**** BEGIN LOGGING AT Tue Apr 17 03:00:00 2012 Apr 17 04:28:03 I want to use a 3G or 4G usb adapter with the beagle to have a stable wireless internet connection. Any ideas about how to best accomplish this? Apr 17 04:29:24 iMammal: er does the adapter have linux support? Apr 17 04:45:13 so is using poll() the accepted way to do interrupts in Linux on BeagleBone? Apr 17 04:53:14 poll by definition is not an interrupt Apr 17 04:57:16 heh Apr 17 04:57:21 poll != interrupt Apr 17 05:00:49 yea...i know. i'd prefer actual interrupts too Apr 17 05:01:00 mranostay: I'll have to find one that does. I'd had cell phones that had linux ACM support but I want to use a usb stick for stability. Apr 17 05:01:28 just can't seem to locate any info on using them in beaglebone. has anyone done this with input gpio? Apr 17 05:02:47 I'd be open to using a boat with a built in 3g or 4g radio if anybody knows of such a thing. Apr 17 05:02:57 s/boat/board/ Apr 17 05:04:14 nemik: wondering that myself Apr 17 05:04:24 tried gpio_request but that doesnt esem to work Apr 17 05:05:48 iMammal: most should work, these instructions (except the pacman stuff) should work on Angstrom too https://wiki.archlinux.org/index.php/USB_3G_Modem Apr 17 05:12:45 f00bard: yea been reading for a while now and can't find much about it either. everyone just seems to use polling or there's some talk about the PRUSS but i don't understand it Apr 17 05:15:06 seems like sprint has a number of usb modem sticks that have linux support. Apr 17 05:16:01 https://wiki.archlinux.org/index.php/USB_3G_Modem Apr 17 05:29:06 iMammal: as long as it isn't a x86 binary blob Apr 17 06:50:58 is there a way for Gpio to work on 1.8v instead of 3.3 ? (i have a device pluged on adc which i give 1.8v) Apr 17 06:58:16 http://www.sparkfun.com/products/8745 Apr 17 07:01:21 aholler: thanks Apr 17 07:05:29 I've just bought a BB-xM, but my screen doesn't have HDMI port. Can I use COM port (RS232) to display on the monitor via terminal? Apr 17 07:42:33 Hi, anyone knows if I can use the Sitara SDK (targeted for windows) in CCS ona windows machine? Apr 17 07:42:57 ....sorry (targeted for Linux)* Apr 17 07:46:03 yes, but you can also poke your eyeballs with sharp sticks Apr 17 07:46:13 that something is possible doesn't mean it's a good idea Apr 17 07:46:20 if you target linux, use linux Apr 17 07:50:13 I don't target Linux but Non-OS development instead using CCS on a windows. Starterware doesn't support all that much for the beaglebone currently and I thought that maybe I could benefit from using the Sitara SDK for Linux...Am I completely off? Apr 17 07:50:49 slightly Apr 17 07:50:53 ok Apr 17 07:52:25 My main problem is that I need to get McASP running on the Beaglebone and the recipe for doing that in the Starterware documentation is meant for the EVM and not the Beaglebone so I can't follow that recipe :( Apr 17 07:56:51 For example following this recipe: http://processors.wiki.ti.com/index.php/StarterWare_McASP ....Starterware for Beaglebone doesnt support McASPPinMuxSetup() Apr 17 07:58:43 why should mcasp differ between bone and evm? Apr 17 07:58:49 its the same cpu Apr 17 08:07:18 in my CCS project I need to include both soc_AM335x.h and the evmAM335x.h and I get an error...I'm completely new with BB and CCS so maybe this is easily remedied Apr 17 08:11:28 In the evmAM335x.h the McASPPinMuxSetup() is listed as a protype function and it isn't in the soc_AM3358.h Apr 17 08:35:28 then steal it Apr 17 09:51:08 hmmm Apr 17 09:51:24 fixing beaglebone regressions turns out to be an exercise in reverting vendor patches Apr 17 09:51:49 dont buy patches from vendors Apr 17 09:55:30 hehe Apr 17 09:56:41 patches! fresh patches! finest produce! applies fine! compiles well! lowest prices! 26 each, buy two, pay 54! Apr 17 09:56:57 can I take an option on that deal Apr 17 09:57:52 I have those fine collateral patch obligations, would you be interested in those? Apr 17 09:57:55 how hard is it to enter "DM3730 TRM" into google? Apr 17 09:58:14 dm8tbr: not "collateral damage" ? Apr 17 09:58:16 N-P-hard? Apr 17 09:58:50 koen: also, I dont get your remark, you mean bitbake does not support ftp? Apr 17 09:58:55 but first you need to find some evil unlocking tool which voids the warranty. Apr 17 10:00:03 av500: only gopher Apr 17 10:01:31 hmm, I should release our GPL stuff on CD-R on demand Apr 17 10:01:38 Hi I would like to check my understanding of C6runlib. I have 3 files main,lib A and lib B main - calls lib A function lib A - use C6Run_malloc and call lib B function (This lib runs on ARM but comunicate with dsp) lib B -can access to mem from dsp side (Runs on dsp) is that right ? main use gcc compiler and lib A, lib B c6run compiler ? Apr 17 10:01:41 I wonder how many would "order"... Apr 17 10:03:14 root Apr 17 10:03:14 * dm8tbr has ordered several times and paid in 'pieces of sweet' ;) Apr 17 10:03:15 uups Apr 17 10:03:33 people I know have actually killed their machines with GPL Warez :-D Apr 17 10:03:41 dm8tbr: I have many moar CD-Rs , so please send Apr 17 