On Wed, 1 Dec 2004 18:32:33 -0800
"Mark Pizzolato" <[EMAIL PROTECTED]> wrote:

> > Every normal POSIX compliant platform provides reentrant and thread
> > safe tmpfile().
> 
> Solaris man pages for tmpnam(), and tempnam() and tmpfile() suggest
> that tmpfile() uses tmpnam() and that tmpnam() is not threadsafe.

No, it doesn't. It just says that "The tmpnam() function is unsafe  in 
multithreaded  applications. The  tempnam()  function  is  safe in
multithreaded applications and should be used instead." No word about
tmpfile().

Also `man -s5 attributes` doesn't list tmpfile() as a MT-unsafe
function.

> exists a real likelyhood that collisions will occur on systems which
> have relatively low resolution updates to the microsecond return from
> gettimeofday().

To protect against potential collisions it uses a last result of
cli_md5buff as a salt.

-- 
   oo    .....         Tomasz Kojm <[EMAIL PROTECTED]>
  (\/)\.........         http://www.ClamAV.net/gpg/tkojm.gpg
     \..........._         0DCA5A08407D5288279DB43454822DC8985A444B
       //\   /\              Thu Dec  2 11:32:11 CET 2004

Attachment: pgpSAtqC2gpfT.pgp
Description: PGP signature

_______________________________________________
http://lists.clamav.net/cgi-bin/mailman/listinfo/clamav-devel

Reply via email to