Re: Ping: random, srandom, and initstate

[Jeff Johnston <jjohnstn at redhat dot com>]

On 05/06/2012 01:46 PM, Joern Rennecke wrote:
> http://sourceware.org/ml/newlib/2012/msg00130.html

Hi Joern,

Sorry for the delay on looking at this.

You don't seem to have incorporated my comments from:

http://sourceware.org/ml/newlib/2010/msg00359.html

I think you should implement the _r routines with glibc extension 
function signatures.

I also think that the reent struct should be used to hold the data for 
the non _r versions and this is passed to the _r versions.  Use the 
small reentrancy strategy of putting aside a pointer in the struct and 
allocating a pointer to an area when first needed.

This would remove the need for locking since threads get their own 
reentrancy struct.

Is there anything I am missing?

-- Jeff J.