xen.git
18 years agoFix/cleanup destroyDevice code path in xend.
kfraser@localhost.localdomain [Mon, 6 Aug 2007 09:12:34 +0000 (10:12 +0100)]
Fix/cleanup destroyDevice code path in xend.

When calling destroyDevice code path (e.g. xm block-detach dom devid),
allow specifying an integer device id or a device name such as xvdN or
/dev/xvdN.  Allowing the /dev/xvdN form is useful when detaching
devices from dom0.  Bootloaders may do this to unmount a disk
previously mounted in dom0.

Move examination of device ID format into the DevController,
permitting device controllers to determine a valid device ID instead
of higher level code.

Signed-off-by: Jim Fehlig <jfehlig@novell.com>
18 years ago[ACM] Support for removing / replacing a policy
kfraser@localhost.localdomain [Mon, 6 Aug 2007 09:11:25 +0000 (10:11 +0100)]
[ACM] Support for removing / replacing a policy

Add support for removing/replacing a policy from the running
system. This operation is only successful if currently running VMs
would also be allowed to run under the new policy. Removing the
current policy means that the default policy is installed, which then
only has support for a single VM label and STE type (SystemManagement)
and is the same policy that the system starts up with when no policy
is chosen.

Signed-off-by: Stefan Berger <stefanb@us.ibm.com>
18 years ago[ACM] Support for running unlabeled domains alongside labeled ones
kfraser@localhost.localdomain [Mon, 6 Aug 2007 09:10:34 +0000 (10:10 +0100)]
[ACM] Support for running unlabeled domains alongside labeled ones

Add support for running unlabeled domains alongside labeled ones, if
the policy contains a VM label with name '__UNLABELED__' and an STE
type with the same name. The ezpolicy tool has been modified to
automatically suggest a policy under which unlabeled domains can
run. The user may delete this, if this is not desired.

Signed-off-by: Stefan Berger <stefanb@us.ibm.com>
18 years agodomain builder: Zero start_info struct.
kfraser@localhost.localdomain [Fri, 3 Aug 2007 11:23:03 +0000 (12:23 +0100)]
domain builder: Zero start_info struct.
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years agoOn debug builds, scrub pages with non-zero poison.
kfraser@localhost.localdomain [Fri, 3 Aug 2007 11:22:33 +0000 (12:22 +0100)]
On debug builds, scrub pages with non-zero poison.
Will flush out guests which are relying on zeroed memory.
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years ago[HVM] Yet another MCA/MCE MSR.
Tim Deegan [Fri, 3 Aug 2007 11:10:35 +0000 (12:10 +0100)]
[HVM] Yet another MCA/MCE MSR.
Signed-off-by: Tim Deegan <Tim.Deegan@xensource.com>
18 years agoes7000: Remove the bulk of the ES7000 specific code from Xen.
kfraser@localhost.localdomain [Fri, 3 Aug 2007 08:04:29 +0000 (09:04 +0100)]
es7000: Remove the bulk of the ES7000 specific code from Xen.

On the Unisys ES7000-One, Xen now uses the bigsmp APIC driver. This patch
has been tested on the ES7000 and a commodity box.

Signed-off-by: Raj Subrahmanian <raj.subrahmanian@unisys.com>
18 years agoMerge with ppc/xen-unstable.hg
kfraser@localhost.localdomain [Thu, 2 Aug 2007 15:03:35 +0000 (16:03 +0100)]
Merge with ppc/xen-unstable.hg

18 years agomerge with xen-unstable.hg
Hollis Blanchard [Thu, 2 Aug 2007 14:54:18 +0000 (09:54 -0500)]
merge with xen-unstable.hg

18 years ago[POWERPC][XEN] Commit missing multiboot files.
Hollis Blanchard [Thu, 2 Aug 2007 14:50:55 +0000 (09:50 -0500)]
[POWERPC][XEN] Commit missing multiboot files.
Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>
18 years agolibelf: Clean up and fix up BSD symtab handling.
kfraser@localhost.localdomain [Thu, 2 Aug 2007 13:18:38 +0000 (14:18 +0100)]
libelf: Clean up and fix up BSD symtab handling.
Should now (hopefully) work for domU and dom0 builders.
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years agoelfloader: Re-add break stmt I stupidly removed from elf parser.
kfraser@localhost.localdomain [Thu, 2 Aug 2007 09:05:14 +0000 (10:05 +0100)]
elfloader: Re-add break stmt I stupidly removed from elf parser.
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years agolibelf: Remove unneeded extra elf_set_verbose() invocation.
kfraser@localhost.localdomain [Thu, 2 Aug 2007 08:53:13 +0000 (09:53 +0100)]
libelf: Remove unneeded extra elf_set_verbose() invocation.
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years agoMerge
Tim Deegan [Thu, 2 Aug 2007 08:02:26 +0000 (09:02 +0100)]
Merge

