Adds 8-, 9- and 10-bit versions of some of the functions used by the h264 decoder.

This patch lets e.g. dsputil_init chose dsp functions with respect to
the bit depth to decode. The naming scheme of bit depth dependent
functions is <base name>_<bit depth>[_<prefix>] (i.e. the old
clear_blocks_c is now named clear_blocks_8_c).

Note: Some of the functions for high bit depth is not dependent on the
bit depth, but only on the pixel size. This leaves some room for
optimizing binary size.

Preparatory patch for high bit depth h264 decoding support.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

6 files changed, 30 insertions, 6 deletions

diff --git a/libavcodec/arm/dsputil_init_arm.c b/libavcodec/arm/dsputil_init_arm.c
index 5c13290..218d162 100644
--- a/libavcodec/arm/dsputil_init_arm.c
+++ b/libavcodec/arm/dsputil_init_arm.c
@@ -75,6 +75,8 @@ static void simple_idct_arm_add(uint8_t *dest, int line_size, DCTELEM *block)
 
 void dsputil_init_arm(DSPContext* c, AVCodecContext *avctx)
 {
+    const int h264_high_depth = avctx->codec_id == CODEC_ID_H264 && avctx->bits_per_raw_sample > 8;
+
     ff_put_pixels_clamped = c->put_pixels_clamped;
     ff_add_pixels_clamped = c->add_pixels_clamped;
 
@@ -95,6 +97,7 @@ void dsputil_init_arm(DSPContext* c, AVCodecContext *avctx)
 
     c->add_pixels_clamped = ff_add_pixels_clamped_arm;
 
+    if (!h264_high_depth) {
     c->put_pixels_tab[0][0] = ff_put_pixels16_arm;
     c->put_pixels_tab[0][1] = ff_put_pixels16_x2_arm;
     c->put_pixels_tab[0][2] = ff_put_pixels16_y2_arm;
@@ -112,6 +115,7 @@ void dsputil_init_arm(DSPContext* c, AVCodecContext *avctx)
     c->put_no_rnd_pixels_tab[1][1] = ff_put_no_rnd_pixels8_x2_arm;
     c->put_no_rnd_pixels_tab[1][2] = ff_put_no_rnd_pixels8_y2_arm;
     c->put_no_rnd_pixels_tab[1][3] = ff_put_no_rnd_pixels8_xy2_arm;
+    }
 
     if (HAVE_ARMV5TE) ff_dsputil_init_armv5te(c, avctx);
     if (HAVE_ARMV6)   ff_dsputil_init_armv6(c, avctx);
diff --git a/libavcodec/arm/dsputil_init_armv6.c b/libavcodec/arm/dsputil_init_armv6.c
index 3209062..fc0f786 100644
--- a/libavcodec/arm/dsputil_init_armv6.c
+++ b/libavcodec/arm/dsputil_init_armv6.c
@@ -72,6 +72,8 @@ int ff_pix_sum_armv6(uint8_t *pix, int line_size);
 
 void av_cold ff_dsputil_init_armv6(DSPContext* c, AVCodecContext *avctx)
 {
+    const int h264_high_depth = avctx->codec_id == CODEC_ID_H264 && avctx->bits_per_raw_sample > 8;
+
     if (!avctx->lowres && (avctx->idct_algo == FF_IDCT_AUTO ||
                            avctx->idct_algo == FF_IDCT_SIMPLEARMV6)) {
         c->idct_put              = ff_simple_idct_put_armv6;
@@ -80,6 +82,7 @@ void av_cold ff_dsputil_init_armv6(DSPContext* c, AVCodecContext *avctx)
         c->idct_permutation_type = FF_LIBMPEG2_IDCT_PERM;
     }
 