10:03:45 so you would probably find you CDR "pirated" Apr 17 10:03:48 :p Apr 17 10:04:00 XorA: damn, can we prevent that? Apr 17 10:04:31 offer a boot(loader) from usb or sd Apr 17 10:04:32 hmm, I have a Click drive, I will send it out on that Apr 17 10:04:53 av500: like the dreamcast record the disks backwards? Apr 17 10:05:08 maybe Apr 17 10:05:42 XorA: what happens if you play such a disk forward? Apr 17 10:05:55 Satan Satan Satan Apr 17 10:06:01 :] Apr 17 10:06:37 I guess that means 1/GPL = BSD Apr 17 10:10:39 * XorA boots Banshee in the stones! Apr 17 10:15:07 * XorA wonders why linux coders fail so badly in doing a ls -R Apr 17 10:16:02 ls -hard Apr 17 10:16:52 I now seem to have 3-4 copies of every track in Banshee, 2-3 of them unplayable Apr 17 10:17:08 it seems every time I start it, it just adds more copies Apr 17 10:20:43 so its like Itunes Apr 17 10:21:04 never used itunes, but I heard the horror stories Apr 17 10:30:57 XorA: from time to time I update it and try to have it index my music collection Apr 17 10:31:10 but it seems that 300GB is NP hard to do Apr 17 10:31:58 other "media players" manage the scan, but then are utterly unusable afterwards Apr 17 10:33:15 I tend to use foobar2000 which scans my whole collection in seconds and is highly usable afterwards Apr 17 10:33:27 but currently with the pulse mess wine audio its all jumpy :-D Apr 17 10:34:37 Ive only got 100G so far though :_) Apr 17 10:35:13 my CDs are all ripped as mp3 though as my ears are too damaged to tell different between that and FLAC :_D Apr 17 10:36:32 same here Apr 17 10:39:25 mp3 is a crime against humanity Apr 17 10:42:23 hehe Apr 17 10:53:10 Can someone please tell me how to read the register in the control module on the AM3358 - I need to read the mode (=[0:7]) of the GPIO periphs_? Apr 17 10:53:19 I'm using CCS BTW?` Apr 17 10:54:49 koen, what do you want for free? Apr 17 10:55:13 just the syntax Apr 17 10:55:57 but how much would it cost me to have you do a simple project that configures the McASP? Apr 17 11:01:00 mdp: I want a pony Apr 17 11:01:12 (well, rather a real horse) Apr 17 11:01:28 * dm8tbr wants a steak sandwich Apr 17 11:01:37 koen: did you try sudo? Apr 17 11:02:37 yes Apr 17 11:02:38 koen, I don't blame you Apr 17 11:04:32 koen, but let's be realistic about the maintenance costs of horses..especially if you are living in a flat :) Apr 17 11:05:11 the horse is a largely obsolete animal Apr 17 11:05:19 It's $10/day for food alone Apr 17 11:06:06 koen, indeed..don't forget time for required grooming maintenance of structures or cost to rent one to protect from the elements..medical expenses Apr 17 11:06:17 wait, are we talkign about horse maintenance or s/w maintenance…or both? Apr 17 11:06:39 medical expenses... what code did you read? Apr 17 11:06:51 m4 or python Apr 17 11:06:56 those make my eyes bleed Apr 17 11:07:03 mru, it will be necessary during the zombie apocalypse..don't discount their value by looking at current society's needs Apr 17 11:07:51 and if all else fails, you can eat the horse Apr 17 11:08:04 koen, stick with perl (of all things), as gregkh points out, you can write it to look C-like :) Apr 17 11:08:35 koen, or, in the case hypothermia threat..you can slice it open with your lightsaber and sleep inside the carcass Apr 17 11:08:58 this is a one-shot trick, mind you Apr 17 11:09:12 unless you can control a zombie horse Apr 17 11:09:52 koen, mostly I think you need to set your expectations low(er) Apr 17 11:10:10 koen, this has helped me get through the days Apr 17 11:10:32 mdp: Sturgeon's Law Apr 17 11:11:10 well, I think that one is optimistic Apr 17 11:11:21 wasn't it only 90%? Apr 17 11:11:36 yes Apr 17 11:11:45 yeah, ok Apr 17 11:11:53 "Sturgeon was an optimist" Apr 17 11:12:05 good t-shirt Apr 17 11:12:28 * mdp is about to regret something Apr 17 11:12:34 he was right, he just failed to mention you need to apply the law recursively Apr 17 11:13:22 vristo: there's an mcasp project in the kernel, ready to go for you. read sound/soc/davinci/davinci-mcasp.c Apr 17 11:13:58 "When faced with either doing something nonsensical or aborting with an error, it will do something nonsensical." Apr 17 11:14:14 so there's mcasp and mcbsp... imagine the confusion when we get mcdsp Apr 17 11:15:05 multi channel differential serial port? Apr 17 11:15:38 how many years to reach mcpos? Apr 17 11:15:50 * mdp waits for it Apr 17 11:16:55 ..dary! Apr 17 11:17:02 lol Apr 17 11:17:28 mdp: some might say we're already there Apr 17 11:17:52 * mdp is relieved somebody else said it Apr 17 11:18:00 Thanks a lot for your generous help. I'm coding the beaglebone from CCS - can I use the information in the file your reference? Apr 17 11:18:15 also, have you ever had to deal with a point-of-sale system? Apr 17 11:18:18 CCS in win7 that is Apr 17 11:19:05 vristo: no idea..I don't use CCS/Eclipse as it's not compatible with my workflow Apr 17 11:19:25 eclipse is not compatible with my brain Apr 17 11:20:38 vristo: I would suggest you go to a CCS e2e forum and post the question