18 years ago[MINIOS] fix x86_64 p2m map and CFLAGS
Tim Deegan [Thu, 2 Aug 2007 08:02:08 +0000 (09:02 +0100)]
[MINIOS] fix x86_64 p2m map and CFLAGS
Signed-off-by: Tim Deegan <Tim.Deegan@xensource.com>
18 years agoxenstore: Redirect stdin/stdout/stderr to /dev/null.
kfraser@localhost.localdomain [Wed, 1 Aug 2007 18:34:56 +0000 (19:34 +0100)]
xenstore: Redirect stdin/stdout/stderr to /dev/null.
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years agoxenstore: Do not assign to stdout/stderr/stdin - they are not
kfraser@localhost.localdomain [Wed, 1 Aug 2007 18:17:08 +0000 (19:17 +0100)]
xenstore: Do not assign to stdout/stderr/stdin - they are not
implemented as variables on all systems.
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years ago[ELF] Load elf symbols when BSD_SYMTAB=yes.
kfraser@localhost.localdomain [Wed, 1 Aug 2007 14:47:54 +0000 (15:47 +0100)]
[ELF] Load elf symbols when BSD_SYMTAB=yes.

When a guest kernel specifies BSD_SYMTAB=yes, then Xen loads the ELF
symbols for it. This works with Xen 3.0.4, but not with Xen 3.1.
During the libelf work between Xen 3.0.4 and Xen 3.1 the loading got
broken in the way, that BSD_SYMTAB gets parsed but not handled.

Signed-off-by: Christoph Egger <Christoph.Egger@amd.com>
18 years agoxenstored: Do not write to stderr if we are daemonised!
kfraser@localhost.localdomain [Wed, 1 Aug 2007 11:55:10 +0000 (12:55 +0100)]
xenstored: Do not write to stderr if we are daemonised!
This fixes client reader-thread deaths in which a 'garbage string' was
being read instead of a well-formed message header.
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years agoxenstored: Remove unused util code.
kfraser@localhost.localdomain [Wed, 1 Aug 2007 11:05:42 +0000 (12:05 +0100)]
xenstored: Remove unused util code.
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years agoblktap: Fix unaligned access to blktap/tapdisk message payloads.
kfraser@localhost.localdomain [Wed, 1 Aug 2007 08:11:08 +0000 (09:11 +0100)]
blktap: Fix unaligned access to blktap/tapdisk message payloads.
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years agoFix file descriptor leak in blktapctrl
kfraser@localhost.localdomain [Wed, 1 Aug 2007 08:04:20 +0000 (09:04 +0100)]
Fix file descriptor leak in blktapctrl

The blktapctrl process is responsible for spawning individual tapdisk
processes. It does this using the 'system' method, but unfortunately
none of its file descriptors have the close-on-exec flag set. The
parent blktapctl process opens a couple of unix domain sockets
per-tapdisk it spawns. So the first tapdisk get 2 FDs leaked to it,
the second gets 4 FDs leaked to it, the 3rd gets 6 and so on. The use
of 'system' also unneccessarily invokes the shell.

Replace system with fork/execvp, and explicitly close all file handles
up to _SC_OPEN_MAX.

Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
18 years agoMerge with ppc/xen-unstable.hg.
kfraser@localhost.localdomain [Wed, 1 Aug 2007 07:59:03 +0000 (08:59 +0100)]
Merge with ppc/xen-unstable.hg.

