On 01/15/2013 06:50 PM, Lucas Stach wrote:
Am Dienstag, den 15.01.2013, 17:53 +0800 schrieb Mark Zhang:
On 01/15/2013 12:05 AM, Thierry Reding wrote:
Add support for the B and C planes which support RGB and YUV pixel formats and can be used as overlays or hardware cursor.
I think "hardware cursor" has specific meaning for Tegra(e.g: Tegra30 has a 32x32 24bpp or 64x64 2bpp hardware cursor). So you may change it to "hardware accelerated cursor"?
According to the TRM no Tegra has ARGB hardware cursor support, but only 2-color. So we talked about doing the hardware cursor by using a plane. If the TRM is wrong in this regard and we can get a ARGB cursor on Tegra 3 it would be nice to know.
Lucas, yes, TRM says "Hardware cursor is supported for 32x32 or for 64x64 2-bpp cursor.", but just as you can see, we can set cursor's foreground & background color by register "DC_DISP_CURSOR_FOREGROUND_0 " & "DC_DISP_CURSOR_BACKGROUND_0".
So I asked the expert in nvidia and here is the explanation of the hardware cursor:
"each pixel in the cursor is encoded by 2 bits. only 3 values are used per pixel: transparent, foreground, background.
when pixel is transparent - no pixel is displayed. (also known as a mask) when pixel is foreground - color of pixel is 24-bit value in DC_DISP_CURSOR_FOREGROUND_0. when pixel is background - color of pixel is 24-bit value in DC_DISP_CURSOR_BACKGROUND_0.
So I would still phrase it as a 2-bit cursor. It's a palette with 2 colors plus a 1-bit alpha. The palette entries are 24-bit."
Mark
Signed-off-by: Thierry Reding thierry.reding@avionic-design.de
[...]
+};
+static int tegra_dc_add_planes(struct drm_device *drm, struct tegra_dc *dc) +{
- unsigned int i;
- int err = 0;
- for (i = 0; i < 2; i++) {
struct tegra_plane *plane;plane = devm_kzalloc(drm->dev, sizeof(*plane), GFP_KERNEL);if (!plane)return -ENOMEM;plane->index = i;I suggest to change this line to: "plane->index = i + 1;". This makes the plane's index be consistent with Tegra's windows number. And also we don't need to worry about passing "plane->index + 1" to some functions which need to know which window is operating on.
Again, if we make WIN_C the root window, we can keep the plane index assignment as is and get rid of the "index + 1" passing.