**** BEGIN LOGGING AT Sun Sep 13 02:59:58 2015 Sep 13 03:03:37 I think it takes only a few minutes Sep 13 03:03:46 definitely not 4 hours Sep 13 03:05:01 what are the leds doing Sep 13 03:08:53 They are just flashing like crazy which makes me think it is still going but it is taking a REALLY long time Sep 13 03:10:04 if I understand correctly (I never used it) the flasher shows a simple sequential pattern of leds (knightrider-style) Sep 13 03:10:43 if led 1 is blinking a heartbeat pattern and the other leds are just doing "stuff" then you have a system running normally Sep 13 03:10:49 *led 0 Sep 13 03:12:39 Well I lost the ability to ssh into it. So I have no idea Sep 13 03:16:24 well that doesn't mean you can't look at the leds and describe what they're doing a bit more specific than "flashing like crazy" Sep 13 03:16:35 also, which download did you use exactly? Sep 13 03:19:28 The latest debian file. led 0 doing a double flash every second or so and led 2 is just rapidly flashing Sep 13 03:21:28 "latest debian file" is not an answer "which download did you use exactly", and the led pattern matches a running system Sep 13 03:21:39 led 0: "heartbeat", led 2: cpu activity Sep 13 03:22:28 http://beagleboard.org/latest-images first and only link under Recommended Debian Image Sep 13 03:24:09 ok, bone-debian-7.8-lxde-4gb-armhf-2015-03-01-4gb Sep 13 03:24:12 that's not a flasher Sep 13 03:24:28 they actually say "For testing, flasher and other Debian images, see ..." Sep 13 03:25:45 then what is that file and what is its purpose? Sep 13 03:26:08 it's a working system that boots directly from SD Sep 13 03:26:15 (no idea why you can't ssh to it) Sep 13 03:27:18 note btw that I'd personally recommend a testing image. debian 7.8 is pretty old Sep 13 03:29:08 honestly I have no idea if I even need to do this. I was just trying to fix issues where apt-get update fails to retrieve package lists and issues where the adc and pwm cant be used together or the adc files lock. I was told to update the beaglebone Sep 13 03:29:49 there's one known issue where apt-get update gives a harmless warning, I don't know if that's the issue you're referring to Sep 13 03:30:09 in any case, development is still pretty fast so lots of stuff changes and gets fixed all the time Sep 13 03:30:32 you can find the latest images here: https://rcn-ee.com/rootfs/bb.org/testing/ Sep 13 03:31:37 no my issue is that apt-get update does nothing because it cant get any of the update packages. Sep 13 03:31:51 but thanks for the link. Sep 13 03:32:17 it seems rcn-ee now also includes usb flashers, nice Sep 13 03:33:20 hum, maybe that means not quite what I think it means, not sure Sep 13 03:34:45 hmm, and rcn-ee seems to have switched to the -ti branch of kernels by default rather than the -bone kernels Sep 13 04:39:01 Why is apache installed by default on the rnc-ee images? I'd really love an option for a completely minimal installation. Sep 13 04:39:37 then get a "console" image Sep 13 04:39:51 https://rcn-ee.com/rootfs/bb.org/testing/2015-09-06/console/ Sep 13 04:40:06 Yep, thats the one I pulled down and it still had apache installed and running by default. Sep 13 04:40:18 It mentions it in the wiki as well, which I missed the first time. Sep 13 04:40:39 huh, I'm quite certain apache wasn't included last time I checked Sep 13 04:41:15 I pulled the BBB-eMMC-flasher-debian-8.1-console-armhf-2015-08-17-2gb.img.xz image Sep 13 04:45:09 I'll pull a new image and load it. I'd just like not to have to remove apache Sep 13 04:45:41 I've so far only used the bone-debian-8.1-console images (flashed via USB) Sep 