**** BEGIN LOGGING AT Sun Jun 09 02:59:59 2013 Jun 09 03:30:13 mastiff: nice Jun 09 03:30:43 can't get rid of soft block on rfkill Jun 09 03:31:05 opkg remove conman Jun 09 03:31:11 * m_billybob shrugs Jun 09 03:32:21 Might try it, but not sure if that will break things. Jun 09 03:32:30 no idea Jun 09 03:32:51 you know i use a different distro, i dotn need to mention it again . . . Jun 09 03:34:16 mranostay, ah you moonlighter you saw you name in few sources . . . Jun 09 03:34:44 moonliter, however it's spellled Jun 09 03:34:59 m_billybob: heh not really Jun 09 03:35:25 * mranostay only started Intel Jun 09 03:35:45 kind of your job there too isnt it ? or do they set you all on a specific path ? Jun 09 03:35:59 well not for arm specifically but . . . Jun 09 03:36:27 mranostay, i do hope you get to call the code you write in your spare time, yours . . . Jun 09 03:36:45 legally yes. Jun 09 03:37:02 esp when i was in CA Jun 09 03:38:13 sounds like a round about way of saying "no" ;) Jun 09 03:38:50 im just teasing anyhow. doesnt matter to me. Jun 09 03:39:00 heh it isn't Texas :P Jun 09 03:39:23 ;) Jun 09 03:39:27 horrible worker's rights iirc Jun 09 03:39:45 depends on the pay I guess Jun 09 03:39:57 it is still effing texas :) Jun 09 03:40:30 I lived in that state, the *state* was a decent place to live, so long as you did not LIVE in the middle of dallas Jun 09 03:40:40 but yeah i get the drift ;) Jun 09 03:41:37 I lived in mesquite, suburb of dallas during the 90's when mom and pop gaming studios did lots of business out of that area Jun 09 03:43:34 some of my grad school buddies work at intel. They always talk about leaving but never do. lol Jun 09 03:43:44 heh Jun 09 03:44:01 yeah well talk is cheap Jun 09 03:45:08 mastiff: nobody leaves intel... Jun 09 03:46:47 add "rfkill unblock bluetooth" to the list of things that are fucked up on BBB/3.8 kernel Jun 09 03:47:05 well I'll be damned Jun 09 03:47:13 I think I fixed spi Jun 09 03:47:16 * m_billybob is still attempting to get isci root working Jun 09 03:47:40 ssi, how is that ? what was wrong Jun 09 03:47:51 something rick mentioned earlier Jun 09 03:47:59 SCLK input lol Jun 09 03:47:59 sclk has to have the receiver enabled Jun 09 03:48:11 I guess I understand why Jun 09 03:48:13 but it's counterintuitive Jun 09 03:48:27 not really Jun 09 03:49:03 SCLK as input is for an alternative SCLK source Jun 09 03:49:13 no, Jun 09 03:49:17 SCLK is generated by the mcspi Jun 09 03:49:18 SCLK internally to the device is controlling MISO /MOSI Jun 09 03:49:31 but when it's reading from the slave, I guess it samples its own SCLK pin Jun 09 03:49:34 even though it's not driven Jun 09 03:50:06 anyway let's see if I can program the flash Jun 09 03:50:24 ssi: so do I need to buy a heat gun now? Jun 09 03:50:40 wal-mart is 24hrs, they stock 'em Jun 09 03:50:40 mastiff: dunno yet Jun 09 03:50:42 but maybe soon :) Jun 09 03:51:09 mastiff, ebay stock em 24/7 ~$50 last i checked ;) Jun 09 03:51:33 we have a cheap chinese knock off here, that works excellent was bought several years ago though Jun 09 03:51:35 yeah but walmart he can be desoldering tonight :) Jun 09 03:51:44 delivery time....... Jun 09 03:51:47 mastiff: do you have another chip? Jun 09 03:51:51 or are you going to try to reuse that one Jun 09 03:51:57 reuse that one Jun 09 03:52:00 eep Jun 09 03:52:05 it's gonna be tricky ;) Jun 09 03:52:08 I have another board if it's a disaster Jun 09 03:52:14 with another chip? Jun 09 03:52:14 :D Jun 09 03:52:45 Plenty of extra passives, would require another digikey order for the ICs Jun 09 03:53:24 well this is interesting Jun 09 03:53:27 The pins can be adjusted with a razer blade if you're careful. Jun 09 03:53:38 now I can't seem to config the fpga Jun 09 03:54:00 never mind, I'm being stupid Jun 09 03:54:12 How many maragaritas??? Jun 09 03:54:25 bcc_pgm: Sending WREN Jun 09 03:54:25 bcc_pgm: Sending Bulk Erase Jun 09 03:54:25 bcc_pgm: Bulk Erase complete (8988 checks) Jun 09 03:54:26 bcc_pgm: Sending PP Jun 09 03:54:28 heh i remember that software publisher btw Jun 09 03:54:30 bcc_pgm: Programmed 149516 bytes Jun 09 03:54:31 bam Jun 09 03:54:34 "ssi" Jun 09 03:55:09 haha ok next hurdle Jun 09 03:55:16 they published most of the games i played in the 90's on the PC Jun 09 03:55:17 fpga doesn't autoconfigure from the flash on powerup Jun 09 03:55:49 ssi: what are you doing with the FPGA anyways? Jun 09 03:55:54 nothing yet! Jun 09 03:56:14 .... Jun 09 03:56:17 heheh Jun 09 03:56:21 * m_billybob knows Jun 09 03:56:27 yea you know one of my reasons Jun 09 03:56:38 the other.... Jun 09 03:56:44 learning. Jun 09 03:56:54 either that or "free iPorn" Jun 09 03:56:57 there's a company called mesa electronics that makes fpga PCI and PCIe cards with daughterboards for machine control Jun 09 03:57:08 ah Jun 09 03:57:09 the fpga code is open source Jun 09 03:57:24 I want to modify it to run in this s3a, and be able to use their daughterboards with the BCC Jun 09 03:57:32 and use the BBB/BCC combo as a small machine control platform Jun 09 03:58:02 Maybe that would be a good platform for an autopilot on my sailboat. Jun 09 03:58:07 could be Jun 09 03:58:38 digital radio would be a good project for FPGA too Jun 09 03:58:42 so having bidirectional SPI is kind of important for me for both of these Jun 09 03:58:53 m_billybob: that's what eric designed this board for... that's his pet project Jun 09 03:58:53 soft radio . . . its been done though Jun 09 03:58:59 I'm planning on using my FPGA and BBB with hackRF Jun 09 03:59:10 ssi, he a fan of whats he face from OR ? Jun 09 03:59:16 whats her face* Jun 09 03:59:23 Geri or whatever Jun 09 03:59:30 I don't know what anyone's face is, and I don't know what eric is or isn't a fan of :D Jun 09 04:00:01 * m_billybob shrugs Jun 09 04:00:03 that youtube engineer lady? Jun 09 04:00:40 self taught i thin kbut yeah Jun 09 04:01:16 http://comments.gmane.org/gmane.linux.ports.arm.omap/90699 Jun 09 04:01:30 that long thread is where someone mentions the need to have sclk muxed as input Jun 09 04:02:47 m_billybob, thats what I am interested in... Jun 09 04:02:50 FreeDV Jun 09 04:03:32 This 2 1/2 hour long decon talk on building FPGA devices looks interesting. https://www.youtube.com/watch?v=ekU460CJCiQ Jun 09 04:03:37 defcon Jun 09 04:05:07 ka6sox, soft radio ? Jun 09 04:05:24 more like soft modem/codec Jun 09 04:05:40 opensource standard for radio voice modem Jun 09 04:06:15 * mranostay ears perk up Jun 09 04:06:15 I'd like to let the FPGA do the modem and the pru do the codec Jun 09 04:06:30 mranostay, what is the name of your cat Jun 09 04:07:45 ka6sox, i think some o that stuff i neat Jun 09 04:07:54 is neat Jun 09 04:08:06 yeah im multitasking again . . . Jun 09 04:08:19 the problem with solving my spi issue Jun 09 04:08:21 fighting an arguementative sd card atm Jun 09 04:08:27 is now I have to actually write some verilog to rx/tx spi Jun 09 04:08:37 heh Jun 09 04:08:44 ka6sox: tiger Jun 09 04:08:54 mranostay, kewl Jun 09 04:09:15 ssi, can you submit this as the spidev-spi0 standard? Jun 09 04:09:23 the dt? Jun 09 04:09:28 sure Jun 09 04:09:29 where? Jun 09 04:09:46 create a patches/spi Jun 09 04:09:51 and put the patch in there Jun 09 04:10:08 beagleboard/kernel 3.8 branch Jun 09 04:10:14 then issue a PR against that Jun 09 04:10:24 ok Jun 09 04:10:40 so "fork" to your own and then create the patch Jun 09 04:10:52 yea I know how to fork and do pull requests Jun 09 04:10:58 sweet Jun 09 04:10:58 what I'm not so sure about is creating a patch Jun 09 04:10:59 heheh Jun 09 04:11:09 wtf Jun 09 04:11:11 $sudo dd if=/dev/zero of=${DISK} bs=1024 count=1024 Jun 09 04:11:11 dd: opening `/dev/sdb': Read-only file system Jun 09 04:11:35 http://crna.cc/BVzwA0I9W9BbVNM Jun 09 04:11:38 oh, look at the format Jun 09 04:11:48 that the rest of the dt patches are. Jun 09 04:11:54 I'm five hours in trying to get this to work. Jun 09 04:12:01 d'oh bumped the lick switch on the adapter Jun 09 04:12:07 lock switch* Jun 09 04:12:21 Why am I scared of that link? Jun 09 04:12:36 mastiff, nope...its a good one Jun 09 04:12:38 ka6sox: I also need to generalize the dts a little bit Jun 09 04:12:46 right Jun 09 04:12:52 but gimme a minute, I'll get something pushed to my fork, and I'll have you look at it Jun 09 04:13:05 I'm working on the combined which I will call bcc-00A0.dts Jun 09 04:13:06 it's friendly, don't worry: http://crna.cc/cat/about-me Jun 09 04:13:06 :D Jun 09 04:14:10 isn't my website pretty? Jun 09 04:14:16 * mranostay reads the ACPI spec and notes he is out of beer Jun 09 04:14:47 * m_billybob still has 4 beers left from over a week ago Jun 09 04:15:03 I approve of the pixelated design. Jun 09 04:15:18 *mastiff fills wine glass........ Jun 09 04:15:43 no emeb Jun 09 04:16:00 Kristina: I like how the kernel told you to go rtfm Jun 09 04:16:04 they probably won't like it but I guess I'll create an FPGA one Jun 09 04:16:13 ssi: that's uboot Jun 09 04:16:14 ssi, I wasn't going to mention that but it is funny. Jun 09 04:16:23 Kristina: touche Jun 09 04:16:26 I can't get the kernel to run because some shit is getting corrupted Jun 09 04:16:43 ka6sox: you're going to create an fpga patches dir? Jun 09 04:16:47 well, master_cpu is getting corrupted Jun 09 04:17:04 ssi, yes, but yours is the generic spidev-spi0 Jun 09 04:17:19 ka6sox: I have the beginnings of an id eeprom bin for bcc Jun 09 04:17:22 it'll need a bit of tweaking Jun 09 04:17:28 right Jun 09 04:17:40 we should work on this either tonight (maybe too late) or tomorrow Jun 09 04:17:50 I'm here til I run out of steam :) Jun 09 04:17:55 which last night was about five hours from now Jun 09 04:18:04 that sentence parsed weird, but you know what I mean Jun 09 04:18:06 I'd ideally like to have both the spi0-spidev Jun 09 04:18:19 I'll make an spi1 one as well Jun 09 04:18:29 ssi, good plan Jun 09 04:18:33 push both of them up Jun 09 04:18:51 I'm not terribly confident about the chipselect part of this Jun 09 04:18:55 but I'm sure we'll get it figured out Jun 09 04:19:05 I'll "borrow" your fragments and also the gpmc ones and create the combined Jun 09 04:19:06 I know cs0 on spi0 is working correctly, and bbb doesn't pinout cs1 on spi0 Jun 09 04:19:26 yes, iirc thats a choice that was made by the hardware designer Jun 09 04:19:26 are you going to include them? or just copypasta Jun 09 04:19:46 I'm going to create a combined with the right fragment orders for the bcc Jun 09 04:19:53 ok Jun 09 04:20:28 so it will be a bit of copy pasta and then rework to get it right Jun 09 04:20:30 ka6soc: can you elaborate on the fragment orders part? I don't understand that yet. Jun 09 04:20:56 this went from being enjoyable to annoying Jun 09 04:21:08 IRC or the project Kristina ? Jun 09 04:21:10 mastiff: there is name complete with Tab Jun 09 04:21:25 Kristina: sorry to disappoint Jun 09 04:21:32 ka6sox: trying to get darwin to run on this thing Jun 09 04:21:35 I have a challenge. nothign to do with beaghles, but you guys are clever Jun 09 04:21:51 mrpackethead: setting us up for failure, eh) Jun 09 04:21:52 i have a microcontroller device Jun 09 04:21:57 or crazy Jun 09 04:21:58 why is flyswatter2 not letting me do semihost btw? Jun 09 04:22:12 that you can talk to via a web GUI Jun 09 04:22:14 bdi3000 does :/ Jun 09 04:22:22 i'm trying to cut down every feature to make it cheaper Jun 09 04:22:29 to make Jun 09 04:22:38 if its not "functional" its coming off Jun 09 04:22:47 clearification needed mrpackethead Jun 09 04:22:54 clarification too Jun 09 04:23:13 "microcontroller talk via web UI" does not compute. Jun 09 04:23:32 Kristina: how goes the MacOSX on ARM project? :P Jun 09 04:23:49 mranostay: I've had Darwin working on ARM for like a year now. Jun 09 04:24:04 mrpackethead: did you see how the new arduinos are incorporating a wifi router with many web APIs? Is that similar to what you are talking about? Jun 09 04:24:09 mastiff, since our "cape" dts will be replacing both the eMMC and HDMI I want to make sure it unloads them in the right order to not have issues Jun 09 04:24:39 mranostay: http://crna.cc/lyDVg2ew8cd4TYn Jun 09 04:24:53 ka6sox: I'll want to make a version of the dts that doesn't use gpmc as well... I have an application where I'll want to retain hdmi Jun 09 04:25:13 ssi, then just use the spi0 generic then Jun 09 04:25:17 yeah that's true Jun 09 04:25:24 ka6sox: yes I read something about how numbering changes with loading order but where is the detailed info on that located? Jun 09 04:25:29 this is why I'm differentiating Jun 09 04:25:41 now that I'm WORLD CLASS DEVICE TREE OVERLAY MASTER Jun 09 04:25:43 I can do whatever I want! Jun 09 04:25:57 ssi, tell him Jun 09 04:25:59 hey it's emeb! Jun 09 04:26:04 emeb_mac: I got spi fixed :D Jun 09 04:26:20 Almost time for me to buy a heat gun.... Jun 09 04:26:21 ssi: excellent! Jun 09 04:26:27 emeb_mac: I was able to blow the flash Jun 09 04:26:36 recently I've upgraded my fork to xnu 2050 Jun 09 04:26:37 emeb_mac: but it doesn't autoconfigure on powerup once I've got something in flash Jun 09 04:26:43 because my original one was 14xx Jun 09 04:26:44 ssi, when you have your generic spi0-spidev up there I'll add that bit to my dts for the bcc Jun 09 04:27:12 ssi: odd - wonder what went wrong? Jun 09 04:27:21 interesting issue Jun 09 04:27:27 and one that makes sense Jun 09 04:27:29 debug darwin kernel is 8.8mb Jun 09 04:27:35 can't upload over serial Jun 09 04:27:45 so I have to take the sd card out every time Jun 09 04:27:49 Kristina, how many kexts does that have? Jun 09 04:27:50 and tftp is slow Jun 09 04:27:50 emeb_mac: I have no idea... I haven't yet investigated how the config at boot stuff works Jun 09 04:28:03 I guess the POR state of the port expander is such that the fpga is configured from flash Jun 09 04:28:10 and then the tool changes the expander to use SPI instead? Jun 09 04:28:14 ka6sox: it has armPlatformExpert and a PBA8 kext. Jun 09 04:28:27 the one for omap3? only armPlatformExpert so far. Jun 09 04:28:29 so pretty simple then Jun 09 04:28:33 mrpackethead: Here's a link to the info I saw with the new arduino stuff. https://www.youtube.com/watch?v=jqDG2yiVplM Jun 09 04:28:51 ka6sox: what else would you want? Jun 09 04:29:02 not much else Jun 09 04:29:10 ok a bunch of nake bikers went past my apartment... oh portland Jun 09 04:29:11 okay back to DTS stuff Jun 09 04:29:16 right Jun 09 04:29:20 *naked Jun 09 04:29:25 nakey bikers? Jun 09 04:29:33 I wonder how hard would porting the sgx stuff to darwin be Jun 09 04:29:46 Kristina, I would not even want to try Jun 09 04:30:03 ssi: it is portland Jun 09 04:30:14 was there a bird on it? Jun 09 04:30:26 I think it's services4 stuff that has to be ported Jun 09 04:30:38 the rest of the ddk is kind of platform independent Jun 09 04:31:26 I'm starting to wonder if hacking around binary blobs would be easier than trying to compile this myself. Jun 09 04:32:34 ssi, I have been struggling with creating an fpga directory and not having objections...so instead I'll create one for sdp Jun 09 04:32:45 this ddk isn't even for omap :/ Jun 09 04:32:51 same sgx core though Jun 09 04:32:55 should be compatible Jun 09 04:32:56 I think Jun 09 04:32:58 that's... software defined peripheral? Jun 09 04:33:03 yes Jun 09 04:33:03 seems reasonable Jun 09 04:33:25 srry got called away Jun 09 04:33:37 ok, so, heres the problem. Jun 09 04:33:40 software define all things! Jun 09 04:33:50 i want to rip off all the dip switches that set "configuration" modes Jun 09 04:33:51 i'm going to bed Jun 09 04:33:56 on my PIC micro based solution Jun 09 04:34:00 nite Kristina Jun 09 04:34:03 dip wwitch = money Jun 09 04:34:11 indeed Jun 09 04:34:21 all the configurations can be done via a web page Jun 09 04:34:26 however.. Jun 09 