mirror of
https://github.com/Mercury-Language/mercury.git
synced 2025-12-15 13:55:07 +00:00
Estimated hours taken: 400
Branches: main
This diff implements stack slot optimization for the LLDS back end based on
the idea that after a unification such as A = f(B, C, D), saving the
variable A on the stack indirectly also saves the values of B, C and D.
Figuring out what subset of {B,C,D} to access via A and what subset to access
via their own stack slots is a tricky optimization problem. The algorithm we
use to solve it is described in the paper "Using the heap to eliminate stack
accesses" by Zoltan Somogyi and Peter Stuckey, available in ~zs/rep/stackslot.
That paper also describes (and has examples of) the source-to-source
transformation that implements the optimization.
The optimization needs to know what variables are flushed at call sites
and at program points that establish resume points (e.g. entries to
disjunctions and if-then-elses). We already had code to compute this
information in live_vars.m, but this code was being invoked too late.
This diff modifies live_vars.m to allow it to be invoked both by the stack
slot optimization transformation and by the code generator, and allows its
function to be tailored to the requirements of each invocation.
The information computed by live_vars.m is specific to the LLDS back end,
since the MLDS back ends do not (yet) have the same control over stack
frame layout. We therefore store this information in a new back end specific
field in goal_infos. For uniformity, we make all the other existing back end
specific fields in goal_infos, as well as the similarly back end specific
store map field of goal_exprs, subfields of this new field. This happens
to significantly reduce the sizes of goal_infos.
To allow a more meaningful comparison of the gains produced by the new
optimization, do not save any variables across erroneous calls even if
the new optimization is not enabled.
compiler/stack_opt.m:
New module containing the code that performs the transformation
to optimize stack slot usage.
compiler/matching.m:
New module containing an algorithm for maximal matching in bipartite
graphs, specialized for the graphs needed by stack_opt.m.
compiler/mercury_compile.m:
Invoke the new optimization if the options ask for it.
compiler/stack_alloc.m:
New module containing code that is shared between the old,
non-optimizing stack slot allocation system and the new, optimizing
stack slot allocation system, and the code for actually allocating
stack slots in the absence of optimization.
Live_vars.m used to have two tasks: find out what variables need to be
saved on the stack, and allocating those variables to stack slots.
Live_vars.m now does only the first task; stack_alloc.m now does
the second, using code that used to be in live_vars.m.
compiler/trace_params:
Add a new function to test the trace level, which returns yes if we
want to preserve the values of the input headvars.
compiler/notes/compiler_design.html:
Document the new modules (as well as trace_params.m, which wasn't
documented earlier).
compiler/live_vars.m:
Delete the code that is now in stack_alloc.m and graph_colour.m.
Separate out the kinds of stack uses due to nondeterminism: the stack
slots used by nondet calls, and the stack slots used by resumption
points, in order to allow the reuse of stack slots used by resumption
points after execution has left their scope. This should allow the
same stack slots to be used by different variables in the resumption
point at the start of an else branch and nondet calls in the then
branch, since the resumption point of the else branch is not in effect
when the then branch is executed.
If the new option --opt-no-return-calls is set, then say that we do not
need to save any values across erroneous calls.
Use type classes to allow the information generated by this module
to be recorded in the way required by its invoker.
Package up the data structures being passed around readonly into a
single tuple.
compiler/store_alloc.m:
Allow this module to be invoked by stack_opt.m without invoking the
follow_vars transformation, since applying follow_vars before the form
of the HLDS code is otherwise final can be a pessimization.
Make the module_info a part of the record containing the readonly data
passed around during the traversal.
compiler/common.m:
Do not delete or move around unifications created by stack_opt.m.
compiler/call_gen.m:
compiler/code_info.m:
compiler/continuation_info.m:
compiler/var_locn.m:
Allow the code generator to delete its last record of the location
of a value when generating code to make an erroneous call, if the new
--opt-no-return-calls option is set.
compiler/code_gen.m:
Use a more useful algorithm to create the messages/comments that
we put into incr_sp instructions, e.g. by distinguishing between
predicates and functions. This is to allow the new scripts in the
tool directory to gather statistics about the effect of the
optimization on stack frame sizes.
library/exception.m:
Make a hand-written incr_sp follow the new pattern.
compiler/arg_info.m:
Add predicates to figure out the set of input, output and unused
arguments of a procedure in several different circumstances.
Previously, variants of these predicates were repeated in several
places.
compiler/goal_util.m:
Export some previously private utility predicates.
compiler/handle_options.m:
Turn off stack slot optimizations when debugging, unless
--trace-optimized is set.