18 years ago[IA64] Fix tools build.
kfraser@localhost.localdomain [Wed, 1 Aug 2007 07:57:55 +0000 (08:57 +0100)]
[IA64] Fix tools build.
Signed-off-by: Alex Williamson <alex.williamson@hp.com>
18 years agohvm: Large-scale cleanups and fixes to event deliver logic.
kfraser@localhost.localdomain [Tue, 31 Jul 2007 17:59:00 +0000 (18:59 +0100)]
hvm: Large-scale cleanups and fixes to event deliver logic.
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years agohvmloader: Avoid bogus use of NULL pointer.
kfraser@localhost.localdomain [Tue, 31 Jul 2007 14:09:45 +0000 (15:09 +0100)]
hvmloader: Avoid bogus use of NULL pointer.
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years ago[HVM] Inject #PF when mmio instruction fetch fails
Tim Deegan [Tue, 31 Jul 2007 10:37:27 +0000 (11:37 +0100)]
[HVM] Inject #PF when mmio instruction fetch fails
instead of crashing the guest.  This can happen if one vcpu pages out
another vcpu's kernel text page while the other is performing an mmio op.
Signed-off-by: Tim Deegan <Tim.Deegan@xensource.com>
18 years agohvm: Define common (across VMX and SVM) set of event types.
kfraser@localhost.localdomain [Tue, 31 Jul 2007 09:11:47 +0000 (10:11 +0100)]
hvm: Define common (across VMX and SVM) set of event types.
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years agohvm: Fix CR0 handling, which I broke in 15652.
kfraser@localhost.localdomain [Tue, 31 Jul 2007 08:47:11 +0000 (09:47 +0100)]
hvm: Fix CR0 handling, which I broke in 15652.
Thanks to Eric Liu for spotting this.
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years agovmx: Simplify event-injection logic.
kfraser@localhost.localdomain [Tue, 31 Jul 2007 08:39:14 +0000 (09:39 +0100)]
vmx: Simplify event-injection logic.
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years ago[POWERPC][XEN] Use multiboot2 data structures when present.
Hollis Blanchard [Mon, 30 Jul 2007 22:10:45 +0000 (17:10 -0500)]
[POWERPC][XEN] Use multiboot2 data structures when present.
Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>
18 years ago[POWERPC][XEN] Deal with "shadow" -> "paging" fallout.
Hollis Blanchard [Mon, 30 Jul 2007 18:50:14 +0000 (13:50 -0500)]
[POWERPC][XEN] Deal with "shadow" -> "paging" fallout.
Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>
18 years ago[POWERPC][XEN] Fix build by adding proper guards to asm/numa.h
Hollis Blanchard [Mon, 30 Jul 2007 18:47:02 +0000 (13:47 -0500)]
[POWERPC][XEN] Fix build by adding proper guards to asm/numa.h
Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>
18 years ago[ACM] Some more fixes
kfraser@localhost.localdomain [Mon, 30 Jul 2007 15:03:16 +0000 (16:03 +0100)]
[ACM] Some more fixes

 - don't reload the policy if it has been loaded
 - don't always load the policy in the test suite when the policy is
   already loaded
 - skip tests 07 and 09 when ACM is not enabled and xm is not using the
   Xen-API
 - fix a problem when trying to remove an invalid label

Signed-off-by: Stefan Berger <stefanb@us.ibm.com>
18 years ago[IA64] Disable ACPI SRAT,SLIT table of dom0.
kfraser@localhost.localdomain [Mon, 30 Jul 2007 10:28:16 +0000 (11:28 +0100)]
[IA64] Disable ACPI SRAT,SLIT table of dom0.

  On some ia64 NUMA machine, we cannot boot dom0.
  This issue is caused by different infomation LSAPIC and SRAT.
  Xen-ia64 modify LSAPIC IDs of dom0, but it does not modify SRAT.
  So we decide disabling SRAT, SLIT of dom0 as first step of NUMA
  work.

Signed-off-by: Akio Takebe <takebe_akio@jp.fujitsu.com>
18 years agoAdd acpi_table_disable() into xen/drivers/acpi/tables.c
kfraser@localhost.localdomain [Mon, 30 Jul 2007 10:27:48 +0000 (11:27 +0100)]
Add acpi_table_disable() into xen/drivers/acpi/tables.c
This function can disable a ACPI table by updating the table header.

Signed-off-by: Akio Takebe <takebe_akio@jp.fujitsu.com>
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years agoMove generate_acpi_checksum() from ia64 code to common.
kfraser@localhost.localdomain [Mon, 30 Jul 2007 09:59:27 +0000 (10:59 +0100)]
Move generate_acpi_checksum() from ia64 code to common.
Signed-off-by: Akio Takebe <takebe_akio@jp.fujitsu.com>
18 years agohvm: Clean up CR0 handling.
kfraser@localhost.localdomain [Fri, 27 Jul 2007 08:43:07 +0000 (09:43 +0100)]
hvm: Clean up CR0 handling.

Upper 32 bits should #GP if set. Lower 32 bits should have reserved
bits silently cleared.

Check Intel VMX MSRs to check for compatibility with our CR0
requirements.

Signed-off-by: Eric E Liu <eric.e.liu@intel.com>
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years agoImplement Xen API method Console.set_other_config.
kfraser@localhost.localdomain [Fri, 27 Jul 2007 08:06:58 +0000 (09:06 +0100)]
Implement Xen API method Console.set_other_config.
Signed-off-by: Jim Fehlig <jfehlig@novell.com>
18 years agoImplement missing Xen API method Console.get_other_config.
kfraser@localhost.localdomain [Fri, 27 Jul 2007 08:06:30 +0000 (09:06 +0100)]
Implement missing Xen API method Console.get_other_config.
Signed-off-by: Jim Fehlig <jfehlig@novell.com>
18 years agoFix Xen API console methods that use undefined variables.
kfraser@localhost.localdomain [Fri, 27 Jul 2007 08:06:10 +0000 (09:06 +0100)]
Fix Xen API console methods that use undefined variables.
Signed-off-by: Jim Fehlig <jfehlig@novell.com>
18 years ago[Xend] Do network startup earlier.
kfraser@localhost.localdomain [Fri, 27 Jul 2007 08:04:35 +0000 (09:04 +0100)]
[Xend] Do network startup earlier.

