Patchwork [RFC,v2,2/3] Target FP printing: Simplify and fix ui_printf

login
register
mail settings
Submitter Ulrich Weigand
Date Oct. 5, 2017, 5:31 p.m.
Message ID <20171005173131.B6199D835F0@oc3748833570.ibm.com>
Download mbox | patch
Permalink /patch/23361/
State New
Headers show

Comments

Ulrich Weigand - Oct. 5, 2017, 5:31 p.m.
[RFC v2][2/3] Target FP printing: Simplify and fix ui_printf

This patch adds support for handling format strings to both
floatformat_to_string and decimal_to_string, and then uses
those routines to implement ui_printf formatted printing.

There is already a subroutine printf_decfloat that ui_printf uses to
handle decimal FP.  This is renamed to printf_floating and updated
to handle both binary and decimal FP.  This includes the following
set of changes:

- printf_decfloat currently parses the format string again to determine
  the intended target format.  This seems superfluous since the common
  parsing code in parse_format_string already did this, but then did
  not pass the result on to its users.  Fixed by splitting the decfloat_arg
  argument class into three distinct classes, and passing them through.

- Now we can rename printf_decfloat to printf_floating and also call it
  for the argument classes representing binary FP types.

- The code will now use the argclass to detect the type the value should
  be printed at, and converts the input value to this type if necessary.
  To remain compatible with current behavior, for binary FP the code
  instead tries to re-interpret the input value as a FP type of the
  same size if that exists.  (Maybe this behavior is more confusing
  than useful -- but this can be changed later if we want to ...)

- Finally, we can use floatformat_to_string / decimal_to_string passing
  the format string to perform the formatted output using the desired
  target FP type.

Note that we no longer generate different code depending on whether or not
the host supports "long double" -- this check is obsolete anyway since C++11
mandates "long double", and in any case a %lg format string is intended to
refer to the *target* long double type, not the host version.

Note also that formatted printing of DFP numbers may not work correctly,
since it attempts to use the host printf to do so (and makes unwarranted
assumptions about the host ABI while doing so!).  This is no change to
the current behavior -- I simply moved the code from printf_decfloat to
the decimal_to_string routine in dfp.c.  If we want to fix it in the
future, that is a more appropriate place anyway.

Bye,
Ulrich


ChangeLog:

	* common/format.h (enum argclass): Replace decfloat_arg by
	dec32float_arg, dec64float_arg, and dec128float_arg.
	* common/format.c (parse_format_string): Update to return
	new decimal float argument classes.

	* printcmd.c (printf_decfloat): Rename to ...
	(printf_floating): ... this.  Add argclass argument, and use it
	instead of parsing the format string again.  Add support for
	binary floating-point values, using floatformat_to_string.
	Convert value to the target format if it doesn't already match.
	(ui_printf): Call printf_floating instead of printf_decfloat,
	also for double_arg / long_double_arg.  Pass argclass.

	* dfp.c (decimal_to_string): Add format string argument.
	* dfp.h (decimal_to_string): Likewise.

	* doublest.c (floatformat_to_string): Add format string argument.
	* doublest.h (floatformat_to_string): Likewise.

Patch

Index: binutils-gdb/gdb/common/format.c
===================================================================
--- binutils-gdb.orig/gdb/common/format.c
+++ binutils-gdb/gdb/common/format.c
@@ -274,8 +274,12 @@  parse_format_string (const char **arg)
 	  case 'g':
 	  case 'E':
 	  case 'G':
-	    if (seen_big_h || seen_big_d || seen_double_big_d)
-	      this_argclass = decfloat_arg;
+	    if (seen_double_big_d)
+	      this_argclass = dec128float_arg;
+	    else if (seen_big_d)
+	      this_argclass = dec64float_arg;
+	    else if (seen_big_h)
+	      this_argclass = dec32float_arg;
 	    else if (seen_big_l)
 	      this_argclass = long_double_arg;
 	    else
Index: binutils-gdb/gdb/common/format.h
===================================================================
--- binutils-gdb.orig/gdb/common/format.h
+++ binutils-gdb/gdb/common/format.h
@@ -35,7 +35,8 @@  enum argclass
     literal_piece,
     int_arg, long_arg, long_long_arg, ptr_arg,
     string_arg, wide_string_arg, wide_char_arg,
-    double_arg, long_double_arg, decfloat_arg
+    double_arg, long_double_arg,
+    dec32float_arg, dec64float_arg, dec128float_arg
   };
 