Add a new dump format useful for debugging --optimize-saved-vars.
compiler/hlds_llds.m:
New module for handling all the stuff specific to the LLDS back end
in HLDS goal_infos.
compiler/hlds_goal.m:
Move all the relevant stuff into the new back end specific field
in goal_infos.
compiler/notes/allocation.html:
Update the documentation of store maps to reflect their movement
into a subfield of goal_infos.
compiler/*.m:
Minor changes to accomodate the placement of all back end specific
information about goals from goal_exprs and individual fields of
goal_infos into a new field in goal_infos that gathers together
all back end specific information.
compiler/use_local_vars.m:
Look for sequences in which several instructions use a fake register
or stack slot as a base register pointing to a cell, and make those
instructions use a local variable instead.
Without this, a key assumption of the stack slot optimization,
that accessing a field in a cell costs only one load or store
instruction, would be much less likely to be true. (With this
optimization, the assumption will be false only if the C compiler's
code generator runs out of registers in a basic block, which for
the code we generate should be unlikely even on x86s.)
compiler/options.m:
Make the old option --optimize-saved-vars ask for both the old stack
slot optimization (implemented by saved_vars.m) that only eliminates
the storing of constants in stack slots, and the new optimization.
Add two new options --optimize-saved-vars-{const,cell} to turn on
the two optimizations separately.
Add a bunch of options to specify the parameters of the new
optimizations, both in stack_opt.m and use_local_vars.m. These are
for implementors only; they are deliberately not documented.
Add a new option, --opt-no-return-cells, that governs whether we avoid
saving variables on the stack at calls that cannot return, either by
succeeding or by failing. This is for implementors only, and thus
deliberately documented only in comments. It is enabled by default.
compiler/optimize.m:
Transmit the value of a new option to use_local_vars.m.
doc/user_guide.texi:
Update the documentation of --optimize-saved-vars.
library/tree234.m:
Undo a previous change of mine that effectively applied this
optimization by hand. That change complicated the code, and now
the compiler can do the optimization automatically.
tools/extract_incr_sp:
A new script for extracting stack frame sizes and messages from
stack increment operations in the C code for LLDS grades.
tools/frame_sizes:
A new script that uses extract_incr_sp to extract information about
stack frame sizes from the C files saved from a stage 2 directory
by makebatch and summarizes the resulting information.
tools/avg_frame_size:
A new script that computes average stack frame sizes from the files
created by frame_sizes.
tools/compare_frame_sizes:
A new script that compares the stack frame size information
extracted from two different stage 2 directories by frame_sizes,
reporting on both average stack frame sizes and on specific procedures
that have different stack frame sizes in the two versions.
600 lines
21 KiB
Mathematica
600 lines
21 KiB
Mathematica
%-----------------------------------------------------------------------------%
|
|
% Copyright (C) 1997-2002 The University of Melbourne.
|
|
% This file may only be copied under the terms of the GNU General
|
|
% Public License - see the file COPYING in the Mercury distribution.
|
|
%-----------------------------------------------------------------------------%
|
|
%
|
|
% term_traversal.m
|
|
%
|
|
% Main author: crs.
|
|
% Significant rewrite by zs.
|
|
%
|
|
% This module contains the code used to traverse procedure bodies
|
|
% for both passes of termination analysis.
|
|
%
|
|
% For details, please refer to the papers mentioned in termination.m.
|
|
%
|
|
%-----------------------------------------------------------------------------%
|
|
|
|
:- module transform_hlds__term_traversal.
|
|
|
|
:- interface.
|
|
|
|
:- import_module transform_hlds__term_util, transform_hlds__term_errors.
|
|
:- import_module hlds__hlds_module, hlds__hlds_pred, hlds__hlds_goal.
|
|
:- import_module parse_tree__prog_data.
|
|
:- import_module list, bag, map, std_util, set.
|
|
|
|
:- type traversal_info
|
|
---> ok(
|
|
set(path_info),
|
|
% Information about the paths we have
|
|
% followed. With a conjunction of
|
|
% length N, each of whose elements is
|
|
% a branched control structure, the
|
|
% number of paths through the
|
|
% conjunction is 2^N. The reason why
|
|
% we use a set of path_infos instead
|
|
% of a list is that this can postpone
|
|
% the representation getting too big
|
|
% if (as is at least moderately likely)
|
|
% many of the paths have identical
|
|
% properties.
|
|
list(term_errors__error)
|
|
% Have we processed a call to a
|
|
% procedure whose maybe termination
|
|
% info was yes(can_loop(_))?
|
|
% If yes, record the error here.
|
|
% (This is not an error in pass 1,
|
|
% but we want to find this out in
|
|
% pass 1 so we can avoid doing pass 2.)
