**** BEGIN LOGGING AT Wed Apr 24 02:59:58 2013 Apr 24 03:16:41 Hi, is anyone around to talk about gsoc ideas? :D Apr 24 04:22:27 zamN, probably Apr 24 04:24:57 i am computer that has gained sentience Apr 24 04:27:37 what is 26+26? Apr 24 04:29:08 dear sir and/or madame it is 52 Apr 24 04:29:14 * BigSwole explodes Apr 24 04:29:21 lol Apr 24 04:29:30 should i directly message a mentor then? Apr 24 04:30:06 zamN, are you interested in a particular idea...or do you have something else in mind? Apr 24 04:31:24 I'm interested in a couple of the ideas listed on the ideas page but am not sure if I have the technical abilities for them. That is why I wanted to talk to the mentors for each one Apr 24 04:31:35 which ones? Apr 24 04:31:47 Library of Arduino-compatible functions for Linux Apr 24 04:31:58 actually its 54 Apr 24 04:32:09 Russ, he failed miserably Apr 24 04:32:22 clearly Apr 24 04:32:42 and Android-based boot host Apr 24 04:33:03 zamN, neither of those is me...so I would refer you to the Mailing List... Apr 24 04:33:26 Russ, I think I wouldn't even put him in the Graphing Calculator category... Apr 24 04:33:29 heh, yeah Apr 24 04:33:32 or try ping'ing in the beagle-gsoc channel Apr 24 04:34:00 what mailing list should i post to ka6sox? And wouldn't it be better to just wait for them on irc rather than send something to a mailing list and thus bothering unecessary people with my problems (heh)? Apr 24 04:34:18 zamN, what Timezone are you in? Apr 24 04:34:20 est Apr 24 04:34:40 then tomorrow morning your time would be a good time to ping them Apr 24 04:34:46 okay, thanks ka6sox Apr 24 04:34:52 in #beagle-gsoc or here. Apr 24 04:34:55 morning? Apr 24 04:35:02 moaning Apr 24 04:35:05 isn't afternoon more like it? Apr 24 04:35:10 mourning Apr 24 04:35:17 * mranostay dances in Apr 24 04:35:21 ds2, okay afternoon Apr 24 04:35:29 mranostay: when you headed to PDX? Apr 24 04:35:56 ka6sox: some of them are at PDT at the moment Apr 24 04:36:07 well depends on various factors but no later than May 20th :P Apr 24 04:36:22 ds2, at or *in* Apr 24 04:36:35 this may be the lightest move i've done yet Apr 24 04:36:37 was not aware of there being a beagle gsoc channel o_O lol Apr 24 04:36:58 zamN, its new...and here has more people to interact with. Apr 24 04:37:00 ka6sox: what's the difference? Apr 24 04:37:07 both I'd hope Apr 24 04:37:26 Pacific Dance Theatre? Apr 24 04:37:42 * mranostay does ballet for ka6sox Apr 24 04:37:56 *facepalm* Apr 24 04:38:03 mranostay: cost of living there will be a lot more reasonable I'd guess. Apr 24 04:38:15 this Hoegaarden is damn tasty Apr 24 04:38:20 emeb_mac: well yes Apr 24 04:38:23 mranostay: still doing x86 up there? Apr 24 04:38:34 ssssh :) Apr 24 04:38:41 whats x86? Apr 24 04:38:42 mranostay: mmm Hoegaarden. Yumm Apr 24 04:39:07 clearly I'm missing something Apr 24 04:39:43 white or wheat? Apr 24 04:40:05 (and I don't mean bread) Apr 24 04:41:10 wheat Apr 24 04:42:08 I guess you *must* use the correct glass... Apr 24 04:42:38 i'm not Apr 24 04:42:47 * mranostay is such a rebel Apr 24 04:43:04 *must* be hexagonal Apr 24 04:45:55 I don't wanna go... Apr 24 04:46:08 don't want to participate in Security Theatre again :P Apr 24 04:46:24 it's such fun! Apr 24 04:46:40 I flew for the first time in a few years a couple weeks ago. Apr 24 04:47:02 wasn't too bad - the new mmwave scanners were pretty simple. Apr 24 04:47:49 I wonder how would carbon fibre interact with them Apr 24 04:48:01 no, this is worse...2 weeks ago I could go unescorted..now I need an escort :P Apr 24 04:49:42 the bloody dish is 30m in diameter..I'm not gonna walk away with it :p Apr 24 04:50:59 ka6sox: your pockets do look pretty big.... Apr 24 04:52:17 tardis pockets Apr 24 04:52:25 https://specialcomp.com/beaglebone/ <--- is this real.. $39.00? Apr 24 04:52:47 stripped down Apr 24 04:55:55 more like a greyhound or a whippet Apr 24 04:56:33 last Hoegaarden Apr 24 04:56:36 * mranostay cries Apr 24 04:58:32 how many? Apr 24 05:00:08 isn't the A5 rev the one with the mislabled processor? Apr 24 05:00:19 thurbad: A4 Apr 24 05:00:36 emeb_mac: by the empty bottles the count would be 4 Apr 24 05:08:35 beagbone fuchsia! Apr 24 05:10:06 does this mean I can say the color now? Apr 24 05:11:21 a wonderfull beaglebone JIHAD everyone! Apr 24 05:11:40 damn timezones Apr 24 05:12:03 av500: make me breakfast ;) Apr 24 05:17:40 mranostay: bogarts all the Hoegaardens Apr 24 05:18:13 i bought them Apr 24 05:18:16 so yes Apr 24 05:19:02 we need more beaglebone pinks Apr 24 05:19:20 ds2, you can get the soldermask in any color you want Apr 24 05:19:24 as long as it's black Apr 24 05:19:37 Russ: you can get pink Apr 24 05:19:49 take a current beaglebone... a knife and.... Apr 24 05:19:55 blood + white == pink Apr 24 05:20:16 pretty sure you'd need a beaglebone white for that Apr 24 05:21:19 Make your own. Apr 24 05:21:19 But even at 1000 parts Apr 24 05:21:19 i cant' buy the BOM for $39! Apr 24 05:21:29 Horrah, for the Raspberry PI Apr 24 05:21:46 they are not making it at 1000 pieces Apr 24 05:21:59 no, thats for sure. Apr 24 05:22:08 Hate the PI, its applied some compwetitive pressure for sure Apr 24 05:22:17 the bbb is a pi eater Apr 24 05:22:49 the Pi is a still born doorstop Apr 24 05:43:40 ewww Apr 24 05:52:47 'morning Apr 24 05:53:52 gm koen Apr 24 06:09:35 SPI data communication in android ICS, beaglebone Apr 24 06:09:38 ? Apr 24 06:11:49 &awaz Apr 24 06:15:58 babudec04 android uses linuxkernel so spi communication is the same Apr 24 06:29:37 SPI data communication in android ICS, beaglebone ? how to configure ? Apr 24 06:41:07 morning Apr 24 06:50:12 morning Apr 24 06:55:37 good moaning Apr 24 06:56:42 hi panto Apr 24 06:56:48 hi woglinde Apr 24 07:09:58 moaning panto Apr 24 07:10:19 hi ka6sox Apr 24 07:11:41 is the entire ram reset during reboots Apr 24 07:12:16 can we store something in the ram which can be accessed in the next subsequent reboot? Apr 24 07:12:17 * panto looks for a reset pin in his DRAM datasheets Apr 24 07:12:29 nope Apr 24 07:12:44 now getting your bootloader+os combo to not step on it is another matter Apr 24 07:13:37 exactly! that was my question. Apr 24 07:14:16 so it can be done Apr 24 07:14:40 what is the RTC story on BBB? is there one? the datasheet suggests there is one built into the chip Apr 24 07:15:10 there is one, but there's no battery feeding it Apr 24 07:19:47 panto: does this involve hacking both the x-loader and the kernel Apr 24 07:20:02 there's no x-loader Apr 24 07:20:04 there's u-boot Apr 24 07:22:35 are you referring to the address where u-boot resides or u-boot making the modifications to the location where this "address" that needs to be protected Apr 24 07:23:52 * av500 stirs Apr 24 07:24:23 hey av500 Apr 24 07:24:40 crashovrd: there is, you need battery for it Apr 24 07:24:57 crashovrd: and i believe that there is no way to install battery on BBB Apr 24 07:25:03 nope Apr 24 07:25:20 people use an i2c rtc module usually Apr 24 07:25:31 never buy first release hardware! Apr 24 07:25:43 crashovrd: first release had a way to install battery Apr 24 07:25:50 so there is RTC module on the chip, but there is no way to use it Apr 24 07:25:50 * crashovrd twitters, facebooks and change.org petitions for a battery leads Apr 24 07:25:50 heh Apr 24 07:26:19 also hold your breath Apr 24 07:26:34 RTCs are obsolete Apr 24 07:26:36 use NTP Apr 24 07:26:52 well, it is fed by the power rail that supplies the standby voltages Apr 24 07:26:59 oh! lets have the date/time at kernel boot argument again Apr 24 07:27:14 so if you have an external battery cape and suspend it should work Apr 24 07:27:56 so the drivers are using it? it just needs some power (like suspend) to work? at next boot it will know the date/time? Apr 24 07:28:17 yes Apr 24 07:28:29 you can check it yourself Apr 24 07:28:36 no i cant Apr 24 07:28:38 set the hwclock, and reboot Apr 24 07:28:40 i dont have one Apr 24 07:28:42 :P Apr 24 07:28:43 lol Apr 24 07:39:51 As a faithful member of the cargo cult, I have solved network issues by increasing vm.min_free_kbytes. Then after reading the errata for the OTG-chip, I disabled dma on musb. Now I wonder if I can turn vm.min_free_kbytes down again. Can't remember if the issues were related. Apr 24 07:40:03 I'm saying this out loud in case someone knows :) Apr 24 07:42:44 try it out Apr 24 07:43:55 tasslehoff: first rule of the cargo cult.... Apr 24 07:45:33 vmfree bytes is for ehci, not musb Apr 24 07:46:53 Hey! Does the Beaglebone Black have any kind of display output? Apr 24 07:47:19 hdmi Apr 24 07:47:27 and leds Apr 24 07:47:33 koen: not in my case I think. we have the smsc95xx (and only that) on musb Apr 24 07:47:58 koen: we call that meta-cargo culting Apr 24 07:48:03 tasslehoff, why? Apr 24 07:50:08 woglinde: the thing is that 1 week before we're supposed to release we "suddenly" experience oom-killers. I'm decreasing vm.min_free_kbytes again to free memory. Since we have limited test-time left I thought I'd consult you guys as well :) Apr 24 07:52:49 good morning, "can i haz beaglebone black now"-lemmings and "i'm too lazy to google"rs Apr 24 07:53:49 panto: we use ehci for another device that needs more power than musb can provide, and use musb+smsc95xx on board to provide networking. (not sure if that answers your Q). Apr 24 07:56:31 tasslehoff I do not think someonehere has made some stresstests for software over a couple of days Apr 24 07:57:36 or stresstests with usb involved Apr 24 07:58:01 stress test with musb... Apr 24 08:00:17 panto minutes hours or days if it fails Apr 24 08:01:17 * panto looks at his magic 8 ball Apr 24 08:01:24 'results may vary' Apr 24 08:05:12 woglinde: oh we have stresstests and stuff, but I just thought it worth checking if someone *knew* something about it. I hoped maybe some factual knowledge had surfaced since the last times this pained me. Apr 24 08:05:32 * tasslehoff wonders if 'factual knowledge' is a valid expression Apr 24 08:05:59 yes Apr 24 08:06:07 you can knowledge of a rumor Apr 24 08:07:44 and u can have jumbo shrimp Apr 24 08:17:45 hi all, how is GPU memory controlled on the BeagleBone Black? Apr 24 08:21:02 hi all, just read about the beagleboard black and it is an impressive piece of kit for that price. I have two projects in mind and I am in need of some advice :) Apr 24 08:21:51 first project is home automation, second is a lan connected media-center (no local storage apart from os) Apr 24 08:23:31 my questions: 1) is it 'easy' to connect extra hardware (like arduino) to the bb-black? 