Use GNU make pattern rules to handle multiple file output from bison.

With a pattern rule, the recipe will only be executed once, even when
the rule has multiple targets. Using this to handle the output from
bison is included as an example in the GNU make manual.

This fixes the makefiles so that bison-generated header files will be
regenerated if they are deleted.

.gitattributes

@@ -0,0 +1,2 @@
+# gperf in MSYS chokes on DOS line endings
+*.gperf text eol=lf

BUGS.txt

@@ -116,13 +116,17 @@ programs.
 
 RESEARCHING EXISTING/PAST BUGS, AND FILING REPORTS
 
-The URL <http://sourceforge.net/tracker/?group_id=149850> is the main
-bug tracking system. Once you believe you have found a bug, you may
-browse the bugs database for existing bugs that may be related to
-yours. You might find that your bug has already been fixed in a later
-release or snapshot. If that's the case, then you are set. Also,
-consider if you are reporting a bug or really asking for a new
-feature, and use the appropriate tracker.
+The URL <https://sourceforge.net/p/iverilog/bugs/> is the main
+bug tracking system, although some users have reported bugs at
+<https://github.com/steveicarus/iverilog/issues/>. Once you believe
+you have found a bug, you may browse the bugs database for existing
+bugs that may be related to yours. You might find that your bug has
+already been fixed in a later release or snapshot. If that's the case,
+then you are set. Also, consider if you are reporting a bug or really
+asking for a new feature, and use the appropriate tracker.
+
+ system (although you will also find bug rep
+
 
 The bug database supports basic keyword searches, and you can
 optionally limit your search to active bugs, or fixed bugs. You may
@@ -145,7 +149,7 @@ version from git. Please see the developer documentation for more
 detailed instructions -- <http://iverilog.wikia.com/wiki/>.
 
 When you make a patch, submit it to the "Patches" tracker at
-<http://sourceforge.net/tracker/?group_id=149850>. Patches added to
+<https://sourceforge.net/p/iverilog/patches/>. Patches added to
 the "Patches" tracker enter the developer workflow, are checked,
 applied to the appropriate git branch, and are pushed. Then the
 tracker item is closed.
@@ -158,7 +162,7 @@ clarification before applying it.)
 
 COPYRIGHT ISSUES
 
-Icarus Verilog is Copyright (c) 1998-2008 Stephen Williams except
+Icarus Verilog is Copyright (c) 1998-2018 Stephen Williams except
 where otherwise noted. Minor patches are covered as derivative works
 (or editorial comment or whatever the appropriate legal term is) and
 folded into the rest of ivl. However, if a submission can reasonably

HName.cc

@@ -40,7 +40,7 @@ hname_t::hname_t(perm_string text, int num)
       number_[0] = num;
 }
 
-hname_t::hname_t(perm_string text, vector<int>&nums)
+hname_t::hname_t(perm_string text, const vector<int>&nums)
 : name_(text), number_(nums)
 {
 }

HName.h

@@ -42,7 +42,7 @@ class hname_t {
       hname_t ();
       explicit hname_t (perm_string text);
       explicit hname_t (perm_string text, int num);
-      explicit hname_t (perm_string text, std::vector<int>&nums);
+      explicit hname_t (perm_string text, const std::vector<int>&nums);
       hname_t (const hname_t&that);
       ~hname_t();
 
@@ -56,12 +56,12 @@ class hname_t {
 
       size_t has_numbers() const;
       int peek_number(size_t idx) const;
+      const std::vector<int>&peek_numbers() const;
 
     private:
       perm_string name_;
-	// If the number is anything other than INT_MIN, then this is
-	// the numeric part of the name. Otherwise, it is not part of
-	// the name at all.
+	// If this vector has size, then the numbers all together make
+	// up part of the hierarchical name.
       std::vector<int> number_;
 
     private: // not implemented
@@ -82,6 +82,11 @@ inline int hname_t::peek_number(size_t idx) const
       return number_[idx];
 }
 
+inline const std::vector<int>& hname_t::peek_numbers(void) const
+{
+      return number_;
+}
+
 inline size_t hname_t::has_numbers() const
 {
       return number_.size();

Makefile.in

@@ -118,8 +118,8 @@ O = main.o async.o design_dump.o discipline.o dup_expr.o elaborate.o \
     pform_class_type.o pform_string_type.o pform_struct_type.o pform_types.o \
     symbol_search.o sync.o sys_funcs.o verinum.o verireal.o target.o \
     Attrib.o HName.o Module.o PClass.o PDelays.o PEvent.o PExpr.o PGate.o \
-    PGenerate.o PPackage.o PScope.o PSpec.o PTask.o PUdp.o PFunction.o PWire.o \
-    Statement.o AStatement.o $M $(FF) $(TT)
+    PGenerate.o PModport.o PPackage.o PScope.o PSpec.o PTask.o PUdp.o \
+    PFunction.o PWire.o Statement.o AStatement.o $M $(FF) $(TT)
 
 all: dep config.h _pli_types.h version_tag.h ivl@EXEEXT@ version.exe iverilog-vpi.man
 	$(foreach dir,$(SUBDIRS),$(MAKE) -C $(dir) $@ && ) true
@@ -252,11 +252,9 @@ lexor.o: lexor.cc parse.h
 
 parse.o: parse.cc
 
-# Build this in two steps to avoid parallel build issues (see pr3462585)
-parse.cc: $(srcdir)/parse.y
-	$(YACC) --verbose -t -p VL -d -o $@ $<
-parse.h: parse.cc
-	mv parse.cc.h $@ 2>/dev/null || mv parse.hh $@
+# Use pattern rules to avoid parallel build issues (see pr3462585)
+parse%cc parse%h: $(srcdir)/parse%y
+	$(YACC) --verbose -t -p VL --defines=parse.h -o parse.cc $<
 
 syn-rules.cc: $(srcdir)/syn-rules.y
 	$(YACC) --verbose -t -p syn_ -o $@ $<
@@ -292,7 +290,7 @@ endif
 version_tag.h version:
 	@if $(GIT_PRESENT) && test -d $(srcdir)/.git; then \
 	    echo "Using git-describe for VERSION_TAG"; \
-	    tmp=`$(GIT) --git-dir $(srcdir)/.git describe --always --dirty \
+	    tmp=`(cd $(srcdir) && $(GIT) describe --always --dirty) \
 		   | sed -e 's;\(.*\);#define VERSION_TAG "\1";'`; \
 	    echo "$$tmp" | diff - version_tag.h > /dev/null 2>&1 || echo "$$tmp" > version_tag.h || exit 1; \
 	elif test -r $(srcdir)/version_tag.h; then \

Module.cc

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 1998-2010 Stephen Williams ([email protected])
+ * Copyright (c) 1998-2016 Stephen Williams ([email protected])
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -35,9 +35,6 @@ Module::Module(LexicalScope*parent, perm_string n)
       program_block = false;
       uc_drive = UCD_NONE;
       timescale_warn_done = false;
-      time_unit = 0;
-      time_precision = 0;
-      time_from_timescale = false;
 }
 
 Module::~Module()
@@ -94,13 +91,15 @@ perm_string Module::get_port_name(unsigned idx) const
 {
 
       assert(idx < ports.size());
-      if (ports[idx] == 0) {
+      if (ports[idx] == 0 || ports[idx]->name.str() == 0) {
               /* It is possible to have undeclared ports. These
                  are ports that are skipped in the declaration,
                  for example like so: module foo(x ,, y); The
                  port between x and y is unnamed and thus
-                 inaccessible to binding by name. */
-            return perm_string::literal("");
+                 inaccessible to binding by name. Port references
+		 that aren't simple or escaped identifiers are
+		 also inaccessible to binding by name. */
+            return perm_string::literal("unnamed");
       }
       return ports[idx]->name;
 }

Module.h

@@ -1,7 +1,7 @@
 #ifndef IVL_Module_H
 #define IVL_Module_H
 /*
- * Copyright (c) 1998-2014 Stephen Williams ([email protected])
+ * Copyright (c) 1998-2016 Stephen Williams ([email protected])
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -35,6 +35,7 @@ class PExpr;
 class PEIdent;
 class PGate;
 class PGenerate;
+class PModport;
 class PSpecPath;
 class PTask;
 class PFunction;
@@ -47,6 +48,10 @@ class NetScope;
  * A module is a named container and scope. A module holds a bunch of
  * semantic quantities such as wires and gates. The module is
  * therefore the handle for grasping the described circuit.
+ *
+ * SystemVerilog introduces program blocks and interfaces. These have
+ * much in common with modules, so the Module class is used to represent
+ * these containers as well.
  */
 
 class Module : public PScopeExtra, public LineInfo {
@@ -64,7 +69,7 @@ class Module : public PScopeExtra, public LineInfo {
 
     public:
 	/* The name passed here is the module name, not the instance
-	   name. This make must be a permallocated string. */
+	   name. This name must be a permallocated string. */
       explicit Module(LexicalScope*parent, perm_string name);
       ~Module();
 
@@ -81,6 +86,11 @@ class Module : public PScopeExtra, public LineInfo {
 	   restrictions and slightly modify scheduling semantics. */
       bool program_block;
 
+	/* This is true if the module represents a interface
+	   instead of a module/cell. Interfaces have different
+	   content restrictions and some extra allowed items. */
+      bool is_interface;
+
       enum UCDriveType { UCD_NONE, UCD_PULL0, UCD_PULL1 };
       UCDriveType uc_drive;
 
@@ -111,10 +121,6 @@ class Module : public PScopeExtra, public LineInfo {
 
       map<perm_string,PExpr*> attributes;
 
-	/* These are the timescale for this module. The default is
-	   set by the `timescale directive. */
-      int time_unit, time_precision;
-      bool time_from_timescale;
       bool timescale_warn_done;
 
 	/* The module has a list of generate schemes that appear in
@@ -125,6 +131,11 @@ class Module : public PScopeExtra, public LineInfo {
 	   unless they are instantiated, implicitly or explicitly. */
       std::map<perm_string,Module*> nested_modules;
 
+	/* An interface can contain one or more named modport lists.
+           The parser will ensure these don't appear in modules or
+           program blocks. */
+      map<perm_string,PModport*> modports;
+
       list<PSpecPath*> specify_paths;
 
 	// The mod_name() is the name of the module type.

PDelays.cc

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 1999-2011 Stephen Williams ([email protected])
+ * Copyright (c) 1999-2017 Stephen Williams ([email protected])
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -26,11 +26,6 @@
 # include  "verinum.h"
 # include  "netmisc.h"
 
-bool dly_used_no_timescale = false;
-bool dly_used_timescale = false;
-bool display_ts_dly_warning = true;
-
-
 PDelays::PDelays()
 {
       delete_flag_ = true;
@@ -80,19 +75,7 @@ static NetExpr*calculate_val(Design*des, NetScope*scope, PExpr*expr)
 {
       NetExpr*dex = elab_and_eval(des, scope, expr, -1);
 
-	/* Print a warning if we find default and `timescale based
-	 * delays in the design, since this is likely an error. */
-      if (scope->time_from_timescale()) dly_used_timescale = true;
-      else dly_used_no_timescale = true;
-
-      if (display_ts_dly_warning &&
-          dly_used_no_timescale && dly_used_timescale) {
-	    cerr << "warning: Found both default and "
-	            "`timescale based delays. Use" << endl;
-	    cerr << "         -Wtimescale to find the "
-	            "module(s) with no `timescale." << endl;
-	    display_ts_dly_warning = false;
-      }
+      check_for_inconsistent_delays(scope);
 
 	/* If the delay expression is a real constant or vector
 	   constant, then evaluate it, scale it to the local time

PExpr.cc

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 1998-2012 Stephen Williams <[email protected]>
+ * Copyright (c) 1998-2016 Stephen Williams <[email protected]>
  * Copyright CERN 2013 / Stephen Williams ([email protected])
  *
  *    This source code is free software; you can redistribute it
@@ -126,9 +126,8 @@ PEBinary::~PEBinary()
 
 void PEBinary::declare_implicit_nets(LexicalScope*scope, NetNet::Type type)
 {
-      assert(left_ && right_);
-      left_->declare_implicit_nets(scope, type);
-      right_->declare_implicit_nets(scope, type);
+      if (left_) left_->declare_implicit_nets(scope, type);
+      if (right_) right_->declare_implicit_nets(scope, type);
 }
 
 bool PEBinary::has_aa_term(Design*des, NetScope*scope) const
@@ -137,7 +136,7 @@ bool PEBinary::has_aa_term(Design*des, NetScope*scope) const
       return left_->has_aa_term(des, scope) || right_->has_aa_term(des, scope);
 }
 
-PECastSize::PECastSize(unsigned si, PExpr*b)
+PECastSize::PECastSize(PExpr*si, PExpr*b)
 : size_(si), base_(b)
 {
 }

PExpr.h

@@ -1,7 +1,7 @@
 #ifndef IVL_PExpr_H
 #define IVL_PExpr_H
 /*
- * Copyright (c) 1998-2014 Stephen Williams <[email protected]>
+ * Copyright (c) 1998-2016 Stephen Williams <[email protected]>
  * Copyright CERN 2013 / Stephen Williams ([email protected])
  *
  *    This source code is free software; you can redistribute it
@@ -436,7 +436,7 @@ class PEIdent : public PExpr {
 						   const perm_string&) const;
       bool elaborate_lval_net_packed_member_(Design*, NetScope*,
 					     NetAssign_*,
-					     const perm_string&) const;
+					     const name_component_t&) const;
       bool elaborate_lval_darray_bit_(Design*, NetScope*,
 				       NetAssign_*) const;
 
@@ -917,8 +917,10 @@ class PECallFunction : public PExpr {
       virtual bool has_aa_term(Design*des, NetScope*scope) const;
 
       virtual NetExpr*elaborate_expr(Design*des, NetScope*scope,
-				     unsigned expr_wid,
-                                     unsigned flags) const;
+				     ivl_type_t type, unsigned flags) const;
+
+      virtual NetExpr*elaborate_expr(Design*des, NetScope*scope,
+				     unsigned expr_wid, unsigned flags) const;
 
       virtual unsigned test_width(Design*des, NetScope*scope,
 				  width_mode_t&mode);
@@ -969,7 +971,7 @@ class PECallFunction : public PExpr {
 class PECastSize  : public PExpr {
 
     public:
-      explicit PECastSize(unsigned expr_wid, PExpr*base);
+      explicit PECastSize(PExpr*size, PExpr*base);
       ~PECastSize();
 
       void dump(ostream &out) const;
@@ -982,7 +984,7 @@ class PECastSize  : public PExpr {
 				  width_mode_t&mode);
 
     private:
-      unsigned size_;
+      PExpr* size_;
       PExpr* base_;
 };
 
@@ -997,6 +999,9 @@ class PECastType  : public PExpr {
 
       void dump(ostream &out) const;
 
+      virtual NetExpr*elaborate_expr(Design*des, NetScope*scope,
+				     ivl_type_t type, unsigned flags) const;
+
       virtual NetExpr*elaborate_expr(Design*des, NetScope*scope,
 				     unsigned expr_wid, unsigned flags) const;
 

PModport.cc

@@ -0,0 +1,31 @@
+/*
+ * Copyright (c) 2015 Stephen Williams ([email protected])
+ *
+ *    This source code is free software; you can redistribute it
+ *    and/or modify it in source code form under the terms of the GNU
+ *    General Public License as published by the Free Software
+ *    Foundation; either version 2 of the License, or (at your option)
+ *    any later version.
+ *
+ *    This program is distributed in the hope that it will be useful,
+ *    but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *    GNU General Public License for more details.
+ *
+ *    You should have received a copy of the GNU General Public License
+ *    along with this program; if not, write to the Free Software
+ *    Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
+ */
+
+# include "config.h"
+
+# include  "PModport.h"
+
+PModport::PModport(perm_string n)
+: name_(n)
+{
+}
+
+PModport::~PModport()
+{
+}

PModport.h

@@ -0,0 +1,52 @@
+#ifndef IVL_PModport_H
+#define IVL_PModport_H
+/*
+ * Copyright (c) 2015 Stephen Williams ([email protected])
+ *
+ *    This source code is free software; you can redistribute it
+ *    and/or modify it in source code form under the terms of the GNU
+ *    General Public License as published by the Free Software
+ *    Foundation; either version 2 of the License, or (at your option)
+ *    any later version.
+ *
+ *    This program is distributed in the hope that it will be useful,
+ *    but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *    GNU General Public License for more details.
+ *
+ *    You should have received a copy of the GNU General Public License
+ *    along with this program; if not, write to the Free Software
+ *    Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
+ */
+
+# include  "LineInfo.h"
+# include  "PScope.h"
+# include  "StringHeap.h"
+# include  "netlist.h"
+# include  <vector>
+
+/*
+ * The PModport class represents a parsed SystemVerilog modport list.
+ */
+class PModport : public LineInfo {
+
+    public:
+	// The name is a perm-allocated string. It is the simple name
+	// of the modport, without any scope.
+      explicit PModport(perm_string name);
+      ~PModport();
+
+      perm_string name() const { return name_; }
+
+      typedef pair <NetNet::PortType,PExpr*> simple_port_t;
+      map<perm_string,simple_port_t> simple_ports;
+
+    private:
+      perm_string name_;
+
+    private: // not implemented
+      PModport(const PModport&);
+      PModport& operator= (const PModport&);
+};
+
+#endif /* IVL_PModport_H */

PScope.cc

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2008,2010 Stephen Williams ([email protected])
+ * Copyright (c) 2008,2016 Stephen Williams ([email protected])
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -19,21 +19,9 @@
 
 # include  "PScope.h"
 
-PScope::PScope(perm_string n, LexicalScope*parent)
-: LexicalScope(parent), name_(n)
+bool LexicalScope::var_init_needs_explicit_lifetime() const
 {
-}
-
-PScope::PScope(perm_string n)
-: LexicalScope(0), name_(n)
-{
-}
-
-PScope::~PScope()
-{
-    for(map<perm_string, data_type_t*>::iterator it = typedefs.begin();
-        it != typedefs.end(); ++it)
-        delete it->second;
+      return false;
 }
 
 PWire* LexicalScope::wires_find(perm_string name)
@@ -45,17 +33,26 @@ PWire* LexicalScope::wires_find(perm_string name)
 	    return (*cur).second;
 }
 
-PScopeExtra::PScopeExtra(perm_string n, LexicalScope*parent)
-: PScope(n, parent)
+PScope::PScope(perm_string n, LexicalScope*parent)
+: LexicalScope(parent), name_(n)
 {
+      time_unit = 0;
+      time_precision = 0;
+      time_from_timescale = false;
 }
 
-PScopeExtra::PScopeExtra(perm_string n)
-: PScope(n)
+PScope::~PScope()
+{
+    for(map<perm_string, data_type_t*>::iterator it = typedefs.begin();
+        it != typedefs.end(); ++it)
+        delete it->second;
+}
+
+PScopeExtra::PScopeExtra(perm_string n, LexicalScope*parent)
+: PScope(n, parent)
 {
 }
 
 PScopeExtra::~PScopeExtra()
 {
 }
-

PScope.h

@@ -1,7 +1,7 @@
 #ifndef IVL_PScope_H
 #define IVL_PScope_H
 /*
- * Copyright (c) 2008-2014 Stephen Williams ([email protected])
+ * Copyright (c) 2008-2016 Stephen Williams ([email protected])
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -36,6 +36,7 @@ class PProcess;
 class PClass;
 class PTask;
 class PWire;
+class Statement;
 
 class Design;
 class NetScope;
@@ -52,10 +53,14 @@ class NetScope;
 class LexicalScope {
 
     public:
-      explicit LexicalScope(LexicalScope*parent) : parent_(parent) { }
+      enum lifetime_t { INHERITED, STATIC, AUTOMATIC };
+
+      explicit LexicalScope(LexicalScope*parent) : default_lifetime(INHERITED), parent_(parent) { }
 	// A virtual destructor is so that dynamic_cast can work.
       virtual ~LexicalScope() { }
 
+      lifetime_t default_lifetime;
+
       struct range_t {
 	      // True if this is an exclude
 	    bool exclude_flag;
@@ -108,6 +113,9 @@ class LexicalScope {
         // creating implicit nets.
       map<perm_string,LineInfo*> genvars;
 
+	// Variable initializations in this scope
+      vector<Statement*> var_inits;
+
 	// Behaviors (processes) in this scope
       list<PProcess*> behaviors;
       list<AProcess*> analog_behaviors;
@@ -117,6 +125,8 @@ class LexicalScope {
 
       LexicalScope* parent_scope() const { return parent_; }
 
+      virtual bool var_init_needs_explicit_lifetime() const;
+
     protected:
       void dump_typedefs_(ostream&out, unsigned indent) const;
 
@@ -130,6 +140,10 @@ class LexicalScope {
 
       void dump_wires_(ostream&out, unsigned indent) const;
 
+      void dump_var_inits_(ostream&out, unsigned indent) const;
+
+      bool elaborate_var_inits_(Design*des, NetScope*scope) const;
+
     private:
       LexicalScope*parent_;
 };
@@ -145,12 +159,17 @@ class PScope : public LexicalScope {
 	// modules do not nest in Verilog, the parent must be nil for
 	// modules. Scopes for tasks and functions point to their
 	// containing module.
-      PScope(perm_string name, LexicalScope*parent);
-      PScope(perm_string name);
+      explicit PScope(perm_string name, LexicalScope*parent =0);
       virtual ~PScope();
 
       perm_string pscope_name() const { return name_; }
 
+	/* These are the timescale for this scope. The default is
+	   set by the `timescale directive or, in SystemVerilog,
+	   by timeunit and timeprecision statements. */
+      int time_unit, time_precision;
+      bool time_from_timescale;
+
     protected:
       bool elaborate_sig_wires_(Design*des, NetScope*scope) const;
 
@@ -168,14 +187,13 @@ class PScope : public LexicalScope {
 class PScopeExtra : public PScope {
 
     public:
-      PScopeExtra(perm_string, LexicalScope*parent);
-      PScopeExtra(perm_string);
+      explicit PScopeExtra(perm_string, LexicalScope*parent =0);
       ~PScopeExtra();
 
 	/* Task definitions within this module */
       std::map<perm_string,PTask*> tasks;
       std::map<perm_string,PFunction*> funcs;
-	/* class definitions within this module. */
+	/* Class definitions within this module. */
       std::map<perm_string,PClass*> classes;
 	/* This is the lexical order of the classes, and is used by
 	   elaboration to choose an elaboration order. */

PTask.cc

@@ -30,6 +30,11 @@ PTaskFunc::~PTaskFunc()
 {
 }
 
+bool PTaskFunc::var_init_needs_explicit_lifetime() const
+{
+      return default_lifetime == STATIC;
+}
+
 void PTaskFunc::set_ports(vector<pform_tf_port_t>*p)
 {
       assert(ports_ == 0);
@@ -41,7 +46,7 @@ void PTaskFunc::set_this(class_type_t*type, PWire*this_wire)
       assert(this_type_ == 0);
       this_type_ = type;
 
-	// Push a synthethis argument that is the "this" value.
+	// Push a synthesis argument that is the "this" value.
       if (ports_==0)
 	    ports_ = new vector<pform_tf_port_t>;
 

PTask.h

@@ -41,6 +41,8 @@ class PTaskFunc : public PScope, public LineInfo {
       PTaskFunc(perm_string name, LexicalScope*parent);
       ~PTaskFunc();
 
+      bool var_init_needs_explicit_lifetime() const;
+
       void set_ports(std::vector<pform_tf_port_t>*p);
 
       void set_this(class_type_t*use_type, PWire*this_wire);

README.txt

@@ -1,5 +1,5 @@
 		THE ICARUS VERILOG COMPILATION SYSTEM
-		Copyright 2000-2004 Stephen Williams
+		Copyright 2000-2019 Stephen Williams
 
 
 1.0 What is ICARUS Verilog?
@@ -46,8 +46,10 @@ on a UNIX-like system:
 	  to work. MSVC++ 5 and 6 are known to definitely *not* work.
 
 	- bison and flex
+          It has been reported that bison 2.3 on MacOS generates broken
+          code, but bison 3.0.4 works.
 
-	- gperf 2.7
+	- gperf 3.0 or later
 	  The lexical analyzer doesn't recognize keywords directly,
 	  but instead matches symbols and looks them up in a hash
 	  table in order to get the proper lexical code. The gperf
@@ -56,7 +58,7 @@ on a UNIX-like system:
 	  A version problem with this program is the most common cause
 	  of difficulty. See the Icarus Verilog FAQ.
 
-	- readline 4.2
+	- readline 4.2 or later
 	  On Linux systems, this usually means the readline-devel
 	  rpm. In any case, it is the development headers of readline
 	  that are needed.
@@ -67,7 +69,7 @@ on a UNIX-like system:
 If you are building from git, you will also need software to generate
 the configure scripts.
 
-	- autoconf 2.53
+	- autoconf 2.53 or later
 	  This generates configure scripts from configure.in. The 2.53
 	  or later versions are known to work, autoconf 2.13 is
 	  reported to *not* work.

Statement.cc

@@ -115,6 +115,11 @@ PBlock::~PBlock()
 	    delete list_[idx];
 }
 
+bool PBlock::var_init_needs_explicit_lifetime() const
+{
+      return default_lifetime == STATIC;
+}
+
 PChainConstructor* PBlock::extract_chain_constructor()
 {
       if (list_.empty())

Statement.h

@@ -1,7 +1,7 @@
 #ifndef IVL_Statement_H
 #define IVL_Statement_H
 /*
- * Copyright (c) 1998-2014 Stephen Williams ([email protected])
+ * Copyright (c) 1998-2020 Stephen Williams ([email protected])
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -108,7 +108,8 @@ class PAssign_  : public Statement {
       NetAssign_* elaborate_lval(Design*, NetScope*scope) const;
       NetExpr* elaborate_rval_(Design*, NetScope*, ivl_type_t lv_net_type,
 			       ivl_variable_type_t lv_type,
-			       unsigned lv_width) const;
+			       unsigned lv_width,
+			       bool force_unsigned =false) const;
       NetExpr* elaborate_rval_(Design*, NetScope*, ivl_type_t ntype) const;
 
       NetExpr* elaborate_rval_obj_(Design*, NetScope*,
@@ -182,6 +183,8 @@ class PBlock  : public PScope, public Statement {
 
       BL_TYPE bl_type() const { return bl_type_; }
 
+      bool var_init_needs_explicit_lifetime() const;
+
 	// This is only used if this block is the statement list for a
 	// constructor. We look for a PChainConstructor as the first
 	// statement, and if it is there, extract it.
@@ -234,6 +237,9 @@ class PCallTask  : public Statement {
 					 NetNet*net,
 					 perm_string method_name,
 					 const char*sys_task_name) const;
+      NetProc*elaborate_queue_method_(Design*des, NetScope*scope,
+				      NetNet*net,
+				      const char*sys_task_name) const;
       bool test_task_calls_ok_(Design*des, NetScope*scope) const;
 
       PPackage*package_;

aclocal.m4

@@ -223,8 +223,9 @@ AC_SUBST(strip_dynamic)
 # -------------
 AC_DEFUN([AX_C99_STRTOD],
 [# On MinGW we need to jump through hoops to get a C99 compliant strtod().
+ # mingw-w64 doesn't need this, and the 64-bit version doesn't support it.
 case "${host}" in
-    *-*-mingw*)
+    *-pc-mingw32)
         LDFLAGS+=" -Wl,--undefined=___strtod,--wrap,strtod,--defsym,___wrap_strtod=___strtod"
         ;;
 esac

compiler.h

@@ -1,7 +1,7 @@
 #ifndef IVL_compiler_H
 #define IVL_compiler_H
 /*
- * Copyright (c) 1999-2014 Stephen Williams ([email protected])
+ * Copyright (c) 1999-2016 Stephen Williams ([email protected])
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -80,6 +80,10 @@ extern unsigned recursive_mod_limit;
 /* Implicit definitions of wires. */
 extern bool warn_implicit;
 
+/* Warn if dimensions of port or var/net are implicitly taken from
+   the input/output/inout declaration. */
+extern bool warn_implicit_dimensions;
+
 /* inherit timescales across files. */
 extern bool warn_timescale;
 
@@ -138,7 +142,8 @@ extern int build_library_index(const char*path, bool key_case_sensitive);
 
 /* This is the generation of Verilog that the compiler is asked to
    support. Then there are also more detailed controls for more
-   specific language features. */
+   specific language features. Note that the compiler often assumes
+   this is an ordered list. */
 enum generation_t {
       GN_VER1995  = 1,
       GN_VER2001_NOCONFIG  = 2,
@@ -182,25 +187,24 @@ extern bool gn_strict_ca_eval_flag;
    standard expression width rules. */
 extern bool gn_strict_expr_width_flag;
 
-/* If variables can be converted to uwires by a continuous assignment
-   (assuming no procedural assign, then return true. This will be true
-   for SystemVerilog */
-static inline bool gn_var_can_be_uwire(void)
+/* If this flag is true, then don't add a for-loop control variable
+   to an implicit event_expression list if it is only used inside the
+   loop. */
+extern bool gn_shared_loop_index_flag;
+
+static inline bool gn_system_verilog(void)
 {
-      if (generation_flag == GN_VER2005_SV ||
-          generation_flag == GN_VER2009  ||
-          generation_flag == GN_VER2012)
+      if (generation_flag >= GN_VER2005_SV)
 	    return true;
       return false;
 }
 
-static inline bool gn_system_verilog(void)
+/* If variables can be converted to uwires by a continuous assignment
+   (assuming no procedural assign), then return true. This will be true
+   for SystemVerilog */
+static inline bool gn_var_can_be_uwire(void)
 {
-      if (generation_flag == GN_VER2005_SV ||
-          generation_flag == GN_VER2009  ||
-          generation_flag == GN_VER2012)
-	    return true;
-      return false;
+      return gn_system_verilog();
 }
 
 static inline bool gn_modules_nest(void)

config.h.in

@@ -1,7 +1,7 @@
 #ifndef IVL_config_H                                           /* -*- c++ -*- */
 #define IVL_config_H
 /*
- * Copyright (c) 2001-2014 Stephen Williams ([email protected])
+ * Copyright (c) 2001-2015 Stephen Williams ([email protected])
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -55,6 +55,7 @@
 # undef HAVE_FSEEKO
 /* And this is needed by the fst files (copied from GTKWave). */
 # undef HAVE_LIBPTHREAD
+# undef HAVE_REALPATH
 
 /*
  * Define this if you want to compile vvp with memory freeing and

config.sub

@@ -1,44 +1,40 @@
 #! /bin/sh
 # Configuration validation subroutine script.
-#   Copyright (C) 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
-#   2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008
-#   Free Software Foundation, Inc.
+#   Copyright 1992-2015 Free Software Foundation, Inc.
 
-timestamp='2009-04-17'
+timestamp='2015-03-08'
 
-# This file is (in principle) common to ALL GNU software.
-# The presence of a machine in this file suggests that SOME GNU software
-# can handle that machine.  It does not imply ALL GNU software can.
-#
-# This file is free software; you can redistribute it and/or modify
-# it under the terms of the GNU General Public License as published by
-# the Free Software Foundation; either version 2 of the License, or
+# This file is free software; you can redistribute it and/or modify it
+# under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 3 of the License, or
 # (at your option) any later version.
 #
-# This program is distributed in the hope that it will be useful,
-# but WITHOUT ANY WARRANTY; without even the implied warranty of
-# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
-# GNU General Public License for more details.
+# This program is distributed in the hope that it will be useful, but
+# WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+# General Public License for more details.
 #
 # You should have received a copy of the GNU General Public License
-# along with this program; if not, write to the Free Software
-# Foundation, Inc., 51 Franklin Street - Fifth Floor, Boston, MA
-# 02110-1301, USA.
+# along with this program; if not, see <http://www.gnu.org/licenses/>.
 #
 # As a special exception to the GNU General Public License, if you
 # distribute this file as part of a program that contains a
 # configuration script generated by Autoconf, you may include it under
-# the same distribution terms that you use for the rest of that program.
+# the same distribution terms that you use for the rest of that
+# program.  This Exception is an additional permission under section 7
+# of the GNU General Public License, version 3 ("GPLv3").
 
 
-# Please send patches to <[email protected]>.  Submit a context
-# diff and a properly formatted ChangeLog entry.
+# Please send patches to <[email protected]>.
 #
 # Configuration subroutine to validate and canonicalize a configuration type.
 # Supply the specified configuration type as an argument.
 # If it is invalid, we print an error message on stderr and exit with code 1.
 # Otherwise, we print the canonical config type on stdout and succeed.
 
+# You can get the latest version of this script from:
+# http://git.savannah.gnu.org/gitweb/?p=config.git;a=blob_plain;f=config.sub;hb=HEAD
+
 # This file is supposed to be the same for all GNU packages
 # and recognize all the CPU types, system types and aliases
 # that are meaningful with *any* GNU software.
@@ -72,8 +68,7 @@ Report bugs and patches to <[email protected]>."
 version="\
 GNU config.sub ($timestamp)
 
-Copyright (C) 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000, 2001,
-2002, 2003, 2004, 2005, 2006, 2007, 2008 Free Software Foundation, Inc.
+Copyright 1992-2015 Free Software Foundation, Inc.
 
 This is free software; see the source for copying conditions.  There is NO
 warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE."
@@ -120,13 +115,18 @@ esac
 # Here we must recognize all the valid KERNEL-OS combinations.
 maybe_os=`echo $1 | sed 's/^\(.*\)-\([^-]*-[^-]*\)$/\2/'`
 case $maybe_os in
-  nto-qnx* | linux-gnu* | linux-dietlibc | linux-newlib* | linux-uclibc* | \
-  uclinux-uclibc* | uclinux-gnu* | kfreebsd*-gnu* | knetbsd*-gnu* | netbsd*-gnu* | \
+  nto-qnx* | linux-gnu* | linux-android* | linux-dietlibc | linux-newlib* | \
+  linux-musl* | linux-uclibc* | uclinux-uclibc* | uclinux-gnu* | kfreebsd*-gnu* | \
+  knetbsd*-gnu* | netbsd*-gnu* | netbsd*-eabi* | \
   kopensolaris*-gnu* | \
   storm-chaos* | os2-emx* | rtmk-nova*)
     os=-$maybe_os
     basic_machine=`echo $1 | sed 's/^\(.*\)-\([^-]*-[^-]*\)$/\1/'`
     ;;
+  android-linux)
+    os=-linux-android
+    basic_machine=`echo $1 | sed 's/^\(.*\)-\([^-]*-[^-]*\)$/\1/'`-unknown
+    ;;
   *)
     basic_machine=`echo $1 | sed 's/-[^-]*$//'`
     if [ $basic_machine != $1 ]
@@ -149,10 +149,13 @@ case $os in
 	-convergent* | -ncr* | -news | -32* | -3600* | -3100* | -hitachi* |\
 	-c[123]* | -convex* | -sun | -crds | -omron* | -dg | -ultra | -tti* | \
 	-harris | -dolphin | -highlevel | -gould | -cbm | -ns | -masscomp | \
-	-apple | -axis | -knuth | -cray)
+	-apple | -axis | -knuth | -cray | -microblaze*)
 		os=
 		basic_machine=$1
 		;;
+	-bluegene*)
+		os=-cnk
+		;;
 	-sim | -cisco | -oki | -wec | -winbond)
 		os=
 		basic_machine=$1
@@ -167,10 +170,10 @@ case $os in
 		os=-chorusos
 		basic_machine=$1
 		;;
- 	-chorusrdb)
- 		os=-chorusrdb
+	-chorusrdb)
+		os=-chorusrdb
 		basic_machine=$1
- 		;;
+		;;
 	-hiux*)
 		os=-hiuxwe2
 		;;
@@ -215,6 +218,12 @@ case $os in
 	-isc*)
 		basic_machine=`echo $1 | sed -e 's/86-.*/86-pc/'`
 		;;
+	-lynx*178)
+		os=-lynxos178
+		;;
+	-lynx*5)
+		os=-lynxos5
+		;;
 	-lynx*)
 		os=-lynxos
 		;;
@@ -239,20 +248,28 @@ case $basic_machine in
 	# Some are omitted here because they have special meanings below.
 	1750a | 580 \
 	| a29k \
+	| aarch64 | aarch64_be \
 	| alpha | alphaev[4-8] | alphaev56 | alphaev6[78] | alphapca5[67] \
 	| alpha64 | alpha64ev[4-8] | alpha64ev56 | alpha64ev6[78] | alpha64pca5[67] \
 	| am33_2.0 \
-	| arc | arm | arm[bl]e | arme[lb] | armv[2345] | armv[345][lb] | avr | avr32 \
+	| arc | arceb \
+	| arm | arm[bl]e | arme[lb] | armv[2-8] | armv[3-8][lb] | armv7[arm] \
+	| avr | avr32 \
+	| be32 | be64 \
 	| bfin \
-	| c4x | clipper \
+	| c4x | c8051 | clipper \
 	| d10v | d30v | dlx | dsp16xx \
-	| fido | fr30 | frv \
+	| e2k | epiphany \
+	| fido | fr30 | frv | ft32 \
 	| h8300 | h8500 | hppa | hppa1.[01] | hppa2.0 | hppa2.0[nw] | hppa64 \
+	| hexagon \
 	| i370 | i860 | i960 | ia64 \
 	| ip2k | iq2000 \
+	| k1om \
+	| le32 | le64 \
 	| lm32 \
 	| m32c | m32r | m32rle | m68000 | m68k | m88k \
-	| maxq | mb | microblaze | mcore | mep | metag \
+	| maxq | mb | microblaze | microblazeel | mcore | mep | metag \
 	| mips | mipsbe | mipseb | mipsel | mipsle \
 	| mips16 \
 	| mips64 | mips64el \
@@ -266,36 +283,55 @@ case $basic_machine in
 	| mips64vr5900 | mips64vr5900el \
 	| mipsisa32 | mipsisa32el \
 	| mipsisa32r2 | mipsisa32r2el \
+	| mipsisa32r6 | mipsisa32r6el \
 	| mipsisa64 | mipsisa64el \
 	| mipsisa64r2 | mipsisa64r2el \
+	| mipsisa64r6 | mipsisa64r6el \
 	| mipsisa64sb1 | mipsisa64sb1el \
 	| mipsisa64sr71k | mipsisa64sr71kel \
+	| mipsr5900 | mipsr5900el \
 	| mipstx39 | mipstx39el \
 	| mn10200 | mn10300 \
 	| moxie \
 	| mt \
 	| msp430 \
-	| nios | nios2 \
+	| nds32 | nds32le | nds32be \
+	| nios | nios2 | nios2eb | nios2el \
 	| ns16k | ns32k \
-	| or32 \
+	| open8 | or1k | or1knd | or32 \
 	| pdp10 | pdp11 | pj | pjl \
-	| powerpc | powerpc64 | powerpc64le | powerpcle | ppcbe \
+	| powerpc | powerpc64 | powerpc64le | powerpcle \
 	| pyramid \
+	| riscv32 | riscv64 \
+	| rl78 | rx \
 	| score \
 	| sh | sh[1234] | sh[24]a | sh[24]aeb | sh[23]e | sh[34]eb | sheb | shbe | shle | sh[1234]le | sh3ele \
 	| sh64 | sh64le \
 	| sparc | sparc64 | sparc64b | sparc64v | sparc86x | sparclet | sparclite \
 	| sparcv8 | sparcv9 | sparcv9b | sparcv9v \
-	| spu | strongarm \
-	| tahoe | thumb | tic4x | tic80 | tron \
-	| v850 | v850e \
+	| spu \
+	| tahoe | tic4x | tic54x | tic55x | tic6x | tic80 | tron \
+	| ubicom32 \
+	| v850 | v850e | v850e1 | v850e2 | v850es | v850e2v3 \
+	| visium \
 	| we32k \
-	| x86 | xc16x | xscale | xscalee[bl] | xstormy16 | xtensa \
+	| x86 | xc16x | xstormy16 | xtensa \
 	| z8k | z80)
 		basic_machine=$basic_machine-unknown
 		;;
-	m6811 | m68hc11 | m6812 | m68hc12)
-		# Motorola 68HC11/12.
+	c54x)
+		basic_machine=tic54x-unknown
+		;;
+	c55x)
+		basic_machine=tic55x-unknown
+		;;
+	c6x)
+		basic_machine=tic6x-unknown
+		;;
+	leon|leon[3-9])
+		basic_machine=sparc-$basic_machine
+		;;
+	m6811 | m68hc11 | m6812 | m68hc12 | m68hcs12x | nvptx | picochip)
 		basic_machine=$basic_machine-unknown
 		os=-none
 		;;
@@ -305,6 +341,21 @@ case $basic_machine in
 		basic_machine=mt-unknown
 		;;
 
