diff --git a/include/core/SkMath.h b/include/core/SkMath.h
index 7a4b707..42f830c 100644
--- a/include/core/SkMath.h
+++ b/include/core/SkMath.h
@@ -12,68 +12,51 @@
 
 #include "SkTypes.h"
 
+/**
+ *  Computes numer1 * numer2 / denom in full 64 intermediate precision.
+ *  It is an error for denom to be 0. There is no special handling if
+ *  the result overflows 32bits.
+ */
+int32_t SkMulDiv(int32_t numer1, int32_t numer2, int32_t denom);
+
+/**
+ *  Computes (numer1 << shift) / denom in full 64 intermediate precision.
+ *  It is an error for denom to be 0. There is no special handling if
+ *  the result overflows 32bits.
+ */
+int32_t SkDivBits(int32_t numer, int32_t denom, int shift);
+
+/**
+ *  Return the integer square root of value, with a bias of bitBias
+ */
+int32_t SkSqrtBits(int32_t value, int bitBias);
+
+///////////////////////////////////////////////////////////////////////////////
+
 //! Returns the number of leading zero bits (0...32)
 int SkCLZ_portable(uint32_t);
 
-/** Computes the 64bit product of a * b, and then shifts the answer down by
-    shift bits, returning the low 32bits. shift must be [0..63]
-    e.g. to perform a fixedmul, call SkMulShift(a, b, 16)
-*/
-int32_t SkMulShift(int32_t a, int32_t b, unsigned shift);
+#if defined(__arm__)
+    #define SkCLZ(x)    __builtin_clz(x)
+#endif
 
-/** Computes numer1 * numer2 / denom in full 64 intermediate precision.
-    It is an error for denom to be 0. There is no special handling if
-    the result overflows 32bits.
-*/
-int32_t SkMulDiv(int32_t numer1, int32_t numer2, int32_t denom);
+#ifndef SkCLZ
+    #define SkCLZ(x)    SkCLZ_portable(x)
+#endif
 
-/** Computes (numer1 << shift) / denom in full 64 intermediate precision.
-    It is an error for denom to be 0. There is no special handling if
-    the result overflows 32bits.
-*/
-int32_t SkDivBits(int32_t numer, int32_t denom, int shift);
-
-/** Return the integer square root of value, with a bias of bitBias
-*/
-int32_t SkSqrtBits(int32_t value, int bitBias);
-
-/** Return the integer square root of n, treated as a SkFixed (16.16)
-*/
-#define SkSqrt32(n)         SkSqrtBits(n, 15)
-
-/** Return the integer cube root of value, with a bias of bitBias
+/**
+ *  Returns (value < 0 ? 0 : value) efficiently (i.e. no compares or branches)
  */
-int32_t SkCubeRootBits(int32_t value, int bitBias);
-
-/** Returns -1 if n < 0, else returns 0
-*/
-#define SkExtractSign(n)    ((int32_t)(n) >> 31)
-
-/** If sign == -1, returns -n, else sign must be 0, and returns n.
-    Typically used in conjunction with SkExtractSign().
-*/
-static inline int32_t SkApplySign(int32_t n, int32_t sign) {
-    SkASSERT(sign == 0 || sign == -1);
-    return (n ^ sign) - sign;
-}
-
-/** Return x with the sign of y */
-static inline int32_t SkCopySign32(int32_t x, int32_t y) {
-    return SkApplySign(x, SkExtractSign(x ^ y));
-}
-
-/** Returns (value < 0 ? 0 : value) efficiently (i.e. no compares or branches)
-*/
 static inline int SkClampPos(int value) {
     return value & ~(value >> 31);
 }
 
 /** Given an integer and a positive (max) integer, return the value
-    pinned against 0 and max, inclusive.
-    @param value    The value we want returned pinned between [0...max]
-    @param max      The positive max value
-    @return 0 if value < 0, max if value > max, else value
-*/
+ *  pinned against 0 and max, inclusive.
+ *  @param value    The value we want returned pinned between [0...max]
+ *  @param max      The positive max value
+ *  @return 0 if value < 0, max if value > max, else value
+ */
 static inline int SkClampMax(int value, int max) {
     // ensure that max is positive
     SkASSERT(max >= 0);
@@ -86,62 +69,33 @@
     return value;
 }
 
