**** BEGIN LOGGING AT Sat Nov 24 02:59:59 2012 Nov 24 06:45:34 * mranostay does the ds2 dance Nov 24 06:47:32 hmmmmmmm Nov 24 06:48:07 got this damn driver working :) Nov 24 06:48:14 always blame hardware! Nov 24 06:48:27 * mranostay was missing a pulldown on the blanking Nov 24 06:48:38 d'oh! Nov 24 06:53:10 emeb_mac: how is it in FPGA land? Nov 24 06:53:55 sidetracked Nov 24 06:54:28 how's your project? Nov 24 06:54:40 which one? :) Nov 24 07:04:01 whichever one you're working on right now. Nov 24 07:32:17 gruzi Nov 24 07:41:34 <_av500_> grutzi? Nov 24 07:43:26 _av500_: weird swiss dude :P Nov 24 07:44:14 <_av500_> yeah Nov 24 07:45:24 <_av500_> mranostay: what driver? Nov 24 07:47:12 _av500_: custom capebus one Nov 24 07:47:45 <_av500_> ? Nov 24 07:48:25 i'm not weird, i just like chocolate! Nov 24 07:51:52 _av500_: spi vfd driver Nov 24 07:55:52 <_av500_> ah Nov 24 07:56:07 <_av500_> what kind of vfd? Nov 24 07:56:12 <_av500_> a VCR blinking 12? Nov 24 07:57:17 _av500_: soviet tube Nov 24 08:00:24 <_av500_> pic? Nov 24 08:01:45 https://www.siberian-shop.com/iv18-vfd-nixie-tube-large-multisegment-newnos-p-215.html Nov 24 08:02:15 guys, c++ is easy Nov 24 08:02:36 cplusplus.com has amazing tutorial Nov 24 08:03:37 er ok Nov 24 08:04:01 * _av500_ overloads the ; operator Nov 24 08:04:19 heh Nov 24 08:11:46 james-jones: no white space sensitive language is easy Nov 24 08:12:23 james-jones: if you dont believe me, have a look at http://compsoc.dur.ac.uk/whitespace/ Nov 24 08:13:10 hmm does the spi subsystem probe the bus? Nov 24 08:14:09 there aint no probe Nov 24 08:14:26 assert cs and see whether anything responds Nov 24 08:14:38 ofc polarities have to be correct Nov 24 08:15:02 KotH what did i just readÉ Nov 24 08:15:05 ? Nov 24 08:15:11 lol Nov 24 08:15:38 mranostay: the usual way is to tell your system what is connected on which CS in DT Nov 24 08:15:53 KotH: which i'm doing :) Nov 24 08:15:55 mranostay: if devices are autodetectable (eg SPI flash), then the kernel will do that Nov 24 08:16:12 mranostay: generally thoug, devices are not autodetectable Nov 24 08:18:18 no i just see junk on my shift register on startup.. just wondering Nov 24 08:18:45 hmm.. Nov 24 08:18:49 reset condition not met? Nov 24 08:19:20 let me rephrase that: do you reset the shift register at startu? Nov 24 08:19:39 is this code valid? Nov 24 08:19:39 m = operation (7, 5, addition); Nov 24 08:19:49 int operation (int x, int y, int (*functocall)(int,int)) { Nov 24 08:20:02 int addition (int a, int b) Nov 24 08:20:13 m passes the name addition Nov 24 08:20:13 in c? Nov 24 08:20:16 c++ Nov 24 08:20:16 sure Nov 24 08:20:23 ok so a function name is the pointer to it Nov 24 08:20:25 hmm.. sould be as well Nov 24 08:20:27 ? Nov 24 08:20:29 havent written any c++ in ages Nov 24 08:20:43 uhmm.. get yourself a good c++ book Nov 24 08:20:53 or read the standard Nov 24 08:21:00 will do just reading through a tutorial Nov 24 08:21:14 they used: int (*minus)(int,int) = subtraction; Nov 24 08:21:21 n = operation (20, m, minus); Nov 24 08:21:24 which i understand Nov 24 08:21:30 oh... _THAT'S_ why you said c++ is easy.. you haven't learned all it's peculiarities yet! Nov 24 08:21:35 but the first snippet i posted wasn't really explained Nov 24 08:21:40 :) lol Nov 24 08:22:08 i've skimmed them over the past 2 years or so Nov 24 08:22:15 never got heavy into pointers to pointers or void pointers Nov 24 08:22:17 but it seems ok Nov 24 08:23:10 ohhh Nov 24 08:23:13 i have no reason to write c++ Nov 24 08:23:23 i see now what they do, they define the variable in the function parameter list Nov 24 08:23:32 the pointer stuff in c++ is easy Nov 24 08:23:39 once you understand pointers, that is Nov 24 08:24:04 classes, templates, multiple inheritance etc make it complicated Nov 24 08:25:31 i can do classes Nov 24 08:25:41 and maybe the slightest of multiple inheritance Nov 24 08:25:42 from java Nov 24 08:25:47 dunno about templates though Nov 24 08:26:06 java is a lot more simple than c++ in that regard Nov 24 08:26:12 ah Nov 24 08:26:16 java has very few booby traps compared to c++ Nov 24 08:26:33 garbage collection right Nov 24 08:26:45 is one of them? Nov 24 08:26:50 memory managment is the least of your worries usually Nov 24 08:27:27 or to put it differently: if you dont know when and where your clasess/buffer/etc is freed, then you have a design problem anyways Nov 24 08:27:34 no matter whether you have gc or not Nov 24 08:28:03 alrighty, so whats the problemo Nov 24 08:28:24 understanding and accessing the data structures? Nov 24 08:28:36 this class and template stuff? Nov 24 08:28:43 sorry dude, you cannot learn programming by asking other peoples or read books... that can only help Nov 24 08:29:01 alrighty Nov 24 08:29:02 to learn what programming means, where the problems are and how you get around them, you have to write code, lots of code Nov 24 08:29:13 oh.. and read other peoples code as well Nov 24 08:29:19 but i feel like can't you just stick to some basic rules and don't try fancy coding, everything will be better no? Nov 24 08:29:58 * KotH usually recomends "Code Reading: The Open Source Perspective" by spinellis Nov 24 08:30:11 life isnt that easy Nov 24 08:30:25 and if you just stick to