summaryrefslogtreecommitdiff
path: root/gcc/doc/c-tree.texi
diff options
context:
space:
mode:
authorChao-ying Fu <fu@mips.com>2007-08-08 22:29:12 +0000
committerChao-ying Fu <chaoyingfu@gcc.gnu.org>2007-08-08 22:29:12 +0000
commit325217edf0332af8e5e52a32eb9b3cd862646375 (patch)
tree844eff658d57c57ce2ecfa8982a6e96a0c7c3392 /gcc/doc/c-tree.texi
parentad10fdc2b781c481fe7793dc9eb735d0bb493921 (diff)
downloadgcc-325217edf0332af8e5e52a32eb9b3cd862646375.tar.gz
tree.def (FIXED_POINT_TYPE): New type.
* tree.def (FIXED_POINT_TYPE): New type. (FIXED_CST): New constant. (FIXED_CONVERT_EXPR): New expr. * doc/c-tree.texi (Types): Document FIXED_POINT_TYPE. (Expressions): Document FIXED_CST and FIXED_CONVERT_EXPR. * tree.h (struct tree_base): Add saturating_flag. Remove one bit of spare for saturating_flag. (NUMERICAL_TYPE_CHECK): Support FIXED_POINT_TYPE. (NON_SAT_FIXED_POINT_TYPE_P, SAT_FIXED_POINT_TYPE_P, FIXED_POINT_TYPE_P): Define. (TYPE_SATURATING): Define. (TREE_FIXED_CST_PTR, TREE_FIXED_CST): Define. (struct tree_fixed_cst): New. (TYPE_IBIT, TYPE_FBIT): Define. (tree_node): Add fixed_cst. (enum tree_index): Add new enumeration values of TI_SAT_SFRACT_TYPE, TI_SAT_FRACT_TYPE, TI_SAT_LFRACT_TYPE, TI_SAT_LLFRACT_TYPE, TI_SAT_USFRACT_TYPE, TI_SAT_UFRACT_TYPE, TI_SAT_ULFRACT_TYPE, TI_SAT_ULLFRACT_TYPE, TI_SFRACT_TYPE, TI_FRACT_TYPE, TI_LFRACT_TYPE, TI_LLFRACT_TYPE, TI_USFRACT_TYPE, TI_UFRACT_TYPE, TI_ULFRACT_TYPE, TI_ULLFRACT_TYPE, TI_SAT_SACCUM_TYPE, TI_SAT_ACCUM_TYPE, TI_SAT_LACCUM_TYPE, TI_SAT_LLACCUM_TYPE, TI_SAT_USACCUM_TYPE, TI_SAT_UACCUM_TYPE, TI_SAT_ULACCUM_TYPE, TI_SAT_ULLACCUM_TYPE, TI_SACCUM_TYPE, TI_ACCUM_TYPE, TI_LACCUM_TYPE, TI_LLACCUM_TYPE, TI_USACCUM_TYPE, TI_UACCUM_TYPE, TI_ULACCUM_TYPE, TI_ULLACCUM_TYPE, TI_QQ_TYPE, TI_HQ_TYPE,_TYPE, TI_SQ_TYPE, TI_DQ_TYPE, TI_TQ_TYPE, TI_UQQ_TYPE, TI_UHQ_TYPE, TI_USQ_TYPE, TI_UDQ_TYPE, TI_UTQ_TYPE, TI_SAT_QQ_TYPE, TI_SAT_HQ_TYPE, TI_SAT_SQ_TYPE, TI_SAT_DQ_TYPE, TI_SAT_TQ_TYPE, TI_SAT_UQQ_TYPE, TI_SAT_UHQ_TYPE, TI_SAT_USQ_TYPE, TI_SAT_UDQ_TYPE, TI_SAT_UTQ_TYPE, TI_HA_TYPE, TI_SA_TYPE, TI_DA_TYPE, TI_TA_TYPE, TI_UHA_TYPE, TI_USA_TYPE, TI_UDA_TYPE, TI_UTA_TYPE, TI_SAT_HA_TYPE, TI_SAT_SA_TYPE, TI_SAT_DA_TYPE, TI_SAT_TA_TYPE, TI_SAT_UHA_TYPE, TI_SAT_USA_TYPE, TI_SAT_UDA_TYPE, TI_SAT_UTA_TYPE. (sat_short_fract_type_node, sat_fract_type_node, sat_long_fract_type_node, sat_long_long_fract_type_node, sat_unsigned_short_fract_type_node, sat_unsigned_fract_type_node, sat_unsigned_long_fract_type_node, sat_unsigned_long_long_fract_type_node, short_fract_type_node, fract_type_node, long_fract_type_node, long_long_fract_type_node, unsigned_short_fract_type_node, unsigned_fract_type_node, unsigned_long_fract_type_node, unsigned_long_long_fract_type_node, sat_short_accum_type_node, sat_accum_type_node, sat_long_accum_type_node, sat_long_long_accum_type_node