2) is the bb-black powerfull enough to do full-hd video playback? Apr 24 08:23:58 1) it is easy, you might now even need an arduino - the bone is much faster Apr 24 08:24:02 2) full-hd video no Apr 24 08:24:11 maybe 720p Apr 24 08:24:26 heh Apr 24 08:24:30 maybe D1 Apr 24 08:24:32 not 720p Apr 24 08:24:41 I have faith in MRU Apr 24 08:24:45 ok, so the media-center project is better of with something else powering it Apr 24 08:24:50 yes Apr 24 08:25:03 the original beagle could do 720p because it had hw video overlays Apr 24 08:25:09 the bone lacks that Apr 24 08:25:22 and no 3d support for black, so you can't use shaders Apr 24 08:25:44 it's s/w - you can do miracles just with s/w as long as the h/w stays constant and people learn it's in and outs :) Apr 24 08:25:49 to my knowledge most low-power systems that do video playback rely on special hw to do it Apr 24 08:26:02 yes, they do Apr 24 08:26:02 that works ok until we get a new/better codec Apr 24 08:26:15 koen, and 3d support might come Apr 24 08:26:17 for a hd playback thing get an rpi and install raspbmc Apr 24 08:26:18 at some point Apr 24 08:26:31 probably the only sane use for an rpi :) Apr 24 08:26:38 also considering the rpi Apr 24 08:26:38 yep Apr 24 08:26:40 panto galluim driver? Apr 24 08:26:42 how is GPU memory handled, if running headless is all 512 MB RAM available to the CPU? Apr 24 08:26:49 you said sane and rpi in the same breath? Apr 24 08:26:49 yes Apr 24 08:26:53 IRQ: yes Apr 24 08:27:09 woglinde, there's a closed source binary blob Apr 24 08:27:09 thx Apr 24 08:27:12 ka6sox: it's an STB SoC, so using it as an STB is well, sane :) Apr 24 08:27:25 koen, +1 Apr 24 08:27:35 they might at some point make it work Apr 24 08:27:36 +11 Apr 24 08:27:45 panto I did not look up what the bone has at all for gpu Apr 24 08:28:04 I don't care about the gpu all that much tbh Apr 24 08:28:24 haha so why talking about 3d Apr 24 08:28:29 until now i used a normal pc as mediacenter, nice and flexible but at a price (and lots of BTU) Apr 24 08:28:48 woglinde, shaders can accelerate video Apr 24 08:29:21 so there is no PowerVR SGX on BBB? Apr 24 08:29:27 there is Apr 24 08:29:36 the 46 pin headers are used to connect to other hw i assume? Apr 24 08:29:44 but no drivers to use it? Apr 24 08:29:51 but getting it to work on a very new kernel is another matter Apr 24 08:29:52 or are 'the pins not connected' too? Apr 24 08:29:59 panto there is much much todo even vaapi do not get it right Apr 24 08:30:00 nico_, yes Apr 24 08:30:11 woglinde, yep Apr 24 08:30:49 I tought imagination is in the process to open some sgx stuff Apr 24 08:30:52 panto , developing my own hardware is a bit new for me, any good starting points? Apr 24 08:30:54 imo all that people would want is to run parts of libav/ffmpeg/gstreamer in the different cores that are available Apr 24 08:30:56 we will see Apr 24 08:31:18 nico_, I'm a s/w guy really Apr 24 08:31:21 maybee tegra and mali will put some more pressuare on it Apr 24 08:31:27 you don't want me to tell you how to do that Apr 24 08:31:32 :D Apr 24 08:32:39 panto, same here... programming np, hardware thats new but it seems nice to build something physical for a change Apr 24 08:32:51 and have it do something else than produce smoke and bad smell Apr 24 08:33:00 heh Apr 24 08:33:42 woglinde, from the parts of the glue code I've seen of the SGX sdk, just shame is enough for them not to do it Apr 24 08:33:50 speaking of that... Apr 24 08:34:01 thanks for the advice, i'm off reading more on the bb-black Apr 24 08:34:07 * ka6sox starts a new pull and tests to see if panto fixed the Blue Smoke issue. Apr 24 08:36:20 ka6sox, you'll need a magic cmdline to keep emmc at reset Apr 24 08:37:16 okay...what is the incantation? Apr 24 08:37:18 :) Apr 24 08:37:32 capemgr.extra_override=BB-BONE-RST2 Apr 24 08:37:59 that will load a test cape that will grab and hold the emmc reset line low Apr 24 08:38:17 ty Apr 24 08:38:46 when you have a real cape that uses the mmc1 pins, it will automatically do it Apr 24 08:38:56 make sure you don't boot from it :) Apr 24 08:41:28 blue smoke problem? Apr 24 08:41:38 sounds like a serious problem Apr 24 08:41:56 not a serial problem Apr 24 08:42:24 * letothe2nd likes electric smoke, though. Apr 24 08:42:33 ewww Apr 24 08:42:35 * woglinde wonders that koth did not catch the discussion Apr 24 08:59:52 [kernel] koenkooi pushed 2 new commits to 3.8: http://git.io/NAXNTg Apr 24 08:59:52 kernel/3.8 38a64b0 Koen Kooi: 3.8: fix interrupt latency on am335x... Apr 24 08:59:52 kernel/3.8 91a013c Koen Kooi: 3.8: update GPMC patches... Apr 24 09:04:29 woglinde: i read once in a while about problems with emmc and stuff about smoking... Apr 24 09:04:37 woglinde: but i didnt think it was a software problem Apr 24 09:04:59 woglinde: smoking electronics is most often a problem of not well designed hardware Apr 24 09:05:37 and, quite frankly, i didnt care much about it, so didnt follow it either :) Apr 24 09:06:07 * KotH is more into laser swords and photon torpedos Apr 24 09:14:43 hallo,sorry for my bad english, but has someone a link to the original sd-card Image for BeagleBoardXM rev. B, thanks Apr 24 10:08:53 http://pastebin.com/qn812tEb Apr 24 10:08:54 sigh Apr 24 10:12:51 cookie? Apr 24 10:13:00 might help Apr 24 10:15:18 panto: just map page 0 :) Apr 24 10:15:30 lol Apr 24 10:15:52 didn't AIX used to do that? Apr 24 10:16:01 or was it a different unix Apr 24 10:25:25 anyone here ever tried rust (the programming lang) ? Apr 24 10:27:44 KotH: I had a look but it seemed a bit scary Apr 24 10:28:11 jackmitchell: why? Apr 24 10:28:20 KotH: it wasn't C ;) Apr 24 10:28:23 *g* Apr 24 10:28:36 I intent to have a go at some point Apr 24 10:29:44 i wanted to have a look at it for quite some time now, but somehow never got around Apr 24 10:29:51 now i'm reading the lwn article on it from last week Apr 24 10:29:53 yeah, same Apr 24 10:30:19 I've also beeen meaning to have a plat with google go Apr 24 10:30:24 s/plat/play Apr 24 10:31:15 yeah.. that too.. after i had a look at smalltalk :) Apr 24 10:31:38 bah Apr 24 10:31:41 away Apr 24 10:32:04 panto: enjoy your lunch Apr 24 10:32:44 jackmitchell: when go came out, there was a blog post somewhere that compared go to some old language... this language had all the features of go plus some other stuff... that language was smalltalk Apr 24 10:33:43 smalltalk is nice Apr 24 10:34:13 by the water cooler Apr 24 10:36:21 i rather do that at dinner :) Apr 24 10:38:08 funny facts on smalltalk is that it contains objects created 30 years ago and still living Apr 24 10:45:20 message passing makes me a sad panda Apr 24 10:46:29 crashovrd why? Apr 24 10:47:06 dynamic languages and such move errors from compile to runtime Apr 24 10:47:49 if you think vtables make C++ slow, try calling a function to call a function Apr 24 10:47:54 but you can catch programming errors at runtime and handle them ;) Apr 24 10:48:32 does anybody know what speed the PRU runs at on the beaglebone black? is it CPU/2 or 500mhz? Apr 24 10:48:37 200mhz Apr 24 10:49:00 it didn't get a speed boost with the rest of the chip? Apr 24 10:49:13 its faster downhill Apr 24 10:49:22 minix for the PRU any time soon? Apr 24 10:49:31 keso_: a faster PRU doesnt give you much Apr 24 10:49:56 a faster PRU would let me control more WS2811 LED strips off a single board ;) Apr 24 10:50:07 more? Apr 24 10:50:14 how many do you control? Apr 24 10:50:17 keso_: the speed limit is defined by the I/O pins and creating single ended I/O pins that go over a couple 100MHz is a) difficult b) an EMI nightmare Apr 24 10:50:36 huh, ok Apr 24 10:50:59 keso_: use WS2801 :) Apr 24 10:51:09 then you can bitbang it Apr 24 10:51:25 WS2801 is low density for what I'm looking for... I'm looking at the WS2811-embedded onboard the 5050 LEDs Apr 24 10:51:36 I know Apr 24 10:51:37 keso_: for reference: with the PRU you can toggle the pins at 100MHz, which gives you a spectrum with signal components well into the GHz range Apr 24 10:51:54 actually I only need to drive a 800khz signal to control a strip... but would like to control multiple strips in parallel off a single PRU Apr 24 10:52:04 so use several pins Apr 24 10:52:07 there are a lot Apr 24 10:52:19 why do you need a faster pru for that? Apr 24 10:52:27 DDR :) Apr 24 10:52:28 yeah, that's the idea, but a faster PRU lets me juggle more pins Apr 24 10:52:38 o_0 Apr 24 10:52:58 if 200MHz is not enough to toggle 16 pins with 800kHz, then you are doing it wrong... very wrong Apr 24 10:52:58 if you get them in the same register bank then you control them all in parallel for free, no? Apr 24 10:53:09 i may be talking out my ass.. i dunno, I still haven't grok'd the ws2811 signalling. It's weird. Apr 24 10:53:25 anyways.. time for some kebab! Apr 24 10:53:57 keso_: I am driving something similar to WS2801 from an MSP430 Apr 24 10:54:01 er, 2811 Apr 24 10:54:04 at 16mhz Apr 24 10:54:29 so I am sure a 200mhz PRU can do it Apr 24 10:55:15 same here: http://bleaklow.com/2012/12/02/driving_the_ws2811_at_800khz_with_a_16mhz_avr.html Apr 24 10:55:55 oh, i'm sure it can too. what I want to do is control thousands of pixels off a single board tho. at 800khz signalling, that limits you to a max of around 1000 pixels in a single string @30fps. so you need parallel strings Apr 24 10:56:05 i want to basically have a memory mapped frame buffer of LED pixels Apr 24 10:56:26 if you do, please make it a linux framebuffer driver :) Apr 24 10:56:39 :S Apr 24 10:57:04 keso_: but you get parallel for free if you use the GPIOs for one register, no? Apr 24 10:57:16 a GPIO "bank" Apr 24 10:57:40 just prepare the needed bit pattern in advance and write it out Apr 24 10:58:18 yes, needs more memory Apr 24 10:58:24 yeah but i'm not sure you can do it with the weird signaling. It's not like a synchronous serial or anything.. you have to do strange bit transitions Apr 24 10:58:41 I know Apr 24 10:58:51 but the time for 1 bit is constant, no? Apr 24 10:58:56 if i prepare the bit pattern in advance yeah, might be able to do that. Good idea. Would take lots more memory but the beaglebone has it Apr 24 10:59:52 it's not really constant, no. at least I don't think it is. a bit frame is constant but where it transitions mid-frame to signal either a 1 or 0 is different Apr 24 10:59:55 defio Apr 24 10:59:58 the chain I have has 0 = low for 250ns and 1 = low for 3x 250ns Apr 24 11:00:13 and contant bit time Apr 24 11:00:32 so I have 4 slots, either send 0111 or 0001 Apr 24 11:00:45 ws2811 should be similar Apr 24 11:00:52 so yes, 4.x the mem needed Apr 24 11:00:55 4x Apr 24 11:00:56 yeah, that might work Apr 24 11:01:11 but there the PRU is only pumping out data, nothing else Apr 24 11:01:51 then use the 2nd PRU to read from mem and do the bit pattern Apr 24 11:03:32 if i can, i want to use the second PRU to drive even more strips! :D Apr 24 11:03:41 or that Apr 24 11:03:49 then prepare the patterns from the ARM Apr 24 11:11:38 to get audio out of beagle NG do I need a cape or do I connect audio jack to header pins? Apr 24 11:11:54 HDMI has audio Apr 24 11:12:02 if you need analog audio you have to get a cape Apr 24 11:12:18 or a metal plate in your head Apr 24 11:12:22 and the right PRU SW Apr 24 11:15:42 keso_: if you look at the 2811 timing, its split in 3 parts, common high, low for 0 or 1, common low Apr 24 11:16:02 part 1 and 3 are fixed, and the middle part depends just on your bit pattern for 1 bit Apr 24 11:16:14 but it can be easily written to several GPIOs at once Apr 24 11:19:06 av500: yeah, the more I'm thinking about it, the more it seems that's the way to go Apr 24 11:24:52 any tips how to create a wireless router using bb? Apr 24 11:25:00 yes, don't Apr 24 11:25:51 why not? Apr 24 11:26:45 add a wifi cape Apr 24 11:26:49 run ddwrt or so Apr 24 11:26:52 there are other cheaper options, plus you only get a single real h/w ethernet interface Apr 24 11:27:05 but sure, go knock yourself out if it's for you Apr 24 11:27:11 get a TPlink 703 and run openwrt Apr 24 11:27:20 $20 Apr 24 11:28:57 thanks but i want full linux Apr 24 11:29:22 for a router all you need is a kernel and a handful of userspace tools Apr 24 11:29:33 search the raspberry Pi forums. its done the same way on BB Apr 24 11:30:12 ahem, IIRC there's an recipe in angstrom to generate a router image Apr 24 11:30:30 you make it sound so complicated Apr 24 11:30:58 * panto used to do routers on another life Apr 24 11:31:05 :) Apr 24 11:31:45 thanks. ill look on pi and angstrom Apr 24 11:31:48 what is a full linux? Apr 24 11:32:08 for a wifi "router" all you need is a properly configured kernel, iproute2, and hostapd Apr 24 11:32:38 i want to run all kinds of things in the router ... perhaps I wasnt clear Apr 24 11:32:54 u were clear. just wrong audience Apr 24 11:32:56 :P Apr 24 11:32:58 o.O Apr 24 11:33:28 buy a x86 computer Apr 24 11:33:46 srsly RPi forums have good instrunctions on it Apr 24 11:33:56 ? Apr 24 11:33:58 along with recommendations of which wifi dongle to use Apr 24 11:34:02 x86 too expensive to run Apr 24 11:34:09 why would you go there for instructions on running "ip addr add ...; ip route add ..."