Add code to generate SMBIOS tables to hvmloader.
Signed-off-by: Andrew D. Ball <aball@us.ibm.com>

Make room in e820map for SMBIOS tables.
Signed-off-by: Andrew D. Ball <aball@us.ibm.com>

Add utilities needed for SMBIOS generation to hvmloader.
Signed-off-by: Andrew D. Ball <aball@us.ibm.com>

[HVMLOADER] Include xen.h for __HYPERVISOR* constants.

Signed-off-by: Andrew D. BAll <aball@us.ibm.com>

[LIBXC] Remove unused xc_evtchn_status().

Signed-off-by: John Levon <john.levon@sun.com>

[LINUX] Mark /dev/mem regions as VM_PFNMAP.
Signed-off-by: Keir Fraser <keir@xensource.com>

[LINUX] Oney 'nosmp' and 'max_cpus=' command line options.
Based on a patch from Jan Beulich.
Signed-off-by: Keir Fraser <keir@xensource.com>

[NET] back: Fix an __init that should be an __exit in loopback driver.

This was pointed out by the newer build system used in 2.6.18-rc4.

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

[LINUX][ Fix x86/64 entry.S to use paravirtualized sti/cli properly.

This fixes
- one instance of using XEN_BLOCK_EVENTS where XEN_UNBLOCK_EVENTS was
  meant
- four instances of left-over cli-s that should have been replaced
  with XEN_BLOCK_EVENTS
- one instance of slightly wrong placement (without resulting in ill
  behavior, but needlessly making the diff to native more complex)

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

[XEN] grant tables: update pair of entries using a union to ensure endian neutrality.

The first two members of a grant entry are updated as a combined pair.
The following patch uses a union so updates can happen in an endian neutral fashion.

Signed-off-by: Jimi Xenidis <jimix@watson.ibm.com>

[LINUX] Provide a handy macro for detrmining if we are domain0.
Signed-off-by: Keir Fraser <keir@xensource.com>

[XENBUS] Avoid direct use of xen_start_info. It's unevailable when
building as a separate driver for an HVM guest.
Signed-off-by: Steven Smith <ssmith@xensource.com>
Signed-off-by: Keir Fraser <keir@xensource.com>

[NET] linux: Kill the WARN_ON() calls for checksum fixups.

This changeset is from upstream Linux.  We need it so Xen users
using NAT are not unnecessarily alarmed by these WARN_ON errors
especially since they're not in a position to do anything about
it.  Original changelog:

        [NET]: Kill the WARN_ON() calls for checksum fixups.

        We have a more complete solution in the works, involving
        the seperation of CHECKSUM_HW on input vs. output, and
        having netfilter properly do incremental checksums.

        But that is a very involved patch and is thus 2.6.19
        material.

        What we have now is infinitely better than the past,
        wherein all TSO packets were dropped due to corrupt
        checksums as soon at the NAT module was loaded.  At
        least now, the checksums do get fixed up, it just
        isn't the cleanest nor most optimal solution.

Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

[LINUX] Split useful bits of page.h into new header maddr.h.

Split the bits of page.h which are needed by PV-on-HVM drivers out
into a separate header file, so that it's a bit easier to get the xenolinux
machine address functions when compiled against a native kernel.

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

[HVM][LINUX] Add a header file with wrappers for hvm_op hypercall.
Signed-off-by: Steven Smith <ssmith@xensource.com>

[HVM] Add stubs to Linux for the new hvm_op hypercall.
Signed-off-by: Steven Smith <ssmith@xensource.com>

[NET] back: Tidy up and remove communications via references to global structures.
Signed-off-by: Steven Smith <ssmith@xensource.com>

[HVM] More info when failing to emulate instruction group 1.
Signed-off-by: Keir Fraser <keir@xensource.com>

[XEN] Remove sync_pagetable_state(). No longer needed.
Signed-off-by: Keir Fraser <keir@xensource.com>

[XEN] Add some missing put_page()s to __gnttab_copy.

Signed-off-by: Steven Smith <sos22@cam.ac.uk>

[XEN] Add a missing access check to __gnttab_copy.

Signed-off-by: Steven Smith <sos22@cam.ac.uk>

[XEN] Explicitly include domain_page.h for map_domain_page() and fix a
warning about arithmetic on 'void *' pointers.
Based on a patch from Jimi Xenidis <jimix@watson.ibm.com>
Signed-off-by: Keir Fraser <keir@xensource.com>

[XEN] Use gnttab_clear_flag() in new grant-table copy operation.
From: Jimi Xenidis <jimix@watson.ibm.com>
Signed-off-by: Keir Fraser <keir@xensource.com>

[XEN] Clean up e820 definitions. Define shared set in hvm public dir.
Signed-off-by: Keir Fraser <keir@xensource.com>

[HVM] HVMOP_get_param return parameter value within the provided parameter struct,
not as a direct return value.
Signed-off-by: Keir Fraser <keir@xensource.com>

[XEN] Add a warning about changeable size of shared_info structure.

sizeof(shared_info_t) can change even in compatible hypervisor
versions.
Add a clear warning about this, and fix the few places relying on it.
In addition, add some padding to arch_shared_info, to allow future
additions to the structure without breaking known offsets of members
following it in the shared info.

