* fold-const.c (tree_expr_nonzero_p): Add function prototype.
(fold) <EQ_EXPR>: Move tree_expr_nonzero_p optimization from
fold_relational_const to here, i.e. "(x | 5) == 0" -> false.
(fold) (UNEQ_EXPR>: Add optimizations for unordered comparisons
of the form "x op x" where op is UNLE, UNGE, UNEQ or LTGT.
(fold_relational_const): Tidy up handling of floating point
comparisons by calling real_compare. Remove tree_expr_nonzero_p
transformation; fold_relational_const assumes constant operands.
From-SVN: r84916
* fold-const.c (fold) <PLUS_EXPR>: Canonicalize X + -C as X - C for
floating point additions, to keep real immediate constant positive.
<MINUS_EXPR>: For floating point subtractions, only transform X - -C
into X + C, and leave positive real constants as X - C.
From-SVN: r84535
* builtins.c (fold_builtin_fputs): Don't bother converting the
return type to integer_type_node, as we've already checked that
the result will be ignored.
* tree-eh.c (tree_could_trap_p): Add support for -ftrapv such
that signed addition, subtraction, multiplication, division,
remainder, negation and absolute value may potentially trap.
* fold-const.c (fold_ignored_result): New function to strip
non-side-effecting tree nodes from an expression whose result
is ignored.
(fold_convert): Call fold_ignored_result when casting a value
to VOID_TYPE.
(omit_one_operand): Call fold_ignored_result on the "omitted"
operand when building a COMPOUND_EXPR.
(pedantic_omit_one_operand): Likewise.
* tree.h (fold_ignored_result): Prototype here.
* tree-ssa-ccp.c (ccp_fold_builtin): Call fold_ignored_result
when we're going to ignore the result.
From-SVN: r84525
PR tree-optimization/16437
* c-common.c (shorten_compare): Don't mark result of conversion to
narrower signed type as overflowing.
* fold-const.c (decode_field_reference): Determine whether
signedness comes from outer type using precision rather than size.
testsuite:
* gcc.c-torture/execute/bitfld-4.c: New test.
From-SVN: r84498
* builtins.c (fold_builtin_strcpy): Merge functionality from
simplify_builtin_strcpy. Add additional len argument. No longer
static. Remove function prototype.
(fold_builtin_strncpy): Likewise integrate functionality from
simplify_builtin_strncpy. Add additional slen argument. No
longer static. Remove function prototype.
(simplify_builtin_strcy, simplify_builtin_strncpy): Delete.
(simplify_builtin_fputs): Rename to fold_builtin_fputs. Change
types of "ignore" and "unlocked" parameters to bool.
(fold_builtin_1): Add additional ignore argument. Call renamed
fold_builtin_fputs to simplify GCC "fputs" and "fputs_unlocked"
builtins. Update arguments to fold_builtin_strncpy and
fold_builtin_strcpy. Add function prototype.
(fold_builtin): Add additional Boolean ignore argument to pass
to fold_builtin_1.
(simplify_builtin): Call fold_builtin_fputs, fold_builtin_strcpy
and fold_builtin_strncpy instead of simplify_builtin_fputs,
simplify_builtin_strcpy and simplify_builtin_strncpy respectively.
* expr.h (simplify_builtin_fputs, simplify_builtin_strcpy,
simplify_builtin_strncpy): Delete function prototypes.
* tree.h (fold_builtin_fputs, fold_builtin_strcpy,
fold_builtin_strncpy): Add function prototypes here.
(fold_builtin): Update function prototype with new "bool ignore".
* tree-ssa-ccp.c (ccp_fold): Update call to fold_builtin.
(ccp_fold_builtin): Update call to fold_builtin. Call
fold_builtin_fputs, fold_builtin_strcpy and fold_builtin_strncpy
instead of simplify_builtin_fputs, simplify_builtin_strcpy and
simplify_builtin_strncpy respectively.
* fold-const.c (fold): Update call to fold_builtin.
From-SVN: r84302
* fold-const.c (fold) <TRUNC_MOD_EXPR>: Optimize unsigned modulus
by a power of two into a bit-wise AND, i.e. "X % C" as "X & (C-1)".
Normalize "X % C" as "X % -C" for signed modulus and negative C.
Optimize "X % -Y" as "X % Y" for signed modulus.
<EQ_EXPR>: Recursively call "fold" when transforming "(X % Y) == 0"
into "((unsigned) X % Y) == 0".
From-SVN: r84122
2004-06-28 Paolo Bonzini <bonzini@gnu.org>
* fold-const.c (fold_cond_expr_with_comparison): Add ARG1
parameter. Use it instead of ARG00 to produce the result.
From-SVN: r83768
* fold-const.c (fold) <BIT_IOR_EXPR>: Optimize ~X|X and X|~X as -1.
<BIT_XOR_EXPR>: Optimize ~X|X and X|~X as -1.
<BIT_AND_EXPR>: Optimize ~X&X and X&~X as 0.
<TRUTH_AND_EXPR, TRUTH_ANDIF_EXPR>: Optimize !X&&X and X&&!X as false.
<TRUTH_OR_EXPR, TRUTH_ORIF_EXPR>: Optimize !X||X and !X||X as true.
<TRUTH_XOR_EXPR>: Optimize !X^X and X^X! as true. Now that
TRUTH_XOR_EXPR is a commutative tree code, don't test whether arg0
is a constant.
* gcc.dg/20040527-1.c: New test case.
From-SVN: r83733
2004-06-23 Andrew Pinski <apinski@apple.com>
PR middle-end/15988
* fold-const.c (fold_convert): Types which are compatible
can be converted with only a NOP_EXPR.
2004-06-23 Andrew Pinski <apinski@apple.com>
PR middle-end/15988
* g++.dg/opt/ptrmem4.C: New test.
From-SVN: r83585
2004-06-23 Eric Christopher <echristo@redhat.com>
* fold-const.c (make_range): Cleanup type checking through function.
Remove orig_type. Replace with checks to exp_type and arg0_type.
Clarify comment when converting from unsigned to signed.
From-SVN: r83576
* alias.c (adjust_offset_for_component_ref): Use
component_ref_field_offset.
* c-decl.c (build_array_declarator): Add news args for ARRAY_REF.
* c-gimplify.c (gimplify_expr_stmt): Use alloc_stmt_list.
(gimplify_decl_stmt): Call gimplify_type_sizes for type.
For decl, call gimplify_one_sizepos and use statement list.
(gimplify_compound_literal_expr): New arg PRE_P.
Add statement to PRE_P list and return DECL.
(c_gimplify_expr, case COMPOUND_LITERAL_EXPR): Add arg to
gimplify_compound_literal_expr.
* c-tree.h (getdecls): Deleted.
* c-typeck.c (build_component_ref): Add operand for COMPONENT_REF.
(build_array_ref): Add two operands for ARRAY_REF.
(build_unary_op): Set TREE_INVARIANT and TREE_CONSTANT for
COMPOUND_LITERAL_EXPR.
* coverage.c (tree_coverage_counter_ref): Add new operands
for ARRAY_REF.
* emit-rtl.c (component_ref_for_mem_expr): Add new operand
for COMPONENT_REF.
(set_mem_attributes_minus_bitpos): Use array_ref_low_bound
and array_ref_element_size.
(widen_memory_access):Use component_ref_field_offset.
* explow.c (update_nonlocal_goto_save_area): Add two operands
for ARRAY_REF.
* expr.c (array_ref_element_size, array_ref_low_bound): New functions.
(component_ref_field_offset): Likewise.
(get_inner_reference): Use them.
(expand_expr_real_1, case ARRAY_REF): Use array_ref_low_bound.
* fold-const.c (fold, case EQ_EXPR): Properly handle DECL_SIZE.
(fold_read_from_constant_string): Use array_ref_low_bound.
Verify that result is a character type.
(build_fold_indirect_ref): Add two operands for ARRAY_REF.
* function.c (expand_function_start): Likewise.
* gimple-low.c (expand_var_p): Delete duplicated line.
* gimplify.c: Add static decls for local functions.
(cgraph.h): Now included.
(create_tmp_var): Remove check for ARRAY_TYPE.
(copy_if_shared_r): Look at bounds and sizes of types.
(build_and_jump): Return alloc_stmt_list instead of build_empty_stmt.
(gimplify_exit_expr, shortcut_cond_expr): Likewise.
(gimplify_save_expr, gimple_push_cleanup): Likewise.
(gimplify_init_constructor): Likewise.
WANT_VALUE now bool.
If empty list with no result wanted, return GS_UNHANDLED.
Add additional operands for ARRAY_REF and COMPONENT_REF.
(canonicalize_component_ref): Convert to &array[L].
(gimplify_array_ref_to_plus): Use array_ref_element_size and
array_ref_lower_bound.
(build_addr_expr_with_type, build_addr_expr): New functions.
(gimplify_compound_lval): WANT_LVALUE now bool.
Major rework to allow handle_component_p and initialize and
gimplify new operands for ARRAY_REF, ARRAY_RANGE_REF, and
COMPONENT_REF.
(gimplify_array_ref): Deleted.
(gimplify_self_mod_expr): WANT_VALUE now bool.
