**** BEGIN LOGGING AT Fri Aug 01 03:00:00 2014 Aug 01 03:55:07 Excuse me, I'm curious... does the Beaglebone Black have any USB problems I should be made aware of before I buy one? Aug 01 03:55:37 BoardNConfyoozd: I've been seeing the discussion in #sparkfun Aug 01 03:56:58 the most common issues seen are devices needing to be plugged in before power up... though most devices are discovered after doing a few 'lsusb', but the only reliable method on the shipped kernel right now is to have the device connected at power-up. Aug 01 03:57:37 Are there any other USB issues? Aug 01 03:57:50 Most small embedded boards are notorious for not sourcing a whole ton of power to the USB ports, so if you want to drive anything substantial, you'll want a powered hub. That's all I'm aware of. Aug 01 03:58:10 I think the BBB is better in that respect than most other boards, though. Aug 01 03:58:12 And does that kernel problem relate to other Linux distros, like, say, ArchLinux? Aug 01 03:58:19 myself: the BeagleBone Black will pass through quite a bit of power from the 5V DC jack. Aug 01 03:58:58 The correct current for a USB port is 500 mA. Aug 01 03:59:00 Ooooh, is that just happening when people are trying to run the board from an upstream PC via USB? That makes sense. Aug 01 03:59:01 BoardNConfyoozd: most distros base off the BeagleBoard.org 3.8 kernel, but newer kernels fix several issues, including the dynamic detection. Aug 01 03:59:26 myself: yeah, you'll run out of power pretty quick when powering the board off of USB. Aug 01 03:59:43 (if you start attaching USB devices to the host port of the bone) Aug 01 03:59:51 Yeah, you can't power USB off of a USB supply. :P Aug 01 04:00:08 If you get a 5V 2A supply, any 500mA USB device should work with plenty of spare power. Aug 01 04:00:27 So dynamic detection is the only real issue here? Aug 01 04:00:34 hello? Aug 01 04:00:41 WAIT. Plug the board's own USB back into its USB ports... INIFNITE POWAAAA! Aug 01 04:00:52 The 3.8 kernel also lacks DMA support (still uses FIFOs in the USB controller, but CPU moves the data) Aug 01 04:00:58 newer kernels add DMA support. Aug 01 04:01:21 Would the ArchLinux distro count as a "newer kernel"? Aug 01 04:02:20 the result of not having DMA support is that some high-bandwidth devices will chew up the bandwidth. In practice, I've found the 3.8 kernel on BeagleBone Black to have much higher throughput than on say, a Raspberry Pi. Paul Struffrogen (creator of the Teensy) has documented this finding. Aug 01 04:02:32 * jkridner googles to see which kernel Arch is using. Aug 01 04:05:07 jkridner: (or anybody) Do you know of any changes in the ethernet driver from 3.8.x to 3.15.x that would significantly improve network performace? Aug 01 04:06:00 jbdatko: I haven't heard about Ethernet performance improvements, but there might be some. USB and MMC/SD for sure. Aug 01 04:07:13 BoardNConfyoozd: sorry about the disconnection. Aug 01 04:07:28 Internets, right? XD Aug 01 04:07:31 jkridner: Ok, thanks. Aug 01 04:07:37 Always disconnecting on us... XD Aug 01 04:07:56 I swear, there must be a demon whose sole purpose is to cause lag and disconnections... Aug 01 04:12:37 BoardNConfyoozd: felipebalbi is an expert on the controller... you can accept what he says as true enough. Aug 01 04:14:14 Did you figure out which kernel the ArchLinux distro uses? Aug 01 04:16:15 nope, not yet. I suspect they just grab our 3.8.13-boneX series. Aug 01 04:16:43 back in April that seemed to be the case: https://bbs.archlinux.org/viewtopic.php?pid=1404046 Aug 01 04:17:25 With the provided Debian distro, you can switch kernels pretty easily to