From 6e4009d4cdf5927bdaedf58fcfc5e813b14c366b Mon Sep 17 00:00:00 2001
From: Janne Grunau <janne-libav@jannau.net>
Date: Sat, 22 Feb 2014 18:27:10 +0100
Subject: [PATCH] arm: dcadsp: implement decode_hf as external NEON asm

---
 libavcodec/arm/dcadsp_init_arm.c |  7 +++++++
 libavcodec/arm/dcadsp_neon.S     | 29 +++++++++++++++++++++++++++++
 2 files changed, 36 insertions(+)

diff --git a/libavcodec/arm/dcadsp_init_arm.c b/libavcodec/arm/dcadsp_init_arm.c
index d91c787d79..540048415f 100644
--- a/libavcodec/arm/dcadsp_init_arm.c
+++ b/libavcodec/arm/dcadsp_init_arm.c
@@ -49,6 +49,12 @@ void ff_synth_filter_float_neon(FFTContext *imdct,
                                 float out[32], const float in[32],
                                 float scale);
 
+void ff_decode_hf_neon(float dst[DCA_SUBBANDS][8],
+                       const int32_t vq_num[DCA_SUBBANDS],
+                       const int8_t hf_vq[1024][32], intptr_t vq_offset,
+                       int32_t scale[DCA_SUBBANDS][2],
+                       intptr_t start, intptr_t end);
+
 av_cold void ff_dcadsp_init_arm(DCADSPContext *s)
 {
     int cpu_flags = av_get_cpu_flags();
@@ -61,6 +67,7 @@ av_cold void ff_dcadsp_init_arm(DCADSPContext *s)
     if (have_neon(cpu_flags)) {
         s->lfe_fir[0] = ff_dca_lfe_fir0_neon;
         s->lfe_fir[1] = ff_dca_lfe_fir1_neon;
+        s->decode_hf  = ff_decode_hf_neon;
     }
 }
 
diff --git a/libavcodec/arm/dcadsp_neon.S b/libavcodec/arm/dcadsp_neon.S
index 735c4c28e5..70580cdeec 100644
--- a/libavcodec/arm/dcadsp_neon.S
+++ b/libavcodec/arm/dcadsp_neon.S
@@ -20,6 +20,35 @@
 
 #include "libavutil/arm/asm.S"
 
+function ff_decode_hf_neon, export=1
+        push            {r4-r5,lr}
+        add             r2,  r2,  r3
+        ldr             r3,       [sp, #12]
+        ldrd            r4,  r5,  [sp, #16]
+        add             r3,  r3,  r4, lsl #3
+        add             r1,  r1,  r4, lsl #2
+        add             r0,  r0,  r4, lsl #5
+
+1:      ldr_post        lr,  r1,  #4
+        add             r4,  r4,  #1
+        add             lr,  r2,  lr, lsl #5
+        cmp             r4,  r5
+        vld1.32         {d7},     [r3]!
+        vld1.8          {d0},     [lr,:64]
+        vcvt.f32.s32    d7,  d7,  #4
+        vmovl.s8        q1,  d0
+        vmovl.s16       q0,  d2
+        vmovl.s16       q1,  d3
+        vcvt.f32.s32    q0,  q0
+        vcvt.f32.s32    q1,  q1
+        vmul.f32        q0,  q0,  d7[0]
+        vmul.f32        q1,  q1,  d7[0]
+        vst1.32         {q0-q1},  [r0,:128]!
+        bne             1b
+
+        pop             {r4-r5,pc}
+endfunc
+
 function ff_dca_lfe_fir0_neon, export=1
         push            {r4-r6,lr}
         mov             r3,  #32                @ decifactor