Xend tries to rebuild the network configuration such as bridges and
vlan interfaces  (in those bridges) when starting. Unfortunately this
fails on the first startup since the network script has not run, yet,
and for example created peth0. When restarting xend it works due to
the peth0 being there then. This patch moves the initialization of the
network to an earlier time in the xend initialization process to
mitigate this problem.

Signed-off-by: Stefan Berger <stefanb@us.ibm.com>
18 years ago[XEND] Fix for creating VLANs using the Xen-API
kfraser@localhost.localdomain [Fri, 27 Jul 2007 08:03:35 +0000 (09:03 +0100)]
[XEND] Fix for creating VLANs using the Xen-API

Recent changes to the setup of the peth0 interface have resulted in a
change of its MAC address. Previously it seems to have had a MAC
address of 'fe:ff:ff:ff:ff:ff', but now it has the same MAC address as
eth0. As a consequence to this the Xen-API code to create VLANs
(PIF.create_VLAN("peth0",...)) does not work anymore, since peth0 can
not be identified according to this previous criteria of fake MAC
address. This patch fixes this issue by identifying it through the
prefix 'peth'.

Signed-off-by: Stefan Berger <stefanb@us.ibm.com>
18 years ago[ACM] Check a domain's authorization to run.
kfraser@localhost.localdomain [Fri, 27 Jul 2007 08:01:15 +0000 (09:01 +0100)]
[ACM] Check a domain's authorization to run.

A domain is only authorized to run if it has a superset of Simple Type
Enforcement Types in its VM label compared to that of Domain-0, which
itself may not have all STEs available in a policy. This patch adds a
check for this into Xend and the necessary code support into Xen.

Signed-off-by: Stefan Berger <stefanb@us.ibm.com>
18 years ago[HVM] Save/restore: don't leak shared-memory segments after HVM live-migrate.
Tim Deegan [Thu, 26 Jul 2007 11:00:32 +0000 (12:00 +0100)]
[HVM] Save/restore: don't leak shared-memory segments after HVM live-migrate.
Signed-off-by: Tim Deegan <Tim.Deegan@xensource.com>
18 years agoxenstored: Fix socket-based connection teardown. Reading zero bytes
kfraser@localhost.localdomain [Wed, 25 Jul 2007 08:51:14 +0000 (09:51 +0100)]
xenstored: Fix socket-based connection teardown. Reading zero bytes
*should* in fact cause the connection to be destroyed. Fix this with a
little extra code in the readfd() handler.
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years agoxenstore: Small cleanups and fixes.
kfraser@localhost.localdomain [Tue, 24 Jul 2007 17:28:48 +0000 (18:28 +0100)]
xenstore: Small cleanups and fixes.
 1. readfd/writefd account for EINTR/EAGAIN errno returns.
 2. Handle zero return from ->read() and ->write() handlers
    symmetrically.
 3. Fix some indentation issues (use hard tabs).
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years agoAdd domain name check and UUID check to 'xm new' command.
kfraser@localhost.localdomain [Tue, 24 Jul 2007 17:05:04 +0000 (18:05 +0100)]
Add domain name check and UUID check to 'xm new' command.

Add a domain name check and a UUID check to xm new command. The check
logic is as follows:

 - If the UUID is not specified
       - If a VM with same name exists
           => Update the config for that existing VM
       - Else no vm with same name exists
           => Define a brand new VM with auto-generated UUID
 - Else UUID is specified
       - If a VM with same UUID exists
             - If name is different
                   => Error
             - Else if name is same
                   => Update the config for that existing VM
       - Else no VM with same UUID exists
             - If name is different
                   => Define a branch new VM with that name
             - Else if name is same
                   => Error

Signed-off-by: Masaki Kanno <kanno.masaki@jp.fujitsu.com>
18 years ago[HVM] Live-migration fixups after 15639:c585f993385c
Tim Deegan [Tue, 24 Jul 2007 14:47:36 +0000 (15:47 +0100)]
[HVM] Live-migration fixups after 15639:c585f993385c
Use the same xenstore area for log-dirty commands as for save/continue,
and avoid a race condition by acking the save command after the save.
Signed-off-by: Tim Deegan <Tim.Deegan@xensource.com>
18 years agoMerge
Tim Deegan [Tue, 24 Jul 2007 13:53:06 +0000 (14:53 +0100)]
Merge