+	strongarm | thumb | xscale)
+		basic_machine=arm-unknown
+		;;
+	xgate)
+		basic_machine=$basic_machine-unknown
+		os=-none
+		;;
+	xscaleeb)
+		basic_machine=armeb-unknown
+		;;
+
+	xscaleel)
+		basic_machine=armel-unknown
+		;;
+
 	# We use `pc' rather than `unknown'
 	# because (1) that's what they normally are, and
 	# (2) the word "unknown" tends to confuse beginning users.
@@ -319,25 +370,31 @@ case $basic_machine in
 	# Recognize the basic CPU types with company name.
 	580-* \
 	| a29k-* \
+	| aarch64-* | aarch64_be-* \
 	| alpha-* | alphaev[4-8]-* | alphaev56-* | alphaev6[78]-* \
 	| alpha64-* | alpha64ev[4-8]-* | alpha64ev56-* | alpha64ev6[78]-* \
-	| alphapca5[67]-* | alpha64pca5[67]-* | arc-* \
+	| alphapca5[67]-* | alpha64pca5[67]-* | arc-* | arceb-* \
 	| arm-*  | armbe-* | armle-* | armeb-* | armv*-* \
 	| avr-* | avr32-* \
+	| be32-* | be64-* \
 	| bfin-* | bs2000-* \
-	| c[123]* | c30-* | [cjt]90-* | c4x-* | c54x-* | c55x-* | c6x-* \
-	| clipper-* | craynv-* | cydra-* \
+	| c[123]* | c30-* | [cjt]90-* | c4x-* \
+	| c8051-* | clipper-* | craynv-* | cydra-* \
 	| d10v-* | d30v-* | dlx-* \
-	| elxsi-* \
+	| e2k-* | elxsi-* \
 	| f30[01]-* | f700-* | fido-* | fr30-* | frv-* | fx80-* \
 	| h8300-* | h8500-* \
 	| hppa-* | hppa1.[01]-* | hppa2.0-* | hppa2.0[nw]-* | hppa64-* \
+	| hexagon-* \
 	| i*86-* | i860-* | i960-* | ia64-* \
 	| ip2k-* | iq2000-* \
+	| k1om-* \
+	| le32-* | le64-* \
 	| lm32-* \
 	| m32c-* | m32r-* | m32rle-* \
 	| m68000-* | m680[012346]0-* | m68360-* | m683?2-* | m68k-* \
 	| m88110-* | m88k-* | maxq-* | mcore-* | metag-* \
+	| microblaze-* | microblazeel-* \
 	| mips-* | mipsbe-* | mipseb-* | mipsel-* | mipsle-* \
 	| mips16-* \
 	| mips64-* | mips64el-* \
@@ -351,32 +408,42 @@ case $basic_machine in
 	| mips64vr5900-* | mips64vr5900el-* \
 	| mipsisa32-* | mipsisa32el-* \
 	| mipsisa32r2-* | mipsisa32r2el-* \
+	| mipsisa32r6-* | mipsisa32r6el-* \
 	| mipsisa64-* | mipsisa64el-* \
 	| mipsisa64r2-* | mipsisa64r2el-* \
+	| mipsisa64r6-* | mipsisa64r6el-* \
 	| mipsisa64sb1-* | mipsisa64sb1el-* \
 	| mipsisa64sr71k-* | mipsisa64sr71kel-* \
+	| mipsr5900-* | mipsr5900el-* \
 	| mipstx39-* | mipstx39el-* \
 	| mmix-* \
 	| mt-* \
 	| msp430-* \
-	| nios-* | nios2-* \
+	| nds32-* | nds32le-* | nds32be-* \
+	| nios-* | nios2-* | nios2eb-* | nios2el-* \
 	| none-* | np1-* | ns16k-* | ns32k-* \
+	| open8-* \
+	| or1k*-* \
 	| orion-* \
 	| pdp10-* | pdp11-* | pj-* | pjl-* | pn-* | power-* \
-	| powerpc-* | powerpc64-* | powerpc64le-* | powerpcle-* | ppcbe-* \
+	| powerpc-* | powerpc64-* | powerpc64le-* | powerpcle-* \
 	| pyramid-* \
-	| romp-* | rs6000-* \
+	| rl78-* | romp-* | rs6000-* | rx-* \
 	| sh-* | sh[1234]-* | sh[24]a-* | sh[24]aeb-* | sh[23]e-* | sh[34]eb-* | sheb-* | shbe-* \
 	| shle-* | sh[1234]le-* | sh3ele-* | sh64-* | sh64le-* \
 	| sparc-* | sparc64-* | sparc64b-* | sparc64v-* | sparc86x-* | sparclet-* \
 	| sparclite-* \
-	| sparcv8-* | sparcv9-* | sparcv9b-* | sparcv9v-* | strongarm-* | sv1-* | sx?-* \
-	| tahoe-* | thumb-* \
-	| tic30-* | tic4x-* | tic54x-* | tic55x-* | tic6x-* | tic80-* | tile-* \
+	| sparcv8-* | sparcv9-* | sparcv9b-* | sparcv9v-* | sv1-* | sx?-* \
+	| tahoe-* \
+	| tic30-* | tic4x-* | tic54x-* | tic55x-* | tic6x-* | tic80-* \
+	| tile*-* \
 	| tron-* \
-	| v850-* | v850e-* | vax-* \
+	| ubicom32-* \
+	| v850-* | v850e-* | v850e1-* | v850es-* | v850e2-* | v850e2v3-* \
+	| vax-* \
+	| visium-* \
 	| we32k-* \
-	| x86-* | x86_64-* | xc16x-* | xps100-* | xscale-* | xscalee[bl]-* \
+	| x86-* | x86_64-* | xc16x-* | xps100-* \
 	| xstormy16-* | xtensa*-* \
 	| ymp-* \
 	| z8k-* | z80-*)
@@ -401,7 +468,7 @@ case $basic_machine in
 		basic_machine=a29k-amd
 		os=-udi
 		;;
-    	abacus)
+	abacus)
 		basic_machine=abacus-unknown
 		;;
 	adobe68k)
@@ -451,6 +518,9 @@ case $basic_machine in
 		basic_machine=i386-pc
 		os=-aros
 		;;
+        asmjs)
+		basic_machine=asmjs-unknown
+		;;
 	aux)
 		basic_machine=m68k-apple
 		os=-aux
@@ -467,11 +537,24 @@ case $basic_machine in
 		basic_machine=bfin-`echo $basic_machine | sed 's/^[^-]*-//'`
 		os=-linux
 		;;
+	bluegene*)
+		basic_machine=powerpc-ibm
+		os=-cnk
+		;;
+	c54x-*)
+		basic_machine=tic54x-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	c55x-*)
+		basic_machine=tic55x-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	c6x-*)
+		basic_machine=tic6x-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
 	c90)
 		basic_machine=c90-cray
 		os=-unicos
 		;;
-        cegcc)
+	cegcc)
 		basic_machine=arm-unknown
 		os=-cegcc
 		;;
@@ -503,7 +586,7 @@ case $basic_machine in
 		basic_machine=craynv-cray
 		os=-unicosmp
 		;;
-	cr16)
+	cr16 | cr16-*)
 		basic_machine=cr16-unknown
 		os=-elf
 		;;
@@ -661,7 +744,6 @@ case $basic_machine in
 	i370-ibm* | ibm*)
 		basic_machine=i370-ibm
 		;;
-# I'm not sure what "Sysv32" means.  Should this be sysv3.2?
 	i*86v32)
 		basic_machine=`echo $1 | sed -e 's/86.*/86-pc/'`
 		os=-sysv32
@@ -700,6 +782,9 @@ case $basic_machine in
 		basic_machine=m68k-isi
 		os=-sysv
 		;;
+	leon-*|leon[3-9]-*)
+		basic_machine=sparc-`echo $basic_machine | sed 's/-.*//'`
+		;;
 	m68knommu)
 		basic_machine=m68k-unknown
 		os=-linux
@@ -719,8 +804,15 @@ case $basic_machine in
 		basic_machine=ns32k-utek
 		os=-sysv
 		;;
+	microblaze*)
+		basic_machine=microblaze-xilinx
+		;;
+	mingw64)
+		basic_machine=x86_64-pc
+		os=-mingw64
+		;;
 	mingw32)
-		basic_machine=i386-pc
+		basic_machine=i686-pc
 		os=-mingw32
 		;;
 	mingw32ce)
@@ -748,6 +840,10 @@ case $basic_machine in
 		basic_machine=powerpc-unknown
 		os=-morphos
 		;;
+	moxiebox)
+		basic_machine=moxie-unknown
+		os=-moxiebox
+		;;
 	msdos)
 		basic_machine=i386-pc
 		os=-msdos
@@ -755,10 +851,18 @@ case $basic_machine in
 	ms1-*)
 		basic_machine=`echo $basic_machine | sed -e 's/ms1-/mt-/'`
 		;;
+	msys)
+		basic_machine=i686-pc
+		os=-msys
+		;;
 	mvs)
 		basic_machine=i370-ibm
 		os=-mvs
 		;;
+	nacl)
+		basic_machine=le32-unknown
+		os=-nacl
+		;;
 	ncr3000)
 		basic_machine=i486-ncr
 		os=-sysv4
@@ -823,6 +927,12 @@ case $basic_machine in
 	np1)
 		basic_machine=np1-gould
 		;;
+	neo-tandem)
+		basic_machine=neo-tandem
+		;;
+	nse-tandem)
+		basic_machine=nse-tandem
+		;;
 	nsr-tandem)
 		basic_machine=nsr-tandem
 		;;
@@ -905,9 +1015,10 @@ case $basic_machine in
 		;;
 	power)	basic_machine=power-ibm
 		;;
-	ppc)	basic_machine=powerpc-unknown
+	ppc | ppcbe)	basic_machine=powerpc-unknown
 		;;
-	ppc-*)	basic_machine=powerpc-`echo $basic_machine | sed 's/^[^-]*-//'`
+	ppc-* | ppcbe-*)
+		basic_machine=powerpc-`echo $basic_machine | sed 's/^[^-]*-//'`
 		;;
 	ppcle | powerpclittle | ppc-le | powerpc-little)
 		basic_machine=powerpcle-unknown
@@ -932,7 +1043,11 @@ case $basic_machine in
 		basic_machine=i586-unknown
 		os=-pw32
 		;;
-	rdos)
+	rdos | rdos64)
+		basic_machine=x86_64-pc
+		os=-rdos
+		;;
+	rdos32)
 		basic_machine=i386-pc
 		os=-rdos
 		;;
@@ -1001,6 +1116,9 @@ case $basic_machine in
 		basic_machine=i860-stratus
 		os=-sysv4
 		;;
+	strongarm-* | thumb-*)
+		basic_machine=arm-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
 	sun2)
 		basic_machine=m68000-sun
 		;;
@@ -1057,20 +1175,8 @@ case $basic_machine in
 		basic_machine=t90-cray
 		os=-unicos
 		;;
-	tic54x | c54x*)
-		basic_machine=tic54x-unknown
-		os=-coff
-		;;
-	tic55x | c55x*)
-		basic_machine=tic55x-unknown
-		os=-coff
-		;;
-	tic6x | c6x*)
-		basic_machine=tic6x-unknown
-		os=-coff
-		;;
 	tile*)
-		basic_machine=tile-unknown
+		basic_machine=$basic_machine-unknown
 		os=-linux-gnu
 		;;
 	tx39)
@@ -1140,6 +1246,9 @@ case $basic_machine in
 	xps | xps100)
 		basic_machine=xps100-honeywell
 		;;
+	xscale-* | xscalee[bl]-*)
+		basic_machine=`echo $basic_machine | sed 's/^xscale/arm/'`
+		;;
 	ymp)
 		basic_machine=ymp-cray
 		os=-unicos
@@ -1237,9 +1346,12 @@ esac
 if [ x"$os" != x"" ]
 then
 case $os in
-        # First match some system type aliases
-        # that might get confused with valid system types.
+	# First match some system type aliases
+	# that might get confused with valid system types.
 	# -solaris* is a basic system type, with this one exception.
+	-auroraux)
+		os=-auroraux
+		;;
 	-solaris1 | -solaris1.*)
 		os=`echo $os | sed -e 's|solaris1|sunos4|'`
 		;;
@@ -1260,30 +1372,31 @@ case $os in
 	# Each alternative MUST END IN A *, to match a version number.
 	# -sysv* is not here because it comes later, after sysvr4.
 	-gnu* | -bsd* | -mach* | -minix* | -genix* | -ultrix* | -irix* \
-	      | -*vms* | -sco* | -esix* | -isc* | -aix* | -sunos | -sunos[34]*\
-	      | -hpux* | -unos* | -osf* | -luna* | -dgux* | -solaris* | -sym* \
-	      | -kopensolaris* \
+	      | -*vms* | -sco* | -esix* | -isc* | -aix* | -cnk* | -sunos | -sunos[34]*\
+	      | -hpux* | -unos* | -osf* | -luna* | -dgux* | -auroraux* | -solaris* \
+	      | -sym* | -kopensolaris* | -plan9* \
 	      | -amigaos* | -amigados* | -msdos* | -newsos* | -unicos* | -aof* \
-	      | -aos* | -aros* \
+	      | -aos* | -aros* | -cloudabi* \
 	      | -nindy* | -vxsim* | -vxworks* | -ebmon* | -hms* | -mvs* \
 	      | -clix* | -riscos* | -uniplus* | -iris* | -rtu* | -xenix* \
 	      | -hiux* | -386bsd* | -knetbsd* | -mirbsd* | -netbsd* \
-	      | -openbsd* | -solidbsd* \
+	      | -bitrig* | -openbsd* | -solidbsd* \
 	      | -ekkobsd* | -kfreebsd* | -freebsd* | -riscix* | -lynxos* \
 	      | -bosx* | -nextstep* | -cxux* | -aout* | -elf* | -oabi* \
 	      | -ptx* | -coff* | -ecoff* | -winnt* | -domain* | -vsta* \
 	      | -udi* | -eabi* | -lites* | -ieee* | -go32* | -aux* \
 	      | -chorusos* | -chorusrdb* | -cegcc* \
-	      | -cygwin* | -pe* | -psos* | -moss* | -proelf* | -rtems* \
-	      | -mingw32* | -linux-gnu* | -linux-newlib* | -linux-uclibc* \
-	      | -uxpv* | -beos* | -mpeix* | -udk* \
+	      | -cygwin* | -msys* | -pe* | -psos* | -moss* | -proelf* | -rtems* \
+	      | -mingw32* | -mingw64* | -linux-gnu* | -linux-android* \
+	      | -linux-newlib* | -linux-musl* | -linux-uclibc* \
+	      | -uxpv* | -beos* | -mpeix* | -udk* | -moxiebox* \
 	      | -interix* | -uwin* | -mks* | -rhapsody* | -darwin* | -opened* \
 	      | -openstep* | -oskit* | -conix* | -pw32* | -nonstopux* \
 	      | -storm-chaos* | -tops10* | -tenex* | -tops20* | -its* \
 	      | -os2* | -vos* | -palmos* | -uclinux* | -nucleus* \
 	      | -morphos* | -superux* | -rtmk* | -rtmk-nova* | -windiss* \
 	      | -powermax* | -dnix* | -nx6 | -nx7 | -sei* | -dragonfly* \
-	      | -skyos* | -haiku* | -rdos* | -toppers* | -drops*)
+	      | -skyos* | -haiku* | -rdos* | -toppers* | -drops* | -es* | -tirtos*)
 	# Remember, each alternative MUST END IN *, to match a version number.
 		;;
 	-qnx*)
@@ -1322,7 +1435,7 @@ case $os in
 	-opened*)
 		os=-openedition
 		;;
-        -os400*)
+	-os400*)
 		os=-os400
 		;;
 	-wince*)
@@ -1371,7 +1484,7 @@ case $os in
 	-sinix*)
 		os=-sysv4
 		;;
-        -tpf*)
+	-tpf*)
 		os=-tpf
 		;;
 	-triton*)
@@ -1407,15 +1520,14 @@ case $os in
 	-aros*)
 		os=-aros
 		;;
-	-kaos*)
-		os=-kaos
-		;;
 	-zvmoe)
 		os=-zvmoe
 		;;
 	-dicos*)
 		os=-dicos
 		;;
+	-nacl*)
+		;;
 	-none)
 		;;
 	*)
@@ -1438,10 +1550,10 @@ else
 # system, and we'll never get to this point.
 
 case $basic_machine in
-        score-*)
+	score-*)
 		os=-elf
 		;;
-        spu-*)
+	spu-*)
 		os=-elf
 		;;
 	*-acorn)
@@ -1453,8 +1565,23 @@ case $basic_machine in
 	arm*-semi)
 		os=-aout
 		;;
-        c4x-* | tic4x-*)
-        	os=-coff
+	c4x-* | tic4x-*)
+		os=-coff
+		;;
+	c8051-*)
+		os=-elf
+		;;
+	hexagon-*)
+		os=-elf
+		;;
+	tic54x-*)
+		os=-coff
+		;;
+	tic55x-*)
+		os=-coff
+		;;
+	tic6x-*)
+		os=-coff
 		;;
 	# This must come before the *-dec entry.
 	pdp10-*)
@@ -1474,14 +1601,11 @@ case $basic_machine in
 		;;
 	m68000-sun)
 		os=-sunos3
-		# This also exists in the configure program, but was not the
-		# default.
-		# os=-sunos4
 		;;
 	m68*-cisco)
 		os=-aout
 		;;
-        mep-*)
+	mep-*)
 		os=-elf
 		;;
 	mips*-cisco)
@@ -1508,7 +1632,7 @@ case $basic_machine in
 	*-ibm)
 		os=-aix
 		;;
-    	*-knuth)
+	*-knuth)
 		os=-mmixware
 		;;
 	*-wec)
@@ -1613,7 +1737,7 @@ case $basic_machine in
 			-sunos*)
 				vendor=sun
 				;;
-			-aix*)
+			-cnk*|-aix*)
 				vendor=ibm
 				;;
 			-beos*)

configure.in

@@ -82,7 +82,7 @@ CFLAGS="$iverilog_temp_cflags"
 
 AC_SUBST(DEPENDENCY_FLAG, [-MD])
 AC_SUBST(WARNING_FLAGS, ["-Wall $iverilog_wextra_flag -Wshadow"])
-AC_SUBST(WARNING_FLAGS_CC, [""])
+AC_SUBST(WARNING_FLAGS_CC, ["-Wstrict-prototypes"])
 AC_SUBST(WARNING_FLAGS_CXX, [""])
 fi
 
@@ -223,6 +223,10 @@ case "${host}" in
 	CPPFLAGS="-mieee $CPPFLAGS"
 	CFLAGS="-mieee $CFLAGS"
 	;;
+    *-*-mingw*)
+	CXXFLAGS="-D__USE_MINGW_ANSI_STDIO=1 $CXXFLAGS"
+	CFLAGS="-D__USE_MINGW_ANSI_STDIO=1 $CFLAGS"
+	;;
 esac
 
 # Do some more operating system specific setup. We put the file64_support
@@ -238,6 +242,9 @@ case "${host}" in
 esac
 AC_SUBST(file64_support)
 
+# fstapi.c (from GTKWave) needs this define.
+AC_CHECK_FUNCS(realpath)
+
 # Check that these functions exist. They are mostly C99
 # functions that older compilers may not yet support.
 AC_CHECK_FUNCS(fopen64)

cppcheck.sup

@@ -1,3 +1,3 @@
 // These are correct and are used to find the base (zero) pin.
-thisSubtraction:netlist.h:4938
-thisSubtraction:netlist.h:4947
+thisSubtraction:netlist.h:4976
+thisSubtraction:netlist.h:4985

design_dump.cc

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 1998-2014 Stephen Williams ([email protected])
+ * Copyright (c) 1998-2015 Stephen Williams ([email protected])
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -677,8 +677,12 @@ void NetConst::dump_node(ostream&o, unsigned ind) const
 void NetFF::dump_node(ostream&o, unsigned ind) const
 {
       o << setw(ind) << "" << "LPM_FF: " << name()
-	<< " scope=" << scope_path(scope())
-	<< " aset_value=" << aset_value_ << endl;
+	<< " scope=" << scope_path(scope());
+      if (negedge_)
+	    o << " negedge";
+      else
+	    o << " posedge";
+      o << " aset_value=" << aset_value_ << endl;
 
       dump_node_pins(o, ind+4);
       dump_obj_attr(o, ind+4);
@@ -1375,6 +1379,7 @@ void NetScope::dump(ostream&o) const
       if (is_cell()) o << " (cell)";
       if (nested_module()) o << " (nested)";
       if (program_block()) o << " (program)";
+      if (is_interface()) o << " (interface)";
       o << " " << children_.size() << " children, "
 	<< classes_.size() << " classes" << endl;
 
@@ -1889,7 +1894,7 @@ void Design::dump(ostream&o) const
 	    cur->second->dump(o);
       }
 
-      o << "$ROOT CLASSESS:" << endl;
+      o << "$ROOT CLASSES:" << endl;
       for (map<perm_string,netclass_t*>::const_iterator cur = classes_.begin()
 		 ; cur != classes_.end() ; ++cur) {
 	    cur->second->dump_scope(o);

driver-vpi/Makefile.in

@@ -49,6 +49,7 @@ endif
 
 CPPFLAGS = $(INCLUDE_PATH) @CPPFLAGS@ @DEFS@
 CFLAGS = @WARNING_FLAGS@ @WARNING_FLAGS_CC@ @CFLAGS@
+CXXFLAGS = @WARNING_FLAGS@ @WARNING_FLAGS_CXX@ @CXXFLAGS@
 LDFLAGS = @LDFLAGS@
 
 O = main.o res.o
@@ -80,11 +81,12 @@ config.h: $(srcdir)/config.h.in Makefile
 	sed -e 's;@IVLCC@;@CC@;' -e 's;@IVLCXX@;@CXX@;' \
 	    -e 's;@SUFFIX@;$(suffix);g' \
 	    -e 's;@IVLCFLAGS@;$(CFLAGS);' \
+	    -e 's;@IVLCXXFLAGS@;$(CXXFLAGS);' \
 	    -e 's;@SHARED@;@shared@;' $< > $@
 
 # Windows specific...
 res.rc: $(srcdir)/res.rc.in ../version.exe
-	sed -e 's;@PRODUCTVERSION@;'`../version.exe '%M,%m,%n,0'`';' \
+	sed -e 's;@PRODUCTVERSION@;'`../version.exe '%M,%n,0,0'`';' \
 	$(srcdir)/res.rc.in > $@
 
 res.o: res.rc

driver-vpi/config.h.in

@@ -4,6 +4,7 @@
 #define IVERILOG_VPI_CC        "@IVLCC@"
 #define IVERILOG_VPI_CXX       "@IVLCXX@"
 #define IVERILOG_VPI_CFLAGS    " @IVLCFLAGS@"
+#define IVERILOG_VPI_CXXFLAGS  " @IVLCXXFLAGS@"
 #define IVERILOG_VPI_LDFLAGS   "@SHARED@"
 #define IVERILOG_VPI_LDLIBS    "-lveriuser@SUFFIX@ -lvpi@SUFFIX@"
 #define IVERILOG_SUFFIX        "@SUFFIX@"

driver-vpi/main.c

@@ -1,5 +1,6 @@
 /*
  * Copyright (c) 2002 Gus Baldauf ([email protected])
+ * Copyright (c) 2015 Martin Whitaker
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -20,7 +21,7 @@
 /*
  *	iverilog-vpi.c
  *
- *	this program provides the functionality of iverilog-vpi.sh under Win32
+ *	this program provides the functionality of iverilog-vpi.sh under Windows
  */
 
 #include <stdio.h>
@@ -32,7 +33,7 @@
 
 #include <windows.h>
 
-static void setup_ivl_environment();
+static void setup_ivl_environment(void);
 static void assign(char **ptr, char *str);
 
 /* The compile options: compiler, flags, etc. are in here */
@@ -43,16 +44,17 @@ static void assign(char **ptr, char *str);
 static struct global_strings {
 	char *pCCSRC;		/* list of C source files (*.c) */
 	char *pCXSRC;		/* list of C++ source files (*.cc, *.cpp) */
-	char *pOBJ;			/* list of object files */
-	char *pLIB;			/* list of library files */
-	char *pINCS;			/* list of include directories */
-	char *pDEFS;			/* list of definitions */
-	char *pOUT;			/* output file name (.vpi extension), if 0 length then no source files specified */
+	char *pOBJ;		/* list of object files */
+	char *pLIB;		/* list of library files */
+	char *pINCS;		/* list of include directories */
+	char *pDEFS;		/* list of definitions */
+	char *pOUT;		/* output file name (.vpi extension), if 0 length then no source files specified */
 	char *pMINGW;		/* path to MinGW directory */
-	char *pIVL;			/* path to IVL directory */
-	char *pCFLAGS;		/* CFLAGS option */
-	char *pLDLIBS;		/* LDLIBS option */
-	char *pNewPath;		/* new PATH environment variable setting */
+	char *pIVL;		/* path to IVL directory */
+	char *pCCFLAGS;		/* compiler flags for compiling C source files */
+	char *pCXFLAGS;		/* compiler flags for compiling C++ source files */
+	char *pLDLIBS;		/* linker flags for final linking stage */
+	char *pCCNAME;		/* base name of compiler */
 	char *pLD;		/* what to use for a linker */
 } gstr;
 
@@ -75,21 +77,21 @@ static void myExit(int exitVal)
 	deInitDynString(gstr.pOUT);
 	deInitDynString(gstr.pMINGW);
 	deInitDynString(gstr.pIVL);
-	deInitDynString(gstr.pCFLAGS);
+	deInitDynString(gstr.pCCFLAGS);
+	deInitDynString(gstr.pCXFLAGS);
 	deInitDynString(gstr.pLDLIBS);
-	deInitDynString(gstr.pNewPath);
-	free(gstr.pLD);
+	deInitDynString(gstr.pCCNAME);
+	deInitDynString(gstr.pLD);
 
 	exit(exitVal);
 }
 
 /* display usage summary and exit */
 
-static void usage()
+static void usage(void)
 {
-	fprintf(stderr,"usage: iverilog-vpi" IVERILOG_SUFFIX " [src and obj files]...\n");
-	fprintf(stderr,"   or  iverilog-vpi" IVERILOG_SUFFIX " -mingw=dir\n");
-	fprintf(stderr,"   or  iverilog-vpi" IVERILOG_SUFFIX " -ivl=dir\n");
+	fprintf(stderr, "usage: iverilog-vpi" IVERILOG_SUFFIX " [options] [src and obj files]...\n");
+	fprintf(stderr, "   or  iverilog-vpi" IVERILOG_SUFFIX " -mingw=dir\n");
 	myExit(1);
 }
 
@@ -98,7 +100,7 @@ static void initDynString(char **str)
 	*str = (char *) malloc(1);
 
 	if (!*str) {
-		fprintf(stderr,"error: out of memory\n");
+		fprintf(stderr, "error: out of memory\n");
 		myExit(4);
 	}
 
@@ -107,8 +109,10 @@ static void initDynString(char **str)
 
 /* initialize dynamic memory buffers */
 
-static void init()
+static void init(void)
 {
+	char *ptr;
+
 	initDynString(&gstr.pCCSRC);
 	initDynString(&gstr.pCXSRC);
 	initDynString(&gstr.pOBJ);
@@ -118,9 +122,17 @@ static void init()
 	initDynString(&gstr.pOUT);
 	initDynString(&gstr.pMINGW);
 	initDynString(&gstr.pIVL);
-	initDynString(&gstr.pCFLAGS);
+	initDynString(&gstr.pCCFLAGS);
+	initDynString(&gstr.pCXFLAGS);
 	initDynString(&gstr.pLDLIBS);
-	initDynString(&gstr.pNewPath);
+	initDynString(&gstr.pCCNAME);
+	initDynString(&gstr.pLD);
+
+	  /* Get the base name of the C compiler. */
+	assign(&gstr.pCCNAME, IVERILOG_VPI_CC);
+	ptr = strchr(gstr.pCCNAME, ' ');
+	if (ptr != NULL) *ptr = '\0';
+
 	  /* By default use the C compiler to link the programs. */
 	assign(&gstr.pLD, IVERILOG_VPI_CC);
 }
@@ -136,7 +148,7 @@ static int endsIn (char *end, char *str)
 
 	ext = str + (strlen(str) - strlen(end));
 
-	return stricmp(end,ext) ? 0 : 1;
+	return stricmp(end, ext) ? 0 : 1;
 }
 
 /* return true if "str" begins with "prefix", case insensitive */
@@ -146,35 +158,35 @@ static int startsWith (char *prefix, char *str)
 	if (strlen(prefix) >= strlen(str))
 		return 0;
 
-	return strnicmp(prefix,str,strlen(prefix)) ? 0 : 1;
+	return strnicmp(prefix, str, strlen(prefix)) ? 0 : 1;
 }
 
 /* append "app" to "ptr", allocating memory as needed    */
 /*   if count is zero, then copy all characters of "app" */
 
-static void appendn (char **ptr, char *app, int count)
+static void appendn (char **ptr, char *app, size_t count)
 {
 	char *nptr = (char *) realloc(*ptr, strlen(*ptr) +
-                                            (count?count:strlen(app)) + 1);
+                                            (count ? count : strlen(app)) + 1);
 
 	if (nptr == NULL) {
-		fprintf(stderr,"error: out of memory\n");
+		fprintf(stderr, "error: out of memory\n");
 		free(*ptr);
 		myExit(4);
 	}
 	*ptr = nptr;
 
 	if (count)
-		strncat(*ptr,app,count);
+		strncat(*ptr, app, count);
 	else
-		strcat(*ptr,app);
+		strcat(*ptr, app);
 }
 
 /* append "app" to "ptr", allocating memory as needed    */
 
 static void append (char **ptr, char *app)
 {
-	appendn(ptr,app,0);
+	appendn(ptr, app, 0);
 }
 
 /* if the string does not end with a backslash, add one */
@@ -182,36 +194,36 @@ static void append (char **ptr, char *app)
 static void appendBackSlash(char **str)
 {
 	if ((*str)[strlen(*str)-1] != '\\')
-		append(str,"\\");
+		append(str, "\\");
 }
 
 /* copy count characters of "str" to "ptr", allocating memory as needed */
 /*   if count is zero, then copy all characters of "str"                */
 
-static void assignn (char **ptr, char *str, int count)
+static void assignn (char **ptr, char *str, size_t count)
 {
-	char *nptr = (char *) realloc(*ptr, (count?count:strlen(str)) + 1);
+	char *nptr = (char *) realloc(*ptr, (count ? count : strlen(str)) + 1);
 
 	if (nptr == NULL) {
-		fprintf(stderr,"error: out of memory\n");
+		fprintf(stderr, "error: out of memory\n");
 		free(*ptr);
 		myExit(4);
 	}
 	*ptr = nptr;
 
 	if (count) {
-		strncpy(*ptr,str,count);
+		strncpy(*ptr, str, count);
 		(*ptr)[count] = 0;
 	}
 	else
-		strcpy(*ptr,str);
+		strcpy(*ptr, str);
 }
 
 /* copy count characters of "str" to "ptr", allocating memory as needed */
 
 static void assign (char **ptr, char *str)
 {
-	assignn(ptr,str,0);
+	assignn(ptr, str, 0);
 }
 
 /* get a copy of a Icarus Verilog registry string key */
@@ -223,11 +235,11 @@ static int GetRegistryKey(char *key, char **value)
 	char *regKeyBuffer;
 	DWORD regKeyType, regKeySize;
 
-	lrv = RegOpenKeyEx(HKEY_LOCAL_MACHINE,"Software\\Icarus Verilog",0,KEY_QUERY_VALUE,&hkKey);
+	lrv = RegOpenKeyEx(HKEY_LOCAL_MACHINE, "Software\\Icarus Verilog", 0, KEY_QUERY_VALUE, &hkKey);
 	if (lrv != ERROR_SUCCESS)
 		return 0;
 
-	lrv = RegQueryValueEx(hkKey,key,NULL,&regKeyType,NULL,&regKeySize);
+	lrv = RegQueryValueEx(hkKey, key, NULL, &regKeyType, NULL, &regKeySize);
 	if ((lrv != ERROR_SUCCESS) || (regKeyType != REG_SZ) || (!regKeySize)) {
 		lrv = RegCloseKey(hkKey);
 		return 0;
@@ -236,7 +248,7 @@ static int GetRegistryKey(char *key, char **value)
 	regKeyBuffer = (char *) malloc(regKeySize+1);
 	if (!regKeyBuffer) {
 		lrv = RegCloseKey(hkKey);
-		fprintf(stderr,"error: out of memory\n");
+		fprintf(stderr, "error: out of memory\n");
 		myExit(4);
 	}
 	regKeyBuffer[regKeySize] = 0;		/* makes sure there is a trailing NULL */
@@ -252,7 +264,7 @@ static int GetRegistryKey(char *key, char **value)
 
 	RegCloseKey(hkKey);
 
-	assign(value,regKeyBuffer);
+	assign(value, regKeyBuffer);
 	free(regKeyBuffer);
 
 	return 1;
@@ -262,27 +274,30 @@ static int GetRegistryKey(char *key, char **value)
 
 static void SetRegistryKey(char *key, char *value)
 {
+	long lrv;
 	HKEY hkKey;
 	DWORD res;
 
-	if (RegCreateKeyEx(
-		HKEY_LOCAL_MACHINE,
-		"Software\\Icarus Verilog",
-		0,
-		"",
-		REG_OPTION_NON_VOLATILE,
-		KEY_ALL_ACCESS,NULL,
-		&hkKey,
-		&res) != ERROR_SUCCESS)
-			return;
+	lrv = RegCreateKeyEx(HKEY_LOCAL_MACHINE, "Software\\Icarus Verilog", 0, "",
+			     REG_OPTION_NON_VOLATILE, KEY_ALL_ACCESS, NULL, &hkKey, &res);
+	if (lrv != ERROR_SUCCESS) {
+		char message[1024];
+		FormatMessage(FORMAT_MESSAGE_FROM_SYSTEM, NULL, lrv, LANG_USER_DEFAULT,
+			      message, sizeof(message), NULL);
+		fprintf(stderr, "error: couldn't write to registry - %s\n", message);
+		if (lrv == ERROR_ACCESS_DENIED) {
+			fprintf(stderr, "       try running as administrator\n");
+		}
+		return;
+	}
 
 	/* This needs an unsigned char *, but for MinGW the char is signed. */
 	RegSetValueEx(hkKey, key, 0, REG_SZ, (unsigned char *) value,
 	              strlen(value)+1);
 	RegCloseKey(hkKey);
 
-	printf("info:  storing %s in Windows' registry entry\n",value);
-	printf("       HKEY_LOCAL_MACHINE\\Software\\Icarus Verilog\\%s\n",key);
+	printf("info:  storing %s in Windows' registry entry\n", value);
+	printf("       HKEY_LOCAL_MACHINE\\Software\\Icarus Verilog\\%s\n", key);
 }
 
 /* parse the command line, assign results to global variable strings */
@@ -299,7 +314,6 @@ static int parse(int argc, char *argv[])
 	char lib_option[]   = "-l";
 	char inc_option[]   = "-I";
 	char mingw_option[] = "-mingw=";
-	char ivl_option[]   = "-ivl=";
 	char def_option[]   = "-D";
 
 	if (argc == 1) return 0;
@@ -339,7 +353,7 @@ static int parse(int argc, char *argv[])
 		  /* Check for compiled object files */
 		else if (endsIn(dot_o_ext, argv[idx])) {
 			++srcFileCnt;
-			append(&gstr.pOBJ," ");
+			append(&gstr.pOBJ, " ");
 			append(&gstr.pOBJ, argv[idx]);
 			if (!*gstr.pOUT)
 			   assignn(&gstr.pOUT, argv[idx],
@@ -349,10 +363,6 @@ static int parse(int argc, char *argv[])
 		else if (startsWith(mingw_option, argv[idx]))
 			assignn(&gstr.pMINGW, argv[idx]+sizeof(mingw_option)-1,
 			        strlen(argv[idx])-(sizeof(mingw_option)-1));
-		  /* Check for the -ivl   option */
-		else if (startsWith(ivl_option, argv[idx]))
-			assignn(&gstr.pIVL, argv[idx]+sizeof(ivl_option)-1,
-			        strlen(argv[idx])-(sizeof(ivl_option)-1));
 		  /* Check for the --name option */
 		else if (startsWith(name_option, argv[idx])) {
 			assignn(&gstr.pOUT, argv[idx]+sizeof(name_option)-1,
@@ -376,7 +386,7 @@ static int parse(int argc, char *argv[])
 		  /* Check for the --cflags option */
 		else if (stricmp("--cflags", argv[idx]) == 0) {
 			setup_ivl_environment();
-			printf("%s\n", gstr.pCFLAGS);
+			printf("%s\n", gstr.pCCFLAGS);
 			myExit(0);
 		}
 		  /* Check for the --ldflags option */
@@ -393,7 +403,7 @@ static int parse(int argc, char *argv[])
 		  /* Check for the --install-dir option */
 		else if (stricmp("--install-dir", argv[idx]) == 0) {
 			setup_ivl_environment();
-			printf("%s\\\\lib\\\\ivl" IVERILOG_SUFFIX "\\\\.\n", gstr.pIVL);
+			printf("%s\\lib\\ivl" IVERILOG_SUFFIX "\\.\n", gstr.pIVL);
 			myExit(0);
 		}
 		  /* This is different than iverilog-vpi.sh, we don't
@@ -402,15 +412,16 @@ static int parse(int argc, char *argv[])
 	}
 
 	  /* In case there is a --name without source/object files */
-	if (0 == srcFileCnt) assign(&gstr.pOUT,"");
+	if (0 == srcFileCnt) assign(&gstr.pOUT, "");
 
-	  /* Normally it's an error if there are no source or object files */
-	  /* Unless we are setting the IVL or MinGW registry entries */
-	if (!*gstr.pOUT) {
-		if (!*gstr.pMINGW && !*gstr.pIVL) return 0;
-	} else
+	if (*gstr.pOUT) {
 		  /* We have a valid result file so add the .vpi extension */
 		append(&gstr.pOUT, ".vpi");
+	} else {
+		  /* Unless we are setting the MinGW registry entry, it's
+		     an error if there are no source or object files */
+		if (!*gstr.pMINGW) return 0;
+	}
 
 	return 1;
 }
@@ -421,58 +432,27 @@ static void checkMingwDir(char *root)
 {
 	int irv;
 	struct _stat stat_buf;
-
-	char *path, *comp, *cp;
-	initDynString(&path);
-	assign(&path,gstr.pMINGW);
-	appendBackSlash(&path);
-	append(&path,"bin\\");
-	/* Get just the compiler name (the first word) */
-	comp = strdup(IVERILOG_VPI_CC);
-	cp = strchr(comp, ' ');
-	if (cp != NULL) *cp = '\0';
-        append(&path, comp);
-	append(&path,".exe");
-        free(comp);
-
-	irv = _stat(path,&stat_buf);
-	deInitDynString(path);
-
-	if (irv) {
-		fprintf(stderr,"error: %s does not appear to be the valid root directory\n",root);
-		fprintf(stderr,"       of MinGW.  Use the -mingw option of iverilog-vpi.exe to\n");
-		fprintf(stderr,"       point to the MinGW root directory.  For a Windows command\n");
-		fprintf(stderr,"       shell the option would be something like -mingw=c:\\mingw\n");
-		fprintf(stderr,"       For a Cygwin shell the option would be something like\n");
-		fprintf(stderr,"       -mingw=c:\\\\mingw\n");
-		myExit(5);
-	}
-}
-
-/* do minimal check that the Icarus Verilog root directory looks valid */
-
-static void checkIvlDir(char *root)
-{
-	int irv;
-	struct _stat stat_buf;
-
 	char *path;
+
 	initDynString(&path);
-	assign(&path,gstr.pIVL);
+	assign(&path, gstr.pMINGW);
 	appendBackSlash(&path);
-	append(&path,"bin\\vvp" IVERILOG_SUFFIX ".exe");
+	append(&path, "bin\\");
+        append(&path, gstr.pCCNAME);
+	append(&path, ".exe");
 
 	irv = _stat(path,&stat_buf);
 	deInitDynString(path);
 
 	if (irv) {
-		fprintf(stderr,"error: %s does not appear to be the valid root directory of\n",root);
-		fprintf(stderr,"       Icarus Verilog.  Use the -ivl option of iverilog-vpi" IVERILOG_SUFFIX " to\n");
-		fprintf(stderr,"       point to the Icarus Verilog root directory.  For a Windows\n");
-		fprintf(stderr,"       command shell the option would be something like -ivl=c:\\iverilog\n");
-		fprintf(stderr,"       For a Cygwin shell the option would be something like\n");
-		fprintf(stderr,"       -ivl=c:\\\\iverilog\n");
-		myExit(6);
+		fprintf(stderr, "error: %s\n", root);
+		fprintf(stderr, "       does not appear to be the valid root directory of\n");
+		fprintf(stderr, "       MinGW.  Use the -mingw option of iverilog-vpi.exe to\n");
+		fprintf(stderr, "       point to the MinGW root directory.  For a Windows command\n");
+		fprintf(stderr, "       shell the option would be something like -mingw=c:\\mingw\n");
+		fprintf(stderr, "       For a Cygwin shell the option would be something like\n");
+		fprintf(stderr, "       -mingw=c:\\\\mingw\n");
+		myExit(5);
 	}
 }
 
