**** BEGIN LOGGING AT Fri Jul 19 02:59:58 2013 Jul 19 08:30:51 * vvu can't find the coffee. this is a cursed day Jul 19 08:32:30 uh oh Jul 19 08:33:25 not really nice Jul 19 08:34:52 so lucky having 2 BBB, building on one running on the other Jul 19 08:35:21 you are missing a performance enhancing drug Jul 19 08:35:47 anyway i need to compile a 70 liner program with no external libs Jul 19 08:38:52 * vvu is reminded to purchase a case for the beagle. desk is full of electric conductibility materials Jul 19 10:25:39 vvu: altoid tins are better :) Jul 19 10:33:18 true but need to hack em to fit a BBB and the connectors Jul 19 11:25:28 yeah .. i use it just while taking it along someplace.. lego case would be cool too Jul 19 12:33:45 Panto, ping Jul 19 12:33:52 pong Jul 19 12:33:56 Hi Jul 19 12:35:08 I've built your kernel but an having trouble combing the uImage with bbb .dts, maybe something is missing from your repo that I need? Jul 19 12:35:40 no, nothing is missing Jul 19 12:35:54 make uImage-dtb.am335x-boneblack works Jul 19 12:36:08 make dtbs create the base dtb Jul 19 12:36:37 I think I didn't do that step Jul 19 12:37:10 mluckham-mobile, I use this: make -j 12 && make LOADADDR=0x80008000 uImage && make dtbs && make LOADADDR=0x80008000 uImage-dtb.am335x-bone && make LOADADDR=0x80008000 u Jul 19 12:37:10 Image-dtb.am335x-boneblack && make INSTALL_MOD_PATH=${PWD}/modinst modules_install firmware_install Jul 19 12:38:13 Tx Jul 19 12:38:22 I'll try it shortly Jul 19 12:38:54 What is -j 12? Jul 19 12:39:37 mluckham-mobile: number of simultaneousjobs Jul 19 12:39:44 Ah Jul 19 12:40:11 Great for when you have a machine with 64 cores sitting around. Jul 19 12:40:33 jj2baile, I only have 8 cores Jul 19 12:40:46 rule of thumb is core * 3/2 Jul 19 12:41:10 I tend to do cores*3/4 Jul 19 12:41:37 perhaps i just arbitrarily chose that factor at some point Jul 19 12:45:44 Panto, what is the difference between your kernel and stock as related to vring-pru communication? Jul 19 12:46:20 the standard kernel assumes a coherent memory interface like that on a normal smp system Jul 19 12:46:29 we don't have that Jul 19 12:46:39 coherency must be enforced manually Jul 19 12:48:15 Which means, pru can't access non-pru memory directly? Jul 19 12:48:49 nope Jul 19 12:49:22 it means that when it accesses it along with ARM, there must be proper cache synchronization Jul 19 12:49:52 so depending on the direction of the transfer to/from ARM, the cache must be flushed/invalidated accordingly Jul 19 12:51:30 I'll look for that in your code, but how much could that Jul 19 12:51:56 Impact performance/throughput? Jul 19 12:52:13 there's no other way Jul 19 12:52:32 Another question Jul 19 12:52:43 without it you'll just get corrupted data Jul 19 12:53:28 Are vrings bidirectional or will our open ocd implementation use two vrings? Jul 19 12:54:12 they are uni-directional Jul 19 12:54:39 there is a twist about buffer management, but you don't care about that from the PRU side Jul 19 12:55:15 So there will need to be a way to synchronize communication Jul 19 12:55:48 When talking to the pru/cpld and getting responses Jul 19 12:57:23 that's the job of the pru Jul 19 12:57:35 afk for 10 Jul 19 12:57:58 Thanks for your help Jul 19 13:07:49 np Jul 19 13:07:50 back Jul 19 13:34:40 panto: have any experience with serial communication? Jul 19 13:35:19 you mean rs232? Jul 19 13:35:21 yep Jul 19 13:35:35 i have some trouble with my read function Jul 19 13:35:41 with the read function Jul 19 13:35:51 what kind of trouble? Jul 19 13:36:29 it reads even if nothing is sent Jul 19 13:36:34 and it reads nothing Jul 19 13:36:47 shouldn't it wait until something is on the line ? Jul 19 13:36:50 what are the settings you've used to open the serial port? Jul 19 13:37:03 is the baud rate correct? Jul 19 13:37:22 i have android on ther other end Jul 19 13:37:26 using g_serial Jul 19 13:37:28 on the BBB Jul 19 13:37:29 do you ask for breaks to be delivered? Jul 19 13:37:38 vvu, it doesn't matter Jul 19 13:37:58 and how can i simulate baud rate on my android device i have only bulk transfer of data to usb device Jul 19 13:38:15 i am sending raw data, no breaks no nothing from android to BBB Jul 19 13:38:52 are you