(gimplify_modify_expr): Gimplify to_p and from_p later.
Factor out code into gimplify_modify_expr_rhs and call twice.
Move variable-size code earlier and handle PLACEHOLDER_EXPR.
(gimplify_modify_expr_rhs, gimplify_variable_sized_compare): New fns.
(gimplify_addr_expr, case VIEW_CONVERT_EXPR): New case.
(gimplify_expr, case ARRAY_REF): Delete special case.
Instead handle like COMPONENT_REF; also do ARRAY_RANGE_REF,
IMAGPART, and REALPART the same way.
(gimplify_expr, case VIEW_CONVERT_EXPR): New case.
(gimplify_expr): Call gimplify_variable_sized_compare if applicable.
Call alloc_stmt_list instead of build_empty_stmt.
Deal with _REF that's volatile.
(gimplify_type_sizes, gimplify_one_sizepos): New functions.
(unshare_body, unvisit_body): New functions.
(gimplify_body): Call them.
* stmt.c (expand_stack_alloc): Don't expand TYPE_MAX_VALUE.
* stor-layout.c (get_pending_sizes): Don't change SAVE_EXPR_CONTEXT.
* tree-alias-common.c (get_alias_var): Also skip ARRAY_RANGE_REF.
* tree-cfg.c (tree_node_can_be_shared): Treat ARRAY_RANGE_REF
like ARRAY_REF.
(verify_expr, case ADDR_EXPR): Use handled_component_p.
* tree-dfa.c (get_virtual_var): Likewise.
* tree-dump.c (dequeue_and_dump, case COMPONENT_REF, ARRAY_REF):
New cases to dump new operands; likewise for ARRAY_RANGE_REF.
* tree-eh.c (tree_could_trap, case ARRAY_RANGE_REF): Like ARRAY_REF.
* tree-gimple.c (is_gimple_addr_expr_arg): Add ARRAY_RANGE_REF
and INDIRECT_REF.
(get_base_address): Use handled_component_p.
* tree-gimple.h (gimplify_type_sizes, gimplify_one_sizepos): New.
* tree-line.c (walk_tree): Walk more things for types and decls.
* tree-mudflap.c (mf_build_check_statement_for): Add new operands
for ARRAY_REF and COMPONENT_REF.
(mx_xform_derefs_1): Clean up usage of decl sizes.
* tree-nested.c (build_addr): Use handled_component_p.
(walk_stmts, case CATCH_EXPR): Add missing "break".
(get_static_chain, get_frame_field): Add new operand for COMPONENT_REF.
(finalize_nesting_tree_1): Likewise.
(convert_nonlocal_reference, case ARRAY_RANGE_REF): Like ARRAY_REF
and process additional operands.
(convert_local_reference): Likewise.
* tree-outof-ssa.c (discover_nonconstant_array_refs_r): Treat
ARRAY_RANGE_REF similarly to ARRAY_REF.
* tree-pretty-print.c (dump_generic_node, case QUAL_UNION_TYPE): Handle
like RECORD_TYPE.
(dump_generic_node, case COMPONENT_REF): Print offset operand.
(dump_generic_node, case ARRAY_RANGE_REF): Treat like ARRAY_REF
and print lower bound and element size for both.
(op_prio, case ARRAY_RANGE_REF): Like ARRAY_REF.
* tree-sra.c (csc_build_component_ref): Add new operand.
(scalarize_call_expr): Use get_base_address.
* tree-ssa-ccp.c (widen_bitfield): Clean up size handling.
(maybe_fold_offset_to_array_ref): Rework to handle input having an
ARRAY_REF, refine handling of lower bound, and add new operands
for ARRAY_REF.
(maybe_fold_to_component_ref): Add new operand for COMPONENT_REF.
(maybe_fold_stmt_indirect): Only fold *&B to B if types match.
(maybe_fold_stmt_addition): Only handle constant lower bound.
* tree-ssa-operands.c (get_expr_operands): Minor rearrangements.
Treat ARRAY_REF and ARRAY_RANGE_REF the same; look at extra operands.
Look at new offset operand of COMPONENT_REF.
* tree-ssa.c (set_is_used): Use handled_component_p.
* tree.c (substitute_in_expr, case COMPONENT_REF): Add new operand.
(stabilize_reference, case COMPONENT_REF): Likewise.
(stabilize_reference, case ARRAY_RANGE_REF, ARRAY_REF): Similarly.
(recompute_tree_invariant_for_addr_expr): Completely rework to
be more precise. Also set TREE_SIDE_EFFECTS.
(build1_stat, case ARRAY_EXPR): Don't handle TREE_SIDE_EFFECTS here.
(build2_stat, build3_stat, build4_stat): For references,
propagate TREE_THIS_VOLATILE.
(get_unwidened): Add new operand for COMPONENT_REF.
(get_narrower): Likewise; use host_integerp for DECL_SIZE.
* tree.def (COMPONENT_REF): Add new operand.
(ARRAY_REF, ARRAY_RANGE_REF): Add two new operands.
* tree.h (array_ref_element_size, array_ref_low_bound): New decls.
(component_ref_field_offset): Likewise.
* config/alpha/alpha.c (alpha_va_start): Add new op for COMPONENT_REF.
(alpha_gimplify_va_arg): Likewise.
* config/i386/i386.c (ix86_va_start, ix86_gimplify_va_arg): Likewise.
* config/i860/i860.c (i860_va_start, i860_va_arg): Likewise.
* config/iq2000/iq2000.c (iq2000_va_arg): Likewise.
* config/mips/mips.c (mips_va_start, mips_va_arg): Likewise.
* config/rs6000/rs6000.c (rs6000_va_start, rs6000_gimplify_va_arg):
Likewise.
* config/s390/s390.c (s390_va_start, s390_gimplify_va_arg): Likewise.
* config/sh/sh.c (sh_va_start, sh_va_arg): Likewise.
* config/stormy16/stormy16.c (xstormy1_expand_builin_va_start):
Likewise.
(xstormy16_expand_builtin_va_arg): Likewise.
* config/xtensa/xtensa.c (xtensa_va_start, xtensa_va_arg): Likewise.
* cp/call.c (build_vfield_ref): Add new operand for COMPONENT_REF.
(build_new_method_call): Likewise.
* cp/decl.c (local_variable_p_walkfn): Don't walk into types.
* cp/decl2.c (grok_array_decl): Add new operands for ARRAY_REF.
(build_anon_union_vars): Add new operand for COMPONENT_REF.
* cp/init.c (buld_new): Add new operand for ARRAY_REF.
* cp/method.c (do_build_copy_constructor): New op for COMPONENT_REF.
(do_build_assign_ref): Likewise.
* cp/parser.c (cp_parser_direct_new_declarator): Add new operands
for ARRAY_REF.
(cp_parser_direct_declarator): Likewise.
* cp/pt.c (tsubst): Likewise.
(tsubst_copy, tsubst_copy_and_build): Likewise; also add new operand
for COMPONENT_REF.
* cp/semantics.c (finish_non_static_data_member): Add new operand
for COMPONENT_REF.
* cp/typeck.c (build_class_member_access_expr): Likewise.
(build_class_member_access_expr, finish_class_member_access_expr):
Likewise.
(build_ptrmemfunc_access_expr): Likewise.
(build_array_ref): Add new operands for ARRAY_REF.
* cp/typeck2.c (split_nonconstant_init_1): Likewise; COMPONENT_REF too.
* cp/tree.c (count_trees_r, no_linkage_helper): Don't walk in types.
* fortran/f95-lang.c (LANG_HOOKS_GIMPLE_BEFORE_INLINING): Deleted.
* fortran/trans-array.c (gfc_conv_descriptor_data): Add operand
for COMPONENT_REF.
(gfc_conv_descriptor_offset, gfc_conv_descriptor_dtype): Likewise.
(gfc_conv_descriptor_dimension, gfc_conv_descriptor_stride): Likewise.
(gfc_conv_descriptor_lbound, gfc_conv_descriptor_ubound): Likewise.
* fortran/trans-common.c (create_common): Likewise.
* fortran/trans-expr.c (gfc_conv_component_ref): Likewise.
* fortran/trans-io.c (set_parameter_value): Likewise.
(set_parameter_ref, set_string, set_flag, io_result): Likewise.
(transfer_expr): Likewise.
* fortran/trans-decl.c (gfc_trans_auto_character_variable):
Set up to get DECL_SIZE and DECL_SIZE_UNIT gimplified.
(gfc_simplify_function): New function.
(gfc_generate_function-code): Properly handle nested functions.
* fortran/trans.c (gfc_build_array_ref): Add two new operands
for ARRAY_REF.
* java/class.c (build_class_ref): Add new operand for COMPONENT_REF.
(build_static_field_ref): Likewise and add new operands for ARRAY_REF.
* java/constants.c (build_ref_from_constant_pool): Likewise.
* java/expr.c (build_java_array_length_access): Likewise.
(build_get_class, build_field_ref, build_known_method_ref): Likewise.
(invoke_build_dtable, build_invokevirtual): Likewise.
(build_invokeinterface, java_expand_expr): Likewise.
(emit_init_test_initialization): Likewise.
* java/java-gimplify.c (java_gimplify_new_array_init): Likewise.