04:34:29 one serial eeprom is way cheaper than a bunch of switches :P Jun 09 04:34:31 if you loose track of the device Jun 09 04:34:36 emeb_mac, what were you saying about needing i2c address 0x39 for the expander on the BCC? Jun 09 04:34:46 and you don't know what its IP address is Jun 09 04:35:06 how can you get it back into a "default" state Jun 09 04:35:21 ka6sox: I think it's 0x38 currently? Jun 09 04:35:32 the expander? Jun 09 04:35:36 or the eeprom Jun 09 04:35:36 yes Jun 09 04:35:38 the expander Jun 09 04:35:40 okay Jun 09 04:35:42 eeprom is 54-57 Jun 09 04:35:45 depending on switches Jun 09 04:35:57 is that i2c2? Jun 09 04:36:05 yes it's on the i2c2 bus Jun 09 04:36:13 pins 19/20 Jun 09 04:36:17 P9.19/20 Jun 09 04:36:34 glad you guys are sorting this otu Jun 09 04:36:35 today Jun 09 04:36:40 which is already set up in the default tree... are you going to overlay it to ensure it's correct? Jun 09 04:36:47 no Jun 09 04:36:55 because its part of the base.dts Jun 09 04:37:06 actually beaglebone.dst Jun 09 04:37:09 er dts Jun 09 04:37:11 I guess nothing can commandeer i2c2 since it's part of capemgr? Jun 09 04:37:21 I was just thinking what happens if someoen further up the chain tries to change it Jun 09 04:37:30 but maybe that's not possible Jun 09 04:37:41 have to ask panto on that one..but I think we can use 0x38 Jun 09 04:37:50 oh the address should be fine Jun 09 04:37:55 those drivers are generally reeentrant Jun 09 04:38:01 I think the only reserved addresses are 54-57 Jun 09 04:38:04 address conflicts are my only issue Jun 09 04:38:05 right Jun 09 04:38:13 what I meant was if the i2c2 bus becomes not an i2c2 bus Jun 09 04:38:23 like if someone tries to pinmux uart1 onto those pins Jun 09 04:38:27 iirc eric was able to use it Jun 09 04:38:38 it works fine against the base tree Jun 09 04:38:42 then capebus is blown up Jun 09 04:38:45 yes Jun 09 04:38:46 and all bets are off Jun 09 04:38:50 ok gotcha Jun 09 04:39:01 should the cape eeprom mark the i2c2 pins as "I use these"? Jun 09 04:39:12 no Jun 09 04:39:23 beaglebone.dts has a fragment for this Jun 09 04:39:40 I don't fully grok how the eeprom pin map factors in Jun 09 04:39:48 I know how the effective fragment works Jun 09 04:40:03 the pinmux grab is done early Jun 09 04:40:08 er what's it called Jun 09 04:40:18 exclusive-use Jun 09 04:40:33 I'll talk to panto about this Jun 09 04:40:39 and I realize if one of our overlays tried to mark i2c2 as exclusive-use, it'd conflict with beaglebone.dts Jun 09 04:40:42 and probably fail to load Jun 09 04:40:43 because he wrote cape manager. Jun 09 04:41:03 (it is taking everything I have not to type capemangler Jun 09 04:41:04 ) Jun 09 04:42:08 mastiff, was it you who were having issues unloading (removing) HDMI from the bbb or was that rickta Jun 09 04:42:43 I haven't tried that yet, but I want to. I saw that others were having issues doing so. Jun 09 04:42:58 I marked it in uEnv.txt to be removed Jun 09 04:43:03 I dunno if it worked correctly or not Jun 09 04:43:20 I assume so: Jun 09 04:43:21 4: ff:P-O-L Bone-LT-eMMC-2G,00A0,Texas Instrument,BB-BONE-EMMC-2G Jun 09 04:43:21 5: ff:P-O-- Bone-Black-HDMI,00A0,Texas Instrument,BB-BONELT-HDMI Jun 09 04:43:25 lack of L means not loaded, yeah? Jun 09 04:43:30 was me ka6sox panto told me just to load teh dts that takes the pin prior to hdmi loading Jun 09 04:43:51 well that'd be too easy :P Jun 09 04:43:56 I'm having issues with rfkill soft block on bluetooth. "rfkill unblock bluetooth" doesn't remove the soft block on BBB Jun 09 04:44:10 m_billybob, so you need to load one of the overlays that steals that pin Jun 09 04:44:40 ka6sox, well assuming you needed thsoe pins already sure. but perhaps ask panto when he's aroudn to make sure Jun 09 04:45:00 maybe i misunderstood, but i do not think so Jun 09 04:45:09 they key word here is "overlay" Jun 09 04:45:21 that should make it easier to understand Jun 09 04:45:44 ka6sox: I'm not sure where the spi0-spidev.dts file should end up in the kernel source after my patch creates it Jun 09 04:45:47 more like more confusing ( to me ) but ok . . . Jun 09 04:45:51 ka6sox: firmware/capes? Jun 09 04:45:57 ssi, yes Jun 09 04:46:24 also locate the dtsi as well... Jun 09 04:46:26 .dtsi Jun 09 04:46:30 hm there's already a BB-SPI0-00A0.dts in there Jun 09 04:46:39 orly? Jun 09 04:46:46 and we have been fighting this? Jun 09 04:46:47 yeah Jun 09 04:46:50 it doesn't configure spidev Jun 09 04:47:00 it DOES however mark sclk as input :( Jun 09 04:47:08 aha Jun 09 04:47:25 ssi, we need to create a BB-SPI0-00A1.dts then Jun 09 04:47:37 or something to differentiate which is which Jun 09 04:47:50 well should ours have spidev in the name? Jun 09 04:47:57 BB-SPI0-SPIDEV-00A0.dtsi? Jun 09 04:48:04 probably better to call it BB-SPI0_SPIDEEV Jun 09 04:48:15 I think _ would be better Jun 09 04:48:17 ok Jun 09 04:48:20 with or without version? Jun 09 04:48:25 with version Jun 09 04:48:31 thats the format Jun 09 04:49:18 so a cape could call for BB-SPI0-00A0.dts or BB-SPI0_SPIDEV-00A0.dts Jun 09 04:49:38 ok Jun 09 04:50:44 ssi, so for your case you would put: BB-SPI0_SPIDEV-00A0.dts Jun 09 04:50:56 and I would put BB-BCC-00A0.dts Jun 09 04:51:12 and yours would only load the spidev spi0 Jun 09 04:51:16 right Jun 09 04:51:25 and mine would load both GPMC and SPI0_SPIDEV Jun 09 04:52:34 back... Jun 09 04:52:46 emeb_mac, read backlog :) Jun 09 04:52:50 * m_billybob expect exact steps Jun 09 04:52:52 :P Jun 09 04:53:09 expects too but meh i digress. Jun 09 04:53:13 ka6sox: we need 0x38 on I2C for the expander Jun 09 04:53:35 okay...I'll ask panto monday Jun 09 04:54:33 ssi: yeah - reset state of the I2C port expander should come up with the output pulling such that the FPGA is in SPI flash config mode Jun 09 04:54:36 m_billybob, we are some of us are bumbling our way down the DT road...others are waiting till it becomes smoother. Jun 09 04:54:44 so it self-configures from flash at power up Jun 09 04:54:51 emeb_mac: I'll have to figure out how to troubleshoot that Jun 09 04:55:04 emeb_mac: oh wait... you can't load a bitfile directly on flash can you Jun 09 04:55:06 emeb_mac, I'm concerned about thia Jun 09 04:55:07 have to load a bin or something, right? Jun 09 04:55:29 this/that Jun 09 04:55:44 ka6sox: DT doesn't really give you a way to mark i2c addresses as reserved, does it Jun 09 04:55:47 it's all just "by convention" Jun 09 04:56:08 ssi, nope, 50-54 is cnvention Jun 09 04:56:24 ssi: when you program the flash it's the same bitfile that you load directly to the FPGA Jun 09 04:56:26 54-57 Jun 09 04:56:36 emeb_mac: oh... well in that case then I'm back to stumped :) Jun 09 04:56:56 emeb_mac, I'm concerned on the BBB that we could crash with the eMMC if we configure immediately after reset is lifted on teh bus Jun 09 04:56:57 ssi: I don't have a verify function written in the bcc_lib tho Jun 09 04:57:03 bah never mind it worked fine that time Jun 09 04:57:05 that would be handy Jun 09 04:57:11 maybe I'll write a verify Jun 09 04:57:46 ka6sox: you mean the FPGA would load the eMMC bus? Jun 09 04:57:51 yes Jun 09 04:57:54 *could* Jun 09 04:58:06 does emmc use i2c also? Jun 09 04:58:14 ohh nm Jun 09 04:58:15 that's correct - if the FPGA pins are setup wrong. Jun 09 04:58:30 if the fpga is configured immediately, and it has a program that loads the emmc bus Jun 09 04:58:36 right Jun 09 04:58:42 emeb_mac, for next revision, we need to reverse this Jun 09 04:58:48 so the answer is "don't do that" Jun 09 04:59:02 since the Bone doesn't have a bus_reset Jun 09 04:59:04 ka6sox: how do you propose to reverse that? Jun 09 04:59:09 just SYS_RESETn Jun 09 04:59:11 the answer is set it up so that you have to ask the expander explicitly to configure from flash Jun 09 04:59:22 let me think about this. Jun 09 04:59:26 but then you might need a driver for the board Jun 09 04:59:38 I'm "claritin clear" but have a headache now. Jun 09 05:00:12 the only thing would then be a way to hold the FPGA in reset till we whack the expander Jun 09 05:00:34 but thats later..much later Jun 09 05:00:35 ka6sox: might as well just leave the spi flash off then Jun 09 05:00:48 and have the BBB load it up explicitly Jun 09 05:00:59 we need the flash to set DT correctly Jun 09 05:01:13 when booting Jun 09 05:01:16 if you can't trust the spi flash to contain a compatible FPGA bitstream then better to not have it Jun 09 05:01:37 ka6sox: different flash Jun 09 05:01:42 oh right Jun 09 05:01:46 ka6sox: there's a serial eeprom for capebus, and then the config flash Jun 09 05:01:58 emeb_mac, I'm holding out for Grey Jun 09 05:02:02 then it won't matter Jun 09 05:02:02 config flash isn't really necessary Jun 09 05:02:05 but it's nice to have Jun 09 05:02:10 hee hee Jun 09 05:02:27 my A4 is Grey now Jun 09 05:02:30 ssi: where is the .dts that worked? Jun 09 05:02:38 Grey??? Jun 09 05:02:38 emeb_mac: it's about to be in my kernel tree fork Jun 09 05:02:52 as soon as someone helps me figure out how to create a patch file :( Jun 09 05:03:04 I don't know how to diff a file against no file Jun 09 05:03:05 haha Jun 09 05:03:05 ssi: I take it that you can't use a stock kernel then Jun 09 05:03:11 emeb_mac: sure you can Jun 09 05:03:13 it's very simple Jun 09 05:03:15 lemme pastebin it Jun 09 05:03:42 so you didn't need to patch the spidev driver source? Jun 09 05:03:47 nope Jun 09 05:03:51 nice Jun 09 05:04:00 https://gist.github.com/ianmcmahon/5737725 Jun 09 05:04:15 i'm struggling to come up with a good no button reset idea Jun 09 05:04:24 oh well, never mind Jun 09 05:04:43 mrpackethead: unplug then plug Jun 09 05:04:54 how does that help? Jun 09 05:05:14 mrpackethead: are there PMICs with soft-reset? Jun 09 05:05:16 ssi, please add your copyright line and add that its an SPIDEV DTS Jun 09 05:05:23 yeah sry Jun 09 05:05:49 ssi, can reset it, but that woudl require a button Jun 09 05:05:50 SPIDEV SPI0 virtual cape Jun 09 05:06:03 or a header or somethign Jun 09 05:06:42 i have an ethernetport Jun 09 05:07:02 Can't you use one button for different functions based on how long the button is held until release? Jun 09 05:07:11 https://gist.github.com/ianmcmahon/5737731 Jun 09 05:07:13 ssi: thanks for that. I'll try to see if I can make it work here Jun 09 05:07:30 emeb_mac: should work fine Jun 09 05:08:15 ssi: hope so - I need to learn how to use it though Jun 09 05:08:40 I assume that I need to run it through dtc to generate a .dtbo Jun 09 05:08:58 compile it: dtc -O dtb -@ -b 0 -o BB-SPI0_SPIDEV-00A0.dtbo BB-SPI0_SPIDEV-00A0.dts Jun 09 05:09:00 then use capemgr or something to apply thit Jun 09 05:09:07 -O is output format Jun 09 05:09:16 -@ is relocatable symbols (makes it overlay compatible) Jun 09 05:09:20 -b 0 is cpu 0 Jun 09 05:09:23 -o is output file Jun 09 05:09:36 put the resulting .dtbo in /lib/firmware Jun 09 05:09:52 kk Jun 09 05:09:54 then echo "BB-SPI0_SPIDEV" > /sys/devices/bone_capemgr.*/slots Jun 09 05:10:00 then ls /dev/spidev* Jun 09 05:10:20 now I've noticed that the device nodes don't necessarily correspond to the SPIs Jun 09 05:10:27 if you load a SPI1 overlay, then a SPI0 overlay, Jun 09 05:10:38 spidev1 will be SPI1, and spidev2 will be SPI0 Jun 09 05:11:50 mranostay: no ideas Jun 09 05:11:58 mranostay: im dissapoited Jun 09 05:12:41 no true complete power off Jun 09 05:13:07 ssi: thx - I'll try that. Jun 09 05:13:59 ka6sox: so all these patch files have a number in front of them Jun 09 05:14:02 who arbits those numbers? :P Jun 09 05:14:26 ssi: you mean 001-*? Jun 09 05:14:29 yep Jun 09 05:14:39 I'm trying to create a new one Jun 09 05:14:45 dunno what to name it :) Jun 09 05:14:55 thats the patch order Jun 09 05:15:02 oh they're just numbered within the directory? Jun 09 05:15:04 so put yours at the end Jun 09 05:15:08 yes Jun 09 05:15:12 ok then it's 001 :D Jun 09 05:16:25 ssi: can I just copy/paste the text to create a BB-SPI0_SPIDEV-00A0.dtbo file, or do I have to do something with gist? Don't really understand gist. Jun 09 05:16:54 mastiff: gist is just pastebin Jun 09 05:17:00 but attached to my name and slightly more permanent Jun 09 05:17:04 so yes just copy/paste Jun 09 05:17:08 but in a minute you can pull it directly from my fork Jun 09 05:17:13 av500: +1 on the double ROT-13 Jun 09 05:17:20 well, a patchfile Jun 09 05:17:23 which is maybe less useful :P Jun 09 05:17:37 mranostay: my favorite strong encryption Jun 09 05:17:57 ssi: you posted two different links to the .dts - whats the diff? Jun 09 05:18:10 I tried compiling the 2nd one and got errors Jun 09 05:18:32 emeb_mac: I changed the comment at the top is all Jun 09 05:18:34 what's the error? Jun 09 05:18:45 I guess I should have attempted to build these :) Jun 09 05:18:50 kk Jun 09 05:18:55 here's what I get: Error: BB-SPI0_SPIDEV-00A0.dts:59.33-34 syntax error Jun 09 05:18:55 FATAL ERROR: Unable to parse input tree Jun 09 05:19:27 nuke the comment at the end of the line and try again Jun 09 05:20:40 same result. Jun 09 05:20:46 k hang on one sec Jun 09 05:20:47 complains about char 33,34 Jun 09 05:21:25 oh Jun 09 05:21:28 semicolon on the line above Jun 09 05:21:32 compatible = "linux,spidev"; Jun 09 05:21:42 that's line 58 in my file Jun 09 05:21:43 yep Jun 09 05:21:47 fixed it Jun 09 05:24:07 ssi: is there a pastebin to troll^H^H^H^H review? :) Jun 09 05:24:22 https://gist.github.com/ianmcmahon/5737731 Jun 09 05:24:28 note the missing semicolon on line 58 Jun 09 05:24:40 spidev0 didn't appear when I loaded it Jun 09 05:24:46 should be spidev1 Jun 09 05:24:50 should be spidev1.0 probably Jun 09 05:24:58 here's pastebin of syslog: http://pastebin.com/hv5myw1T Jun 09 05:25:25 nothing at all in /dev/spi* Jun 09 05:26:05 what kernel? Jun 09 05:26:26 Linux beaglebone 3.8.13 #1 SMP Mon May 27 18:04:20 CEST 2013 armv7l GNU/Linux Jun 09 05:26:31 the "no spi-max-frequency property" thing is suspicious Jun 09 05:26:46 ya Jun 09 05:27:21 here lemme give you the "old" dts Jun 09 05:28:03 https://gist.github.com/ianmcmahon/5737754 Jun 09 05:28:18 they should be close to identical, but check that one Jun 09 05:30:09 is the source supposed to have .dts or .dtbo? Jun 09 05:30:22 tried - got a "file exists" when trying to load it Jun 09 05:30:33 do I need to unload the prior one? Jun 09 05:30:35 you'll probably have to reboot Jun 09 05:30:37 you can try unloading Jun 09 05:30:42 how? Jun 09 05:30:43 echo "-6" > .../slots Jun 09 05:30:47 assuming it loaded in slot 6 Jun 09 05:30:52 which I think your paste showed it did Jun 09 05:30:55 mastiff: .dts Jun 09 05:30:59 mastiff: compile it to a .dtbo Jun 09 05:31:10 emeb_mac: capemgr always crashes when I unload Jun 09 05:31:12 ok, had it reversed Jun 09 05:31:14 you will likely need to reboot Jun 09 05:31:46 ssi, did you see the post that mdp gave us for what is needed for the DTS? Jun 09 05:31:56 ssi: looks like the BBB crashed Jun 09 05:32:06 ka6sox: no Jun 09 05:32:12 emeb_mac: on unloading/ Jun 09 05:32:16 emeb_mac: that's what usually happens to me Jun 09 05:32:29 http://www.mjmwired.net/kernel/Documentation/devicetree/bindings/spi/spi-bus.txt Jun 09 05:34:03 ka6sox: I did actually see that Jun 09 05:34:03 hmmm - my ssh hung that time Jun 09 05:34:07 didn't realize it was mdp's Jun 09 05:34:23 it was the one he suggested we lookat Jun 09 05:34:28 oh I see Jun 09 05:34:31 yeah I saw that pretty early Jun 09 05:34:50 I still don't understand what "cells" are Jun 09 05:35:45 the gpio chipselect info is excellent Jun 09 05:35:56 I suspect we are talking about the number of CS on the part Jun 09 05:36:07 so in this case SPI0 = <1> Jun 09 05:36:14 and for SPI1 = <2> Jun 09 05:36:18 meh - shutdown -r doesn't reboot Jun 09 05:36:27 need to run into the other room... Jun 09 05:36:31 ok Jun 09 05:37:16 oh I'm wrong Jun 09 05:37:24 I'm doing it wrong Jun 09 05:37:33 Error: BB-SPI0_SPIDEV-00A0.dts:59.1-2 syntax error FATAL ERROR: Unable to parse input tree Jun 09 05:37:36 ssi, I'm wrong Jun 09 05:37:56 it how many pins it takes to define a CS Jun 09 05:38:08 that is 1 in both cases (as they are not multiplexed) Jun 09 05:38:43 ok Jun 09 05:39:00 each spi1_x device gets its own CS, so they each independently have 1 address cell Jun 09 05:39:13 mastiff: check for that missing semicolon Jun 09 05:39:51 yup, line 58 not 68 facepalm Jun 09 05:40:05 ssi: OK that earlier file w/o comments gave me a spidev1.0 port Jun 09 05:41:35 emeb_mac: ok interesting Jun 09 05:41:42 I'll mess with the new one and see what's wrong with it Jun 09 05:41:59 maybe it doesn't like the comments that come after the max_freq line? Jun 09 05:42:10 shouldn't be a problem Jun 09 05:42:18 thinking maybe the discrepancy in compatible Jun 09 05:42:19 famous last words. :) Jun 09 05:43:04 heheh Jun 09 05:44:20 ssi, can you post a new gistfile? Jun 09 05:44:31 gimme a few, I need to shakeout the problems with it Jun 09 05:44:40 I was goign to help :) Jun 09 05:45:13 yea one ser Jun 09 05:45:15 sec Jun 09 05:45:20 I had to reboot the bbb Jun 09 05:45:21 do I echo this exactly or use the actual filename with .dtbo extension? then echo "BB-SPI0_SPIDEV" > /sys/devices/bone_capemgr.