18 years ago[HVM] Control qemu's state-save via xenstore, instead of SIGUSR1
Tim Deegan [Tue, 24 Jul 2007 13:52:16 +0000 (14:52 +0100)]
[HVM] Control qemu's state-save via xenstore, instead of SIGUSR1
This lets us verify that qemu has indeed stopped processing before
we start saving guest memory.  Also allow qemu to continue processing
after the save has happened, instead of exiting immediately.
Signed-off-by: Tim Deegan <Tim.Deegan@xensource.com>
18 years agoxenstored: Fairly round-robin schedule work across all connections.
kfraser@localhost.localdomain [Tue, 24 Jul 2007 13:50:50 +0000 (14:50 +0100)]
xenstored: Fairly round-robin schedule work across all connections.
Avoids total starvation under some workloads.
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years agoxenstored: Guarantee to fire @releaseDomain watch when a domain is destructed.
kfraser@localhost.localdomain [Tue, 24 Jul 2007 13:50:05 +0000 (14:50 +0100)]
xenstored: Guarantee to fire @releaseDomain watch when a domain is destructed.
Previously this would be missed on some bail paths within xenstored
which would talloc_free() the connection.
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years ago[HVM] Shadow: release shadow lock during emulation path
Tim Deegan [Tue, 24 Jul 2007 10:10:08 +0000 (11:10 +0100)]
[HVM] Shadow: release shadow lock during emulation path
and retake it only for the write-back at the end.
Signed-off-by: Tim Deegan <Tim.Deegan@xensource.com>
18 years ago[PVFB] Fix shift key for graphical vnc display
kfraser@localhost.localdomain [Mon, 23 Jul 2007 09:03:17 +0000 (10:03 +0100)]
[PVFB] Fix shift key for graphical vnc display

There is a problem in the input of the key in the VNC connection on
the PV domain. When client's keyboard is not the same as the kind of
the keyboard of PVFB and GuestOS, it is not possible to input it correctly.

This patch handled the state of shift from the set keymap. When
client's keyboard is not same as the kind of PVFB/GuestOS, it is
possible to input it correctly. It was confirmed to input it correctly
mutually with this patch between en-us and ja.

Signed-off-by: Takanori Kasai <kasai.takanori@jp.fujitsu.com>
18 years ago[IOEMU] Fix shift key for graphical vnc display
kfraser@localhost.localdomain [Mon, 23 Jul 2007 09:02:37 +0000 (10:02 +0100)]
[IOEMU] Fix shift key for graphical vnc display

There is a problem in the input of the key in the VNC connection on
the HVM domain. When client's keyboard is not the same as the kind of
the keyboard of qemu-dm and GuestOS, it is not possible to input it
correctly.

   VNC client     qemu-dm & GuestOS
--------------+-----------------------
    ja             en-us               ==> NG
    en-us          en-us               ==> OK

Originally, the same keysym-code between client and qemu-dm is
transmitted. However, even if it is the same character, the state of
shift is different according to the kind of keyboard.

ex.
   "=" charactor
---------------------
en-us :  "="
ja    :  shift + "-"

Therefore, it is necessary to handle the state of the shift by setting
qemu-dm and GuestOS. There is information on whether shift is
necessary for each key for the keymap of qemu-dm.

ex.
 VNC client        : ja
 qemu-dm & GuestOS : en-us
 input key         : "="
 event client to qemu-dm :
  shift(push) >> "="(push) >> "="(release) >> shift(release)
 event qemu-dm to guest :
  shift(push) >> shift(release) >> "="(push) >> "="(release) >>
 shift(push) >>
shift(release)

This patch handled the state of shift from the set keymap.
When client's keyboard is not same as the kind of qemu-dm/GuestOS,
it is possible to input it correctly.
It was confirmed to input it correctly mutually with this patch
between en-us and ja.

Signed-off-by: Takanori Kasai <kasai.takanori@jp.fujitsu.com>
18 years ago[ACM] Check boundary conditions of passed parameters before accessing
kfraser@localhost.localdomain [Mon, 23 Jul 2007 08:56:49 +0000 (09:56 +0100)]
[ACM] Check boundary conditions of passed parameters before accessing
an array with them.
Signed-off-by: Stefan Berger <stefanb@us.ibm.com>
18 years ago[ACM] Coding style cleanups.
kfraser@localhost.localdomain [Mon, 23 Jul 2007 08:56:00 +0000 (09:56 +0100)]
[ACM] Coding style cleanups.
Signed-off-by: Stefan Berger <stefanb@us.ibm.com>
18 years ago[ACM] Allow the loadpolicy operation once
kfraser@localhost.localdomain [Mon, 23 Jul 2007 08:45:23 +0000 (09:45 +0100)]
[ACM] Allow the loadpolicy operation once

This patch allows the loadpolicy operation to only happen once, then
require an update until the default policy has been installed (again).