@@ -480,68 +460,98 @@ static void checkIvlDir(char *root)
 
 #define IVL_REGKEY_MINGW "MingwDir"
 
-static void setup_mingw_environment()
+static void setup_mingw_environment(void)
 {
-	char *pOldPATH = getenv("PATH");		/* get current path */
+	char buffer[1];		/* doesn't matter how big this is, as we don't use the result */
+	char *path;
 
 	if (*gstr.pMINGW) {
 		checkMingwDir(gstr.pMINGW);
-		SetRegistryKey(IVL_REGKEY_MINGW,gstr.pMINGW);
-	} else if (!GetRegistryKey(IVL_REGKEY_MINGW,&gstr.pMINGW)) {
-			fprintf(stderr,"error: can not locate the MinGW root directory, use the -mingw option of\n");
-			fprintf(stderr,"       iverilog-vpi.exe to point to the MinGW root directory.  For\n");
-			fprintf(stderr,"       a Windows command shell the option would be something like\n");
-			fprintf(stderr,"       -mingw=c:\\mingw  For a Cygwin shell the option would be\n");
-			fprintf(stderr,"       something like -mingw=c:\\\\mingw\n");
-			myExit(5);
-		}
+		if (!*gstr.pOUT) SetRegistryKey(IVL_REGKEY_MINGW, gstr.pMINGW);
+
+	} else if (GetRegistryKey(IVL_REGKEY_MINGW, &gstr.pMINGW)) {
+		checkMingwDir(gstr.pMINGW);
+
+	} else if (SearchPath(NULL, gstr.pCCNAME, ".exe", sizeof(buffer), buffer, NULL)) {
+		return;
+
+	} else {
+		fprintf(stderr, "error: cannot locate the MinGW C compiler - either add its location\n");
+		fprintf(stderr, "       to the PATH environment variable or use the -mingw option of\n");
+		fprintf(stderr, "       iverilog-vpi.exe to point to the MinGW root directory.  For\n");
+		fprintf(stderr, "       a Windows command shell the option would be something like\n");
+		fprintf(stderr, "       -mingw=c:\\mingw  For a Cygwin shell the option would be\n");
+		fprintf(stderr, "       something like -mingw=c:\\\\mingw\n");
+		myExit(5);
+	}
 
 	  /* Create new path with MinGW in it */
-	assign(&gstr.pNewPath,"PATH=");
-	append(&gstr.pNewPath,gstr.pMINGW);
-	appendBackSlash(&gstr.pNewPath);
-	append(&gstr.pNewPath, "\\");
-	append(&gstr.pNewPath,"bin;");
-	append(&gstr.pNewPath,pOldPATH);
+	initDynString(&path);
+	assign(&path, "PATH=");
+	append(&path, gstr.pMINGW);
+	appendBackSlash(&path);
+	append(&path, "bin;");
+	append(&path, getenv("PATH"));
 
 	  /* Place new path in environment */
-	_putenv(gstr.pNewPath);
+	_putenv(path);
+	deInitDynString(path);
 }
 
-/* see if we can find iverilog root */
+/* find the iverilog root and initialise the compiler options */
 
-#define IVL_REGKEY_IVL   "InstallDir"
-
-static void setup_ivl_environment()
+static void setup_ivl_environment(void)
 {
-	if (*gstr.pIVL) {
-		checkIvlDir(gstr.pIVL);
-		SetRegistryKey(IVL_REGKEY_IVL,gstr.pIVL);
-	} else if (!GetRegistryKey(IVL_REGKEY_IVL,&gstr.pIVL)) {
-			fprintf(stderr,"error: can not locate the Icarus Verilog root directory, use the -ivl option\n");
-			fprintf(stderr,"       of iverilog-vpi" IVERILOG_SUFFIX " to point to the Icarus Verilog root directory.\n");
-			fprintf(stderr,"       For a Windows command shell the option would be something like\n");
-			fprintf(stderr,"       -ivl=c:\\iverilog  For a Cygwin shell the option would be something\n");
-			fprintf(stderr,"       like -ivl=c:\\\\iverilog\n");
-			myExit(6);
-		}
+	char path[4096];
+	char *ptr;
 
-	  /* Build up the CFLAGS option string */
-	assign(&gstr.pCFLAGS,IVERILOG_VPI_CFLAGS " -I\"");
-	append(&gstr.pCFLAGS,gstr.pIVL);
-	appendBackSlash(&gstr.pCFLAGS);
-	append(&gstr.pCFLAGS,"\\include\\\\iverilog\"" IVERILOG_SUFFIX);
+	  /* Extract the Icarus Verilog root directory from the path to the
+	     command. The command path will look something like this:
+
+		C:\iverilog\bin\iverilog-vpi.exe
+
+	     The corresponding root directory is
+
+		C:\iverilog
+
+	     so we chop off the file name and the last directory. */
+	GetModuleFileName(NULL, path, sizeof(path));
+	ptr = strrchr(path, '\\');
+	if (!ptr) {
+		fprintf(stderr, "error: couldn't find start of program name in command path '%s'\n", path);
+		myExit(6);
+	}
+	*ptr = 0;
+	ptr = strrchr(path, '\\');
+	if (!ptr) {
+		fprintf(stderr, "error: couldn't find start of bin directory in command path '%s'\n", path);
+		myExit(6);
+	}
+	*ptr = 0;
+	assign(&gstr.pIVL, path);
+
+	  /* Build up the CCFLAGS option string */
+	assign(&gstr.pCCFLAGS, IVERILOG_VPI_CFLAGS " -I\"");
+	append(&gstr.pCCFLAGS, gstr.pIVL);
+	appendBackSlash(&gstr.pCCFLAGS);
+	append(&gstr.pCCFLAGS, "include\\iverilog\"" IVERILOG_SUFFIX);
+
+	  /* Build up the CXFLAGS option string */
+	assign(&gstr.pCXFLAGS, IVERILOG_VPI_CXXFLAGS " -I\"");
+	append(&gstr.pCXFLAGS, gstr.pIVL);
+	appendBackSlash(&gstr.pCXFLAGS);
+	append(&gstr.pCXFLAGS, "include\\iverilog\"" IVERILOG_SUFFIX);
 
 	  /* Build up the LDFLAGS option string */
-	assign(&gstr.pLDLIBS,"-L\"");
-	append(&gstr.pLDLIBS,gstr.pIVL);
+	assign(&gstr.pLDLIBS, "-L\"");
+	append(&gstr.pLDLIBS, gstr.pIVL);
 	appendBackSlash(&gstr.pLDLIBS);
-	append(&gstr.pLDLIBS,"\\lib\" " IVERILOG_VPI_LDLIBS);
+	append(&gstr.pLDLIBS, "lib\" " IVERILOG_VPI_LDLIBS);
 }
 
 /* compile source modules */
 
-static void compile(char *pSource, char **pObject, int *compile_errors, char *compiler)
+static void compile(char *pSource, char *pFlags, char **pObject, int *compile_errors, char *compiler)
 {
 	char *ptr1 = pSource;
 	char *ptr2 = strchr(ptr1, ' ');
@@ -568,7 +578,7 @@ static void compile(char *pSource, char **pObject, int *compile_errors, char *co
 		append(&buf, " ");
 		append(&buf, gstr.pDEFS);
 		append(&buf, " ");
-		append(&buf, gstr.pCFLAGS);
+		append(&buf, pFlags);
 		append(&buf, " ");
 		append(&buf, gstr.pINCS);
 		append(&buf, " ");
@@ -593,27 +603,22 @@ static void compile(char *pSource, char **pObject, int *compile_errors, char *co
 
 /* using the global strings, compile and link */
 
-static void compile_and_link()
+static void compile_and_link(void)
 {
 	char *buf=0;
 	int iRet, compile_errors = 0;
 
 	  /* To make the output match iverilog-vpi.sh do not print out the
 	   * root directories */
-
 //	printf("MinGW root directory:  %s.\n", gstr.pMINGW);
-	checkMingwDir(gstr.pMINGW);
-
-//	printf("Icarus Verilog root directory:  %s.\n", gstr.pIVL);
-	checkIvlDir(gstr.pIVL);
 
 	  /* compile the C source files (*.c) */
-	compile(gstr.pCCSRC, &gstr.pOBJ, &compile_errors, IVERILOG_VPI_CC );
+	compile(gstr.pCCSRC, gstr.pCCFLAGS, &gstr.pOBJ, &compile_errors, IVERILOG_VPI_CC );
 	  /* compile the C++ source files (*.cc, *.cpp) */
-	compile(gstr.pCXSRC, &gstr.pOBJ, &compile_errors, IVERILOG_VPI_CXX);
+	compile(gstr.pCXSRC, gstr.pCXFLAGS, &gstr.pOBJ, &compile_errors, IVERILOG_VPI_CXX);
 
 	if (compile_errors) {
-		fprintf(stderr,"iverilog-vpi: %d file(s) failed to compile.\n",
+		fprintf(stderr, "iverilog-vpi: %d file(s) failed to compile.\n",
 		        compile_errors);
 		myExit(2);
 	}
@@ -644,7 +649,7 @@ int main(int argc, char *argv[])
 {
 	init();
 
-	if (!parse(argc,argv)) usage();
+	if (!parse(argc, argv)) usage();
 
 	setup_mingw_environment();
 	setup_ivl_environment();

driver/Makefile.in

@@ -85,10 +85,9 @@ iverilog@EXEEXT@: $O
 cflexor.c: $(srcdir)/cflexor.lex
 	$(LEX) -s -t $< > $@
 
-# Build this in two steps to avoid parallel build issues (see pr3462585)
-cfparse.c: $(srcdir)/cfparse.y
-	$(YACC) --verbose -t -p cf -d -o $@ $<
-cfparse.h: cfparse.c
+# Use pattern rules to avoid parallel build issues (see pr3462585)
+cfparse%c cfparse%h: $(srcdir)/cfparse%y
+	$(YACC) --verbose -t -p cf -d -o cfparse.c $<
 
 %.o: %.c
 	$(CC) $(CPPFLAGS) $(CFLAGS) @DEPENDENCY_FLAG@ -c $< -o $*.o

driver/cflexor.lex

@@ -4,7 +4,7 @@
 
 %{
 /*
- * Copyright (c) 2001-2014 Stephen Williams ([email protected])
+ * Copyright (c) 2001-2017 Stephen Williams ([email protected])
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -22,11 +22,13 @@
  *    Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  */
 
-# include  "cfparse.h"
 # include  "cfparse_misc.h"
+# include  "cfparse.h"
 # include  "globals.h"
 # include  <string.h>
 
+char *current_file = NULL;
+
 static int comment_enter;
 static char* trim_trailing_white(char*txt, int trim);
 
@@ -124,7 +126,8 @@ int cmdfile_stack_ptr = 0;
 "-c" { return TOK_Dc; }
 "-f" { return TOK_Dc; }
 
-  /* Notice the -v flag. */
+  /* Notice the -l or -v flag. */
+"-l" { return TOK_Dv; }
 "-v" { return TOK_Dv; }
 
   /* Notice the -y flag. */
@@ -251,7 +254,7 @@ void destroy_lexor(void)
 {
 # ifdef FLEX_SCANNER
 #   if YY_FLEX_MAJOR_VERSION >= 2 && YY_FLEX_MINOR_VERSION >= 5
-#     if defined(YY_FLEX_SUBMINOR_VERSION) && YY_FLEX_SUBMINOR_VERSION >= 9
+#     if YY_FLEX_MINOR_VERSION > 5 || defined(YY_FLEX_SUBMINOR_VERSION) && YY_FLEX_SUBMINOR_VERSION >= 9
     yylex_destroy();
 #     endif
 #   endif

driver/cfparse_misc.h

@@ -39,6 +39,6 @@ int cferror(const char *);
 int cfparse(void);
 void switch_to_command_file(const char *);
 void destroy_lexor(void);
-char *current_file;
+extern char *current_file;
 
 #endif /* IVL_cfparse_misc_H */

driver/iverilog.man.in

@@ -1,14 +1,15 @@
-.TH iverilog 1 "February 26th, 2014" "" "Version %M.%m.%n %E"
+.TH iverilog 1 "Oct 2nd, 2016" "" "Version %M.%n%E"
 .SH NAME
 iverilog - Icarus Verilog compiler
 
 .SH SYNOPSIS
 .B iverilog
 [\-ESVv] [\-Bpath] [\-ccmdfile|\-fcmdfile] [\-Dmacro[=defn]]
-[\-Pparameter=value] [\-pflag=value]
-[\-dname] [\-g1995|\-g2001|\-g2005|\-g2005-sv|\-g2009|\-g2012|\-g<feature>]
+[\-Pparameter=value] [\-pflag=value] [\-dname]
+[\-g1995\:|\-g2001\:|\-g2005\:|\-g2005-sv\:|\-g2009\:|\-g2012\:|\-g<feature>]
 [\-Iincludedir] [\-mmodule] [\-M[mode=]file] [\-Nfile] [\-ooutputfilename]
-[\-stopmodule] [\-ttype] [\-Tmin/typ/max] [\-Wclass] [\-ypath] sourcefile
+[\-stopmodule] [\-ttype] [\-Tmin/typ/max] [\-Wclass] [\-ypath] [\-lfile]
+sourcefile
 
 .SH DESCRIPTION
 .PP
@@ -30,7 +31,7 @@ different set of programs. The path given is used to locate
 \fIivlpp\fP, \fIivl\fP, code generators and the VPI modules.
 .TP 8
 .B -c\fIfile\fP -f\fIfile\fP
-These flags specifies an input file that contains a list of Verilog
+These flags specify an input file that contains a list of Verilog
 source files. This is similar to the \fIcommand file\fP of other
 Verilog simulators, in that it is a file that contains the file names
 instead of taking them on the command line. See \fBCommand Files\fP below.
@@ -133,12 +134,29 @@ parameter assignment is evaluated as a lossless expression, as is any
 expression containing an unsized constant number, and unsized constant
 numbers are not truncated to integer width.
 .TP 8
+.B -gshared-loop-index\fI|\fP-gno-shared-loop-index
+Enable or disable (default) the exclusion of for-loop control variables
+from implicit event_expression lists. When enabled, if a for-loop control
+variable (loop index) is only used inside the for-loop statement, the
+compiler will not include it in an implicit event_expression list it
+calculates for that statement or any enclosing statement. This allows
+the same control variable to be used in multiple processes without risk
+of entering an infinite loop caused by each process triggering all other
+processes that use the same varaible. For strict compliance with the
+standards, this behaviour should be disabled.
+.TP 8
 .B -I\fIincludedir\fP
 Append directory \fIincludedir\fP to list of directories searched
 for Verilog include files. The \fB\-I\fP switch may be used many times
 to specify several directories to search, the directories are searched
 in the order they appear on the command line.
 .TP 8
+.B -l\fIfile\fP
+Add the specified file to the list of source files to be compiled,
+but mark it as a library file. All modules contained within that
+file will be treated as library modules, and only elaborated if
+they are instantiated by other modules in the design.
+.TP 8
 .B -M\fIpath\fP
 This is equivalent to \fB\-Mall=path\fP. Preserved for backwards
 compatibility.
@@ -213,6 +231,12 @@ a reference to a key temporary file that passes information to the
 compiler proper.  To keep that file from being deleted at the end
 of the process, provide a file name of your own in the environment
 variable \fBIVERILOG_ICONFIG\fP.
+
+If the selected target is \fIvvp\fP, the \fB\-v\fP switch is appended
+to the shebang line in the compiler output file, so directly executing
+the compiler output file will turn on verbose messages in \fIvvp\fP.
+This extra verbosity can be avoided by using the \fIvvp\fP command to
+indirectly execute the compiler output file.
 .TP 8
 .B -V
 Print the version of the compiler, and exit.
@@ -240,7 +264,7 @@ that contain Verilog source files.  During elaboration, the compiler
 notices the instantiation of undefined module types. If the user
 specifies library search directories, the compiler will search the
 directory for files with the name of the missing module type. If it
-finds such a file, it loads it as a Verilog source file, they tries
+finds such a file, it loads it as a Verilog source file, then tries
 again to elaborate the module.
 
 Library module files should contain only a single module, but this is
@@ -286,8 +310,13 @@ after a \fB\-Wall\fP argument to suppress isolated warning types.
 
 .TP 8
 .B all
-This enables the implicit, portbind, select\-range, timescale, and
-sensitivity\-entire\-array warning categories.
+This enables the anachronisms, implicit, portbind, select\-range,
+timescale, and sensitivity\-entire\-array warning categories.
+
+.TP 8
+.B anachronisms
+This enables warnings for use of features that have been deprecated
+or removed in the selected generation of the Verilog language.
 
 .TP 8
 .B implicit
@@ -391,6 +420,15 @@ A \fB\-c\fP or \fB\-f\fP token prefixes a command file, exactly like it
 does on the command line. The cmdfile may be on the same line or the
 next non-comment line.
 
+.TP 8
+.B -l\ \fIfile\fP -v\ \fIfile\fP
+A \fB\-l\fP token prefixes a library file in the command file,
+exactly like it does on the command line. The parameter to the \fB\-l\fP
+flag may be on the same line or the next non-comment line. \fB\-v\fP is
+an alias for \fB\-l\fP, provided for compatibility with other simulators.
+
+Variables in the \fIfile\fP are substituted.
+
 .TP 8
 .B -y\ \fIlibdir\fP
 A \fB\-y\fP token prefixes a library directory in the command file,
@@ -514,14 +552,14 @@ Steve Williams ([email protected])
 
 .SH SEE ALSO
 vvp(1),
-.BR "<http://www.icarus.com/eda/verilog/>"
+.BR "<http://iverilog.icarus.com/>"
 
 Tips on using, debugging, and developing the compiler can be found at
 .BR "<http://iverilog.wikia.com/>"
 
 .SH COPYRIGHT
 .nf
-Copyright \(co  2002\-2014 Stephen Williams
+Copyright \(co  2002\-2016 Stephen Williams
 
 This document can be freely redistributed according to the terms of the
 GNU General Public License version 2.0

driver/main.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2000-2014 Stephen Williams (steve@icarus.com)
+ * Copyright (c) 2000-2016 Stephen Williams (steve@icarus.com)
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -44,7 +44,7 @@ const char HELP[] =
 "                [-M [mode=]depfile] [-m module]\n"
 "                [-N file] [-o filename] [-p flag=value]\n"
 "                [-s topmodule] [-t target] [-T min|typ|max]\n"
-"                [-W class] [-y dir] [-Y suf] source_file(s)\n"
+"                [-W class] [-y dir] [-Y suf] [-l file] source_file(s)\n"
 "\n"
 "See the man page for details.";
 
@@ -128,6 +128,7 @@ const char*gen_icarus = "icarus-misc";
 const char*gen_io_range_error = "io-range-error";
 const char*gen_strict_ca_eval = "no-strict-ca-eval";
 const char*gen_strict_expr_width = "no-strict-expr-width";
+const char*gen_shared_loop_index = "no-shared-loop-index";
 const char*gen_verilog_ams = "no-verilog-ams";
 
 /* Boolean: true means use a default include dir, false means don't */
@@ -496,6 +497,7 @@ static void process_warning_switch(const char*name)
       if (strcmp(name,"all") == 0) {
 	    process_warning_switch("anachronisms");
 	    process_warning_switch("implicit");
+	    process_warning_switch("implicit-dimensions");
 	    process_warning_switch("portbind");
 	    process_warning_switch("select-range");
 	    process_warning_switch("timescale");
@@ -506,6 +508,9 @@ static void process_warning_switch(const char*name)
       } else if (strcmp(name,"implicit") == 0) {
 	    if (! strchr(warning_flags, 'i'))
 		  strcat(warning_flags, "i");
+      } else if (strcmp(name,"implicit-dimensions") == 0) {
+	    if (! strchr(warning_flags, 'd'))
+		  strcat(warning_flags, "d");
       } else if (strcmp(name,"portbind") == 0) {
 	    if (! strchr(warning_flags, 'p'))
 		  strcat(warning_flags, "p");
@@ -538,6 +543,12 @@ static void process_warning_switch(const char*name)
 		  cp[0] = cp[1];
 		  cp += 1;
 	    }
+      } else if (strcmp(name,"no-implicit-dimensions") == 0) {
+	    char*cp = strchr(warning_flags, 'd');
+	    if (cp) while (*cp) {
+		  cp[0] = cp[1];
+		  cp += 1;
+	    }
       } else if (strcmp(name,"no-portbind") == 0) {
 	    char*cp = strchr(warning_flags, 'p');
 	    if (cp) while (*cp) {
@@ -568,6 +579,9 @@ static void process_warning_switch(const char*name)
 		  cp[0] = cp[1];
 		  cp += 1;
 	    }
+      } else {
+	    fprintf(stderr, "Ignoring unknown warning class "
+		    "%s\n", name);
       }
 }
 
@@ -717,6 +731,12 @@ static int process_generation(const char*name)
       else if (strcmp(name,"no-strict-expr-width") == 0)
 	    gen_strict_expr_width = "no-strict-expr-width";
 
+      else if (strcmp(name,"shared-loop-index") == 0)
+	    gen_shared_loop_index = "shared-loop-index";
+
+      else if (strcmp(name,"no-shared-loop-index") == 0)
+	    gen_shared_loop_index = "no-shared-loop-index";
+
       else if (strcmp(name,"verilog-ams") == 0)
 	    gen_verilog_ams = "verilog-ams";
 
@@ -742,7 +762,8 @@ static int process_generation(const char*name)
 		            "    icarus-misc | no-icarus-misc\n"
 		            "    io-range-error | no-io-range-error\n"
 		            "    strict-ca-eval | no-strict-ca-eval\n"
-		            "    strict-expr-width | no-strict-expr-width\n");
+		            "    strict-expr-width | no-strict-expr-width\n"
+		            "    shared-loop-index | no-shared-loop-index\n");
 
 	    return 1;
       }
@@ -907,7 +928,7 @@ int main(int argc, char **argv)
 	}
       }
 
-      while ((opt = getopt(argc, argv, "B:c:D:d:Ef:g:hI:M:m:N:o:P:p:Ss:T:t:vVW:y:Y:")) != EOF) {
+      while ((opt = getopt(argc, argv, "B:c:D:d:Ef:g:hl:I:M:m:N:o:P:p:Ss:T:t:vVW:y:Y:")) != EOF) {
 
 	    switch (opt) {
 		case 'B':
@@ -962,6 +983,10 @@ int main(int argc, char **argv)
 		  process_include_dir(optarg);
 		  break;
 
+		case 'l':
+		  process_file_name(optarg, 1);
+		  break;
+
 		case 'M':
 		  if (process_depfile(optarg) != 0)
                         return -1;
@@ -1044,19 +1069,10 @@ int main(int argc, char **argv)
 
       if (version_flag || verbose_flag) {
 	    printf("Icarus Verilog version " VERSION " (" VERSION_TAG ")\n\n");
-	    printf("Copyright 1998-2013 Stephen Williams\n\n");
+	    printf("Copyright 1998-2015 Stephen Williams\n\n");
 	    puts(NOTICE);
       }
 
-      if (synth_flag) {
-	    fprintf(stderr, "Warning: Synthesis is not currently being "
-	                    "maintained and may not\n");
-	    fprintf(stderr, "         function correctly. V0.8 was the "
-	                    "last release branch to\n");
-	    fprintf(stderr, "         have active synthesis development "
-	                    "and support!\n");
-      }
-
 	/* Make a common conf file path to reflect the target. */
       snprintf(iconfig_common_path, sizeof iconfig_common_path, "%s%c%s%s.conf",
 	      base, sep, targ, synth_flag? "-s" : "");
@@ -1104,6 +1120,7 @@ int main(int argc, char **argv)
       fprintf(iconfig_file, "generation:%s\n", gen_io_range_error);
       fprintf(iconfig_file, "generation:%s\n", gen_strict_ca_eval);
       fprintf(iconfig_file, "generation:%s\n", gen_strict_expr_width);
+      fprintf(iconfig_file, "generation:%s\n", gen_shared_loop_index);
       fprintf(iconfig_file, "generation:%s\n", gen_verilog_ams);
       fprintf(iconfig_file, "generation:%s\n", gen_icarus);
       fprintf(iconfig_file, "warnings:%s\n", warning_flags);

elab_expr.cc

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 1999-2014 Stephen Williams (steve@icarus.com)
+ * Copyright (c) 1999-2018 Stephen Williams (steve@icarus.com)
  * Copyright CERN 2013 / Stephen Williams (steve@icarus.com)
  *
  *    This source code is free software; you can redistribute it
@@ -90,7 +90,7 @@ static NetBranch* find_existing_implicit_branch(NetNet*sig, NetNet*gnd)
 
 NetExpr* elaborate_rval_expr(Design*des, NetScope*scope, ivl_type_t lv_net_type,
 			     ivl_variable_type_t lv_type, unsigned lv_width,
-			     PExpr*expr, bool need_const)
+			     PExpr*expr, bool need_const, bool force_unsigned)
 {
       if (debug_elaborate) {
 	    cerr << expr->get_fileline() << ": elaborate_rval_expr: "
@@ -135,7 +135,7 @@ NetExpr* elaborate_rval_expr(Design*des, NetScope*scope, ivl_type_t lv_net_type,
       }
 
       return elab_and_eval(des, scope, expr, context_wid, need_const,
-			   false, lv_type);
+			   false, lv_type, force_unsigned);
 }
 
 /*
@@ -657,10 +657,13 @@ NetExpr* PEBComp::elaborate_expr(Design*des, NetScope*scope,
 	  case 'E': /* === */
 	  case 'N': /* !== */
 	    if (lp->expr_type() == IVL_VT_REAL ||
-		rp->expr_type() == IVL_VT_REAL) {
+		lp->expr_type() == IVL_VT_STRING ||
+		rp->expr_type() == IVL_VT_REAL ||
+		rp->expr_type() == IVL_VT_STRING) {
 		  cerr << get_fileline() << ": error: "
 		       << human_readable_op(op_)
-		       << " operator may not have REAL operands." << endl;
+		       << " operator may not have REAL or STRING operands."
+		       << endl;
 		  des->errors += 1;
 		  return 0;
 	    }
@@ -672,10 +675,7 @@ NetExpr* PEBComp::elaborate_expr(Design*des, NetScope*scope,
       NetExpr*tmp = new NetEBComp(op_, lp, rp);
       tmp->set_line(*this);
 
-      tmp = pad_to_width(tmp, expr_wid, *this);
-      tmp->cast_signed(signed_flag_);
-
-      return tmp;
+      return pad_to_width(tmp, expr_wid, signed_flag_, *this);
 }
 
 unsigned PEBLogic::test_width(Design*, NetScope*, width_mode_t&)
@@ -713,10 +713,7 @@ NetExpr*PEBLogic::elaborate_expr(Design*des, NetScope*scope,
       NetExpr*tmp = new NetEBLogic(op_, lp, rp);
       tmp->set_line(*this);
 
-      tmp = pad_to_width(tmp, expr_wid, *this);
-      tmp->cast_signed(signed_flag_);
-
-      return tmp;
+      return pad_to_width(tmp, expr_wid, signed_flag_, *this);
 }
 
 unsigned PEBLeftWidth::test_width(Design*des, NetScope*scope, width_mode_t&mode)
@@ -785,8 +782,11 @@ unsigned PEBLeftWidth::test_width(Design*des, NetScope*scope, width_mode_t&mode)
                   right_ = tmp;
             }
             NetEConst*rc = dynamic_cast<NetEConst*> (rp);
-            if (rc && (r_width < sizeof(long)*8))
-                  r_val = rc->value().as_long();
+	      // Adjust the expression width that can be converter depending
+	      // on if the R-value is signed or not.
+	    unsigned c_width = sizeof(long)*8;
+	    if (! right_->has_sign()) c_width -= 1;
+	    if (rc && (r_width <= c_width)) r_val = rc->value().as_long();
 
 	    if (debug_elaborate && rc) {
 		  cerr << get_fileline() << ": PEBLeftWidth::test_width: "
@@ -1029,8 +1029,7 @@ NetExpr*PEBShift::elaborate_expr_leaf(Design*des, NetExpr*lp, NetExpr*rp,
 		  tmp->set_line(*this);
 		  tmp = new NetESelect(lp, tmp, 1);
 		  tmp->set_line(*this);
-		  tmp = pad_to_width(tmp, expr_wid, *this);
-		  tmp->cast_signed(true);
+		  tmp = pad_to_width(tmp, expr_wid, true, *this);
 
                   delete rp;
 		  return tmp;
@@ -1072,7 +1071,7 @@ unsigned PECallFunction::test_width_sfunc_(Design*des, NetScope*scope,
 
       if (name=="$ivlh_to_unsigned") {
 	    ivl_assert(*this, parms_.size() == 2);
-	      // The Icarus Verilog specific $ivl_unsigned() system
+	      // The Icarus Verilog specific $ivlh_to_unsigned() system
 	      // task takes a second argument which is the output
 	      // size. This can be an arbitrary constant function.
 	    PExpr*pexpr = parms_[1];
@@ -1094,6 +1093,11 @@ unsigned PECallFunction::test_width_sfunc_(Design*des, NetScope*scope,
 	    bool rc = eval_as_long(value, nexpr);
 	    ivl_assert(*this, rc && value>=0);
 
+	      // The argument type/width is self-determined and doesn't
+	      // affect the result type/width.
+	    width_mode_t arg_mode = SIZED;
+	    parms_[0]->test_width(des, scope, arg_mode);
+
 	    expr_width_ = value;
 	    signed_flag_= false;
 	    return expr_width_;
@@ -1324,7 +1328,7 @@ unsigned PECallFunction::test_width_method_(Design*des, NetScope*scope,
       if (net == 0)
 	    return 0;
 
-	// Look fonr built in string attributes.
+	// Look for built in string attributes.
       if (net->data_type()==IVL_VT_STRING) {
 
 	    if (method_name == "len") {
@@ -1403,23 +1407,7 @@ NetExpr*PECallFunction::cast_to_width_(NetExpr*expr, unsigned wid) const
 		 << " from expr_width()=" << expr->expr_width() << endl;
       }
 
-        /* If the expression is a const, then replace it with a new
-           const. This is a more efficient result. */
-      if (NetEConst*tmp = dynamic_cast<NetEConst*>(expr)) {
-            tmp->cast_signed(signed_flag_);
-            if (wid > tmp->expr_width()) {
-                  tmp = new NetEConst(verinum(tmp->value(), wid));
-                  tmp->set_line(*this);
-                  delete expr;
-            }
-            return tmp;
-      }
-
-      NetESelect*tmp = new NetESelect(expr, 0, wid);
-      tmp->cast_signed(signed_flag_);
-      tmp->set_line(*this);
-
-      return tmp;
+      return cast_to_width(expr, wid, signed_flag_, *this);
 }
 
 /*
@@ -1443,7 +1431,7 @@ NetExpr* PECallFunction::elaborate_sfunc_(Design*des, NetScope*scope,
 
 	    PExpr*expr = parms_[0];
 	    ivl_assert(*this, expr);
-	    NetExpr*sub = expr->elaborate_expr(des, scope, expr_width_, flags);
+	    NetExpr*sub = expr->elaborate_expr(des, scope, expr->expr_width(), flags);
 	    return cast_to_width_(sub, expr_wid);
       }
 
@@ -1458,6 +1446,13 @@ NetExpr* PECallFunction::elaborate_sfunc_(Design*des, NetScope*scope,
 		  return 0;
 	    }
 
+            if (!type_is_vectorable(expr_type_)) {
+	          cerr << get_fileline() << ": error: The argument to "
+		       << name << " must be a vector type." << endl;
+	          des->errors += 1;
+	          return 0;
+            }
+
 	    if (debug_elaborate) {
 		  cerr << get_fileline() << ": PECallFunction::elaborate_sfunc_: "
 		       << name << " expression is the argument cast to expr_wid=" << expr_wid << endl;
@@ -1586,7 +1581,7 @@ NetExpr* PECallFunction::elaborate_sfunc_(Design*des, NetScope*scope,
 
       if (missing_parms > 0) {
 	    cerr << get_fileline() << ": error: The function " << name
-		 << " has been called with empty parameters." << endl;
+		 << " has been called with missing/empty parameters." << endl;
 	    cerr << get_fileline() << ":      : Verilog doesn't allow "
 		 << "passing empty parameters to functions." << endl;
 	    des->errors += 1;
@@ -1595,10 +1590,7 @@ NetExpr* PECallFunction::elaborate_sfunc_(Design*des, NetScope*scope,
       if (missing_parms || parm_errors)
             return 0;
 
-      NetExpr*tmp = pad_to_width(fun, expr_wid, *this);
-      tmp->cast_signed(signed_flag_);
-
-      return tmp;
+      return pad_to_width(fun, expr_wid, signed_flag_, *this);
 }
 
 NetExpr* PECallFunction::elaborate_access_func_(Design*des, NetScope*scope,
@@ -1655,10 +1647,7 @@ NetExpr* PECallFunction::elaborate_access_func_(Design*des, NetScope*scope,
       NetExpr*tmp = new NetEAccess(branch, nature);
       tmp->set_line(*this);
 
-      tmp = pad_to_width(tmp, expr_wid, *this);
-      tmp->cast_signed(signed_flag_);
-
-      return tmp;
+      return pad_to_width(tmp, expr_wid, signed_flag_, *this);
 }
 
 /*
@@ -2178,6 +2167,14 @@ NetExpr* PECallFunction::elaborate_expr(Design*des, NetScope*scope,
       return elaborate_base_(des, scope, dscope, expr_wid, flags);
 }
 
+NetExpr* PECallFunction::elaborate_expr(Design*des, NetScope*scope,
+					ivl_type_t type, unsigned flags) const
+{
+      const netdarray_t*darray = dynamic_cast<const netdarray_t*>(type);
+      assert(darray);
+      return elaborate_expr(des, scope, darray->element_type()->packed_width(), flags);
+}
+
 NetExpr* PECallFunction::elaborate_base_(Design*des, NetScope*scope, NetScope*dscope,
 					 unsigned expr_wid, unsigned flags) const
 {
@@ -2259,10 +2256,10 @@ NetExpr* PECallFunction::elaborate_base_(Design*des, NetScope*scope, NetScope*ds
 	    NetEUFunc*func = new NetEUFunc(scope, dscope, eres, parms, need_const);
 	    func->set_line(*this);
 
-            NetExpr*tmp = pad_to_width(func, expr_wid, *this);
-            tmp->cast_signed(signed_flag_);
+	    if(res->darray_type())
+	        return func;
 
-	    return tmp;
+            return pad_to_width(func, expr_wid, signed_flag_, *this);
       }
 
       cerr << get_fileline() << ": internal error: Unable to locate "
@@ -2291,6 +2288,21 @@ unsigned PECallFunction::elaborate_arguments_(Design*des, NetScope*scope,
       const unsigned parm_count = parms.size() - parm_off;
       const unsigned actual_count = parms_.size();
 
+	/* The parser can't distinguish between a function call with
+	   no arguments and a function call with one empty argument,
+	   and always supplies one empty argument. Handle the no
+	   argument case here. */
+      if ((parm_count == 0) && (actual_count == 1) && (parms_[0] == 0))
+	    return 0;
+
+      if (actual_count > parm_count) {
+	    cerr << get_fileline() << ": error: "
+		 << "Too many arguments (" << actual_count
+		 << ", expecting " << parm_count << ")"
+		 << " in call to function." << endl;
+	    des->errors += 1;
+      }
+
       for (unsigned idx = 0 ; idx < parm_count ; idx += 1) {
 	    unsigned pidx = idx + parm_off;
 	    PExpr*tmp = (idx < actual_count) ? parms_[idx] : 0;
@@ -2321,7 +2333,7 @@ unsigned PECallFunction::elaborate_arguments_(Design*des, NetScope*scope,
 	    } else if (def->port_defe(pidx)) {
 		  if (! gn_system_verilog()) {
 			cerr << get_fileline() << ": internal error: "
-			     <<"Found (and using) default function argument "
+			     << "Found (and using) default function argument "
 			     << "requires SystemVerilog." << endl;
 			des->errors += 1;
 		  }
@@ -2335,7 +2347,7 @@ unsigned PECallFunction::elaborate_arguments_(Design*des, NetScope*scope,
 
       if (missing_parms > 0) {
 	    cerr << get_fileline() << ": error: The function " << path_
-		 << " has been called with empty parameters." << endl;
+		 << " has been called with missing/empty parameters." << endl;
 	    cerr << get_fileline() << ":      : Verilog doesn't allow "
 		 << "passing empty parameters to functions." << endl;
 	    parm_errors += 1;
@@ -2484,103 +2496,185 @@ NetExpr* PECallFunction::elaborate_expr_method_(Design*des, NetScope*scope,
 
 unsigned PECastSize::test_width(Design*des, NetScope*scope, width_mode_t&)
 {
-      expr_width_ = size_;
+      ivl_assert(*this, size_);
+      ivl_assert(*this, base_);
+
+      NetExpr*size_ex = elab_and_eval(des, scope, size_, -1, true);
+      NetEConst*size_ce = dynamic_cast<NetEConst*>(size_ex);
+      expr_width_ = size_ce ? size_ce->value().as_ulong() : 0;
+      delete size_ex;
+      if (expr_width_ == 0) {
+	    cerr << get_fileline() << ": error: Cast size expression "
+		    "must be constant and greater than zero." << endl;
+	    des->errors += 1;
+	    return 0;
+      }
 
       width_mode_t tmp_mode = PExpr::SIZED;
       base_->test_width(des, scope, tmp_mode);
 
-      return size_;
+      if (!type_is_vectorable(base_->expr_type())) {
+	    cerr << get_fileline() << ": error: Cast base expression "
+		    "must be a vector type." << endl;
+	    des->errors += 1;
+	    return 0;
+      }
+
+      expr_type_   = base_->expr_type();
+      min_width_   = expr_width_;
+      signed_flag_ = base_->has_sign();
+
+      return expr_width_;
 }
 
 NetExpr* PECastSize::elaborate_expr(Design*des, NetScope*scope,
-				    unsigned, unsigned) const
+				    unsigned expr_wid, unsigned flags) const
 {
-      NetExpr*sub = base_->elaborate_expr(des, scope, base_->expr_width(), NO_FLAGS);
-      NetESelect*sel = new NetESelect(sub, 0, size_);
-      sel->set_line(*this);
+      flags &= ~SYS_TASK_ARG; // don't propagate the SYS_TASK_ARG flag
 
-      return sel;
+      ivl_assert(*this, size_);
+      ivl_assert(*this, base_);
+
+	// A cast behaves exactly like an assignment to a temporary variable,
+	// so the temporary result size may affect the sub-expression width.
+      unsigned cast_width = base_->expr_width();
+      if (cast_width < expr_width_)
+            cast_width = expr_width_;
+
+      NetExpr*sub = base_->elaborate_expr(des, scope, cast_width, flags);
+      if (sub == 0)
+	    return 0;
+
+	// Perform the cast. The extension method (zero/sign), if needed,
+	// depends on the type of the base expression.
+      NetExpr*tmp = cast_to_width(sub, expr_width_, base_->has_sign(), *this);
+
+	// Pad up to the expression width. The extension method (zero/sign)
+	// depends on the type of enclosing expression.
+      return pad_to_width(tmp, expr_wid, signed_flag_, *this);
 }
 
-unsigned PECastType::test_width(Design*des, NetScope*scope, width_mode_t&wid)
+unsigned PECastType::test_width(Design*des, NetScope*scope, width_mode_t&)
 {
       ivl_type_t t = target_->elaborate_type(des, scope);
-      base_->test_width(des, scope, wid);
 