a newer one that has less USB issues, but doesn't have all the add-on cape support. Aug 01 04:17:37 Do you know a hard-real-time OS that runs on Beaglebone Black? Aug 01 04:17:59 I'm not as worried about cape support just yet. Aug 01 04:18:58 kernels and distros aren't always so tightly coupled, so you should be able to use what distro you want with what kernel you want Aug 01 04:19:16 does anyone know if the u-boot partition for the beagleboneblack with the MLO needs to be fat32 or fat16? marked bootable? Aug 01 04:20:09 BoardNConfyoozd: I recommend using http://machinekit.io with the Linux Xenomai patch for hard-real-time, though there are several RTOSes that run and it is possible to run a good set of drivers with Linux PREEMPT_RT. What is your desired RTOS? Aug 01 04:20:28 i swear i had the process down, but when it boots, it just falls into the "CCCC" repeated over and over, searching for a boot image from serail... Aug 01 04:20:33 vagrantc: it does need to be marked bootable. Aug 01 04:20:40 I don't know yet, honestly. Aug 01 04:20:46 I don't think FAT32/FAT16 matter... Aug 01 04:21:06 If there were a locally available class, I would probably just stick with whatever the class taught, until I had clearer requirements. Aug 01 04:21:14 you can get away with not marking the partition bootable if you put MLO in the boot sector (need to avoid overwriting the partition table). Aug 01 04:21:23 But I've heard good things about Xenomai, in that it's free. Aug 01 04:21:25 managed to get the debian-installer daily built images to run out of the box, but my manual process for installing u-boot didn't work out. Aug 01 04:21:33 vagrantc: I did a post on http://beagleboard.org/discuss somewhere on how to dd to the boot sector. Aug 01 04:21:45 jkridner: ah, that would be much more ideal! Aug 01 04:22:15 no messing around with filesystems and making sure the MLO occupies the first inode or whatever that was... Aug 01 04:22:15 vagrantc: When I discussed doing that as the default, concern was expressed over not having the file visible. Aug 01 04:22:51 i don't see what advantage that gives, it seems to only give the ability to break the system. Aug 01 04:26:21 there are so many ways to break the MLO on a filesystem... Aug 01 04:26:59 on other boards, where you dd the SPL and u-boot to the partition, it's seem much less prone to error... Aug 01 04:27:15 musb and DMA? hehe Aug 01 04:27:25 Good luck and pray to your diety of choice. Aug 01 04:30:44 vagrantc: yeah, it seems the user shouldn't be updating MLO if they can't figure out how to 'dd' in the first place Aug 01 04:31:04 vagrantc: maybe I'll fight back on the objection, but it would be nice if you found the thread and chimed in. Aug 01 04:33:37 jkridner: having a hard time finding the thread... which list/forum was it posted to? Aug 01 04:34:52 yeah, musb/dma issues still plague the debian kernel builds for armmp, unfortunately... Aug 01 04:36:38 which build a single kernel for many boards, but the usb support for bbb is pretty much broken, as mainline linux doesn't support dma with multiplatform kernels... Aug 01 04:41:18 vagrantc: subject "microSD layout 2.x" Aug 01 04:41:42 vagrantc: I think it might have been off-list. should be raised on the main list. Aug 01 04:42:57 ah, that thread. Aug 01 04:43:57 vagrantc: I also forwarded the instructions for writing MLO to the boot sector... Aug 01 04:44:09 that too needs to be brought back into focus on the main list. Aug 01 04:46:27 jkridner: i didn't find the instructions in that thread Aug 01 04:47:00 jkridner: last post that i saw on that was july 3rd Aug 01 04:48:00 I sent an additional