linux: vmalloc_sync_all() is not a no-op when running in PAE mode on Xen.
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>

Fix 64-bit build.
Signed-off-by: Keir Fraser <keir@xensource.com>

[TAPDISK] add tapdisk support for image chaining

Enables tapdisk to chain an arbitrary number of VDIs, propagating reads of
holes in children to their parent images.  Introduces two new functions to the
tapdisk interface to facilitate this.  Modifies the QCoW plugin to take
advantage of these changes, thus providing support for arbitrarily long chains
of QCoW image types.

x86-64: fix Lx_DISALLOW_MASK

x86-64's Lx_DISALLOW_MASK values weren't properly taking care of the
NX bit, L3 and L4 were needlessly customized, and the compat mode L3
one wasn't really matching native PAE.

Signed-off-by: Jan Beulich <jbeulich@novell.com>

linux/x86_64: miscellaneous initialization code cleanup
Signed-off-by: Jan Beulich <jbeulich@novell.com>

hvm: Processor Objects require unique IDs.
Signed-off-by: Keir Fraser <keir@xensource.com>

hvm: Dynamically generate the correct number of ACPI Processor Objects
in an SSDT.
Signed-off-by: Keir Fraser <keir@xensource.com>

[XEN] Set p2m page count-info fields properly on allocation.
Signed-off-by: Tim Deegan <Tim.Deegan@xensource.com>

[HVM] Save/restore: explicitly name MSRs in save format
Signed off by: Mats Petersson (Mats.Petersson@amd.com)

Fix grant-table initialisation so that domain destruction does not barf.
Signed-off-by: Keir Fraser <keir@xensource.com>

linux: remove {lock,unlock}_vm_area(). Instead use vmalloc_sync_all()
in alloc_vm_area().
Signed-off-by: Keir Fraser <keir@xensource.com>

Fix vncdisplay config option for HVM guests.
Signed-off-by: Takanori Kasai <kasai.takanori@jp.fujitsu.com>

hvm: Fix acpi processor id in MADT LAPIC entries.
Signed-off-by: Keir Fraser <keir@xensource.com>

ia64: Fix build failure
Signed-off-by: Alex WIlliamson <alex.williamson@hp.com>

[HVM] Save/restore: remove u8 from publich header
Signed-off-by: Tim Deegan <Tim.Deegan@xensource.com>

Remove unnecessary check from HVM builder.
Thanks to Guy Zana for spotting this.
Signed-off-by: Keir Fraser <keir@xensource.com>

Lift physical address restriction in save/restore code.
Bump this to 44 bits for x86-32 and 52 bits for x86-64.

Signed-off-by: Jan Beulich <jbeulich@novell.com>

Propagate error return from elf_xen_parse().
From: Jan Beulich <jbeulich@novell.com>
Signed-off-by: Keir Fraser <keir@xensource.com>

x86: fix NMI delivery to dom0
Signed-off-by: Jan Beulich <jbeulich@novell.com>

Bring TPM driver files to 2.6.18 level. Remove the tpm.c so that the
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>

[ACM] Update the xensec_ezpolicy ACM policy generation tool so
that it works with wxPython under Python 2.5.
Signed-off by: Reiner Sailer <sailer@us.ibm.com>

minios: build system cleanups.

- Added ARCH_LDFLAGS for architecture specific LDFLAGS
- Fixed build dependencies after changing makerule files
- Fixed ARCH_CFLAGS for 64bit guest, added ARCH_ASFLAGS
- Couple of variable renames

Signed-off-by: Dietmar Hahn <dietmar.hahn@fujitsu-siemens.com>
Signed-off-by: Grzegorz Milos <gm281@cam.ac.uk>

Disable GCC stack protector for ROMBIOS.
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

linux: miscellaneous changes after upgrade through 2.6.17.
Signed-off-by: Jan Beulich <jbeulich@novell.com>

linux: e820 handling adjustments
In certain places, machine_e820 should be used rather than e820.
Signed-off-by: Jan Beulich <jbeulich@novell.com>

linux: user user_mode_vm() in place of user_mode() where necessary.
Signed-off-by: Jan Beulich <jbeulich@novell.com>

linux: avoid need to modify include/linux/pfn.h
Signed-off-by: Jan Beulich <jbeulich@novell.com>

linux: early_printk adjustments

This is a little more than a pure merge with 2.6.17, to remove a few
more non-Xen items.

Signed-off-by: Jan Beulich <jbeulich@novell.com>

linux: proper mutex use in balloon driver
Mutexes are supposed to not use the old semaphore technique anymore.
Signed-off-by: Jan Beulich <jbeulich@novell.com>

[VTPM] Add a missing initialization of a state variable in the TIS
device model so that the first response received after a resume is not
aborted.