-/** Given a positive value and a positive max, return the value
-    pinned against max.
-    Note: only works as long as max - value doesn't wrap around
-    @return max if value >= max, else value
-*/
-static inline unsigned SkClampUMax(unsigned value, unsigned max) {
-#ifdef SK_CPU_HAS_CONDITIONAL_INSTR
-    if (value > max) {
-        value = max;
-    }
-    return value;
-#else
-    int diff = max - value;
-    // clear diff if diff is positive
-    diff &= diff >> 31;
-
-    return value + diff;
-#endif
-}
-
-///////////////////////////////////////////////////////////////////////////////
-
-#if defined(__arm__)
-    #define SkCLZ(x)    __builtin_clz(x)
-#endif
-
-#ifndef SkCLZ
-    #define SkCLZ(x)    SkCLZ_portable(x)
-#endif
-
-///////////////////////////////////////////////////////////////////////////////
-
-/** Returns the smallest power-of-2 that is >= the specified value. If value
-    is already a power of 2, then it is returned unchanged. It is undefined
-    if value is <= 0.
-*/
+/**
+ *  Returns the smallest power-of-2 that is >= the specified value. If value
+ *  is already a power of 2, then it is returned unchanged. It is undefined
+ *  if value is <= 0.
+ */
 static inline int SkNextPow2(int value) {
     SkASSERT(value > 0);
     return 1 << (32 - SkCLZ(value - 1));
 }
 
-/** Returns the log2 of the specified value, were that value to be rounded up
-    to the next power of 2. It is undefined to pass 0. Examples:
-         SkNextLog2(1) -> 0
-         SkNextLog2(2) -> 1
-         SkNextLog2(3) -> 2
-         SkNextLog2(4) -> 2
-         SkNextLog2(5) -> 3
-*/
+/**
+ *  Returns the log2 of the specified value, were that value to be rounded up
+ *  to the next power of 2. It is undefined to pass 0. Examples:
+ *  SkNextLog2(1) -> 0
+ *  SkNextLog2(2) -> 1
+ *  SkNextLog2(3) -> 2
+ *  SkNextLog2(4) -> 2
+ *  SkNextLog2(5) -> 3
+ */
 static inline int SkNextLog2(uint32_t value) {
     SkASSERT(value != 0);
     return 32 - SkCLZ(value - 1);
 }
 
-/** Returns true if value is a power of 2. Does not explicitly check for
-    value <= 0.
+/**
+ *  Returns true if value is a power of 2. Does not explicitly check for
+ *  value <= 0.
  */
 static inline bool SkIsPow2(int value) {
     return (value & (value - 1)) == 0;
@@ -149,10 +103,11 @@
 
 ///////////////////////////////////////////////////////////////////////////////
 
-/** SkMulS16(a, b) multiplies a * b, but requires that a and b are both int16_t.
-    With this requirement, we can generate faster instructions on some
-    architectures.
-*/
+/**
+ *  SkMulS16(a, b) multiplies a * b, but requires that a and b are both int16_t.
+ *  With this requirement, we can generate faster instructions on some
+ *  architectures.
+ */
 #ifdef SK_ARM_HAS_EDSP
     static inline int32_t SkMulS16(S16CPU x, S16CPU y) {
         SkASSERT((int16_t)x == x);
@@ -176,39 +131,10 @@
     #endif
 #endif
 
