**** BEGIN LOGGING AT Tue Sep 22 02:59:59 2015 Sep 22 04:12:50 Has anyone got node-wekbkit to work on BeagleBone Black Debian Wheezy? Sep 22 05:16:39 hello Sep 22 05:16:46 i have a doubt Sep 22 05:17:03 i want to bye beagle bone black board Sep 22 05:17:37 I am a self dependent researcher and developer technologist Sep 22 05:18:26 i will publish my research work after implementing with beagle boad in reputed journal Sep 22 05:18:37 is there any problem with license Sep 22 05:18:41 ??' Sep 22 05:19:11 it depends quite a bit on what you're doing... Sep 22 05:19:35 "is there any problem with license" is a very broad question that is essentially impossible to answer. Sep 22 05:20:11 is it open source? Sep 22 05:34:22 is what open source? Sep 22 05:38:29 reverse engineering sgx blobs -> problematic (though kinda cool). reverse engineering the kernel, or glibc -> unproblematic (but rather böring) Sep 22 05:38:39 is there more to it? :) Sep 22 06:25:12 I want to play mp4 vedio on beaglebone black with debian os. Suggest the best mp4 player. Sep 22 06:27:25 Suggest mp4 player which can be run from command line. Sep 22 06:47:14 wahahaha... "Remember: Small is beautiful!" (on the U-Boot site) Sep 22 07:19:28 Lalit_: ffplay/avplay Sep 22 07:19:37 from ffmpeg/libav packages Sep 22 11:11:46 lol... http://haskell.spreadshirt.com/warm-fuzzy-thing-A6501562 Sep 22 11:15:55 * zmatt tries to reconstruct how that definition worked... Sep 22 11:17:44 ah, indeed it is Sep 22 12:18:19 Hi good evening to all !!! Sep 22 12:19:02 Have any one tested bbb android kitkat ? Sep 22 12:19:45 so far only indians try to use android with bbb Sep 22 12:20:00 they drop in, ask the very same question you just did and disapear again Sep 22 12:20:06 so, in the end, we don't know Sep 22 12:21:54 bbb kitkat available ? Sep 22 12:22:05 i am unable to find Sep 22 12:22:18 any source Sep 22 12:23:39 narasimha: there is kitkat source from google Sep 22 12:23:47 there is a TI effort to make android run on BBB Sep 22 12:23:52 but it might not have kitkat Sep 22 12:23:58 google for "android rowboat" Sep 22 12:26:40 any help Sep 22 12:26:52 for bbb android Sep 22 12:26:54 yes, you were offered help, did you read it? Sep 22 12:26:57 kitkat support Sep 22 12:27:03 asking the same question again is insulting Sep 22 12:27:21 sorry Sep 22 12:28:02 how to add usb auto mount support in android Sep 22 12:28:46 any one have idea on this can suggest please Sep 22 12:29:33 narasimha_: you should ask in some android channel Sep 22 12:40:37 Hi, reading TRM the interrupt number of ADC_TSC_GENINT is 16 (to register a handler on adc's IRQ) but nothing is mentionned about which ADC IRQ is connected to this "ADC_TSC_GENINT" ? I mean i'm missing something, what's the link between this IRQ number and the ADC ? I'm looking to register an handler on FIFO1TRESHOLD IRQ Sep 22 13:04:35 What exactly is triggering ADC_TSC_GENINT interrupt ? Sep 22 13:17:09 it's the combined irq from the ADC Sep 22 13:18:07 it's asserted if any of the ADC's irqs is asserted and enabled Sep 22 13:18:19 except the async pendown event I think, not sure Sep 22 13:18:53 this isn't really specific to the ADC, nearly every peripheral uses an irq combiner Sep 22 13:20:19 Hum, ok, so if any of the IRQSTATUS register from the ADC is up this will trigger the IRQ n°16 Sep 22 13:20:30 if you enable it, yes Sep 22 13:21:29 all newer "Highlander 0.9 compliant" (apparently some TI internal standard) peripherals have four registers related to the irqs: rawstatus, status, enable, disable Sep 22 13:21:42 I'm not sure but i've read somewhere that i've to add 16 to the irq nbr to register the handler in the kernel module, is this true ? registering IRQ nbr 16 return error but registering 32 is working, but my handler seems never called Sep 22 13:22:10 adding 16 sounds like advice for a Cortex-M Sep 22 13:22:40 there irq n maps to vector n+16 (the first 16 are used for exceptions and other cpu-internal stuff) Sep 22 13:22:53 but that doesn't apply to the cortex-A Sep 22 13:24:06 but you should leave irq numbering to device tree... your module should just ask for its irq and the kernel will arrange for it Sep 22 13:24:15 But i'm facing an error