Signed-off-by: Stefan Berger <stefanb@us.ibm.com>

libxc: Fix xc_vcpu_{set,get}_affinity (Xen-bugzilla No.900).
http://bugzilla.xensource.com/bugzilla/show_bug.cgi?id=900

Signed-off-by: Masaki Kanno <kanno.masaki@jp.fujitsu.com>

Dynamic grant-table sizing.
Signed-off-by: Christopher CLark <christopher.clark@cl.cam.ac.uk>
Signed-off-by: Andrei Petrov <andrei.petrov@xensource.com>
Signed-off-by: Keir Fraser <keir@xensource.com>

Remove unuseful and noisy seg-fixup warning message.
Signed-off-by: Keir Fraser <keir@xensource.com>

Remove old elf-parsing code from tools and from Xen.
Signed-off-by: Keir Fraser <keir@xensource.com>

Protect self.domid from being empty

Export PRI_xen_pfn rather than PRIpfn: less polluting of namespace.
libxc continues to use PRIpfn internally, #defined to PRI_xen_pfn.

Signed-off-by: Keir Fraser <keir@xensource.com>

Merge with xenppc-unstable.hg

libelf: use xen code style.
Signed-off-by: Keir Fraser <keir@xensource.com>

linux: quieten some driver printk logging.
Signed-off-by: Keir Fraser <keir@xensource.com>

Another domctl fix to insert missing copy_to_guest.
Signed-off-by: Gerd Hoffmann <kraxel@suse.de>

Merge

[XEN] Snapshot PAE l3es when they are shadowed.
We don't update the shadows so we mustn't look at the guest l3es
or we'll be confused by them if they change.
Signed-off-by: Tim Deegan <Tim.Deegan@xensource.com>

[XEN] Re-sync PAE l3 shadows when PV guests change them
-in particular, for PAE Solaris guests
Signed-off-by: Tim Deegan <Tim.Deegan@xensource.com>

Make libelf not fail on unknown elf notes.
Signed-off-by: Gerd Hoffmann <kraxel@suse.de>

Add missing breaks to {set,get}_address_size domctls.
Signed-off-by: Gerd Hoffmann <kraxel@suse.de>

Fix DEC instruction decode. Turn into SUB, not OR. :-)
Signed-off-by: Keir Fraser <keir@xensource.com>

Support INC/DEC in mmio decoder. Opcodes 0xFE and 0xFF.
Signed-off-by: Keir Fraser <keir@xensource.com>

[XEN] New paging-assistance interface.
Adds paging_* functions, wrapping the existing shadow_* ones,
so that later hardware paging assistance can be hooked in at this level.
Separates p2m maintenance code into its own files.
Adjusts shadow code to comply with the new interfaces.
Changes all callers in arch/x86 to use the new interfaces.

Signed-off-by: Wei Huang <Wei.Huang2@amd.com>
Signed-off-by: Tim Deegan <Tim.Deegan@xensource.com>

[LINUX] Call ctrl_alt_del() to trigger a reboot so we benefit from the
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>

[POWERPC][XEN] Use libelf for dom0 construction.
- Remove old 32-bit ELF loader hack.
Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>

[LIBELF] Define PRIpfn next to xen_pfn_t.
Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>

[POWERPC][XEN] Opt out of new libxc domain builder.
PowerPC needs more flexibility than the new framework allows.
Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>

[LIBELF] Print phdr addresses when loading.
Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>

[POWERPC][XEN] Fix RCU crash.
Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>

[POWERPC][XEN] strlcpy() fallout.
- Implement strlcpy() for the dom0 firmware.
- Remove strncpy() from dom0 firmware.
- Correct buffer length in device tree copying.
Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>

[HVM] Fill in mmio decode gaps for Grp1 immediate opcodes.
Signed-off-by: Keir Fraser <keir@xensource.com>

Print all by default in debug builds.

[HVM] Save/restore: disentangle max_pfn from nr_pages.
These have been used interchangeably, which is OK for PV domains but
not for HVM.  This fixes an over-allocation by 256MB when restoring
HVM guests that have more than 4GB of RAM.
Signed-off-by: Tim Deegan <Tim.Deegan@xensource.com>

Add missing Makefile in x86_64/compat so that it builds on x86_64.

Signed-off-by: Alastair Tse <atse@xensource.com>

Fix non-x86 build.

Fix x86/64 build.
Signed-off-by: Keir Fraser <keir@xensource.com>

x86: hide assembly labels

Prevent the fault recovery labels to clutter the symbol table and the
disassembly.

Signed-off-by: Jan Beulich <jbeulich@novell.com>

[HVM][SVM] Do not advertise FFXSR feature.
Signed-off-by: Keir Fraser <keir@xensource.com>

Use printk log-level limiting even during Xen boot.
Signed-off-by: Alex Williamson <alex.williamson@hp.com>

Small cleanups to PV-on-HVM support code.
Signed-off-by: Keir Fraser <keir@xensource.com>

Merge with xen-ia64-unstable.hg.