* java/parse.y (make_qualifed_name, build_array_ref): Likewise.
* objc/ojbc-act.c (generate_static_references): Add additional
operands to ARRAY_REF.
(generate_strings, build_method_prototype_list_template): Likewise.
(generate_protocol_list): Likewise.
From-SVN: r83474
* fold-const.c (operand_equal_p): Pass flags in recursive calls for
binary and relational operations. Add support for TRUTH_ANDIF_EXPR,
TRUTH_ORIF_EXPR, TRUTH_AND_EXPR, TRUTH_OR_EXPR and TRUTH_XOR_EXPR.
* tree.c (commutative_tree_code): Also list UNORDERED_EXPR,
ORDERED_EXPR, UNEQ_EXPR, LTGT_EXPR, TRUTH_AND_EXPR, TRUTH_OR_EXPR
and TRUTH_XOR_EXPR.
From-SVN: r83433
2004-06-21 Paolo Bonzini <bonzini@gnu.org>
* fold-const.c (fold_cond_expr_with_comparison):
New function, extracted from fold.
(fold): Extract code to fold A op B ? A : C, use
it to fold A op B ? C : A. Really optimize
A & N ? N : 0 where N is a power of two. Avoid
relying on canonicalization and recursion for
foldings of COND_EXPR to happen.
From-SVN: r83428
* fold-const.c (fold) <UNORDERED_EXPR, ORDERED_EXPR, UNLT_EXPR,
UNLE_EXPR, UNGT_EXPR, UNGE_EXPR, UNEQ_EXPR, LTGT_EXPR>: Add
constant folding for unordered comparison tree nodes. If both
operands are real constants, call fold_relational_const. If either
operand is a NaN, evaluate the other for side-effects and return a
constant. Optimize (double)float1 CMP (double)float2 into the
equivalent float1 CMP float2.
(nondestructive_fold_binary_to_constant) <UNORDERED_EXPR,
ORDERED_EXPR, UNLT_EXPR, UNLE_EXPR, UNGT_EXPR, UNGE_EXPR, UNEQ_EXPR,
LTGT_EXPR>: Call fold_relational_const for constant operands.
(fold_relational_const): Add support for unordered comparison tree
nodes. Don't constant fold "ordered" floating point comparisons
against NaN if when flag_trapping_math is set.
* gcc.dg/unordered-2.c: New test case.
* gcc.dg/unordered-3.c: New test case.
From-SVN: r83379
* fold-const.c (build_range_check): If !in_p and recursive call
fails, exit immediately. If high - low overflows and etype is
a signed type, retry with unsigned etype.
(merge_ranges): If !in0_p and !in1_p, handle even range2 adjacent
to range1 at TYPE_MAX_VALUE and TYPE_MIN_VALUE.
* gcc.dg/range-test-1.c: New test.
* g++.dg/opt/range-test-1.C: New test.
From-SVN: r83369
* fold-const.c (swap_tree_comparison): No longer static.
(tree_swap_operands_p): Similarly. Return true if both operands
are SSA_NAMEs and the first operand has a higher version number than
the second operand.
* tree.h (swap_tree_comparison): Prototype.
(tree_swap_operands_p): Prototype.
* tree-ssa-operands.c (get_expr_operands): For commutative
operators and relational comparisons, canonicalize the
order of the operands.
* gcc.dg/tree-ssa/20040615-1.c: New test.
* gcc.dg/tree-ssa/20030824-1.c: Update expected output to
be less sensitive to operand ordering.
* gcc.dg/tree-ssa/20030824-2.c: Likewise.
From-SVN: r83224
2004-06-13 Andrew Pinski <pinskia@physics.uc.edu>
* fold-const.c (fold_checksum_tree <case 't'>): Only
look at TREE_VALUES if the EXPR is an ENUMERAL_TYPE.
Only look at TYPE_MIN_VALUE and TYPE_MAX_VALUE if
EXPR is an INTEGERAL_TYPE or a scalar float type.
From-SVN: r83068
* fold-const.c (omit_two_operands): New function.
* tree.h (omit_two_operands): Prototype here.
* builtins.c (fold_builtin_unordered_cmp): New function to lower
C99 unordered comparison builtins to the appropriate tree nodes.
(fold_builtin_1): Use fold_builtin_unordered_cmp to lower
BUILT_IN_ISGREATER, BUILT_IN_ISGREATEREQUAL, BUILT_IN_ISLESS,
BUILT_IN_ISLESSEQUAL and BUILT_IN_ISLESSGREATER. Manually lower
BUILT_IN_ISUNORDERED comparisons to an UNORDERED_EXPR tree node.
(simplify_builtin_memcmp, simplify_builtin_strncmp,
simplify_builtin_strncat, simplify_builtin_strspn): Use the new
omit_two_operands function to build the required COMPOUND_EXPRs.
From-SVN: r83040
* fold-const.c (fold_inf_compare): Avoid creating non-gimple
code when we are in gimple form.
* gcc.c-torture/compile/20040610-1.c: New test.
From-SVN: r82958
* fold-const.c (fold_abs_const): Make extern.
* tree.h (fold_abs_const): Prototype here.
* builtins.c (fold_builtin_fabs): New function to transform
fabs, fabsf and fabsl builtins into ABS_EXPR tree nodes.
(fold_builtin_abs): New function to transform abs, labs, llabs
and imaxabs builtins into ABS_EXPR tree nodes.
(expand_builtin): Fall back to a function call for abs, labs,
llabs and imaxabs builtins that survive constant folding.
(fold_builtin_1): Call fold_builtin_fabs for FABS, FABSF and
FABSL, and fold_builtin_abs for ABS, LABS, LLABS and IMAXABS.
From-SVN: r82916
* target.h: Add gimplify_va_arg_expr hook.
* target-def.h: Add TARGET_GIMPLIFY_VA_ARG_EXPR.
* fold-const.c (build_fold_addr_expr)
(build_fold_addr_expr_with_type): Move from gimplify.c.
* tree.h: Declare them.
* gimplify.c (gimplify_and_add): New fn.
(build_addr_expr, build_addr_expr_with_type): Move to fold-const.c.
(gimplify_array_ref_to_plus, gimplify_modify_expr)
(gimplify_expr): Use build_fold_*.
(copy_if_shared_r): Only mark VA_ARG_EXPR volatile if we
don't know how to gimplify it.
* builtins.c (std_gimplify_va_arg_expr): New fn.
(dummy_object): New static fn.
(gimplify_va_arg_expr): New fn.
(stabilize_va_list): Use build_fold_*.
* tree-gimple.h: Declare new fns.
* config/i386/i386.c (TARGET_GIMPLIFY_VA_ARG_EXPR): Define.
(ix86_gimplify_va_arg): New fn.
* config/i386/ia64.c (TARGET_GIMPLIFY_VA_ARG_EXPR): Define.
(ia64_gimplify_va_arg): New fn.
* config/i386/rs6000.c (rs6000_gimplify_va_arg): New fn.
(TARGET_GIMPLIFY_VA_ARG_EXPR): Define.
* config/i386/sparc.c (sparc_gimplify_va_arg): New fn.
* alias.c (get_varargs_alias_set): Just return 0 for now.
* c-objc-common.c (c_tree_printer): Improve handling of %T.
From-SVN: r82838
2004-06-08 Andrew Pinski <pinskia@physics.uc.edu>
* fold-const.c (fold_convert): Treat OFFSET_TYPE like
POINTER_TYPE and INTEGER_TYPE.
From-SVN: r82792
* real.c (real_copysign): New function to implement libm's copysign.
* real.h (real_copysign): Prototype here.
* fold-const.c (tree_expr_nonnegative_p): The result of sqrt, sqrtf
and sqrtl can be negative, as sqrt(-0.0) = -0.0. Correct whitespace.
* builtins.c (fold_builtin_isascii, fold_builtin_toascii,
fold_builtin_isdigit): Add function prototypes.
(fold_builtin_copysign): New function to fold copysign, copysignf
and copysignl. Optimize copysign(x,x) as x. Evaluate copysign of
constant arguments at compile-time using real_copysign. Fold
copysign(X,Y) as fabs(X) if Y is always non-negative.
(fold_builtin_1): Correct minor whitespace/style issues. Call
fold_builtin_copysign for BUILT_IN_COPYSIGN{,F,L}.
* gcc.dg/builtins-41.c: New test case.
* gcc.dg/builtins-42.c: New test case.
From-SVN: r82721
PR middle-end/15069
* fold-const.c (fold_single_bit_test): Only perform "(X & C) != 0"
into "X < 0" (where C is the signbit) if X's type is a full mode.
* g++.dg/opt/fold3.C: New test case.
From-SVN: r82490
* fold-const.c (combine_comparisons, optimize_bit_field_compare,
range_binop, fold_truthop, fold_binary_op_with_conditional_arg,
fold_mathfn_compare, fold_inf_compare, fold,
fold_relational_hi_lo, nondestructive_fold_binary_to_constant):
Use constant_boolean_node where appropriate. Don't bother using
fold_convert on the second argument to omit_one_operand.