when calling request_irq on number 16 Sep 22 13:24:35 did you make sure the normal tscadc driver is _not_ loaded? Sep 22 13:25:08 Ye, i'm writing my own, actually bind on timer4 and seems working Sep 22 13:25:57 cat /proc/interrupt list no IRQ 16 or 32 Sep 22 13:27:37 I don't really know how irqs work in linux, but I do know you shouldn't be hardcoding a number in the driver; declare it in your device-tree node Sep 22 13:28:34 override the "compatible" property of the adc with one that your driver matches on Sep 22 13:28:51 Ye i will but i'm actually testing the driver Sep 22 13:28:55 (or just replace the whole node) Sep 22 13:29:09 Hum ye good idea i'm going to look at the device tree wrote for the original driver Sep 22 13:29:33 if you didn't override or remove it, most likely the kernel is reserving the irq declared in that node Sep 22 13:30:07 I'm guessing you're then also manually enabling the adc's clock in your driver now? Sep 22 13:31:01 (since the kernel won't do that for you if it doesn't know your driver is associated with the DT node for the adc) Sep 22 13:31:30 I'm actually doing this manually yes Sep 22 13:31:58 I know it's not a good practice but i'm doing some test Sep 22 13:32:39 I understand, and I'm not shy of bypassing the kernel if I need to Sep 22 13:33:40 but if you're writing a driver anyway, using DT is actually the easy way: you just ask for your resources (clocks, memory regions, irqs) and the kernel will dig in device-tree to see what they are and get them for you Sep 22 13:34:05 and it avoids accidental conflicts Sep 22 13:37:04 Ok thanks :) Sep 22 13:37:30 btw you can also test from userspace Sep 22 13:38:10 using uio_pdrv_genirq Sep 22 13:39:20 then you also declare one or more memory regions in device tree, and an irq, and you get a device which you can open in userspace, mmap() the memory regions, and read() to wait for an irq Sep 22 13:49:50 Gwak: http://pastebin.com/GrHwgYiR Sep 22 13:53:24 if you also need to interact with other parts of the kernel (e.g. timers) or need DMA then you'll need a kernel module anyway, but for testing with a peripheral this is easier than a kernel module Sep 22 13:54:43 (the device tree node in that example reuses the existing tscadc node, which already declares the associated irq and other hardware resources) Sep 22 14:40:50 I'm trying to bind a handler on the interrupt 16 (adc_tsc), but i've "16: 3770 INTC 68 Level gp_timer" and registering with IRQF_SHARED return device or ressource busy, why a timer is binded on this interrupt ? Sep 22 14:45:47 and what's the number after INTC ? the 1st number is the interrupt nbr isn't it ? Sep 22 14:52:08 Gwak: because interrupt 68 is indeed timer 2 Sep 22 14:52:22 the 16 seems to be some internal number used by linux Sep 22 14:52:26 it's not the irq line Sep 22 14:53:39 looking at /proc/interrupts it is actually clear that there's no relationship between the irq numbers dynamically allocated by linux and the physical irq lines Sep 22 14:54:42 the linux ones are also global over all irq controllers (it views GPIO peripherals also as irq controllers to allow individual GPIOs to be used as irq lines) Sep 22 14:55:03 which means that whatever function you're using to request an irq is the wrong one Sep 22 14:56:16 Hum ok I thought the 1st number was the IRQ number and this explained why i've error 16 (ressource busy) when registering the handler Sep 22 14:57:04 request_irq(IRQ_ADC, adc_handler, IRQF_SHARED, THIS_MODULE->name, NULL); Sep 22 14:57:13 this is wrong ? ... Sep 22 14:57:19 IRQ_ADC = 16 Sep 22 14:58:45 if you use device tree then you use platform_get_irq(pdev, 0) to get the irq number to pass to request_irq Sep 22 14:58:59 How can this return ressource busy and nothing appears in /proc/interrupts related to interrupt 16 .. Sep 22 14:59:30 the first argument of request_irq is wrong Sep 22 15:00:09 ok i'll try to bind the DT correctly but i doubt this will solve the problem because interrupt nbr in DT is 16 too Sep 22 15:00:34 ? it's not the IRQ nbr for the 1st arg in request_irq Sep 22 15:01:30 no, 16 is the irq number associated with the interrupt controller (also defined in DT using interrupt-parent, though I think it allows that property to be omitted for backwards compatibility) Sep 22 15:02:21 also, apparently one should use devm_request_irq instead of request_irq .. according to these slides http://haifux.org/lectures/323/haifux-devres.pdf Sep 22 15:04:09 Thanks I was reading old stuff, I can't wait for 4th edition of "Linux Device Driver" everything seems outdated for kernel 3.8+ and 4 Sep 22 15:04:10 >.