*/slots Jun 09 05:45:33 echo that exactly Jun 09 05:45:34 it's a part number Jun 09 05:45:36 that's in the dts Jun 09 05:45:45 ok, no need for version number Jun 09 05:45:48 right Jun 09 05:46:57 I suppose that if I were feeling adventurous then I could test this on my BCC Jun 09 05:47:31 go for it Jun 09 05:47:38 works on mine Jun 09 05:47:51 first have to pry a BCC off of one of my BBWs Jun 09 05:48:19 crowbar?? Jun 09 05:48:32 just work around the edges Jun 09 05:48:34 it's not that bad :) Jun 09 05:48:54 I've had to have mine on and off a buch of times Jun 09 05:49:04 cause unlike the bbw, if I need to get at the serial port the cape's gotta come off Jun 09 05:49:52 ka6sox: at some point we need to get with panto and fix this null ptr dereference on cape unload Jun 09 05:50:08 yes, we should beat up on him monday :0 Jun 09 05:50:13 he he Jun 09 05:50:28 ka6sox: here's the current dts that doesn't work right: https://gist.github.com/ianmcmahon/5737791 Jun 09 05:51:43 oh I see it Jun 09 05:51:49 underscores instead of dashes Jun 09 05:51:58 derp Jun 09 05:52:12 yep Jun 09 05:53:36 Ok i see BB-SPI0_SPIDEV in slot 7. What am I supposed to see with "then ls /dev/spidev*" Jun 09 05:53:50 do you have device nodes? Jun 09 05:54:19 did it return something like /dev/spidev1.0? Jun 09 05:54:34 no such file or directory Jun 09 05:54:39 dmesg Jun 09 05:54:45 you may be using the broken fragment Jun 09 05:55:23 [89770.642945] bone-capemgr bone_capemgr.9: part_number 'BB-SPI0_SPIDEV', version 'N/A' [89770.666698] bone-capemgr bone_capemgr.9: slot #7: generic override [89770.673259] bone-capemgr bone_capemgr.9: bone: Using override eeprom data at slot 7 [89770.681271] bone-capemgr bone_capemgr.9: slot #7: 'Override Board Name,00A0,Override Manuf,BB-SPI0_SPIDEV' [89770.708001] bone-capemgr bone_capemgr.9: slot #7: Requesting part number/versio Jun 09 05:55:24 check the syslog - did it complain about max freq? Jun 09 05:55:40 lol, sorry for the text bomb Jun 09 05:55:57 mastiff: you didn't get to the part I was worried about Jun 09 05:55:58 the error Jun 09 05:56:04 anyway here's the new improved corrected dts Jun 09 05:56:04 https://gist.github.com/ianmcmahon/5737796 Jun 09 05:56:54 ok, I was a few steps behind you guys, so I need to unload this one first before echoing in the newly compiled replacement? Jun 09 05:57:21 maybe Jun 09 05:57:31 go ahead and try it first Jun 09 05:57:50 ok Jun 09 06:00:07 I have to rerun the patches in the kernel tree :/ Jun 09 06:00:57 echo: write error: File exists Jun 09 06:01:03 ok Jun 09 06:01:03 reboot Jun 09 06:01:06 then try it Jun 09 06:02:06 why is it the later it gets, the more time you have to spend waiting on some mindless task to execute? :/ Jun 09 06:02:07 damn you now I'm gonna get jacked by conman again lol Jun 09 06:03:23 ssi: new .dts worked for me Jun 09 06:04:57 looks good: root@beaglebone:~# ls /dev/spidev* /dev/spidev1.0 Jun 09 06:05:31 excellent for both of you Jun 09 06:06:00 i'm happy to say that I understand 99% o fwhat's in that dts now :) Jun 09 06:07:14 wish I did Jun 09 06:07:20 it's not too bad Jun 09 06:07:29 I understand maybe 50%, but the fog is clearing. Jun 09 06:08:03 ssi: what was the main thing you had to change to get the spi rx working? Jun 09 06:08:19 emeb_mac: pinctrl, 0x150 0x10 to 0x150 0x30 Jun 09 06:08:25 (drum roll please) Jun 09 06:08:32 just setting the input_enable flag on the pinmux Jun 09 06:08:32 ah - so it was just GPIO setup Jun 09 06:08:43 no, letting the clock be used by the input too Jun 09 06:08:43 gpio isn't the right term Jun 09 06:08:47 it's padconfig Jun 09 06:08:51 right Jun 09 06:09:03 IIRC that was an issue way back on original Beagle too. Jun 09 06:09:15 yeah it makes sense now that I know what's going on Jun 09 06:09:18 it just wasn't obvious Jun 09 06:09:21 that stumped Crofton on some of his original experiments Jun 09 06:11:01 So where do I go from here to get data from my spi thermocouple amplifier module? Jun 09 06:11:21 well you can work with it two ways Jun 09 06:11:30 one is just reading/writing the device node Jun 09 06:11:37 that way is half-duplex Jun 09 06:11:58 the other is to write C code that uses ioctl to run spi transactions Jun 09 06:11:58 that's similar to reading and writing GPIO? Jun 09 06:12:01 you'll get full duplex that way Jun 09 06:12:21 yea it's similar, but you don't have to bitbang the line Jun 09 06:12:25 the driver does it for you Jun 09 06:13:51 you can study the bcc_lib code to see how to use spidev in C Jun 09 06:14:11 I've got full-duplex spi transfers there Jun 09 06:14:23 ssi, so spidev_spi0 and spidev_spi1? Jun 09 06:14:25 good suggestion, haven't had time to look at that code yet. Jun 09 06:17:10 emeb_mac: So for programming the FPGA you use the xilinx software, and for communicating with the FPGA we use the bcc_lib? Jun 09 06:17:15 ka6sox: https://github.com/ianmcmahon/kernel/commit/87a0dfa023e7ee27682e05576b63c182cc3311b4 Jun 09 06:17:34 mastiff: you can program it from bcc_tol Jun 09 06:17:39 bcc_tool Jun 09 06:17:42 mastiff^: Xilinx software is used to compile the design Jun 09 06:17:56 ssi, now issue the PR and it should get there tomorrow night our time Jun 09 06:18:00 bcc_lib and bcc_tool are used to load the design into the hardware Jun 09 06:18:06 and to program the spi flash Jun 09 06:18:09 ka6sox: who gets it? Jun 09 06:18:19 ka6sox: and are they going to be welcoming of it? :P Jun 09 06:18:41 we can get it done Jun 09 06:18:44 haha ok Jun 09 06:18:49 i'm timid :P Jun 09 06:18:51 if you are up to it... Jun 09 06:19:03 switch to spidev_spi1 Jun 09 06:19:09 they're both in there Jun 09 06:19:49 was supposed to both be in there... maybe it's not? Jun 09 06:20:31 thats only spi0 Jun 09 06:20:39 I swear to god Jun 09 06:20:41 add the other file and repush Jun 09 06:20:43 on my local filesystem it's both Jun 09 06:20:52 forgot to git add Jun 09 06:20:56 the second one Jun 09 06:21:11 no, it's one file Jun 09 06:21:17 one patch file that creates two dts's Jun 09 06:21:30 and it's committed and pushed Jun 09 06:21:31 so wtf Jun 09 06:21:37 not this one kemosabe Jun 09 06:21:54 I know what's showing up on github Jun 09 06:22:00 I know it's wrong Jun 09 06:22:04 what I don't know is why it's wrong Jun 09 06:22:07 push again? Jun 09 06:22:09 :) Jun 09 06:22:12 who knows Jun 09 06:22:14 So I have a good reverse polarity story. Jun 09 06:22:29 Working on fixing a Porsche 911. Jun 09 06:23:26 https://github.com/ianmcmahon/kernel/blob/3.8/patches/spi/001-spi0_spidev-virtual-cape-dts.patch Jun 09 06:23:27 Owner disconnected the battery for the winter and connected to a trickle charger, but hooked it up backwards. Jun 09 06:23:30 that looks better Jun 09 06:24:04 Now the car batter has +12v from the negative terminal. lol Jun 09 06:24:14 had no idea that could even occur, but it can Jun 09 06:24:35 force opposite polarity on the battery? Jun 09 06:24:38 yup Jun 09 06:24:41 lol Jun 09 06:24:43 weird Jun 09 06:24:55 I didn't think it was possible, but it is. Jun 09 06:25:48 When he reconnected the battery the car wouldn't start, but the blower comes on without key in ignition. Jun 09 06:26:14 Just discovered the RP issue today. Not sure how much damage it caused. Could be expensive for him. Jun 09 06:26:26 ugh Jun 09 06:26:31 Or it may fire right up with a new batter. will see tommorow. Jun 09 06:27:05 mastiff^, emeb_mac ssi: can we test the combined tomorrow. Jun 09 06:27:10 and fix that if necessary? Jun 09 06:27:14 whenever you're ready Jun 09 06:27:20 you have a gpmc fragment working? Jun 09 06:27:37 I needed the SPI working first Jun 09 06:27:49 now I can programme the FPGA Jun 09 06:27:57 ah gotcha Jun 09 06:27:58 I'll be around, and will get a heat gun in the morning to hopefully rotate my FPGA. Jun 09 06:28:00 (well I could have used my pod. Jun 09 06:28:02 and you'll need an FPGA design that talks to GPMC Jun 09 06:28:11 to test it Jun 09 06:28:12 yeah Jun 09 06:28:20 and I know nothing about gpmc Jun 09 06:28:22 but i'm interested to learn Jun 09 06:28:30 and it has to work w/o eMMC active of course Jun 09 06:28:36 right Jun 09 06:28:44 so you need to develop on a uSD-booting system Jun 09 06:28:48 so we need angstrom images that boot off the boot Jun 09 06:28:58 there are uSD images someone said now Jun 09 06:29:08 my main development platform is debian off usd Jun 09 06:29:38 so this is going to take a bit...ucf, vhdl + the DT... Jun 09 06:29:50 emeb_mac: How does this chip look for a software radio cape? http://www.ti.com/product/afe7225&lpos=Middle_Container&lid=Alternative_Devices Jun 09 06:30:02 I've got a .ucf for the BCC ready Jun 09 06:30:12 I'll put it up somewhere Jun 09 06:30:56 we just need a single 16bit register we can write and then read Jun 09 06:31:00 mastiff^: I've looked at those before Jun 09 06:31:29 TI sent me some, but I haven't made test boards yet. Need to learn more first. Jun 09 06:31:34 [kernel] ianmcmahon opened pull request #44: SPIDEV virtual capes for SPI0 and SPI1 (3.8...3.8) http://git.io/_eRlyQ Jun 09 06:31:37 pull request sent Jun 09 06:31:40 they cost about 4x what the FPGA does Jun 09 06:32:07 oh neat Jun 09 06:32:08 haha Jun 09 06:32:44 and they need a high-speed serdes I/O that the S3A won't support Jun 09 06:32:51 for now we should just simulate a SRAM Jun 09 06:32:54 so I can't use 'em Jun 09 06:32:58 AH, good to know. Jun 09 06:33:18 oh wait - looks like they do parallel too Jun 09 06:33:50 Unfortunately the insulator on the antenna for my sailboat costs 20x the FPGA cost, so I compare prices relative to the insulator. Jun 09 06:33:57 you could prolly make a decent radio w/ that Jun 09 06:35:15 Near term I'm looking at using a couple hackRFs for experimental radio stuff, and I have a new Standard Horizons HF for the usual marine radio com. Jun 09 06:35:38 I wouldn't mind screwing with SDR stuff at some point Jun 09 06:35:58 would be fun to try to make open source aviation nav/com radios ;) Jun 09 06:36:18 questionably legal, but fun Jun 09 06:36:35 ssi: I'm working aviation nav/com right now. Jun 09 06:36:42 yea? Jun 09 06:36:46 the waveforms are very interesting Jun 09 06:36:47 I'd like to learn more about that Jun 09 06:36:56 doing VOR, ILS, ADF Jun 09 06:37:00 awesome Jun 09 06:37:08 Get a ham license, tons of priveleges Jun 09 06:37:17 n4ml, amateur extra :P Jun 09 06:37:21 me too Jun 09 06:37:31 not extra I mean general Jun 09 06:37:34 all the nav waveforms are RX only, so no license needed. Jun 09 06:37:34 sorry Jun 09 06:37:35 lol Jun 09 06:37:50 emeb_mac: yea but using them under ifr is questionably legal :) Jun 09 06:37:55 non-tso'd and all Jun 09 06:38:17 I'm not clear on how that shakes out under experimental rules, but I think under IFR you still have to have TSO'd radios Jun 09 06:38:20 as long as you have tso for main ones Jun 09 06:38:24 yeah Jun 09 06:38:32 Marine and aviation are very different environments. Jun 09 06:39:10 I'd like to do a GMSK modem for the beacons for DGPS Jun 09 06:39:16 would be awesome to make some remote radios that work with my HXr remote radio interface Jun 09 06:39:36 I've got a good relationship with GRT, so they'd probably support me on the interface side Jun 09 06:42:06 Hey, I have connected a LED with resistance to GPIO_39 and DGND Jun 09 06:42:30 and the LED always blinks even I actually done echo >> 0 /sys/class/gpio39/value Jun 09 06:43:47 the >> is too much! Jun 09 06:44:03 echo 0 > blah Jun 09 06:44:24 Use a FET Jun 09 06:44:34 paint it blue Jun 09 06:45:00 ka6sox: isn't that black? Jun 09 06:45:26 Only on the bottom. Paint the top green if you're inland. Jun 09 06:46:23 echo 0 > /sys/class/gpio/gpio39/value is the same Jun 09 06:46:31 And mix in iron oxide, silicon carbide, and graphite so the paint abosrbs radar. Jun 09 06:46:38 are you sure you have the right gpio number? Jun 09 06:46:54 I hope so Jun 09 06:47:16 according to the header image yes Jun 09 06:47:30 P8 #4 => GPIO_39 Jun 09 06:47:49 if it is wrong wouldn't it not blink at all? Jun 09 06:48:10 not necessarily :) Jun 09 06:48:41 Try P9-16 I have that hooked up on my BBB to turn on a FET that an led is connected to. Jun 09 06:48:43 but yeah 39 sounds about right? Jun 09 06:48:52 It's GPIO 58 Jun 09 06:50:02 sorry GPIO 51 Jun 09 06:50:29 ok wait Jun 09 06:50:45 he's saying that P9-16 is gpio 51 Jun 09 06:50:48 which also sounds correct Jun 09 06:50:58 okay guys...I'm beat...testing tomorrow. Jun 09 06:51:02 Many of the pins on P8 are used by other devices and that will cause you trouble. Jun 09 06:51:03 night :D Jun 09 06:51:06 later Jun 09 06:51:18 nn Jun 09 06:51:41 ok GPIO51 does not work.. Jun 09 06:51:52 is there a "safe" GPIO Pin Jun 09 06:51:53 did you hook your led up to P9-16? Jun 09 06:52:04 did you enable the gpio? Jun 09 06:52:06 No Jun 09 06:52:10 I'll try this out on BBB + BCC tomorrow Jun 09 06:52:10 P9-16 not Jun 09 06:52:15 Don't hook it up directly or it may fry things. Jun 09 06:52:16 still on P9-4 Jun 09 06:52:39 p8-4 you mean? Jun 09 06:52:48 ah yes Jun 09 06:52:49 emeb_mac: Great, I'll be around and will get the heat gun for rework in the morning. Jun 09 06:52:57 ok I change to to p9-16 Jun 09 06:52:59 mastiff^: I wish you the best of luck there Jun 09 06:54:05 Yeah I need a large batch of luck. Jun 09 06:54:40 or a small batch of new chips :) Jun 09 06:54:55 ahh this works Jun 09 06:54:58 I can now disable it Jun 09 06:55:02 there ya go Jun 09 06:55:12 thank you !! Jun 09 06:55:23 True, I have nothing to lose though. Got an extra BCC board because I assumed a fuckup Jun 09 06:55:31 to have it constantly on I need to use a conductor Jun 09 06:55:37 ? Jun 09 06:55:39 yea I ordered two boards and two sets of components Jun 09 06:55:41 plus extra passives Jun 09 06:56:31 I'm ordering more hackrf boards, you can get in on that if you can hand placing a few hundred 0402 components. Jun 09 06:56:50 heh Jun 09 06:56:52 sounds miserable Jun 09 06:57:23 how much are the boards? Jun 09 06:57:33 It's not that bad if you don't do what I did and order them in a format where digikey puts each individual 0402 in it's own bag that you have to open lol Jun 09 06:58:55 They are 3"x6" 4-layer boards. I'm getting them either from OSH Park 3 for $140, or from another company at 20 for $700. Jun 09 06:59:04 BOM will run you around $115 Jun 09 06:59:52 well if you get 20 for $700, I'll buy two ;) Jun 09 06:59:56 Good project though, 30mhz-6Ghz radio, and at least 500 people are working on the open source project. Jun 09 07:00:50 I could also get 100 for $800+ shipping and taxes from Germany.. Jun 09 07:01:00 that sounds like the value Jun 09 07:01:47 Not many people want to buy PCBs until they find out if they made the waiting list for a fully assembled board Jun 09 07:01:56 well I'm not on a waiting list :P Jun 09 07:02:13 But it's half duplex, so myself and many other will end up wantin two or more. Jun 09 07:02:27 does av500 like strawberries? Jun 09 07:03:43 mastiff^: well let me know what you decide... I'll buy at least two, and if you buy 100 of them and get the price down, I may buy a few. I also may know another person that might want some Jun 09 07:04:15 I'm going to bed Jun 09 07:04:17 ttyl! :D Jun 09 07:04:25 later. Good work today Jun 09 07:04:30 thx Jun 09 07:23:53 hi Jun 09 07:24:02 hi~ Jun 09 07:24:09 hello? Jun 09 07:50:30 <_Sy_> Morning.... Jun 09 07:53:52 <_Sy_> anyone here? Jun 09 08:19:38 does anyone here know how to return the BBB to stock software on the 2gig flash? Jun 09 08:21:59 monday: use the flasher, latest is 6/6 Jun 09 08:22:21 what is that? I'm brand new to BB Jun 09 08:23:02 dm8tbr: I thought this was the latest image Angstrom-Cloud9-IDE-GNOME-eglibc-ipk-v2012.12-beaglebone.rootfs.tar.gz Jun 09 08:23:10 not sure though. Jun 09 08:23:17 <_Sy_> modany: What O/S are you running on your host system Jun 09 08:23:21 <_Sy_> *monday Jun 09 08:23:28 that's the raw filesystem content Jun 09 08:23:31 I was about to mod the OS quite a bit but thought I better know how to return it to stock 1st. Jun 09 08:23:45 my host system is ubuntu Jun 09 08:25:00 <_Sy_> This page has links to everything you will need: http://circuitco.com/support/index.php?title=BeagleBoneBlack#Image_Files Jun 09 08:25:29 http://downloads.angstrom-distribution.org/demo/beaglebone/Angstrom-Cloud9-IDE-GNOME-eglibc-ipk-v2012.12-beaglebone-2013.06.06.img.xz Jun 09 08:25:36 that's the flasher image Jun 09 08:26:28 is that a multipart file? Jun 09 08:26:42 dm8tbr: that xz extention ?? Jun 09 08:26:50 let me check it out. Jun 09 08:27:05 <_Sy_> You will need the tool to unpack it Jun 09 08:33:17 _Sy_: lzcat or xz Jun 09 08:33:31 I see. I've never used xz before. Jun 09 08:33:36 love learning new stuff. Jun 09 08:34:02 <_Sy_> monday, I use www.unziplite.com Jun 09 08:34:19 _Sy_: do you know if you can write to this without an sd card? Jun 09 08:35:08 <_Sy_> Unpack the xz file on your host system, that will contain the img which you can then use the flasher tool to prepare sd card. Jun 09 08:36:10 I wonder I get full access to /dev/mmcblk0 while I'm connected via usb on my host. Jun 09 08:37:01 monday: it requires a µSD card Jun 09 08:37:06 <_Sy_> Don't do this on the BBB itself, prepare the image on a PC and flash to SD card using an external SD card writer Jun 09 08:37:16 i'm trying to write to the onboard 2gig emmc Jun 09 08:37:38 <_Sy_> ok, you do that using a PC connected to BBB via USB Jun 09 08:37:45 yeah Jun 09 08:37:53 <_Sy_> running the flasher Jun 09 08:38:03 _Sy_: yeah that is what I was going for. Jun 09 08:38:13 If I have access to the entire emmc Jun 09 08:38:45 <_Sy_> the flasher will take care of everything for you, once you select the image it will erase the emmc Jun 09 08:39:20 then I can just xz --decompress --stdout | cat > /dev/emmc Jun 09 08:39:57 <_Sy_> not sure why you would do that instead of using the flasher, which does it for you Jun 09 08:40:10 _Sy_: do you know where the flasher is? Jun 09 08:40:17 I'm looking for it now. Jun 09 08:40:49 monday: the image is the flasher. it goes onto a µSD card and will flash the eMMC if the BBB is booted from that µSD Jun 09 08:41:47 <_Sy_> http://www.crashcourse.ca/wiki/index.php/BBB_software_update_process Jun 09 08:47:54 _Sy_: wow that flasher process seems complicated. Jun 09 08:48:17 <_Sy_> tell me about it :) Jun 09 08:48:53 _Sy_: I thougth the ...img.xz file was the entire file system compressed and if I just used dd or cat straight to the emmc I would be cool. Jun 09 08:49:18 _Sy_: I still have more reading to do. Jun 09 08:49:38 <_Sy_> I haven't done it for a couple of weeks now....this is only the beginnging Jun 09 08:49:46 _Sy_: documentation says sudo dd if=BBB-eMMC-flasher-2013.05.20.img of=/dev/sdX bs=1M Jun 09 08:50:01 <_Sy_> yep, sounds correct Jun 09 08:50:08 instead of mounting the image and checking it. Jun 09 08:50:44 I just was hoping I could write the image file straight to the emmc if I had the BBB connected to my host via USB Jun 09 08:51:28 but what I don't understand is its a live system so I can't just write to the the entire device. I'm offered a single partition. Jun 09 08:51:44 monday: yes that doesn't work Jun 09 08:51:47 I better power it down and connect via usb and see. Jun 09 08:51:55 then check this out. Jun 09 08:52:09 you might get away with it if you'd craft a special kernel + initrd that would expose the whole eMMC over USB Jun 09 08:52:17 not really hard, but someone would have to do it Jun 09 08:54:38 * dm8tbr doesn't have a BBB yet, else he'd try Jun 09 08:54:59 actually I could probably try to on a BBW, should be about the same Jun 09 08:55:27 (the µSD ~= eMMC for that special case) Jun 09 08:56:08 dm8tbr: yeah I see /dev/sde on my host and that has a single partition. Jun 09 08:56:45 dm8tbr: so when I boot to a micro sd then that must free teh emmc Jun 09 08:56:50 koen: has anyone done that yet? a simple boot-over-USB into a kernel + initrd that exposes mmcblk0 as gadget block device? Jun 09 08:57:21 someone has on the bbw, or maybe bb xm Jun 09 08:57:28 was just reading about that today Jun 09 08:57:35 have a link handy? Jun 09 08:57:43 * m_billybob is researching iscsi root Jun 09 08:57:48 maybe . . Jun 09 08:58:22 I have my resources, but they are about different kinds of OMAP based hardware Jun 09 08:58:29 https://groups.google.com/forum/?fromgroups=#!topic/beagleboard/xyrvq8GN4IU Jun 09 08:58:47 mainly just uEnv.txt stuff that ive been looking at Jun 09 08:58:56 those guys are doing nfs root Jun 09 08:59:05 one looks to be using usbeth Jun 09 09:00:05 <_Sy_> Oh yes!!!! success....managed to get mysql.service to start after ntpdate.service, now both BBB's stay in perfect sync. even after a power down Jun 09 09:01:29 _Sy_ did you post on beaglefu ? cause i responded but didnt hear back from the guy . . . Jun 09 09:01:41 whoever it is. Jun 09 09:01:59 <_Sy_> sorry on what subect? Jun 09 09:02:14 Linaro + code:blocks Jun 09 09:02:26 <_Sy_> will check now, is there a response? Jun 09 09:02:43 Yes, if that was you asking on their Jun 09 09:02:49 <_Sy_> it was Jun 09 09:02:53 on there* i gave a solution Jun 09 09:03:03 we'll see if it works Jun 09 09:03:24 <_Sy_> reading it now, thanks Jun 09 09:03:29 np Jun 09 09:03:52 i usually read the forums about 5 times a day sometimes more when expecting responces back Jun 09 09:05:13 _Sy_ grats on the ntpdatte ting too :) Jun 09 09:05:40 thats somethign i need to iron out myself eventualy. ive got possible fixes that ive not yet looked into Jun 09 09:05:45 <_Sy_> ty, bit of a relief, would really like to sort out remote debugging and compiling tho....then I can start to write some code. Jun 09 09:05:56 <_Sy_> also, there is the micro SD issue... Jun 09 09:05:58 gdb wil work Jun 09 09:06:04 however Jun 09 09:06:10 ther are other options too Jun 09 09:06:38 black magic probe is one, that may work Jun 09 09:06:49 dont know much about it but people i know been talkign about it for other platforms Jun 09 09:06:59 <_Sy_> I was thinking of sacrificing a chicken :) Jun 09 09:07:04 heh Jun 09 09:07:13 learn how to use GDB through cli Jun 09 09:07:16 it not too hard Jun 09 09:07:37 there is also the tried and psuedo true printf() Jun 09 09:08:05 <_Sy_> the frustarting thing is its like I'm trying to climb out of a well and someone is adding more bricks to the top, so the well just keeps getting deeper and deeper... Jun 09 09:08:29 well you're not alone . . . baby steps Jun 09 09:08:46 <_Sy_> Really question why TI thought this was ready for release when so much of it doesn't work correctly yet. Jun 09 09:08:50 could be worse you ould be doing what I am right now lol Jun 09 09:09:54 dm8tbr, i take it that wasnt helpful, the link i gave. Jun 09 09:10:10 <_Sy_> The BBB is excellent, price is fantastic, but let down by web-site, documentation and OS support. Jun 09 09:10:34 expect TI kernel support quarter 4 Jun 09 09:10:46 by then the community will have passed them by 500x times Jun 09 09:11:38 TI sent me a couple experimental chips in an unmarked package with no packing list, no docs, and I can't find anything on their website with the part #. Jun 09 09:11:41 yay Jun 09 09:11:55 _Sy_ anyhow for the executable problem let me know if that cures it Jun 09 09:12:10 <_Sy_> will do... Jun 09 09:12:37 chmod a+x ./file will most definately cure the problem Jun 09 09:12:49 so long as you're running debian . . . Jun 09 09:13:20 If anyone knows what this chips is let me know: xaz42j86x Jun 09 09:14:17 Your search - texas instruments xaz42j86x - did not match any documents. Jun 09 09:14:22 mastiff^: maybe that's a serial number, not a chip symbol? Jun 09 09:14:23 nor google Jun 09 09:14:52 the other markings make less sense as a chip part# Jun 09 09:15:08 mastiff: I mean, do they even have a part# if it's an experimental chip? Jun 09 09:15:49 Yeah usually the x* parts are listed on the site as new parts at least. Jun 09 09:16:20 I doubt it's anything particularly special just a fuck up as far as posting docs, etc. Jun 09 09:16:31 QFN64 package Jun 09 09:17:36 could be a msp430 ;) Jun 09 09:17:41 lol Jun 09 09:17:56 Don't get me wrong though I love TI's samples :) Jun 09 09:18:43 why does the !@#$ing synopsys core trigger a transfer finished interrupt :S Jun 09 09:18:49 *doesn't Jun 09 09:20:50 ug think i may call it soon Jun 09 09:20:57 5th kernel ive compiled today Jun 09 09:21:04 well, maybe 4th Jun 09 09:21:42 iscsi root might not be in my cards soon Jun 09 09:32:19 m_billybob have a whisky and a song http://www.youtube.com/watch?v=AfUZAMoQXoc Jun 09 09:32:29 Forget how many kernels, its too many Jun 09 09:52:23 <_Sy_> problem....in ubuntu terminal I have the toolchain installed in /arm/arm-linux-gnueabihf Jun 09 09:53:05 <_Sy_> If I go to /arm/arm-linux-gnueabihf/bin and execute gcc it finds the application gcc and tells me that I need to supply a file, so far so good....but... Jun 09 09:53:39 <_Sy_> If I go to my workspace project folder /home/simon/workspace/hello_world/bbb_debug and type Jun 09 09:55:02 <_Sy_> and its the same if I omit ../hello_world.c, its not finding the application gcc even tho the parth is explicitly specified Jun 09 09:55:04 <_Sy_> whats going on? Jun 09 09:56:41 typo? Jun 09 09:58:03 <_Sy_> I used tab to auto complete, and have checked everything looks ok, just looking at the ownership on the executables, all set to 1001, could that be it ? Jun 09 10:00:08 <_Sy_> group ownership on all files is now root:root, but no improvement, if I go to the folder and type gcc it works....but not from any other location. Jun 09 10:03:10 <_Sy_> mmmm....its not picking up on the arm gcc at all, just typed which gcc and its acutlly running /usr/bin/gcc Jun 09 10:03:21 it's gcc in /arm/arn-linux-gnueabihf/bin or arm-eabi-none-gcc ? Jun 09 10:03:48 *arm Jun 09 10:03:55 err Jun 09 10:03:57 <_Sy_> the only folder I have is /arm/arm-linux-gnueabihf/bin Jun 09 10:04:02 arm-linux-gnueabi-gcc Jun 09 10:04:07 tried doing ls in /arm/arm-linux-gnueabihf/bin ? Jun 09 10:05:38 <_Sy_> yep, that works fine, lists all the files in it: ar, as, c++, g++, gcc, gfortran, ld, ld.bfd, ld.gold, nm, objcopy, objdump, ranlib, strip Jun 09 10:07:35 anything useful in dmesg? Jun 09 10:07:51 also, do this command in that directory: ldd gcc Jun 09 10:08:55 <_Sy_> Running ldd gcc, it comes back with not a dyanmic executable Jun 09 10:09:26 ls -l gcc? Jun 09 10:10:20 <_Sy_> -rwxr-xr-x 1 root root 477656 Mar 14 00:05 gcc Jun 09 10:11:01 you're messing with the wrong files Jun 09 10:11:46 executables should be named as such Jun 09 10:11:49 arm-linux-gnueabihf-gcc Jun 09 10:11:56 etc etc Jun 09 10:12:16 <_Sy_> I have those in /arm/bin, but it reports exaclty the same for those too Jun 09 10:13:27 <_Sy_> ldd arm-linux-gnueabihf-gcc "not a dynamic executable" Jun 09 10:13:40 google that Jun 09 10:13:48 find out what it means Jun 09 10:14:34 <_Sy_> ls -l arm-linux-gnueabihf-gcc "lrwxrwxrws 1 root root 29 Mar 14 01:12 arm-linux-gnueabihf-gcc -> arm-linux-gnueabihf-gcc-4.7.3" Jun 09 10:15:12 anyhow it's 3am here im off to bed Jun 09 10:15:18 good luck with that Jun 09 10:16:00 <_Sy_> ty, nite Jun 09 10:19:12 <_Sy_> any ideas ? Jun 09 10:23:18 <_Sy_> If I go to the folder where arm-linux-gnueabihf-gcc resides and try to launch it I get "The program 'arm-linux-gnueabihf-gcc' is currently not installed. You can install it by typing: sudo apt-get install gcc-arm-linux-gnueabihf" Jun 09 10:24:30 <_Sy_> However, if I do that it looks like it trys to install it but then reports: "gcc-arm-linux-gnueabihf" is already the newest version. Jun 09 10:24:39 <_Sy_> so I go around in the same endless loop.... Jun 09 10:33:54 <_Sy_> just performed "sudo apt-get remove gnu-arm-linux-gnueabihf" and it has removed an installation. Will try to install the toolchain again from the archive. Jun 09 10:35:39 _Sy_: try arm-linux-gnueabihf-gcc-4.7 Jun 09 10:35:52 without a path Jun 09 10:36:30 <_Sy_> from anywhere or the bin folder? Jun 09 10:36:35 from anywhere Jun 09 10:37:29 <_Sy_> I comes back with: No command 'arm-linux-gnueabihf-gcc-4.7' found, did you mean: Jun 09 10:38:03 <_Sy_> Command 'arm-linux-gnueabihf-gcc-4.6' from package 'gcc-4.6-arm-linux-gnueabihf' (universe) Jun 09 10:38:10 <_Sy_> also lists 4.5 and 4.4 Jun 09 10:38:17 which version of ubuntu? Jun 09 10:38:42 ok, try 4.6 instead of 4.7 Jun 09 10:39:08 <_Sy_> I will have to install it Jun 09 10:39:08 I use Ubuntu 13.04 and that has 4.7 Jun 09 10:39:18 you may already have it installed, just try it Jun 09 10:40:01 <_Sy_> Ok, I'll create a new virtual machine and update the ubuntu image, the image I'm using is 12-04.2 Jun 09 10:40:21 how about just typing arm-linux-gnueabihf-gcc-4.6 instead Jun 09 10:40:22 :> Jun 09 10:40:56 <_Sy_> I'd like to be up to date... I'll try 4.6 now and if that works then build an updated system Jun 09 10:49:54 <_Sy_> The download of the ISO is going to take quite a long time, will report back if it resolves any issues... Jun 09 10:51:34 I've never seen anyone that bad at problem-solving :S Jun 09 10:51:48 heh Jun 09 10:52:21 what will he do when he runs into a bug in his code? Jun 09 10:52:31 he'll ask us :P Jun 09 10:52:49 right Jun 09 10:53:09 but, seriously, he complains about Polish immigrants in England, but I think that's last of his worries at this moment :P Jun 09 10:53:34 I know a few polish immigrants in england Jun 09 10:53:39 honest, hard-working people Jun 09 10:55:46 well, as long as you don't run into our hillbilly types ;) Jun 09 10:57:03 there are hillbillies of all nationalities Jun 09 10:58:29 no doubt Jun 09 10:59:28 so picking on 'immigrants from X' is stupid Jun 09 11:00:03 even if some scum from poland were just convicted of murder here Jun 09 11:00:23 well, he didn't specificially mention Poland - it's just that due to our !