thread to you. didn't see it? Aug 01 04:48:05 jkridner: the good news, the debian installer daily builds work to support network booted bbb installs almost out of the box. Aug 01 04:48:30 I realize that the rev. C isn't that old, but is there going to be a new Beaglebone any time soon that supports 1080p at 60 Hz or so? Aug 01 04:48:57 dd if=MLO of=/dev/sdX bs=1 count=80 Aug 01 04:48:57 dd if=MLO of=/dev/sdX bs=512 seek=1 skip=1 Aug 01 04:49:02 jkridner: got it. Aug 01 04:49:10 vagrantc: ^^^ will avoid writing over the partition table... Aug 01 04:49:42 jkridner: needs to be written in both places? or that's two ways of writing the same thing? Aug 01 04:49:59 You need to make sure it has a "CH" in it, which will also help with having no valid info in MLO at the location of the partition table. Aug 01 04:50:37 vagrantc: it writes the first 80 bytes, then skips bytes 80-511, starting again at byte 512. Aug 01 04:50:49 ah, fun. Aug 01 04:51:00 BoardNConfyoozd: no, not likely. Aug 01 04:51:19 jkridner: i wonder if i couldn't get an image that just pads the middle 80-511 bytes. Aug 01 04:51:23 I suppose they would have to re-do the hardware of the entire board? Aug 01 04:51:29 vagrantc: the byte skipping prevents trashing the partition table. Aug 01 04:51:29 jkridner: to simplify the dd process. Aug 01 04:51:44 vagrantc: no, you need to not touch those bytes. Aug 01 04:51:47 got it. Aug 01 04:52:14 the content in them is fine for MLO, but the card partitions would be wiped out. :-) Aug 01 04:52:48 right Aug 01 04:53:38 jkridner: so the two dd commands... does that essentially handle the "CH" thing? Aug 01 04:53:56 vagrantc: no, that is something that comes up as you build MLO... Aug 01 04:54:09 I think MLO might always have it, but I'd need to check. Aug 01 04:54:13 need to build the MLO in some special way? Aug 01 04:54:26 vagrantc: I'm not sure if it is the default or not. Aug 01 04:54:52 well, there's nothing like experimentation to figure things out (at least partially) Aug 01 04:55:48 jkridner: no, never got that. Aug 01 04:56:19 vagrantc: http://git.denx.de/?p=u-boot.git;a=blob;f=tools/omapimage.c;h=1e0c16479681aec2d8621eabd953ff9a8cdb203c;hb=HEAD shows that it knows how to build a CH header Aug 01 04:56:24 jkridner: the "all that is holy to #beagle" one... Aug 01 04:56:37 vagrantc: now, I need to see how omapimage is called in the Makefile. Aug 01 04:56:49 * vagrantc fires up the bbb Aug 01 04:56:51 vagrantc: oh. I thought you were on that list. :( Aug 01 04:57:06 jkridner: it didn't look like it was posted to the list Aug 01 04:57:07 vagrantc: feel free to raise the topic on the main list. Aug 01 04:58:20 it does make it tricker to force boot off of SD... but you can wipe it with dd instead of "rm MLO" Aug 01 05:01:14 vagrantc: http://git.denx.de/?p=u-boot.git;a=commit;h=3decb14abe76d244ba98fd158ef95f89e7e37d70 seems to indicate that MLO is built with a CH header Aug 01 05:02:43 jkridner: so those dd lines should work with eMMC or uSD card, yes? Aug 01 05:03:07 yes. Aug 01 05:03:55 with eMMC, there is technically yet another mechanism that is possible, though I have no idea how to use it, which is to use the boot partitions. Aug 01 05:04:15 anyway, the 'dd' of MLO to the boot sector will work fine. Aug 01 05:04:27 * jkridner goes to sleep Aug 01 05:04:30 * vagrantc tries for sure Aug 01 05:04:32 * vagrantc waves Aug 01 05:08:43 U-Boot SPL 2014.04 (Jul 28 2014 - 23:46:34) Aug 01 05:08:43 ### ERROR ### Please RESET the board ### Aug 01 05:08:45 :( Aug 01 05:09:12 doesn't know where to find u-boot.img, i'm guessing Aug 01 05:09:39 for another day Aug 01 05:12:17 Is there a free