+    if (!h264_high_depth) {
     c->put_pixels_tab[0][0] = ff_put_pixels16_armv6;
     c->put_pixels_tab[0][1] = ff_put_pixels16_x2_armv6;
     c->put_pixels_tab[0][2] = ff_put_pixels16_y2_armv6;
@@ -100,6 +103,7 @@ void av_cold ff_dsputil_init_armv6(DSPContext* c, AVCodecContext *avctx)
 
     c->avg_pixels_tab[0][0] = ff_avg_pixels16_armv6;
     c->avg_pixels_tab[1][0] = ff_avg_pixels8_armv6;
+    }
 
     c->add_pixels_clamped = ff_add_pixels_clamped_armv6;
     c->get_pixels = ff_get_pixels_armv6;
diff --git a/libavcodec/arm/dsputil_init_neon.c b/libavcodec/arm/dsputil_init_neon.c
index ef5b63b..9e456f3 100644
--- a/libavcodec/arm/dsputil_init_neon.c
+++ b/libavcodec/arm/dsputil_init_neon.c
@@ -173,6 +173,8 @@ void ff_apply_window_int16_neon(int16_t *dst, const int16_t *src,
 
 void ff_dsputil_init_neon(DSPContext *c, AVCodecContext *avctx)
 {
+    const int h264_high_depth = avctx->codec_id == CODEC_ID_H264 && avctx->bits_per_raw_sample > 8;
+
     if (!avctx->lowres) {
         if (avctx->idct_algo == FF_IDCT_AUTO ||
             avctx->idct_algo == FF_IDCT_SIMPLENEON) {
@@ -190,6 +192,7 @@ void ff_dsputil_init_neon(DSPContext *c, AVCodecContext *avctx)
         }
     }
 
+    if (!h264_high_depth) {
     c->clear_block  = ff_clear_block_neon;
     c->clear_blocks = ff_clear_blocks_neon;
 
@@ -213,12 +216,14 @@ void ff_dsputil_init_neon(DSPContext *c, AVCodecContext *avctx)
 
     c->avg_pixels_tab[0][0] = ff_avg_pixels16_neon;
     c->avg_pixels_tab[1][0] = ff_avg_pixels8_neon;
+    }
 
     c->add_pixels_clamped = ff_add_pixels_clamped_neon;
     c->put_pixels_clamped = ff_put_pixels_clamped_neon;
     c->put_signed_pixels_clamped = ff_put_signed_pixels_clamped_neon;
 
     if (CONFIG_H264_DECODER) {
+        if (!h264_high_depth) {
         c->put_h264_chroma_pixels_tab[0] = ff_put_h264_chroma_mc8_neon;
         c->put_h264_chroma_pixels_tab[1] = ff_put_h264_chroma_mc4_neon;
         c->put_h264_chroma_pixels_tab[2] = ff_put_h264_chroma_mc2_neon;
@@ -294,6 +299,7 @@ void ff_dsputil_init_neon(DSPContext *c, AVCodecContext *avctx)
         c->avg_h264_qpel_pixels_tab[1][13] = ff_avg_h264_qpel8_mc13_neon;
         c->avg_h264_qpel_pixels_tab[1][14] = ff_avg_h264_qpel8_mc23_neon;
         c->avg_h264_qpel_pixels_tab[1][15] = ff_avg_h264_qpel8_mc33_neon;
+        }
     }
 
     if (CONFIG_VP3_DECODER) {
diff --git a/libavcodec/arm/dsputil_iwmmxt.c b/libavcodec/arm/dsputil_iwmmxt.c
index 758d9cb..6db1837 100644
--- a/libavcodec/arm/dsputil_iwmmxt.c
+++ b/libavcodec/arm/dsputil_iwmmxt.c
@@ -155,6 +155,7 @@ static void nop(uint8_t *block, const uint8_t *pixels, int line_size, int h)
 void ff_dsputil_init_iwmmxt(DSPContext* c, AVCodecContext *avctx)
 {
     int mm_flags = AV_CPU_FLAG_IWMMXT; /* multimedia extension flags */
+    const int h264_high_depth = avctx->codec_id == CODEC_ID_H264 && avctx->bits_per_raw_sample > 8;
 
     if (avctx->dsp_mask) {
         if (avctx->dsp_mask & AV_CPU_FLAG_FORCE)
@@ -167,6 +168,7 @@ void ff_dsputil_init_iwmmxt(DSPContext* c, AVCodecContext *avctx)
 
     c->add_pixels_clamped = add_pixels_clamped_iwmmxt;
 
