On Fri, Nov 30, 2007 at 11:40:52AM -0800, Chuck McIntyre wrote:
> On Nov 29, 2007 2:09 AM, Rafał Kupka <rkupka+Listy.Xen@xxxxxxxxxxxxx> wrote:
> Thanks for your help, see my responses below.
> > > (XEN) Physical RAM map:
> > > (XEN) 0000000000000000 - 000000000009dc00 (usable)
> > > (XEN) 0000000000100000 - 00000000dcff0000 (usable)
> > > (XEN) System RAM: 3535MB (3620404kB)
> > > (XEN) Xen heap: 14MB (14340kB)
> > > (XEN) found SMP MP-table at 000ff780
> > > (XEN) DMI present.
> > Xen gets memory map from grub, unlike linux.
> Do you have a reference for this? Is this intentional?
Yes, it's intentional. Xen hypervisor (unlike linux) uses GRUB multiboot
protocol instead of custom one.
Look at description of mmap_* fields.
> understanding is that grub doesn't have any way to map around PCI BIOS
> created RAM holes because it is such a low level (boot) piece of
> software. Is this, then, a bug? Is there a way around it?
The whole e820 memory map is a BIOS thing.
> > Yes, it's this strange bug in grub I spot some time ago.
> > Please look at Debian bug #419994:
> > http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=419994
> I'm not sure it's a grub bug though, since it works correctly in a
> non-xen kernel.
Non Xen kernel gets memory map differently, as I said before.
> Should I try a different boot loader?
Not much choice to boot Xen. AFAIK it's grub or grub2.
Don't know if grub2 is affected.
> > Can you try patch (with volatile unsigned long cont) and tell if it make
> > any difference?
> Will try this, and see, just to be sure, in the meantime, if you could
> provide reference for your assertion that Xen uses the boot loader to
> get its RAM map, that would be very helpful.
Look at Xen hypervisor source.
Great software without the knowledge to run it is pretty useless.
(Linux Gazette #1)
Xen-users mailing list