Signed-off-by: John Levon <john.levon@sun.com>

[XM-TEST] Fix network tests for -route and -nat

vif-nat and vif-route need to know domU ethX IP addresses in order to set
up the appropriate routes.The new network test infrastructure is not
providing the IP address on the vif line in /tmp/xm-test.conf; adding
it turns out to be a one-liner.

Tested with xm-test running with -bridge, -route, and -nat network
scripts.

Signed-off-by: Jim Dykman <dykman@us.ibm.com>

[NET] front: Restore 16 bytes of reservation

By replacing dev_alloc_skb with alloc_skb we need to do the 16-byte
reservation ourselves.  Also, it's unnecessary to align the size
when calling alloc_skb since the alignment is an implementation
detail of alloc_skb and it already takes care of it for us.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

[LINUX][XENBUS] Wait for devices to connect when frontend driver is loaded as a module.

Thsi fixes problems where a driver is loaded from a script, and the
following parts of the script depend on devices having connected
by the time the modprobe/insmod returns.

Based on a tested patch from Rik van Riel at Red Hat.

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

New xm uptime command.

From Alex Brett <alex.brett@xensource.com>

Signed-off-by: Ewan Mellor <ewan@xensource.com>

[IA64] Add two extra empty files and add a comment to every empty
source file so that they can be handled by patch/diff.
Signed-off-by: Keir Fraser <keir@xensource.com>

[XEN] Remove batched writable pagetable logic.

Benchmarks show it provides little or no benefit (except
on synthetic benchmarks). Also it is complicated and
likely to hinder efforts to reduce lockign granularity.

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

[NET] front: Do not allocate unnecessary page-sized main data area
for receive skbuffs. The bulk data is stored in fragments.
Signed-off-by: Keir Fraser <keir@xensource.com>

[XEN] Fix bug in spurious pagefault detection which could
be exploited by unprivileged guests. Thanks to Matt Yourst
for finding this and providing the patch.
Signed-off-by: Keir Fraser <keir@xensource.com>

[HVM] Remove unused apic_enabled field from hvm_info_table.
Signed-off-by: Keir Fraser <keir@xensource.com>

Revert inadvertent code change from "replace tabs with spaces" changeset.

Signed-off-by: Ewan Mellor <ewan@xensource.com>

[XEN] Add a new type of grant copy operation.
Based on an original patch from Rolf Neugebauer at Intel.
Signed-off-by: Steven Smith <ssmith@xensource.com>
Signed-off-by: Keir Fraser <keir@xensource.com>

Add support for sending event channel interrupts to HVM guests.
Signed-off-by: Steven Smith <ssmith@xensource.com>

[LIBXC] clean specified files instead of wildcarding
Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>

[LIBXC] replace tabs with spaces
Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>

[HVM] Fix some IOAPIC and LAPIC device model bugs.

Fix some boundary checking errors.
Fix the confusion using variables 'level' and 'trig_mode'.
Also fix some other misc mistakes and do some cleanup.

Signed-off-by: Dexuan Cui <dexuan.cui@intel.com>

[HVM] Rmove unused field pae_enabled from hvm_info_table.
Signed-off-by: Xin Li <xin.b.li@intel.com>

[MINIOS] A first step to re-architecture mini-os for a port to ia64.
Create architecture specific directories below mini-os for sources
and below mini-os/include for headers.

Signed-off-by: Dietmar Hahn <dietmar.hahn@fujitsu-siemens.com>

[CONSOLE] Always specify CON_ENABLED in Linux ocnsole driver.

I cannot think of any circumstance that the Xen console device would
not have CON_ENABLED.  With out this, DomUs that panic() before the
init process has started (and maybe even after) do not get output hi
priority log_levels (like panic()) from LOG_BUF().

Signed-off-by: Jimi Xenidis <jimix@watson.ibm.com>

Fix 2 issues when creating HVM guests:
 1) xend can't start due to xend-config.sxp
 2) qemu dm option vncunused should be effective only
    when vnc is turned on.

Signed-off-by: Xin Li <xin.b.li@intel.com>

[qemu patches] Update patches for changeset 11048:27bef4f3c1d3.

Signed-off-by: Christian Limpach <Christian.Limpach@xensource.com>

[qemu] Move acpi to devfn 2, so that it matches the fadt.