, sat_unsigned_short_accum_type_node, sat_unsigned_accum_type_node, sat_unsigned_long_accum_type_node, sat_unsigned_long_long_accum_type_node, short_accum_type_node, accum_type_node, long_accum_type_node, long_long_accum_type_node, unsigned_short_accum_type_node, unsigned_accum_type_node, unsigned_long_accum_type_node, unsigned_long_long_accum_type_node, qq_type_node, hq_type_node, sq_type_node, dq_type_node, tq_type_node, uqq_type_node, uhq_type_node, usq_type_node, udq_type_node, utq_type_node, sat_qq_type_node, sat_hq_type_node, sat_sq_type_node, sat_dq_type_node, sat_tq_type_node, sat_uqq_type_node, sat_uhq_type_node, sat_usq_type_node, sat_udq_type_node, sat_utq_type_node, ha_type_node, sa_type_node, da_type_node, ta_type_node, uha_type_node, usa_type_node, uda_type_node, uta_type_node, sat_ha_type_node, sat_sa_type_node, sat_da_type_node, sat_ta_type_node, sat_uha_type_node, sat_usa_type_node, sat_uda_type_node, sat_uta_type_node): New macro. (make_fract_type, make_accum_type): Declare. (make_signed_fract_type, make_unsigned_fract_type, make_sat_signed_fract_type, make_sat_unsigned_fract_type, make_signed_accum_type, make_unsigned_accum_type, make_sat_signed_accum_type, make_sat_unsigned_accum_type, make_or_reuse_signed_fract_type, make_or_reuse_unsigned_fract_type, make_or_reuse_sat_signed_fract_type, make_or_reuse_sat_unsigned_fract_type, make_or_reuse_signed_accum_type, make_or_reuse_unsigned_accum_type, make_or_reuse_sat_signed_accum_type, make_or_reuse_sat_unsigned_accum_type): New macro. (fixed_zerop): Declare. * defaults.h (SHORT_FRACT_TYPE_SIZE, FRACT_TYPE_SIZE, LONG_FRACT_TYPE_SIZE, LONG_LONG_FRACT_TYPE_SIZE, SHORT_ACCUM_TYPE_SIZE, ACCUM_TYPE_SIZE, LONG_ACCUM_TYPE_SIZE, LONG_LONG_ACCUM_TYPE_SIZE): Define. * treestruct.def: Add TS_FIXED_CST. * Makefile.in (c-pretty-print.o): Add dependence on fixed-value.h. (tree.o): Likewise. (tree-dump.o): Likewise. (print-tree.o): Likewise. (tree-pretty-print.o): Likewise. (fold-const.o): Likewise. * tree-complex.c (some_nonzerop): Handle FIXED_CST. * tree-gimple.c (is_gimple_formal_tmp_rhs): Handle FIXED_CST. (is_gimple_min_invariant): Handle FIXED_CST. * stor-layout.c (int_mode_for_mode): Handle MODE_FRACT, MODE_UFRACT, MODE_ACCUM, MODE_UACCUM, MODE_VECTOR_FRACT, MODE_VECTOR_UFRACT, MODE_VECTOR_ACCUM, MODE_VECTOR_UACCUM. (layout_type): Handle FIXED_POINT_TYPE. (make_fract_type, make_accum_type): New functions. * tree-browser.c (browse_tree): Handle FIXED_POINT_TYPE. * tree-dump.c (fixed-value.h): New include. (dump_fixed): New function. (dequeue_and_dump): Handle FIXED_POINT_TYPE and FIXED_CST. * tree-inline.c (remap_type_1): Handle FIXED_POINT_TYPE. (estimate_num_insns_1): Handle FIXED_CST and FIXED_CONVERT_EXPR. * tree-pretty-print.c (fixed-value.h): New include. (dump_generic_node): Handle FIXED_POINT_TYPE, FIXED_CST, and