:06:00 av500, since I'm unconcerned with BOM cost, my solution to this is to add more h/w…like I told the guy on the list Apr 23 14:06:04 KISS Apr 23 14:06:08 I know of someone who wanted 8+ channels Apr 23 14:06:25 surely there must be some simple pwm chip controlled over i2c one can buy Apr 23 14:06:32 lol Apr 23 14:06:35 mru: sure Apr 23 14:06:44 tlc5940 is popular Apr 23 14:07:05 but you add cost Apr 23 14:07:07 but i2c is maybe not fast enough to controll steppers Apr 23 14:07:07 probably cheaper than spending many engineering hours on software Apr 23 14:07:19 av500: put it on spi then Apr 23 14:07:53 av500, that's why you go spi Apr 23 14:08:03 I spi with my little i Apr 23 14:08:05 mru, right, 5940 will drive at 30MHz Apr 23 14:08:28 mru, and oh, btw, you fail for considering s/w cost in a project plan…fail fail fail! Apr 23 14:09:01 bom is all that matters :) Apr 23 14:09:05 resolution of the pwm's matters too Apr 23 14:09:19 also, who designed their horrible blog theme Apr 23 14:09:21 mdp: so 5940 and an arm926 then Apr 23 14:09:30 should be cheaper than an omap3 Apr 23 14:09:34 mdp: spot on Apr 23 14:09:46 or an arm926 with enough onchip pwm channels Apr 23 14:09:57 yeah, omap3 is overkill for 95% of the simple tasks people are trying :) Apr 23 14:10:43 av500, we are the poster children for not considering s/w cost..so I'm allowed to say that :P Apr 23 14:11:08 all hw-people think sw doesn't cost anything ;) Apr 23 14:11:31 all biz people can think is what will be the final cost in volume Apr 23 14:11:34 mdp: I failed many times to explain my CEO the "years of pain ahead" for the 10¢ he saved on the BOM :) Apr 23 14:11:54 aholler: didn't you get the memo that software is free? Apr 23 14:12:04 free SW is free Apr 23 14:12:04 this BOM, is that a big or little endian marker? Apr 23 14:12:15 never endian Apr 23 14:12:18 av500, yes, maintenance and support gets some fixed line item on a product cycle ongoing costs thing. Apr 23 14:12:20 koen: what the hell does dhclient use openssl for? Apr 23 14:12:24 MOB is the big endian version Apr 23 14:12:29 it never gets a cost associated with poor design decisions :) Apr 23 14:12:49 but it just doesn't cost the hw-people, they are ok if it isn't there department which has the costs. Apr 23 14:13:03 XorA: I don't know Apr 23 14:13:47 XorA: some kind of client authentication scheme Apr 23 14:13:48 aholler, that's why most orgs with a h/w design team have a process that carefully excludes s/w team input…too costly to allow that input Apr 23 14:14:16 mdp: yeah, sw-people just get in when the hw is ready to deliver. Apr 23 14:14:17 aholler, this is one thing that seems to be constant in the industry regardless of where I've been Apr 23 14:14:46 I had that too too often ;) Apr 23 14:14:53 see "Real Men Have Fabs" mentality Apr 23 14:14:59 mdp: heh, you must have worked here :) Apr 23 14:15:20 but the fun thing is that all this has come back to haunt the HW team in the end Apr 23 14:15:21 koen: fouth option, use udhcpc if you havent mounted /usr yet Apr 23 14:15:25 it's best that we just learn our place :) Apr 23 14:15:27 koen: or maybe thats fith Apr 23 14:15:35 who needs /usr? Apr 23 14:15:56 XorA: yeah, I went with 4) and added the bbappends, they can get deleted when a proper fix hits the trees Apr 23 14:15:59 make OpenSSL base is scary as hell Apr 23 14:16:05 mru: who need any other binary beside /sbin/init ? Apr 23 14:16:15 who needs userspace? Apr 23 14:16:22 panto, I don't need userspace either Apr 23 14:16:24 mru, +1 Apr 23 14:16:26 /sbin/ioquake3 Apr 23 14:16:30 that's where all the crappy code lives Apr 23 14:16:32 thats all you ever need Apr 23 14:16:35 kernelspace ought to be enough for anyone Apr 23 14:16:41 modprobe ioquake3 Apr 23 14:16:47 and recent cpus have hyperspace too Apr 23 14:17:09 * panto waits for the discussion to degenerate to the point where someone will note that he used to run Linux on an abacus 50 years ago... Apr 23 14:17:09 who needs a kernel, non-os is state of the art Apr 23 14:17:24 tux webserver ftw Apr 23 14:17:57 panto, I used to make beowulf clusters out of those hand-keyed IPLed abacus linux systems Apr 23 14:18:07 panto, over token ring Apr 23 14:18:13 mdp: the winnar is you... Apr 23 14:18:19 :P Apr 23 14:18:24 koen: BTW I agree with the revert patch, OpenSSL has no place in /lib/ but I guess when Lennert is finished all of gnome will be in /lib :-D Apr 23 14:18:54 I thought he is working on getting all to /usr Apr 23 14:18:59 XorA: fedora is actually moving it to /usr :) Apr 23 14:19:11 I was told that sysfs is the future, and I don't see it worked in the discussion Apr 23 14:19:41 * XorA waits for the systemd bonobo plugin :-D Apr 23 14:19:46 panto, speaking of hack value…it's nearing time to revive PLIP over parport for use with beagle* Apr 23 14:19:55 why not put everything in a kernel-managed dynamically populated apptmpfs? Apr 23 