there…most people on #beagle probably can't help you with it Apr 17 11:22:02 Thanks! the davinci-mcasp.c file looks interesting...I think I will try my luck with Linux. I was just a little scared with dealing with ALSA and Linux in general but if I can figure out how to get the Mcasp kernal running I should be happy Apr 17 11:22:07 * koen looks at http://git.kernel.org/?p=utils/util-linux/util-linux.git;a=commitdiff;h=8fb6a2de79540d85567c40e0e5e8d42a612f3af2 Apr 17 11:25:32 wtf, is lennart capable of writing something that doesn't use dbus, gconf, and half of gnome? Apr 17 11:43:44 gm all Apr 17 11:44:16 mru: next project is gnome-shell-bios :-D Apr 17 11:44:24 wb jkridner___ Apr 17 11:44:56 XorA: written in forth…that would be crazy if somebody wrote firmware in forth!!!! Apr 17 11:45:04 totally Apr 17 11:45:14 "never been done before" Apr 17 11:45:29 Jupiter 1 BIOS :_D Apr 17 11:45:34 revolutionary Apr 17 11:45:53 * XorA still needs to get a Jupiter 1 Apr 17 11:47:49 thankfully DT provides a comfort blankee for me so I feel like I'm back in the CHRP mac clone days again Apr 17 11:47:51 Jupiter Ace sorry Apr 17 11:50:03 mdp: can one write a forth machine in DT? Apr 17 11:52:52 jkridner___: welcome back Apr 17 11:54:11 av500: I don't think it's possible... yet Apr 17 11:57:52 thanks koen, mdp Apr 17 12:35:50 av500, one can do anything in DT Apr 17 12:43:40 mdp: make me a sandwich in DT Apr 17 12:44:52 av500, shouldn't that be: sudo mdp: make me a sandwich in DT? Apr 17 12:44:59 * mdp starts a binding to describe the sandwich hardware Apr 17 12:45:26 I only respond to "Siri" Apr 17 12:45:51 how about ramirez? Apr 17 12:46:00 possibly Apr 17 12:47:24 av500, we need to take this sandwich binding to lkml..I think that we need to have a generic binding, but then some platform specific portion due to cultural differences in the makeup of the sandwich h/w Apr 17 12:47:54 and we need to handle the case of the Dagwood sandwich which make have N slices of bread in it's towering form factor Apr 17 12:47:59 * mdp boggles Apr 17 12:48:17 * mdp wonders how this ties into the pinmux binding Apr 17 12:51:24 mdp: also, do we want "stable" sandwiches to get backport from more "experimental" ones? Apr 17 12:53:07 maybe Apr 17 12:53:16 now I'm worried about the SandwichCape Apr 17 12:53:43 perhaps we should store DT fragments in the eeprom on there Apr 17 12:53:51 koen, hey, is DT ready yet? :) Apr 17 12:54:23 it smell funny Apr 17 12:54:28 smells* Apr 17 13:00:09 * koen pokes jkridner___ Apr 17 13:00:52 mdp, thanks for the edma sample application, unfortunately, after fixing 2 includes to make it build, it hangs my beaglebone when I insmod it... Apr 17 13:01:09 mdp: did such a tree fall upon you or where does that obesession come from? Apr 17 13:07:05 jsabeaudry: progress! Apr 17 13:07:28 aholler, sandwich obsession? Apr 17 13:07:56 ya even if I cant make that sample app run, perhaps I can learn a thing or two from it Apr 17 13:09:23 jsabeaudry: yeah, I don't have any experience with that stuff, maybe somebody else around here does Apr 17 13:10:27 keep in mind that it was for ti814x/816x too, probably dependent on some default configuration of channels or otherwise Apr 17 13:22:08 I can't even get the first printk in the __init, would someone be so kind as to compile and insmod this TI EDMA Sample on the beaglebone? http://pastebin.ca/2137187 Apr 17 13:24:11 ok so Apr 17 13:24:16 any PRU interested people here Apr 17 13:24:21 because we have a serious problem. Apr 17 13:29:29 TI seems to have pulled all the documentation on the PRU subsystem ENTIRELY now Apr 17 13:29:33 it DISAPPEARED from the TRM Apr 17 13:30:52 it's possible that it's not sinister... and that the documentation on the PRUSS is going to be in the next version of the SDK (with pasm v2) tha twill be freely available for download. Apr 17 13:31:01 or TI is screwing us in the A, if you get my drift. Apr 17 13:33:52 it is moving to an addendum Apr 17 13:33:58 I'll be releasing the addendum soon. Apr 17 13:34:03 are you telling me not to panic? Because it's too late!!! Apr 17 13:34:07 I've been reviewing the candidate releases. Apr 17 13:34:20 oh. okay. whew. Apr 17 13:34:25 I knew TI wasn't evil all along! Apr 17 13:35:09 I'll ping for the final approval to release. Apr 17 13:35:25 The last question I was asked about was the license for the documentation... Apr 17 13:35:36 I'm thinking I'll recommend CC-BY-SA Apr 17 13:40:49 Do you think the new documetnation is likely to have instruction encodings ? Apr 17 13:41:06 yes Apr 17 13:41:22 that will really help me with my pru debugger project Apr 17 13:41:59 SilicaGel: I have old versions of the trm if you need them Apr 17 13:42:52 a, b, and c Apr 17 13:43:03 I kept them too, :) Apr 17 13:43:19 you've learned :) Apr 17 13:43:46 yeah. Until now though I haven't really ever had a reason I HAD to go back to a previous revision Apr 17 13:48:09 SilicaGel: http://en.wikipedia.org/wiki/Hanlon's_razor Apr 17 13:49:05 wow Apr 17 13:49:06 haha Apr 17 13:50:10 that wikipedia page is incomplete Apr 17 