[m-rev.] for review: foreign code with determinism failure

[Zoltan Somogyi]

For review by anyone.

compiler/pragma_c_gen.m:
	We have traditionally generated the same code for model_semi
	pragma foreign code, regardless of its actual determinism. This diff
	changes that; for foreign code whos determinism is failure, it does not
	check SUCCESS_INDICATOR, but fails directly instead.

	The objective of this change is to speed up deep profiling, which
	inserts many calls to procedures implemented in foreign code whose
	determinism is failure.

Zoltan.

Index: pragma_c_gen.m
===================================================================
RCS file: /home/mercury1/repository/mercury/compiler/pragma_c_gen.m,v
retrieving revision 1.44
diff -u -b -r1.44 pragma_c_gen.m
--- pragma_c_gen.m	2001/04/03 03:20:14	1.44
+++ pragma_c_gen.m	2001/04/20 04:47:40
@@ -79,6 +79,13 @@
 %
 %	if (!r1) MR_GOTO_LABEL(fail_label);
 %
+% In the case of a pragma c_code with determinism failure, the above
+% is followed by
+%
+%    <code to fail>
+%
+% and the <check of r1> is empty.
+%
 % The code we generate for nondet pragma_c_code assumes that this code is
 % the only thing between the procedure prolog and epilog; such pragma_c_codes
 % therefore cannot be inlined. The code of the procedure is of one of the
@@ -333,9 +340,10 @@
 
 :- pred pragma_c_gen__ordinary_pragma_c_code(code_model::in,
 	pragma_foreign_proc_attributes::in, pred_id::in, proc_id::in,
-	list(prog_var)::in, list(maybe(pair(string, mode)))::in, list(type)::in,
-	string::in, maybe(prog_context)::in, hlds_goal_info::in,
-	code_tree::out, code_info::in, code_info::out) is det.
+	list(prog_var)::in, list(maybe(pair(string, mode)))::in,
+	list(type)::in, string::in, maybe(prog_context)::in,
+	hlds_goal_info::in, code_tree::out, code_info::in, code_info::out)
+	is det.
 
 pragma_c_gen__ordinary_pragma_c_code(CodeModel, Attributes,
 		PredId, ProcId, ArgVars, ArgDatas, OrigArgTypes,
@@ -377,7 +385,11 @@
 		code_info__save_variables(OutArgsSet, _, SaveVarsCode)
 	),
 
+	{ goal_info_get_determinism(GoalInfo, Detism) },
 	( { CodeModel = model_semi } ->
+		% We want to clear r1 even for Detism = failure,
+		% since code with such detism may still assign to
+		% SUCCESS_INDICATOR (i.e. r1).
 		code_info__reserve_r1(ReserveR1_Code)
 	;
 		{ ReserveR1_Code = empty }
@@ -408,6 +420,9 @@
 	% into some other location.
 	%
 	( { CodeModel = model_semi } ->
+		% We want to clear r1 even for Detism = failure,
+		% since code with such detism may still assign to
+		% SUCCESS_INDICATOR (i.e. r1).
 		code_info__clear_r1(ClearR1_Code)
 	;
 		{ ClearR1_Code = empty }
@@ -477,7 +492,7 @@
 	%
 	% <for semidet code, check of r1>
 	%
-	( { CodeModel = model_semi } ->
+	( { CodeModel = model_semi, Detism \= failure } ->
 		code_info__get_next_label(FailLabel),
 		{ CheckR1_Comp = pragma_c_fail_to(FailLabel) },
 		{ MaybeFailLabel = yes(FailLabel) }
@@ -538,6 +553,11 @@
 	%	<code to fail>
 	%	skip_label:
 	%
+	% for code with determinism failure, we need to insert the failure
+	% handling code here:
+	%
+	%	<code to fail>
+	%
 	( { MaybeFailLabel = yes(TheFailLabel) } ->
 		code_info__get_next_label(SkipLabel),
 		code_info__generate_failure(FailCode),
@@ -552,6 +572,8 @@
 			tree(FailCode,
 			     SkipLabelCode)))
 		}
+	; { Detism = failure } ->
+		code_info__generate_failure(FailureCode)
 	;
 		{ FailureCode = empty }
 	),
--------------------------------------------------------------------------
mercury-reviews mailing list
post:  mercury-reviews at cs.mu.oz.au
administrative address: owner-mercury-reviews at cs.mu.oz.au
unsubscribe: Address: mercury-reviews-request at cs.mu.oz.au Message: unsubscribe
subscribe:   Address: mercury-reviews-request at cs.mu.oz.au Message: subscribe
--------------------------------------------------------------------------