api: std_types: add fractional number type

Added odp_fract_u64_t type which can be used to define
fractional numbers accurately. A fractional to double
conversion function was added to help with this typical
conversion.

Signed-off-by: Petri Savolainen <petri.savolainen@nokia.com>
Reviewed-by: Jere Leppänen <jere.leppanen@nokia.com>
Reviewed-by: Jerin Jacob <jerinj@marvell.com>

1 files changed, 34 insertions, 0 deletions

diff --git a/include/odp/api/spec/std_types.h b/include/odp/api/spec/std_types.h
index bf3eb77a6..4b2af87ef 100644
--- a/include/odp/api/spec/std_types.h
+++ b/include/odp/api/spec/std_types.h
@@ -1,5 +1,6 @@
 /* Copyright (c) 2013-2018, Linaro Limited
  * Copyright (c) 2021, ARM Limited
+ * Copyright (c) 2021, Nokia
  * All rights reserved.
  *
  * SPDX-License-Identifier:     BSD-3-Clause
@@ -74,6 +75,39 @@ typedef struct ODP_ALIGNED(16) odp_u128_s {
 } odp_u128_t;
 
 /**
+ * Unsigned 64 bit fractional number
+ *
+ * The number is composed of integer and fraction parts. The fraction part is composed of
+ * two terms: numerator and denominator. Value of the number is sum of the integer and fraction
+ * parts: value = integer + numer/denom. When the fraction part is zero, the numerator is zero and
+ * the denominator may be zero.
+ */
+typedef struct odp_fract_u64_t {
+		/** Integer part */
+		uint64_t integer;
+
+		/** Numerator of the fraction part */
+		uint64_t numer;
+
+		/** Denominator of the fraction part. This may be zero when the numerator
+		 *  is zero. */
+		uint64_t denom;
+
+} odp_fract_u64_t;
+
+/**
+ * Convert fractional number (u64) to double
+ *
+ * Converts value of the unsigned 64 bit fractional number to a double-precision
+ * floating-point value.
+ *
+ * @param fract  Pointer to a fractional number
+ *
+ * @return Value of the fractional number as double
+ */
+double odp_fract_u64_to_dbl(const odp_fract_u64_t *fract);
+
+/**
  * @}
  */