On Sun, Apr 29, 2012 at 09:48:01PM +0800, Liu Yuan wrote: > On 04/29/2012 09:24 PM, Christoph Hellwig wrote: > > > Having it global before was ugly but survivable, the problem is > > overloading the meaning in your patch - in the parent it contains the > > fork() result, which is the child pid, using one and the same variable > > for the parent pid in the child is bound to cause confusion > > > No, my patch just fixes what you forgot to do. See your below commit: > cc5e7ee93a67acc535af3659e9664f9ef381cbf6 Please check if you're fine with the fix I sent. It explicitly mentions that, and contains precautions to make the code more robust against accidentally reintroducing the bug by making it a lot more obvious. |