|
|
)
|
|
; error(
|
|
list(term_errors__error),
|
|
% Errors which are fatal in both
|
|
% passes.
|
|
list(term_errors__error)
|
|
% Have we processed a call to a
|
|
% procedure whose maybe termination
|
|
% info was yes(can_loop(_))?
|
|
% If yes, record the error here.
|
|
% (This is not an error in pass 1,
|
|
% but we want to find this out in
|
|
% pass 1 so we can avoid doing pass 2.)
|
|
).
|
|
|
|
:- type path_info
|
|
---> path_info(
|
|
pred_proc_id, % The identify of the procedure
|
|
% that this path is within.
|
|
maybe(pair(pred_proc_id, prog_context)),
|
|
% If no, path was started at the end
|
|
% of the procedure given by field 1.
|
|
% If yes, the arg names the procedure
|
|
% at the call to which the path started
|
|
% and the context of the call.
|
|
% In pass 1, all starts should be no.
|
|
% In pass 2, all starts should be yes.
|
|
int,
|
|
list(pred_proc_id),
|
|
bag(prog_var)
|
|
% These three fields describe the
|
|
% right hand side of the inequation
|
|
% we are propagating.
|
|
).
|
|
|
|
:- type traversal_params.
|
|
|
|
:- pred init_traversal_params(module_info::in, functor_info::in,
|
|
pred_proc_id::in, prog_context::in, map(prog_var, type)::in,
|
|
used_args::in, used_args::in, int::in, int::in,
|
|
traversal_params::out) is det.
|
|
|
|
:- pred traverse_goal(hlds_goal::in, traversal_params::in,
|
|
traversal_info::in, traversal_info::out) is det.
|
|
|
|
:- pred upper_bound_active_vars(list(path_info)::in, bag(prog_var)::out) is det.
|
|
|
|
:- implementation.
|
|
|
|
:- import_module hlds__hlds_data, check_hlds__type_util.
|
|
:- import_module bool, int, require.
|
|
|
|
traverse_goal(Goal, Params, Info0, Info) :-
|
|
Goal = GoalExpr - GoalInfo,
|
|
(
|
|
goal_info_get_determinism(GoalInfo, Detism),
|
|
determinism_components(Detism, _, at_most_zero)
|
|
->
|
|
cannot_succeed(Info0, Info1)
|
|
;
|
|
Info1 = Info0
|
|
),
|
|
traverse_goal_2(GoalExpr, GoalInfo, Params, Info1, Info).
|
|
|
|
:- pred traverse_goal_2(hlds_goal_expr::in, hlds_goal_info::in,
|
|
traversal_params::in, traversal_info::in, traversal_info::out) is det.
|
|
|
|
traverse_goal_2(unify(_Var, _RHS, _UniMode, Unification, _Context),
|
|
_GoalInfo, Params, Info0, Info) :-
|
|
(
|
|
Unification = construct(OutVar, ConsId, Args, Modes, _, _, _),
|
|
(
|
|
unify_change(OutVar, ConsId, Args, Modes, Params,
|
|
Gamma, InVars, OutVars0)
|
|
->
|
|
bag__insert(OutVars0, OutVar, OutVars),
|
|
record_change(InVars, OutVars, Gamma, [], Info0, Info)
|
|
;
|
|
% length(Args) is not necessarily equal to length(Modes)
|
|
% for higher order constructions.
|
|
Info = Info0
|
|
)
|
|
;
|
|
Unification = deconstruct(InVar, ConsId, Args, Modes, _, _),
|
|
(
|
|
unify_change(InVar, ConsId, Args, Modes, Params,
|
|
Gamma0, InVars0, OutVars)
|
|
->
|
|
bag__insert(InVars0, InVar, InVars),
|
|
Gamma is 0 - Gamma0,
|
|
record_change(InVars, OutVars, Gamma, [], Info0, Info)
|
|
;
|
|
error("higher order deconstruction")
|
|
)
|
|
;
|
|
Unification = assign(OutVar, InVar),
|
|
bag__init(Empty),
|
|
bag__insert(Empty, InVar, InVars),
|
|
bag__insert(Empty, OutVar, OutVars),
|
|
record_change(InVars, OutVars, 0, [], Info0, Info)
|
|
;
|
|
Unification = simple_test(_InVar1, _InVar2),
|
|
Info = Info0
|
|
;
|
|
Unification = complicated_unify(_, _, _),
|
|
error("Unexpected complicated_unify in termination analysis")
|
|
).