-      if(const netdarray_t*use_darray = dynamic_cast<const netdarray_t*> (t)) {
+      width_mode_t tmp_mode = PExpr::SIZED;
+      base_->test_width(des, scope, tmp_mode);
+
+      if (const netdarray_t*use_darray = dynamic_cast<const netdarray_t*>(t)) {
 	    expr_type_  = use_darray->element_base_type();
 	    expr_width_ = use_darray->element_width();
-      }
 
-      else if(const netstring_t*use_string = dynamic_cast<const netstring_t*> (t)) {
+      } else if (const netstring_t*use_string = dynamic_cast<const netstring_t*>(t)) {
 	    expr_type_  = use_string->base_type();
 	    expr_width_ = 8;
-      }
 
-      else {
+      } else {
 	    expr_type_  = t->base_type();
 	    expr_width_ = t->packed_width();
       }
+      min_width_   = expr_width_;
+      signed_flag_ = t->get_signed();
 
-      signed_flag_= t->get_signed();
-      min_width_ = expr_width_;
       return expr_width_;
 }
 
 NetExpr* PECastType::elaborate_expr(Design*des, NetScope*scope,
-				    unsigned, unsigned) const
+                                    ivl_type_t type, unsigned flags) const
 {
-      NetExpr*expr = base_->elaborate_expr(des, scope, base_->expr_width(), NO_FLAGS);
+    const netdarray_t*darray = NULL;
+    const netvector_t*vector = NULL;
 
-      if(dynamic_cast<const real_type_t*>(target_)) {
-          return cast_to_real(expr);
+    // Casting array of vectors to dynamic array type
+    if((darray = dynamic_cast<const netdarray_t*>(type)) &&
+            (vector = dynamic_cast<const netvector_t*>(darray->element_type()))) {
+        PExpr::width_mode_t mode = PExpr::SIZED;
+        unsigned use_wid = base_->test_width(des, scope, mode);
+        NetExpr*base = base_->elaborate_expr(des, scope, use_wid, NO_FLAGS);
+
+        assert(vector->packed_width() > 0);
+        assert(base->expr_width() > 0);
+
+        // Find rounded up length that can fit the whole casted array of vectors
+        int len = base->expr_width() + vector->packed_width() - 1;
+        if(base->expr_width() > (unsigned)vector->packed_width()) {
+            len /= vector->packed_width();
+        } else {
+            len /= base->expr_width();
+        }
+
+        // Number of words in the created dynamic array
+        NetEConst*len_expr = new NetEConst(verinum(len));
+        return new NetENew(type, len_expr, base);
+    }
+
+    // Fallback
+    return elaborate_expr(des, scope, (unsigned) 0, flags);
+}
+
+NetExpr* PECastType::elaborate_expr(Design*des, NetScope*scope,
+				    unsigned expr_wid, unsigned flags) const
+{
+      flags &= ~SYS_TASK_ARG; // don't propagate the SYS_TASK_ARG flag
+
+	// A cast behaves exactly like an assignment to a temporary variable,
+	// so the temporary result size may affect the sub-expression width.
+      unsigned cast_width = base_->expr_width();
+      if (type_is_vectorable(base_->expr_type()) && (cast_width < expr_width_))
+	    cast_width = expr_width_;
+
+      NetExpr*sub = base_->elaborate_expr(des, scope, cast_width, flags);
+      if (sub == 0)
+	    return 0;
+
+      if (dynamic_cast<const real_type_t*>(target_)) {
+	    return cast_to_real(sub);
       }
 
-      if(const atom2_type_t*atom = dynamic_cast<const atom2_type_t*>(target_)) {
-          if(base_->expr_width() > expr_width_) {
-              cerr << get_fileline() << ": cast type is not wide enough to store the result." << endl;
-              ivl_assert(*this, 0);
-          }
+      NetExpr*tmp = 0;
+      if (dynamic_cast<const atom2_type_t*>(target_)) {
+	    tmp = cast_to_int2(sub, expr_width_);
+      }
+      if (const vector_type_t*vec = dynamic_cast<const vector_type_t*>(target_)) {
+	    switch (vec->base_type) {
+		case IVL_VT_BOOL:
+		  tmp = cast_to_int2(sub, expr_width_);
+		  break;
 
-          if(base_->has_sign() != atom->signed_flag) {
-              cerr << get_fileline() << ": cast type and subject differ in signedness." << endl;
-              ivl_assert(*this, 0);
-          }
+		case IVL_VT_LOGIC:
+		  tmp = cast_to_int4(sub, expr_width_);
+		  break;
 
-          // That is how you both resize & cast to integers
-          return new NetECast('2', expr, expr_width_, expr->has_sign());
+		default:
+		  break;
+	    }
+      }
+      if (tmp) {
+            if (!type_is_vectorable(tmp->expr_type()))
+                  return tmp;
+
+	    if (tmp == sub) {
+		    // We already had the correct base type, so we just need to
+		    // fix the size. Note that even if the size is already correct,
+                    // we still need to isolate the sub-expression from changes in
+                    // the signedness pushed down from the main expression.
+		  tmp = cast_to_width(sub, expr_width_, sub->has_sign(), *this);
+	    }
+	    return pad_to_width(tmp, expr_wid, signed_flag_, *this);
       }
 
-      if(const vector_type_t*vec = dynamic_cast<const vector_type_t*>(target_)) {
-          switch(vec->base_type) {
-              case IVL_VT_BOOL:
-                  return cast_to_int2(expr, expr_width_);
+      if (dynamic_cast<const string_type_t*>(target_)) {
+	    if (base_->expr_type() == IVL_VT_STRING)
+		  return sub; // no conversion
 
-              case IVL_VT_LOGIC:
-                  return cast_to_int4(expr, expr_width_);
-
-              default:
-                  break;  /* Suppress warnings */
-          }
+	    if (base_->expr_type() == IVL_VT_LOGIC
+	     || base_->expr_type() == IVL_VT_BOOL)
+		  return sub; // handled by the target as special cases
       }
 
-      else if(dynamic_cast<const string_type_t*>(target_)) {
-          if(base_->expr_type() == IVL_VT_STRING)
-            return expr;        // no conversion
-
-          if((base_->expr_type() != IVL_VT_BOOL) &&
-                (base_->expr_type() != IVL_VT_LOGIC)) {
-              cerr << get_fileline() << ": cannot be casted to string." << endl;
-              ivl_assert(*this, false);
-          }
-
-          return expr;
-      }
-
-      cerr << get_fileline() << ": sorry: I don't know how to cast expression." << endl;
-      ivl_assert(*this, false);
-
-      return expr;
+      cerr << get_fileline() << ": sorry: This cast operation is not yet supported." << endl;
+      des->errors += 1;
+      return 0;
 }
 
 unsigned PEConcat::test_width(Design*des, NetScope*scope, width_mode_t&)
@@ -2678,6 +2772,7 @@ NetExpr* PEConcat::elaborate_expr(Design*, NetScope*,
 		  tmp->set_line(*this);
 		  return tmp;
 	    }
+	    // fallthrough
 	  default:
 	    cerr << get_fileline() << ": internal error: "
 		 << "I don't know how to elaborate(ivl_type_t)"
@@ -2772,17 +2867,16 @@ NetExpr* PEConcat::elaborate_expr(Design*des, NetScope*scope,
 	    concat->set(idx, parms[off+idx]);
       }
 
-      if (wid_sum == 0 && concat_depth < 2) {
-	    cerr << get_fileline() << ": error: Concatenation may not "
-	         << "have zero width in this context." << endl;
+      if (wid_sum == 0) {
+	    cerr << get_fileline() << ": error: Concatenation/replication "
+	         << "may not have zero width in this context." << endl;
 	    des->errors += 1;
 	    concat_depth -= 1;
 	    delete concat;
 	    return 0;
       }
 
-      NetExpr*tmp = pad_to_width(concat, expr_wid, *this);
-      tmp->cast_signed(signed_flag_);
+      NetExpr*tmp = pad_to_width(concat, expr_wid, signed_flag_, *this);
 
       concat_depth -= 1;
       return tmp;
@@ -2822,6 +2916,25 @@ NetExpr* PEFNumber::elaborate_expr(Design*, NetScope*, unsigned, unsigned) const
 bool PEIdent::calculate_packed_indices_(Design*des, NetScope*scope, NetNet*net,
 					list<long>&prefix_indices) const
 {
+      unsigned dimensions = net->unpacked_dimensions() + net->packed_dimensions();
+      switch (net->data_type()) {
+	  case IVL_VT_STRING:
+	  case IVL_VT_DARRAY:
+	  case IVL_VT_QUEUE:
+	    dimensions += 1;
+	  default:
+	    break;
+      }
+      if (path_.back().index.size() > dimensions) {
+	    cerr << get_fileline() << ": error: the number of indices ("
+		 << path_.back().index.size()
+		 << ") is greater than the number of dimensions ("
+		 << dimensions
+		 << ")." << endl;
+	    des->errors += 1;
+	    return false;
+      }
+
       list<index_component_t> index;
       index = path_.back().index;
       ivl_assert(*this, index.size() >= net->unpacked_dimensions());
@@ -2850,7 +2963,7 @@ bool PEIdent::calculate_bits_(Design*des, NetScope*scope,
       NetEConst*msb_c = dynamic_cast<NetEConst*>(msb_ex);
       if (msb_c == 0) {
 	    cerr << index_tail.msb->get_fileline() << ": error: "
-		  "Bit select expressionsmust be constant."
+		  "Bit select expressions must be constant."
 		 << endl;
 	    cerr << index_tail.msb->get_fileline() << ":      : "
                   "This msb expression violates the rule: "
@@ -3354,29 +3467,41 @@ NetExpr* PEIdent::elaborate_expr(Design*des, NetScope*scope,
       }
 
       if (net == 0) {
-	    cerr << get_fileline() << ": internal error: "
-		 << "Expecting idents with ntype to be signals." << endl;
+            cerr << get_fileline() << ": error: Unable to bind variable `"
+	         << path_ << "' in `" << scope_path(use_scope) << "'" << endl;
 	    des->errors += 1;
 	    return 0;
       }
 
-      if (! ntype->type_compatible(net->net_type())) {
-	    cerr << get_fileline() << ": internal_error: "
-		 << "net type doesn't match context type." << endl;
+      if (const netdarray_t*array_type = dynamic_cast<const netdarray_t*> (ntype)) {
+            if (array_type->type_compatible(net->net_type())) {
+                  NetESignal*tmp = new NetESignal(net);
+                  tmp->set_line(*this);
+                  return tmp;
+            }
 
-	    cerr << get_fileline() << ":               : "
-		 << "net type=";
+              // Icarus allows a dynamic array to be initialised with a
+              // single elementary value, so try that next.
+	    ntype = array_type->element_type();
+      }
+
+      if (! ntype->type_compatible(net->net_type())) {
+	    cerr << get_fileline() << ": error: the type of the variable '"
+		 << path_ << "' doesn't match the context type." << endl;
+
+	    cerr << get_fileline() << ":      : " << "variable type=";
 	    if (net->net_type())
 		  net->net_type()->debug_dump(cerr);
 	    else
 		  cerr << "<nil>";
 	    cerr << endl;
 
-	    cerr << get_fileline() << ":               : "
-		 << "context type=";
+	    cerr << get_fileline() << ":      : " << "context type=";
 	    ivl_assert(*this, ntype);
 	    ntype->debug_dump(cerr);
 	    cerr << endl;
+	    des->errors += 1;
+	    return 0;
       }
       ivl_assert(*this, ntype->type_compatible(net->net_type()));
 
@@ -3416,7 +3541,7 @@ NetExpr* PEIdent::elaborate_expr(Design*des, NetScope*scope,
       indices_flags idx_flags;
       indices_to_expressions(des, scope, this,
 			     use_comp.index, net->unpacked_dimensions(),
-			     need_const, net->unpacked_count(),
+			     need_const,
 			     idx_flags,
 			     unpacked_indices,
 			     unpacked_indices_const);
@@ -3680,10 +3805,7 @@ NetExpr* PEIdent::elaborate_expr(Design*des, NetScope*scope,
 
             if (!tmp) return 0;
 
-            tmp = pad_to_width(tmp, expr_wid, *this);
-            tmp->cast_signed(signed_flag_);
-
-            return tmp;
+            return pad_to_width(tmp, expr_wid, signed_flag_, *this);
       }
 
 	// If the identifier names a signal (a register or wire)
@@ -3719,10 +3841,7 @@ NetExpr* PEIdent::elaborate_expr(Design*des, NetScope*scope,
 		       << ", tmp=" << *tmp << endl;
 	    }
 
-            tmp = pad_to_width(tmp, expr_wid, *this);
-            tmp->cast_signed(signed_flag_);
-
-            return tmp;
+            return pad_to_width(tmp, expr_wid, signed_flag_, *this);
       }
 
 	// If the identifier is a named event
@@ -3769,7 +3888,7 @@ NetExpr* PEIdent::elaborate_expr(Design*des, NetScope*scope,
       }
 
 	// Maybe this is a method attached to an enumeration name? If
-	// this is system verilog, then test to see if the name is
+	// this is SystemVerilog, then test to see if the name is
 	// really a method attached to an object.
       if (gn_system_verilog() && found_in==0 && path_.size() >= 2) {
 	    pform_name_t use_path = path_;
@@ -3822,9 +3941,7 @@ NetExpr* PEIdent::elaborate_expr(Design*des, NetScope*scope,
 							       member_comp);
 			if (!tmp) return 0;
 
-			tmp = pad_to_width(tmp, expr_wid, *this);
-			tmp->cast_signed(signed_flag_);
-			return tmp;
+			return pad_to_width(tmp, expr_wid, signed_flag_, *this);
 		  }
 
 		  if (net->class_type() != 0) {
@@ -4403,7 +4520,7 @@ NetExpr* PEIdent::elaborate_expr_param_(Design*des,
 		       << "Elaborate parameter <" << path_
 		       << "> as enumeration constant." << *etmp << endl;
 	    tmp = etmp->dup_expr();
-            tmp = pad_to_width(tmp, expr_wid, *this);
+            tmp = pad_to_width(tmp, expr_wid, signed_flag_, *this);
 
       } else {
 	    perm_string name = peek_tail_name(path_);
@@ -4489,7 +4606,7 @@ NetExpr* PEIdent::elaborate_expr_net_word_(Design*des, NetScope*scope,
       indices_flags idx_flags;
       indices_to_expressions(des, scope, this,
 			     name_tail.index, net->unpacked_dimensions(),
-			     need_const, net->unpacked_count(),
+			     need_const,
 			     idx_flags,
 			     unpacked_indices,
 			     unpacked_indices_const);
@@ -4578,7 +4695,8 @@ NetExpr* PEIdent::elaborate_expr_net_part_(Design*des, NetScope*scope,
 {
       list<long> prefix_indices;
       bool rc = calculate_packed_indices_(des, scope, net->sig(), prefix_indices);
-      ivl_assert(*this, rc);
+      if (!rc)
+	    return 0;
 
       long msv, lsv;
       bool parts_defined_flag;
@@ -4718,7 +4836,8 @@ NetExpr* PEIdent::elaborate_expr_net_idx_up_(Design*des, NetScope*scope,
 {
       list<long>prefix_indices;
       bool rc = calculate_packed_indices_(des, scope, net->sig(), prefix_indices);
-      ivl_assert(*this, rc);
+      if (!rc)
+	    return 0;
 
       NetExpr*base = calculate_up_do_base_(des, scope, need_const);
 
@@ -4820,7 +4939,8 @@ NetExpr* PEIdent::elaborate_expr_net_idx_do_(Design*des, NetScope*scope,
 {
       list<long>prefix_indices;
       bool rc = calculate_packed_indices_(des, scope, net->sig(), prefix_indices);
-      ivl_assert(*this, rc);
+      if (!rc)
+	    return 0;
 
       NetExpr*base = calculate_up_do_base_(des, scope, need_const);
 
@@ -4908,7 +5028,8 @@ NetExpr* PEIdent::elaborate_expr_net_bit_(Design*des, NetScope*scope,
 {
       list<long>prefix_indices;
       bool rc = calculate_packed_indices_(des, scope, net->sig(), prefix_indices);
-      ivl_assert(*this, rc);
+      if (!rc)
+	    return 0;
 
       const name_component_t&name_tail = path_.back();
       ivl_assert(*this, !name_tail.index.empty());
@@ -4918,6 +5039,8 @@ NetExpr* PEIdent::elaborate_expr_net_bit_(Design*des, NetScope*scope,
       ivl_assert(*this, index_tail.lsb == 0);
 
       NetExpr*mux = elab_and_eval(des, scope, index_tail.msb, -1, need_const);
+      if (!mux)
+	    return 0;
 
       if (const netdarray_t*darray = net->sig()->darray_type()) {
 	      // Special case: This is a select of a dynamic
@@ -5234,9 +5357,8 @@ NetExpr* PENewArray::elaborate_expr(Design*des, NetScope*scope,
 	      // expression. Elaborate the expression as an element
 	      // type. The run-time will assign this value to each element.
 	    const netarray_t*array_type = dynamic_cast<const netarray_t*> (ntype);
-	    ivl_type_t elem_type = array_type->element_type();
 
-	    init_val = init_->elaborate_expr(des, scope, elem_type, flags);
+	    init_val = init_->elaborate_expr(des, scope, array_type, flags);
       }
 
       NetENew*tmp = new NetENew(ntype, size, init_val);
@@ -5487,6 +5609,10 @@ unsigned PENumber::test_width(Design*, NetScope*, width_mode_t&mode)
 
 NetExpr* PENumber::elaborate_expr(Design*des, NetScope*, ivl_type_t ntype, unsigned) const
 {
+        // Icarus allows dynamic arrays to be initialised with a single value.
+      if (const netdarray_t*array_type = dynamic_cast<const netdarray_t*> (ntype))
+            ntype = array_type->element_type();
+
       const netvector_t*use_type = dynamic_cast<const netvector_t*> (ntype);
       if (use_type == 0) {
 	    cerr << get_fileline() << ": internal error: "
@@ -5593,7 +5719,6 @@ unsigned PETernary::test_width(Design*des, NetScope*scope, width_mode_t&mode)
       } else if (tru_type == IVL_VT_LOGIC || fal_type == IVL_VT_LOGIC) {
 	    expr_type_ = IVL_VT_LOGIC;
       } else {
-	    ivl_assert(*this, tru_type == fal_type);
 	    expr_type_ = tru_type;
       }
       if (expr_type_ == IVL_VT_REAL) {
@@ -5989,7 +6114,7 @@ NetExpr* PEUnary::elaborate_expr(Design*des, NetScope*scope,
 		  }
 		  tmp->set_line(*this);
 	    }
-            tmp = pad_to_width(tmp, expr_wid, *this);
+            tmp = pad_to_width(tmp, expr_wid, signed_flag_, *this);
 	    break;
 
 	  case '&': // Reduction AND
@@ -6007,7 +6132,7 @@ NetExpr* PEUnary::elaborate_expr(Design*des, NetScope*scope,
 	    }
 	    tmp = new NetEUReduce(op_, ip);
 	    tmp->set_line(*this);
-            tmp = pad_to_width(tmp, expr_wid, *this);
+            tmp = pad_to_width(tmp, expr_wid, signed_flag_, *this);
 	    break;
 
 	  case '~':

elab_lval.cc

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2000-2014 Stephen Williams (steve@icarus.com)
+ * Copyright (c) 2000-2016 Stephen Williams (steve@icarus.com)
  * Copyright CERN 2012-2013 / Stephen Williams (steve@icarus.com)
  *
  *    This source code is free software; you can redistribute it
@@ -138,6 +138,9 @@ NetAssign_* PEConcat::elaborate_lval(Design*des,
 		  continue;
 	    }
 
+	      /* A concatenation is always unsigned. */
+	    tmp->set_signed(false);
+
 	      /* Link the new l-value to the previous one. */
 	    NetAssign_*last = tmp;
 	    while (last->more)
@@ -157,7 +160,7 @@ NetAssign_*PEIdent::scan_lname_for_nested_members_(Design*des, NetScope*scope,
 	    return 0;
 
       pform_name_t use_path = cur_path;
-      perm_string tmp_name = peek_tail_name(use_path);
+      name_component_t tail = use_path.back();
       use_path.pop_back();
 
       NetNet*       reg = 0;
@@ -171,18 +174,32 @@ NetAssign_*PEIdent::scan_lname_for_nested_members_(Design*des, NetScope*scope,
 		  return 0;
 
 	    tmp = new NetAssign_(tmp);
-	    tmp->set_property(tmp_name);
+	    tmp->set_property(tail.name);
 	    return tmp;
       }
 
-      if (reg->struct_type()) {
+      if (reg->struct_type() && reg->struct_type()->packed()) {
+	    NetAssign_*tmp = new NetAssign_(reg);
+	    elaborate_lval_net_packed_member_(des, scope, tmp, tail);
+	    return tmp;
+      }
+#if 0
+      if (reg->struct_type() && reg->struct_type()->packed()) {
 	    cerr << get_fileline() << ": sorry: "
-		 << "I don't know what to do with struct " << use_path << endl;
+		 << "I don't know what to do with packed struct " << use_path
+		 << " with member " << tail << "." << endl;
+	    return 0;
+      }
+#endif
+      if (reg->struct_type() && !reg->struct_type()->packed()) {
+	    cerr << get_fileline() << ": sorry: "
+		 << "I don't know what to do with unpacked struct " << use_path
+		 << " with member " << tail << "." << endl;
 	    return 0;
       }
 
       if (reg->class_type()) {
-	    return elaborate_lval_net_class_member_(des, scope, reg, tmp_name);
+	    return elaborate_lval_net_class_member_(des, scope, reg, tail.name);
       }
 
       return 0;
@@ -273,13 +290,13 @@ NetAssign_* PEIdent::elaborate_lval(Design*des,
       }
 
 	// We are processing the tail of a string of names. For
-	// example, the verilog may be "a.b.c", so we are processing
+	// example, the Verilog may be "a.b.c", so we are processing
 	// "c" at this point. (Note that if method_name is not nil,
 	// then this is "a.b.c.method" and "a.b.c" is a struct or class.)
       const name_component_t&name_tail = path_.back();
 
 	// Use the last index to determine what kind of select
-	// (bit/part/etc) we are processing. For example, the verilog
+	// (bit/part/etc) we are processing. For example, the Verilog
 	// may be "a.b.c[1][2][<index>]". All but the last index must
 	// be simple expressions, only the <index> may be a part
 	// select etc., so look at it to determine how we will be
@@ -316,7 +333,8 @@ NetAssign_* PEIdent::elaborate_lval(Design*des,
 
       if (reg->struct_type() && !method_name.nil()) {
 	    NetAssign_*lv = new NetAssign_(reg);
-	    elaborate_lval_net_packed_member_(des, use_scope, lv, method_name);
+	    name_component_t tmp_name (method_name);
+	    elaborate_lval_net_packed_member_(des, use_scope, lv, tmp_name);
 	    return lv;
       }
 
@@ -384,6 +402,7 @@ NetAssign_* PEIdent::elaborate_lval(Design*des,
 	/* No select expressions. */
 
       NetAssign_*lv = new NetAssign_(reg);
+      lv->set_signed(reg->get_signed());
 
       return lv;
 }
@@ -470,7 +489,7 @@ NetAssign_* PEIdent::elaborate_lval_method_class_member_(Design*des,
 
 		  if (debug_elaborate) {
 			cerr << get_fileline() << ": PEIdent::elaborate_lval_method_class_member_: "
-			     << "Found initialzers for property " << class_type->get_prop_name(pidx) << endl;
+			     << "Found initializers for property " << class_type->get_prop_name(pidx) << endl;
 		  }
 	    }
       }
@@ -528,8 +547,6 @@ NetAssign_* PEIdent::elaborate_lval_net_word_(Design*des,
 	    return 0;
       }
 
-      unsigned array_need_words = reg->unpacked_count();
-
 	// Make sure there are enough indices to address an array element.
       const index_component_t&index_head = name_tail.index.front();
       if (index_head.sel == index_component_t::SEL_PART) {
@@ -547,7 +564,7 @@ NetAssign_* PEIdent::elaborate_lval_net_word_(Design*des,
       indices_flags flags;
       indices_to_expressions(des, scope, this,
 			     name_tail.index, reg->unpacked_dimensions(),
-			     false, array_need_words,
+			     false,
 			     flags,
 			     unpacked_indices,
 			     unpacked_indices_const);
@@ -674,20 +691,24 @@ bool PEIdent::elaborate_lval_net_bit_(Design*des,
 		  unsigned long lwid;
 		  bool rcl = reg->sb_to_slice(prefix_indices, lsb, loff, lwid);
 		  ivl_assert(*this, rcl);
-		  cerr << get_fileline() << ": warning: L-value packed array "
-		       << "select of " << reg->name();
-		  if (reg->unpacked_dimensions() > 0) cerr << "[]";
-		  cerr << " has an undefined index." << endl;
-
+		  if (warn_ob_select) {
+			cerr << get_fileline()
+			     << ": warning: L-value packed array select of "
+			     << reg->name();
+			if (reg->unpacked_dimensions() > 0) cerr << "[]";
+			cerr << " has an undefined index." << endl;
+		  }
 		  lv->set_part(new NetEConst(verinum(verinum::Vx)), lwid);
 		  return true;
 	      // The index is undefined and this is a bit select.
 	    } else {
-		  cerr << get_fileline() << ": warning: L-value bit select of "
-		       << reg->name();
-		  if (reg->unpacked_dimensions() > 0) cerr << "[]";
-		  cerr << " has an undefined index." << endl;
-
+		  if (warn_ob_select) {
+			cerr << get_fileline()
+			     << ": warning: L-value bit select of "
+			     << reg->name();
+			if (reg->unpacked_dimensions() > 0) cerr << "[]";
+			cerr << " has an undefined index." << endl;
+		  }
 		  lv->set_part(new NetEConst(verinum(verinum::Vx)), 1);
 		  return true;
 	    }
@@ -784,12 +805,10 @@ bool PEIdent::elaborate_lval_net_bit_(Design*des,
 	      // Constant bit select that does something useful.
 	    long loff = reg->sb_to_idx(prefix_indices,lsb);
 
-	    if (loff < 0 || loff >= (long)reg->vector_width()) {
-		  cerr << get_fileline() << ": error: bit select "
+	    if (warn_ob_select && (loff < 0 || loff >= (long)reg->vector_width())) {
+		  cerr << get_fileline() << ": warning: bit select "
 		       << reg->name() << "[" <<lsb<<"]"
 		       << " is out of range." << endl;
-		  des->errors += 1;
-		  return 0;
 	    }
 
 	    if (reg->type()==NetNet::UNRESOLVED_WIRE) {
@@ -855,10 +874,13 @@ bool PEIdent::elaborate_lval_net_part_(Design*des,
       ivl_assert(*this, reg);
 
       if (! parts_defined_flag) {
-	    cerr << get_fileline() << ": warning: L-value part select of "
-	         << reg->name();
-	    if (reg->unpacked_dimensions() > 0) cerr << "[]";
-	    cerr << " has an undefined index." << endl;
+	    if (warn_ob_select) {
+		  cerr << get_fileline()
+		       << ": warning: L-value part select of "
+		       << reg->name();
+		  if (reg->unpacked_dimensions() > 0) cerr << "[]";
+		  cerr << " has an undefined index." << endl;
+	    }
 	      // Use a width of two here so we can distinguish between an
 	      // undefined bit or part select.
 	    lv->set_part(new NetEConst(verinum(verinum::Vx)), 2);
@@ -921,11 +943,11 @@ bool PEIdent::elaborate_lval_net_part_(Design*des,
       }
 
 	/* If the part select extends beyond the extremes of the
-	   variable, then report an error. Note that loff is
+	   variable, then output a warning. Note that loff is
 	   converted to normalized form so is relative the
 	   variable pins. */
 
-      if (loff < 0 || moff >= (long)reg->vector_width()) {
+      if (warn_ob_select && (loff < 0 || moff >= (long)reg->vector_width())) {
 	    cerr << get_fileline() << ": warning: Part select "
 		 << reg->name() << "[" << msb<<":"<<lsb<<"]"
 		 << " is out of range." << endl;
@@ -1117,8 +1139,9 @@ NetAssign_* PEIdent::elaborate_lval_net_class_member_(Design*des, NetScope*scope
 
 bool PEIdent::elaborate_lval_net_packed_member_(Design*des, NetScope*scope,
 						NetAssign_*lv,
-						const perm_string&member_name) const
+						const name_component_t&member_comp) const
 {
+      const perm_string&member_name = member_comp.name;
       NetNet*reg = lv->sig();
       ivl_assert(*this, reg);
 
@@ -1178,7 +1201,12 @@ bool PEIdent::elaborate_lval_net_packed_member_(Design*des, NetScope*scope,
       if (!name_tail.index.empty())
 	    use_sel = name_tail.index.back().sel;
 
-      ivl_assert(*this, use_sel == index_component_t::SEL_NONE || use_sel == index_component_t::SEL_BIT);
+      if (use_sel != index_component_t::SEL_NONE && use_sel != index_component_t::SEL_BIT) {
+	    cerr << get_fileline() << ": sorry: Assignments to part selects of "
+		    "a struct member are not yet supported." << endl;
+	    des->errors += 1;
+	    return false;
+      }
 
       if (! name_tail.index.empty()) {
 

elab_net.cc

@@ -328,39 +328,66 @@ bool PEIdent::eval_part_select_(Design*des, NetScope*scope, NetNet*sig,
 		      return false;
 		}
 
-		long lidx_tmp = sig->sb_to_idx(prefix_indices, lsb);
-		long midx_tmp = sig->sb_to_idx(prefix_indices, msb);
-		  /* Detect reversed indices of a part select. */
-		if (lidx_tmp > midx_tmp) {
-		      cerr << get_fileline() << ": error: Part select "
-			   << sig->name() << "[" << msb << ":"
-			   << lsb << "] indices reversed." << endl;
-		      cerr << get_fileline() << ":      : Did you mean "
-			   << sig->name() << "[" << lsb << ":"
-			   << msb << "]?" << endl;
-		      long tmp = midx_tmp;
-		      midx_tmp = lidx_tmp;
-		      lidx_tmp = tmp;
-		      des->errors += 1;
-		}
+		if (prefix_indices.size()+1 < sig->packed_dims().size()) {
+		      // Here we have a slice that doesn't have enough indices
+		      // to get to a single slice. For example:
+		      //    wire [9:0][5:1] foo
+		      //      ... foo[4:3] ...
+		      // Make this work by finding the indexed slices and
+		      // creating a generated slice that spans the whole
+		      // range.
+		      long loff, moff;
+		      unsigned long lwid, mwid;
+		      bool lrc;
+		      lrc = sig->sb_to_slice(prefix_indices, lsb, loff, lwid);
+		      ivl_assert(*this, lrc);
+		      lrc = sig->sb_to_slice(prefix_indices, msb, moff, mwid);
+		      ivl_assert(*this, lrc);
+		      ivl_assert(*this, lwid == mwid);
 
-		  /* Warn about a part select that is out of range. */
-		if (midx_tmp >= (long)sig->vector_width() || lidx_tmp < 0) {
-		      cerr << get_fileline() << ": warning: Part select "
-			   << sig->name();
-		      if (sig->unpacked_dimensions() > 0) {
-			    cerr << "[]";
+		      if (moff > loff) {
+			    lidx = loff;
+			    midx = moff + mwid - 1;
+		      } else {
+			    lidx = moff;
+			    midx = loff + lwid - 1;
 		      }
-		      cerr << "[" << msb << ":" << lsb
-			   << "] is out of range." << endl;
-		}
-		  /* This is completely out side the signal so just skip it. */
-		if (lidx_tmp >= (long)sig->vector_width() || midx_tmp < 0) {
-		      return false;
-		}
+		} else {
+		      long lidx_tmp = sig->sb_to_idx(prefix_indices, lsb);
+		      long midx_tmp = sig->sb_to_idx(prefix_indices, msb);
 
-		midx = midx_tmp;
-		lidx = lidx_tmp;
+		        /* Detect reversed indices of a part select. */
+		      if (lidx_tmp > midx_tmp) {
+			    cerr << get_fileline() << ": error: Part select "
+			        << sig->name() << "[" << msb << ":"
+			        << lsb << "] indices reversed." << endl;
+			    cerr << get_fileline() << ":      : Did you mean "
+			        << sig->name() << "[" << lsb << ":"
+			        << msb << "]?" << endl;
+			    long tmp = midx_tmp;
+			    midx_tmp = lidx_tmp;
+			    lidx_tmp = tmp;
+			    des->errors += 1;
+		      }
+
+		        /* Warn about a part select that is out of range. */
+		      if (midx_tmp >= (long)sig->vector_width() || lidx_tmp < 0) {
+			    cerr << get_fileline() << ": warning: Part select "
+			         << sig->name();
+			    if (sig->unpacked_dimensions() > 0) {
+				cerr << "[]";
+			    }
+			    cerr << "[" << msb << ":" << lsb
+			         << "] is out of range." << endl;
+		}
+		        /* This is completely out side the signal so just skip it. */
+		      if (lidx_tmp >= (long)sig->vector_width() || midx_tmp < 0) {
+			    return false;
+		      }
+
+		      midx = midx_tmp;
+		      lidx = lidx_tmp;
+		}
 		break;
 	  }
 
@@ -630,7 +657,7 @@ NetNet* PEIdent::elaborate_lnet_common_(Design*des, NetScope*scope,
 	    indices_flags flags;
 	    indices_to_expressions(des, scope, this,
 				   path_tail.index, sig->unpacked_dimensions(),
-				   true, sig->unpacked_count(),
+				   true,
 				   flags,
 				   unpacked_indices,
 				   unpacked_indices_const);

elab_scope.cc

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2000-2014 Stephen Williams (steve@icarus.com)
+ * Copyright (c) 2000-2017 Stephen Williams (steve@icarus.com)
  * Copyright CERN 2013 / Stephen Williams (steve@icarus.com)
  *
  *    This source code is free software; you can redistribute it
@@ -54,6 +54,15 @@
 # include  <cassert>
 # include  "ivl_assert.h"
 
+
+void set_scope_timescale(Design*des, NetScope*scope, PScope*pscope)
+{
+      scope->time_unit(pscope->time_unit);
+      scope->time_precision(pscope->time_precision);
+      scope->time_from_timescale(pscope->time_from_timescale);
+      des->set_precision(pscope->time_precision);
+}
+
 typedef map<perm_string,LexicalScope::param_expr_t>::const_iterator mparm_it_t;
 
 static void collect_parm_item_(Design*des, NetScope*scope, perm_string name,
@@ -201,16 +210,15 @@ static void elaborate_scope_enumeration(Design*des, NetScope*scope,
       verinum one_value ((uint64_t)1, enum_width);
       one_value.has_sign(enum_type->signed_flag);
 	// Find the maximum allowed enumeration value.
-      verinum min_value (0);
       verinum max_value (0);
       if (enum_type->signed_flag) {
-	    min_value = -pow(verinum(2), verinum(enum_width-1));
 	    max_value = pow(verinum(2), verinum(enum_width-1)) - one_value;
       } else {
 	    max_value = pow(verinum(2), verinum(enum_width)) - one_value;
       }
-      min_value.has_sign(true);
       max_value.has_sign(enum_type->signed_flag);
+	// Variable to indicate when a defined value wraps.
+      bool implicit_wrapped = false;
 	// Process the enumeration definition.
       for (list<named_pexpr_t>::const_iterator cur = enum_type->names->begin()
 		 ; cur != enum_type->names->end() ;  ++ cur, name_idx += 1) {
@@ -229,6 +237,8 @@ static void elaborate_scope_enumeration(Design*des, NetScope*scope,
 			continue;
 		  }
 		  cur_value = val_const->value();
+		    // Clear the implicit wrapped flag if a parameter is given.
+		  implicit_wrapped = false;
 
 		    // A 2-state value can not have a constant with X/Z bits.
 		  if (enum_type->base_type==IVL_VT_BOOL &&
@@ -335,24 +345,8 @@ static void elaborate_scope_enumeration(Design*des, NetScope*scope,
 		  continue;
 	    }
 
-	      // Cast any undefined bits to zero so the comparisons below
-	      // return just true (1) or false (0).
-	    verinum two_state_value = cur_value;
-	    two_state_value.cast_to_int2();
-
-	      // The enumeration value must fit into the enumeration bits.
-	    if (!cur_value.is_defined()) {
-		  if (cur_value.len() > (unsigned long)use_enum->packed_width()) {
-			cerr << use_enum->get_fileline()
-			     << ": error: Enumeration name " << cur->name
-			     << " value=" << cur_value
-			     << " is too wide for enumeration base type." << endl;
-			des->errors += 1;
-		  }
-
-	    } else if ((two_state_value > max_value) ||
-                (cur_value.has_sign() && (two_state_value < min_value))) {
-
+	      // Check to see if an implicitly wrapped value is used.
+	    if (implicit_wrapped) {
 		  cerr << use_enum->get_fileline()
 		       << ": error: Enumeration name " << cur->name
 		       << " has an inferred value that overflowed." << endl;
@@ -385,6 +379,7 @@ static void elaborate_scope_enumeration(Design*des, NetScope*scope,
 	      // In case the next name has an implicit value,
 	      // increment the current value by one.
 	    if (cur_value.is_defined()) {
+		  if (cur_value == max_value) implicit_wrapped = true;
 		  cur_value = cur_value + one_value;
 	    }
       }
@@ -514,8 +509,10 @@ static void elaborate_scope_class(Design*des, NetScope*scope, PClass*pclass)
 
       netclass_t*use_base_class = 0;
       if (base_class) {
-	    ivl_assert(*pclass, scope);
-	    use_base_class = scope->find_class(base_class->name);
+	    if (scope)
+		  use_base_class = scope->find_class(base_class->name);
+	    if (use_base_class == 0)
+		  use_base_class = des->find_class(base_class->name);
 	    if (use_base_class == 0) {
 		  cerr << pclass->get_fileline() << ": error: "
 		       << "Base class " << base_class->name
@@ -537,6 +534,7 @@ static void elaborate_scope_class(Design*des, NetScope*scope, PClass*pclass)
       class_scope->set_class_def(use_class);
       use_class->set_class_scope(class_scope);
       use_class->set_definition_scope(scope);
+      set_scope_timescale(des, class_scope, pclass);
 
 	// Collect the properties, elaborate them, and add them to the
 	// elaborated class definition.
@@ -609,13 +607,9 @@ static void elaborate_scope_classes(Design*des, NetScope*scope,
 
 void elaborate_rootscope_classes(Design*des)
 {
-      if (pform_classes.empty())
-	    return;
-
-      for (map<perm_string,PClass*>::iterator cur = pform_classes.begin()
-		 ; cur != pform_classes.end() ; ++ cur) {
-	    blend_class_constructors(cur->second);
-	    elaborate_scope_class(des, 0, cur->second);
+      for (size_t idx = 0 ; idx < pform_classes.size() ; idx += 1) {
+            blend_class_constructors(pform_classes[idx]);
+            elaborate_scope_class(des, 0, pform_classes[idx]);
       }
 }
 
@@ -668,8 +662,10 @@ static void elaborate_scope_task(Design*des, NetScope*scope, PTask*task)
       task_scope->is_auto(task->is_auto());
       task_scope->set_line(task);
 
-      if (scope==0)
+      if (scope==0) {
+	    set_scope_timescale(des, task_scope, task);
 	    des->add_root_task(task_scope, task);
+      }
 
       if (debug_scopes) {
 	    cerr << task->get_fileline() << ": elaborate_scope_task: "
@@ -733,8 +729,10 @@ static void elaborate_scope_func(Design*des, NetScope*scope, PFunction*task)
       task_scope->is_auto(task->is_auto());
       task_scope->set_line(task);
 
