dsputil: Move APE-specific bits into apedsp

5 files changed, 102 insertions, 45 deletions

diff --git a/libavcodec/arm/Makefile b/libavcodec/arm/Makefile
index 9d5b6aab5b..13025af9c1 100644
--- a/libavcodec/arm/Makefile
+++ b/libavcodec/arm/Makefile
@@ -24,6 +24,7 @@ OBJS-$(CONFIG_VP3DSP)                  += arm/vp3dsp_init_arm.o
 
 OBJS-$(CONFIG_AAC_DECODER)             += arm/aacpsdsp_init_arm.o       \
                                           arm/sbrdsp_init_arm.o
+OBJS-$(CONFIG_APE_DECODER)             += arm/apedsp_init_arm.o
 OBJS-$(CONFIG_DCA_DECODER)             += arm/dcadsp_init_arm.o
 OBJS-$(CONFIG_FLAC_DECODER)            += arm/flacdsp_init_arm.o        \
                                           arm/flacdsp_arm.o
@@ -97,6 +98,7 @@ NEON-OBJS-$(CONFIG_VP3DSP)             += arm/vp3dsp_neon.o
 
 NEON-OBJS-$(CONFIG_AAC_DECODER)        += arm/aacpsdsp_neon.o           \
                                           arm/sbrdsp_neon.o
+NEON-OBJS-$(CONFIG_APE_DECODER)        += arm/apedsp_neon.o
 NEON-OBJS-$(CONFIG_DCA_DECODER)        += arm/dcadsp_neon.o             \
                                           arm/synth_filter_neon.o
 NEON-OBJS-$(CONFIG_RV30_DECODER)       += arm/rv34dsp_neon.o
diff --git a/libavcodec/arm/apedsp_init_arm.c b/libavcodec/arm/apedsp_init_arm.c
new file mode 100644
index 0000000000..47ea034359
--- /dev/null
+++ b/libavcodec/arm/apedsp_init_arm.c
@@ -0,0 +1,38 @@
+/*
+ * Copyright (c) 2011 Mans Rullgard <mans@mansr.com>
+ *
+ * This file is part of Libav.
+ *
+ * Libav is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * Libav is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with Libav; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include <stdint.h>
+
+#include "libavutil/attributes.h"
+#include "libavutil/cpu.h"
+#include "libavutil/arm/cpu.h"
+#include "libavcodec/apedsp.h"
+
+int32_t ff_scalarproduct_and_madd_int16_neon(int16_t *v1, const int16_t *v2,
+                                             const int16_t *v3, int len, int mul);
+
+av_cold void ff_apedsp_init_arm(APEDSPContext *c)
+{
+    int cpu_flags = av_get_cpu_flags();
+
+    if (have_neon(cpu_flags)) {
+        c->scalarproduct_and_madd_int16 = ff_scalarproduct_and_madd_int16_neon;
+    }
+}
diff --git a/libavcodec/arm/apedsp_neon.S b/libavcodec/arm/apedsp_neon.S
new file mode 100644
index 0000000000..7cfbf43c6d
--- /dev/null
+++ b/libavcodec/arm/apedsp_neon.S
@@ -0,0 +1,62 @@
+/*
+ * ARM NEON optimised integer operations
+ * Copyright (c) 2009 Kostya Shishkov
+ *
+ * This file is part of Libav.
+ *
+ * Libav is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * Libav is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with Libav; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include "libavutil/arm/asm.S"
+
+@ scalarproduct_and_madd_int16(/*aligned*/v0,v1,v2,order,mul)
+function ff_scalarproduct_and_madd_int16_neon, export=1
+        vld1.16         {d28[],d29[]}, [sp]
+        vmov.i16        q0,  #0
+        vmov.i16        q1,  #0
+        vmov.i16        q2,  #0
+        vmov.i16        q3,  #0
+        mov             r12, r0
+
+1:      vld1.16         {d16-d17}, [r0,:128]!
+        vld1.16         {d18-d19}, [r1]!
+        vld1.16         {d20-d21}, [r2]!
+        vld1.16         {d22-d23}, [r0,:128]!
+        vld1.16         {d24-d25}, [r1]!
+        vld1.16         {d26-d27}, [r2]!
+        vmul.s16        q10, q10,  q14
+        vmul.s16        q13, q13,  q14
+        vmlal.s16       q0,  d16,  d18
+        vmlal.s16       q1,  d17,  d19
+        vadd.s16        q10, q8,   q10
+        vadd.s16        q13, q11,  q13
+        vmlal.s16       q2,  d22,  d24
+        vmlal.s16       q3,  d23,  d25
+        vst1.16         {q10},     [r12,:128]!
+        subs            r3,  r3,   #16
+        vst1.16         {q13},     [r12,:128]!
+        bne             1b
+
+        vpadd.s32       d16, d0,   d1
+        vpadd.s32       d17, d2,   d3
+        vpadd.s32       d18, d4,   d5
+        vpadd.s32       d19, d6,   d7
+        vpadd.s32       d0,  d16,  d17
+        vpadd.s32       d1,  d18,  d19
+        vpadd.s32       d2,  d0,   d1
+        vpaddl.s32      d3,  d2
+        vmov.32         r0,  d3[0]
+        bx              lr
+endfunc
diff --git a/libavcodec/arm/dsputil_init_neon.c b/libavcodec/arm/dsputil_init_neon.c
index 16e052dddd..c9bdaa5a78 100644
--- a/libavcodec/arm/dsputil_init_neon.c
+++ b/libavcodec/arm/dsputil_init_neon.c
@@ -44,9 +44,6 @@ void ff_vector_clip_int32_neon(int32_t *dst, const int32_t *src, int32_t min,
 