13:52:17 there's another guy who insists he's the one who came up with it: http://bilclarke.blogspot.co.uk/2006/01/credit-where-its-due.html Apr 17 13:52:36 so is the PRU the only way to get proper GPIO interrupts in the kernel? Apr 17 13:56:03 nemik, what do you mean by proper? Apr 17 13:56:16 nemik, what is wrong with standard gpio interrupts? Apr 17 13:56:48 In TI's emda sample they call edma_write_slot on a channel !? Apr 17 13:58:46 oh, there is something big I dont understand, they also call edma_free_slot on a channel Apr 17 13:59:23 jsabeaudry: from everything i read, the 'standard' way seems to be using poll() Apr 17 13:59:48 unless i'm missing other ways to use them in userland? Apr 17 14:00:23 oh sorry though you were talking about "in the kernel" Apr 17 14:01:02 nemik, what is improper about poll() in userspace? Apr 17 14:01:40 yeah, nemik, you can poll or select or whatever. or if it's just one, read I think Apr 17 14:02:16 poll() is the opposite of an interrupt Apr 17 14:02:21 i saw somethign come across the beagle ... well it's on facebook so I don't really know what the hell it is Apr 17 14:02:25 i mean getting a proper interrupt Apr 17 14:02:34 where somebody made an optical shaft decoder using just gpio interrupts Apr 17 14:02:43 well you're in userspace Apr 17 14:02:58 traditional interrupts don't really make sense, the only mechanism linux really has for something like that is signals, and that'd be an awful use of signals Apr 17 14:03:06 nemik, you're getting a proper interrupt! Apr 17 14:03:11 SilicaGel: sure, but why couldn't i get an event for it? like when a key is pressed Apr 17 14:03:31 you are getting an event Apr 17 14:03:34 yeah Apr 17 14:03:46 but it's polling constantly to observe it, no? Apr 17 14:04:00 your userspace code does not have to poll Apr 17 14:04:25 your thread will simply be woken up by the kernel when using poll() Apr 17 14:04:33 yeah. poll and select can block Apr 17 14:05:08 would it be fast enough to detect something like a weigand signal? Apr 17 14:05:11 the pattern for asynchronous events and callbacks in linux typically is to have a service thread (LWP) do the poll and/or select and they in turn make the "asynchronous" call Apr 17 14:05:19 do not confuse the term "poll" with the behavior of "poll()" Apr 17 14:06:10 nemik, well now you are talking response latency in the context of userspace…that's a completely different question Apr 17 14:06:46 nemik, you need to decide what your response time is and then see if you can handle the latency of the context switch Apr 17 14:06:55 mdp: yea good point, i suppose i could write a kernel driver for it and just shuttle the output i get/process to userland instead of each event Apr 17 14:07:05 much smarter Apr 17 14:07:32 it's better to use linux as intended instead of fighting it, trust me Apr 17 14:10:12 mdp: so in that driver, would that still use poll() too? is it a kernel-level utility as well? sorry i'm not familiar with it Apr 17 14:10:54 nemik: there exists a gpio-keys driver Apr 17 14:11:20 iirc, I'm not weigand card reader guru, but iirc, the protocol timing is extremely delicate..you have no chance to do this in userspace without moving to rt-preempt Apr 17 14:11:48 oh you're wanting to use GPIO to read weigand? that's what i'm working on too =X Apr 17 14:12:06 f00bard: yea, for RFID readers Apr 17 14:12:17 it's not that delicate, it's basically self clocking, pulses come in on the 0 line or the 1 line Apr 17 14:12:24 nemik: same! Apr 17 14:12:27 SilicaGel should sell you a PRU implementation Apr 17 14:12:28 mdp: yea, it's not too bad Apr 17 14:13:17 yeah, could be some other pos-related protocol I was told about..my bad Apr 17 14:13:21 i do this on a dedicated micro currently with interrupts but it'd be nicer if it could all be done on the AM35x Apr 17 14:13:29 i saw someboyd else on the mailing list was talking about some PRU application they were working on that was like a 16 bit wide SPI Apr 17 14:14:02 nemik: SPI might be another option for RFID, but you have to have the cards to reprogram the reader (assuming HID) Apr 17 14:14:19 and parsing weigand is much easier ^^ Apr 17 14:15:36 * mdp reads timing diagram Apr 17 14:16:24 yes, 2mS intervals between pulses is doable…but you risk losing data in user space Apr 17 14:16:28 f00bard: yup, HID readers. the wiegand also sends their keypresses so it's nice all in one and its LEDs and beeps can also be done with GPIO. i currently use atmega or NXP micros and their interrupts to get the wiegand and then just send it over serial Apr 17 14:16:52 without read further I assume it has a form of crc so you'll just deal with retransmits in some fashion Apr 17 14:17:23 mdp: yea parity bit at front and back of the payload, at least for RFID Apr 17 14:17:38 haha yo uknow Apr 17 14:17:42 with a uio driver Apr 17 14:17:45 you know how many interrupts you lost Apr 17 14:17:52 so if those interrupts only occur on a GPIO bit changing Apr 17 14:17:57 then you wouldn't really ever lose anything Apr 17 14:18:13 except timing. You know if the