|
|
|
|
traverse_goal_2(conj(Goals), _, Params, Info0, Info) :-
|
|
list__reverse(Goals, RevGoals),
|
|
traverse_conj(RevGoals, Params, Info0, Info).
|
|
|
|
traverse_goal_2(par_conj(Goals), _, Params, Info0, Info) :-
|
|
list__reverse(Goals, RevGoals),
|
|
traverse_conj(RevGoals, Params, Info0, Info).
|
|
|
|
traverse_goal_2(switch(_, _, Cases), _, Params, Info0, Info) :-
|
|
traverse_switch(Cases, Params, Info0, Info).
|
|
|
|
traverse_goal_2(disj(Goals), _, Params, Info0, Info) :-
|
|
traverse_disj(Goals, Params, Info0, Info).
|
|
|
|
traverse_goal_2(not(Goal), _, Params, Info0, Info) :-
|
|
% Since goal cannot bind any active variables,
|
|
% we don't need to traverse Goal for pass1,
|
|
% but it shouldn't hurt either.
|
|
traverse_goal(Goal, Params, Info0, Info).
|
|
|
|
traverse_goal_2(some(_Vars, _, Goal), _GoalInfo, Params, Info0, Info) :-
|
|
traverse_goal(Goal, Params, Info0, Info).
|
|
|
|
traverse_goal_2(if_then_else(_, Cond, Then, Else), _, Params, Info0, Info) :-
|
|
traverse_conj([Then, Cond], Params, Info0, Info1),
|
|
traverse_goal(Else, Params, Info0, Info2),
|
|
combine_paths(Info1, Info2, Params, Info).
|
|
|
|
traverse_goal_2(foreign_proc(_, CallPredId, CallProcId, Args, _,_,_),
|
|
GoalInfo, Params, Info0, Info) :-
|
|
params_get_module_info(Params, Module),
|
|
module_info_pred_proc_info(Module, CallPredId, CallProcId, _,
|
|
CallProcInfo),
|
|
proc_info_argmodes(CallProcInfo, CallArgModes),
|
|
partition_call_args(Module, CallArgModes, Args, _InVars, OutVars),
|
|
goal_info_get_context(GoalInfo, Context),
|
|
error_if_intersect(OutVars, Context, pragma_foreign_code, Info0, Info).
|
|
|
|
traverse_goal_2(generic_call(_, _, _, _), GoalInfo, Params, Info0, Info) :-
|
|
%
|
|
% For class method calls, we could probably analyse further
|
|
% than this, since we know that the method being called must come
|
|
% from one of the instance declarations, and we could potentially
|
|
% (globally) analyse these.
|
|
%
|
|
% Aditi builtins are not guaranteed to terminate
|
|
% - all of them cause the transaction to abort if an error occurs
|
|
% (e.g. if the database server dies).
|
|
% - all except `aditi_insert' execute a user-specified goal
|
|
% which could possibly loop. Analysis of the termination of
|
|
% goals executed bottom-up is not yet implemented.
|
|
%
|
|
% The error message for `generic_call's other than higher-order calls
|
|
% could be better.
|
|
%
|
|
goal_info_get_context(GoalInfo, Context),
|
|
add_error(Context, horder_call, Params, Info0, Info).
|
|
|
|
traverse_goal_2(call(CallPredId, CallProcId, Args, _, _, _),
|
|
GoalInfo, Params, Info0, Info) :-
|
|
goal_info_get_context(GoalInfo, Context),
|
|
params_get_module_info(Params, Module),
|
|
params_get_ppid(Params, PPId),
|
|
CallPPId = proc(CallPredId, CallProcId),
|
|
|
|
module_info_pred_proc_info(Module, CallPredId, CallProcId, _,
|
|
CallProcInfo),
|
|
proc_info_argmodes(CallProcInfo, CallArgModes),
|
|
proc_info_get_maybe_arg_size_info(CallProcInfo, CallArgSizeInfo),
|
|
proc_info_get_maybe_termination_info(CallProcInfo, CallTerminationInfo),
|
|
|
|
partition_call_args(Module, CallArgModes, Args, InVars, OutVars),
|
|
|
|
% Handle existing paths
|
|
(
|
|
CallArgSizeInfo = yes(finite(CallGamma, OutputSuppliers)),
|
|
remove_unused_args(InVars, Args, OutputSuppliers, UsedInVars),
|
|
record_change(UsedInVars, OutVars, CallGamma, [], Info0, Info1)
|
|
;
|
|
CallArgSizeInfo = yes(infinite(_)),
|
|
error_if_intersect(OutVars, Context,
|
|
inf_termination_const(PPId, CallPPId), Info0, Info1)
|
|
;
|
|
CallArgSizeInfo = no,
|
|
% We should get to this point only in pass 1.
