**** BEGIN LOGGING AT Thu Feb 14 03:00:00 2013 Feb 14 03:04:05 _chase_: hey we have been waiting for you to log into irc Feb 14 03:04:16 _chase_: a serious burning question for you Feb 14 03:04:46 <_chase_> prpplague: ok Feb 14 03:04:49 _chase_: got slides? Feb 14 03:04:52 hehehe Feb 14 03:05:02 * prpplague jokes with _chase_ Feb 14 03:05:08 <_chase_> prpplague: yes, but i don't have Kevin's buy-in Feb 14 03:05:14 <_chase_> so let's blame him :) Feb 14 03:05:30 _chase_: hehe Feb 14 03:05:56 <_chase_> but I will get his attention soon. We are almost done with our meetings in India Feb 14 03:06:05 _chase_: just joking with you Feb 14 03:06:13 _chase_: everyone is behind this time around Feb 14 03:06:40 <_chase_> so no Bill Gates? Feb 14 03:06:45 <_chase_> I'm disappointed Feb 14 03:07:00 hehe we were able to postpone the "gates of hell" opening Feb 14 03:17:41 prpplague: +1 Feb 14 03:18:38 mranostay: 20 minutes not wasted? Feb 14 03:19:56 prpplague: i don't get why this meme is so awesome but it is :) Feb 14 03:20:06 hehe Feb 14 03:20:57 mranostay: i say the same thing about this one - http://www.youtube.com/watch?v=zn7-fVtT16k&feature=share&list=PLQ-7WiWmOuK-55mfcd_tdcvy-57VMCkOW Feb 14 03:29:09 heh Feb 14 03:33:57 prpplague: i think koen- has the most flamebait talk title :) Feb 14 04:06:46 Anyone cross compile node.js for arm, I have a question about the supporting install files Feb 14 04:12:19 how is this a #beagle question? Feb 14 04:12:33 because node.js is badass rockstar tech Feb 14 04:12:42 oh right i forgot Feb 14 04:12:45 thanks Russ Feb 14 04:15:19 beagle is my target, was trying to find some overlap, nobody around to help in #Node.js Feb 14 04:28:35 eephillip, better luck waiting in node.js Feb 14 04:33:14 yeah no biggie, maybe manana Feb 14 08:15:57 a good, JIHAD morning everyone.. may your bombs have the shapes of hearts, today! Feb 14 08:35:04 morning Feb 14 09:23:31 for the msp430 enthusiasts here... and those who like to drive others crazy: http://renaud.schleck.free.fr/horloge_vetinari.php Feb 14 09:32:15 KotH: cool! I like it! thnx. Looks like I need to find some more spare time! Feb 14 10:11:38 :q Feb 14 10:11:45 doh... Feb 14 10:24:10 I guess your favorite cmdline editor isn't nano... Feb 14 10:25:21 does anyone but ubuntu users use nano? Feb 14 10:26:30 mm... I do use ubuntu... Feb 14 10:28:17 why nano Feb 14 10:28:24 pico Feb 14 10:28:37 yocto? Feb 14 10:35:50 VIM! Feb 14 11:02:07 What would everyones favourite filesystem for nand flash be from ext4, jffs2, ubifs, yaffs2 or maybe samsungs new f2fs? Feb 14 11:25:42 jackmitchell ubifs yaffs2 or f2fs Feb 14 11:25:52 but should make some speed tests Feb 14 11:26:55 isn't f2fs ment for nand-using-devices like usb-sticks / sd-cards (so with built-in wear-levelling etc) so not really comparable with the other 'raw-nand' filesystems? Feb 14 11:27:23 tsjsieb and? Feb 14 11:27:33 in the end speed and safety matters Feb 14 11:27:42 and you have to evaluate it Feb 14 11:28:22 I wouldn't like it on my raw-NAND like I wouldn't like ubifs on my sd-card... Feb 14 11:28:25 nilfs2 is pretty fast too (but still immature afaik and it does a lot of stuff in userspace) Feb 14 11:28:31 woglinde: fsfs doesnt have any handling of badblocks, it relies on an underlying layer to handle that... afaik Feb 14 11:30:38 hmm, I'm interested mostly in not corrupting data on hard reset Feb 14 11:31:12 I know there are systems where you can have a ramfs type layer which then propogates down Feb 14 11:31:33 what hardware doesn't handle its own bad blocks these days? Feb 14 11:31:38 (other than raw flash) Feb 14 11:32:03 (but yes, fsfs isn't for raw flash) Feb 14 11:32:22 okay Feb 14 11:32:27 than yaffs2 or ubifs Feb 14 11:32:31 till later Feb 14 11:32:36 ubi in that case Feb 14 11:32:37 I recommend ubifs Feb 14 11:33:27 ok, cheers everyone - will look at that one first Feb 14 11:34:08 I recommend ubifs in combination with not hard resetting to often Feb 14 11:42:47 ubifs or ubifs Feb 14 11:42:56 mdp: :) Feb 14 11:48:34 or squashfs on ubiro Feb 14 11:48:54 er, ubiblk Feb 14 11:50:43 jffs2 still has perf advantages on nor parts 16MiB and smaller Feb 14 11:51:06 any recommendations for punched cards? Feb 14 11:51:21 and its non-deterministic scrubbing algo isn't a concern there :) Feb 14 11:51:37 punched cards are write once Feb 14 11:51:44 so i would go for something simple Feb 14 11:51:46 with little overheads Feb 14 11:51:48 like fat Feb 14 11:51:55 mdp, I see http://elinux.org/Flash_Filesystem_Benchmarks_2.6.39 Feb 14 11:51:55 jacekowski: you can add holes to cards Feb 14 11:52:06 mru: write once Feb 14 11:52:09 not read only Feb 14 11:52:18 you can add holes more than once Feb 14 11:52:28 you can close the holes with tape Feb 14 11:52:53 not easily Feb 14 11:53:06 tape might pass enough light to make it misread the holes Feb 14 11:53:10 punch cards have full support for subpage writes Feb 14 11:53:34 jacekowski: use metal tape Feb 14 11:53:43 metal is hard to punch Feb 14 11:53:57 use black tape ? Feb 14 11:54:30 like those stickers used to cover the write-protect cutout on old floppy discs Feb 14 11:54:46 yeah Feb 14 11:54:58 those were more often mechanical than light based Feb 14 11:55:18 yeah, light based tape is rare Feb 14 11:55:32 i'm talking about floppies Feb 14 11:55:32 the 5" and 8" ones are light based Feb 14 11:55:39 3.5" are mechanical Feb 14 11:55:44 tape used in colossus was light based Feb 14 11:56:03 http://en.wikipedia.org/wiki/Colossus_computer Feb 14 11:56:25 Russ, that's a 2 year newer version of what I'm citing from memory. previous study had a matrix of part types, cpu speeds, fs sizes, and read write size patterns Feb 14 11:56:54 https://lh6.googleusercontent.com/-qwOkjnU7wFc/UFnZ94X9GJI/AAAAAAAACe0/npCGdGZcjiA/s0/ Feb 14 11:56:57 it nicely showed that jffs2 worked best for what it was designed for Feb 14 11:57:04 what about the new version of ubi with faster start times? Feb 14 11:57:40 ubi no longer has to scan all erase blocks Feb 14 11:57:52 Russ, yeah, have only reviewed that...it changes everything Feb 14 11:58:14 strangely, almost had a customer pay us to do that...darn Feb 14 11:58:46 as we were playing games with deferred mounting to deal with the scan Feb 14 11:58:54 glad to see it done Feb 14 11:59:49 I think I'll stick with pcie work...nand is fugly :) Feb 14 12:49:39 mdp: finally got my awesome openbench logic sniffer Feb 14 12:49:44 mdp: let me play with it for a while :-) Feb 14 12:49:51 heh, nice email on pandaboard list Feb 14 12:50:19 "I want to make the *Counter Program for Pandaboard ES.* Please suggest step for that." Feb 14 12:51:24 hello Feb 14 12:52:24 echo Feb 14 12:55:01 mru: 1) open editor 2) write Counter Program for Panaboard ES 3) compile 4) run program on pandaboar Feb 14 12:55:40 mru: ofc you can make that a one point exact step answer as well: 1) write, compile and run Counter Program Feb 14 12:57:43 he didn't say he wanted to run it Feb 14 12:58:23 and he didnt say it would have to be binary form ... Feb 14 12:59:10 he didn't say a whole lot of anything, in fact Feb 14 12:59:10 while true; do i=$(($i+1));echo $i;sleep 1;done Feb 14 12:59:47 pandaborat! Feb 14 13:00:07 http://pandaborad.org/ Feb 14 13:01:38 mru: that was sick. I'm calling WWF Feb 14 13:01:54 I am merely the messenger Feb 14 13:03:36 tha's like the beagleboar Feb 14 13:12:56 * KotH hates it when people sort functions in reverse order and use forward declarations Feb 14 13:13:39 there's one thing that's worse: ifdefs around forward declarations Feb 14 13:13:57 ? Feb 14 13:14:03 why? how? what for? Feb 14 13:14:10 #ifdef FOO Feb 14 13:14:13 void foo(int); Feb 14 13:14:15 #endif Feb 14 13:14:54 well..........kill them Feb 14 13:14:58 with fire Feb 14 13:15:01 from orbit Feb 14 13:15:02 with nukes Feb 14 13:15:24 or just with an axe if you have nothing at hand Feb 14 13:16:02 I'm partial to OADS Feb 14 13:16:11 works too Feb 14 13:16:29 although it doesnt give you much bang for the buck, it's funny enough Feb 14 13:19:05 Right, I forgot who