Initial-checkin for ON2 Software AVC/H264 decoder

o when neon is present, the performance gain of On2 AVC software decoder
  over PV software decoder is more than 30%.

o In addition, it fixes some known PV software decoder issues like missing
  output frames

o allow both pv and on2 software avc to be available for easy comparision

o change output frames from 8 to 16

Change-Id: I567ad1842025ead7092f0c47e3513d6d9ca232dd

1 files changed, 87 insertions, 0 deletions

diff --git a/media/libstagefright/codecs/on2/h264dec/omxdl/reference/vc/m4p2/src/omxVCM4P2_DCT8x8blk.c b/media/libstagefright/codecs/on2/h264dec/omxdl/reference/vc/m4p2/src/omxVCM4P2_DCT8x8blk.c
new file mode 100644
index 0000000..a077ac8
--- /dev/null
+++ b/media/libstagefright/codecs/on2/h264dec/omxdl/reference/vc/m4p2/src/omxVCM4P2_DCT8x8blk.c
@@ -0,0 +1,87 @@
+/**
+ * 
+ * File Name:  omxVCM4P2_DCT8x8blk.c
+ * OpenMAX DL: v1.0.2
+ * Revision:   9641
+ * Date:       Thursday, February 7, 2008
+ * 
+ * (c) Copyright 2007-2008 ARM Limited. All Rights Reserved.
+ * 
+ * 
+ *
+ * Description:
+ * Contains modules for 8x8 block DCT
+ * 
+ */
+ 
+#include <math.h>
+#include "omxtypes.h"
+#include "armOMX.h"
+
+#include "armCOMM.h"
+#include "armVCM4P2_DCT_Table.h"
+
+/**
+ * Function:  omxVCM4P2_DCT8x8blk   (6.2.4.4.1)
+ *
+ * Description:
+ * Computes a 2D forward DCT for a single 8x8 block, as defined in 
+ * [ISO14496-2]. 
+ *
+ * Input Arguments:
+ *   
+ *   pSrc - pointer to the start of the linearly arranged input buffer; must 
+ *            be aligned on a 16-byte boundary.  Input values (pixel 
+ *            intensities) are valid in the range [-255,255]. 
+ *
+ * Output Arguments:
+ *   
+ *   pDst - pointer to the start of the linearly arranged output buffer; must 
+ *            be aligned on a 16-byte boundary. 
+ *
+ * Return Value:
+ *    
+ *    OMX_Sts_NoErr - no error 
+ *    OMX_Sts_BadArgErr - bad arguments, returned if:
+ *    -    pSrc or pDst is NULL. 
+ *    -    pSrc or pDst is not 16-byte aligned. 
+ *
+ */
+
+OMXResult omxVCM4P2_DCT8x8blk (const OMX_S16 *pSrc, OMX_S16 *pDst)
+{
+    OMX_INT x, y, u, v;
+    
+    /* Argument error checks */
+    armRetArgErrIf(pSrc == NULL, OMX_Sts_BadArgErr);
+    armRetArgErrIf(!armIs16ByteAligned(pSrc), OMX_Sts_BadArgErr);
+    armRetArgErrIf(pDst == NULL, OMX_Sts_BadArgErr);
+    armRetArgErrIf(!armIs16ByteAligned(pDst), OMX_Sts_BadArgErr);
+
+
+    for (u = 0; u < 8; u++)
+    {
+        for (v = 0; v < 8; v++)
+        {
+            OMX_F64 sum = 0.0;
+            for (x = 0; x < 8; x++)
+            {
+                for (y = 0; y < 8; y++)
+                {
+                    sum += pSrc[(x * 8) + y] *
+                       armVCM4P2_preCalcDCTCos[x][u] *
+                       armVCM4P2_preCalcDCTCos[y][v];
+                }
+            }
+            pDst[(u * 8) + v]= armRoundFloatToS16 (sum);            
+        }
+    }
+
+    return OMX_Sts_NoErr;
+}
+
+
+
+/* End of file */
+
+