|
|
% In pass 2, OutputSuppliersMap will be empty,
|
|
% which will lead to a runtime abort in map__lookup.
|
|
params_get_output_suppliers(Params, OutputSuppliersMap),
|
|
map__lookup(OutputSuppliersMap, CallPPId, OutputSuppliers),
|
|
remove_unused_args(InVars, Args, OutputSuppliers, UsedInVars),
|
|
record_change(UsedInVars, OutVars, 0, [CallPPId], Info0, Info1)
|
|
),
|
|
|
|
% Did we call a non-terminating procedure?
|
|
(
|
|
CallTerminationInfo = yes(can_loop(_))
|
|
->
|
|
called_can_loop(Context, can_loop_proc_called(PPId, CallPPId),
|
|
Params, Info1, Info2)
|
|
;
|
|
Info2 = Info1
|
|
),
|
|
|
|
% Did we call a procedure with some procedure-valued arguments?
|
|
(
|
|
% This is an overapproximation, since it includes
|
|
% higher order outputs. XXX
|
|
params_get_var_types(Params, VarTypes),
|
|
horder_vars(Args, VarTypes)
|
|
->
|
|
add_error(Context, horder_args(PPId, CallPPId), Params,
|
|
Info2, Info3)
|
|
;
|
|
Info3 = Info2
|
|
),
|
|
|
|
% Do we start another path?
|
|
(
|
|
params_get_rec_input_suppliers(Params, RecInputSuppliersMap),
|
|
map__search(RecInputSuppliersMap, CallPPId, RecInputSuppliers)
|
|
->
|
|
% We should get to this point only in pass 2, and then
|
|
% only if this call is to a procedure in the current SCC.
|
|
% In pass 1, RecInputSuppliersMap will be empty.
|
|
|
|
compute_rec_start_vars(Args, RecInputSuppliers, Bag),
|
|
PathStart = yes(CallPPId - Context),
|
|
NewPath = path_info(PPId, PathStart, 0, [], Bag),
|
|
add_path(NewPath, Info3, Info)
|
|
;
|
|
Info = Info3
|
|
).
|
|
|
|
traverse_goal_2(shorthand(_), _, _, _, _) :-
|
|
% these should have been expanded out by now
|
|
error("traverse_goal_2traverse_goal_2: unexpected shorthand").
|
|
|
|
%-----------------------------------------------------------------------------%
|
|
|
|
% Traverse_conj should be invoked with a reversed list of goals.
|
|
% This is to keep stack consumption down.
|
|
|
|
:- pred traverse_conj(list(hlds_goal)::in, traversal_params::in,
|
|
traversal_info::in, traversal_info::out) is det.
|
|
|
|
traverse_conj([], _, Info, Info).
|
|
traverse_conj([Goal | Goals], Params, Info0, Info) :-
|
|
traverse_goal(Goal, Params, Info0, Info1),
|
|
traverse_conj(Goals, Params, Info1, Info).
|
|
|
|
:- pred traverse_disj(list(hlds_goal)::in, traversal_params::in,
|
|
traversal_info::in, traversal_info::out) is det.
|
|
|
|
traverse_disj([], _, _, ok(Empty, [])) :-
|
|
set__init(Empty).
|
|
traverse_disj([Goal | Goals], Params, Info0, Info) :-
|
|
traverse_goal(Goal, Params, Info0, Info1),
|
|
traverse_disj(Goals, Params, Info0, Info2),
|
|
combine_paths(Info1, Info2, Params, Info).
|
|
|
|
:- pred traverse_switch(list(case)::in, traversal_params::in,
|
|
traversal_info::in, traversal_info::out) is det.
|
|
|
|
traverse_switch([], _, _, ok(Empty, [])) :-
|
|
set__init(Empty).
|
|
traverse_switch([case(_, Goal) | Cases], Params, Info0, Info) :-
|
|
traverse_goal(Goal, Params, Info0, Info1),
|
|
traverse_switch(Cases, Params, Info0, Info2),
|
|
combine_paths(Info1, Info2, Params, Info).
|
|
|
|
%-----------------------------------------------------------------------------%
|
|
|
|
:- pred cannot_succeed(traversal_info::in, traversal_info::out) is det.
|
|
|
|
cannot_succeed(error(Errors, CanLoop), error(Errors, CanLoop)).
|
|
cannot_succeed(ok(_, CanLoop), ok(Empty, CanLoop)) :-
|
|
set__init(Empty).
|
|
|
|
:- pred add_path(path_info::in, traversal_info::in, traversal_info::out) is det.
|
|
|
|
add_path(_, error(Errors, CanLoop), error(Errors, CanLoop)).
