rtl.c (rtx_code_size): Check CONST_FIXED to calcualte correct sizes in DEF_RTL_EXPR.

	* rtl.c (rtx_code_size): Check CONST_FIXED to calcualte correct sizes
	in DEF_RTL_EXPR.
	(copy_rtx): Handle CONST_FIXED.
	(rtx_equal_p): Likewise.
	* rtl.h (fixed_value.h): New include.
	(rtx_def): Add a new field of fixed_value to u.
	(XCNMPFV): Define for accessing fixed_value.
	(CONST_FIXED_VALUE, CONST_FIXED_VALUE_HIGH, CONST_FIXED_VALUE_LOW):
	Define.
	* rtl.def (CONST_FIXED): New constant.
	(SS_MULT, US_MULT, SS_DIV, US_DIV, FRACT_CONVERT,
	UNSIGNED_FRACT_CONVERT, SAT_FRACT, UNSIGNED_SAT_FRACT, US_NEG,
	US_ASHIFT): New codes.
	* doc/rtl.texi (Expressions): Document const_fixed, us_neg, ss_mult,
	us_mult, ss_div, us_div, us_ashift, fract_convert, sat_fract,
	unsigned_fract_convert, unsigned_sat_fract): Document them.
	* varasm.c (assemble_integer): Extend to support fixed-point constants
	by using different machine classes.
	(decode_addr_const): Handle FIXED_CST.
	(const_hash_1): Likewise.
	(compare_constant): Likewise.
	(copy_constant): Likewise.
	(const_rtx_hash_1): Handle CONST_FIXED.
	(output_constant_pool_2): Handle MODE_FRACT, MODE_UFRACT, MODE_ACCUM,
	MODE_UACCUM, MODE_VECTOR_FRACT, MODE_VECTOR_UFRACT, MODE_VECTOR_ACCUM,
	MODE_VECTOR_UACCUM.
	(initializer_constant_valid_p): Handle FIXED_CST.
	(output_constant): Support FIXED_POINT_TYPE.
	* gengenrtl.c (excluded_rtx): Check CONST_FIXED to exclude.
	* cse.c (hash_rtx): Support CONST_FIXED.
	(exp_equiv_p): Likewise.
	(cannon_reg): Likewise.
	(fold_rtx): Likewise.
	(equiv_constant): Likewise.
	(cse_process_notes_1): Likewise.
	(count_reg_usage): Likewise.
	* cselib.c (entry_and_rtx_equal_p): Check CONST_FIXED.
	(rtx_equal_for_cselib_p): Handle CONST_FIXED.
	(wrap_constant): Check CONST_FIXED.
	(cselib_hash_rtx): Support CONST_FIXED.
	(cselib_subst_to_values): Likewise.
	* df-scan.c (df_uses_record): Likewise.
	* gcse.c (want_to_gcse_p): Likewise.
	(oprs_unchanged_p): Likewise.
	(oprs_not_set_p): Likewise.
	(compute_transp): Likewise.
	(extract_mentioned_regs_helper): Likewise.
	* genemit.c (gen_exp): Likewise.
	* local-alloc.c (equiv_init_varies_p): Likewise.
	(contains_replace_regs): Likewise.
	(memref_referenced_p): Likewise.
	* loop-invariant.c (check_maybe_invariant): Likewise.
	(hash_invariant_expr_1): Likewise.
	(invariant_expr_equal_p): Likewise.
	* postreload-gcse.c (oprs_unchanged_p): Likewise.
	* regclass.c (reg_scan_mark_refs): Likewise.
	* regrename.c (scan_rtx): Likewise.
	* resource.c (mark_referenced_resources): Likewise.
	(mark_set_resources): Likewise.
	* rtlanal.c (rtx_unstable_p): Likewise.
	(rtx_varies_p): Likewise.
	(count_occurrences): Likewise.
	(reg_mentioned_p): Likewise.
	(modified_between_p): Likewise.
	(modified_in_p): Likewise.
	(volatile_insn_p): Likewise.
	(volatile_refs_p): Likewise.
	(side_effects_p): Likewise.
	(may_trap_p_1): Likewise.
	(inequality_comparisons_p): Likewise.
	(computed_jump_p_1): Likewise.
	(commutative_operand_precedence): Likewise.
	* sched-deps.c (sched_analyze_2): Likewise.
	* sched-vis.c (print_value): Likewise.
	* reload.c (operands_match_p): Likewise.
	(subst_reg_equivs): Likewise.
	* reload1.c (eliminate_regs_1): Likewise.
	(elimination_effects): Likewise.
	(scan_paradoxical_subregs): Likewise.
	* alias.c (rtx_equal_for_memref_p): Likewise.
	* Makefile.in (RTL_BASE_H): Add fixed-value.h.
	* emit-rtl.c (const_fixed_htab): New hash table.
	(const_fixed_htab_hash, const_fixed_htab_eq, lookup_const_fixed):
	Declare.
	(const_fixed_htab_hash, const_fixed_htab_eq, lookup_const_fixed,
	const_fixed_from_fixed_value): New functions.
	(verify_rtx_sharing): Handle CONST_FIXED.
	(copy_rtx_if_shared_1): Likewise.
	(reset_used_flags): Likewise.
	(set_used_flags): Likewise.
	(copy_insn_1): Likewise.
	(init_emit_once): Create const_fixed_htab.
	Store fixed-point scalar and vector zero and one to const_tiny_rtx.