14:20:14 why call it tmp? /appfs Apr 23 14:20:17 you forgot to work 'cloud enabled' in there Apr 23 14:20:19 make it it lock waking Apr 23 14:20:23 isn't that what android does? Apr 23 14:20:39 aholler, you can't say that name here Apr 23 14:20:41 * mdp shuns aholler Apr 23 14:20:49 android does everything (but weirdly) Apr 23 14:20:53 * XorA is mildly amused about Linux copying windows so closely now :-D Apr 23 14:21:20 instead of w everything will get a g in front Apr 23 14:21:26 XorA: s/linux/lennart/ Apr 23 14:21:37 But I will stay using k* ;) Apr 23 14:21:38 Lennux Apr 23 14:21:54 egotistical bastard Apr 23 14:22:05 egostical Apr 23 14:22:14 binary logging, a registry, device tree, requiring 3D to draw 2D objects Apr 23 14:22:15 ergotistical Apr 23 14:22:29 "lit, a new source code control tool" Apr 23 14:22:31 erotistical Apr 23 14:22:33 cogito ergo lennart Apr 23 14:23:09 XorA: wait till they deprecate the shell.... Apr 23 14:23:30 the only thing missing is an Intel puppet project to back the whole vision Apr 23 14:23:42 av500: hehe Apr 23 14:23:56 they still have the meego/tizen dead horse to flog around Apr 23 14:24:16 panto, systemd needs FT community manager and 50+ program managers Apr 23 14:24:54 * panto read FT as Financial Times Apr 23 14:25:00 hehe Apr 23 14:25:29 yes, that makes me feel like I'm on a .eu->.us flight…must have my FT! Apr 23 14:26:15 step out of the US and USA Today is not good enough for you? Elitist! Apr 23 14:26:35 panto, I'll let you know if I can somehow get the retro parport project worked into our day job..there's got to be a way Apr 23 14:26:58 panto, I do struggle with the lack of cartoons in FT Apr 23 14:27:16 mdp: is the front page not funny enough? Apr 23 14:27:22 :) Apr 23 14:27:42 av500: please don't make fun of economists - their lives are drab enough as it is Apr 23 14:27:57 oddly enough, the british airways 'business' magazine usually has some interesting articles Apr 23 14:28:11 economics: The science with the ability to predict the past with 100% accuracy Apr 23 14:39:46 panto: science? Apr 23 14:40:29 ok, "science" Apr 23 14:42:02 3.3.3 almost booted, looks like something more got fixed, still not enough ;) Apr 23 14:42:34 3.3.4RC! Apr 23 14:43:10 never, that will eat my fs, damage my brain and whatever else Apr 23 14:44:10 well, there is also linux-arm :) Apr 23 14:44:24 but that might be fucked Apr 23 14:45:07 h, recovered state with another try, just "booting the kernel" ;) Apr 23 14:45:14 totally fucked Apr 23 14:45:46 does rmk has an rpi? Apr 23 14:46:01 ffmpeg has an arpi Apr 23 14:46:19 :) Apr 23 14:46:40 an advanced rpi, like omap5? Apr 23 14:46:53 $15 Apr 23 14:47:11 that's how you measure advancements Apr 23 14:48:23 the daily stock ubuntu precise image is booting my xm Apr 23 14:48:28 mdp: in units of mountain view espressos? Apr 23 14:49:07 "LXDE theme for user “pi” is now “mist” (performance improvement)" Apr 23 14:50:14 coffee should be banned from bay area Apr 23 14:50:29 XorA: it will crash the economy of the whole state Apr 23 14:50:53 they just cant handle the caffeine and are far too hyper Apr 23 14:51:06 and it probably causes cancer as well :-D Apr 23 14:51:31 speaking of coffee, I need to out and buy more beans Apr 23 14:51:35 you can pry the coffee make maker for my cold jittery hands Apr 23 14:51:50 magic beans?!? Apr 23 14:52:16 jelly beans Apr 23 15:00:57 what is equivalent c code for "cat /sys/class/gpio/gpio37/value"? Apr 23 15:02:18 fp = fopen("/sys/class/gpio/gpio37/value"); puts(gets(fp)); fclose(fp) Apr 23 15:02:23 char buf[4096]; int fd = open("/sys/class/gpio/gpio37/value", O_RDONLY); int s = read(fd, buf, sizeof(buf)); write(0, buf, s); Apr 23 15:02:42 err, you need a buffer Apr 23 15:02:45 mru: 4096 is enough for everybody? Apr 23 15:02:49 yes Apr 23 15:02:54 sysfs files are limited to one page Apr 23 15:03:04 thanks a lot Apr 23 15:03:06 ha, so PAGE_SIZE! Apr 23 15:03:07 err, char buf[BUFSIZ]; gets(buf, fp); Apr 23 15:03:17 av500: PAGE_SIZE is 4096 on arm Apr 23 15:03:36 yes Apr 23 15:03:43 y u hate portable code? Apr 23 15:04:08 panto: what if BUFSIZ is smaller than the file? Apr 23 15:04:12 shouldn't you use get_page_size? Apr 23 15:04:15 then you have a very nice buffer overflow Apr 23 15:04:27 then the kernel ownz you! Apr 23 15:04:34 getpagesize(), that it Apr 23 15:04:35 mru: my version only reads one line Apr 23 15:04:45 koen: sysconf(_SC_PAGESIZE), sure Apr 23 15:04:52 but BUFSIZ is guaranteed to have a minimum size IIRC Apr 23 15:04:57 panto: is not Apr 23 15:05:17 BUFSIZ: Size of buffers. This shall expand to a positive value. Apr 23 15:05:23 * panto puts on his language lawyer hat Apr 23 15:05:29 let me check Apr 23 15:05:40 that was a quote from the spec Apr 23 15:06:14 and the positive requirement is a posix addition Apr 23 15:06:26 c99 only requires that it be an integer constant expression Apr 23 15:07:00 "that is the