> Uh, did you by any chance forget to allocate VM in that range?

Here is the code. Is there something wrong in VM allocation?

        bsd_flags = FORK_SHAREVM;
        tf = (struct frame *)p->p_md.md_regs;

        child_stack_addr = (u_long)(tf->f_regs[SP] - IRIX_SPROC_STACK_OFFSET);
        child_stack_size = (u_long)p->p_vmspace->vm_minsaddr
            - (u_long)p->p_vmspace->vm_maxsaddr;

        error = fork1(p, bsd_flags, SIGCHLD, (void *)child_stack_addr,
            child_stack_size, NULL, NULL, retval, &p2);
        if (error)
                return error;

        tf = (struct frame *)p2->p_md.md_regs;

        tf->f_regs[PC] = (unsigned long)SCARG(uap, entry);

        tf->f_regs[A0] = (unsigned long)SCARG(uap, arg);
        tf->f_regs[A1] = tf->f_regs[S3]; /* XXX ? */

        bzero(&vmc, sizeof(vmc));
        vmc.ev_addr = trunc_page(child_stack_addr - child_stack_size);
        vmc.ev_len = round_page(child_stack_size);
        vmc.ev_prot = UVM_PROT_RWX;
        vmc.ev_flags = UVM_FLAG_COPYONW|UVM_FLAG_FIXED|UVM_FLAG_OVERLAY;
        vmc.ev_proc = vmcmd_map_zero;
        error = (*vmc.ev_proc)(p, &vmc);

        retval[0] = (register_t)p2->p_pid;
        retval[1] = 0;

        return error;

-- 
Emmanuel Dreyfus.
"Le 80x86 n'est pas si complexe - il n'a simplement pas de sens"
(Mike Johnson, responsable de la conception x86 chez AMD) 
manu@netbsd.org