xen.git
19 years agoAdd missing spin_unlock_irq() at xen/arch/x86/irq.c
kaf24@firebug.cl.cam.ac.uk [Fri, 5 May 2006 12:41:35 +0000 (13:41 +0100)]
Add missing spin_unlock_irq() at xen/arch/x86/irq.c

Changeset 9889:42a8e3101c6c reorganized the code on this file, and
missed this spin_unlock_irq().  Without this patch, my machine hangs
completely during boot.  With this, it works.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Chris Wright <chrisw@sous-sol.org>
19 years agoDon't try to write to pmd pages, there is no writeable-pmd support.
cl349@firebug.cl.cam.ac.uk [Thu, 4 May 2006 23:27:10 +0000 (00:27 +0100)]
Don't try to write to pmd pages, there is no writeable-pmd support.

Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
19 years agoSVM patch for 64bit hv, to reset the ss, es, ds host selectors to NULL
kaf24@firebug.cl.cam.ac.uk [Thu, 4 May 2006 20:24:39 +0000 (21:24 +0100)]
SVM patch for 64bit hv, to reset the ss, es, ds host selectors to NULL
during a context switch to the SVM domain's vcpu. This patch also
initializes the tlb_control to 1 for the initial do_launch().
Signed-off-by: Tom Woller <thomas.woller@amd.com>
19 years agoLinux: upgrade patches to 2.6.16.13.
cl349@firebug.cl.cam.ac.uk [Thu, 4 May 2006 16:44:14 +0000 (17:44 +0100)]
Linux: upgrade patches to 2.6.16.13.

Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
--HG--
rename : patches/linux-2.6.16/device_bind.patch => patches/linux-2.6.16.13/device_bind.patch
rename : patches/linux-2.6.16/i386-mach-io-check-nmi.patch => patches/linux-2.6.16.13/i386-mach-io-check-nmi.patch
rename : patches/linux-2.6.16/net-csum.patch => patches/linux-2.6.16.13/net-csum.patch
rename : patches/linux-2.6.16/pmd-shared.patch => patches/linux-2.6.16.13/pmd-shared.patch
rename : patches/linux-2.6.16/rename-TSS_sysenter_esp0-SYSENTER_stack_esp0.patch => patches/linux-2.6.16.13/rename-TSS_sysenter_esp0-SYSENTER_stack_esp0.patch
rename : patches/linux-2.6.16/smp-alts.patch => patches/linux-2.6.16.13/smp-alts.patch
rename : patches/linux-2.6.16/x86-increase-interrupt-vector-range.patch => patches/linux-2.6.16.13/x86-increase-interrupt-vector-range.patch
rename : patches/linux-2.6.16/xenoprof-generic.patch => patches/linux-2.6.16.13/xenoprof-generic.patch

19 years agoLinux: upgrade to 2.6.16.13.
cl349@firebug.cl.cam.ac.uk [Thu, 4 May 2006 16:38:25 +0000 (17:38 +0100)]
Linux: upgrade to 2.6.16.13.

Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
19 years agoRemove unused LINUX_SRCS.
cl349@firebug.cl.cam.ac.uk [Thu, 4 May 2006 15:49:58 +0000 (16:49 +0100)]
Remove unused LINUX_SRCS.

Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
19 years agoAdd a HOWTO to help writing tests, includes domains with networking.
stekloff@dyn9047022152.beaverton.ibm.com [Thu, 4 May 2006 13:22:38 +0000 (14:22 +0100)]
Add a HOWTO to help writing tests, includes domains with networking.

Signed-off-by: Daniel Stekloff <dsteklof@us.ibm.com>
19 years agoAdd new networking infrastructure to Xm-Test. The goal is to make
stekloff@dyn9047022152.beaverton.ibm.com [Thu, 4 May 2006 13:22:29 +0000 (14:22 +0100)]
Add new networking infrastructure to Xm-Test. The goal is to make
creating domains with networking very easy. This patch:

1) Adds new XenDevice class, with the XenNetDevice subclass. These
   classes represent devices for xm-test and are tied to XenDomains.
   This can eventually be used for block devices as well. Currently,
   devices must be added to domains prior to starting the domain. The
   attach and detach needs to be handled.

2) Adds a new NetConfig class to handle configuring the network
   environment in which the tests run. This patch only handles ranges
   of IPs in a bridged environment. DHCP needs to be added as well as
   handling NAT and routed environments.

3) Modifies XenDomain class to handle XenDevices.

4) Adds new configuration options for defining a range of IPs, their
   network address, and their netmask.

5) Removes the old Network.py and Network class.

6) Modifies the existing tests to use the new infrastructure.

7) Adds some documentation to help creating domains.

Signed-off-by: Daniel Stekloff <dsteklof@us.ibm.com>
19 years agoSecond step toward a new network infrastructure, move XmConsole to be
emellor@leeni.uk.xensource.com [Thu, 4 May 2006 13:22:17 +0000 (14:22 +0100)]
Second step toward a new network infrastructure, move XmConsole to be
in XenDomain. The devices we will add, like network, will need to run
console commands when a domain is started. So I have made console be
a part of domain object, which it's tied to anyway.

