http://sourceware.org/ml/gdb-patches/2016-02/msg00843.html Subject: [PATCH v2 1/6] fortran: allow multi-dimensional subarrays From: Christoph Weinmann Add an argument count for subrange expressions in Fortran. Based on the counted value calculate a new array with the elements specified by the user. First parse the user input, secondly copy the desired array values into the return array, thirdly re-create the necessary ranges and bounds. 1| program prog 2| integer :: ary(10,5) = (/ (i,i=1,10) (j, j=1,5) /) 3| end program prog (gdb) print ary(2:4,1:3) old> Syntax error in expression near ':3' new> $3 = ( ( 21, 31, 41) ( 22, 32, 42) ( 23, 33, 43) ) 2013-11-25 Christoph Weinmann * eval.c (multi_f77_subscript): Remove function. * eval.c (evaluate_subrange_expr): When evaluating an array or string expression, call value_f90_subarray. * eval.c (value_f90_subarray): Add argument parsing and compute result array based on user input. * f-exp.y: Increment argument counter for every subrange expression entered by the user. * valops.c (value_slice): Call value_slice_1 with additional default argument. * valops.c (value_slice_1): Add functionality to copy and return result values based on input. * value.h: Add function definition. Signed-off-by: Christoph Weinmann --- gdb/eval.c | 309 ++++++++++++++++++++++++++++++++++++++++++++++------------- gdb/f-exp.y | 2 + gdb/valops.c | 157 ++++++++++++++++++++++++------ gdb/value.h | 2 + 4 files changed, 375 insertions(+), 95 deletions(-) diff --git a/gdb/eval.c b/gdb/eval.c index 78ad946..c9f325f 100644 --- a/gdb/eval.c +++ b/gdb/eval.c @@ -399,29 +399,253 @@ init_array_element (struct value *array, struct value *element, return index; } +/* Evaluates any operation on Fortran arrays or strings with at least + one user provided parameter. Expects the input ARRAY to be either + an array, or a string. Evaluates EXP by incrementing POS, and + writes the content from the elt stack into a local struct. NARGS + specifies number of literal or range arguments the user provided. + NARGS must be the same number as ARRAY has dimensions. */ + static struct value * -value_f90_subarray (struct value *array, - struct expression *exp, int *pos, enum noside noside) +value_f90_subarray (struct value *array, struct expression *exp, + int *pos, int nargs, enum noside noside) { - int pc = (*pos) + 1; + int i, dim_count = 0; LONGEST low_bound, high_bound; struct type *range = check_typedef (TYPE_INDEX_TYPE (value_type (array))); - enum f90_range_type range_type - = (enum f90_range_type) longest_to_int (exp->elts[pc].longconst); - - *pos += 3; + struct value *new_array = array; + struct type *array_type = check_typedef (value_type (new_array)); + struct type *temp_type; + + /* Local struct to hold user data for Fortran subarray dimensions. */ + struct subscript_store + { + /* For every dimension, we are either working on a range or an index + expression, so we store this info separately for later. */ + enum + { + SUBSCRIPT_RANGE, /* e.g. "(lowbound:highbound)" */ + SUBSCRIPT_INDEX /* e.g. "(literal)" */ + } kind; + + /* We also store either the lower and upper bound info, or the index + number. Before evaluation of the input values, we do not know if we are + actually working on a range of ranges, or an index in a range. So as a + first step we store all input in a union. The array calculation itself + deals with this later on. */ + union + { + struct subscript_range + { + enum f90_range_type f90_range_type; + LONGEST low, high; + } + range; + LONGEST number; + }; + } *subscript_array; + + /* Check if the number of arguments provided by the user matches + the number of dimension of the array. A string has only one + dimension. */ + if (nargs != calc_f77_array_dims (value_type (new_array))) + error (_("Wrong number of subscripts")); + + subscript_array = alloca (sizeof (*subscript_array) * nargs); + + /* Parse the user input into the SUBSCRIPT_ARRAY to store it. We need + to evaluate it first, as the input is from left-to-right. The + array is stored from right-to-left. So we have to use the user + input