Subject: Re: thread fork [was Re: NetBSD 2.0 release date]
To: Nathan J. Williams <>
From: Kamal R Prasad <>
List: tech-kern
Date: 11/11/2003 11:44:23
>To the best of my knowledge, the current implementation of fork() in a
>thread is correct. How do you think it fails? The permitted behavior
>of a child forked from a threaded process is quite limited, and it's
>easy to invoke undefined behavior.

I had a misunderstanding that fork() will inherit threads in the child 
process [but Christian Limpach corrected me]. The POSIX requirement is 
that the fork() should be async-signal-safe.


Kamal R. Prasad
AIX Support & Test, IBM India Software Labs
Golden Enclave, Airport Road, Bangalore-560017, India
Phone : +91-80-5094963,  Internal Ext   : 2963
In theory, there is no difference between theory and practice. In 
practice, there is.