it was yesterday on here mru or someone else, but they suggested that marking the SD partition as active for my little problem, I did it explicitly with diskpart, and boom it worked Feb 14 13:20:04 for good suggestions we all like to be held responsible... Feb 14 13:20:55 I should send in a suggestion to TI, as that little peice of advice could save alot of people time Feb 14 13:21:03 it's in the docs... Feb 14 13:21:41 mru, do you know where in the docs, because I mainly used the document that came in the starterware package Feb 14 13:22:06 the booting section of the trm should mention it Feb 14 13:22:21 okay I will check Feb 14 13:22:37 but anyway, that little bit of advice made it work Feb 14 13:23:11 I was about to go back to FPGA's. I will actually later on, I am just using the beagle bone for initial prototyping Feb 14 13:23:54 anyway, super thanks Feb 14 13:28:24 HappyHippo: you are now required to send mru a bag of happyhippos Feb 14 13:29:02 coming from an FPGA background, I am used to, less reading, and more logic gates, flip flops, etc etc Feb 14 13:31:01 well.. fpgas are not half as complex as a modern uC... and modern SoC are in a totally different category Feb 14 13:31:25 hmm... s/complex/complicated/ Feb 14 13:31:34 same thing Feb 14 13:33:42 depends, altera MAX are very simple things, however, fpga's will give you super duper timing precision when you want it, they will run everything at once, if you stick several filters, frequency counters, dsp cores, etc etc, it will all run at once, Feb 14 13:33:42 * KotH thought there is a slight difference Feb 14 13:33:46 mm though I never think about VHDL as 'hot half as complex' Feb 14 13:34:04 also, most modern FPGAs have system on chip libraries Feb 14 13:34:28 HappyHippo: you're in a channel where half of the people are EEs and some have been doing chip design when VHDL was the new kid on the block.... Feb 14 13:34:29 at my old work place, we used to throw several NIOS processors on one FPGA Feb 14 13:34:46 good Feb 14 13:34:54 so you guys would understand what I am on about Feb 14 13:35:07 we understand Feb 14 13:35:23 but we value RTFM highly nevertheless Feb 14 13:36:16 I do like RTFM as well, I just couldnt find that one little critical piece Feb 14 13:36:37 be glad it's TIs docs and not freescales :) Feb 14 13:37:02 well, its why I am using a beaglebone and not arduino Feb 14 13:37:20 arduinos are AVRs which are from atmel Feb 14 13:37:32 still better docs than freescale Feb 14 13:38:10 thats true, but, arduino have designed their system and boards around their "wiring" language, they dont document any of their lower stuff Feb 14 13:38:39 use the source, luke Feb 14 13:38:42 for the lower stuff, you have to use atmel studio Feb 14 13:39:13 so I thought this way, why not just use a complete product, and go straight to the beaglebone Feb 14 13:41:42 mru: hmm.. according to oxford there is a slight difference between complex and complicated, while the latter has also the meaning of "intricated", the former does not... Feb 14 13:42:01 mru: is that just a dictionary difference or does it happen in real use as well? Feb 14 13:47:16 * mdp notes that TI also has exact steps on the wiki: http://processors.wiki.ti.com/index.php/Sitara_Linux_SDK_create_SD_card_script Feb 14 13:47:49 mdp: dont confuse the n00bs with facts! Feb 14 13:47:53 ok, well, it says, "Common Steps" but we know what they really mean Feb 14 13:49:37 I didnt see that document Feb 14 13:51:16 I am new to TI, and I only had the beagle bone for a few days Feb 14 13:51:33 it shall deliver a lifetime of happiness Feb 14 13:51:50 mdp: he is already a happy hippo Feb 14 13:51:51 :) Feb 14 13:52:08 hmm.. do you have happyhippos in the us? Feb 14 13:52:12 in just a few days! Feb 14 13:52:21 no, we only have hungry hungry hippos Feb 14 13:52:22 or is that an europe only thing? Feb 14 13:52:52 you only get happyhippos in the land of milk and honey there Feb 14 13:53:34 <_av500_> gm Feb 14 13:53:38 <_av500_> from Sofia Feb 14 13:53:41 <_av500_> no Loren Feb 14 13:54:41 stop by olimex and grab a handful of a10 prototype boards plz Feb 14 13:54:46 thx! Feb 14 13:55:03 heh Feb 14 13:55:07 _av500_: and bring me some of the imx23 boards, thanks Feb 14 13:55:11 mdp, they're not in sofia Feb 14 13:55:25 I thought they were there, ahhh Feb 14 13:56:21 Plovdiv? Feb 14 13:56:36 plovdiv? Feb 14 13:56:45 my family comes from there Feb 14 13:57:24 yes Feb 14 13:57:27 <_av500_> KotH: you are a bulgarian japanoswiss turk? Feb 14 13:57:31 mdp: previously known as filibe... before the bulgarians invaded the region Feb 14 13:57:48 panto, only two cities I know, figured it would be one of the bigees ;) Feb 14 13:57:56 KotH, you could be greek you know Feb 14 13:58:11 * panto is from the north too (just 100km away) Feb 14 13:58:20 panto: you know fairly well, that there were no greeks or bulgarians in that region prior to WWI Feb 14 13:58:21 considered this? Feb 14 13:58:40 KotH, ahem, no way Feb 14 13:59:05 panto: you should study your own history then Feb 14 13:59:11 here is a question, is there a nice text book to read on the topic of programming ARM processors? Feb 14 13:59:14 there were ethnic greeks/bulgrarians - there was no greek/bulgarian state Feb 14 13:59:21 HappyHippo: the ARM ARM Feb 14 13:59:31 <_av500_> and the LEG LEG Feb 14 13:59:44 _av500_: and he PROSTHETIC PROSTHETIC Feb 14 14:02:31 HappyHippo: it depends, I assume you mean to program them in assembly? Feb 14 14:02:57 HappyHippo: otherwise your generic programming book would do Feb 14 14:07:07 jackmitchell : more for something that will bridge the gap between the C programming language, and the getting use out of the hardware, for example, a programmer will know C, but he might not know how the timer counters, PWM, prescalers of the under lying hardware work, and how to interface with the underlying hardware Feb 14 14:07:59 HappyHippo: that's different than programming ARM processors Feb 14 14:08:01 HappyHippo: I think that's probably something to the AM335x Technical Reference Manual would be helpful Feb 14 14:09:30 I will give that document a good read, I am guessing its one of those brace yourself, read it 100 times over Feb 14 14:09:56 at ~2500 pages I would just pick out the bits that are specific to your cause Feb 14 14:10:08 mdp: not fun... OLS client is crashing :-s Feb 14 14:10:20 oh nos Feb 14 14:11:03 mdp: don't wanna debug java code right now Feb 14 14:11:16 mdp: also disappointed that they don't have a simple streaming mode Feb 14 14:11:19 felipebalbi: don't wanna debug java code ever Feb 14 14:12:01 jackmitchell: yeah, should've said that... but I really want my OLS working :-) Feb 14 14:12:18 enough of that was done at uni to cause I lifetime of seething hate ;) Feb 14 14:12:53 what's OLS? Logic Analyser? Feb 14 14:13:02 jackmitchell: yeah, openbench logic sniffer Feb 14 14:13:25 is OLS a client to an API, or is OLS the API? Feb 14 14:13:30 hmm... looks like it's some permission crap. I'm part of dialout which can read/write to /dev/ttyACM0, but only works as root Feb 14 14:13:39 jackmitchell: OLS is the hw Feb 14 14:13:51 felipebalbi: you want to switch to sigrok for a flexible streaming interface Feb 14 14:13:52 and the driver is the Java? Feb 14 14:14:05 mdp: what is the state of sigrok? Feb 14 14:14:11 http://dangerousprototypes.com/docs/Open_Bench_Logic_Sniffer Feb 14 14:14:11 but pulseview is not quite what the java client is Feb 14 14:14:19 jackmitchell: it implements USB CDC ACM Feb 14 14:14:21 mdp: last time i tried (a year or two ago) it looked very beta Feb 14 14:14:30 KotH: The State of Sigrok is "strong" ;) Feb 14 14:14:31 jackmitchell: and the Java client uses simple read/write on that port Feb 14 14:14:49 mdp: the democratic state of sigrok? ;) Feb 14 14:14:50 mdp: sigrok... let me try that one Feb 14 14:14:51 KotH: depends on what you are doing...sigrok-cli works well Feb 14 14:15:01 mdp: and the gui frontends? Feb 14 14:15:24 KotH: pulseview is immature relative to the jawi's java client for OLS Feb 