On Thu, 2008-11-06 at 01:21 +0200, ext David Brownell wrote:

On Wednesday 05 November 2008, Jarkko Nikula wrote:

quoted

quoted

+     r = gpio_request(beagle_display_data_dvi.panel_reset_gpio,

"DVI reset GPIO");

quoted

quoted

+     if(r < 0) {

GPIO label "DVI reset" is enough.

Surely this ** doesn't work ** at least against the OMAP git tree
which already has:

Well, it surely does! =)

But good catch, I'll fix it. There's probably a warning on kernel log
about "can't get a gpio".

        omap_cfg_reg(J25_34XX_GPIO170);
        gpio_request(170, "DVI_nPD");
        /* REVISIT leave DVI powered down until it's needed ... */
        gpio_direction_output(170, true);

Label matches the DVI chip, not board schematics.  Agreed,
it's not a reset.

I'd like to see any FB/DSS/... updates have the property that
the TFP410 chip is powered down *unless* both (a) the frame
buffer/DSS/... driver is active, and (b) the DVI channel is
in use.

So for example a kernel without FB/DSS/... configured, that
DVI chip is never powered up.  And when only the S-Video channel
is used, DVI isn't powered up either.

Well, my drivers do not enable the DVI chip unless DVI is in use. It is
in use by default, but you can turn it off. And if the whole DSS is not
compiled in, then no body will turn the DVI chip on.

However, the bootloader turns the DVI chip on (at least the default
beagle board u-boot). If the kernel by default turns the DVI chip off,
and then later on when the DSS starts, we'll have a glitch on the
display. Not that there currently is support for glitchless transition
from bootloader display to kernel display, but still...

- Dave

 Tomi