Verify new entries when merging.

Minor tweaks on map/mergeverify2.
Closes CATLFISH-36.

3 files changed, 92 insertions, 7 deletions

diff --git a/src/catlfish.erl b/src/catlfish.erl
index ed75495..d98e741 100644
--- a/src/catlfish.erl
+++ b/src/catlfish.erl
@@ -5,7 +5,7 @@
 -export([add_chain/3, entries/2, entry_and_proof/2]).
 -export([known_roots/0, update_known_roots/0]).
 -export([init_cache_table/0]).
--export([entryhash_from_entry/1]).
+-export([entryhash_from_entry/1, chain_from_entry/1]).
 -include_lib("eunit/include/eunit.hrl").
 
 -define(PROTOCOL_VERSION, 0).
@@ -252,7 +252,7 @@ deserialise_extra_data(ExtraData) ->
             [E | deserialise_extra_data(Rest)]
     end.
 
-entryhash_from_entry(Entry) ->
+chain_from_entry(Entry) ->
     {MTLText, ExtraDataPacked} = unpack_entry(Entry),
     {ExtraData, <<>>} = decode_tls_vector(ExtraDataPacked, 3),
     MTL = deserialise_mtl(MTLText),
@@ -266,7 +266,10 @@ entryhash_from_entry(Entry) ->
             precert_entry ->
                 Chain
         end,
-    crypto:hash(sha256, Data).
+    Data.
+
+entryhash_from_entry(Entry) ->
+    crypto:hash(sha256, chain_from_entry(Entry)).
 
 %% Private functions.
 -spec unpack_entry(binary()) -> {binary(), binary()}.
diff --git a/tools/merge.py b/tools/merge.py
index f9c93d9..ce3bf0b 100755
--- a/tools/merge.py
+++ b/tools/merge.py
@@ -16,9 +16,11 @@ import hashlib
 import urlparse
 import os
 import yaml
+import select
+import struct
 from certtools import build_merkle_tree, create_sth_signature, \
     check_sth_signature, get_eckey_from_file, timing_point, http_request, \
-    get_public_key_from_file
+    get_public_key_from_file, get_leaf_hash, decode_certificate_chain
 
 parser = argparse.ArgumentParser(description="")
 parser.add_argument('--config', help="System configuration", required=True)
@@ -61,7 +63,7 @@ def write_chain(key, value):
         try:
             os.makedirs(path)
         except Exception, e:
-            print e
+            pass
     else:
         path = chainsdir
     f = open(path + "/" + filename, "w")
@@ -131,7 +133,7 @@ def sendlog(node, baseurl, submission):
         return json.loads(result)
     except urllib2.HTTPError, e:
         print "ERROR: sendlog", e.read()
-        sys.exit(1)
+        return None
     except ValueError, e:
         print "==== FAILED REQUEST ===="
         print submission
@@ -206,6 +208,32 @@ for storagenode in storagenodes:
     new_entries.update(new_entries_per_node[storagenode["name"]])
     entries_to_fetch[storagenode["name"]] = []
 
+def unpack_entry(entry):
+    pieces = []
+    while len(entry):
+        (length,) = struct.unpack(">I", entry[0:4])
+        data = entry[4:4+length]
+        entry = entry[4+length:]
+        pieces.append(data)
+    return pieces
+
+import subprocess
+
+def verify_entry(verifycert, entry, hash):
+    unpacked = unpack_entry(entry)
+    mtl = unpacked[0]
+    assert hash == get_leaf_hash(mtl)
+    s = struct.pack(">I", len(entry)) + entry
+    verifycert.stdin.write(s)
+    result_length_packed = verifycert.stdout.read(4)
+    (result_length,) = struct.unpack(">I", result_length_packed)
+    result = verifycert.stdout.read(result_length)
+    assert len(result) == result_length
+    (error_code,) = struct.unpack("B", result[0:1])
+    if error_code != 0:
+        print >>sys.stderr, result[1:]
+        sys.exit(1)
+
 timing_point(timing, "get new entries")
 
 new_entries -= certsinlog
@@ -221,6 +249,8 @@ for hash in new_entries:
             entries_to_fetch[storagenode["name"]].append(hash)
             break
 