Based on feedback from: Wang, Winston L <winston.l.wang@intel.com>
Signed-off-by: Christian Limpach <Christian.Limpach@xensource.com>

[qemu patches] Update patches for changesets 11045:bd11c4855c5d and 11046:8a615d28759e.

Signed-off-by: Christian Limpach <Christian.Limpach@xensource.com>

[qemu] HVM acpi port init update.
Removed the acpi port depency on USB exist.

Based on a patch from: Winston Wang <winston.l.wang@intel.com>
Signed-off-by: Christian Limpach <Christian.Limpach@xensource.com>

[qemu] Disable qemu's acpi support.

Based on a patch from: Wang, Winston L <winston.l.wang@intel.com>
Signed-off-by: Christian Limpach <Christian.Limpach@xensource.com>

Merge with xen-ia64-unstable.

Merge with xenppc-unstable.

merge with xen-unstable.hg

[IA64] catch up new hypercall HYPERVISR_hvm_op for IPF (libxc)

append xc_set_hvm_param for IPF

Signed-off-by: Tsunehisa Doi <Doi.Tsunehisa@jp.fujitsu.com>
Signed-off-by: Tomonari Horikoshi <t.horikoshi@jp.fujitsu.com>

[IA64] catch up new hypercall HYPERVISR_hvm_op for IPF

append HYPERVISOR_hvm_op for IPF

Signed-off-by: Tsunehisa Doi <Doi.Tsunehisa@jp.fujitsu.com>
Signed-off-by: Tomonari Horikoshi <t.horikoshi@jp.fujitsu.com>

merge with xen-unstable.hg

[qemu patches] Update patches for changeset 10998:bd04004865ba.

Signed-off-by: Christian Limpach <Christian.Limpach@xensource.com>

[qemu] Add vncunused option.
If the port used for the requested display number is in use, try
additional ports until a free port is found.

Signed-off-by: Christian Limpach <Christian.Limpach@xensource.com>

[IA64] fix a fetch code bug

Fetch code may fail, if there is no corresponding tlb entry
in THASH-VTLB. This patch adds "retry mechanism" to resolve
this issue.

Signed-off-by: Anthony Xu <anthony.xu@intel.com>

merge

Make ballon and backend drivers fail gracefully if they are unable to initialize
due to missing hypervisor support.

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

Fix BUG_ON on switch_mm if XENFEAT_writable_page_tables is enabled.

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

[qemu patches] Update patches for changeset 10993:dc7b56b8cfb5.

Signed-off-by: Christian Limpach <Christian.Limpach@xensource.com>

[hvm/qemu] Add HVM buffered IO writes framework to accelerate stdvga speed.
With this patch, win2k VMX guest boots fine on xen.

Signed-off-by: Zhu Han <zhu.han@intel.com>
Signed-off-by: Eddie Dong <eddie.dong@intel.com>
Signed-off-by: Xin Li <xin.b.li@intel.com>
Signed-off-by: Christian Limpach <Christian.Limpach@xensource.com>

Build tools/firmware only on x86.
Signed-off-by: Keir Fraser <keir@xensource.com>

[XEN][IA64] IA64 build fixes after recent PER_CPU changes.
Signed-off-by: Alex Williamson <alex.williamson@hp.com>

Fix the unit test to match the new way of allocating VNC ports.

Signed-off-by: Ewan Mellor <ewan@xensource.com>

Added a couple of unit tests for the sxp parsing.

Signed-off-by: Ewan Mellor <ewan@xensource.com>

Fix reference to buf in parsing of ocaml escapes.

Signed-off-by: Ewan Mellor <ewan@xensource.com>

Better diagnosis of the failure when Xend crashes.  Thanks to Hollis Blanchard
for diagnosis.

Signed-off-by: Ewan Mellor <ewan@xensource.com>

[LIBXC] put PowerPC loader in libxenguest, not libxenctrl
Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>

[LIBXC] remove relocated file from Makefile
Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>

[IA64] Remove scheduler heartbeat

Signed-off-by: Daniel Miles <daniel.t.miles@hp.com>

[IA64] Virtualize dom0 lid

Because dom0 can't write eid/id into iosapics, lid can be virtualized.

Signed-off-by: Tristan Gingold <tristan.gingold@bull.net>

[IA64] bug fix in ioports_deny_access

Signed-off-by: Tristan Gingold <tristan.gingold@bull.net>

[IA64] enable ioports_deny_access for dom0

Signed-off-by: Tristan Gingold <tristan.gingold@bull.net>

[IA64] remove CONFIG_DOMAIN0_CONTIGUOUS and dom0_start

Remove unused build option CONFIG_DOMAIN0_CONTIGUOUS, thereby
making dom0_start obsolete.

