A 3.0.2 compatibility requirement slipped my attention before
submitting the original page table handling changes.
Signed-off-by: Jan Beulich <jbeulich@novell.com>
Ensure that all and only those page table entries that have their
present bit set undergo p2m/m2p translation in all relevant places.
This should fix migration issues with _PAGE_PROTNONE pages which
previously could retain MFNs in PTEs while having the present bit
clear (and thus were not getting (un)canonicalized during save/
restore).
Many thanks to Keir Fraser for his analysis, suggestions, and
corrections.
Signed-off-by: Jan Beulich <jbeulich@novell.com>
Also consolidate time-xen.c in giving get_time_values_from_xen() a
'cpu' parameter to match other functions (this consolidation could of
course also be done the other way around, but I think this way the
resulting code can be more efficient).
Signed-off-by: Jan Beulich <jbeulich@novell.com>
Also vmalloc_sync_all() is only defined for x86 architectures so make
use of it in driver code conditional on CONFIG_X86.
Signed-off-by: Keir Fraser <keir@xensource.com>
one from the kernel can be used and reduce changes to a minimum
required to compile the xen-specific driver.
Signed-off-by: Stefan Berger <stefanb@us.ibm.com>
generic infrastructure for cad_pid etc.
Add a ctrl_alt_del() implemenation to the PV on HVM compatibility
layer since ctrl_alt_del() is not exported on native kernels.
Signed-off-by: Ian Campbell <ian.campbell@xensource.com>
This makes all domain boot messages come out through xencons. Previously
the hpsim console allowed an early console via Xen. Also, make xencons
the preferred console unless a console is explicitly provided.
Signed-off-by: Alex Williamson <alex.williamson@hp.com>
This patch creates two separate input devices for keyboard and mouse
events. The reason for this is to separate them in the linux input
layer and allow them being routed different ways.
Use case: Configure the X-Server like this to get the mouse
events directly from the linux input layer, which has the major
advantage that absolute coordinates work correctly:
Section "InputDevice"
Driver "evdev"
Identifier "Mouse"
Option "Device" "/dev/input/event<nr>"
EndSection
This makes the keyboard stop working though in case mouse and
keyboard events are coming through the same input device, at least
with older Xorg (6.9) versions.
Signed-off-by: Gerd Hoffmann <kraxel@suse.de>
Assign fake pci product identifiers down from 0xffff. This may help
clarify that the devices are not actually PCI. Also it reduces
probability of accidental conflict if further real PCI assignments
occur in future (0x0002, 0x0003, ... would be obvious next identifiers
to allocate).
Signed-off-by: Keir Fraser <keir@xensource.com>
- use symbolics instead of raw numbers (going to help the 2.6.20 port,
as those numbers change there)
- fold instructions where possible
- move fixup table into .rodata
Signed-off-by: Jan Beulich <jbeulich@novell.com>
machine_halt() stops the vcpus, but no longer destroys the domain.
This should be more in-line with the behavior on native hardware.
Signed-off-by: Akio Takebe <takebe_akio@jp.fujitsu.com>