From-SVN: r82468
gcc/ChangeLog:
2004-05-27 Paolo Bonzini <bonzini@gnu.org>
Roger Sayle <roger@eyesopen.com>
PR rtl-optimization/15649
Add LTGT_EXPR and improve pretty-printing of unordered
comparisons.
* c-common.c (c_common_truthvalue_conversion):
Handle LTGT_EXPR.
* c-typeck.c (build_binary_op): Likewise.
* dojump.c (do_jump): Likewise.
* expr.c (expand_expr_real_1, do_store_flag): Likewise.
* predict.c (tree_predict_by_opcode): Likewise.
* real.c (real_compare): Likewise.
* tree-cfg.c (verify_expr): Likewise.
* tree-inline.c (estimate_num_insns_1): Likewise.
* tree-pretty-print.c (dump_generic_node): Likewise.
Handle ORDERED_EXPR, UNORDERED_EXPR.
(op_symbol): Print unordered comparisons differently
than ordered ones.
* tree.def (LTGT_EXPR): New '<' tree code.
* doc/c-tree.texi (Expressions): Document floating-point
comparison nodes.
Fold comparisons between floating point values.
* fold-const.c (enum comparison_code): New, from
#define'd constants. Define compcodes for unordered
comparisons and for invalid transformations.
(invert_tree_comparison): Add "honor_nans" parameter.
(fold_truthop): Revamp to work on floating-point types too.
(comparison_to_compcode): Support unordered comparisons.
Use new enum comparison_code.
(compcode_to_comparison): Likewise.
(combine_compcodes): New function.
(invert_truthvalue): Let invert_tree_comparison decide
whether it is valid to fold the comparison. Fold ORDERED
and UNORDERED even if flag_unsafe_math_optimizations is off,
and the remaining even if flag_unsafe_math_optimizations
is off but we are under -fno-trapping-math.
(fold_relational_const): Integer modes do not honor NaNs.
gcc/testsuite/ChangeLog:
2004-05-27 Paolo Bonzini <bonzini@gnu.org>
* gcc.c-torture/compare-fp-1.c, gcc.c-torture/compare-fp-2.c,
gcc.c-torture/compare-fp-3.c, gcc.c-torture/compare-fp-4.c,
gcc.c-torture/compare-fp-3.x, gcc.c-torture/compare-fp-4.x,
gcc.c-torture/pr15649-1.c: New.
Co-Authored-By: Roger Sayle <roger@eyesopen.com>
From-SVN: r82365
PR middle-end/3074
* fold-const.c (strip_compound_expr): Delete function.
(count_cond): Delete function.
(fold_binary_op_with_conditional_arg): Only perform transformations
"a + (b?c:d) -> b ? a+c : a+d" and "(b?c:d) + a -> b ? c+a : d+a"
when a is constant. This greatly simplifies this routine.
* tree.c (saved_expr_p): Delete function.
* tree.h (saved_expr_p): Delete function prototype.
* gcc.dg/pr3074-1.c: New test case.
* gcc.dg/sequence-pt-1.c: Remove an XFAIL.
From-SVN: r82071
* fold-const.c (fold_div_compare): New function to optimize X/C1 op C2
where op is a comparison operator and C1 and C2 are integer constants
into a range check.
(fold): Call fold_div_compare.
* gcc.c-torture/execute/divcmp-1.c: New test case.
* gcc.c-torture/execute/divcmp-2.c: New test case.
* gcc.c-torture/execute/divcmp-3.c: New test case.
From-SVN: r81645
* fold-const.c (fold): Prefer fold_convert (negate_expr (...)) to
fold (build1 (NEGATE_EXPR, ...)). Optimize X / -1 as -X and
X % -1 as 0.
From-SVN: r81177
* fold-const.c (fold_binary_op_with_conditional_arg): Tweak
calling convention to allow a NULL_TREE to be returned. Factor
sanity checks from callers, return NULL_TREE when appropriate.
(fold): Handle COMPOUND_EXPR operands of binary expressions
before COND_EXPR operands. Use reorder_operands_p(a,b) to check
whether a op (b,c) can be rewritten as (b, a op c). Simplify
calls to fold_binary_op_with_conditional_arg.
From-SVN: r80609
* expr.c (get_inner_reference): Use DECL_UNSIGNED, not TREE_UNSIGNED.
* stor-layout.c (layout_decl): Likewise.
* tree.c (get_narrower): Likewise and also use BIT_FIELD_REF_UNSIGNED.
* fold-const.c (make_bit_field_ref): Use BIT_FIELD_REF_UNSIGNED.
* print-tree.c (print_node): Handle various used of unsigned_flag.
* tree.def (BIT_FIELD_REF): Update comment.
* tree.h (TREE_UNSIGNED): Deleted.
(DECL_UNSIGNED, BIT_FIELD_REF_UNSIGNED): New macros.
* cp/class.c (VTT_TOP_LEVEL_P): Use unsigned_flag directly.
* f/com.c (ffe_truthvalue_conversion, case COMPONENT_REF):
Use DECL_UNSIGNED and integer_onep.
From-SVN: r80293
* fold-const.c (tree_expr_nonnegative_p): Handle BIT_XOR_EXPR like
BIT_IOR_EXPR; A^B is nonnegative when A and B are nonnegative.
* gcc.dg/compare8.c: Add an additional test for XOR.
From-SVN: r79979
* fold-const.c (tree_expr_nonnegative_p): A&B is nonnegative when
A is nonnegative or B is nonnegative. Similarly A|B is nonnegative
when both A and B are nonnegative.
(tree_expr_nonzero_p): A|B is nonzero when A is nonzero or B is
nonzero.
From-SVN: r79870
2004-03-23 Kazu Hirata <kazu@cs.umass.edu>
PR optimization/14669
* fold-const.c (fold): Only unwiden integer comparisons for equality
or inequality operators, or when the signedness is the same.
* g++.dg/opt/fold2.C: New test case.
From-SVN: r79859
* fold-const.c (fold): Remove variable "invert".
Move the handling of relational expressions that can be folded
to a constant ...
(fold_relational_const): ... here.
(tree_expr_nonzero_p): New.
From-SVN: r79745
* fold-const.c (negate_expr_p) <RSHIFT_EXPR>: We can optimize
-((int)X>>C) where C is an integer constant one bit less than the
size of X into (unsigned)X>>C. Similarly for unsigned->signed.
(negate_expr) <RSHIFT_EXPR>: Implement the above transformations.
* simplify-rtx.c (simplify_unary_operation): Also implement the
above transformations at the RTL level.
* gcc.c-torture/execute/20040311-1.c: New test case.
From-SVN: r79334
* fold-const.c (fold): An equality comparison of a non-weak object
against zero has a known result. Similarly an equality comparison
of the address of two non-weak, unaliased symbols has a known result.
* ggc-page.c (struct page_entry): New field PREV.
(ggc_alloc): Update PREV field appropriately.
(sweep_pages): Likewise.
(ggc_free): Likewise. Use PREV field rather than loop to
improve ggc_free performance.
cp/
* init.c (build_vec_delete_1): Convert 2nd argument to NE_EXPR to
the proper type.
From-SVN: r78713
* fold-const.c (fold): Revert 2004-02-25 change. Use the original
operands to build a tree with swapped operands.
* expr.c (expand_expr_real) <MAX_EXPR>: Consistently use the
'unsignedp' predicate to specify the signedness.
Co-Authored-By: Roger Sayle <roger@eyesopen.com>
From-SVN: r78558
PR c/14092
* gcc.dg/pr14092-1.c: New testcase.
* fold-const.c (fold) <NEGATE_EXPR>: Convert result of
negate_expr back to the original type.
From-SVN: r77602
* fold-const.c (fold) <NOP_EXPR>: Use the original type conversion
tree code rather than call fold_convert, which doesn't specify a
default floating point to integer conversion.
* gcc.c-torture/compile/20040209-1.c: New test case.
From-SVN: r77535
PR middle-end/13696
* fold-const.c (fold_convert): New function to provide type
conversion to the middle-end without using convert.
(negate_expr, associate_trees, size_diffop, omit_one_operand,
operand_equal_for_comparison_p, pedantic_omit_one_operand,
invert_truthvalue, optimize_bit_field_compare, range_binop,
decode_field_reference, make_range, build_range_check, unextend,
fold_truthop, extract_muldiv_1, fold_mathfn_compare,
fold_binary_op_with_conditional_arg, fold_inf_compare,
fold_single_bit_test, fold, multiple_of_p): Replace all calls to
convert with calls to fold_convert.
* g++.dg/opt/fold1.C: New test case.
From-SVN: r77464
* fold-const.c (negate_expr_p, negate_expr): Optimize -(A+B) into
either (-A)-B or (-B)-A, if A or B is easily negated respectively.
(fold) <MINUS_EXPR>: Optimize (A*C) - (B*C) -> (A-B)*C for both
integer types and floating point with unsafe_math_optimizations.
Add similar optimization for (A*C1) - (A*C2) -> A*(C1-C2).
Optimize A - B as A + (-B), if B is easily negated.
From-SVN: r77453
* tree.c (commutative_tree_code, associative_tree_code): New
functions.
(iterative_hash_expr): Use commutative_tree_code.
* tree.h (commutative_tree_code, associative_tree_code): Declare.
