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

Re: [Xen-users] Confusion about Kernel in Installation process

So wrote Mark Williamson on Monday, 16 May 2005:
> The build process produces three images used to boot Xen domains:
> * xen.gz - Xen itself.  You need to boot this in your grub.conf / menu.lst
> * vmlinuz-2.6.11-xen0 - a "domain 0" (host) enabled kernel.  You need to pass 
> this to Xen in your grub.conf / menu.lst
> * vmlinuz-2.6.11-xenU - a guest-only kernel.  You can boot guest domains from 
> this instead of using the -xen0 kernel.  The only difference is it's slightly 
> smaller because it includes only "virtual" drivers.

# find ./ -name vmlinuz-2.6.11-xenU*   comes up empty.  Although the
domain 0 kernel was built I cannot find anything that looks like a Linux
kernel image under the linux-2.6.11-xen-sparse tree.  Could it be that my
build failed?

> > My question is: Why should the user not just patch the kernel with
> > whatever required patches there are and then compile the kernel himself?
> > The make script seems to be nice but unfortunately I have no idea what
> > it is actually doing.  For example, are there specific options in the
> > Linux kernel which need to be disabled or enabled for the Xen kernels?
> > Why not just instruct the user to patch the kernel with the appropriate
> > patches and to dis(en)able whatever options are necessary?
> You can cd linux-2.6.11-xen0 and do "make ARCH=xen menuconfig" to configure.  
> You may want to copy your existing .config into that directory first.
> Alternatively, stick your config under "dist/install/boot/config-2.6.11-xen0" 
> and it should get applied automatically.

Here are the default values for Xen in the kernel config for domain 0:

[ ] Privileged Guest (domain 0) (NEW)
[ ] Physical device access (NEW)
[*] Grant table substrate for block drivers (NEW)
[*] Block-device frontend driver (NEW)
[*] Network-device frontend driver (NEW)
[ ]   Pipelined transmitter (DANGEROUS) (NEW)
[ ] Block device tap driver (NEW) 
[ ] Fake shadow mode (NEW)
[*] Scrub memory before freeing it to Xen (NEW)
    Processor Type (X86)  --->

For the kernel for domain 0 I would want to select "Privileged Guest"
and "Physical device access" (automatically selected when "Privileged
Guest" is selected)?  For the kernels for virtual domains should these
two options not be selected?

Xen-users mailing list



Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.