 /* A format piece is a section of the format string that may include a
Index: binutils-gdb/gdb/printcmd.c
===================================================================
--- binutils-gdb.orig/gdb/printcmd.c
+++ binutils-gdb/gdb/printcmd.c
@@ -2288,86 +2288,78 @@  printf_wide_c_string (struct ui_file *st
 }
 
 /* Subroutine of ui_printf to simplify it.
-   Print VALUE, a decimal floating point value, to STREAM using FORMAT.  */
+   Print VALUE, a floating point value, to STREAM using FORMAT.  */
 
 static void
-printf_decfloat (struct ui_file *stream, const char *format,
-		 struct value *value)
+printf_floating (struct ui_file *stream, const char *format,
+		 struct value *value, enum argclass argclass)
 {
-  const gdb_byte *param_ptr = value_contents (value);
-
-#if defined (PRINTF_HAS_DECFLOAT)
-  /* If we have native support for Decimal floating
-     printing, handle it here.  */
-  fprintf_filtered (stream, format, param_ptr);
-#else
-  /* As a workaround until vasprintf has native support for DFP
-     we convert the DFP values to string and print them using
-     the %s format specifier.  */
-  const char *p;
-
   /* Parameter data.  */
   struct type *param_type = value_type (value);
   struct gdbarch *gdbarch = get_type_arch (param_type);
   enum bfd_endian byte_order = gdbarch_byte_order (gdbarch);
 
-  /* DFP output data.  */
-  struct value *dfp_value = NULL;
-  gdb_byte *dfp_ptr;
-  int dfp_len = 16;
-  gdb_byte dec[16];
-  struct type *dfp_type = NULL;
-
-  /* Points to the end of the string so that we can go back
-     and check for DFP length modifiers.  */
-  p = format + strlen (format);
-
-  /* Look for the float/double format specifier.  */
-  while (*p != 'f' && *p != 'e' && *p != 'E'
-	 && *p != 'g' && *p != 'G')
-    p--;
-
-  /* Search for the '%' char and extract the size and type of
-     the output decimal value based on its modifiers
-     (%Hf, %Df, %DDf).  */
-  while (*--p != '%')
+  /* Determine target type corresponding to the format string.  */
+  struct type *fmt_type;
+  switch (argclass)
     {
-      if (*p == 'H')
-	{
-	  dfp_len = 4;
-	  dfp_type = builtin_type (gdbarch)->builtin_decfloat;
-	}
-      else if (*p == 'D' && *(p - 1) == 'D')
-	{
-	  dfp_len = 16;
-	  dfp_type = builtin_type (gdbarch)->builtin_declong;
-	  p--;
-	}
-      else
-	{
-	  dfp_len = 8;
-	  dfp_type = builtin_type (gdbarch)->builtin_decdouble;
-	}
+      case double_arg:
+	fmt_type = builtin_type (gdbarch)->builtin_double;
+	break;
+      case long_double_arg:
+	fmt_type = builtin_type (gdbarch)->builtin_long_double;
+	break;
+      case dec32float_arg:
+	fmt_type = builtin_type (gdbarch)->builtin_decfloat;
+	break;
+      case dec64float_arg:
+	fmt_type = builtin_type (gdbarch)->builtin_decdouble;
+	break;
+      case dec128float_arg:
+	fmt_type = builtin_type (gdbarch)->builtin_declong;
+	break;
+      default:
+	gdb_assert_not_reached ("unexpected argument class");
     }
 
-  /* Conversion between different DFP types.  */
-  if (TYPE_CODE (param_type) == TYPE_CODE_DECFLOAT)
-    decimal_convert (param_ptr, TYPE_LENGTH (param_type),
-		     byte_order, dec, dfp_len, byte_order);
-  else
-    /* If this is a non-trivial conversion, just output 0.
-       A correct converted value can be displayed by explicitly
-       casting to a DFP type.  */
-    decimal_from_string (dec, dfp_len, byte_order, "0");
+  /* To match the traditional GDB behavior, the conversion is
+     done differently depending on the type of the parameter:
+
+     - if the parameter has floating-point type, it's value
+       is converted to the target type;
+
+     - otherwise, if the parameter has a type that is of the
+       same size as a built-in floating-point type, the value
+       bytes are interpreted as if they were of that type, and
+       then converted to the target type (this is not done for
+       decimal floating-point argument classes);
+
+     - otherwise, if the source value has an integer value,
+       it's value is converted to the target type;
 