bit is now LOW and you missed 3 interrupts, well, you can fgiure out exactly what you missed Apr 17 14:18:43 hmm. This might be the first thing I have considered about UIO that I actually like. (Generally I dislike UIO) Apr 17 14:18:48 brb Apr 17 14:19:08 SilicaGel: found some info about that here: https://groups.google.com/forum/?fromgroups#!topic/beagleboard/qpZSG7cjNSw is that what you mean? Apr 17 14:21:06 SilicaGel: that's exactly what you need too…without that missed interrupt count that UIO kindly has, you have catastrophic failure if you silently miss one data bit due to a missed deadline Apr 17 14:22:01 it appears that it is only transmitted once is why I say that Apr 17 14:22:12 mdp: yea, exactly Apr 17 14:24:39 nemik, http://www.kernel.org/doc/htmldocs/uio-howto.html Apr 17 14:24:46 nemik: yeah. That's actually own own post :) Apr 17 14:24:48 mdp: thank you, i'm reading that now actually Apr 17 14:24:59 SilicaGel: thanks! Apr 17 14:25:01 from way back in feb, which I can barely remember! Apr 17 14:26:04 nemik, note that serious industrial i/o type users often are using it in conjunction with rt kernels to achieve tighter scheduling requirements.. Apr 17 14:26:21 hmm, I think we need kernel-run, like c6run Apr 17 14:26:32 where c6run takes your code and transparently runs it on the DSP Apr 17 14:26:41 kernel-run would run it on the kernel side Apr 17 14:26:51 and make me a sandwich? Apr 17 14:26:56 sure Apr 17 14:27:15 sudo make me a sandwich Apr 17 14:27:23 userspace sandwiches for all Apr 17 14:27:28 f00bard: I have that on my wall! Apr 17 14:27:29 I bet in-kernel-python would beat the crap out of nodejs Apr 17 14:27:42 f00bard: a framed print, signed by the author! Apr 17 14:27:49 =D Apr 17 14:27:55 mdp: thanks, i suppose it's another thing to consider but i think in this case and the quantaties we're targetting, the cost of the extra micro and UART connections is cheaper than porting everything over to a new kernel Apr 17 14:28:14 SilicaGel: my friend got me this one as a gift http://xkcd.com/386/ Apr 17 14:28:18 there is a reason why the bone cpu has 6 uarts Apr 17 14:28:27 they are all to talk to micros that talk to hardware Apr 17 14:28:29 ahahaha that is an awesome one Apr 17 14:28:34 indeed, moving to -rt should be a last resort Apr 17 14:29:05 av500, perl is the only blessed scripting language for kernel related items Apr 17 14:29:38 this is more of a nice-to-have to reduce part count, the micros aren't really that expensive and even the cheapest msp430's or avrs could do this. but it's quite tempting to take advantage of all the IO that exists on the AM35x! Apr 17 14:29:40 So if one wanted to build a BBone-powered robot, the best approach would be to put in an Arduino to do the low level hardware interfaces and let the BBone just do the more complex stuff? Apr 17 14:29:44 av500, I thought that 6 uarts were to hang 6 vt100s off and have a low-cost timesharing *nux machine Apr 17 14:30:02 woah, what an idea Apr 17 14:30:12 I'm full of them Apr 17 14:30:23 All playing "Zork" Apr 17 14:30:32 Mojito: until user-land IO gets better, yes that would be the fastest approach to get started. not sure about 'best' Apr 17 14:30:35 always with the games! Apr 17 14:30:45 wasn't there a multiplayer one Apr 17 14:30:51 Moria or something Apr 17 14:30:53 mdp: so 6 poor kids can learn to program at the same time, hmm, that's only ~15$ per kid!!!!!!! Apr 17 14:31:09 Where you getting these free VT100s? :) Apr 17 14:31:14 Mojito: Zork and running PennMUSH Apr 17 14:31:18 mangband? or phantasia? or something? Apr 17 14:31:31 there was something that was like hack except you could see other players running around and you could gamg up on monsters Apr 17 14:31:31 Mojito: the same place r-pi gets freew HDMI monitors.... Apr 17 14:31:40 Mojito: you get them at the university surplus place..cheap! Apr 17 14:32:19 or any any goverment office Apr 17 14:32:23 at any* Apr 17 14:32:27 or maybe they can hook up to this inexpensive system via their laptop and a usb serial converter Apr 17 14:32:28 try the DMV Apr 17 14:32:35 mdp: +1 Apr 17 14:32:36 * mdp concocts more good ideas Apr 17 14:32:48 i met with a prospective client the other day and he actually used the word "timesharing" Apr 17 14:33:04 with regard to computing (not vacation homes) Apr 17 14:33:12 SilicaGel: was he checking his FB account a lot while talking to you? Apr 17 14:33:27 haha no Apr 17 14:34:29 * mdp runs to marketing (all 1850km) to pitch the idea in person Apr 17 14:34:35 Using a $1000 laptop to access a timesharing service running on an $89 processor sounds like a good business plan Apr 17 14:35:02 Mojito: they tell me this is why I'm stuck in engineering :( Apr 17 14:35:19 * mdp shakes his fist Apr 17 14:35:25 hahahaahahaha Apr 17 14:36:47 SilicaGel: tell him you can provide him a multitasking OS too Apr 17 14:36:58 http://26-26-54.hardwarebug.org/45 Apr 17 14:36:59 DJWillis: go back to your new fanbois! Apr 17 14:54:01 Progress on the EDMA front, it works in incremental mode, but if I happen to set FIFO mode at