#include <recog.h>
Data Fields | |
| rtx_insn * | insn |
| rtx | from |
| rtx | to |
| unsigned int | num_replacements |
| uint16_t | result_flags: 16 |
| uint16_t | should_unshare: 1 |
| uint16_t | should_check_mems: 1 |
| uint16_t | should_note_simplifications: 1 |
| uint16_t | spare: 13 |
| const char * | failure_reason |
| unsigned int | mem_depth = 0 |
| unsigned int | assoc_count = 0 |
Static Public Attributes | |
| static const uint16_t | UNSIMPLIFIED = 1U << 0 |
| static const uint16_t | FIRST_SPARE_RESULT = 1U << 1 |
| static const unsigned int | max_assoc_count = 64 |
A class for substituting one rtx for another within an instruction, or for recursively simplifying the instruction as-is. Derived classes can record or filter certain decisions.
|
inline |
Try to simplify INSN without performing a substitution.
References insn, insn_propagation(), and NULL_RTX.
Referenced by apply_to_rvalue_1(), and insn_propagation().
|
inline |
Try to replace FROM with TO in INSN. SHARED_P is true if TO is shared with other instructions, false if INSN can use TO directly.
References failure_reason, false, from, insn, num_replacements, result_flags, should_check_mems, should_note_simplifications, should_unshare, spare, and to.
Try to process the lvalue expression at *LOC. Return true on success; leave the caller to clean up on failure.
References apply_to_mem_1(), apply_to_rvalue_1(), failure_reason, from, GET_CODE, MEM_P, read_modify_subreg_p(), reg_overlap_mentioned_p(), REG_P, SUBREG_P, SUBREG_REG, and XEXP.
Referenced by apply_to_pattern_1().
Try to process the address of memory expression MEM. Return true on success; leave the caller to clean up on failure.
References apply_to_rvalue_1(), check_mem(), simplify_context::mem_depth, num_validated_changes(), should_check_mems, and XEXP.
Referenced by apply_to_lvalue_1(), and apply_to_rvalue_1().
Like apply_to_rvalue, but specifically for the case where *LOC is in a note. This never changes the INSN_CODE.
References apply_to_rvalue(), insn, and INSN_CODE.
Apply this insn_propagation object's simplification or substitution to the instruction pattern at LOC.
References apply_to_pattern_1(), cancel_changes(), num_changes, and num_validated_changes().
Referenced by fixup_debug_use().
Try to process the instruction pattern at *LOC. Return true on success; leave the caller to clean up on failure.
References apply_to_lvalue_1(), apply_to_pattern_1(), apply_to_rvalue_1(), ASM_OPERANDS_INPUT, ASM_OPERANDS_INPUT_LENGTH, COND_EXEC_CODE, COND_EXEC_TEST, GET_CODE, i, SET, SET_DEST, SET_SRC, XEXP, XVECEXP, and XVECLEN.
Referenced by apply_to_pattern(), and apply_to_pattern_1().
Apply this insn_propagation object's simplification or substitution to the rvalue expression at LOC.
References apply_to_rvalue_1(), cancel_changes(), num_changes, and num_validated_changes().
Referenced by apply_to_note(), and apply_to_rvalue_1().
Try to process the rvalue expression at *LOC. Return true on success; leave the caller to clean up on failure.