14 14:16:22 * koen- would like iio support in sigrok Feb 14 14:16:33 you have the source! Feb 14 14:17:24 KotH: one fun thing to do is to gather data in VCD format and view in gtkwave ;) Feb 14 14:17:43 or output ascii art wave forms and not use a gui at all Feb 14 14:18:15 KotH: http://sigrok.org/wiki/Input_output_formats Feb 14 14:18:26 mdp: guess I need to upgrade my ols, sigrok says it doesn't support continuous mode Feb 14 14:18:29 oh well Feb 14 14:19:44 mdp: but gtkwave cannot do protocol analysis, can it? Feb 14 14:19:54 no Feb 14 14:19:58 :-/ Feb 14 14:20:03 all depends on what you are doing Feb 14 14:20:14 usually: debugging of flacky hardware :) Feb 14 14:20:25 for pru debug I mostly just watched that bit output Feb 14 14:20:51 was more convenient than the java client when I had no protocol and was just watching lines twiddle Feb 14 14:20:57 don't need a UI for that Feb 14 14:21:24 if you have just 1 or two signals that only toggle that's ok Feb 14 14:21:41 but then i just use scope and am done Feb 14 14:21:46 mdp: hex output seems to be easily scriptable too Feb 14 14:21:47 felipebalbi: yeah, I don't use continuous mode, however, I recall there being an fpga update with the in it Feb 14 14:22:05 our usualy suspect is an i2c or spi device that doesnt behave as it should Feb 14 14:22:12 or an SD IO card... Feb 14 14:22:16 felipebalbi: yes, this is one of the kernel test techniques I plan to bring up at the bof at ELC Feb 14 14:22:29 and there, having some form of protocol analysis is not only nice to have, but save a damn lot of time Feb 14 14:22:34 felipebalbi: very easy to test the full range of a pwm driver in this manner Feb 14 14:22:39 for example Feb 14 14:22:39 mdp: cool... Feb 14 14:22:44 mdp: right :-) Feb 14 14:22:48 mdp: easily automated too Feb 14 14:22:53 yep Feb 14 14:22:56 "automatable" if that's a word Feb 14 14:23:06 full automagic Feb 14 14:23:07 if you can say it, it's a word Feb 14 14:23:27 mru: did you watch too much marry popins? Feb 14 14:23:28 mru: I can pronounce flsndfr, but it doesn't mean it's a word :-p Feb 14 14:23:29 :) Feb 14 14:23:43 if you can say it and others understand it, it's a word Feb 14 14:23:56 the english language does not have a central defining authority Feb 14 14:25:33 mdp: btw, it would be much nicer if ols was using one of the FTDI USB HS UART FIFOs Feb 14 14:25:43 mdp: instead of a FS PIC Feb 14 14:26:10 that's quite a string of acronyms Feb 14 14:26:16 :-p Feb 14 14:26:38 still, for 50 bucks, it's a great bit of kit Feb 14 14:27:06 great does not describe it :) Feb 14 14:27:21 KotH: google it then Feb 14 14:27:28 felipebalbi: i have one on my desk :) Feb 14 14:27:43 KotH: k.. so you know how great it is Feb 14 14:27:48 juup$ Feb 14 14:28:00 specially when you consider the price and compare to similar tools Feb 14 14:28:05 juup Feb 14 14:28:19 though, i must say that the asix sigma2 is worth it's 200usd Feb 14 14:28:20 take the Totalphase USB I2C/SPI sniffer for instance Feb 14 14:28:26 400 USDs and it only has a fraction of the features Feb 14 14:28:27 s/it's/its/ Feb 14 14:28:42 felipebalbi: go for the sigma2 Feb 14 14:28:58 felipebalbi: the software that comes with it is quite good as well Feb 14 14:29:01 felipebalbi: yes, it has maximum bang for the buck Feb 14 14:29:21 KotH: windows only Feb 14 14:29:28 KotH: that's quite a no-no for me Feb 14 14:29:32 I fully expect more OSHW and more capable designs to appear in the future Feb 14 14:29:32 * KotH knows Feb 14 14:29:33 mdp: right Feb 14 14:29:41 this design is already a little old Feb 14 14:29:42 felipebalbi: that's why i bought the OLS for home use Feb 14 14:29:50 but to me, it's just a tool Feb 14 14:30:18 mdp: yeah... it's pretty cool. I heard some guys were developing a USB sniffer (OSHW) Feb 14 14:30:22 openvizla or something Feb 14 14:30:28 isn't this discussion getting a bit serious? Feb 14 14:30:33 http://openvizsla.org/ Feb 14 14:30:46 mru: the internet is serious busines Feb 14 14:30:50 it's already friday in parts of the world Feb 14 14:31:08 cool Feb 14 14:31:17 mdp: too bad they're so slow Feb 14 14:31:29 mdp: their website has been under construction for almost a year now Feb 14 14:31:54 mdp: they got 81k on kickstarter and... nothing Feb 14 14:32:24 if I got 81k for free, I'd scram too Feb 14 14:32:31 that would be a nice "little" project though.. Feb 14 14:32:42 verilog usb packet sniffer and packet generator Feb 14 14:33:27 felipebalbi: i am not sure they can really hold that low price Feb 14 14:33:42 felipebalbi: yeah, this is a common case with kickstarter ;) Feb 14 14:33:44 felipebalbi: an usb sniffer is a hole different category than OLS Feb 14 14:34:00 KotH: yeah, I don't mind it being more expensive Feb 14 14:34:10 felipebalbi: and knowing what the beagle480 costs.... (and that thing rocks!) Feb 14 14:34:17 KotH: if it's open source and works fine, I rather pay for an open source tool than closed Feb 14 14:34:30 felipebalbi: same here Feb 14 14:34:31 KotH: I have a beagle5000 sitting on my desk. Awesome tool Feb 14 14:34:40 felipebalbi: i'm envious! Feb 14 14:35:05 * mdp notes that OLS is already a usb sniffer...just not HS/SS ;) Feb 14 14:35:05 KotH: but if there was and open source tool for the same price with the same features, I'd go for that without thinking twice Feb 14 14:35:25 felipebalbi: uhm.. i'd go for oss only if it shows the same maturity :) Feb 14 14:35:27 mdp: with 24K sample memory ? there's not a lot you can do with that :-p Feb 14 14:35:37 felipebalbi: hehe Feb 14 14:35:54 mdp: beagle5000 has 4GiB DDR3 RAM :-) Feb 14 14:36:00 felipebalbi: the nice thing about totalphase is, they continued developing the software and you didnt had to pay extra for the updates Feb 14 14:36:07 felipebalbi: I think the real development of that protocol decoder for sigrok is done on the more capable platforms Feb 14 14:36:19 KotH: right... I think they're on the right track. compare to LeCroy's analyzers Feb 14 14:36:27 *nod* Feb 14 14:36:29 KotH: a voyager M3i starts out at around 25K Feb 14 14:36:36 felipebalbi: not to talk about their nice linux support Feb 14 14:36:46 felipebalbi: and support questions are answered quickly Feb 14 14:36:47 KotH: linux, mac and windows.. you can't beat that Feb 14 14:36:57 vms? Feb 14 14:37:09 * KotH usually uses the linux soft, because windows gets bogged down by the high usb traffic Feb 14 14:37:11 mdp: you mean they bounce raw data to PC and decode on the PC itself ? Feb 14 14:37:22 yes Feb 14 14:37:26 mdp: could be... Feb 14 14:37:37 mru: everyone but a few guys doing CTs want vms to die Feb 14 14:37:41 mdp: I should try to sniff my usb3 sniffer and see what kind of data it's giving my PC :-p Feb 14 14:37:53 hehe Feb 14 14:37:56 mdp: but I'm afraid if I do that I'll open the gates to hell Feb 14 14:38:27 I'm getting a cool piece of OS hardware soon Feb 14 14:38:43 I'm finally going for a ColorHug Feb 14 14:38:54 find me a good linux-based pcie bus analayzer while you are looking at all this stuff. Feb 14 14:39:04 been wanting one for a while but never had *enough* of a need for one, but now I do :) Feb 14 14:39:25 the VMETRO stuff I used in the past is top notch but all Windows...but you do what you need to do Feb 14 14:39:34 jackmitchell: that looks like a nice device :-) Feb 14 14:40:10 mdp: I don't think you'll ever find a linux-based PCIe analyzer :-( Feb 14 14:40:18 mdp: they're all windows based, unfortunately Feb 14 14:40:25 mdp: LeCroy has cool tools for that too Feb 14 14:40:26 jackmitchell: let me know how well it works Feb 14 14:40:31 yeah, test equipment hasn't progressed in this area very well Feb 14 14:40:36 jackmitchell: i thought about buying one of those as well Feb 14 14:40:54 will do, they've just had a new shipment of PCB's yesterday so they should have them tested and shipping in a week Feb 14 14:40:56 felipebalbi: some people claimed in 1998 there'd never be a something linux friendly in JTAG Feb 14 14:41:22 and then there was Abatron about the next year ;) Feb 14 14:41:25 mdp: hah! Feb 14 14:41:35 