On Fri, Jul 03, 2009 at 01:19:10PM -0600, Eric Blake wrote: >-----BEGIN PGP SIGNED MESSAGE----- >Hash: SHA1 > >According to Christopher Faylor on 7/3/2009 9:17 AM: >> Is there some reason why we're not just using the newlib version of all >> of these functions? I have stared at the code in mktemp.cc and the only >> thing I see that seems to be Cygwin specific is the arc4random function. >> Is the security that this provides the only reason not to use newlib? >> >> That is probably a good enough reason right there but I was just >> wondering. > >Well, before today, cygwin had mkdtemp but newlib didn't. But you are >correct that after today, the only substantial difference is getpid() vs. >arc4random(). For mkstemp, this is not an issue. But guess which one is >more predictable, and thus makes for a less secure mktemp (even though we >already have a compiler warning that mktemp is insecure)? > >Maybe it would be worth pushing the arc4random approach to newlib?
I'm not sure exactly how you'd do that. Obviously you can use /dev/random on linux. For windows, you'd have to port windows-specific functions to newlib. I don't think there is any precedent for that. Sounds like more work than it's worth. Nevermind. cgf