* fold-const.c (operand_equal_p): Use commutative_tree_code
rather than inlining the commutativity check.
(fold): Likewise.
Co-Authored-By: Roger Sayle <roger@eyesopen.com>
From-SVN: r77152
* fold-const.c (fold_binary_op_with_conditional_arg): Only
build a COMPOUND_EXPR if 'arg' is really a SAVE_EXPR.
Co-Authored-By: Olivier Hainque <hainque@act-europe.fr>
From-SVN: r76434
* fold-const.c (fold_convert): Rename to fold_convert_const.
(fold_convert_const): Change arguments to take a tree_code,
a type and the operand/expression to be converted. Return
NULL_TREE if no simplification is possible. Add support for
FIX_CEIL_EXPR and FIX_FLOOR_EXPR in addition to FIX_TRUNC_EXPR.
(fold): Handle FIX_CEIL_EXPR and FIX_FLOOR_EXPR.
Adjust call to fold_convert to match new fold_convert_const.
Avoid modifying the tree passed to fold in-place.
From-SVN: r76237
* fold-const.c (fold) <COND_EXPR>: Don't require strict type
equality, instead just prevent replacing a COND_EXPR of non-void
type by one of its operands of void type.
From-SVN: r75239
PR fortran/12632
* fold-const.c (fold) <COND_EXPR>: Don't fold a constant condition,
if the type of the selected branch doesn't match its' parent.
* com.c (ffecom_subscript_check_): Take as an extra argument the
(possibly NULL) decl of the array. Don't create unnecessary tree
nodes if the array index is known to be safe at compile-time.
If the array index is unsafe, force the array decl into memory to
avoid RTL expansion problems.
(ffecom_array_ref_): Update calls to ffecom_subscript_check_.
(ffecom_char_args_x_): Likewise.
* g77.dg/12632.f: New test case.
From-SVN: r75203
* fold-const.c (negate_mathfn_p): New function to determine whether
a built-in mathematical function is sign preserving, f(-x) == -f(x).
Add support for BUILT_IN_ASIN, BUILT_IN_ASINF and BUILT_IN_ASINL.
(tree_swap_operands_p): Change API to take an additional argument
indicating that the swapped operands evaluate in reverse order.
Canonicalize VAR_DECLs and PARM_DECLs last if we can, i.e. neither
operand side-effects or we don't care about flag_evaluation_order.
(reorder_operands_p): New function to check whether its safe to
evaluate the given operands in reverse order.
(negate_expr_p): We can always negate integer constants unless
we honor -ftrapv and the signed type would overflow. Only allow
-(A-B) into B-A if reorder_operands_p says that its OK. Allow
negation of COMPLEX_CST if both real and imaginary parts can be
negated. Allow negation through floating point extensions and
sign-preserving built-in functions.
(negate_expr): Move the code to negate integers from "fold" to
here. Always negate integer constants unless we honor -ftrapv
and the signed type would overflow. Always negate real constants
unless we honor -ftrapping-math. Only convert -(A-B) into B-A
if allowed by reorder_operands_p. Add support for COMPLEX_CST.
Optimize negation through floating point extensions and
sign-preserving built-in functions (as defined by negate_mathfn_p).
(fold): Adjust calls to tree_swap_operands_p.
(fold <NEGATE_EXPR>): Move the remaining negation optimizations
to negate_expr_p/negate_expr.
(fold <MINUS_EXPR>): Use reorder_operands_p to check whether we're
allowed to convert (-A) - B into (-B) - A.
From-SVN: r72381
PR optimization/9325, PR java/6391
* fold-const.c (fold_convert): For floating point to integer
conversions, return the maximum/minimum representable integer
value if the real constant overflows the destination type.
* tree.c (real_value_from_int_cst): Allow the type to be NULL,
meaning don't truncate the result to a floating point mode.
Simplify the logic by calling real_from_integer directly.
* simplify-rtx.c (simplify_unary_operation): Implement the
same semantics for folding floating point to integer conversions
in RTL.
* gcc.c-torture/execute/20031003-1.c: New test case.
From-SVN: r72079
* fold-const.c (tree_swap_operands_p): New function to determine
the prefered ordering of operands.
(fold): Numerous clean-ups. Use tree_swap_operands_p when swapping
operands to commutative, comparison or ternary operators. Replace
uses of TREE_SET_CODE with recursive call to fold. Remove duplicate
transformation of A ? B : C into !A ? C : B.
From-SVN: r71438
* builtins.c (real_dconstp, fold_builtin_logarithm,
fold_builtin_exponent): New, split out from fold_builtin. Also
generalize to add log2, log10, exp2 and exp10/pow10 equivalents.
* emit-rtl.c (dconst3, dconst10, dconstthird): New.
(init_emit_once): Initialize new dconsts, use ARRAY_SIZE in lieu
of hardcoded array size.
* fold-const.c (fold): Add cases for exp2, exp10 and pow10.
(tree_expr_nonnegative_p): Likewise.
* real.h (dconst3, dconst10, dconstthird): New.
testsuite:
* gcc.dg/torture/builtin-explog-1.c: New testcase.
From-SVN: r71252
PR middle-end/12002
* tree.h (SCALAR_FLOAT_TYPE_P, COMPLEX_FLOAT_TYPE_P): New macros.
(FLOAT_TYPE_P): Define in terms of these two new macros.
* fold-const.c (fold <PLUS_EXPR>): Don't convert x+x into x*2.0
for complex floating point types.
* g77.f-torture/compile/12002.f: New test case.
From-SVN: r70821
* fold-const.c (fold <MULT_EXPR>): Optimize (C1/X)*C2 into
(C1*C2)/X when unsafe math optimizations are allowed.
(fold <RDIV_EXPR>): Optimize C1/(X*C2) into (C1/C2)/X with unsafe
math optimizations. Minor code clean-ups. Recursively call
fold when constructing sub-expressions.
* gcc.dg/20030826-1.c: New test case.
From-SVN: r70807
PR middle-end/11984
* fold-const.c (fold <PLUS_EXPR>): Check for integer constant
operands before calling tree_int_cst_lt when performing associative
transformations.
* gcc.dg/20030820-1.c: New test case.
From-SVN: r70618
* fold-const.c (negate_expr_p): MULT_EXPRs and RDIV_EXPRs are easy
to negate if either operand is easy to negate, if we don't care
about sign-dependent rounding.
(negate_expr): Make the logic to negate a REAL_CST explicit.
Attempt to negate a MULT_EXPR or RDIV_EXPR by negating an operand
that's easy to negate, if we don't honor sign-dependent rounding.
(fold <MULT_EXPR>): Optimize -A * B as A * -B if B is easy to
negate, and the symmetric A * -B as -A * B if A is easy to negate.
(fold <RDIV_EXPR>): Likewise, optimize -A/B and C/-D as A/-B and
-C/D if B and C are cheap to negate. Add an explicit rule to
optimize X/-1.0 as -X when we don't care about signaling NaNs.
From-SVN: r70455
PR middle-end/11771
* fold-const.c (negate_expr_p <MINUS_EXPR>): Change to match the
logic in negate_expr, i.e. we don't invert (A-B) for floating
point types unless flag_unsafe_math_optimizations.
* gcc.c-torture/compile/20030804-1.c: New test case.
From-SVN: r70159
* fold-const.c (fold <PLUS_EXPR>): Transform x+x into x*2.0.
Optimize x*c+x and x+x*c into x*(c+1) and x*c1+x*c2 into x*(c1+c2)
for floating point expressions with -ffast-math.
(fold <MULT_EXPR>): Don't transform x*2.0 into x+x.
* expmed.c (expand_mult): Wrap long line. Expand x*2.0 as x+x.
* gcc.dg/20030804-1.c: New test case.
From-SVN: r70158
* fold-const.c (fold <MULT_EXPR>): Optimize both x*pow(x,c) and
pow(x,c)*x as pow(x,c+1) for constant values c. Optimize x*x
as pow(x,2.0) when the latter will be expanded back into x*x.
(fold <RDIV_EXPR>): Optimize pow(x,c)/x as pow(x,c-1).
* builtins.c (expand_builtin_pow): Ignore flag_errno_math as
pow can never set errno when used with an integer exponent.
Always use expand_powi when exponent is -1, 0, 1 or 2.
(fold_builtin): Don't rewrite pow(x,2.0) as x*x nor pow(x,-2.0)
as 1.0/(x*x). This avoids unbounded recursion as we now prefer
the pow forms of these expressions.
* gcc.dg/builtins-27.c: New test case.
From-SVN: r70030
* fold-const.c (const_binop): Avoid performing the FP operation at
compile-time, if either operand is NaN and we honor signaling NaNs,
or if we're dividing by zero and either flag_trapping_math is set
or the desired mode doesn't support infinities.
(fold_initializer): New function to fold an expression ignoring any
potential run-time exceptions or traps.
* tree.h (fold_initializer): Prototype here.
* c-typeck.c (build_binary_op): Move to the end of the file so
that intializer_stack is in scope. If constructing an initializer,
i.e. when initializer_stack is not NULL, use fold_initializer to
fold expressions.
* simplify-rtx.c (simplify_binary_operation): Likewise, avoid
performing FP operations at compile-time, if they would raise an
exception at run-time.
