Subject: Re: fork1 problem in Memory disk root file system
To: None <port-arm@netbsd.org>
From: Toru Nishimura <locore64@alkyltechnology.com>
List: port-arm
Date: 08/14/2005 03:36:27
Jigar Shah asked;

> 1)  file sys/kern/init_main.c  fork1 function doesnot execute start_init
> process(user space process) ,
> 
> fork1( l, 0, SIGCHLD, NULL , 0, start_init, NULL, NULL, &initproc)
>
> can anyone tell me how & when start_init process has been executed??
>
> 2) even in above file (sys/kern/init_main.c) kthread_create1 doesnt =
> create kernel thread

I have been involved with bootstrapping NetBSD port to new HWs for
years and I have never been trapped in the way you experience.
I suspect you may have made serious malfunction(s) which are
spoiling for the whole NetBSD machinery.

- kernel address space arrangement.
- available RAM memory deposited to pmap.
- hardclock() distribution.

One of peculiar aspect of ARM architecture in which many have been
trapped so far is ARM VIVT (virtual address indexed, virtual address
tagged) cache.  Most of engineers do not understand the design
implication.  It's unlikely the case this time, though.

Toru Nishimura/ALKYL Technology