Signed-off-by: Daniel Stekloff <dsteklof@us.ibm.com>
19 years agoFirst step to network infrastructure, add a type to XenDomain to identify
stekloff@dyn9047022152.beaverton.ibm.com [Thu, 4 May 2006 13:20:11 +0000 (14:20 +0100)]
First step to network infrastructure, add a type to XenDomain to identify
it as HVM or PV rather than using the config time option. The forthcoming
XenNetDevice will need to check. This is also the first step toward running
PV and HVM domains at the same time.

Signed-off-by: Daniel Stekloff <dsteklof@us.ibm.com>
19 years agoMerged.
emellor@leeni.uk.xensource.com [Thu, 4 May 2006 13:19:19 +0000 (14:19 +0100)]
Merged.

19 years agoChange Mini-OS so that it uses stddef.h to define size_t and
kaf24@firebug.cl.cam.ac.uk [Thu, 4 May 2006 10:24:19 +0000 (11:24 +0100)]
Change Mini-OS so that it uses stddef.h to define size_t and
NULL. This problem fixes errors that occur when linking Mini-OS with
ANSI standard code that uses stddef.h.
Signed-off-by: Keir Fraser <keir@xensource.com>
19 years agoThis patch
kaf24@firebug.cl.cam.ac.uk [Thu, 4 May 2006 10:22:19 +0000 (11:22 +0100)]
This patch
- moves the TPM frontend completely into the char/tpm directory where it
is a plug-in to the generic TPM driver
- removes a now obsolete include file
- adapts part of the documentation
- fixes some locking problems where copy_to/from_user was called with
IRQs blocked

Signed-off-by: Stefan Berger <stefanb@us.ibm.com>
19 years agoFix fault handler for both 32 and 64 bit architecture. Clean up unused code.
kaf24@firebug.cl.cam.ac.uk [Thu, 4 May 2006 10:19:27 +0000 (11:19 +0100)]
Fix fault handler for both 32 and 64 bit architecture. Clean up unused code.
Signed-off-by: Grzegorz Milos <gm281@cam.ac.uk>
Signed-off-by: Aravindh Puthiyaparambil
<aravindh.puthiyaparambil@unisys.com>

19 years agoSVM patch to cleanup the host save area allocation and deallocation,
kaf24@firebug.cl.cam.ac.uk [Thu, 4 May 2006 10:14:45 +0000 (11:14 +0100)]
SVM patch to cleanup the host save area allocation and deallocation,
including removing  memory leaks concerning these areas.  Also fixes
problem where the HSA MSR was not initialized properly for cores>0.

Signed-off-by: Tom Woller <thomas.woller@amd.com>
19 years agoFix xenbus userspace device transaction tracking.
cl349@firebug.cl.cam.ac.uk [Thu, 4 May 2006 09:25:27 +0000 (10:25 +0100)]
Fix xenbus userspace device transaction tracking.
If a transaction end command fails, the semaphore which keeps track
of whether we're in a transaction or not was not getting updated.

Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
19 years agoFix mismerge of vmx vm86 patch: 9915:e1409c2ace46afd67b54de1f5c0b2b6d3721ca3b
kaf24@firebug.cl.cam.ac.uk [Wed, 3 May 2006 12:33:01 +0000 (13:33 +0100)]
Fix mismerge of vmx vm86 patch: 9915:e1409c2ace46afd67b54de1f5c0b2b6d3721ca3b
Signed-off-by: Keir Fraser <keir@xensource.com>
19 years agoInstead of knowing explicitly about the pygrub entry option, allow
kaf24@firebug.cl.cam.ac.uk [Wed, 3 May 2006 10:02:30 +0000 (11:02 +0100)]
Instead of knowing explicitly about the pygrub entry option, allow
passing arbitrary options to the bootloader and deprecate the bootentry
option.

Signed-off-by: Jeremy Katz <katzj@redhat.com>
19 years agoIntroduce some finer-grained locking to prevent calling
kaf24@firebug.cl.cam.ac.uk [Wed, 3 May 2006 10:00:57 +0000 (11:00 +0100)]
Introduce some finer-grained locking to prevent calling
copy_from/to_user while preventing interrupts and fix allocation of
memory from within a task and other issues.

Signed-off-by: Stefan Berger <stefanb@us.ibm.com>
19 years agoClean up xen-mkbuildtree-pre for ia64 xenlinux.
kaf24@firebug.cl.cam.ac.uk [Wed, 3 May 2006 09:59:54 +0000 (10:59 +0100)]
Clean up xen-mkbuildtree-pre for ia64 xenlinux.
Signed-off-by: Dan Magenheimer <dan.magenheimer@hp.com>
19 years agoUse native value for SPLIT_PTLOCK_CPUS except on xen/x86.
kaf24@firebug.cl.cam.ac.uk [Wed, 3 May 2006 09:58:43 +0000 (10:58 +0100)]
Use native value for SPLIT_PTLOCK_CPUS except on xen/x86.
Signed-off-by: Keir Fraser <keir@xensource.com>
19 years agoEnsure segment bases are consistent with their
kaf24@firebug.cl.cam.ac.uk [Wed, 3 May 2006 09:56:19 +0000 (10:56 +0100)]
Ensure segment bases are consistent with their
selectors for VMX guests in VM86 mode.