-      if (scope==0)
+      if (scope==0) {
+	    set_scope_timescale(des, task_scope, task);
 	    des->add_root_task(task_scope, task);
+      }
 
       if (debug_scopes) {
 	    cerr << task->get_fileline() << ": elaborate_scope_func: "
@@ -807,7 +805,7 @@ void elaborate_rootscope_tasks(Design*des)
 	    }
 
 	    cerr << cur->second->get_fileline() << ": internal error: "
-		 << "elabortae_rootscope_tasks does not understand "
+		 << "elaborate_rootscope_tasks does not understand "
 		 << "this object," << endl;
 	    des->errors += 1;
       }
@@ -1169,7 +1167,7 @@ bool PGenerate::generate_scope_loop_(Design*des, NetScope*container)
 	    container->genvar_tmp_val = genvar;
 	    delete step;
 	    delete test_ex;
-	    test_ex = elab_and_eval(des, container, loop_test, -1);
+	    test_ex = elab_and_eval(des, container, loop_test, -1, true);
 	    test = dynamic_cast<NetEConst*>(test_ex);
 	    assert(test);
       }
@@ -1756,18 +1754,15 @@ void PGModule::elaborate_scope_mod_instances_(Design*des, Module*mod, NetScope*s
 	      // scope searches will continue into the parent scope.
 	    NetScope*my_scope = new NetScope(sc, use_name, NetScope::MODULE,
 					     bound_type_? true : false,
-					     mod->program_block);
+					     mod->program_block,
+					     mod->is_interface);
 	    my_scope->set_line(get_file(), mod->get_file(),
 	                       get_lineno(), mod->get_lineno());
 	    my_scope->set_module_name(mod->mod_name());
 
 	    instances[idx] = my_scope;
 
-	      // Set time units and precision.
-	    my_scope->time_unit(mod->time_unit);
-	    my_scope->time_precision(mod->time_precision);
-	    my_scope->time_from_timescale(mod->time_from_timescale);
-	    des->set_precision(mod->time_precision);
+	    set_scope_timescale(des, my_scope, mod);
 
 	      // Look for module parameter replacements. The "replace" map
 	      // maps parameter name to replacement expression that is
@@ -1806,8 +1801,12 @@ void PGModule::elaborate_scope_mod_instances_(Design*des, Module*mod, NetScope*s
 	      // so the mapping into the replace list is much easier.
 	    if (parms_) {
 		  assert(overrides_ == 0);
-		  for (unsigned jdx = 0 ;  jdx < nparms_ ;  jdx += 1)
-			replace[parms_[jdx].name] = parms_[jdx].parm;
+		  for (unsigned jdx = 0 ;  jdx < nparms_ ;  jdx += 1) {
+		          // No expression means that the parameter is not
+		          // replaced.
+			if (parms_[jdx].parm)
+			      replace[parms_[jdx].name] = parms_[jdx].parm;
+		  }
 
 	    }
 

elab_sig.cc

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2000-2014 Stephen Williams (steve@icarus.com)
+ * Copyright (c) 2000-2019 Stephen Williams (steve@icarus.com)
  * Copyright CERN 2012 / Stephen Williams (steve@icarus.com)
  *
  *    This source code is free software; you can redistribute it
@@ -47,6 +47,8 @@
 
 using namespace std;
 
+#if 0
+/* These functions are not currently used. */
 static bool get_const_argument(NetExpr*exp, verinum&res)
 {
       switch (exp->expr_type()) {
@@ -73,8 +75,6 @@ static bool get_const_argument(NetExpr*exp, verinum&res)
       return true;
 }
 
-#if 0
-/* This function is not currently used. */
 static bool get_const_argument(NetExpr*exp, long&res)
 {
       verinum tmp;
@@ -715,16 +715,23 @@ void PTaskFunc::elaborate_sig_ports_(Design*des, NetScope*scope,
 		  continue;
 	    }
 
-	      // If the port has a default expression that can be used
-	      // as a value when the caller doesn't bind, then
-	      // elaborate that expression here. This expression
-	      // should evaluate down do a constant.
+	      // If the port has a default expression, elaborate
+	      // that expression here.
 	    if (ports_->at(idx).defe != 0) {
-		  tmp_def = elab_and_eval(des, scope, ports_->at(idx).defe, -1, true);
-		  if (tmp_def==0) {
-			cerr << get_fileline() << ": error: Unable to evaluate "
-			     << *ports_->at(idx).defe
-			     << " as a port default (constant) expression." << endl;
+		  if (tmp->port_type() == NetNet::PINPUT) {
+			tmp_def = elab_and_eval(des, scope, ports_->at(idx).defe,
+						-1, scope->need_const_func());
+			if (tmp_def == 0) {
+			      cerr << get_fileline()
+				   << ": error: Unable to evaluate "
+				   << *ports_->at(idx).defe
+				   << " as a port default expression." << endl;
+			      des->errors += 1;
+			}
+		  } else {
+			cerr << get_fileline() << ": sorry: Default arguments "
+			        "for subroutine output or inout ports are not "
+			        "yet supported." << endl;
 			des->errors += 1;
 		  }
 	    }
@@ -750,6 +757,11 @@ void PTaskFunc::elaborate_sig_ports_(Design*des, NetScope*scope,
 		       << "Function arguments must be input ports." << endl;
 		  des->errors += 1;
 	    }
+	    if (tmp->unpacked_dimensions() != 0) {
+		  cerr << get_fileline() << ": sorry: Subroutine ports with "
+			  "unpacked dimensions are not yet supported." << endl;
+		 des->errors += 1;
+	    }
       }
 }
 
@@ -864,13 +876,13 @@ static ivl_type_s*elaborate_type(Design*des, NetScope*scope,
       return 0;
 }
 
-static netparray_t* elaborate_parray_type(Design*des, NetScope*scope,
+static netparray_t* elaborate_parray_type(Design*des, NetScope*scope, const LineInfo*li,
 					  parray_type_t*data_type)
 {
 
       vector<netrange_t>packed_dimensions;
-      bool bad_range = evaluate_ranges(des, scope, packed_dimensions, * data_type->dims);
-      ivl_assert(*data_type, !bad_range);
+      bool dimensions_ok = evaluate_ranges(des, scope, li, packed_dimensions, * data_type->dims);
+      ivl_assert(*data_type, dimensions_ok);
 
       ivl_type_s*element_type = elaborate_type(des, scope, data_type->base_type);
 
@@ -909,15 +921,26 @@ bool test_ranges_eeq(const vector<netrange_t>&lef, const vector<netrange_t>&rig)
  */
 NetNet* PWire::elaborate_sig(Design*des, NetScope*scope) const
 {
+	// This sets the vector or array dimension size that will
+	// cause a warning. For now, these warnings are permanently
+	// enabled.
+      const long warn_dimension_size = 1 << 30;
+
       NetNet::Type wtype = type_;
       bool is_implicit_scalar = false;
       if (wtype == NetNet::IMPLICIT) {
 	    wtype = NetNet::WIRE;
 	    is_implicit_scalar = true;
       }
+	// Certain contexts, such as arguments to functions, presume
+	// "reg" instead of "wire". The parser reports these as
+	// IMPLICIT_REG. Also, certain cases, such as:
+	//   fun(string arg1) ...
+	// are implicitly NOT scalar, so detect that case here.
       if (wtype == NetNet::IMPLICIT_REG) {
 	    wtype = NetNet::REG;
-	    is_implicit_scalar = true;
+	    if (data_type_!=IVL_VT_STRING)
+		  is_implicit_scalar = true;
       }
 
       unsigned wid = 1;
@@ -964,7 +987,24 @@ NetNet* PWire::elaborate_sig(Design*des, NetScope*scope) const
       }
 
       if (port_set_ || net_set_) {
-	    bool bad_range = false;
+
+	    if (warn_implicit_dimensions
+		&& port_set_ && net_set_
+		&& net_.empty() && !port_.empty()) {
+		  cerr << get_fileline() << ": warning: "
+		       << "var/net declaration of " << basename()
+		       << " inherits dimensions from port declaration." << endl;
+	    }
+
+	    if (warn_implicit_dimensions
+		&& port_set_ && net_set_
+		&& port_.empty() && !net_.empty()) {
+		  cerr << get_fileline() << ": warning: "
+		       << "Port declaration of " << basename()
+		       << " inherits dimensions from var/net." << endl;
+	    }
+
+	    bool dimensions_ok = true;
 	    vector<netrange_t> plist, nlist;
 	    /* If they exist get the port definition MSB and LSB */
 	    if (port_set_ && !port_.empty()) {
@@ -972,7 +1012,7 @@ NetNet* PWire::elaborate_sig(Design*des, NetScope*scope) const
 			cerr << get_fileline() << ": PWire::elaborate_sig: "
 			     << "Evaluate ranges for port " << basename() << endl;
 		  }
-		  bad_range |= evaluate_ranges(des, scope, plist, port_);
+		  dimensions_ok &= evaluate_ranges(des, scope, this, plist, port_);
 		  nlist = plist;
 		    /* An implicit port can have a range so note that here. */
 		  is_implicit_scalar = false;
@@ -980,13 +1020,13 @@ NetNet* PWire::elaborate_sig(Design*des, NetScope*scope) const
             assert(port_set_ || port_.empty());
 
 	    /* If they exist get the net/etc. definition MSB and LSB */
-	    if (net_set_ && !net_.empty() && !bad_range) {
+	    if (net_set_ && !net_.empty() && dimensions_ok) {
 		  nlist.clear();
 		  if (debug_elaborate) {
 			cerr << get_fileline() << ": PWire::elaborate_sig: "
 			     << "Evaluate ranges for net " << basename() << endl;
 		  }
-		  bad_range |= evaluate_ranges(des, scope, nlist, net_);
+		  dimensions_ok &= evaluate_ranges(des, scope, this, nlist, net_);
 	    }
             assert(net_set_ || net_.empty());
 
@@ -1042,7 +1082,11 @@ NetNet* PWire::elaborate_sig(Design*des, NetScope*scope) const
 
 	    packed_dimensions = nlist;
 	    wid = netrange_width(packed_dimensions);
-
+	    if (wid > warn_dimension_size) {
+		  cerr << get_fileline() << ": warning: Vector size "
+		          "is greater than " << warn_dimension_size
+		       << "." << endl;
+	    }
       }
 
       unsigned nattrib = 0;
@@ -1063,6 +1107,7 @@ NetNet* PWire::elaborate_sig(Design*des, NetScope*scope) const
 	      // put all the packed dimensions there.
 	    if (use_lidx==0 && use_ridx==0) {
 		  netvector_t*vec = new netvector_t(packed_dimensions, data_type_);
+		  vec->set_signed(get_signed());
 		  packed_dimensions.clear();
 		  ivl_assert(*this, netdarray==0);
 		  netdarray = new netdarray_t(vec);
@@ -1073,6 +1118,7 @@ NetNet* PWire::elaborate_sig(Design*des, NetScope*scope) const
 	      // declaration, which is the dimensions [null:<nil>].
 	    if (use_ridx==0 && dynamic_cast<PENull*>(use_lidx)) {
 		  netvector_t*vec = new netvector_t(packed_dimensions, data_type_);
+		  vec->set_signed(get_signed());
 		  packed_dimensions.clear();
 		  ivl_assert(*this, netdarray==0);
 		  netdarray = new netqueue_t(vec);
@@ -1081,38 +1127,14 @@ NetNet* PWire::elaborate_sig(Design*des, NetScope*scope) const
 
 	      // Cannot handle dynamic arrays of arrays yet.
 	    ivl_assert(*this, netdarray==0);
-	    ivl_assert(*this, use_lidx && use_ridx);
-
-	    NetExpr*lexp = elab_and_eval(des, scope, use_lidx, -1, true);
-	    NetExpr*rexp = elab_and_eval(des, scope, use_ridx, -1, true);
-
-	    if ((lexp == 0) || (rexp == 0)) {
-		  cerr << get_fileline() << ": internal error: There is "
-		       << "a problem evaluating indices for ``"
-		       << name_ << "''." << endl;
-		  des->errors += 1;
-		  return 0;
-	    }
-
-	    bool const_flag = true;
-	    verinum lval, rval;
-	    const_flag &= get_const_argument(lexp, lval);
-	    const_flag &= get_const_argument(rexp, rval);
-	    delete rexp;
-	    delete lexp;
 
 	    long index_l, index_r;
-	    if (! const_flag) {
-		  cerr << get_fileline() << ": error: The indices "
-		       << "are not constant for array ``"
-		       << name_ << "''." << endl;
-		  des->errors += 1;
-                    /* Attempt to recover from error, */
-	          index_l = 0;
-	          index_r = 0;
-	    } else {
-		  index_l = lval.as_long();
-		  index_r = rval.as_long();
+	    evaluate_range(des, scope, this, *cur, index_l, index_r);
+
+	    if (abs(index_r - index_l) > warn_dimension_size) {
+		  cerr << get_fileline() << ": warning: Array dimension "
+		          "is greater than " << warn_dimension_size
+		       << "." << endl;
 	    }
 
 	    unpacked_dimensions.push_back(netrange_t(index_l, index_r));
@@ -1145,6 +1167,14 @@ NetNet* PWire::elaborate_sig(Design*des, NetScope*scope) const
 
       NetLogic*pull = 0;
       if (wtype == NetNet::SUPPLY0 || wtype == NetNet::SUPPLY1) {
+	    if (debug_elaborate) {
+		  cerr << get_fileline() << ": debug: "
+		       << "Generate a SUPPLY pull for the ";
+		  if (wtype == NetNet::SUPPLY0) cerr << "supply0";
+		  else cerr << "supply1";
+		  cerr << " net." << endl;
+	    }
+
 	    NetLogic::TYPE pull_type = (wtype==NetNet::SUPPLY1)
 		  ? NetLogic::PULLUP
 		  : NetLogic::PULLDOWN;
@@ -1155,14 +1185,6 @@ NetNet* PWire::elaborate_sig(Design*des, NetScope*scope) const
 	    pull->pin(0).drive1(IVL_DR_SUPPLY);
 	    des->add_node(pull);
 	    wtype = NetNet::WIRE;
-
-	    if (debug_elaborate) {
-		  cerr << get_fileline() << ": debug: "
-		       << "Generate a SUPPLY pull for the ";
-		  if (wtype == NetNet::SUPPLY0) cerr << "supply0";
-		  else cerr << "supply1";
-		  cerr << " net." << endl;
-	    }
       }
 
 
@@ -1194,7 +1216,7 @@ NetNet* PWire::elaborate_sig(Design*des, NetScope*scope) const
 		  cerr << " in scope " << scope_path(scope) << endl;
 	    }
 
-	    sig = new NetNet(scope, name_, wtype, use_type);
+	    sig = new NetNet(scope, name_, wtype, unpacked_dimensions, use_type);
 
       } else if (enum_type_t*enum_type = dynamic_cast<enum_type_t*>(set_data_type_)) {
 	    list<named_pexpr_t>::const_iterator sample_name = enum_type->names->begin();
@@ -1231,7 +1253,7 @@ NetNet* PWire::elaborate_sig(Design*des, NetScope*scope) const
 
 	      // The trick here is that the parray type has an
 	      // arbitrary sub-type, and not just a scalar bit...
-	    netparray_t*use_type = elaborate_parray_type(des, scope, parray_type);
+	    netparray_t*use_type = elaborate_parray_type(des, scope, this, parray_type);
 	      // Should not be getting packed dimensions other than
 	      // through the parray type declaration.
 	    ivl_assert(*this, packed_dimensions.empty());

elab_type.cc

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2012-2014 Stephen Williams (steve@icarus.com)
+ * Copyright (c) 2012-2019 Stephen Williams (steve@icarus.com)
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -17,11 +17,13 @@
  *    Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  */
 
+# include  "PExpr.h"
 # include  "pform_types.h"
 # include  "netlist.h"
 # include  "netclass.h"
 # include  "netdarray.h"
 # include  "netenum.h"
+# include  "netqueue.h"
 # include  "netparray.h"
 # include  "netscalar.h"
 # include  "netstruct.h"
@@ -32,6 +34,44 @@
 
 using namespace std;
 
+/*
+ * Some types have a list of ranges that need to be elaborated. This
+ * function elaborates the ranges referenced by "dims" into the vector
+ * "ranges".
+ */
+static void elaborate_array_ranges(Design*des, NetScope*scope,
+				   vector<netrange_t>&ranges,
+				   const list<pform_range_t>*dims)
+{
+      if (dims == 0)
+	    return;
+
+      for (list<pform_range_t>::const_iterator cur = dims->begin()
+		 ; cur != dims->end() ; ++ cur) {
+
+	    NetExpr*me = elab_and_eval(des, scope, cur->first, 0, true);
+
+	    NetExpr*le = elab_and_eval(des, scope, cur->second, 0, true);
+
+	      /* If elaboration failed for either expression, we
+		 should have already reported the error, so just
+		 skip the following evaluation to recover. */
+
+	    long mnum = 0, lnum = 0;
+	    if ( me && ! eval_as_long(mnum, me) ) {
+		  assert(0);
+		  des->errors += 1;
+	    }
+
+	    if ( le && ! eval_as_long(lnum, le) ) {
+		  assert(0);
+		  des->errors += 1;
+	    }
+
+	    ranges.push_back(netrange_t(mnum, lnum));
+      }
+}
+
 /*
  * Elaborations of types may vary depending on the scope that it is
  * done in, so keep a per-scope cache of the results.
@@ -43,8 +83,8 @@ ivl_type_s* data_type_t::elaborate_type(Design*des, NetScope*scope)
 	    use_definitions = des;
 
       map<Definitions*,ivl_type_s*>::iterator pos = cache_type_elaborate_.lower_bound(use_definitions);
-      if (pos->first == use_definitions)
-	    return pos->second;
+	  if (pos != cache_type_elaborate_.end() && pos->first == use_definitions)
+	     return pos->second;
 
       ivl_type_s*tmp = elaborate_type_raw(des, scope);
       cache_type_elaborate_.insert(pos, pair<NetScope*,ivl_type_s*>(scope, tmp));
@@ -120,33 +160,7 @@ ivl_type_s* enum_type_t::elaborate_type_raw(Design*des, NetScope*scope) const
 ivl_type_s* vector_type_t::elaborate_type_raw(Design*des, NetScope*scope) const
 {
       vector<netrange_t> packed;
-
-      if (pdims.get()) {
-	    for (list<pform_range_t>::const_iterator cur = pdims->begin()
-		       ; cur != pdims->end() ; ++ cur) {
-
-		  NetExpr*me = elab_and_eval(des, scope, cur->first, 0, true);
-
-		  NetExpr*le = elab_and_eval(des, scope, cur->second, 0, true);
-
-		    /* If elaboration failed for either expression, we
-		       should have already reported the error, so just
-		       skip the following evaluation to recover. */
-
-		  long mnum = 0, lnum = 0;
-		  if ( me && ! eval_as_long(mnum, me) ) {
-			assert(0);
-			des->errors += 1;
-		  }
-
-		  if ( le && ! eval_as_long(lnum, le) ) {
-			assert(0);
-			des->errors += 1;
-		  }
-
-		  packed.push_back(netrange_t(mnum, lnum));
-	    }
-      }
+      elaborate_array_ranges(des, scope, packed, pdims.get());
 
       netvector_t*tmp = new netvector_t(packed, base_type);
       tmp->set_signed(signed_flag);
@@ -171,13 +185,14 @@ ivl_type_s* string_type_t::elaborate_type_raw(Design*, NetScope*) const
       return &netstring_t::type_string;
 }
 
-ivl_type_s* parray_type_t::elaborate_type_raw(Design*des, NetScope*) const
+ivl_type_s* parray_type_t::elaborate_type_raw(Design*des, NetScope*scope) const
 {
-      cerr << get_fileline() << " : sorry: "
-           << "Packed arrays are not currently supported in this context."
-           << endl;
-      des->errors += 1;
-      return 0;
+      vector<netrange_t>packed;
+      elaborate_array_ranges(des, scope, packed, dims.get());
+
+      ivl_type_t etype = base_type->elaborate_type(des, scope);
+
+      return new netparray_t(packed, etype);
 }
 
 netstruct_t* struct_type_t::elaborate_type_raw(Design*des, NetScope*scope) const
@@ -232,10 +247,20 @@ ivl_type_s* uarray_type_t::elaborate_type_raw(Design*des, NetScope*scope) const
 	    return res;
       }
 
-      vector<netrange_t> dimensions;
-      bool bad_range = evaluate_ranges(des, scope, dimensions, *dims);
+	// Special case: if the dimension is null:nil. this is a queue.
+      if (cur->second==0 && dynamic_cast<PENull*>(cur->first)) {
+	    cerr << get_fileline() << ": sorry: "
+		 << "SV queues inside classes are not yet supported." << endl;
+	    des->errors += 1;
 
-      if (bad_range) {
+	    ivl_type_s*res = new netqueue_t(btype);
+	    return res;
+      }
+
+      vector<netrange_t> dimensions;
+      bool dimensions_ok = evaluate_ranges(des, scope, this, dimensions, *dims);
+
+      if (!dimensions_ok) {
 	    cerr << get_fileline() << " : warning: "
 		 << "Bad dimensions for type here." << endl;
       }

emit.cc

@@ -507,7 +507,6 @@ int Design::emit(struct target_t*tgt) const
       for (map<NetScope*,PTaskFunc*>::const_iterator scope = root_tasks_.begin()
 		 ; scope != root_tasks_.end() ; ++ scope) {
 	    scope->first->emit_scope(tgt);
-	    scope->first->emit_defs(tgt);
       }
 
 	// enumerate package scopes
@@ -521,7 +520,6 @@ int Design::emit(struct target_t*tgt) const
 	    const NetScope*use_scope = cur->second->class_scope();
 	    cur->second->emit_scope(tgt);
 	    tgt->class_type(use_scope, cur->second);
-	    cur->second->emit_defs(tgt);
       }
 
 	// enumerate root scopes
@@ -548,6 +546,12 @@ int Design::emit(struct target_t*tgt) const
 
 	// emit task and function definitions
       bool tasks_rc = true;
+      for (map<NetScope*,PTaskFunc*>::const_iterator scope = root_tasks_.begin()
+		 ; scope != root_tasks_.end() ; ++ scope)
+	    tasks_rc &= scope->first->emit_defs(tgt);
+      for (map<perm_string,netclass_t*>::const_iterator cur = classes_.begin()
+		 ; cur != classes_.end() ; ++cur)
+	    tasks_rc &= cur->second->emit_defs(tgt);
       for (map<perm_string,NetScope*>::const_iterator scope = packages_.begin()
 		 ; scope != packages_.end() ; ++ scope )
 	    tasks_rc &= scope->second->emit_defs(tgt);

eval_tree.cc

@@ -697,7 +697,9 @@ NetEConst* NetEBComp::eval_eqeqeq_(bool ne_flag, const NetExpr*le, const NetExpr
 	// If the left value is longer check it against the pad bit.
       if (res == verinum::V1) {
 	    verinum::V pad = verinum::V0;
-	    if (is_signed) pad = rv.get(rv.len()-1);
+	    if (is_signed)
+		  pad = rv.get(rv.len()-1);
+
 	    for (unsigned idx = cnt ;  idx < lv.len() ;  idx += 1)
 		  if (lv.get(idx) != pad) {
 			res = verinum::V0;
@@ -708,12 +710,15 @@ NetEConst* NetEBComp::eval_eqeqeq_(bool ne_flag, const NetExpr*le, const NetExpr
 	// If the right value is longer check it against the pad bit.
       if (res == verinum::V1) {
 	    verinum::V pad = verinum::V0;
-	    if (is_signed) pad = lv.get(lv.len()-1);
+	    if (is_signed)
+		  pad = lv.get(lv.len()-1);
+
 	    for (unsigned idx = cnt ;  idx < rv.len() ;  idx += 1) {
-		  if (rv.get(idx) != pad)
+		  if (rv.get(idx) != pad) {
 			res = verinum::V0;
 			break;
 		  }
+	    }
       }
 
       if (ne_flag) {
@@ -1097,6 +1102,7 @@ NetEConst* NetEBShift::eval_arguments_(const NetExpr*l, const NetExpr*r) const
 		  break;
 		case 'r':
                   lv.has_sign(false);
+		  // fallthrough
 		case 'R':
 		  val = cast_to_width(lv >> shift, wid);
 		  break;
@@ -1548,6 +1554,7 @@ NetEConst* NetEUReduce::eval_arguments_(const NetExpr*ex) const
 
 	  case 'A':
 		invert = true;
+		// fallthrough
 	  case '&': {
 		res = verinum::V1;
 		for (unsigned idx = 0 ;  idx < val.len() ;  idx += 1)
@@ -1557,6 +1564,7 @@ NetEConst* NetEUReduce::eval_arguments_(const NetExpr*ex) const
 
 	  case 'N':
 		invert = true;
+		// fallthrough
 	  case '|': {
 		res = verinum::V0;
 		for (unsigned idx = 0 ;  idx < val.len() ;  idx += 1)
@@ -1566,6 +1574,7 @@ NetEConst* NetEUReduce::eval_arguments_(const NetExpr*ex) const
 
 	  case 'X':
 		invert = true;
+		// fallthrough
 	  case '^': {
 		  /* Reduction XOR. */
 		unsigned ones = 0, unknown = 0;
@@ -1617,6 +1626,7 @@ NetExpr* NetECast::eval_arguments_(const NetExpr*ex) const
 			res_val = cast_to_width(res_val, expr_width());
 		  res = new NetEConst(res_val);
 	    }
+	    // fallthrough
 	  case 'v':
 	    if (const NetECReal*val = dynamic_cast<const NetECReal*>(ex)) {
 		  verinum res_val(val->value().as_double(), false);
@@ -2009,6 +2019,12 @@ static bool check_dimension(const NetExpr*dim_expr, long &dim)
 static bool get_array_info(const NetExpr*arg, long dim,
                            long &left, long &right, bool&defer)
 {
+      if (const NetEConstParam*param = dynamic_cast<const NetEConstParam*>(arg)) {
+	assert(dim == 1);
+	left = param->expr_width() - 1;
+	right = 0;
+	return false;
+      }
 	/* The argument must be a signal that has enough dimensions. */
       const NetESignal*esig = dynamic_cast<const NetESignal*>(arg);
       if (esig == 0) return true;

expr_synth.cc

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 1999-2013 Stephen Williams (steve@icarus.com)
+ * Copyright (c) 1999-2018 Stephen Williams (steve@icarus.com)
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -335,6 +335,7 @@ NetNet* NetEBComp::synthesize(Design*des, NetScope*scope, NetExpr*root)
 		  des->errors += 1;
 		  return 0;
 	    }
+	    // fallthrough
 	  case 'e': // ==
 	    connect(dev->pin_AEB(), osig->pin(0));
 	    break;
@@ -353,6 +354,7 @@ NetNet* NetEBComp::synthesize(Design*des, NetScope*scope, NetExpr*root)
 		  des->errors += 1;
 		  return 0;
 	    }
+	    // fallthrough
 	  case 'n': // !=
 	    connect(dev->pin_ANEB(), osig->pin(0));
 	    break;
@@ -386,7 +388,8 @@ NetNet* NetEBPow::synthesize(Design*des, NetScope*scope, NetExpr*root)
       powr->set_line(*this);
       des->add_node(powr);
 
-      powr->set_signed( has_sign() );
+        // The lpm_pwr object only cares about the signedness of the exponent.
+      powr->set_signed( right_->has_sign() );
 
       connect(powr->pin_DataA(), lsig->pin(0));
       connect(powr->pin_DataB(), rsig->pin(0));
@@ -1138,18 +1141,29 @@ NetNet* NetESelect::synthesize(Design *des, NetScope*scope, NetExpr*root)
       if (sub->vector_width() == expr_width())
 	    return sub;
 
-	// The vector_width is not exactly right, so the source is
-	// probably asking for padding. Create nodes to do sign
-	// extension or 0 extension, depending on the has_sign() mode
-	// of the expression.
-
       netvector_t*net_vec = new netvector_t(expr_type(), expr_width()-1, 0);
       net_vec->set_signed(has_sign());
       NetNet*net = new NetNet(scope, scope->local_symbol(),
 			      NetNet::IMPLICIT, net_vec);
       net->set_line(*this);
       net->local_flag(true);
-      if (has_sign()) {
+
+	// It may still happen that the expression is wider than the selection,
+	// and there was no part select created earlier (size casting).
+      if(expr_width() < sub->vector_width()) {
+	    NetPartSelect*sel = new NetPartSelect(sub, 0, expr_width(),
+                                                  NetPartSelect::VP, has_sign());
+	    sel->set_line(*this);
+	    des->add_node(sel);
+
+	    connect(net->pin(0), sel->pin(0));
+
+	// The vector_width is not exactly right, so the source is
+	// probably asking for padding. Create nodes to do sign
+	// extension or 0 extension, depending on the has_sign() mode
+	// of the expression.
+
+      } else if (has_sign()) {
 	    NetSignExtend*pad = new NetSignExtend(scope,
 						  scope->local_symbol(),
 						  expr_width());
@@ -1166,7 +1180,6 @@ NetNet* NetESelect::synthesize(Design *des, NetScope*scope, NetExpr*root)
 	    cat->set_line(*this);
 	    des->add_node(cat);
 
-	    assert(expr_width() > sub->vector_width());
 	    unsigned pad_width = expr_width() - sub->vector_width();
 	    verinum pad((uint64_t)0, pad_width);
 	    NetConst*con = new NetConst(scope, scope->local_symbol(),
@@ -1278,6 +1291,8 @@ NetNet* NetESignal::synthesize(Design*des, NetScope*scope, NetExpr*root)
 						  net_->vector_width()-1, 0);
 	    NetNet*tmp = new NetNet(scope, scope->local_symbol(),
 				    NetNet::IMPLICIT, tmp_vec);
+	    tmp->set_line(*this);
+	    tmp->local_flag(true);
 	    verinum tmp_val ((uint64_t)scope->genvar_tmp_val, net_->vector_width());
 	    NetConst*tmp_const = new NetConst(scope, scope->local_symbol(), tmp_val);
 	    tmp_const->set_line(*this);

iverilog-vpi.man.in

@@ -1,10 +1,10 @@
-.TH iverilog-vpi 1 "April 17th, 2009" "" "Version %M.%m.%n %E"
+.TH iverilog-vpi 1 "Jan 29th, 2017" "" "Version %M.%n%E"
 .SH NAME
 iverilog-vpi - Compile front end for VPI modules
 
 .SH SYNOPSIS
 .B iverilog-vpi
-[\-\-name=\fIname\fP]
+[options]
 \fIsourcefile\fP...
 
 .SH DESCRIPTION
@@ -15,9 +15,9 @@ list of C or C++ source files, and generates as output a linked VPI
 module. See the \fBvvp\fP(1) man page for a description of how the
 linked module is loaded by a simulation.
 
-The output is named after the first source file. For example, if the
-first source file is named \fIfoo.c\fP, the output becomes
-\fIfoo.vpi\fP.
+By default the output is named after the first source file. For
+example, if the first source file is named \fIfoo.c\fP, the output
+becomes \fIfoo.vpi\fP.
 
 .SH OPTIONS
 \fIiverilog\-vpi\fP accepts the following options:
@@ -26,10 +26,15 @@ first source file is named \fIfoo.c\fP, the output becomes
 Include the named library in the link of the VPI module. This allows
 VPI modules to further reference external libraries.
 
+.TP 8
+.B -L\fIdirectory\fP
+Add \fIdirectory\fP to the list of directories that will be searched
+for library files.
+
 .TP 8
 .B -I\fIdirectory\fP
-Add \fIdirectory\fP to the list of directories that will be search for
-header files.
+Add \fIdirectory\fP to the list of directories that will be searched
+for header files.
 
 .TP 8
 .B -D\fIdefine\fP
@@ -41,46 +46,38 @@ Normally, the output VPI module will be named after the first source
 file passed to the command. This flag sets the name (without the .vpi
 suffix) of the output vpi module.
 
-.TP 8
-.B --install-dir
-This flag causes the program to print the install directory for VPI
-modules, then exit. It is a convenience for makefiles or automated
-plug-in installers.
-
-.TP 8
-.B --cflags, --ldflags and --ldlibs
-These flags provide compile time information.
-
 .SH "PC-ONLY OPTIONS"
 
-The PC port of \fIiverilog\-vpi\fP includes two special flags needed to
-support the more intractable development environment. These flags help
-the program locate parts that it needs.
+When built as a native Windows program (using the MinGW toolchain),
+by default \fIiverilog\-vpi\fP will attempt to locate the MinGW tools
+needed to compile a VPI module on the system path (as set by the PATH
+environment variable). As an alternative, the user may specify the
+location of the MinGW tools via the following option.
 
 .TP 8
 .B -mingw=\fIpath\fP
-Tell the program the root of the Mingw compiler tool suite. The
+Tell the program the root of the MinGW compiler tool suite. The
 \fBvvp\fP runtime is compiled with this compiler, and this is the
-compiler that \fIiverilog\-vpi\fP expects to use to compile your source
-code. This is normally not needed, and if you do use it, it is only
-needed once. The compiler will save the \fIpath\fP in the registry for
-use later.
+compiler that \fIiverilog\-vpi\fP expects to use to compile your
+source code. If this option accompanies a list of files, it will
+apply to the current build only. If this option is provided on its
+own, \fIiverilog\-vpi\fP will save the \fIpath\fP in the registry
+and use that path in preference to the system path for subsequent
+operations, avoiding the need to specify it on the command line
+every time.
+
+.SH "INFORMATIONAL OPTIONS"
+
+\fIiverilog\-vpi\fP includes additional flags to let Makefile gurus
+peek at the configuration of the \fIiverilog\fP installation.  This way,
+Makefiles can be written that handle complex VPI builds natively, and
+without hard-coding values that depend on the system and installation.
+If used at all, these options must be used one at a time, and without
+any other options or directives.
 
 .TP 8
-.B -ivl=\fIpath\fP
-Set for the use during compilation the root if the Icarus Verilog
-install. This is the place where you installed Icarus Verilog when you
-ran the installer. This flag is also only needed once, and the path is
-stored in the registry for future use.
-
-.SH "UNIX-ONLY OPTIONS"
-
-The UNIX version of \fIiverilog\-vpi\fP includes additional flags to
-let Makefile gurus peek at the configuration of the \fIiverilog\fP
-installation.  This way, Makefiles can be written that handle complex VPI
-builds natively, and without hard-coding values that depend on the
-system and installation.  If used at all, these options must be
-used one at a time, and without any other options or directives.
+.B --install-dir
+Print the install directory for VPI modules.
 
 .TP 8
 .B --cflags
@@ -95,11 +92,6 @@ Print the linker flags (LDFLAGS) needed to link a VPI module.
 .B --ldlibs
 Print the libraries (LDLIBS) needed to link a VPI module.
 
-.TP 8
-.B -m32
-On 64bit systems that support it (and support vvp32) this flag
-requests a 32bit vpi binary instead of the default 64bit binary.
-
 .P
 Example GNU makefile that takes advantage of these flags:
 .IP "" 4
@@ -123,12 +115,12 @@ Steve Williams ([email protected])
 
 .SH SEE ALSO
 iverilog(1), vvp(1),
-.BR "<http://www.icarus.com/eda/verilog/>",
-.BR "<http://www.mingw.org>",
+.BR "<http://iverilog.icarus.com/>",
+.BR "<http://mingw-w64.yaxm.org/>",
 
 .SH COPYRIGHT
 .nf
-Copyright \(co  2002\-2009 Stephen Williams
+Copyright \(co  2002\-2017 Stephen Williams
 
 This document can be freely redistributed according to the terms of the
 GNU General Public License version 2.0

iverilog-vpi.sh

@@ -35,6 +35,7 @@ CCSRC=
 CXSRC=
 OBJ=
 LIB=
+LIBDIR=
 OUT=
 INCOPT=
 DEFS=
@@ -81,6 +82,9 @@ do
     -l*) LIB="$LIB $parm"
 	 ;;
 
+    -L*) LIBDIR="$LIBDIR $parm"
+	 ;;
+
     -I*) INCOPT="$INCOPT $parm"
 	 ;;
 
@@ -148,4 +152,4 @@ then
 fi
 
 echo "Making $OUT from $OBJ..."
-exec $LD -o $OUT $LDFLAGS $OBJ $LIB $LDLIBS
+exec $LD -o $OUT $LDFLAGS $LIBDIR $OBJ $LIB $LDLIBS

ivl.def

@@ -125,6 +125,7 @@ ivl_lpm_enable
 ivl_lpm_file
 ivl_lpm_lineno
 ivl_lpm_name
+ivl_lpm_negedge
 ivl_lpm_q
 ivl_lpm_scope
 ivl_lpm_select

ivl_alloc.h

@@ -26,6 +26,7 @@
 #  include <stdio.h>
 #endif
 
+#if defined(__GNUC__)
 /*
  * Define a safer version of malloc().
  */
@@ -82,5 +83,6 @@
       } \
       __ivl_rtn; \
 })
+#endif
 
 #endif /* IVL_ivl_alloc_H */

ivl_target.h

@@ -1,7 +1,7 @@
 #ifndef IVL_ivl_target_H
 #define IVL_ivl_target_H
 /*
- * Copyright (c) 2000-2014 Stephen Williams (steve@icarus.com)
+ * Copyright (c) 2000-2015 Stephen Williams (steve@icarus.com)
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -41,6 +41,12 @@
 # define __attribute__(x)
 #endif
 
+#if defined(__cplusplus) && defined(_MSC_VER)
+# define ENUM_UNSIGNED_INT : unsigned int
+#else
+# define ENUM_UNSIGNED_INT
+#endif
+
 _BEGIN_DECL
 
 /*
@@ -201,7 +207,7 @@ typedef enum ivl_dis_domain_e {
       IVL_DIS_CONTINUOUS = 2
 } ivl_dis_domain_t;
 
-typedef enum ivl_drive_e {
+typedef enum ivl_drive_e ENUM_UNSIGNED_INT {
       IVL_DR_HiZ    = 0,
       IVL_DR_SMALL  = 1,
       IVL_DR_MEDIUM = 2,
@@ -243,7 +249,7 @@ typedef enum ivl_expr_type_e {
       IVL_EX_UNARY = 14
 } ivl_expr_type_t;
 
-typedef enum ivl_select_type_e {
+typedef enum ivl_select_type_e ENUM_UNSIGNED_INT {
       IVL_SEL_OTHER = 0,
       IVL_SEL_IDX_UP = 1,
       IVL_SEL_IDX_DOWN = 2
@@ -300,7 +306,7 @@ typedef enum ivl_lpm_type_e {
       IVL_LPM_CONCAT = 16,
       IVL_LPM_CONCATZ = 36, /* Transparent concat */
       IVL_LPM_CMP_EEQ= 18, /* Case EQ (===) */
-      IVL_LPM_CMP_EQX= 37, /* Windcard EQ (==?) */
+      IVL_LPM_CMP_EQX= 37, /* Wildcard EQ (==?) */
       IVL_LPM_CMP_EQZ= 38, /* casez EQ */
       IVL_LPM_CMP_EQ = 10,
       IVL_LPM_CMP_GE =  1,
@@ -345,7 +351,7 @@ typedef enum ivl_path_edge_e {
 
 /* Processes are initial, always, or final blocks with a statement. This is
    the type of the ivl_process_t object. */
-typedef enum ivl_process_type_e {
+typedef enum ivl_process_type_e ENUM_UNSIGNED_INT {
       IVL_PR_INITIAL = 0,
       IVL_PR_ALWAYS  = 1,
       IVL_PR_FINAL   = 2
@@ -431,7 +437,7 @@ typedef enum ivl_sfunc_as_task_e {
 
 /* This is the type of a variable, and also used as the type for an
    expression. */
-typedef enum ivl_variable_type_e {
+typedef enum ivl_variable_type_e ENUM_UNSIGNED_INT {
       IVL_VT_VOID    = 0,  /* Not used */
       IVL_VT_NO_TYPE = 1,  /* Place holder for missing/unknown type. */
       IVL_VT_REAL    = 2,
@@ -1239,8 +1245,10 @@ extern unsigned    ivl_lpm_lineno(ivl_lpm_t net);
  * width of a general power is the XXXX of the widths of the
  * inputs.
  *
- * Power may be signed. If so, the output should be sign extended
- * to fill in its result.
+ * Power may be signed. This indicates the type of the exponent. The
+ * base will always be treated as unsigned. The compiler must ensure
+ * the width of the base is equal to the width of the output to
+ * obtain the correct result when the base is really a signed value.
  *
  * - Part Select (IVL_LPM_PART_VP and IVL_LPM_PART_PV)
  * There are two part select devices, one that extracts a part from a
@@ -1285,8 +1293,8 @@ extern unsigned    ivl_lpm_lineno(ivl_lpm_t net);
  * magnitude compare, the signedness does matter. In any case, the
  * result of the compare is always unsigned.
  *
- * The EQX and EQZ nodes are windcard compares, where xz bits (EQX) or
- * z bits (EQZ) in the data(1) operand are treated as windcards. no
+ * The EQX and EQZ nodes are wildcard compares, where xz bits (EQX) or
+ * z bits (EQZ) in the data(1) operand are treated as wildcards. no
  * bits in the data(0) operand are wild. This matches the
  * SystemVerilog convention for the ==? operator.
  *
@@ -1418,6 +1426,7 @@ extern ivl_signal_t ivl_lpm_array(ivl_lpm_t net);
   /* IVL_LPM_PART IVL_LPM_SUBSTITUTE */
 extern unsigned ivl_lpm_base(ivl_lpm_t net);
   /* IVL_LPM_FF */
+extern unsigned    ivl_lpm_negedge(ivl_lpm_t net);
 extern ivl_nexus_t ivl_lpm_clk(ivl_lpm_t net);
   /* IVL_LPM_UFUNC */
 extern ivl_scope_t  ivl_lpm_define(ivl_lpm_t net);
@@ -2312,7 +2321,7 @@ extern const char* ivl_type_prop_name(ivl_type_t net, int idx);
 extern ivl_type_t  ivl_type_prop_type(ivl_type_t net, int idx);
 
 
-#if defined(__MINGW32__) || defined (__CYGWIN32__)
+#if defined(__MINGW32__) || defined (__CYGWIN__)
 #  define DLLEXPORT __declspec(dllexport)
 #else
 #  define DLLEXPORT
@@ -2342,4 +2351,6 @@ typedef const char* (*target_query_f) (const char*key);
 
 _END_DECL
 
+#undef ENUM_UNSIGNED_INT
+
 #endif /* IVL_ivl_target_H */

ivlpp/ivlpp.txt

@@ -95,7 +95,7 @@ keys, with their corresponding values, are:
       This is exactly the same as the "-Dname=<value>" described above.
 