-  dfp_value = value_from_decfloat (dfp_type, dec);
+     - otherwise, an error is raised.
+
+     In either case, the result of the conversion is a byte buffer
+     formatted in the target format for the target type.  */
+
+  if (TYPE_CODE (fmt_type) == TYPE_CODE_FLT)
+    {
+      param_type = float_type_from_length (param_type);
+      if (param_type != value_type (value))
+	value = value_from_contents (param_type, value_contents (value));
+    }
 
-  dfp_ptr = (gdb_byte *) value_contents (dfp_value);
+  value = value_cast (fmt_type, value);
 
   /* Convert the value to a string and print it.  */
-  std::string str = decimal_to_string (dfp_ptr, dfp_len, byte_order);
+  std::string str;
+  if (TYPE_CODE (fmt_type) == TYPE_CODE_FLT)
+    str = floatformat_to_string (floatformat_from_type (fmt_type),
+				 value_contents (value), format);
+  else
+    str = decimal_to_string (value_contents (value),
+			     TYPE_LENGTH (fmt_type), byte_order, format);
   fputs_filtered (str.c_str (), stream);
-#endif
 }
 
 /* Subroutine of ui_printf to simplify it.
@@ -2550,43 +2542,6 @@  ui_printf (const char *arg, struct ui_fi
                                 obstack_base (&output));
 	    }
 	    break;
-	  case double_arg:
-	    {
-	      struct type *type = value_type (val_args[i]);
-	      DOUBLEST val;
-	      int inv;
-
-	      /* If format string wants a float, unchecked-convert the value
-		 to floating point of the same size.  */
-	      type = float_type_from_length (type);
-	      val = unpack_double (type, value_contents (val_args[i]), &inv);
-	      if (inv)
-		error (_("Invalid floating value found in program."));
-
-              fprintf_filtered (stream, current_substring, (double) val);
-	      break;
-	    }
-	  case long_double_arg:
-#ifdef HAVE_LONG_DOUBLE
-	    {
-	      struct type *type = value_type (val_args[i]);
-	      DOUBLEST val;
-	      int inv;
-
-	      /* If format string wants a float, unchecked-convert the value
-		 to floating point of the same size.  */
-	      type = float_type_from_length (type);
-	      val = unpack_double (type, value_contents (val_args[i]), &inv);
-	      if (inv)
-		error (_("Invalid floating value found in program."));
-
-	      fprintf_filtered (stream, current_substring,
-                                (long double) val);
-	      break;
-	    }
-#else
-	    error (_("long double not supported in printf"));
-#endif
 	  case long_long_arg:
 #ifdef PRINTF_HAS_LONG_LONG
 	    {
@@ -2612,9 +2567,14 @@  ui_printf (const char *arg, struct ui_fi
               fprintf_filtered (stream, current_substring, val);
 	      break;
 	    }
-	  /* Handles decimal floating values.  */
-	  case decfloat_arg:
-	    printf_decfloat (stream, current_substring, val_args[i]);
+	  /* Handles floating-point values.  */
+	  case double_arg:
+	  case long_double_arg:
+	  case dec32float_arg:
+	  case dec64float_arg:
+	  case dec128float_arg:
+	    printf_floating (stream, current_substring, val_args[i],
+			     fpieces[fr].argclass);
 	    break;
 	  case ptr_arg:
 	    printf_pointer (stream, current_substring, val_args[i]);
Index: binutils-gdb/gdb/dfp.c
===================================================================
--- binutils-gdb.orig/gdb/dfp.c
+++ binutils-gdb/gdb/dfp.c
@@ -146,12 +146,23 @@  decimal_to_number (const gdb_byte *from,
    16 bytes for decimal128.  */
 std::string
 decimal_to_string (const gdb_byte *decbytes, int len,
-		   enum bfd_endian byte_order)
+		   enum bfd_endian byte_order, const char *format)
 {
   gdb_byte dec[16];
 
   match_endianness (decbytes, len, byte_order, dec);
 
+  if (format != nullptr)
+    {
+      /* We don't handle format strings (yet).  If the host printf supports
+	 decimal floating point types, just use this.  Otherwise, fall back
+	 to printing the number while ignoring the format string.  */
+#if defined (PRINTF_HAS_DECFLOAT)
+      /* FIXME: This makes unwarranted assumptions about the host ABI!  */
+      return string_printf (format, dec);
+#endif
+    }
+
   std::string result;
   result.resize (MAX_DECIMAL_STRING);
 
