FHuettig's devicetree.dtb, solved the problem (ssh's are not respodnig in most of the powering) of booting in parallella headless microserver, conected only with power source and ethernet.
Statistics: Posted by nis — Wed Dec 10, 2014 8:00 pm
]]>
The order of initializing the MIO pins. In my experience the order of initialization of the MIO-pins is rather random. Maybe the MIOs are initialized in a suboptimal order, better said changing the order could solve the problem. E.g. defining the clock signal first(and maybe causing an edge) before defining the data signal could cause such problem.
The second possibility (only guessing here) is, that uboot fiddels with the MIO configuration. This would be really a problem for the USB-PHY.
hope this info helps.I am also planning to look deeper into this matter, but I have still some other projects in the queue.
BR
Joerg
Statistics: Posted by ESI — Sun Aug 17, 2014 10:27 am
]]>
]]>
]]>
]]>
]]>
]]>
]]>
As Alberto said the serial cable -should- not be required for booting, I for one have never needed it. However, I think you may be experiencing the headless equivalent of the "Parallella Perpetual Penguin Problem." This problem occurred on some boards when nothing was connected to the UART pins, and the boot process would get stuck with the two penguins displayed on the screen. It never happened when the UART RX line was pulled high by an external device. The solution was to change the boot arguments and remove the UART as the console.
I've updated the headless device tree with the changes we made to the hdmi dev-tree, and attached it here. If you are not using the UART I expect this new tree will solve your problem, copy it to the BOOT partition of your SD card, and rename it "devicetree.dtb" (extract the dtb file from the zip first!). I've tested it on my board and it seems to work, but my board didn't show the problem you describe. Please let me know if this helps, and I'll have it merged into the regular distribution.
The new devtree still includes the UART, it just doesn't assign it as the root console. If at some later point you want to use the UART as a console, you can either:
1) Switch back to the previous devtree, but then you'll again have to make sure there is always something connected to the serial pins at boot time.
2) Keep the new devtree and use agetty to re-assign /dev/ttyPS0 as a login console, you just won't get boot messages on it.
Also, I made one more change to this file, removing one of the GPIOs which was assigned as a reset pin for the USB PHY. If you are using USB devices with your headless system, please let me know if this new devtree causes problems with USB. I want to reclaim that GPIO for general use.
Cheers.
Statistics: Posted by FHuettig — Tue Aug 12, 2014 4:17 am
]]>
Do you know that you can read the log files from the SD card by plugging it into another computer?
That's a way of seeing the error messages from booting without the serial cable attached.
If there are no clear error messages, you can also compare the log files when booting with and without it to see what's different.
Statistics: Posted by AlbertoGP — Mon Aug 11, 2014 8:25 am
]]>
I think I did follow every recommendations from the howto and from the forum posts,
still ... I have this annoying issue I would like to fix
Statistics: Posted by SuperSlot — Mon Aug 11, 2014 7:34 am
]]>
I did look into all the posts about booting issues,
my board is cooled by a big fan, heasinks on both chip, good power supply (with current display monitor).
Linux image from june 11, I also removed /etc/udev/rules.d/70-persistent-net.rules
still ... if I have the RS232/USB cable attached it does boot every time,
if I remove it ... does not boot ... I need to reset and unplug power multiple times...
am I missing something?
is the rss3s cable mandatory for headless config?
thanks
Statistics: Posted by SuperSlot — Sat Aug 09, 2014 9:49 am
]]>