@#$ right-wing government and opening of job market, Britain and Ireland had mass immigration from Poland and not other new-EU countries Jun 09 11:00:47 sure, there's plenty of scumbags around Jun 09 11:01:18 well, I don't have a problem with honest people moving somewhere for work Jun 09 11:01:44 of course I'm biased, being such a person Jun 09 11:01:54 well, I like to think I'm reasonably honest at least Jun 09 11:02:03 or at least I like others to think I am Jun 09 11:04:04 I've moved for many reasons - incompetent government, lack of *interesting* IT jobs, low wages in Poland, housing problems (i.e. all I could afford in Poland was a small flat, and that's senior software engineer salary - they're better built than most Irish houses, but still, you have to deal with neighbours, which sometimes sucks) Jun 09 11:04:55 I'm happy renting a small flat Jun 09 11:05:13 do you have a family? Jun 09 11:05:16 no Jun 09 11:05:32 if I did, I might think differently of course Jun 09 11:06:46 2 small rooms for a family with 2 kids is simply a pain in the ass, especially if you need peace and quiet to concentrate because you work on non-work projects in spare time Jun 09 11:07:02 oh, definitely Jun 09 11:07:19 you need a proper dungeon to keep the kids in Jun 09 11:07:39 Heh, you need a proper dungeon to keep the TV in ;) Jun 09 11:08:27 Also, what I worked on in Poland: webcrap. What I work on in Ireland: network probes used by NASDAQ, Goldman Sachs and other supervillains. Jun 09 11:49:39 kfoltman, love the name "super-villians" Jun 09 11:50:20 mru, I tossed mine in a cave...cheaper than a proper dungeon. Jun 09 11:50:35 heh, speaking of problem solving: I was getting intermittent failures after 30 control packets sent to my USB device implementation on STM32 (which uses Synopsys OTG core). Turns out I was supposed to flush the TX FIFO when disabling the IN endpoint :| Jun 09 11:50:50 whoops Jun 09 11:50:58 took me a whole morning to find that out :S Jun 09 11:51:27 ugh Jun 09 11:51:30 mourning Jun 09 11:51:57 more like moaning... that's all you can do about that core Jun 09 11:51:58 moaning ka6sox Jun 09 11:52:43 kfoltman: he's probably mourning proper comms interfaces Jun 09 11:54:34 or the Synopsys designers, after enough programmers have to deal with this mess Jun 09 11:55:04 but that wouldn't be mourning Jun 09 11:55:38 you think people from Synopsys don't have families/friends Jun 09 11:56:15 kfoltman, are they super-villians too? Jun 09 11:56:16 I don't think they do Jun 09 11:56:27 and it's made them bitter with the world Jun 09 11:56:39 and this is their revenge Jun 09 11:57:19 ka6sox: no, I don't think so - they're just not good at idiot-proofing and documenting their VLSI designs Jun 09 12:00:51 It's just that the core fails in a fairly weird way, as in, it works fine for 30 packets, no difference in status register values (at least those I've checked), then - also with no different with status register values - the next IN packet inserted to the queue (and the queue always contains only one packet) never gets picked up and never triggers an input-transaction-completed interrupt Jun 09 12:01:33 bitbang it Jun 09 12:01:41 and both in successful and unsuccessful case, the FIFO space register says 0x200 (= empty) Jun 09 12:01:43 haha Jun 09 12:01:47 suuure Jun 09 12:01:56 I'm not an Allwinner Jun 09 12:02:22 also, there's no fex implementation for stm32 ;) Jun 09 12:02:37 rpi almost bitbangs usb Jun 09 12:03:11 interrupt per microframe Jun 09 12:06:29 mru: I'm fairly sure it uses a similar core as stm32 (Synopsys, except I'm using the FS one and theirs is HS) Jun 09 12:10:13 if FS=10 Jun 09 12:10:21 and HS=480 Jun 09 12:10:27 what does that make USB3? Jun 09 12:11:04 bad idea? Jun 09 12:11:31 signal rate is 5G Jun 09 12:11:39 no idea what useful throughput you get Jun 09 12:12:05 I've always found that the USB framing is ugly for data thruput Jun 09 12:12:19 it is Jun 09 12:12:26 as is the half duplex Jun 09 12:12:26 well, I don't like bit stuffing Jun 09 12:12:28 and the polling Jun 09 12:13:21 so what you are saying is that musb being broken is a "Good Thing" as then you don't have to deal with it. Jun 09 12:13:22 :) Jun 09 12:14:44 usb3 is actually completely different from usb2 Jun 09 12:36:09 ssi: so it was that input mode on the clk? .. just verified it here also Jun 09 12:36:32 and I did it on spi1 for good measure Jun 09 12:36:42 now .. to figure out where i read that Jun 09 12:38:58 Hi experts.. is it normal not to have the omap folder under /sys/device/platforms in the BB white latest angstrom distribution? Jun 09 12:43:12 WhileTrue, I suspect so. Jun 09 12:43:20 ssi, i think it was here i read it https://patchwork.kernel.org/patch/1859291/ Jun 09 12:44:35 Rickta59, we found a non-spidev implmentationin capes/firmware after we got done Jun 09 12:45:03 heh .. sorry i can't parse that Jun 09 12:45:32 there was a non-spidev version of the DTS that ssi was working on already in the repo Jun 09 12:45:39 with that input set Jun 09 12:45:43 k got it Jun 09 12:45:51 i knew i had seen it someplace Jun 09 12:46:21 as i wasn't trying to do rx/tx i didn't really dig too deep Jun 09 12:46:37 what i wanted to do was working so i stopped Jun 09 12:46:51 right Jun 09 12:47:50 so .. how does sharing of the spi device work with linux? Jun 09 12:48:11 if the proper way is to make a device driver Jun 09 12:48:34 won't the device driver try to get exclusive access to it? Jun 09 12:49:14 say i wanted to put a small spi driver display and also something else Jun 09 12:49:35 and use different cs pins to enable and disable them as i use them? Jun 09 12:53:36 am I here? Jun 09 12:54:15 is that a philosophical question? Jun 09 12:54:37 you IRC, therefore you are. Jun 09 12:54:45 mru, was that directed to me? Jun 09 12:54:57 'I might think, therefore I'm not really sure if I am' Jun 09 12:57:26 is it possible to change the pinctrl setting programmatically or must it be done in the DT? Jun 09 12:58:07 specifically the PULLUP/PULLDOWN on a pin Jun 09 12:59:04 Rickta59, the way we are told to do this is via a DTBO Jun 09 12:59:23 omap_mux is not in the new images to do that. Jun 09 15:01:31 hi! I'm going to buy an AC adapter for my beaglebone Jun 09 15:01:58 <_av500_> make sure it outputs DC Jun 09 15:02:49 thanks _av500_, I've found this on amazon: Netzteil 5V 4A Jun 09 15:02:54 http://www.amazon.it/Adattatore-AC-4000mA-connettore-1mm/dp/B004S7U4IO/ref=sr_1_cc_1?s=aps&ie=UTF8&qid=1370789867&sr=1-1-catcorr&keywords=AC+4000mA Jun 09 15:03:25 <_av500_> seems fine Jun 09 15:03:58 I'm most interested in using the beaglebone as a server, with some usb peripherals attached (tv tuners, remote keyboards dongles, wifi..etc) Jun 09 15:05:49 Can an alimented usb hub be a valid substitute for a dedicated AC adapter? Jun 09 15:06:50 I mean, considering that I will use only usb peripherals, do I need a dedicated AC adapter at all or can I use a powered usb hub? Jun 09 15:08:54 As long as my pheripals don't exeed 500mA each, an alimented usb adapter should be able to sustain their needs and I can use the microusb cable to power the beaglebone, am I right? Jun 09 15:10:35 *alimented usb hub Jun 09 15:16:50 morn Jun 09 15:17:40 cool Jun 09 15:17:41 http://crna.cc/MY7lDnKO2kOcie9 Jun 09 15:17:46 it took me 5 hours .... Jun 09 15:17:54 I had 0 && there ..... Jun 09 15:17:57 fuccccckkkkk Jun 09 15:18:17 so the zeroed things weren't being zeroed Jun 09 15:18:20 fuuuuccccckkk Jun 09 15:27:47 Kristina: it's always those little bugs that your brain just can't see Jun 09 15:29:29 sometimes it's a good idea to stop looking for a while Jun 09 15:29:34 agreed Jun 09 15:29:45 and come back with fresh brain and less pre-conceptions about what your own code does ;) Jun 09 15:30:16 someone here has reported issues using a powered usb hub (https://groups.google.com/forum/#!topic/beagleboard/SbEpcNvtiA4) I guess I'm better using a dedicated AC adaptor then Jun 09 15:31:49 hey ssi, i saw you ramblings last night .. glad you got it working Jun 09 15:31:54 your Jun 09 15:31:58 yeah Jun 09 15:32:15 i confirmed it here also .. works fine with both spi0 and spi1 Jun 09 15:32:17 you were right but I was stubborn ;) Jun 09 15:32:33 i wasn't sure .. but i thought i had read it somepace Jun 09 15:32:36 lace Jun 09 15:32:46 http://comments.gmane.org/gmane.linux.ports.arm.omap/90699 Jun 09 15:32:49 that's where I read it Jun 09 15:32:51 seemed bizzare to me also Jun 09 15:33:02 it mostly makes sense to me now that I'm thinking about it Jun 09 15:33:05 but what about the BBB doesn't so i just went with it Jun 09 15:33:37 i tried to find a reference in the TRM but I didn't see anything that indicated it needed to be setup as input Jun 09 15:33:46 cause setting a pad as input doesn't really make it an input as opposed to an output Jun 09 15:33:50 it just enables the receiver on the pad Jun 09 15:35:01 so later I need to write some verilog that talks spi, and test duplex comms with the fpga Jun 09 15:35:09 do you know which mcSPI register is getting set when that is configured that way? Jun 09 15:35:44 configuring the pin as input? Jun 09 15:35:50 that's not a mcspi register, it's a padcontrol register Jun 09 15:35:58 padcontrol is a kernel thing? Jun 09 15:36:06 no, it's an omap thing Jun 09 15:36:16 which is a peripheral ? Jun 09 15:36:23 or a data structure in the kernel Jun 09 15:36:37 I guess you could consider it a peripheral Jun 09 15:36:52 it's the 44e100000 chunk of the control module Jun 09 15:37:11 -0 Jun 09 15:37:12 heh Jun 09 15:37:42 AHA Jun 09 15:37:44 it is in the TRM Jun 09 15:37:49 SPIx_SCLK Jun 09 15:37:49 I/O (1) Jun 09 15:37:50 SPI serial clock (output when master, input when slave) Jun 09 15:37:54 (1) This output signal is also used as a re-timing input. The associated CONF___RXACTIVE bit for the output clock must be set to 1 to enable the clock input back to the module. Jun 09 15:38:03 there is a FOOTNOTE Jun 09 15:38:05 :D Jun 09 15:38:25 page 3996 Jun 09 15:38:27 24.2.3 Jun 09 15:39:09 missed that .. thanks Jun 09 15:39:26 that is a tiny ass footnote Jun 09 15:40:59 hey emeb Jun 09 16:01:05 ssi: hey yourself. do you ever sleep? :) Jun 09 16:12:54 mourning Jun 09 16:15:43 hola dude Jun 09 16:16:02 Yo! What up? Jun 09 16:16:24 are we ready to play some GPMC? Jun 09 16:17:45 heh Jun 09 16:17:56 you want a .ucf file? Jun 09 16:19:01 can you put it up on github with the other bcc stuff? Jun 09 16:20:14 sure Jun 09 16:20:25 gotta refresh my memory on how that works... Jun 09 16:27:06 here: https://github.com/emeb/bcc_hardware/blob/master/xilinx/bcc.ucf Jun 09 16:28:09 * emeb_mac is surprised to have remembered his minimal git skills Jun 09 16:31:31 got it Jun 09 16:31:48 just uncomment the pins you need in any particular design. Jun 09 16:32:00 (and comment out the ones you don't) Jun 09 17:16:24 has anyone compiled opecv2.4 from the source tarball for beaglebone black? Jun 09 17:17:23 I am wondering what would happen if I did this under the default angstrom distro Jun 09 17:19:51 <_av500_> an explosion? Jun 09 17:20:32 I am thinking it would probably clobber something Jun 09 17:23:14 <_av500_> you could just try Jun 09 17:23:35 I guess, as damange would be limited since reflashing is possible.. Jun 09 17:24:00 WOuld it be easier to just run straight debian or ubuntu on it (standardised packages, etc) Jun 09 17:33:19 perhaps it is safer to cross-compile on my ubuntu box and copy the files into the python extension directories. I have angstrom toolchain.. Jun 09 17:33:52 If someone here knows what commands were used to compile opencv for beaglebone, it would be helpful Jun 09 17:37:01 it seems there is a new toolchain as of 29-May - angstrom-eglibc-x86_64-armv7a-vfp-neon-v2012.12-toolchain.gz Jun 09 17:37:36 <_av500_> bminuk: you need libs to compile against Jun 09 17:37:53 <_av500_> so either you build it on your PC using angstrom Jun 09 17:37:56 <_av500_> or on the BBB Jun 09 17:38:09 <_av500_> but just a cross toolchain is not enough Jun 09 17:38:14 <_av500_> by why not just install openCV? Jun 09 17:38:18 <_av500_> angstrom has it already Jun 09 17:38:42 the one in angstrom doesn't have the python extensions installed Jun 09 17:39:05 unless they fixed it recently Jun 09 17:40:48 Can the BBlack deliver enough power to an external usb HDD? I've got two amps going in @ 5V and am only running SSH and console. Jun 09 17:41:00 Arch Linux - atm I plug in and nothing happens, no listing in lsblk etc. Jun 09 17:41:38 who is the maintainer for opencv on beaglebone? Jun 09 17:45:35 heh Jun 09 17:45:49 maintainer? Jun 09 17:45:55 cellardoor: I'd recommend using and external hub Jun 09 17:46:09 powered external hub Jun 09 17:46:15 ok, the guy who issued the commands to build the thing Jun 09 17:49:16 Hi all, any idea how to enable Gigabit Ethernet on the Beaglebone Black? Jun 09 17:52:52 I found that there are angstrom packages for opencv 2.4, but the python bindings are for opencv 2.3... Jun 09 17:54:46 hm leikr watch looks cool Jun 09 17:55:03 bminuk yes koen disabled them for 2.4 Jun 09 17:55:20 Hello, just noticed a minor bug, not sure where to file it. Basically, if I call my device tree fragment file cape-bone-lcd1800-00A0.dtbo then I cannot install it, I get "-bash: echo: write error: No such file or directory" but if I rename the file to cape-bone-lcd18-00A0.dtbo then it installs fine. I've tried a few different filenames, it doesn't seem to like me having more than two numbers, Jun 09 17:55:20 so lcd18 is ok, but lcd1800 is not. Not a big deal, I can rename my cape file, but wanted to let people know just in case Jun 09 17:55:28 convince him to enable them again for newer versions Jun 09 17:56:25 that is with the latest eMMC flasher (06/06) Jun 09 17:57:14 did koen give a reason? Jun 09 17:57:19 bminuk no Jun 09 17:57:28 only he had some problems Jun 09 17:58:22 is koen the maintainer for all of angstrom, or just the beaglebone? Jun 09 17:59:58 all anstrom Jun 09 18:00:13 hm 300 euros for the leikr watch Jun 09 18:00:19 with a8 Jun 09 18:11:58 does anybody have references on gigabit support for the beaglebone black. I can not find anythign on google Jun 09 18:12:18 the chip clearly states in the datasheet that it is capable of Gigabit ethernet Jun 09 18:12:46 however the beaglebone black only supports 100MBit. Why is this? How can we harness the onchip gigabit? Jun 09 18:14:59 presumably the magnetics, or board can't cope Jun 09 18:15:24 it may take anything from swapping out the mag jack to adding a GE phy chip Jun 09 18:15:43 it's not going to be software Jun 09 18:16:43 I wonder why this wasn't included, a true shame : ( Jun 09 18:16:47 gigabit ethernet would have been nice Jun 09 18:17:08 presumably, it was one, or several dollars more Jun 09 18:17:35 :( I would gladly pay an extra 40 $ for that functionality Jun 09 18:17:39 and then there is the issue that it cannot generally sustain much over 100mbut Jun 09 18:17:43 bit Jun 09 18:18:19 why not? Jun 09 18:18:29 for ge, it's likely you can stream data from memory, but not much more Jun 09 18:18:54 there isn't a way of reading data that fast into the beagle Jun 09 18:19:06 not to saturate Jun 09 18:21:04 beaglebone with latest image. How could I assign a static IP to eth0? Jun 09 18:21:37 ifconfig Jun 09 18:21:51 and after reboot? Jun 09 18:22:53 you could set ethaddr in uboot Jun 09 18:23:04 err, ipaddr. sry Jun 09 18:23:37 via "/boot/uEnv.txt"? Jun 09 18:24:22 but i'm not sure if linux uses that address also. never tried that. Jun 09 18:25:30 personally i don't use a uEnv.txt so idk. Jun 09 18:26:04 i simply make setenv ipaddr=x.x.x.x and saveenv Jun 09 18:26:41 i still have no cable to direct conntect to the uboot sequence. Jun 09 18:27:55 if you need a static ip only under linux it may be sufficient to set it in the /etc/network/interfaces file. Jun 09 18:28:00 And my local "dhcp-server" on the DSL-Router is buggy Jun 09 18:28:15 the file /etc/network/interfaces do not exist :( Jun 09 18:28:38 root@beaglebone:/etc# cat /etc/angstrom-version Angstrom v2012.12 (Core edition) Built from branch: angstrom-staging-yocto1.3 Revision: 93f9f674059d8003a84ef1caad39339c2d940d4d Target system: arm-angstrom-linux-gnueabi Jun 09 18:29:39 i usually run DEBIAN or RH/SLES Jun 09 18:30:00 5.1GB of stuff takes a LONG time to download Jun 09 18:31:49 on the "image", the