**** BEGIN LOGGING AT Tue Feb 06 03:00:04 2018 Feb 06 05:32:31 it appears the bbb driver signing expired on the 26/1/2018 and won't install on win 10 anymore Feb 06 05:45:50 since what seems to be the demise of capemgr, is there a way to load/remove dtbo's at run time? Feb 06 06:56:29 hii Feb 06 08:26:39 kenrestivo: configfs should still work Feb 06 08:27:19 kenrestivo: I once wrote some scripts to simplify that, you can find them in the bin dir of https://github.com/mvduin/overlay-utils Feb 06 08:28:40 of course there's also still the option of disabling u-boot overlays to go back to the way things were Feb 06 08:29:42 to be fair, I don't think removing overlays at runtime ever worked very reliably? iirc it was an excellent way to cause kernel panics Feb 06 08:30:12 maybe that got better over time Feb 06 09:40:12 Hi all, can anyone point me to a good web ressource which explains how to debug code on the IPU coprocessor while running Linux? Feb 06 09:45:10 you can use jtag debugging on IPU without disturbing the rest of the system Feb 06 09:47:54 when making a debug configuration in CCS, you can select which cores and other debug resources it should attach to Feb 06 09:48:40 right. Feb 06 09:49:23 http://processors.wiki.ti.com/images/b/bf/Ccs5_nondebuggable_devices.JPG Feb 06 09:49:46 that shows a good example, albeit for a different SoC Feb 06 09:50:47 do you have a tutorial at hand. I have a configuration which does not let me connect to the target Feb 06 09:50:48 (Error -1170 @ 0x0) Unable to access the DAP Feb 06 09:52:09 I don't. I also don't have any bbx15 to test on Feb 06 09:52:55 is that the only line of error output? if there's more, please show all of it (using a paste service like pastebin.com) Feb 06 09:54:46 https://pastebin.com/embed_js/mNVU8PT2 Feb 06 09:55:03 so, for reference, the screenshot shows this SoC has two cortex-M3 cores among other stuff (which they've named the "RTOS" core and the "ISS" core). in this example you'd select the core(s) to debug, parent DAP, and optionally icepick. the rest of the checkboxes must be unchecked Feb 06 09:56:12 right, one of CCS's generic "something is wrong, I don't understand it, HALP, *flails arms around wildly*" error messages Feb 06 09:56:50 where typically none of the things it suggests are actually useful Feb 06 09:57:21 is the cortex-M3 currently running your program? Feb 06 09:58:48 the A15 is currently running Linux. Feb 06 09:58:57 one concern comes to mind is that when no program is loaded onto it, linux may be keeping its clock gated, the subsystem reset asserted, or even its power off. ICEPick *should* have the ability to override these but I don't know whether that actually works in the AM572x nor whether it requires any special configuration in CCS Feb 06 10:00:20 these are very good questions. Feb 06 10:00:37 so you may need to have linux load some small dummy program onto the cortex-m4 Feb 06 10:01:02 there are other workarounds obviously, but this would be the simplest to test this theory Feb 06 10:01:31 (I said cortex-m3 earlier, I meant m4 of course) Feb 06 10:01:41 I am trying this Feb 06 10:13:26 I've tried loading the same binary as running on IPU2 and got the following error on dmesg Feb 06 10:13:27 https://pastebin.com/embed_js/a9zfRvSX Feb 06 10:15:00 sorry, I don't know anything about remoteproc really Feb 06 10:15:19 no problem, thanks for your help. Feb 06 10:15:20 43 MB of dma memory? what the hell does it need that for? Feb 06 10:16:00 that sounds rather excessive Feb 06 10:16:04 I think it does video accelaration or something Feb 06 10:16:39 is it just me or is this topic not documented very well??? Feb 06 10:18:01 I don't remember which of the two IPU instances is used for managing video acceleration, but obviously you should use the other ipu instance Feb 06 10:18:23 I have no idea, I've never search for documentation on this topic so I can't tell you Feb 06 10:18:30 *searched Feb 06 10:19:04 the IPU1 is for general purpose Feb 06 10:21:42 so how is the video acceleration thing relevant here? Feb 06 10:23:30 it is not, I just want to connect to the IPU1 via JTAG Feb 06 10:25:26 and I tried to load a binary to the IPU1 and just had the binary of the IPU2 at hand. So I copied it Feb 06 10:29:07 but that binary will request a huge amount of memory which is likely not reserved for use with the IPU1 ;) Feb 06 10:29:53 that indeed sounds like a very bad idea Feb 06 10:30:15 also because if it were to run, it would probably make a mess of whatever ipu2 is managing Feb 06 10:31:43 hmm, right. Feb 06 12:39:52 I am having issues to connect to my BBB wifi Feb 06 12:40:17 Please kindly provide with support, Thank you Feb 06 12:44:37 Hello Feb 06 14:13:08 I just installed CCS 6.2.0 and everything works out of the box :D Feb 06 14:13:33 :-) Feb 06 14:14:10 Don't know why 7.4 is not working Feb 06 14:15:15 BeagleBone Black? Feb 06 14:15:19 JTAG or serial? Feb 06 14:16:32 BeagleBoard X15 with JTAG (XDS100V3) Feb 06 14:16:49 ah. Feb 06 14:16:57 very much should work out-of-box then. Feb 06 14:17:27 can you share any details of where it fails? Feb 06 14:17:40 same configuration steps? Feb 06 14:17:43 i've been fighting with this beast for about 4 days. Feb 06 14:18:04 * jkridner[m] is trying to gather enough info for a