Difference between -ti and bone* kernel?

Hello all,

I’ve noticed some bad HDMI jitter/flickering when using the -bone kernel, which is present both in 3.14 and latest 4.1. However this flickering is not present when using the -ti kernel, regardless of the version; the -ti version is used by default in rcn’s netinstall script. Unfortunately, building sgx modules fails with the -ti kernel (“error: implicit declaration of function fb2display”) but works fine, using the same SDK, when using -bone kernel as compiled with rcn’s bb-kernel repo (branch am33x-3.14).

Therefore I’m wondering what is ultimately the difference between those two (isn’t it a waste of effort to maintain two separate kernels?), and if any of you had a solution for this build issue. Hopefully I’d like to use the kernel with the least bugs, which seems to be -ti so far regarding HDMI. Nonetheless I’d like to thank Robert C. Nelson for his work and in particular the very useful netinstall/bb-kernel repos.

Best regards,

Tziang

Hello all,

I've noticed some bad HDMI jitter/flickering when using the -bone kernel,
which is present both in 3.14 and latest 4.1. However this flickering is
not present when using the -ti kernel, regardless of the version; the -ti
version is used by default in rcn's netinstall script. Unfortunately,
building sgx modules fails with the -ti kernel ("error: implicit
declaration of function fb2display") but works fine, using the same SDK,
when using -bone kernel as compiled with rcn's bb-kernel repo (branch
am33x-3.14).

"flickering" should be gone in am33x-4.4+ as tilcdc patches went mainline a
couple months back and i back ported it..

Therefore I'm wondering what is ultimately the difference between those
two (isn't it a waste of effort to maintain two separate kernels?), and if
any of you had a solution for this build issue. Hopefully I'd like to use
the kernel with the least bugs, which seems to be -ti so far regarding
HDMI. Nonetheless I'd like to thank Robert C. Nelson for his work and in
particular the very useful netinstall/bb-kernel repos.

It's only a mess right now, as we have a couple transitions in flight..

uio_pruss -> remoteproc_pruss
old_sgx_drivers -> new_sgx_drivers

A lot of users, use uio_pruss and the old_sgx_drivers, thus they are
currently stuck in the "bone" kernel's..

at some point, remoteproc_pruss will fulfill uio_pruss users needs (and
also go mainline), thus solving that problem..

sgx = no comment...

Regards,

sgx = no comment…

Regards,

heh yeah, it’s only been 3 years TI . . .just for the beaglebone black platform.

Robert, TI kernel still has usable power management ? Or is this no longer a difference between the two ?

v4.1.x-ti has the best..

v4.4.x-bone & v4.4.x-ti are about the exact same.. (at the moment)

at idle:

  current CPU frequency is 300 MHz (asserted by call to hardware).
  cpufreq stats: 300 MHz:60.64%, 600 MHz:2.75%, 800 MHz:0.00%, 1000
MHz:36.60% (14)
root@beaglebone:~# uname -r
4.4.8-ti-r22

5v @ 0.15~0.18 Amps on my jumpy power supply...

echo mem > /sys/power/state

5v @ 0.05 Amps

bbb: rev a5a, just a ftdi serial cape attached..

Regards,

OK cool. I’m still wanting to stick with 4.1.xbone but since for the given project I’m working on I do not need uio_pruss - Maybe I’ll use the TI kernel instead.

I had also noticed in the past that the TI kernels used slightly less memory at boot too. But only because the bone kernel was creating a few smallish ramdisk’s at boot. I think the difference was only a few Megabytes of used RAM.

Thanks Robert :slight_smile:

Hello Robert,

Thanks for the fast reply. I’m unsure what you mean by “sgx = no comment”. Is sgx no longer officially supported by TI?

Also I would like to provide more light about my building issues with the stock Graphics SDK provided by TI:

  • issue 1: “error: implicit declaration of function fb2display” occurred because I forgot to add TI_PLATFORM=ti335x. The complete command line is
    make GRAPHICS_INSTALL_DIR=/home/tzsw/Graphics_SDK_5_01_01_02 KERNEL_INSTALL_DIR=/home/tzsw/ti-linux-kernel BUILD=release OMAPES=8.x FBDEV=yes TI_PLATFORM=ti335x all

  • issue 2: “fatal error: omapfb/omapfb.h: No such file or directory” is a known issue with the TI Graphics SDK. Does TI not test its SDK against its kernel? Robert C Nelson’s ti-sdk-pvr repository on github has the fix, in particular the 5.01.01.01-ti branch: https://github.com/RobertCNelson/ti-sdk-pvr/commits/5.01.01.01-ti

Tziang