From-SVN: r69533
PR c/11449
* fold-const.c (sign_bit_p): Return EXP if VAL is the sign bit
of HOST_WIDE_INT.
(fold_single_bit_test): If sign_bit_p() fails, assume that the
bit being tested is not a sign bit.
PR c/11449
* gcc.c-torture/compile/20030707-1.c: New.
* config/h8300/h8300.md (a peephole2): New.
From-SVN: r69184
* expr.c (do_store_flag): Pass in the correct result type
when calling fold_single_bit_test.
* fold-const.c (fold_single_bit_test): Use result_type for the
result when folding a sign bit test.
From-SVN: r68940
* real.c (real_trunc, real_floor, real_ceil): New functions
to implement trunc, floor and ceil respectively.
* real.h (real_trunc, real_floor, real_ceil): Prototype here.
* builtins.c (integer_valued_real_p): New function to test if
a floating point expression has an integer valued result.
(fold_trunc_transparent_mathfn): Optimize foo(foo(x)) as
foo(x) where foo is an integer rounding function. Similarly,
optimize foo(bar(x)) as bar(x), and foo((double)(int)x) as
(double)(int)x when both foo and bar are integer rounding
functions and we don't need to honor errno.
(fold_builtin_trunc, fold_builtin_floor, fold_builtin_ceil):
New functions to fold trunc, floor and ceil.
(fold_builtin): Use fold_builtin_trunc to fold BUILT_IN_TRUNC*,
fold_builtin_floor to fold BUILT_IN_FLOOR* and fold_builtin_ceil
to fold BUILT_IN_CEIL*.
* fold-const.c (tree_expr_nonnegative_p): Handle FLOAT_EXPR and
the remaining integer rounding functions.
* gcc.dg/builtins-25.c: New testcase.
* gcc.dg/builtins-26.c: New testcase.
From-SVN: r68903
PR target/10700
* fold-const.c (extract_muldiv_1): There's nothing that can be done
if the expression is a SAVE_EXPR.
* gcc.c-torture/compile/20030703-1.c: New test case.
From-SVN: r68877
* expr.c (do_store_flag): Remove special case folding for
single bit tests. Instead call back into the commonized folder
routine.
* fold-const.c (fold_single_bit_test): New function, mostly
extracted from do_store_flag, with an additional case extracted
from fold.
(fold): Call fold_single_bit_test appropriately.
* tree.h (fold_single_bit_test): Prototype.
From-SVN: r68867
PR optimization/11210
* expr.c (handled_component_p) [NOP_EXPR]: Add ??? note
about the behaviour with regard to bitfields.
* fold-const (decode_field_reference): Record outermost type in
case the expression is a NOP. Strip all NOPs. Set the signedness
to that of the outermost type (if any) when the bitsize is equal
to the size of the type.
From-SVN: r68823
* fold-const.c (operand_equal_p): Consider two calls to "const"
functions with identical non-volatile arguments to be equal.
Consider the FUNCTION_DECL for the "__builtin_foo" form of a
built-in function to be equal to the "foo" form.
Co-Authored-By: Jeff Law <law@redhat.com>
From-SVN: r68011
PR middle-end/168
* fold-const.c (tree_expr_nonnegative_p): Handle addition
and multiplication of zero extensions, floating point division,
and integer<->fp, fp<->fp and zero extension conversions.
The built-in ceil and floor functions preserve signedness.
* gcc.dg/20030612-1.c: New test case.
From-SVN: r67850
* fold-const.c (tree_expr_nonnegative_p): Add support for
floating point constants, addition and multiplication.
* gcc.dg/builtins-21.c: New test case.
From-SVN: r67828
* builtins.c (fold_builtin): Optimize cos(-x) as cos(x).
* fold-const.c (fold <NEGATE_EXPR>): Convert -f(x) into f(-x)
when x is easily negated and f is sin, tan or atan.
(fold <MULT_EXPR>): Optimize tan(x)*cos(x) and cos(x)*tan(x) as
sin(x) with flag_unsafe_math_optimizations.
(fold <RDIV_EXPR>): With flag_unsafe_math_optimizations fold
sin(x)/cos(x) as tan(x) and cos(x)/sin(x) as 1.0/tan(x).
* gcc.dg/builtins-20.c: New test case.
From-SVN: r67701
* fold-const.c (fold <EQ_EXPR>): Don't fold x == x only if x
is a floating point type *and* we currently honor NaNs.
(fold <NE_EXPR>): Likewise.
From-SVN: r67700
* fold-const.c (fold <ABS_EXPR>): Re-fold the result of folding
fabs(-x) into fabs(x). Use tree_expr_nonnegative_p to determine
when the ABS_EXPR (fabs or abs) is not required.
(tree_expr_nonnegative_p): Move the logic that sqrt and exp are
always nonnegative from fold to here. Additionally, cabs and fabs
are always non-negative, and pow and atan are non-negative if
their first argument is non-negative.
* builtins.c (fold_builtin_cabs): New function to fold cabs{,f,l}.
Evaluate cabs of a constant at compile-time. Convert cabs of a
non-complex argument into fabs. Convert cabs(z) into
sqrt(z.r*z.r + z.i*z.i) at the tree-level with -ffast-math or
-funsafe-math-optimizations or -ffast-math.
(fold_builtin): Convert BUILT_IN_FABS{,F,L} into an ABS_EXPR.
Fold BUILT_IN_CABS{,F,L} using fold_builtin_cabs.
* gcc.dg/builtins-2.c: Add some more tests.
* gcc.dg/builtins-18.c: New test case.
* gcc.dg/builtins-19.c: New test case.
From-SVN: r67541
* flags.h (flag_wrapv): New flag controlling overflow semantics.
* toplev.c (flag_wrapv): Declare the variable with default false.
(lang_independent_options): New option "-fwrapv" to set the above.
* fold-const.c (extract_muldiv_1): Disable optimization of (2*x)/2
as x, when signed arithmetic overflow wraps around.
(fold): Optimize "-A - B" as "-B - A" if overflow wraps around.
* loop.c (basic_induction_var): Ignore BIVs that rely on undefined
overflow when flag_wrapv is true.
* java/lang.c (java_init_options): Prescribe wrap-around two's
complement arithmetic overflow by setting flag_wrapv.
* doc/invoke.texi: Document new -fwrapv command line option.
* doc/c-tree.texi: Mention that the overflow semantics of
NEGATE_EXPR, PLUS_EXPR, MINUS_EXPR and MULT_EXPR is dependent
upon both flag_wrapv and flag_trapv.
* gcc.dg/fwrapv-1.c: New test case.
* gcc.dg/fwrapv-2.c: New test case.
* libjava.lang/Overflow.java: New test.
* libjava.lang/Overflow.out: New file.
From-SVN: r67270
* real.c (real_maxval): New function to return the largest finite
value representable in a given mode (i.e. FLT_MAX and DBL_MAX).
* real.h (real_maxval): Prototype here.
* fold-const.c (fold_inf_compare): Transform comparisons against
+-Infinity into comparisons against DBL_MAX (or equivalent).
* gcc.c-torture/execute/ieee/inf-2.c: New test case.
From-SVN: r67112
* fold-const.c (fold, case PLUS_EXPR and case MULT_EXPR): Restore
a number of conversions required for type consistency and previously
stripped off by STRIP_NOPS.
From-SVN: r65913
* tree.c (skip_simple_arithmetics_at, saved_expr_p): New functions.
(save_expr): Replace loop by call to skip_simple_arithmetics_at.
* tree.h: Add prototypes for the two new functions.
* fold-const.c (fold_binary_op_with_conditional_arg): Replace test
updates introduced in the previous revision by call to saved_expr_p.
* stor-layout.c (put_pending_size): Use skip_simple_arithmetics_at.
From-SVN: r65702
* fold-const.c (fold): Transform (c1 - x) cmp c2, where cmp is a
comparison operation and c1/c2 are floating point constants into
x swap(cmp) (c1 - c2).
* gcc.dg/20030414-2.c: New test case.
From-SVN: r65584
* fold-const.c (fold): Use "fold" following build in more places.
Optimize sqrt(x)*sqrt(x) as x, pow(x,y)*pow(z,y) as pow(x*z,y),
pow(x,y)*pow(x,z) as pow(x,y+z) and x/pow(y,z) as x*pow(y,-z).
* gcc.dg/builtins-11.c: New test case.
From-SVN: r65387
* fold-const.c (fold_inf_compare): New function to simplify FP
comparisons against +Infinity or -Infinity.
(fold): Optimize floating point comparisons against Infs and NaNs.
* gcc.c-torture/execute/ieee/fp-cmp-6.c: New test case.
* gcc.c-torture/execute/ieee/fp-cmp-7.c: New test case.
From-SVN: r64945
PR c/8224
* fold-const.c (extract_muldiv_1): Don't pass through type conversions
when signedness changes for division or modulus.
PR c/8224
* gcc.dg/20030323-1.c: New test.
From-SVN: r64760
* fold-const.c (fold_mathfn_compare): New function to simplify
comparisons against built-in math functions. Fold comparisons
of sqrt against constants.
(fold): Call fold_mathfn_compare when appropriate.