    I:<dir>
-      This is exctly the same as "-I<dir>".
+      This is exactly the same as "-I<dir>".
 
    relative include:<flag>
       The <flag> can be "true" or "false". This enables "relative

ivlpp/main.c

@@ -1,5 +1,5 @@
 const char COPYRIGHT[] =
-          "Copyright (c) 1999-2011 Stephen Williams ([email protected])";
+          "Copyright (c) 1999-2011,2015 Stephen Williams ([email protected])";
 /*
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU

lexor.lex

@@ -4,7 +4,7 @@
 
 %{
 /*
- * Copyright (c) 1998-2014 Stephen Williams ([email protected])
+ * Copyright (c) 1998-2015 Stephen Williams ([email protected])
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -51,6 +51,22 @@
  */
 extern YYLTYPE yylloc;
 
+char* yytext_string_filter(const char*str, size_t str_len)
+{
+      if (str == 0) return 0;
+      char*buf = new char[str_len+1];
+      for (size_t idx = 0 ; idx < str_len ; idx += 1) {
+	    if (str[idx] == 0) {
+		  VLerror(yylloc, "error: Found nil (\\000) in string literal, replacing with space (\\015) character.");
+		  buf[idx] = ' ';
+	    } else {
+		  buf[idx] = str[idx];
+	    }
+      }
+      buf[str_len] = 0;
+      return buf;
+}
+
 char* strdupnew(char const *str)
 {
        return str ? strcpy(new char [strlen(str)+1], str) : 0;
@@ -227,12 +243,12 @@ TU [munpf]
 <CSTRING>\\\\ { yymore(); /* Catch \\, which is a \ escaping itself */ }
 <CSTRING>\\\" { yymore(); /* Catch \", which is an escaped quote */ }
 <CSTRING>\n   { BEGIN(0);
-                yylval.text = strdupnew(yytext);
+                yylval.text = yytext_string_filter(yytext, yyleng);
 		VLerror(yylloc, "Missing close quote of string.");
 		yylloc.first_line += 1;
 		return STRING; }
 <CSTRING>\"   { BEGIN(0);
-                yylval.text = strdupnew(yytext);
+      yylval.text = yytext_string_filter(yytext, yyleng);
 		yylval.text[strlen(yytext)-1] = 0;
 		return STRING; }
 <CSTRING>.    { yymore(); }
@@ -432,18 +448,28 @@ TU [munpf]
       return SYSTEM_IDENTIFIER; }
 
 
-\'[sS]?[dD][ \t]*[0-9][0-9_]*  { yylval.number = make_unsized_dec(yytext);
-                            return BASED_NUMBER; }
-\'[sS]?[dD][ \t]*[xzXZ?]_* { yylval.number = make_undef_highz_dec(yytext);
-                             return BASED_NUMBER; }
-\'[sS]?[bB][ \t]*[0-1xzXZ_\?]+ { yylval.number = make_unsized_binary(yytext);
-                        return BASED_NUMBER; }
-\'[sS]?[oO][ \t]*[0-7xzXZ_\?]+ { yylval.number = make_unsized_octal(yytext);
-                        return BASED_NUMBER; }
-\'[sS]?[hH][ \t]*[0-9a-fA-FxzXZ_\?]+ { yylval.number = make_unsized_hex(yytext);
-                              return BASED_NUMBER; }
+\'[sS]?[dD][ \t]*[0-9][0-9_]* {
+      yylval.number = make_unsized_dec(yytext);
+      return BASED_NUMBER;
+}
+\'[sS]?[dD][ \t]*[xzXZ?]_* {
+      yylval.number = make_undef_highz_dec(yytext);
+      return BASED_NUMBER;
+}
+\'[sS]?[bB][ \t]*[0-1xzXZ?][0-1xzXZ?_]* {
+      yylval.number = make_unsized_binary(yytext);
+      return BASED_NUMBER;
+}
+\'[sS]?[oO][ \t]*[0-7xzXZ?][0-7xzXZ?_]* {
+      yylval.number = make_unsized_octal(yytext);
+      return BASED_NUMBER;
+}
+\'[sS]?[hH][ \t]*[0-9a-fA-FxzXZ?][0-9a-fA-FxzXZ?_]* {
+      yylval.number = make_unsized_hex(yytext);
+      return BASED_NUMBER;
+}
 \'[01xzXZ] {
-      if (generation_flag < GN_VER2005_SV) {
+      if (!gn_system_verilog()) {
 	    cerr << yylloc.text << ":" << yylloc.first_line << ": warning: "
 		 << "Using SystemVerilog 'N bit vector.  Use at least "
 		 << "-g2005-sv to remove this warning." << endl;
@@ -469,7 +495,7 @@ TU [munpf]
 
   /* This rule handles scaled time values for SystemVerilog. */
 [0-9][0-9_]*(\.[0-9][0-9_]*)?{TU}?s {
-      if(generation_flag & (GN_VER2005_SV | GN_VER2009 | GN_VER2012)) {
+      if (gn_system_verilog()) {
 	    yylval.text = strdupnew(yytext);
 	    return TIME_LITERAL;
       } else REJECT; }
@@ -847,7 +873,7 @@ verinum*make_unsized_binary(const char*txt)
 	    ptr += 1;
       }
 
-      assert((tolower(*ptr) == 'b') || (generation_flag >= GN_VER2005_SV));
+      assert((tolower(*ptr) == 'b') || gn_system_verilog());
       if (tolower(*ptr) == 'b') {
 	    ptr += 1;
       } else {
@@ -1596,7 +1622,7 @@ void destroy_lexor()
 {
 # ifdef FLEX_SCANNER
 #   if YY_FLEX_MAJOR_VERSION >= 2 && YY_FLEX_MINOR_VERSION >= 5
-#     if defined(YY_FLEX_SUBMINOR_VERSION) && YY_FLEX_SUBMINOR_VERSION >= 9
+#     if YY_FLEX_MINOR_VERSION > 5 || defined(YY_FLEX_SUBMINOR_VERSION) && YY_FLEX_SUBMINOR_VERSION >= 9
     yylex_destroy();
 #     endif
 #   endif

libmisc/LineInfo.cc

@@ -27,6 +27,11 @@ LineInfo::LineInfo()
 {
 }
 
+LineInfo::LineInfo(const LineInfo&that) :
+    file_(that.file_), lineno_(that.lineno_)
+{
+}
+
 LineInfo::~LineInfo()
 {
 }

libmisc/LineInfo.h

@@ -36,6 +36,7 @@ using namespace std;
 class LineInfo {
     public:
       LineInfo();
+      LineInfo(const LineInfo&that);
       virtual ~LineInfo();
 
 	// Get a fully formatted file/lineno

libmisc/StringHeap.cc

@@ -32,8 +32,8 @@ static unsigned string_pool_count = 0;
 StringHeap::StringHeap()
 {
       cell_base_ = 0;
-      cell_ptr_ = HEAPCELL;
-      cell_count_ = 0;
+      cell_size_ = 0;
+      cell_ptr_  = 0;
 }
 
 StringHeap::~StringHeap()
@@ -45,20 +45,24 @@ StringHeap::~StringHeap()
 const char* StringHeap::add(const char*text)
 {
       unsigned len = strlen(text);
-      assert((len+1) <= HEAPCELL);
-
-      unsigned rem = HEAPCELL - cell_ptr_;
+      unsigned rem = cell_size_ - cell_ptr_;
       if (rem < (len+1)) {
-	    cell_base_ = (char*)malloc(HEAPCELL);
+	      // release any unused memory
+	    if (rem > 0) {
+		  cell_base_ = (char*)realloc(cell_base_, cell_ptr_);
+		  assert(cell_base_ != 0);
+	    }
+	      // start new cell
+	    cell_size_ = (len+1) > DEFAULT_CELL_SIZE ? len+1 : DEFAULT_CELL_SIZE;
+	    cell_base_ = (char*)malloc(cell_size_);
+	    cell_ptr_  = 0;
+	    assert(cell_base_ != 0);
 #ifdef CHECK_WITH_VALGRIND
 	    string_pool_count += 1;
 	    string_pool = (char **) realloc(string_pool,
 	                                    string_pool_count*sizeof(char **));
 	    string_pool[string_pool_count-1] = cell_base_;
 #endif
-	    cell_ptr_ = 0;
-	    cell_count_ += 1;
-	    assert(cell_base_ != 0);
       }
 
       char*res = cell_base_ + cell_ptr_;
@@ -66,7 +70,7 @@ const char* StringHeap::add(const char*text)
       cell_ptr_ += len;
       cell_base_[cell_ptr_++] = 0;
 
-      assert(cell_ptr_ <= HEAPCELL);
+      assert(cell_ptr_ <= cell_size_);
 
       return res;
 }

libmisc/StringHeap.h

@@ -78,11 +78,11 @@ class StringHeap {
       perm_string make(const char*);
 
     private:
-      enum { HEAPCELL = 0x10000 };
+      static const unsigned DEFAULT_CELL_SIZE = 0x10000;
 
       char*cell_base_;
+      unsigned cell_size_;
       unsigned cell_ptr_;
-      unsigned cell_count_;
 
     private: // not implemented
       StringHeap(const StringHeap&);

libveriuser/Makefile.in

@@ -47,7 +47,7 @@ LDRELOCFLAGS = @LDRELOCFLAGS@
 
 LDTARGETFLAGS = @LDTARGETFLAGS@
 
-CPPFLAGS = $(INCLUDE_PATH) @CPPFLAGS@ @DEFS@ @PICFLAG@
+CPPFLAGS = $(INCLUDE_PATH) @CPPFLAGS@ @DEFS@ -DICARUS_VPI_CONST=const @PICFLAG@
 CFLAGS = @WARNING_FLAGS@ @WARNING_FLAGS_CC@ @CFLAGS@
 
 A = a_close.o a_compare_handles.o a_configure.o a_fetch_argc.o \

libveriuser/a_fetch_param.c

@@ -36,7 +36,7 @@ double acc_fetch_paramval(handle object)
 		  fprintf(pli_trace, "acc_fetch_paramval(%s) --> \"%s\"\n",
 			  vpi_get_str(vpiName, object), val.value.str);
 	    }
-	    return (double) (long)val.value.str;
+	    return (double) (intptr_t)val.value.str;
 
 	  default:
 	    vpi_printf("XXXX: parameter %s has type %d\n",

libveriuser/getsimtime.c

@@ -115,7 +115,7 @@ PLI_INT32 tf_getlongtime(PLI_INT32 *high)
 }
 
 /*
- * This function is not defined in the IEE standard, but is provided for
+ * This function is not defined in the IEEE standard, but is provided for
  * compatibility with other simulators. On platforms that support this,
  * make it a weak symbol just in case the user has defined their own
  * function for this.

libveriuser/priv.h

@@ -31,6 +31,6 @@ extern char* __acc_newstring(const char*txt);
 /*
  * Trace file for logging ACC and TF calls.
  */
-FILE* pli_trace;
+extern FILE* pli_trace;
 
 #endif /* IVL_priv_H */

libveriuser/veriusertfs.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2002-2014 Michael Ruff (mruff at chiaro.com)
+ * Copyright (c) 2002-2018 Michael Ruff (mruff at chiaro.com)
  *                         Michael Runyan (mrunyan at chiaro.com)
  *
  *    This source code is free software; you can redistribute it
@@ -42,8 +42,9 @@ typedef struct t_pli_data {
       int	paramvc;	/* parameter number for misctf */
 } s_pli_data, *p_pli_data;
 
-static PLI_INT32 compiletf(char *);
-static PLI_INT32 calltf(char *);
+static PLI_INT32 compiletf(ICARUS_VPI_CONST PLI_BYTE8 *);
+static PLI_INT32 calltf(ICARUS_VPI_CONST PLI_BYTE8 *);
+static PLI_INT32 sizetf(ICARUS_VPI_CONST PLI_BYTE8 *);
 static PLI_INT32 callback(p_cb_data);
 
 /*
@@ -150,7 +151,7 @@ void veriusertfs_register_table(p_tfcell vtable)
 	    tf_data.tfname = tf->tfname;
 	    tf_data.compiletf = compiletf;
 	    tf_data.calltf = calltf;
-	    tf_data.sizetf = (PLI_INT32 (*)(PLI_BYTE8 *))tf->sizetf;
+	    tf_data.sizetf = sizetf;
 	    tf_data.user_data = (char *)data;
 
 	    if (pli_trace) {
@@ -180,7 +181,7 @@ void veriusertfs_register_table(p_tfcell vtable)
  * This function calls the veriusertfs checktf and sets up all the
  * callbacks misctf requires.
  */
-static PLI_INT32 compiletf(char *data)
+static PLI_INT32 compiletf(ICARUS_VPI_CONST PLI_BYTE8*data)
 {
       p_pli_data pli;
       p_tfcell tf;
@@ -260,7 +261,7 @@ static PLI_INT32 compiletf(char *data)
 /*
  * This function is the wrapper for the veriusertfs calltf routine.
  */
-static PLI_INT32 calltf(char *data)
+static PLI_INT32 calltf(ICARUS_VPI_CONST PLI_BYTE8*data)
 {
       int rc = 0;
       p_pli_data pli;
@@ -283,6 +284,32 @@ static PLI_INT32 calltf(char *data)
       return rc;
 }
 
+/*
+ * This function is the wrapper for the veriusertfs sizetf routine.
+ */
+static PLI_INT32 sizetf(ICARUS_VPI_CONST PLI_BYTE8*data)
+{
+      int rc = 32;
+      p_pli_data pli;
+      p_tfcell tf;
+
+      /* cast back from opaque */
+      pli = (p_pli_data)data;
+      tf = pli->tf;
+
+      /* execute sizetf */
+      if (tf->sizetf) {
+	    if (pli_trace) {
+		  fprintf(pli_trace, "Call %s->sizetf(%d, %d)\n",
+			  tf->tfname, tf->data, reason_sizetf);
+	    }
+
+	    rc = tf->sizetf(tf->data, reason_sizetf);
+      }
+
+      return rc;
+}
+
 /*
  * This function is the wrapper for all the misctf callbacks
  */

link_const.cc

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2000-2013 Stephen Williams (steve@icarus.com)
+ * Copyright (c) 2000-2015 Stephen Williams (steve@icarus.com)
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -45,6 +45,14 @@ bool Nexus::drivers_constant() const
 		  return false;
 	    }
 
+	      /* If we are connected to a tran, there may be a driver
+		 on the other side of the tran. We could try checking
+		 for this, but for now, be pessimistic. */
+	    if (dynamic_cast<const NetTran*>(cur->get_obj())) {
+		  driven_ = VAR;
+		  return false;
+	    }
+
 	    Link::DIR cur_dir = cur->get_dir();
 	    if (cur_dir == Link::INPUT)
 		  continue;
@@ -267,11 +275,17 @@ vector<bool> Nexus::driven_mask(void) const
 	    }
 
 	    const NetPartSelect*obj_ps = dynamic_cast<const NetPartSelect*>(obj);
-	    if (obj_ps && obj_ps->dir()==NetPartSelect::VP)
-		  continue;
-	    if (obj_ps && cur->get_pin()!=1)
-		  continue;
-	    if (obj_ps) {
+	    if(obj_ps) {
+		  if (obj_ps->dir()==NetPartSelect::VP) {
+			if(cur->get_pin() != 0)
+			      continue;
+			for (size_t idx = 0 ; idx < mask.size() ; idx += 1)
+			      mask[idx] = true;
+			return mask;
+		  } else {
+			if (cur->get_pin() != 1)
+			      continue;
+		  }
 		  for (unsigned idx = 0 ; idx < obj_ps->width() ; idx += 1) {
 			size_t bit = idx + obj_ps->base();
 			ivl_assert(*obj, bit < mask.size());

load_module.cc

@@ -54,11 +54,13 @@ extern FILE *depend_file;
  * Use the type name as a key, and search the module library for a
  * file name that has that key.
  */
-bool load_module(const char*type)
+bool load_module(const char*type, int&parser_errors)
 {
       char path[4096];
       char*ltype = strdup(type);
 
+      parser_errors = 0;
+
       for (char*tmp = ltype ; *tmp ;  tmp += 1)
 	    *tmp = tolower(*tmp);
 
@@ -71,7 +73,8 @@ bool load_module(const char*type)
 	    if (cur == lcur->name_map.end())
 		  continue;
 
-	    sprintf(path, "%s%c%s", lcur->dir, dir_character, (*cur).second);
+	    snprintf(path, sizeof(path), "%s%c%s",
+		     lcur->dir, dir_character, (*cur).second);
 
 	    if(depend_file) {
                   if (depfile_mode == 'p') {
@@ -98,7 +101,7 @@ bool load_module(const char*type)
 		  if (verbose_flag)
 			cerr << "...parsing output from preprocessor..." << endl << flush;
 
-		  pform_parse(path, file);
+		  parser_errors = pform_parse(path, file);
 		  pclose(file);
 		  free(cmdline);
 
@@ -109,14 +112,14 @@ bool load_module(const char*type)
 
 		  FILE*file = fopen(path, "r");
 		  assert(file);
-		  pform_parse(path, file);
+		  parser_errors = pform_parse(path, file);
 		  fclose(file);
 	    }
 
 	    if (verbose_flag)
 		  cerr << "... Load module complete." << endl << flush;
 
-	    return true;
+	    return parser_errors == 0;
       }
 
 

main.cc

@@ -1,5 +1,5 @@
 const char COPYRIGHT[] =
-          "Copyright (c) 1998-2014 Stephen Williams ([email protected])";
+          "Copyright (c) 1998-2017 Stephen Williams ([email protected])";
 
 /*
  *    This source code is free software; you can redistribute it
@@ -68,7 +68,7 @@ extern "C" int optind;
 extern "C" const char*optarg;
 #endif
 
-#if defined(__CYGWIN32__) && !defined(HAVE_GETOPT_H)
+#if defined(__CYGWIN__) && !defined(HAVE_GETOPT_H)
 extern "C" int getopt(int argc, char*argv[], const char*fmt);
 extern "C" int optind;
 extern "C" const char*optarg;
@@ -108,6 +108,7 @@ bool gn_assertions_flag = true;
 bool gn_io_range_error_flag = true;
 bool gn_strict_ca_eval_flag = false;
 bool gn_strict_expr_width_flag = false;
+bool gn_shared_loop_index_flag = false;
 bool gn_verilog_ams_flag = false;
 
 /*
@@ -153,6 +154,7 @@ FILE *depend_file = NULL;
  * These are the warning enable flags.
  */
 bool warn_implicit  = false;
+bool warn_implicit_dimensions = false;
 bool warn_timescale = false;
 bool warn_portbinding = false;
 bool warn_inf_loop = false;
@@ -343,6 +345,12 @@ static void process_generation_flag(const char*gen)
       } else if (strcmp(gen,"no-strict-expr-width") == 0) {
 	    gn_strict_expr_width_flag = false;
 
+      } else if (strcmp(gen,"shared-loop-index") == 0) {
+	    gn_shared_loop_index_flag = true;
+
+      } else if (strcmp(gen,"no-shared-loop-index") == 0) {
+	    gn_shared_loop_index_flag = false;
+
 	  } else {
       }
 }
@@ -678,6 +686,9 @@ static void read_iconfig_file(const char*ipath)
 		      case 'i':
 			warn_implicit = true;
 			break;
+		      case 'd':
+			warn_implicit_dimensions = true;
+			break;
 		      case 'l':
 			warn_inf_loop = true;
 			break;
@@ -859,6 +870,7 @@ int main(int argc, char*argv[])
 #          if defined(HAVE_TIMES)
 	    times_flag = true;
 #          endif
+	    flags["VVP_EXTRA_ARGS"] = strdup(" -v");
 	    break;
 	  case 'V':
 	    version_flag = true;
@@ -921,16 +933,22 @@ int main(int argc, char*argv[])
       switch (generation_flag) {
         case GN_VER2012:
 	  lexor_keyword_mask |= GN_KEYWORDS_1800_2012;
+	  // fallthrough
         case GN_VER2009:
 	  lexor_keyword_mask |= GN_KEYWORDS_1800_2009;
+	  // fallthrough
         case GN_VER2005_SV:
 	  lexor_keyword_mask |= GN_KEYWORDS_1800_2005;
+	  // fallthrough
         case GN_VER2005:
 	  lexor_keyword_mask |= GN_KEYWORDS_1364_2005;
+	  // fallthrough
         case GN_VER2001:
 	  lexor_keyword_mask |= GN_KEYWORDS_1364_2001_CONFIG;
+	  // fallthrough
         case GN_VER2001_NOCONFIG:
 	  lexor_keyword_mask |= GN_KEYWORDS_1364_2001;
+	  // fallthrough
         case GN_VER1995:
 	  lexor_keyword_mask |= GN_KEYWORDS_1364_1995;
       }
@@ -1025,9 +1043,8 @@ int main(int argc, char*argv[])
 		  pform_dump(out, cur->second);
 	    }
 	    out << "PFORM DUMP $ROOT CLASSES:" << endl;
-	    for (map<perm_string,PClass*>::iterator cur = pform_classes.begin()
-		       ; cur != pform_classes.end() ; ++ cur) {
-		  pform_dump(out, cur->second);
+	    for (size_t idx = 0 ; idx < pform_classes.size() ; idx += 1) {
+		  pform_dump(out, pform_classes[idx]);
 	    }
 	    out << "PFORM DUMP PACKAGES:" << endl;
 	    for (map<perm_string,PPackage*>::iterator pac = pform_packages.begin()
@@ -1106,7 +1123,7 @@ int main(int argc, char*argv[])
 
 	/* Decide if we are going to allow system functions to be called
 	 * as tasks. */
-      if (generation_flag >= GN_VER2005_SV) {
+      if (gn_system_verilog()) {
 	    def_sfunc_as_task = IVL_SFUNC_AS_TASK_WARNING;
       }
 

mingw.txt

@@ -1,282 +1,4 @@
-Please NOTE:
+Please see the Icarus Verilog Wiki for instruction on building and installing
+Icarus Verilog as a native Windows application using the MinGW tools:
 
-These directions are slightly outdated and need to be updated.
-The plan is to rewrite all this using the latest MinGW at
-some time in the not too distant future (CR Aug. 2009)
-
-
-MINGW PORT OF ICARUS VERILOG
-
-    Copyright 2006 Stephen Williams  <[email protected]>
-
-
-Icarus Verilog source can be compiled with the mingw C/C++ compilers
-to get a Windows binary that does not require the POSIX compatibility
-cruft of the Cygwin.dll library. The resulting program can be run with
-or without Cygwin, so this is the preferred Windows distribution form.
-The configure scripts automatically detect that the compilers in use
-are the mingw compilers and will configure the Makefiles appropriately.
-
-The mingw patch doesn't contain tools beyond the compiler, but there
-is the "msys" package that the makers of Mingw publish that has enough
-extra tools to get most everything else working. There are a few extra
-packages needed beyond mingw and msys, and the following instructions
-explain how to get them and install them.
-
-* Some Preliminary Comments  -- PLEASE READ ME --
-
-The Windows port of Icarus Verilog is the most difficult of all the
-ports. The Windows system off the shelf is completely stripped, devoid
-of any support for software development. Everything needed to compile
-Icarus Verilog must be collected from various sources and stitched
-together by hand. Normal human beings with a basic understanding of
-software development can do this, but some patience (and access to the
-Internet) is required. You may choose to print these instructions. FOR
-BEST RESULTS, FOLLOW THESE INSTRUCTIONS CAREFULLY.
-
-NOTE that if you have Cygwin installed, it is best to not use a cygwin
-window to do the build, as the Cygwin tools will intermix with the
-mingw tools such that it is hard to be sure you are using the right
-compiler. Thus, it is recommended that these steps be *not* done in a
-Cygwin window. Use an MSYS window instead, and be careful that your
-msys/mingw tools are not masked by paths that point to Cygwin binaries.
-
-I have no plans to intentionally support MSVC++ compilation. Don't ask.
-
-* Summary of Prerequisites
-
-This is a brief list of prerequisite packages, along with the URL
-where each can be found. In most cases, the specific version is not
-critical, but these are the versions I use.
-
-     msys-1.0			<http://www.mingw/org>
-     msysDTK-1.0		<http://www.mingw.org>
-     Mingw32-5.x		<http://www.mingw.org>
-        ... including the sub-packages binutils, gcc-core and gcc-g++
-	if you are given the option.
-     readline-5.0-1             <http://gnuwin32.sourceforge.net>
-     bzip2-1.0.3		<http://gnuwin32.sourceforge.net>
-     zlib-1.2.3			<http://gnuwin32.sourceforge.net>
-     gperf-3.0.1		<http://gnuwin32.sourceforge.net>
-     bison-2.1			<http://gnuwin32.sourceforge.net>
-     flex-2.5.4a		<http://gnuwin32.sourceforge.net>
-
-The above table lists the packages required. It is convenient to
-install them in the above order. Many of these packages are also
-collected into the directory:
-
-     <ftp://icarus.com/pub/eda/verilog/win32-build-parts>
-
-Incidentally, besides Mingw32, none of these packages are needed after
-installation of Icarus Verilog is complete. These are only needed to
-build the compiler. The Mingw32 package can be used to compile VPI
-modules if you choose.
-
-* Install MSYS and msysDTK
-
-The msys package is available from the mingw download site. This is
-not the compiler but a collection of *nix tools ported to Windows and
-wrapped in a convenient installer. The msys package is all the various
-basic tools (shells, file utils, etc) and the msysDTK is extra
-developer tools other than the compiler.
-
-Download the msys-1.x.x.exe and msysdtc-1.x.x.exe binaries. These are
-self-installing packages. Install msys first, and then msysDTC. Most
-likely, you want to install them in c:/msys. (The msysDTK is installed
-in the same location, as it is an add-on.)
-
-This install should be easy and reliable.
-
-The installation will leave an "msys" icon on your desktop and in the
-mingw sub-menu of your Start menu. This icon brings up a shell window
-(a command line) that has paths all set up for executing msys and
-mingw commands. This is what you will want to use while executing
-commands below.
-
-* Install Mingw32
-
-The obvious step 2, then, is install the mingw compilers. These can be
-found at the web page <http://www.mingw.org>. The Mingw-5.x.x package
-is a convenient remote installer. Download this program and run
-it. The installer will ask which components you want to install. You
-need only the base C compiler and the C++ compiler. (You may install
-other languages if you wish.)
-
-When I install Mingw32 (using the installer) I typically set a
-destination directory of d:\mingw or the like. You will be using
-that path later.
-
-    NOTES:
-      If you intend to compile VPI modules for Icarus Verilog, you
-      need Mingw32, even if you are using a precompiled binary. VPI
-      modules only require Mingw32, and none of the other libraries.
-
-Finally, as part of installing the mingw32 compilers, remember to add
-the mingw/bin directory to your path. You will need that to be able to
-find the compilers later.
-
-* Install GnuWin32 Packages
-
-The GnuWin32 project is a collections of open source programs and
-libraries ported to Windows. These also work well with the Mingw
-compiler, and in fact Icarus Verilog uses a few pieces from this
-collection.
-
-You will need these gnuwin32 packages to compile Icarus Verilog:
-
-  <http://gnuwin32.sourceforge.net>
-    bzip2-1.0.3.exe
-    zlib-1.2.3.exe
-    gperf-3.0.1.exe
-    bison-2.1.exe
-    flex-2.5.4a.exe
-    readline-5.0-1.exe
-
-I suggest creating a common directory for holding all your gnuwin32
-packages. I use C:\gnuwin32. The download page at the gnuwin32 site
-has a "setup" link for each of these packages. Click the setup to
-download the installer for each of the desired programs, then execute
-the downloaded .exe files to invoke the installer. Install into the
-c:\gunwin32 directory.
-
-   NOTES:
-      You need the binaries and the developer files, but you do not
-      need the source to these packages. The installer gives you the
-      choice.
-
-After you are done installing the gnuwin32 tools, you should add the
-c:\gnuwin32\bin directory (assuming you installed in c:\gnuwin32) to
-your Windows path. The msys shell will pick up your Windows path.
-
-* Unpack Icarus Verilog source
-
-Unpack the compressed tar file (.tar.gz) of the source with a command
-like this:
-
-	$ gunzip -d verilog-xxxxxxxx.tar.gz | tar xvf -
-
-This will create a directory "verilog-xxxxxxxx" that contains all the
-source for Icarus Verilog. Descend into that directory, as that is
-where we will work from now on.
-
-	$ cd verilog-xxxxxxxx
-
-    NOTES:
-      The exact name of the file will vary according to the
-      snapshot. The 20030303 name is only an example.
-
-      Unpack the source into a directory that has no spaces. The
-      makefiles included in the source get confused by white space in
-      directory names.
-
-* Preconfigure Icarus Verilog (Not normally needed)
-
-Under certain cases, you may need to "preconfigure" the Icarus Verilog
-source tree. You should only need to do this if you are getting the
-Icarus Verilog source tree from git, or you are using an existing
-source tree that you've patched to cause configure.in files to change.
-
-    NOTE: If you are building from a fresh, bundled source tree that
-      you downloaded from an FTP site, then SKIP THIS STEP. Go on to
-      the "Configure Icarus Verilog" step below.
-
-First, remove any autom4te.cache directories that may exist in your
-source tree. These can make a mess of autoconf runs. Then, generate
-configure scripts with this command:
-
-        $ sh autoconf.sh
-
-This script will run the "autoconf" command (part of the msysDTK) to
-generate all the necessary "configure" scripts. This will take a few
-minutes. This should go smoothly.
-
-* Configure Icarus Verilog
-
-Now we are all set to configure and compile Icarus Verilog. Choose a
-destination path where you will want to install the binaries. I chose
-on my system the directory "D:\iverilog". This choice is not
-permanent, so don't get too much angst over it. Just choose a name
-without white space.
-
-Now, configure the source to make the makefiles and configuration
-details. Run these commands:
-
-	$ CPPFLAGS="-Ic:/gnuwin32/include
-	$ LDFLAGS="-Lc:/gnuwin32/lib
-	$ export CPPFLAGS LDFLAGS
-	$ ./configure --prefix=c:/iverilog
-
-    NOTES:
-      The CPPFLAGS and LDFLAGS variables tell configure where
-      the gnuwin32 packages are. The configure program will
-      write these values into the Makefiles, so you only need to
-      keep these variables long enough for the configure program
-      to work.
-
-      Your PATH variable was set in the previous step.
-
-      Use forward slashes as directory characters. All the various
-      tools prefer the forward slash.
-
-Substitute your chosen directory for the prefix. This will cause the
-makefiles to build and the source code to configure. The configure
-program will detect that this is a mingw environment and set things up
-to build properly.
-
-(For a prefix, use the drive letter notation; the mingw compiled parts
-require it, and the Cygwin tools should be able to handle it. You may
-need to check or upgrade your Cygwin installation if this causes
-problems for you.)
-
-
-* Compile Icarus Verilog
-
-This, believe it or not, should be the easy part:
-
-	$ make
-
-It could take a while. Now is a good time to go get some coffee or
-take a tea break.
-
-
-* Install Icarus Verilog
-
-If the compile ran OK, then next you install Icarus Verilog in the
-directory you have chosen. When you are ready, install like this:
-
-	$ make install
-
-This is part of what the configure program did for you. The Makefiles
-now know to put the files under the D:\iverilog directory (or whatever
-directory you chose) and away you go.
-
-You may find that you need to put some of the prerequisite DLLs into
-the d:\iverilog\bin directory. These include:
-
-      c:\mingw\bin\mingw10.dll
-      c:\gnuwin32\bin\readline.dll
-      c:\gnuwin32\bin\history.dll
-      c:\gnuwin32\bin\bzip2.dll
-      c:\gnuwin32\bin\zlib.dll
-
-If you already have these in your Windows path (i.e. your system32
-directory) then you do not need to copy them into the iverilog
-directory. However, prepackaged Icarus Verilog binaries include these
-files.
-
-* Running Icarus Verilog
-
-Finally, put the C:\iverilog\bin directory in your Windows path, and
-you should be able to run the iverilog and vvp commands to your
-heart's content.
-
-Currently, the iverilog.exe uses the path to itself to locate the
-libraries and modules associated with itself. In other words, if you
-execute the C:\iverilog\bin\iverilog.exe program, it will locate its
-subparts in the C:\iverilog directory and subdirectories below
-that. This means you can move the Icarus Verilog installation by
-simply moving the root directory and all its contents.
-
-The vvp.pdf and iverilog.pdf files document the main commands. View
-these with Acrobat reader, or any other viewer capable of displaying
-PDF format files.
+  http://iverilog.wikia.com/wiki/Installation_Guide#Compiling_on_MS_Windows_.28MinGW.29

net_assign.cc

@@ -43,6 +43,8 @@ NetAssign_::NetAssign_(NetAssign_*n)
 : nest_(n), sig_(0), word_(0), base_(0), sel_type_(IVL_SEL_OTHER)
 {
       more = 0;
+      signed_ = false;
+      turn_sig_to_wire_on_release_ = false;
 }
 
 NetAssign_::NetAssign_(NetNet*s)
@@ -51,6 +53,7 @@ NetAssign_::NetAssign_(NetNet*s)
       lwid_ = sig_->vector_width();
       sig_->incr_lref();
       more = 0;
+      signed_ = false;
       turn_sig_to_wire_on_release_ = false;
 }
 

net_design.cc

@@ -101,11 +101,12 @@ uint64_t Design::scale_to_precision(uint64_t val,
       return val;
 }
 
-NetScope* Design::make_root_scope(perm_string root, bool program_block)
+NetScope* Design::make_root_scope(perm_string root, bool program_block,
+				  bool is_interface)
 {
       NetScope *root_scope_;
       root_scope_ = new NetScope(0, hname_t(root), NetScope::MODULE,
-				 false, program_block);
+				 false, program_block, is_interface);
 	/* This relies on the fact that the basename return value is
 	   permallocated. */
       root_scope_->set_module_name(root_scope_->basename());
@@ -846,6 +847,7 @@ NetFuncDef* Design::find_function(NetScope*scope, const pform_name_t&name)
               // the function's signals have been elaborated. If this is
               // the case, elaborate them now.
             if (func->elab_stage() < 2) {
+		  func->need_const_func(true);
                   const PFunction*pfunc = func->func_pform();
                   assert(pfunc);
                   pfunc->elaborate_sig(this, func);

net_func_eval.cc

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2012-2013 Stephen Williams (steve@icarus.com)
+ * Copyright (c) 2012-2018 Stephen Williams (steve@icarus.com)
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -23,6 +23,10 @@
 # include  <typeinfo>
 # include  "ivl_assert.h"
 
+#if __cplusplus < 201103L
+#define unique_ptr auto_ptr
+#endif
+
 using namespace std;
 
 /*
@@ -88,6 +92,10 @@ NetExpr* NetFuncDef::evaluate_function(const LineInfo&loc, const std::vector<Net
 	// fills in the context_map with local variables held by the scope.
       scope()->evaluate_function_find_locals(loc, context_map);
 
+	// Execute any variable initialization statements.
+      if (const NetProc*init_proc = scope()->var_init())
+	    init_proc->evaluate_function(loc, context_map);
+
       if (debug_eval_tree && proc_==0) {
 	    cerr << loc.get_fileline() << ": NetFuncDef::evaluate_function: "
 		 << "Function " << scope_path(scope())
@@ -209,6 +217,98 @@ bool NetProc::evaluate_function(const LineInfo&,
       return false;
 }
 
+void NetAssign::eval_func_lval_op_real_(const LineInfo&loc,
+					verireal&lv, verireal&rv) const
+{
+      switch (op_) {
+	  case '+':
+	    lv = lv + rv;
+	    break;
+	  case '-':
+	    lv = lv - rv;
+	    break;
+	  case '*':
+	    lv = lv * rv;
+	    break;
+	  case '/':
+	    lv = lv / rv;
+	    break;
+	  case '%':
+	    lv = lv % rv;
+	    break;
+	  default:
+	    cerr << "Illegal assignment operator: "
+		 << human_readable_op(op_) << endl;
+	    ivl_assert(loc, 0);
+      }
+}
+
+void NetAssign::eval_func_lval_op_(const LineInfo&loc,
+				   verinum&lv, verinum&rv) const
+{
+      unsigned lv_width = lv.len();
+      bool lv_sign = lv.has_sign();
+      switch (op_) {
+	  case 'l':
+	  case 'R':
+	      // The left operand is self-determined.
+	    break;
+	  case 'r':
+	      // The left operand is self-determined, but we need to
+	      // cast it to unsigned to get a logical shift.
+	    lv.has_sign(false);
+	    break;
+          default:
+	      // The left operand must be cast to the expression type/size
+	    lv.has_sign(rv.has_sign());
+	    lv = cast_to_width(lv, rv.len());
+      }
+      switch (op_) {
+	  case '+':
+	    lv = lv + rv;
+	    break;
+	  case '-':
+	    lv = lv - rv;
+	    break;
+	  case '*':
+	    lv = lv * rv;
+	    break;
+	  case '/':
+	    lv = lv / rv;
+	    break;
+	  case '%':
+	    lv = lv % rv;
+	    break;
+	  case '&':
+	    for (unsigned idx = 0 ; idx < lv.len() ; idx += 1)
+		  lv.set(idx, lv[idx] & rv[idx]);
+	    break;
+	  case '|':
+	    for (unsigned idx = 0 ; idx < lv.len() ; idx += 1)
+		  lv.set(idx, lv[idx] | rv[idx]);
+	    break;
+	  case '^':
+	    for (unsigned idx = 0 ; idx < lv.len() ; idx += 1)
+		  lv.set(idx, lv[idx] ^ rv[idx]);
+	    break;
+	  case 'l':
+	    lv = lv << rv.as_unsigned();
+	    break;
+	  case 'r':
+	    lv = lv >> rv.as_unsigned();
+	    break;
+	  case 'R':
+	    lv = lv >> rv.as_unsigned();
+	    break;
+	  default:
+	    cerr << "Illegal assignment operator: "
+		 << human_readable_op(op_) << endl;
+	    ivl_assert(loc, 0);
+      }
+      lv = cast_to_width(lv, lv_width);
+      lv.has_sign(lv_sign);
+}
+
 bool NetAssign::eval_func_lval_(const LineInfo&loc,
 				map<perm_string,LocalVar>&context_map,
 				const NetAssign_*lval, NetExpr*rval_result) const
@@ -269,18 +369,55 @@ bool NetAssign::eval_func_lval_(const LineInfo&loc,
 	    ivl_assert(loc, base + lval->lwidth() <= old_lval->expr_width());
 
 	    NetEConst*lval_const = dynamic_cast<NetEConst*>(old_lval);
+	    ivl_assert(loc, lval_const);
 	    verinum lval_v = lval_const->value();
 	    NetEConst*rval_const = dynamic_cast<NetEConst*>(rval_result);
-	    verinum rval_v = cast_to_width(rval_const->value(), lval->lwidth());
+	    ivl_assert(loc, rval_const);
+	    verinum rval_v = rval_const->value();
 
