hevc: Add support for bitdepth 10 for IDCT DC

Signed-off-by: Martin Storsjö <martin@martin.st>

libavcodec/arm/hevc_idct.S

@@ -30,26 +30,29 @@ const trans, align=4
         .short 57, 43, 25, 9
 endconst
 
-function ff_hevc_idct_4x4_dc_8_neon, export=1
+.macro idct_4x4_dc bitdepth
+function ff_hevc_idct_4x4_dc_\bitdepth\()_neon, export=1
         ldrsh           r1, [r0]
-        ldr             r2, =0x20
+        ldr             r2, =(1 << (13 - \bitdepth))
         add             r1, #1
         asr             r1, #1
         add             r1, r2
-        asr             r1, #6
+        asr             r1, #(14 - \bitdepth)
         vdup.16         q0, r1
         vdup.16         q1, r1
         vst1.16         {q0, q1}, [r0, :128]
         bx              lr
 endfunc
+.endm
 
-function ff_hevc_idct_8x8_dc_8_neon, export=1
+.macro idct_8x8_dc bitdepth
+function ff_hevc_idct_8x8_dc_\bitdepth\()_neon, export=1
         ldrsh           r1, [r0]
-        ldr             r2, =0x20
+        ldr             r2, =(1 << (13 - \bitdepth))
         add             r1, #1
         asr             r1, #1
         add             r1, r2
-        asr             r1, #6
+        asr             r1, #(14 - \bitdepth)
         vdup.16         q8, r1
         vdup.16         q9, r1
         vmov.16         q10, q8
@@ -61,14 +64,16 @@ function ff_hevc_idct_8x8_dc_8_neon, export=1
         vstm            r0, {q8-q15}
         bx              lr
 endfunc
+.endm
 
-function ff_hevc_idct_16x16_dc_8_neon, export=1
+.macro idct_16x16_dc bitdepth
+function ff_hevc_idct_16x16_dc_\bitdepth\()_neon, export=1
         ldrsh           r1, [r0]
-        ldr             r2, =0x20
+        ldr             r2, =(1 << (13 - \bitdepth))
         add             r1, #1
         asr             r1, #1
         add             r1, r2
-        asr             r1, #6
+        asr             r1, #(14 - \bitdepth)
         vdup.16         q8, r1
         vdup.16         q9, r1
         vmov.16         q10, q8
@@ -83,14 +88,16 @@ function ff_hevc_idct_16x16_dc_8_neon, export=1
         vstm            r0, {q8-q15}
         bx              lr
 endfunc
+.endm
 
-function ff_hevc_idct_32x32_dc_8_neon, export=1
+.macro idct_32x32_dc bitdepth
+function ff_hevc_idct_32x32_dc_\bitdepth\()_neon, export=1
         ldrsh           r1, [r0]
-        ldr             r2, =0x20
+        ldr             r2, =(1 << (13 - \bitdepth))
         add             r1, #1
         asr             r1, #1
         add             r1, r2
-        asr             r1, #6
+        asr             r1, #(14 - \bitdepth)
         mov             r3, #16
         vdup.16         q8, r1
         vdup.16         q9, r1
@@ -105,6 +112,7 @@ function ff_hevc_idct_32x32_dc_8_neon, export=1
         bne             1b
         bx              lr
 endfunc
+.endm
 
 .macro sum_sub out, in, c, op
   .ifc \op, +
@@ -496,8 +504,16 @@ tr_16x4 secondpass_10, 20 - 10
 .ltorg
 
 idct_4x4 8
+idct_4x4_dc 8
 idct_4x4 10
+idct_4x4_dc 10
 idct_8x8 8
+idct_8x8_dc 8
 idct_8x8 10
+idct_8x8_dc 10
 idct_16x16 8
+idct_16x16_dc 8
 idct_16x16 10
+idct_16x16_dc 10
+idct_32x32_dc 8
+idct_32x32_dc 10

libavcodec/arm/hevcdsp_init_arm.c

@@ -29,6 +29,10 @@ void ff_hevc_idct_4x4_dc_8_neon(int16_t *coeffs);
 void ff_hevc_idct_8x8_dc_8_neon(int16_t *coeffs);
 void ff_hevc_idct_16x16_dc_8_neon(int16_t *coeffs);
 void ff_hevc_idct_32x32_dc_8_neon(int16_t *coeffs);
+void ff_hevc_idct_4x4_dc_10_neon(int16_t *coeffs);
+void ff_hevc_idct_8x8_dc_10_neon(int16_t *coeffs);
+void ff_hevc_idct_16x16_dc_10_neon(int16_t *coeffs);
+void ff_hevc_idct_32x32_dc_10_neon(int16_t *coeffs);
 
 void ff_hevc_idct_4x4_8_neon(int16_t *coeffs, int col_limit);
 void ff_hevc_idct_8x8_8_neon(int16_t *coeffs, int col_limit);
@@ -53,6 +57,11 @@ av_cold void ff_hevc_dsp_init_arm(HEVCDSPContext *c, int bit_depth)
             c->idct[2] = ff_hevc_idct_16x16_8_neon;
         }
         if (bit_depth == 10) {
+            c->idct_dc[0] = ff_hevc_idct_4x4_dc_10_neon;
+            c->idct_dc[1] = ff_hevc_idct_8x8_dc_10_neon;
+            c->idct_dc[2] = ff_hevc_idct_16x16_dc_10_neon;
+            c->idct_dc[3] = ff_hevc_idct_32x32_dc_10_neon;
+
             c->idct[0] = ff_hevc_idct_4x4_10_neon;
             c->idct[1] = ff_hevc_idct_8x8_10_neon;
             c->idct[2] = ff_hevc_idct_16x16_10_neon;