? Apr 24 11:34:11 thats cool crash thanks Apr 24 11:34:44 crashovrd, I guarantee you that a wifi router run on RPi with a USB dongle it will suck Apr 24 11:35:01 panto why? *g* Apr 24 11:35:01 omg, i didnt tell him to use a PI Apr 24 11:35:11 i told him that the instructions for doing it were the same Apr 24 11:35:16 pi isnt suitable I heard Apr 24 11:35:22 and implied there was additional helpful info Apr 24 11:35:32 cause when you pump packets you need a certain amount of horsepower Apr 24 11:35:43 other than 1) compile a kernel 2) make a distro 3) hack xyz Apr 24 11:35:56 :) Apr 24 11:36:05 and because the rpi drops usb frames Apr 24 11:36:12 bb sounds better for "full linux" wifi router I have in mind Apr 24 11:36:15 jsut a few Apr 24 11:36:48 cant it turn them into video frames and process them through the decoder instead ? Apr 24 11:36:55 having to go through usb for said packets, and correct me if I'm wrong, isn't the RPi ethernet on USB too Apr 24 11:37:04 it is Apr 24 11:37:04 yes Apr 24 11:37:13 so you have to make sure it loses packets in sync Apr 24 11:37:24 I friggin guarantee then, you will suffer Apr 24 11:37:26 on eth and wlan Apr 24 11:37:30 yes pi ether IS on usb Apr 24 11:37:56 unless you're still on dial-up, which in that case it's good enough Apr 24 11:38:22 lol no. adsl Apr 24 11:38:37 hope you like waiting Apr 24 11:39:19 so in a nutshell, yeah bb would make a better router that rpi, but it would still be retarded compared to just getting another router and putting ddwrt on it Apr 24 11:39:37 however, if you want to do some other stuff too, on top of being a router, the bb would be better Apr 24 11:39:56 in the time it took to have this 'debate' it would be already done, up and running on RPi Apr 24 11:39:58 :P Apr 24 11:40:07 im not confident using ddwrt ... would prefer to invest in something I know like deb/rh even freebsd Apr 24 11:40:14 done poorly, yes Apr 24 11:40:22 but half-assing it, that's the RPi way :) Apr 24 11:40:42 rpi isnt designed for real work I think Apr 24 11:40:43 ah right we had it yesterday with the usb Apr 24 11:41:09 rpi is designed to get rid of old chip stock without losing money Apr 24 11:41:15 mru, +1 Apr 24 11:41:36 that's going to be fun when the old stock dries up Apr 24 11:41:37 well, i guess broadcom makes some money actually ... Apr 24 11:41:40 but it lead the way in cost Apr 24 11:41:45 on the video decoder side Apr 24 11:41:48 led Apr 24 11:41:56 one might even call the way they're funneling it through this "charity" foundation a scam Apr 24 11:42:16 :D Apr 24 11:42:27 dunno, this 'foundation' looks like to have deep connections to UK Apr 24 11:42:41 it has deeper ties to bcm Apr 24 11:42:47 yeah Apr 24 11:42:52 smells like a desperate attempt to restart the old british home computer industry Apr 24 11:43:02 yes panto Apr 24 11:43:03 desperate? Apr 24 11:43:08 arm is going strong these days Apr 24 11:43:17 and it's as british as ever Apr 24 11:43:17 but not exactly desperate Apr 24 11:43:18 mru, not talking about arm Apr 24 11:43:39 mru but its nothing besides intel or samsung Apr 24 11:43:46 arm is doing great, but there's no manufacturing done in the .uk Apr 24 11:43:48 looking at the ebid Apr 24 11:43:57 i wasnt hear to talk about pi tho ;) Apr 24 11:44:01 samsung is a different kind of beast Apr 24 11:44:04 here Apr 24 11:44:18 nikem: we will do the talking Apr 24 11:44:46 tiny arm boards must have a very bright future Apr 24 11:44:56 o.O Apr 24 11:44:56 VERY bright Apr 24 11:45:13 they have a very bright past in your phone already Apr 24 11:45:14 they have the future since 4 5 years Apr 24 11:45:17 tiny arm boards, aka phones Apr 24 11:45:44 i mean tiny arm boards that can be used by ppl like me Apr 24 11:45:54 you cant use your phone ? Apr 24 11:46:06 then your manufacturer surely sucks Apr 24 11:46:10 if you can't use a phone, the beagle is not for you Apr 24 11:46:16 i mean for building things Apr 24 11:46:47 instead of using x86 boxes Apr 24 11:46:49 nikem, ahem, there are people in this channel have been using tiny arm boards to build things for close to two decades :) Apr 24 11:47:10 or even phones (or phone boards) Apr 24 11:47:28 ok panto maybe I steps into a nest of aces :) Apr 24 11:47:37 stepped Apr 24 11:47:55 the only difference is that you can do it now too as a hobbyist Apr 24 11:48:04 instead of having to be employed by a company to do it Apr 24 11:49:51 anybody know how reliable the cycle timings are for reading host memory from the PRU? I see a mentioned of 'stalled PRU cycles' in the datasheet. Are there some circumstances where reading host memory takes longer than others? Apr 24 11:50:11 contention Apr 24 11:50:31 constipation Apr 24 11:50:55 confirmation Apr 24 11:53:10 http://www.theregister.co.uk/2013/04/24/beagleboard_launches_black_board_computer/ Apr 24 11:53:19 "Black-eyed Pies" Apr 24 11:53:34 2 shades of grey Apr 24 11:54:20 blackbeard Apr 24 11:54:31 lol, @1st comment Apr 24 11:54:37 "The board has rounded corners Apr 24 11:54:37 Surely that's just asking for a lawsuit." Apr 24 11:54:48 heh Apr 24 11:55:07 the comments on that article should make it clear there's no winning over rpi lies Apr 24 11:55:17 ". And at the end of the day Cortex-A8 isn't *that* much faster than ARM1" Apr 24 11:55:24 ...So you upgrade from ARM11 to Cortex-A8, lose about 15% of the clock speed and lose the GPU?... Apr 24 11:55:42 wtf Apr 24 11:55:42 so 1GHz is 15% less than 700MHz? Apr 24 11:56:14 hehe Apr 24 11:58:00 one comment even says the black has no video out Apr 24 11:58:17 * koen proposes to bitbang s-video Apr 24 11:58:25 koen: can be done Apr 24 11:58:30 koen: svideo out is essential Apr 24 11:58:31 yep Apr 24 11:58:35 koen: s-video? why not go all the way to fbas? Apr 24 11:58:36 for poor uk kids to use the spare TV Apr 24 11:58:43 and learn computering Apr 24 11:58:49 pfst, use the RF aerial input Apr 24 11:58:58 yes Apr 24 12:07:08 what's this whole "no gpu" shit? Apr 24 12:07:14 lies Apr 24 12:08:11 well, there is a GPU Apr 24 12:08:16 but so for, no driver Apr 24 12:08:18 same thing Apr 24 12:09:17 was there a driver for the pi gpu? Apr 24 12:09:19 no Apr 24 12:09:25 so? Apr 24 12:10:02 yes Apr 24 12:10:04 there was Apr 24 12:10:14 otherwise no xbmc Apr 24 12:10:40 the pi has no gpu, it *is* a gpu Apr 24 12:10:46 with a puny arm core inside Apr 24 12:10:49 i came here after seeing the register article. i dropped the idea of being able to use rpi instead of x86 servers once i decided it had sacrificed too much to get the cost down to academic/hobbyist level Apr 24 12:12:14 pi vs bone is getting boring Apr 24 12:12:30 av500: it was documented? Apr 24 12:12:38 bone pile? Apr 24 12:12:38 av500: then i must have missed that Apr 24 12:12:59 nikem: hmm? Apr 24 12:13:12 nikem: what did they sacrifice that you would need? Apr 24 12:13:13 vim vs emacs Apr 24 12:13:38 rpi eth is run over usb Apr 24 12:13:49 KotH: did I say documented? Apr 24 12:14:00 nikem: well.. it's a phone cpu. phones do not have ethernet connections Apr 24 12:14:02 its a binary blob as usual Apr 24 12:14:14 KotH: no, its a STB cpu Apr 24 12:14:24 nikem: while stuff like the am3xx are more in the automotive market Apr 24 12:14:27 av500: STB? Apr 24 12:14:34 STD Apr 24 12:14:37 set top box Apr 24 12:15:03 ah.. Apr 24 12:15:11 then i'm surprised it doesnt have ethernet Apr 24 12:15:28 no need Apr 24 12:15:39 why? Apr 24 12:15:45 there is a ton of countries all over the world where all you need is a satellite dish Apr 24 12:16:05 ah.. Apr 24 12:16:15 then i would rather use a blackfin ;) Apr 24 12:16:21 o_O Apr 24 12:16:27 odd choice Apr 24 12:17:05 why? for decoding video, it's not bad Apr 24 12:17:35 KotH: http://www.youtube.com/watch?v=gRB45Jv6lW8 Apr 24 12:18:49 av500: no youtube at work Apr 24 12:19:34 its the monty python equivalent yo *deskhead* Apr 24 12:19:37 *to Apr 24 12:19:51 with religious touch Apr 24 12:19:52 ah.. Apr 24 12:20:54 av500, mru: so, what's wrong with the blackfin? Apr 24 12:21:47 if you need a DSP its fine Apr 24 12:21:55 why does an STB need a dsp? Apr 24 12:22:00 unless its made by TI Apr 24 12:22:05 then of course it needs a DSP Apr 24 12:22:13 KotH: what dev tools would you use with bfin? Apr 24 12:22:39 fish composer studio Apr 24 12:23:47 you have a choice of gcc, which is horribly buggy and produces slow code, or adi's commercial compiler which only runs on windows and uses a different abi Apr 24 12:23:55 and is no doubt horribly buggy too Apr 24 12:24:23 when paired up with an arm core, dsp tools being a bit quirky matters less Apr 24 12:25:20 you don't want to run your user interface on the bfin core Apr 24 12:25:33 or any dsp Apr 24 12:27:46 but blackfins don't have an application processor Apr 24 12:28:09 koen: don't take note of The Register comments (or the register in general) it's the equivalent of Phoronix for people who like to think they care about technology Apr 24 12:28:27 av500: dsp's are cool! everything needs a dsp! Apr 24 12:28:53 jackmitchell: the register is funnier than phoronix Apr 24 12:29:13 +1 Apr 24 12:29:34 I only read it for comedy value, intentional or otherwise Apr 24 12:29:36 who read comments at all Apr 24 12:31:28 I only read them for comedy value, intentional or otherwise Apr 24 12:35:12 you should work not read Apr 24 12:36:31 where's the comedy value in working? Apr 24 12:41:20 working? who said i'm working? Apr 24 12:41:30 all i did today was rebooting windows every couple of minuts Apr 24 12:41:34 minutes* Apr 24 12:41:43 that's more than I've done so far Apr 24 12:41:57 * jackmitchell finally got net-snmp to play ball \o/ Apr 24 12:41:57 i didnt even have to do that. its been rebooting on its own today Apr 24 12:42:46 mru: are you at least paid for this? Apr 24 12:43:13 I believe they expect me to do _something_ Apr 24 12:45:45 * KotH is tempted to go home and get a computer he can work with Apr 24 12:46:02 this is fucking stupid Apr 24 12:46:18 * KotH mumbles something about stupid decisions from higher ups Apr 24 12:47:34 they decided you have to use windows? Apr 24 12:47:48 yes Apr 24 12:48:12 and even better, i have to install all development tools in a vm, so it can be transfered onto someone else computer if there is a need Apr 24 12:48:15 well... Apr 24 12:48:26 * ogra_ resigned for such a decision from his last company Apr 24 12:48:28 good thinking Apr 24 12:48:32 i probably wasted 20% of my time with stupid vm problems over the past half year Apr 24 12:48:58 and now i got a new msp430 launchpad (because i fried the other one) and i cannot get it working with the vm Apr 24 12:50:15 ogra_: usually, the people around here are sane and listen to my opinion Apr 24 12:50:32 yeah, was like that for me as well ... Apr 24 12:50:38 until they stopped Apr 24 12:50:56 well. the moment they stop that, i'm gone Apr 24 12:51:04 right Apr 24 12:59:23 thank god for small companies where they don't have enough management to make such stupid decisions Apr 24 13:03:06 Unfortunately, there is often a reason why small companies are (still) small.... (e.g. stupid decisions!) Apr 24 13:03:22 burn! Apr 24 13:03:35 or they cater to niche markets, like us.... Apr 24 13:03:39 growing huge is not always the goal Apr 24 13:03:43 Sony don't want to build a special 100 off system Apr 24 13:05:23 teralaser: this is a small company Apr 24 13:05:25 teralaser: we are 11 Apr 24 13:25:57 hm, why is the beagleboard populated with 10nF (0.01uF) instead of 100nF ? Apr 24 13:26:00 cheaper? Apr 24 13:26:57 they had a case that corresponds better with the black Apr 24 13:28:57 hm, just seems a weird systematic typo on the schematic Apr 24 13:30:06 teralaser: probably higher self-resonance frequency Apr 24 13:30:16 most probably* Apr 24 13:36:11 well they are both 0402 ... so ... but yes. Apr 24 13:36:53 the BOM says 100nF, so I am figuring some dude made a mistake in the Orcad Schematics I stole :P Apr 24 13:37:11 teralaser: resonance frequency is a function of case size and capacitance Apr 24 13:38:01 and no matter what you do, there will always be a spot where there is a resonance for your PCB, so you might as well use mostly 100nF Apr 24 13:38:31 Hi, newbie to beagle here with a couple of questions if anyone has a minute Apr 24 13:40:55 I'm looking at getting a BeagleBone Black and adding an LCD4 or LCD3 but I need GPIO and PWM outputs, in fact at least 5 pwm outputs Apr 24 13:41:16 Anyone got any suggestions on how to do that as the screens use a lot of the io lines Apr 24 13:44:45 PeteS, use external I2C pwm chips Apr 24 13:45:11 teralaser: not really Apr 24 13:45:22 teralaser: depends on which frequency you are working with Apr 24 13:45:55 teralaser: and given proper design, the resonances on your PCB will not matter, because there are no signals that will excite that resonance Apr 24 13:46:15 teralaser: yes, that's not easy, but it's not rocket science either Apr 24 13:46:32 