-	    for (unsigned idx = 0 ; idx < rval_v.len() ; idx += 1)
-		  lval_v.set(idx+base, rval_v[idx]);
+	    verinum lpart(verinum::Vx, lval->lwidth());
+	    if (op_) {
+		  for (unsigned idx = 0 ; idx < lpart.len() ; idx += 1)
+			lpart.set(idx, lval_v[base+idx]);
+
+		  eval_func_lval_op_(loc, lpart, rval_v);
+	    } else {
+		  lpart = cast_to_width(rval_v, lval->lwidth());
+	    }
+	    for (unsigned idx = 0 ; idx < lpart.len() ; idx += 1)
+		  lval_v.set(idx+base, lpart[idx]);
 
 	    delete base_result;
 	    delete rval_result;
 	    rval_result = new NetEConst(lval_v);
       } else {
-	    rval_result = fix_assign_value(lval->sig(), rval_result);
+	    if (op_ == 0) {
+		  rval_result = fix_assign_value(lval->sig(), rval_result);
+	    } else if (dynamic_cast<NetECReal*>(rval_result)) {
+		  NetECReal*lval_const = dynamic_cast<NetECReal*>(old_lval);
+		  ivl_assert(loc, lval_const);
+		  verireal lval_r = lval_const->value();
+		  NetECReal*rval_const = dynamic_cast<NetECReal*>(rval_result);
+		  ivl_assert(loc, rval_const);
+		  verireal rval_r = rval_const->value();
+
+		  eval_func_lval_op_real_(loc, lval_r, rval_r);
+
+		  delete rval_result;
+		  rval_result = new NetECReal(lval_r);
+	    } else {
+		  NetEConst*lval_const = dynamic_cast<NetEConst*>(old_lval);
+		  ivl_assert(loc, lval_const);
+		  verinum lval_v = lval_const->value();
+		  NetEConst*rval_const = dynamic_cast<NetEConst*>(rval_result);
+		  ivl_assert(loc, rval_const);
+		  verinum rval_v = rval_const->value();
+
+		  eval_func_lval_op_(loc, lval_v, rval_v);
+
+		  delete rval_result;
+		  rval_result = new NetEConst(lval_v);
+	    }
       }
 
       if (old_lval)
@@ -318,6 +455,13 @@ bool NetAssign::evaluate_function(const LineInfo&loc,
       NetEConst*rval_const = dynamic_cast<NetEConst*>(rval_result);
       ivl_assert(*this, rval_const);
 
+      if (op_) {
+	    cerr << get_fileline() << ": sorry: Assignment operators "
+		    "inside a constant function are not currently "
+		    "supported if the LHS is a concatenation." << endl;
+	    return false;
+      }
+
       verinum rval_full = rval_const->value();
       delete rval_result;
 
@@ -369,6 +513,10 @@ bool NetBlock::evaluate_function(const LineInfo&loc,
 	      // Now collect the new locals.
 	    subscope_->evaluate_function_find_locals(loc, local_context_map);
 	    use_local_context_map = true;
+
+	      // Execute any variable initialization statements.
+	    if (const NetProc*init_proc = subscope_->var_init())
+		  init_proc->evaluate_function(loc, local_context_map);
       }
 
 	// Now use the local context map if there is any local
@@ -874,7 +1022,7 @@ NetExpr* NetESignal::evaluate_function(const LineInfo&loc,
 NetExpr* NetETernary::evaluate_function(const LineInfo&loc,
 				map<perm_string,LocalVar>&context_map) const
 {
-      auto_ptr<NetExpr> cval (cond_->evaluate_function(loc, context_map));
+      unique_ptr<NetExpr> cval (cond_->evaluate_function(loc, context_map));
 
       switch (const_logical(cval.get())) {
 

net_link.cc

@@ -311,6 +311,20 @@ void Nexus::count_io(unsigned&inp, unsigned&out) const
       }
 }
 
+bool Nexus::has_floating_input() const
+{
+      bool found_input = false;
+      for (const Link*cur = first_nlink() ;  cur ; cur = cur->next_nlink()) {
+	    if (cur->get_dir() == Link::OUTPUT)
+		  return false;
+
+	    if (cur->get_dir() == Link::INPUT)
+		  found_input = true;
+      }
+
+      return found_input;
+}
+
 bool Nexus::drivers_present() const
 {
       for (const Link*cur = first_nlink() ;  cur ; cur = cur->next_nlink()) {

net_nex_input.cc

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2002-2013 Stephen Williams (steve@icarus.com)
+ * Copyright (c) 2002-2020 Stephen Williams (steve@icarus.com)
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -261,7 +261,8 @@ NexusSet* NetAssign_::nex_input(bool rem_out)
 
 NexusSet* NetAssignBase::nex_input(bool rem_out)
 {
-      NexusSet*result = rval_->nex_input(rem_out);
+	// For the deassign and release statements there is no R-value.
+      NexusSet*result = rval_ ? rval_->nex_input(rem_out) : new NexusSet;
 
 	/* It is possible that the lval_ can have nex_input values. In
 	   particular, index expressions are statement inputs as well,
@@ -404,6 +405,17 @@ NexusSet* NetDoWhile::nex_input(bool rem_out)
       return result;
 }
 
+NexusSet* NetEvWait::nex_input(bool rem_out)
+{
+      NexusSet*result;
+      if (statement_)
+	    result = statement_->nex_input(rem_out);
+      else
+	    result = new NexusSet;
+
+      return result;
+}
+
 NexusSet* NetForce::nex_input(bool)
 {
       cerr << get_fileline() << ": internal warning: NetForce::nex_input()"
@@ -427,6 +439,15 @@ NexusSet* NetForLoop::nex_input(bool rem_out)
       result->add(*tmp);
       delete tmp;
 
+      if (gn_shared_loop_index_flag) {
+	    tmp = new NexusSet();
+	    for (unsigned idx = 0 ; idx < index_->pin_count() ; idx += 1)
+		tmp->add(index_->pin(idx).nexus(), 0, index_->vector_width());
+
+	    result->rem(*tmp);
+	    delete tmp;
+      }
+
       return result;
 }
 

net_proc.cc

@@ -56,6 +56,17 @@ void NetBlock::append(NetProc*cur)
       }
 }
 
+void NetBlock::prepend(NetProc*cur)
+{
+      if (last_ == 0) {
+	    last_ = cur;
+	    cur->next_ = cur;
+      } else {
+	    cur->next_ = last_->next_;
+	    last_->next_ = cur;
+      }
+}
+
 const NetProc* NetBlock::proc_first() const
 {
       if (last_ == 0)

net_scope.cc

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2000-2014 Stephen Williams (steve@icarus.com)
+ * Copyright (c) 2000-2017 Stephen Williams (steve@icarus.com)
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -110,8 +110,10 @@ void Definitions::add_class(netclass_t*net_class)
  * in question.
  */
 
-NetScope::NetScope(NetScope*up, const hname_t&n, NetScope::TYPE t, bool nest, bool prog)
-: type_(t), name_(n), nested_module_(nest), program_block_(prog), up_(up)
+NetScope::NetScope(NetScope*up, const hname_t&n, NetScope::TYPE t, bool nest,
+		   bool program, bool interface)
+: type_(t), name_(n), nested_module_(nest), program_block_(program),
+  is_interface_(interface), up_(up)
 {
       events_ = 0;
       lcounter_ = 0;
@@ -137,6 +139,7 @@ NetScope::NetScope(NetScope*up, const hname_t&n, NetScope::TYPE t, bool nest, bo
 	    time_from_timescale_ = false;
       }
 
+      var_init_ = 0;
       switch (t) {
 	  case NetScope::TASK:
 	    task_ = 0;
@@ -262,19 +265,20 @@ bool NetScope::auto_name(const char*prefix, char pad, const char* suffix)
       assert(self != up_->children_.end());
       assert(self->second == this);
 
-      char tmp[32];
-      int pad_pos = strlen(prefix);
-      int max_pos = sizeof(tmp) - strlen(suffix) - 1;
-      strncpy(tmp, prefix, sizeof(tmp));
-      tmp[31] = 0;
+	// This is to keep the pad attempts from being stuck in some
+	// sort of infinite loop. This should not be a practical
+	// limit, but an extreme one.
+      const size_t max_pad_attempts = 32 + strlen(prefix);
+
+      string use_prefix = prefix;
 
 	// Try a variety of potential new names. Make sure the new
 	// name is not in the parent scope. Keep looking until we are
 	// sure we have a unique name, or we run out of names to try.
-      while (pad_pos <= max_pos) {
+      while (use_prefix.size() <= max_pad_attempts) {
 	      // Try this name...
-	    strcat(tmp + pad_pos, suffix);
-	    hname_t new_name(lex_strings.make(tmp));
+	    string tmp = use_prefix + suffix;
+	    hname_t new_name(lex_strings.make(tmp.c_str()), name_.peek_numbers());
 	    if (!up_->child(new_name)) {
 		    // Ah, this name is unique. Rename myself, and
 		    // change my name in the parent scope.
@@ -283,7 +287,9 @@ bool NetScope::auto_name(const char*prefix, char pad, const char* suffix)
 		  up_->children_[name_] = this;
 		  return true;
 	    }
-	    tmp[pad_pos++] = pad;
+
+	      // Name collides, so try a different name.
+	    use_prefix = use_prefix + pad;
       }
       return false;
 }
@@ -294,16 +300,16 @@ bool NetScope::auto_name(const char*prefix, char pad, const char* suffix)
  */
 bool NetScope::replace_parameter(perm_string key, PExpr*val, NetScope*scope)
 {
-      bool flag = false;
+      if (parameters.find(key) == parameters.end())
+	    return false;
 
-      if (parameters.find(key) != parameters.end()) {
-	    param_expr_t&ref = parameters[key];
-	    ref.val_expr = val;
-            ref.val_scope = scope;
-	    flag = true;
-      }
+      param_expr_t&ref = parameters[key];
+      if (ref.local_flag)
+	    return false;
 
-      return flag;
+      ref.val_expr = val;
+      ref.val_scope = scope;
+      return true;
 }
 
 bool NetScope::make_parameter_unannotatable(perm_string key)
@@ -372,6 +378,8 @@ NetScope::param_ref_t NetScope::find_parameter(perm_string key)
 	// To get here the parameter must already exist, so we should
 	// never get here.
       assert(0);
+	// But return something to avoid a compiler warning.
+      return idx;
 }
 
 NetScope::TYPE NetScope::type() const

netclass.cc

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2012-2014 Stephen Williams (steve@icarus.com)
+ * Copyright (c) 2012-2017 Stephen Williams (steve@icarus.com)
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -165,7 +165,8 @@ bool netclass_t::test_for_missing_initializers() const
 NetScope*netclass_t::method_from_name(perm_string name) const
 {
       NetScope*task = class_scope_->child( hname_t(name) );
-      if (task == 0) return 0;
+      if ((task == 0) && super_)
+	    task = super_->method_from_name(name);
       return task;
 
 }

netdarray.h

@@ -1,7 +1,7 @@
 #ifndef IVL_netdarray_H
 #define IVL_netdarray_H
 /*
- * Copyright (c) 2012-2014 Stephen Williams (steve@icarus.com)
+ * Copyright (c) 2012-2015 Stephen Williams (steve@icarus.com)
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -33,6 +33,9 @@ class netdarray_t : public netarray_t {
 	// IVL_VT_DARRAY for dynamic arrays?
       ivl_variable_type_t base_type() const;
 
+	// A dynamic array may have a type that is signed.
+      inline bool get_signed() const { return element_type()->get_signed(); }
+
 	// This is the base_type() of the element of the array. We
 	// need this in some cases in order to get the base type of
 	// the element, and not the IVL_VT_DARRAY of the array itself.

netlist.cc

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 1998-2013 Stephen Williams (steve@icarus.com)
+ * Copyright (c) 1998-2017 Stephen Williams (steve@icarus.com)
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -557,7 +557,7 @@ void NetNet::calculate_slice_widths_from_packed_dims_(void)
       ivl_assert(*this, ! slice_wids_.empty());
       slice_wids_[0] = netrange_width(slice_dims_);
       vector<netrange_t>::const_iterator cur = slice_dims_.begin();
-      for (size_t idx = 1 ; idx < slice_wids_.size() ; idx += 1) {
+      for (size_t idx = 1 ; idx < slice_wids_.size() ; idx += 1, cur++) {
 	    slice_wids_[idx] = slice_wids_[idx-1] / cur->width();
       }
 }
@@ -1153,8 +1153,8 @@ unsigned NetReplicate::repeat() const
  *     ...
  */
 
-NetFF::NetFF(NetScope*s, perm_string n, unsigned width__)
-: NetNode(s, n, 8), width_(width__)
+NetFF::NetFF(NetScope*s, perm_string n, bool negedge__, unsigned width__)
+: NetNode(s, n, 8), negedge_(negedge__), width_(width__)
 {
       pin_Clock().set_dir(Link::INPUT);
       pin_Enable().set_dir(Link::INPUT);
@@ -1170,6 +1170,11 @@ NetFF::~NetFF()
 {
 }
 
+bool NetFF::is_negedge() const
+{
+      return negedge_;
+}
+
 unsigned NetFF::width() const
 {
       return width_;
@@ -2149,6 +2154,14 @@ ivl_variable_type_t NetEUFunc::expr_type() const
       return IVL_VT_VOID;
 }
 
+const netenum_t* NetEUFunc::enumeration() const
+{
+      if (result_sig_)
+	    return result_sig_->enumeration();
+
+      return 0;
+}
+
 NetUTask::NetUTask(NetScope*def)
 : task_(def)
 {
@@ -2438,6 +2451,20 @@ NetETernary::~NetETernary()
       delete false_val_;
 }
 
+const netenum_t* NetETernary::enumeration() const
+{
+	// If the condition can evaluate to an ambiguous value,
+	// the result may be blended, and so is not guaranteed
+	// to be a valid enumeration value.
+      if (cond_->expr_type() != IVL_VT_BOOL)
+	    return 0;
+
+      if (true_val_->enumeration() != false_val_->enumeration())
+	    return 0;
+
+      return true_val_->enumeration();
+}
+
 const NetExpr* NetETernary::cond_expr() const
 {
       return cond_;
@@ -2719,12 +2746,27 @@ DelayType NetProc::delay_type() const
 
 DelayType NetBlock::delay_type() const
 {
-      DelayType result = NO_DELAY;
+	// A join_none has no delay.
+      if (type() == PARA_JOIN_NONE) return NO_DELAY;
 
-      for (const NetProc*cur = proc_first(); cur; cur = proc_next(cur)) {
-	    DelayType dt = cur->delay_type();
-            if (dt > result) result = dt;
-            if (dt == DEFINITE_DELAY) break;
+      DelayType result;
+	// A join_any has the minimum delay.
+      if (type() == PARA_JOIN_ANY) {
+	    result = DEFINITE_DELAY;
+	    for (const NetProc*cur = proc_first(); cur; cur = proc_next(cur)) {
+		  DelayType dt = cur->delay_type();
+		  if (dt < result) result = dt;
+		  if (dt == NO_DELAY) break;
+	    }
+
+	// A begin or join has the maximum delay.
+      } else {
+	    result = NO_DELAY;
+	    for (const NetProc*cur = proc_first(); cur; cur = proc_next(cur)) {
+		  DelayType dt = cur->delay_type();
+		  if (dt > result) result = dt;
+		  if (dt == DEFINITE_DELAY) break;
+	    }
       }
 
       return result;
@@ -2738,10 +2780,11 @@ DelayType NetCase::delay_type() const
 
       for (unsigned idx = 0; idx < nstmts; idx += 1) {
 	    if (!expr(idx)) def_stmt = true;
+	    DelayType dt = stat(idx) ? stat(idx)->delay_type() : NO_DELAY;
             if (idx == 0) {
-		  result = stat(idx)->delay_type();
+		  result = dt;
             } else {
-		  result = combine_delays(result, stat(idx)->delay_type());
+		  result = combine_delays(result, dt);
             }
       }
 
@@ -2764,6 +2807,7 @@ DelayType NetCondit::delay_type() const
  */
 DelayType NetDoWhile::delay_type() const
 {
+      ivl_assert(*this, proc_);
       return proc_->delay_type();
 }
 
@@ -2774,11 +2818,13 @@ DelayType NetEvWait::delay_type() const
 
 DelayType NetForever::delay_type() const
 {
+      ivl_assert(*this, statement_);
       return statement_->delay_type();
 }
 
 DelayType NetForLoop::delay_type() const
 {
+      ivl_assert(*this, statement_);
       return get_loop_delay_type(condition_, statement_);
 }
 
@@ -2791,9 +2837,10 @@ DelayType NetPDelay::delay_type() const
 		  return DEFINITE_DELAY;
 	    } else {
 		  if (statement_) {
-			return statement_->delay_type();
+			return combine_delays(ZERO_DELAY,
+			                      statement_->delay_type());
 		  } else {
-			return NO_DELAY;
+			return ZERO_DELAY;
 		  }
 	    }
       }
@@ -2801,6 +2848,7 @@ DelayType NetPDelay::delay_type() const
 
 DelayType NetRepeat::delay_type() const
 {
+      ivl_assert(*this, statement_);
       return get_loop_delay_type(expr_, statement_);
 }
 
@@ -2816,5 +2864,6 @@ DelayType NetUTask::delay_type() const
 
 DelayType NetWhile::delay_type() const
 {
+      ivl_assert(*this, proc_);
       return get_loop_delay_type(cond_, proc_);
 }

netlist.h

@@ -1,7 +1,7 @@
 #ifndef IVL_netlist_H
 #define IVL_netlist_H
 /*
- * Copyright (c) 1998-2014 Stephen Williams (steve@icarus.com)
+ * Copyright (c) 1998-2015 Stephen Williams (steve@icarus.com)
  * Copyright CERN 2013 / Stephen Williams (steve@icarus.com)
  *
  *    This source code is free software; you can redistribute it
@@ -90,6 +90,12 @@ class netvector_t;
 struct target;
 struct functor_t;
 
+#if defined(__cplusplus) && defined(_MSC_VER)
+# define ENUM_UNSIGNED_INT : unsigned int
+#else
+# define ENUM_UNSIGNED_INT
+#endif
+
 ostream& operator << (ostream&o, ivl_variable_type_t val);
 
 extern void join_island(NetPins*obj);
@@ -102,8 +108,7 @@ class Link {
       friend class NexusSet;
 
     public:
-      enum DIR { PASSIVE, INPUT, OUTPUT };
-
+      enum DIR ENUM_UNSIGNED_INT { PASSIVE, INPUT, OUTPUT };
     private: // Only NetPins can create/delete Link objects
       Link();
       ~Link();
@@ -383,6 +388,10 @@ class Nexus {
 	   is a variable, but also if this is a net with a force. */
       bool assign_lval() const;
 
+	/* This method returns true if there are any inputs
+	   attached to this nexus but no drivers. */
+      bool has_floating_input() const;
+
 	/* This method returns true if there are any drivers
 	   (including variables) attached to this nexus. */
       bool drivers_present() const;
@@ -619,7 +628,7 @@ class NetDelaySrc  : public NetObj {
 class PortType
 {
 public:
-    enum Enum { NOT_A_PORT, PIMPLICIT, PINPUT, POUTPUT, PINOUT, PREF };
+	enum Enum ENUM_UNSIGNED_INT { NOT_A_PORT, PIMPLICIT, PINPUT, POUTPUT, PINOUT, PREF };
 
     /*
      * Merge Port types (used to construct a sane combined port-type
@@ -647,7 +656,7 @@ struct PortInfo
 class NetNet  : public NetObj, public PortType {
 
     public:
-      enum Type { NONE, IMPLICIT, IMPLICIT_REG, INTEGER, WIRE, TRI, TRI1,
+      enum Type ENUM_UNSIGNED_INT { NONE, IMPLICIT, IMPLICIT_REG, INTEGER, WIRE, TRI, TRI1,
 		  SUPPLY0, SUPPLY1, WAND, TRIAND, TRI0, WOR, TRIOR, REG,
 		  UNRESOLVED_WIRE };
 
@@ -708,7 +717,7 @@ class NetNet  : public NetObj, public PortType {
 	/* This method returns a reference to the packed dimensions
 	   for the vector. These are arranged as a list where the
 	   first range in the list (front) is the left-most range in
-	   the verilog declaration. These packed dims are compressed
+	   the Verilog declaration. These packed dims are compressed
 	   to represent the dimensions of all the subtypes. */
       const std::vector<netrange_t>& packed_dims() const { return slice_dims_; }
 
@@ -765,7 +774,7 @@ class NetNet  : public NetObj, public PortType {
       unsigned peek_eref() const;
 
 	// Assignment statements count their lrefs here. And by
-	// asignment statements, we mean BEHAVIORAL assignments.
+	// assignment statements, we mean BEHAVIORAL assignments.
       void incr_lref();
       void decr_lref();
       unsigned peek_lref() const { return lref_count_; }
@@ -918,7 +927,8 @@ class NetScope : public Definitions, public Attrib {
 
 	/* Create a new scope, and attach it to the given parent. The
 	   name is expected to have been permallocated. */
-      NetScope(NetScope*up, const hname_t&name, TYPE t, bool nest=false, bool prog=false);
+      NetScope(NetScope*up, const hname_t&name, TYPE t, bool nest=false,
+	       bool program=false, bool interface=false);
       ~NetScope();
 
 	/* Rename the scope using the name generated by inserting as
@@ -1008,11 +1018,19 @@ class NetScope : public Definitions, public Attrib {
 
 	// Nested modules have slightly different scope search rules.
       inline bool nested_module() const { return nested_module_; }
-	// Program blocks have elaboration constraints.
+	// Program blocks and interfaces have elaboration constraints.
       inline bool program_block() const { return program_block_; }
+      inline bool is_interface() const { return is_interface_; }
       TYPE type() const;
       void print_type(ostream&) const;
 
+	// This provides a link to the variable initialisation process
+	// for use when evaluating a constant function. Note this is
+	// only used for static functions - the variable initialization
+	// for automatic functions is included in the function definition.
+      void set_var_init(const NetProc*proc) { var_init_ = proc; }
+      const NetProc* var_init() const { return var_init_; }
+
       void set_task_def(NetTaskDef*);
       void set_func_def(NetFuncDef*);
       void set_class_def(netclass_t*);
@@ -1221,6 +1239,8 @@ class NetScope : public Definitions, public Attrib {
       bool nested_module_;
 	// True if the scope is a program block
       bool program_block_;
+	// True if the scope is an interface
+      bool is_interface_;
 
       perm_string file_;
       perm_string def_file_;
@@ -1241,6 +1261,8 @@ class NetScope : public Definitions, public Attrib {
 
       vector<PortInfo> ports_;
 
+      const NetProc*var_init_;
+
       union {
 	    NetTaskDef*task_;
 	    NetFuncDef*func_;
@@ -1618,9 +1640,10 @@ class NetModulo  : public NetNode {
 class NetFF  : public NetNode {
 
     public:
-      NetFF(NetScope*s, perm_string n, unsigned vector_width);
+      NetFF(NetScope*s, perm_string n, bool negedge, unsigned vector_width);
       ~NetFF();
 
+      bool is_negedge() const;
       unsigned width() const;
 
       Link& pin_Clock();
@@ -1652,6 +1675,7 @@ class NetFF  : public NetNode {
       virtual void functor_node(Design*des, functor_t*fun);
 
     private:
+      bool negedge_;
       unsigned width_;
       verinum aset_value_;
       verinum sset_value_;
@@ -2212,7 +2236,7 @@ class NetPartSelect  : public NetNode {
  *
  *      wire [7:0] foo = NetSubstitute(bar, bat, off);
  *
- * meaus that bar is a vector the same width as foo, bat is a narrower
+ * means that bar is a vector the same width as foo, bat is a narrower
  * vector. The off is a constant offset into the bar vector. This
  * looks something like this:
  *
@@ -2283,7 +2307,7 @@ class NetBUFZ  : public NetNode {
  *
  *     0   -- Output (always returns 0 or 1)
  *     1   -- Input
- *     2   -- Input (windcard input for EQX and EQZ variants)
+ *     2   -- Input (wildcard input for EQX and EQZ variants)
  */
 class NetCaseCmp  : public NetNode {
 
@@ -2614,8 +2638,10 @@ class NetProc : public virtual LineInfo {
 	// picked off by e.g. condit statements as set/reset inputs to
 	// the flipflop being generated.
       virtual bool synth_sync(Design*des, NetScope*scope,
+			      bool&ff_negedge,
 			      NetNet*ff_clock, NetBus&ff_ce,
 			      NetBus&ff_aclr,  NetBus&ff_aset,
+			      vector<verinum>&ff_aset_value,
 			      NexusSet&nex_map, NetBus&nex_out,
 			      const std::vector<NetEvProbe*>&events);
 
@@ -2719,6 +2745,12 @@ class NetAssign_ {
       void set_property(const perm_string&name);
       inline perm_string get_property(void) const { return member_; }
 
+	// Determine if the assigned object is signed or unsigned.
+	// This is used when determining the expression type for
+	// a compressed assignment statement.
+      bool get_signed() const { return signed_; }
+      void set_signed(bool flag) { signed_ = flag; }
+
 	// Get the width of the r-value that this node expects. This
 	// method accounts for the presence of the mux, so it is not
 	// necessarily the same as the pin_count().
@@ -2769,6 +2801,7 @@ class NetAssign_ {
 	// member/property if signal is a class.
       perm_string member_;
 
+      bool signed_;
       bool turn_sig_to_wire_on_release_;
 	// indexed part select base
       NetExpr*base_;
@@ -2836,6 +2869,8 @@ class NetAssign : public NetAssignBase {
 				     map<perm_string,LocalVar>&ctx) const;
 
     private:
+      void eval_func_lval_op_real_(const LineInfo&loc, verireal&lv, verireal&rv) const;
+      void eval_func_lval_op_(const LineInfo&loc, verinum&lv, verinum&rv) const;
       bool eval_func_lval_(const LineInfo&loc, map<perm_string,LocalVar>&ctx,
 			   const NetAssign_*lval, NetExpr*rval_result) const;
 
@@ -2882,6 +2917,7 @@ class NetBlock  : public NetProc {
       NetScope* subscope() const { return subscope_; }
 
       void append(NetProc*);
+      void prepend(NetProc*);
 
       const NetProc*proc_first() const;
       const NetProc*proc_next(const NetProc*cur) const;
@@ -2895,8 +2931,10 @@ class NetBlock  : public NetProc {
 		       NetBus&accumulated_nex_out);
 
       bool synth_sync(Design*des, NetScope*scope,
+		      bool&ff_negedge,
 		      NetNet*ff_clk, NetBus&ff_ce,
 		      NetBus&ff_aclr,NetBus&ff_aset,
+		      vector<verinum>&ff_aset_value,
 		      NexusSet&nex_map, NetBus&nex_out,
 		      const std::vector<NetEvProbe*>&events);
 
@@ -3036,8 +3074,10 @@ class NetCondit  : public NetProc {
 		       NetBus&accumulated_nex_out);
 
       bool synth_sync(Design*des, NetScope*scope,
+		      bool&ff_negedge,
 		      NetNet*ff_clk, NetBus&ff_ce,
 		      NetBus&ff_aclr,NetBus&ff_aset,
+		      vector<verinum>&ff_aset_value,
 		      NexusSet&nex_map, NetBus&nex_out,
 		      const std::vector<NetEvProbe*>&events);
 
@@ -3306,6 +3346,7 @@ class NetEvWait  : public NetProc {
 	// process? This method checks.
       virtual bool is_synchronous();
 
+      virtual NexusSet* nex_input(bool rem_out = true);
       virtual void nex_output(NexusSet&out);
 
       virtual bool synth_async(Design*des, NetScope*scope,
@@ -3313,8 +3354,10 @@ class NetEvWait  : public NetProc {
 			       NetBus&accumulated_nex_out);
 
       virtual bool synth_sync(Design*des, NetScope*scope,
+			      bool&ff_negedge,
 			      NetNet*ff_clk, NetBus&ff_ce,
 			      NetBus&ff_aclr,NetBus&ff_aset,
+			      vector<verinum>&ff_aset_value,
 			      NexusSet&nex_map, NetBus&nex_out,
 			      const std::vector<NetEvProbe*>&events);
 
@@ -3671,6 +3714,7 @@ class NetEUFunc  : public NetExpr {
       const NetScope* func() const;
 
       virtual ivl_variable_type_t expr_type() const;
+      virtual const netenum_t* enumeration() const;
       virtual void dump(ostream&) const;
 
       virtual void expr_scan(struct expr_scan_t*) const;
@@ -4537,6 +4581,8 @@ class NetETernary  : public NetExpr {
       NetETernary(NetExpr*c, NetExpr*t, NetExpr*f, unsigned wid, bool signed_flag);
       ~NetETernary();
 
+      const netenum_t* enumeration() const;
+
       const NetExpr*cond_expr() const;
       const NetExpr*true_expr() const;
       const NetExpr*false_expr() const;
@@ -4747,7 +4793,8 @@ class Design : public Definitions {
 
       const char* get_flag(const string&key) const;
 
-      NetScope* make_root_scope(perm_string name, bool program_block);
+      NetScope* make_root_scope(perm_string name, bool program_block,
+				bool is_interface);
       NetScope* find_root_scope();
       std::list<NetScope*> find_root_scopes() const;
 
@@ -4990,4 +5037,5 @@ inline unsigned Link::get_pin() const
 	    return pin_;
 }
 
+#undef ENUM_UNSIGNED_INT
 #endif /* IVL_netlist_H */

netmisc.cc

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2001-2014 Stephen Williams (steve@icarus.com)
+ * Copyright (c) 2001-2017 Stephen Williams (steve@icarus.com)
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -203,7 +203,7 @@ static NetExpr* make_add_expr(NetExpr*expr, long val)
       }
 
       verinum val_v (val, expr->expr_width());
-      val_v.has_sign(true);
+      val_v.has_sign(expr->has_sign());
 
       NetEConst*val_c = new NetEConst(val_v);
       val_c->set_line(*expr);
@@ -236,7 +236,7 @@ static NetExpr* make_add_expr(const LineInfo*loc, NetExpr*expr1, NetExpr*expr2)
 static NetExpr* make_sub_expr(long val, NetExpr*expr)
 {
       verinum val_v (val, expr->expr_width());
-      val_v.has_sign(true);
+      val_v.has_sign(expr->has_sign());
 
       NetEConst*val_c = new NetEConst(val_v);
       val_c->set_line(*expr);
@@ -249,7 +249,26 @@ static NetExpr* make_sub_expr(long val, NetExpr*expr)
 }
 
 /*
- * Multiple an existing expression by a signed positive number.
+ * Subtract a signed constant from an existing expression.
+ */
+static NetExpr* make_sub_expr(NetExpr*expr, long val)
+{
+      verinum val_v (val, expr->expr_width());
+      val_v.has_sign(expr->has_sign());
+
+      NetEConst*val_c = new NetEConst(val_v);
+      val_c->set_line(*expr);
+
+      NetEBAdd*res = new NetEBAdd('-', expr, val_c, expr->expr_width(),
+                                  expr->has_sign());
+      res->set_line(*expr);
+
+      return res;
+}
+
+
+/*
+ * Multiply an existing expression by a signed positive number.
  * This does a lossless multiply, so the arguments will need to be
  * sized to match the output size.
  */
@@ -258,7 +277,7 @@ static NetExpr* make_mult_expr(NetExpr*expr, unsigned long val)
       const unsigned val_wid = ceil(log2((double)val)) ;
       unsigned use_wid = expr->expr_width() + val_wid;
       verinum val_v (val, use_wid);
-      val_v.has_sign(true);
+      val_v.has_sign(expr->has_sign());
 
       NetEConst*val_c = new NetEConst(val_v);
       val_c->set_line(*expr);
@@ -434,17 +453,41 @@ NetExpr *normalize_variable_slice_base(const list<long>&indices, NetExpr*base,
 	    -- pcur;
       }
 
-      long sb;
-      if (pcur->get_msb() >= pcur->get_lsb())
-	    sb = pcur->get_lsb();
-      else
-	    sb = pcur->get_msb();
-
+      long sb = min(pcur->get_lsb(), pcur->get_msb());
       long loff;
       reg->sb_to_slice(indices, sb, loff, lwid);
 
-      base = make_mult_expr(base, lwid);
-      base = make_add_expr(base, loff);
+      unsigned min_wid = base->expr_width();
+      if ((sb < 0) && !base->has_sign()) min_wid += 1;
+      if (min_wid < num_bits(pcur->get_lsb())) min_wid = pcur->get_lsb();
+      if (min_wid < num_bits(pcur->get_msb())) min_wid = pcur->get_msb();
+      base = pad_to_width(base, min_wid, *base);
+      if ((sb < 0) && !base->has_sign()) {
+	    NetESelect *tmp = new NetESelect(base, 0 , min_wid);
+	    tmp->set_line(*base);
+	    tmp->cast_signed(true);
+            base = tmp;
+      }
+
+      if (pcur->get_msb() >= pcur->get_lsb()) {
+	    if (pcur->get_lsb() != 0)
+		  base = make_sub_expr(base, pcur->get_lsb());
+	    base = make_mult_expr(base, lwid);
+	    min_wid = base->expr_width();
+	    if (min_wid < num_bits(loff)) min_wid = num_bits(loff);
+	    if (loff != 0) min_wid += 1;
+	    base = pad_to_width(base, min_wid, *base);
+	    base = make_add_expr(base, loff);
+      } else {
+	    if (pcur->get_msb() != 0)
+		  base = make_sub_expr(base, pcur->get_msb());
+	    base = make_mult_expr(base, lwid);
+	    min_wid = base->expr_width();
+	    if (min_wid < num_bits(loff)) min_wid = num_bits(loff);
+	    if (loff != 0) min_wid += 1;
+	    base = pad_to_width(base, min_wid, *base);
+	    base = make_sub_expr(loff, base);
+      }
       return base;
 }
 
@@ -480,8 +523,6 @@ void indices_to_expressions(Design*des, NetScope*scope,
 			    const list<index_component_t>&src, unsigned count,
 			      // True if the expression MUST be constant.
 			    bool need_const,
-			      // Total words in target array
-			    unsigned need_addr,
 			      // These are the outputs.
 			    indices_flags&flags,
 			    list<NetExpr*>&indices, list<long>&indices_const)
@@ -692,7 +733,7 @@ NetExpr* make_canonical_index(Design*des, NetScope*scope,
       indices_flags flags;
       indices_to_expressions(des, scope, loc,
 			     src, src.size(),
-			     need_const, stype->static_dimensions().size(),
+			     need_const,
 			     flags,
 			     indices_expr, indices_const);
 
@@ -804,9 +845,10 @@ NetExpr* condition_reduce(NetExpr*expr)
 }
 
 static NetExpr* do_elab_and_eval(Design*des, NetScope*scope, PExpr*pe,
-				 int context_width, bool need_const, bool annotatable,
-				 bool force_expand,
-				 ivl_variable_type_t cast_type)
+				 int context_width, bool need_const,
+				 bool annotatable, bool force_expand,
+				 ivl_variable_type_t cast_type,
+				 bool force_unsigned)
 {
       PExpr::width_mode_t mode = PExpr::SIZED;
       if ((context_width == -2) && !gn_strict_expr_width_flag)
@@ -826,6 +868,11 @@ static NetExpr* do_elab_and_eval(Design*des, NetScope*scope, PExpr*pe,
       if ((pe->expr_type() != IVL_VT_REAL) && (expr_width < pos_context_width))
             expr_width = pos_context_width;
 
+	// If this is the RHS of a compressed assignment, the LHS also
+	// affects the expression type (signed/unsigned).
+      if (force_unsigned)
+	    pe->cast_signed(false);
+
       if (debug_elaborate) {
             cerr << pe->get_fileline() << ": elab_and_eval: test_width of "
                  << *pe << endl;
@@ -893,7 +940,7 @@ static NetExpr* do_elab_and_eval(Design*des, NetScope*scope, PExpr*pe,
             }
       }
 
-	// If the context_width sent is is actually the minimim width,
+	// If the context_width sent is is actually the minimum width,
 	// then raise the context_width to be big enough for the
 	// lossless expression.
       if (force_expand && context_width > 0) {
@@ -912,15 +959,16 @@ static NetExpr* do_elab_and_eval(Design*des, NetScope*scope, PExpr*pe,
 
 NetExpr* elab_and_eval(Design*des, NetScope*scope, PExpr*pe,
 		       int context_width, bool need_const, bool annotatable,
-		       ivl_variable_type_t cast_type)
+		       ivl_variable_type_t cast_type, bool force_unsigned)
 {
       return do_elab_and_eval(des, scope, pe, context_width,
-			      need_const, annotatable, false, cast_type);
+			      need_const, annotatable, false,
+			      cast_type, force_unsigned);
 }
 
 /*
  * This variant of elab_and_eval does the expression losslessly, no
- * matter what the generation of verilog. This is in support of
+ * matter what the generation of Verilog. This is in support of
  * certain special contexts, notably index expressions.
  */
 NetExpr* elab_and_eval_lossless(Design*des, NetScope*scope, PExpr*pe,
@@ -928,7 +976,8 @@ NetExpr* elab_and_eval_lossless(Design*des, NetScope*scope, PExpr*pe,
 				 ivl_variable_type_t cast_type)
 {
       return do_elab_and_eval(des, scope, pe, context_width,
-			      need_const, annotatable, true, cast_type);
+			      need_const, annotatable, true,
+			      cast_type, false);
 }
 
 NetExpr* elab_and_eval(Design*des, NetScope*scope, PExpr*pe,
@@ -988,50 +1037,94 @@ NetExpr* elab_sys_task_arg(Design*des, NetScope*scope, perm_string name,
       return tmp;
 }
 
-bool evaluate_ranges(Design*des, NetScope*scope,
+bool evaluate_range(Design*des, NetScope*scope, const LineInfo*li,
+		    const pform_range_t&range, long&index_l, long&index_r)
+{
+      bool dimension_ok = true;
+
+        // Unsized and queue dimensions should be handled before calling
+        // this function. If we find them here, we are in a context where
+        // they are not allowed.
+      if (range.first == 0) {
+            cerr << li->get_fileline() << ": error: "
+                    "An unsized dimension is not allowed here." << endl;
+            dimension_ok = false;
+            des->errors += 1;
+      } else if (dynamic_cast<PENull*>(range.first)) {
+            cerr << li->get_fileline() << ": error: "
+                    "A queue dimension is not allowed here." << endl;
+            dimension_ok = false;
+            des->errors += 1;
+      } else {
+            NetExpr*texpr = elab_and_eval(des, scope, range.first, -1, true);
+            if (! eval_as_long(index_l, texpr)) {
+                  cerr << range.first->get_fileline() << ": error: "
+                          "Dimensions must be constant." << endl;
+                  cerr << range.first->get_fileline() << "       : "
+                       << (range.second ? "This MSB" : "This size")
+                       << " expression violates the rule: "
+                       << *range.first << endl;
+                  dimension_ok = false;
+                  des->errors += 1;
+            }
+            delete texpr;
+
+            if (range.second == 0) {
+                    // This is a SystemVerilog [size] dimension. The IEEE
+                    // standard does not allow this in a packed dimension,
+                    // but we do. At least one commercial simulator does too.
+                  if (!dimension_ok) {
+                        // bail out
+                  } else if (index_l > 0) {
+                        index_l = index_l - 1;
+                        index_r = 0;
+                  } else {
+                        cerr << range.first->get_fileline() << ": error: "
+                                "Dimension size must be greater than zero." << endl;
+                        cerr << range.first->get_fileline() << "       : "
+                                "This size expression violates the rule: "
+                             << *range.first << endl;
+                        dimension_ok = false;
+                        des->errors += 1;
+                  }
+            } else {
+                  texpr = elab_and_eval(des, scope, range.second, -1, true);
+                  if (! eval_as_long(index_r, texpr)) {
+                        cerr << range.second->get_fileline() << ": error: "
+                                "Dimensions must be constant." << endl;
+                        cerr << range.second->get_fileline() << "       : "
+                                "This LSB expression violates the rule: "
+                             << *range.second << endl;
+                        dimension_ok = false;
+                        des->errors += 1;
+                  }
+                  delete texpr;
+            }
+      }
+
+        /* Error recovery */
+      if (!dimension_ok) {
+            index_l = 0;
+            index_r = 0;
+      }
+
+      return dimension_ok;
+}
+
+bool evaluate_ranges(Design*des, NetScope*scope, const LineInfo*li,
 		     vector<netrange_t>&llist,
 		     const list<pform_range_t>&rlist)
 {
-      bool bad_msb = false, bad_lsb = false;
+      bool dimensions_ok = true;
 
       for (list<pform_range_t>::const_iterator cur = rlist.begin()
 		 ; cur != rlist.end() ; ++cur) {
-	    long use_msb, use_lsb;
-
-	    NetExpr*texpr = elab_and_eval(des, scope, cur->first, -1, true);
-	    if (! eval_as_long(use_msb, texpr)) {
-		  cerr << cur->first->get_fileline() << ": error: "
-			"Range expressions must be constant." << endl;
-		  cerr << cur->first->get_fileline() << "       : "
-			"This MSB expression violates the rule: "
-		       << *cur->first << endl;
-		  des->errors += 1;
-		  bad_msb = true;
-	    }
-
-	    delete texpr;
-
-	    texpr = elab_and_eval(des, scope, cur->second, -1, true);
-	    if (! eval_as_long(use_lsb, texpr)) {
-		  cerr << cur->second->get_fileline() << ": error: "
-			"Range expressions must be constant." << endl;
-		  cerr << cur->second->get_fileline() << "       : "
-			"This LSB expression violates the rule: "
-		       << *cur->second << endl;
-		  des->errors += 1;
-		  bad_lsb = true;
-	    }
-
-	    delete texpr;
-
-	      /* Error recovery */
-	    if (bad_lsb) use_lsb = 0;
-	    if (bad_msb) use_msb = use_lsb;
-
-	    llist.push_back(netrange_t(use_msb, use_lsb));
+            long index_l, index_r;
+            dimensions_ok &= evaluate_range(des, scope, li, *cur, index_l, index_r);
+            llist.push_back(netrange_t(index_l, index_r));
       }
 