-/** Return a*b/255, truncating away any fractional bits. Only valid if both
-    a and b are 0..255
-*/
-static inline U8CPU SkMulDiv255Trunc(U8CPU a, U8CPU b) {
-    SkASSERT((uint8_t)a == a);
-    SkASSERT((uint8_t)b == b);
-    unsigned prod = SkMulS16(a, b) + 1;
-    return (prod + (prod >> 8)) >> 8;
-}
-
-/** Return a*b/255, rounding any fractional bits. Only valid if both
-    a and b are 0..255
+/**
+ *  Return a*b/((1 << shift) - 1), rounding any fractional bits.
+ *  Only valid if a and b are unsigned and <= 32767 and shift is > 0 and <= 8
  */
-static inline U8CPU SkMulDiv255Round(U8CPU a, U8CPU b) {
-    SkASSERT((uint8_t)a == a);
-    SkASSERT((uint8_t)b == b);
-    unsigned prod = SkMulS16(a, b) + 128;
-    return (prod + (prod >> 8)) >> 8;
-}
-
-/** Return (a*b)/255, taking the ceiling of any fractional bits. Only valid if
-    both a and b are 0..255. The expected result equals (a * b + 254) / 255.
- */
-static inline U8CPU SkMulDiv255Ceiling(U8CPU a, U8CPU b) {
-    SkASSERT((uint8_t)a == a);
-    SkASSERT((uint8_t)b == b);
-    unsigned prod = SkMulS16(a, b) + 255;
-    return (prod + (prod >> 8)) >> 8;
-}
-
-/** Return a*b/((1 << shift) - 1), rounding any fractional bits.
-    Only valid if a and b are unsigned and <= 32767 and shift is > 0 and <= 8
-*/
 static inline unsigned SkMul16ShiftRound(unsigned a, unsigned b, int shift) {
     SkASSERT(a <= 32767);
     SkASSERT(b <= 32767);
@@ -217,10 +143,14 @@
     return (prod + (prod >> shift)) >> shift;
 }
 
-/** Just the rounding step in SkDiv255Round: round(value / 255)
+/**
+ *  Return a*b/255, rounding any fractional bits. Only valid if both
+ *  a and b are 0..255
  */
-static inline unsigned SkDiv255Round(unsigned prod) {
-    prod += 128;
+static inline U8CPU SkMulDiv255Round(U8CPU a, U8CPU b) {
+    SkASSERT((uint8_t)a == a);
+    SkASSERT((uint8_t)b == b);
+    unsigned prod = SkMulS16(a, b) + 128;
     return (prod + (prod >> 8)) >> 8;
 }
 
diff --git a/src/core/Sk64.cpp b/src/core/Sk64.cpp
index f350656..c530ed8 100644
--- a/src/core/Sk64.cpp
+++ b/src/core/Sk64.cpp
@@ -1,4 +1,3 @@
-
 /*
  * Copyright 2006 The Android Open Source Project
  *
@@ -7,7 +6,7 @@
  */
 
 #include "Sk64.h"
-#include "SkMath.h"
+#include "SkMathPriv.h"
 
 #define shift_left(hi, lo)          \
     hi = (hi << 1) | (lo >> 31);    \
diff --git a/src/core/SkBitmapProcState_matrix.h b/src/core/SkBitmapProcState_matrix.h
index 469cbf3..317c984 100644
--- a/src/core/SkBitmapProcState_matrix.h
+++ b/src/core/SkBitmapProcState_matrix.h
@@ -7,6 +7,7 @@
  */
 
 #include "SkMath.h"
+#include "SkMathPriv.h"
 
 #define SCALE_NOFILTER_NAME     MAKENAME(_nofilter_scale)
 #define SCALE_FILTER_NAME       MAKENAME(_filter_scale)
diff --git a/src/core/SkBitmapProcState_shaderproc.h b/src/core/SkBitmapProcState_shaderproc.h
index a3a8a99..ead57f1 100644
--- a/src/core/SkBitmapProcState_shaderproc.h
+++ b/src/core/SkBitmapProcState_shaderproc.h
@@ -6,7 +6,7 @@
  * found in the LICENSE file.
  */
 
-
+#include "SkMathPriv.h"
 
 #define SCALE_FILTER_NAME       MAKENAME(_filter_DX_shaderproc)
 
diff --git a/src/core/SkBlitRow_D16.cpp b/src/core/SkBlitRow_D16.cpp
index c815468..5e675e5 100644
--- a/src/core/SkBlitRow_D16.cpp
+++ b/src/core/SkBlitRow_D16.cpp
@@ -8,6 +8,7 @@
 #include "SkBlitRow.h"
 #include "SkColorPriv.h"
 #include "SkDither.h"
+#include "SkMathPriv.h"
 
 ///////////////////////////////////////////////////////////////////////////////
 
diff --git a/src/core/SkConfig8888.cpp b/src/core/SkConfig8888.cpp
index 10a1b36..f889f20 100644
--- a/src/core/SkConfig8888.cpp
+++ b/src/core/SkConfig8888.cpp
@@ -1,4 +1,5 @@
 #include "SkConfig8888.h"
+#include "SkMathPriv.h"
 
 namespace {
 
diff --git a/src/core/SkCordic.cpp b/src/core/SkCordic.cpp
index 14d34fb..56cb460 100644
--- a/src/core/SkCordic.cpp
+++ b/src/core/SkCordic.cpp
@@ -1,4 +1,3 @@
-
 /*
  * Copyright 2006 The Android Open Source Project
  *
@@ -6,9 +5,8 @@
  * found in the LICENSE file.
  */
 