* gcc.dg/builtins-6.c: New test case.
From-SVN: r64619
* fold-const.c (omit_one_operand): No longer static.
* tree.h (omit_one_operand): Prototype here.
(div_and_round_double): Keep fold-const.c prototypes together.
* builtins.c (builtin_mathfn_code): Handle binary built-in
funtions, such as "pow" and "atan2".
(fold_builtin): Optimize both pow(x,0.0) and pow(1.0,y) to 1.0.
Simplify optimizations using "type" the builtin's return type.
* gcc.dg/builtins-5.c: New test case.
From-SVN: r63273
* fold-const.c (negate_expr_p): New function to determine whether
an expression can be negated cheaply.
(fold) [MINUS_EXPR]: Use it to determine whether to transform
-A - B into -B - A for floating point types.
From-SVN: r63059
* fold-const.c (extract_muldiv_1): Rename from extract_muldiv;
rearrange mult arguments for less recursion.
(extract_muldiv): New. Prevent runaway recursion.
Co-Authored-By: Richard Henderson <rth@redhat.com>
From-SVN: r62963
PR f/9258
* global.c (struct allocno): Add no_stack_reg.
(global_conflicts): Set no_stack_reg.
(find_reg): Use it.
* convert.c (convert_to_real): Fold - and abs only when profitable.
* fold-const.c (fold): Fold truncates in - and abs.
* gcc.c-torture/compile/20030115-1.c: New test.
* gcc.dg/i386-fpcvt-1.c: New test.
* gcc.dg/i386-fpcvt-2.c: New test.
From-SVN: r61329
* fold-const.c (fold) [COND_EXPR]: Avoid NOP_EXPRs better.
* integrate.c (copy_decl_for_inlining): Don't clear the rtl for
static/external decls.
cp/
* call.c (build_conditional_expr): Stabilize lvalues properly.
* cvt.c (ocp_convert): Don't build NOP_EXPRs of class type.
* tree.c (lvalue_p_1): Don't allow sloppy NOP_EXPRs as lvalues.
Don't allow CALL_EXPR or VA_ARG_EXPR, either.
* call.c (convert_like_real): Call decl_constant_value for an
IDENTITY_CONV even if there are no more conversions.
* cvt.c (build_up_reference): Don't push unnamed temps.
* decl2.c (do_namespace_alias): Namespace aliases are DECL_EXTERNAL.
* dump.c (cp_dump_tree): Don't try to dump class-specific fields
for a backend struct.
* except.c (wrap_cleanups_r, build_throw): Make
MUST_NOT_THROW_EXPRs void.
* init.c (expand_default_init): Update to handle MUST_NOT_THROW_EXPR.
* init.c (build_vec_delete_1): Pre-evaluate the base address.
* init.c (get_temp_regvar): Simplify logic.
* tree.c (cp_copy_res_decl_for_inlining): Only do debug tweaks if
our replacement is a decl.
From-SVN: r60851
PR c/8639
* fold-const.c (extract_muldiv): Don't propagate division unless
both arguments are multiples of C.
* gcc.c-torture/execute/20021119-1.c: New.
From-SVN: r59466
2002-11-05 Andrew Haley <aph@redhat.com>
* fold-const.c (fold): Don't transform (a0 op compound(a1,a2))
to (compound(a1,a0 op a2)) if a0 or a1 have side effects.
From-SVN: r58822
* fold-const.c (fold_binary_op_with_conditional_arg): Improve
handling of cases where one or both branches of the conditional
have void type, i.e. throw an exception or don't return.
(fold): Only apply (and undo) type conversion to the non-void
branches of a COND_EXPR.
* f/com.c (ffecom_subscript_check_): Cast the failure branch
of the bounds check COND_EXPR to void, to indicate noreturn.
(ffe_truthvalue_conversion): Only apply truth value conversion
to the non-void branches of a COND_EXPR.
From-SVN: r58661
PR c++/7209
* fold_const.c (fold_binary_op_with_conditional_arg): Always
build compound_expr if we used save_expr.
testsuite:
* g++.dg/expr/cond1.C: New test.
From-SVN: r58405
* fold-const.c (make_range): Only narrow to signed range if
the signed range is smaller than the unsigned range.
* gcc.c-torture/execute/20020911-1.c: New.
From-SVN: r57050
* fold-const.c (associate_trees): Only optimize NEGATE_EXPR in one
of the operands into MINUS_EXPR if code is PLUS_EXPR.
* gcc.c-torture/execute/20020805-1.c: New test.
From-SVN: r56058
* Makefile.in: rtlanal.o now depends upon real.h.
* flags.h [flag_signaling_nans]: New flag.
[HONOR_SNANS]: New macro.
* toplev.c [flag_signaling_nans]: Initialize to false.
(f_options): Add processing for "-fsignaling-nans".
(set_fast_math_flags): Clear flag_signaling_nans with -ffast-math.
(process_options): flag_signaling_nans implies flag_trapping_math.
* c-common.c (cb_register_builtins): Define __SUPPORT_SNAN__
when -fsignaling-nans. First step to implementing WG14's N965.
* fold-const.c (fold) [MULT_EXPR]: Conditionalize transforming
1.0 * x into x, and -1.0 * x into -x on !HONOR_SNANS.
[RDIV_EXPR]: Conditionalize x/1.0 into x on !HONOR_SNANS.
* simplify-rtx.c (simplify_relational_operation): Conditionalize
transforming abs(x) < 0.0 into false on !HONOR_SNANS.
* rtlanal.c: #include real.c for TARGET_FLOAT_FORMAT definitions
required by HONOR_SNANS. (may_trap_p): Floating point DIV, MOD,
UDIV, UMOD, GE, GT, LE, LT and COMPARE may always trap with
-fsignaling_nans. EQ and NE only trap for flag_signaling_nans
not flag_trapping_math (i.e. HONOR_SNANS but not HONOR_NANS).
* doc/invoke.texi: Document new -fsignaling-nans compiler option.
From-SVN: r55804
PR opt/4046
* fold-const.c (fold) [COND_EXPR]: Simplify A ? 0 : 1 to !A,
A ? B : 0 to A && B and A ? B : 1 into !A || B if both A and
B are truth values.
From-SVN: r55153
* fold-const.c (comparison_to_compcode): New function to convert
an comparison TREE CODE into a bit-based representation.
(compcode_to_comparison): New function to convert from this bit
based representation back to a comparison TREE CODE.
(fold_truthop): Simplify (x<y) && (x==y) and related composite
comparisons.
* gcc.c-tortuture/execute/compare-1.c: New test case.
* gcc.c-tortuture/execute/compare-2.c: New test case.
* gcc.c-tortuture/execute/compare-3.c: New test case.
From-SVN: r54647
* convert.c (convert_to_integer) [LSHIFT_EXPR]: Don't pass
the truncation down when the target type is signed.
[trunc1]: Use unsigned arithmetic for LSHIFT_EXPR.
* fold-const.c (extract_muldiv) [NOP_EXPR]: Don't pass through
the conversion if the target type is a smaller type.
From-SVN: r54627
* fold-const.c (fold) [EQ_EXPR]: Place both integer and real
constants last in comparisons. Optimize (x+1.0)>0.0 into the
equivalent x > -1.0 when -ffast-math.
* gcc.dg/20020607-2.c: New test case.
From-SVN: r54356
* tree.c (real_minus_onep): New function to test for -1.0.
* fold-const.c (fold) [MULT_EXPR]: Optimize -1.0*x into -x.
* gcc.dg/fnegate-1.c: New test case.
From-SVN: r54149
* tree.h: Forward-declare struct realvaluetype.
(struct tree_real_cst): Point to the REAL_VALUE_TYPE, do not
contain it.
(TREE_REAL_CST_PTR): New accessor.
(TREE_REAL_CST): Update.
* real.h: Include machmode.h.
(realvaluetype): Make it struct realvaluetype, not a typedef.
(build_real): Prototype here.
* tree.c: Include real.h.
(build_real): Allocate the REAL_VALUE_TYPE as a separate
object in GC memory, set TREE_REAL_CST_PTR to point to it.
(build_real_from_int_cst): Use build_real.
* ggc-common.c (ggc_mark_trees): Mark TREE_REAL_CST_PTR of a
REAL_CST.
* builtins.c, c-common.c, c-lex.c, dwarf2out.c, expr.c,
fold-const.c, print-tree.c, real.c, cp/mangle.c, cp/tree.c,
f/bld.c, f/com.c, f/expr.c, f/target.c, java/decl.c,
java/jcf-parse.c, java/parse.y, java/typeck.c: Include real.h.
* Makefile.in, cp/Make-lang.in, f/Make-lang.in,
java/Make-lang.in: Update dependency lists.
From-SVN: r53959
2002-05-15 Jakub Jelinek <jakub@redhat.com>
* fold-const.c (fold): Fix a typo.
2002-05-15 Eric Botcazou <ebotcazou@multimania.com>
* fold-const.c (fold) [LT_EXPR]: Move the transformation of a
comparison against the highest or lowest integer value before
the 'X >= CST to X > (CST - 1)' and 'X < CST to X <= (CST - 1)'
transformation and that of an unsigned comparison against 0
right after.