FIXED_CONVERT_EXPR. * tree-scalar-evolution.c (get_scalar_evolution): Handle FIXED_CST. * tree-ssa-loop-im.c (for_each_index): Handle FIXED_CST. * tree-ssa-pre.c (poolify_tree): Handle FIXED_CST. * tree-ssa-reassoc.c (break_up_subtract_bb): We can do reassociation for non-saturating fixed-point types. (reassociate_bb): Likewise. * emit-rtl.c (fixed-value.h): New include. (fconst0, fconst1): New array. (init_emit_once): Initialize fconst0 and fconst1 for fixed-point modes. * tree-vect-generic.c expand_vector_operation): Support MODE_VECTOR_FRACT, MODE_VECTOR_UFRACT, MODE_VECTOR_ACCUM, and MODE_VECTOR_UACCUM. (type_for_widest_vector_mode): Add one parameter for the saturating flag. Check scalar FRACT, UFRACT, ACCUM, and UACCUM mode to select their vector mode. Pass the satp parameter to type_for_mode for fixed-point types. (expand_vector_operations_1): Pass the saturating flag to type_for_widest_vector_mode. Support MODE_VECTOR_FRACT, MODE_VECTOR_UFRACT, MODE_VECTOR_ACCUM, and MODE_VECTOR_UACCUM. * tree-vect-transform.c (vect_is_simple_cond): Support FIXED_CST. (vectorizable_condition): Likewise. * tree.c (fixed-value.h): New include. (tree_code_size): Support FIXED_CST. (build_fixed): New function. (build_one_cst): Support FIXED_POINT_TYPE for accum types. (fixed_zerop): New function. (tree_node_structure): Support FIXED_CST. (type_contains_placeholder_1): Support FIXED_POINT_TYPE. (build_type_attribute_qual_variant): Handle FIXED_POINT_TYPE. (type_hash_eq): Handle FIXED_POINT_TYPE. (simple_cst_equal): Support FIXED_CST. (iterative_hash_expr): Handle FIXED_CST. (get_unwidened): Make sure type is not FIXED_POINT_TYPE. (get_narrower): Likewise. (variably_modified_type_p): Handle FIXED_POINT_TYPE. (make_or_reuse_fract_type, make_or_reuse_accum_type): New functions. (build_common_tree_nodes_2): Use MAKE_FIXED_TYPE_NODE_FAMILY and MAKE_FIXED_MODE_NODE macros to initialize fixed-point type nodes. (build_vector_type_for_mode): Handle MODE_VECTOR_FRACT, MODE_VECTOR_UFRACT, MODE_VECTOR_ACCUM, MODE_VECTOR_UACCUM. (initializer_zerop): Support FIXED_CST. (walk_tree): Handle FIXED_CST and FIXED_POINT_TYPE. * dwarf2out.c (base_type_die): Use DW_ATE_signed_fixed or DW_ATE_unsigned_fixed to describe FIXED_POINT_TYPE. (is_base_type): Handle FIXED_POINT_TYPE. (add_type_attribute): Handle FIXED_POINT_TYPE. (gen_type_die_with_usage): Handle FIXED_POINT_TYPE. * print-tree.c (fixed-value.h): New include. (print_node_brief): Support FIXED_CST. (print_node): Support FIXED_POINT_TYPE and FIXED_CST. * c-pretty-print.c (fixed-value.h): New include. (pp_c_type_specifier): Handle FIXED_POINT_TYPE. Need to pass TYPE_SATURATING to c_common_type_for_mode for fixed-point modes. (pp_c_direct_abstract_declarator): Handle FIXED_POINT_TYPE. Support fixed-point types for inner items in