tech-kern archive

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

Re: pmf(9) and Xen domU suspending?

Joerg Sonnenberger wrote:
OK, That makes things a lot easier. If you have two more days, I can
work out the details with Jean-Yves directly.
Great, thanks.

Question regarding sysmon events though: in dev/sysmon/sysmon_power.c:965, the PSWITCH_TYPE_SLEEP case seems to catch the event for the SLEEP, but ends in a NOOP (when comparing it to other entries like RESET or POWERDOWN, which calls cpu_reboot(...) ). Does that mean that the SLEEP event should be handled somewhere else?

On Sat, Mar 29, 2008 at 06:21:22PM +0100, Jean-Yves Migeon wrote:
I am working on suspend/resume functionality for port-xen under the supervision of Manuel Bouyer (bouyer@) and Stoned Elipot (seb@). Currently, I am mostly trying to get familiar with NetBSD internals, and Xen.
You mean the Xen eqivalent of suspend-to-RAM? Is this for domU or dom0?
I do not know the exact behaviour regarding kernel state for a suspend-to-RAM. Xen manipulates MMU directly, the domu is not (well, there are some corner cases like writable page tables, but it adds up complexity when restoring a VM).

I suppose that we will have some differences though, as a domu uses mostly "pseudo" physical frame numbers (Xen adds another level for memory mappings). And these mappings can eventually change upon a restore (between the pseudo physical addresses and the real machine ones).

For suspend-to-ram, I guess the OS keeps its PDs and PTs entries intact, since there is no reason for them to change between suspend and resume.

(As always, I may be wrong though).

Thanks for the replies :)

Jean-Yves Migeon

Home | Main Index | Thread Index | Old Index