[Valgrind-developers] valgrind: r5449 - branches/VALGRIND_3_1_BRANCH/cachegrind

[<sv...@va...>]

Author: sewardj
Date: 2005-12-27 14:46:50 +0000 (Tue, 27 Dec 2005)
New Revision: 5449

Log:
Merge r5396 (fix for: cg_annotate's --sort option broken).  Needs check.

Modified:
   branches/VALGRIND_3_1_BRANCH/cachegrind/cg_annotate.in


Modified: branches/VALGRIND_3_1_BRANCH/cachegrind/cg_annotate.in
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
--- branches/VALGRIND_3_1_BRANCH/cachegrind/cg_annotate.in	2005-12-27 14:=
43:31 UTC (rev 5448)
+++ branches/VALGRIND_3_1_BRANCH/cachegrind/cg_annotate.in	2005-12-27 14:=
46:50 UTC (rev 5449)
@@ -1,9 +1,7 @@
-#! @PERL@ -w
+#! @PERL@
=20
 ##--------------------------------------------------------------------##
-##--- The cache simulation framework: instrumentation, recording   ---##
-##--- and results printing.                                        ---##
-##---                                               cg_annotate.in ---##
+##--- Cachegrind's annotator.                       cg_annotate.in ---##
 ##--------------------------------------------------------------------##
=20
 #  This file is part of Cachegrind, a Valgrind tool for cache
@@ -30,10 +28,53 @@
 #  The GNU General Public License is contained in the file COPYING.
=20
 #-----------------------------------------------------------------------=
-----
-# Annotator for cachegrind.=20
-#
-# File format is described in /docs/techdocs.html.
-#
+# The file format is simple, basically printing the cost centre for ever=
y
+# source line, grouped by files and functions:
+#=20
+#   file         ::=3D desc_line* cmd_line events_line data_line+ summar=
y_line
+#   desc_line    ::=3D "desc:" ws? non_nl_string
+#   cmd_line     ::=3D "cmd:" ws? cmd
+#   events_line  ::=3D "events:" ws? (event ws)+
+#   data_line    ::=3D file_line | fn_line | count_line
+#   file_line    ::=3D ("fl=3D" | "fi=3D" | "fe=3D") filename
+#   fn_line      ::=3D "fn=3D" fn_name
+#   count_line   ::=3D line_num ws? (count ws)+
+#   summary_line ::=3D "summary:" ws? (count ws)+
+#   count        ::=3D num | "."
+#=20
+# where
+#   'non_nl_string' is any string not containing a newline.
+#   'cmd' is a string holding the command line of the profiled program.
+#   'filename' and 'fn_name' are strings.
+#   'num' and 'line_num' are decimal integers.
+#   'ws' is whitespace.
+#=20
+# The contents of the "desc:" lines are printed out at the top
+# of the summary.  This is a generic way of providing simulation
+# specific information, eg. for giving the cache configuration for
+# cache simulation.
+#=20
+# Counts can be "." to represent "N/A", eg. the number of write misses f=
or an
+# instruction that doesn't write to memory.
+#=20
+# The number of counts in each 'line' and the 'summary_line' should not =
exceed
+# the number of events in the 'event_line'.  If the number in each 'line=
' is
+# less, cg_annotate treats those missing as though they were a "." entry=
.
+#=20
+# A 'file_line' changes the current file name.  A 'fn_line' changes the
+# current function name.  A 'count_line' contains counts that pertain to=
 the
+# current filename/fn_name.  A 'file_line' and a 'fn_line' must appear
+# before any 'count_line's to give the context of the first 'count_line'=
.
+#=20
+# Each 'file_line' should be immediately followed by a 'fn_line'.  "fi=3D=
"
+# 'file_lines' are used to switch filenames for inlined functions; "fe=3D=
"
+# 'file_lines' are similar, but are put at the end of a basic block in w=
hich
+# the file name hasn't been switched back to the original file name.  (f=
i
+# and fe lines behave the same, they are only distinguished to help
+# debugging.)  [Nb: "fi=3D" and "fe=3D" have not been produced by Cacheg=
rind for
+# some time, they are no longer necessary.]
+
+#-----------------------------------------------------------------------=
-----
 # Performance improvements record, using cachegrind.out for cacheprof, d=
oing no
 # source annotation (irrelevant ones removed):
 #                                                               user tim=
e
@@ -57,6 +98,7 @@
 #16. Finding count lengths by int((length-1)/3), not by
 #    commifying (halves the number of commify calls)            1.68s --=
> 1.47s
=20
+use warnings;
 use strict;
=20
 #-----------------------------------------------------------------------=
-----
@@ -187,19 +229,26 @@
                 @show_events =3D split(/,/, $1);
=20
             # --sort=3DA,B,C
+            #   Nb: You can specify thresholds individually, eg.
+            #   --sort=3DA:99,B:95,C:90.  These will override any --thre=
shold
+            #   argument.
             } elsif ($arg =3D~ /^--sort=3D(.*)$/) {
                 @sort_events =3D split(/,/, $1);
+                my $th_specified =3D 0;
                 foreach my $i (0 .. scalar @sort_events - 1) {
-                    if ($sort_events[$i] =3D~#/.*:(\d+)$/) {
-                                            /.*:([\d\.]+)%?$/) {
+                    if ($sort_events[$i] =3D~ /.*:([\d\.]+)%?$/) {
                         my $th =3D $1;
                         ($th >=3D 0 && $th <=3D 100) or die($usage);
                         $sort_events[$i] =3D~ s/:.*//;
                         $thresholds[$i] =3D $th;
+                        $th_specified =3D 1;
                     } else {
                         $thresholds[$i] =3D 0;
                     }
                 }
+                if (not $th_specified) {
+                    @thresholds =3D ();
+                }
=20
             # --threshold=3DX (tolerates a trailing '%')
             } elsif ($arg =3D~ /^--threshold=3D([\d\.]+)%?$/) {