Signed-off-by: David Lively <dlively@virtualiron.com>
19 years agoTurn page-structure layout assertions into BUILD_BUG_ON().
kaf24@firebug.cl.cam.ac.uk [Wed, 3 May 2006 06:44:20 +0000 (07:44 +0100)]
Turn page-structure layout assertions into BUILD_BUG_ON().
Signed-off-by: Keir Fraser <keir@xensource.com>
19 years agoXm-tests 01_memset_basic_pos.test and 03_memset_random_pos.test start
emellor@leeni.uk.xensource.com [Tue, 2 May 2006 17:17:59 +0000 (18:17 +0100)]
Xm-tests 01_memset_basic_pos.test and 03_memset_random_pos.test start
domUs with 64MB of memory,
and assume that all of that memory is available in the domU. They then
perform various xm memset tests
based on the assumed starting conditions.

When netback & blkback are comiled into the domU kernel, they eat about
5MB. This mismatch will cause
01_memset_basic_pos.test to fail every time. 03_memset_random_pos.test
will probably fail, but can
pass with some lucky numbers from the random number generator.

This patch changes both tests to read the starting memory allocation from
the balloon driver in the domU.

Signed-off-by: Jim Dykman <dykman@us.ibm.com>
19 years agoBalloon driver should hijack the ->lru list field rather than
kaf24@firebug.cl.cam.ac.uk [Tue, 2 May 2006 16:23:21 +0000 (17:23 +0100)]
Balloon driver should hijack the ->lru list field rather than
adding another list field to every page structure.
Signed-off-by: Keir Fraser <keir@xensource.com>
19 years agoRemove i386/x8664 differences in vmx reg store/load routines.
kaf24@firebug.cl.cam.ac.uk [Tue, 2 May 2006 16:17:15 +0000 (17:17 +0100)]
Remove i386/x8664 differences in vmx reg store/load routines.
Signed-off-by: Keir Fraser <keir@xensource.com>
19 years agoVirtual net drivers advertise multicast capabilities.
kaf24@firebug.cl.cam.ac.uk [Tue, 2 May 2006 14:46:15 +0000 (15:46 +0100)]
Virtual net drivers advertise multicast capabilities.
Signed-off-by: Keir Fraser <keir@xensource.com>
19 years agoThis patch changes to format in which EIP is printed to be consistent on 64-bit regar...
kaf24@firebug.cl.cam.ac.uk [Tue, 2 May 2006 14:36:07 +0000 (15:36 +0100)]
This patch changes to format in which EIP is printed to be consistent on 64-bit regardless of whether the addresses
leading bits are set or clear. It additionally changes the formatting so that trailing spaces are avoided, and it
suppresses the printing of empty records. It also prevents combining records with identical EIP but from different
domains.

It further changes the type of some variables from plain int to unsigned int, as that is yielding more efficient code
on x86-64 (signed 32-bit array indices require explicit sign extension, whereas in most cases an extra copy can be
avoided when the index type is unsigned, since all 32-bit operations already zero-extend their results).

Signed-off-by: Jan Beulich <jbeulich@novell.com>
19 years agoThis patch conditionalizes some output from perfc_printall(), thus making relevant...
kaf24@firebug.cl.cam.ac.uk [Tue, 2 May 2006 14:33:47 +0000 (15:33 +0100)]
This patch conditionalizes some output from perfc_printall(), thus making relevant information more compact and easier
legible. It additionally changes the formatting so that trailing spaces are avoided.

Also changing the type of some variables from plain int to unsigned int, as that is yielding more efficient code on
x86-64 (signed 32-bit array indices require explicit sign extension, whereas in most cases an extra copy can be avoided
when the index type is unsigned, since all 32-bit operations already zero-extend their results).

Signed-off-by: Jan Beulich <jbeulich@novell.com>
19 years agoFix perfc array range on reset.
kaf24@firebug.cl.cam.ac.uk [Tue, 2 May 2006 14:21:19 +0000 (15:21 +0100)]
Fix perfc array range on reset.

Signed-off-by: Jan Beulich <jbeulich@novell.com>
19 years agoMove Linux-specific privcmd code into private libxc implementations.
kaf24@firebug.cl.cam.ac.uk [Tue, 2 May 2006 14:12:06 +0000 (15:12 +0100)]
Move Linux-specific privcmd code into private libxc implementations.
Make header path for kernel's privcmd/evtchn headers generic.
Remove pointless xi_*() interface that was using private libxc
interfaces.

Signed-off-by: John Levon <john.levon@sun.com>
19 years agoRemove obsolete privcmd_blkmsg struct.
kaf24@firebug.cl.cam.ac.uk [Tue, 2 May 2006 08:18:55 +0000 (09:18 +0100)]
Remove obsolete privcmd_blkmsg struct.
Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>
19 years agoFix ia64 build. do_physdev_op* declared incorrectly
kaf24@firebug.cl.cam.ac.uk [Tue, 2 May 2006 08:16:02 +0000 (09:16 +0100)]
Fix ia64 build. do_physdev_op* declared incorrectly
Signed-off-by: Dan Magenheimer <dan.magenheimer@hp.com>
19 years agoFix x86/64 version of Mini-OS. It encompasses the following:
kaf24@firebug.cl.cam.ac.uk [Tue, 2 May 2006 08:12:39 +0000 (09:12 +0100)]
Fix x86/64 version of Mini-OS. It encompasses the following:
a) 64-bit switch_to scheduler macro (by Aravindh Puthiyaparambil)
b) implements 64-bit hypervisor_callback
c) fixes thread creation issues (thread_starter used to perform
initialisation)

Signed-off-by: Grzegorz Milos <gm281@cam.ac.uk>
19 years agoClean up compiler.h to define a few more things and update
kaf24@firebug.cl.cam.ac.uk [Mon, 1 May 2006 16:44:51 +0000 (17:44 +0100)]
Clean up compiler.h to define a few more things and update
for gcc4.