lauterbach supports linux Feb 14 14:41:38 can't beat that Feb 14 14:41:42 <_av500_> indeed Feb 14 14:41:45 and lauterbach as well Feb 14 14:41:46 mru: took them a while though, didn't it ? Feb 14 14:41:55 <_av500_> people have told me they prefer lauterbach to the TI TRM Feb 14 14:41:58 don't know, wasn't using that stuff then Feb 14 14:42:01 <_av500_> coz it has less errors Feb 14 14:42:13 _av500_: :-) Feb 14 14:42:17 _av500_: I have colleagues in dallas that will only use lauterbach ;) Feb 14 14:42:26 mdp: wasnt openocd written around that time as well? Feb 14 14:42:30 they cost a small fortune though don't they? Feb 14 14:42:37 <_av500_> jackmitchell: only money Feb 14 14:42:39 KotH: later Feb 14 14:42:44 yeah, I'm quite happy with my flyswatter2 :-p Feb 14 14:42:56 * _av500_ hugs printk Feb 14 14:42:57 but openocd isn't really easy to productize Feb 14 14:43:04 KotH: some people were trying to work with macgraigor's wigglers back then, but they sucked Feb 14 14:43:13 if you compare openocd with trace32... well, there's no comparisson Feb 14 14:43:24 KotH: everybody I knew doing linux work had a bdi or lauterbach Feb 14 14:43:47 I've been using my bdi2000 recently for ti814x u-boot work ;) Feb 14 14:44:12 just avoiding yet another usb connected tool is reason enough Feb 14 14:44:12 mdp: nice :-) with gdb ? Feb 14 14:44:25 mdp: don't say that... :-( Feb 14 14:44:27 heh Feb 14 14:44:34 felipebalbi: no, gdb is for computer scientists ;) Feb 14 14:44:37 mdp: USB pays for my kid's milk Feb 14 14:44:57 mdp: does BDI have their own UI too ? I haven't see it Feb 14 14:45:08 I use the cli clear/set breakpoints, memory dump, and and objdump of spl/u-boot in the next window Feb 14 14:45:18 mdp: or were you telneting into BDI ? Feb 14 14:45:23 felipebalbi: +1 on paying for chilluns ;) Feb 14 14:45:47 felipebalbi: they have several different firmware options..some that go with various proprietary RTOS debug environments Feb 14 14:45:57 and then bdiGDB Feb 14 14:46:05 bdi is the one to beat for linux dev Feb 14 14:46:08 so you slap whatever gdb + UI you want on top of that Feb 14 14:46:15 nothing came close for more than a decade Feb 14 14:46:22 but yes, I just telnet in Feb 14 14:46:26 why would anyone voluntarily use gdb? Feb 14 14:46:31 I used bdi3000 back at nokia for about a month or so Feb 14 14:46:40 but I was using telnet :-p Feb 14 14:46:41 hehe Feb 14 14:46:47 felipebalbi: for bringup, there's not much need to run gdb on top of it Feb 14 14:46:48 re Feb 14 14:46:59 some people that need a debugger like to use it to walk through a kernel driver Feb 14 14:47:16 but they quickly discover the pain of following code that's optimized Feb 14 14:47:21 oh it has some uses Feb 14 14:47:27 yes Feb 14 14:47:29 mdp: isn't ftrace nicer for that ? Feb 14 14:47:38 hit a debug breakpoint, fire up gdb Feb 14 14:47:44 traditional debuggers are usually only useful for the most trivial things Feb 14 14:47:45 see what went gaga Feb 14 14:48:00 *g* Feb 14 14:48:00 agree fully Feb 14 14:48:16 panto, I think the most useful part is if you have a set of kernel specific macros to dump key structures in a readable format Feb 14 14:48:27 most of the stuff the debugger will catch, is more easier caught by a fast reboot cycle + printk Feb 14 14:49:11 mdp, it's a tradeoff Feb 14 14:49:13 yep, so like felipebalbi says, ftrace is an effective tool for an entire other class of bugs Feb 14 14:49:37 the time spend creating the macros for the debugger, can be spend creating dump functions in the kernel Feb 14 14:49:48 <_av500_> omg, osgi Feb 14 14:50:05 yeah, I once used some scripting to break and have it upload some buffers in a graphics overlay driver I was writing..that's another useful feature of jtag Feb 14 14:50:23 scripts the whole thing to shove the buffer to the host, transform, verify qualitatively that it's doing the right thing Feb 14 14:50:35 av500 osgi rockz Feb 14 14:50:47 panto, hurray for jtag is all I can say ;) Feb 14 14:51:26 mdp, can't argue with that - I just feel that the actual number of use cases is very small Feb 14 14:51:48 now, if you could combine jtag + logic analyzer that would be a another matter Feb 14 14:52:07 indeed, that's why I don't use jtag much these days..it's rare to do any bringup for me Feb 14 14:52:35 driver work just doesn't need it Feb 14 14:52:56 for example in your overlay debug use case, what would be ideal would be to have the logic analyzer decoding the video output and upon detecting an abnormality present the sequence of driver commands to the video hardware that generated it Feb 14 14:52:59 once you have a working printf, you don't need anything else Feb 14 14:53:13 mdp: you're not doing bare metal software then? Feb 14 14:54:11 uhm did not know mru favours printk Feb 14 14:54:19 mru, +1 Feb 14 14:54:34 mru: +1 Feb 14 14:54:37 KotH: all I do is rewrite existing s/w for a living Feb 14 14:54:39 * panto awards mru +1 too Feb 14 14:54:40 ;) Feb 14 14:54:41 woglinde: of course I prefer writing correct code in the first place Feb 14 14:54:50 if printk() is taking too long, then use trace_printk() Feb 14 14:54:58 hm hm Feb 14 14:55:02 but reboot time Feb 14 14:55:09 I like debugger more Feb 14 14:55:48 if actual i/o is too slow, log stuff to a buffer and dump it later when you have time Feb 14 15:00:47 hi NishanthMenon Feb 14 15:00:58 woglinde, hi Feb 14 15:15:23 gm Feb 14 15:15:31 interesting google doodle today Feb 14 15:16:00 #exactsteps please Feb 14 15:16:13 54 Feb 14 15:16:14 plain old logo here Feb 14 15:17:51 NishanthMenon: hehe. Feb 14 15:18:33 pity that i'm going to have to cancel valentine's day tickets to a play or sell them Feb 14 15:19:10 dont let them dump you.. just show the beaglebone ;) Feb 14 15:20:05 NishanthMenon: no red colored valentine's day cape here :-P . time to call cco Feb 14 15:20:13 hehe Feb 14 15:20:51 I have a beaglebone black... feels appropriate Feb 14 15:21:28 aka the-worst-kept-sekrit-ever Feb 14 15:21:31 * NishanthMenon thinks if red is just overrated :) (no pub intended at taylor) Feb 14 15:23:20 * KotH hands mru some swiss chocolate hearts Feb 14 15:25:36 * jackmitchell eagerly awaiting his bbb Feb 14 15:29:06 <_av500_> watch BBB onlyine Feb 14 15:29:09 <_av500_> online Feb 14 15:29:42 can we have a version with a black bunny? Feb 14 15:31:09 jackmitchell: are you aware of the inside joke? http://www.bigbuckbunny.org/ Feb 14 15:31:19 fine, the kid is satisfy Feb 14 15:31:21 I haven't seen you around much Feb 14 15:31:32 with his starwars trading cards Feb 14 15:32:05 jkridner: not entirely no... Feb 14 15:32:21 it is a free video that most of us have seen ad naseum. Feb 14 15:32:26 jkridner: A prior conversation I have missed Feb 14 15:32:32 called Big Buck Bunny == BBB Feb 14 15:32:49 it clicks! Feb 14 15:32:57 BBB reminds of the test video torture :( Feb 14 15:34:04 we even managed to uncover a bug in the early cortex-a8 revs with it Feb 14 15:34:43 * mdp notes BBB as a testing tool Feb 14 15:34:52 I'm anxious to see the Black video wall. Take that no video accelerator! Feb 14 15:35:36 I've been contemplating how to do yuv to rgb conversion w/o the sgx Feb 14 15:35:51 the pru might be fast enough to do it Feb 14 15:36:30 mru: this is also of interest to me, although I don't need realtime Feb 14 15:38:49 mru: why take the sgx out of the loop? Feb 14 15:38:56 why not? Feb 14 15:38:58 to make it 100% open source? Feb 14 15:39:20 Sounds cool Feb 14 15:39:48 If you wanted to beat-up another TI C compiler, there is now one for the PRU. Feb 14 15:39:59 wtf Feb 14 15:40:06 last alpha wasn't very usable, but I could try to get you on the list for the next one. Feb 14 15:40:07 jkridner: announced publicly now? Feb 14 15:40:12 no. Feb 14 15:40:15 I guess so ;) Feb 14 15:40:17 you just did Feb 14 15:40:17 jkridner, heh Feb 14 15:40:19 why would anyone want a compiler for the pru? Feb 14 15:40:31 mru, this is what I kinda implied some time back ;) Feb 14 15:40:33 mru, there are uses Feb 14 15:40:35 didn't you know, this isn't a public forum? Feb 14 15:40:40 * jkridner runs to erase the logs Feb 14 15:40:43 jkridner: right! ;) Feb 14 15:41:00 jkridner: is that you poking around my hard drive? Feb 14 15:41:04 remember the printf talk a little bit earlier? Feb 14 15:41:05 * jkridner poitns out that BeagleBot is down. Feb 14 15:41:20 imagine having the capability to printf from your PRU program Feb 14 15:41:23 secure channel mode engaged! Feb 14 15:41:32 panto: printf does not require a compiler Feb 14 15:41:45 I have often called printf from assembly Feb 14 15:41:46 mru, I know Feb 14 15:41:47 panto, I want a unicorn Feb 14 15:41:58 that's not the point I was making Feb 14 15:42:02 mdp: I have a unicorn Feb 14 15:42:07 omg! Feb 14 15:42:18 there is the problem that PRU programs are worlds apart from the ARM user-space that drives them Feb 14 15:42:22 you have dragons over there too Feb 14 15:42:23 just go to N Korea Feb 14 15:42:26 or so the map says Feb 14 15:42:38 mdp: do a reverse dns on the machine hosting 26-26-54 Feb 14 15:42:40 you have to use some kind of PRU specific interface to communicate Feb 14 15:43:17 what if you could call a function to execute on the PRU, and the PRU code could call a function back to the ARM user-space program Feb 14 15:43:19 * jkridner points out that mru is good at thrashing compilers before BeagleBot comes bakc Feb 14 15:43:31 oh, hi BeagleBot. nothing said here. please move along. Feb 14 15:43:57 mru, lol, I see! real unicorn. I'm jealous Feb 14 15:44:05 you can do it with assembly, but it's too painful; there's no calling convention like with C Feb 14 15:44:17 why would you want that? Feb 14 15:44:34 and you can impose conventions without using C Feb 14 15:44:58 you know, once you did that, we could build some JNI and then some new android classes to drive all this Feb 14 15:45:21 but why would you want to do rpc to the pru? Feb 14 15:45:22 get some hot PRU app action going Feb 14 15:45:33 mru, why re-invent the wheel? Feb 14 15:45:50 the C function call ABI is all you need Feb 14 15:46:17 why tie it up in a synchronous interface? Feb 14 15:46:27 what you want is a command queue Feb 14 15:46:34 mru, cause it's easier to develop for Feb 14 15:46:56 so we're only doing easy things now? Feb 14 15:46:58 if you need a command queue you can build it on top of that Feb 14 15:47:08 wasteful Feb 14 15:47:09 mru, no Feb 14 15:47:19 smart Feb 14 15:47:38 what is the overhead of the command queue? Feb 14 15:47:48 most of the times is in the noise Feb 14 15:48:01 yes, it's very low if done right Feb 14 15:48:07 take one example that you know very well Feb 14 15:48:13 video encoding Feb 14 15:48:16 a blocking rpc call has much higher overhead Feb 14 15:48:47 now blocking calls can be easily implemented on top of async commmand queues Feb 14 15:49:20 mru, yes, but not as transparently Feb 14 15:49:51 but you _want_ the things to run in parallel, no? Feb 14 15:50:00 the whole point is not having to do any argument mangling Feb 14 15:50:04 otherwise you might as well do it all on the arm Feb 14 15:50:16 mru, it's trivial to run things in parallel - spawn more threads Feb 14 15:50:54 overhead Feb 14 15:51:02 in the noise :) Feb 14 15:51:35 maybe I'll get something done by next's ELC Feb 14 15:51:49 a concrete example should be clear Feb 14 15:52:27 too bad you can't demonstrate your existing implementation on zsp ;) Feb 14 15:52:40 mdp, it's dead jim :) Feb 14 15:54:20 mru, my point is that absolutely the best performance is not the end goal every time Feb 14 15:55:33 if you have power to spare Feb 14 15:56:03 panto, it went *poof* after this http://www.eetasia.com/ART_8800497381_1034362_NP_617a526a.HTM Feb 14 15:56:18 but then I'm a lazy c++ coder :P Feb 14 15:56:28 thurbad, with the way the PRU/DSP SDKs work, it is so painful that most people never run anything on them Feb 14 15:57:00 panto, it's more that PRU devel is discouraged and obscured Feb 14 15:57:03 I don't disagree with that Feb 14 15:57:23 coming up with a simple way to use them (even not being absolutely the most performant way), people will use them Feb 14 15:57:37 so they'll have tapped that power that they could use Feb 14 15:58:16 mdp, killed by magnum Feb 14 15:58:19 panto: mdp: I would use the PRU, however the level of integration it requires for a platform we may or may not use makes the risk to great Feb 14 15:58:39 jackmitchell: that *is* a risk regardless of the s/w Feb 14 15:58:40 so a C compiler which could bring the initial effort to a minimum would be welcomed Feb 14 15:58:58 the compiler is the least of the worries for me Feb 14 15:58:58 jackmitchell: you risk lock-in designing around a TI specific part, no doubt Feb 14 15:58:59 jackmitchell, how about if you could just port your ARM bit twiddling code to PRU Feb 14 15:59:00 indeed, but at the moment it's not even worth the effort to try Feb 14 15:59:14 with almost no changes (*) Feb 14 15:59:25 panto: exactly, then it would be worth it Feb 14 15:59:28 * Restrictions may apply, and/or the devil appears Feb 14 15:59:54 and that's why it exists..some people want it Feb 14 16:00:12 mdp: the TI lock in isn't a problem as I use mainly generic linux interfaces, however if I spent a few months on the PRU, that would get us locked in Feb 14 16:00:25 jackmitchell: right Feb 14 16:00:34 jackmitchell, right Feb 14 16:00:37 if it was just an 'easy' addition that would be weeks of tuning, I would use it Feb 14 16:00:44 but what if that months was just a couple of days? Feb 14 16:00:44 s/would/wouldn't Feb 14 16:00:56 panto: I'd be on it by now Feb 14 16:01:16 of course, the effort depends on the task ;) Feb 14 16:01:28 you guys are talking in complete undefined terms Feb 14 16:01:36 mdp, for most of the PRU stuff, it's just bit twiddling Feb 14 16:01:40 the initial time required, and the lock in to the processor is why I essentially pretend it's not there Feb 14 16:02:13 panto, sure, ask mranostay how long it took him to learn and solve a problem using PRU Feb 14 16:02:31 "as is" Feb 14 16:02:36 hey, it's the punk - suffering is good for his career :) Feb 14 16:02:43 it wasn't much time Feb 14 16:02:57 about a week I guess Feb 14 16:03:06 the real pain comes in complex management of resources..which C doesn't solve Feb 14 16:03:52 quite frankly, the pain of moving from any SoC to another is a type of lock-in ;) Feb 14 16:04:07 the arm core is nearly insignificant in the system Feb 14 16:04:36 mdp: +1 on arm being insignificant Feb 14 16:04:39 mdp: yes, that's why my current requirements is oe-core support and at least near mainline comapability Feb 14 16:04:59 jackmitchell: hah, then avoid arm :-p Feb 14 16:05:07 lol Feb 14 16:05:11 heh, it could be mips for all I care Feb 14 16:05:41 jackmitchell: that's be cool, indeed. But boy MIPS has some pretty weird stuff on their ISA :-p Feb 14 16:05:43 or ppc :) Feb 14 16:05:57 never played with power, but I'd love to :-) Feb 14 16:06:05 panto: have you had experience with power ? Feb 14 16:06:10 I know mdp has Feb 14 16:06:11 heh Feb 14 16:06:13 it's just another cpu Feb 14 16:06:23 err, how do you think mdp and I hooked up? :) Feb 14 16:06:30 mru: MIPS is too, how can I put it, academic Feb 14 16:06:32 panto, um, huh? ;) Feb 14 16:06:32 both linux ppc kernel devs Feb 14 16:06:39 mips is a fine cpu Feb 14 16:06:53 nothing academic about it Feb 14 16:06:56 as long as the kernel integration gives me generic spi and gpio implementations (that work ;))then I'm home and dry Feb 14 16:06:57 mips is fine Feb 14 16:07:02 ppc is fine too Feb 14 16:07:13 but don't have the numbers that ARM has Feb 14 16:07:23 mru: initialize a 32-bit value to a 32-bit register Feb 14 16:07:47 felipebalbi: two instructions Feb 14 16:07:49 I think a lot of the distaste for arm may arise from the historical pile of crap sw that has sat on it Feb 14 16:07:52 felipebalbi, it's just a purer risc Feb 14 16:08:04 mru: three, no ? mov, lsl, mov Feb 14 16:08:26 addi + lui Feb 14 16:08:36 * panto awards mru +1 Feb 14 16:08:40 mru: indeed... :-) Feb 14 16:08:46 i stand corrected Feb 14 16:08:46 or ori if