sure the serial gadget works correctly? Jul 19 13:39:15 yep Jul 19 13:39:20 http://pastebin.com/AvrPWp2d Jul 19 13:39:23 this are my settings Jul 19 13:39:33 if i do a while(1) read();prinf() Jul 19 13:39:39 i get the data Jul 19 13:40:36 err, I think you've opened the device in a mode to return even if no data received Jul 19 13:40:50 that VTIME = 0 doesn't look good Jul 19 13:41:19 commented it out a few moments ago and same behaviour] Jul 19 13:41:57 DTR, RTS can influence it ? Jul 19 13:42:19 are you calling cfmakeraw? Jul 19 13:42:31 where did you get that settings you use? Jul 19 13:42:42 tcsetattr Jul 19 13:42:54 no, tcsetattr applies the settings Jul 19 13:43:09 there is a family of posix tty control functions that operate on the setting structure Jul 19 13:43:12 mhmh i found an rs232 lib Jul 19 13:43:43 panto: http://www.teuniz.net/RS-232/ Jul 19 13:44:25 cfmakeraw(&tio_new); Jul 19 13:44:26 tio_new.c_cc[VMIN] = 1; /* block until n bytes are received */ Jul 19 13:44:41 that's part of some of my own serial control code I keep Jul 19 13:45:00 VMIN=0 means return immediately (even when 0 bytes received) Jul 19 13:45:12 thx :) Jul 19 13:45:25 worked Jul 19 13:46:25 ok Jul 19 13:50:31 panto: can this work only for one read then behave the same ? Jul 19 13:54:16 nvmind got it working Jul 19 15:53:14 Panto, I'm getting a build error Jul 19 15:53:34 with my tree and config? Jul 19 15:53:53 No rule to make target uImage-dtb.am335x-bone Jul 19 15:54:39 what is the kernel you're using? Jul 19 15:54:43 I used your instructions from yesterday to build .config from defconfig Jul 19 15:54:56 are you using the correct branch? Jul 19 15:55:27 the master branch is vanilla mainline Jul 19 15:55:40 if it's my tree, the correct branch is the latest not-capebus-v33 Jul 19 15:56:23 Good question, git://github.com/pantoniou/linux_bbxm.git Jul 19 15:56:50 yes, but it's not the master branch the one you have to use Jul 19 15:58:19 Ok Jul 19 18:33:23 Panto, uImages build now and boot up on bbb Jul 19 18:34:50 mluckham-mobile, good Jul 19 18:35:59 Has been a long time since using command line, never used git at all (only svn) Jul 19 18:36:47 Now I am trying pru example_app on bbb Jul 19 18:43:39 ok Jul 19 18:43:48 let me know how it goess Jul 19 18:43:51 *goes Jul 19 18:43:57 have to take off, it's friday night :) Jul 19 18:45:41 Have fun :) Jul 19 19:00:57 hiya mike Jul 19 19:12:44 Hi ka6sox Jul 19 19:13:19 I built panto's kernel and it works on bbb Jul 19 19:13:29 excellent Jul 19 19:13:37 got instructions someplace? Jul 19 19:13:39 :) Jul 19 19:14:09 Yes, just a sec Jul 19 19:14:14 :) Jul 19 19:17:10 Http://pastebin.com/0sigzjEh Jul 19 19:19:38 can we publish this in elinux.org? Jul 19 19:19:57 (and mod when it goes into the standard kernel)? Jul 19 19:20:24 Is a bit of a mess and work in progress but I don't see why not Jul 19 19:21:26 what toolchain did you setup and where? Jul 19 19:21:31 Modprobe uio_pruss didn't work for me Jul 19 19:22:45 Https://launchpad.net/gcc-arm-embedded Jul 19 19:23:10 interesting Jul 19 19:23:21 as this is a Cortex-A8 Jul 19 19:23:48 That is on my ubuntu system Jul 19 19:24:07 right, but the target is Cortex-A8 Jul 19 19:24:16 Pruss examples were built on bbb Jul 19 19:24:31 and this toolchain is for Cortex-M and Cortex-R Jul 19 19:25:07 Pre-built GNU toolchain from ARM Cortex-M & Cortex-R processors (Cortex-M0/M0+/M3/M4, Cortex-R4/R5/R7). Jul 19 19:25:52 I always specify ARCH=arm CROSS_COMPILE=arm-none-eabi- Jul 19 19:27:11 So why does it work? Or does it .. Jul 19 19:27:23 it will compile Jul 19 19:27:29 however its not for the right target Jul 19 19:27:30 http://elinux.org/Toolchains#Linaro_.28ARM.29 Jul 19 19:27:52 The kernel runs on bbb ... Jul 19 19:28:15 But if it is wrong I'll get another Jul 19 19:29:03 Learning lots of stuff fast, too fast to grok in fullness Jul 19 19:29:03 interesting Jul 19 19:29:43 I don't know enough about the cortex-* line to comment other than I'm sure there are missing registers. Jul 19 19:31:14 I had code sorcery first, didn't trust the way it installed (for stellaris launchpad) Jul 19 19:31:38 the Linaro ARM toolchains seem to work. Jul 19 19:31:51 I'll get it then **** ENDING LOGGING AT Sat Jul 20 02:59:57 2013