diff options
| author | Robert Morell <rmorell@nvidia.com> | 2011-07-15 18:47:06 -0700 |
|---|---|---|
| committer | Varun Colbert <vcolbert@nvidia.com> | 2011-08-29 17:49:50 -0700 |
| commit | d4f30a049a1d97b350c415c3f40d5fa57724c8d4 (patch) | |
| tree | c44eb2954929f2f0e1587d0f6db92d78c6bbacb0 /arch | |
| parent | 9ce011627ad7255153cf78d2c7e5b09a21d16101 (diff) | |
video: tegra: dc: Separate allocations for U and V
Currently, dc_ext only takes a single nvmap memory ID per overlay, even
in the YUV case (the U and V planes are expected to be differentiated
using an offset from the beginning of the nvmap allocation).
This is problematic for some software flows, such as certain video
interlacing algorithms that will vary the luma plane while keeping the
chrome plane constant.
This change allows dc_ext clients to specify a different nvmap
allocation for each of the Y, U, and V planes. If a YUV surface is
used and no U or V plane allocation is specified, the old behavior is
preserved: the U and V offsets are assumed to be within the same
allocation as Y.
Note: this changes the behavior of the offset parameter: the old code
added offset to offset_u and offset_v when using it. The new code
treats all three offsets as relative to the beginning of the allocation.
It also fixes a bug in the code where offset was applied twice to the Y
plane. I believe this is safe because the presence of this bug means
that no existing clients are using offset != 0 (or if they are, they're
already broken).
Signed-off-by: Robert Morell <rmorell@nvidia.com>
Bug 850882
Change-Id: I230e03db25baaae73a3bdc0d45a2aec162b87fa4
Reviewed-on: http://git-master/r/41471
Reviewed-by: Varun Colbert <vcolbert@nvidia.com>
Tested-by: Varun Colbert <vcolbert@nvidia.com>
Diffstat (limited to 'arch')
| -rw-r--r-- | arch/arm/mach-tegra/include/mach/dc.h | 5 |
1 files changed, 2 insertions, 3 deletions
diff --git a/arch/arm/mach-tegra/include/mach/dc.h b/arch/arm/mach-tegra/include/mach/dc.h index 397e9d147721..d291f29268d7 100644 --- a/arch/arm/mach-tegra/include/mach/dc.h +++ b/arch/arm/mach-tegra/include/mach/dc.h @@ -370,9 +370,8 @@ struct tegra_dc_win { void *virt_addr; dma_addr_t phys_addr; - unsigned offset; - unsigned offset_u; - unsigned offset_v; + dma_addr_t phys_addr_u; + dma_addr_t phys_addr_v; unsigned stride; unsigned stride_uv; fixed20_12 x; |