Difference between revisions of "Panda How to kernel 3 0 rcx"

From eLinux.org
Jump to: navigation, search
Line 29: Line 29:
  
 
The oops occurs when the EHCI driver tries to access the hardware which has no power.
 
The oops occurs when the EHCI driver tries to access the hardware which has no power.
 +
 +
There is also an issue with the setting up of the wl12xx:
 +
 +
[    0.181457] error setting wl12xx data
 +
 +
Which is a problem new to 3.0-rc1, still present in linux-omap-2.6.git.
  
  

Revision as of 08:34, 1 June 2011

Introduction

The 3.0-rc1 kernel has been released.

rc1

There have been regressions since 2.6.39.

Building

Building 3.0-rc1 is basically the same as How to build 2.6.38 kernel, but of course you need the -rc1 sources. The omap2plus_defconfig works once the mmc ptch is applied. Enabling USB results in very ugly kernel oopses.

MMC Patch

A patch is available which addresses the mmc issue noted above:

0002-omap4-pandaboard-fix-mmc-support.patch

Apply it like so: (from inside the kernel sources directory)

patch -p1 < 0002-omap4-pandaboard-fix-mmc-support.patch


This issue has been fixed in linux-omap-2.6.git, but the oopses are still present when EHCI is enabled in .config.

It looks like they are related to VUSB power not being set up correctly as evidenced by:

[    0.243896] machine_constraints_voltage: VUSB: failed to apply 3300000uV constraint
[    0.245208] twl_reg twl_reg.46: can't register VUSB, -22
[    0.245239] twl_reg: probe of twl_reg.46 failed with error -22

which occurs after the board-omap4panda.c file init's the various regulators, and happens whether the EHCI driver is selected in the .config or not.

The oops occurs when the EHCI driver tries to access the hardware which has no power.

There is also an issue with the setting up of the wl12xx:

[    0.181457] error setting wl12xx data

Which is a problem new to 3.0-rc1, still present in linux-omap-2.6.git.


DVI Patch

Apply it like so: (from inside the kernel sources directory)

patch -p1 < 0001-omap4-pandaboard-fix-dvi-support.patch

Then use either of the following configs

make ARCH=arm omap2plus_defconfig
        or
make ARCH=arm panda_dvi_defconfig

Then compile with

make ARCH=arm CROSS_COMPILE=Path_to_your/arm-2009q3/bin/arm-none-linux-gnueabi- uImage
     or
make ARCH=arm CROSS_COMPILE=Path_to_your/arm-2010q1/bin/arm-none-linux-gnueabi- uImage


i2c character device driver

Ever since 2.6.38, the i2c character device driver support from user space has been broken for OMAP44xx.
* A fix has been submitted for the linux-omap-2.6 branch, but it may be a while till it gets into an -rcx.
See http://git.kernel.org/?p=linux/kernel/git/tmlind/linux-omap-2.6.git;a=commit;h=39fe1a6fafe1e85c183379af9f3ceda7cd24bd65 for the commit.
* A quick fix for this issue is available File:Omap44xx-i2c-fix.patch

I2C Patch

Apply it like so: (from inside the kernel sources directory)

patch -p1 < Omap44xx-i2c-fix.patch

It applies with some fuzzes and offsets, but does apply successfuly. Then compile as above.


Testing: No testing possible yet till the booting problems are resolved

After booting run fbtest to see a nice test pattern from the dvi port.

Fbtest.jpg

Then you can run i2cdetect and the results should look like this:

# i2cdetect -y -r 1
    0  1  2  3  4  5  6  7  8  9  a  b  c  d  e  f
00:          -- -- -- -- -- -- -- -- -- -- -- -- -- 
10: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- 
20: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- 
30: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- 
40: -- -- -- -- -- -- -- -- UU UU UU UU -- -- -- -- 
50: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- 
60: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- 
70: -- -- -- -- -- -- -- --


Stay tuned as the 3.0-rcx series evolves, where there needs to be some serious bug squashing.