bug-hurd
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Xen domU vs. more than 652 MiB of RAM


From: Thomas Schwinge
Subject: Re: Xen domU vs. more than 652 MiB of RAM
Date: Thu, 1 Oct 2009 18:22:56 +0200
User-agent: Mutt/1.5.11

Hello!

On Thu, Oct 01, 2009 at 05:46:18PM +0200, Samuel Thibault wrote:
> Thomas Schwinge, le Thu 01 Oct 2009 17:24:55 +0200, a écrit :
> > On Thu, Oct 01, 2009 at 05:23:21PM +0200, Samuel Thibault wrote:
> > > Thomas Schwinge, le Thu 01 Oct 2009 17:00:43 +0200, a écrit :
> > > > > There's a bootstrap issue: Xen only provides 512KiB of spare bootstrap
> > > > > memory, which is not so much to build a pagetable covering 100s of 
> > > > > MiB.
> > > > 
> > > > But why did it work until now?
> > > 
> > > Because PAE makes page tables twice bigger.
> > 
> > Again: this system was already using PAE.
> 
> Ah, and with PAE you managed to boot with more memory?

Yes.

> How big is your
> kernel?  Actually, kernels without debugging symbols & such don't have
> the issue because while 512KiB is a minimum, the granularity is 4MiB, of
> which the kernel may or may not consume a lot. without debugging
> symbols, the kernel typically only consumes 1 or 2 MiB, which leaves
> plenty of room for the pagetable.

All kernel are configured with ``--enable-platform=xen --enable-kdb''.
Here are the sizes of the previous, functioning combo:

    -rwxr-xr-x 1 tschwinge tschwinge 2443087 Apr 21 23:11 /boot/gnumach-xen..
    -rw-r--r-- 1 tschwinge root      1142784 Apr 21 23:14 /boot/hurd-modules..

And this is the new stuff:

    -rwxr-xr-x 1 tschwinge tschwinge  474396 Oct  1 18:01 /boot/gnumach-xen
    -rwxr-xr-x 1 tschwinge tschwinge 2437411 Oct  1 16:42 /boot/gnumach-xen.
    -rw-r--r-- 1 tschwinge root      1167360 Oct  1 13:40 /boot/hurd-modules

`gnumach-xen' is the stripped version of `gnumach-xen.', without your
patch.  Which one I use of these two doesn't make a difference.  But
then, are the debug section being mapped at all?

But, and I guess that's what you meant: if I build a ``--disable-kdb''
kernel, which is 386364 stripped, then I can go somewhere above 700, but
below 800 MiB.  Your patch is meant to cancel this limitation?


Regards,
 Thomas

Attachment: signature.asc
Description: Digital signature


reply via email to

[Prev in Thread] Current Thread [Next in Thread]