man pages are missing :-( Jun 09 18:34:42 koboldmaki, its an *embedded* image... Jun 09 18:34:53 so things like manpages are usually not included Jun 09 18:35:27 connman is the "magic"... but how could a gui be magic?!? Jun 09 18:35:37 ups... connman-properties Jun 09 18:38:25 hi ! Jun 09 18:38:40 I can't manage to get gpio interrupts to work in userspace :-( Jun 09 18:38:51 i set the "edge" correctly Jun 09 18:39:05 but every "poll" return immediatly (with a return value of -1) Jun 09 18:39:20 duh, i mean timeout value of -1 Jun 09 18:41:16 are the pins muxed properly to start with? Jun 09 18:41:45 Hello Jun 09 18:41:52 yeah Jun 09 18:41:58 i can read the values Jun 09 18:42:05 using sysfs Jun 09 18:42:42 /proc/sys/debug/gpio doesn't indicate that an interrupt is configured Jun 09 18:43:22 gpio-36 (sysfs ) in hi Jun 09 18:43:24 :-/ Jun 09 18:43:45 and /sys/class/gpio/gpio36/edge is set on "falling" Jun 09 18:44:14 this is 3.2.42-psp27 Jun 09 18:47:29 hmmm /proc/interrupts show the interrupt Jun 09 18:47:35 doesn't help my code though Jun 09 18:56:44 The Keyboard and Mouse will not power up when connected to BBB USB host port. It worked fine for a couple of weeks but recently become unhappy. Could this be a problem with the distributed Angstrom Software? Jun 09 18:59:41 Chuck__: powering from wall wart? Jun 09 19:00:06 yes 5vdc 2 amp from adafruit Jun 09 19:01:49 The hub I am using is powered too. Jun 09 19:10:20 Could I have mucked up the power management chip by running minix from the sd card using a 3.3v interface cable on the debug serial header? Jun 09 19:14:31 Chuck__, you know you *could* put ubuntu or debian on an SD card and see if the same problems exist on that/those distros Jun 09 19:14:38 if you're wondering about Angstrom Jun 09 19:14:52 you may boot from an SD card too Jun 09 19:16:58 m_billybob, I tried an ubuntu image and have the same results. Jun 09 19:17:41 you may want to contact Gerald on the groups then and see if he has any solutions Jun 09 19:17:57 beaglebone google groups Jun 09 19:18:13 damn it doesn't work :-( Jun 09 19:20:29 grrrr Jun 09 19:21:02 12:20] * Disconnected Jun 09 19:21:03 [12:20] * Attempting to rejoin channel #beagle Jun 09 19:21:39 ssi, you here? Jun 09 19:21:51 Chuck__ anyhow yes, ask on the beagleboard user groups, see what Gerald has to say. You may also want to say somethign about thinking abotu an RMA Jun 09 19:25:46 [15:17] < Kristina> | Hm. Jun 09 19:25:46 [15:18] < Kristina> | I get a data abort at 0x8064bbb4 but when I read it using a debugger, it reads it just fine. Jun 09 19:25:49 [15:18] < Kristina> | And I turned off icache and dcache. Jun 09 19:25:52 [15:18] < Kristina> | But it still does that. Jun 09 19:26:39 Can anyone suggest what the problem might be? Jun 09 19:27:55 Is there some OMAP specific thing I'm missing out on? Jun 09 19:29:00 After /w 2 Jun 09 19:29:02 Er. Jun 09 19:30:10 And I did clear the TLB after doing the mappings. Jun 09 19:30:13 <_av500_> Kristina: ? Jun 09 19:30:32 _av500_: Ignore that line with /w 2 Jun 09 19:31:12 <_av500_> the debugger reads phys or virtual? Jun 09 19:31:22 virt Jun 09 19:31:26 <_av500_> since 0x8 is the physical dram Jun 09 19:31:30 <_av500_> no? Jun 09 19:31:41 > mdw 0x7064bbb4 Jun 09 19:31:41 Address translation failure Jun 09 19:31:41 in procedure 'mdw' Jun 09 19:31:47 It is virtual. Jun 09 19:32:25 woohoo slackware running on my beaglebone black Jun 09 19:33:00 This is the whole crash: http://crna.cc/uFbIcAkjwAThKhs Jun 09 19:33:25 but the thing is, 0x8064bbb4 is valid Jun 09 19:33:36 and is in the translation table Jun 09 19:34:31 and to disable the caches I did: Jun 09 19:34:32 bic r5, #(1 << 2) /* dcache */ Jun 09 19:34:36 bic r5, #(1 << 12) /* icache */ Jun 09 19:34:43 mcr p15, 0, r5, c1, c0, 0 Jun 09 19:34:59 <_av500_> Kristina: thanks for dark blue on black Jun 09 19:35:11 <_av500_> you should add red too :) Jun 09 19:35:12 anyone else running slack on their bbb? Jun 09 19:35:14 _av500_: I like it :) Jun 09 19:35:20 <_av500_> Kristina: my eyes dont Jun 09 19:35:36 _av500_: it gets more readable if you select all of it :) Jun 09 19:35:38 <_av500_> blue has a different focus distance Jun 09 19:35:52 _av500_: http://crna.cc/H2mmfJNrzo5eFc7 Jun 09 19:35:54 happy? Jun 09 19:36:03 _av500_: or just copy-paste to text editor :P Jun 09 19:36:08 <_av500_> sure Jun 09 19:36:11 <_av500_> ::effort:: Jun 09 19:36:28 anyway, and suggestions? Jun 09 19:36:31 <_av500_> Kristina: "crna" as in black? Jun 09 19:36:34 don't want to make potential helpers put in extra effort :P Jun 09 19:37:12 leave my poor website alone :( Jun 09 19:37:21 <_av500_> just wondering Jun 09 19:37:26 at least fuckface was in green, heh Jun 09 19:37:43 colorful no matter what color it is in ;) Jun 09 19:38:04 :( Jun 09 19:38:06 there are kids in here! Jun 09 19:38:13 <_av500_> mranostay: then leave Jun 09 19:38:15 can anyone help me with the actual problem? :P Jun 09 19:38:18 lol Jun 09 19:38:29 _av500_: i knew that was coming Jun 09 19:38:42 "fuckface" is the platform verification code for xnu-arm :D Jun 09 19:38:44 <_av500_> Kristina: ok, what linux kernel is that? Jun 09 19:38:51 _av500_: it's darwin Jun 09 19:38:52 * m_billybob grins Jun 09 19:38:56 mach kernel... darwin stuff? Jun 09 19:38:57 it says it's darwin Jun 09 19:38:58 _av500_: troll! Jun 09 19:39:02 heh Jun 09 19:39:04 Darwin Kernel Version 12.2.0: Sun 9 Jun 2013 19:40:14 BST; Jun 09 19:40:16 :( Jun 09 19:40:24 damn Jun 09 19:40:41 <_av500_> Kristina: not much to say from that log Jun 09 19:40:48 <_av500_> I trust your mappings are correct Jun 09 19:40:59 _av500_: then what could be the problem? Jun 09 19:41:04 mastiff* you around? Jun 09 19:42:14 yo Jun 09 19:42:18 <_av500_> Kristina: sorry, no idea atm Jun 09 19:42:22 did I actually disable the caches? or is there some sort of an omap smc thing that you need to call as well? Jun 09 19:42:58 debugger says MMU: enabled, D-Cache: disabled, I-Cache: disabled Jun 09 19:43:34 omap3 has a 100% compatible tt structure with the armv7 vmsa spec, right? Jun 09 19:43:46 <_av500_> omap3 is arm cortex a8 Jun 09 19:44:01 so that's a y Jun 09 19:44:03 yes Jun 09 19:44:13 <_av500_> I never read it was anything special Jun 09 19:44:18 what about that weird smc thing? Jun 09 19:44:53 hiya Mastiff whats up today? Jun 09 19:45:57 maybe the cache isn't actually disabled? Jun 09 19:46:59 Kristina: any way to confirm that? Jun 09 19:47:09 no idea Jun 09 19:47:10 Kristina: what if it's something trivial that isn't cache-related? Jun 09 19:47:21 such as? Jun 09 19:47:28 (not that I'd know, I know exactly nothing about ARM MMU) Jun 09 19:47:42 <_av500_> Kristina: in your boot process is that the forst time a mapped address is hit? Jun 09 19:47:46 <_av500_> first Jun 09 19:47:56 nope Jun 09 19:48:07 it's hit by the bootloader when it maps the kernel in Jun 09 19:48:30 <_av500_> the bootloader enables the mmu? Jun 09 19:48:42 hey ka6sox, my mission for today is to rotate hte FPGA on my BCC by 180 degrees. Xilinx decided it was a good idea to put circles on two corners of the chip, and I used the large circle for alignment,. Then had a major face palm when I looked at emeb's finished device. Jun 09 19:48:47 _av500_: yes Jun 09 19:48:57 <_av500_> mastiff: ouch Jun 09 19:49:29 Kristina: is there an instruction that can check if an address can be accessed without a page fault? is it possible for you to add a bit of code to run that instruction on every page in some address range to give you a memory map of some sort, of accessible/inaccessible pages? Jun 09 19:49:53 mastiff: jaysus Jun 09 19:49:53 kfoltman: there isn't an instruction like that Jun 09 19:50:13 <_av500_> mastiff: I once had to rotate a chip by 45 degrees Jun 09 19:50:42 Kristina: (there is one on x86, hence my question) so the only way to check it involves being blown up to pieces? :/ Jun 09 19:50:53 yes Jun 09 19:51:03 First I have to go buy a heat gun. Also working on fixing a Porsche 911 that had the battery connected with reverse polarity. Jun 09 19:51:15 <_av500_> mastiff: you like 180 ? Jun 09 19:51:52 The genious owner (not me) disconnected the battery for the winter, and hooked up a trickle charger backwards. Now that battery has +12V on the negative terminal. Jun 09 19:52:04 Had no idea that could even happen lol Jun 09 19:52:06 <_av500_> nice Jun 09 19:52:10 uh, i'm going cycling now, when I come back, I should probably look at what's in IFAR Jun 09 19:52:15 er, IFSR Jun 09 19:52:24 <_av500_> safe ride Jun 09 19:52:51 What are you guys working on today? Jun 09 19:53:32 !%$% USB (but it's not BBB USB, just a future peripherial I want to connect to it) Jun 09 19:58:40 I'm at a stage where things mostly work. Mostly. It's "just" fixing all the sloppy work and missed nuances in the spec. Jun 09 19:58:56 mastiff I"m working on the BCC dts right now, hoping to test to see that we don't blow up things but disable the eMMC by loading the GPMC stuff Jun 09 19:59:41 <_Sy_> hello, had a lot of success l8r today...thank you kflotman, installed ubuntu 13.04 and updated with toolchain, project compiled and built with 0 errors and 0 warnings. Jun 09 19:59:46 <_Sy_> Have posted on http://forum.beaglefu.com/index.php?/topic/82-codeblocks-linaro-for-windows-quick-rundown/#entry257 Jun 09 20:00:06 <_Sy_> Sorry *kfoltman Jun 09 20:02:59 anyone running a 3.9 kernel yet on the bbb? Jun 09 20:04:28 if I try to access processor registers from a userland program, would linux let me or would i get a segfault? Jun 09 20:04:37 _Sy_: I think the next step would be getting some good Linux book, and possibly some good embedded Linux book, if there's any (I have "Embedded Linux Primer" but much of it is outdated) Jun 09 20:05:42 <_Sy_> Ty, will definately have to get something like that...use linux at work, but only know the basic stuff, use QNX too...but admittedly my linux knowledge is limited. Jun 09 20:10:11 ka6sox-away: that sounds excellent, if I'm very lucky my FPGA will be rotated by time you guys have that sorted out. Jun 09 20:12:02 <_Sy_> Ty, will definately have to get something like that...use linux at work, but only know the basic stuff, use QNX too...but admittedly my linux knowledge is limited. Jun 09 20:12:06 <_Sy_> oops, sorry Jun 09 20:13:00 <_Sy_> I also managed to get mysql.service to start after ntpdate.service which was another big step forward, now both my BBB's synchronise and keep mysql HA running, works really well. Jun 09 20:37:21 <_Sy_> one step forward....still not there yet tho...although I can now compile a project using the toolchain in ubuntu with 0 errors, when I transfer the executable to the target although it has the correct permissions it will not execute....all it is supposed to do is display "hello world" Jun 09 20:37:42 <_Sy_> The target BBB is running angstrom Jun 09 20:37:55 which toolchain are you using _Sy_ ? Jun 09 20:38:15 the linaro binary? Jun 09 20:38:27 <_Sy_> the tool chain I have is gcc-4.7-arm-linux-gnueabihf Jun 09 20:38:46 that make code that runs on the BBB when it is running ubuntu Jun 09 20:39:15 http://www.angstrom-distribution.org/toolchains/ Jun 09 20:39:27 <_Sy_> thank you, will get it and try that. Jun 09 20:39:46 <_Sy_> which one, they're are several files Jun 09 20:40:02 you could grab the one for i686-linux-armv7a-linux .. with or without qt Jun 09 20:40:19 ka6sox-away: I'm back Jun 09 20:40:51 not sure, but you might have to point at newer angstrom libs and headers Jun 09 20:41:13 <_Sy_> How about this one angstrom-eglibc-x86_64-armv7a-v2012.05-toolchain.tar.bz2 ? Jun 09 20:41:48 that would probably work Jun 09 20:41:56 what is your host machine? Jun 09 20:42:46 <_Sy_> My dev. system is Windows 7 x64, but I'm running Oracle VirtualBox with Ubuntu 13.04 Jun 09 20:43:35 should be fine Jun 09 20:43:49 so your host machine is a ubunto on a x65 Jun 09 20:43:52 so your host machine is a ubunto on a x64 Jun 09 20:43:57 so your host machine is a ubuntu on a x64 Jun 09 20:43:59 getting closer Jun 09 20:44:00 hahaha Jun 09 20:44:08 3rd times the charm Jun 09 20:45:56 <_Sy_> yes Jun 09 20:46:03 <_Sy_> ubuntu x64 Jun 09 20:46:52 <_Sy_> getting closer will be sorted so I can actually start writing some code tomorrow :) Jun 09 20:47:40 anyone have any links to samples of making a loadable device driver that uses spi? Jun 09 20:47:40 <_Sy_> been a bit of an uphill struggle, but have learnt alot along the way... Jun 09 20:48:00 Rickta59: I think he spent enough time trying to get the current toolchain to work, no point in getting all OCD on toolchains - either eabihf or eabi should work Jun 09 20:48:12 is that true? Jun 09 20:48:14 kfoltman: Jun 09 20:48:20 i got the same results Jun 09 20:48:28 using the linaro one Jun 09 20:48:39 switched to the angstrom one and it just worked Jun 09 20:48:39 I think the Ubuntu toolchain *is* the linaro one Jun 09 20:48:56 he said he was running angstrom on the BBBB Jun 09 20:49:10 it might be as simple as missing dynamic library Jun 09 20:49:14 Rickta59: I dont have any specific examples of one, but I might be interestd to help you with it Jun 09 20:49:20 i don't think so Jun 09 20:49:20 Sy: can you do ldd on your binary on the target? Jun 09 20:49:25 Rickta59: what do you need a driver for? Jun 09 20:49:32 doing an ldd on the binary in angstrom gives nothing Jun 09 20:49:38 although it isn't static Jun 09 20:49:47 doing it in ubuntu shows libc etc.. Jun 09 20:50:12 i want to take the thing i did for spidev and learn how to map those things to a loadable driver Jun 09 20:50:44 Hi Everyone, I'm trying to enable GPIO2_1 as an input, but this seems not possible using a device tree fragment (I'm setting 0x2f according to the pinmux.tking.org and the reference manual. So, then I tried to set it using the PRU, since it has access to the entire address space too. But, I just can't see any input. Works fine for GPIO1, but GPIO2 doesn't work. I'm wondering if some register Jun 09 20:50:44 I've not thought of needs setting. I've set the address CONF_GPMC_CLK (which is GPIO2_1) to 0x2f, set the bit 1 in GPIO1 | GPIO_OE, and then read address GPIO | GPIO_DATAIN and tested bit 1. It's always low. So I think some other register also needs configuring, but I can't see what Jun 09 20:51:56 so ws2811 ssi Jun 09 20:52:37 aha Jun 09 20:52:50 not sure if the led driver makes sense for it Jun 09 20:52:57 does that let you address an array of leds? Jun 09 20:53:03 I don't know anything about it Jun 09 20:53:30 k Jun 09 20:53:54 shabaz: have you read the data sheet for the processor? Jun 09 20:54:25 I've read the GPIO chapter, but I must be missing something :-( Jun 09 20:54:26 shabaz: setting OE and the pinmux value is all you need to do Jun 09 20:54:35 http://www.jumpnowtek.com/?option=com_content&view=article&id=57&Itemid=62 Jun 09 20:54:39 i came across that .. Jun 09 20:54:41 shabaz: make triple sure you're using the correct values for the pin you're physically attached to Jun 09 20:54:54 but gear towards a different board and linux version i think Jun 09 20:55:32 but probably close to what i want Jun 09 20:55:46 what board is your ws2811 on? Jun 09 20:55:54 ie what are you trying to do with it Jun 09 20:55:58 gnd/vcc/spi_d1 Jun 09 20:56:01 wires no board Jun 09 20:56:07 it is led strip Jun 09 20:56:18 ok Jun 09 20:56:43 i have it working with spidev Jun 09 20:56:52 but i want to understand the proper way to do it Jun 09 20:57:00 with a device driver Jun 09 20:57:06 depending on what you're trying to do with it, that may BE the proper way to do it Jun 09 20:57:19 yeah just trying to conform to biases Jun 09 20:57:41 well if you had a spi LCD, you might want a device driver so you had a char device for it Jun 09 20:57:44 I guess Jun 09 20:58:08 yeah that is part of it i want to be able to understand how to in general interface to spi device