References apply_to_mem_1(), apply_to_rvalue(), apply_to_rvalue_1(), asm_noperands(), cancel_changes(), changes, CONSTANT_P, copy_rtx(), failure_reason, FOR_EACH_SUBRTX, from, gcc_assert, gcc_unreachable, GET_CODE, GET_MODE, GET_RTX_CLASS, GET_RTX_FORMAT, HARD_REGISTER_P, i, insn, insn_propagation(), simplify_context::mem_depth, note_simplification(), NULL_RTX, num_changes, num_replacements, num_validated_changes(), paradoxical_subreg_p(), PATTERN(), REG_CAN_CHANGE_MODE_P, REG_NREGS, reg_overlap_mentioned_p(), REG_P, register_asm_p(), REGNO, result_flags, RTX_AUTOINC, RTX_BIN_ARITH, RTX_BITFIELD_OPS, RTX_COMM_ARITH, RTX_COMM_COMPARE, RTX_COMPARE, RTX_CONST_OBJ, rtx_equal_p(), RTX_EXTRA, RTX_INSN, RTX_MATCH, RTX_OBJ, RTX_TERNARY, RTX_UNARY, should_note_simplifications, should_unshare, simplify_context::simplify_binary_operation(), simplify_context::simplify_gen_binary(), simplify_context::simplify_relational_operation(), simplify_context::simplify_subreg(), simplify_context::simplify_ternary_operation(), simplify_context::simplify_unary_operation(), split_const(), SUBREG_BYTE, subreg_lowpart_offset(), SUBREG_REG, swap_commutative_operands_p(), to, UNSIMPLIFIED, validate_change(), validate_subreg(), validate_unshare_change(), XEXP, XVECEXP, and XVECLEN.
Referenced by apply_to_lvalue_1(), apply_to_mem_1(), apply_to_pattern_1(), apply_to_rvalue(), and apply_to_rvalue_1().
Referenced by apply_to_mem_1().
|
inherited |
Generates a subreg to get the least significant part of EXPR (in mode INNER_MODE) to OUTER_MODE.
References simplify_gen_subreg(), and subreg_lowpart_offset().
Referenced by lowpart_subreg(), simplify_binary_operation_1(), simplify_gen_vec_select(), simplify_relational_operation_1(), and simplify_subreg().
Referenced by apply_to_rvalue_1().
|
privateinherited |
Subroutine of simplify_binary_operation to simplify a commutative, associative binary operation CODE with result mode MODE, operating on OP0 and OP1. CODE is currently one of PLUS, MULT, AND, IOR, XOR, SMIN, SMAX, UMIN or UMAX. Return zero if no simplification or canonicalization is possible.
References assoc_count, GET_CODE, max_assoc_count, NULL_RTX, simplify_binary_operation(), simplify_gen_binary(), swap_commutative_operands_p(), and XEXP.
Referenced by simplify_binary_operation_1().
|
inherited |
Simplify a binary operation CODE with result mode MODE, operating on OP0 and OP1. Return 0 if no simplification is possible. Don't use this for relational operations such as EQ or LT. Use simplify_relational_operation instead.
References avoid_constant_pool_reference(), gcc_assert, GET_RTX_CLASS, NULL_RTX, RTX_COMM_ARITH, RTX_COMM_COMPARE, RTX_COMPARE, simplify_binary_operation_1(), simplify_const_binary_operation(), simplify_gen_binary(), and swap_commutative_operands_p().
Referenced by insn_propagation::apply_to_rvalue_1(), simplify_associative_operation(), simplify_binary_operation(), simplify_binary_operation_1(), simplify_binary_operation_series(), simplify_gen_binary(), and simplify_plus_minus().
|
privateinherited |
Subroutine of simplify_binary_operation. Simplify a binary operation CODE with result mode MODE, operating on OP0 and OP1. If OP0 and/or OP1 are constant pool references, TRUEOP0 and TRUEOP1 represent the actual constants.
