**** BEGIN LOGGING AT Wed Apr 25 02:59:58 2012 Apr 25 05:22:58 <_av500_> gm Apr 25 05:23:47 not yet Apr 25 05:24:26 <_av500_> always Apr 25 05:25:25 this is the day that won't end Apr 25 05:25:44 why couldn't i be an english major? :) Apr 25 05:26:42 <_av500_> coz you dont like cab driving? Apr 25 05:29:06 touche mr _av500_ touche Apr 25 06:03:18 Hello everybody. I have a rather specific question, since I saw some vids about the beaglebone. It looks like you can code in more then 1 language, but is there a favorite OS and/or programming language for coding very precise timings? I mean hereby a way to control stepper motors. Is I want there to pulse a pin with exactly 10kHz, then it should be quite stable and not be between 9.8 and 10.2kHz. I heard you normally need a realtime o Apr 25 06:03:59 no, you use the pwm functionality Apr 25 06:04:32 euhm, pwm to control steppers ? Apr 25 06:04:40 yeah Apr 25 06:04:59 you can run a realtime OS on the bone Apr 25 06:05:05 pwm would not work on a stepper Apr 25 06:05:18 ok let me rephrase. For constant pulses pwm might do something Apr 25 06:05:22 but its not constant Apr 25 06:05:32 educa: what kind of stepper driver? Apr 25 06:05:36 pulse/direction? Apr 25 06:05:40 yes Apr 25 06:05:42 4 wire direct? Apr 25 06:05:56 I come from arduino platform (actually avr with gcc) Apr 25 06:05:59 its pulse direction Apr 25 06:06:00 offloading to the PRU Apr 25 06:06:05 but how accurate do you need it? Apr 25 06:06:27 Well I need to get upto 20 or 30 khz signals Apr 25 06:06:34 but to describe accuracy Apr 25 06:06:35 hmmm Apr 25 06:06:42 lets say I coded it the simple way Apr 25 06:06:46 with some kind of delay Apr 25 06:06:55 and I say I want in my code a delay of 1 microsecond Apr 25 06:07:03 the question is not so much what langauge but what hw block Apr 25 06:07:07 then I would need that delay to be faitly accurate Apr 25 06:07:23 you could setup timers but linux is not a realtime OS Apr 25 06:07:29 so it would be best effort. Apr 25 06:07:41 how many steppers do you need? Apr 25 06:07:51 I like the fact that beaglebone looks to have sooo much cpu power Apr 25 06:07:56 2 motors at max Apr 25 06:08:02 its for a cnc laser cutter Apr 25 06:08:14 ah Apr 25 06:08:23 <_av500_> timer+FIQ Apr 25 06:08:40 PRU might be easier Apr 25 06:08:44 Maybe I could of course use the beagleboard for the steering and an extra arduino for high precision step generation Apr 25 06:09:01 for 2 of them, I wonder if you could use the McASP Apr 25 06:09:23 PRU, FIQ, hmmm thats latin at the moment. Is there some manual I could read that explains these? Apr 25 06:09:36 TRM :) Apr 25 06:09:42 If possible I would like to code native executables, not something intrepreted Apr 25 06:09:48 PRU is a microcontroller on the AM33x processor...there are 2 of them, IIRC Apr 25 06:09:58 FIQ is the fast interrupt signal on ARM processors Apr 25 06:10:26 unless you are writing in java/python/node.js, it is likely to be native Apr 25 06:10:35 C is the easiest and sanest Apr 25 06:10:38 I prefer gcc textmode Apr 25 06:10:43 yes Apr 25 06:10:47 educa: how many people have you talked to on this Apr 25 06:10:48 ? Apr 25 06:11:20 for beaglebone, nobody yet. For arduino I have everything setup but arduino just is too slow for the speeds I need Apr 25 06:11:26 there are quite a few of us wanting to do CNCish stuffon the AM33x Apr 25 06:11:41 educa: how many watts laser and what kind of materials? Apr 25 06:11:56 My laser will be 80 watts Apr 25 06:12:03 machine is 90% built now Apr 25 06:12:14 hmmm so that would need pretty fast feed speeds Apr 25 06:12:20 has a travel of 48x24" plates Apr 25 06:12:22 open loop steppers? Apr 25 06:12:32 whoa...big Apr 25 06:12:36 lead screws or belts? Apr 25 06:12:39 well I need 50khz to go 1 meter / sec Apr 25 06:12:46 the up-down table is screw Apr 25 06:12:49 x and z are belts Apr 25 06:12:54 arrrrgggg Apr 25 06:12:59 how many IPM? Apr 25 06:13:00 all is made in pure aluminium and extremely rigid Apr 25 06:13:09 alum and rigid heheh Apr 25 06:13:25 fastest engraving speed around 40"/second Apr 25 06:13:27 if you are doing up down, don't you need 3 motors? Apr 25 06:13:37 40IPS!? Apr 25 06:13:46 yes, but up-down never moves at same time as xy Apr 25 06:13:55 2400IPM?! holy crap... Apr 25 06:13:56 40IPS is MAX engraving speed Apr 25 06:14:02 yes its fast isn't it Apr 25 06:14:14 hope you got good kinetics Apr 25 06:14:18 motors are 3.1Nm Nema23 motors Apr 25 06:14:24 and why are you using steppers? Apr 25 06:14:37 I'd be worry about steppers loosing torque at those speeds Apr 25 06:14:46 everything runs on thk ball bearing slides Apr 25 06:15:02 torque is not an isue with 3.1Nm motors on this machine Apr 25 06:15:08 * mranostay scrolls up at the mad sciencist talk Apr 25 06:15:16 its not a cnc mill where a lot of forces are put on the head Apr 25 06:15:19 its light Apr 25 06:15:20 it is an issue when it drops down to 0 Apr 25 06:15:30 and the table moves backwards or it skips steps Apr 25 06:15:44 at 2400IPM, I'd personally go with a servo setup Apr 25 06:15:50 ds2 for that speed my motor only has to do 10 rpm so thats not a problem at all Apr 25 06:15:56 mranostay: CNC stuff is neat :) Apr 25 06:16:09 ds2: no doubt Apr 25 06:16:12 educa: oh you must have course gears? Apr 25 06:16:33 course gears ? Apr 25 06:16:48 pulleys are directly mounted onto the motor shaft Apr 25 06:16:53 educa: I'd personally look into using the PRU then FIQ and failing that maybe using the McASP hw + some logic Apr 25 06:16:56 40 tooth t5 pulleys Apr 25 06:17:05 well, course pulleys then Apr 25 06:17:12 these are timing belts, right? Apr 25 06:17:14 and what is course ? Apr 25 06:17:21 X and Y are on timing belts yes Apr 25 06:17:45 don't have a number there... I am more familiar with mills and screws Apr 25 06:17:47 for these high speed engravings with timing belts you will sure need some kind of backlash compensation Apr 25 06:18:10 yep or never engrave curves ;) Apr 25 06:18:11 lets say my