VECTOR_TYPE. (pp_c_direct_declarator): Likewise. (pp_c_declarator): Likewise. (pp_c_fixed_constant): New function. (pp_c_constant): Handle FIXED_CST. (pp_c_primary_expression): Likewise. (pp_c_expression): Likewise. * fold-const.c (fixed-value.h): New include. (negate_expr_p): Return true for FIXED_CST. (fold_negate_expr): Support FIXED_CST. (split_tree): Support FIXED_CST. (const_binop): Support FIXED_CST. (fold_convert_const_int_from_fixed): New function to convert from fixed to int. (fold_convert_const_real_from_fixed): New function to convert from fixed to real. (fold_convert_const_fixed_from_fixed): New function to convert from fixed to another fixed. (fold_convert_const_fixed_from_int): New function to convert from int to fixed. (fold_convert_const_fixed_from_real): New function to convert from real to fixed. (fold_convert_const): Support conversions from fixed to int, from fixed to real, from fixed to fixed, from int to fixed, and from real to fixed. (fold_convert): Support FIXED_CST and FIXED_POINT_TYPE. (operand_equal_p): Support FIXED_CST. (make_range): For fixed-point modes, we need to pass the saturating flag as the 2nd parameter. (tree_swap_operands_p): Handle FIXED_CST. (fold_plusminus_mult_expr): For fract modes, we cannot generate constant 1. (fold_unary): Support FIXED_CONVERT_EXPR. (fold_binary): Handle FIXED_CST. Make sure the type is not saturating, before associating operations. Ex: A + B + C, A * B * C, (A1 * C1) +/- (A2 * C2). (tree_expr_nonnegative_warnv_p): Handle FIXED_CST. (fold_negate_const): Support FIXED_CST. (fold_relational_const): Support FIXED_CST. * gimplify.c (omp_firstprivatize_type_sizes): Handle FIXED_POINT_TYPE. (gimplify_expr): Handle FIXED_CST. (gimplify_type_sizes): Handle FIXED_POINT_TYPE. * ipa-prop.c (ipa_callsite_compute_param): Support FIXED_CST. * ipa-type-escape.c (type_to_consider): Handle FIXED_POINT_TYPE. * doc/tm.texi (Type Layout): Document SHORT_FRACT_TYPE_SIZE, FRACT_TYPE_SIZE, LONG_FRACT_TYPE_SIZE, LONG_LONG_FRACT_TYPE_SIZE, SHORT_ACCUM_TYPE_SIZE, ACCUM_TYPE_SIZE, LONG_ACCUM_TYPE_SIZE, LONG_LONG_ACCUM_TYPE_SIZE. * dbxout.c (dbxout_type): Handle FIXED_POINT_TYPE. * c-aux-info.c (gen_type): Handle FIXED_POINT_TYPE. * tree-sra.c (is_sra_scalar_type): Support FIXED_POINT_TYPE. * expmed.c (extract_bit_field): Support MODE_FRACT, MODE_UFRACT, MODE_ACCUM, and MODE_UACCUM. * tree-vectorizer.c (vect_is_simple_reduction): Check for saturating fixed-point types to disable reduction. * explow.c (promote_mode): Support FIXED_POINT_TYPE. From-SVN: r127306
Diffstat (limited to 'gcc/doc/c-tree.texi')
-rw-r--r--gcc/doc/c-tree.texi32
1 files changed, 32 insertions, 0 deletions
diff --git a/gcc/doc/c-tree.texi b/gcc/doc/c-tree.texi
index 4134007c24f..a53329a7a6b 100644
--- a/gcc/doc/c-tree.texi
+++ b/gcc/doc/c-tree.texi
@@ -271,6 +271,7 @@ The elements are indexed from zero.