+verifycert = subprocess.Popen(["../verifycert.erl"],
+                               stdin=subprocess.PIPE, stdout=subprocess.PIPE)
 
 added_entries = 0
 for storagenode in storagenodes:
@@ -229,6 +259,7 @@ for storagenode in storagenodes:
         entries = get_entries(storagenode["name"], "https://%s/" % storagenode["address"], chunk)
         for hash in chunk:
             entry = entries[hash]
+            verify_entry(verifycert, entry, hash)
             write_chain(hash, entry)
             add_to_logorder(hash)
             logorder.append(hash)
@@ -237,6 +268,8 @@ for storagenode in storagenodes:
 timing_point(timing, "add entries")
 print "added", added_entries, "entries"
 
+verifycert.communicate(struct.pack("I", 0))
+
 tree = build_merkle_tree(logorder)
 tree_size = len(logorder)
 root_hash = tree[-1][0]
@@ -277,7 +310,15 @@ for frontendnode in frontendnodes:
     print "current position", curpos
     entries = [base64.b64encode(entry) for entry in logorder[curpos:]]
     for chunk in chunks(entries, 1000):
-        sendlogresult = sendlog(nodename, nodeaddress, {"start": curpos, "hashes": chunk})
+        for trynumber in range(5, 0, -1):
+            sendlogresult = sendlog(nodename, nodeaddress, {"start": curpos, "hashes": chunk})
+            if sendlogresult == None:
+                if trynumber == 1:
+                    sys.exit(1)
+                select.select([], [], [], 10.0)
+                print "tries left:", trynumber
+                continue
+            break
         if sendlogresult["result"] != "ok":
             print "sendlog:", sendlogresult
             sys.exit(1)
diff --git a/verifycert.erl b/verifycert.erl
new file mode 100755
index 0000000..b9a3753
--- /dev/null
+++ b/verifycert.erl
@@ -0,0 +1,41 @@
+#!/usr/bin/env escript
+%% -*- erlang -*-
+%%! -pa ebin -pa lib/catlfish-0.6.0-dev.ez/catlfish-0.6.0-dev/ebin -pa lib/lager-2.1.1.ez/lager-2.1.1/ebin
+
+write_reply(Bin) ->
+    Length = size(Bin),
+    file:write(standard_io, <<Length:32, Bin/binary>>).
+
+verify(RootCerts, DBEntry) ->
+    try
+        Chain = catlfish:chain_from_entry(DBEntry),
+        %% XXX: doesn't verify that MTL is derived from Chain
+        case x509:normalise_chain(RootCerts, Chain) of
+            {ok, _} ->
+                write_reply(<<0:8>>);
+            {error, Reason} ->
+                ReasonBin = list_to_binary(io_lib:format("~p", [Reason])),
+            write_reply(<<1:8, ReasonBin/binary>>)
+        end
+    catch
+        Type:What ->
+            [CrashFunction | Stack] = erlang:get_stacktrace(),
+            ErrorBin = list_to_binary(io_lib:format("Crash: ~p ~p~n~p~n~p~n", [Type, What, CrashFunction, Stack])),
+            write_reply(<<2:8, ErrorBin/binary>>)
+    end.
+
+loop(RootCerts) ->
+    {ok, LengthBin} = file:read(standard_io, 4),
+    <<Length:32>> = list_to_binary(LengthBin),
+    case Length of
+        0 ->
+            none;
+        _ ->
+            {ok, DBEntry} = file:read(standard_io, Length),
+            verify(RootCerts, list_to_binary(DBEntry)),
+            loop(RootCerts)
+    end.
+
+main(_) ->
+    Certs = x509:read_pemfiles_from_dir("tests/known_roots/"),
+    loop(Certs).