Signed-off-by: Stefan Berger <stefanb@us.ibm.com>
18 years agoFix IA64 build
Tim Deegan [Fri, 20 Jul 2007 13:10:40 +0000 (14:10 +0100)]
Fix IA64 build
Signed-off-by: Alex Williamson <alex.williamson@hp.com>
18 years ago[HVM] HAP: mark domains as HAP-enabled, not shadow-enabled.
Tim Deegan [Fri, 20 Jul 2007 11:00:25 +0000 (12:00 +0100)]
[HVM] HAP: mark domains as HAP-enabled, not shadow-enabled.
Thanks to Wei Huang.
Signed-off-by: TIm Deegan <Tim.Deegan@xensource.com>
18 years ago[XEN] replace shadow_* with paging_* in common code
Tim Deegan [Fri, 20 Jul 2007 10:36:36 +0000 (11:36 +0100)]
[XEN] replace shadow_* with paging_* in common code
Signed-off-by: Tim Deegan <Tim.Deegan@xensource.com>
18 years agoXen PM: Clean up printk usage.
kfraser@localhost.localdomain [Fri, 20 Jul 2007 09:47:11 +0000 (10:47 +0100)]
Xen PM: Clean up printk usage.
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years agoXen PM: Remove useless __sync_lazy_execstate().
kfraser@localhost.localdomain [Fri, 20 Jul 2007 09:41:31 +0000 (10:41 +0100)]
Xen PM: Remove useless __sync_lazy_execstate().
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years agoXen PM: Enter all ACPI S states on boot CPU.
kfraser@localhost.localdomain [Fri, 20 Jul 2007 09:38:51 +0000 (10:38 +0100)]
Xen PM: Enter all ACPI S states on boot CPU.
Signed-off-by: Kevin Tian <kevin.tian@intel.com>
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years agoFix hypercall migration. schedule_tail() is not required
kfraser@localhost.localdomain [Fri, 20 Jul 2007 08:38:27 +0000 (09:38 +0100)]
Fix hypercall migration. schedule_tail() is not required
after set affinity, or else continue stub runs on current
cpu immediately before migrating is checked when switching
out.

Signed-off-by Kevin Tian <kevin.tian@intel.com>
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years ago[MAN] Explanation of recent extensions to xm security subcommands
kfraser@localhost.localdomain [Thu, 19 Jul 2007 16:18:20 +0000 (17:18 +0100)]
[MAN] Explanation of recent extensions to xm security subcommands
Signed-off-by: Stefan Berger <stefanb@usa.ibm.com>
18 years agoImprove xm vcpu-list command for inactive managed domains.
kfraser@localhost.localdomain [Thu, 19 Jul 2007 16:17:25 +0000 (17:17 +0100)]
Improve xm vcpu-list command for inactive managed domains.
Signed-off-by: Masaki Kanno <kanno.masaki@jp.fujitsu.com>
18 years agoxend: Give back memory to dom0 after a failed auto-balloon attempt.
kfraser@localhost.localdomain [Thu, 19 Jul 2007 16:15:49 +0000 (17:15 +0100)]
xend: Give back memory to dom0 after a failed auto-balloon attempt.
Signed-off-by: Frank van der Linden <frank.vanderlinden@sun.com>
18 years agoFix 64-bit build.
kfraser@localhost.localdomain [Thu, 19 Jul 2007 16:09:47 +0000 (17:09 +0100)]
Fix 64-bit build.
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years agoUpdate xen-api docs.
kfraser@localhost.localdomain [Thu, 19 Jul 2007 16:00:31 +0000 (17:00 +0100)]
Update xen-api docs.
Signed-off-by: Stefan Berger <stefanb@us.ibm.com>
18 years ago[xend / libxen] Add support for labeling of virtual network interfaces.
kfraser@localhost.localdomain [Thu, 19 Jul 2007 15:59:48 +0000 (16:59 +0100)]
[xend / libxen] Add support for labeling of virtual network interfaces.

This patch adds labeling of virtual network interfaces to xend and
makes this manageable through the Xen-API.  It's a feature that is
only usable if ACM is enabled in Xen and xend is used through the
xen-api. A labeled virtual network interface will be plugged into a
bridge where other domains with the same-labeled network interface are
connected to, so that only same-colored domains can communicate with
each other. The bridge should be connected to the outside world using
VLAN for isolation, extending the isolation beyond the local machine.
If a virtual machine is labeled with a VM label that only has one
Simple Type Enforcement Type then it is not necessary to label the
virtual network interface, but the color of the network interface is
determined from the VM's label. If, however, a virtual machine is
labeled with a VM label that has multiple Simple Type Enforcement
Types, then the explicit labeling of each virtual network interface is
required. To specify the label of a network interface, the vif line in
the VM's configuration file has been extended with parameters similar
use for specifying the label of the VM:

vif = ['policy=<policy name>,label=<resource label>']

This labels the VIF of the virtual machine for usage under the policy
'policy name' and labels it with the label 'resource label'.