the simple stuff, all your projects will be simple as well Nov 24 08:30:37 first thing to learn: programming is a craft, not a science Nov 24 08:30:47 and a craft is learned by duing it Nov 24 08:30:51 doing* Nov 24 08:31:12 hey man Nov 24 08:31:13 you start with something simple and work on your skills until you can do the complicated stuff as well Nov 24 08:31:17 i have been programming, just not in c++ Nov 24 08:31:33 doesn't that count? :) Nov 24 08:31:41 how many lines of code have you written in which programming languages? Nov 24 08:32:27 500 java, 800 verilog, 100 perl, >1000 matlab, 250 C Nov 24 08:33:00 eh.. Nov 24 08:33:02 i can't speak for in-depth knowledge, but my code always works Nov 24 08:33:31 rule of thumb: you cannot say you know anything about a programming language unless you have written more than 10kloc in it Nov 24 08:33:44 and even then you know very little of its more intricate problems Nov 24 08:34:13 alright Nov 24 08:34:28 i was jesting when i said it was easy Nov 24 08:34:35 i dont want to discourage you, but programming is really really writing code, lots of it Nov 24 08:34:46 what should the code do though? Nov 24 08:34:53 anything Nov 24 08:35:04 anything that is fun for you to code Nov 24 08:35:04 i could write tons of mathematical models and algorithms but i don't feel i learn anything new there Nov 24 08:35:24 if you write for yourself alone, then yes Nov 24 08:35:59 as i said programming is a craft...that measn you have to do different things to learn different stuff Nov 24 08:36:22 i have a question Nov 24 08:36:23 looking at how other people solved similar problems helps learning as well Nov 24 08:36:33 * KotH has hundreds of questions Nov 24 08:36:38 ;) Nov 24 08:36:38 kinda off topic Nov 24 08:36:50 when i plug in my webcam, windows finds the driver Nov 24 08:36:51 how? Nov 24 08:37:35 usb contains a descriptor that tells the host what kind of device is connected Nov 24 08:37:45 the OS has a table that links this device id to the driver Nov 24 08:38:08 it uses a general camera driver then? Nov 24 08:38:24 i dont think there is a general camera driver... though i'm not sure Nov 24 08:38:43 because i plugged in two different brands and i can imagine i plug in many others and it will work Nov 24 08:38:46 but you can ask windows what driver it uses Nov 24 08:38:46 i could be wrong Nov 24 08:39:11 lets say im using linux Nov 24 08:39:13 how would that work Nov 24 08:39:22 exactly the same way Nov 24 08:39:57 so the linux system comes with a bunch of drivers Nov 24 08:40:18 well, all OS i guess Nov 24 08:40:22 you should really get yourself a linux book as i have told you 2 days ago Nov 24 08:40:31 ok so linux book and c++ book? Nov 24 08:40:41 i saw good words about "Accelerated C++" Nov 24 08:40:42 lots of books, actually :) Nov 24 08:40:50 * KotH doesnt know that one Nov 24 08:41:00 for c++ i usally recomend "c++ primer plus" by prata Nov 24 08:41:27 why program c++? :) Nov 24 08:41:56 because he wants to use opencv Nov 24 08:42:35 accelerated c++ also had 4.5 rating off of 102 votes Nov 24 08:43:34 i heard stories from a prof Nov 24 08:43:40 employers would be like "do you just know matlab?" Nov 24 08:43:45 apparently c is welcomed Nov 24 08:44:14 employers want an employee being productive from the first day on Nov 24 08:44:27 that doesnt work if he doesnt know the programming language they use Nov 24 08:44:57 and if the employee only knows languages that are unrelated to the languages they use, then it will take him a long time to learn those Nov 24 08:45:01 yeah but how long does it take to pick up the basics of a new language? 2 weeks? Nov 24 08:45:18 depends on how much experience you have, on what languages you know and how good you are Nov 24 08:45:59 how much software organization do you need to know Nov 24 08:46:09 picking up c# took me 3 weeks or so... erlang took me 3 months.... haskell... 6 months and counting Nov 24 08:46:19 like i don't fully understand the whole chain from my C code to 1s and 0s, is that a problem? Nov 24 08:46:38 yes and no. yes you should know that to be a good programmer, no you can ignore it in the beginning Nov 24 08:46:49 and you are at the beginning Nov 24 08:48:13 ok what kind of book do i need to understand that Nov 24 08:48:18 operating system book Nov 24 08:48:19 ? Nov 24 08:48:48 h&p is the standard book in that area Nov 24 08:50:46 you can try http://www.nand2tetris.org/ which covers approximately the same stuff, but follows only one possible path and doesnt give you the details of alternative approaches as h&p does Nov 24 08:52:11 computer architecture: a quantitative approach Nov 24 08:52:17 that book? by henessy and patterson? Nov 24 08:53:06 no, the other one "computer organisation and design" Nov 24 08:53:22 computer architecture is about hardware only Nov 24 08:53:31 very informative, but not what you want Nov 24 08:54:00 this link you gave me... thank you :) Nov 24 08:54:20 np Nov 24 08:54:29 i just learned about it recently myself Nov 24 08:54:39 havent read that book yet, so i cannot say how good it is Nov 24 08:54:46 but the stuff they have online looks quite decent Nov 24 09:26:17 mranostay: the reboot handler is missing form mainline Nov 24 09:26:40 koen: nice... Nov 