size of the buffer used by the setbuf function" Apr 23 15:07:11 so maybe that implicitly ensures it's >0 Apr 23 15:08:08 I always use it as a constant that declares some semi-efficient block size for IO Apr 23 15:08:44 yeah, in theory it's just a positive integer - so I concede Apr 23 15:09:21 uclibc on blackfin sets it to 256 Apr 23 15:10:34 just use std::string Apr 23 15:10:36 which should be enough for everybody Apr 23 15:11:22 would it be correct to use something like "int = system(cat /sys/class/gpio/gpio37/value)"? Apr 23 15:11:32 nope Apr 23 15:12:11 it is not possible to assing system output to a variable? Apr 23 15:12:19 sure its possible Apr 23 15:12:48 its also possible to read the man page for system Apr 23 15:16:43 are man pages accessible through social networks? Apr 23 15:18:11 one should the make friends to all out there ;) Apr 23 15:18:16 s/the/them/ Apr 23 15:19:35 Does somebody know performance limits for gpio ports on BeagleBone? I`d like to solder some 10 MSPS ADC to GPIO and try some DSP, but don`t know max throughput of ports Apr 23 15:19:50 fooblya_monad: wont work Apr 23 15:20:46 why? Apr 23 15:21:39 because of performance limits for gpio ports on BeagleBone Apr 23 15:21:52 :) Apr 23 15:22:13 use an spi adc Apr 23 15:49:32 av500, I had some more time to work on my IRQ latency problem, did some more measures, it seems that 4/5 times the delay is exactly 127 ms, Apr 23 15:49:54 4/5 times that it fails* Apr 23 15:50:21 more than 99% of the time the irq is served under 1ms Apr 23 15:51:30 statistically you are on the safe side, go write that term paper :) Apr 23 15:51:30 hmm, gregkh had announce 3.3.3, just not with the word stable in the subject. Apr 23 15:51:44 aholler: so 3.3.3 is unstable? Apr 23 15:51:59 it's just "Linux 3.3.3" Apr 23 15:52:24 not sure if that means something ;) Apr 23 15:52:41 oh my god, some one call phoronix Apr 23 15:52:46 he's taking over linux development Apr 23 15:52:54 they'll want to hear about this Apr 23 15:53:17 linus retires? Apr 23 15:53:29 retries Apr 23 15:53:37 this time with micro kernel... Apr 23 15:53:38 abort? Apr 23 15:53:40 doing it right Apr 23 15:54:00 * Russ looks up a picture of a quetzal Apr 23 15:56:50 koen, did you want to mirror the demo @ OSUOSL? Apr 23 15:57:20 ka6sox: that's a jkridner question, he's in charge of beagle mirrors :) Apr 23 15:57:34 okay..sorry Apr 23 15:57:54 jkridner, ^^^ Apr 23 15:58:14 Russ: 3.3.3 contains something for you: USB: pl2303: fix DTR/RTS being raised on baud rate change Apr 23 15:58:41 since I don't change baud rate... Apr 23 15:59:02 on 3.3.0 though, changing the baud rate to zero does drop DTR as expected Apr 23 15:59:18 hmm, 115200 is the default on boot-up? Apr 23 15:59:26 for? Apr 23 16:00:47 whatever, if you connect with 115200 you have to change from the default one. so the description reads if it used to raise DTR Apr 23 16:01:05 opening a port raises DTR anyway Apr 23 16:03:36 does anyone even connect those pins any more? Apr 23 16:04:03 I do, arduino does Apr 23 16:04:03 some people even still try to send data over serials ;) Apr 23 16:04:06 x10 does Apr 23 16:04:46 aholler: you mean a 2nd data channel over DTR? Apr 23 16:05:15 rts was mentioned too. Apr 23 16:05:27 yes Apr 23 16:06:09 and I know of a lot people which are confusing hw-handshake (rts/cts) with checksumms Apr 23 16:06:29 i guess you can use it as a 1bit checksum Apr 23 16:06:40 I know a lot of people who if presented with two things will confuse them Apr 23 16:06:50 .oO(enable hw-handshake and all errors are gone) Apr 23 16:06:54 hey, everyone knows that on serial, 16 bit checksums are enough for anybody Apr 23 16:08:23 pfft, checksums are for people that cant afford proper cables :P Apr 23 16:08:52 they just have to be short enough Apr 23 16:09:07 ;) Apr 23 16:09:30 pure, oxygen-free copper Apr 23 16:09:39 from denon Apr 23 16:09:44 silver ftw Apr 23 16:09:50 connected in the right direction Apr 23 16:09:52 well, silver plated Apr 23 16:09:53 and directional ethernet cables Apr 23 16:27:26 mru, aren't those from Monster? Apr 23 16:27:58 don't remember Apr 23 16:28:12 and then there are the audio-grade wall power socket Apr 23 16:28:32 ka6sox: the ethernet one is from Denon Apr 23 16:28:45 av500, ah Apr 23 16:29:13 d3 Apr 23 16:29:16 aargh Apr 23 16:29:24 http://www.amazon.com/Denon-AKDL1-Dedicated-Link-Cable/dp/B000I1X6PM Apr 23 16:31:26 whoa, why don't I have those ethernet cables Apr 23 16:31:59 I'm just glad I discovered Monoprice...otherwise I'd be paying those kinds of prices. Apr 23 16:34:29 my $0.99 x-over cable is working fine for DRBD and @ 1Gb even :D Apr 23 17:11:16 does anyone know any replacements for x-loader? Apr 23 17:11:31 anything named MLO Apr 23 17:11:43 u-boot Apr 23 17:12:01 doesn't x-loader load u-boot? Apr 23 17:12:10 yes Apr 23 17:12:16 but these days, even uboot can load uboot Apr 23 17:12:17 it doesn't have to be