Frequent hang on compiling linux kernel

I am trying to just compile the mainline kernel for some performance benchmarks. It seems that the board hangs every time i do so. It seems to hang when the cpu reaches 70c. However this might be totally coincidental. But the board hangs up completely and consistently every time.

Hello ric96, not sure if this is related, but see here.

seems to be the same.
70c is too little to trip the cpu, I am using one of the latest armbian builds on my sdcard. Is their any other work around or do I need to add a fan and heatsink.

Try the latest Ubuntu desktop image, I had same issue with armbian and the other images. With the latest it’s fine.

look at:

http://docs.khadas.com/basics/FirmwareResources/

this one:

http://www.mediafire.com/file/b6mgy8wjrkmot46/Vim_Ubuntu-mate-16.04_V170220.7z

its only to flash not for sd card

Their seems to be a bunch of image floating around, can you link me to the right one?
Thnx

look at:

http://docs.khadas.com/basics/FirmwareResources/

this one:

its only to flash not for sd card

1 Like

Did you mean that you wanna get mainline linux kernel running on Khadas VIM?

That would be nice, but not what I meant. I just wanted to check how much time it would take to compile the kernel. To get some benchmarks.

OK, I see. So your problem has been resolved with the new Ubuntu-Mate ROM, right?

In fact, there are some developers have get Linux-4.11 running on Khadas VIM, and they are also push Khadas VIM with official Linux supported, so VIM will got mainline linux supported soon.

Yes the problem has been solved with the latest official ubuntu mate build.
And it’s good to hear that manline kernel will be getting support.

The reason for the drop of the core at a temperature of 70 degrees in the images Armbian module “mali.ko”. Remove it from the system and lows will be.

1 Like

@balbes150
I think that is not really necessary, I have the kernel (from khadas github) with the mali driver for frame-buffer from amlogic buildroot that is working over 70 celsius,
these are the steps:

  1. just disable the gpu cooling and gpucore into the linux thermal fsys,
  2. then is necessary a little patch to the gpu cooling (the code return NULL instead int…)
  3. then another little patch to the thermal driver from amlogic (I think that this is not necessary , I have to check beacuse I have changed that code for debug)

after these 2 or 3 steps when the temp go over the first trip point (70) all should be fine,
the gpufreq_get_requested_power is no more called.
The reason could be the mali driver that is not handling the request correctly.

1 Like