Signed-off-by: Keir Fraser <keir@xensource.com>
19 years agoA few functions should be static in shutdown.c.
kaf24@firebug.cl.cam.ac.uk [Mon, 1 May 2006 16:41:02 +0000 (17:41 +0100)]
A few functions should be static in shutdown.c.

Signed-off-by: Keir Fraser <keir@xensource.com>
19 years agoInclude string.h in the hypercall header file to ensure that memcpy() is defined.
kaf24@firebug.cl.cam.ac.uk [Mon, 1 May 2006 16:02:43 +0000 (17:02 +0100)]
Include string.h in the hypercall header file to ensure that memcpy() is defined.

Signed-off-by: Keir Fraser <keir@xensource.com>
19 years agoThe Xen netif drivers won't complile if NET is absent.
kaf24@firebug.cl.cam.ac.uk [Mon, 1 May 2006 14:30:53 +0000 (15:30 +0100)]
The Xen netif drivers won't complile if NET is absent.

Signed-Off-By: Horms <horms@verge.net.au>
19 years agoCompilation fix on ia64. Include linux/vmalloc.h for struct vm_area.
kaf24@firebug.cl.cam.ac.uk [Mon, 1 May 2006 14:29:38 +0000 (15:29 +0100)]
Compilation fix on ia64. Include linux/vmalloc.h for struct vm_area.

Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
19 years agoCompilation fix tpmback.c on ia64. Use pfn_to_mfn instead of
kaf24@firebug.cl.cam.ac.uk [Mon, 1 May 2006 14:29:08 +0000 (15:29 +0100)]
Compilation fix tpmback.c on ia64. Use pfn_to_mfn instead of
phys_to_machine_mapping.

Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
19 years agoChange semantics of grant transfers for vp guests so that the
kaf24@firebug.cl.cam.ac.uk [Mon, 1 May 2006 14:28:01 +0000 (15:28 +0100)]
Change semantics of grant transfers for vp guests so that the
operation automatically gets you a fresh page at the same
pseudo-physical address as Keir suggested.

Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
19 years agoDebugging Xend with XEND_DAEMONIZE set to False doesn't work as
kaf24@firebug.cl.cam.ac.uk [Sun, 30 Apr 2006 08:52:59 +0000 (09:52 +0100)]
Debugging Xend with XEND_DAEMONIZE set to False doesn't work as
expected.  Xend actually creates two children and the XEND_DAEMONIZE
flag only prevents one of the children from being created.  This
patch causes none of the children to be created.  With this patch,
Xend functionality is performed in the process executed from the
command line (i.e., /usr/sbin/xend).  This patch makes it possible to
debug Xend with pdb.

Signed-off-by: Randy Thelen <rthelen@netapp.com>
19 years agoAllow architectures to define their own privcmd_mmap() by
kaf24@firebug.cl.cam.ac.uk [Sun, 30 Apr 2006 08:52:13 +0000 (09:52 +0100)]
Allow architectures to define their own privcmd_mmap() by
introducing HAVE_ARCH_PRIVCMD_MMAP.

Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
19 years agoIntroduce XEN_IA64_DOM0_NON_VP to simplify some ifdefs
kaf24@firebug.cl.cam.ac.uk [Sun, 30 Apr 2006 08:50:49 +0000 (09:50 +0100)]
Introduce XEN_IA64_DOM0_NON_VP to simplify some ifdefs
from defined(__ia64__) && !defined(CONFIG_XEN_IA64_DOM0_VP)
to defined(XEN_IA64_DOM0_NON_VP).

Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
19 years agoBalloon driver: when in autotranslate mode, pages are returned
kaf24@firebug.cl.cam.ac.uk [Sun, 30 Apr 2006 08:47:51 +0000 (09:47 +0100)]
Balloon driver: when in autotranslate mode, pages are returned
to xen without traversing init_mm.
This change is needed for xen/ia64 vp model.

Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
19 years agoFix blktap "bad page state" bug.
kaf24@firebug.cl.cam.ac.uk [Sun, 30 Apr 2006 08:39:04 +0000 (09:39 +0100)]
Fix blktap "bad page state" bug.

Signed-off-by: Andrew Warfield <andrew.warfield@cl.cam.ac.uk>
Signed-off-by: Ewan Mellor <ewan@xensource.com>
19 years agoballoon_alloc_empty_page_range() should set the reference count
kaf24@firebug.cl.cam.ac.uk [Sun, 30 Apr 2006 08:32:21 +0000 (09:32 +0100)]
balloon_alloc_empty_page_range() should set the reference count
on every page structure before returning.

Signed-off-by: Keir Fraser <keir@xensource.com>
19 years agoRemove definition of NR_PIRQS and replace all uses with
kaf24@firebug.cl.cam.ac.uk [Sun, 30 Apr 2006 08:26:01 +0000 (09:26 +0100)]
Remove definition of NR_PIRQS and replace all uses with
NR_IRQS. Allows removal of redundant check in pirq_guest_bind().

Signed-off-by: Keir Fraser <keir@xensource.com>
19 years agoDefine new event-channel and physdev hypercalls with a more extensible
kaf24@firebug.cl.cam.ac.uk [Sun, 30 Apr 2006 08:16:15 +0000 (09:16 +0100)]
Define new event-channel and physdev hypercalls with a more extensible
interface (the legacy hypercalls would break if subcommands with large
argument structures were added, as it would grow the size of the
union of all argument structures).