u-boot Apr 23 17:12:18 or can u-boot be configured without x-loader? Apr 23 17:12:23 yes Apr 23 17:12:30 spl Apr 23 17:12:46 mini-uboot loads maxi-uboot Apr 23 17:13:11 which loads a slightly more complex state machine named Linux Apr 23 17:13:11 cool, I'm going to swim around boot loaders for a bit Apr 23 17:13:34 whats a state machine? Apr 23 17:14:27 av500: hey, i know how much you love openmax and how it works ""the same way"" for everyone ... i've been offered an internship to implement openmax api on another SoC, i could increase entropy even more :D Apr 23 17:14:37 \o/ Apr 23 17:15:20 phh: we recently had to revisit the omx stuff you did Apr 23 17:15:28 the guy is slowly recovering Apr 23 17:15:43 av500: hum which part ? Apr 23 17:17:27 openmax does work the same way for everyone: not at all Apr 23 17:17:50 mru +1 Apr 23 17:32:04 phh: the prodtest one Apr 23 17:48:59 av500: what's wrong about it ? Apr 23 17:49:02 av500: it's fairly simple Apr 23 17:51:17 phh: not much, just adapting it to new ducati side Apr 23 17:51:40 Hello! I'm using Beagleboard B4 and image usedd is ubuntu-11.10-preinstalled-desktop-armel+omap.img This is my uboot log - http://pastebin.com/MG55iCvJ I'm always finding it to be on that last line. When I insert card to host system and do fdisk -l , it says - http://pastebin.com/wKqhydD6 Can anyone help me identify the problem? Apr 23 17:52:49 did you fsck the filesystem? Apr 23 18:01:40 are you holding the usr button when initially booting? Apr 23 18:05:14 thurbad: yes Apr 23 18:07:45 Ok, now I tried it again. It says - http://pastebin.com/BLawpvP0 Apr 23 18:09:44 uuuh, you definitely dont want to run ubuntu-desktop in a B series beagle Apr 23 18:09:56 it will run out of ram all the time Apr 23 18:10:34 ogra_, eh, its all good Apr 23 18:10:35 xfce Apr 23 18:10:55 ubuntu-11.10-preinstalled-desktop-armel+omap.img doesnt contain xcfe Apr 23 18:11:05 himcesjf: > Apr 23 18:11:15 *xfce Apr 23 18:11:19 apt-get install Apr 23 18:11:55 i doubt even the installer will properly finish Apr 23 18:12:16 use ubuntu-server (note its serial only) and then install xfce on it Apr 23 18:12:41 is there an armel ubuntu-server? Apr 23 18:12:44 I did it with my b4 Apr 23 18:12:46 the desktop image has the fully graphical installer ... i doubt that will survive in 256M Apr 23 18:13:41 ka6sox, for 11.10 there is Apr 23 18:13:57 I didn't use the graphical installer Apr 23 18:14:03 (in 12.04 armel is dead) Apr 23 18:14:16 Russ, then you didnt use ubuntu-11.10-preinstalled-desktop-armel+omap.img Apr 23 18:14:19 :) Apr 23 18:14:42 this one expands itself to the full size of the card and then reboots into the graphical ubuntu installer Apr 23 18:14:55 ah, then I used the rcn-ee one Apr 23 18:14:57 Ok, I'll try http://cdimages.ubuntu.com/releases/oneiric/release/ubuntu-11.10-preinstalled-server-armel+omap.img.gz Apr 23 18:15:11 yeah, thats way better for a B beagle Apr 23 18:15:18 desktop works from C4 on Apr 23 18:15:33 but i wouldnt suggest anyone to try it on a B Apr 23 18:15:47 Only ubuntu-xfce and rcn-ee image work on BB B4? Apr 23 18:16:23 I mean ubuntu server and installin xfce later Apr 23 18:16:26 well, its a 256M board ... unity or gnome will eat all your ram even before you start an app Apr 23 18:16:40 yeah, that should work fine Apr 23 18:16:57 sudo apt-get install xorg xfce lightdm Apr 23 18:16:59 also you might consider compressed ram pages Apr 23 18:17:02 that should get you all you need Apr 23 18:17:28 Ah, I thought xubuntu-desktop package Apr 23 18:17:33 or that, yes Apr 23 18:17:40 but thats more bloated than just xfce Apr 23 18:18:32 xubuntu shares a lot of desktop backend components with the normal ubuntu desktop, that adds a lot of stuff compared to a plain xfce Apr 23 18:19:08 if you want it realy lightweight, try lubuntu-desktop :) Apr 23 18:19:11 *really Apr 23 18:19:41 ratpoison FTW Apr 23 18:19:42 that will give you a fully usable desktop with less than 64M used when idling Apr 23 18:20:02 and should leave some headspace for i.e. running a browser Apr 23 18:20:15 Hmmm Apr 23 18:21:07 ogra_, what do you mean "in 12.04 armel is dead"? Apr 23 18:21:25 with 12.04 ubuntu switched to hardfloat by default Apr 23 18:22:04 so armhf is the new armel :) Apr 23 18:22:25 yea, I don't understand what they mean by hardfloat and soft float... Apr 23 18:22:36 Could anyone brief me on that? Apr 23 18:22:43 remember the 486 ? there were 486 SX and DX ... Apr 23 18:23:00 one had a math coprocessor to do math operations in hardware ... Apr 23 18:23:06 the other had to do it in SW Apr 23 18:23:15 ye .. Apr 23 18:23:20 armel vs armhf is somewhat similar Apr 23 18:23:40 not really Apr 23 18:23:45 it changes the hw :) Apr 23 18:23:53 :S Apr 23 18:23:53 armel does float operations in SW while armhf can lod that off to the SoC Apr 23 18:24:09 stop spouting crap Apr 23 18:24:14 now there are two sorts of armel. Apr 23 18:24:19 