you prefer Feb 14 16:08:51 * mranostay scrolls up Feb 14 16:09:01 mranostay: you're "the punk" apparently Feb 14 16:09:09 mdp: +1 for crappy sw on arm Feb 14 16:09:14 that is his name Feb 14 16:09:20 * ka6sox goes back to finding where he can save 4 clock ticks and make his State Machine work :P Feb 14 16:09:50 ka6sox which sate machine? Feb 14 16:10:23 woglinde, its a VHDL State Machine that is too slow... Feb 14 16:10:57 ka6sox: do you know why the ssh host key on emerald changed? Feb 14 16:10:57 (by 2 clock cycles) Feb 14 16:11:08 koen- look again Feb 14 16:11:59 the Proxy had the IP address, its been bought out from in front of it again. Feb 14 16:12:28 ah Feb 14 16:13:27 ka6sox than gogogo Feb 14 16:13:43 panto, often times the cost of setting up the thread, managing the thread, and then merging when its done are not worth it. Feb 14 16:14:34 ka6sox and what do you do than? Feb 14 16:14:58 optimize your code... Feb 14 16:15:03 ? Feb 14 16:15:13 ka6sox, agreed; but my premise is that most of the times it's worth it Feb 14 16:15:39 ka6sox: or call an optimisation consultant... Feb 14 16:15:41 getting more performance out of it can come at a later time (or not at all) Feb 14 16:15:49 woglinde, for the VHDL? I need to see where I can parallelize (or drop a term) Feb 14 16:16:23 no, I meant the thread stuff Feb 14 16:16:28 mru, +1 Feb 14 16:16:45 Optimize your inline code Feb 14 16:16:49 mru, right +1 Feb 14 16:16:55 bbl Feb 14 16:18:10 and do you know what said consultant will tell you? Feb 14 16:19:06 "my rate is ..." Feb 14 16:19:25 yes, that's one thing he'll say Feb 14 16:20:12 first thing Feb 14 16:21:20 sounds like you've had experience with consultants Feb 14 16:21:35 mru, lol, he was one of them :) Feb 14 16:22:03 not the same thing Feb 14 16:22:36 http://www.despair.com/consulting.html :) Feb 14 16:22:52 mru, he wasn't always a boy-scout Feb 14 16:26:23 anybody want a linux kernel engineer job in san diego? can fwd recruiter spam if desired Feb 14 16:26:41 afk Feb 14 16:27:10 mdp not bangalore? Feb 14 16:27:22 woglinde: surprisingly not ;) Feb 14 16:27:32 * mru wants an engineering job where the engineering is more important than using right cover sheet for the tps report Feb 14 16:27:40 unpossible Feb 14 16:27:57 mru, it's called work for yourself Feb 14 16:28:17 jayabharath: yes, I've come to that realisation Feb 14 16:28:50 definitely don't do upstream work...it's mostly tps reports Feb 14 16:29:11 mru make your company? Feb 14 16:29:15 +own Feb 14 16:29:34 makers make Feb 14 16:30:02 maker's mark? Feb 14 16:30:31 troll comany nr.1 limited inc Feb 14 16:30:38 hi mr plague Feb 14 16:30:47 gm prpplague Feb 14 16:30:53 hi prpplague Feb 14 16:31:21 greetings all Feb 14 16:32:09 hm should I connect to prpplague on linked in Feb 14 16:32:29 woglinde: I did, and it got me NO jobs. none. Feb 14 16:32:41 hehe Feb 14 16:32:46 biot haha Feb 14 16:32:57 all I got was this lousy t-shirt! Feb 14 16:33:35 hm I will try Feb 14 16:33:38 in 5 minutes Feb 14 16:39:07 hm I wanted to grep to openjdk code for vor voilatile use Feb 14 16:39:10 *sigh* Feb 14 16:39:36 local firmware group poll about topics for a meetings...top choice was "how to use c++ in device drivers" Feb 14 16:39:50 easy: not Feb 14 16:39:56 hi prpplague Feb 14 16:40:10 mdp point them to vmware Feb 14 16:40:18 mdp: where are you working? Feb 14 16:40:24 mdp: $consultingcompany? Feb 14 16:40:40 oh koth is back Feb 14 16:40:58 KotH: I'm just reading the northeast ohio firmware group's poll page Feb 14 16:41:02 they do some meetups Feb 14 16:43:21 prpplague: embedded linux driver position up here if you want to relocate ;) Feb 14 16:43:41 mdp: just accepted an offer Feb 14 16:43:50 w00t! Feb 14 16:43:52 prpplague: where? Feb 14 16:43:53 ! Feb 14 16:43:56 congrats! Feb 14 16:43:59 I hope Feb 14 16:44:11 prpplague yeah Feb 14 16:44:45 congrats prpplague Feb 14 16:45:27 *drum roll* Feb 14 16:45:51 rum troll Feb 14 16:46:04 [troll is now drunk] Feb 14 16:46:11 troll screech Feb 14 16:56:00 * prpplague sends emails to all the open offers Feb 14 16:56:56 *g* Feb 14 17:15:06 woglinde: i was never gone Feb 14 18:08:09 prpplague: just don't beg in the emails :P Feb 14 18:08:18 hehe Feb 14 18:18:52 prpplague: "Feel free to hire the lesser candidate, good luck" Feb 14 18:19:34 hehe Feb 14 18:24:35 hey i got an email :P Feb 14 18:25:16 lol Feb 14 18:25:38 "feel free to hire the punk"? Feb 14 18:27:40 hm intressing wargaming bought gas powered games Feb 14 18:30:10 eh? Feb 14 18:30:36 wtf is a gas powered game? Feb 14 18:31:28 perhaps one involving air guns Feb 14 18:34:37 prpplague: congrats on the new job! Feb 14 18:35:14 koen-, prpplague: is something official? Feb 14 18:35:34 or, is koen- speculating? Feb 14 18:35:36 yes but not announced Feb 14 18:35:58 today is a great day for non-announcements on #beagle Feb 14 18:36:06 btw, Black is a secret. Feb 14 18:36:57 could i have a cooler secret? Feb 14 18:37:06 boo. Feb 14 18:37:10 mranostay: what colour would you like yours? Feb 14 18:37:22 mranostay, no, but you can have a secret cooler if you like Feb 14 18:39:21 a black one? Feb 14 18:47:52 mru: hot pink Feb 14 18:48:38 mranostay gaming company Feb 14 18:49:06 mranostay: I'd keep that secret too Feb 14 18:50:14 mranostay the made some games for microsoft Feb 14 18:50:49 and lately trying to servive via kickstarter Feb 14 18:52:08 av500: the picture isn't showing in your g+ post Feb 14 18:53:11 hm... is it normal that a new pc costs 600eur and the graphic card you'd choose wouldnt even cost 20eur? Feb 14 18:53:42 or have i not paid attention to pc prices for toooooooo long? Feb 14 18:53:47 no idea, haven't bought a new pc in a long time Feb 14 18:54:13 mru: s/colour/color/ fixed that for you :P Feb 14 18:54:20 graphics card used to cost 100eur up... at least the usable ones Feb 14 18:54:37 only bought laptops in the last few years Feb 14 18:54:47 and my laptops tend to cost significantly more Feb 14 18:54:55 mru: having 32g ram sounds nice :) Feb 14 18:54:56 prpplague, when is the deadline for the slides today? Feb 14 18:55:06 panto: yesterday Feb 14 18:55:14 mru, you're mea Feb 14 18:55:15 *mean Feb 14 18:55:24 I sent mine this morning Feb 14 18:55:30 panto: being right is mean? Feb 14 18:55:32 for a very loose definition of morning Feb 14 18:55:52 KotH, yah Feb 14 18:57:15 panto: \pi is _NOT_ 4! Feb 14 18:57:42 so let's fix that Feb 14 18:57:57 panto: and if you had stayed within the ottoman empire, you would not be broke now ;-> Feb 14 18:58:20 being free > being broke Feb 14 18:58:42 they are not mutually exclusive Feb 14 18:59:06 KotH, you know who came up with all this \pi business right? Feb 14 18:59:20 panto: mranostay ? Feb 14 18:59:41 what? Feb 14 19:00:16 that Feb 14 19:00:17 mdp: https://plus.google.com/101339419642360856354/posts/CdqXzHkykC2 Feb 14 19:00:19 nothing, we started blaming you for everything now Feb 14 19:00:27 started? Feb 14 19:00:40 mranostay: panto blames you for the whole pi busines Feb 14 19:01:04 ok, we blamed him before, but behind his back Feb 14 19:01:08 now it's all out in the open Feb 14 19:01:29 prpplague, nice! Feb 14 19:01:46 prpplague: sweet Feb 14 19:01:54 leds == goodness Feb 14 19:15:11 oh prpplague has a picture of edelstoff Feb 14 19:16:07 hehe Feb 14 19:16:16 LetoThe2nd: gave me that one iirc Feb 14 19:17:19 good thing is I can buy here helles and edelstoff Feb 14 19:23:40 hmmm so what is a good barebones USB drive linux to use? Feb 14 19:25:26 ? Feb 14 19:25:45 woglinde: something that boots off a USB drive :) Feb 14 19:27:33 barebonebox Feb 14 19:28:44 toybox! Feb 14 19:29:20 mranostay? that depends on bios or firmware or bootloader Feb 14 19:29:42 I guess a usb-boot capable system is to be assumed Feb 14 19:30:04 heh yeah Feb 14 19:30:05 or does he man an usbstick? Feb 14 19:30:16 or a computer in formfactor of usbstick Feb 14 19:30:19 I dont get it Feb 14 19:31:45 usbstick Feb 14 19:32:04 hm I boot grub on some Feb 14 19:33:36 isn't syslinux usually the boot loader of choice there Feb 14 19:33:45 