Wandboard imx 3.0.35 4.0.0 test

[wolfgar]

Hi John

Here are a few trivial commits I made on your test branch to be able to compile and to give it a try...

[johnweber]

Thanks Stephan - I merged these. I'll be testing them over the next day or so
before merging this into the main branch.

One thing I want to do is to upstream the gpu-viv patch. It does give warnings
and for certain kernel configs it can cause a hard build error.

John

On 7/31/13 6:54 PM, wolfgar wrote:

Hi John

Here are a few trivial commits I made on your test branch to be able to compile
and to give it a try...

---

    You can merge this Pull Request by running

git pull https://github.com/wolfgar/linux wandboard_imx_3.0.35_4.0.0_test

Or view, comment on, or merge it at:

#1

    Commit Summary

• Fix uninitialized variables warnings

• Fix compilation when mipi-csi option is disabled

• Adapt video memory limit according to wandboard version

  File Changes
  

• M arch/arm/mach-mx6/board-wand.c
  https://github.com/johnweber/linux/pull/1/files#diff-0 (20)

• M drivers/mxc/gpu-viv/hal/kernel/gc_hal_kernel.c
  https://github.com/johnweber/linux/pull/1/files#diff-1 (2)

• M drivers/mxc/gpu-viv/hal/kernel/gc_hal_kernel_command_vg.c
  https://github.com/johnweber/linux/pull/1/files#diff-2 (2)

• M drivers/mxc/gpu-viv/hal/kernel/gc_hal_kernel_video_memory.c
  https://github.com/johnweber/linux/pull/1/files#diff-3 (2)

  Patch Links:
  

• https://github.com/johnweber/linux/pull/1.patch

• https://github.com/johnweber/linux/pull/1.diff

[johnweber]

Stephan - I merged everything into the wandboard_imx_3.0.35_4.0.0 branch from test, and added a commit to fix a problem in the MIPI CSI exclusion code, and reorganized the mipi csi init code. I've tested it on Wandboard Solo so far (with/without) the MIPI included.

[wolfgar]

OK, I switch to wandboard_imx_3.0.35_4.0.0 branch
Sorry for the missing trailing "2" : At least it worked when mipi was disabled ;)

[johnweber]

To add to the difficulty, there is no difference between the MX6 solo and DL in the CPU type detection, so we can't distinguish between WB Dual and WB solo here.

This function is important because it appears to be telling the kernel how much memory it can occupy, similar to the mem= argument. We need a way to tell the difference between S and DL in order to set the memory limit this way. Any ideas?

[wolfgar]

I have solved it that way : wolfgar@f77f226

[johnweber]

I wonder if there is an easier way of getting the memory config, but this works.
I merged this into _test. Thanks!

On 8/3/13 9:06 PM, wolfgar wrote:

I have solved it that way : wolfgar/linux@f77f226
wolfgar@f77f226

—
Reply to this email directly or view it on GitHub
#1 (comment).

[wolfgar]

Hi John,
Regarding memory, this information comes from bootloader atags (or dts if used). It is, to my mind, the best source as it is up to the bootloader to know the amount of memory or to be able to probe it. I am not aware of another way to grab this information at this stage of boot (but obviously, it does not mean that there is no other way...)