actually it can do it in hardware. Apr 23 18:24:23 mranostay, feel free to explain it more easily Apr 23 18:24:31 err Apr 23 18:24:32 mru, Apr 23 18:24:39 (sorry mranostay) Apr 23 18:26:11 just because hard float instructions are used doesn't mean that they'll get executed by dedicated hardware Apr 23 18:26:17 Is it related to GCC floating-point? Apr 23 18:26:24 ogra_: just different calling convention for float arguments Apr 23 18:27:33 http://wiki.debian.org/ArmHardFloatPort Apr 23 18:28:42 so how is the FPU not a piece of hardware ? Apr 23 18:28:58 ogra_, what FPU? Apr 23 18:29:10 on what SoC? Apr 23 18:29:11 I'll order 12! Apr 23 18:29:23 "A lot of modern ARM boards and devices ship with a floating-point unit (FPU) but the current Debian armel port doesn't take much advantage of it." Apr 23 18:29:44 "A new ARM port requiring the presence of a FPU would help squeeze the most performance juice out of hardware with a FPU. " Apr 23 18:29:46 even that page contains crap Apr 23 18:29:59 that page is mostly crap Apr 23 18:30:04 fpu isn't really a unit anymore Apr 23 18:30:12 Russ: that's beside the point Apr 23 18:30:37 all widely used chips with arm11 or later have hw fpu as part of the cpu Apr 23 18:31:06 don't modern ARMs have like 2 floating point capable blocks? Apr 23 18:31:10 ogra_: the only difference is that hardfp uses the vfp registers to pass float arguments Apr 23 18:31:11 no Apr 23 18:31:13 "all widely" is another way of saying most but not all Apr 23 18:31:28 I challenge you to find an arm11 without vfp Apr 23 18:31:33 NEON can't do fp? Apr 23 18:31:36 I want mmx Apr 23 18:31:43 which obviously won't work on a cpu without vfp since it doesn't have the registers Apr 23 18:31:46 ds2: neon and vfp are rather tightly coupled Apr 23 18:31:58 mru: Could you suggest me some resource which explains armel, armfhs etc ? Apr 23 18:32:08 jannau, well, try to explain it to someone who doesnt know the term vfp now :P Apr 23 18:32:11 so how does HF deal with those things *without* vfp? or are debian/ubuntu just ignoring them now? Apr 23 18:32:20 mru: fair enough... but they take different instructions Apr 23 18:32:26 ka6sox: by not running at all Apr 23 18:32:41 ka6sox, they dont run on them Apr 23 18:33:11 hmmm...so why should they care about i386 anymore under that logic... Apr 23 18:33:15 and by developer decision armhf in debian and ubuntu are armv7 only atm Apr 23 18:33:20 ogra_: softfp passes the arguments on stack so a cpu without fpu can use software floating point emulation Apr 23 18:33:32 ka6sox, ubuntu is i686 Apr 23 18:33:37 ogra_: vfp is just the name of the arm fpu Apr 23 18:33:40 doesnt run on i386 Apr 23 18:33:48 couldn't you do a snail pace mode by kernel emulating hf on say... arm9's? Apr 23 18:33:57 jannau, still, my mom wouldnt understand that :) Apr 23 18:34:41 ds2: yes, kernel has that Apr 23 18:34:41 math by traps ;) Apr 23 18:34:42 ogra_: they don't care Apr 23 18:35:49 * ogra_ leaves it to jannau then to explain el vs hf to oragsy Apr 23 18:35:52 and there is little reason to care since it usually makes very little difference Apr 23 18:37:04 ogra_: No, its all right.. I'm trying to understand it myself with some new words above :) Apr 23 18:37:10 so ARMv6 is no longer supported... Apr 23 18:37:25 ka6sox, you theoretically could support it Apr 23 18:37:38 *cough* thorny dessert board *cough* Apr 23 18:38:02 yet another reason to *not* use a desktop distro on an embedded device :P Apr 23 18:38:04 and afaik there is someone working on rebuilding debian armhf for the RPi Apr 23 18:38:28 (which is v6) Apr 23 18:38:49 ka6sox, define "embedded device" ? :) Apr 23 18:38:54 thats going to be painful. Apr 23 18:39:01 do you call a beagle XM still embedded ? Apr 23 18:39:39 or the pandaboard ... Apr 23 18:40:14 embedded is what happens when you throw BGAs around Apr 23 18:41:01 BGAs in 4 layer boards :( Apr 23 18:42:00 yes, the line is blurring. Apr 23 18:42:22 so now we have OABI EABI and something else? Apr 23 18:42:39 OABI ? is there anyone still using it ? Apr 23 18:43:43 mru, so the msm7200 isn't widely used? Apr 23 18:44:59 Russ: the msm* series is not arm[^v]* anything Apr 23 18:45:08 msm7200 is arm11 Apr 23 18:45:49 arm1136 Apr 23 18:46:35 hmm Apr 23 18:46:39 and what about plx's arm11 based NAS chips? Apr 23 18:47:46 * ogra_ wonders what Russ tries to say Apr 23 18:47:50 blah... another abomination Apr 23 18:48:03 (arm11 cores without an fpu) Apr 23 18:48:26 ka6sox, EABI, OABI, and Thumb ABI http://cross-lfs.org/view/clfs-embedded/arm/cross-tools/abi.html Apr 23 18:48:26 yes, what about them ? Apr 23 18:49:25 * _av500_ prefers LOLABI Apr 23 18:49:28 Russ, debian armel will run just fine on them, just armhf wont Apr 23 18:50:13 (ubuntu arm was always targeted for v7 anyway and dropped pre-v7 support after the second release it had) Apr 23 18:51:08 poor rpi ;) Apr 23 18:53:03 good thing I am using OE