Jun 09 20:58:10 I mean I guess you could write a device driver where you write a float from 0 to 1 in it, and it light sup that much of the strip or something Jun 09 20:58:11 s Jun 09 20:58:14 I just dunno what your application is Jun 09 20:58:26 you could make anything out of it Jun 09 20:58:30 exactly; that's all I thought I needed to do. I have triple-checked, but I'll do it again. Very bizarre. I know it's easy to make a mistake, so I did do my best to ensure it was accurate. I'm using pin 18 on port 8, connected via a resistor (4.7k) to 3.3V or 0V, to test the input. This worked on GPIO1. Pin 18 should be GPIO2_1 i.e. CONF_GPMC_CLK and the configuration in the device tree is <0x8c Jun 09 20:58:30 0x2f>. I have set the OE to 1, i.e. GPIO2 | GPIO_OE bit 1 is set high Jun 09 20:58:31 a display? Jun 09 20:58:44 a string of lights and chase them Jun 09 20:58:55 so it might be better as an fb device Jun 09 20:59:07 or maybe it is just ambient light Jun 09 20:59:17 16:58 < Rickta59> you could make anything out of it Jun 09 20:59:19 that's my point Jun 09 20:59:19 you just slam rgb data at it Jun 09 20:59:23 spidev lets you make anything out of it Jun 09 20:59:40 you'd write a specific device driver if you wanted to interact with it in a specific way Jun 09 21:00:05 well i guess there is a couple general uses Jun 09 21:00:07 as a display Jun 09 21:00:24 as a status lights Jun 09 21:00:40 you get a wall of leds using spi Jun 09 21:01:22 or you could use it to handle a panel of individual lights Jun 09 21:02:03 shabaz: some processors have a pio status register. dunno if the ti have. usually you can see there if a pio port is muxed to a peripheral or gpio. maybe that helps? Jun 09 21:02:17 i also want to understand spi and devices drives so i can use it to interface to some standalone chips Jun 09 21:02:29 maybe an lpc1114 or a msp430 Jun 09 21:03:10 maybe instead of flogging myself to do time sensitive stuff on the linux i'd just send the data down to a chip and let it handle that Jun 09 21:03:25 Rickta59: I implemented SPI using the PRU, to drive an LCD Jun 09 21:03:32 link? Jun 09 21:03:41 on 3.8? Jun 09 21:03:43 shabaz: why use the PRU, when mcspi already knows how to do it? Jun 09 21:03:54 partly because it isn't continues Jun 09 21:03:57 contiuous Jun 09 21:04:16 at least you can't do continuous with spidev Jun 09 21:04:20 maybe with a driver Jun 09 21:04:26 I don't know what you mean by continuous Jun 09 21:04:33 no gaps on the clock Jun 09 21:04:39 consistent ticks Jun 09 21:04:49 spi isn't supposed to have a continuous clock Jun 09 21:04:52 It was an LCD I was unfamiliar with, and wasn't sure I was going to stick with it. So I implemented with the PRU so if I needed to tweak the protocol, I could Jun 09 21:04:59 right so i'm abusing it Jun 09 21:05:01 link is here: http://www.element14.com/community/community/knode/single-board_computers/next-gen_beaglebone/blog/2013/05/28/bbb--connecting-up-an-lcd Jun 09 21:05:03 it only run the clock after it asserts cs Jun 09 21:05:20 i want to assert CS and keep it asserted unto my ioctl returns Jun 09 21:05:47 and a lot of chips can handle gapless spi Jun 09 21:05:56 back to back using a fifo Jun 09 21:06:05 thanks shabaz Jun 09 21:06:20 no prob Jun 09 21:08:17 so i guess the ws2811 isn't the best example of why i want to understand this stuff Jun 09 21:08:42 i'd really like to be able to setup a mcu as a slave and read and write from it Jun 09 21:12:14 so is that something that interests you ssi? Jun 09 21:13:26 it is Jun 09 21:13:40 I am about to write an SPI slave module in verilog for this fpga Jun 09 21:13:46 and I might try to write a driver for interacting with it Jun 09 21:13:48 perfect Jun 09 21:13:51 but I don't really know how I'd do it yet :) Jun 09 21:14:05 I'm kinda in the same boat you are Jun 09 21:14:10 I want to understand it, but I don't know where to start yet Jun 09 21:14:13 yeah fgpa/mcu .. Jun 09 21:14:28 did you see that link i posted Jun 09 21:14:38 http://www.jumpnowtek.com/index.php?option=com_content&view=article&id=57&Itemid=62 Jun 09 21:14:43 yeah Jun 09 21:14:52 i think that is guts of what has to be done Jun 09 21:15:09 i'm fuzzy on how the DT fits into kernel drivers Jun 09 21:15:55 i'm guessing you have to query the kernel for it Jun 09 21:16:03 and it give you back something Jun 09 21:16:33 that article is non-DT Jun 09 21:16:37 right Jun 09 21:16:57 but the driver ties into the DT using compatible string and the of_paramater stuff Jun 09 21:17:08 then the DT can figure out what driver to load based on the tree Jun 09 21:17:12 and configure the driver Jun 09 21:17:45 k Jun 09 21:17:53 the driver will likely do direct register writes to mcspi Jun 09 21:18:13 I'll have to look at spidev more closely Jun 09 21:18:53 before I get into that mess, I need to get something on the fpga that I can communicate with using spidev Jun 09 21:19:00 sure Jun 09 21:19:04 which is what i did Jun 09 21:19:10 then I'd need to figure out how I'd want to interact with linux to talk to the fpga Jun 09 21:19:14 and i could probably do all i want to do with spidev Jun 09 21:19:15 and base my driver work around that interface Jun 09 21:19:23 I'm sure you could do what you need with spidev Jun 09 21:19:28 but I understand wanting to do it fancier :) Jun 09 21:20:19 http://xillybus.com/tutorials/device-tree-zynq-1 Jun 09 21:20:22 also found that one Jun 09 21:20:42 and i need to read it... Jun 09 21:20:47 haven't gone through it Jun 09 21:21:04 i'm wondering how different 2.6 kernel is from 3.x Jun 09 21:22:49 there may be all the info i need in that article Jun 09 21:24:35 Is the BBB able to do 1080p video? If not, will it at some point? Jun 09 21:25:10 Prolly not the most common usecase for it, but I haven't got mine right now, since it's at the studio for taking pictures. Jun 09 21:26:23 Thihi: nope, SoC limitation Jun 09 21:27:00 Ok, thanks. Jun 09 21:28:07 I'm haven't been able to interact with my BBB through the local webserver, with the IP given. Is there something else other than installing the drivers that I need to do to be able to interact with it through "http://beaglebone.local/"? Jun 09 21:28:14 so kfoltman did you know the answer to the questiona bout how to make the linaro cross compile for angstrom? Jun 09 21:29:00 the linaro downloadable package for x86 Jun 09 21:29:25 i read someplace that the abi for angtrom and ubuntu were different Jun 09 21:29:27 Rickta59: probably not - but I think I'd at least look for 1) ABI compatibility 2) library version compatibility Jun 09 21:29:57 what does 'file' say about 1) binary that works, 2) binary that doesn't work Jun 09 21:30:11 and the same with ldd - perhaps some shared libs can't be found Jun 09 21:30:14 file both say arm exe Jun 09 21:30:21 like different versions of libc Jun 09 21:30:36 yeah on ubuntu it was libc Jun 09 21:30:56 -sh: ldd: command not found Jun 09 21:31:00 on angstrom Jun 09 21:31:03 maybe it can be opkged Jun 09 21:31:45 right i went down this path Jun 09 21:31:53 wget: res_query.c:251: __libc_res_nquery: Assertion `hp != hp2' failed. Jun 09 21:32:15 and just ended up downloading the angstrom cross compiler Jun 09 21:32:45 mastiff, any reason to treat the BCC as anything other than a piece of RAM out there? Jun 09 21:33:41 ka6sox-away: are you thinking that you'll use the gpmc interface to read/write to registers, kinda like you do on the soc? Jun 09 21:34:09 registers and buffers Jun 09 21:34:13 I think there's definitely some issue here. Basically I cannot get the DT fragment to set 0x8c (which is GPIO2_1) to 0x2f (which is input with no pullup/pulldown). I have no problem with GPIO1. The problem occurs with GPIO2 only. Then, if I try to do the pinmux via the PRU (which has access to the entire address space), and then set OE to 1, I'm convinced that I'm still reading 0, regardless Jun 09 21:34:14 of if the pin is high or low. I am setting the pin (Connector P8, pin 18) high/low with a 4.7k resistor, and a multimeter does show that the pin goes high/low. But reading the port is always low. This is so frustrating, because GPIO1 is good, but there are not enough input pins without using some GPIO2 pins too Jun 09 21:34:28 ssi, registers and buffers Jun 09 21:34:32 does it conflict with soemthing shabaz ? Jun 09 21:34:40 gotcha Jun 09 21:35:34 Rickta59: I could try copying it over from the host system (the ARM version obviously), I'd try it but I can't connect to the damn thing using ssh Jun 09 21:35:53 yeah i just tried that Jun 09 21:36:06 Connecting to feeds.angstrom-distribution.org|140.211.169.179|:80... connected. Jun 09 21:36:06 HTTP request sent, awaiting response... 404 Not Found Jun 09 21:37:00 * tries an okpg update Jun 09 21:37:56 yeah so that works Jun 09 21:38:01 so /bin/busybox Jun 09 21:38:13 it also says libc.so.6 Jun 09 21:38:22 and ld-linux.so.3 Jun 09 21:38:23 It looks like the pin goes to nothing except the header, according to the schematic. I've read through the GPIO section in the 4000-page ref manual, but not sure if there could be another module which overrides to be honest :-( I do think some register may need attention, something that may have already been set for GPIO1 but not for GPIO2, but I can't see what :-( I'll post my device tree Jun 09 21:38:24 fragment on a server, see if anyone can spot if there is an error in it (I think the fragment looks good though, there are only two bytes, i.e. <0x8c 0x2f> and it worked if I use a GPIO1 value. Jun 09 21:39:03 Rickta59: /bin/busybox? huh? Jun 09 21:39:16 yeah figured it would be there Jun 09 21:39:39 you're running ldd on /bin/busybox, right? Jun 09 21:39:49 yes ldd /bin/busybox Jun 09 21:39:49 and it says that it's linked to libc.so.6 and ld-linux.so.3 Jun 09 21:40:02 right this is with angstrom Jun 09 21:40:02 ok, what is your linaro-compiled binary linked to? Jun 09 21:40:40 give me a few to get setup again Jun 09 21:41:26 mranostay: did I miss a ping from you? Jun 09 21:45:58 Rickta59: so, it's linked to ld-linux-armhf Jun 09 21:46:15 http://pastebin.com/sVQxZnHY Jun 09 21:46:46 Rickta59: yes, ldd doesn't work, I've used strings instead Jun 09 21:47:03 sooo Jun 09 21:47:17 arm-linux-gnueabihf-gcc* doesn't work, but arm-linux-gnueabi-gcc* does! Jun 09 21:47:32 from Ubuntu Jun 09 21:48:43 what does it mean, the angstrom libs use softfp calling convention? Jun 09 21:49:09 it puts a wrapper around fp calls Jun 09 21:49:11 _Sy_: here's your answer - use gnueabi, not gnueabihf (both are available in ubuntu) Jun 09 21:49:24 for angstrom Jun 09 21:49:35 wmat: yes. get my mail? Jun 09 21:50:11 did he use an apt-get to install those? Jun 09 21:50:16 AFAIK, yes Jun 09 21:50:20 i was using the ones from the linaro download directory Jun 09 21:50:31 and I was definitely using apt-get Jun 09 21:50:32 granted my ubuntu is oldold Jun 09 21:50:48 well, he just upgraded to 13.04 I think, and I had 13.04 since long ago Jun 09 21:51:13 so that gnueabi is using a less optimized more common libc/libm? Jun 09 21:51:30 more common libeverything :) Jun 09 21:51:31 i've looked at code targetted at neon and it is really compact Jun 09 21:51:34 hf == hard float Jun 09 21:51:45 tried tellign you all that a couple days ago Jun 09 21:51:52 for anstrom hf wotn work Jun 09 21:51:56 hf == passing float values in FPU registers instead of integer registers Jun 09 21:52:16 ah ok well could be that im thinking of armhf versus armel Jun 09 21:52:22 ( debian names ) Jun 09 21:52:22 m_billybob: that might explain why my DSP code is now "fucking slow" instead of just "slow" Jun 09 21:52:24 that is fine for stuff you write Jun 09 21:52:37 on angstrom Jun 09 21:52:41 but the minute you link with a non neon libm .. you lose all that no? Jun 09 21:53:14 Rickta59: I guess you need to use a single ABI everywhere, all libs + executable Jun 09 21:53:23 otherwise it just Won't Work Jun 09 21:53:34 or fail in mysterious and unexplained ways Jun 09 21:53:40 i think that toolchain i pointed him at the angstrom one Jun 09 21:53:50 it is compiled with hf and optimized libs Jun 09 21:53:55 i could be wrong Jun 09 21:54:02 thats what koen told me a few days ago, that libc is not the same and somethign abotu abi issues to contend with Jun 09 21:54:32 Rick, so the binary i compiled for you, that didnt work on ubuntu either ? Jun 09 21:54:43 i forget Jun 09 21:54:46 what did i tell you? Jun 09 21:54:59 think we just talked about angstrom Jun 09 21:55:09 if you bitbake up angstrom you get a new optimized toolchain Jun 09 21:55:18 but thought i saw you say somethign about it wouldnt run on ubuntu to someone else Jun 09 21:55:29 yeah not sure at this point Jun 09 21:55:38 you always seem to catch me when i have angstrom running Jun 09 21:55:39 :) Jun 09 21:55:44 at this point im of the mind "does it really matter?" Jun 09 21:55:47 anyway, need to go play some HoMM3 before I'm at risk of dying from all the derp collected over this weekend :/ Jun 09 21:55:57 nice to know Jun 09 21:56:27 im still tryign ot get a workign netboot going . . . Jun 09 21:56:29 i'd rather do dev in eclipse on my host Jun 09 21:56:38 or atleast networked /root Jun 09 21:56:41 and be able to target both ubuntu and angstrom Jun 09 21:56:48 nfs netboot? Jun 09 21:56:53 or you going with isci now Jun 09 21:56:57 workign on it now Jun 09 21:57:07 iscsi is going ot be a slower more painful project Jun 09 21:57:10 :) Jun 09 21:57:23 bb i just a bit need to go start the coals Jun 09 21:57:31 i may do nfs /root, and iscsi /home Jun 09 21:57:37 nfs implemented in FPGA Jun 09 21:57:58 hmm ? Jun 09 21:58:05 ok, that might be the worst project ever Jun 09 21:58:07 kfoltman for what purpose ? Jun 09 21:58:13 m_billybob: self-torture I suppose Jun 09 21:58:26 you mean like me with my quest for iscsi /root ? Jun 09 21:58:30 heh Jun 09 21:58:31 all the madness of NFS + all the madness of TCP/IP + all the madness of FPGA, all in one :) Jun 09 21:58:46 ah, well toss iscsi in there somewhere too Jun 09 21:58:51 make it a real challenge ;) Jun 09 21:58:56 and LVDS Jun 09 21:59:05 * m_billybob chokes Jun 09 21:59:21 * kfoltman sends m_billybob some adrenaline Jun 09 21:59:37 you know nfs home over the g-ether gadget could be nice Jun 09 21:59:42 err nfs root Jun 09 22:00:24 saw some guys on the groups with settings that indicated they might be doing just that Jun 09 22:00:25 oh nice, my eMMC partition is now full Jun 09 22:00:34 no HoMM3 for me today, it seems :( Jun 09 22:01:06 kfoltman, you dont have a networked /home yet ? Jun 09 22:01:26 no Jun 09 22:01:30 i have workign exact stepsw for both iscsi and nfs . . .just need to write up a blog Jun 09 22:01:55 iscsi is of course more of a pain to setup Jun 09 22:02:02 but a little quicker too Jun 09 22:03:40 anyhow ive got ~45 minutes before all my VM prereq's are met. may as well go outside and get some work done. bbiaw Jun 09 22:12:13 I've quadruple-confirmed! I'm sure there is something weird going on here with GPIO2. I've placed my device tree fragment file here: http://badcafe.co.uk/2013/06/10/device-tree-setting-gpio2_1-as-an-input/ but when I install the DT fragment file, it does not change the pinmux to input. Even if I then try to do it via the PRU, I still read 0 regardless of the input being set high or low. Jun 09 22:23:08 shabaz, pastebin the fragment please Jun 09 22:24:47 shabaz, the dts you are using. Jun 09 22:30:47 Hi ka6sox, it is http://pastebin.com/ZxAztbEm Jun 09 22:37:27 did you look in syslog for errors during install? Jun 09 22:40:50 DFSR = 0x8 Jun 09 22:40:58 * Kristina looks in the armarm Jun 09 22:41:03 01000 Synchronous external abort Jun 09 22:41:13 What ... Jun 09 23:02:49 [19:01] < Kristina> | ah, ok, this is a hardware bug Jun 09 23:02:50 [19:01] < Kristina> | grrr Jun 09 23:02:50 [19:02] < Kristina> | strex is sometimes