24 09:26:42 mranostay: there was a patch posted, but it requires linux-omap changes, so that's 3.8/3.9 material Nov 24 09:54:25 something for the crazy texans here: http://i.imgur.com/JfC7n.jpg?2163 Nov 24 09:55:52 <_av500_> KotH: tautology much? Nov 24 09:56:18 you mean all crazy people are texans? Nov 24 09:56:29 <_av500_> that too :) Nov 24 09:57:53 hmm.. that means i have to move to texas, once i finish my studies Nov 24 09:58:04 * KotH changes subject Nov 24 09:58:39 KotH is a texas expat? Nov 24 09:58:58 i'm not crazy! i just like chocolate! Nov 24 10:00:10 KotH: btw, I gave a piece of your chocolate to our cook in the canteen. he was like 'OMG OMG awesome'. we immediately started discussing how to replicate that chocolate here too. Nov 24 10:00:21 rotfl Nov 24 10:00:37 good luck with that Nov 24 10:03:35 <_av500_> replicate? Nov 24 10:03:39 <_av500_> clone a swiss cow Nov 24 10:09:42 KotH: we'll not be able to match the quality of swiss chocolate obviously... Nov 24 10:12:22 whats the point then? Nov 24 10:12:36 if you cannot match the taste and quality? Nov 24 10:27:20 KotH: well, how expensive would it be to get the stuff here including p&p? Nov 24 10:27:48 depends how much you want :) Nov 24 10:28:34 up to 500g is 20chf shipping Nov 24 10:28:54 500g of chocolate costs anything between 4chf and 100chf Nov 24 10:31:13 I'd be specifically interested in that rose-pepper (schinus terbinthifolius) chocolate Nov 24 10:32:26 uhmm.. dont remember how much that did cost Nov 24 10:32:50 prolly somewhere in the range of 3-10chf/100g Nov 24 10:34:22 meh.. they have a website with a list of all chocolate, but no prices Nov 24 10:35:43 ah.. found a list of prices Nov 24 10:35:58 that one is not on it (special chocolate) but similar ones are 7chf/100g Nov 24 10:36:12 so it should be around that as well.. probably a bit higher Nov 24 10:56:02 koth the chocolate dealer Nov 24 10:56:42 i see it as a mission to bring peace to the world! Nov 24 10:58:08 so you fixed them all at barcelona Nov 24 10:58:21 that only works if everyone has enough cocolate though Nov 24 10:58:56 gm ogra Nov 24 10:59:01 hey Nov 24 10:59:57 given that the average swiss consumes only about 12kg/year, and the world consumes much less... i guess it is not yet much of a problem ot give everyone enough cocolate Nov 24 11:02:50 world peace/domination by chocolate. I approve of this! Nov 24 11:06:11 hmm so including shipping that would be approximately 9EUR /100g for 500g batches Nov 24 14:22:44 I am in an airplane! Nov 24 14:23:15 you have my deepest sympathies Nov 24 14:26:40 better than being ON an airplane Nov 24 14:26:56 or _under_ one Nov 24 14:27:24 or for that matter, being one Nov 24 14:30:32 wow, they've bitbanged an entire distro: http://pibanglinux.org/ Nov 24 14:31:30 first "forking" and now "bitbanging" sheesh.... Nov 24 14:51:36 :P Nov 24 14:54:06 THis interweb in plane will never catch on Nov 24 14:57:26 Crofton: traveling again? Nov 24 14:57:57 yes Nov 24 14:58:05 back to Mountain View for more beatings Nov 24 14:58:35 joy and happiness. Nov 24 14:59:52 you beater or beatee? Nov 24 15:00:16 we'll see :) Nov 24 15:19:13 he crofton Nov 24 15:51:46 what's the best way to build qt embedded apps from either ubuntu host or ubuntu beagleboard enviroment? Nov 24 16:08:45 xenome install g++ and follow the embedded instructions Nov 24 16:09:00 <_av500_> are they embedded inside g++? Nov 24 16:09:07 <_av500_> strings g++ Nov 24 16:09:42 <_av500_> mru: wow: Furthermore, PiBang has forked the Raspi-Config program to add the ability to change usernames and gives users the ability to change the hostname. Nov 24 16:10:16 advanced! Nov 24 16:10:31 <_av500_> i_can_have_a_leet_host_name Nov 24 16:10:37 Where does uboot keep the /proc/cmdline text? It certainly doesn't appear to be uenv.txt Nov 24 16:10:51 <_av500_> monttyle: it can read it from there Nov 24 16:11:05 <_av500_> and I guess a default is compiled in Nov 24 16:11:15 <_av500_> unless you compile it into the kernel Nov 24 16:12:28 <_av500_> uboot can also read it from nand Nov 24 16:13:12 Half the options that ARE in uenv.txt don't go into /proc/cmdline, either. If they don't go there, what are they doing? Nov 24 16:13:34 monttyle than your uboot is fucked up Nov 24 16:14:02 This is the default card that comes with beagleboard. Nov 24 16:14:45 In fact, I have many cards and images, and none of them so far seem to obey how uboot is "supposed" to work. Nov 24 16:15:51 <_av500_> note that uenv.txt works different from boot.scr Nov 24 16:16:09 av500 and he didnt say which device Nov 24 16:16:28 so its shopping time Nov 24 16:16:49 * KotH just came back from shopping Nov 24 16:17:04 I understand that it's not grub (though even grub's not what it used to be). Nov 24 16:19:26 koth chocolate? Nov 24 16:20:05 The problem seems to be that uboot doesn't just have default values. It has default RULES, which can do all sorts of things depending on what they are. And they're embedded in a binary file. Nov 24 16:20:20 uboot has many ways Nov 24 16:20:42 woglinde: nope cheese this time Nov 24 16:20:56 koth which one? Nov 24 16:21:05 swiss asortment Nov 24 16:21:35 asortment Nov 24 16:21:36 17:20 < woglinde> uboot has many ways <- sounds zen'y Nov 24 16:21:51 