Also, based on a patch from Kevin Tian, add a new physdev op to
signal EOI for a particular irq.

Signed-off-by: Keir Fraser <keir@xensource.com>
19 years agoWait for the device to be connected on the guest (with a 10s timeout)
vhanquez@kneesa.uk.xensource.com [Fri, 28 Apr 2006 15:44:00 +0000 (16:44 +0100)]
Wait for the device to be connected on the guest (with a 10s timeout)
before detaching it. Fix a race condition that dom0 was tearing down the
block device while the guest was putting it in a Connected state.

Signed-off-by: Vincent Hanquez <vincent@snarc.org>
19 years agoxen/ia64 with dom0 vp model doesn't need the blkbkack ia64 workarounds
kaf24@firebug.cl.cam.ac.uk [Fri, 28 Apr 2006 13:48:50 +0000 (14:48 +0100)]
xen/ia64 with dom0 vp model doesn't need the blkbkack ia64 workarounds
any more.  xen/ia64 dom0 vp model is a big change so that xen/ia64
with the old model (P=M) still remains as the compile option.
Hopefully the work around will be eliminated in the future completely.

Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
19 years agoMakefile changes for xen/ia64 with dom0 vp model needs.
kaf24@firebug.cl.cam.ac.uk [Fri, 28 Apr 2006 13:47:42 +0000 (14:47 +0100)]
Makefile changes for xen/ia64 with dom0 vp model needs.
xenLinux/ia64 with dom0 vp model needs arch/i386/kernel/swiotlb.o

Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
19 years agoxen/ia64 with dom0 vp model needs direct_remap_pfn_range() to be called
kaf24@firebug.cl.cam.ac.uk [Fri, 28 Apr 2006 13:42:53 +0000 (14:42 +0100)]
xen/ia64 with dom0 vp model needs direct_remap_pfn_range() to be called
for IOCTL_PRIVCMD_MMAPBATCH.

Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
19 years agoDefine 8 hypercall numbers for arch-specific purposes.
kaf24@firebug.cl.cam.ac.uk [Fri, 28 Apr 2006 13:38:39 +0000 (14:38 +0100)]
Define 8 hypercall numbers for arch-specific purposes.

Signed-off-by: Keir Fraser <Keir.Fraser@cl.cam.ac.uk>
Signed-off-by: Tian Kevin <kevin.tian@intel.com>
Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
19 years agoRecent GUEST_HANDLE related cset broke ia64 compilation.
kaf24@firebug.cl.cam.ac.uk [Fri, 28 Apr 2006 13:35:34 +0000 (14:35 +0100)]
Recent GUEST_HANDLE related cset broke ia64 compilation.

Signed-off-by Kevin Tian <kevin.tian@intel.com>

19 years agoWhen boot device is cdrom but the cdrom is not bootable, the guest appears
kaf24@firebug.cl.cam.ac.uk [Fri, 28 Apr 2006 13:33:40 +0000 (14:33 +0100)]
When boot device is cdrom but the cdrom is not bootable, the guest appears
to hang. This patch fixes this by filling in second- and third-priority
fallback boot devices.

Signed-off by: Chen Jun <chenjunb@lenovo.com>

19 years agoRevert the xm-test ramdisk patch until we build new initrds.
kaf24@firebug.cl.cam.ac.uk [Fri, 28 Apr 2006 13:31:47 +0000 (14:31 +0100)]
Revert the xm-test ramdisk patch until we build new initrds.
Signed-off-by: Keir Fraser <keir@xensource.com>
19 years agoAllow for arch specific virq definitions.
kaf24@firebug.cl.cam.ac.uk [Fri, 28 Apr 2006 13:27:38 +0000 (14:27 +0100)]
Allow for arch specific virq definitions.

Signed-off-by Kevin Tian <kevin.tian@intel.com>

19 years agoThis patch fixes the buildroot for the xm-test ramdisk. The buildroot xm-test
kaf24@firebug.cl.cam.ac.uk [Fri, 28 Apr 2006 13:13:56 +0000 (14:13 +0100)]
This patch fixes the buildroot for the xm-test ramdisk. The buildroot xm-test
was getting is no longer there, they've pruned the snapshots. This patch
adds a new buildroot and updates the buildroot configuration. I also
incremented the xm-test release version to 0.8.

Signed-off-by: Daniel Stekloff <dsteklof@us.ibm.com>
19 years agoInstead of relying on xm create to always run the bootloader, make sure
kaf24@firebug.cl.cam.ac.uk [Fri, 28 Apr 2006 13:11:57 +0000 (14:11 +0100)]
Instead of relying on xm create to always run the bootloader, make sure
we run it if we get into domain creation with a bootloader set but no
image.  This could happen if someone creates a domain config via the
XML-RPC or sxp interfaces.

Signed-off-by: Jeremy Katz <katzj@redhat.com>
19 years agoThe features for a domain are more a property of the domain than the
kaf24@firebug.cl.cam.ac.uk [Fri, 28 Apr 2006 13:10:59 +0000 (14:10 +0100)]
The features for a domain are more a property of the domain than the
image.  Move them into the domain info so that the image is just the
pieces needed for booting (eg, kernel + initrd + args)

Signed-off-by: Jeremy Katz <katzj@redhat.com>
19 years agoThe bootloader code doesn't actually have to deal with the vcpu config
kaf24@firebug.cl.cam.ac.uk [Fri, 28 Apr 2006 13:09:35 +0000 (14:09 +0100)]
The bootloader code doesn't actually have to deal with the vcpu config
anymore as that's just in the domaininfo.  So stop doing so.

