[Fedora-xen] Latest Xen RPMS testing in FC5.

[Stephen C. Tweedie]

Hi,

On Thu, 2006-03-09 at 14:32 +0800, You, Yongkang wrote:

> I couldn't reproduce the failed udev problem in either IA32 or x86_64
> again. But I met it once on both IA32 and x86_64 platforms yesterday.
> If I catch it next time, I will record the detail and send to you.

Thanks.  It may be related to a driver problem that we saw sporadically
before and have also been unable to reproduce in 1.2032 --- it's
possible that this was related to one of the other fixes that went into
the underlying kernel when we moved from git8 to git9.

> For the x86_64 problem of hang xen0 if destroying HVM(VMX), our
> developer shared a information maybe usefully. From the failed info of
> serial out, it seemed debug=y was opened when making Xen. And VMX will
> meet some problem if debug option is opened. Could you have a double
> check? And let the debug=n? :)

What's the problem?  I'd rather fix it than disable debug, and we're
really close to the scheduled FC5 release date now so I don't want to be
making significant changes at this point.

> For enabling serial console in Xen and Xen0, we still meet the bug
> https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=180126 . We have
> to disable acpi to enable serial console operation and output, or Xen0
> couldn't boot up. Do you have some updates?

No -- it always works for me.  But I did notice something from the
bugzilla that might be important, and I just added:
        
                kernel /boot/xen.gz-2.6.15-1.29_FC5 com1=115200,8n1
                module /boot/vmlinuz-2.6.15-1.29_FC5hypervisor ro root=/dev/sda1 
        console=tty0 console=ttyS0 sync_console
        
        This is wrong, the "sync_console" needs to be on the "kernel" hypervisor option
        list, not the xen0 "module" list.  Can you reproduce with that change? 

Cheers,
 Stephen