whole laserhead weight around 300 grams Apr 25 06:18:26 300grams... Apr 25 06:18:29 engraving is raster engraving in staight lines Apr 25 06:18:49 0.2 pounds aprox ? Apr 25 06:18:49 2.2 oz?! Apr 25 06:19:08 this is the mirror that you are moving, right? Apr 25 06:19:08 possibly, dunno about oz :) Apr 25 06:19:17 16 oz to a 1 lb Apr 25 06:19:20 mirror/lens assemby yes Apr 25 06:19:41 educa: hang around...there are several other folks looking at this. they might have other ideas Apr 25 06:20:12 educa: what are you using to drive it? G code or something more custom? Apr 25 06:20:28 I'll write my own kind of stuff Apr 25 06:20:43 actualy HPGL parser would be enough allready Apr 25 06:21:04 I see Apr 25 06:21:13 don't like EMC? :) Apr 25 06:22:02 it runs on my most expensive pc, but thats a $3000 one and a little too expensive to run only a cnc laser Apr 25 06:22:20 port it to the bone Apr 25 06:22:34 you just need to get RTAI working on it...that's all Apr 25 06:22:53 educa: do you have postings on your build? Apr 25 06:23:17 I'll be back tonigh, promised Apr 25 06:23:20 have to run now Apr 25 06:23:24 but I have postings yes Apr 25 06:23:31 I'll show them thn Apr 25 06:23:34 tnx for info Apr 25 06:23:41 have fun Apr 25 06:26:38 always a good way to get help in this channel Apr 25 06:26:40 mention lasers Apr 25 06:33:59 hehe Russ Apr 25 06:34:06 I'm temporarily back by the way Apr 25 06:35:01 If I understand right then beaglebone has a lot of io pins available Apr 25 06:35:16 yes, and those nifty PRUs Apr 25 06:35:45 so what I actually want to do it set a pin high, call some function to halt execution for x microseconds and then go on Apr 25 06:36:00 it MAY be blocking delay even Apr 25 06:36:02 then you probably want to let the PRU do it Apr 25 06:36:06 doesn't have to be threaded Apr 25 06:36:37 ok, and now for the dummy question. Where do I find info about this PRU? Apr 25 06:36:49 the SRM and TRM Apr 25 06:37:14 actually just the TRM I think Apr 25 06:38:01 TRM stands for ? Apr 25 06:38:08 http://www.ti.com/lit/pdf/spruh73 Apr 25 06:38:12 technical reference manual Apr 25 06:39:58 'The Programmable Real-Time Unit and Industrial Communication Subsystem (PRU-ICSS) consists of dual 32-bit RISC cores (Programmable Real-Time Units, or PRUs), memories, interrupt controller, and internal peripherals that enable additional peripheral interfaces and protocols.' Apr 25 06:40:46 wow 4290 pages :) nice read Apr 25 06:40:57 hmm...but a little skimpy on pru Apr 25 06:41:21 http://www.ti.com/litv/pdf/spry136a Apr 25 06:41:50 yeah page 180 and thats all there is there :) Apr 25 06:45:54 For additional information, please visit www.ti.com/sprc940-pru Apr 25 06:46:51 follow the boucing links... Apr 25 06:47:21 http://processors.wiki.ti.com/index.php/Programmable_Realtime_Unit_Subsystem (pointed to by the user manual which consists only of an abstract) Apr 25 06:48:37 lets see...there are some slides Apr 25 06:48:50 it seems this chip would be quite handly for interfacing to the original tuxscreen com board Apr 25 06:50:09 it looks like an extra difficult step to gte accurate timing Apr 25 06:50:24 its really what you probably need though Apr 25 06:50:34 see if ds2 will help out Apr 25 06:51:19 the slides seem to have everything to get going Apr 25 06:51:20 I wonder if it wouldn't be a lot easier for me to just code the cpu intensive stuff on beagle with bcc and then connect 1 atmel chip externally on a io pin to give the exact timing pulses. That would be a fairly easy and very cheap solution Apr 25 06:51:39 It would be very complex to get beagle up and going without an os Apr 25 06:51:47 er, what's bcc? Apr 25 06:51:53 gcc sorry Apr 25 06:52:41 have to realy go now, talk to you later Apr 25 06:52:44 tnx 4 info Apr 25 06:52:48 cya Apr 25 06:58:43 <_tasslehoff_> koen: users are losers. a console-image works fine on my system, so it must be something I've added in my rootfs recipe that makes it hang. Apr 25 08:25:13 hi everybody - I am doing MACHINE=beaglebone ./oebb.sh bitbake systemd-image - And I cannot find sources for beaglebone afterwards... Apr 25 08:25:46 they might get deleted Apr 25 08:25:51 Need to build realtek 8192cu driver... Apr 25 08:26:14 I put this additional think in autoconf ... uhmmm what was it.. just a sec... Apr 25 08:26:58 yes commented: INHERIT += "rm_work" Apr 25 08:27:07 in local.conf Apr 25 08:28:29 I have: /setup-scripts/build/tmp-angstrom_v2012_05-eglibc/work/beaglebone-angstrom-linux-gnueabi/linux-ti33x-psp-3.2-r10d+gitre8004dad869568692ca2a45c04e8464ce48cc4d7, but no sources there? Does anybody know where to look for ? or what to change to get sources ? Apr 25 08:33:23 <_sundar_> av500, any idea on OCF linux? Apr 25 08:34:07 ??? Apr 25 08:34:10 <_sundar_> av500, am trying to get AES 128 CTR mode for TI8168. Apr 25 08:34:31 <_sundar_> the OCF framework doesn't have this mode ported from OpenBSD Apr 25 08:34:42 <_sundar_> but the driver supports that Apr 25 08:35:57 <_sundar_> just trying to figure out how this OCF is getting the supported drivers and supported algorithms from the kernel's crypto API Apr 25 08:48:50 zhivko: go to the angstrom website and read the frontpage Apr 25 08:55:37 hello there. please excuse the following question - it's a little urgent Apr 25 08:55:46 i un-tar-ed the linux root file system i created with buildroot and copied it to the 2nd partition of the flash rom, but the linux kernel complains about not being able to mount the fs. can you give me a quick hint? Apr 25 08:57:08 paste the bootlog Apr 25 08:57:19 using a paste-service, not here Apr 25 08:58:02 and urgent or quick or similiar won't help you here Apr 25 08:58:33 it just might have the oposite result Apr 25 08:59:39 guess i just need some "root=" argument, but where? Apr 25 08:59:52 in uEnv.txt Apr 25 09:00:22 mmcroot= or similiar Apr 25 09:05:13 angstrom (the one distributed with the board) works without it, which is what confuses me. what shall i state as root or mmcroot as long as /dev