then... Apr 23 18:53:28 on my ARM1136 Apr 23 18:55:06 is there a way to modify the existing SMSC PHY on the beaglebone to add another network interface to it? Apr 23 18:55:56 nemik, not without modifying the PCB or getting VERY creative with jumper wires Apr 23 18:55:57 or would i have to mux some GPIO's to enable the second RMII interface and put in a chip with a switch there? sorry, don't know much about these PHYs... Apr 23 18:56:34 nemik, second Ethernet peripheral isn't pinned out to the headers in a usable way, regardless of pin mux Apr 23 18:56:35 bradfa: oh yea, i know that. but i'm just wondering of the SMSC chip could be configured at all to do that. or do i need to use another PHY entirely Apr 23 18:56:50 nemik, you'd need another PHY Apr 23 18:57:04 bradfa: ah, damnit. alright. thank you! Apr 23 18:57:39 nemik, I was also interested in adding another native Ethernet interface to my Bone, but no such luck (as far as I know) Apr 23 19:03:51 bradfa: well one quick and easy way is to use a USB-ethernet adapter Apr 23 19:05:13 as far as i know, the usb-cdc-acm kernel driver is installed and detects them on the angstrom builds Apr 23 19:05:29 or could be installed easily enough Apr 23 19:05:46 nemik, yes, USB Ethernet is an option (see Panda, xM, etc) but they have their own drawbacks, of which I was not willing to take Apr 23 19:06:35 latency is not as good as most internal Ethernet MACs Apr 23 19:09:07 layout your own board Apr 23 19:09:31 ds2, Yes, that's what we're doing Apr 23 19:26:23 GFP_DMA or GFP_DMA32 on beaglebone? Apr 23 19:31:41 ds2: yea, we're doing the same. just gotta find a better replacement PHY now Apr 23 20:08:02 Hmmm, is it possible that scattergather mappings are not completely functionnal on bone? I get data corruption except if I always DMA block of PAGE_SIZE? Apr 23 20:25:51 I'd try to help except I am interested in deleting the ethernet Apr 23 20:39:36 ds2, deleting the ethernet? Apr 23 20:39:50 I'm not sure what you mean by that Apr 23 20:42:02 jsabeaudry, GFP_DMA, GFP_DMA32, and GFP_KERNEL all grab from zone normal in this case, so technically doesn't matter Apr 23 20:43:32 jsabeaudry: it is possible if there are not demonstrateable existing drivers with SG working Apr 23 20:44:18 mdp, Would you says that is unlikely or very unlikely ? ;) Apr 23 20:44:56 let's just say that I trust *nothing* from some sources until it's personally tested Apr 23 20:47:54 At this point, I'm pretty sure that I'm screwing up with the dma_map_sg and unmap... It seems I can get the heartbeat led to stop blinking if I'm very wrong with the parameter so the data corruption must come from a slightly wrong parameter :) Apr 23 20:48:43 put that into the shutdown code Apr 23 20:49:12 I always found it annoying that the heartbeat never stops on shutdown Apr 23 20:49:29 jsabeaudry: I'll admit that I find it unlikely that the edma driver is so broken to have a corner case triggered by simple sg mappings Apr 23 20:50:49 aholler, hehe Apr 23 21:08:32 rcn-ee: hi, what does "igevp2 (no video)" mean here : https://github.com/RobertCNelson/stable-kernel ? Apr 23 21:14:23 rcn-ee: I mean, is "video" == "dss2" or "video" == "sgx" in this statement? Apr 23 21:14:56 rcn-ee: or maybe DSP codecs? Apr 23 21:42:23 hey guys. I am having trouble connecting a pico projector to my gumstix board and I was wondering if anyone here could help me with it. I have two projectors (Optoma PK301 and DLP Development Kit Pico Projector V2) and I can't get any output on either projector. I have set the dvimode to 640x480MR-16@60 but that doesn't seem to work. Anyone have any ideas what the problem could be? Apr 23 21:44:12 joejacobs: play with the MR flags? Apr 23 21:44:34 you mean remove them? Apr 23 21:44:49 well they mean different things which I cant remember off top of head Apr 23 21:44:59 R is reduced timings I think Apr 23 21:45:00 ssvb, it means on my igepv2, i currently get no video out.. Apr 23 21:45:49 checked that the tfp410 is enabled? Apr 23 21:45:55 i.e. DVI_PUP is sane Apr 23 21:46:09 rcn-ee: at least on my igepv2, your kernel gets a working framebuffer, but a lot of other things are broken Apr 23 21:46:18 the Pico dev kit is reasonably tolerant of bad timings Apr 23 21:46:37 rcn-ee: hence I wanted to clarify what that notice was supposed to mean Apr 23 21:46:50 ds2: DVI_PUP? not even sure what that is! Apr 23 21:47:31 DVI_PUP is a GPIO that is used to enable/disable the DVI transmitter Apr 23 21:47:39 ssvb, it could just a be a board problem with my board, i picked it up on the first run, and then always ran it headless as a gcc/kernel regression tester.. Apr 23 21:47:41 it is available on the LCD header so you can measure it Apr 23 21:48:30 so other then ehci, ethernet lan, and serial not much gets tested on it.. Apr 23 21:50:09 ssvb, and their patchset is still on 2.6.37.. http://www.isee.biz/gitweb/?p=pub/scm/linux-omap-2.6.git;a=summary