KotH : Not going to argue with you, sorry. Apr 24 13:46:47 damn! Apr 24 13:46:48 ;) Apr 24 13:46:51 hehe Apr 24 13:48:06 hadn't thought of that panto Apr 24 13:48:46 although we're thinking about making the PRU generate PWMs Apr 24 13:48:55 but that's not going to happen immediately Apr 24 13:50:55 What does the PRU do right now? Apr 24 13:51:37 be default, nothing Apr 24 13:51:38 by Apr 24 13:52:13 generate questions on IRC Apr 24 13:53:57 The other thing I need is Dallas one wire but it sounds like that is available on a pin already using the default distro Apr 24 13:54:49 I need to check if the same pin is used by the LCD capes though Apr 24 13:55:19 if need be, you can do 1-wire by hand Apr 24 13:57:27 I'm assuming it would be easy enough to modify the normal w1 code using a bit bashing driver. I think linux has one, it would just need pointing at a pin Apr 24 13:57:46 PeteS, there's already a cape that does that Apr 24 13:57:57 search in firmware/capes and find out Apr 24 13:58:13 a cape for what panto? Apr 24 13:58:18 and we call it bit banging here :) Apr 24 13:58:33 weather cape, that has an 1wire peripheral on it Apr 24 13:58:41 bit bashing, bit banging all the same :) Apr 24 14:00:21 names are critical Apr 24 14:00:24 we like to bang Apr 24 14:00:36 more bit banged for the buck! Apr 24 14:01:47 When I looked in the compatability page it said the weather cape wasn't compatible with any of the LCD capes Apr 24 14:02:02 maybe we need compatibility capes Apr 24 14:02:07 that shuffle GPIOs around Apr 24 14:02:16 a pin router cape Apr 24 14:02:51 I think it's just the sheer number of pins used up for the LCD capes Apr 24 14:03:48 yeah Apr 24 14:03:55 no way around it, they need pins Apr 24 14:04:23 indeed Apr 24 14:04:29 is there a pinExpanderCape? Apr 24 14:04:55 how many capes can fit on the head of a pin Apr 24 14:05:55 I think the Adafruit -16-Channel 12-bit PWM/Servo Driver - I2C interface - PCA9685 sorts out the PWM issue as long as an i2c bus is free to use Apr 24 14:06:09 isn't one i2c used by the capes themselves for eeprom Apr 24 14:07:03 one can always bitbang i2c Apr 24 14:07:08 PeteS, it's not used, you can add more i2c devices there Apr 24 14:07:19 there's I2C0, I2C1, I2C2 Apr 24 14:07:19 using a 1-wire GPIO extender Apr 24 14:07:27 connected by a PRU GPIO Apr 24 14:07:27 I2C2 is the one it has the eeproms Apr 24 14:07:42 the baseboard eeprom (and others) are on I2C0 Apr 24 14:07:47 I2C1 is totally free to use Apr 24 14:09:37 the datasheet for the LCD4 shows i2c1 connected, but it looks like it just goes to a header. Can't find any photos to corroberate this though Apr 24 14:09:51 schematics should tell Apr 24 14:10:28 sorry I meant schematics Apr 24 14:19:19 Looks like the beaglebone black will be the choice for me, I was going to get a Pi until I found beaglebone. A screen is important for my project :) Apr 24 14:20:12 beagle, always a good choice... unless you are a cat lover Apr 24 14:21:25 PeteS : isnt there a little memory i2c to identify the LCD4 ? Apr 24 14:22:31 It looks like it yes Apr 24 14:22:54 I have one LCD4 on my table Apr 24 14:24:00 teralaser is there a header somewhere labeled J2? Apr 24 14:24:22 yeah it's not mounted Apr 24 14:24:57 it goes out underneath the LCD4 but outside the beaglebone, so you can mount it Apr 24 14:25:00 Oh, wonder why they bothered using the pins, well I2C1 is free then :) Apr 24 14:25:29 Thank you very useful info that I haven't been able to find :) Apr 24 14:25:53 :) Apr 24 14:27:58 Will spend some time tonight working out what GPIO I will have free, I probably don't need more that 3 or 4 as long as DOW, i2c, and the screen can be connected :) Apr 24 14:30:58 Thanks for the help, chat again soon probably when I have more silly questions :) Apr 24 14:35:46 from BBB_SRM.pdf A5A: "The memory used is the MT41K512M16HA-125 from Micron. It will operate at a clock frequency of 303MHz yielding an effective rate of 606MHZ on the DDR3L bus allowing for 1.32GB/S of DDR3L memory bandwidth" Apr 24 14:36:03 Isn't this really slow? Apr 24 14:36:16 it's 400MHz IIRC Apr 24 14:36:28 and define slow please Apr 24 14:36:55 worse than any other similar board Apr 24 14:37:08 which similar board is that? Apr 24 14:39:08 the original Beagleboard, Raspberry Pi, Cubieboard, Apr 24 14:39:32 have you ever tried to make your own h/w design based on those board? Apr 24 14:39:35 *boards Apr 24 14:39:41 but 400MHz sounds better :) Apr 24 14:39:43 the bone isn't intended for hd video processing Apr 24 14:40:20 all those boards you mention are only intended for cheap mobile development Apr 24 14:40:31 you can't really make your own based on them Apr 24 14:41:33 also, usually TI fucks up the CPU memory access anyway Apr 24 14:41:39 so DDR speed does noteven matter :) Apr 24 14:42:42 heh Apr 24 14:47:16 av500, unpossible Apr 24 14:48:41 panto, did the magic 400Mhz timings make it in now? Tartarus_ ? Apr 24 14:49:01 Yes, they did Apr 24 14:49:09 av500: broadcom has also messed it up pretty badly with their external L2 cache for ARM11 Apr 24 14:49:20 In fact, it seems the "303" MHz timings would be more magical as that's not supported on that PG Apr 24 14:49:35 heh Apr 24 14:49:36 ssvb: people care for arm11? Apr 24 14:49:43 not people Apr 24 14:50:02 mdp: DM270, DM320, DM6441, OMAP3 all limited by DDR bandwidth Apr 24 14:50:05 mdp: OMAP4430 was really impressive Apr 24 14:50:15 then omap4 comes along and is now limited by CPU to DDR speed :) Apr 24 14:51:38 av500, yeah, I recall we evaluated omap3 for a project and found it horrible..in my previous life. Apr 24 14:52:00 av500, I was just claiming lies as I must act the role of the company man now ;) Apr 24 14:52:08 there are no issue Apr 24 14:52:21 av500: arm11? all the best for the children Apr 24 14:52:33 yes Apr 24 14:53:03 clearance sale Apr 24 14:55:30 .o0(if the code and the comments disagree....) Apr 24 14:55:49 if? Apr 24 14:56:23 often, there are no comments to disagree Apr 24 14:56:37 that's preferable to comments that lie Apr 24 14:57:29 i have here a piece of code that says "switches of (external) chip and the spi subsystem" what it really does is toggle the CS pin Apr 24 15:01:51 Oh. i _finally_ get what '.o0' is Apr 24 15:02:18 the opposite of Oo. Apr 24 15:02:47 I. Wah. Apr 24 15:04:11 (eh soed ,ti teg t'nsoed eh)0o. Apr 24 15:05:41 KotH: I can decipher your cleverly coded messages! Apr 24 15:08:18 KotH: damn, you'd better start using double rot13 Apr 24 15:08:41 mru: nah.. i just send someone with a bomb belt over Apr 24 15:08:44 mru: much more effective Apr 24 15:12:21 jkridner, apparently you have magical pixie dust because I all of a sudden see some direction to do upstream *again*. lol Apr 24 15:13:56 SPRUH73h seems to have had its PRU documentation removed -- just 2 pages of almost nothing, so I assume it's been moved to some other TI document. Does someone have the relevant link please? Apr 24 15:14:06 fiola: get spruh73c Apr 24 15:14:21 Searching, thanks Apr 24 15:14:28 fiola: at least mdp would tell you to do that Apr 24 15:14:31 ;) Apr 24 15:14:50 just google the code, it's the first hit Apr 24 15:15:12 I have the code, was looking for the docs. Apr 24 15:15:13 8+3 sekrit codez Apr 24 15:15:27 yes, google "spruh73c" Apr 24 15:15:34 Yup, doing Apr 24 15:17:18 it has the answers you seek Apr 24 15:18:24 Are we in transition or something, with a dedicated PRU TRM coming out? A zillion people are going to be interested with the release of BBB. Apr 24 15:18:35 no Apr 24 15:19:01 mdp: let me guess, the other answer one might seek are hidden in the koans? Apr 24 15:19:20 yes Apr 24 15:19:50 hrm...we really need to create a "PRU koans" in the spirit of vim/git koans Apr 24 15:20:09 that's a big enough waste of time that it should be considered Apr 24 15:20:32 You sure? The release announcement made special play of the PRUs, a bullet point feature. It sounded to me like they were giving it pride of place (which I think it deserves, though experience will tell). Apr 24 15:22:47 fiola, beagleboard.org is not TI Apr 24 15:22:55 TI writes the chip trm Apr 24 15:23:01 bug your local TI rep Apr 24 15:23:51 well, "bug your local TI rep" is no answer for a BBB customer Apr 24 15:23:57 ? Apr 24 15:23:59 if PRU is featured Apr 24 15:24:05 one would expect its documented Apr 24 15:24:15 av500, it is, just in old docs that have been redacted since Apr 24 15:24:22 :) Apr 24 15:25:14 bradfa: That's a good point, but you should have suggested "bug beagleboard.org" since they've highlighted the PRU as a bullet point and make a point of open documentation. Apr 24 15:25:27 fiola, it's open docs, just not the current ones Apr 24 15:25:39 fiola, and there's a "community" set of code taht can be used Apr 24 15:25:47 people are using the pru Apr 24 15:26:29 I know, I've seen examples. But looking in the 'h' TRM caused a WTF moment :P Apr 24 15:26:34 https://github.com/beagleboard/am335x_pru_package/tree/master/Documentation Apr 24 15:26:35 * av500 sings "get into the PRU" Apr 24 15:27:21 I see pru docs within beagleboard.org's github, what's the problem? Apr 24 15:27:36 and rev c of the trm has TI docs in it Apr 24 15:27:42 how's that not open? Apr 24 15:27:57 * bradfa goes and eats, hunger frustrates me Apr 24 15:28:57 No problem at all, if they're detailed. I asked to find them, don't be so touchy. :P Apr 24 15:30:27 fiola: bradfa isnt touchy Apr 24 15:30:47 fiola: but dont let mranostay touch you ;) Apr 24 15:30:54 lol Apr 24 15:31:16 keep your creeper cards at hand when mranostay is around. Apr 24 15:31:42 I'll put one PRU on mranostay detection duty Apr 24 15:31:44 * KotH has a dozen of them, if anyone needs some spare ones Apr 24 15:31:58 fiola, we've tried to be clear that spruh73c is the canonical documentation to use. Apr 24 15:32:27 mdp: Thanks, I'll help spread that word around too Apr 24 15:32:29 a little birdie tells me that "the community" may be getting this properly collected on elinux wiki *finally* Apr 24 15:32:53 Excellent Apr 24 15:32:53 as, for a variety of reasons, TIers cannot be involved in that Apr 24 15:32:57 Excuse me - I'm currently trying to decide between Rasp Pi and BBone Black for a road traffic monitoring project, I believe there will shortly be a camera module for the pi, but I'm wondering what kind of camera cape or module may be available for the BBone Black - preferably without tying up the USB port? Can anyone help please? Apr 24 15:33:17 fiola, I hope that will solve the problem finally Apr 24 15:34:24 andybets: that largely depends on what you are doing and how you plan to do it exactly Apr 24 15:34:27 andybets: does your project involve 3D graphics or hd video decoding? Apr 24 15:34:36 if not, the blackbone is always the better choice Apr 24 15:34:52 andybets: and what mru just said Apr 24 15:36:13 thanks... it's a little open ended... we may do some limited computer vision on the stream if we can, otherwise just record to h264 to an SD card I guess Apr 24 15:37:06 * KotH wouldnt use a beagle for recording video Apr 24 15:37:15 i'd rather use a full blown pc Apr 24 15:37:27 video gets very big very quickly Apr 24 15:37:32 andybets: And if your application makes heavy use of USB then RPi may not work at all, because its USB controller is effectively broken and drops USB events under some circumstances. So don't commit to a design until you've tested it. Fortunately both boards are so cheap that buying both won't break the bank. Apr 24 15:37:34 Pi Envy(tm) Apr 24 15:37:39 modules/camera that provide h264 would be interesting, as would something providing raw/jpeg for CV purposes Apr 24 15:38:12 if size/cost doesn't matter maybe a mini itx box? Apr 24 15:38:15 BBB wont encode much video Apr 24 15:38:22 andybets: there are cameras that give you mjpeg Apr 24 15:38:34 yes, over USB Apr 24 15:38:40 andybets: i have never heard of a camera that does h264... but i have not been looking for that either Apr 24 15:39:02 av500: details! ;) Apr 24 15:39:19 some logitech webcams seem to do h264, but I'd prefer an option where the USB is kept clear for other purposes if possible Apr 24 15:40:13 BBB is not the best choice for video encoding Apr 24 15:40:35 there are camera capes via