+    if (!h264_high_depth) {
     c->clear_blocks = clear_blocks_iwmmxt;
 
     c->put_pixels_tab[0][0] = put_pixels16_iwmmxt;
@@ -204,4 +206,5 @@ void ff_dsputil_init_iwmmxt(DSPContext* c, AVCodecContext *avctx)
     c->avg_no_rnd_pixels_tab[1][1] = avg_no_rnd_pixels8_x2_iwmmxt;
     c->avg_no_rnd_pixels_tab[1][2] = avg_no_rnd_pixels8_y2_iwmmxt;
     c->avg_no_rnd_pixels_tab[1][3] = avg_no_rnd_pixels8_xy2_iwmmxt;
+    }
 }
diff --git a/libavcodec/arm/h264dsp_init_arm.c b/libavcodec/arm/h264dsp_init_arm.c
index e55a9f4..e914640 100644
--- a/libavcodec/arm/h264dsp_init_arm.c
+++ b/libavcodec/arm/h264dsp_init_arm.c
@@ -92,8 +92,9 @@ void ff_h264_idct8_add4_neon(uint8_t *dst, const int *block_offset,
                              DCTELEM *block, int stride,
                              const uint8_t nnzc[6*8]);
 
-static void ff_h264dsp_init_neon(H264DSPContext *c)
+static void ff_h264dsp_init_neon(H264DSPContext *c, const int bit_depth)
 {
+    if (bit_depth == 8) {
     c->h264_v_loop_filter_luma   = ff_h264_v_loop_filter_luma_neon;
     c->h264_h_loop_filter_luma   = ff_h264_h_loop_filter_luma_neon;
     c->h264_v_loop_filter_chroma = ff_h264_v_loop_filter_chroma_neon;
@@ -125,9 +126,10 @@ static void ff_h264dsp_init_neon(H264DSPContext *c)
     c->h264_idct8_add       = ff_h264_idct8_add_neon;
     c->h264_idct8_dc_add    = ff_h264_idct8_dc_add_neon;
     c->h264_idct8_add4      = ff_h264_idct8_add4_neon;
+    }
 }
 
-void ff_h264dsp_init_arm(H264DSPContext *c)
+void ff_h264dsp_init_arm(H264DSPContext *c, const int bit_depth)
 {
-    if (HAVE_NEON) ff_h264dsp_init_neon(c);
+    if (HAVE_NEON) ff_h264dsp_init_neon(c, bit_depth);
 }
diff --git a/libavcodec/arm/h264pred_init_arm.c b/libavcodec/arm/h264pred_init_arm.c
index 87cc558..142f4b0 100644
--- a/libavcodec/arm/h264pred_init_arm.c
+++ b/libavcodec/arm/h264pred_init_arm.c
@@ -42,8 +42,13 @@ void ff_pred8x8_0lt_dc_neon(uint8_t *src, int stride);
 void ff_pred8x8_l00_dc_neon(uint8_t *src, int stride);
 void ff_pred8x8_0l0_dc_neon(uint8_t *src, int stride);
 
-static void ff_h264_pred_init_neon(H264PredContext *h, int codec_id)
+static void ff_h264_pred_init_neon(H264PredContext *h, int codec_id, const int bit_depth)
 {
+    const int high_depth = bit_depth > 8;
+
+    if (high_depth)
+        return;
+
     h->pred8x8[VERT_PRED8x8     ] = ff_pred8x8_vert_neon;
     h->pred8x8[HOR_PRED8x8      ] = ff_pred8x8_hor_neon;
     if (codec_id != CODEC_ID_VP8)
@@ -69,7 +74,7 @@ static void ff_h264_pred_init_neon(H264PredContext *h, int codec_id)
         h->pred16x16[PLANE_PRED8x8  ] = ff_pred16x16_plane_neon;
 }
 
-void ff_h264_pred_init_arm(H264PredContext *h, int codec_id)
+void ff_h264_pred_init_arm(H264PredContext *h, int codec_id, bit_depth)
 {
-    if (HAVE_NEON)    ff_h264_pred_init_neon(h, codec_id);
+    if (HAVE_NEON)    ff_h264_pred_init_neon(h, codec_id, bit_depth);
 }