-
 #include "SkCordic.h"
-#include "SkMath.h"
+#include "SkMathPriv.h"
 #include "Sk64.h"
 
 // 0x20000000 equals pi / 4
diff --git a/include/core/SkFDot6.h b/src/core/SkFDot6.h
similarity index 100%
rename from include/core/SkFDot6.h
rename to src/core/SkFDot6.h
diff --git a/src/core/SkFloat.cpp b/src/core/SkFloat.cpp
index ec785f3..dffb0d7 100644
--- a/src/core/SkFloat.cpp
+++ b/src/core/SkFloat.cpp
@@ -8,7 +8,7 @@
 
 
 #include "SkFloat.h"
-#include "SkMath.h"
+#include "SkMathPriv.h"
 
 #define EXP_BIAS    (127+23)
 
diff --git a/src/core/SkFloatBits.cpp b/src/core/SkFloatBits.cpp
index 378f98d..fe4a1b9 100644
--- a/src/core/SkFloatBits.cpp
+++ b/src/core/SkFloatBits.cpp
@@ -1,12 +1,12 @@
-
 /*
  * Copyright 2011 Google Inc.
  *
  * Use of this source code is governed by a BSD-style license that can be
  * found in the LICENSE file.
  */
+
 #include "SkFloatBits.h"
-#include "SkMath.h"
+#include "SkMathPriv.h"
 
 /******************************************************************************
     SkFloatBits_toInt[Floor, Round, Ceil] are identical except for what they
diff --git a/src/core/SkMath.cpp b/src/core/SkMath.cpp
index 7b75305..a149625 100644
--- a/src/core/SkMath.cpp
+++ b/src/core/SkMath.cpp
@@ -1,4 +1,3 @@
-
 /*
  * Copyright 2008 The Android Open Source Project
  *
@@ -6,8 +5,7 @@
  * found in the LICENSE file.
  */
 
