[Patch][binutils] Generic support for parsing numbers in bfloat16 format [5/X]

Hi,

This patch is part of a series that adds support for Armv8.6-A
(Matrix Multiply and BFloat16 extensions).

This patch contains some general refactoring of the atof_ieee
function, exposing a function that allows a higher level of control
over the format of IEEE-like floating point numbers.
This has been done in order to be able to add a directive for assembling
floating point literals in the bfloat16 format in the following patches.

Committed on behalf of Mihail Ionescu.

Tested on arm-none-eabi, arm-none-linux-gnueabihf, aarch64-none-elf
and aarch64-none-linux-gnuwith no issues.

gas/ChangeLog:

2019-10-21  Mihail Ionescu  <mihail.ionescu@arm.com>
2019-10-21  Barnaby Wilks  <barnaby.wilks@arm.com>

	* as.h (atof_ieee_detail): Add prototype for atof_ieee_detail function.
	(atof_ieee): Move some code into the atof_ieee_detail function.
	(atof_ieee_detail): Add function that provides a higher level of control over generating
	IEEE-like numbers.

Is it ok for trunk?

Regards,
Mihail

diff --git a/gas/ChangeLog b/gas/ChangeLog
index 57ee5cd1795316c1beda2c03e166b966feeb4649..3b94f2975a04b013a6652edd8c2bae3d9f240a24 100644 (file)
--- a/gas/ChangeLog
+++ b/gas/ChangeLog
@@ -1,3 +1,11 @@
+2019-11-07  Mihail Ionescu  <mihail.ionescu@arm.com>
+2019-11-07  Barnaby Wilks  <barnaby.wilks@arm.com>
+
+       * as.h (atof_ieee_detail): Add prototype for atof_ieee_detail function.
+       (atof_ieee): Move some code into the atof_ieee_detail function.
+       (atof_ieee_detail): Add function that provides a higher level of
+       control over generating IEEE-like numbers.
+
 2019-11-07  Mihail Ionescu  <mihail.ionescu@arm.com>
 2019-11-07  Matthew Malcomson  <matthew.malcomson@arm.com>
 

diff --git a/gas/as.h b/gas/as.h
index 5b604a45342205187bef6d96a6bb1f1f35c5423d..d996697bed8784260a8f5400ac0c72254e4d0387 100644 (file)
--- a/gas/as.h
+++ b/gas/as.h
@@ -484,6 +484,7 @@ char * app_push (void);
 #define MAX_LITTLENUMS 6
 
 char * atof_ieee (char *, int, LITTLENUM_TYPE *);
+char * atof_ieee_detail (char *, int, int, LITTLENUM_TYPE *, FLONUM_TYPE *);
 const char * ieee_md_atof (int, char *, int *, bfd_boolean);
 const char * vax_md_atof (int, char *, int *);
 char * input_scrub_include_file (const char *, char *);

diff --git a/gas/config/atof-ieee.c b/gas/config/atof-ieee.c
index 944c8a46ff486ddc4cedc7fe4679b2e1d862a5f5..03f90bea75d8e29c5ed9b6a97c44c75dfd63eef9 100644 (file)
--- a/gas/config/atof-ieee.c
+++ b/gas/config/atof-ieee.c
@@ -146,29 +146,30 @@ make_invalid_floating_point_number (LITTLENUM_TYPE *words)
   words[4] = (LITTLENUM_TYPE) -1;
   words[5] = (LITTLENUM_TYPE) -1;
 }
-\f
-/* Warning: This returns 16-bit LITTLENUMs.  It is up to the caller to
-   figure out any alignment problems and to conspire for the
-   bytes/word to be emitted in the right order.  Bigendians beware!  */
 
-/* Note that atof-ieee always has X and P precisions enabled.  it is up
-   to md_atof to filter them out if the target machine does not support
-   them.  */
+/* Build a floating point constant at str into a IEEE floating
+   point number.  This function does the same thing as atof_ieee
+   however it allows more control over the exact format, i.e.
+   explicitly specifying the precision and number of exponent bits
+   instead of relying on this infomation being deduced from a given type.
 
-/* Returns pointer past text consumed.  */
+   If generic_float_info is not NULL then it will be set to contain generic
+   infomation about the parsed floating point number.
 
+   Returns pointer past text consumed. */
 char *
-atof_ieee (char *str,                  /* Text to convert to binary.  */
-          int what_kind,               /* 'd', 'f', 'x', 'p'.  */
-          LITTLENUM_TYPE *words)       /* Build the binary here.  */
+atof_ieee_detail (char * str,
+                 int precision,
+                 int exponent_bits,
+                 LITTLENUM_TYPE * words,
+                 FLONUM_TYPE * generic_float_info)
 {
   /* Extra bits for zeroed low-order bits.
      The 1st MAX_PRECISION are zeroed, the last contain flonum bits.  */
   static LITTLENUM_TYPE bits[MAX_PRECISION + MAX_PRECISION + GUARD];
   char *return_value;
+
   /* Number of 16-bit words in the format.  */
-  int precision;
-  long exponent_bits;
   FLONUM_TYPE save_gen_flonum;
 
   /* We have to save the generic_floating_point_number because it
@@ -189,6 +190,45 @@ atof_ieee (char *str,                      /* Text to convert to binary.  */
 
   memset (bits, '\0', sizeof (LITTLENUM_TYPE) * MAX_PRECISION);
 
+  generic_floating_point_number.high
+    = generic_floating_point_number.low + precision - 1 + GUARD;
+
+  if (atof_generic (&return_value, ".", EXP_CHARS,
+                   &generic_floating_point_number))
+    {
+      make_invalid_floating_point_number (words);
+      return NULL;
+    }
+
+  if (generic_float_info)
+    *generic_float_info = generic_floating_point_number;
+
+  gen_to_words (words, precision, exponent_bits);
+
+  /* Restore the generic_floating_point_number's storage alloc (and
+     everything else).  */
+  generic_floating_point_number = save_gen_flonum;
+
+  return return_value;
+}
+
+/* Warning: This returns 16-bit LITTLENUMs.  It is up to the caller to
+   figure out any alignment problems and to conspire for the
+   bytes/word to be emitted in the right order.  Bigendians beware!  */
+
+/* Note that atof-ieee always has X and P precisions enabled.  it is up
+   to md_atof to filter them out if the target machine does not support
+   them.  */
+
+/* Returns pointer past text consumed.  */
+char *
+atof_ieee (char *str,                  /* Text to convert to binary.  */
+          int what_kind,               /* 'd', 'f', 'x', 'p'.  */
+          LITTLENUM_TYPE *words)       /* Build the binary here.  */
+{
+  int precision;
+  long exponent_bits;
+
   switch (what_kind)
     {
     case 'h':
@@ -232,22 +272,7 @@ atof_ieee (char *str,                      /* Text to convert to binary.  */
       return (NULL);
     }
 
-  generic_floating_point_number.high
-    = generic_floating_point_number.low + precision - 1 + GUARD;
-
-  if (atof_generic (&return_value, ".", EXP_CHARS,
-                   &generic_floating_point_number))
-    {
-      make_invalid_floating_point_number (words);
-      return NULL;
-    }
-  gen_to_words (words, precision, exponent_bits);
-
-  /* Restore the generic_floating_point_number's storage alloc (and
-     everything else).  */
-  generic_floating_point_number = save_gen_flonum;
-
-  return return_value;
+  return atof_ieee_detail (str, precision, exponent_bits, words, NULL);
 }
 
 /* Turn generic_floating_point_number into a real float/double/extended.  */