LG monitor behaving incorrectly when used in conjunction with the Panda board and HDMI

[Lee Jones]

The OMAP4 pre-installed image does not display on the LG W2261VP monitor via HDMI.

The image tested was here:
    http://cdimage.ubuntu.com/ubuntu-netbook/ports/daily-preinstalled/20100713/

-------------------------------------------------------------------------------------------------------------------------

When using a bare kernel and rootfs from rootstock with the following kernel cmdline:
    setenv bootargs vram=32M, omapfb.vram=0:8M console=tty2 console=ttyO2,115200n8 \
    noinitrd mem=463M root=/dev/mmcblk0p2 rootwait ip=none; mmcinit 0; fatload mmc 0 \
    0x80200000 uImage; bootm 0x80200000

When the monitor is -
    _on_ : Monitor enters Power Saving Mode (PSM)
   _PSM_ : Monitor comes out of, then re-enters PSM

When adding the kernel cmdline arguments:
    omapdss.hdmimode=0 omapdss.hdmicode=4

The exact same behaviour is displayed.

-------------------------------------------------------------------------------------------------------------------------

Some days after reporting this behaviour to the OEM it was reported back to me that there were
seemingly issues with long kernel cmdlines. So I chopped mine down and issued the following:
    setenv bootargs omapdss.hdmimode=0 omapdss.hdmicode=4 vram=32M, omapfb.vram=0:8M \
    console=tty2; mmcinit 0; fatload mmc 0 0x80200000 uImage; bootm 0x80200000

This provided some relic shakey video mode (640 x 480):
    4(0x4) {640, 480, 25175, 96, 16, 48, 2 , 11, 31}

Any other mode I tried on the kernel cmdline was either shakey and produced 'out of range'
errors or placed the monitor straight into Power Saving Mode.

-------------------------------------------------------------------------------------------------------------------------

Next I found out the driver used sysfs, so I decided to use it.

I issued the following command:
    echo 4 | sudo tee -a /sys/devices/platform/omapdss/display0/custom_edid_timing

This worked seamlessly.

-------------------------------------------------------------------------------------------------------------------------

So how do we make this work with the CD image, which has no display related cmdline args?

The OEM and I are currently working to solve this issue.

[Lee Jones (lag)]

Placing 'omapdss.debug=1' on the kernel cmdline provides lots of useful messages which can be used to debug the issue further.

It seems the values the monitor provides via its EDID are non-standard, thus confusing the DSS HDMI driver.

The monitor is providing the following values:
    1920 1080 setting with 138.5 MHz

[Lee Jones (lag)]

This bug also affects Philips 244E.

[Ricardo Salveti (rsalveti)]

The fixes are all available already.

Currently I'm maintaining them at my kernel tree, to make our image work with my LG monitor:
http://kernel.ubuntu.com/git?p=rsalveti/ubuntu-maverick.git;a=shortlog;h=refs/heads/rsalveti-ti-omap4

You can find the deb file at:
http://people.canonical.com/~rsalveti/maverick/kernel/linux-image-2.6.34-903-omap4_2.6.34-903.7rsalveti1_armel.deb

For next 2.6.35 kernel you can also see that the fixes are applied already:
http://kernel.ubuntu.com/git?p=roc/ubuntu-maverick.git;a=shortlog;h=refs/heads/ti-omap4

[Lee Jones (lag)]

I'm well aware that the fixes are available. I was actively involved in their production. However, we are waiting for our kernel to be rebased onto L24.9. If we persist in segregating fixes from the main release, the rebase will become more difficult.

The short answer is; I'm afraid you'll have to be patient and wait.

Out of interest does this issue still occur with ES2.0 and ES2.1?

[Ricardo Salveti (rsalveti)]

Sure, I was just documenting the issue by pointing the patches.

Tim Gardner uploaded the new tree yesterday, that contains all the fixes for this issue. The only problem is that it should be tested only with ES2.X boards.

The link:
http://kernel.ubuntu.com/git?p=ubuntu/ubuntu-maverick.git;a=shortlog;h=refs/heads/ti-omap4