References a, as_a(), avoid_constant_pool_reference(), b, BITS_PER_WORD, cfun, CLZ_DEFINED_VALUE_AT_ZERO, COMPARISON_P, CONST0_RTX, const0_rtx, CONST1_RTX, const1_rtx, CONST_DOUBLE_AS_FLOAT_P, const_double_from_real_value(), CONST_DOUBLE_REAL_VALUE, CONST_INT_P, CONST_SCALAR_INT_P, const_true_rtx, const_vec_duplicate_p(), CONST_VECTOR_ELT, CONSTM1_RTX, constm1_rtx, contains_symbolic_reference_p(), copy_rtx(), count, d1, dconst0, dconst1, dconst2, dconstm1, DECIMAL_FLOAT_MODE_P, exact_log2(), wi::exact_log2(), f1, FLOAT_MODE_P, gcc_assert, gcc_unreachable, gen_const_vec_series(), GEN_INT, gen_int_mode(), gen_int_shift_amount(), rtl_hooks::gen_lowpart_no_emit, gen_rtx_CONST_VECTOR(), gen_vec_duplicate(), GET_CODE, GET_MODE, GET_MODE_BITSIZE(), GET_MODE_CLASS, GET_MODE_INNER, GET_MODE_MASK, GET_MODE_NUNITS(), GET_MODE_PRECISION(), GET_MODE_SIZE(), GET_MODE_UNIT_BITSIZE, GET_MODE_UNIT_PRECISION, HONOR_NANS(), HONOR_SIGN_DEPENDENT_ROUNDING(), HONOR_SIGNED_ZEROS(), HONOR_SNANS(), HOST_BITS_PER_WIDE_INT, HOST_WIDE_INT_1U, HWI_COMPUTABLE_MODE_P(), i, i1, immed_wide_int_const(), IN_RANGE, INTEGRAL_MODE_P, INTVAL, is_a(), known_eq, known_ge, known_lt, lowpart_subreg(), wi::mask(), match_plus_neg_pattern(), maybe_ge, mem_depth, wi::minus_one(), mode_signbit_p(), wi::neg(), neg_poly_int_rtx(), nonzero_bits(), NULL_RTX, wi::one(), optimize_function_for_speed_p(), paradoxical_subreg_p(), plus_constant(), plus_minus_operand_p(), poly_int_rtx_p(), real_arithmetic(), real_convert(), real_equal(), REAL_VALUE_NEGATIVE, REAL_VALUE_TYPE, reverse_rotate_by_imm_p(), reversed_comparison(), rtvec_alloc(), RTVEC_ELT, rtx_equal_p(), SCALAR_FLOAT_MODE_P, SCALAR_INT_MODE_P, wi::set_bit_in_zero(), set_src_cost(), SHIFT_COUNT_TRUNCATED, side_effects_p(), simplify_associative_operation(), simplify_binary_operation(), simplify_binary_operation_series(), simplify_byte_swapping_operation(), simplify_distributive_operation(), simplify_gen_binary(), simplify_gen_relational(), simplify_gen_unary(), simplify_logical_relational_operation(), simplify_plus_minus(), simplify_rotate_op(), simplify_unary_operation(), STORE_FLAG_VALUE, subreg_lowpart_p(), subreg_memory_offset(), SUBREG_P, SUBREG_REG, poly_int< N, C >::to_constant(), trunc_int_for_mode(), UINTVAL, unwrap_const_vec_duplicate(), val_signbit_p(), valid_for_const_vector_p(), vec_duplicate_p(), vec_series_highpart_p(), vec_series_lowpart_p(), VECTOR_MODE_P, word_mode, WORD_REGISTER_OPERATIONS, XEXP, XVECEXP, XVECLEN, and y.
Referenced by simplify_binary_operation().
|
privateinherited |
Subroutine of simplify_binary_operation_1 that looks for cases in which OP0 and OP1 are both vector series or vector duplicates (which are really just series with a step of 0). If so, try to form a new series by applying CODE to the bases and to the steps. Return null if no simplification is possible. MODE is the mode of the operation and is known to be a vector integer mode.
References const0_rtx, gen_vec_series(), GET_MODE_INNER, NULL_RTX, simplify_binary_operation(), vec_duplicate_p(), and vec_series_p().
Referenced by simplify_binary_operation_1().
|
privateinherited |
Subroutine of simplify_binary_operation to simplify a binary operation CODE that can commute with byte swapping, with result mode MODE and operating on OP0 and OP1. CODE is currently one of AND, IOR or XOR. Return zero if no simplification or canonicalization is possible.
