Changeset 1894

Timestamp:

Aug 6, 2022, 9:35:08 PM (3 years ago)

Author:

Gregg Young

Message:

Update mapsymw.pl

File:

• trunk/debugtools/mapsymw.pl (modified) (14 diffs)

trunk/debugtools/mapsymw.pl

@@ -1 +1 @@
 #!perl -w
-# mapsymw - mapsym wrapper Watcom map files
-# $Id: $
-# Copyright (c) 2007, 2010 Steven Levine and Associates, Inc.
+# mapsymw - mapsym wrapper form OpenWatcom map files
+# Generate IBM format map file and run mapsym
+# Writes IBM format map file to OpenWatcom map file directory
+
+# Copyright (c) 2007-2019 Steven Levine and Associates, Inc.
 # All rights reserved.
 
@@ -9 +11 @@
 # gnugpl2.txt or at http://www.gnu.org/licenses/licenses.html#GPL
 
-# 02 Jul 07 SHL Baseline
-# 02 Jul 07 SHL Adapt from mapsymb.pl
-# 28 Jul 07 SHL Relax module name detect
-# 30 Jul 07 SHL Auto-trim libstdc++ symbols from libc06x maps
-# 09 Aug 07 SHL Generate dummy symbol for interior segments with no symbols
-# 08 Nov 07 SHL Drop leading keywords from function definitions
-# 14 Dec 08 SHL Ensure symbols sorted by value - some apps care
-# 03 May 10 SHL Comments
-# 14 Jun 10 SHL Avoid missing C++ symbols
+# 2007-07-02 SHL Baseline
+# 2008-12-14 SHL Ensure symbols sorted by value - some apps care
+# 2012-03-19 SHL Segment names must be uppercase for pmdf
+# 2014-06-13 SHL Correct typos
+# 2014-11-01 SHL Generate ibm map files in watcom map file directory
+# 2015-01-31 SHL Use unix slashes to keep kLIBC sh.exe happy
+# 2019-03-02 SHL More C++ logic
+# 2019-03-03 SHL Rework dup logic and sym length limiting
+# 2019-03-22 SHL Disable debug code
+# 2019-04-09 SHL Comments
 
 # mapsym requires each segment to have at least 1 symbol
@@ -31 +34 @@
 use POSIX qw(strftime);
 use Getopt::Std;
-use File::Spec;
 use File::Basename;
 
-our $g_version = '0.3';
+our $g_version = '0.4';
 
 our $g_cmdname;
-our $g_tmpdir;
 our @g_mapfiles;                        # All map files
 our $g_mapfile;                         # Current .map file name
@@ -50 +51 @@
 
 foreach $g_mapfile (@g_mapfiles) {
+  $g_mapfile =~ tr|\\|/|;       # 2015-01-31 SHL Unix slashes for sh.exe
   &mapsym;
 }
@@ -57 +59 @@
 # end main
 
-#=== initialize() Intialize globals ===
+#=== initialize() Initialize globals ===
 
 sub initialize {
 
   &set_cmd_name;
-  &get_tmp_dir;
 
 } # initialize
@@ -79 +80 @@
   open MAPFILE, $g_mapfile or die "open $g_mapfile $!";
 
-  my $g_wrkfile = File::Spec->catfile($g_tmpdir, "$mapid.map");
+  my $g_wrkfile = $g_mapfile . '-ibm';
+  $g_wrkfile =~ s/-watcom-ibm$/-ibm/;   # In case renamed from .map
   unlink $g_wrkfile || die "unlink $g_wrkfile $!" if -f $g_wrkfile;
   open WRKFILE, ">$g_wrkfile" or die "open $g_wrkfile $!";
@@ -99 +101 @@
   while (<MAPFILE>) {
 
-    chomp;                      # EOL
+    chomp;                              # EOL
 
     if (/Executable Image: (\S+)\.\w+$/) {
@@ -177 +179 @@
 
         my $seginfo;
-        if (defined($segsinfo{$1})) {
-          $seginfo = $segsinfo{$1};
-        }
-        else {
+        if (defined($segsinfo{$segnum})) {
+          $seginfo = $segsinfo{$segnum};
+        } else {
           $seginfo = {max_offset => 0,
                       symcnt => 0};
@@ -190 +191 @@
         $seginfo->{symcnt}++;
 
-        $segsinfo{$1} = $seginfo;
+        $segsinfo{$segnum} = $seginfo;
 
         $segaddr = "$segnum:$offset";
 
