From 284ea790d89441fa1e6b2d72d3c1ed6d61972f0b Mon Sep 17 00:00:00 2001
From: Justin Ruggles <justin.ruggles@gmail.com>
Date: Sat, 22 Sep 2012 18:13:57 -0400
Subject: dsputil: move vector_fmul_scalar() to AVFloatDSPContext in libavutil

---
 libavutil/arm/float_dsp_init_neon.c |  4 ++++
 libavutil/arm/float_dsp_neon.S      | 38 +++++++++++++++++++++++++++++++++++++
 2 files changed, 42 insertions(+)

(limited to 'libavutil/arm')

diff --git a/libavutil/arm/float_dsp_init_neon.c b/libavutil/arm/float_dsp_init_neon.c
index 3ca0288b31..88eb4b3d2a 100644
--- a/libavutil/arm/float_dsp_init_neon.c
+++ b/libavutil/arm/float_dsp_init_neon.c
@@ -29,8 +29,12 @@ void ff_vector_fmul_neon(float *dst, const float *src0, const float *src1, int l
 void ff_vector_fmac_scalar_neon(float *dst, const float *src, float mul,
                                 int len);
 
+void ff_vector_fmul_scalar_neon(float *dst, const float *src, float mul,
+                                int len);
+
 void ff_float_dsp_init_neon(AVFloatDSPContext *fdsp)
 {
     fdsp->vector_fmul = ff_vector_fmul_neon;
     fdsp->vector_fmac_scalar = ff_vector_fmac_scalar_neon;
+    fdsp->vector_fmul_scalar = ff_vector_fmul_scalar_neon;
 }
diff --git a/libavutil/arm/float_dsp_neon.S b/libavutil/arm/float_dsp_neon.S
index 4aa6f838dd..6d7bd5236e 100644
--- a/libavutil/arm/float_dsp_neon.S
+++ b/libavutil/arm/float_dsp_neon.S
@@ -108,3 +108,41 @@ NOVFP   vdup.32         q15, r2
         bx              lr
         .unreq          len
 endfunc
+
+function ff_vector_fmul_scalar_neon, export=1
+VFP     len .req r2
+NOVFP   len .req r3
+VFP     vdup.32         q8,  d0[0]
+NOVFP   vdup.32         q8,  r2
+        bics            r12, len, #15
+        beq             3f
+        vld1.32         {q0},[r1,:128]!
+        vld1.32         {q1},[r1,:128]!
+1:      vmul.f32        q0,  q0,  q8
+        vld1.32         {q2},[r1,:128]!
+        vmul.f32        q1,  q1,  q8
+        vld1.32         {q3},[r1,:128]!
+        vmul.f32        q2,  q2,  q8
+        vst1.32         {q0},[r0,:128]!
+        vmul.f32        q3,  q3,  q8
+        vst1.32         {q1},[r0,:128]!
+        subs            r12, r12, #16
+        beq             2f
+        vld1.32         {q0},[r1,:128]!
+        vst1.32         {q2},[r0,:128]!
+        vld1.32         {q1},[r1,:128]!
+        vst1.32         {q3},[r0,:128]!
+        b               1b
+2:      vst1.32         {q2},[r0,:128]!
+        vst1.32         {q3},[r0,:128]!
+        ands            len, len, #15
+        it              eq
+        bxeq            lr
+3:      vld1.32         {q0},[r1,:128]!
+        vmul.f32        q0,  q0,  q8
+        vst1.32         {q0},[r0,:128]!
+        subs            len, len, #4
+        bgt             3b
+        bx              lr
+        .unreq          len
+endfunc
-- 
cgit v1.2.3