broken on omap3 Jun 09 23:04:00 S bit + STREX = external abort Jun 09 23:32:05 I've just installed ubuntu-raring 13.04 armhf (using http://rcn-ee.net/deb/flasher/raring/BBB-eMMC-flasher-ubuntu-13.04-2013-05-29.img.xz). The system boot up and everything seems mostly fine Jun 09 23:33:51 what i'm not able to get running is PRU. I've downloaded am335x_pru_package , compiled everything without errors. Then I loaded uio_pruss kernel module and tried the examples, now if fails with "prussdrv_open open failed" - anyone know anything about this? Jun 09 23:50:06 droogie, you should probably ask RCN on his comments page for the post Jun 09 23:50:32 or ask on the beagleboard google groups Jun 09 23:50:47 he'll most likely answer either Jun 10 00:08:21 m_billybob: found a solution, apparently there are many people having problems loading pruss Jun 10 00:11:46 yeah ive noticed that Jun 10 00:11:55 wow its 5PM here already . . . Jun 10 00:12:31 guess times flies when you're out in a 3 acre garden burning down weeds with a weed wacker Jun 10 00:40:03 Im working on using raspberry pis as dmx and audio sources in a haunted house. And im wondering if the black would be better suited for this. I know there's Android expansion boards for DMX output but can the BeagleBone black be set up to output Digital 5.1 audio? Jun 10 00:40:23 *arduino Jun 10 00:43:48 unfluf, you will have to disable some of the onboard resources to do 5.1 audio Jun 10 00:44:39 OH MY GOD I GOT IT TO GET PAST VM INIT Jun 10 00:45:14 Now it crashes where I want it to crash. Jun 10 00:45:19 panic(cpu 0 caller 0x800f8599): "Misconfigured primary clock (RTC frequency is 0)"@/SourceCache/xnu-2050.18.24/osfmk/arm/rtclock.c:141 Jun 10 00:48:32 ssi, ping? Jun 10 00:50:29 Ka6sox, like what? All it has to do is run a python script to generate dmx stream, and run digital audio realativly in sync Jun 10 00:51:04 And a ssh tunnle. Jun 10 00:58:54 unfluf, try posting to the mailinglist/googlegroups/forums about this... Jun 10 00:59:23 so helpfull..... Jun 10 01:04:21 such ca smart ass Jun 10 01:12:43 m_billybob, I'm up to my ears in DTS stuff...didn't want to get into a discussion that I couldn't give my full attention to. Jun 10 01:15:06 does uboot sort out the clock selectors for GPTIMERs? Jun 10 01:15:14 or does the kernel need to do that? Jun 10 01:19:21 ka6sox-away him, not you Jun 10 01:19:25 ka6sox-away: how goes it? Jun 10 01:20:45 mastiff, I'm making progress on the DTS but its not looking like today I'll have it sorted out. Jun 10 01:21:18 reading timings on the Block Ram in the FPGA to set the DT values Jun 10 01:23:30 I will say that if you find the right documentation (in the kernel sources) it does tell you what you need to set. Jun 10 01:25:10 ka6sox-away, besides, his question could have been easily answered by a simple google session. Jun 10 01:25:32 ( yeah im busy too) still fighting non mmc /root Jun 10 01:25:51 mastiff: got one of my BCCs working on the BBB. Jun 10 01:26:15 picked up a heat gun, and refueld with a big plate of nachos, so about ready to do pcb surgery Jun 10 01:26:26 rotate yer FPGA? Jun 10 01:26:28 emeb: nice! Jun 10 01:26:30 yeah Jun 10 01:26:36 good luck! Jun 10 01:26:41 be careful, etc. Jun 10 01:27:22 I think i'll try holding the board by the FPGA with tweezers, and apply heat until the pcb falls. Jun 10 01:27:32 Then clean up, and resolder. Jun 10 01:28:18 Probably need to armor up the tweezers hand with leather glove Jun 10 01:32:04 mastiff: be careful that the pins heat evenly - if some pins hold on then you might rip pads up. Jun 10 01:35:45 Hmm, that's true. Might be better to heat thoroughly with the pcb resting on a surface and lift off the FPGA after I'm sure all sides are free. I do have a junk board to practice on though.. Jun 10 01:40:19 ya Jun 10 01:42:36 pong! Jun 10 01:44:11 ssi: got my BCC working on BBB Jun 10 01:44:14 excellent Jun 10 01:44:25 I started working on an spi transciever in verilog Jun 10 01:44:28 programming spi flash works Jun 10 01:44:35 gooood Jun 10 01:44:48 but BBB behaves differently - prevents FPGA from self-configuring at powerup Jun 10 01:44:55 really? it's working for me Jun 10 01:45:06 BCC configures fine on BBW, but not on BBB Jun 10 01:45:13 er wait Jun 10 01:45:18 it self-configures if I plug the cape into a running board Jun 10 01:45:23 lemme try bouncing the board Jun 10 01:45:32 whoa - you hotplugged it? Jun 10 01:45:47 yea Jun 10 01:45:49 it doesn't mind Jun 10 01:45:51 been doing a lot of that :P Jun 10 01:46:10 you're braver than me. Jun 10 01:46:14 it configures at board powerup too Jun 10 01:46:19 orly? Jun 10 01:46:25 yea safety isn't really my thing Jun 10 01:46:26 yarly Jun 10 01:46:45 doesn't on my BBB, but does on my BBW. Jun 10 01:47:05 thinking that the BBB loads the SPI0_SCLK / CCLK line & keeps FPGA offline. Jun 10 01:47:06 oddi Jun 10 01:47:52 just a theory though - haven't probed to test. Jun 10 01:48:19 hiya ssi Jun 10 01:48:48 Other thing - I redesigned the flash_prog and blinky designs so they don't load the eMMC bus Jun 10 01:48:50 hey Jun 10 01:49:12 setup the entire GPMC bus hookup of the FPGA as inputs. Jun 10 01:49:13 emeb: the board that I've got on my bbb at the moment doesn't have a header populated on P8 Jun 10 01:49:16 for what it's worth :P Jun 10 01:50:15 aha! Jun 10 01:50:16 ssi, 4hrs of reading and I think I have a small handle on what we need to do. Jun 10 01:50:19 a clew Jun 10 01:51:06 achew !? Jun 10 01:51:29 emeb: I'll try the other board, which has all headers pop'd Jun 10 01:51:33 ka6sox-away: ok enlighten me! Jun 10 01:52:16 okay the gpmc is used for MTD devices (we knew that) but they have a one called mtd-ram Jun 10 01:52:51 supposed to be for NVRAM but for our purposes its purfect Jun 10 01:53:08 ssi: good clue - I plugged my BCC into the BBB part-way with only P9 hooked up and it configured fine. Jun 10 01:53:17 pushed P8 in and it hung. Jun 10 01:53:29 interesting Jun 10 01:53:34 38 pins to isolate now :) Jun 10 01:54:06 only 24 active pins on P8 used Jun 10 01:54:16 even better Jun 10 01:55:36 ssi, I know how to assemble the GPMC stuff now, and from the looks of it we are the first GPMC board with DT on BB* Jun 10 01:56:19 general purpose * * ? Jun 10 01:56:36 memory controller Jun 10 01:56:41 ah Jun 10 01:56:53 ka6sox-away: well that's exciting, I'd like to be part of some real contribution Jun 10 01:57:05 I was a bit disappointed to find out that my spi work was largely already solved :/ Jun 10 01:57:32 did you see the comment about how this will help another guy to your PR? Jun 10 01:57:36 yes Jun 10 01:57:42 so thats good. Jun 10 01:57:44 yeah Jun 10 01:58:19 gpmc is currently not built for our kernels as I think they had no way to test it. Jun 10 01:58:30 ssi, do you have ds099? Jun 10 01:58:42 I don't know what that is? Jun 10 01:58:50 (datasheet for the S3a we are using) Jun 10 01:58:56 oh Jun 10 01:59:00 I haven't dug into it yet Jun 10 01:59:08 lemme grab it Jun 10 01:59:16 there is a good section on the Block Ram... Jun 10 01:59:25 including instantiation as a DPram Jun 10 01:59:32 yeah Jun 10 01:59:48 back around the turn of the century I did a bunch of work with an s3 200 and then later an s3e 500 Jun 10 01:59:54 and I did a little bit with block rams Jun 10 01:59:57 but it's been so many ages :P Jun 10 02:00:06 ayup Jun 10 02:00:22 its 2 column so we can create a 16bit DPram Jun 10 02:00:22 OK - here's my theory: FPGA pins 29-31 are hooked to GPMC. Those are "Variant Select" pins during SPI config and will change what kind of SPI flash the FPGA is expecting. Jun 10 02:00:23 ds099 loooks like an s3 datasheet Jun 10 02:00:43 whoops Jun 10 02:00:46 worng one Jun 10 02:00:53 on BBW they float, but on BBB they're being driven by eMMC Jun 10 02:01:04 b00m Jun 10 02:01:09 emeb: that'd do it Jun 10 02:01:21 confusion Jun 10 02:02:02 emeb, see if booting off the sD solves this Jun 10 02:02:31 (for an otherwise stock BBB) Jun 10 02:03:36 ka6sox-away: tried holding down reset and boot buttons during powerup - neither helps. Jun 10 02:06:25 don't see any way around this w/ the current board Jun 10 02:06:50 emeb, want to try my more "radical" approach? Jun 10 02:06:58 whazzat? Jun 10 02:07:13 oh - disable emmc with the jumper. Jun 10 02:07:19 ya Jun 10 02:07:26 which jumper is that? Jun 10 02:07:37 break out your slobbering iron Jun 10 02:07:42 its not too bad Jun 10 02:08:06 output of the G1 to ground (the one that holds the eMMC in reset) Jun 10 02:08:42 are you doing this to an A4 or A5A? Jun 10 02:08:43 what's G1? Jun 10 02:08:49 ka6sox-away: doubt that'll help though Jun 10 02:09:05 if it's in reset it shouldn't drive those pins should it? Jun 10 02:09:07 if they aren't being driven during boot it will help. Jun 10 02:09:08 need to keep the AM33xx off the bus to allow the FPGA to configure. Jun 10 02:09:25 if you boot instead from sD that shouldn't be an issue Jun 10 02:09:51 ssi: that's what I was thinking too. Jun 10 02:10:39 schematic pg3 lower right Jun 10 02:10:49 U3 Jun 10 02:11:02 OpenDrain driver Jun 10 02:11:46 thats the eMMC DIAF pin. Jun 10 02:12:03 lol Jun 10 02:12:10 :) Jun 10 02:12:47 I got on r162 Jun 10 02:13:30 I'm not saying we should tell people this is the permanent fix... Jun 10 02:13:44 (well...mayb on 2ond thought...) Jun 10 02:13:48 j/k Jun 10 02:13:56 you just need to pull pin 4 to ground? Jun 10 02:14:08 or lift R162 and take rstn to ground Jun 10 02:14:55 its open drain so you can just jumper it to ground Jun 10 02:17:00 i guess you could reasonably take it to ground from either side of r162, huh Jun 10 02:17:20 mastiff, one of the best things I invested in was a hot air gun with nozzles for different SMC footprints. Jun 10 02:17:30 but no room for that on the boat. Jun 10 02:17:34 ssi, yes Jun 10 02:17:57 yeah that wouldn't be too bad Jun 10 02:18:01 R162 is on the bottom side Jun 10 02:18:15 solder wick ;) Jun 10 02:18:31 run a small kynar wire up to P8.1 Jun 10 02:18:37 There's room, I have a cabinet for electronics development. Jun 10 02:24:48 First practice chip came up nicely. Jun 10 02:25:12 what are you using for practice chips? Jun 10 02:25:29 Arduinos or Pi? Jun 10 02:25:34 j/k Jun 10 02:25:35 hahaha Jun 10 02:26:11 heading downstairs... Jun 10 02:29:22 Not wasting my arduinos, and wouldn't waste money on a pi. Jun 10 02:29:33 I bought a pi Jun 10 02:29:35 not sure why I bothered :P Jun 10 02:29:37 Set off the damn smoke alarm though. Jun 10 02:30:22 disabled Jun 10 02:30:36 "the ventilator shaft" Jun 10 02:31:20 btw, keeping pressure on while reflowing is bad™ Jun 10 02:31:43 whatever the last pin to reflow is can get all that pulling pressure and you can pull up pad pretty easily Jun 10 02:31:59 just tap it every now and then with tweezers Jun 10 02:32:09 as soon as it moves freely, pull it up with the tweezers Jun 10 02:32:13 Russ, +1 Jun 10 02:32:27 Russ, have you reflowed BGA? Jun 10 02:32:32 yes, with my toaster Jun 10 02:32:45 what kind of yield on 1.0mm parts? Jun 10 02:33:01 if mastiff is talking about a tqfp, all you have to do after lifting it, apply flux, rotate, set it back down, apply heat Jun 10 02:33:05 ka6sox, 100% Jun 10 02:33:18 100TQFP Jun 10 02:33:19 1mm is huge compared to say, a 0.5mm qfn Jun 10 02:33:26 ya Jun 10 02:33:45 ya, I've had to rotate some tqfps with my heat gun, went pretty easy, I think the were 64 pin Jun 10 02:33:51 honestly the hardest parts on the bcc are the stupid little 0.65mm 6 pin DCK guys Jun 10 02:33:56 all the more reason to get the paste printer working Jun 10 02:34:16 that'd be nice, no more stencils Jun 10 02:34:20 ka6sox: I'd rather just build a laser cutter Jun 10 02:34:24 is it like an inkjet? Jun 10 02:34:27 cut my own stencils :P Jun 10 02:34:41 Russ, trying to see if thats possible even Jun 10 02:34:56 I've seen videos of automated paste printers Jun 10 02:35:04 get the mix right so it will flow and be able to be "extruded" Jun 10 02:35:09 stepper controlled syringe with a fine nozzle Jun 10 02:35:17 ya Jun 10 02:35:46 https://www.youtube.com/watch?v=3ThMLs-Cuek like this? Jun 10 02:36:34 Lifted. Pads look intact. Jun 10 02:36:38 excellent Jun 10 02:36:49 any pads that you're likely to lift are not connected anyway :P Jun 10 02:37:01 Small amount of goo out the side of the board, not much. Jun 10 02:37:06 I managed to lift every single NC pad on the port expander on one of my boards :( Jun 10 02:37:10 Russ, yes Jun 10 02:37:33 Russ: yea that's not too bad Jun 10 02:37:50 I have three repraps sitting here, I could convert one to a paste printer if I had a past extruder Jun 10 02:37:55 registration would take some thought Jun 10 02:38:23 ssi, would need a Jig is all Jun 10 02:38:35 yes, but also repraps don't tend to have terribly repeatable x/y homing Jun 10 02:38:38 because they don't need it Jun 10 02:38:45 no feedback loop Jun 10 02:39:05 the lack of feedback isn't the issue, it's the poorly repeatable home switches Jun 10 02:39:23 one of mine has lead screws so I can define a good zero Jun 10 02:39:52 ahh if you watch his second video he has software that compensates for position and angle Jun 10 02:40:25 which ishow the guy at makerfaire with the pick and place did his too Jun 10 02:42:23 ssi, I missed that demo as my son wanted to do something else Jun 10 02:43:25 okay i'm getting sidetracked...I need to finish up this GPMC dts Jun 10 02:43:35 anything I can do to help? Jun 10 02:44:10 need to gather all the pins used and get the names and modes so that I can set the pinmux Jun 10 02:44:31 wanna take the AD's and I'll do the controls? Jun 10 02:44:37 the AD should be bang bang Jun 10 02:45:12 the controls I gotta think about... Jun 10 02:45:16 sorry back up Jun 10 02:45:17 ADs? Jun 10 02:45:29 Address/Data Lines Jun 10 02:45:35 oh ok sry Jun 10 02:45:42 ok I'm looking at the scheme now Jun 10 02:45:48 alright I'll work on it Jun 10 02:45:56 do you have a git repo you're working in? Jun 10 02:46:19 I'll probably push to eric's bcc one Jun 10 02:46:27 keep all the doco there. Jun 10 02:46:54 for right now a gist would be good with those Jun 10 02:47:19 dunno if this helps or not but : pinmux.tking.org Jun 10 02:48:18 where did you guys buy the headers? Digikey was out when I ordered the BOM. Jun 10 02:49:03 mastiff: I had some on hand Jun 10 02:49:15 ka6sox: do you know what we want for pullup/down? Jun 10 02:51:35 uh... Jun 10 02:51:49 ssi, thats a good question Jun 10 02:54:03 I don't think its necessary Jun 10 02:54:20 unless we want to do that for some of the controls till things get setup Jun 10 02:54:32 things like WE Jun 10 02:55:44 emeb_mac, do we have enough i2c expansion to create a OE input to the FPGA? Jun 10 02:56:05 there are a couple free lines on the expander Jun 10 02:56:20 free and wired or just free. Jun 10 02:56:27 just free Jun 10 02:56:32 but could airwire to the expansion header Jun 10 02:56:36 or to an fpga pin Jun 10 02:57:04 for several designs I had to create a pin that essentially was an OE to prevent crashes on the bus. Jun 10 02:57:16 ok I've got pinmux set for AD0-15 Jun 10 02:57:19 set for input, pulldown Jun 10 02:57:25 that can be changed easily Jun 10 02:57:28 and fast? Jun 10 02:57:30 yep Jun 10 02:57:50 you beat me...I'm still looking @ the doco for the various other control pins Jun 10 02:57:56 https://gist.github.com/ianmcmahon/5746255 Jun 10 02:58:01 thanks Jun 10 02:58:44 ka6sox: I think there are a few unused pins on the i2c port expander Jun 10 02:58:50 gpmc might actually be really useful for me Jun 10 02:59:00 I don't think there are any spare FPGA inputs though... Jun 10 02:59:11 lemme check... Jun 10 02:59:29 okay...I should look at the schematic again...I need to see what you do with reset Jun 10 02:59:55 connect VDD to VSS directly? :D **** ENDING LOGGING AT Mon Jun 10 02:59:58 2013