version of CCS for Beaglebone? Aug 01 05:15:19 hrm. doesn't boot from SD now either... Aug 01 07:37:30 hi, anyone here? Aug 01 07:37:44 im interested if anyone has managed to run a BBB off battery power Aug 01 07:37:48 as i cant shut it down Aug 01 07:38:24 it was not designed to be optimized for very low power consumption Aug 01 07:38:46 that said some people succeeded in using it with LiIon batteries Aug 01 07:39:43 mine has a constant 40mA draw after "shutdown" Aug 01 07:40:05 going to break out i2c0 and attach a logic analyser today Aug 01 07:40:39 i dont see any fixes in the kernel Aug 01 07:42:32 no idea about that Aug 01 07:42:35 tbh this is so annoying im tempted to add a micro to hijack the i2c bus Aug 01 08:08:51 hello Aug 01 08:09:14 anyone there with some VNC experience with BeagleBone black? Aug 01 08:36:24 heh, we haven't had TII ask for help here in a while... Aug 01 08:39:25 Hi Aug 01 08:39:34 how to disable bbb cape ? Aug 01 08:39:37 on boot? Aug 01 08:39:46 uEnv.txt thing does nothing Aug 01 08:40:59 expand on "thing" Aug 01 08:41:13 "does not work" is not a problem description Aug 01 08:41:14 maybe it needs its internet Aug 01 08:42:09 ah Aug 01 08:42:15 cape_disable=capemgr.disable_partno=BB-BONE-EMMC-2G Aug 01 08:42:19 I will try that Aug 01 08:43:52 does nothing Aug 01 08:43:56 :( Aug 01 08:44:16 I add to uEnv.txt the following line cape_disable=capemgr.disable_partno=BB-BONE-EMMC-2G,BB-BONELT-HDMI,BB-BONELT-HDMIN Aug 01 08:44:32 and after booting I still see those capes in cat /sys/devices/bone_capemgr.9/slots Aug 01 08:45:08 mkad: and do you see that on /proc/cmdline? Aug 01 08:45:38 yes Aug 01 08:45:49 cat /proc/cmdline Aug 01 08:45:49 console=ttyO0,115200n8 capemgr.disable_partno=BB-BONE-EMMC-2G,BB-BONELT-HDMI,BB-BONELT-HDMIN root=/dev/mmcblk0p2 ro rootfstype=ext4 rootwait fixrtc quiet Aug 01 08:46:30 looking at that I think you should consider buying a blue-steel board Aug 01 08:46:47 but no, no idea why it still loads that Aug 01 08:48:37 but... Aug 01 08:48:38 4: ff:P-O-- Bone-LT-eMMC-2G,00A0,Texas Instrument,BB-BONE-EMMC-2G Aug 01 08:48:47 ff:P-0 ? Aug 01 08:49:05 I wonder what does it mean Aug 01 08:49:29 oh I know Aug 01 08:49:35 maybe I can mv away dtb files Aug 01 08:51:26 tbr, is L for loaded ? Aug 01 08:53:43 no idea Aug 01 08:55:37 citing "Now, when the list of overlays are displayed, the Bone-Black-HDMI & Bone-Black-HDMIN interfaces are shown as not loaded (L): " Aug 01 08:55:40 so it works Aug 01 08:55:52 tbr, thanks for patience Aug 01 08:58:12 :) Aug 01 11:08:37 Hi Aug 01 11:08:54 is there a working pwm code for pru? Aug 01 11:09:09 all examples I find are mess or dont work Aug 01 11:21:09 can u recommend reading how to get along with prus? Aug 01 11:46:44 hi.. does boneblack boot from the Nandflash? Aug 01 11:46:59 I want to format the NAND flash and put the kernel and filesystem Aug 01 11:47:04 into the nand Aug 01 11:49:30 hacker1, it boots from emmc Aug 01 11:50:12 hacker1, I guess u can flash it, there is an image flasher which boots from sd and installs OS into emmc, maybe you could modify it Aug 01 13:01:15 hello! Aug 01 13:10:44 ehlo Aug 01 13:11:45 freestyle! Aug 01 13:50:13 I am running ubuntu 14.04 on a beagleboard xM, is it useful to install xserver-xorg-video-armsoc or is there a different package that's appropriate? Aug 01 13:56:06 honestly, I dont know Aug 01 14:00:43 xserver-xorg pulls in xserver-xorg-video-all anyway so I guess it's a moot point. Aug 01 17:16:04 jkridner: so, dd'ing the MLO to the eMMC didn't work out so well Aug 01 17:16:15 no? Aug 01 17:16:31 22:08 < vagrantc> U-Boot SPL 2014.04 (Jul 28 2014 - 23:46:34) Aug 