|
|
add_path(Path, ok(Paths0, CanLoop), ok(Paths, CanLoop)) :-
|
|
set__insert(Paths0, Path, Paths).
|
|
|
|
:- pred add_error(prog_context::in, termination_error::in,
|
|
traversal_params::in, traversal_info::in, traversal_info::out) is det.
|
|
|
|
add_error(Context, Error, Params, error(Errors0, CanLoop),
|
|
error(Errors, CanLoop)) :-
|
|
Errors1 = [Context - Error | Errors0],
|
|
params_get_max_errors(Params, MaxErrors),
|
|
list__take_upto(MaxErrors, Errors1, Errors).
|
|
add_error(Context, Error, _, ok(_, CanLoop),
|
|
error([Context - Error], CanLoop)).
|
|
|
|
:- pred called_can_loop(prog_context::in, termination_error::in,
|
|
traversal_params::in, traversal_info::in, traversal_info::out) is det.
|
|
|
|
called_can_loop(Context, Error, Params, error(Errors, CanLoop0),
|
|
error(Errors, CanLoop)) :-
|
|
CanLoop1 = [Context - Error | CanLoop0],
|
|
params_get_max_errors(Params, MaxErrors),
|
|
list__take_upto(MaxErrors, CanLoop1, CanLoop).
|
|
called_can_loop(Context, Error, Params, ok(Paths, CanLoop0),
|
|
ok(Paths, CanLoop)) :-
|
|
CanLoop1 = [Context - Error | CanLoop0],
|
|
params_get_max_errors(Params, MaxErrors),
|
|
list__take_upto(MaxErrors, CanLoop1, CanLoop).
|
|
|
|
:- pred combine_paths(traversal_info::in, traversal_info::in,
|
|
traversal_params::in, traversal_info::out) is det.
|
|
|
|
combine_paths(error(Errors1, CanLoop1), error(Errors2, CanLoop2), Params,
|
|
error(Errors, CanLoop)) :-
|
|
params_get_max_errors(Params, MaxErrors),
|
|
list__append(Errors1, Errors2, Errors3),
|
|
list__take_upto(MaxErrors, Errors3, Errors),
|
|
list__append(CanLoop1, CanLoop2, CanLoop3),
|
|
list__take_upto(MaxErrors, CanLoop3, CanLoop).
|
|
combine_paths(error(Errors1, CanLoop1), ok(_, CanLoop2), Params,
|
|
error(Errors1, CanLoop)) :-
|
|
params_get_max_errors(Params, MaxErrors),
|
|
list__append(CanLoop1, CanLoop2, CanLoop3),
|
|
list__take_upto(MaxErrors, CanLoop3, CanLoop).
|
|
combine_paths(ok(_, CanLoop1), error(Errors2, CanLoop2), Params,
|
|
error(Errors2, CanLoop)) :-
|
|
params_get_max_errors(Params, MaxErrors),
|
|
list__append(CanLoop1, CanLoop2, CanLoop3),
|
|
list__take_upto(MaxErrors, CanLoop3, CanLoop).
|
|
combine_paths(ok(Paths1, CanLoop1), ok(Paths2, CanLoop2), Params,
|
|
Info) :-
|
|
params_get_max_errors(Params, MaxErrors),
|
|
list__append(CanLoop1, CanLoop2, CanLoop3),
|
|
list__take_upto(MaxErrors, CanLoop3, CanLoop),
|
|
set__union(Paths2, Paths1, Paths),
|
|
params_get_max_paths(Params, MaxPaths),
|
|
(
|
|
% Don't try to track the state of too many paths;
|
|
% doing so can require too much memory.
|
|
set__count(Paths, Count),
|
|
Count =< MaxPaths
|
|
->
|
|
Info = ok(Paths, CanLoop)
|
|
;
|
|
params_get_context(Params, Context),
|
|
Info = error([Context - too_many_paths], CanLoop)
|
|
).
|
|
|
|
%-----------------------------------------------------------------------------%
|
|
|
|
:- pred compute_rec_start_vars(list(prog_var)::in, list(bool)::in,
|
|
bag(prog_var)::out) is det.
|
|
|
|
compute_rec_start_vars([], [], Out) :-
|
|
bag__init(Out).
|
|
compute_rec_start_vars([_|_], [], _Out) :-
|
|
error("Unmatched vars in compute_rec_start_vars\n").
|
|
compute_rec_start_vars([], [_|_], _Out) :-
|
|
error("Unmatched vars in compute_rec_start_vars\n").