 int32_t ff_scalarproduct_int16_neon(const int16_t *v1, const int16_t *v2, int len);
 
-int32_t ff_scalarproduct_and_madd_int16_neon(int16_t *v1, const int16_t *v2,
-                                             const int16_t *v3, int len, int mul);
-
 av_cold void ff_dsputil_init_neon(DSPContext *c, AVCodecContext *avctx,
                                   unsigned high_bit_depth)
 {
@@ -73,6 +70,4 @@ av_cold void ff_dsputil_init_neon(DSPContext *c, AVCodecContext *avctx,
     c->vector_clip_int32 = ff_vector_clip_int32_neon;
 
     c->scalarproduct_int16 = ff_scalarproduct_int16_neon;
-
-    c->scalarproduct_and_madd_int16 = ff_scalarproduct_and_madd_int16_neon;
 }
diff --git a/libavcodec/arm/int_neon.S b/libavcodec/arm/int_neon.S
index 3d2faffa48..42f37392e1 100644
--- a/libavcodec/arm/int_neon.S
+++ b/libavcodec/arm/int_neon.S
@@ -48,43 +48,3 @@ function ff_scalarproduct_int16_neon, export=1
         vmov.32         r0,  d3[0]
         bx              lr
 endfunc
-
-@ scalarproduct_and_madd_int16(/*aligned*/v0,v1,v2,order,mul)
-function ff_scalarproduct_and_madd_int16_neon, export=1
-        vld1.16         {d28[],d29[]}, [sp]
-        vmov.i16        q0,  #0
-        vmov.i16        q1,  #0
-        vmov.i16        q2,  #0
-        vmov.i16        q3,  #0
-        mov             r12, r0
-
-1:      vld1.16         {d16-d17}, [r0,:128]!
-        vld1.16         {d18-d19}, [r1]!
-        vld1.16         {d20-d21}, [r2]!
-        vld1.16         {d22-d23}, [r0,:128]!
-        vld1.16         {d24-d25}, [r1]!
-        vld1.16         {d26-d27}, [r2]!
-        vmul.s16        q10, q10,  q14
-        vmul.s16        q13, q13,  q14
-        vmlal.s16       q0,  d16,  d18
-        vmlal.s16       q1,  d17,  d19
-        vadd.s16        q10, q8,   q10
-        vadd.s16        q13, q11,  q13
-        vmlal.s16       q2,  d22,  d24
-        vmlal.s16       q3,  d23,  d25
-        vst1.16         {q10},     [r12,:128]!
-        subs            r3,  r3,   #16
-        vst1.16         {q13},     [r12,:128]!
-        bne             1b
-
-        vpadd.s32       d16, d0,   d1
-        vpadd.s32       d17, d2,   d3
-        vpadd.s32       d18, d4,   d5
-        vpadd.s32       d19, d6,   d7
-        vpadd.s32       d0,  d16,  d17
-        vpadd.s32       d1,  d18,  d19
-        vpadd.s32       d2,  d0,   d1
-        vpaddl.s32      d3,  d2
-        vmov.32         r0,  d3[0]
-        bx              lr
-endfunc