>	/* Set the process idea of the break to the real value */	<- added
>	((struct linux_emuldata*)(p->p_emuldata))->p_break = 		<-
>	    p->p_vmspace->vm_daddr + ctob(p->p_vmspace->vm_dsize);	<-

	oops, the analysis may not be correct, but the function is defintely
	doing something bad.

Well, this touches p_emuldata, but it has been allocated just before if it
was not already allocated... 
If something bad happens, then it's because p_vmspace on vm_addr, or vm_dsize
is stale, I don't see anything else that could be wrong.

Maybe you can tell me more about the problem? I have not been reading the
initial discussion

--
Emmanuel Dreyfus
manu@netbsd.org