sure Nov 24 16:21:59 zen is all you need in embedded Nov 24 16:22:23 Is there a way to get these embedded rules back out, so I can put them in uenv.txt where they belong? Nov 24 16:26:13 use the source code and one of the tousand blogs article for it Nov 24 16:26:15 Serial login looks okay at first, showing the angstrom banner, but when I login I get binary garbage. Is this a known problem? Nov 24 16:28:23 now rest of family is ready Nov 24 16:28:23 woglinde: Also, I'll take that as a 'no, the embedded rules are stuck indelibly and you will never know exactly what they were doing'. Fine. Nov 24 16:28:26 till later Nov 24 16:28:49 monttyle o.O source code does mostly predictable Nov 24 16:30:39 No bootloader is going to have dev_dsp="dvi" built into the source. I'm trying to find out what options got compiled into this. Nov 24 16:35:13 if I had more time to waste, I'd build a u-boot with that option built in Nov 24 16:37:13 If you were building a distribution though, I hope you'd document that Nov 24 16:38:27 My goal at this point is to build a system which does what I tell it to. So far I only have systems which do as they please or what other people have told them to. Nov 24 17:29:40 ok, anyone know how i can get uart2 to work like uart0? Nov 24 17:29:58 i need to send commands to the beagleboard to shut it down Nov 24 17:30:13 from serial/uart2 Nov 24 17:31:39 You mean, you want that serial port to have a login screen? Nov 24 17:34:06 My /etc/inittab has a line like '1:2345:respawn:/sbin/getty 115200 ttyS2 ' causing it to run getty on /dev/ttyS2, which lets people login on it. Nov 24 17:36:35 i see 2 possible solutions: make serial2 work like serial0 Nov 24 17:37:03 or use some script to echo commands from usart to bash Nov 24 17:38:05 sabesto: So you want to join two serial devices together, rather than use them separately? Nov 24 17:39:15 basicly, its in a box with a beagle + wifi router and an AVR inside Nov 24 17:39:29 the avr is connected to a switch and the serial2 port Nov 24 17:40:28 when i hold the switch for 5s the avr will send a shutdown command to the beagle Nov 24 17:40:43 Ah, so it needs to read the port, but it is NOT reading shell commands, just text Nov 24 17:41:38 i could send plain shell commands too Nov 24 17:41:55 thats maybe the easiest for now Nov 24 17:41:58 Would having a normal login screen available on that port do? Nov 24 17:42:06 yeah Nov 24 17:42:18 for now, i will write a c++ program later Nov 24 17:42:32 i just need a dirty solution for now Nov 24 17:42:43 Like I said, that's controlled by /etc/inittab. There's probably a line in yours already to run the 'getty' login program on your other serial port. Just make another line like it. Nov 24 17:42:57 ok, i will check, thanks Nov 24 17:42:58 When you make your C++ program, have /etc/inittab run it instead of getty. Nov 24 17:43:40 yeah Nov 24 17:44:07 the box must be sent to a big show in germany by tomorrow :P Nov 24 17:44:17 just cutting power corrupts the SD-card Nov 24 17:44:28 now and then Nov 24 17:45:57 Ouch. I had to drive 500km to fix a disk corruption problem like that just yesterday. The only partition that seems to get corrupted is /var/cache/squid, and I don't care about its contents on reboot, so I made a real brute force solution this time. Set it noauto in /etc/fstab, and on every boot, mkfs.ext3 /dev/sda6 ; mount /var/cache/squid before squid starts :p Nov 24 17:46:52 It's a good argument for not keeping everything on one partition. Nov 24 17:48:13 monttyle: well, thanks for your help, i will try this in an hour. need to get some food Nov 24 17:48:22 Good luck. Nov 24 17:50:13 Oh -- don't forget that your C++ program will probably need to set baud rate and such. That's usually getty's job. If you have stty, the fastest solution would probably be system("stty 115200 etc etc etc"); Nov 24 18:19:21 woglinde: so it's better to just build on the beagle vice cross compiling? Nov 24 18:50:13 wog's not here. I'd consider it preferable. Far fewer variables to go wrong. Nov 24 18:52:11 speak of the devil Nov 24 18:52:45 777 Nov 24 18:52:49 re Nov 24 18:53:39 Is there anything special about the u-boot files? Do they need to reside at sector zero or somesuch? Or is the bootstrapper able to just pick them via names from a fat filesystem? Nov 24 18:54:05 uenv.txt can be read from fat partition Nov 24 18:55:24 I know it can be, I'm asking whether the arrangement is any deeper than the file level. I don't think that beagleboard uses the boot-sector the way x86 does. Nov 24 18:56:59 So assuming I've set up the partitions the right way -- correct number of heads, sectors, and cylinders in the manner ppcboot wants -- is having the right files in the right places all uboot needs? Nov 24 18:57:08 CHS does not matter Nov 24 18:57:13 sd cards are not cylindrical Nov 24 18:57:48 It doesn't physically matter, no. It's just a number. It does matter to the boot system, though; it's picky. Nov 24 18:58:00 it's just a number and _nothing_ so much as glances at it Nov 24 18:58:33 Then why do all the tutorials I see for making boot cards for beagleboard show quite specifically how to set the number of heads, sectors, and cylinders? Nov 24 18:58:39 because they lie Nov 24 18:59:17 Okay. And