From-SVN: r53493
PR opt/3995
* fold-const.c (sign_bit_p): New function.
(fold) [EQ_EXPR]: Use this to convert (A & C) == 0 into A >= 0 and
(A & C) != 0 into A < 0, when constant C is the sign bit of A's type.
Reapply fold when converting (A & C) == C into (A & C) != 0.
(fold_binary_op_with_conditional_arg): Fix typo in comment.
testsuite/
* gcc.c-torture/execute/20020506-1.c: New test case.
From-SVN: r53241
* fold-const.c (lshift-double): Cast the high word to an unsigned
HOST_WIDE_INT when extracting sign bit to avoid compiler warning.
(div_and_round_double): Cast carry to a signed HOST_WIDE_INT to
avoid compiler warning. (fold): Remove redundant code from
BIT_AND_EXPR as integer operands are canonicalized to be arg1.
From-SVN: r53228
2002-04-18 David S. Miller <davem@redhat.com>
* config/sparc/sparc.h (BRANCH_COST): Define.
* fold-const.c (BRANCH_COST): Don't provide default here, expr.h
does it.
From-SVN: r52500
* fold-const.c (fold) [NOP_EXPR]: Convert (T)(x&c) into ((T)x&(T)c)
for integer constant c (if x has unsigned type or sign bit is not
set in c). This folds the zero/sign extension into the bit-wise and
operation.
* gcc.c-torture/compile/20020415-1.c: New.
Co-Authored-By: Jakub Jelinek <jakub@redhat.com>
From-SVN: r52465
* toplev.c: Don't include setjmp.h. Kill float_handler_set,
float_handled, float_handler, float_signal, set_float_handler,
and do_float_handler. Set handler for SIGFPE to crash_signal.
* toplev.h: Don't prototype do_float_handler.
* c-lex.c: Fold parse_float into lex_number. Make warning
about portability of hex float constants more informative, and
don't issue it on top of a syntax error.
* fold-const.c: Fold const_binop_1 and fold_convert_1 into
their callers.
* real.h: Define REAL_VALUE_ABS here...
* simplify-rtx.c: ... not here. Fold check_fold_consts,
simplify_unary_real, simplify_binary_real, and
simplify_binary_is2orm1 into their callers.
* tree.c: Fold build_real_from_int_cst_1 into caller.
* doc/tm.texi: Document REAL_VALUE_ABS and REAL_VALUE_NEGATIVE.
* tsystem.h: Include float.h here...
* libgcc2.c: ... not here.
java:
* lex.c: Change java_perform_atof to take normal parameters
instead of a pointer to a parameter block. Call it directly
from java_lex.
testsuite:
* gcc.dg/c90-hexfloat-1.c: Adjust error regexps.
From-SVN: r51336
* real.h: Don't define REAL_INFINITY or REAL_IS_NOT_DOUBLE.
Always make REAL_VALUE_TYPE a struct containing an array of
HOST_WIDE_INT, not a double. Tidy up the code deciding how
big it is. Don't declare or use union real_extract.
* emit-rtl.c (init_emit_once), varasm.c (immed_real_const_1,
decode_rtx_const, output_constant_pool), config/a29k/a29k.c
(print_operand), config/arm/arm.c (output_move_double),
config/arm/arm.md (consttable_4, consttable_8),
config/romp/romp.c (output_fpops), config/s390/s390.h
(ASM_OUTPUT_SPECIAL_POOL_ENTRY), config/xtensa/xtensa.c
(xtensa_output_literal): Don't use union real_extract.
* config/dsp16xx/dsp16xx.c (print_operand), config/i860/i860.c
(sfmode_constant_to_ulong), config/ns32k/merlin.h
(PRINT_OPERAND), config/ns32k/ns32k.c (print_operand),
config/pdp11/pdp11.h (PRINT_OPERAND), config/we32k/we32k.h
(PRINT_OPERAND): Don't use local version of union
real_extract.
* config/convex/convex.c (check_float_value), config/vax/vax.c
(vax_float_literal), config/m88k/m88k.md (divdf3),
config/dsp16xx/dsp16xx.md (fixuns_trunchfhi2),
config/pdp11/pdp11.c (output_move_quad): Don't do host
arithmetic on target floating point quantities.
* config/a29k/a29k.md, config/dsp16xx/dsp16xx.c
(output_dsp16xx_float_const): Don't test HOST_FLOAT_FORMAT.
* fold-const.c (fold), simplify-rtx.c (simplify_binary_real):
Use MODE_HAS_INFINITIES rather than #ifdef REAL_INFINITY.
* real.c (earith): Test INFINITY rather than REAL_INFINITY;
NANS implies INFINITY, so can drop #ifdef NANS inside #ifndef
INFINITY.
* print-rtl.c (print_rtx): Disable code which needs
floating-point emulator.
* libgcc2.c: Include float.h and use DBL_MANT_DIG,
FLT_MANT_DIG, to define DF_SIZE and SF_SIZE, rather than
depending on HOST_FLOAT_FORMAT to be defined properly.
* ch/grant.c, cp/error.c: Always use REAL_VALUE_TO_DECIMAL;
don't test REAL_IS_NOT_DOUBLE.
* config/1750a/1750a.c (get_double, float_label): Delete.
(print_operand): Delete huge commented-out chunk. Use
REAL_VALUE_TO_DECIMAL.
* config/1750a/1750a-protos.h: Delete prototypes of deleted
functions.
* config/convex/convex.h: Always set TARGET_FLOAT_FORMAT to
IEEE_FLOAT_FORMAT.
* config/i370/i370.h (PRINT_OPERAND [TARGET_HLASM version]):
Use REAL_VALUE_TO_DECIMAL as ELF version does.
* config/m88k/m88k.c (real_power_of_2_operand,
legitimize_operand): Take the REAL_VALUE_TYPE and/or union
real_extract out of the union; run the input through
REAL_VALUE_TO_TARGET_DOUBLE, then plug the pair of longwords
from that into the union.
* config/pdp11/pdp11.c (output_move_double): Rearrange
parentheses to make automatic indenter happy.
* doc/tm.texi (Cross-compilation): Rename node to "Floating
Point" and rewrite to describe current situation. Also adjust
documentation of REAL_VALUE_TO_TARGET_SINGLE and friends to
match code.
* doc/rtl.texi: Adjust cross reference.
From-SVN: r51210
* defaults.h (MODE_HAS_NANS, MODE_HAS_INFINITIES): New.
(MODE_HAS_SIGNED_ZEROS, MODE_HAS_SIGN_DEPENDENT_ROUNDING): New.
* flags.h (HONOR_NANS, HONOR_INFINITIES, HONOR_SIGNED_ZEROS): New.
(HONOR_SIGN_DEPENDENT_ROUNDING): New.
* builtins.c (expand_builtin_mathfn): Use HONOR_NANS.
* c-common.c (truthvalue_conversion): Reduce x - y != 0 to x != y
unless x and y could be infinite.
(expand_unordered_cmp): New, mostly split from expand_tree_builtin.
Check that the common type of both arguments is a real, even for
targets without unordered comparisons. Allow an integer argument
to be compared against a real.
(expand_tree_builtin): Use expand_unordered_cmp.
* combine.c (combine_simplify_rtx): Use the new HONOR_... macros.
* cse.c (fold_rtx): Likewise. Fix indentation.
* fold-const.c (fold_real_zero_addition_p): New.
(fold): Use it, and the new HONOR_... macros.
* ifcvt.c (noce_try_minmax): Use the new HONOR_... macros.
* jump.c (reversed_comparison_code_parts): After searching for
the true comparison mode, use HONOR_NANS to decide whether it
can be safely reversed.
(reverse_condition_maybe_unordered): Remove IEEE check.
* simplify-rtx.c (simplify_binary_operation): Use the new macros
to decide which simplifications are valid. Allow the following
simplifications for IEEE: (-a + b) to (b - a), (a + -b) to (a - b),
and (a - -b) to (a + b).
(simplify_relational_operation): Use HONOR_NANS.
* doc/tm.texi: Document the MODE_HAS_... macros.
From-SVN: r50401
* tree.def (VIEW_CONVERT_EXPR): New tree code.
* tree.h (handled_component_p): New declaration.
* alias.c (handled_component_p): Move function from here ...
* expr.c (handled_component_p): ... to here and non longer static.
(is_zeros_p, case VIEW_CONVERT_EXPR): New case.
(store_field): Remove unneeded MEM_COPY_ATTRIBUTES call.
(get_inner_reference): Handle VIEW_CONVERT_EXPR.
(expand_expr, case VAR_DECL): Clean up handling of re-layout case.
(expand_expr, case VIEW_CONVERT_EXPR): New case.
(expand_expr, cse ADDR_EXPR): No need to copy if already
BIGGEST_ALIGNMENT.
* fold-const.c (fold, case VIEW_CONVERT_EXPR): New case.
* varasm.c (initializer_constant_value_p, case VIEW_CONVERT_EXPR): New.
(output_constant): Handle VIEW_CONVERT_EXPR.
* dwarf2out.c (loc-descriptor_from_tree, add_bound_info):
Add new case for VIEW_CONVERT_EXPR.
From-SVN: r47179
