#400955 still problematic

[Michele Baldessari]

Hi Fabian et all,

(sorry if this shows up through b.d.o as well, I wasn't sure if I had to
reopen the bug before adding more info or not)

Anyways, here's more info:

The bug (or a related base64 bug) is unfortunately still present. To
reproduce:
$ sasl-sample-client
Waiting for mechanism list from server...
S:
TlRMTSBMT0dJTiBQTEFJTiBBTk9OWU1PVVMgRElHRVNULU1ENSBHU1NBUEkgQ1JBTS1NRDU=
sasl-sample-client: Decoding data from base64: bad protocol / cancel

The above base64 is perfectly decoded:
kdc:~/cyrus-sasl2-2.1.22.dfsg1# python
Python 2.4.4 (#2, Oct 20 2006, 00:23:25)
[GCC 4.1.2 20061015 (prerelease) (Debian 4.1.1-16.1)] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> import base64
>>> print base64.b64decode("TlRMTSBMT0dJTiBQTEFJTiBBTk9OWU1PVVMgRElHRVNULU1ENSBHU1NBUEkgQ1JBTS1NRDU=")
NTLM LOGIN PLAIN ANONYMOUS DIGEST-MD5 GSSAPI CRAM-MD5

I tracked it down on
https://bugzilla.andrew.cmu.edu/cgi-bin/cvsweb.cgi/src/sasl/lib/saslutil.c.diff?r1=1.43&r2=1.44
This change broke the decoding, because if I revert sasl_decode64 to the
1.43 version, recompile and install the new .so in /usr/lib, I correctly
get:

$ sasl-sample-client
Waiting for mechanism list from server...
S:
TlRMTSBMT0dJTiBQTEFJTiBBTk9OWU1PVVMgRElHRVNULU1ENSBHU1NBUEkgQ1JBTS1NRDU=
recieved 53 byte message
Choosing best mechanism from: NTLM LOGIN PLAIN ANONYMOUS DIGEST-MD5
GSSAPI CRAM-MD5
Using mechanism DIGEST-MD5
Sending initial response...
C: RElHRVNULU1ENQ==
Waiting for server reply...

So for now reverting to 1.43 does help. I'll see if I can see what's
wrong in the 1.44 version of saslutil.c. But that will take place in a
few days.

hth,
Michele