in reverse order. Later on, we need the input information to + re-calculate the output array. For multi-dimensional arrays, we + can be dealing with any possible combination of ranges and indices + for every dimension. */ + for (i = 0; i < nargs; i++) + { + struct subscript_store *index = &subscript_array[i]; - if (range_type == LOW_BOUND_DEFAULT || range_type == BOTH_BOUND_DEFAULT) - low_bound = TYPE_LOW_BOUND (range); - else - low_bound = value_as_long (evaluate_subexp (NULL_TYPE, exp, pos, noside)); + /* The user input is a range, with or without lower and upper bound. + E.g.: "p arry(2:5)", "p arry( :5)", "p arry( : )", etc. */ + if (exp->elts[*pos].opcode == OP_F90_RANGE) + { + int pc = (*pos) + 1; + struct subscript_range *range; + + index->kind = SUBSCRIPT_RANGE; + range = &index->range; + + *pos += 3; + range->f90_range_type = longest_to_int (exp->elts[pc].longconst); + + /* If a lower bound was provided by the user, the bit has been + set and we can assign the value from the elt stack. Same for + upper bound. */ + if ((range->f90_range_type == HIGH_BOUND_DEFAULT) + || range->f90_range_type == NONE_BOUND_DEFAULT) + range->low = value_as_long (evaluate_subexp (NULL_TYPE, exp, + pos, noside)); + if ((range->f90_range_type == LOW_BOUND_DEFAULT) + || range->f90_range_type == NONE_BOUND_DEFAULT) + range->high = value_as_long (evaluate_subexp (NULL_TYPE, exp, + pos, noside)); + } + /* User input is an index. E.g.: "p arry(5)". */ + else + { + struct value *val; - if (range_type == HIGH_BOUND_DEFAULT || range_type == BOTH_BOUND_DEFAULT) - high_bound = TYPE_HIGH_BOUND (range); - else - high_bound = value_as_long (evaluate_subexp (NULL_TYPE, exp, pos, noside)); + index->kind = SUBSCRIPT_INDEX; + + /* Evaluate each subscript; it must be a legal integer in F77. This + ensures the validity of the provided index. */ + val = evaluate_subexp_with_coercion (exp, pos, noside); + index->number = value_as_long (val); + } + + } + + /* Traverse the array from right to left and evaluate each corresponding + user input. VALUE_SUBSCRIPT is called for every index, until a range + expression is evaluated. After a range expression has been evaluated, + every subsequent expression is also treated as a range. */ + for (i = nargs - 1; i >= 0; i--) + { + struct subscript_store *index = &subscript_array[i]; + struct type *index_type = TYPE_INDEX_TYPE (array_type); + + switch (index->kind) + { + case SUBSCRIPT_RANGE: + { + + /* When we hit the first range specified by the user, we must + treat any subsequent user entry as a range. We simply + increment DIM_COUNT which tells us how many times we are + calling VALUE_SLICE_1. */ + struct subscript_range *range = &index->range; + + /* If no lower bound was provided by the user, we take the + default boundary. Same for the high bound. */ + if ((range->f90_range_type == LOW_BOUND_DEFAULT) + || (range->f90_range_type == BOTH_BOUND_DEFAULT)) + range->low = TYPE_LOW_BOUND (index_type); + + if ((range->f90_range_type == HIGH_BOUND_DEFAULT) + || (range->f90_range_type == BOTH_BOUND_DEFAULT)) + range->high = TYPE_HIGH_BOUND (index_type); + + /* Both user provided low and high bound have to be inside the + array bounds. Throw an error if not. */ + if (range->low < TYPE_LOW_BOUND (index_type) + || range->low > TYPE_HIGH_BOUND (index_type) + || range->high < TYPE_LOW_BOUND (index_type) + || range->high > TYPE_HIGH_BOUND (index_type)) + error (_("provided bound(s) outside array bound(s)")); + + /* DIM_COUNT counts every user argument that is treated as a range. + This is necessary for expressions like 'print array(7, 8:9). + Here the first argument is a literal, but must be treated as a + range argument to allow the correct output representation. */ + dim_count++; + + new_array + = value_slice_1 (new_array, + longest_to_int (range->low), + longest_to_int (range->high - range->low + 1), + dim_count); + } + break; + + case SUBSCRIPT_INDEX: + { + /* DIM_COUNT only stays '0' when no range argument was processed + before, starting from the last dimension. This way we can + reduce the number of dimensions