some point, nothing will work until a reboot (even if I edma_clean_channel) Apr 17 14:56:15 * mdp reports another 100% happy silicon customer to marketing Apr 17 14:56:34 Could perhaps be related to SDOCM00084262 Apr 17 14:57:09 Anyone with access to the status of that ticket here? Apr 17 14:57:19 also check SDOCM0008434444468957456234326784623785-Z2 Apr 17 14:59:10 jsabeaudry: those look "promising" Apr 17 15:01:37 Well if the DMA is non-functionnal it's a big problem for me Apr 17 15:02:58 the good news is that it's functional in the use case of spi-omap2-mcspi.c Apr 17 15:03:17 at least in koen's tree that I moved to Apr 17 15:04:19 mdp, I'll take a look if that does FIFO Source address mode Apr 17 15:10:09 it does not Apr 17 15:10:36 ya not even destination fifo mode Apr 17 15:18:01 ah my bad, was not aware of the 256 bits alignment rule Apr 17 15:18:31 assumed alignment rule was the same as the fifo width Apr 17 15:34:15 ah this is not the whole answer, even with a properly aligned address it fails Apr 17 16:04:49 jsabeaudry: despite the pain of using the RationalToolThatShallNotBeNamed, I found that issue and the status is unresolved..fwiw Apr 17 16:07:55 mdp, thanks for the info, I think I have found a workaround, it seems you can fake a non-incr dma by using an AB sync dma with a bidx of 0 or something of the sort... Apr 17 16:13:35 ah, this explains the burning smell: http://www.bbc.co.uk/news/uk-england-hampshire-17741364 Apr 17 16:14:26 how does one set scrap metal alight? Apr 17 16:14:38 somebody dropped thermite on the scrap yard? Apr 17 16:15:06 jsabeaudry: that is what mcspi does, fwiw..maybe that was your reference Apr 17 16:17:58 I want to install a minimal debian on my pandaboard es, but I want the root to be on my SSD-drive via USB(have adapter). I dont know where to start, even though I have googled. Please help Apr 17 16:24:54 Haikarainen, http://elinux.org/BeagleBoardDebian#Debian_NetInstall then just choose the usb harddrive during partition selection.. Apr 17 16:26:05 rcn-ee omg thanks! how well supported is this with panda? Apr 17 16:26:50 running on 24/7 on 4 panda/panda_es and tested with 2other's on a weekly basis.. ;) Apr 17 16:28:09 (i saw no point in creating X number of "X"Board wiki pages, that page just started with the Beagle, but the script supports a half dozen other boards).. Apr 17 16:28:15 rcn-ee thanks a bunch man! all hw supported? Apr 17 16:30:09 define 'all'... it's a pure mainline kernel.. currently based off v3.2.... so usb/wifi/bluetooth/ethernet/display.. (v3.3 added kms support, but isn't surviving my stress testing so i haven't pushed it yet.) (and v3.4-rc added audio support).. .so no 3d graphic support yet.. Apr 17 16:30:10 all hw ever conceived by mankind! Apr 17 16:30:31 Haikarainen: and btw, this is not the panda channel Apr 17 16:31:07 ah, my bad for not replying on the #pandaboard.. he spamed both. ;) Apr 17 16:31:24 not getting an immediate answer on one channel is not reason to spam another one Apr 17 16:31:25 booo panda! Apr 17 16:31:28 and I dont mean rcn-ee Apr 17 16:32:49 wtf is a gist? Apr 17 16:33:26 av500: i think a gist is a large cyst Apr 17 16:34:02 *the* gist? Apr 17 16:36:57 https://gist.github.com/2391604 Apr 17 16:37:03 wtf is a gist? Apr 17 16:38:01 http://oxforddictionaries.com/definition/gist Apr 17 17:09:22 does the bone expose gpmc on the headers? Apr 17 17:10:44 Crofton|work: yes… see SRM starting at pg 49 for which pins and pin mux modes you need to set Apr 17 17:11:26 ah Apr 17 17:11:31 I looked once and did not see it Apr 17 17:15:52 kkeller, perfect for our ISACape :) Apr 17 17:16:04 A15 !!!! Apr 17 17:17:35 A20 Apr 17 17:18:15 You sank my battleship! Apr 17 17:19:29 sorry, it was in the way Apr 17 17:20:30 I was actually talking about http://www.engadget.com/2012/04/17/arm-announces-new-quad-core-cortex-a15-hard-macro-variant/ Apr 17 17:21:11 what does hard macro mean? Apr 17 17:21:17 no microcode? Apr 17 17:21:20 no Apr 17 17:21:29 hard macro means a silicon level template Apr 17 17:21:32 it means that some IP core that's in there is Apr 17 17:21:33 yeah Apr 17 17:21:36 soft means vhdl Apr 17 17:21:39 as opposed to being RTL or Verilog or something Apr 17 17:21:42 yeah Apr 17 17:21:46 ic, thanks Apr 17 17:21:51 firm usually means RTL Apr 17 17:21:52 I think Apr 17 17:22:10 do I need a hard macro to run debian hf? Apr 17 17:22:34 no, but you need one to mate with memory Apr 17 17:27:41 *groan* Apr 17 18:25:58 why oh why do a growing number of people call and IM me directly and start out with "hi, how are you?" when what they want is BeagleBoard support? Apr 17 18:26:08 don't they understand how upset that makes me? Apr 17 18:26:54 if they just came here and asked... perhaps even singling me out... and just asked, without asking to ask... how much nicer would my life be? Apr 17 18:27:08 think politics. Apr 17 18:27:26 jkridner: it is also a general introduction in many cultures Apr 17 18:28:25 jkridner: don't worry, i normally start of with: hey, you rat @#$@#$ so you don't need to get upset