Apr 23 21:50:35 rcn-ee: ok, I see, I have a dm3730 based board, and I first took the vendor provided X-Loader+kernel, and then replaced the kernel with yours Apr 23 21:51:01 i have the omap3430 720mhz version. .;) Apr 23 21:52:19 ds2: ok...well i'm gonna fiddle around with the MR flags first and see if I can get anything by changing that. how do I check if the DVI_PUP is enabled/sane? I am googling it right now and I see some thing related to u-boot??? Apr 23 21:52:49 rcn-ee: yes, I noticed that their master branch is a bit old and there are some patches which are still are not in upstream (though some of them look like just reshuffling of code and not real fixes) Apr 23 21:52:59 simpliest way is to stick a logicprobe/DMM/scope on that signal and measure it Apr 23 21:53:15 look at the schematic and the datasheet Apr 23 21:53:40 yeah, they were very active for awhile in u-boot/mainline... but then after 2.6.37 they just kinda dissappeared.. Apr 23 21:54:29 they even sent a few patches directly to me back in 2.6.35.. Apr 23 21:55:20 rcn-ee: regarding SGX stuff, you have a special repackaging script for 2.6.37 kernel, which uses older blobs. Are there some problems with the newer blobs on 2.6.37? Apr 23 21:57:20 ssvb, off the top of my head.. i don't remember, looks back the git log.. Apr 23 21:58:36 rcn-ee: ok, I'll check, right now I'm using their 2.6.37 kernel plus the SGX modules compiled from SDK sources and wondered if anything bad should be expected Apr 23 21:58:50 ds2: forgot to mention this...when I use the dlp dev kit I get a flicker while the board is booting up. just a bunch of horizontal lines that interfere with the boot image but nothing after that. nothing for the optoma though Apr 23 21:59:20 i thought that meant a framerate problem but I have tried both 50hz and 60 hz and i get the same problem with both Apr 23 21:59:26 oh ssvb, it's because, i pull in the sgx kernel headers into the *.deb build... so the 2.6.37->2.6.38 switch was a good time to bump the driver... (If you build the kernel modules externally, it should work just fine..) Apr 23 22:00:31 XorA: just tried removing the MR...no difference. just the same flicker during the initial part of the boot sequence then nothing Apr 23 22:00:33 i was trying to make things easier for users, aka they'd just have to isntall the lib's.. as the kernel modules were already prebuilt in the kernel.. Apr 23 22:03:06 rcn-ee: I understand :) actually I think that your scripts are a great idea. I just was a bit sad that igepv2 support is in a bad shape currently and wondered if anything could be improved Apr 23 22:04:09 other then video, we also need to figure out how to get the broadcom sdio wifi to work? Apr 23 22:04:44 or wait, it was some marvel one.. Apr 23 22:04:50 rcn-ee: trying to do it with 2 strikes? Apr 23 22:06:09 rcn-ee: my board is a NOWIFI variant, I got it when they had a special discount and sold such boards for just 95 euro (before VAT and shipping) :) Apr 23 22:06:35 it's probally the same board, just not populated.. Apr 23 22:18:59 I am trying to understand the scans sent to the TAP router (mentioned in ICEPICK-c) before DAP is added to the scan chain. irscan $jrc 7 -endstate IRPAUSE drscan $jrc 8 0x89 -endstate DRPAUSE As per my understanding, Instruction 7 selects a data register in which a dr-scan sends a 8-bit value 0x89. What does 0x89 stand for? Apr 23 22:22:13 Hi people. Could someone tell me how much RAM does the beagleboard have or supports? Apr 23 22:22:34 is the beagleboard.org web page down? Apr 23 22:22:44 and digikey and mouser and... also down? Apr 23 22:23:15 ds2: i am on beagleboard.org...seems fine to me Apr 23 22:23:43 yup, site is up Apr 23 22:24:55 is it in an unreadable font? Apr 23 22:25:13 definitely readable... Apr 23 22:25:52 anyone else feel the trend, lets ask questions instead of looking it up? ;) Apr 23 22:26:09 then what part of the amount of memory listed is wrong? Apr 23 22:29:57 Could someone tell me how much RAM does the beagleboard have or supports? Apr 23 22:32:50 http://beagleboard.org/hardware Apr 23 22:33:16 http://beagleboard.org/hardware-xM Apr 23 22:34:22 (note that revisions Cx of the original beagleboard have 256M instead of 128M() Apr 23 22:49:24 hmm..ktest.pl is doing a git bisect bad whenever a build fails Apr 23 22:49:40 I have bisect type as boot, and bisect skip as 1 Apr 23 22:52:26 # Are we looking for where it worked, not failed? Apr 23 22:52:30 if ($reverse_bisect) { Apr 23 22:52:34 $ret = !