GPMC, so usb is free Apr 24 15:40:36 BBB has an independent Ethernet path, not over USB, if you want to keep USB clear. Apr 24 15:40:39 but you have raw video Apr 24 15:40:58 yes, eth camera is possible Apr 24 15:41:06 they tend to do h264 Apr 24 15:41:14 * mdp ponders porting uARM to 6502 to boot linux Apr 24 15:41:30 mdp: do it! Apr 24 15:41:46 if all you want is basically a camera recording as h264, I don't see much reason to roll your own design Apr 24 15:41:55 companies like axis already do that just fine Apr 24 15:42:05 KotH, I just remembered the 13 year old april 1 lkml post of linux on 6502 Apr 24 15:42:19 KotH, and it got me thinking Apr 24 15:42:22 thanks... av500... any cape links? Apr 24 15:42:29 6502 is too bleeding edge, let's keep a solid foundation and use 8080. Apr 24 15:42:33 4004 Apr 24 15:42:41 pffft Apr 24 15:42:50 http://www.google.de/search?sourceid=chrome&ie=UTF-8&q=beaglebone+camera+cape Apr 24 15:42:58 I don't work on that crap...only godprocessors Apr 24 15:43:06 mru... we're using Axis cams at the moment... looking for a cheaper arrangement where we can have a dozen or more units deployed Apr 24 15:43:07 not goodprocessors? Apr 24 15:43:14 Hehe. I've never used a 4004, but I used an 8008, a couple of years before 8080. Apr 24 15:43:33 suddenly I dont feel old any more Apr 24 15:43:35 thanks av500 Apr 24 15:43:52 mdp: it would almost certainly be more efficient to emulate something simpler than arm Apr 24 15:44:01 finger? Apr 24 15:44:03 hand? Apr 24 15:44:09 thumb :) Apr 24 15:44:15 nail Apr 24 15:44:20 hammer Apr 24 15:44:20 hammer Apr 24 15:44:27 mru wins Apr 24 15:44:29 I don't know which of the linux-supported architectures is the simplest Apr 24 15:44:30 :) Apr 24 15:44:35 x86 Apr 24 15:44:38 mru: emulating a transputer would be pretty cool. That had a unique architecture. Apr 24 15:44:39 its best supported Apr 24 15:44:55 fiola: easy to emulate for sure Apr 24 15:44:58 mru, I was trying to be a little lazy by reusing that AVR implementation Apr 24 15:45:15 but linux doesn't run on that Apr 24 15:45:16 mru, but do it all via bone host i/o Apr 24 15:45:37 mru: aye, extremely RISCy, I loved its tiny but extensible instruction set. Apr 24 15:45:51 I came to loathe it Apr 24 15:46:03 or rather the st20 incarnation of it Apr 24 15:46:06 From before the days when "RISC" was an accepted term. Apr 24 15:48:35 how do I know that I am compiling a kernel for a cheap android tablet Apr 24 15:48:45 drivers/input/touchscreen is just warnings Apr 24 15:50:06 andybets: rolling your own sollution will nearly always be more expensive unless you go into massproduction Apr 24 15:50:30 for the masseffect Apr 24 15:50:52 or if it's so specialised that there isn't an off the shelf solution Apr 24 15:51:04 yeah... Apr 24 15:51:11 but it doesnt sound so specialized yet Apr 24 15:52:00 or using off the shelf parts would end up with unused bits Apr 24 15:52:32 it is pretty specialised. There are quite a few contraints. Battery life, weights, dimensions etc. Apr 24 15:53:22 andybets: oh.. running from battery... Apr 24 15:53:32 andybets: then you dont want to do encoding at all Apr 24 15:53:43 andybets: or anything that needs cpu Apr 24 15:54:11 andybets: unless you have a 200kWh battery, like the other project we are doing :) Apr 24 15:54:38 yeah... 20Ah lithium batteries... the Axis cameras and some webcams can do the encoding in this situation Apr 24 15:54:59 how long does it have to run between recharges? Apr 24 15:55:25 12-48 hours depending on the use Apr 24 15:55:41 that sounds doable, even with off the shelf components Apr 24 15:55:46 get a cell phone and glue some oversized battery pack to it. Apr 24 15:56:01 being able to sleep at certain times is another thing that the BB or RP solution may offer Apr 24 15:56:18 andybets: research project? Apr 24 15:56:20 camera is there, android is there, etc. Apr 24 15:56:33 ;) our clients wouldn't like that Apr 24 15:57:03 temporary traffic surveys Apr 24 15:57:14 then enclose it in a sealed box Apr 24 15:57:18 yeah. Apr 24 15:57:32 and put some scary warning stickers on it Apr 24 15:57:47 [WARNING: SCARY] Apr 24 15:57:49 not sure a cellphone would record that long... but it is a thought! Apr 24 15:57:58 given the paltform itself, it would be considerably less hackish AND way cheaper than anything else you've mentioned so far. Apr 24 15:58:25 if you want to record longer, spend $3 on a micro sd card to insert. Apr 24 15:58:59 here are some good ones: http://store.wondermark.com/products/hyperbolic Apr 24 15:59:28 h264 may be a requirement... do any phone record that? Apr 24 15:59:36 transcoding may not be feasible Apr 24 15:59:38 do any not? Apr 24 15:59:59 above logic does not apply if you are going to sell >10k units per year or so, bu if you need a cheap platform that does not rely too much on a huge variety of different parts for your hackup, a cell phone is pretty much a commodity item. Apr 24 16:00:32 fiola, sorry for being touchy, was a frustrating morning, I didn't mean any disrespect to you Apr 24 16:00:35 mind, many different parts for hackup usually mean that at least one is always out of stock, being replaced or whatever. Apr 24 16:01:06 Apr 24 16:02:13 bradfa: hunger is a bad thing :P Apr 24 16:02:25 not really Apr 24 16:02:32 if not for hunger, eating would be much less pleasant Apr 24 16:02:39 lol Apr 24 16:02:54 mru: you live in england, eating is always a pain Apr 24 16:02:58 and I do enjoy a good meal Apr 24 16:03:18 KotH: nonsense Apr 24 16:03:22 there's good food in england Apr 24 16:03:25 yes Apr 24 16:03:28 immigrants make it Apr 24 16:03:30 you've just got to know where to find it Apr 24 16:03:53 Eating would only be a pain here (in UK) only if we stuck to British cuisine, which doesn't exist. Apr 24 16:03:59 As av500 says :P Apr 24 16:04:37 UK is a food aggregator from its ex-empire :-) Apr 24 16:04:43 my local pub serves a nice sunday roast... Apr 24 16:05:07 mru: i gladly let me introduce to good food when i'm in the uk next time Apr 24 16:05:28 i agree , wasn't aware that android phones/tablets could do real-time h264 encoding ? are you sure? Apr 24 16:05:29 mru, on a wednesday ? Apr 24 16:05:59 no, on wednesdays they do fake mexican food Apr 24 16:06:08 heh Apr 24 16:06:22 entire staff looks very british to me Apr 24 16:06:26 Buh, such unnecessary constraints that sunday should only fall on a sunday Apr 24 16:06:42 andybets: are you joking? Apr 24 16:06:47 if course they do Apr 24 16:06:50 of Apr 24 16:06:53 that said, the burrito isn't bad Apr 24 16:07:03 sorry... we all have our specialisms! Apr 24 16:07:15 metal and spice, for the record. Apr 24 16:07:19 its all that "full HD wardrobe fails" on youtube :) Apr 24 16:07:33 though most are iOS Apr 24 16:10:23 letothe2nd: next time we meet, i'd like to have a bit of your spice Apr 24 16:10:55 KotH: hrhr Apr 24 16:11:25 ok... I see I was fooled by the container format, and google's webm emphasis... that's news to me Apr 24 16:11:38 so thanks Apr 24 16:12:22 np Apr 24 16:14:45 that really does make the tablet in enclosure idea seem vaguely sensible... I may need a better camera quality than that, for some things (low-light), but it'll work for some of it perhaps. Apr 24 16:16:25 hi there, having created a proposal on gsoc 2013. anyone can give some advices about making the 11-week long schedule? Apr 24 16:17:01 code like a demon Apr 24 16:17:29 :-), nice advice. Apr 24 16:18:02 or playing with the board like a demon,yeah? Apr 24 16:19:47 XorA: you mean daemon Apr 24 16:20:05 KotH: depends on your religion Apr 24 16:20:42 any mentors here? Apr 24 16:20:58 or can I have your time plz. Apr 24 16:20:59 mostly tormentors Apr 24 16:21:00 XorA: my religion demands the heads of infidels! :-) Apr 24 16:21:26 KotH: I expect to see row upon row of Pi fans then :-D Apr 24 16:21:32 mru, what does tormentors mean Apr 24 16:21:42 we torment people Apr 24 16:22:01 :-( Apr 24 16:22:35 guess what, time to sleep:) Apr 24 16:23:21 we can probably torment you in dreams as well! Apr 24 16:23:47 * KotH has done that occasionally Apr 24 16:23:55 see you there then:-) Apr 24 16:24:20 XorA: i dont dirty my hands with such lowly beings... they are left for the lower casts of our religion Apr 24 16:24:32 morning Apr 24 16:24:47 en wunderschöne guete morge herr mranostay Apr 24 16:25:04 KotH: *castes :P Apr 24 16:25:35 domo Apr 24 16:26:08 * KotH sighs Apr 24 16:26:24 why are some people writing bad code although they should know better? Apr 24 16:26:42 because they _don't_ know better Apr 24 16:26:48 even though they should Apr 24 16:27:01 or it's simply against the corporate policy Apr 24 16:27:44 sometimes i feel like being the most experienced programmer around... although i'm a very bad coder :-( Apr 24 16:28:10 there's a simple explanation for that Apr 24 16:28:16 hybris? Apr 24 16:28:19 the good coders don't get to code Apr 24 16:28:52 I consider myself reasonably good, and I can't remember the last real line of code I wrote Apr 24 16:29:14 * KotH sighs Apr 24 16:29:19 but i dont want to code! Apr 24 16:29:24 i want to do hardware! Apr 24 16:29:27 re Apr 24 16:29:28 * KotH cries Apr 24 16:29:38 and I don't want to run spec2k6 Apr 24 16:29:41 koth sw is better than hw Apr 24 16:29:45 It's ok mru, I can't remember the last time I wrote something other than a script to compensate for the deficient network architecture at my college. Apr 24 16:30:14 woglinde: most software could be written by a reasonably intelligent student after a year of training Apr 24 16:30:24 woglinde: that's neither a challange nor fullfilling Apr 24 16:30:28 trained monkeys write boot loaders Apr 24 16:30:30 KotH: most of that software would be of Oracle quality. Apr 24 16:30:46 woglinde: and you can paper over a lot of shitty code to get it somehow work Apr 24 16:31:10 woglinde: that's much more difficult with hardware... if there is a bug in your hw design, the whole thing might blow up Apr 24 16:31:16 literally Apr 24 16:31:32 * Hardolaf has only had one capacitor explode on him in the last six months :D Apr 24 16:31:39 KotH: then you paper it over with software Apr 24 16:31:40 koth I think I could do hw too after a year of training Apr 24 16:31:42 I do actually mean on me... Apr 24 16:31:46 koth so there is no point Apr 24 16:31:56 woglinde: but i dont mean that there is no good code or interesting software projects. they are just hard to come by ... Apr 24 16:32:22 they're not, but it's hard to get paid to work on them Apr 24 16:32:23 woglinde: hmm... right... Apr 24 16:33:06 Bedrock Linux is a great example of an interesting project, sadly, no one makes any money from it. Apr 24 16:33:11 * KotH quits his job and becomes landscape gardener Apr 24 16:33:24 KotH, that could actually be interesting. Apr 24 16:33:33 You'd get more exercise definitely. Apr 24 16:34:11 * KotH wonders whether he should do a phd and go into research Apr 24 16:34:21 ideas are the things that afford opportunities for interesting programming work aren't they? Apr 24 16:34:39 * Hardolaf wonders if he should actually care about his calculus final today. Apr 24 16:34:48 KotH: it's even harder to get paid there Apr 24 16:34:55 andybets: for most software projects, you dont need bright ideas Apr 24 16:34:59 have a good idea, spend siz months developing it, then sell it... that's the aim, that I'm always on the edge of Apr 24 16:35:17 andybets: most stuff has been done again and again and any programmer worth his salt knows how to do such projects Apr 24 16:35:30 andybets: after the design is done, it's mostly just typing the stuff Apr 24 16:35:38 most stuff yes... that why ideas are the key Apr 24 16:35:54 ideas dont get you paid, ideas cannot be sold Apr 24 16:35:57 you need a product Apr 24 16:36:08 and those are often very boring on the inside Apr 24 16:36:17 you have an idea, then develop it, then sell it... simple in theory Apr 24 16:36:24 ideas get people persecuted. Apr 24 16:36:34 that's true too Apr 24 16:36:52 mru: not that much for an EE Apr 24 16:36:57 If you want money, make a product; if you want to make a difference, come up with an idea. Apr 24 16:37:10 mru: there are so few EEs that do phds that you have a good chance