References CONST_SCALAR_INT_P, GET_CODE, NULL_RTX, simplify_gen_binary(), simplify_gen_unary(), and XEXP.
Referenced by simplify_binary_operation_1().
|
privateinherited |
Recognize expressions of the form (X CMP 0) ? VAL : OP (X) where OP is CLZ or CTZ and VAL is the value from CLZ_DEFINED_VALUE_AT_ZERO or CTZ_DEFINED_VALUE_AT_ZERO respectively and return OP (X) if the expression can be simplified to that or NULL_RTX if not. Assume X is compared against zero with CMP_CODE and the true arm is TRUE_VAL and the false arm is FALSE_VAL.
References as_a(), CLZ_DEFINED_VALUE_AT_ZERO, CONST_INT_P, CTZ_DEFINED_VALUE_AT_ZERO, GET_CODE, GET_MODE, INTVAL, NULL_RTX, rtx_equal_p(), and XEXP.
Referenced by simplify_ternary_operation().
|
privateinherited |
|
privateinherited |
Subroutine of simplify_binary_operation_1. Un-distribute a binary operation CODE with result mode MODE, operating on OP0 and OP1. e.g. simplify (xor (and A C) (and (B C)) to (and (xor (A B) C). Returns NULL_RTX if no simplification is possible.
References gcc_assert, GET_CODE, GET_RTX_CLASS, NULL_RTX, RTX_COMM_ARITH, rtx_equal_p(), side_effects_p(), simplify_gen_binary(), and XEXP.
Referenced by simplify_binary_operation_1().
|
inherited |
Make a binary operation by properly ordering the operands and seeing if the expression folds.
References GET_RTX_CLASS, RTX_COMM_ARITH, simplify_binary_operation(), and swap_commutative_operands_p().
Referenced by insn_propagation::apply_to_rvalue_1(), simplify_associative_operation(), simplify_binary_operation(), simplify_binary_operation_1(), simplify_byte_swapping_operation(), simplify_distributive_operation(), simplify_gen_binary(), simplify_merge_mask(), simplify_relational_operation_1(), simplify_subreg(), simplify_ternary_operation(), simplify_truncation(), and simplify_unary_operation_1().
|
inherited |
Likewise, for relational operations. CMP_MODE specifies mode comparison is done in.
References simplify_relational_operation().
Referenced by simplify_binary_operation_1(), simplify_gen_relational(), simplify_logical_relational_operation(), simplify_merge_mask(), simplify_relational_operation(), simplify_relational_operation_1(), simplify_subreg(), simplify_ternary_operation(), and simplify_unary_operation_1().
|
inherited |
Make a SUBREG operation or equivalent if it folds.
References CONST_DOUBLE_AS_FLOAT_P, CONST_FIXED_P, CONST_SCALAR_INT_P, gen_rtx_SUBREG(), GET_CODE, GET_MODE, MODE_COMPOSITE_P, NULL_RTX, simplify_subreg(), and validate_subreg().
Referenced by lowpart_subreg(), simplify_gen_subreg(), simplify_subreg(), and simplify_truncation().
|
inherited |
Likewise for ternary operations.
References simplify_ternary_operation().
Referenced by simplify_gen_ternary(), simplify_merge_mask(), simplify_ternary_operation(), simplify_truncation(), and simplify_unary_operation_1().
|
inherited |
Make a unary operation by first seeing if it folds and otherwise making the specified operation.
References simplify_unary_operation().
Referenced by simplify_binary_operation_1(), simplify_byte_swapping_operation(), simplify_gen_unary(), simplify_merge_mask(), simplify_relational_operation_1(), simplify_truncation(), and simplify_unary_operation_1().
Generate RTX to select element at INDEX out of vector OP.
References gcc_assert, GEN_INT, gen_rtvec(), GET_MODE, GET_MODE_INNER, GET_MODE_SIZE(), known_eq, lowpart_subreg(), subreg_lowpart_offset(), and VECTOR_MODE_P.