< Sep 22 15:04:11 anyway, hardcoding the irq number also means you somehow have to locate the interrupt controller and have it map allocate a linux irq number for it for you Sep 22 15:04:36 yes I also noticed documentation tends to lag behind the rapid kernel development Sep 22 15:05:31 I've actually given up on trying to understand device driver development for now, so far I can do everything I need to from userspace Sep 22 15:08:23 Ye but for jitter and speed trouble i've to make my own driver for the BBB, I've intensely tried to do it the proper way but books are old, and device tree are quite annoying and not well documented Sep 22 15:09:21 I'm going to bind my driver on the adc tsc description in the DT I think this will solve a lot of my problem, thanks for your help Sep 22 15:09:42 my approach to adc performance was configuring DMA Sep 22 15:12:13 apparently irq latency is pretty bad in the kernel too, although the -rt kernels apparently reduce it a lot Sep 22 15:14:05 i'm using a PREEMPT enabled kernel, and binding the ADC on TIMER4 solve the jitter problem, I will see but I think I may not need the RT patch Sep 22 15:15:45 you use a timer to start the adc? why not just insert appropriate delay in the adc config and set it to run continuously? Sep 22 15:17:09 I use the timer4 to trigger the HW_event on the adc and start the sampling on 6 analog input, this every 1ms Sep 22 15:17:20 Hey guys, I'm looking at https://github.com/beagleboard/image-builder and something doesn't look quite right. Is tehre anyone on that knows a bit more about it, and can explain something to me? Sep 22 15:18:08 I can do this without a timer ? Sep 22 15:22:02 Gwak: sure, if you configure the step timings right, then set it to SW enabled, continuous Sep 22 15:23:31 but regardless of whether you do it that way or use a timer as hw trigger, receiving the adc fifo irq and reading the data from the fifo is no longer timing-critical Sep 22 15:24:04 so I'd personally use uio for that, assuming the final destination of the data is in userspace and not in the kernel Sep 22 15:26:56 according to my notes, total time of a step is open_delay + 2^averaging * ( 15 + sample_delay ) where open_delay, sample_delay, and averaging are the values configured in the registers Sep 22 15:27:25 (in adc cycles, which is 24 MHz / ( 1 + adc_clkdiv ) Sep 22 15:27:28 ) Sep 22 16:35:34 hello everyone...I have a troubleshooting question: I've connected a mouse and keyboard to the BBB, which is connected to my desktop. The signals are going straight through to the PC instead of the BBB. Any ideas? Thanks! Sep 22 16:37:47 how are your mouse and keyboard connected to the BBB Sep 22 16:38:40 via unpowered usb hub Sep 22 16:38:57 How is your BBB connected to the computer? Sep 22 16:39:17 And the hub Sep 22 16:39:18 via the usb power cable. Sep 22 16:39:43 the keyboard mouse is connected to the hubm, which is connected to the BBB, which is connected to the desktop Sep 22 16:40:17 hmmm, the hub is not connected to the desktop is it? Sep 22 16:40:22 nope Sep 22 16:40:22 ehh, are you sure there's no mixup of cables involved? since the BBB isn't even *capable* of passing through those signals even if you wanted to Sep 22 16:40:46 zmatt: that's what I thought, too. but it's definitely connected the way I describe. Sep 22 16:41:10 o.O Sep 22 16:41:13 It is possible there is a usb_gadget driver being loaded, but I doubt it... Sep 22 16:41:24 awww crap...on triple check it looks like it's totally my fault. Sep 22 16:41:35 abferm: would still require an elaborate non-default setup Sep 22 16:41:37 hang on...update in like 60 seconds Sep 22 16:41:42 radams78: ghehe Sep 22 16:43:02 well, that explains why I had enough power...cable mix up. accidentally connected the hub to the desktop when messing