add: debugging tools

2 files changed, 195 insertions, 0 deletions

diff --git a/tools/debugging/convert_trace.py b/tools/debugging/convert_trace.py
new file mode 100644
index 0000000..289ccd1
--- /dev/null
+++ b/tools/debugging/convert_trace.py
@@ -0,0 +1,82 @@
+#!/usr/bin/python
+"""
+es wird benoetigt:
+
+# Disassemblerlisting
+zpu-elf-objdump --disassemble --source greth.elf > greth.diss
+
+# Simulationsoutput
+# Simulator -> trace.txt
+
+convert_trace.py sucht die zu einer trace-zeile 
+passende Zeilennummer aus dem Disassemblerlisting
+
+
+# das Ausgabeformat kann von vim/quickfix gelesen werden
+# Achtung! Ggf lange Laufzeit
+# (ca. 3min fuer 1 MB trace.txt)
+convert_trace.py > greth_conv.txt
+
+
+Verwendung in vim:
+
+"" """"""""""""""""""""
+" quickfix-Handling
+" Alt-2 bzw. Alt-3 um durch die Fehlerliste zu gehen
+set errorformat +=$f:%l:%m
+noremap <M-2> :cp<CR>
+noremap <M-3> :cn<CR>
+
+"""
+
+import profile
+
+import string
+import re
+
+
+def get_line_number( src, pattern):
+    lineno = 0
+    for m in re.finditer( pattern, src):
+        start = m.start()
+        lineno = src.count('\n', 0, start)+2
+        #offset = start - src.rfind('\n',0 ,start)
+        #word = m.group(0)
+        #print "greth.diss(%s,%s): %s" % (lineno, offset, word)
+        break
+    return lineno 
+
+
+def do_stuff():
+    
+
+    src = open('greth.diss').read()
+    
+    with open("trace.txt", "rU") as fobj:
+
+        line_count = 0
+
+        for line in fobj:
+
+            #if line_count < 10000:
+            #    line_count += 1
+            #else:
+            #   break
+
+            line.strip()
+
+            if line.lstrip().startswith("#"):
+                continue
+            if line.lstrip() == '':
+                continue
+
+            zuordnung = line.split(" ")
+            pc = string.atoi( zuordnung[0], 16)
+            search_str = "\s+%x:.*" % (pc)
+            
+            print "greth.diss:%d:%s" % ( get_line_number( src, search_str), line.rstrip('\r\n'))
+
+
+profile.run('do_stuff()')
+
+
diff --git a/tools/debugging/statistic_trace.py b/tools/debugging/statistic_trace.py
new file mode 100644
index 0000000..02b55f1
--- /dev/null
+++ b/tools/debugging/statistic_trace.py
@@ -0,0 +1,113 @@
+#!/usr/bin/python
+
+"""
+
+Erzeugt eine Statistik in welcher Programmfunktion die Taktzyklen
+verbraten werden.
+
+benoetigt den Simulationstrace -> trace.txt
+und eine Liste mit den Symbolen (sortiert)
+
+dabei sollten die Symbole mit 'N' und 't' ignoriert werden
+
+zpu-elf-nm --numeric-sort greth.elf | grep -v " N " | grep -v " t " > greth.symbols
+
+
+Aufruf:
+statistic_trace.py > greth_statistic.txt
+
+"""
+
+import string
+import re
+
+
+def get_symbol_line( src, number):
+
+    lastline = ""
+    lastaddr = -1
+
+    for lineno, line in enumerate( src):
+
+        line = line.strip()
+        
+        try:
+            addr = string.atoi( "0x0"+line.split(" ")[0], 16)
+
+        except ValueError:
+            #print "ignore: %s" % line
+            addr = -1
+
+        if (number >= lastaddr) and (number < addr):
+            result = " ".join( lastline.split(" ")[1:] )
+            #return result
+            return lastline.split(" ")[2]
+            #return lastaddr
+
+        lastline = line
+        lastaddr = addr
+
+#   lineno = 0
+#   for m in re.finditer( pattern, src):
+#       start = m.start()
+#       lineno = src.count('\n', 0, start)+2
+#       #offset = start - src.rfind('\n',0 ,start)
+#       #word = m.group(0)
+#       #print "greth.diss(%s,%s): %s" % (lineno, offset, word)
+#       break
+#   return lineno 
+    return ""
+#    return 0 
+
+
+
+def sortkey( (k,v)):
+    return (-v, k)
+
+
+
+def do_stuff():
+    
+    statistik = {}
+
+    src = open('greth.symbols').readlines()
+    
+    with open("trace.txt", "rU") as fobj:
+
+        line_count = 0
+
+        for line in fobj:
+
+#            if line_count < 100:
+#                line_count += 1
+#            else:
+#                print "attention debug shortcut!"
+#                break
+
+            line = line.strip()
+
+            if line.startswith("#"):
+                continue
+            if len(line) == 0:
+                continue
+
+            zuordnung = line.split(" ")
+            pc = string.atoi( zuordnung[0], 16)
+            #print "%s --- %s" % ( get_symbol_line( src, pc), line)
+            value = get_symbol_line( src, pc)
+            if not value in statistik:
+                statistik[ value ] = 1
+            else:
+                statistik[ value ] += 1
+
+
+    items = statistik.items()
+    items.sort( key=sortkey)
+
+
+    for value, key in items:
+        print "%30s , %s" % (value, key)
+
+do_stuff()
+
+