Referenced by simplify_gen_vec_select().
|
privateinherited |
Simplify a logical operation CODE with result mode MODE, operating on OP0 and OP1, in the case where both are relational operations. Assume that OP0 is inverted if INVERT0_P is true. Return 0 if no such simplification is possible.
References COMPARISON_P, comparison_to_mask(), const0_rtx, const_true_rtx, GET_CODE, GET_MODE, GET_MODE_CLASS, HONOR_NANS(), IN_RANGE, mask_to_comparison(), mask_to_unsigned_comparison(), relational_result(), rtx_equal_p(), side_effects_p(), simplify_gen_relational(), unsigned_comparison_to_mask(), and XEXP.
Referenced by simplify_binary_operation_1(), and simplify_relational_operation_1().
Try to simplify X given that it appears within operand OP of a VEC_MERGE operation whose mask is MASK. X need not use the same vector mode as the VEC_MERGE, but it must have the same number of elements. Return the simplified X on success, otherwise return NULL_RTX.
References BINARY_P, COMPARISON_P, gcc_assert, GET_CODE, GET_MODE, GET_MODE_NUNITS(), GET_RTX_CLASS, known_eq, NULL_RTX, rtx_equal_p(), RTX_TERNARY, side_effects_p(), simplify_gen_binary(), simplify_gen_relational(), simplify_gen_ternary(), simplify_gen_unary(), simplify_merge_mask(), UNARY_P, VECTOR_MODE_P, and XEXP.
Referenced by simplify_merge_mask(), and simplify_ternary_operation().
|
privateinherited |
Simplify and canonicalize a PLUS or MINUS, at least one of whose operands may be another PLUS or MINUS. Rather than test for specific case, we do this by a brute-force method and do all possible simplifications until no more changes occur. Then we rebuild the operation. May return NULL_RTX when no changes were made.
References arg_pointer_rtx, CASE_CONST_SCALAR_INT, changed, CONST_INT_P, CONSTANT_P, CONSTM1_RTX, fixed_regs, frame_pointer_rtx, gcc_assert, GET_CODE, GET_MODE, global_regs, i, INTVAL, neg_poly_int_rtx(), NULL_RTX, plus_constant(), poly_int_rtx_p(), REG_P, REGNO, simplify_binary_operation(), simplify_const_binary_operation(), simplify_plus_minus_op_data_cmp(), stack_pointer_rtx, swap_commutative_operands_p(), and XEXP.
Referenced by simplify_binary_operation_1().
|
inherited |
Like simplify_binary_operation except used for relational operators. MODE is the mode of the result. If MODE is VOIDmode, both operands must not also be VOIDmode. CMP_MODE specifies in which mode the comparison is done in, so it is the mode of the operands. If CMP_MODE is VOIDmode, it is taken from the operands or, if both are VOIDmode, the operands are compared in "infinite precision".
References avoid_constant_pool_reference(), const0_rtx, GET_CODE, GET_MODE, GET_MODE_CLASS, NULL_RTX, relational_result(), simplify_const_relational_operation(), simplify_gen_relational(), simplify_relational_operation_1(), swap_commutative_operands_p(), swap_condition(), and XEXP.
Referenced by insn_propagation::apply_to_rvalue_1(), simplify_gen_relational(), simplify_relational_operation(), and simplify_ternary_operation().
|
privateinherited |
This part of simplify_relational_operation is only used when CMP_MODE is not in class MODE_CC (i.e. it is a real comparison). MODE is the mode of the result, while CMP_MODE specifies in which mode the comparison is done in, so it is the mode of the operands.