with me Apr 17 18:29:28 i personally "like" those who will spam you for hours in qry if you once tried to help them.. Apr 17 18:29:58 LetoThe2nd: hehe Apr 17 18:30:06 * jkridner always prefers socializing over a beer face-to-face and using e-mail/IM/IRC to simply let me know what you want! Apr 17 18:31:10 you know, all your best friends are on facebook ;) Apr 17 18:31:11 jkridner: so please then, send money, beer and hot chicks. you can also add a few beagleboards if you happen to have some lying around. Apr 17 18:31:20 jkridner: i want a quad core arm board with 8gb ram and sata, for $19.99 Apr 17 18:31:34 (and free shipping) Apr 17 18:31:40 prpplague: bring your own hdmi monitor, mokay? Apr 17 18:31:53 mdp: m'lay Apr 17 18:31:55 mdp: m'kay Apr 17 18:32:01 that does come with hand holding? Apr 17 18:32:02 prpplague: does it need to work? Apr 17 18:32:22 jkridner: as long as the leds blink Apr 17 18:32:46 can do. I'll probably get around to it in around 2232. Apr 17 18:32:48 prpplague: m'''''kay Apr 17 18:32:58 jkridner: hehe Apr 17 18:33:06 prpplague: a guitar player i know actually purchases gear based on the number of blinking leds that are mounted on it. Apr 17 18:33:52 .... now what was I doing before the last three calls and two IMs anyway? :-/ Apr 17 18:34:02 going for beer? Apr 17 18:34:09 hm, beer. Apr 17 18:34:14 jkridner: i'll go throw old sdram chips at gerald while i wait for the board Apr 17 18:35:42 gm Apr 17 18:35:55 my new car has a good beer cooler built in :) Apr 17 18:36:20 big well over the spare to toss in a couple six packs and ice Apr 17 18:36:45 djlewis: hello open container? Apr 17 18:36:55 not while parked. Apr 17 18:37:10 oO( "hello, computer?" ) Apr 17 18:37:13 i think the container means the can Apr 17 18:41:12 hi alexandersalas. welcome to #beagle. Apr 17 18:41:47 jkridner: hi! :) Apr 17 18:42:29 Thank you jkridner Apr 17 19:21:48 does beagleXM prints 'CCCC...' chars when it is trying to read bootloader from serial port? Apr 17 19:25:10 cehh, i'm pretty sure the am335x parts are the first to do that.. Apr 17 19:25:12 Hi, I have problem with c6runlib-cc can somebody give me advice please ? Apr 17 19:26:04 I can compile examples (emqbit), but when I try to compile my dsp lib (c6runlib-cc -c -O3 DSP.c -o DSP.c) compiler end up in endless loop ( CPU100%) ( in c file isjust one simple c function) what am I doing wrong ? Apr 17 19:37:57 rcn-ee: thanks Apr 17 19:42:01 if I enable modem support cdc acm, thats just the kernel right? Apr 17 19:44:01 just in case you don't know, we don't see where you enable this Apr 17 19:44:51 menuconfig Apr 17 19:46:20 from the kernel I assume, so it is very likely that this just enables compilation of the driver in kernel ;) Apr 17 19:46:48 thats what I mean lol Apr 17 19:49:43 man Apr 17 19:49:56 the SRM says that the mounting holes on the bone are Apr 17 19:50:01 oh crap I'm friggin reading it wrong Apr 17 19:50:26 that's the problem, i'm looking at the radius of the edge of the board, not the radius of the hole. no wonder it is ridiculously big. Apr 17 19:50:56 sorry about being vague, so All you have to do is make a new uImage when you use menuconfig to make changes to the kernel? Apr 17 19:52:42 borillion: no, you need the modules too Apr 17 19:53:23 does anybody know how big those holes are SUPPOSED to be? dwery? Apr 17 19:54:18 for one thing I'm working on, I just brought up the gerbers in gerbv and took some measurements Apr 17 19:55:23 since the SRM drawing only show the outer mechanicals Apr 17 19:57:18 2mm radius iirc Apr 17 19:57:21 SilicaGel, http://ladyada.net/products/bonebox/ Apr 17 19:57:32 look at the screw sizes there Apr 17 19:57:44 borillion++ good thinking Apr 17 19:58:11 The diam of a #4 screw is about 0.112". The average size of the clearance hole that the screw will fit in is about 0.120" Apr 17 19:58:23 0.120 inches = 3.04800 millimeters Apr 17 19:59:39 SilicaGel: holes? Apr 17 19:59:57 i think we figured it out. Just trying to figure out how big to make the holes! Apr 17 19:59:58 no hole son my capes, sorry Apr 17 20:00:03 yep, that's the measurement I see from the gerbers Apr 17 20:00:21 since I remembered wrong :) Apr 17 20:02:38 hmm so the only other thing i have to figure out is how do you make it so that in eagle you can place a hol eon the board but not have it be on the schematic. It works if I do it using a hole from the holes library, but not if I use my own hole, so something must be peculiar about the way they make theirs. Apr 17 20:05:41 SilicaGel: use the button marked as "hole" in the pcb editor Apr 17 20:06:23 yeah. but theirs also has a pad around it Apr 17 20:06:28 they seem to have both Apr 17 20:06:46 one called 3,2 and another one called 3,2-PAD Apr 17 20:07:11 at first Ithought their "pad" wasn't really a pad, it was just a polygon Apr 17 20:07:16 but it looks like it's a real genuine pad Apr 17 20:07:42 uhm... then it should be in the holes library Apr 17 20:07:55 is that magical? that it came from the holes library? Apr 17 20:08:05 aholler: modules? Apr 17 20:08:29 I