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

[LIBXC] move xc_ppc_linux_build.c into a powerpc64 directory
Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>

[XEN] Assorted further PER_CPU- or read_mostly-ifications.
Signed-off-by: Keir Fraser <keir@xensource.com>

[XEN] Implement proper __read_mostly memory arrangemengt for x86.
Signed-off-by: Keir Fraser <keir@xensource.com>

[XEN] Make tracing structures and buffers explicitly PER_CPU.
Signed-off-by: Keir Fraser <keir@xensource.com>

[XEN] Make multicall info explicitly PER_CPU.
Also remove cacheline alignment in a few places. We leave
the common timer and schedule_data structures aligned because
they may be accessed randomly by remoted CPUs: at least we can
contain cacheline bouncing to just the offending structures.
Signed-off-by: Keir Fraser <keir@xensource.com>

[XEN] Make per-cpu pending EOI interrupt info explicitly PER_CPU.
Signed-off-by: Keir Fraser <keir@xensource.com>

[XEN] Make per-cpu mm information explicitly PER_CPU.
Signed-off-by: Keir Fraser <keir@xensource.com>

[XEN] Make per-cpu time information explicitly PER_CPU.
Signed-off-by: Keir Fraser <keir@xensource.com>

[XEN] Make per-cpu schedule data explicitly PER_CPU.
Signed-off-by: Keir Fraser <keir@xensource.com>

[XEN] Place per-cpu timer info in PER_CPU space.
Signed-of-fby: Keir Fraser <keir@xensource.com>

[XEN] Make tlbflush_time a PER_CPU variable.
Signed-off-by: Keir Fraser <keir@xensource.com>

[qemu patches] Refresh patches.

Signed-off-by: Christian Limpach <Christian.Limpach@xensource.com>

[qemu patches] Update patches for changeset 10974:415614d3a1ee.

Signed-off-by: Christian Limpach <Christian.Limpach@xensource.com>

[hvm/qemu] Flip the device model over to using the new Xen event channels
support.

Signed-off-by: Steven Smith <ssmith@xensource.com>

[hvm] Move the mapping of the shared IO page slightly earlier
in the initialisation sequence for HVM guests, so that we can
write vp_eport into the shared after creating the port.

Signed-off-by: Steven Smith <ssmith@xensource.com>

[XENSTAT] Ignore domains that are being created/destroyed during get_info call.

Currently, xenstat_get_node exits with a failure and returns NULL if a
VM is deleted between the time the list of VMs is obtained and the time
it attempts to collect information about the VMs. This patch modifies xenstat
to ignore domains that transition during the call.

Signed-off-by: Simon Graham <Simon.Graham@stratus.com>

[HVM][VMX] Cleanup vmx vmexit handler.

Current code doesn't handle vmexits caused by triple fault, INIT, SIPI,
etc, which are listed in IA32 Architecture System Programming Guide 2,
Appendix I VMX Basic Exit Reasons, and calls __hvm_bug() when not
handling such cases, actually domain crash should be good enough.

Also various coding style cleanups.

Signed-off-by: Xin Li <xin.b.li@intel.com>

[XEND] Daemonise vncviewer when it is spawned from xend.

There are various reasons it should be daemonized, but the particular
problem we hit was that YaST called "xm create" and waited on output on
stdout/stderr; xm then spawned vncviewer (which never closed its
inherited stdout and stderr); xm then would exit, but YaST still had
open file descriptors, and therefore waited forever.  It would be
possible to work around in YaST, but it seemed cleaner to daemonize
vncviewer.

Signed-off-by: Charles Coffing <ccoffing@novell.com>

[PCI] back: New virtual pci backend: slot

This backend use a slot per pci device.  Contrary to vpci two functions from
one slot appear as two slots. This is useful on ia64 because it doesn't scan
functions > 0 if function 0 does not exist.

Signed-off-by: Tristan Gingold <tristan.gingold@bull.net>

[XEND] Fix fallback and error path for domain destruction by domid,
or of non-existent domain.

I saw the following error message. (A domain of domain name XXX does
not exist.)

# xm list
Name                              ID Mem(MiB) VCPUs State  Time(s)
Domain-0                           0     1019     2 r-----    29.8
# xm destroy XXX
Error: an integer is required

This patch fixes it to print the following error message.

# xm list
Name                              ID Mem(MiB) VCPUs State  Time(s)
Domain-0                           0     1019     2 r-----    35.8
# xm destroy XXX
Error: the domain 'XXX' does not exist.

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

[HVM][VMX] Move hvm interface init_hypercall_page initialization into vmx_setup_hvm_funcs.
Signed-off-by: Xin Li <xin.b.li@intel.com>

[LIBXC] Do not gzclose() after close() on a gzdopen()'ed fd.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>

[HVMLOADER] Only install hvmloader if it is built.
This reverts a broken change in changeset 10948 that broke
the ia64 build.
Signed-off-by: Al Stone <ahs3@fc.hp.com>