References COMPARISON_P, CONST0_RTX, const0_rtx, const1_rtx, CONST_INT_P, CONST_SCALAR_INT_P, CONSTANT_P, constm1_rtx, copy_rtx(), GET_CODE, GET_MODE, GET_MODE_PRECISION(), GET_MODE_SIZE(), INTEGRAL_MODE_P, INTVAL, is_a(), is_int_mode(), lowpart_subreg(), nonzero_bits(), NULL, NULL_RTX, partial_subreg_p(), reversed_comparison_code(), rtx_equal_p(), SCALAR_INT_MODE_P, side_effects_p(), simplify_gen_binary(), simplify_gen_relational(), simplify_gen_unary(), simplify_logical_relational_operation(), simplify_rtx(), STORE_FLAG_VALUE, subreg_lowpart_p(), UINTVAL, and XEXP.
Referenced by simplify_relational_operation().
|
privateinherited |
|
inherited |
Simplify SUBREG:OUTERMODE(OP:INNERMODE, BYTE) Return 0 if no simplifications are possible.
References adjust_address_nv, COMPARISON_P, CONST0_RTX, CONST_DOUBLE_AS_FLOAT_P, CONST_FIXED_P, CONST_INT_P, CONST_POLY_INT_P, const_poly_int_value(), CONST_SCALAR_INT_P, poly_int< NUM_POLY_INT_COEFFS, wide_int >::from(), gcc_assert, gen_rtx_REG_offset(), gen_rtx_SUBREG(), gen_vec_duplicate(), GET_CODE, GET_MODE, GET_MODE_ALIGNMENT, GET_MODE_BITSIZE(), GET_MODE_CLASS, GET_MODE_INNER, GET_MODE_NUNITS(), GET_MODE_PRECISION(), GET_MODE_SIZE(), GET_MODE_UNIT_SIZE, HARD_REGISTER_NUM_P, HARD_REGISTER_P, have_insn_for(), HOST_BITS_PER_WIDE_INT, HOST_WIDE_INT_1U, immed_wide_int_const(), int_mode_for_size(), INTVAL, is_a(), known_eq, known_ge, known_gt, known_le, known_lt, lowpart_subreg(), maybe_ge, maybe_gt, MEM_ADDR_SPACE, MEM_ALIGN, MEM_P, MEM_VOLATILE_P, mode_dependent_address_p(), NULL_RTX, ORIGINAL_REGNO, paradoxical_subreg_p(), REG_P, REGNO, SCALAR_INT_MODE_P, SET, SIGNED, simplify_const_vector_byte_offset(), simplify_const_vector_subreg(), simplify_gen_binary(), simplify_gen_relational(), simplify_gen_subreg(), simplify_immed_subreg(), simplify_subreg(), simplify_subreg_regno(), simplify_truncation(), SUBREG_BYTE, subreg_highpart_offset(), subreg_lowpart_offset(), subreg_lowpart_p(), subreg_lsb_1(), subreg_memory_offset(), SUBREG_PROMOTED_GET, SUBREG_PROMOTED_SET, SUBREG_PROMOTED_SIGN, SUBREG_PROMOTED_VAR_P, SUBREG_REG, UINTVAL, validate_subreg(), vec_duplicate_p(), VECTOR_MODE_P, word_mode, and XEXP.
Referenced by insn_propagation::apply_to_rvalue_1(), simplify_gen_subreg(), simplify_subreg(), and simplify_subreg().
|
inherited |
Simplify CODE, an operation with result mode MODE and three operands, OP0, OP1, and OP2. OP0_MODE was the mode of OP0 before it became a constant. Return 0 if no simplifications is possible.
References avoid_constant_pool_reference(), COMPARISON_P, CONST0_RTX, const0_rtx, CONST_INT_P, CONST_VECTOR_ELT, CONST_VECTOR_NUNITS, gcc_assert, gcc_unreachable, gen_int_mode(), gen_rtx_CONST_VECTOR(), GET_CODE, GET_MODE, GET_MODE_NUNITS(), GET_MODE_PRECISION(), HONOR_NANS(), HONOR_SIGNED_ZEROS(), HOST_BITS_PER_WIDE_INT, HOST_WIDE_INT_1U, HWI_COMPUTABLE_MODE_P(), i, IN_RANGE, INTVAL, is_a(), known_eq, may_trap_p(), NULL, NULL_RTX, paradoxical_subreg_p(), reversed_comparison_code(), rtvec_alloc(), RTVEC_ELT, rtx_equal_p(), side_effects_p(), simplify_cond_clz_ctz(), simplify_gen_binary(), simplify_gen_relational(), simplify_gen_ternary(), simplify_merge_mask(), simplify_relational_operation(), simplify_unary_operation(), STORE_FLAG_VALUE, subreg_lowpart_p(), SUBREG_REG, swap_commutative_operands_p(), UINTVAL, VECTOR_MODE_P, XEXP, and XVECEXP.