is unknown to the kernel? (this would be my last question ;-)) Apr 25 09:06:43 the stuff in uEnv.txt is just env-var of u-boot. u-boot changes this to a root=foo in the kernel commandline Apr 25 09:07:17 and your command line should be visible in the bootlogs, therefor I said paste it Apr 25 09:08:29 and there is a default in u-boot, therefor you don't need it. just try printenv in the u-boot cmdline Apr 25 09:08:39 paste early, paste often Apr 25 09:08:50 don't eat the paste Apr 25 09:09:05 especially not the yellow paste Apr 25 09:09:52 hasta pasta Apr 25 09:11:59 I wonder who you can copy something into a ROM Apr 25 09:12:08 it being Read Only Memory and all Apr 25 09:12:22 burn only memory Apr 25 09:12:30 i like pasta Apr 25 09:12:36 have some Apr 25 09:12:36 maybe that's JohnCope's problem, trying to write to ROM Apr 25 09:12:58 it overheats my positronic brain thinking about it Apr 25 09:12:59 no, it's not >:-/ Apr 25 09:13:06 use a white edding paint marker Apr 25 09:13:23 JohnCope: pastebin a boot log Apr 25 09:14:38 <_tasslehoff_> koen: /usr/bin/pvrsrvinit: error while loading shared libraries: libdrm.so.2 <- libgles-omap3 on a console-image Apr 25 09:15:56 thank you, and see you ladder Apr 25 09:16:07 ??? Apr 25 09:16:41 those are the urgent ones Apr 25 09:17:17 but he wanted to see our ladders ! Apr 25 09:17:20 <_tasslehoff_> I see your ladder and raise you one Apr 25 09:17:31 * ogra_ wont show his ladder to anyone !! Apr 25 09:18:09 it's best you don't need it anymore, just be on the top Apr 25 09:18:21 ladder 8 ftw! Apr 25 09:18:49 koen, it overheats my brain to think you have such easy access to a 3d printer that your print pucks for your bone Apr 25 09:19:20 I already though too about buying one ;) Apr 25 09:20:06 but then, I have enough Lego ;) Apr 25 09:20:42 ...but alumide Apr 25 09:21:47 the next board desing should fit into lego grids, the bb doesn't :( Apr 25 09:22:12 http://www.apotome.com/grids.html Apr 25 09:29:43 hm, alumide looks sexy Apr 25 09:30:19 but the price not so "$1.59 per cubic centimeter" Apr 25 09:30:40 (from shapeways.com) Apr 25 09:33:51 Russ: and those pucks took a lot of revisions to print right (really thin walls) and fit into the altoids tin Apr 25 09:38:25 heh, new discovery, if I power cycle my xM for 1 second or less there is a chance the microsd card will come up in a bad state Apr 25 09:38:43 'mmc0: unrecognised SCR structure version 15' Apr 25 09:39:10 update the mmc-firmware, it's just an arm ;) Apr 25 09:39:44 its been screwing up my bisects Apr 25 09:40:21 hmm, an usb-hd would be faster Apr 25 09:40:39 yes, if usb worked reliably across the range of commits I was bisecting Apr 25 09:40:57 aholler: the TI mmc driver is so horrible coded because the kernel people used nfsroot and never really tested it Apr 25 09:41:02 you just need usb in u-boot to load the kernel Apr 25 09:41:29 the kernel is loading via tftp Apr 25 09:41:39 the problem is mounting the rootfs Apr 25 09:41:46 ah dammit Apr 25 09:41:52 hmm, why not nfs? Apr 25 09:41:56 by increasing it to 3 seconds, I've just made it more rare Apr 25 09:41:59 aholler, again, usb Apr 25 09:42:08 oh, sorry, forgot, no eth Apr 25 09:42:35 rootfs over serial ;) Apr 25 09:43:24 Russ: in the pre-xM days I used a zippy board to get working ethernet Apr 25 09:43:33 Russ: you could use a rootfs inside the kernel Apr 25 09:43:43 which lead to a lot of "usb fails when I use the DSP" bugreports Apr 25 09:44:04 should at least be enough to test if the kernel boots Apr 25 09:44:27 I could make you one. Apr 25 09:44:34 I make them here Apr 25 09:45:00 maybe I'll just make the success line 'I want to scream" Apr 25 09:45:41 busybox + 10 lines initscript is usually enough to get a working and usable rootfs Apr 25 09:46:19 I have a small rootfs that chain loads another kernel, initrd, and bootargs over dhcp/tftp Apr 25 09:46:28 for boards that don't have tftp support in u-boot Apr 25 09:46:54 ok, its time for sleep Apr 25 09:47:02 initrd is foo ;) Apr 25 09:47:14 some tests need one Apr 25 10:00:02 After building angstrom for beaglebone, where do I find the patched sources of kernel? Is it build/tmp/sysroots/beaglebone/kernel? Apr 25 10:06:40 running gag #1 Apr 25 10:08:05 cd Apr 25 10:08:19 oops sorry wrong window :D Apr 25 10:16:50 After building angstrom for beaglebone, where do I find the patched sources of kernel? Is it build/tmp/sysroots/beaglebone/kernel? Apr 25 10:44:01 Is StarterWare only be used throught JTAG? Apr 25 10:45:17 <_av500_> no Apr 25 10:45:26 <_av500_> its software Apr 25 10:45:31 <_av500_> it runs on the cpu Apr 25 10:45:35 <_av500_> however it got there Apr 25 10:46:14 so just have to compile program and put it on the SD? Apr 25 10:48:21 <_av500_> maybe you need a way to get it from sd into main memory Apr 25 10:48:34 <_av500_> thats where boot loaders come in handy Apr 25 10:48:35 ok Apr 25 10:49:15 So what the JTAg is it for on the bb xM? Apr 25 10:49:35 On the bb mainpage, they talk about serial Apr 25 10:49:47 its for doing jtag Apr 25 10:49:53 if you need jtag Apr 25 10:49:56 debugging you mean Apr 25 10:50:07 yes, thats one use case Apr 25 10:50:50 but it's useless when you use an OS as Angstrom, right? Apr 25 10:51:04 I have Ångström (2012.02.14) and I'm trying to get the kernel headers to compile a kernel module. Tried installing kernel-headers and kernel-dev, but I still can't find the headers. Tried: "git clone git://github.com/Angstrom-distribution/setup-scripts.git" and "MACHINE=beaglebone ./oebb.sh config beaglebone", but get the error: "No manual entry for git-log. Your installed version of git is too old, it lacks --no-abbrev-commit. Apr 25 10:51:05 not useless Apr 25 10:51:14 but maybe not needed by the average user Apr 25 10:51:35 thomas_fogh: Your installed version of git is too old, it lacks --no-abbrev-commit. Apr 25 10:51:53 av500: I have version 1.7.7 installed Apr 25 10:52:21 because if I want to debug on the beagleboard when I wrote a program, I can just