big report. Feb 06 14:18:18 yuck Feb 06 14:19:30 it would not let me access memory for releasing the M4 from reset. There is a thread https://e2e.ti.com/support/development_tools/code_composer_studio/f/81/p/661696/2433037#2433037 Feb 06 14:21:19 i wonder if the unit got changed to only expose one of the M4s. Feb 06 14:21:29 s/unit/init/ Feb 06 14:21:56 i know there was talk that developement was only going to be supported on one of them... Feb 06 14:22:12 and the other was to be reserved for TI codecs. Feb 06 14:29:16 yes there was. The second M4 is only for video acceleration, but I used only the first one Feb 06 14:58:09 jkridner[m]: one of the dual-M4 subsystems you mean Feb 06 14:58:23 there are IPUs, each with two M4 cores Feb 06 14:58:26 *two IPUs Feb 06 14:59:29 jkridner[m]: still seems silly to fully reserve it when not everyone might care about TI codec support Feb 06 15:00:24 darkdust: a lot of what I'm reading there sounds like a bad idea Feb 06 15:01:09 you're running linux on the A15. you should not connect to the A15 with the debugger, run the "EnableAllCores" init script, or mess with the pmic Feb 06 15:01:30 I think I have fried my AM338 Processor in BBB. Is it possible that, if allowed more than 1.8V in analog input pin without the proper divider, might fry the whole processor chip ? Feb 06 15:01:43 any wisdom here ? Feb 06 15:02:06 Guest31378: overvoltage can most definitely kill the entire processor Feb 06 15:02:43 what are the symptoms? power leds remains off or just flashes briefly? Feb 06 15:03:01 zmatt: what sounds like a bad idea? Feb 06 15:03:30 darkdust: the three things I just mentioned Feb 06 15:03:59 zmatt: only one of the leds glows and turns off immeditely when powered the board Feb 06 15:04:19 Guest31378: presumably you mean the power led Feb 06 15:04:38 yup. thats the power led Feb 06 15:04:44 I am not connecting to the A15 anymore. Just connecting to the IPU1 Feb 06 15:04:48 yep, that's the pmic shutting down due to overcurrent, caused by an internal short-circuit in the processor Feb 06 15:04:59 darkdust: good :) I was just commenting on what I read there Feb 06 15:05:48 Yes, I actually tried to connect to the A15 but I did not know that the BeagleBoard X15 ships with a linux preinstalled. Feb 06 15:06:11 thanks. will try if some tinkering with pmic helps or not Feb 06 15:06:27 ??? Feb 06 15:07:48 well that was weird Feb 06 15:09:15 if the processor has an internal short-circuit what should tinkering with the pmic help??? Feb 06 15:09:44 ¯\_(ツ)_/¯ Feb 06 15:10:40 we'll see if he comes back in a few Feb 06 15:58:23 sorry, boarded a plane. Feb 06 15:59:27 did we work out that the scan chain was the X15 M4 issue that darkdust is having? Feb 06 16:02:20 zmatt: I will need to confirm, but I think the theory is to simply not market it as available so that it is always free for running the codecs in the case it is needed, but I sympathize with that being over simplistic. Feb 06 16:02:52 going to “airplane mode” Feb 06 16:24:09 zmatt: here's the same test run with the i2c read in a while(1) loop in C Feb 06 16:24:12 https://imagebin.ca/v/3qn2H7W8vE0e Feb 06 16:26:24 with a usleep(1000) Feb 06 16:34:02 jkridner: Now I am confused. The CCS 6.2 was able to connect to the IPU1 and the 7.3 wasn't. Then I installed the RTOS SDK on the 6.2 and it started throwing errors. Feb 06 16:34:52 I've also tested with the 7.3 with RTOS SDK and without - same effect Feb 06 17:24:37 zmatt: I didn’t know there were 4 M4s. I thought there were only 2. It must have been “fixed” in collateral pretty early then. Feb 06 17:58:13 hey zmatt, think you looked into this - on the bone, do you know if SYS_5V is sufficient to key off to sequence things? Feb 06 18:16:37 darkdust: what kind of errors? Feb 06 18:54:30 jkridner: "Unable to access the DAP." Feb 06 19:33:07 ds2: what do you mean? Feb 06 19:40:05 ds2: for power domain related yada? no, sys_5v is the very first thing that's up, so it's not a safe cue for driving stuff into the am335x Feb 06 19:42:50 (and when powering down it is, idiotically, also the first supply to get cut, although it should hopefully coast on caps long enough to sustain the rest; except when battery-powered, then sys_5v is always-on) Feb 06 19:43:29 anyway, bbl Feb 06 22:13:51 Hey, I'm looking for some advice for unbricking an original beagleboard Feb 06 22:59:28 did you screw with the eeprom? Feb 06 23:00:26 original beagleboard had no eeprom. just NAND. Feb 06 23:00:44 beagleboard != beaglebone Feb 06 23:01:11 Still, unbricking for either is straightforward Feb 06 23:01:22 i was thinking x15, sorry, I haven't used any of the others. Feb 06 23:02:12 ah. that one does indeed have EEPROM and eMMC flash. Feb 06 23:12:50 the recommended unbricking procedure for all boards is to use a high-pressure pump to reinsert the magic smoke Feb 06 23:13:29 (if original magic smoke has been lost, be sure to use high-quality replacement) Feb 07 00:07:38 synthentic magic smoke Feb 07 00:27:25 Anyone know why Beaglebone Black's memory is 512mB but schematic says 4Gig DDR3? (not flash) Feb 07 00:41:06 I do, but you left, so you won't Feb 07 00:41:52 (well I guess you'll probably eventually discover ram is sized in bits rather than in bytes) **** ENDING LOGGING AT Wed Feb 07 03:00:02 2018