From-SVN: r127725

1 files changed, 70 insertions, 6 deletions

diff --git a/gcc/doc/rtl.texi b/gcc/doc/rtl.texi
index 247a22e..83fa089 100644
--- a/gcc/doc/rtl.texi
+++ b/gcc/doc/rtl.texi
@@ -1492,6 +1492,15 @@ bits but small enough to fit within twice that number of bits (GCC
 does not provide a mechanism to represent even larger constants).  In
 the latter case, @var{m} will be @code{VOIDmode}.
 
+@findex const_fixed
+@item (const_fixed:@var{m} @var{addr})
+Represents a fixed-point constant of mode @var{m}.
+The data structure, which contains data with the size of two
+@code{HOST_BITS_PER_WIDE_INT} and the associated fixed-point mode,
+is access with the macro @code{CONST_FIXED_VALUE}.  The high part of data
+is accessed with @code{CONST_FIXED_VALUE_HIGH}; the low part is accessed
+with @code{CONST_FIXED_VALUE_LOW}.
+
 @findex const_vector
 @item (const_vector:@var{m} [@var{x0} @var{x1} @dots{}])
 Represents a vector constant.  The square brackets stand for the vector
@@ -2013,24 +2022,35 @@ still known.
 
 @findex neg
 @findex ss_neg
+@findex us_neg
 @cindex negation
 @cindex negation with signed saturation
+@cindex negation with unsigned saturation
 @item (neg:@var{m} @var{x})
 @itemx (ss_neg:@var{m} @var{x})
+@itemx (us_neg:@var{m} @var{x})
 These two expressions represent the negation (subtraction from zero) of
 the value represented by @var{x}, carried out in mode @var{m}.  They
 differ in the behavior on overflow of integer modes.  In the case of
 @code{neg}, the negation of the operand may be a number not representable
 in mode @var{m}, in which case it is truncated to @var{m}.  @code{ss_neg}
-ensures that an out-of-bounds result saturates to the maximum or minimum
-representable value.
+and @code{us_neg} ensure that an out-of-bounds result saturates to the
+maximum or minimum signed or unsigned value.
 
 @findex mult
+@findex ss_mult
+@findex us_mult
 @cindex multiplication
 @cindex product
+@cindex multiplication with signed saturation
+@cindex multiplication with unsigned saturation
 @item (mult:@var{m} @var{x} @var{y})
+@itemx (ss_mult:@var{m} @var{x} @var{y})
+@itemx (us_mult:@var{m} @var{x} @var{y})
 Represents the signed product of the values represented by @var{x} and
 @var{y} carried out in machine mode @var{m}.
+@code{ss_mult} and @code{us_mult} ensure that an out-of-bounds result
+saturates to the maximum or minimum signed or unsigned value.
 
 Some machines support a multiplication that generates a product wider
 than the operands.  Write the pattern for this as
@@ -2046,14 +2066,19 @@ For unsigned widening multiplication, use the same idiom, but with
 @code{zero_extend} instead of @code{sign_extend}.
 
