[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Problems with unified heimdal/samba domain

"James F. Hranicky" <jfh@cise.ufl.edu> writes:

> On Sat, 11 Dec 2004 23:48:33 +0100
> Love <lha@stacken.kth.se> wrote:
>> So, I think its the libhdb backend that do the memset and not the caller,
>> does it still work for you if its done that way (ie only the samba change).
>> I've commited all other changes than the memset and the compare below
> I'm sorry, which memset? The one I commented out, or the ones I added
> to initialize some of the hdb_entry structs?

All of them, I didn't put in any of the memset nor the removal.

> The initializations I put in were to prevent junk pointers from getting
> free()ed when hdb_free_entry is called.

That was a side effect of you commenting out the first memset.

>> When is this needed ? I think the caller is wrong if it passes a NULL
>> principal to this function, when does it happen ?
> I believe I put that in originally to catch the null pointer passed in
> due to the erroneous memset() call in LDAP_message2entry(), but it 
> seemed like a decent check to keep in.

I think that was because the the LDAP_message2entry failed to fail to fill
in the mandatory field ent->principal, now it will fail then it can't fill
it in.

Not keeping the check in is for the same reason strcmp or memcmp doesn't
check their argument.

Now that I've fixed OpenLDAP's SASL to deal with NetBSD I've tested the
samba stuff and fixed some more problems related to deletions of attributes
before adding new ones.

Also, can you tell me what the key is for the sambaNTPassword entry is
after setting the password to "foo", without quotes using samba (not
heimdal). Is it




note the nibble order. I think Heimdal does the wrong order right now.


PGP signature