Referenced by insn_propagation::apply_to_rvalue_1(), simplify_gen_ternary(), and simplify_ternary_operation().
|
privateinherited |
|
privateinherited |
Try to simplify a MODE truncation of OP, which has OP_MODE. Only handle cases where the truncated value is inherently an rvalue. RTL provides two ways of truncating a value: 1. a lowpart subreg. This form is only a truncation when both the outer and inner modes (here MODE and OP_MODE respectively) are scalar integers, and only then when the subreg is used as an rvalue. It is only valid to form such truncating subregs if the truncation requires no action by the target. The onus for proving this is on the creator of the subreg -- e.g. the caller to simplify_subreg or simplify_gen_subreg -- and typically involves either TRULY_NOOP_TRUNCATION_MODES_P or truncated_to_mode. 2. a TRUNCATE. This form handles both scalar and compound integers. The first form is preferred where valid. However, the TRUNCATE handling in simplify_unary_operation turns the second form into the first form when TRULY_NOOP_TRUNCATION_MODES_P or truncated_to_mode allow, so it is generally safe to form rvalue truncations using: simplify_gen_unary (TRUNCATE, ...) and leave simplify_unary_operation to work out which representation should be used. Because of the proof requirements on (1), simplify_truncation must also use simplify_gen_unary (TRUNCATE, ...) to truncate parts of OP, regardless of whether the outer truncation came from a SUBREG or a TRUNCATE. For example, if the caller has proven that an SImode truncation of: (and:DI X Y) is a no-op and can be represented as a subreg, it does not follow that SImode truncations of X and Y are also no-ops. On a target like 64-bit MIPS that requires SImode values to be stored in sign-extended form, an SImode truncation of: (and:DI (reg:DI X) (const_int 63)) is trivially a no-op because only the lower 6 bits can be set. However, X is still an arbitrary 64-bit number and so we cannot assume that truncating it too is a no-op.
References adjust_address_nv, BITS_PER_WORD, CONST_INT_P, constm1_rtx, gcc_assert, GEN_INT, GET_CODE, GET_MODE, GET_MODE_BITSIZE(), GET_MODE_MASK, GET_MODE_PRECISION(), GET_MODE_SIZE(), GET_MODE_UNIT_PRECISION, INTVAL, is_a(), MEM_ADDR_SPACE, MEM_P, MEM_VOLATILE_P, mode_dependent_address_p(), NULL_RTX, REG_P, SCALAR_INT_MODE_P, shift, simplify_gen_binary(), simplify_gen_subreg(), simplify_gen_ternary(), simplify_gen_unary(), subreg_lowpart_offset(), subreg_lowpart_p(), SUBREG_REG, trunc_int_for_mode(), UINTVAL, WORD_REGISTER_OPERATIONS, and XEXP.
Referenced by simplify_subreg(), and simplify_unary_operation_1().
|
inherited |
Try to simplify a unary operation CODE whose output mode is to be MODE with input operand OP whose mode was originally OP_MODE. Return zero if no simplification can be made.
References avoid_constant_pool_reference(), simplify_const_unary_operation(), and simplify_unary_operation_1().
Referenced by insn_propagation::apply_to_rvalue_1(), simplify_binary_operation_1(), simplify_gen_unary(), simplify_ternary_operation(), simplify_unary_operation(), and simplify_unary_operation_1().
|
privateinherited |
Perform some simplifications we can do even if the operands aren't constant.
