Xen users - Nov 2005 - FreeBSD 5.3 as DomU - new kernel with sysinstall embedded

On 16/11/05, Kip Macy <kip.macy@gmail.com> wrote:
> The way the initial memory space gets setup there is no way MFS could work.
> It wasn''t an anticipated use. I guess somewhere either in FreeBSD
GNATS or
> xen''s bugzilla there should be i386-xen FreeBSD bugs filed for me
to track.
I intend to file a bug report sooner or later, but in the meantime I
decided to toy around with the kernel and the mfsroot from
5.3-release, and I did eventually come up with something suitable. By
a combination of stripping down both the DOMU kernel and the
sysinstall floppy, I managed to get the pair together in a size that
works and doesn''t caush an immediate crash under Xen.  I''ve
FTP
installed 10 different guests with it now :-)

Details and the files are at http://txrx.org/xen/

Regards

Chris

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

On Mon, Nov 21, 2005 at 08:59:34PM -0600, Chris Brookes
wrote:
> On 16/11/05, Kip Macy <kip.macy@gmail.com> wrote:
> 
> Details and the files are at http://txrx.org/xen/
> 
In response to some of the open questions on your website.

apmd is what you want for power managment, never tried it 
under xen though.

I do not believe you need the extra = "boot.netif.ip=192.168.0.221"
line in your config, although the layout of the args line in the 
actual config does seem importent, try "xm create -n -f <your
config>"
and see what you get. The following works for me using a raw disk
partition under FC3. However if the args line spans multiple lines, 
or as I recall the comma is missing, the root disk config seems to get 
lost.

(vm
    (name FreeBSD-3)
    (memory 150)
    (cpu -1)
    (restart always)
    (image (linux (kernel /boot/kernel-xbd)
    (args '',vfs.root.mountfrom=ufs:/dev/xbd0a'')
        ))
    (device (vbd (uname phy:ida/c1d0p2) (dev ida/c1d0p2) (mode w)))
    (device (vif (mac aa:00:00:42:8a:b5)))
)

I wouldn''t descibe this as stable but it does work well enough to 
compile new kernels in the guest, building complex ports reliably
crashes or hangs the system.

I have found kernels compiled using the sparse directory and scripts
in the 2.0.7 source distribution *ARE* more stable than those built 
from the source on Kip Macey website. (You need bash and wget 
installed to use those scripts)  

DHCP does seem to work.

I have found NFS root guest domains to be more unstable.




_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

On 22/11/05, Geoff Buckingham <geoffb@chuggalug.clues.com> wrote:
> apmd is what you want for power managment, never tried it
> under xen though.
Just tried it quickly, and it won''t run. Will take a proper look later.
> I do not believe you need the extra =
"boot.netif.ip=192.168.0.221"
You are absolutely right: you just need a "," at the start of the
extra= line. That must be a real simple bug somewhere...
> I wouldn''t descibe this as stable but it does work well enough to
> compile new kernels in the guest, building complex ports reliably
> crashes or hangs the system.
Compiling complex ports is exactly what I''ve been doing while
experiencing crashes...
> I have found kernels compiled using the sparse directory and scripts
> in the 2.0.7 source distribution *ARE* more stable than those built
I''ll give that a try, thanks :-)
> DHCP does seem to work.
So noted, HOWTO updated. Thanks for the feedback.

Regards

Chris

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

On 22/11/05, Chris Brookes <cbrookes@gmail.com>
wrote:
> Compiling complex ports is exactly what I''ve been doing while
> experiencing crashes...
I just tried a FreeBSD kernel compiled from the xen-sparse source in
Xen 2.0.7-testing-src. It seemed like it was coping with the ports
build (bash3) and the kernel compile I threw at it, until it exploded.
Much to my surprise it did leave some info though:

./mkbuiltins -D . caller.def
cc -c  -DHAVE_CONFIG_H -DSHELL   -I. -I..  -I.. -I../include -I../lib
-I.   -I/usr/local/includ               e -O -pipe caller.c || ( rm -f
caller.c ; exit 1 )


Fatal trap 12: page fault while in kernel mode
fault virtual address   = 0xc029bc70
fault code              = supervisor read, page not present
instruction pointer     = 0x819:0xc019129e
stack pointer           = 0x821:0xc4004c2c
frame pointer           = 0x821:0xc4004c34
code segment            = base 0x90c, limit 0xc01bae51, type 0x4
                        = DPL 2, pres 1, def32 1, gran 0
processor eflags        = interrupt enabled, resume, IOPL = 0
current process         = 12519 (wget)
trap number             = 12
panic: page fault
Uptime: 34m31s
Cannot dump. No dump device defined.
Automatic reboot in 15 seconds - press a key on the console to abort
--> Press a key on the console to reboot,
--> or switch off the system now.
Rebooting...

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

On Tue, Nov 22, 2005 at 03:39:13PM -0600, Chris Brookes
wrote:
> On 22/11/05, Chris Brookes <cbrookes@gmail.com> wrote:
> > Compiling complex ports is exactly what I''ve been doing while
> > experiencing crashes...
> 
> I just tried a FreeBSD kernel compiled from the xen-sparse source in
> Xen 2.0.7-testing-src. It seemed like it was coping with the ports
> build (bash3) and the kernel compile I threw at it, until it exploded.
> Much to my surprise it did leave some info though:
> 
As I said it is less unstable, not stable.
> 
> Fatal trap 12: page fault while in kernel mode
> fault virtual address   = 0xc029bc70
> fault code              = supervisor read, page not present
> instruction pointer     = 0x819:0xc019129e
> stack pointer           = 0x821:0xc4004c2c
> frame pointer           = 0x821:0xc4004c34
> code segment            = base 0x90c, limit 0xc01bae51, type 0x4
>                         = DPL 2, pres 1, def32 1, gran 0
> processor eflags        = interrupt enabled, resume, IOPL = 0
> current process         = 12519 (wget)
> trap number             = 12
> panic: page fault
> Uptime: 34m31s
try

"nm -n <your kernel> |grep c019129e"  then c019129, c019129
which may help find where this happened

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

On 23/11/05, Geoff Buckingham <geoffb@chuggalug.clues.com> wrote:
> "nm -n <your kernel> |grep c019129e"  then c019129, c019129
> which may help find where this happened
Seems that the crash happened in function mcl_queue_pt_update  in
xen_machdep.c.It looks like that function''s already recieved some
attention in patch
http://www.fsmware.com/xenofreebsd/5.3/050418/freebsd.patch ...

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

On Wed, Nov 23, 2005 at 06:58:45PM -0600, Chris Brookes
wrote:
> On 23/11/05, Geoff Buckingham <geoffb@chuggalug.clues.com> wrote:
> 
> > "nm -n <your kernel> |grep c019129e"  then c019129,
c019129
> > which may help find where this happened
> 
> Seems that the crash happened in function mcl_queue_pt_update  in
> xen_machdep.c.It looks like that function''s already recieved some
> attention in patch
> http://www.fsmware.com/xenofreebsd/5.3/050418/freebsd.patch ...
Hmm, it seems it ma be worth fetching the 2.0-tetsing code, then seeing 
if any of the web site patches still need to be applied...

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users