Signed-off-by: Jeremy Katz <katzj@redhat.com>
19 years agoWhen building with FORTIFY_SOURCE to ensure that return codes of common
kaf24@firebug.cl.cam.ac.uk [Fri, 28 Apr 2006 13:08:40 +0000 (14:08 +0100)]
When building with FORTIFY_SOURCE to ensure that return codes of common
functions are checked to avoid some bugs, a few warnings pop up and
become errors due to -Werror.  Attached checks the return codes
(or at least stores them to a dummy variable).

Signed-off-by: Jeremy Katz <katzj@redhat.com>
19 years agoRename {SG}ET_XEN_GUEST_HANDLE to {sg}et_Xen_guest_handle, as they
kaf24@firebug.cl.cam.ac.uk [Fri, 28 Apr 2006 08:40:07 +0000 (09:40 +0100)]
Rename {SG}ET_XEN_GUEST_HANDLE to {sg}et_Xen_guest_handle, as they
are function macros rather than define/declare macros.

Signed-off-by: Keir Fraser <keir@xensource.com>
19 years agoRelax BUG_ON() state ment in increase_reservation() in balloon.c.
kaf24@firebug.cl.cam.ac.uk [Fri, 28 Apr 2006 07:30:52 +0000 (08:30 +0100)]
Relax BUG_ON() state ment in increase_reservation() in balloon.c.
When xen_feature(XENFEAT_auto_translated_physmap) = 1,
phys_to_machine_mapping_valid(pfn) always returns true.
Relaxed BUG_ON() for xen_feature(XENFEAT_auto_translated_physmap) = 1 case.

Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
19 years agoFix memory_op hypercall mlock/munlock wrapping after the
kaf24@firebug.cl.cam.ac.uk [Thu, 27 Apr 2006 18:19:59 +0000 (19:19 +0100)]
Fix memory_op hypercall mlock/munlock wrapping after the
GET/SET_XEN_GUEST_HANDLE patch.

Signed-off-by: Keir Fraser <keir@xensource.com>
19 years agoAdd BUILD_BUG_ON() and a handful of users.
kaf24@firebug.cl.cam.ac.uk [Thu, 27 Apr 2006 14:55:09 +0000 (15:55 +0100)]
Add BUILD_BUG_ON() and a handful of users.
Signed-off-by: Jan Beulich <jbeulich@novell.com>
19 years agoAdd bounds check to get_mfn_from_gpfn().
kaf24@firebug.cl.cam.ac.uk [Thu, 27 Apr 2006 13:13:42 +0000 (14:13 +0100)]
Add bounds check to get_mfn_from_gpfn().
From: Jan Beulich
Signed-off-by: Keir Fraser <keir@xensource.com>
19 years agoDefine __XEN_TOOLS__ when building tools. Use this to
kaf24@firebug.cl.cam.ac.uk [Thu, 27 Apr 2006 13:06:41 +0000 (14:06 +0100)]
Define __XEN_TOOLS__ when building tools. Use this to
automatically build against latest Xen interface version,
and to specifically define GET_XEN_GUEST_HANDLE().

Signed-off-by: Keir Fraser <keir@xensource.com>
19 years agoBump Xen interface version number to 0x00030201 for new
kaf24@firebug.cl.cam.ac.uk [Thu, 27 Apr 2006 13:03:22 +0000 (14:03 +0100)]
Bump Xen interface version number to 0x00030201 for new
structural guest handles.
Signed-off-by: Keir Fraser <keir@xensource.com>
19 years agoWrap all GUEST_HANDLEs in structures, and define SET/GET macros to access them
kaf24@firebug.cl.cam.ac.uk [Thu, 27 Apr 2006 12:58:50 +0000 (13:58 +0100)]
Wrap all GUEST_HANDLEs in structures, and define SET/GET macros to access them
in Linux and libxc.

Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>
19 years agoRename *GUEST_HANDLE to *XEN_GUEST_HANDLE.
kaf24@firebug.cl.cam.ac.uk [Thu, 27 Apr 2006 12:38:21 +0000 (13:38 +0100)]
Rename *GUEST_HANDLE to *XEN_GUEST_HANDLE.
Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>
19 years agoOccasionally large smp machines fail to reboot properly and die under
kaf24@firebug.cl.cam.ac.uk [Thu, 27 Apr 2006 08:56:21 +0000 (09:56 +0100)]
Occasionally large smp machines fail to reboot properly and die under
an IPI storm of smp_call_function() to machine_reboot.  Only the boot
processor needs to run machine_restart, so send an IPI to CPU0.

Original patch by: Ryan Harper <ryanh@us.ibm.com>

Signed-off-by: Keir Fraser <keir@xensource.com>
19 years agoThis patch adds the ability to have restricted write access to some
kaf24@firebug.cl.cam.ac.uk [Thu, 27 Apr 2006 08:49:48 +0000 (09:49 +0100)]
This patch adds the ability to have restricted write access to some
of the structures on the capability list. Specifically, this patch adds
support for reading data through the Vital Product Data capability
structure and for manipulating power management of a card. A driver
domain is not allowed to enable Power Management Events (the PME trace
may be shared across devices in many domains), but it can put a device
in its control to sleep or query it for power usage statistics. This
code could possibly be expanded in the future to add support for AGP,
PCI-X, and MSI/MSIX (all of which are controlled through structures on
the capability list).

