Re: booting l4-Hurd, memory unaligned

[Marcus Brinkmann]

At Tue, 5 Apr 2005 07:48:23 +0200,
"Yoshinori K. Okuji" <address@hidden> wrote:
> 
> On Tuesday 05 April 2005 04:56 am, Derek Davies wrote:
> > Grub supplies the initial memory map information to laden, so its
> > possible this is a grub bug.  Then again, the BIOS supplies the info
> > to grub, so it could be a BIOS bug.  I suppose there's some remote
> > possibility that its an l4/hurd bug too :)
> 
> It is a bug in laden. It should not check the alignment of a memory region, 
> when it does not intend to map the region. Current code checks all memory 
> maps even if they are reserved for BIOS thus not used by laden.

Laden needs to build a memory map for the L4 kernel.  This map has as
its basic unit the minimal page size of the system architecture.
 
> Actually, I guess the length returned by the BIOS looks bogus. Possibly it is 
> a bug in the BIOS. But, strictly speaking, there is no such rule that memory 
> maps must be aligned for BIOS. The rule is defined by laden, so the BIOS may 
> or may not follow it. Therefore, the length is valid in a sense.

Yes, I suppose that in any case, the rule: "Be liberal in what you
accept, and strict in what you output" needs to be applied here.

This means that laden must do the necessary rounding.  It probably
shouldn't even warn, as there is nothing a user can do about it
anyway.

The problem for me is that there are a zillion of BIOSes, and have
only one or two for testing.  I don't think there is much of a
specification for memory maps in BIOSes, is there?

Thanks,
Marcus