         # Convert C++ symbols to something mapsym will accept
-        # warn "$sym\n";
+        # warn "\n$sym\n";
 
         $_ = $sym;
 
-        # s/\bIdle\b/    /;     # Drop Idle keyword - obsolete done later
-        s/\(.*\).*$//;          # Drop (...) tails
-
-        s/::~/__x/;             # Replace ::~ with __x
-        s/::/__/;               # Replace :: with __
-
-        s/[<,]/_/g;             # Replace < and , with _
-        s/[>]//g;               # Replace > with nothing
-        s/[\[\]]//g;            # Replace [] with nothing
-        # s/_*$//;              # Drop trailing _
-        # s/\W+\w//;            # Drop leading keywords (including Idle)
-        s/\b.*\b\s+//g;         # Drop leading keywords (including Idle)
+        s/\s+\(.*\)//;                  # Drop (...) comments preceeded by whitespace
+
+        # Make C++ compatible with mapsym
+
+        # Drop keywords and decorations
+        s/\b(near|int|short|unsigned|long|void|char|const|wchar_t)\b\s*[*&]?\s*//g;
+
+        # s/<[^<>]+>//g;                # Replace < > templates with nothing
+
+        s/std::\s*/std/g;               # Replace std:: with std
+        s/::~/__x/g;                    # Replace destructor ::~ with __x
+        s/::/__/g;                      # Replace :: with __
+
+        s/[\[\]()<>,]/ /g;              # Replace [ ] < > and , with spaces
+        s/\s+/_/g;                      # Replace whitespace with single _
+        s/__+/_/g;                      # Replace multiple _ with single _
 
         # Drop leading and trailing _ to match source code
 
-        s/^_//;                 # Drop leading _ (cdecl)
-        s/_$//;                 # Drop trailing _ (watcall)
-
-        # warn "$_\n";
+        s/^_//;                         # Drop leading _ (cdecl)
+        s/_$//;                         # Drop trailing _ (watcall)
+
+        $_ = substr($_, 0, 63) if length($_) > 63;              # limit
+
+        # warn "\n$_\n";
 
         # Prune some libc symbols to avoid mapsym overflows
@@ -261 +268 @@
       if (defined($segsinfo{$segnum})) {
         $seginfo = $segsinfo{$segnum};
-      }
-      else {
+      } else {
         $seginfo = {max_offset => 0,
                     symcnt => 0};
@@ -268 +274 @@
       if ($seginfo->{symcnt} == 0) {
         warn "Segment $segnum has no symbols - generating dummy symbol\n";
-        $_ = "Seg${segnum}_dummy";
+        $_ = "SEG${segnum}_dummy";
         if ($is32bit) {
           $segaddr = "$segnum:00010000";
@@ -278 +284 @@
       } elsif ($is32bit && $seginfo->{max_offset} < 0x10000) {
         warn "32 bit segment $segnum is smaller than 64K - generating dummy symbol\n";
-        $_ = "Seg${segnum}_dummy";
+        $_ = "SEG${segnum}_dummy";
         $segaddr = "$segnum:00010000";
         $syms{$segaddr} = $_;
@@ -284 +290 @@
       }
     }
-  } # foreach
+  } # foreach $segnum
 
   # Generate symbols by value listing
-
-  my $lastsym = '';
-  my $seq = 0;
+  # FIXME to warn if may overflow 64K limit
+
   @keys = sort keys %syms;
+  my %used;
   foreach $segaddr (@keys) {
     my $sym = $syms{$segaddr};
     my $imp = substr($segaddr, 0, 4) eq '0000' ? 'Imp' : '';
-    if ($sym ne $lastsym) {
-      $lastsym = $sym;
-      $seq = 0;
+    # Avoid dups
+    if (defined $used{$sym}) {
+      my $seq = $used{$sym};
+      $seq++;
+      $used{$sym} = $seq;
+      $sym = "${sym}_$seq";
     } else {
-      $seq++;
-      $sym = "${sym}_$seq";
+      $used{$sym} = 0;
     }
     printf WRKFILE $symfmt, $segaddr, $imp, $sym;
-  }
+  } # foreach
 
   close WRKFILE;
@@ -416 +424 @@
 } # set_cmd_name
 
-#=== get_tmp_dir() Get TMP dir name with trailing backslash, set Gbl. ===
-
-sub get_tmp_dir {
-
-  $g_tmpdir = File::Spec->tmpdir();
-  die "Need to have TMP or TMPDIR or TEMP defined" unless $g_tmpdir;
-
-} # get_tmp_dir
-
 # The end