This patch also corrects some formatting issues in the PCI backend and
adds some comments to the code regarding permissive mode.

Signed-off-by: Ryan Wilson <hap9@epoch.ncsc.mil>
19 years agoCheck a write() return value in xenbaked.
kaf24@firebug.cl.cam.ac.uk [Thu, 27 Apr 2006 08:47:05 +0000 (09:47 +0100)]
Check a write() return value in xenbaked.

From: Jeremy Katz

Signed-off-by: Keir Fraser <keir@xensource.com>
19 years agoJust allocate unbound irq only from dynirq range, since
kaf24@firebug.cl.cam.ac.uk [Thu, 27 Apr 2006 08:43:49 +0000 (09:43 +0100)]
Just allocate unbound irq only from dynirq range, since
pirq range is reserved for physical devices. This saves
unnecessary checks on pirq range.

Signed-off-by Kevin Tian <kevin.tian@intel.com>

19 years agoMerge with xen-ia64-unstable.hg
kaf24@firebug.cl.cam.ac.uk [Thu, 27 Apr 2006 08:32:28 +0000 (09:32 +0100)]
Merge with xen-ia64-unstable.hg

19 years ago[IA64] Make VTI domain boot
awilliam@xenbuild.aw [Thu, 27 Apr 2006 02:55:42 +0000 (20:55 -0600)]
[IA64] Make VTI domain boot

Signed-off-by: Kevin Tian <kevin.tian@intel.com>
Signed-off-by: Anthony Xu <anthony.tian@intel.com>
19 years agomerge with xen-unstable.hg
awilliam@xenbuild.aw [Wed, 26 Apr 2006 19:17:37 +0000 (13:17 -0600)]
merge with xen-unstable.hg

19 years agoThis patch addresses CVE-2006-1056 (information leak from
kaf24@firebug.cl.cam.ac.uk [Wed, 26 Apr 2006 19:01:37 +0000 (20:01 +0100)]
This patch addresses CVE-2006-1056 (information leak from
fxsave/fxrstor on AMD CPUs) and also adjusts 64-bit handling so that
full 64-bit RIP/RDP values get saved/restored. More fine-grained
handling may be needed if 32-bit processes are expected to properly
see their selectors (native Linux doesn't currently do that either,
but there is a patch to adjust it there).

Original patch: Jan Beulich (based on Linux original by Andi Kleen)

Signed-off-by: Keir Fraser <keir@xensource.com>
19 years ago[IA64] not quite ready for --fatal-warnings
awilliam@xenbuild.aw [Wed, 26 Apr 2006 18:55:18 +0000 (12:55 -0600)]
[IA64] not quite ready for --fatal-warnings

The build fails for me with gcc-3.4.5/gas-2.16.91

Signed-off-by: Alex Williamson <alex.williamson@hp.com>
19 years ago[IA64] remove remaining asm warnings
awilliam@xenbuild.aw [Wed, 26 Apr 2006 18:50:42 +0000 (12:50 -0600)]
[IA64] remove remaining asm warnings

Fix all remaining warnings generated by gas.
Enforce the no gas warnings by using --fatal-warnings gas flag.

Signed-off-by: Tristan Gingold <tristan.gingold@bull.net>
19 years ago[IA64] cleanup
awilliam@xenbuild.aw [Wed, 26 Apr 2006 18:44:45 +0000 (12:44 -0600)]
[IA64] cleanup

Cleanup: add static, remove unused code/variables/macros.
Simplifications.

Signed-off-by: Tristan Gingold <tristan.gingold@bull.net>
19 years ago[IA64] Fix vm_summary info in VTi domain
awilliam@xenbuild.aw [Wed, 26 Apr 2006 18:40:56 +0000 (12:40 -0600)]
[IA64] Fix vm_summary info in VTi domain

This patch fixed vm_summary info and provide correct max_itr_entry,
max_dtr_entry,impl_va_msb, rid_size and so on.

Signed-off-by: Zhang xiantao <xiantao.zhang@intel.com>
19 years agoMake xen_features and force_evtchn_callback() non-GPL
kaf24@firebug.cl.cam.ac.uk [Wed, 26 Apr 2006 16:41:19 +0000 (17:41 +0100)]
Make xen_features and force_evtchn_callback() non-GPL
symbols as they are used in a variety of ubiquitous
kernel macros.

