It is not a flakey DC jack. But yes the symptoms do mimic that kind of failure. Same hardware works fine with Debian 8.1/kernel 3.14 and prior.
It happens under all loading /activity, but easy to reproduce with the board idling, minimum console software load. It does not happen with the board powered by the USB port. It only happens when powered by the +5V barrel jack. No indication in syslog as to what happened. == I note that you were working in area changing ticks to jiffies or something like that. I personally suspect that there was some code dependent on the previous timing system that was used to time some of the USB power switch over mechanisms, that was not updated to deal with the timing system changes you made. For instance, the On-the-Go USB system periodically pushes power out the USB port to see if someone has connected a USB device. The PMIC sees this and can interpret this as power has appeared on the USB connector and to attempt a switchover. It really needs to wait some time period to see that it is stable input power rather than a power probe pulse generated by the BBB itself. If your timing system changes modified this switchover timing, the system could attempt to switch to a non-existant power source and die, causing a power-on reboot. This theory is supported by experimental tests run by William that says if the USB power line is shorted to ground, then the unit does not reboot. I am just summarizing several weeks of observations by all the guys on this thread. --- Graham On Mon, Aug 10, 2015 at 2:00 PM, Felipe Balbi <[email protected]> wrote: > Hi, > > On Mon, Aug 10, 2015 at 01:56:04PM -0500, Robert Nelson wrote: > > On Mon, Aug 10, 2015 at 1:05 PM, Robert Nelson <[email protected]> > wrote: > > > On Mon, Aug 10, 2015 at 11:23 AM, Nuno Gonçalves <[email protected]> > wrote: > > >> 3 weeks ago I asked for it "For what is worth I believe the OMAP and > > >> PMIC reset reason registers should be part of the boot log so future > > >> reboot problems can be sorted.". > > >> > > >> But I just asked, I didn't actually code! > > >> > > >> Not sure if it is easy to retrieve both, or if PMIC is that much > > >> relevant for this issues. > > >> > > >> Anyway, this will surely take out black magic from the equation when > > >> debugging this problems. > > > > > > Okay here's something quick/dirty... > > > > > > > http://rcn-ee.homeip.net:81/farm/testing/v4.1.x/MLO-am335x_evm-v2015.07-r3 > > > > http://rcn-ee.homeip.net:81/farm/testing/v4.1.x/u-boot-am335x_evm-v2015.07-r3.img > > > > > > #patch: > > > > https://github.com/RobertCNelson/Bootloader-Builder/blob/master/patches/v2015.07/0001-am335x_evm-uEnv.txt-bootz-n-fixes.patch#L21 > > > > > > make sure the eMMC doesn't have it's own bootloader: > > > > > > U-Boot 2015.07-00001-gd8d32cf (Aug 10 2015 - 12:54:52 -0500) > > > > > > Watchdog enabled > > > I2C: ready > > > DRAM: 512 MiB > > > Reset Source: Global external warm reset has occurred. > > > Reset Source: Power-on reset has occurred. > > > MMC: OMAP SD/MMC: 0, OMAP SD/MMC: 1 > > > Using default environment > > > > > > dd if=/dev/zero of=/dev/mmcblk1 bs=1M count=10 > > > > > > sudo dd if=MLO-am335x_evm-v2015.07-r3 of=/dev/sde count=1 seek=1 > bs=128k > > > sudo dd if=u-boot-am335x_evm-v2015.07-r3.img of=/dev/sde count=2 > seek=1 bs=384k > > > > > > I need to fire up x86, that's attached to the bbb's (right now i can't > > > log serial) > > > > Okay.... within a few minutes the first board rebooted... > > > > "Reset Source: Power-on reset has occurred." .... > > weird, this is likely either flakey DC Jack connection, or PMIC going > bonkers. > > How do you guys reproduce this ? Is the board idle or is it running ? > Which gadget driver ? Is USB cable connected to a host ? A desktop > host or wired back into the host port of BBB itself ? > > I wanna try to reproduce it here. > > -- > balbi > -- For more options, visit http://beagleboard.org/discuss --- You received this message because you are subscribed to the Google Groups "BeagleBoard" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