to get a decent place Apr 24 16:37:25 I would have made that an else if, but I feel like they aren't mutually exclusive. Apr 24 16:37:27 there's no such thing as making a difference in the grand scheme of things Apr 24 16:37:39 unless you find a way to prevent the heat death of the universe Apr 24 16:37:41 shh mru, don't discourage people... Apr 24 16:37:48 mru: but it unless you end up at a great place, you'll be writing applications and proposals to get money, most of the time Apr 24 16:38:07 and where's the fun in that? Apr 24 16:38:14 Become like my faculty advisor: work on the same project for 20 years. Apr 24 16:38:20 mru: there isnt much fun in that Apr 24 16:38:22 With no useful results. Apr 24 16:38:29 mru: that's one of the reasons why i'm not in research Apr 24 16:38:42 Hardolaf: sounds typical for academia Apr 24 16:38:56 Hardolaf: er.. you know that most of research is a wild goose chase? Apr 24 16:39:01 mru: to be honest, it did take him 2.5 years to process data at one point. Apr 24 16:39:03 KotH: yes. Apr 24 16:40:20 KotH: even less fun is when you push students out the door with a degree, and the following year they return for a class reunion and are already earning more than you. Apr 24 16:40:40 Especially at this university! Apr 24 16:42:43 And what really grinds is when it's the fairly bad students that you maybe shouldn't have awarded a degree because you know full well that they don't *understand*, despite having passed. Apr 24 16:42:46 koth the trick is to reuse your code in new projects Apr 24 16:43:37 koth and the other trick is to make it opensource so you can reuse it at your new job Apr 24 16:44:08 woglinde: i have never written such code, otherwise i would use mine Apr 24 16:44:27 koth so you are clearly making something wrong Apr 24 16:44:38 woglinde: i cannot opensource something i do not own and the company is not interested in that (they consider the code their prime asset) Apr 24 16:44:44 woglinde: looks like Apr 24 17:26:20 ugh Apr 24 17:26:24 PRU does not have a long jump? Apr 24 17:26:35 what is this 1968? Apr 24 17:28:35 cmicali, PRU is not a general purpose CPU Apr 24 17:28:53 you got to make concessions to bang the bits fast Apr 24 17:28:58 indeed Apr 24 17:28:59 indeed you do Apr 24 17:29:10 with 8k instruction ram, how long a jump can you make? Apr 24 17:29:27 whatever you do...don't foot fault Apr 24 17:29:36 hah Apr 24 17:29:48 what happens if you run off the end? Apr 24 17:29:55 there be dragons Apr 24 17:30:08 +1 Apr 24 17:30:09 av500: we found your dragons Apr 24 17:30:16 not really sure, i'm abusing macros and an outer loop is saying "Operand 1 relative jump out of range" Apr 24 17:30:38 the jmp parameter says it's 16 bits long, so not really sure what the deal is Apr 24 17:30:55 maybe fewer bits for the quick branch instructions Apr 24 17:31:13 looking at the insn encoding will tell you Apr 24 17:33:41 yeah Apr 24 17:33:42 that's it Apr 24 17:33:48 9 bit offset Apr 24 17:40:27 that's a reasonable jump Apr 24 17:45:26 jump, to the left? Apr 24 17:50:02 <_av500_> ahoi Apr 24 18:00:42 panto: ping Apr 24 18:00:56 panto: ping Apr 24 18:00:58 felipebalbi, pong Apr 24 18:00:59 :) Apr 24 18:01:04 ping Apr 24 18:01:17 pong Apr 24 18:01:24 panto: hi, about am335x touchscreen driver. Any clues about what's missing ? (apart from DT and runtime PM) Apr 24 18:01:42 panto: also, are there any bone capes which I could use to play with that ? Apr 24 18:01:46 felipebalbi, err, we got it working Apr 24 18:01:53 panto, upstream Apr 24 18:01:55 panto: yeah, I figured that much :-) Apr 24 18:02:01 ah, that :) Apr 24 18:02:10 felipebalbi, keep in mind "got it working" means in the beagleboard.org vendor tree Apr 24 18:02:18 so let's all get on the same page first Apr 24 18:02:30 panto: oh, so there are patches in beagleboard.org tree ? we can just cherry-pick for mainline, if you don't mind Apr 24 18:02:35 ok, so, there's an LCD3 cape with touch screen Apr 24 18:02:48 felipebalbi, you can, but, I had to fix other parts as well Apr 24 18:03:06 the mfd driver for touch/adc comes to mind Apr 24 18:03:24 panto: no issues, as long as you point me to the tree, I'll cherry-pick the patches, fix any conflicts and push it upstream (of course I'll keep authorships in place ;-) Apr 24 18:03:35 panto: yeah, that mfd part is a bit 'weird', to say the least Apr 24 18:03:37 but fair enough Apr 24 18:03:46 firmware/capes/cape-bone-lcd3-00A2.dts Apr 24 18:03:52 can i use keil software with it Apr 24 18:04:15 <_av500_> with a burrito? Apr 24 18:04:50 felipebalbi, one of these days I'll have to rebase against 3.9 and clean up Apr 24 18:04:55 I'd appreciate it if you did it :) Apr 24 18:05:17 panto: sure, just point me to the git tree and I'll help you guys out, since that helps us out too :-p Apr 24 18:05:39 panto, told you TI cares about upstream again ;) Apr 24 18:05:42 ok, so there's the beagleboard.org tree, but that's in form of patches Apr 24 18:06:22 btw, BB black is a sexy board, ain't it ? or maybe I've been close to HW for far too long :-p Apr 24 18:06:38 koen's tree, has a sikrit branch that's for internal use which has the tree in normal form Apr 24 18:06:40 felipebalbi, only if you have an A6 Apr 24 18:07:04 my A4 is castrated Apr 24 18:07:19 git://github.com/koenkooi/linux.git branch 3.8-for-panto-rebase Apr 24 18:07:53 alright Apr 24 18:07:56 I'll go there Apr 24 18:08:15 it's in a known good working state Apr 24 18:08:26 so not only builds, but it works too! :) Apr 24 18:08:28 that makes our life a whole lot easier :-) Apr 24 18:08:44 try to do something with the omap device mess please Apr 24 18:08:46 panto: builds *and* works ? that's gotta be a plus :-p Apr 24 18:09:03 most of the drivers can not be unregistered without crashing Apr 24 18:09:04 panto: we're trying... lots of barriers to break though :-( Apr 24 18:09:52 k, I'm hungry, bb after dinner Apr 24 18:10:02 alright, let me try to get my hands around an LCD3 cape Apr 24 18:11:22 felipebalbi: where are you located nowadays? Apr 24 18:12:11 * koen realizes he sent all his spare capes to panto Apr 24 18:12:18 koen: still in Finland Apr 24 18:13:04 koen: fetching your tree right now ;-) Apr 24 18:13:13 I'll check tomorrow if I have working lcd capes to spare that I can send you Apr 24 18:15:50 koen: thanks, that'd help validating Apr 24 18:16:45 koen: you sent items to greece? Apr 24 18:16:56 i hope you don't expect to get them back :P Apr 24 18:17:12 yes, stimulating the economy Apr 24 18:19:00 <_av500_> mranostay: its not about getting them back Apr 24 18:19:08 <_av500_> its about owning greece after a while Apr 24 18:19:12 <_av500_> we did it with submarines Apr 24 18:20:37 I don't understand why BB black got 1GHz ARM CPU(AM3359) and the previous BB got the same micro but the maximum speed up to 720 MHz Apr 24 18:21:07 pg1.0 vs pg2.1 Apr 24 18:21:17 _av500_: going to cart away islands again? :P Apr 24 18:21:26 different silicon revs Apr 24 18:23:15 I see, ty Apr 24 18:27:25 <_av500_> also, that speed difference matters little Apr 24 18:28:50 1000MHz looks better IMHO Apr 24 18:29:10 0Hz looks even better Apr 24 18:29:21 it should be properly advertised as a quad-core SoC too Apr 24 18:29:23 dc! Apr 24 18:29:36 yes Apr 24 18:29:41 0Hz @ 0A Apr 24 18:29:45 unzip the quad core fury *now* Apr 24 18:31:10 ds2: That would be an ideal device for computing the halting problem. Apr 24 18:32:39 <_av500_> http://26-26-54.hardwarebug.org/14 Apr 24 18:39:30 Hello, I am interested to work on "sysfs entries for IIO and PWM" for GSOC'13. Any pointers on what needs to be done will be helpful. Apr 24 18:40:14 rtdin subscribe to the beagle gsoc mailinglist and repeat your question there Apr 24 18:40:55 did u mean beagle gsoc google group ? Apr 24 18:41:43 I have already asked the question in that google group. Apr 24 18:48:33 okay Apr 24 18:48:37 so be paitent Apr 24 18:49:51 okay Apr 24 18:58:11 <_av500_> koen: has jkridner elpoed? Apr 24 18:58:15 <_av500_> eloped? Apr 24 18:58:23 <_av500_> eaten by a grue? Apr 24 18:58:29 designwest Apr 24 18:58:36 <_av500_> ah Apr 24 18:58:42 <_av500_> right Apr 24 18:58:48 he should be doing workshops with the baconboards this moment Apr 24 18:58:56 if not, he's adding moustaches to people Apr 24 19:00:04 prpplague, gerald and clint are there as well Apr 24 19:02:54 he might run out of people to add mustashes to Apr 24 19:03:31 so soccer now Apr 24 19:03:39 <_av500_> beer now Apr 24 19:03:55 hm no Apr 24 19:04:04 lol Apr 24 19:04:05 goal Apr 24 19:04:17 looks like german final Apr 24 19:30:02 koen: When you experimented with kernelshark, did you run it on host or target? Apr 24 19:31:44 think kernelshark would make a nice usecase for BBB hdmi =) Apr 24 19:33:08 do it cheaper on RPi Apr 24 19:34:10 but I thought rpi was full of binary blobs without debug info, difficult to trace Apr 24 19:34:39 debug info is overrated Apr 24 19:34:49 joel_, like the BBB SGX binary blob? Apr 24 19:35:05 yes like that one Apr 24 19:35:06 that's completely different Apr 24 19:35:09 right! Apr 24 19:35:11 that's _our_ binary blob Apr 24 19:35:29 tame and gentle Apr 24 19:35:43 brought here to Vegas to thrill audiences until it eats us alive Apr 24 19:57:43 what matter is if we can create MINIX drivers for it Apr 24 19:58:09 to save an other piece of hardware Apr 24 19:58:15 RMS does not aprove of the torvalds version of linux Apr 24 19:59:15 Is there any other kind? Apr 24 19:59:58 RMS just doesn't approve in general. Apr 24 20:00:35 does the BBB provide any kind ot serial over usb e.g. does the u-boot implement the serial gadget? or will do you expect people to get the serial cable when it doesn't work? Apr 24 20:00:50 RMS does approve of Linux and its licensing, it's under a GPL license after all. He just wishes it were GPLv3. Apr 24 20:00:58 keesj: you dont have a serial cable? Apr 24 20:01:23 not a 3.3 ttl level no Apr 24 20:01:43 olimex sells them for 5 bucks Apr 24 20:01:55 or get an el-cheapo ftdi board Apr 24 20:02:05 * keesj rembers the days where topic number one on #beagle was how to connect a serial on the beagleboard :P Apr 24 20:02:26 eh.. Apr 24 20:02:29 I do have a flyswatter (1) laying around Apr 24 20:02:48 " can we please stay off topic?" Apr 24 20:05:06 or if you're crazy and you want low-intrusive debug, just print to the ETB. once you figure out how to Apr 24 20:05:06 olimex is dangerous (I might order a non TI dev board) Apr 24 20:05:21 olimex is cool Apr 24 20:05:53 i dont know how those guys make money, but the boards work great (ok, they are easy to build), reasonably priced and the support from olimex is good Apr 24 20:07:22 I talked to Tsvetan @ embedded world. He sounded like a good guy Apr 24 20:07:44 hm next year I should go to embedded world too Apr 24 20:07:59 And they're friendly and accessible. I don't know how they can create so many boards and support them though, there's about a million. Apr 24 20:08:10 they started small Apr 24 20:08:15 with one single arm7 board Apr 24 20:08:15 <_av500_> keesj: yes Apr 24 20:08:46 <_av500_> keesj: afaik uboot does usb serial too Apr 24 20:08:54 <_av500_> at least the kernel does Apr 24 20:09:09 <_av500_> unlike minix Apr 24 20:10:47 I just wonder how people will use the BBB and if they will plug a serial cable or not. Apr 24 20:11:00 <_av500_> even on BB many did not Apr 24 20:11:11 <_av500_> since it comes with web UI per default Apr 24 20:11:33 right. Apr 24 20:12:05 <_av500_> and now there is also hdmi Apr 24 20:12:12 so they are going to expect to "flash" their device over either usb or using an sdcard Apr 24 20:12:17 <_av500_> flash? Apr 24 20:12:30 <_av500_> mayn wont flash at all Apr 24 20:12:31 put stuff on the eMMC Apr 24 20:13:29 really? that would be nice. will they use a little javascript and that's it? Apr 24 20:13:38 <_av500_> I have no idea Apr 24 20:14:08 <_av500_> I guess koen or jkridner know more about the typical bone user Apr 24 20:14:25 <_av500_> once they come here, we will tell them to get a serial :) Apr 24 20:14:37 <_av500_> hmm, I should sell some on ebay Apr 24 20:15:58 <_av500_> $2.60 on aliexpress Apr 24 20:16:19 <_av500_> $1.48 Apr 24 20:18:22 they're really cheap Apr 24 20:18:32 <_av500_> yeah PL2303 