Remove unnecessarily modified a.out.h.
Signed-off-by: Jan Beulich <jbeulich@novell.com>

Move skb_pull_rcsum() to its usual place in skbuff.c

Fix a typo in console_steal().
Signed-off-by: Kevin Tian <kevin.tian@intel.com>

Fix xen-detect build.
Signed-off-by: Keir Fraser <keir@xensource.com>

Add example xen-detect.c code for detecting application execution in
Xen PV or HVM contexts, using the Xen CPUID extensions.
Signed-off-by: Keir Fraser <keir@xensource.com>

Fix localtime setting for PV domains.

Signed-off-by: Atsushi SAKAI <sakaia@jp.fujitsu.com>

[HVM] Save/restore: don't try to set the console ring MFN
in a HVM restore.  In particular, don't print an uninitialized stack
variable, which causes XMLRPC exceptions in xend.
Signed-off-by: Tim Deegan <Tim.Deegan@xensource.com>

Merge

[HVM] Save/restore: save pmtimer count register
Also remove the repeating timer from pmtimer.c because it doesn't do anything.
Signed-off-by: Tim Deegan <Tim.Deegan@xensource.com>

[XEN] Fix typos in comment describing 32on64 memory layout

Signed-off-by: Ian Campbell <ian.campbell@xensource.com>

[XEN] kexec: add compatability shim for kexec in 32on64 mode.

Signed-off-by: Ian Campbell <ian.campbell@xensource.com>

[XEN] kexec: Remove asm/kexec.h. Move the single inline function into
arch specific machine_kexec.c with the other arch specific kexec
functions.

IA64 already had a stub in both kexec.h and machine_kexec.c.

Signed-off-by: Ian Campbell <ian.campbell@xensource.com>

[XEN] When removing pages, drop shadow refs before complaining about refcount.
Signed-off-by: Tim Deegan <Tim.Deegan@xensource.com>

[POWERPC][XEN] Fix mismerge.

[blktap] Allow HVM booting from blktap device(s)

Signed-off-by: wim colgate <wim@xensource.com>

merge with xen-unstable.hg

[HVM] Save/restore: don't try to save the MMIO memory-hole at 4GB.
Also remove two integer arrays that were set to a[i] == i.
Signed-off-by: Tim Deegan <Tim.Deegan@xensource.com>

[XEND] Open save/restore files with O_LARGEFILE if possible
Signed-off-by: Tim Deegan <Tim.Deegan@xensource.com>

[IA64] Clean up error messages from lookup_domain_mpa()

Guest I/O port accesses aren't very interesting, so reduce their log level

Signed-off-by: Alex Williamson <alex.williamson@hp.com>

Improve hotplug script error reporting via xenstore.

Use this to detect and report up-front in vif-bridge script whether
the bridge device exists or not.

Signed-off-by: Daniel P. Berrange <berrange@redhat.com>

Dom0 VNC server should not drop any of teh 8 supported mouse button
event types.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>

pvfb: Split mouse and keyboard into separate devices.

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>

hvm: Add PCI subsystem ID to emulated devices.

This patch adds the PCI Subsystem ID 5853:0001 to the virtual
devices emulated by qemu (ide, nic, vga).

Signed-off-by: Gerd Hoffmann <kraxel@suse.de>

[XEN] Remove bogus assertion in shadow emulation path
Signed-off-by: Tim Deegan <Tim.Deegan@xensource.com>

[HVM] Save/restore: dynamically calculate the size of the save buffer
Signed-off-by: Tim Deegan <Tim.Deegan@xensource.com>

[HVM] Save/restore: move save/restore routines out of intercept.c
Signed-off-by: Tim Deegan <Tim.Deegan@xensource.com>

Revert 13805: losetup -r is not portable
which breaks CD support on Debian Sarge, FC4, probably others.
Signed-off-by: Tim Deegan <Tim.Deegan@xensource.com>

linux: Fix softlockup interaction with noidlehz.

next_timer_interrupt() must check for when timer ISR is due to wake up
the softlockup thread.

Signed-off-by: Kevin Tian <kevin.tian@intel.com>
Signed-off-by: Keir Fraser <keir@xensource.com>

Update .hgignore list

linux: Zap /proc/interrupts count when a dynamic IRQ is unbound.
Signed-off-by: Keir Fraser <keir@xensource.com>

linux/i386: Improve critical region handling code

- 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>

hvm: vcpu reset support for x86 INIT IPI, needed for CPU hotplug.
Signed-off-by: Xin Li <xin.b.li@intel.com>

minios: Fix netfront for ia64.
Signed-off-by: Dietmar Hahn <dietmar.hahn@fujitsu-siemens.com>

Fix up xentrace record extraction in xentrace_format for 64-bit
architectures.
Signed-off-by: Thomas Friebel <thomas.friebel@amd.com>