References ADDR_SPACE_GENERIC, as_a(), COMPARISON_P, const0_rtx, CONST1_RTX, const1_rtx, CONST_DOUBLE_AS_FLOAT_P, CONST_INT_P, CONST_SCALAR_INT_P, CONSTANT_P, CONSTM1_RTX, constm1_rtx, convert_memory_address_addr_space_1(), DECIMAL_FLOAT_MODE_P, exact_int_to_float_conversion_p(), false_rtx, gcc_assert, gen_int_mode(), gen_int_shift_amount(), rtl_hooks::gen_lowpart_no_emit, gen_vec_duplicate(), gen_vec_series(), GET_CODE, GET_MODE, GET_MODE_CLASS, GET_MODE_INNER, GET_MODE_MASK, GET_MODE_NUNITS(), GET_MODE_PRECISION(), GET_MODE_UNIT_PRECISION, GET_MODE_UNIT_SIZE, HONOR_SIGN_DEPENDENT_ROUNDING(), HONOR_SIGNED_ZEROS(), HOST_BITS_PER_WIDE_INT, HWI_COMPUTABLE_MODE_P(), int_mode_for_size(), INTVAL, is_a(), known_eq, MEM_ADDR_SPACE, MEM_VOLATILE_P, mode_dependent_address_p(), mode_signbit_p(), nonzero_bits(), NULL, num_sign_bit_copies(), paradoxical_subreg_p(), partial_subreg_p(), plus_constant(), ptr_mode, REG_P, REG_POINTER, reversed_comparison(), reversed_comparison_code(), rtx_equal_p(), side_effects_p(), simplify_gen_binary(), simplify_gen_relational(), simplify_gen_ternary(), simplify_gen_unary(), simplify_truncation(), simplify_unary_operation(), SRP_SIGNED, SRP_UNSIGNED, STORE_FLAG_VALUE, subreg_lowpart_p(), SUBREG_PROMOTED_SET, SUBREG_PROMOTED_SIGNED_P, SUBREG_PROMOTED_UNSIGNED_P, SUBREG_PROMOTED_VAR_P, SUBREG_REG, target_default_pointer_address_modes_p(), targetm, true_rtx, TRULY_NOOP_TRUNCATION_MODES_P, truncated_to_mode(), val_signbit_known_clear_p(), vec_duplicate_p(), vec_series_p(), VECTOR_MODE_P, and XEXP.
Referenced by simplify_unary_operation().
|
inherited |
Referenced by simplify_associative_operation().
| const char* insn_propagation::failure_reason |
Referenced by apply_to_lvalue_1(), apply_to_rvalue_1(), fixup_debug_use(), and insn_propagation().
|
static |
| rtx insn_propagation::from |
Referenced by apply_to_lvalue_1(), apply_to_rvalue_1(), and insn_propagation().
| rtx_insn* insn_propagation::insn |
Referenced by apply_to_note(), apply_to_rvalue_1(), insn_propagation(), and insn_propagation().
|
staticinherited |
Referenced by simplify_associative_operation().
|
inherited |
| unsigned int insn_propagation::num_replacements |
Referenced by apply_to_rvalue_1(), and insn_propagation().
| uint16_t insn_propagation::result_flags |
Referenced by apply_to_rvalue_1(), and insn_propagation().
| uint16_t insn_propagation::should_check_mems |
Referenced by apply_to_mem_1(), and insn_propagation().
| uint16_t insn_propagation::should_note_simplifications |
Referenced by apply_to_rvalue_1(), and insn_propagation().
| uint16_t insn_propagation::should_unshare |
Referenced by apply_to_rvalue_1(), and insn_propagation().
| uint16_t insn_propagation::spare |
Referenced by insn_propagation().
| rtx insn_propagation::to |
Referenced by apply_to_rvalue_1(), and insn_propagation().
|
static |
Referenced by apply_to_rvalue_1().
1.13.2