13 04:46:17 How would I flash via usb? Sep 13 04:46:36 Should I expect a device to show up in dmesg? Sep 13 04:47:19 And the bone- images are base images? Sep 13 04:48:16 bone-console is a minimal image for the BBB Sep 13 04:48:41 So I shoudl be able to dd the image directly to the beagle? Sep 13 04:49:04 Do I boot it first, hold the boot button, how? Sep 13 04:49:39 Also, why so many images? Sep 13 04:50:03 yes, though I'd recommend shrinking the ext4 partition to minimal size first, then after flashing recreate the partition (same offset, max size) and expand the image again Sep 13 04:50:56 for USB flashing I use BBBlfs... its scripts are broken, but the "usb_flasher" tool itself works and makes the BBB appear via USB as mass-storage device Sep 13 04:51:33 Why is that method preferable to the flasher? Sep 13 04:52:13 more direct, no need to prepare an sd card Sep 13 04:54:06 The results should be identical though, correct? Sep 13 04:58:16 Also, is there a way to use the SD card as storage? Sep 13 04:59:30 (and shrinking the fs before flashing saves a lot of time: you only have to flash the space actually used) Sep 13 04:59:35 sure Sep 13 05:00:07 some people even run entirely from SD (using the eMMC interferes with some CAPEs) Sep 13 05:02:29 I just checked the latest console image -- no apache is isntalled Sep 13 05:02:47 Cool, I'll try out your method of install tomorrow Sep 13 05:03:06 Sounds somewhat complicated Sep 13 05:16:49 I should probably just fix the BBBlfs scripts and send those as patch-request Sep 13 05:17:42 cool, using chroot and qemu-arm-static you can actually "enter" the image Sep 13 05:17:47 yes please :) Sep 13 12:02:50 why does the beaglebone images not use upstream debian kernels? Sep 13 12:30:50 <_av500_> patches needed I guess Sep 13 12:50:36 it appears to be some ti-patchset. however arch/arm/boot/dts/am335x-boneblack.dts exists in upstream kernel. if it is just non essential stuff (like graphics) i'd rather use the upstream version Sep 13 13:48:18 @/b 2 Sep 13 14:54:14 nks: there are quite a bit more changes than that afaik Sep 13 14:54:38 but feel free to build your own kernel Sep 13 15:02:22 nks: I think most of the changes do eventually land upstream, but for example 3.8.13-bone kernels have DT overlays, which didn't make it to mainstream until.. I think 4.1 ? Sep 13 15:11:30 nks: https://github.com/RobertCNelson/linux-stable-rcn-ee/releases contains the linux trees for various releases, you can compare them to see what's different Sep 13 15:17:04 for example, diff between 4.1.6 and 4.1.6-ti-r15 Sep 13 15:17:05 https://github.com/RobertCNelson/linux-stable-rcn-ee/compare/4ff62ca06c0c0b084f585f7a2cfcf832b21d94fc...d89bc272f41c6e0c335a68b2caa1d82fd8fc5ef5 Sep 13 15:17:43 clearly not all patches are beaglebone-related but also stuff for the beagleboard-x15 and such, which makes sense Sep 13 15:18:30 the -bone kernels are closer to mainline: Sep 13 15:18:35 https://github.com/RobertCNelson/linux-stable-rcn-ee/compare/4ff62ca06c0c0b084f585f7a2cfcf832b21d94fc...c9aa25bc26d25afc8a211a9dcf13ed6e2ba2058e Sep 13 15:25:54 and 4.2-bone5 is closer still to mainline 4.2 Sep 13 15:25:57 https://github.com/RobertCNelson/linux-stable-rcn-ee/compare/v4.2...e285ab70171a1dca6f9fd7d319c9d275d76b7dbd Sep 13 16:55:59 Hello Sep 13 16:56:15 I am trying to set up cross compile on my Beaglebone Black Sep 13 16:56:33 gcc-linaro-arm-linux-gnueabihf-4.8 on my Ubuntu machine Sep 13 16:56:44 and gcc-linaro-arm-linux-gnueabihf-4.6 on my Beaglebone Running Debian Sep 13 16:56:46 Is that okay? Sep 13 16:57:05 I am not able to do with via eclipse. Sep 13 19:55:50 