[2621] in Kerberos_V5_Development
Re: Kerberos NT source code
daemon@ATHENA.MIT.EDU (Ezra Peisach)
Wed Oct 22 11:31:03 1997
To: Assar Westerlund <assar@sics.se>
Cc: Ken Hornstein <kenh@cmf.nrl.navy.mil>, krbdev@MIT.EDU, epeisach@MIT.EDU
In-Reply-To: Your message of "22 Oct 1997 17:05:00 +0200."
<5l7mb5et43.fsf@assaris.sics.se>
Date: Wed, 22 Oct 1997 11:30:19 EDT
From: Ezra Peisach <epeisach@MIT.EDU>
The question is whether the krb5_context is a per process or a per thread
context.... If it is per thread - i.e. a new one for each thread, I suspect
that many of the problems could be trivially solved.
If the context is a per process context, then access to the context must
be blocked (mutex lock?) and that will affect lots of places in the code...
i.e. do you block when you want to modify a field, or do you hold
a lock at the beginning of the high level functions such that the lower levels
can have a consistant state, etc... Not an easy question...
Ezra
