Boot problem with the new version of l4linux on Xscale

Adam Lackorzynski adam at
Sat Oct 14 12:19:11 CEST 2006

On Sat Oct 14, 2006 at 12:37:08 +0900, WanKeun Jo wrote:
> I tried to run Fiasco-L4Linux on Xscale board. I'm using L4Linux-2.6.17 version.
> First, I made a bootstap image (=bootstrap.raw) and loaded on Xscale and run it.
> But some server programs' ranges were overlapped, so I fixed some server programs'
> DEFAULT_RELOC_arm address, and then started working.

Ok, fine.

> During the boot process, "Warning: Unable to open an initial console" is warned and halt. 
> I checked on boot process after ram disk is loaded, dev file path lookup fails..
> So as to mount dev file system again after loading ram disk, I added "devfs options" to fs/Kconfig.
> And then the warning message is not presented.

You need to to have some files under /dev in your ramdisk, esp. console,
then this message goes away. Enabling devfs is no option as it's gone in

> But after that, I got a new error message
> ---------------------------------------------------------------------------------------------
> Assertion failed: 'space() == current_space()'  in /home/jowkju/project/Microkernel/Fiasco/Fiasco_Xscale/l4/kernel/fiasco/src/kern/thread-
> pagefault.cpp:82
> at f0005400
> ---------------------------------------------------------------------------------------------
> This error may occur during the "init" program in ram disk execute.
> However, there may not be errors in init program because it works well
> in the 2.6.16 version.
> And maybe the problem occurs during the context switching between
> "init" program and other threads...

It's from Fiasco and has primarily nothing to do with L4Linux in the
first place. There was a subtle bug recently which has been fixed in CVS
for some time now. This bug triggers the above assertion. Go to
src/kern/arm/ivt.S and check if swi_entry looks like this.

	exceptionframe 4

If not, i.e. switch_to_kernel is missing, add it. It should make the
assertion go away.

Adam                 adam at

More information about the l4-hackers mailing list