Apr 24 20:18:34 but their quality follows suit on most Apr 24 20:18:49 this discussion is way too on topic! Apr 24 20:18:56 http://www.templaworld.net/wp-content/uploads/2010/11/efdakihug-925x524.jpg (NSFW!) Apr 24 20:19:42 <_av500_> um Apr 24 20:20:05 <_av500_> KotH: isnt the typical user male? Apr 24 20:20:29 in europe, almost all are male...not so much in .jp Apr 24 20:20:33 <_av500_> and the pillow of the fairer kind? Apr 24 20:20:50 <_av500_> but pillow is a good hint Apr 24 20:20:54 <_av500_> 'nächtle Apr 24 20:20:57 schlaf gut Apr 24 20:21:00 Hi is there anyway to play 1080p movies on Beaglebone black ? Apr 24 20:21:07 yes Apr 24 20:21:16 i wonder about the low res in wiki Apr 24 20:21:33 1.280 x 1.024 Apr 24 20:21:56 * KotH does the same as _av500_ Apr 24 20:22:07 night boys, girls, and n00bs Apr 24 20:22:15 ka6sox-away, +1 :) Apr 24 20:22:51 night Apr 24 20:23:10 keesj/KotH, love my olimex a13 board passionately. great quality board..nice design Apr 24 20:23:28 syntax no Apr 24 20:23:46 bone is not made for xmbc Apr 24 20:25:01 mdp: how's that DMA driver coming along? :) Apr 24 20:25:11 mdp: hovered over the "add to basket" like a fly over sh*t Apr 24 20:25:28 thx woglinde Apr 24 20:25:33 I just don't know how hard it will be to port Apr 24 20:25:58 first need the BeagleBone and BBB working Apr 24 20:46:34 Does anyone know hot to enable the ADC on the beagle bone? Apr 24 20:46:39 how to* Apr 24 20:48:20 heh Apr 24 20:48:26 AIN i assume? Apr 24 20:49:21 well I use to access it through /sys/devices/platform/omap/tsc/ain1 on the 3.2 angstrom image Apr 24 20:49:36 but under the new image from april 2013 that isn't available anymore Apr 24 20:51:08 and I can't seem to find any documentation about it under kernel 3.8 (which is what ships with the latest demo image) Apr 24 21:05:58 JG i think you may have to use IIO Apr 24 21:07:48 cmicali: that's what I thought to, and I was using the TI AM335x docs for IIO and they show how to access the raw devices, but I don't have them under the directory (e.g. raw_volt_ain1) Apr 24 21:08:32 JG: hm - i saw a patch go into the 3.8 branch that said it enabled iio for the built-in adc, but i'm not sure if it's present/enabled in the latest image Apr 24 21:08:38 JG: sorry can't be more help Apr 24 21:09:35 cmicali: well that might explain it, I've found info before the release of the latest image, and I've been looking at github for any patches Apr 24 21:09:45 maybe I can continue to hunt down Apr 24 21:09:47 https://github.com/beagleboard/kernel/commit/41dbd166a6394be45389e25459cd127854615366 Apr 24 21:09:52 that was the commit i was thinking of Apr 24 21:10:16 i would not be surprised Apr 24 21:10:31 if the patch made it in but the tiadc device in the DTS is not enabled by default Apr 24 21:11:13 JG, cmicali, yeah, that's the idea Apr 24 21:11:29 they're not enabled by default because capes might use them Apr 24 21:12:05 load the iio helper and you will get them Apr 24 21:12:22 panto: what's the recommended way to enable that - i have a local kernel build that i've just overridden stuff in am335x-bone.dts Apr 24 21:12:26 which i'm sure is probably not the right thing to do Apr 24 21:12:31 panto: I hate to ask it because I'm new to linux but do I have to recompile the kernel? Apr 24 21:12:50 the image that ships with the black should have it all included Apr 24 21:14:09 panto: so if I understand what you're saying if I enable the iio helper in the am335x-bon.dts file (assuming I set it to okay) reboot then the adc will be visible? Apr 24 21:14:18 echo cape-bone-iio >/sys/devices/bone_capemgr*/slots Apr 24 21:14:32 you don't need to modify the dts Apr 24 21:14:41 you can load the dts fragment dynamically Apr 24 21:14:56 eeebee da eeba de deeba. Apr 24 21:15:44 tl;dr - # echo cape-bone-iio >/sys/devices/bone_capemgr*/slots; cat /sys/devices/ocp.2/helper*/AIN0 Apr 24 21:16:52 I'll give it a go once I reload the image (I reverted earlier today) Apr 24 21:17:17 that's the slowpoke ADC method Apr 24 21:17:26 if you need higher performance you have to go iio Apr 24 21:17:45 I do need higher performance Apr 24 21:18:03 have to use an iio client then Apr 24 21:18:23 you still need to load the adc driver Apr 24 21:18:45 right now I have different piece of hardware in the proof that sample at 100 kSps (per channel) Apr 24 21:19:15 panto: is there any documentation about how to do that because I should probably read up on things a bit Apr 24 21:19:25 jg so you want 200k/s? Apr 24 21:19:37 Documentation/devicetree Apr 24 21:20:14 well at the very least I'd like to get 40k - 50k out of 4 channels, but I'm not sure if that's possible Apr 24 21:20:19 for the iio helper, no, but it should be pretty obvious Apr 24 21:20:56 jg i was trying to get higher rates, and i was not able to get reliable perf out of iio Apr 24 21:21:00 too tired Apr 24 21:21:02 time to go Apr 24 21:21:04 jg at least at 1msps Apr 24 21:21:20 thank you pant! Apr 24 21:21:24 panto* Apr 24 21:21:26 panto +1 Apr 24 21:21:59 cmicali: 1msps? on the bone? Apr 24 21:22:24 I thought the chip was only capable of 200 kSps (rolling) Apr 24 21:22:29 jg that's what i was trying for, tried a kernel driver/spi first, then iio, now using the PRU and it's working OK Apr 24 21:23:03 so you're able to hit 1,000,000 samples per second on a channel (multiple channels) Apr 24 21:23:04 sorry, external adc Apr 24 21:23:07 oh ok Apr 24 21:23:20 that computes ;) Apr 24 21:23:51 maybe the internal one is different, but i found too much jitter using IIO Apr 24 21:23:53 currently I'm working with an NI-9215 for the prototype, but really I think I can get by with much less sampling Apr 24 21:24:01 I see Apr 24 21:24:48 it's worth trying http://wiki.analog.com/resources/tools-software/linux-software/iio_cmdsrv Apr 24 21:24:57 this is what i used to make sure things were working Apr 24 21:25:13 then wrote a small program to sample continuously Apr 24 21:25:38 ok I'll give that a look Apr 24 21:26:19 hopefully I'll get something to work with soon as I just got the bbone, and I'm curious as to what it can do and if it'll work for me Apr 24 21:26:44 what is your use case? Apr 24 21:27:19 sound ranging (very generically) Apr 24 21:29:47 heh, this week's Amp Hour throws Cloud9 on BBB under the bus. If only I could +1 that podcast while driving ;) Apr 24 21:40:30 I wished someone from CircuitCo would respond to emails Apr 24 21:40:35 its insane Apr 24 21:41:17 heh Apr 24 21:43:17 hey you Apr 24 21:44:58 Mr. anostay! Apr 24 21:48:37 mdp: listening now Apr 24 21:50:25 ha ha. Apr 24 21:53:59 Newark showing "May 13" shipdate for a BBB Apr 24 22:15:25 _troll_: troololololol Apr 24 22:15:36 hm? Apr 24 22:16:11 mru: since when are you in ch? Apr 24 22:16:42 I'm not Apr 24 22:17:05 nor is _troll_ Apr 24 22:33:49 ch is not china, cn is... isn't it? Apr 24 22:34:05 ch is switzerland Apr 24 22:34:08 ah Apr 24 22:34:33 strange lettering Apr 24 22:34:47 not really Apr 24 22:34:54 look up the official name of the country Apr 24 22:35:04 it's switzerland :P Apr 24 22:35:08 nope Apr 24 22:35:15 yes, yes it is Apr 24 22:35:24 :P Apr 24 22:45:54 Would this screen be able to display BeagleBone video data? https://www.laptopscreen.com/English/model/Amazon/KINDLE~FIRE~HD~8.9~TABLET~32GB/ Apr 24 22:48:42 Or should I stick to something like this: http://www.chalk-elec.com/?page_id=1280#!/~/product/category=3094859&id=13939433 Apr 24 22:49:19 I'm not sure if the first one would be able to display BeagleBoard on a touchscreen... so i need some advice Apr 24 22:51:34 Any advice on building a tablet with BeagleBone Black? Apr 24 23:00:02 thurbad, Confoederatio Helvetica, Apr 24 23:00:36 nuh uh.. helvetica is a font :P Apr 24 23:00:57 itym typeface Apr 24 23:02:42 funny that the name is Latin (not one of the 4 languages spoken in the country) Apr 24 23:03:10 using one of those wouldn't be fair Apr 24 23:03:37 mru, didn't think about that Apr 24 23:38:31 how can i connect an LCD to my BeagleBone? Apr 24 23:42:34 cellotape Apr 24 23:42:39 but gaffa is better Apr 24 23:43:16 I was going to suggest wires Apr 24 23:44:21 he did say 'connect' after all Apr 24 23:44:32 i know i'm a new guy but what should I use? would this screen be ok: https://www.laptopscreen.com/English/model/Amazon/KINDLE~FIRE~HD~8.9~TABLET~32GB/ Apr 24 23:44:58 does it have a compatible interface? Apr 24 23:45:51 I'm not sure... where could I find out? Apr 24 23:46:23 I also did find this one that works with BeagleBone: http://www.chalk-elec.com/?page_id=1280#!/~/product/category=3094859&id=13939433 Apr 24 23:46:33 However the price is a little steep. Apr 24 23:46:52 I know there is a cape for the LCD screen, would that work with the first link? Apr 24 23:47:33 http://youtu.be/RRA_eVAyH5c Apr 24 23:47:49 I cod'nt see any trolls. Apr 24 23:48:06 ajain, theres LCD capes, that work out of the box Apr 24 23:48:10 buy one of htose Apr 24 23:49:06 i was thinking about those too... however i am wanting to make a tablet and these screens already have the tablet frame around it giving it the tablet look Apr 24 23:49:42 you can buy a tablet for almost nothing Apr 24 23:49:47 why build one Apr 24 23:49:56 and then you'll need to power it Apr 24 23:51:06 I want to build one because I want to be an electronics engineer and thought this would be a good project Apr 24 23:52:37 if you want to be an engineer, you should go to an engineering school Apr 24 23:52:58 Yeah, but I thought it would be a cool project Apr 24 23:53:16 have you tryed to get 2 leds to blink, running of a beaglebone? Apr 24 23:54:13 not yet i've done it on arduino Apr 24 23:54:22 well, small steps Apr 24 23:54:35 ajain: It's very cool that you want to build something, even if it's already available. Don't let naysayers put you off. Apr 24 23:54:48 blah. Apr 24 23:54:56 fiola: acehivable steps Apr 24 23:55:01 build on experience Apr 24 23:55:15 mean that such projects are buildable Apr 24 23:55:25 To fail is fine. You learn mostly from failures, far less from successes. Apr 24 23:55:27 its a core competency of being an EE. Apr 24 23:55:43 to be able to turn big problems into a set of small ones Apr 24 23:56:34 That's definitely true. Decomposing a problem is important, or most non-trivial things are unmanageable. Apr 24 23:57:11 i know its possible to build one. it can be done in raspberry pi. all you have to do is download raspbian and hook it up to the screen. you have a tablet. but i want to make one in beaglebone because i can use android. wouldn't i just have to download the Android SDK and plug in the BeagleBone into the screen, with the cape from this link: http://www.chalk-elec.com/?page_id=1280#!/~/product/category=3094859&id=13415465 Apr 24 23:57:20 for this screen: http://www.chalk-elec.com/?page_id=1280#!/~/product/category=3094859&id=13939433 Apr 24 23:57:31 hello Apr 24 23:57:37 well, try it and see what happens Apr 24 23:57:39 I need help Apr 24 23:57:53 you can install debian on teh beagle Apr 24 23:58:00 Don't know, never tried Android on my BB. Apr 24 23:58:00 raspbian is debian Apr 24 23:58:08 with a fruity flavor Apr 24 23:58:12 lol Apr 24 23:58:24 who knows about cloud9?? Apr 24 23:58:27 do you think it would work though? if not i probably won't do it Apr 24 23:58:38 well, the only way i would know Apr 24 23:58:45 would be for me to do your project for you Apr 24 23:58:47 Raspian is recompiled for hardfp. Apr 24 23:59:04 hardfp? Apr 24 23:59:09 cloud9 is not working with me Apr 24 23:59:23 @mrpackethead do you think it's at least possible? Apr 24 23:59:25 even I put port 3000 Apr 24 23:59:34 It is probably possible Apr 24 23:59:41 any one ? Apr 24 23:59:41 i cnat' see anything that woudl say impossible Apr 25 00:00:41 Hello comminity Apr 25 00:00:58 hello Apr 25 00:01:10 any one can help with cloud9 Apr 25 00:01:12 ???? Apr 25 00:01:14 ajain: http://www.gurucoding.com/en/rpi_cross_compiler/diff_hardfp_softfp.php Apr 25 00:02:09 tablets from beagles is so yesterday Apr 25 00:02:10 seems like no one wants to help Apr 25 00:02:26 I have a question to ask, can beagle black function like a tipical computer (with linux os) including GUI etc ? Apr 25 00:02:30 sorry karzan i wish i could help but i dont know a lot about cloud9 Apr 25 00:02:56 thanks ajan fro the respond Apr 25 00:03:03 people wanting to use cloud9 need a different sort of help Apr 25 00:03:07 ajain: The standard distros won't support hardfp on old architectures like ARM11 because they'd have to maintain yet another complete set of dozens of thousands of package binaries, which is why Raspbian exists, they recompiled everything for hardfp. Apr 25 00:03:34 I can not access cloud9 Apr 25 00:03:44 Gotcha. Apr 25 00:04:54 fiola: do you think i could make a tablet with android in beaglebone? Apr 25 00:05:42 Any link to small inch lcds to utilize with the board ? Apr 25 00:06:53 There is this one http://www.chalk-elec.com/?page_id=1280#!