Signed-off-by: Keir Fraser <keir@xensource.com>
19 years agoxenLinux/ia64 has its own dma_map_page(), dma_unmap_page(), and it needs
yamahata@valinux.co.jp [Wed, 26 Apr 2006 09:43:16 +0000 (10:43 +0100)]
xenLinux/ia64 has its own dma_map_page(), dma_unmap_page(), and it needs
linux default implementations of dma_declare_coheremnt_memory() and its
families which is defined in include/linux/dma-mapping.h.
So those in pci-dma-xen.c are unnecessary. #ifdef out them.
PATCHNAME: pci-dma-xen-common

Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
19 years ago[IA64] remove RELOC_HIDE
awilliam@xenbuild.aw [Wed, 26 Apr 2006 05:38:09 +0000 (23:38 -0600)]
[IA64] remove RELOC_HIDE

This is now defined in xen/include/xen/compiler.h

Signed-off-by: Alex Williamson <alex.williamson@hp.com>
19 years agomerge with xen-unstable.hg
awilliam@xenbuild.aw [Wed, 26 Apr 2006 05:35:55 +0000 (23:35 -0600)]
merge with xen-unstable.hg

19 years ago[IA64] Fix the total memory info with xm info command
awilliam@xenbuild.aw [Wed, 26 Apr 2006 04:55:22 +0000 (22:55 -0600)]
[IA64] Fix the total memory info with xm info command

This small patch intends to provide correct total memory info for
control panel and fixed hardcode for that. The total memory info doesn't
include the memory FW used.

Signed-off-by : Zhang Xiantao  <xiantao.zhang@intel.com>

19 years ago[IA64] Reseve memory of domain0 (fix dom0 boot panic)
awilliam@xenbuild.aw [Wed, 26 Apr 2006 04:52:49 +0000 (22:52 -0600)]
[IA64] Reseve memory of domain0 (fix dom0 boot panic)

Our patch fix domain0 boot panic on large memory system.
(e.g. amount of installed memory is 16GB)

Memory of domain0 is not reserved now.
Our patch can make this memory reserved.
And we clean up initrd_start of domain0.

Signed-off-by: Kouya Shimura <kouya@jp.fujitsu.com>
Signed-off-by: Akio Takebe <takebe_akio@jp.fujitsu.com>
19 years ago[IA64] Fixed "Oops: time tick before it's due" issue
awilliam@xenbuild.aw [Wed, 26 Apr 2006 04:38:25 +0000 (22:38 -0600)]
[IA64] Fixed "Oops: time tick before it's due" issue

This patch fixed following issue,
"Oops: time tick before it's due" in VTI-domain

Signed-off-by: Anthony Xu <anthony.xu@intel.com>
19 years ago[IA64] FPH enabling + cleanup
awilliam@xenbuild.aw [Wed, 26 Apr 2006 04:35:41 +0000 (22:35 -0600)]
[IA64] FPH enabling + cleanup

Move contents of switch_to macro from xensystem.h to context_switch function.
Initialize FPU on all processors.  FPH is always enabled in Xen.
Speed up context-switch (a little bit!) by not enabling/disabling FPH.
Cleanup (unused function/variablesi/fields, debug printf...)
vmx_ia64_switch_to removed (was unused).

Signed-off-by: Tristan Gingold <tristan.gingold@bull.net>
19 years ago[IA64] smp boot speed-up (sal cache flush, itc/itv messages)
awilliam@xenbuild.aw [Wed, 26 Apr 2006 04:32:14 +0000 (22:32 -0600)]
[IA64] smp boot speed-up (sal cache flush, itc/itv messages)

Use sal_cache_flush to emulate SAL_CACHE_FLUSH instead of fc.
Only prints one vcpu_set_itc message, do not print set itv message.

Signed-off-by: Tristan Gingold <tristan.gingold@bull.net>
19 years ago[IA64] missed chunk of Kevin's hypercall cleanup patch
awilliam@xenbuild.aw [Wed, 26 Apr 2006 04:30:07 +0000 (22:30 -0600)]
[IA64] missed chunk of Kevin's hypercall cleanup patch

Missed this chunk of Kevin's patch when merging with dom0vp changes

Signed-off-by Kevin Tian <kevin.tian@intel.com>

19 years ago[IA64] Hypercall cleanup
awilliam@xenbuild.aw [Wed, 26 Apr 2006 04:27:18 +0000 (22:27 -0600)]
[IA64] Hypercall cleanup

Clean up xen_hypercall to query hypercall table instead.

Signed-off-by Kevin Tian <kevin.tian@intel.com>

19 years ago[IA64] Use16M page size in identity mapping
awilliam@xenbuild.aw [Wed, 26 Apr 2006 04:10:05 +0000 (22:10 -0600)]
[IA64] Use16M page size in identity mapping

Signed-off-by: Anthony Xu <anthony.xu@intel.com>
19 years ago[IA64] cleanup vtlb code
awilliam@xenbuild.aw [Wed, 26 Apr 2006 02:53:38 +0000 (20:53 -0600)]
[IA64] cleanup vtlb code

This patch is to clean up vtlb code.

Signed-off-by: Anthony Xu <anthony.xu@intel.com>
19 years ago[IA64] emulate ldfp8 in mmio
awilliam@localhost [Tue, 25 Apr 2006 23:05:16 +0000 (17:05 -0600)]
[IA64] emulate ldfp8 in mmio

1. emulate ldpf8 in mmio
2. handle floating point register rotation in functions setfpreg and getfpreg

Signed-off-by: Anthony Xu <anthony.xu@intel.com>
19 years ago[IA64] linux: arch/ia64/xen/makefile change
awilliam@localhost [Tue, 25 Apr 2006 22:55:09 +0000 (16:55 -0600)]
[IA64] linux: arch/ia64/xen/makefile change

update linux-2.6-xen-sparse/arch/ia64/xen/Makefile

Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
19 years ago[IA64] dma paravirtualization
awilliam@localhost [Tue, 25 Apr 2006 22:53:27 +0000 (16:53 -0600)]
[IA64] dma paravirtualization

Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>