|
|
compute_rec_start_vars([Var | Vars], [RecInputSupplier | RecInputSuppliers],
|
|
Out) :-
|
|
compute_rec_start_vars(Vars, RecInputSuppliers, Out1),
|
|
( RecInputSupplier = yes ->
|
|
bag__insert(Out1, Var, Out)
|
|
;
|
|
Out = Out1
|
|
).
|
|
|
|
%-----------------------------------------------------------------------------%
|
|
|
|
% unify_change is invoked for unifications of the form X = f(Yi),
|
|
% with the first argument giving the identity of X, the second the
|
|
% identity of f, the third and fourth the identity and modes of the Yi.
|
|
% unify_change returns the norm of f and the bags of input and output
|
|
% variables among the Yi. It is up to the caller to look after the
|
|
% sign of the norm of f and after the membership of X in either the
|
|
% input or output bags. The predicate fails if invoked on a higher
|
|
% order unification.
|
|
|
|
:- pred unify_change(prog_var::in, cons_id::in, list(prog_var)::in,
|
|
list(uni_mode)::in, traversal_params::in, int::out, bag(prog_var)::out,
|
|
bag(prog_var)::out) is semidet.
|
|
|
|
unify_change(OutVar, ConsId, Args0, Modes0, Params, Gamma, InVars, OutVars) :-
|
|
params_get_functor_info(Params, FunctorInfo),
|
|
params_get_var_types(Params, VarTypes),
|
|
map__lookup(VarTypes, OutVar, Type),
|
|
\+ type_is_higher_order(Type, _, _, _),
|
|
( type_to_ctor_and_args(Type, TypeCtor, _) ->
|
|
params_get_module_info(Params, Module),
|
|
functor_norm(FunctorInfo, TypeCtor, ConsId, Module,
|
|
Gamma, Args0, Args, Modes0, Modes),
|
|
split_unification_vars(Args, Modes, Module, InVars, OutVars)
|
|
;
|
|
error("variable type in traverse_goal_2")
|
|
).
|
|
|
|
%-----------------------------------------------------------------------------%
|
|
|
|
:- pred record_change(bag(prog_var)::in, bag(prog_var)::in, int::in,
|
|
list(pred_proc_id)::in, traversal_info::in, traversal_info::out) is det.
|
|
|
|
record_change(_, _, _, _, error(Errors, CanLoop), error(Errors, CanLoop)).
|
|
record_change(InVars, OutVars, Gamma, CalledPPIds, ok(Paths0, CanLoop),
|
|
ok(NewPaths, CanLoop)) :-
|
|
set__to_sorted_list(Paths0, PathsList0),
|
|
set__init(NewPaths0),
|
|
record_change_2(PathsList0, InVars, OutVars, Gamma, CalledPPIds,
|
|
NewPaths0, NewPaths).
|
|
|
|
:- pred record_change_2(list(path_info)::in, bag(prog_var)::in,
|
|
bag(prog_var)::in, int::in, list(pred_proc_id)::in,
|
|
set(path_info)::in, set(path_info)::out) is det.
|
|
|
|
record_change_2([], _, _, _, _, PathSet, PathSet).
|
|
record_change_2([Path0 | Paths0], InVars, OutVars, CallGamma, CallPPIds,
|
|
PathSet0, PathSet) :-
|
|
Path0 = path_info(ProcData, Start, Gamma0, PPIds0, Vars0),
|
|
( bag__intersect(OutVars, Vars0) ->
|
|
% The change produces some active variables.
|
|
Gamma is CallGamma + Gamma0,
|
|
list__append(CallPPIds, PPIds0, PPIds),
|
|
bag__subtract(Vars0, OutVars, Vars1),
|
|
bag__union(InVars, Vars1, Vars),
|
|
Path = path_info(ProcData, Start, Gamma, PPIds, Vars)
|
|
;
|
|
% The change produces no active variables.
|
|
Path = Path0
|
|
),
|
|
set__insert(PathSet0, Path, PathSet1),
|
|
record_change_2(Paths0, InVars, OutVars, CallGamma, CallPPIds,
|
|
PathSet1, PathSet).
|
|
|
|
%-----------------------------------------------------------------------------%
|
|
|
|
:- pred error_if_intersect(bag(prog_var)::in, prog_context::in,
|
|
termination_error::in, traversal_info::in, traversal_info::out) is det.
|
|
|
|
error_if_intersect(_, _, _, error(Errors, CanLoop), error(Errors, CanLoop)).
|
|
error_if_intersect(OutVars, Context, ErrorMsg, ok(Paths, CanLoop), Info)
|
|
:-
|
|
(
|
|
set__to_sorted_list(Paths, PathList),
|
|
some_active_vars_in_bag(PathList, OutVars)
|
|
->
|
|
Info = error([Context - ErrorMsg], CanLoop)
|
|
;
|
|
Info = ok(Paths, CanLoop)
|
|
).