01 17:16:32 22:08 < vagrantc> ### ERROR ### Please RESET the board ### Aug 01 17:17:08 was it introduced in more recent u-boot ? i.e. 2014.07 ? Aug 01 17:18:11 or maybe the MLO is not finding the u-boot.img on the partition? Aug 01 17:19:17 can the u-boot.img also be written directly to the boot sector? Aug 01 17:20:25 * jkridner is checking out http://makezine.com/projects/internet-speedometer/ Aug 01 17:24:17 vagrantc: well, you still need u-boot.img... Aug 01 17:24:46 it is possible to modify the MLO/SPL to look for it elsewhere, but the default is to look for u-boot.img on the first partition (FAT). Aug 01 17:25:03 the fact that you have stuff printing shows the MLO bootload was successful. Aug 01 17:26:22 right Aug 01 17:26:42 so, essentially, i'm probably running into the same problem that caused the MLO to not load from the fat partition Aug 01 17:27:40 what really gets me, is i can't think of what i'm doing differently in the partitioning layout... it's worked for me dozens of other times. Aug 01 17:31:40 well, MLO knows how to parse FAT reasonably well... you can recompile it to look on an EXT2/3/4. I've been encouraging rcn-ee to do just that, leaving only a uEnv.txt and the getting started info on the FAT partition. Aug 01 17:59:52 could the position of the fat partition matter? i.e. starting a 1024kb vs. 512kb ? Aug 01 18:09:02 well, how 'bout that. moving the fat partition earlier got it to work with MLO on the filesystem. still no dice with the dd to raw device approach Aug 01 18:13:10 vagrantc, the bootrom is pretty picky... ;) Aug 01 18:13:20 rcn-ee: clearly. Aug 01 18:14:43 so debian-installer's habit of leaving a little extra unpartitioned space at the beginning of the disk seem to make installing u-boot from debian-installer a bit of a hassle. Aug 01 18:15:28 how much does it leave? a 0 - 1Mb hole is fine and usually what i leave.. Aug 01 18:16:10 i looked like ~1MB Aug 01 18:16:18 is the boot flag turned on? Aug 01 18:16:21 yup Aug 01 18:16:42 its it sfdisk 2.17..x? Aug 01 18:17:03 the ROMBL used to be even more tricky on OMAP3/4 Aug 01 18:17:21 pain in the ass on the omap34xx ;) Aug 01 18:18:05 rcn-ee: sfdisk in jessie is in util-linux 2.20.1, currently Aug 01 18:18:18 but i think debian-installer uses parted Aug 01 18:18:37 yeap, jessie's works fine.. humm, i never had much luck with parted.. Aug 01 18:19:08 Does MLO get copied atleast first, before anything else after it's formated? Aug 01 18:19:20 well, i did the MLO copying manually Aug 01 18:19:25 okay Aug 01 18:20:41 i guess before maybe i manually created the partition as well ... not sure at this point. Aug 01 18:21:04 vagrantc, http://pastebin.com/hzB7s2cJ (dump from parted, ona working disk) Aug 01 18:21:38 * vagrantc will run through d-i again Aug 01 18:22:37 once i get a process down, i'll submit patches where needed to make it work out of the box Aug 01 18:25:23 hrm. u-boot 2014.07 seems to be missing autodetectfdt :( Aug 01 18:25:41 findfdt Aug 01 18:26:28 another one of those u-boot variables that needs to unified.. Aug 01 18:26:50 no doubt. Aug 01 18:39:01 rcn-ee: for the u-boot partition, did you settle on 16MB? 4MB? i forget all the nuances with the different layouts Aug 01 18:40:16 12Mb.. fat16 didn't want to go smaller... could go the "fat12" route... Aug 01 18:41:10 i've also been doing fat32, i think... Aug 01 18:42:01 which didn't seem to mind smaller partitions... but maybe that's caused me all manner of other headache Aug 01 18:42:22 it should still boot with that.. I'm pretty sure angstrom always defaulted to fat32.. I've just always set it as fat16.. Aug 01 18:43:28 i know mkfs.vfat