$ret; Apr 23 22:52:39 } Apr 23 22:53:08 where ret is -1 on build failure Apr 23 22:53:10 nice Apr 23 23:54:06 so with the beaglebone the devices showing up in linux /dev/ttyUSB0 is JTAG and /dev/ttyUSB1 is Serial? Apr 23 23:55:55 try and find out Apr 23 23:56:16 :) Apr 23 23:57:20 I did, i think i might have accidently connected screen to the JTAG interface and caused it to reboot while it was expanding the filesystem Apr 23 23:58:30 wanted a second opinion if that might have been what fubared my root fs Apr 23 23:58:32 its ok, the beaglebone doesn't ship with a jtag chip on it, so ttyUSB0 is connected to nothing Apr 23 23:59:04 er, wait nevermind Apr 23 23:59:12 it does, its just the extra connector that isn't there Apr 24 00:02:02 Russ: thanks for the beacon video!! Apr 24 00:03:11 np Apr 24 00:03:49 Russ: hehe, i think i need to get with you before ELC-E with the i2c stuff i'm making Apr 24 00:04:03 now all I need to do is put the beagle and beacon on a motor, like a fan Apr 24 00:04:22 sounds like fun, I'll definitely be at ELC-E Apr 24 00:04:34 er wait Apr 24 00:04:37 I'm thinking of the other one Apr 24 00:05:22 ELC-E would be nice though Apr 24 00:06:26 Russ: hehe, you should pressure your *cough* employer to go..... Apr 24 00:06:59 suppose I have a while to figure out a talk Apr 24 00:08:02 Russ: something like this: http://www.youtube.com/watch?v=UKswotx0tnU Apr 24 00:09:00 you know, when the video was loading, I was thinking, it can't be bacon with a fan Apr 24 00:09:39 it was the first (non-kevin) hit for 'bacon fan' Apr 24 00:11:01 what was the conference coming up in Irvine Apr 24 00:27:52 interesting place to have it Apr 24 00:30:29 I swear I saw one, now I can't for the life of me find it Apr 24 00:31:39 irvine is a boring place Apr 24 00:32:00 ah, but it is a place I'll be very close to Apr 24 00:32:13 there's a broadcom office there Apr 24 00:32:21 go troll them about the rpi Apr 24 00:32:29 heh Apr 24 00:32:40 there are broadcom offices here Apr 24 00:32:58 the people I had to deal with back in the day were in irvine Apr 24 00:33:03 http://www.broadcom.com/company/locations/chandler.php Apr 24 00:33:08 bill or bob or something Apr 24 00:33:22 ya, sure, I'll show up and ask for bill or bob or something Apr 24 00:33:44 one of those useless manager types Apr 24 00:33:50 always friendly but never got results Apr 24 00:34:56 possibly the clueless engineers were in part to blame for that Apr 24 00:36:56 mru: perfect storm :) Apr 24 00:38:24 there's the sucker Apr 24 00:38:44 ah, a different elc, I was looking at this when I was much too tired Apr 24 01:12:13 hi folks. got a beaglebone question. after a lot of frustration I had my usb wifi adapter working. decided I wanted PWM; upgraded my kernel; wifi's gone. well, not unexpected. but I could use some help clawing my way back to connectivity -- connman still makes no sense to me Apr 24 01:12:46 * mranostay takes a shot Apr 24 01:13:14 vodka? Apr 24 01:13:51 so: dmesg mentions nothing about firmware, but it more or less looks like things are okay: http://pastebin.com/YgG1SXBi Apr 24 01:14:29 iwconfig and ifconfig say nothing about wlan0; but ifup wlan0 delivers "ifup: interface wlan0 already configured" Apr 24 01:16:09 prior to the kernel upgrade `ifup wlan0` brought things up fine -- had a pre and post-init that launched wpa_supplicant. I believe I tried to get back to that place by killing connman and in the process eliminating my ability to even get the ethernet iface up. reflashed to SD backup, re-upgraded the kernel, and here I am. any suggestions on how to proceed without repeating that cycle? Apr 24 01:17:32 * mranostay takes a shot Apr 24 01:17:44 rum? Apr 24 01:20:43 * prpplague pops in Apr 24 01:20:48 did someone say rum? Apr 24 01:20:57 pop goes the weasel Apr 24 01:21:18 hehe Apr 24 01:24:01 bunch of functional alcoholics in here :) Apr 24 01:24:33 * mru insists he is quite procedural Apr 24 01:24:38 except when using emacs Apr 24 01:26:50 * mranostay smacks mru Apr 24 01:27:46 hmm, a http://wiki.multimedia.cx/index.php?title=Smacker Apr 24 01:28:43 not clicking that at work Apr 24 01:28:54 it's perfectly safe Apr 24 02:00:39 isn't smacker a predecessor of bink or somesuch? Apr 24 02:00:47 same company Apr 24 02:45:12 *gasp* no beer in fridge! Apr 24 02:45:38 who could have pulled off such a dastardly crime! Apr 24 02:45:50 me it seems Apr 24 02:47:31 past you is a jerk Apr 24 02:48:52 present one isn't much better :) Apr 24 02:49:37 hmm...next ktest problem Apr 24 02:49:47 when it runs it, it leaves stdin open Apr 24 02:50:18 haven't rewritten in it python yet? :) Apr 24 02:50:21 later, when it tries to kill conmux console, it doesn't die until I type `~$' Apr 24 02:50:25 I KNOW Apr 24 02:50:36 how can we modify kernel to add support for some specific LCD? Apr 24 02:51:59 I mean, to set timming, and the default value of hsync, vsync values to low Apr 24 02:52:21 we? Apr 24 02:52:44 have you attempted google first? :) Apr 24 02:53:05 ironically enough, conmux console is also perl Apr 24 02:59:00 ah...ktest.pl is sending SIGINT, but conmux-console has made that not work Apr 24 02:59:26 SIGKILL ftw Apr 24 02:59:28 when conmux tries to write to its output file descriptor after receiving ~$, then it dies **** ENDING LOGGING AT Tue Apr 24 02:59:58 2012