errorr "_create_session: Operation not permitted" on recent mol snapshots

[Martin]

I just compiled a recent mol-rsync snapshot and when I run startmol I get:
rs6000:/etc/mol# startmol --test
Mac-on-Linux 0.9.70-rc2 [Feb 18 2004 01:51]
Copyright (C) 1997-2004 Samuel Rydh
Starting MOL session 0
_create_session: Operation not permitted

I also tried mol-0.9.70-rc2 but it gives me the same error.
Machine is a rs6000 with 604 processor running debian. I have run mol 0.95 and mol 0.98 succesfully on it  although I never got osx 10.0 or 10.1 to work.

Which parameters I give it (--test --osx --cdboot) don't seem to matter.

I narrowed the problem down to where I think it happens in the source(see below), but I don't know why.

Does anybody have a clue?
greetings,
Martin

I have added the lines <<errormessage appears here I Think>> and
<<problem is here I think>> in the code below.

src/cpu/ppc/init.c line 150:
int
open_session( void )
{
	struct timeval tv;
	
	check_kmod_version();

	gettimeofday( &tv, NULL );
	srandom( tv.tv_usec + get_tbl() );

	if( _create_session(g_session_id) < 0 ) {
		switch( errno ) {

      if( _create_session(g_session_id) < 0 ) {
		switch( errno ) {
		case EMOLINUSE:
			printm("---> Session %d is already running (needs to be killed?)\n", g_session_id);
			break;

		case EMOLINVAL:
			printm("The session number (%d) is out of bounds\n", g_session_id );
			break;

		case EMOLSECURITY:
			printm("*****************************************************\n"
			       "* SECURITY ALERT. Somebody (other than MOL) has\n"
			       "* tried to utilize the MOL infrastructure. Due to\n"
			       "* security reasons, a reboot is necessary in order to\n"
			       "* get MOL working again.\n"
			       "*****************************************************\n");
			break;

		case EMOLGENERAL:
			printm("_create_session() failed due to an unspecified error\n");
			break;
                 default:
<<errormessage appears here I think>>	perrorm("_create_session");
			break;
		}
		return 1;
	}

src/include/wrapper.h line 40:
static inline int _create_session( int sessnum ) {
	return mol_ioctl( MOL_IOCTL_CREATE_SESSION, sessnum, 0, 0 ); }

src/kmod/Linux/dev.c line 249:
static int
mol_ioctl( struct inode *inode, struct file *file, unsigned int cmd, unsigned long arg )
{
	mol_ioctl_pb_t pb;
	kernel_vars_t *kv;
	int ret;

	/* fast path */
	if( cmd == MOL_IOCTL_SMP_SEND_IPI ) {
		send_ipi();
		return 0;
	}

	if( copy_from_user(&pb, (void*)arg, sizeof(pb)) )
		return -EFAULT;

	switch( cmd ) {
	case MOL_IOCTL_GET_INFO:
		return get_info( (mol_kmod_info_t*)pb.arg1, pb.arg2 );

	case MOL_IOCTL_CREATE_SESSION:
<<problem is here I think>>		if( !(file->f_mode & FMODE_WRITE) || !capable(CAP_SYS_ADMIN) )
<<problem is here I think>>			return -EPERM;
		ret = -EINVAL;
		down( &initmutex );
		if( (uint)pb.arg1 < MAX_NUM_SESSIONS && !file->private_data ) {
			if( !(ret=initialize_session(pb.arg1)) ) {
				kv = g_sesstab->kvars[pb.arg1];
				init_MUTEX( &kv->ioctl_sem );
				file->private_data = kv;
			}
		}
		up( &initmutex );
		return ret;