On Thu, 17 Jul 2003 21:26:46 +0200, "Malte S. Stretz" <[EMAIL PROTECTED]> writes:

> On Thursday 17 July 2003 20:56 CET Scott A Crosby wrote:
> > In any case, if 5.8.1 it goes out with the fix, its going to be
> > interesting how many latent bugs the fix exposes. The perl core has
> > fixed 4 so far, and identified at least two external packages that
> > break their self-tests.
> >
> > SA, being one premire application of perl should test with RC1, or RC2
> > with hash randomization to see if you can tease out any latent bugs so
> > they're fixed before 5.8.1 goes live.
> 
> *sigh* So I've got to compile another Perl. I'll have a try. But could give 
> a short explanation (or pointer) on how to use it? The second mail you 
> referred to uses different values for PERL_HASH_SEED...

I'm not entirely sure. I think it exists within the
documentation. (Grep it for HASH.)

What's occuring is to avoid an attacker deliberately causing collisions
in the hash table, the hash funciton is now being randomized. With
that code enabled, 'keys %hash' will return the keys in a different
order on *each* run of Perl. This has exposed latent
bugs. Unfortunately, that nondeterminism makes it difficult to debug
problems, so they offer some debugging aids. One of them makes perl
print out the current hash seed, another lets you set a hash seed
manually. Thus, if you have a nonrepeatible failure, you can fix the
hash seed and see if it repeats. Also, if you can find two hash seeds
where one succeeds and the other fails, you now know that hash
nondeterminism is the cause of the bug.

Scott


-------------------------------------------------------
This SF.net email is sponsored by: VM Ware
With VMware you can run multiple operating systems on a single machine.
WITHOUT REBOOTING! Mix Linux / Windows / Novell virtual machines at the
same time. Free trial click here: http://www.vmware.com/wl/offer/345/0
_______________________________________________
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/spamassassin-talk

Reply via email to