diff options
| author | Martin Storsjö <martin@martin.st> | 2021-05-13 09:33:21 +0300 |
|---|---|---|
| committer | Jean-Baptiste Kempf <jb@videolan.org> | 2021-05-13 11:49:02 +0300 |
| commit | 3aac025204602810c5bf33cbad6ac1bf157487cc (patch) | |
| tree | 54bcb2f2c39f7e4d0377f51c5bdbc952ed89a214 /src/arm | |
| parent | 1cf1b309bb2c40e25582f2ace6917d84e45f354b (diff) | |
arm64: filmgrain16: Guard against out of range pixels in the gather function
In 16 bpc, the pixels are 16 bit integers, but valid pixels only
are up to 12 bits, and the scaling buffer only contains 4096
elements.
The src pixels are, normally, supposed to be valid pixels, but when
processing blocks of 32 pixels at a time, it can operate on
uninitialized pixels past the right edge.
Before: Cortex A53 A72 A73 Apple M1
fgy_32x32xn_16bpc_neon: 10372.5 8194.4 8612.1 24.2
After:
fgy_32x32xn_16bpc_neon: 10837.9 8469.5 8885.1 24.6
Diffstat (limited to 'src/arm')
| -rw-r--r-- | src/arm/64/film_grain16.S | 8 |
1 files changed, 8 insertions, 0 deletions
diff --git a/src/arm/64/film_grain16.S b/src/arm/64/film_grain16.S index be40388..a72164c 100644 --- a/src/arm/64/film_grain16.S +++ b/src/arm/64/film_grain16.S @@ -188,6 +188,7 @@ function fgy_loop_neon .macro fgy ox, oy L(loop_\ox\oy): 1: + mov w16, #0xfff ld1 {v0.8h, v1.8h, v2.8h, v3.8h}, [x1], x2 // src .if \ox ld1 {v20.4h}, [x4], x9 // grain_lut old @@ -198,8 +199,15 @@ L(loop_\ox\oy): .if \ox && \oy ld1 {v14.4h}, [x8], x9 // grain_lut top old .endif + dup v4.8h, w16 ld1 {v16.8h, v17.8h, v18.8h, v19.8h}, [x5], x9 // grain_lut + // Make sure that uninitialized pixels out of range past the right + // edge are in range; their actual values shouldn't matter. + and v0.16b, v0.16b, v4.16b + and v1.16b, v1.16b, v4.16b + and v2.16b, v2.16b, v4.16b + and v3.16b, v3.16b, v4.16b bl gather_neon .if \ox |