use GDB throught serial or ethernet? Or can I use the JTag to debug it? Apr 25 10:52:31 either one Apr 25 10:52:40 but if you can use gdb, use it Apr 25 10:52:44 much simpler Apr 25 10:54:28 ok, so Jtag give you R/W access to the RAM. In this case, it's possible to upload a program directly into the memory and execute it, without OS/Starterware Apr 25 10:55:13 yes Apr 25 10:56:02 interesting. But no one is using this, isn't it? And nothing is available to do that? Apr 25 10:56:32 (I don't want to do this, but only by curiosity) Apr 25 10:57:02 maybe not no one Apr 25 10:57:05 but not many Apr 25 10:58:05 people like convenience Apr 25 10:59:04 Concerning linux debugging using Jtag, the kernel must give all the information about the processus to the host, like where is the memory stack into the memory Apr 25 10:59:24 Hello! Could anyone suggest me few good resources for a beginner which explains about OS porting, ARM architecture and Beagle board? Apr 25 10:59:50 oragsy: depends Apr 25 10:59:58 what OS do you want to port? Apr 25 11:00:07 Windows Apr 25 11:00:14 Linux* Apr 25 11:00:18 thats done Apr 25 11:00:21 done Apr 25 11:00:27 I actually would like to know the basics of porting Apr 25 11:00:40 if you have to ask, its not for you Apr 25 11:00:57 sorry to be blunt Apr 25 11:02:11 https://www.google.com/search?q=linux+porting+guide Apr 25 11:02:14 From what I know porting involved having a bootloader, linux kernel and applications on the machine/system one would like to have an OS/application running... Is that right? Apr 25 11:03:01 involves* Apr 25 11:03:03 normally apps are not "ported" Apr 25 11:03:28 the whole idea of an OS is to abstract away the hardware Apr 25 11:06:35 After building angstrom for beaglebone, where do I find the patched sources of kernel? Is it build/tmp/sysroots/beaglebone/kernel? Apr 25 11:08:37 <_av500_> no idea, but should be easy to find Apr 25 11:08:45 <_av500_> not a so many places a kernel can hide Apr 25 11:11:10 <_av500_> after reading OE manual, as far as I understood, it should be in build/tmp/work/beaglebone-ang..../linux-ti33x-psp..../ but I do not find it there. Apr 25 11:11:56 <_av500_> did you comment that rm_work stuff? Apr 25 11:12:10 <_av500_> otherwise source are deleted after build Apr 25 11:12:14 <_av500_> OE is very cleanly Apr 25 11:13:42 no did not comment rm_work.. actually still trying to understand the build process.. I should comment that in sources/meta-ti/recipes-kernel/linux/ ?? Apr 25 11:14:22 https://www.google.com/search?q=rm_work Apr 25 11:20:15 Doesn't anyone know where I can get the source for kernel 3.2.5+ ? Can only find 3.2.5 Apr 25 11:21:17 i guess its a patched 3.2.5 Apr 25 11:22:15 the + means there's stuff following the 3.2.5 tag Apr 25 11:24:00 I've installed Ångström (2012.02.14) and run opkg update/upgrade (as recommended). Maybe I shouldn't do the upgrade? Apr 25 11:40:04 prasant: don't comment anything, go to the angstrom website and read the news articles Apr 25 11:40:31 look for "kernel workflow" Apr 25 11:40:58 I know, it's thinking out of the box, looking at the angstrom website for angstrom info Apr 25 11:41:06 that's why I'm a rocket scientist Apr 25 11:44:52 koen: I did read the kernel worklow as well. I had the sources in the tmp/build/angstrom-linux/linux (something) but when I try to build the entire image kernel sources was gone, so I though may be the sources are moved somewhere else (the build system seems so weird that I'm thinking weird now) anyways.. How do I get Uboot sources if I follow kernel workflow? for that I would need to comment out rm_work right? Apr 25 11:45:21 no need to comment anything Apr 25 11:45:33 follow that workflow document Apr 25 11:47:19 koen: There are SD Images released for beaglebone, toolchain releases, why there is no separate git for kernel and uboot? Apr 25 11:48:00 just to make it easier for anyone new to atleast kick start the work :-) Apr 25 11:49:01 because >90% of the people that say they "just want a git tree for the kernel" then come back saying they don't have a toolchain Apr 25 11:49:10 or that their toolchain is broken Apr 25 11:49:15 or ask what crosscompiling is Apr 25 11:49:30 and the kernel isn't developed in git anyway Apr 25 11:50:16 pfft, cross compiling ... real men build natively :)= Apr 25 11:50:31 * koen powers down measurement equipement and goes out to find lunch Apr 25 11:50:37 I'd like to see you build a kernel on a cortex-m3 Apr 25 11:50:49 or an msp430 Apr 25 11:50:50 ogra_: good luck with that m3 firmware Apr 25 11:51:11 just install ubuntu on it :) Apr 25 11:51:13 * ogra_ hides Apr 25 11:51:37 this native building nonsense was started by debian with the notion that it was somehow more pure/free that way Apr 25 11:51:46 nah Apr 25 11:52:12 thats just an excuse for "dpkg cant resolve build deps in a cross way" Apr 25 11:52:46 but I have been told by some debian die-hards that it _is_ more free to build natively Apr 25 11:52:57 nonsense Apr 25 11:53:11 most things debian people say is nonsense to me Apr 25 11:53:24 are you saying there are different kinds of it? Apr 25 11:53:25 and if you look at the work of linaro xdeb is already there ... Apr 25 11:53:47 just needs devs to touch essentially all packages to make them work with it Apr 25 11:53:51 it was not linaro people who told me cross-compiling was unfree Apr 25 11:53:57 You don't see people cross-building their x86 kernels on ARM, do you? Apr 25 11:54:06 Mojito: of course not Apr 25 11:54:09 x86 machines are faster Apr 25 11:54:23 except a few atoms I guess Apr 25 11:54:46 these new calxeda chips should certainly be faster than an atom Apr 25 11:57:07 which makes cross building useless ... Apr 25 11:57:39 a high-end intel is still much faster Apr 25 11:57:40 if you can build on a 2000 CPU calxeda cluster you will likely be faster than cross building on your desktop PC Apr 25 11:58:47 Speed isn't everything. Building on ARM you get all the other ARM benefits too, like less power