around with stuff...Thanks everyone! Sorry for the newb error. Sep 22 18:25:53 https://github.com/beagleboard/image-builder doesn't appear to put a uEnv.txt file in the tarballs it creates... am I doing something wrong? Sep 22 19:16:35 abferm: rcn has a lot of repos, it's not always clear which is for what :P Sep 22 20:02:18 kilroi: I would assume that the image-builder would make a fully functioning image... Sep 22 20:04:44 Woops, sorry kilroi, wrong nick Sep 22 20:04:59 zmatt: I would assume that the image-builder would make a fully functioning image... Sep 22 20:23:31 abferm .. for most it does :D Sep 22 20:27:15 hi Sep 22 20:28:09 veremit: hmm... I got it to finish the process using the eewiki_bare_debian_jessie_armhf config file, but there was no uEnv.txt file in /boot/. using the bb.org-debian-jessie-console config file I get to the very end and it exits with code 1, but no error messages, and nothing in ./deploy/ Sep 22 20:28:42 hmm.. haven't used it myself alas .. I just use the pre-built stuff Sep 22 20:28:59 anyone here works with video applications? Sep 22 20:29:03 I probably should soon .. RN's auto builds are a bit bloaty Sep 22 20:29:12 marcelos .. meaning what? Sep 22 20:29:25 veremit: any kind of video processing Sep 22 20:29:39 you realise thats a really broad subject .. lol ? Sep 22 20:29:40 either on beagleboard or another platform Sep 22 20:30:09 well, i work specifically with objective video quality evaluation... it's a bit specific Sep 22 20:30:19 that's why i look for video applications in general Sep 22 20:31:01 like .. my company has done ip network camera development with a ti dm368 processor using h264 compression .. but thats just one of a plethora of potential "video applications" Sep 22 20:31:09 i see Sep 22 20:31:13 yeah, that's true Sep 22 20:31:46 and, incidentally .. although there are c ouple of TI nuts that would understand the dm368's existence .. most would not :) Sep 22 20:31:57 basically, i pick up some video sequence and apply a algorithm (based on the pixels luminance) to retrieve a quality score Sep 22 20:32:36 i did it on pc, also on hardware (an fpga board, using vhdl) Sep 22 20:32:50 and now i'm looking for an ARM-based platform to do it Sep 22 20:32:51 the am335x processor doesn't have a lot of video hardware built-on Sep 22 20:33:08 and not a whole lot of fast cpu processing capability Sep 22 20:33:15 certainly not comparable to high-speed fpgas Sep 22 20:33:15 oh, i see Sep 22 20:33:29 however .. I have seen an fpga 'cape' Sep 22 20:33:45 so you'd just have to figure (an) interface(s) Sep 22 20:33:46 i've been considering the beagleboard because of the price and big community support Sep 22 20:34:16 what resolution video are you using? or hoping for? Sep 22 20:34:47 you might be better with a arm board with some form of camera interface, possibly? Sep 22 20:35:08 camera is not necessary Sep 22 20:35:09 and multiple cpu cores or some video acceleration hardware Sep 22 20:35:15 well... to my research, one of the objectives is to evaluate up to what resolutions i can process in real time with some given processor Sep 22 20:35:25 at maximum 1920x1080 pixels Sep 22 20:36:40 the videos that we use for the research at disponible in some video databases. it is kind of necessary to use those specific videos Sep 22 20:36:46 are disponible* Sep 22 20:36:55 if you want to try the beagle .. you'll definitely want to look at the PRU (programmable real-time units) which are special embedded cores in the processor Sep 22 20:37:04 I think the Raspberry Pi has aditional video hardware. Sep 22 20:37:09 hum... i see Sep 22 20:37:19 they are equivalent on price, right? Sep 22 20:37:26 close Sep 22 20:37:26 abferm .. yes I was thinking go for the pi2 with the quad-core Sep 22 20:37:38 as a low-cost starting point Sep 22 20:38:03 I haven't looked at the pi2... quad-core sounds nice. Sep 22 20:38:12 same price .. much better hardware Sep 22 20:38:16 boards specific for video start at some $300, so i think of a beagle or rasp to do some preliminary tests Sep 22 20:38:46 marcelos .. a good plan .. but be prepared to be disappointed with results .. these boards dont have a lot of 'grunt' :) Sep 22 20:38:59 $200 would be my starting point, personally Sep 22 20:39:10 i see veremit Sep 22 20:39:23 but it depends on the resolution/complexity of your video/algorithm Sep 22 20:39:28 won't be avery good test if you are planning on using video specific boards in the future. Sep 22 20:39:39 abferm .. my thoughts precisely Sep 22 20:39:39 i see Sep 22 20:40:01 the hardware required is very different Sep 22 20:40:08 well, i think that's correct Sep 22 20:40:20 if you are just testing your algorithm it sounds like you have done that already. Sep 22 20:40:27 hence a good start is also the fpga route Sep 22 20:40:39 the purpose of my research is to compare the pc/arm/fpga approaches Sep 22 20:41:02 its a legitimate project :) Sep 22 20:41:34 :D thanks Sep 22 20:41:42 it's my masters thesis Sep 22 20:41:55 imo stick with the pi, the hardware video stuff will probably help Sep 22 20:41:55 my undergraduate project was the fpga implementation Sep 22 20:41:55 In that case, you would be fine comparing something that doesn't have video acceleration. That would make it more of a true comparison. Sep 22 20:42:15 oh this is for a thesis nvm Sep 22 20:42:32 That's my opinion anyway. Sep 22 20:43:00 that's a complex point abferm Sep 22 20:43:02 at least for me Sep 22 20:43:17 don't know yet what to do in this matter Sep 22 20:43:45 i'm beginning to study software profiling Sep 22 20:43:47 I also don't know what your funding looks like. Sep 22 20:44:07 my professor said that $300 is kind of ok Sep 22 20:44:17 at least for now Sep 22 20:44:31 I can tell you the dm365/8 is a good video chip for HD video processing :) Sep 22 20:44:32 i'm thinking of this http://microzed.org/product/microzed-embedded-vision-kits Sep 22 20:44:44 A lot of the acceleration hardware is fpga based if I'm not mistaken, meaning you would be comparing fpga to arm+fpga. Sep 22 20:44:50 but the software sdk is a nightmare Sep 22 20:45:33 that's important too... i need a good reference design Sep 22 20:48:00 beagle fpga cape - http://www.element14.com/community/docs/DOC-69129/l/logi-fpga-development-boards-by-valentfx Sep 22 20:48:15 dunno which fpga you would have targetted before .. interesting they also do a Pi version too Sep 22 20:48:24 relatively inexpensive as a try-out Sep 22 20:49:49 interesting, didn't know about this Sep 22 20:51:06 its a cheat really though .. because the processing is done external to the main cpu .. which I think defeats the object of your project :) Sep 22 20:51:46 hahaha yes, kind of Sep 22 20:51:51 which, should, in a nutshell, prove that arm processors aren't good at real-time, high-resolution video processing :) Sep 22 20:52:07 i'll try pure arm and fpga approaches Sep 22 20:52:13 unless they have dedicated hardware to do so .. which is in itself an interesting sub-project Sep 22 20:52:20 yeah, i think that's what i will get in the end Sep 22 20:52:32 but need to do the experiments to confirm the hypothesis Sep 22 20:52:40 but you might have to bite the bullet with the more expensive arm boards :) just fyi! Sep 22 20:53:11 hahaha yeah, that's true. i'll go for a big one then Sep 22 20:53:31 its not so bad if you're not footing the bill lol Sep 22 20:54:30 i'm still to discuss the exact values, but i think i can run for a big one Sep 22 20:54:46 even with the us dollar getting expensiver here Sep 22 20:55:14 well, i got go for now Sep 22 20:55:35 you guys helped me more than did my advisor in the last months. thanks :) Sep 22 20:56:51 welcome Sep 22 20:57:01 been there, also lol Sep 22 20:59:30 ok bbl ... gotta do some stuff :) good luck with your project Sep 22 21:17:43 marcelos computer science grad school 101: any time you have the opportunity to spend someone else’s money to buy toys, do it. Buy the most expensive, awesome toy you can Sep 22 21:18:25 example: one of the other grad students got a fancy graphics card, ostensibly to do some research project with CUDA Sep 22 21:18:35 he really got it so he could play Oblivion on his lab box Sep 22 21:19:06 but he got a paper out of it so it was ok Sep 23 01:57:03 Hi, i'm new using BBB, i need to make a dehydrator from 0 to 100 anyone know some about this ?? **** ENDING LOGGING AT Wed Sep 23 02:59:59 2015