Re: commit: ldap/doc/man/man8 slappasswd.8

["Kurt D. Zeilenga" <Kurt@OpenLDAP.org>]

At 07:22 PM 12/29/2003, Luke Howard wrote:
>>At 06:39 PM 12/29/2003, Luke Howard wrote:
>>>BTW we should check for crypt_r() -- not sure if the API signature is
>>>portable but for platforms that we can support it avoids reentrancy
>>>issues associated with crypt().

The API signature does not appear to be portable.

>>
>>Little point in that... it's all behind a mutex anyways.
>
>Sure, but: plugins may call crypt() without being access to slapd's
>mutex (we have one that does).

Well, how do you protect against other plugins which call
crypt(3)?

>Also, performance... but, probably not a real biggie :-)

I would be surprised that crypt_r(3) would be much faster
than lock+crypt(3)+unlock given the extra work (in slapd(8))
necessary to figure out whether the lock/unlock can be safely
eliminated.

Kurt