consumption. Apr 25 11:59:23 are you sure? Apr 25 11:59:23 i doubt mru is intrested in saving power if he wants his binaries quickly available for testing :) Apr 25 11:59:43 does 1 minute on an i7 use more power than 2h on an omap4? Apr 25 11:59:54 I think not Apr 25 12:00:22 the i7 has a TDP of about 100W Apr 25 12:00:32 omap4 at full speed uses a few W Apr 25 12:03:46 What is the difference when executing a bitbake comand directly and bitbake as argument to oebb.sh ? When bitbake is executed through oebb.sh it complains. Apr 25 12:07:32 cross-compiling and quick is just a dream Apr 25 12:07:49 at least if it comes to whole systems Apr 25 12:08:17 building on a fast machine is always faster than building on a slow one Apr 25 12:08:43 just if don't count the countless hours needed to patch various packages Apr 25 12:08:54 and every version of them Apr 25 12:09:04 most patches carry forward easily Apr 25 12:10:05 and when you're building a dozen or so targets, the time spent fixing a package once in a while is quickly recovered Apr 25 12:13:34 and of course when the target is unable to run a compiler, you have no choice Apr 25 12:14:02 really depends ... Apr 25 12:14:09 if you have a similar target that is .... Apr 25 12:14:19 similar? Apr 25 12:14:29 well, close enough Apr 25 12:14:46 either it's the same, or it's different Apr 25 12:14:52 i can build packages for a beagle A1 here on my panda Apr 25 12:15:03 say your target is a cortex-m3 Apr 25 12:15:14 building for that on an armv7-a would still amount to cross-compiling Apr 25 12:15:27 even though the architectures are similar Apr 25 12:15:32 sure Apr 25 12:16:03 depends on the level of similarity i guess :) Apr 25 12:16:04 and then building on a fast x86 instead is no more difficult Apr 25 12:16:21 indeed Apr 25 12:18:39 and once you've set things up for cross-building to one target, building for another is little extra work Apr 25 12:18:58 unless you use debian as a target OS :) Apr 25 12:19:03 but then you're stupid Apr 25 12:19:06 or a derivative Apr 25 12:19:14 or a deviant Apr 25 12:19:24 or paid for it ;) Apr 25 12:19:35 there are things I will not do even for money Apr 25 12:20:10 so the stuff you do for linaro doesnt get built natively in the end ? Apr 25 12:20:40 I don't care what other people do Apr 25 12:29:04 I just want to say that native building on "slow" machines is a crutch Apr 25 12:29:07 that is all Apr 25 12:33:32 Crofton: and often a broken crutch that's been repaired with duct tape and bits of string Apr 25 12:34:12 I wouldn't go that far Apr 25 12:34:12 well, its a good stress test for the HW (and the OS running on it indeed) Apr 25 12:34:30 ogra_: you mean for the swap storage Apr 25 12:34:41 sure, but I think Fedora and Debian are making some really bad decisions because they are limited by rebuild time Apr 25 12:34:49 depends what you build on what HW :) Apr 25 12:35:08 let's say a 140MHz avr32 with 64MB ram Apr 25 12:35:11 its a technical limitation of the package system, not a time issue Apr 25 12:35:14 some idiot built qt4 on that Apr 25 12:35:20 took two weeks Apr 25 12:35:33 I sat in a Fedora ARM discussion at FUDCON Apr 25 12:35:43 building qt4 on my i7 is a matter of minutes Apr 25 12:35:54 it blew my mind they level ofo hoop jumping Apr 25 12:36:02 an ubuntu cycle is 6 months long ... thats enough time to rebuild the 15000 or so source packages in the archive Apr 25 12:36:28 mru: how many packages do you cross-compile regulary? Apr 25 12:36:32 rofl Apr 25 12:36:37 (a debian cycle isnt even defined in its lenght) Apr 25 12:36:54 you talk of months, we talk of hours Apr 25 12:36:58 nuff said Apr 25 12:37:01 :) Apr 25 12:37:10 aholler: most of my dev boards have about 100 packages installed Apr 25 12:37:34 some of us can't discard archs just because our build system is slow Apr 25 12:37:44 note that until shortly debian even suported 68000 CPUs Apr 25 12:38:09 68000 or 68040? Apr 25 12:38:24 68000 has no mmu... Apr 25 12:38:28 68xxx, never touched that port so i have no idea Apr 25 12:38:46 but i think they also had some plain 68000's as builders Apr 25 12:39:25 not possible Apr 25 12:39:29 you cannot run gcc without an mmu Apr 25 12:43:58 is there way to access d-cache / i-cache directly? Apr 25 12:44:50 yes, set the page attributes to cached Apr 25 12:45:04 then any load or store will access the cache Apr 25 12:46:18 give me some more detailed explanation? how can i set the page attr.? Apr 25 12:46:28 what are you really trying to do? Apr 25 12:47:58 av500: concerning our chat we had before, which tools you can use to program by JTAG? Apr 25 12:48:16 sharing data from dsp to gpp using cache mem. Apr 25 12:48:29 i'm using dsplink Apr 25 12:48:55 oh, then you want to use the dsp local memory Apr 25 12:49:39 configure it as plain ram (not cache) on the dsp and map it on the arm Apr 25 12:49:40 correct. as fast as possilble :) Apr 25 12:49:44 the trm has all the details Apr 25 12:51:04 okaay. i will try them. thx mru Apr 25 13:18:24 Crofton|work: ping Apr 25 13:18:34 pong Apr 25 13:18:41 Crofton|work: did you see the dns mail? Apr 25 13:18:44 router upgrade in process though, I may drop Apr 25 13:18:47 yes Apr 25 13:18:52 I jsut sorted out home dsl Apr 25 13:18:57 I will work on it next Apr 25 13:19:03 XorA: is twl6030-pwm.c part of your area of expertise? Apr 25 13:19:07 Crofton|work: thanks! Apr 25 13:19:51 koen: that will never work Apr 25 13:20:15 koen: PWM can be called from atomic and that file does not handle this Apr 25 13:21:35 <_av500_> mru: the dsp cannot even see cpu cache, no? Apr 25 13:21:46 of course not Apr 25 13:21:54 <_av500_> confused people are confused Apr 25 13:22:14 not even the cpu can see the cpu caches directly Apr 25 13:22:20 other than through some debug interfaces Apr 25 13:27:51 <_av500_> but of course, before you can even talk to the dsp, make sure it will be as fast as possible... Apr 25 13:31:41 do you really talk to the dsp? Apr 25 13:37:50 koen downloads is new? Apr 25 13:40:22 <_av500_> mdp: yes Apr 25 13:40:40 <_av500_> I am a dsp whisperer Apr 25 14:10:37 Hey, I'm a bit stuck building compat-wireless on my beagleboard running Ubuntu with kernel version 3.2.7-x5 Apr 25 14:11:40 I tried using the linux-headers-omap package and symlinking it into /lib/modules/3.2.7-x5/build but the build process now fails on gpio-related stuff in compat-wireless Apr 25 14:12:03 does anyone have any experience with this sort of thing, or at the very least kernel headers on the beagleboard? Apr 25 14:15:35 Crofton|work: I'll enable redirects when DNS is live Apr 25 14:16:38 ok Apr 25 14:23:22 Crofton|work: I've pointed my corporate overlord at the donate page Apr 25 14:27:25 we should point all the angstrom users there also Apr 25 14:27:41 check your mail :) Apr 25 14:27:50 it is the easiest route we have to generate goodwill with money :) Apr 25 14:28:00 ah thanks Apr 25 17:48:25 Anyone here used the sgx on their beagleboard with an oe-core build lately? I'm struggling to get it working, and wonder if it is my fault (as usual). Apr 25 17:48:56 there's a heisenbug Apr 25 17:49:45 tasslehoff: this is on oe-core: http://www.youtube.com/watch?v=osrBbfaZN9w Apr 25 17:50:04 tasslehoff: there's a bug in the code that copies over the right libs for the sgx hw Apr 25 17:50:25 tasslehoff: I've been tracking it on the beaglebone and I have a build that works and one that doesnt Apr 25 17:50:30 koen: I daresay that's more than I've achieved :) Apr 25 17:50:39 tasslehoff: I'm up to the point where I am comparing md5sums :( Apr 25 17:50:43 koen: ah. how does it manifest itself? Apr 25 17:50:48 koen: obvious answer build it twice :-D Apr 25 17:52:16 sounds like a parrallel build type bugget Apr 25 17:52:28 koen: I built the angstrom xfce image + some and tried to run the demos. That failed, then I ran the test-app in <..>/ES3.0/ and it said it couldn't find the powervr module or driver. (taking this from memory since I'm at home without My Source. Apr 25 17:53:19 tasslehoff: the funny thing is that during debugging this I fixed a whole lot of other bugs :) Apr 25 17:54:39 koen: good that its not all despair :) Apr 25 17:55:17 just a lot of headdesking involved Apr 25 17:56:37 I did learn a lot about the power class in the kernel today Apr 25 18:00:21 koen: If there are any patches you need tested, let me know. I'm trying to create a recipe for Ogre 1.8 RC1. Apr 25 18:01:07 koen: the heisenbug is present for both beaglebone and beagleboard? Apr 25 18:01:21 <_av500_> and beaglebug Apr 25 18:34:56 makes me beaglebored, so perhaps I should watch champions league instead Apr 25 18:35:03 ugh Apr 25 18:35:58 at least you're not pandaborad Apr 25 18:36:11 beagleborg will help Apr 25 18:37:14 nail down that beagleboard Apr 25 18:37:35 grow a beaglebeard Apr 25 18:37:52 mru: made me chuckle :) Apr 25 18:39:13 oh my, what have I started :) Apr 25 18:44:29 <_av500_> beaglebards Apr 25 18:53:51 Does anyone know how to handle nmea data from a usb dongle? How do you feed it to google maps and other services that use position? Apr 25 18:53:53 beaglehards Apr 25 18:55:00 <_av500_> borillion: android? Apr 25 18:55:21 yes _av500_ Apr 25 18:55:22 <_av500_> there is a place to hook it up in the framework Apr 25 18:55:30 I should have been more specific :P Apr 25 18:55:49 ahh k, will google more then Apr 25 18:55:50 borillion: simple matter of the software asking for the data from the gps Apr 25 18:55:59 <_av500_> djlewis: not so simple Apr 25 18:56:05 why not Apr 25 18:56:06 <_av500_> sw asks the "system" Apr 25 18:56:11 well, Apr 25 18:56:14 <_av500_> and system know the gps data providers Apr 25 18:56:24 <_av500_> borillion: or grep the source code Apr 25 18:56:31 <_av500_> I bet there is a fake gps for the simulator Apr 25 18:56:40 <_av500_> that should point you to the apis Apr 25 18:56:48 there is something called mock location Apr 25 18:57:03 <_av500_> yep Apr 25 18:57:05 * djlewis plugged in a usb gps and terminal displayed nema data Apr 25 18:57:17 <_av500_> djlewis: great :) Apr 25 18:57:23 <_av500_> now make google maps read your terminal Apr 25 18:57:30 <_av500_> maybe with OCR Apr 25 18:57:39 that is a fine task for the borillion\ Apr 25 18:57:44 djlewis, yes you can plug in dongles all you want, but maps and android are looking somewhere else Apr 25 18:58:18 so you need to direct your apps where to look? Apr 25 18:58:19 I see the data in terminal too, I just had to enable ttyACM in kernel Apr 25 18:59:28 <_av500_> borillion: ./hardware/libhardware/include/hardware/gps.h Apr 25 18:59:42 <_av500_> ./hardware/qcom/gps/loc_api/libloc_api/gps.c Apr 25 19:01:19 <_av500_> ./development/pdk/docs/porting/gps.jd Apr 25 19:01:22 heh looks like a start Apr 25 19:01:32 <_av500_> but then, your find is as good as mine Apr 25 19:05:17 oddly ttyACM devices are for modems Apr 25 19:06:09 that's how they started out, yes Apr 25 19:06:24 ttyACM devices behave mostly like serial ports Apr 25 19:12:50 ahh kk Apr 25 19:15:40 that makes sense, I don't like how the usb touchscreen is patched LOL, the manufacture made two chips with the same devID, so the patch someone made spoofs a different devID loads a new driver and the reverses the x & y axis to fix the issue Apr 25 19:16:04 <_av500_> you can reverse that on a higher level too I guess Apr 25 19:16:06 shouldn't one just fix the driver instead of spoofing ID's Apr 25 19:16:47 * borillion cant figure out usbhid to do it :P Apr 25 19:17:47 tasslehoff: it's present for both since it's the same binary drop Apr 25 19:19:50 can anyone recommend a good crash course for reading patch files? I think I know what its saying but not 100% Apr 25 19:21:24 http://pubs.opengroup.org/onlinepubs/9699919799/utilities/diff.html Apr 25 19:22:52 Hello again Apr 25 19:22:53 use something like meld or kdiff? Apr 25 19:23:27 reading diffs is not hard Apr 25 19:23:35 does anyone know if there is by default a gcc compiler active on a beaglebone with the supplied linux distro on the sd card? Apr 