Signed-off-by: Stefan Berger <stefanb@us.ibm.com>
18 years agoxen: Clean up code style.
kfraser@localhost.localdomain [Thu, 19 Jul 2007 12:39:10 +0000 (13:39 +0100)]
xen: Clean up code style.
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years agoSMP support for Xen PM.
kfraser@localhost.localdomain [Thu, 19 Jul 2007 12:23:33 +0000 (13:23 +0100)]
SMP support for Xen PM.

Add SMP sleep support to Xen. One new utility is created
to allow vcpu continue previous running after migration
to new processor.

Signed-off-by: Kevin Tian <kevin.tian@intel.com>
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years ago[host s3] Retrieve necessary sleep information from plain-text ACPI
kfraser@localhost.localdomain [Thu, 19 Jul 2007 11:53:32 +0000 (12:53 +0100)]
[host s3] Retrieve necessary sleep information from plain-text ACPI
tables (FADT/FACS), and keep one hypercall remained for sleep
notification.

Signed-off-by: Ke Yu <ke.yu@intel.com>
Signed-off-by: Kevin Tian <kevin.tian@intel.com>
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years agovmx: Save host MSR state for SYSCALLs on all CPUs.
kfraser@localhost.localdomain [Thu, 19 Jul 2007 09:59:05 +0000 (10:59 +0100)]
vmx: Save host MSR state for SYSCALLs on all CPUs.
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years agoBuild updates for building upstream trees.
Ian Campbell [Thu, 19 Jul 2007 09:26:00 +0000 (10:26 +0100)]
Build updates for building upstream trees.

* Merge the guts of the various upstream branch buildconfigs into a
  single mk.linux-2.6 which they include.
* Add support for the -git snapshot branch.
* Update .hgignore to ignore all these new trees.
* Update buildconfigs/enable-xen-config to make the upstream defconfig
  compile without questions.

Signed-off-by: Ian Campbell <ian.campbell@xensource.com>
18 years agoMerge
Tim Deegan [Wed, 18 Jul 2007 12:56:21 +0000 (13:56 +0100)]
Merge

18 years ago[HVM] HAP: tidy up page allocation/tracking and monitor-table building.
Tim Deegan [Wed, 18 Jul 2007 12:56:00 +0000 (13:56 +0100)]
[HVM] HAP: tidy up page allocation/tracking and monitor-table building.
Signed-off-by: Tim Deegan <Tim.Deegan@xensource.com>
18 years agohvm: Only complain about short-period periodic tickers.
kfraser@localhost.localdomain [Wed, 18 Jul 2007 10:00:32 +0000 (11:00 +0100)]
hvm: Only complain about short-period periodic tickers.
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years ago[xend][xen-api] Save new memory values in domain config when calling
kfraser@localhost.localdomain [Wed, 18 Jul 2007 09:10:04 +0000 (10:10 +0100)]
[xend][xen-api] Save new memory values in domain config when calling
VM.set_memory_[dynamic|static]_[min|max] via Xen API.
Signed-off-by: Jim Fehlig <jfehlig@novell.com>
18 years ago[Xend] More security-related fixes
kfraser@localhost.localdomain [Wed, 18 Jul 2007 09:09:06 +0000 (10:09 +0100)]
[Xend] More security-related fixes

This patch provides some more fixes related to the recent
security-related extensions to xend.

Signed-off-by: Stefan Berger <stefanb@us.ibm.com>
18 years ago[Xm-Test] Additional tests for the xm-test suite
kfraser@localhost.localdomain [Wed, 18 Jul 2007 09:08:37 +0000 (10:08 +0100)]
[Xm-Test] Additional tests for the xm-test suite

This adds a couple of test cases exercising the new policy management
functionality to the security tests.

Signed-off-by: Stefan Berger <stefanb@us.ibm.com>
18 years ago[HVM] More MCA MSRs to ignore
Tim Deegan [Wed, 18 Jul 2007 09:04:46 +0000 (10:04 +0100)]
[HVM] More MCA MSRs to ignore
Signed-off-by: Tim Deegan <Tim.Deegan@xensource.com>
18 years ago[HVM] Hide contents of IA32_MCG_STATUS MSR from the guest.
Tim Deegan [Wed, 18 Jul 2007 08:06:41 +0000 (09:06 +0100)]
[HVM] Hide contents of IA32_MCG_STATUS MSR from the guest.
It can't clear the MCE, and it probably can't understand it anyway.
Signed-off-by: Tim Deegan <Tim.Deegan@xensource.com>
18 years agohap: Merge the guest-walking functions into one.
kfraser@localhost.localdomain [Tue, 17 Jul 2007 13:09:45 +0000 (14:09 +0100)]
hap: Merge the guest-walking functions into one.
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years agoImprove xm uptime command for inactive managed domains
kfraser@localhost.localdomain [Tue, 17 Jul 2007 09:36:33 +0000 (10:36 +0100)]
Improve xm uptime command for inactive managed domains
Signed-off-by: Masaki Kanno <kanno.masaki@jp.fujitsu.com>
18 years ago[VTPM] Make vTPM hotplug scripts use the UUID of the vTPM instance
kfraser@localhost.localdomain [Tue, 17 Jul 2007 09:33:26 +0000 (10:33 +0100)]
[VTPM] Make vTPM hotplug scripts use the UUID of the vTPM instance