zmatt: thanks Sep 13 19:57:43 nks: it's worth mentioning that rcn seems to have switched from the -bone to the -ti kernel as default in recent images, so presumably they do have added value Sep 13 20:04:47 * tbr prefers the vanilla kernels, run well on the BBB Sep 13 20:07:16 I think RN's have a tendency to get 'bloaty' as he enables a lot of features by default Sep 13 20:09:26 yeah, which makes sense to get stuff working "out of the box" for new users Sep 13 20:09:45 I personally do recompile with a more stripped-down config Sep 13 20:11:15 I suspect you'll definitely want a -ti kernel if power management is important Sep 14 00:57:11 zmatt: hey, you there? Sep 14 00:57:47 last night you said one of the problems with the PMIC circuit was "cpu pin "VDDS" is connected to LDO1 instead of LDO3" Sep 14 00:57:56 why should it be LDO3? Sep 14 00:58:03 zmatt: ^^^^ Sep 14 01:01:13 or anyone else.. Sep 14 01:01:37 na zmatt is the pmic/vdd3v3 dude Sep 14 01:03:08 the known resident expert? Sep 14 01:03:52 mhmm Sep 14 01:04:10 think he's got a page about it on elinux if you want to look ... Sep 14 01:04:54 .. http://elinux.org/BeagleBone_Power_Management .. Sep 14 01:11:33 thanks veremit Sep 14 01:11:40 np :) Sep 14 01:12:13 do you know anything about the ddr3 "leveling program"? Sep 14 01:12:33 namely, what you do with the information it provides? Sep 14 01:13:12 nm. Sep 14 01:13:19 i have a link to the info already... Sep 14 01:15:19 no ^_^ sorry lol Sep 14 01:16:30 zmatt dumped so much data on my last night i'm gasping for air... Sep 14 01:16:35 s/my/me/ Sep 14 01:16:55 http://processors.wiki.ti.com/index.php/AM335x_DDR_PHY_register_configuration_for_DDR3_using_Software_Leveling Sep 14 01:17:36 heh yea Sep 14 01:18:21 yates: the leveling data is passed to the ram init function somewhere in board.c in uboot Sep 14 01:18:41 note that the morons put the actual values in constants named after the RAM chip used Sep 14 01:18:48 except the values are dependent on the pcb Sep 14 01:20:31 see for example https://www.google.com/search?q=MT41K256M16HA125E_RD_DQS Sep 14 01:20:53 *facepalm* Sep 14 01:21:11 (psst: hey matt, some of those morons r us!) Sep 14 01:22:27 the fact that you need to run a program on the target board to determine the values rather then being able to fish them out of the RAM datasheet should have been a hint ;P Sep 14 01:22:48 moreso given that that program requires you to give approximate pcb trace lengths as seed values Sep 14 01:23:21 did you see my LDO question? Sep 14 01:23:23 yes Sep 14 01:23:40 is that covered in one of the links you gave? Sep 14 01:23:43 we woke him up .. lol :P~ Sep 14 01:23:49 yes it is -> https://e2e.ti.com/support/arm/sitara_arm/f/791/t/421756 Sep 14 01:23:50 : Sep 14 01:24:16 except as background info you may need to know that LDO1 is the only LDO enabled in RTC-only sleep Sep 14 01:24:21 *RTC-only mode Sep 14 01:24:51 if vdds is kept powered it leaks like crazy to other power rails Sep 14 01:25:02 when you talk about "RTC-only" Sep 14 01:25:04 mode Sep 14 01:25:06 ... Sep 14 01:25:14 are you talking about the RTC in the AM335x? Sep 14 01:25:15 that's the mode in which only the am335x RTC is powered Sep 14 01:25:17 yes Sep 14 01:25:26 is there one in the TPS65217 too? Sep 14 01:25:35 no Sep 14 01:25:51 ok, right. great! thanks!!! Sep 14 01:25:56 tbh the TPS65217 kinda sucks Sep 14 01:26:17 but it's got the built-in li-ion charger that's semi-convenient Sep 14 01:26:17 did they fix any of this in the TPS65217D? Sep 14 01:26:30 D isn't a revision code, it's a configuration code Sep 14 01:26:52 it isn't the same chip, revised? Sep 