why do the cards I bought for beagleboard have the number of heads, sectors, and cylinders set in that lying way? Nov 24 18:59:41 unfortunately the fields exist in the partition table Nov 24 18:59:43 <_av500_> cargo cult Nov 24 18:59:45 they have to have _some_ value Nov 24 18:59:46 because of cargo cult Nov 24 18:59:48 <_av500_> CHS is irrelevant Nov 24 18:59:53 <_av500_> the ROM code ignores it Nov 24 18:59:54 * dm8tbr high-5-s _av500_ Nov 24 18:59:56 <_av500_> go and try Nov 24 18:59:57 <_av500_> I did Nov 24 19:00:04 <_av500_> wrote 00 or FF, still booting Nov 24 19:00:38 Well, sure. But did you regenerate your partitions to that? Nov 24 19:00:56 wtf do you mean by that? Nov 24 19:01:11 <_av500_> monttyle: here: https://groups.google.com/forum/#!msg/pandaboard/ro5k5r4Cuq4/pLSkA46s_D0J Nov 24 19:01:11 monttyle: if you think you're smarter than we are, go ahead you don't seem to need our help... Nov 24 19:01:12 partitions know only sectors Nov 24 19:01:21 or blocks Nov 24 19:01:34 <_av500_> monttyle: of course not Nov 24 19:01:43 if you think you know better than us, why do you bother asking? Nov 24 19:01:49 <_av500_> I left the partition start and size as is Nov 24 19:01:53 <_av500_> any wrote random CHS Nov 24 19:01:56 <_av500_> still boots Nov 24 19:02:00 <_av500_> hence, CHS is irrelevant Nov 24 19:02:18 <_av500_> the TI document on the rom code says the same Nov 24 19:02:23 <_av500_> adly its not public Nov 24 19:02:24 I don't know everything, I'm not completely ignorant either. I've used PC linux for years, and having a hard time making the transition. Nov 24 19:02:25 <_av500_> sadly Nov 24 19:02:31 <_av500_> bbl Nov 24 19:02:53 I never met a PC that cared either Nov 24 19:03:23 No, PC's use a completely different boot system. Nov 24 19:05:55 Anyway. Once partitions are made, is installing u-boot as simple as copying files, or is it a procedure like grub or lilo where special values need to be written in ordinarily inaccessible places? Nov 24 19:06:07 just copy the files Nov 24 19:06:19 Great, thanks Nov 24 19:27:46 hes truggling now for hours? Nov 24 19:28:40 hello Nov 24 19:29:42 sime3d yes? Nov 24 19:31:15 i am using beagleboard-xm with ubuntu 12.04 Nov 24 19:31:35 that's impressive, I didn't know ubuntu was usable on the beagle Nov 24 19:31:45 and i am try to run a program with opencv Nov 24 19:33:51 i got something like this Nov 24 19:33:52 uvcvideo: Failed to resubmit video URB (-27). Nov 24 19:34:28 welcome to usb hell Nov 24 19:34:39 someone said USB? Nov 24 19:35:03 sime3d : so what is -27 ? Nov 24 19:35:10 $ perror 27 Nov 24 19:35:10 OS error code 27: File too large Nov 24 19:35:23 seems like someone is donitrong Nov 24 19:35:31 find and fix Nov 24 19:35:33 .. Nov 24 19:35:34 profit Nov 24 19:37:23 CareBare\ when you say File too large do you mean the Frame from the webcam is too big or what exactly? Nov 24 19:37:52 could be anything Nov 24 19:38:09 find out where that error is being reported and why Nov 24 19:42:14 i got this error after receiving 50 frame from usb camera then usb port stop working even the ethernet. and by using rs232 i was able to find out this uvcvideo: Failed to resubmit video URB (-27). Nov 24 19:42:46 i google it and find nothing Nov 24 19:43:05 can anyone help me fix it ? Nov 24 19:43:07 that looks like an easily greppable message Nov 24 19:52:56 sime3d : webcams never send files. the fact that you express yourself that way suggests that you aren't experienced with devices and device drivers, and I think that will make it very time consuming for you to fix this problem. Nov 24 19:53:16 sime3d : yes, there are lots of people who can help you fix the problem. Nov 24 19:54:36 sime3d : if you do not want to or can not learn to fix the problem on your own then you have to get someone else to do it for you. the way that usually works is by hiring someone. but consider that the market is starved for anyone who is actually skilled at Linux, so you will have to search hard, or pay a lot of money. Nov 24 19:55:38 sime3d : if you do want to learn then you can get excellent help for free, but the way that arrangement works is that you learn exactly everything you need to learn in order to reach a little bit under the developers that know enough to fix the problem - you need to do that in order to express the problem to them in a way that is useful to them. Nov 24 19:55:47 sime3d : those are your only two options, really Nov 24 19:56:29 sime3d : one takes a lot of money, the other takes a lot of time. if you have neither, your project fails. it's really that easy. Nov 24 20:15:38 you can not get 50 fps on the beagle via usb Nov 24 20:15:41 only 25 Nov 24 20:16:13 I should have bookmark the gumstix forum thread about it Nov 24 20:17:22 <_av500_> CareBear\: I just realized you are the one from the MTP thread Nov 24 20:29:13 av500 yes carebear can take over from us Nov 24 20:30:34 av500 : I am Nov 24 20:32:47 woglinde : do you know why? Nov 24 20:32:53 woglinde : 25fps, that is Nov 24 20:33:03 hello, how to downgrade kernel on BB-xm? I had 3.2.34, downloaded http://rcn-ee.net/deb/squeeze/v3.0.8-x3/ , ran install-me.sh, files marked 3.0.8 are in /boot but board still boots to 3.2.34... debian_squeeze Nov 24 20:33:33 carebear color