@tindex TYPE_MIN_VALUE
@tindex TYPE_MAX_VALUE
@tindex REAL_TYPE
+@tindex FIXED_POINT_TYPE
@tindex COMPLEX_TYPE
@tindex ENUMERAL_TYPE
@tindex BOOLEAN_TYPE
@@ -498,6 +499,19 @@ Used to represent the @code{float}, @code{double}, and @code{long
double} types. The number of bits in the floating-point representation
is given by @code{TYPE_PRECISION}, as in the @code{INTEGER_TYPE} case.
+@item FIXED_POINT_TYPE
+Used to represent the @code{short _Fract}, @code{_Fract}, @code{long
+_Fract}, @code{long long _Fract}, @code{short _Accum}, @code{_Accum},
+@code{long _Accum}, and @code{long long _Accum} types. The number of bits
+in the fixed-point representation is given by @code{TYPE_PRECISION},
+as in the @code{INTEGER_TYPE} case. There may be padding bits, fractional
+bits and integral bits. The number of fractional bits is given by
+@code{TYPE_FBIT}, and the number of integral bits is given by @code{TYPE_IBIT}.
+The fixed-point type is unsigned if @code{TYPE_UNSIGNED} holds; otherwise,
+it is signed.
+The fixed-point type is saturating if @code{TYPE_SATURATING} holds; otherwise,
+it is not saturating.
+
@item COMPLEX_TYPE
Used to represent GCC built-in @code{__complex__} data types. The
@code{TREE_TYPE} is the type of the real and imaginary parts.
@@ -1879,6 +1893,7 @@ This macro returns the attributes on the type @var{type}.
@findex tree_int_cst_lt
@findex tree_int_cst_equal
@tindex REAL_CST
+@tindex FIXED_CST
@tindex COMPLEX_CST
@tindex VECTOR_CST
@tindex STRING_CST
@@ -1907,6 +1922,7 @@ This macro returns the attributes on the type @var{type}.
@tindex NON_LVALUE_EXPR
@tindex NOP_EXPR
@tindex CONVERT_EXPR
+@tindex FIXED_CONVERT_EXPR
@tindex THROW_EXPR
@tindex LSHIFT_EXPR
@tindex RSHIFT_EXPR
@@ -2080,6 +2096,15 @@ its bit-pattern.
FIXME: Talk about how to obtain representations of this constant, do
comparisons, and so forth.
+@item FIXED_CST
+
+These nodes represent fixed-point constants. The type of these constants
+is obtained with @code{TREE_TYPE}. @code{TREE_FIXED_CST_PTR} points to
+to struct fixed_value; @code{TREE_FIXED_CST} returns the structure itself.
+Struct fixed_value contains @code{data} with the size of two
+HOST_BITS_PER_WIDE_INT and @code{mode} as the associated fixed-point
+machine mode for @code{data}.
+
@item COMPLEX_CST
These nodes are used to represent complex number constants, that is a
@code{__complex__} whose parts are constant nodes. The
@@ -2246,6 +2271,13 @@ cases are always indicated explicitly. Similarly, a user-defined
conversion is never represented by a @code{CONVERT_EXPR}; instead, the
function calls are made explicit.
+@item FIXED_CONVERT_EXPR
+These nodes are used to represent conversions that involve fixed-point
+values. For example, from a fixed-point value to another fixed-point value,
+from an integer to a fixed-point value, from a fixed-point value to an
+integer, from a floating-point value to a fixed-point value, or from
+a fixed-point value to a floating-point value.
+
@item THROW_EXPR
These nodes represent @code{throw} expressions. The single operand is
an expression for the code that should be executed to throw the