in jessie like to yell at you if that fat partition is too small.. Aug 01 18:46:32 wheezy's mkfs.vfat didn't seem to complain Aug 01 18:48:35 can we consider that a regression? ;) Also in jessie, when you call mkfs.ext4, if the microSD was already previously partitioned, it'll detect that and ask you if you are sure.. Aug 01 18:48:55 even if you've blown out the first 10Mb of data with /dev/zero. .;) Aug 01 18:49:13 that seems perplexing Aug 01 18:50:03 i find it hilarous. ;) so when i zero it out, https://github.com/RobertCNelson/omap-image-builder/blob/master/tools/setup_sdcard.sh#L239 (it has to be atleast bigger then then first partition to nuke the 2nd's table) Aug 01 19:41:49 jkridner, just got darren's sgx patches to work... Aug 01 19:41:57 sweet Aug 01 19:42:10 features? limitations? java? :-) Aug 01 19:42:31 the running man demo. ;) now i need to cleanup my kernel patches and push out.. Aug 01 19:42:52 i'll start a thread on the mainling list to see if we can get more users to try it.. Aug 01 19:43:14 waaaaaaaat Aug 01 19:43:17 working sgx? Aug 01 19:43:19 in which kernel? Aug 01 19:43:45 nyt-, we were limited to v3.13.x, but now v3.14.x works.. Aug 01 19:44:02 hows device tree in those? Aug 01 19:44:10 i hadn't checked on those in a long time Aug 01 19:44:43 and debian just decided to go with 3.16 for jessie... Aug 01 19:46:32 ti had fixed one of the road blocks for me getting sgx working on v3.13.x, so the next stage is v3.16.x ;) Aug 01 19:46:42 * vagrantc wonders if mounting vfat with utf8 is corrupting the filesystem... Aug 01 19:50:01 in my debian config-delta, i always shove that to iso8859-1 Aug 01 19:50:23 is device tree stuff in the newer kernels yet? Aug 01 19:50:25 hi I short circuited my beaglebone black, does anyone know of tests I could run to figure out what failed? Aug 01 19:50:49 nyt-, mainline boots out of the box on bbb Aug 01 19:52:18 been a while, last i checked it was running 3.8, but upgrading to one of the newer kernels couldnt load any overlays Aug 01 19:52:27 should flash soon Aug 01 19:52:30 and check it out Aug 01 19:54:22 nyt-, no overlay support in mainline. Aug 01 19:56:41 =[ Aug 01 19:58:10 rcn-ee: man demo is in full-screen mode? Aug 01 19:58:42 yeap, full screen.. don't even have xorg on this install. ;) Aug 01 19:58:56 very, very interested in the xorg behavior... Aug 01 19:59:05 not sure if you saw the follow-up from the Oracle guy. Aug 01 20:03:35 yeah xorg w/ accel would be nice Aug 01 20:04:03 just the initial reply from Gary Aug 01 20:04:31 nyt-, we have EGL, so if someone gets the middle stuff working wayland works. ;) Aug 01 20:06:18 shit i wish i had time that sounds like a good effort Aug 01 20:06:27 at this point I wouldn't care about xorg, wayland would be preferred Aug 01 20:06:30 ever since i wrote that ruby lib i got hammered at work and haven't even been able to build something =[ Aug 01 21:20:57 Hi All. I cannot connect to my BB. I have the drivers installed. If I go to the device manager page, I have USB serial converter A and B and Com13 installed and working. Aug 01 21:22:36 If I go to network connections - no USB network devices are listed. Aug 01 21:25:00 All my computers are on fixed IP addresses. They are on 120.20.2.XX for example. Aug 01 21:25:14 Could this be problem ? Aug 01 21:26:34 I can see the files on the BB so I am connected Aug 01 21:47:43 ok, fat16, 12MB, created from debian-installer, manually copying MLO/u-boot.img ... seems to work so far. Aug 01 21:49:16 good to hear... So anything smaller then 12Mb will need fat12... i wonder if vfat handles that just fine? Aug 01 21:50:13 * vagrantc has tested