14 01:26:56 :0 Sep 14 01:27:15 no, A/B/C/D are different configurations of voltage levels and sequencer timings Sep 14 01:27:21 there are revisions of the chip though Sep 14 01:27:33 but dunno if that's clearly marked anywhere Sep 14 01:28:00 nor is there a revision log available anywhere, so no idea what they changed Sep 14 01:28:04 well you recommended using the D last night, right? Sep 14 01:28:19 well it doesn't matter hugely, C or D Sep 14 01:28:36 C defaults to ddr3 voltage while D defaults to ddr3l voltage Sep 14 01:29:03 what is the "l" in ddr3l? Sep 14 01:29:09 low-voltage / low-power Sep 14 01:29:19 ah. Sep 14 01:29:19 the RAM chips on the BBB are compatible with either Sep 14 01:29:27 as long as the voltage has stabilized before the RAM is released from reset Sep 14 01:29:34 so it's only that, not the bugs, that makes the d very slightly better. Sep 14 01:30:05 so u-boot simply reconfigures the TPS to ddr3l voltage levels before ram is initialized Sep 14 01:30:21 if you're using the same RAM chip then yes Sep 14 01:30:37 yes, we are Sep 14 01:30:53 if you're using a RAM chip that only supports ddr or only supports ddr3l, then you'd specifically need the TPS65217C or TPS65217D respectively Sep 14 01:31:11 (B is DDR2) Sep 14 01:31:12 we're using whatever is in the bbb rev. c Sep 14 01:31:47 yeah then it doesn't matter, you can pick either C or D, whichever is cheaper / in stock or whatever Sep 14 01:32:42 I'd lean to D given the choice given that the general direction in voltage levels is that they're getting lower and lower Sep 14 01:33:29 (or well, converging to ~1V ... there are apparently some fundamental problems with going even lower) Sep 14 01:36:04 note that having a usable RTC-only mode (VDDS on LDO3) requires ignoring TI's advice (VDDS on LDO1) Sep 14 01:37:14 but during my many many measurements I found no indication that VDDS on LDO3 causes any problems, while VDDS on LDO1 did show unpleasant leakage during powerdown sequencing Sep 14 01:37:43 and I haven't been able to get any further comment from TI on this Sep 14 01:38:22 gotcha Sep 14 01:39:11 I suspect the weird leakage on VDDS would also disappear if you'd use 1.8V I/O instead of 3.3V I/O Sep 14 01:40:00 stupid question: if LD01 is the only output providing during RTC-only, why would RTC-only work with LDO3? Sep 14 01:40:18 VDDS is not supposed to be powered during RTC-only mode Sep 14 01:41:57 doh! Sep 14 01:41:58 VDDS_RTC is the only supply of the am335x powered in RTC-only mode Sep 14 01:42:11 why are people so interested in the internal RTC? Sep 14 01:42:20 low-power mode Sep 14 01:42:26 super-low-power, even! Sep 14 01:42:33 the internal RTC of the AM335x is not particularly low power Sep 14 01:42:45 how does that relate to the internal RTC? Sep 14 01:42:48 I mean, RTC-only mode is of course much lower power than something like suspend-to-RAM Sep 14 01:43:00 relative to the other AM335x low power modes, it's WAY less... Sep 14 01:43:15 but you can find external RTCs that run on a fraction of a microamp Sep 14 01:43:44 a self contained RTC with its own CR2032 or one of those self contained Maxim (old DS) parts Sep 14 01:44:00 that should put 0 drain on the main battery Sep 14 01:44:41 the rtc isn't the real power issue with our platform, and the internal RTC is easier than plopping down another chip. Sep 14 01:45:14 provided you don't mind fixing the power supply scheme bugs of the bbb Sep 14 01:45:36 how small are you making the board? Sep 14 01:45:49 RTCs come in pretty small packages Sep 14 01:46:09 not real small, size isn't the issue. Sep 14 01:46:32 assembly costs with