BSP update for Marvell Dove kernel LSP 5.1.0

A comparison of all the commits between LSP 5.1.0 and ubuntu kernel shows the following list, I've marked them with '+/-/X', which means:

'+' this commit is missing in ubuntu kernel and will be merged
'-' this commit is either useless in ubuntu kernel or is reverted in the original version and will not be merged
'X' this commit has already been merged in ubuntu kernel

- dove: set version to 5.1.0
+ eth643: fix suspend/resume by deatach/attach phy
+ dove: disable CH7025_COMPOSITE
+ remove dove_avng_x0_defconfig and dove_avng_y1_defconfig
- dove: set version to 5.1.0-rc3
+ enable bmm driver as module in dove_avng_x0_defconfig
+ rt5611: only swap touchscreen Y-axis for ANDROID
+ Fix ADI9889 PM suspend/resume issue
+ refine the comments
+ Remove unused function.
+ Lower wait for vsync's debug message level and double confirm isr before printing the message.
+ Lower debug message's level and mark it
+ Fix warning message when compiling
+ Fix wrong bit operation
+ Added Headset detection feature.
+ fix ADI9889/9389 HDMI audio no sound (SPDIF) issue
+ update dove_avng_x0_defconfig and dove_avng_y1_defconfig
X sound: skip error handling inside soc_pcm_open

+ dove: alway use the galcore name for the gpu driver
- dove: set version to 5.1.0-rc2
+ dove: disable MTD_VERIFY_WRITE and enable MTD test support
+ dove: build dmatest as module
+ Modify dove_gpu's name to galcore for Android to keep align with other Android platform.
+ DVS: Fix compilation error and disable DVS by default for AVD1 board
+ dvs: make changes to dove_rd_avng_setup.c
+ add back rt5630 i2c device in dove-rd-avng-setup.c
+ Fix various settings for new AV-D1 X0 boards
+ Enable two Android PMEM device files: /dev/pmem --> 24M, for Graphics /dev/pmem_adsp --> 64M, for Video
+ Decrease ALSA soc period_size min and max parameter to fix video not fluent on Android issue. Signed-off-by: Steve Guo <email address hidden>
X Swap touchscreen Y axis Signed-off-by: Steve Guo <email address hidden>
+ Disabled some annonying debug information of KG2 due to it flashs console output Signed-off-by: Steve Guo <email address hidden>
+ Fix EDID parsing error in ADI9889 device driver
+ Fix compiling error on Android caused by the patch d4cc16c82748d07258b3808d1f572d69640526bc
+ Disable ADI9889 by default to remove this error during kernel booting: "ERROR at adi9889_error_callback - error code 14"
+ Update standby mode support for AV-D1 X0 board.
X disable some andriod features in dove_avng_x0_defconfig and dove_avng_y1_defconfig
X DVS: Fixed a potential conflict of defining dvs_enable as global for two diffrent baords Signed-off-by: Maen Suleiman <email address hidden>
+ config: small change to align with Ubuntu kernel configuration

- dove: set version to 5.1.0-rc1

X UBUNTU: sub-makefile invocation should use $SHELL and standardise script
X UBUNTU: SAUCE: Enable CPUFREQ by default

- dove: set version to 5.0.0
X UBUNTU: SAUCE: dove: disable conflicting v6 cache fns in zImage head.S

X arm: disable L2 cache in the v7 finish function
X arm: invalidate TLBs when enabling mmu
X UBUNTU: SAUCE: arm: Fix init_atags_procfs() to check tag->hdr.size
X ARM: 5885/1: a...

Merged commits has been pushed to

http://kernel.ubuntu.com/git?p=ycmiao/ubuntu-lucid.git;a=shortlog;h=refs/heads/lp551491

A testing kernel package is built and placed at:

http://people.canonical.com/~ycmiao/linux-image-2.6.32-202-dove_2.6.32-202.14~lp551491_armel.deb

This kernel fails to resume from suspend, still testing...

Despite the log entry, headphone detection does not work on the development boards with Analog Devices AD1980 (AD1888) sound chips. The patches are for a different codec. See bug #551249.

Paul, is galcore.ko inserted or not? This module doesn't support suspend/resume in this version.

Tobin, Paul, could you please check the U-Boot version that you were having the suspend/resume issue? U-Boot is now suspected.

It should be 4.4.0 by default. Eric, the DB you are using is 4.4.2 just because I upgraded it.

I'm running on 4.4.1 test build that I got to try to correct a different
problem I was seeing previously. We have 4.4.2, but there are several
files in the package, can you tell me which one to use for the update?

I'm not sure what's in your uboot package (may include various binaries). Please use the one like u-boot-db88ap510bp_hX00_spi_2CS_gang.bin, which boot from SPI NOR and support 2CS (1GB DDR).

BTW, just make sure you don't have 'no_console_suspend' on the command line, that will cause resume failure at the moment.

Paul & Tobin,

Is it possible to try the kernel in -beta1 to see if it's a regression on your boards, or it might be indeed a HW issue? Thanks.

suspend/resume works fine for me with the kernel in beta1 ( 2.6.32-203.15)

suspend/resume works fine for me with the kernel 2.6.32-204-dove_2.6.32-204.16

Confirming that suspend/resume still works fine for me in the 20100419.1 rc candidate image

Ok, I may have spoken too soon. The first suspend/resumes I did worked fine, but since then I'm now unable to get it to resume after suspending. The only thing I've really done that I can think of, is enable audio. Also, I tried enabling serial console after noticing problems with suspend/resume and I'm not getting any output when I try to resume so far.

I am not seeing these issues with multiple suspend/resume. I have been running a script to automated suspend/resume and also manually tested syspend/resume 10 times without failure. The only issue I have noticed is that I need to restart pulseaudio after resume in order to play audio files through rhythmbox.

updated my uboot to 4.4.2 and so far I have been unable to reproduce the suspend/resume failures on this board.

Now since the suspend/resume regression is no longer there after u-boot updated, I'll close this bug. As the BSP update has already been merged and uploaded, change the status to "Fix Released".