Index: binutils-gdb/gdb/dfp.h
===================================================================
--- binutils-gdb.orig/gdb/dfp.h
+++ binutils-gdb/gdb/dfp.h
@@ -28,7 +28,8 @@ 
 #include "doublest.h"    /* For DOUBLEST.  */
 #include "expression.h"  /* For enum exp_opcode.  */
 
-extern std::string decimal_to_string (const gdb_byte *, int, enum bfd_endian);
+extern std::string decimal_to_string (const gdb_byte *, int, enum bfd_endian,
+				      const char *format = nullptr);
 extern bool decimal_from_string (gdb_byte *, int, enum bfd_endian,
 				 std::string string);
 extern void decimal_from_longest (LONGEST from, gdb_byte *to,
Index: binutils-gdb/gdb/doublest.c
===================================================================
--- binutils-gdb.orig/gdb/doublest.c
+++ binutils-gdb/gdb/doublest.c
@@ -789,10 +789,10 @@  floatformat_from_doublest (const struct
 }
 
 /* Convert the byte-stream ADDR, interpreted as floating-point format FMT,
-   to a string.  */
+   to a string, optionally using the print format FORMAT.  */
 std::string
 floatformat_to_string (const struct floatformat *fmt,
-		       const gdb_byte *in)
+		       const gdb_byte *in, const char *format)
 {
   /* Detect invalid representations.  */
   if (!floatformat_is_valid (fmt, in))
@@ -812,22 +812,50 @@  floatformat_to_string (const struct floa
       return string_printf ("%sinf", sign);
     }
 
-  /* Print the number using a format string where the precision is set
-     to the DECIMAL_DIG value for the given floating-point format.
-     This value is computed as
+  /* Determine the format string to use on the host side.  */
+  std::string host_format;
+  char conversion;
+
+  if (format == nullptr)
+    {
+      /* If no format was specified, print the number using a format string
+	 where the precision is set to the DECIMAL_DIG value for the given
+	 floating-point format.  This value is computed as
 
 		ceil(1 + p * log10(b)),
 
-     where p is the precision of the floating-point format in bits, and
-     b is the base (which is always 2 for the formats we support).  */
-  const double log10_2 = .30102999566398119521;
-  double d_decimal_dig = 1 + floatformat_precision (fmt) * log10_2;
-  int decimal_dig = d_decimal_dig;
-  if (decimal_dig < d_decimal_dig)
-    decimal_dig++;
+	 where p is the precision of the floating-point format in bits, and
+	 b is the base (which is always 2 for the formats we support).  */
+      const double log10_2 = .30102999566398119521;
+      double d_decimal_dig = 1 + floatformat_precision (fmt) * log10_2;
+      int decimal_dig = d_decimal_dig;
+      if (decimal_dig < d_decimal_dig)
+	decimal_dig++;
+
+      host_format = string_printf ("%%.%d", decimal_dig);
+      conversion = 'g';
+    }
+  else
+    {
+      /* Use the specified format, stripping out the conversion character
+         and length modifier, if present.  */
+      size_t len = strlen (format);
+      gdb_assert (len > 1);
+      conversion = format[--len];
+      gdb_assert (conversion == 'e' || conversion == 'f' || conversion == 'g'
+		  || conversion == 'E' || conversion == 'G');
+      if (format[len - 1] == 'L')
+	len--;
+
+      host_format = std::string (format, len);
+    }
 
-  std::string host_format
-    = string_printf ("%%.%d" DOUBLEST_PRINT_FORMAT, decimal_dig);
+  /* Add the length modifier and conversion character appropriate for
+     handling the host DOUBLEST type.  */
+#ifdef HAVE_LONG_DOUBLE
+  host_format += 'L';
+#endif
+  host_format += conversion;
 
   DOUBLEST doub;
   floatformat_to_doublest (fmt, in, &doub);
Index: binutils-gdb/gdb/doublest.h
===================================================================
--- binutils-gdb.orig/gdb/doublest.h
+++ binutils-gdb/gdb/doublest.h
@@ -72,7 +72,8 @@  extern const char *floatformat_mantissa
 					 const bfd_byte *);
 
 extern std::string floatformat_to_string (const struct floatformat *fmt,
-					  const gdb_byte *in);
+					  const gdb_byte *in,
+					  const char *format = nullptr);
 
 /* Return the floatformat's total size in host bytes.  */