25 19:23:41 whats @@ -536,6 +536,7 @@ config TOUCHSCREEN_USB_COMPOSITE ? Apr 25 19:25:19 a change at line 536 in the original file spanning 6 lines, changed to line 536 spanning 7 lines in the output Apr 25 19:25:38 anything after the second @@ is ignored Apr 25 19:26:04 some diff programs put a bit of extra context there for the benefit of the reader Apr 25 19:26:05 ohhhh ok thats easy LOL, Apr 25 19:26:37 koen: is the bug there in 4_03_00_02 as well? Apr 25 19:26:38 a line starting with - is removed, one starting with + is added Apr 25 19:26:46 lines starting with space are unchanged Apr 25 19:27:04 educa: http://26-26-54.hardwarebug.org/30 Apr 25 19:27:06 right, I figured that part out :P Apr 25 19:27:37 koen, that was extremely helpfull, thank you :) Apr 25 19:28:10 maybe I ask this because I'm evaluating if I will buy a beaglebone or not, because I'm looking for hardware to write my soft on to control a cnc lasercutter Apr 25 19:29:13 I need some basic info and answers like RTFM are difficult if you only found a 4500+ page manual of the soc Apr 25 19:29:30 ecuca: most active ppl here are huge trolls, unfortunately Apr 25 19:29:39 trolls = ? Apr 25 19:29:52 sorry, I'm dutch so don't understand all words Apr 25 19:30:08 hm.. not sure how to translate that Apr 25 19:30:31 educa: etters Apr 25 19:30:33 I come from arduino platform (like a lot I guess), but I also know other stuff like atmel avr assembly and Z80 asm(old) Apr 25 19:30:37 ah etters, leukie Apr 25 19:30:41 giving sarcastic answers to what they perceive as dumb questions? Apr 25 19:30:51 well thats the problem Apr 25 19:31:12 if you would ask me something about something I know better then you, I could answer the same way Apr 25 19:31:37 but there is of course always a difference to the regulars and the new ones on a certain chat channel :) Apr 25 19:31:39 and people just stop asking or telling Apr 25 19:31:41 huge trolls: https://lh6.googleusercontent.com/-dRNW7vuYlYs/TzAkOOVf0nI/AAAAAAAAAds/jLfUdRcc-9E/s949/IMG_0104.JPG Apr 25 19:32:07 I am new to beagleboard/bone in that was that I discovered it a few days ago by looking at EMBED and Raspberry Pi Apr 25 19:32:32 do you really want to say now that I should stop talking and go away ? Apr 25 19:32:56 no Apr 25 19:33:02 ah phew :) Apr 25 19:33:02 i think the situation is rather unfortunate Apr 25 19:33:13 if you ask good questions you'll get good answers Apr 25 19:33:25 Is it just an impression or is beaglebone not yet documented very well ? Apr 25 19:33:37 yep.. always someone else's fault Apr 25 19:33:40 yeah, GOOD questions, ok I'll try to do another one Apr 25 19:33:44 if you ask questions where you could find the answer yourself in less than 30s by google or experiment, you might get silly answers Apr 25 19:33:57 well trust me, I googled Apr 25 19:34:22 but somehow its not nice to be told such stuff. I'm not a 14yo kid wanting to code the next superthing Apr 25 19:34:36 ssh to board, run gcc and see if it exists would be a simple experiment Apr 25 19:34:51 if you have one it is very simple mru Apr 25 19:35:05 I look if the board can suit my needs before buying Apr 25 19:35:15 you can always install it Apr 25 19:35:29 educa you can install its easy :P Apr 25 19:35:34 most of the time Apr 25 19:37:03 Lets say that I am looking for hardware to control the steering of 3 stepper motors (max 2 simultateously moving). For that I need to be able to set pins hi/lo and read pins hi/lo + it would be nice to at least have 1 PWM output on the system. I have the impression that all of that is fairly easy on the beagle since pins are threated like files in some way, but my biggest concern is about timing. I need to set pins and then be sure th Apr 25 19:37:09 for for example 13 microseconds Apr 25 19:37:19 and that 13 should be very accurate Apr 25 19:37:27 + 13 is variable of course, its just an example Apr 25 19:38:03 fortunately, my program may be totally halted during that delay. Nothing else has to run while waiting so its 1 thread Apr 25 19:38:15 :-s is lagging or something Apr 25 19:38:15 therefore I think simple gnu c is enough to handle this Apr 25 19:39:09 between you program and the hw is an os called linux Apr 25 19:39:28 aholler, and linux is not a realtime os Apr 25 19:41:06 sorry to interject, "real time kernel" :-? Apr 25 19:41:10 am I right to suppose that this accurate delay is my biggest problem on such platforms ? Apr 25 19:41:30 yeah sorry borillion, you are right Apr 25 19:42:03 there's a stepper cape coming up with an msp430 driving the stepper controllers Apr 25 19:42:06 educa there's no guarantees on interrupt latency Apr 25 19:42:12 it's as hard as you want to make it :) Apr 25 19:42:33 koen, info available aboud that ? Apr 25 19:43:33 steppers are faitly easy, you just need to be able to fire the steps in an accurate timing Apr 25 19:43:52 faitly=fairly Apr 25 19:44:01 * borillion makes things harder than they need be Apr 25 19:44:39 hehe Apr 25 19:47:56 question is are you going to see a delay thats great enough to screw things up? I've never worked with steppers or controllers for them but Id imagine that delay is not so much of a worry. If you where not seeing something with the arduino(sp?) Apr 25 19:48:33 then again its not running a whole os on it Apr 25 19:48:46 the fastest speed at which I will have to give pulses is 50 000 hertz Apr 25 19:48:56 on an arduino you use the timer interrupt of a hardware timer Apr 25 19:50:39 try to write a small linux driver on your pc Apr 25 19:50:56 than decide if you want to get friend with linux Apr 25 19:53:06 here you can find some stuff to read in regard to rt: https://www.osadl.org/Single-View.111+M59e3481cdfe.0.html Apr 25 19:53:50 educa, there are also others doing what your plan : http://webcache.googleusercontent.com/search?q=cache:qTg-zY_cNpIJ:e2e.ti.com/thread/655125.aspx+&cd=5&hl=en&ct=clnk&gl=us Apr 25 19:55:26 educa also here https://groups.google.com/forum/#!topic/beagleboard/NIAMcPbaGuY Apr 25 19:58:24 