Added small python script which can be used to calculate & experiment

with Authenticode page hashes. Calculating everything but the first
page hash seems to be working and is straightforward. If anyone knows
how this is done, p-p-p-please let me know.

misc/pagehash.py

@@ -0,0 +1,139 @@
+#!/usr/bin/python
+
+import struct
+import sys
+import hashlib
+from pyasn1.type import univ
+from pyasn1.codec.ber import encoder, decoder
+
+f = open(sys.argv[1], 'rb')
+b = f.read(1024)
+if b[0:2] != 'MZ':
+    print "Not a DOS file."
+    sys.exit(0)
+pepos = struct.unpack('<I', b[60:64])[0]
+if b[pepos:pepos+4] != 'PE\0\0':
+    print "Not a PE file."
+    sys.exit(0)
+pepos += 4
+
+nsections = struct.unpack('<H', b[pepos+2:pepos+4])[0]
+print "#sections", nsections
+
+magic = struct.unpack('<H', b[pepos+20:pepos+22])[0]
+pe32plus = 0
+if magic == 0x20b:
+    pe32plus = 1
+elif magic == 0x10b:
+    pe32plus = 0
+else:
+    print "Unknown magic", magic
+    sys.exit(0)
+
+sectoralign = struct.unpack('<I', b[pepos+52:pepos+56])[0]
+print "Sector alignment", sectoralign
+
+pos = pepos + 112 + pe32plus*16
+nrvas = struct.unpack('<I', b[pos:pos+4])[0]
+print "#rvas", nrvas
+
+pos += 4
+tpos = pos
+rvas = []
+for i in range(0, nrvas):
+    (p1,p2) = struct.unpack('<II', b[pos:pos+8])
+    rvas.append((p1,p2))
+    pos += 8
+
+sections = []
+for i in range(0, nsections):
+    (vsize,vaddr,rsize,raddr) = struct.unpack('<IIII', b[pos+8:pos+24])
+    pos += 40
+    sections.append((vsize,vaddr,rsize,raddr))
+
+hdrend = pos
+print "End of headers", pos
+print rvas
+print sections
+
+sigpos,siglen = rvas[4]
+if sigpos == 0:
+    print "No signature found"
+    sys.exit(0)
+
+f.seek(sigpos)
+sigblob = f.read(siglen)
+cid_page_hash = "\xa6\xb5\x86\xd5\xb4\xa1\x24\x66\xae\x05\xa2\x17\xda\x8e\x60\xd6"
+oid_ph_v1 = "\x06\x01\x04\x01\x82\x37\x02\x03\x01"
+oid_ph_v2 = "\x06\x01\x04\x01\x82\x37\x02\x03\x02"
+p = sigblob.find(cid_page_hash)
+if p == -1:
+    print "No page hash present"
+    sys.exit(0)
+
+p += len(cid_page_hash)
+sha1 = True
+i = sigblob.find(oid_ph_v1)
+if i == -1:
+    i = sigblob.find(oid_ph_v2)
+    if i == -1:
+        print "No page hash found"
+        sys.exit(0)
+    sha1 = False
+p = i + len(oid_ph_v1)
+
+blob = str(decoder.decode(sigblob[p:])[0].getComponentByPosition(0))
+ph = []
+i = 0
+hashlen = 20
+if not sha1:
+    hashlen = 24
+while i < len(blob):
+    offset = struct.unpack('<I', blob[i:i+4])[0]
+    i += 4
+    data = blob[i:i+hashlen]
+    ph.append((offset,data.encode("hex")))
+    i += hashlen
+
+# Calculating first page hash is not working, the normal authenticode
+# hash is calculated like below, but that's not how it should be done...
+if sha1:
+    md = hashlib.sha1()
+else:
+    md = hashlib.sha256()
+md.update(blob[0:pepos+84])
+md.update(blob[pepos:88:tpos+3*8])
+md.update(blob[tpos+4*8:1024])
+digest = md.hexdigest()
+
+print ""
+print "Checking page hash..."
+print ""
+
+nph = [(0,digest)]
+lastpos = 0
+pagesize = sectoralign # ???
+for vs,vo,rs,ro in sections:
+    l = 0
+    while l < rs:
+        f.seek(ro+l)
+        howmuch = pagesize
+        if rs - l < pagesize:
+            howmuch = rs - l
+        b = f.read(howmuch)
+        if howmuch < pagesize:
+            b = b + '\0' * (pagesize - (rs - l))
+        if sha1:
+            d = hashlib.sha1(b).hexdigest()
+        else:
+            d = hashlib.sha256(b).hexdigest()
+        nph.append((ro+l, d))
+        l += pagesize
+    lastpos = ro + rs
+
+nph.append((lastpos,'0'*(2*hashlen)))
+for i in range(0,len(nph)):
+    x=ph[i]
+    y=nph[i]
+    if x[0] != y[0] or x[1] != y[1]:
+        print "Not matching:", x, "!=", y