conversion Nov 24 20:34:44 thomas because the kernel is maybee booted from the first partition and not from /boot check your uboot setup Nov 24 20:36:20 so how to do the downgrade? Nov 24 20:37:27 I was looking into all the files in the first partition Nov 24 20:37:56 I was expecting to find something similar to grub configuration, but I have no clue how to change the configuration Nov 24 20:37:59 Thomas4 : learn about how the kernel is started, so that you can change the setting in the right place Nov 24 20:38:29 Thomas4 : grub is almost entirely PC-specific Nov 24 20:38:35 Thomas4 : where by PC I mean x86 Nov 24 20:39:11 Thomas4 : boot firmware is a very interesting topic, which on x86 has 35+ years of legacy that you'll have to study for a full understanding. not many have that, unfortunately :\ Nov 24 20:39:22 Thomas4 : beagle is arm Nov 24 20:39:31 I know there is no grub Nov 24 20:39:36 Thomas4 : a different machine, different firmware Nov 24 20:39:38 I was looking for something similar Nov 24 20:39:44 like configuration file Nov 24 20:39:52 why would there be a configuration file? Nov 24 20:39:54 CareBear\: I'd say it's fortunate Nov 24 20:40:02 carebear hm as I said I dont find the thread about this again Nov 24 20:40:06 the asylums would be overflowing if everybody had Nov 24 20:40:24 mru : we would also not have UEFI Nov 24 20:40:48 I hear they're building a new asylum as we speak Nov 24 20:40:59 it's at Microsoft Nov 24 20:41:04 there's one at Intel too Nov 24 20:42:27 Thomas4 : look into the firmware. the firmware is u-boot on the beagle as shipped. u-boot is not quite as large a topic as x86 firmware, but will still take a while to get into Nov 24 20:42:31 Thomas4 : google: u-boot Nov 24 20:42:35 Thomas4 : go from there Nov 24 20:43:56 I am looking all the time Nov 24 20:45:32 <_av500_> woglinde: color conversion? Nov 24 20:45:40 <_av500_> who forces you to color convert? Nov 24 20:49:17 av500 when I find thr thread again I can tell you Nov 24 21:46:47 ho ho hum Nov 24 21:48:06 hi prpplague Nov 24 21:48:43 woglinde: greetings Nov 24 21:50:29 <_av500_> ho prpplague Nov 24 21:51:26 _av500_: saw a fabulous south western hawk at the car park of the local wal-mart Nov 24 21:51:40 _av500_: he/she downed a pigeon right before our eyes Nov 24 21:53:58 <_av500_> saw that on g+ Nov 24 22:00:25 prpplague cool Nov 24 22:25:38 no shortage of pigeons in this world. Nov 24 22:47:51 w00t! Beaglebone FPGA cape is alive - just downloaded a bitstream into the FPGA via SPI0. Nov 24 22:48:14 emeb : you'd better be bitbanging that LED! Nov 24 22:48:36 CareBear\: damn straight. Nov 24 22:48:41 Both of 'em. Nov 24 22:49:37 emeb nice Nov 24 22:49:41 ty Nov 24 22:49:52 what fpga is it? Nov 24 22:50:06 the good answer would be lx9 :) Nov 24 22:50:12 Xilinx XC3S200A - Spartan 3A Nov 24 22:50:17 aw Nov 24 22:50:19 CareBear\: sorry. Nov 24 22:50:34 Hey - Spartan 3A are pretty good. Nov 24 22:50:47 lotsa on-chip RAM & multipliers. Nov 24 22:50:57 I've done some serious work with them. Nov 24 22:50:58 emeb : a week ago I wouldn't have cared much either way. then I saw an open source toolchain for lx9 Nov 24 22:51:16 hm whats the lx9? Nov 24 22:51:22 woglinde : spartan-6 Nov 24 22:51:24 xilinx I know a bit Nov 24 22:51:28 second smallest device in that range Nov 24 22:51:28 ah okay Nov 24 22:51:31 hmpf. I'd be really surprised if you can get much out of that besides their testcase. Nov 24 22:52:20 emeb : tooling is important to me. I'd prefer a sucky toolchain that is pleasant to use but outputs less than stellar results Nov 24 22:53:03 CareBear\: FPGAs are complex enough that you don't want to be fighting lots of bugs in the toolchain. Nov 24 22:53:04 emeb : I certainly want to try to get much out of it. did you follow the development of that project? Nov 24 22:53:21 CareBear\: Nope - just heard of it a few weeks ago. Nov 24 22:53:35 It's a fantastic effort Nov 24 22:53:40 how do you mean FPGAs are complex? Nov 24 22:53:43 it's just hardware Nov 24 22:54:20 *g* Nov 24 22:54:32 CareBear\: getting decent utilization of the available resources requires more than just knowing what bits go where. Nov 24 22:54:45 sure Nov 24 22:54:53 synthesis, place & route, etc are all extremely complex subjects Nov 24 22:55:17 as I said, I would like it, if I could try a new toolchain Nov 24 22:55:18 and the parameters are highly dependent on the underlying hardware architecture. Nov 24 22:55:23 yes of course Nov 24 22:56:03 So, trying to get reasonable performance from a FOSS toolchain without lots of support from the vendor is an uphill battle in the extreme. Nov 24 22:56:06 IBM said what you're saying now, when gcc was new Nov 24 22:56:14 No doube Nov 24 22:56:21 s/e/t/ Nov 24 22:56:41 as for uphill battles and information from vendors, know that I have 11+ years of experience from the coreboot project Nov 24 22:56:58 where this is also a well-known topic :) Nov 24 22:57:25 So you understand that it may be a while before you can get good results from a FOSS FPGA toolchain. Nov 24 22:57:38 stop talking about good results Nov 24 22:58:02 Overall it depends on what you want to do. Do you want to spend your time debugging tools, or do you want to get FPGA designs done. Nov 24 22:58:19 depends Nov 