-      return bad_msb | bad_lsb;
+      return dimensions_ok;
 }
 
 void eval_expr(NetExpr*&expr, int context_width)
@@ -1394,8 +1487,8 @@ bool evaluate_index_prefix(Design*des, NetScope*scope,
 {
       list<index_component_t>::const_iterator icur = indices.begin();
       for (size_t idx = 0 ; (idx+1) < indices.size() ; idx += 1, ++icur) {
-	    assert(icur != indices.end());
-	    assert(icur->sel == index_component_t::SEL_BIT);
+	    ivl_assert(*(icur->msb), icur != indices.end());
+	    ivl_assert(*(icur->msb), icur->sel == index_component_t::SEL_BIT);
 	    NetExpr*texpr = elab_and_eval(des, scope, icur->msb, -1, true);
 
 	    long tmp;
@@ -1406,7 +1499,7 @@ bool evaluate_index_prefix(Design*des, NetScope*scope,
 		  return false;
 	    }
 
-	    prefix_indices .push_back(tmp);
+	    prefix_indices.push_back(tmp);
 	    delete texpr;
       }
 
@@ -1428,7 +1521,7 @@ NetExpr*collapse_array_exprs(Design*des, NetScope*scope,
       indices_flags flags;
       indices_to_expressions(des, scope, loc, indices,
                              net->packed_dimensions(),
-                             false, net->unpacked_count(), flags, exprs, exprs_const);
+                             false, flags, exprs, exprs_const);
       ivl_assert(*loc, exprs.size() == net->packed_dimensions());
 
 	// Special Case: there is only 1 packed dimension, so the
@@ -1616,3 +1709,38 @@ NetScope* find_method_containing_scope(const LineInfo&, NetScope*scope)
 
       return scope;
 }
+
+
+/*
+ * Print a warning if we find a mixture of default and explicit timescale
+ * based delays in the design, since this is likely an error.
+ */
+void check_for_inconsistent_delays(NetScope*scope)
+{
+      static bool used_implicit_timescale = false;
+      static bool used_explicit_timescale = false;
+      static bool display_ts_dly_warning = true;
+
+      if (scope->time_from_timescale())
+	    used_explicit_timescale = true;
+      else
+	    used_implicit_timescale = true;
+
+      if (display_ts_dly_warning &&
+	  used_explicit_timescale &&
+	  used_implicit_timescale) {
+	    if (gn_system_verilog()) {
+		  cerr << "warning: Found both default and explicit "
+			  "timescale based delays. Use" << endl;
+		  cerr << "       : -Wtimescale to find the design "
+			  "element(s) with no explicit" << endl;
+		  cerr << "       : timescale." << endl;
+	    } else {
+		  cerr << "warning: Found both default and "
+			  "`timescale based delays. Use" << endl;
+		  cerr << "       : -Wtimescale to find the "
+			  "module(s) with no `timescale." << endl;
+	    }
+	    display_ts_dly_warning = false;
+      }
+}

netmisc.h

@@ -1,7 +1,7 @@
 #ifndef IVL_netmisc_H
 #define IVL_netmisc_H
 /*
- * Copyright (c) 1999-2014 Stephen Williams (steve@icarus.com)
+ * Copyright (c) 1999-2019 Stephen Williams (steve@icarus.com)
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -60,12 +60,32 @@ inline NetScope* symbol_search(const LineInfo*li,
 }
 
 /*
- * This function transforms an expression by padding the high bits
- * with V0 until the expression has the desired width. This may mean
- * not transforming the expression at all, if it is already wide
- * enough.
+ * This function transforms an expression by either zero or sign extending
+ * the high bits until the expression has the desired width. This may mean
+ * not transforming the expression at all, if it is already wide enough.
+ * The extension method and the returned expression type is determined by
+ * signed_flag.
  */
-extern NetExpr*pad_to_width(NetExpr*expr, unsigned wid, const LineInfo&info);
+extern NetExpr*pad_to_width(NetExpr*expr, unsigned wid, bool signed_flag,
+			    const LineInfo&info);
+/*
+ * This version determines the extension method from the base expression type.
+ */
+inline NetExpr*pad_to_width(NetExpr*expr, unsigned wid, const LineInfo&info)
+{
+      return pad_to_width(expr, wid, expr->has_sign(), info);
+}
+
+/*
+ * This function transforms an expression by either zero or sign extending
+ * or discarding the high bits until the expression has the desired width.
+ * This may mean not transforming the expression at all, if it is already
+ * the correct width. The extension method (if needed) and the returned
+ * expression type is determined by signed_flag.
+ */
+extern NetExpr*cast_to_width(NetExpr*expr, unsigned wid, bool signed_flag,
+			     const LineInfo&info);
+
 extern NetNet*pad_to_width(Design*des, NetNet*n, unsigned w,
                            const LineInfo&info);
 
@@ -181,8 +201,6 @@ extern void indices_to_expressions(Design*des, NetScope*scope,
 				   const list<index_component_t>&src, unsigned count,
 				     // True if the expression MUST be constant.
 				   bool need_const,
-				     // Total array size, for sizing expressions
-				   unsigned need_addr,
 				     // These are the outputs.
 				   indices_flags&flags,
 				   list<NetExpr*>&indices,list<long>&indices_const);
@@ -260,7 +278,8 @@ extern NetExpr* elab_and_eval(Design*des, NetScope*scope,
 			      PExpr*pe, int context_width,
                               bool need_const =false,
                               bool annotatable =false,
-			      ivl_variable_type_t cast_type =IVL_VT_NO_TYPE);
+			      ivl_variable_type_t cast_type =IVL_VT_NO_TYPE,
+			      bool force_unsigned =false);
 
 extern NetExpr* elab_and_eval_lossless(Design*des, NetScope*scope,
 			      PExpr*pe, int context_width,
@@ -299,9 +318,14 @@ extern NetExpr* elaborate_rval_expr(Design*des, NetScope*scope,
 				    ivl_type_t lv_net_type,
 				    ivl_variable_type_t lv_type,
 				    unsigned lv_width, PExpr*expr,
-                                    bool need_const =false);
+				    bool need_const =false,
+				    bool force_unsigned =false);
 
-extern bool evaluate_ranges(Design*des, NetScope*scope,
+extern bool evaluate_range(Design*des, NetScope*scope, const LineInfo*li,
+                           const pform_range_t&range,
+                           long&index_l, long&index_r);
+
+extern bool evaluate_ranges(Design*des, NetScope*scope, const LineInfo*li,
 			    std::vector<netrange_t>&llist,
 			    const std::list<pform_range_t>&rlist);
 /*
@@ -358,10 +382,6 @@ const char *human_readable_op(const char op, bool unary = false);
 enum const_bool { C_NON, C_0, C_1, C_X };
 const_bool const_logical(const NetExpr*expr);
 
-extern bool dly_used_no_timescale;
-extern bool dly_used_timescale;
-extern bool display_ts_dly_warning;
-
 /*
  * When scaling a real value to a time we need to do some standard
  * processing.
@@ -389,4 +409,10 @@ extern void assign_unpacked_with_bufz(Design*des, NetScope*scope,
 
 extern NetPartSelect* detect_partselect_lval(Link&pin);
 
+/*
+ * Print a warning if we find a mixture of default and explicit timescale
+ * based delays in the design, since this is likely an error.
+ */
+extern void check_for_inconsistent_delays(NetScope*scope);
+
 #endif /* IVL_netmisc_H */

netparray.cc

@@ -34,6 +34,12 @@ netparray_t::~netparray_t()
  * The packed width of a packed array is the packed width of the
  * element times the dimension width of the array itself.
  */
+
+bool netparray_t::packed(void) const
+{
+      return true;
+}
+
 long netparray_t::packed_width(void) const
 {
       long cur_width = element_type()->packed_width();

netparray.h

@@ -64,6 +64,7 @@ class netparray_t : public netsarray_t {
 
     public:
 	// Virtual methods from the ivl_type_s type...
+      bool packed(void) const;
       long packed_width(void) const;
       std::vector<netrange_t> slice_dimensions() const;
 

netqueue.h

@@ -1,7 +1,7 @@
 #ifndef IVL__netqueue_H
 #define IVL__netqueue_H
 /*
- * Copyright (c) 2014 Stephen Williams (steve@icarus.com)
+ * Copyright (c) 2014-2015 Stephen Williams (steve@icarus.com)
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -38,6 +38,9 @@ class netqueue_t : public netdarray_t {
 	// IVL_VT_QUEUE for queues.
       ivl_variable_type_t base_type() const;
 
+	// A queue may have a type that is signed.
+      inline bool get_signed() const { return element_type()->get_signed(); }
+
       std::ostream& debug_dump(std::ostream&) const;
 
     private:

nettypes.cc

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2012-2013 Stephen Williams (steve@icarus.com)
+ * Copyright (c) 2012-2016 Stephen Williams (steve@icarus.com)
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -141,19 +141,18 @@ bool prefix_to_slice(const std::vector<netrange_t>&dims,
 
 	// Now similarly go through the prefix numbers, working
 	// through the dimensions until we run out. Accumulate a
-	// growing slice width (acc_wid) that is used to caculate the
+	// growing slice width (acc_wid) that is used to calculate the
 	// growing offset (acc_off).
       list<long>::const_iterator icur = prefix.end();
       do {
 	    -- icur;
 	    acc_wid *= pcur->width();
+	    -- pcur;
 	    if (pcur->get_msb() >= pcur->get_lsb())
 		  acc_off += (*icur - pcur->get_lsb()) * acc_wid;
 	    else
 		  acc_off += (pcur->get_lsb() - *icur) * acc_wid;
 
-	    -- pcur;
-
       } while (icur != prefix.begin());
 
 	// Got our final offset.

nettypes.h

@@ -94,7 +94,7 @@ class netrange_t {
       inline netrange_t() : msb_(LONG_MAX), lsb_(LONG_MAX) { }
 	// Create a properly defined netrange
       inline netrange_t(long m, long l) : msb_(m), lsb_(l) { }
-	// Copy constructure.
+	// Copy constructor.
       inline netrange_t(const netrange_t&that)
       : msb_(that.msb_), lsb_(that.lsb_) { }
 

pad_to_width.cc

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 1999-2011 Stephen Williams (steve@icarus.com)
+ * Copyright (c) 1999-2016 Stephen Williams (steve@icarus.com)
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -24,21 +24,20 @@
 # include  "netmisc.h"
 
 
-/*
- * This function transforms an expression by padding the high bits
- * with V0 until the expression has the desired width. This may mean
- * not transforming the expression at all, if it is already wide
- * enough.
- */
-NetExpr*pad_to_width(NetExpr*expr, unsigned wid, const LineInfo&info)
+NetExpr*pad_to_width(NetExpr*expr, unsigned wid, bool signed_flag,
+		     const LineInfo&info)
 {
-      if (wid <= expr->expr_width())
+      if (wid <= expr->expr_width()) {
+	    expr->cast_signed(signed_flag);
 	    return expr;
+      }
 
 	/* If the expression is a const, then replace it with a wider
 	   const. This is a more efficient result. */
       if (NetEConst*tmp = dynamic_cast<NetEConst*>(expr)) {
-	    verinum oval = pad_to_width(tmp->value(), wid);
+	    verinum oval = tmp->value();
+	    oval.has_sign(signed_flag);
+	    oval = pad_to_width(oval, wid);
 	    tmp = new NetEConst(oval);
 	    tmp->set_line(info);
 	    delete expr;
@@ -46,8 +45,30 @@ NetExpr*pad_to_width(NetExpr*expr, unsigned wid, const LineInfo&info)
       }
 
       NetESelect*tmp = new NetESelect(expr, 0, wid);
+      tmp->cast_signed(signed_flag);
       tmp->set_line(info);
-      tmp->cast_signed(expr->has_sign());
+      return tmp;
+}
+
+NetExpr*cast_to_width(NetExpr*expr, unsigned wid, bool signed_flag,
+		      const LineInfo&info)
+{
+        /* If the expression is a const, then replace it with a new
+           const. This is a more efficient result. */
+      if (NetEConst*tmp = dynamic_cast<NetEConst*>(expr)) {
+            tmp->cast_signed(signed_flag);
+            if (wid != tmp->expr_width()) {
+                  tmp = new NetEConst(verinum(tmp->value(), wid));
+                  tmp->set_line(info);
+                  delete expr;
+            }
+            return tmp;
+      }
+
+      NetESelect*tmp = new NetESelect(expr, 0, wid);
+      tmp->cast_signed(signed_flag);
+      tmp->set_line(info);
+
       return tmp;
 }
 

parse_api.h

@@ -1,7 +1,7 @@
 #ifndef IVL_parse_api_H
 #define IVL_parse_api_H
 /*
- * Copyright (c) 2001-2014 Stephen Williams (steve@icarus.com)
+ * Copyright (c) 2001-2017 Stephen Williams (steve@icarus.com)
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -45,7 +45,7 @@ extern std::map<perm_string,PUdp*>   pform_primitives;
 extern std::map<perm_string,data_type_t*> pform_typedefs;
 extern std::set<enum_type_t*> pform_enum_sets;
 extern std::map<perm_string,PTaskFunc*> pform_tasks;
-extern std::map<perm_string,PClass*> pform_classes;
+extern std::vector<PClass*> pform_classes;
 extern std::map<perm_string,PPackage*> pform_packages;
 
 extern void pform_dump(std::ostream&out, const PClass*pac);

pform.h

@@ -1,7 +1,7 @@
 #ifndef IVL_pform_H
 #define IVL_pform_H
 /*
- * Copyright (c) 1998-2014 Stephen Williams (steve@icarus.com)
+ * Copyright (c) 1998-2016 Stephen Williams (steve@icarus.com)
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -136,6 +136,10 @@ extern void pform_set_default_nettype(NetNet::Type net,
      used to reject statements that cannot exist in program blocks. */
 extern bool pform_in_program_block(void);
 
+  /* Return true if currently processing an interface. This can be
+     used to reject statements that cannot exist in interfaces. */
+extern bool pform_in_interface(void);
+
 /*
  * Look for the given wire in the current lexical scope. If the wire
  * (including variables of any type) cannot be found in the current
@@ -152,23 +156,33 @@ extern PWire* pform_get_make_wire_in_scope(perm_string name, NetNet::Type net_ty
  * are to apply to the scope of that module. The endmodule causes the
  * pform to close up and finish the named module.
  *
- * The program_flag indicates that the module is actually a program
- * block. This has implications during parse and during
+ * The program_block flag indicates that the module is actually a program
+ * block. The is_interface flag indicates that the module is actually
+ * an interface. These flags have implications during parse and during
  * elaboration/code generation.
  */
 extern void pform_startmodule(const struct vlltype&loc, const char*name,
-			      bool program_block, list<named_pexpr_t>*attr);
+			      bool program_block, bool is_interface,
+			      LexicalScope::lifetime_t lifetime,
+			      list<named_pexpr_t>*attr);
 extern void pform_check_timeunit_prec();
 extern void pform_module_set_ports(vector<Module::port_t*>*);
 
-/* This function is used to support the port definition in a
-   port_definition_list. In this case, we have everything needed to
-   define the port, all in one place. */
+/* These functions are used when we have a complete port definition, either
+   in an ansi style or non-ansi style declaration. In this case, we have
+   everything needed to define the port, all in one place. */
 extern void pform_module_define_port(const struct vlltype&li,
 				     perm_string name,
 				     NetNet::PortType,
 				     NetNet::Type type,
 				     data_type_t*vtype,
+				     list<named_pexpr_t>*attr,
+				     bool keep_attr =false);
+extern void pform_module_define_port(const struct vlltype&li,
+				     list<pform_port_t>*ports,
+				     NetNet::PortType,
+				     NetNet::Type type,
+				     data_type_t*vtype,
 				     list<named_pexpr_t>*attr);
 
 extern Module::port_t* pform_module_port_reference(perm_string name,
@@ -180,7 +194,8 @@ extern void pform_endmodule(const char*, bool inside_celldefine,
 extern void pform_start_class_declaration(const struct vlltype&loc,
 					  class_type_t*type,
 					  data_type_t*base_type,
-					  std::list<PExpr*>*base_exprs);
+					  std::list<PExpr*>*base_exprs,
+					  LexicalScope::lifetime_t lifetime);
 extern void pform_class_property(const struct vlltype&loc,
 				 property_qualifier_t pq,
 				 data_type_t*data_type,
@@ -205,7 +220,8 @@ extern void pform_make_udp(perm_string name,
  * Package related functions.
  */
 extern void pform_start_package_declaration(const struct vlltype&loc,
-					    const char*type);
+					    const char*type,
+					    LexicalScope::lifetime_t lifetime);
 extern void pform_end_package_declaration(const struct vlltype&loc);
 extern void pform_package_import(const struct vlltype&loc,
 				 PPackage*pkg, const char*ident);
@@ -213,6 +229,15 @@ extern void pform_package_import(const struct vlltype&loc,
 extern PExpr* pform_package_ident(const struct vlltype&loc,
 				  PPackage*pkg, pform_name_t*ident);
 
+/*
+ * Interface related functions.
+ */
+extern void pform_start_modport_item(const struct vlltype&loc, const char*name);
+extern void pform_end_modport_item(const struct vlltype&loc);
+extern void pform_add_modport_port(const struct vlltype&loc,
+	                           NetNet::PortType port_type,
+	                           perm_string name, PExpr*expr);
+
 /*
  * This creates an identifier aware of names that may have been
  * imported from other packages.
@@ -231,13 +256,20 @@ extern void pform_pop_scope();
  */
 extern LexicalScope* pform_peek_scope();
 
-extern PClass* pform_push_class_scope(const struct vlltype&loc, perm_string name);
+extern PClass* pform_push_class_scope(const struct vlltype&loc, perm_string name,
+				      LexicalScope::lifetime_t lifetime);
+
 extern PFunction*pform_push_constructor_scope(const struct vlltype&loc);
-extern PPackage* pform_push_package_scope(const struct vlltype&loc, perm_string name);
+
+extern PPackage* pform_push_package_scope(const struct vlltype&loc, perm_string name,
+					  LexicalScope::lifetime_t lifetime);
+
 extern PTask*pform_push_task_scope(const struct vlltype&loc, char*name,
-                                   bool is_auto);
+				   LexicalScope::lifetime_t lifetime);
+
 extern PFunction*pform_push_function_scope(const struct vlltype&loc, const char*name,
-                                           bool is_auto);
+					   LexicalScope::lifetime_t lifetime);
+
 extern PBlock*pform_push_block_scope(char*name, PBlock::BL_TYPE tt);
 
 extern void pform_put_behavior_in_scope(AProcess*proc);
@@ -272,7 +304,8 @@ extern void pform_endgenerate();
  */
 extern PGenerate* pform_parent_generate(void);
 
-extern void pform_set_typedef(perm_string name, data_type_t*data_type);
+extern void pform_set_typedef(perm_string name, data_type_t*data_type,
+			      std::list<pform_range_t>*unp_ranges);
 
 /*
  * This function makes a PECallFunction of the named function. Decide
@@ -335,17 +368,17 @@ extern void pform_makewire(const struct vlltype&li,
 			   list<perm_string>*names,
 			   list<named_pexpr_t>*attr);
 
-extern void pform_make_reginit(const struct vlltype&li,
-			       perm_string name, PExpr*expr);
+extern void pform_make_var_init(const struct vlltype&li,
+				perm_string name, PExpr*expr);
 
-  /* Look up the names of the wires, and set the port type,
-     i.e. input, output or inout. If the wire does not exist, create
-     it. The second form takes a single name. */
+/* This function is used when we have an incomplete port definition in
+   a non-ansi style declaration. Look up the names of the wires, and set
+   the port type, i.e. input, output or inout, and, if specified, the
+   range and signedness. If the wire does not exist, create it. */
 extern void pform_set_port_type(const struct vlltype&li,
-				list<perm_string>*names,
-				list<pform_range_t>*range,
-				bool signed_flag,
+				list<pform_port_t>*ports,
 				NetNet::PortType,
+				data_type_t*dt,
 				list<named_pexpr_t>*attr);
 
 extern void pform_set_reg_idx(perm_string name,

pform_dump.cc

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 1998-2014 Stephen Williams (steve@icarus.com)
+ * Copyright (c) 1998-2016 Stephen Williams (steve@icarus.com)
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -318,7 +318,7 @@ void PECallFunction::dump(ostream &out) const
 
 void PECastSize::dump(ostream &out) const
 {
-      out << size_ << "'(";
+      out << *size_ << "'(";
       base_->dump(out);
       out << ")";
 }
@@ -789,6 +789,8 @@ void PBlock::dump(ostream&out, unsigned ind) const
             dump_events_(out, ind+2);
 
 	    dump_wires_(out, ind+2);
+
+	    dump_var_inits_(out, ind+2);
       }
 
       for (unsigned idx = 0 ;  idx < list_.size() ;  idx += 1) {
@@ -926,7 +928,10 @@ void PDisable::dump(ostream&out, unsigned ind) const
 void PDoWhile::dump(ostream&out, unsigned ind) const
 {
       out << setw(ind) << "" << "do" << endl;
-      statement_->dump(out, ind+3);
+      if (statement_)
+	    statement_->dump(out, ind+3);
+      else
+	    out << setw(ind+3) << "" << "/* NOOP */" << endl;
       out << setw(ind) << "" << "while (" << *cond_ << ");" << endl;
 }
 
@@ -989,13 +994,19 @@ void PForeach::dump(ostream&fd, unsigned ind) const
       }
 
       fd << "] /* " << get_fileline() << " */" << endl;
-      statement_->dump(fd, ind+3);
+      if (statement_)
+	    statement_->dump(fd, ind+3);
+      else
+	    fd << setw(ind+3) << "" << "/* NOOP */" << endl;
 }
 
 void PForever::dump(ostream&out, unsigned ind) const
 {
       out << setw(ind) << "" << "forever /* " << get_fileline() << " */" << endl;
-      statement_->dump(out, ind+3);
+      if (statement_)
+	    statement_->dump(out, ind+3);
+      else
+	    out << setw(ind+3) << "" << "/* NOOP */" << endl;
 }
 
 void PForStatement::dump(ostream&out, unsigned ind) const
@@ -1003,7 +1014,10 @@ void PForStatement::dump(ostream&out, unsigned ind) const
       out << setw(ind) << "" << "for (" << *name1_ << " = " << *expr1_
 	  << "; " << *cond_ << "; <for_step>)" << endl;
       step_->dump(out, ind+6);
-      statement_->dump(out, ind+3);
+      if (statement_)
+	    statement_->dump(out, ind+3);
+      else
+	    out << setw(ind+3) << "" << "/* NOOP */" << endl;
 }
 
 void PFunction::dump(ostream&out, unsigned ind) const
@@ -1030,6 +1044,8 @@ void PFunction::dump(ostream&out, unsigned ind) const
 
       dump_wires_(out, ind+2);
 
+      dump_var_inits_(out, ind+2);
+
       if (statement_)
 	    statement_->dump(out, ind+2);
       else
@@ -1045,7 +1061,10 @@ void PRelease::dump(ostream&out, unsigned ind) const
 void PRepeat::dump(ostream&out, unsigned ind) const
 {
       out << setw(ind) << "" << "repeat (" << *expr_ << ")" << endl;
-      statement_->dump(out, ind+3);
+      if (statement_)
+	    statement_->dump(out, ind+3);
+      else
+	    out << setw(ind+3) << "" << "/* NOOP */" << endl;
 }
 
 void PReturn::dump(ostream&fd, unsigned ind) const
@@ -1072,6 +1091,8 @@ void PTask::dump(ostream&out, unsigned ind) const
 
       dump_wires_(out, ind+2);
 
+      dump_var_inits_(out, ind+2);
+
       if (statement_)
 	    statement_->dump(out, ind+2);
       else
@@ -1126,7 +1147,10 @@ void PTrigger::dump(ostream&out, unsigned ind) const
 void PWhile::dump(ostream&out, unsigned ind) const
 {
       out << setw(ind) << "" << "while (" << *cond_ << ")" << endl;
-      statement_->dump(out, ind+3);
+      if (statement_)
+	    statement_->dump(out, ind+3);
+      else
+	    out << setw(ind+3) << "" << "/* NOOP */" << endl;
 }
 
 void PProcess::dump(ostream&out, unsigned ind) const
@@ -1136,7 +1160,10 @@ void PProcess::dump(ostream&out, unsigned ind) const
 
       dump_attributes_map(out, attributes, ind+2);
 
-      statement_->dump(out, ind+2);
+      if (statement_)
+	    statement_->dump(out, ind+2);
+      else
+	    out << setw(ind+2) << "" << "/* NOOP */" << endl;
 }
 
 void AProcess::dump(ostream&out, unsigned ind) const
@@ -1157,7 +1184,10 @@ void AProcess::dump(ostream&out, unsigned ind) const
 
       dump_attributes_map(out, attributes, ind+2);
 
-      statement_->dump(out, ind+2);
+      if (statement_)
+	    statement_->dump(out, ind+2);
+      else
+	    out << setw(ind+2) << "" << "/* NOOP */" << endl;
 }
 
 void PSpecPath::dump(std::ostream&out, unsigned ind) const
@@ -1269,6 +1299,8 @@ void PGenerate::dump(ostream&out, unsigned indent) const
 	    (*idx)->dump(out, indent+2);
       }
 
+      dump_var_inits_(out, indent+2);
+
       for (list<PProcess*>::const_iterator idx = behaviors.begin()
 		 ; idx != behaviors.end() ; ++ idx ) {
 	    (*idx)->dump(out, indent+2);
@@ -1408,6 +1440,14 @@ void LexicalScope::dump_wires_(ostream&out, unsigned indent) const
       }
 }
 
+void LexicalScope::dump_var_inits_(ostream&out, unsigned indent) const
+{
+	// Iterate through and display all the register initializations.
+      for (unsigned idx = 0; idx < var_inits.size(); idx += 1) {
+	    var_inits[idx]->dump(out, indent);
+      }
+}
+
 void PScopeExtra::dump_classes_(ostream&out, unsigned indent) const
 {
 	// Dump the task definitions.
@@ -1564,6 +1604,7 @@ void Module::dump(ostream&out) const
 	    (*gate)->dump(out);
       }
 
+      dump_var_inits_(out, 4);
 
       for (list<PProcess*>::const_iterator behav = behaviors.begin()
 		 ; behav != behaviors.end() ; ++ behav ) {

pform_package.cc

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2012 Stephen Williams (steve@icarus.com)
+ * Copyright (c) 2012-2016 Stephen Williams (steve@icarus.com)
  * Copyright CERN 2013 Stephen Williams (steve@icarus.com)
  *
  *    This source code is free software; you can redistribute it
@@ -35,12 +35,13 @@ map<perm_string,PPackage*> pform_packages;
 
 static PPackage*pform_cur_package = 0;
 
-void pform_start_package_declaration(const struct vlltype&loc, const char*name)
+void pform_start_package_declaration(const struct vlltype&loc, const char*name,
+				     LexicalScope::lifetime_t lifetime)
 {
       ivl_assert(loc, pform_cur_package == 0);
 
       perm_string use_name = lex_strings.make(name);
-      PPackage*pkg_scope = pform_push_package_scope(loc, use_name);
+      PPackage*pkg_scope = pform_push_package_scope(loc, use_name, lifetime);
       FILE_NAME(pkg_scope, loc);
       pform_cur_package = pkg_scope;
 }
@@ -74,6 +75,7 @@ void pform_end_package_declaration(const struct vlltype&loc)
 void pform_package_import(const struct vlltype&, PPackage*pkg, const char*ident)
 {
       LexicalScope*scope = pform_peek_scope();
+      assert(scope);
 
       if (ident) {
 	    perm_string use_ident = lex_strings.make(ident);
@@ -152,6 +154,11 @@ void pform_package_import(const struct vlltype&, PPackage*pkg, const char*ident)
 
 		  scope->imports[cur->first] = pkg;
 	    }
+
+            for (set<enum_type_t*>::const_iterator cur = pkg->enum_sets.begin()
+		       ; cur != pkg->enum_sets.end() ; ++ cur) {
+                  scope->enum_sets.insert(*cur);
+            }
       }
 }
 

pform_pclass.cc

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2012-2014 Stephen Williams (steve@icarus.com)
+ * Copyright (c) 2012-2016 Stephen Williams (steve@icarus.com)
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -36,9 +36,13 @@ static PClass*pform_cur_class = 0;
  * if present, are the "exprs" that would be passed to a chained
  * constructor.
  */
-void pform_start_class_declaration(const struct vlltype&loc, class_type_t*type, data_type_t*base_type, list<PExpr*>*base_exprs)
+void pform_start_class_declaration(const struct vlltype&loc,
+				   class_type_t*type,
+				   data_type_t*base_type,
+				   list<PExpr*>*base_exprs,
+				   LexicalScope::lifetime_t lifetime)
 {
-      PClass*class_scope = pform_push_class_scope(loc, type->name);
+      PClass*class_scope = pform_push_class_scope(loc, type->name, lifetime);
       class_scope->type = type;
       assert(pform_cur_class == 0);
       pform_cur_class = class_scope;
@@ -75,6 +79,7 @@ void pform_class_property(const struct vlltype&loc,
 	    if (! curp->index.empty()) {
 		  list<pform_range_t>*pd = new list<pform_range_t> (curp->index);
 		  use_type = new uarray_type_t(use_type, pd);
+		  FILE_NAME(use_type, loc);
 	    }
 
 	    pform_cur_class->type->properties[curp->name]
@@ -127,7 +132,7 @@ void pform_set_constructor_return(PFunction*net)
 PFunction*pform_push_constructor_scope(const struct vlltype&loc)
 {
       assert(pform_cur_class);
-      PFunction*func = pform_push_function_scope(loc, "new", true);
+      PFunction*func = pform_push_function_scope(loc, "new", LexicalScope::AUTOMATIC);
       return func;
 }
 
@@ -138,7 +143,7 @@ void pform_end_class_declaration(const struct vlltype&loc)
 	// If there were initializer statements, then collect them
 	// into an implicit constructor function.
       if (! pform_cur_class->type->initialize.empty()) {
-	    PFunction*func = pform_push_function_scope(loc, "new@", true);
+	    PFunction*func = pform_push_function_scope(loc, "new@", LexicalScope::AUTOMATIC);
 	    func->set_ports(0);
 	    pform_set_constructor_return(func);
 	    pform_set_this_class(loc, func);

pform_struct_type.cc

@@ -79,8 +79,8 @@ static void pform_set_struct_type(struct_type_t*struct_type, perm_string name, N
 	    return;
       }
 
-	// For now, can only handle packed structs.
-      ivl_assert(*struct_type, 0);
+	// For now, can only handle packed structs. The parser generates
+	// a "sorry" message, so no need to do anything here.
 }
 
 void pform_set_struct_type(struct_type_t*struct_type, list<perm_string>*names, NetNet::Type net_type, list<named_pexpr_t>*attr)

pform_types.h

@@ -1,7 +1,7 @@
 #ifndef IVL_pform_types_H
 #define IVL_pform_types_H
 /*
- * Copyright (c) 2007-2014 Stephen Williams (steve@icarus.com)
+ * Copyright (c) 2007-2018 Stephen Williams (steve@icarus.com)
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -33,6 +33,10 @@
 # include  <map>
 # include  <memory>
 
+#if __cplusplus < 201103L
+#define unique_ptr auto_ptr
+#endif
+
 /*
  * parse-form types.
  */
@@ -50,7 +54,7 @@ typedef named<verinum> named_number_t;
 typedef named<PExpr*> named_pexpr_t;
 
 /*
- * The pform_range_t holds variable diimensions for type
+ * The pform_range_t holds variable dimensions for type
  * declarations. The two expressions are interpreted as the first and
  * last values of the range. For example:
  *
@@ -72,6 +76,21 @@ typedef named<PExpr*> named_pexpr_t;
  */
 typedef std::pair<PExpr*,PExpr*> pform_range_t;
 
+/*
+ * The pform_port_t holds the name and optional unpacked dimensions
+ * and initialization expression for a single port in a list of port
+ * declarations.
+ */
+struct pform_port_t {
+      pform_port_t(perm_string n, list<pform_range_t>*ud, PExpr*e)
+	: name(n), udims(ud), expr(e) { }
+      ~pform_port_t() { }
+
+      perm_string name;
+      list<pform_range_t>*udims;
+      PExpr*expr;
+};
+
 /*
  * Semantic NOTES:
  * - The SEL_BIT is a single expression. This might me a bit select
@@ -102,7 +121,7 @@ struct name_component_t {
 struct decl_assignment_t {
       perm_string name;
       std::list<pform_range_t>index;
-      std::auto_ptr<PExpr> expr;
+      std::unique_ptr<PExpr> expr;
 };
 
 struct pform_tf_port_t {
@@ -155,14 +174,14 @@ struct enum_type_t : public data_type_t {
       ivl_variable_type_t base_type;
       bool signed_flag;
       bool integer_flag; // True if "integer" was used
-      std::auto_ptr< list<pform_range_t> > range;
-      std::auto_ptr< list<named_pexpr_t> > names;
+      std::unique_ptr< list<pform_range_t> > range;
+      std::unique_ptr< list<named_pexpr_t> > names;
       LineInfo li;
 };
 
 struct struct_member_t : public LineInfo {
-      std::auto_ptr<data_type_t> type;
-      std::auto_ptr< list<decl_assignment_t*> > names;
+      std::unique_ptr<data_type_t> type;
+      std::unique_ptr< list<decl_assignment_t*> > names;
       void pform_dump(std::ostream&out, unsigned indent) const;
 };
 
@@ -173,7 +192,7 @@ struct struct_type_t : public data_type_t {
 
       bool packed_flag;
       bool union_flag;
-      std::auto_ptr< list<struct_member_t*> > members;
+      std::unique_ptr< list<struct_member_t*> > members;
 };
 
 struct atom2_type_t : public data_type_t {
@@ -219,7 +238,7 @@ struct vector_type_t : public data_type_t {
       bool reg_flag; // True if "reg" was used
       bool integer_flag; // True if "integer" was used
       bool implicit_flag; // True if this type is implicitly logic/reg
-      std::auto_ptr< list<pform_range_t> > pdims;
+      std::unique_ptr< list<pform_range_t> > pdims;
 };
 
 struct array_base_t : public data_type_t {
@@ -228,7 +247,7 @@ struct array_base_t : public data_type_t {
       : base_type(btype), dims(pd) { }
 
       data_type_t*base_type;
-      std::auto_ptr< list<pform_range_t> > dims;
+      std::unique_ptr< list<pform_range_t> > dims;
 };
 
 /*
@@ -361,4 +380,8 @@ extern std::ostream& operator<< (std::ostream&out, const pform_name_t&);
 extern std::ostream& operator<< (std::ostream&out, const name_component_t&that);
 extern std::ostream& operator<< (std::ostream&out, const index_component_t&that);
 
+#if __cplusplus < 201103L
+#undef unique_ptr
+#endif
+
 #endif /* IVL_pform_types_H */

scripts/MAKE_RELEASE.sh

@@ -4,22 +4,26 @@
 # the number for a snapshot and the path to a temporary directory.
 # for example:
 #
-#    sh scripts/MAKE_RELEASE.sh 0.9.1 ~/tmp
+#    sh scripts/MAKE_RELEASE.sh 10.1 ~/tmp
 #
-# The above assumes that there is a tag "v0_9_1" at the point
+# The above assumes that there is a tag "v10_1" at the point
 # to be released. (The tag has the "v", but the argument to this
 # script does not have the "v"). This script extracts based on the
 # tag, uses the temporary directory to stage intermediate results,
-# and finally creates a file called verilog-0.9.1.tar.gz that
+# and finally creates a file called verilog-10.1.tar.gz that
 # contains the release ready to go.
 #
-# The complete steps to make a release x.y.z generally are:
+# The complete steps to make a release x.y generally are:
 #
-#   git tag -a v0_9_1
+#   Edit version_base.h to suit.
+#
+#   Edit verilog.spec to suit.
+#
+#   git tag -a v10_1
 #                 (Make the tag in the local git repository.)
 #
-#   sh scripts/MAKE_RELEASE.sh 0.9.1 ~/tmp
-#                 (Make the snapshot bundle verilog-0.9.1.tar.gz)
+#   sh scripts/MAKE_RELEASE.sh 10.1 ~/tmp
+#                 (Make the snapshot bundle verilog-10.1.tar.gz)
 #
 #   git push --tags
 #                 (Publish the tag to the repository.)

scripts/devel-stub.conf

@@ -15,7 +15,7 @@ iwidth:32
 sys_func:vpi/system.sft
 sys_func:vpi/v2005_math.sft
 sys_func:vpi/va_math.sft
-warnings:ailnpstv
+warnings:adfilnpstv
 debug:eval_tree
 debug:elaborate
 debug:emit

sv_vpi_user.h

@@ -21,7 +21,7 @@
 
 # include  "vpi_user.h"
 
-#if defined(__MINGW32__) || defined (__CYGWIN32__)
+#if defined(__MINGW32__) || defined (__CYGWIN__)
 #  define DLLEXPORT __declspec(dllexport)
 #else
 #  define DLLEXPORT

symbol_search.cc

@@ -88,7 +88,7 @@ static bool symbol_search(const LineInfo*li, Design*des, NetScope*scope,
 		  return false;
 
 	      // The prefix is found to be a scope, so switch to that
-	      // scope, set the hier_path to turn of upwards searches,
+	      // scope, set the hier_path to turn off upwards searches,
 	      // and continue our search for the tail.
 	    if (recurse.is_scope()) {
 		  scope = recurse.scope;

syn-rules.y

@@ -1,7 +1,7 @@
 
 %{
 /*
- * Copyright (c) 2000-2014 Stephen Williams ([email protected])
+ * Copyright (c) 2000-2015 Stephen Williams ([email protected])
  *
  *    This source code is free software; you can redistribute it
  *    and/or modify it in source code form under the terms of the GNU
@@ -121,11 +121,6 @@ static void hookup_DFF_CE(NetFF*ff, NetESignal*d, NetEvProbe*pclk,
       connect(ff->pin_Clock(), pclk->pin(0));
       if (ce) connect(ff->pin_Enable(), ce->pin(0));
 
-      ff->attribute(perm_string::literal("LPM_FFType"), verinum("DFF"));
-      if (pclk->edge() == NetEvProbe::NEGEDGE)
-	    ff->attribute(perm_string::literal("Clock:LPM_Polarity"), verinum("INVERT"));
-
-
 	/* This lval_ represents a reg that is a WIRE in the
 	   synthesized results. This function signals the destructor
 	   to change the REG that this l-value refers to into a
@@ -165,7 +160,8 @@ static void make_DFF_CE(Design*des, NetProcTop*top,
 
 	    if (a->sig()) {
               // cerr << "new NetFF named " << a->name() << endl;
-              NetFF*ff = new NetFF(top->scope(), a->name(),
+	      bool negedge = pclk->edge() == NetEvProbe::NEGEDGE;
+              NetFF*ff = new NetFF(top->scope(), a->name(), negedge,
 				   a->sig()->vector_width());
               hookup_DFF_CE(ff, d, pclk, ce, a, rval_pinoffset);
               des->add_node(ff);