/~/product/category=3094859&id=13939433 which is a 7-inch Apr 25 00:07:32 ds2: lol Apr 25 00:07:36 danke Apr 25 00:07:38 most LCDs can be made to work Apr 25 00:07:40 That is the one I am probably going to use, but I want to know if there is one like this that is cheaper and has the tablet frame. If it doesn't have the frame could someone still post the link anyway? Apr 25 00:08:09 joel_: found out more about ETB...it is purely an addition to the JTAG stuff Apr 25 00:08:16 Id be more than happy for a how to make your LCD work tut Ds2 Apr 25 00:08:47 ds2: yeah. ETB can be read from JTAG directly.. but it is tiny, still not sure how useful. Apr 25 00:09:40 I did a simple ETB trace on a beaglebone. Apr 25 00:09:45 ajain: I have no idea. In principle, sure, but being easy in principle doesn't mean that there won't be a mountain to climb. On the positive side, the official announcement for BB Black mentioned Android, so success is available at the end. Apr 25 00:09:53 The buffer fills up with tracedata in a jiffie Apr 25 00:09:53 ajain: http://beagleboard.org/Products/BeagleBone%20Black Apr 25 00:09:56 I cant seem to find a cheap non touch small lcd, any clues ? Apr 25 00:09:59 Wuzws: how many units do you need to build? Apr 25 00:10:21 joel_: think it is like 16K...but it is one of the destinations of the ETM. you can get more data via the ETM pins Apr 25 00:10:32 how small is small? Apr 25 00:10:43 just one unit Apr 25 00:10:44 and small as in low resolution or physical size or volume or? Apr 25 00:10:47 7 inch would do Apr 25 00:10:51 ds2: it is around 32KB. have you used ETB for any purpose before? Apr 25 00:10:53 yeah 7inchs minimum Apr 25 00:10:54 7inch? that is trivial Apr 25 00:10:59 size realy Apr 25 00:11:00 *on am335x Apr 25 00:11:11 i haven't used ETB before Apr 25 00:11:17 joel_: nope. I was just talking to folks at ESC Apr 25 00:11:17 that was for ds2 Apr 25 00:11:21 sorry guys i know i don't know a lot of stuff about this Apr 25 00:11:32 7inch is next to trivial Apr 25 00:11:45 you just wire it up Apr 25 00:12:11 ds2: ok. the simple ft2232 can't read from the tracepins (emu0/1) even though they are connected Apr 25 00:12:14 ajain: follow the links, there's tons of reading material from beagleboard.org on BB and from TI on the SoC. Apr 25 00:12:17 ind be ok with less just something I can have visual feedback of the board Apr 25 00:12:26 so you'd need the jtag connector and expensive debugger, or live with the ETB Apr 25 00:12:28 joel_: yes. they run relatively fast since it is real time Apr 25 00:12:44 they were quoting like $10K+ for a device that can read it in real time Apr 25 00:13:00 ds2: for simple traces I guess its useful. I will be using tracing for a real application shortly to know how useful using just the ETB really is Apr 25 00:13:26 joel_: do you know if the ETM can place markers into the ETB? Apr 25 00:13:37 ds2: wow. I know I saw a demo somewhere where the xds560v2 read it in real time without halting ARM Apr 25 00:14:02 ds2: like time stamps? yes the tracedata is timestamped by hardware Apr 25 00:14:03 joel_: jtag isn't fast enough to do it continously for a 1GHz processor Apr 25 00:14:18 no more like - BEGIN trace here, STOP here Apr 25 00:14:24 so I can use it to trace a process or a function Apr 25 00:14:29 thanks fiola i will definitely check it out Apr 25 00:14:46 or some way of triggering and stopping it Apr 25 00:14:52 ds2: yeah you can tell the ETM start and stop address, that triggers the trace instead of just tracing everything and flooding the buffer Apr 25 00:15:07 OOhh Apr 25 00:15:14 seems interesting enough to look into Apr 25 00:15:21 ds2: yeah Apr 25 00:16:06 ds2, sounds about like the price of a RISCTrace that did the same for IBM PPC.. Apr 25 00:16:27 TI's STM implementation can do much more.... L3 introspection, DMA profiling, SoC event collection, SW messaging etc Apr 25 00:16:38 is it possible to boot windows os on the board ? Apr 25 00:16:38 would beaglebone black be too slow to run android? Apr 25 00:17:00 mdp: was that really for the PPC or was it a hold over from RS6000? Apr 25 00:17:23 Beagle bone white runs android just fine Apr 25 00:17:45 but you're stuck with the ETB without a $$$ debugger and also w/o debug software capable of using those features Apr 25 00:17:56 what do you mean joel? Apr 25 00:18:03 i'm sorry i just don't understand Apr 25 00:18:12 I mean its all av500's fault Apr 25 00:18:31 * joel_ pokes av500 Apr 25 00:19:18 ds2, holdover, of course, the name is a giveaway Apr 25 00:19:25 ds2: how are you going to be using ETM? Apr 25 00:19:40 ajain: that's a different conversation, joel's talking about low level processor tracing with JTAG and ETM/ETB, not related to Android. Apr 25 00:19:53 joel_: time machine Apr 25 00:20:00 joel_, yeah, STM (regardless of vendor) is far more useful than the old ETM stuff Apr 25 00:20:32 i see. sorry guys. Apr 25 00:20:47 ds2: actually STM uses the ETM.. ETM is internal to the ARM cortex-a8 and is still used, just newer version (as can be seen in the ARM cortex-a8 TRM) Apr 25 00:21:05 so for tracing STM still needs the ETM's help.. Apr 25 00:21:13 joel_: yeah, I figured that Apr 25 00:21:43 the key being the standard MIPI trace support Apr 25 00:23:06 is there off the shelf PRUSS code to generate NTSC? Apr 25 00:23:12 so i should probably stick with this screen then: http://www.chalk-elec.com/?page_id=1280#!/~/product/category=3094859&id=13939433 Apr 25 00:23:18 joel_, did TI ever open any of their decoders as a part of the linaro STM support effort? Apr 25 00:23:51 mdp: debuggers you mean? Apr 25 00:24:57 somebody I know that worked closely with this stuff at linaro had mentioned the message decoder library etc. TI has Apr 25 00:25:14 TI is working on supporting STM in near future but so far nothing works (atleast on AM33xx) except for ETB. You can't access the traceport (yet) for am33xx Apr 25 00:25:48 not talking about am33xx stuff Apr 25 00:25:56 the stuff where all the work was done already Apr 25 00:26:11 http://processors.wiki.ti.com/index.php/STM_Linux_Device_Driver Apr 25 00:26:56 nm! fount it ;) http://processors.wiki.ti.com/index.php/CToolsLib#Linux_Kernel_Driver_for_STM_printf Apr 25 00:27:06 or Apr 25 00:27:13 http://processors.wiki.ti.com/index.php/CToolsLib Apr 25 00:27:25 yes CToolsLib Apr 25 00:27:56 the linaro project looks stale still Apr 25 00:28:00 guess everybody lost interest Apr 25 00:29:10 I think it is more of a resourcing problem Apr 25 00:29:30 same thing Apr 25 00:30:28 what would really be interesting is using this for low-latency printk Apr 25 00:31:03 and the other usecase is for ex, looking at an ETM trace after crash Apr 25 00:31:09 when is printk latency important? Apr 25 00:31:11 no one else is in interested in using it as a time machine? Apr 25 00:31:27 warm reset, and just read the ETB to see last instructions executed before the crash Apr 25 00:31:42 yes Apr 25 00:31:48 but on a preprocess basis Apr 25 00:32:21 ds2: to enable logging to a circular buffer and not worry about performance Apr 25 00:32:35 joel_, sure, that's the same stuff we were doing 10 years ago on other processors Apr 25 00:32:47 standard debug practice for tough issues Apr 25 00:32:53 joel_: was thinking more of a way for the SEGV handler to dump it Apr 25 00:33:09 getting where you crashed is easy Apr 25 00:33:16 figuring out how you got there is the nightmare Apr 25 00:33:32 joel_, tools for hardware trace usability are always welcome ;) Apr 25 00:34:09 linux does trace dumps on various errors on some systems Apr 25 00:34:19 blackfin for instance Apr 25 00:34:22 very useful Apr 25 00:34:33 trace dumps? besides the usual stack traces? Apr 25 00:34:37 mdp: yes, who wouldn't love cycle accurate function profiling and tracing, all low/non-intrusive Apr 25 00:34:46 ds2: branch trace Apr 25 00:34:46 i.e I can go back into the last 100 or so lines that the code traversed? Apr 25 00:35:07 Oohh that is useful Apr 25 00:35:10 yes, you can see exactly how you got there Apr 25 00:35:16 how deep on BF? Apr 25 00:35:18 joel_, it's necessary as most hard problems can't be debugged via stop mode Apr 25 00:35:27 joel_, as they disappear Apr 25 00:35:33 don't remember how large the buffer is Apr 25 00:35:38 100 entries or so Apr 25 00:35:47 not bad Apr 25 00:35:55 jkridner: you around? :D Apr 25 00:35:59 that can cover a function or two Apr 25 00:36:30 mdp: indeed. can be very useful for that type of debug Apr 25 00:36:32 joel_, we caught an coherency/dma/cache alignment issue once this way that took 5-7 of runtime to trigger Apr 25 00:36:47 it's the type of stuff that keeps a production line down ;) Apr 25 00:36:53 mru: how does it work? what causes the branch information export (in HW)? Apr 25 00:37:03 a trace unit in the cpu Apr 25 00:37:06 s/5-7/5-7 days/ Apr 25 00:37:20 that's the kind of stuff that gets you sent onto a plane with an hour's notice :/ Apr 25 00:37:46 ds2, sounds like a common memory Apr 25 00:37:58 A quick newbie question, can I run beaglebone black with windows os and just utillize it as a pc ? Apr 25 00:38:05 no Apr 25 00:38:22 has to alsways be conected then ? Apr 25 00:38:36 well its still a lot like what the ETM exports, branching info. would be useful to dump. I did see an ETM driver in the kernel that seems to be enabled only for a certain OMAP emu target. Apr 25 00:39:00 any other os with GUI ? Apr 25 00:39:03 yeah, the etm does everything you need Apr 25 00:39:09 mdp: that's interesting Apr 25 00:40:01 arch/arm/kernel/etm.c Apr 25 00:40:54 Hey if I'm interested in working with you guys during gsoc where would be a good place to start (code-wise)? Apr 25 00:41:12 s/if// Apr 25 00:41:25 come up with a concept Apr 25 00:41:47 I like the ideas that you guys came up with on the ideas page Apr 25 00:42:01 but the mentors don't seem to be here atm Apr 25 00:45:13 There is a CONFIG_OMAP3_EMU (arch/arm/mach-omap2/emu.c) that registers the device. So far looks like the only user of the etm driver. Apr 25 00:48:25 mru: just storing the branch info seems useful though instead of every instruction in the path Apr 25 00:49:03 bf's limited to last 16 branches, still Apr 25 00:49:33 not really Apr 25 00:49:40 on ARM you do need instructions Apr 25 00:49:55 conditional execution will do you in otherwise Apr 25 00:50:52 well, when you've limited etb. its a plus Apr 25 00:51:52 depends on debug usecase really... I will over come my fear of the limited etb soon when I see how it plays out Apr 25 00:53:16 ds2: i see what you mean, k Apr 25 00:54:15 any hangups with the black that people are voicing concerns about? Apr 25 00:54:26 * shouldi as in should i buy one Apr 25 00:56:43 yes.... it takes longer then 24hours to get one right now Apr 25 01:00:17 ds2: usecase for ETM time machine Apr 25 01:00:46 ask for one or you going to enumerate them? Apr 25 01:01:05 no I said that could be one, to take you forward 24 hours Apr 25 01:03:39 Hello. I have installed Ubuntu with GUI-cape in my beagleboard xm. When i run "X :1" command for init x-server i obtain a freeze in "loading extension glx" output Apr 25 01:04:12 can somebody help me? Apr 25 01:12:46 [6~[6~[5~[6~[6~[5~[6~[5~[5~[5~[5~[5~[5~ Apr 25 01:13:05 gosh. irc client going nuts here. Apr 25 01:13:06 bbl Apr 25 01:29:12 yo? Apr 25 01:31:05 * mranostay sips beer Apr 25 01:31:08 CHaa: yes? Apr 25 02:43:29 runtime pm can eat up cycles like no one's business Apr 25 02:44:07 hi joel_ Apr 25 02:45:13 mranostay: hey matt **** ENDING LOGGING AT Thu Apr 25 02:59:59 2013