from the result array. + However, if a range has been processed before an index, we + treat the index like a range with equal low- and high bounds + to get the value offset right. */ + if (dim_count == 0) + new_array + = value_subscripted_rvalue (new_array, index->number, + f77_get_lowerbound (value_type + (new_array))); + else + { + /* Check for valid index input. */ + if (index->number < TYPE_LOW_BOUND (index_type) + || index->number > TYPE_HIGH_BOUND (index_type)) + error (_("error no such vector element")); + + dim_count++; + new_array = value_slice_1 (new_array, + longest_to_int (index->number), + 1, /* length is '1' element */ + dim_count); + } + + } + break; + } + } + + /* With DIM_COUNT > 1 we currently have a one dimensional array, but expect + an array of arrays, depending on how many ranges have been provided by + the user. So we need to rebuild the array dimensions for printing it + correctly. + Starting from right to left in the user input, after we hit the first + range argument every subsequent argument is also treated as a range. + E.g.: + "p ary(3, 7, 2:15)" in Fortran has only 1 dimension, but we calculated 3 + ranges. + "p ary(3, 7:12, 4)" in Fortran has only 1 dimension, but we calculated 2 + ranges. + "p ary(2:4, 5, 7)" in Fortran has only 1 dimension, and we calculated 1 + range. */ + if (dim_count > 1) + { + struct value *v = NULL; - return value_slice (array, low_bound, high_bound - low_bound + 1); + temp_type = TYPE_TARGET_TYPE (value_type (new_array)); + + /* Every SUBSCRIPT_RANGE in the user input signifies an actual range in + the output array. So we traverse the SUBSCRIPT_ARRAY again, looking + for a range entry. When we find one, we use the range info to create + an additional range_type to set the correct bounds and dimensions for + the output array. */ + for (i = 0; i < nargs; i++) + { + struct subscript_store *index = &subscript_array[i]; + + if (index->kind == SUBSCRIPT_RANGE) + { + struct type *range_type, *interim_array_type; + + range_type + = create_static_range_type (NULL, + temp_type, + 1, + index->range.high - index->range.low + 1); + + interim_array_type = create_array_type (NULL, + temp_type, + range_type); + + /* For some reason the type code of the contents is missing, so + reset it from the original array. */ + TYPE_CODE (interim_array_type) + = TYPE_CODE (value_type (new_array)); + + v = allocate_value (interim_array_type); + + temp_type = value_type (v); + } + + } + value_contents_copy (v, 0, new_array, 0, TYPE_LENGTH (temp_type)); + return v; + } + + return new_array; } @@ -1810,14 +2034,11 @@ evaluate_subexp_standard (struct type *expect_type, switch (code) { case TYPE_CODE_ARRAY: - if (exp->elts[*pos].opcode == OP_F90_RANGE) - return value_f90_subarray (arg1, exp, pos, noside); - else - goto multi_f77_subscript; + return value_f90_subarray (arg1, exp, pos, nargs, noside); case TYPE_CODE_STRING: if (exp->elts[*pos].opcode == OP_F90_RANGE) - return value_f90_subarray (arg1, exp, pos, noside); + return value_f90_subarray (arg1, exp, pos, 1, noside); else { arg2 = evaluate_subexp_with_coercion (exp, pos, noside); @@ -2222,49 +2443,6 @@ evaluate_subexp_standard (struct type *expect_type, } return (arg1); - multi_f77_subscript: - { - LONGEST subscript_array[MAX_FORTRAN_DIMS]; - int ndimensions = 1, i; - struct value *array = arg1; - - if (nargs > MAX_FORTRAN_DIMS) - error (_("Too many subscripts for F77 (%d Max)"), MAX_FORTRAN_DIMS); - - ndimensions = calc_f77_array_dims (type); - - if (nargs != ndimensions) - error (_("Wrong number of subscripts")); - - gdb_assert (nargs > 0); - - /* Now that we know we have a legal array subscript expression - let us actually find out where this element exists in the array. */ - - /* Take array indices left to right. */ - for (i = 0; i < nargs; i++) - { - /* Evaluate each subscript; it must be a legal integer in F77. */ - arg2 = evaluate_subexp_with_coercion (exp, pos, noside); - - /* Fill in the subscript array. */ - - subscript_array[i] = value_as_long (arg2); - } - - /* Internal type of array is arranged right to left. */ - for (i = nargs; i > 0; i--) - { - struct type *array_type = check_typedef (value_type (array)); - LONGEST index = subscript_array[i - 1]; - - array = value_subscripted_rvalue (array, index, - f77_get_lowerbound (array_type)); - } - - return array; - } - case BINOP_LOGICAL_AND: arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside); if (noside == EVAL_SKIP) @@ -3121,6 +3299,9 @@ calc_f77_array_dims (struct type *array_type) int ndimen = 1; struct type *tmp_type; + if (TYPE_CODE (array_type) == TYPE_CODE_STRING) + return 1; + if ((TYPE_CODE (array_type) != TYPE_CODE_ARRAY)) error (_("Can't get dimensions for a non-array type")); diff --git a/gdb/f-exp.y b/gdb/f-exp.y index 4faac32..9343abb 100644 --- a/gdb/f-exp.y +++ b/gdb/f-exp.y @@ -308,6 +308,8 @@ arglist : subrange arglist : arglist ',' exp %prec ABOVE_COMMA { arglist_len++; } + | arglist ',' subrange %prec ABOVE_COMMA + { arglist_len++; } ; /* There are four sorts of subrange types in F90. */ diff --git a/gdb/valops.c b/gdb/valops.c index 5a244a9..09ea877 100644 --- a/gdb/valops.c +++ b/gdb/valops.c @@ -3759,56 +3759,151 @@ value_of_this_silent (const struct language_defn *lang) struct value * value_slice (struct value *array, int lowbound, int length) { + /* Pass unaltered arguments to VALUE_SLICE_1, plus a CALL_COUNT of '1' as we + are only considering the highest dimension, or we are working on a one + dimensional array. So we call VALUE_SLICE_1 exactly once. */ + return value_slice_1 (array, lowbound, length, 1); +} + +/* CALL_COUNT is used to determine if we are calling the function once, e.g. + we are working on the current dimension of ARRAY, or if we are calling + the function repeatedly. In the later case we need to take elements + from the TARGET_TYPE of ARRAY. + With a CALL_COUNT greater than 1 we calculate the offsets for every element + that should be in the result array. Then we fetch the contents and then + copy them into the result array. The result array will have one dimension + less than the input array, so later on we need to recreate the indices and + ranges in the calling function. */ + +struct value * +value_slice_1 (struct value *array, int lowbound, int length, int call_count) +{ struct type *slice_range_type, *slice_type, *range_type; - LONGEST lowerbound, upperbound; - struct value *slice; - struct type *array_type; + struct type *array_type = check_typedef (value_type (array)); + struct type *elt_type = check_typedef (TYPE_TARGET_TYPE (array_type)); + unsigned int elt_size, elt_offs; + LONGEST elt_stride, ary_high_bound, ary_low_bound; + struct value *v; + int slice_range_size, i = 0, row_count = 1, elem_count = 1; - array_type = check_typedef (value_type (array)); + /* Check for legacy code if we are actually dealing with an array or + string. */ if (TYPE_CODE (array_type) != TYPE_CODE_ARRAY && TYPE_CODE (array_type) != TYPE_CODE_STRING) error (_("cannot take slice of non-array")); - range_type = TYPE_INDEX_TYPE (array_type); - if (get_discrete_bounds (range_type, &lowerbound, &upperbound) < 0) - error (_("slice from bad array or bitstring")); + ary_low_bound = TYPE_LOW_BOUND (TYPE_INDEX_TYPE (array_type)); + ary_high_bound = TYPE_HIGH_BOUND (TYPE_INDEX_TYPE (array_type)); + + /* When we are working on a multi-dimensional array, we need to get the + attributes of the underlying type. */ + if (call_count > 1) + { + elt_type = check_typedef (TYPE_TARGET_TYPE (elt_type)); + row_count = TYPE_LENGTH (array_type) + / TYPE_LENGTH (TYPE_TARGET_TYPE (array_type)); + } + + elem_count = length; + elt_size = TYPE_LENGTH (elt_type); + elt_offs = longest_to_int (lowbound - ary_low_bound); + elt_stride = TYPE_LENGTH (TYPE_INDEX_TYPE (array_type)); + + elt_offs *= elt_size; + + /* Check for valid user input. In case of Fortran this was already done + in the calling function. */ + if (call_count == 1 + && (!TYPE_ARRAY_UPPER_BOUND_IS_UNDEFINED (array_type) + && elt_offs >= TYPE_LENGTH (array_type))) + error (_("no such vector element")); - if (lowbound < lowerbound || length < 0 - || lowbound + length - 1 > upperbound) - error (_("slice out of range")); + /* CALL_COUNT is 1 when we are dealing either with the highest dimension + of the array, or a one dimensional array. Set RANGE_TYPE accordingly. + In both cases we calculate how many rows/elements will be in the output + array by setting slice_range_size. */ + if (call_count == 1) + { + range_type = TYPE_INDEX_TYPE (array_type); + slice_range_size = elem_count; + + /* Check if the array bounds are valid. */ + if (get_discrete_bounds (range_type, &ary_low_bound, &ary_high_bound) < 0) + error (_("slice from bad array or bitstring")); + } + /* When CALL_COUNT is greater than 1, we are dealing with an array of arrays. + So we need to get the type below the current one and set the RANGE_TYPE + accordingly. */ + else + { + range_type = TYPE_INDEX_TYPE (TYPE_TARGET_TYPE (array_type)); + slice_range_size = (ary_low_bound + row_count - 1) * (elem_count); + ary_low_bound = TYPE_LOW_BOUND (range_type); + } /* FIXME-type-allocation: need a way to free this type when we are - done with it. */ - slice_range_type = create_static_range_type ((struct type *) NULL, - TYPE_TARGET_TYPE (range_type), - lowbound, - lowbound + length - 1); + done with it. */ + slice_range_type = create_static_range_type (NULL, TYPE_TARGET_TYPE (range_type), + ary_low_bound, slice_range_size); { - struct type *element_type = TYPE_TARGET_TYPE (array_type); - LONGEST offset - = (lowbound - lowerbound) * TYPE_LENGTH (check_typedef (element_type)); + struct type *element_type; + + /* When CALL_COUNT equals 1 we can use the legacy code for subarrays. */ + if (call_count == 1) + { + element_type = TYPE_TARGET_TYPE (array_type); - slice_type = create_array_type ((struct type *) NULL, - element_type, - slice_range_type); - TYPE_CODE (slice_type) = TYPE_CODE (array_type); + slice_type = create_array_type (NULL, element_type, slice_range_type); + + TYPE_CODE (slice_type) = TYPE_CODE (array_type); + + if (VALUE_LVAL (array) == lval_memory && value_lazy (array)) + v = allocate_value_lazy (slice_type); + else + { + v = allocate_value (slice_type); + value_contents_copy (v, + value_embedded_offset (v), + array, + value_embedded_offset (array) + elt_offs, + elt_size * longest_to_int (length)); + } - if (VALUE_LVAL (array) == lval_memory && value_lazy (array)) - slice = allocate_value_lazy (slice_type); + } + /* When CALL_COUNT is larger than 1 we are working on a range of ranges. + So we copy the relevant elements into the new array we return. */ else { - slice = allocate_value (slice_type); - value_contents_copy (slice, 0, array, offset, - type_length_units (slice_type)); + LONGEST dst_offset = 0; + LONGEST src_row_length = TYPE_LENGTH (TYPE_TARGET_TYPE (array_type)); + + element_type = TYPE_TARGET_TYPE (TYPE_TARGET_TYPE (array_type)); + slice_type = create_array_type (NULL, element_type, slice_range_type); + + TYPE_CODE (slice_type) = TYPE_CODE (TYPE_TARGET_TYPE (array_type)); + + v = allocate_value (slice_type); + for (i = 0; i < longest_to_int (row_count); i++) + { + /* Fetches the contents of ARRAY and copies them into V. */ + value_contents_copy (v, + dst_offset, + array, + elt_offs, + elt_size * elem_count); + elt_offs += src_row_length; + dst_offset += elt_size * elem_count; + } } - set_value_component_location (slice, array); - VALUE_FRAME_ID (slice) = VALUE_FRAME_ID (array); - set_value_offset (slice, value_offset (array) + offset); + set_value_component_location (v, array); + VALUE_REGNUM (v) = VALUE_REGNUM (array); + VALUE_FRAME_ID (v) = VALUE_FRAME_ID (array); + set_value_offset (v, value_offset (array) + elt_offs); } - return slice; + return v; } /* Create a value for a FORTRAN complex number. Currently most of the diff --git a/gdb/value.h b/gdb/value.h index 2eac5ef..3400460 100644 --- a/gdb/value.h +++ b/gdb/value.h @@ -1056,6 +1056,8 @@ extern struct value *varying_to_slice (struct value *); extern struct value *value_slice (struct value *, int, int); +extern struct value *value_slice_1 (struct value *, int, int, int); + extern struct value *value_literal_complex (struct value *, struct value *, struct type *); -- 2.5.0