guess so, it's been a while since my last hole :D Apr 17 20:08:30 ls /lib/modules Apr 17 20:08:57 yeah ok Apr 17 20:09:01 their hole says Apr 17 20:09:42 oh no, theirs does too have a pad ... weird. Looking at the xml, my hole and their hole are very similar Apr 17 20:09:55 you might want to copy and paste the xml :D Apr 17 20:10:56 no Apr 17 20:10:59 you were right in the first place. Apr 17 20:11:17 Their 3,6 I can place, their 3,6-PAD I can NOT place from the board side, meaning if it has a pad, it has to be on the schematic. Apr 17 20:11:23 D O H Apr 17 20:12:19 aholler so when you do make ARCH=arm CROSS_COMPILE=arm-eabi- menuconfig , how do you get the default settings, are they already marked ? Also having downloaded the kernel sources aren't the modules supposed to be in them already? Apr 17 20:15:58 wb sakoman_ Apr 17 20:16:01 SilicaGel: yes, a pad must be on the schematic Apr 17 20:16:15 so your cape is going in production? Apr 17 20:16:44 not this one Apr 17 20:16:56 well Apr 17 20:16:56 it is Apr 17 20:17:01 but it's just for us Apr 17 20:17:45 k Apr 17 20:39:16 borillion: get a config from a running system: zcat /proc/config.gz > .config Apr 17 20:39:24 then run make oldconfig Apr 17 20:41:22 i have a linux question. People say that with Linux, everything is read using files. Apr 17 20:41:22 Does this sound familiar? Apr 17 20:41:22 can I write to /dev/fb0 and see somthing appear on the screen? Apr 17 20:41:47 basically yes. Apr 17 20:41:47 cat /dev/urandom >/dev/fb0 Apr 17 20:42:17 taking the assumption that there is a screen connected and set up properly. Apr 17 20:42:28 else you can write whatever you like and nothing is gonna show up Apr 17 20:50:01 Am I supposed to use optimization flags like -O3 with kernel modules? Apr 17 20:52:30 up to you Apr 17 20:54:13 ok so its legal, nice Apr 17 20:54:22 * mranostay arrests jsabeaudry Apr 17 20:54:25 it just makes tighter asm Apr 17 20:54:48 Perhaps this will reduce abominable irq latency of up to 55ms im getting Apr 17 20:54:56 doubt it Apr 17 20:55:14 (it includes my actual irq code) Apr 17 20:55:19 yeah that isn't really a compile time Apr 17 20:55:22 *thing Apr 17 20:55:32 55ms is an eternity, that's a lot of cycles Apr 17 20:55:47 ya I'm puzzled what takes so long Apr 17 20:55:59 especially at 720 MHz Apr 17 20:58:18 From what I can see at the scope, its anywhere from 0.5 ms to 55 ms Apr 17 20:59:00 Ideally, I would need it to be always under 2ms Apr 17 21:03:22 mranostay, jay6981: Any idea what could influence irq lantency? Apr 17 21:04:51 most irqs are now tasks, so I assume the scheduler Apr 17 21:13:12 awful news Apr 17 21:31:25 Well, I guess it's time to look at the scheduler, because my interrupt code is 5us-15us long Apr 17 22:01:07 I'll give it a try with CONFIG_PREEMPT and CONFIG_HZ=1000 Apr 17 22:09:29 why is my board stopping after ehci-omap.0 supply hsusb1 not found, using dummy regulator ? :| Apr 17 22:09:53 just started doing it? Apr 17 22:11:18 undo whatever you did just before it started doing that Apr 17 22:12:41 anyone familiar with the wl1271? Apr 17 23:42:56 I want program Driver for BeagleBoard, can you share me a documents? Apr 17 23:43:36 holy cow, thats a good one Apr 17 23:43:49 I suggest reading the internet Apr 17 23:44:15 ... like all of it? Apr 17 23:44:37 sure Apr 17 23:44:46 start with http://www.youtube.com/watch?v=iDbyYGrswtg **** ENDING LOGGING AT Tue Apr 17 23:46:14 2012 **** BEGIN LOGGING AT Tue Apr 17 23:48:17 2012 Apr 17 23:59:10 * mdp makes it back in time to catch the best one of the week Apr 17 23:59:22 mdp: hehe Apr 17 23:59:23 copy-paste-archive! Apr 17 23:59:30 mdp: already on G+ Apr 17 23:59:35 w00t! Apr 18 00:00:06 please tell me that's av500 under there...please! Apr 18 00:02:18 Hmmmmm Apr 18 00:03:17 or ds2 Apr 18 00:04:56 <-- buried Apr 18 00:05:06 it could just be a fortune-bot with a compilation of the "best of" 2011 to replay for us Apr 18 00:06:33 isn't that what mru's site is for? :) Apr 18 00:10:40 yes :) but it needs to complete the circle Apr 18 00:11:17 262654bot has entered the room! Apr 18 00:15:29 kkeller, thanks, merged it…I have some other stuff I need to finish before I'll get to the wrapping up the schematic view Apr 18 00:16:24 mdp glad to help out - I already have one of my capes in fritzing :) Apr 18 00:16:35 excellent :) Apr 18 00:17:12 I have some tweaks yet and got the blessing from Boris Houndleroy's sidekick to add the branded logos to it Apr 18 00:17:49 I've been sitting on some half finished stuff wrt openocd docs that simply must get done now Apr 18 00:19:13 I'm in no hurry… if there is more I can do to help, let me know Apr 18 00:59:04 i am seeing a serious problem with the beagle family of boards Apr 18 00:59:21 it is too unrealistically easy to get things to work compare to the other stuff out there @!#$!#@!@%$!@$#!@ Apr 18 01:00:21 ds2: hehe Apr 18 01:10:33 we need a decoy channel i think :) Apr 18 01:30:56 * ds2 *heart* frequency counters Apr 18 01:31:24 versions in the dmm do not cut it **** ENDING LOGGING AT Wed Apr 18 02:59:58 2012