-
-#include "SkMath.h"
+#include "SkMathPriv.h"
 #include "SkCordic.h"
 #include "SkFloatBits.h"
 #include "SkFloatingPoint.h"
diff --git a/src/core/SkMathPriv.h b/src/core/SkMathPriv.h
new file mode 100644
index 0000000..055acc4
--- /dev/null
+++ b/src/core/SkMathPriv.h
@@ -0,0 +1,94 @@
+/*
+ * Copyright 2012 Google Inc.
+ *
+ * Use of this source code is governed by a BSD-style license that can be
+ * found in the LICENSE file.
+ */
+
+#ifndef SkMathPriv_DEFINED
+#define SkMathPriv_DEFINED
+
+#include "SkMath.h"
+
+/** Returns -1 if n < 0, else returns 0
+ */
+#define SkExtractSign(n)    ((int32_t)(n) >> 31)
+
+/** If sign == -1, returns -n, else sign must be 0, and returns n.
+ Typically used in conjunction with SkExtractSign().
+ */
+static inline int32_t SkApplySign(int32_t n, int32_t sign) {
+    SkASSERT(sign == 0 || sign == -1);
+    return (n ^ sign) - sign;
+}
+
+/** Return x with the sign of y */
+static inline int32_t SkCopySign32(int32_t x, int32_t y) {
+    return SkApplySign(x, SkExtractSign(x ^ y));
+}
+
+/** Given a positive value and a positive max, return the value
+ pinned against max.
+ Note: only works as long as max - value doesn't wrap around
+ @return max if value >= max, else value
+ */
+static inline unsigned SkClampUMax(unsigned value, unsigned max) {
+#ifdef SK_CPU_HAS_CONDITIONAL_INSTR
+    if (value > max) {
+        value = max;
+    }
+    return value;
+#else
+    int diff = max - value;
+    // clear diff if diff is positive
+    diff &= diff >> 31;
+    
+    return value + diff;
+#endif
+}
+
+/** Computes the 64bit product of a * b, and then shifts the answer down by
+ shift bits, returning the low 32bits. shift must be [0..63]
+ e.g. to perform a fixedmul, call SkMulShift(a, b, 16)
+ */
+int32_t SkMulShift(int32_t a, int32_t b, unsigned shift);
+
+/** Return the integer square root of n, treated as a SkFixed (16.16)
+ */
+#define SkSqrt32(n)         SkSqrtBits(n, 15)
+
+/** Return the integer cube root of value, with a bias of bitBias
+ */
+int32_t SkCubeRootBits(int32_t value, int bitBias);
+
+///////////////////////////////////////////////////////////////////////////////
+
+/** Return a*b/255, truncating away any fractional bits. Only valid if both
+ a and b are 0..255
+ */
+static inline U8CPU SkMulDiv255Trunc(U8CPU a, U8CPU b) {
+    SkASSERT((uint8_t)a == a);
+    SkASSERT((uint8_t)b == b);
+    unsigned prod = SkMulS16(a, b) + 1;
+    return (prod + (prod >> 8)) >> 8;
+}
+
+/** Return (a*b)/255, taking the ceiling of any fractional bits. Only valid if
+ both a and b are 0..255. The expected result equals (a * b + 254) / 255.
+ */
+static inline U8CPU SkMulDiv255Ceiling(U8CPU a, U8CPU b) {
+    SkASSERT((uint8_t)a == a);
+    SkASSERT((uint8_t)b == b);
+    unsigned prod = SkMulS16(a, b) + 255;
+    return (prod + (prod >> 8)) >> 8;
+}
+
+/** Just the rounding step in SkDiv255Round: round(value / 255)
+ */
+static inline unsigned SkDiv255Round(unsigned prod) {
+    prod += 128;
+    return (prod + (prod >> 8)) >> 8;
+}
+
+#endif
+
diff --git a/src/core/SkXfermode.cpp b/src/core/SkXfermode.cpp
index 74097be..aeed9cc 100644
--- a/src/core/SkXfermode.cpp
+++ b/src/core/SkXfermode.cpp
@@ -10,6 +10,7 @@
 #include "SkXfermode.h"
 #include "SkColorPriv.h"
 #include "SkFlattenableBuffers.h"
+#include "SkMathPriv.h"
 
 SK_DEFINE_INST_COUNT(SkXfermode)
 
diff --git a/tests/ColorTest.cpp b/tests/ColorTest.cpp
index 83e2e3f..86729d2 100644
--- a/tests/ColorTest.cpp
+++ b/tests/ColorTest.cpp
@@ -8,7 +8,7 @@
 #include "Test.h"
 #include "SkColor.h"
 #include "SkColorPriv.h"
-#include "SkMath.h"
+#include "SkMathPriv.h"
 #include "SkRandom.h"
 #include "SkUnPreMultiply.h"
 
diff --git a/tests/MathTest.cpp b/tests/MathTest.cpp
index b41c471..e2e63c5 100644
--- a/tests/MathTest.cpp
+++ b/tests/MathTest.cpp
@@ -7,7 +7,7 @@
  */
 #include "Test.h"
 #include "SkFloatingPoint.h"
-#include "SkMath.h"
+#include "SkMathPriv.h"
 #include "SkPoint.h"
 #include "SkRandom.h"
 #include "SkColorPriv.h"
diff --git a/tests/ReadPixelsTest.cpp b/tests/ReadPixelsTest.cpp
index 730be05..a0eee04 100644
--- a/tests/ReadPixelsTest.cpp
+++ b/tests/ReadPixelsTest.cpp
@@ -10,6 +10,7 @@
 #include "SkCanvas.h"
 #include "SkColorPriv.h"
 #include "SkDevice.h"
+#include "SkMathPriv.h"
 #include "SkRegion.h"
 #if SK_SUPPORT_GPU
 #include "SkGpuDevice.h"
diff --git a/tests/WritePixelsTest.cpp b/tests/WritePixelsTest.cpp
index 688aa5d..5911b80 100644
--- a/tests/WritePixelsTest.cpp
+++ b/tests/WritePixelsTest.cpp
@@ -10,6 +10,7 @@
 #include "SkCanvas.h"
 #include "SkColorPriv.h"
 #include "SkDevice.h"
+#include "SkMathPriv.h"
 #include "SkRegion.h"
 #if SK_SUPPORT_GPU
 #include "SkGpuDevice.h"