Ticket History #5766: MSLSA krb5_cc module fails to check success of UNICODE string conversions

Remove the broken optimization from UnicodeToANSI() that attempted
in a stupid way to avoid to WideCharToMultiByte() call by testing
to see if the second byte (not wchar) was NUL. This test works
for Latin1 but not for anything more interesting. Always call
WideCharToMultiByte().

I commented out the use of WC_NO_BEST_FIT_CHARS in order to
provide compatibility with Windows. Windows converts the user
name without that option.

With the current code a principal name consisting of a single
component equivalent to the Greek character Sigma and a realm
name will be converted to S@REALM exactly as the "WhoAmI"
command does. If WC_NO_BEST_FIT_CHARS was specified, this
string would be converted to "?@REALM".


Commit By: jaltman



Revision: 19970
Changed Files:
U trunk/src/lib/krb5/ccache/cc_mslsa.c

pull up r19969 from trunk

r19969@cathode-dark-space: jaltman | 2007-09-24 13:46:26 -0400
ticket: new
subject: MSLSA krb5_cc module fails to check success of UNICODE string conversions

The MSLSA krb5_cc module was written with an assumption that probably does not
hold true anymore. It assumed that all Kerberos strings although stored in
wide character data structures could in fact be represented in the application's
ANSI code page and that such conversions would not fail.

The UnicodeToANSI() function did not check the result of WideCharToMultiByte()
for success. If the conversion failed, this could result in the caller believing
the contents of the output string buffer were a valid string when instead they
were simply stack garbage.

The UnicodeStringToMITPrinc() and KerbExternalNameToMITPrinc() functions did not
check the return value of krb5_parse_name() for success. If krb5_parse_name()
was passed a pointer to garbage on the stack instead of an actual principal name,
this could result in the caller believing the output krb5_principal * was valid
when instead it was NULL.

The function CacheInfoEx2ToMITCred() is dependent on the success or failure of
UnicodeStringToMITPrinc() assumed it could not fail and did not return a
success or failure indication to its caller.

If Microsoft a formatted ticket contains a Unicode string that can not be
represented in the application's ANSI code page, this could result in a NULL
pointer dereference during a call to krb5_cc_resolve("MSLSA:") or
krb5_cc_retrieve(), or krb5_cc_get_principal().

With the changes in this commit, tickets containing principal names that cannot
be represented in the application's ANSI code page will be hidden from the
application.




Commit By: tlyu



Revision: 20046
Changed Files:
_U branches/krb5-1-6/
U branches/krb5-1-6/src/lib/krb5/ccache/cc_mslsa.c

pull up r19970 from trunk

r19970@cathode-dark-space: jaltman | 2007-09-24 14:49:58 -0400
ticket: 5766

Remove the broken optimization from UnicodeToANSI() that attempted
in a stupid way to avoid to WideCharToMultiByte() call by testing
to see if the second byte (not wchar) was NUL. This test works
for Latin1 but not for anything more interesting. Always call
WideCharToMultiByte().

I commented out the use of WC_NO_BEST_FIT_CHARS in order to
provide compatibility with Windows. Windows converts the user
name without that option.

With the current code a principal name consisting of a single
component equivalent to the Greek character Sigma and a realm
name will be converted to S@REALM exactly as the "WhoAmI"
command does. If WC_NO_BEST_FIT_CHARS was specified, this
string would be converted to "?@REALM".




Commit By: tlyu



Revision: 20047
Changed Files:
_U branches/krb5-1-6/
U branches/krb5-1-6/src/lib/krb5/ccache/cc_mslsa.c