Make the vTPM hotplug script use the UUID of the vTPM instance for
associating it with its instance number rather than the name of the
domain.
Also change the cleanup in the xm test to use the atexit mechanism
rather than explicitly calling a function to clean up.

Signed-off-by: Stefan Berger <stefanb@us.ibm.com>
18 years agoMerge with PPC Xen tree.
kfraser@localhost.localdomain [Tue, 17 Jul 2007 09:20:21 +0000 (10:20 +0100)]
Merge with PPC Xen tree.

18 years agohap: Fix for coding style.
kfraser@localhost.localdomain [Tue, 17 Jul 2007 08:22:41 +0000 (09:22 +0100)]
hap: Fix for coding style.
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years ago[POWERPC][XEN] Use top-level CFLAGS.
Hollis Blanchard [Mon, 16 Jul 2007 19:20:16 +0000 (14:20 -0500)]
[POWERPC][XEN] Use top-level CFLAGS.
- Append to CFLAGS instead of replacing it.
- Fix newly-exposed mixed statement/declaration warnings.

18 years ago[HVM] Qemu rtl8139: correct rx CRC calculation
Tim Deegan [Mon, 16 Jul 2007 13:28:54 +0000 (14:28 +0100)]
[HVM] Qemu rtl8139: correct rx CRC calculation
Signed-off-by: Tim Deegan <Tim.Deegan@xensource.com>
18 years ago[HVM] Qemu rtl8139: correct ring-buffer wrapping.
Tim Deegan [Mon, 16 Jul 2007 13:27:12 +0000 (14:27 +0100)]
[HVM] Qemu rtl8139: correct ring-buffer wrapping.
The rx DMA should never overrun when it hits the end of a 64k buffer
Signed-off-by: Tim Deegan <Tim.Deegan@xensource.com>
18 years ago[XEN] HAP: need to intercept CR0 even when using NPT
Tim Deegan [Mon, 16 Jul 2007 09:36:52 +0000 (10:36 +0100)]
[XEN] HAP: need to intercept CR0 even when using NPT
because we tinker with CR0.TS.  Thanks to Wei Huang for pointing this out.
Signed-off-by: Tim Deegan <Tim.Deegan@xensource.com>
18 years agoRemove unused function reserve_boot_memory() and alloc_boot_low_page().
Keir Fraser [Sat, 14 Jul 2007 09:26:02 +0000 (10:26 +0100)]
Remove unused function reserve_boot_memory() and alloc_boot_low_page().
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years agoxen: Fix 32-bit boot.
Keir Fraser [Sat, 14 Jul 2007 07:19:17 +0000 (08:19 +0100)]
xen: Fix 32-bit boot.
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years ago[POWERPC][XEN] Enable in-guest performance monitoring.
Hollis Blanchard [Fri, 13 Jul 2007 22:28:15 +0000 (17:28 -0500)]
[POWERPC][XEN] Enable in-guest performance monitoring.
- Lazily save and restore the performance monitor counters when switching
  domains.
- Control with the H_PERFMON PAPR hypercall.
- Ignore guest perfmon exceptions that land in Xen.
Signed-off-by: Christian Ehrhardt <ehrhardt@linux.vnet.ibm.com>
Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>
18 years ago[XEN][POWERPC] Fix build after NUMA nodemask changes.
Hollis Blanchard [Fri, 13 Jul 2007 20:23:24 +0000 (15:23 -0500)]
[XEN][POWERPC] Fix build after NUMA nodemask changes.
Signed-off-by: Jerone Young <jyoung5@us.ibm.com>
Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>
18 years agoRemove hard tabs.
kfraser@localhost.localdomain [Fri, 13 Jul 2007 16:08:05 +0000 (17:08 +0100)]
Remove hard tabs.
Signed-off-by: Keir Fraser <keir@xensource.com>
18 years agox86/32: Limit bootstrap 1:1 mapping to 1GB.
kfraser@localhost.localdomain [Fri, 13 Jul 2007 14:42:27 +0000 (15:42 +0100)]
x86/32: Limit bootstrap 1:1 mapping to 1GB.
Signed-off-by: Keir Fraser <keir@xensource.com>