24 22:58:24 Don't get me wrong - I'm glad they're doing this. I like FOSS too. Nov 24 22:58:28 not everyone needs to make money Nov 24 22:58:32 Use it 90% of the time. Nov 24 22:59:01 emeb : tooling is very important to me Nov 24 22:59:15 CareBear\: that's great. Your itch to scratch. Nov 24 22:59:36 my itch is making hardware with the path of least resistance. Nov 24 22:59:42 emeb : and tooling has other qualities than just their results Nov 24 22:59:56 its Nov 24 23:00:21 so in 20-30 years time we might have good open source hw design tools Nov 24 23:00:28 bingo Nov 24 23:00:41 huh? Nov 24 23:00:43 in the meantime what are we supposed to do? Nov 24 23:00:50 who said that there are any good open source software tools? Nov 24 23:01:16 22:56 < CareBear\> IBM said what you're saying now, when gcc was new Nov 24 23:01:33 mru : do you consider gcc good? Nov 24 23:01:51 it's as good as the competition in most cases Nov 24 23:01:57 orly? Nov 24 23:02:06 nobody is as good as mru ;) Nov 24 23:02:32 gcc has improved by leaps and bounds the last 3-4 years Nov 24 23:02:42 emeb : you mentioned path of least resistance - the vendor toolchains have a good deal of resistance to them as well Nov 24 23:03:02 CareBear\: what do you see as the biggest problems with the vendor toolchains? Nov 24 23:05:15 emeb : too many layers, leading to in some cases really difficult integration/automation - and of course it's fun that they are so many gigabytes large Nov 24 23:05:32 CareBear\: they are large no doubt. Nov 24 23:05:50 but my experience is that they can be trimmed down without too much trouble. Nov 24 23:06:12 which ones did you do that with? Nov 24 23:06:17 (I've got an install of Xilinx ISE that fits on a 4GB SD card for my Ubuntu netbook) Nov 24 23:06:18 it's a good idea Nov 24 23:07:03 and if you like standard build processes like make then you can build FPGAs entirely with command line tools. Nov 24 23:07:20 I've got some fairly large chip builds that are all done via make. Nov 24 23:07:33 no need for nasssty guis. :) Nov 24 23:07:49 yes, it works for xilinx, did you do it with others as well? Nov 24 23:08:33 the only problem with the stripping down thing is that it reminds so much of installing slackware from floppy disks Nov 24 23:08:42 CareBear\: haven't tried others. I use some Altera parts, but I'm not the build manager for those. Nov 24 23:08:49 ack Nov 24 23:09:05 microsemi is even more difficult Nov 24 23:09:31 No doubt there are some 2nd-tier folks out there with disagreeable tools. Nov 24 23:09:53 Ideally, over time I'd like to see something like gcc take over for hardware design. Nov 24 23:10:05 oh don't go trolling other brands than your favorite just because they do things differently! :) Nov 24 23:10:20 a standard front-end with back-ends targeting lots of different players. Nov 24 23:10:30 yes, that would be nice Nov 24 23:10:38 it requires cooperation Nov 24 23:10:54 Eventually, if this project gets some legs then it's conceivable they could get support from the vendors. Nov 24 23:11:16 I'm sure the vendors would love to not have to roll their own tools. Nov 24 23:11:50 the big loser would be synopsys Nov 24 23:12:05 In the long run, perhaps. Nov 24 23:12:07 they'd still have to write the backend for the common tool Nov 24 23:12:12 since they would suddenly go out of business :) Nov 24 23:12:18 like they do for gcc these days Nov 24 23:12:21 mru: no doubt. Nov 24 23:12:43 but if the synthesis and place/route infrastructure was already there that makes their job a lot easier. Nov 24 23:13:19 certainly Nov 24 23:13:33 anyway, I am eager to experiment Nov 24 23:13:39 Funny thing is that most of the big synthesis vendors got their start using FOSS code that came out of .edu research. Nov 24 23:14:18 CareBear\: sound like a fun project. Do you have an lx9 devboard to try it out on? Nov 24 23:14:24 emeb : I don't Nov 24 23:14:29 Hmmm. Nov 24 23:14:35 there aren't very many candidates Nov 24 23:14:54 Well, ka6sox and I have been looking at doing a Spartan6 version of this. No big plans yet though. Nov 24 23:15:04 what is on there besides the fpga? Nov 24 23:15:34 and do you currently use 4 or 2 layers? Nov 24 23:15:40 FPGA, power supply, I2C ID PROM, configuration flash memory. Nov 24 23:15:47 it's on 4 layers Nov 24 23:15:57 what footprint do you use? Nov 24 23:16:06 FPGA is hooked up to the GPMC and to the SPI0 bus Nov 24 23:16:32 that's the way to do it, yes Nov 24 23:16:33 This is the VQ100 part. Here's a website with more details: http://ebrombaugh.studionebula.com/embedded/bcc/index.html Nov 24 23:17:24 I don't suppose the lx9 would be pin compatible.. Nov 24 23:17:41 CareBear\: no, LX9 only comes in a qfp144 and BGA packages. Nov 24 23:17:48 mm Nov 24 23:18:03 It's larger footprint means it'll be harder to shoehorn into the available space. Nov 24 23:18:08 yes Nov 24 23:18:21 can you show me a png from pcb? Nov 24 23:18:46 lemme check... Nov 24 23:20:04 http://imagebin.org/237080 <- render from PCB Nov 24 23:20:31 I was hoping for a screenshot with transparency :) Nov 24 23:21:12 Oh, just a moment... Nov 24 23:22:36 http://imagebin.org/237081 Nov 24 23:22:56 oh.. Nov 24 23:23:01 well it's not *too* crowded Nov 24 23:23:46 Could be less if I re-did it with 6/6 