|
|
|
|
:- pred some_active_vars_in_bag(list(path_info)::in,
|
|
bag(prog_var)::in) is semidet.
|
|
|
|
some_active_vars_in_bag([Path | Paths], OutVars) :-
|
|
(
|
|
Path = path_info(_, _, _, _, Vars),
|
|
bag__intersect(Vars, OutVars)
|
|
;
|
|
some_active_vars_in_bag(Paths, OutVars)
|
|
).
|
|
|
|
%-----------------------------------------------------------------------------%
|
|
|
|
upper_bound_active_vars([], ActiveVars) :-
|
|
bag__init(ActiveVars).
|
|
upper_bound_active_vars([Path | Paths], ActiveVars) :-
|
|
upper_bound_active_vars(Paths, ActiveVars1),
|
|
Path = path_info(_, _, _, _, ActiveVars2),
|
|
bag__least_upper_bound(ActiveVars1, ActiveVars2, ActiveVars).
|
|
|
|
%-----------------------------------------------------------------------------%
|
|
|
|
:- type traversal_params
|
|
---> traversal_params(
|
|
module_info,
|
|
functor_info,
|
|
pred_proc_id, % The procedure we are tracing through.
|
|
prog_context, % The context of the procedure.
|
|
map(prog_var, type),
|
|
map(pred_proc_id, list(bool)),
|
|
% Output suppliers of each procedure.
|
|
% Empty during pass 2.
|
|
map(pred_proc_id, list(bool)),
|
|
% Rec input suppliers of each procedure.
|
|
% Empty during pass 1.
|
|
int, % Max number of errors to gather.
|
|
int % Max number of paths to analyze.
|
|
).
|
|
|
|
init_traversal_params(ModuleInfo, FunctorInfo, PredProcId, Context, VarTypes,
|
|
OutputSuppliers, RecInputSuppliers, MaxErrors, MaxPaths,
|
|
Params) :-
|
|
Params = traversal_params(ModuleInfo, FunctorInfo, PredProcId, Context,
|
|
VarTypes, OutputSuppliers, RecInputSuppliers,
|
|
MaxErrors, MaxPaths).
|
|
|
|
:- pred params_get_module_info(traversal_params::in, module_info::out)
|
|
is det.
|
|
:- pred params_get_functor_info(traversal_params::in, functor_info::out)
|
|
is det.
|
|
:- pred params_get_ppid(traversal_params::in, pred_proc_id::out)
|
|
is det.
|
|
:- pred params_get_context(traversal_params::in, prog_context::out)
|
|
is det.
|
|
:- pred params_get_var_types(traversal_params::in, map(prog_var, type)::out)
|
|
is det.
|
|
:- pred params_get_output_suppliers(traversal_params::in,
|
|
map(pred_proc_id, list(bool))::out) is det.
|
|
:- pred params_get_rec_input_suppliers(traversal_params::in,
|
|
map(pred_proc_id, list(bool))::out) is det.
|
|
:- pred params_get_max_errors(traversal_params::in, int::out) is det.
|
|
:- pred params_get_max_paths(traversal_params::in, int::out) is det.
|
|
|
|
params_get_module_info(Params, A) :-
|
|
Params = traversal_params(A, _, _, _, _, _, _, _, _).
|
|
|
|
params_get_functor_info(Params, B) :-
|
|
Params = traversal_params(_, B, _, _, _, _, _, _, _).
|
|
|
|
params_get_ppid(Params, C) :-
|
|
Params = traversal_params(_, _, C, _, _, _, _, _, _).
|
|
|
|
params_get_context(Params, D) :-
|
|
Params = traversal_params(_, _, _, D, _, _, _, _, _).
|
|
|
|
params_get_var_types(Params, E) :-
|
|
Params = traversal_params(_, _, _, _, E, _, _, _, _).
|
|
|
|
params_get_output_suppliers(Params, F) :-
|
|
Params = traversal_params(_, _, _, _, _, F, _, _, _).
|
|
|
|
params_get_rec_input_suppliers(Params, G) :-
|
|
Params = traversal_params(_, _, _, _, _, _, G, _, _).
|
|
|
|
params_get_max_errors(Params, H) :-
|
|
Params = traversal_params(_, _, _, _, _, _, _, H, _).
|
|
|
|
params_get_max_paths(Params, I) :-
|
|
Params = traversal_params(_, _, _, _, _, _, _, _, I).
|
|
|
|
%-----------------------------------------------------------------------------%
|