enough combinations for one day... :) Aug 01 21:51:02 * vagrantc is tempted to try the raw device approach again Aug 01 21:51:05 fat12 died around the days of >32Mb harddrives... you don't mind wasting 11mb? Aug 01 21:51:43 we can use that for the "firmware" directory... that way windows users could add their debian-installer firmware. . ;) Aug 01 21:53:22 zmatt: ping Aug 01 21:55:31 given how finicky it is about where the MLO is, and ideally u-boot.img, i'd rather write them to the raw device so that users can't accidentally delete them... Aug 01 21:56:35 would also probably be easier to configure a debian-installer u-boot hook to write them than having to mess with getting the partitioning exactly right Aug 01 21:57:14 well, no dice on the lastest dd to raw device... Aug 01 21:57:41 let's see if it has somehow broken the fat partition now. Aug 01 21:58:36 as long as that fat partition is not mounted and only MLO/u-boot.img is on it. that should be good enough for most users, as "hidden" as long as they don't plug the card into a windows machine.. Aug 01 21:59:49 Right now, i only mount that partition, if i have to use it as a trampoline, with vmlinuz/initrd.img/dtbs on it. For when u-boot cant' read the final rootfs.. Aug 01 21:59:51 * Laurenceb__ has just fixed BBB shutdown Aug 01 22:00:12 added a cap/resistor network onto pin1 on U4 Aug 01 22:00:32 so the regulator shuts down correctly Aug 01 22:06:46 it seems that booting the MLO from the raw device corrupts the fat filesystem when it tries to read it Aug 01 22:21:54 what's U4 Aug 01 22:22:14 rcn-ee: is the first partition still exported as as MSD over USB? Aug 01 23:02:59 Laurenceb__: pong :P Aug 01 23:05:28 U4 is a linear regulator Aug 01 23:06:02 generating VDD_3V3B from SYS_5V with VDD_3V3AUX as enable Aug 01 23:07:44 ahhhh Aug 01 23:08:57 VDD_3V3AUX is used only for that enable and for the power led Aug 01 23:09:20 which might not be enough to discharge that rail Aug 01 23:11:14 LED's forward voltage at 0 mA is 3V Aug 01 23:12:24 enable input is logic high at >= 2V Aug 01 23:14:24 not sure what he meant by cap/resistor network though, if I understand this right all that's needed is a pulldown resistor Aug 01 23:16:00 ds2: i don't think the first (u-boot) partition will be exported, but there's still a fat partition that's exported Aug 01 23:16:28 wait, the LDO has an internal discharge resistor, so a pulldown shouldn't be necessary Aug 01 23:16:44 * zmatt then has no idea what Laurenceb meant Aug 01 23:20:30 Hmmmmmm Aug 01 23:23:32 What's that humming sound? Aug 01 23:28:28 xformers Aug 01 23:33:33 emeb: http://38.media.tumblr.com/2655874232e0c8790fca00bf835b780a/tumblr_mr1eojBJIt1r8vrhxo1_1280.jpg Aug 01 23:34:40 zmatt: that's a big cicada Aug 01 23:35:01 the ones we get here are about 2" long Aug 01 23:36:24 I don't really encounter any large insects around here... I think I'm fine with that :P Aug 01 23:42:44 ahh wait, they changed the schematic around U4 Aug 01 23:45:39 oh boy, that doesn't look good Aug 01 23:46:20 3V3A now enables 3V3B Aug 01 23:46:36 but there are probably a shitload of leakage paths from 3V3B to 3V3A Aug 01 23:47:24 based on Laurenceb's experience of high current during shutdown, enough leakage to keep the regulator enabled Aug 01 23:47:59 but keeping the 3V3B rail on like that while the remaining rails are off can't be good for the hardware's life expectency Aug 01 23:50:10 can anyone with A6 hardware or later check whether VDD_3V3B (the 3v3 available on the expansion headers) remains on after shutdown (all leds off) ? I have A5A hardware myself **** ENDING LOGGING AT Sat Aug 02 02:59:59 2014