 @findex div
+@findex ss_div
 @cindex division
 @cindex signed division
+@cindex signed division with signed saturation
 @cindex quotient
 @item (div:@var{m} @var{x} @var{y})
+@itemx (ss_div:@var{m} @var{x} @var{y})
 Represents the quotient in signed division of @var{x} by @var{y},
 carried out in machine mode @var{m}.  If @var{m} is a floating point
 mode, it represents the exact quotient; otherwise, the integerized
 quotient.
+@code{ss_div} ensures that an out-of-bounds result saturates to the maximum
+or minimum signed value.
 
 Some machines have division instructions in which the operands and
 quotient widths are not all the same; you should represent
@@ -2065,9 +2090,13 @@ such instructions using @code{truncate} and @code{sign_extend} as in,
 
 @findex udiv
 @cindex unsigned division
+@cindex unsigned division with unsigned saturation
 @cindex division
 @item (udiv:@var{m} @var{x} @var{y})
+@itemx (us_div:@var{m} @var{x} @var{y})
 Like @code{div} but represents unsigned division.
+@code{us_div} ensures that an out-of-bounds result saturates to the maximum
+or minimum unsigned value.
 
 @findex mod
 @findex umod
@@ -2131,18 +2160,21 @@ fixed-point mode.
 
 @findex ashift
 @findex ss_ashift
+@findex us_ashift
 @cindex left shift
 @cindex shift
 @cindex arithmetic shift
 @cindex arithmetic shift with signed saturation
+@cindex arithmetic shift with unsigned saturation
 @item (ashift:@var{m} @var{x} @var{c})
 @itemx (ss_ashift:@var{m} @var{x} @var{c})
-These two expressions represent the result of arithmetically shifting @var{x}
+@itemx (us_ashift:@var{m} @var{x} @var{c})
+These three expressions represent the result of arithmetically shifting @var{x}
 left by @var{c} places.  They differ in their behavior on overflow of integer
 modes.  An @code{ashift} operation is a plain shift with no special behavior
-in case of a change in the sign bit; @code{ss_ashift} saturates to the minimum
-or maximum representable value if any of the bits shifted out differs from the
-final sign bit.
+in case of a change in the sign bit; @code{ss_ashift} and @code{us_ashift}
+saturates to the minimum or maximum representable value if any of the bits
+shifted out differs from the final sign bit.
 
 @var{x} have mode @var{m}, a fixed-point machine mode.  @var{c}
 be a fixed-point mode or be a constant with mode @code{VOIDmode}; which
@@ -2529,6 +2561,38 @@ When @var{m} is a floating point mode, represents the result of
 converting floating point value @var{x} (valid for mode @var{m}) to an
 integer, still represented in floating point mode @var{m}, by rounding
 towards zero.
+
+@findex fract_convert
+@item (fract_convert:@var{m} @var{x})
+Represents the result of converting fixed-point value @var{x} to
+fixed-point mode @var{m}, signed integer value @var{x} to
+fixed-point mode @var{m}, floating-point value @var{x} to
+fixed-point mode @var{m}, fixed-point value @var{x} to integer mode @var{m}
+regarded as signed, or fixed-point value @var{x} to floating-point mode @var{m}.
+When overflows or underflows happen, the results are undefined.
+
+@findex sat_fract
+@item (sat_fract:@var{m} @var{x})
+Represents the result of converting fixed-point value @var{x} to
+fixed-point mode @var{m}, signed integer value @var{x} to
+fixed-point mode @var{m}, or floating-point value @var{x} to
+fixed-point mode @var{m}.
+When overflows or underflows happen, the results are saturated to the
+maximum or the minimum.
+
+@findex unsigned_fract_convert
+@item (unsigned_fract_convert:@var{m} @var{x})
+Represents the result of converting fixed-point value @var{x} to
+integer mode @var{m} regarded as unsigned, or unsigned integer value @var{x} to
+fixed-point mode @var{m}.
+When overflows or underflows happen, the results are undefined.
+
+@findex unsigned_sat_fract
+@item (unsigned_sat_fract:@var{m} @var{x})
+Represents the result of converting unsigned integer value @var{x} to
+fixed-point mode @var{m}.
+When overflows or underflows happen, the results are saturated to the
+maximum or the minimum.
 @end table
 
 @node RTL Declarations