extra BOM items? Sep 14 01:47:06 it's a little historical. we are 99.98 percent routed. had we known this 2 months we may have chosen an external rtc Sep 14 01:47:33 nothing I2C related on the remaining 0.02%? Sep 14 01:47:33 2 months agao Sep 14 01:48:09 how many layers? Sep 14 01:49:16 yates: ah, so this is probably also not a good time to mention that unless you specifically need tps65217 features the tps65910a3 might be a more attractive pmic? Sep 14 01:49:58 zmatt: :) Sep 14 01:50:24 i'm afraid they're gonna hate me for giving them all the bad news you've had the kindness and patience to relay to me the last 24 hours! Sep 14 01:50:58 come now, enough! I wanna get this report done and go watch me "The Wrecking Crew" movie! Larry Knechtel - YEAH!!! Sep 14 01:51:06 s/me/my/ Sep 14 01:51:09 yates: don't worry, we have plenty of expensive mistakes lying around the office too ;) Sep 14 01:51:17 DOH!! Sep 14 01:51:29 it's only time/money, right? Sep 14 01:51:36 a 14-layer pcb based on the DM814x being one of them (designed right before TI abandoned support for it) Sep 14 01:51:46 assholes Sep 14 01:51:54 fallout of OMAPgeddon Sep 14 01:52:04 i think ours is 8 layers, fyi, ds2 Sep 14 01:52:33 it's really not up to me, anyway guys. i'm just the peon sw dude. Sep 14 01:52:59 ah, the "well, this is how the hw turned out, go and find sw workarounds" guy ? Sep 14 01:53:14 ;) Sep 14 01:53:22 i used to do hw, a long time ago in a universe far, far away... Sep 14 01:53:41 i was great at wire-wrap... Sep 14 01:53:43 * yates ducks Sep 14 01:53:46 cool Sep 14 01:54:22 I had that orange TI TTL database every red-blooded American God-fearing electrical engineer had on his desk. Sep 14 01:54:29 s/database/databook/ Sep 14 01:54:46 enough reminiscing... Sep 14 01:55:13 I'm mostly a software guy who'd been close enough to hardware for enough time to pick up quite a bit by osmosis Sep 14 01:55:38 ouch. Sep 14 01:55:57 if you're careful you can inhale resistors these days. Sep 14 01:56:29 /not/ careful! Sep 14 01:56:30 hehe, yeah we have one BBB patched to reconnect VDDS to LDO3 instead of LDO1 ... just a matter of moving one resistor Sep 14 01:56:48 no way I was going to try that one myself Sep 14 01:57:51 I'd probably have turned the whole area into a scorched solder blob Sep 14 01:58:58 i finally bought a nice Weller just a few months back and stopped using my $10 Radio Shack crapper Sep 14 01:59:01 but still one needs to know enough about hw to do firmware programming and such... since linux for aforementioned DM814x was unmaintained I actually did quite a bit of baremetal programming on it Sep 14 01:59:23 * zmatt still misses the freedom of not having a kernel getting between you and the hw Sep 14 01:59:45 yes, those days were nice. or a GOOD os like DSP/BIOS Sep 14 02:00:31 I've never used DSP/BIOS... I can tell you however that some of the code of its successor SYS/BIOS is absolute crap Sep 14 02:03:42 (and I wouldn't touch StarterWare with a ten foot pole) Sep 14 02:04:03 8 layers should be flexible enough to squeeze in an I2C RTC Sep 14 02:04:44 Metcal or bust. Sep 14 02:06:28 yates: the sysbios/family/arm/a8/Mmu.c code for updating a translation table entry is particularly hilarious Sep 14 02:06:49 you'd expect Sep 14 02:07:23 ok, night folks - thanks for the info, matt! Sep 14 02:07:23 flush range from cache if required Sep 14 02:07:27 change entry Sep 14 02:07:31 flush tlb Sep 14 02:07:34 they do Sep 14 02:07:37 disable MMU Sep 14 02:07:39 change entry Sep 14 02:07:41 enable MMU Sep 14 02:07:46 -.- Sep 14 02:08:16 oh well, **** ENDING LOGGING AT Mon Sep 14 02:59:58 2015