thanks, that second one I allready read before Apr 25 20:07:16 just fyi for anyone interested, we have a new openocd-support mailing list started to deal with user(not dev) issues with openocd - http://groups.google.com/group/openocd-support Apr 25 22:15:11 hey there ... i updated my new beagle bone and it broke in the middle ... i am trying to restore the image cannot mount my sd card from my dev box please could i get some assistance Apr 25 22:18:55 green_ are you using linux to mount your sd card ? Apr 25 22:26:03 yes i am using ubuntu Apr 25 22:30:44 i am using linux ... i cannot see my device as a sd drive in my dev box ... i broke it in the middle of an update and trying to restore image ... Apr 25 22:31:58 dang Apr 25 22:32:11 dosen't show up in /media or /dev Apr 25 22:32:13 either? Apr 25 22:32:22 either ... Apr 25 22:32:35 sounds like your foobard it Apr 25 22:32:35 it was .. until i did an upgrade ... Apr 25 22:33:03 I was going to say if maybe it showed up in /dev you could fsck it Apr 25 22:33:04 370418.996034] usb 1-5.2: device not accepting address 90, error -110 [370419.069337] usb 1-5.2: new high speed USB device using ehci_hcd and address 91 Apr 25 22:34:54 how could i confirm it is shows up in dev .... i see the /dev/sd? but they look like files not devices Apr 25 22:35:29 i cannot connect to it using the serial connection though ... and i get a console. Apr 25 22:36:00 meant ... i can connect use screen and get a console. Apr 25 22:44:00 I would get a list of /dev through the terminal, plugin the sd card, then get the list of /dev Apr 25 22:44:03 then compare Apr 25 22:44:09 then pick the difference Apr 25 22:45:48 diff /tmp/x.txt /tmp/y.txt 180c180,183 < drwxr-xr-x 2 root root 3360 2012-04-25 15:44 char --- > crw-rw---- 1 root dialout 188, 1 2012-04-25 15:45 ttyUSB1 > drwxr-xr-x 4 root root 80 2012-04-25 15:45 serial > crw-rw---- 1 root dialout 188, 0 2012-04-25 15:45 ttyUSB0 > drwxr-xr-x 2 root root 3440 2012-04-25 15:45 char Apr 25 22:46:47 sorry the output is ugly ... only ttyUSB changes ... not the disk stuff. Apr 25 23:02:47 I am updating the kernel for a derivative of the beagleboard and am trying to add features to a device driver. I have modified the defconfig files in both the meta-ti subdirectory for my board and in the build tree. When I rebuild the kernel and put the it on the board, my changes do not show up. It seems to me that either I am modifying files that are not really used or am not able to get bitbake to notice the changes. Apr 25 23:03:31 * mru prefers to develop kernels outside any build framework Apr 25 23:15:17 Hi, I have a weird issue with beaglebones Apr 25 23:15:35 I have 16 boards Apr 25 23:15:57 There are similar issues posted to the google group but no obvious answer was posted. I also looked for specific recipes for the driver but didn't find any. Apr 25 23:16:12 abelloni: build a skeleton Apr 25 23:16:25 and while booting the same kernel on each of those, I get 5 that kernel oops Apr 25 23:16:38 just after mounting rootfs Apr 25 23:17:24 same sd card? Apr 25 23:17:26 I tried kernels from Angstrom and from arago Apr 25 23:17:56 actually, same sd card but I'm only using MLO and u-boot on it Apr 25 23:18:14 I'm using tftp for the kernel and NFS as rootfs Apr 25 23:18:25 they are powered using usb Apr 25 23:18:53 which board revision? Apr 25 23:18:57 A5 Apr 25 23:19:27 though, some have the sdcard slot one way and others upside down Apr 25 23:19:37 but all are labeled A5 Apr 25 23:20:00 the issue seems a bit random too Apr 25 23:20:13 sometimes, it is illegal instruction #0 Apr 25 23:20:24 sometimes it is NULL pointer dereference Apr 25 23:21:09 I'm using the defconfig from angstrom Apr 25 23:22:14 and it's consistently happening on a few of the boards, never on the others? Apr 25 23:23:42 I would say yes Apr 25 23:25:20 would that be a power issue ? Apr 25 23:31:08 maybe an issue with the cortexM3 firmware ? Apr 26 00:00:36 does anyone know how to use uart1 through the sys filesystem? Apr 26 00:01:06 what do you want to do? Apr 26 00:01:39 use uart1 to recieve data from a cpld Apr 26 00:01:55 why not use /dev/ttyO0 or whatever it maps to? Apr 26 00:02:09 that's how you normally use a uart Apr 26 00:05:37 i've never done it before Apr 26 00:05:58 do you just pipe to dev/ttyO0 to write? Apr 26 00:06:02 and cat to read? Apr 26 00:06:21 you need to configure the right parameters first Apr 26 00:06:24 like bitrate Apr 26 00:06:39 how do you do that? Apr 26 00:07:44 stty Apr 26 00:07:55 if you want a command line tool Apr 26 00:08:13 man termios for the C interface Apr 26 00:08:36 thank you Apr 26 00:26:03 these xbee units work amazingly well Apr 26 00:49:46 Anyone feel like answering a newbie git question? Apr 26 00:49:54 shoot Apr 26 00:49:55 How would I grab this tree? http://dominion.thruhere.net/git/cgit.cgi/linux-omap/tree/?h=nishant-github/linux-omap-3.0 Apr 26 00:50:36 click the summary tab Apr 26 00:50:42 there's a clone url at the bottom of the page Apr 26 00:51:21 um um, where exactly? Apr 26 00:51:43 * mranostay hands mru a beer Apr 26 00:51:45 right under the 'Clone' heading Apr 26 00:52:42 Ah, under the "summary" Apr 26 00:52:44 Cheers! Apr 26 00:52:49 ppotera: http://bit.ly/ISF042 Apr 26 01:57:06 Hello Apr 26 01:57:50 Does anyone know if the beagloboard-XM supports UHS-I or UHS-II SD cards? Apr 26 01:58:34 I am trying to improve performances, but the only info I've found on the net is that "apparently the SD card reader supports 4 bits" Apr 26 02:38:56 uhs-1 speed is not supported by the H/W (in my case, omap4 hsmmc comtrller) Apr 26 02:40:57 omap4 hsmmc controller support up to 24MB/s in high speed sd Apr 26 02:42:37 vanity: techincally the pll will go up to around 68MHz, but that feature set requires a lot of careful programming and lots of planning for your sd signals Apr 26 02:52:49 i didn't think that far. hmm.. Can hsmmc controller support 68Mhz in 4-bit h/w interface? **** ENDING LOGGING AT Thu Apr 26 03:00:00 2012