rules. That's 8/8 I think Nov 24 23:25:04 the back seems mostly clear Nov 24 23:25:49 yeah - not much there except decoupling caps for the FPGA supplies Nov 24 23:26:07 a few routes Nov 24 23:26:23 *one* quick way would be to put the tq144 on the back with lots of vias up top Nov 24 23:26:52 maybe. Nov 24 23:27:24 I think it would fit on the front though - I've seen other cape-sized boards with tqfp144 parts on them. Nov 24 23:27:41 yes, can fit, but more work Nov 24 23:27:53 we're not afraid of work! Nov 24 23:28:40 different topic: how do you do the outline/mounting hole thing for the cape? do you have a cape footprint? Nov 24 23:29:27 and what about the J102-103 Nov 24 23:29:48 No - I used the measurements in the TRM and did it by hand. Nov 24 23:29:49 I mean: do you have it all packaged up somehow neatly in a single cape $something? Nov 24 23:29:54 nod, ok Nov 24 23:30:24 It would be fairly simple to take this .pcb file and strip it down though. Nov 24 23:30:30 sure Nov 24 23:30:34 use it as a starter for other boards. Nov 24 23:30:47 I think that would be cool for those who want to do capes Nov 24 23:30:56 I was asking more generally though Nov 24 23:32:00 It wouldn't be too hard to convert this outline + connectors into a component. Nov 24 23:32:12 is true Nov 24 23:32:24 text formats ftw Nov 24 23:32:36 then you could slap down a bone on a carrier board for larger projects. Nov 24 23:33:23 yeah - gEDA/PCB has some warts, but it's easy to work with the file formats. Nov 24 23:34:45 I like pcb Nov 24 23:35:37 there are two open hardware minimal lx9 board designs, both using kicad Nov 24 23:37:11 kicad is OK. It does much nicer polygon fills than pcb. Nov 24 23:37:43 I've done a couple designs with kicad but pcb feels more natural to me. Nov 24 23:38:11 and I can run pcb reliably on my Mac, while kicad on the Mac is still very green. Nov 24 23:39:13 I didn't use kicad yet Nov 24 23:39:23 it seems quite french Nov 24 23:39:37 whatever that means. :) Nov 24 23:39:57 well the wording is wacky for one Nov 24 23:40:28 also dealing with a language barrier when trying a new tool isn't awesome :) Nov 24 23:40:39 hmm. never noticed that. Nov 24 23:41:36 bon soir Nov 24 23:42:11 * _troll_ alert! Nov 24 23:43:19 hola mranostay Nov 24 23:46:22 good night Nov 24 23:53:14 hi emeb Nov 24 23:53:49 mranostay: got the FPGA working on the Bone. Nov 25 00:01:52 awesome Nov 25 00:02:14 does someone have a plan to use it for something already? Nov 25 00:05:30 emeb : maybe 3x32 2mm pitch headers would fit on a tq144 spin Nov 25 00:06:56 CareBear\: ka6sox has some apps in mind. He's working on the gpmc driver for it. Nov 25 00:07:27 aha Nov 25 00:08:04 dunno about 2mm headers. I've also done some work w/ 50mil headers that seems OK. Nov 25 00:08:47 they're fine too, but a bit less common, and not as easy to work with Nov 25 00:09:05 the 2mm ones are a sweet spot for me Nov 25 02:28:09 hmmm nifty Nov 25 02:28:22 2 level translators and 6ft of cable later, I2C still works Nov 25 02:32:38 ds2: sweet! Nov 25 02:34:03 ds2: then put a gpio expander on, and i2c-gpio another bus, shift level back to original voltage and see if you can get all the way back to your mainboard Nov 25 02:36:48 Not pushing my luck on it Nov 25 02:37:24 compass works, accelerometer works... next to check is gpio expander and touch screen controller Nov 25 02:40:44 ds2: nice. What are you making? Nov 25 02:40:59 (if you're allowed to say) Nov 25 02:42:44 alan_o: sensor board for the Beagle class and xM and Panda Nov 25 02:43:02 oh cool. Nov 25 02:43:23 complete with a prototyping area for testing other stuff Nov 25 02:43:28 no soldering required for normal use Nov 25 02:44:09 make it have those little spring connectors like those electronic kit boards they make for kids and sell (or used to) in radio shack Nov 25 02:44:17 no need Nov 25 02:44:18 1000-in-1 Nov 25 02:44:34 That would be cool to have one of those on a beagle cape Nov 25 02:44:38 the main chips don't require any soldering Nov 25 02:45:01 no reason why this can't be plugged into a cape; but that'd require soldering Nov 25 02:45:07 all I2C signals are available Nov 25 02:45:19 other 3.3V I2C devices can be easily wired in Nov 25 02:45:51 yeah, a level-shifter proto-board is a good idea Nov 25 02:45:54 even without the sensors Nov 25 02:46:03 for xm and panda that is. Nov 25 02:46:09 I'd have bought one of those..... Nov 25 02:46:12 might still :) Nov 25 02:46:15 it can be used for that Nov 25 02:46:35 and it will coexist with my LCD interfaces Nov 25 02:46:46 no need for stacking or any of the other shannagians Nov 25 02:47:06 do xm boards typically stack? Can't remember Nov 25 02:47:29 I like prpplague's zippy boards. I miss them :( Nov 25 02:47:32 don't matter... this board doesn't care Nov 25 02:47:39 pass through? Nov 25 02:47:53 tired the arguement about male or female connectors Nov 25 02:48:03 yeah Nov 25 02:48:03 no, grabs the signal in a different way Nov 25 02:48:21 doesn't use the expansion connector? Nov 25 02:48:28 the as shipped, default connectors on the xM blows chunks... gives you a 3 inch tall stack to do anything Nov 25 02:48:38 i flip it over to get thinner stack Nov 25 02:48:38 nope **** ENDING LOGGING AT Sun Nov 25 02:59:59 2012