no? Feb 14 19:33:48 grub is more fun Feb 14 19:34:00 syslinux is limited Feb 14 19:34:04 but it works Feb 14 19:34:10 grub works too Feb 14 19:34:17 otherwise I had not done it Feb 14 19:34:19 whatever Feb 14 19:34:31 I've only ever booted off usb with syslinux Feb 14 19:34:50 and I only do it to install a new system or (occasionally) recover a broken one Feb 14 19:34:55 mranostay use kvm or virtualbox to test it Feb 14 19:35:39 mru with grub you can easily boot either a linux recover or windows recover system from the stick Feb 14 19:35:53 and you can partitioning it better Feb 14 19:35:55 ok ok Feb 14 19:36:16 with syslinux I dark remember you can only have one Feb 14 19:36:33 one's enough, if it's the right one Feb 14 19:37:08 I just assumed there was some reason the installation media all use syslinux Feb 14 20:19:20 g'night all Feb 14 20:21:56 *sniff* Feb 14 20:28:56 source code is called source code because it is teh source of all evil Feb 14 20:29:17 object code is called object code becuase it its the object of much fustration Feb 14 20:29:38 and the root user? Feb 14 20:31:17 gm mrpackethead Feb 14 20:31:49 <_av500_> use the root luke! Feb 14 20:52:01 yes, listen to the root Feb 14 20:53:33 jo djlewis Feb 14 20:53:36 pizza? Feb 14 20:59:28 woglinde: hi, no that was tuesday. Feb 14 20:59:57 hehe Feb 14 21:00:02 over here tuesday is burger day Feb 14 21:00:21 * djlewis misses the popeye cartoons Feb 14 21:06:56 hm here is thai day on tuesday Feb 14 21:07:33 odd how these things vary Feb 14 21:08:32 Is anyone experiencing system crashes on Ubuntu 12.10 with kernel 3.5.0.24/3.5.0.23 with the Beaglebone.. the crash is due to the usb hub.. Here is the log file.. http://pastebin.com/bNmwNXXF Feb 14 21:09:15 later... Feb 14 21:09:22 The problem does not reproduce on 12.04 or older systems with 3.2.x and older kernel it seems.. Feb 14 21:10:17 the stack trace mentions xhci Feb 14 21:10:20 does the system have usb3? Feb 14 21:13:32 yes.. It has 2 usb 3.0 ports and one usb 2.0.. Feb 14 21:13:50 try whichever type this wasn't Feb 14 21:16:14 I tried on all the ports.. it still crashes the system.. Feb 14 21:16:25 complain to ubuntu then Feb 14 21:20:50 Will try other kernel versions to see if the problem has been fixed.. Feb 14 22:14:44 sysfs is making me insane Feb 14 22:14:59 * Crofton surces libudev Feb 14 22:15:59 curses Feb 14 22:16:36 *g* Feb 14 22:16:38 ncurses? Feb 14 22:16:39 libsysfs Feb 14 22:20:12 I find /sys/devices/virtual/misc/axi_fpga with libudev Feb 14 22:20:25 but I want to find /sys/devices/amba.0/40000000.axi-fpga/ Feb 14 22:20:32 so I can read some dev attrs Feb 14 22:20:44 well, does it exist? Feb 14 22:22:02 yes Feb 14 22:22:42 http://www.signal11.us/oss/udev/ Feb 14 22:25:18 hmm, that's alan_o stuff Feb 14 22:25:51 yeah Feb 14 22:29:22 I wish I understaodd all these directorier witht eh same name, but diffeent contents Feb 14 22:31:37 sounds like a sysfs training class is in demand! Feb 14 22:31:47 *g* Feb 14 22:37:01 Crofton: sysfs.txt has a little background on the hierarchy and why some of the different contexts exist..if you haven't read that yet Feb 14 22:39:41 mdp, when is the class scheduled? Feb 14 22:40:54 course material is being produced..but it's in line after the DT and systemd classes Feb 14 22:41:26 it's part of our "Thing That Can Help Here(tm)" series Feb 14 22:41:59 okay where do I sign up? Feb 14 22:43:56 an attendee would require to find an instructor...uh oh Feb 14 22:44:04 s/to/us to/ Feb 14 22:47:23 thought over there, burger day is horsey day? Feb 14 22:51:49 ds2: depends on where you get your burgers Feb 14 22:53:25 hm horse meat is nice Feb 14 22:53:39 as bulette Feb 14 22:54:02 nice or not, labelling it as beef is wrong Feb 14 22:55:32 sure Feb 14 22:55:48 but you will not die Feb 14 22:55:55 when eating it Feb 14 22:55:56 nobody said that Feb 14 22:56:06 but it suggests poor quality control somewhere in the chain Feb 14 22:56:25 sure all should be cheap Feb 14 22:56:38 so quality control is cheap too Feb 14 22:56:55 i have a theory that it isn't horse meat at all.... they have been quietly disposing of large numbers of aliens that Torchwood has killed... Feb 14 22:57:43 scary thought Feb 14 22:57:44 prpplague I wonder its in the news in the us Feb 14 22:58:30 it hit reuters Feb 14 23:01:19 good nite Feb 14 23:06:26 prpplague: I'm actually starting to get interested in that HDMI debug board of yorus Feb 14 23:06:28 yours Feb 14 23:06:42 biot: oh? Feb 14 23:06:45 biot: do tell Feb 14 23:06:52 biot: working on the fpga code now Feb 14 23:07:03 biot: trying to get my vhdl uart working Feb 14 23:07:04 I don't need it or anything, but I've long thought that writing an HDMI protocol decoder would be fun, it's kind of interesting Feb 14 23:07:21 biot: ahh Feb 14 23:07:33 but sort of despaired of ever getting anything fast enough to capture it Feb 14 23:07:42 biot: well the board itself decodes the protocol Feb 14 23:07:52 that chip does? Feb 14 23:07:56 biot: so you just need something to decode the standard timings Feb 14 23:07:58 biot: indeed Feb 14 23:08:15 aha ok, so what it outputs to let you sniff is just the raw LVDS pins? Feb 14 23:08:17 biot: you get standard TTL timings on the output side Feb 14 23:08:29 biot: negative Feb 14 23:08:33 biot: not lvds Feb 14 23:08:39 biot: just standard TTL Feb 14 23:09:13 biot: you can hook any standard LA to it on those pins Feb 14 23:09:17 biot: and view the data Feb 14 23:09:29 what would that chip normally be used for? Feb 14 23:09:56 biot: in DVI/HDMI lcd panels Feb 14 23:10:31 biot: generally the same thing you would have in a desktop HDMI/DVI capable panel Feb 14 23:10:47 you could die from eatting horse meat...some of them have horse drugs Feb 14 23:11:01 ds2: which is one of the items they are concerned about now Feb 14 23:11:15 prpplague: so this is for testing whether your device under test is generating good HDMI out? Feb 14 23:11:18 prpplague: yep... surprisingly, no one has really commented on results of testing for that Feb 14 23:11:29 you'd think they can test for that quicker then they can sequence DNA Feb 14 23:11:32 biot: correct, and good timing information Feb 14 23:11:34 ds2: worst thing that could happen, you win the tour de france accidentally Feb 14 23:11:37 *ZING* Feb 14 23:11:53 mdp, thanks. Sounds like I should start there Feb 14 23:12:10 biot: i.e. if you set the Vertical Sync Width to 20, and you check on the LA or scope, and it isn't 20, then you know you have issues Feb 14 23:12:43 biot: you could connect up a standard 100MHz capable LA to my board and actually decode the video Feb 14 23:13:01 prpplague: but wouldn't that sort of thing be driven by a dedicated GPU of sorts? you tell it 20, it's guaranteed to work? Feb 14 23:13:13 I mean I assume you're not designing a GPU Feb 14 23:13:27 nothing is guaranteed Feb 14 23:13:35 esp. when pesky PLLs and stuff is involved Feb 14 23:13:55 biot: sort of, but it is depend on the software in many respects Feb 14 23:14:08 ah, so you'd still want to check the output of even a known working GPU, just on your board with your driver etc Feb 14 23:14:09 biot: for instance, robclark was debugging a new HDMI driver recently Feb 14 23:14:21 biot: so it would work on some displays but not other Feb 14 23:14:52 I'm soooo bored that I came here Feb 14 23:15:02 biot: so after debugging it with my board, they realized that the bottom four bits for the VSW were in one register, and the top four were in another Feb 14 23:15:05 sync sizes are such trivial things Feb 14 23:15:11 for added fun, debug color issues Feb 14 23:15:14 biot: they were writing all 8 bits to the same register Feb 14 23:15:25 ds2: indeed Feb 14 23:15:32 interesting Feb 14 23:15:43 prpplague: the AM335x has a very convoluted video out Feb 14 23:15:50 ds2: seems so Feb 14 23:16:01 biot: you should watch my lcd presentation Feb 14 23:16:06 prpplague: you saw my rants on how their "24 bpp" output is done? Feb 14 23:16:14 yea Feb 14 23:16:19 not to mention, it is often useful to know that