problem with 64-bit builds on big-endian platforms?
Cesar Garcia
Cesar.Garcia at morganstanley.com
Thu Aug 7 18:16:38 EDT 2003
I'm not sure where *all* of the problem are on 64 bit big endian
platforms.
What I did find is that in 1.3.1 the marshalling of the start time in
the sealed portion of a krb4 ticket (krb_create_ticket) will work on
32bit or 64bit on big endian platforms correctly, but not on little
endian platforms. This is based on inspection of code, rather than
actually testing or using it.
krb524d uses code similar to krb_create_ticket (krb524int_krb_create_ticket)
that still has the bug (which is that it copies 4 of the higher
order bytes of an 8 byte start time, which on big endian platforms
is basically 4 bytes of zeros). This was based in testing, finding the
bug, hopefully fixing correctly, and re-testing the code, which
preliminary testing shows it working on 32,64 bit big endian and
32bit little endian machines. I submitted a bug report to
krb5-bugs at mit.edu - see RT/krbdev.mit.edu: Ticket #1714.
I noticed, but haven't investigated problems with krb524init.
This is the only 1.3.1 code that I can speak of, since it's the only
code I've tested.
Hope this helps.
>>>>> "Tim" == Tim Mooney <mooney at dogbert.cc.ndsu.NoDak.edu> writes:
Tim> In December of 2002, Ken Raeburn indicated that there would
Tim> likely be problems if you build krb5 1.2.6 on a 64-bit platform
Tim> that was also big-endian. Can anyone comment on whether that is
Tim> believed to be fixed in later 1.2.x or in 1.3.1?
Tim> If not, and if the nature of the problems are generally known,
Tim> can anyone comment on what those problems are?
Tim> I'm in the situation where I may have to build (at least the
Tim> shared libraries) for both the 32-bit and 64-bit ABIs on Solaris
Tim> on sparc, and I want to know what I'm getting myself into.
Tim> Tim -- Tim Mooney mooney at dogbert.cc.ndsu.NoDak.edu Information
Tim> Technology Services (701) 231-1076 (Voice) Room 242-J6, IACC
Tim> Building (701) 231-8541 (Fax) North Dakota State University,
Tim> Fargo, ND 58105-5164
Tim> _______________________________________________ krbdev mailing
Tim> list krbdev at mit.edu
Tim> https://mailman.mit.edu/mailman/listinfo/krbdev
More information about the krbdev
mailing list