the clock you think you're getting is the one you are actually getting, etc.. Feb 14 23:16:47 robclark: exactly Feb 14 23:16:51 biot: http://www.elinux.org/Elc-lcd Feb 14 23:17:02 it is also good to know that the video lines aren't loaded too much and your so call clock signal hasn't come a sine wave Feb 14 23:17:03 biot: the redwood version is the lastest and up to date Feb 14 23:17:30 the one where you generate that "biot is a moocher" text on the test display? Feb 14 23:17:33 I REMEMBER IT WELL Feb 14 23:17:53 hehe Feb 14 23:18:20 biot: i forgot about that Feb 14 23:18:52 aha, didn't watch this latest version, could do with a refresher Feb 14 23:18:55 * biot watches Feb 14 23:19:05 biot: most of the display driver developers are limited by access to the displays they can test Feb 14 23:19:44 biot: this will allow one device to pretend to be a whole different number of them and have quantitative data to prove that it is working properly Feb 14 23:20:41 isn't it more accurate to say - limited by access to displays with data that they can test with? Feb 14 23:20:49 :D Feb 14 23:21:10 aha, so you'd make your driver generate vsw, pixel clock, etc for a particular display, and then check the output with that board + an LA Feb 14 23:21:37 biot: exactly Feb 14 23:21:43 interesting Feb 14 23:22:16 so ideally, you'd want to automate that -- have the LA analyze the output and report exactly what timings etc it saw Feb 14 23:22:36 so you could have it run through a bunch of them, and just report it all after it's done Feb 14 23:22:38 biot: exactly Feb 14 23:23:00 biot: so marry my current board to a fpga design similar to the OLS Feb 14 23:23:11 biot: but specifically configured for this Feb 14 23:23:23 er, it would need tons of memory though? Feb 14 23:23:35 biot: only if you are going to decode the video Feb 14 23:23:43 sysfs.txt des not answer my questions Feb 14 23:23:47 biot: most concerned with just capturing the timings Feb 14 23:24:02 ah, you're talking about doing the timing analysis with the FPGA, not just shipping it out to a PC Feb 14 23:24:11 biot: yea Feb 14 23:24:17 makes sense Feb 14 23:24:56 so completely different project than I thought it was, but it's interesting again :) Feb 14 23:25:20 biot: hehe, well i haven't made a full attempt at documenting and explaining it as of yet Feb 14 23:25:47 and include a parser that takes the timing description structures so you can turn the screen green or red as go/no go test Feb 14 23:25:48 :D Feb 14 23:27:25 ds2: hehe, already have a flag set out in the uart data to identify timings that don't match what the EDID is set for Feb 14 23:27:31 prpplague: so is the plan to expand that board with an FPGA etc? Feb 14 23:27:39 biot: yea, two versions Feb 14 23:27:50 biot: one for folks who have the tools they need Feb 14 23:28:08 biot: the other for more software developers like robclark that just want some numbers from an easy to use tool Feb 14 23:28:39 what about the FPGA firmware to analyze and report the timing? will you do that? Feb 14 23:28:48 (I have no idea if you do FPGA stuff) Feb 14 23:30:06 isn't the timing analysis basically a few flipflops and a counter? Feb 14 23:30:10 biot: yea working on that now Feb 14 23:30:14 ds2: yea Feb 14 23:30:24 build it with 7400 chips! ;) Feb 14 23:30:46 biot: i have the pixel clock freq counter working, as well as counting the frame refresh rate Feb 14 23:30:58 oh I dunno, you might want to report more precise deviation Feb 14 23:31:10 or hmm, is that relevant for a driver developer Feb 14 23:31:32 biot: i'm going to release the vhdl under an open license, so people will be able to expand it Feb 14 23:31:59 biot: not so much for software dev Feb 14 23:32:10 right Feb 15 00:19:01 I'm getting read errors from an SD card that lock up terminal, is it bad? Feb 15 00:46:28 i would say that is bad Feb 15 00:47:30 bad unless you're rewriting the MMC code at the moment Feb 15 00:59:21 thurbad: and I would say that is bad Feb 15 00:59:37 not if it's fixable Feb 15 01:00:12 I had some when I had to rework our mmc driver Feb 15 01:00:45 I just meant hacking the mmc driver isn't something you'd normally want to be doing Feb 15 01:01:28 unless you spinyour own board that doesn't have a twl4030 based pmic :P Feb 15 01:56:54 busco novio... Feb 15 01:57:21 quien quiera diga 123 Feb 15 01:58:44 Why does my BeagleBone display the wrong time? Feb 15 02:00:24 im using python and when i used time.localtime() it gives me time.gmtime() Feb 15 02:14:17 argh Feb 15 02:14:26 can't seem to focus on this vhdl this evening Feb 15 02:14:32 * prpplague grumbles Feb 15 02:28:21 prpplague: Same here, except not VHDL Feb 15 02:28:22 :) Feb 15 02:29:20 I've had distraction from actual engineering all week, but I did manage to crack the code on how to get Staples to print business cards that don't look crappy :) Feb 15 02:30:03 alan_o: what weirdo has business cards in 2013 :P Feb 15 02:30:18 hehe Feb 15 02:30:23 yeah, OSS people are easy to find Feb 15 02:31:07 alan_o: i usually end up with 490 left of a 500 box before i quit :) Feb 15 02:31:08 mranostay: i am printing yours on saturday Feb 15 02:32:13 hmm.. I should have put #beagle on there Feb 15 02:32:35 of course then people would probably look for it on twitter Feb 15 02:32:51 alan_o: #cjoe and #cplusplus as well Feb 15 02:32:58 hehe Feb 15 02:33:21 Linkedin keeps wanting me to join a C++ professionals group Feb 15 02:33:54 If the level of discussion that's on the Embedded Linux group is any guide, I can't imagine how bad the discussion on C++ Professionals must be Feb 15 02:34:11 alan_o: you already have two strike against you :) Feb 15 02:34:26 don't admit to liking java or nodejs Feb 15 02:34:30 is prpplague printing creeper cards for you as well? Feb 15 02:35:38 alan_o: no, but i am putting a link on the back of the card to survey url. once you start the survey, the only question is "was this speaker a total ass?" YES/NO Feb 15 02:35:54 hehe Feb 15 02:36:00 no button doesn't work Feb 15 02:36:05 hah Feb 15 02:36:59 seriously - please sit down and have a talk with anyone who is trying to deploy creeper cards Feb 15 02:37:39 CareBear\: I was thinking cards that had an adhesive strip on the back that would attach to your badge. Kind of like the ones for "speaker" Feb 15 02:37:42 I think they are a really lame attempt at a solution for an important problem that deserves a better approach Feb 15 02:38:04 I don't think anyone here is serious about these cards. Feb 15 02:38:16 right Feb 15 02:38:25 but they are still being talked about Feb 15 02:38:30 is someone else serious about handing out cards? Feb 15 02:38:56 dunno? Feb 15 02:39:12 You're worried that someone "important" will actually start doing it for real? Feb 15 02:40:17 I'm of the opinion that the joke is harmless because it would be such a horrible idea in reality. Feb 15 02:40:29 but maybe I'm wrong, or an insensitive clod. Feb 15 02:40:54 (odds on the latter are high regardless) Feb 15 02:47:03 well speaking as a person who is border line aspergers, i think they are brilliant, they should have had them in high school for all types of social interaction, hehe Feb 15 02:49:36 * mranostay hands CareBear\ a creeper card Feb 15 02:50:00 prpplague: "Do you like me? check yes or no" :P Feb 15 02:50:12 what's a creeper card? Feb 15 02:50:14 mranostay: no there is quality communication Feb 15 02:51:12 jakllsch: http://singlevoice.net/redyellow-card-project/ Feb 15 02:52:13 oh crap Feb 15 02:52:15 someone was serious Feb 15 02:52:20 hehe Feb 15 02:52:22 any get the problem can not make/remove an entry for the specified session? Feb 15 02:52:34 i got it after I updated the kernel Feb 15 02:52:41 3.2.34 Feb 15 02:52:41 ok, I'm with CareBear\ on this. I see it as funny to joke about, but a disaster to implement in real life. Feb 15 02:52:44 alan_o: i say we all give creeper cards to beckyb at LPC Feb 15 02:53:23 +1 Feb 15 02:53:31 lolcreeper Feb 15 02:54:25 alan_o: just don't be a creeper :P **** ENDING LOGGING AT Fri Feb 15 02:59:58 2013