]> xenbits.xensource.com Git - xen.git/log
xen.git
17 years ago[ACM] Remove aggregate set calculations for Domain-0
Keir Fraser [Sat, 24 Nov 2007 13:39:13 +0000 (13:39 +0000)]
[ACM] Remove aggregate set calculations for Domain-0

Since the pre_domain_create hook is called also for Dom-0, this
special case handling is not necessary any more.

Signed-off-by: Stefan Berger <stefanb@us.ibm.com>
17 years agoFix non-optimized compilation of Xen's memcmp
Keir Fraser [Sat, 24 Nov 2007 13:37:26 +0000 (13:37 +0000)]
Fix non-optimized compilation of Xen's memcmp

Even when using __builtin_memcmp, gcc may emit external references to
memcmp (when not optimizing for instance), so this #define does not
always provide a completely suitable memcmp().

Signed-off-by: Samuel Thibault <samuel.thibault@citrix.com>
Signed-off-by: Keir Fraser <keir.fraser@citrix.com>
17 years ago[Mini-OS] Fix domain blocking race
Keir Fraser [Sat, 24 Nov 2007 13:31:39 +0000 (13:31 +0000)]
[Mini-OS] Fix domain blocking race

A callback which wakes a thread may happen between the moment
schedule() gives hand to the idle thread and the latter blocks the
domain.  Idle hence needs to atomically check that no thread is
running and block, else awoken threads may have to wait up to 10
seconds.

Signed-off-by: Samuel Thibault <samuel.thibault@citrix.com>
17 years ago[Mini-OS] Permit x86_64 trap handlers to return
Keir Fraser [Sat, 24 Nov 2007 13:31:01 +0000 (13:31 +0000)]
[Mini-OS] Permit x86_64 trap handlers to return

Trap handlers sometimes fix the problem, so they need to be able to
return.

Signed-off-by: Samuel Thibault <samuel.thibault@citrix.com>
17 years ago[Mini-OS] Make sure schedule() is called safely
Keir Fraser [Sat, 24 Nov 2007 13:29:20 +0000 (13:29 +0000)]
[Mini-OS] Make sure schedule() is called safely

If a thread tries to sleep from a callback or with callbacks disabled,
Mini-OS will completely lock, so make sure this never happens.

Signed-off-by: Samuel Thibault <samuel.thibault@citrix.com>
17 years ago[Mini-OS] Fix netfront xmit overflow
Keir Fraser [Sat, 24 Nov 2007 13:28:54 +0000 (13:28 +0000)]
[Mini-OS] Fix netfront xmit overflow

Fix xmit overflow by making netfront_xmit sleep until
network_tx_buf_gc() frees an xmit request.

Signed-off-by: Samuel Thibault <samuel.thibault@citrix.com>
17 years ago[Mini-OS] Fix net backend path leak
Keir Fraser [Sat, 24 Nov 2007 13:28:27 +0000 (13:28 +0000)]
[Mini-OS] Fix net backend path leak

Free backend path after use.

Signed-off-by: Samuel Thibault <samuel.thibault@citrix.com>
17 years ago[Mini-OS] Make wake_up callback-safe
Keir Fraser [Sat, 24 Nov 2007 13:27:59 +0000 (13:27 +0000)]
[Mini-OS] Make wake_up callback-safe

As it may compete with event handlers waking wait queues, wake_up
needs to be callback-safe.

Signed-off-by: Samuel Thibault <samuel.thibault@citrix.com>
17 years ago[Mini-OS] Fix init_waitqueue_entry
Keir Fraser [Sat, 24 Nov 2007 13:27:32 +0000 (13:27 +0000)]
[Mini-OS] Fix init_waitqueue_entry

Fix initialization of thread_list in init_waitqueue_entry().

Signed-off-by: Samuel Thibault <samuel.thibault@xensource.com>
17 years ago[Mini-OS] Add init_SEMAPHORE
Keir Fraser [Sat, 24 Nov 2007 13:27:08 +0000 (13:27 +0000)]
[Mini-OS] Add init_SEMAPHORE

Add init_SEMAPHORE() to dynamically initialize semaphores.

Signed-off-by: Samuel Thibault <samuel.thibault@citrix.com>
17 years ago[Mini-OS] Make semaphores callback-safe
Keir Fraser [Sat, 24 Nov 2007 13:26:39 +0000 (13:26 +0000)]
[Mini-OS] Make semaphores callback-safe

One may want to use semaphores in event handlers to wake threads
waiting for a resource, so semaphores then need to be callback-safe.

Signed-off-by: Samuel Thibault <samuel.thibault@citrix.com>
17 years agox86, hvm: Config option to allow vmxassist to be disabled.
Keir Fraser [Sat, 24 Nov 2007 13:23:22 +0000 (13:23 +0000)]
x86, hvm: Config option to allow vmxassist to be disabled.
hvmloader is modified to dynamically detect this, allowing possibility
of optional full vmxassist replacement in 3.2 stable branch in future.
Currently 'vmxassist=y' is not much use since no replacement is
implemented.

Signed-off-by: Keir Fraser <keir.fraser@citrix.com>
17 years ago[Mini-OS] Add strrchr()
Keir Fraser [Fri, 23 Nov 2007 16:42:44 +0000 (16:42 +0000)]
[Mini-OS] Add strrchr()

Add strrchr(), useful e.g. for grabbing the last part of a xenbus
path.

Signed-off-by: Samuel Thibault <samuel.thibault@citrix.com>
17 years ago[Mini-OS] Add ENOTSUP
Keir Fraser [Fri, 23 Nov 2007 16:42:23 +0000 (16:42 +0000)]
[Mini-OS] Add ENOTSUP

Add missing ENOTSUP by defining it to EOPNOTSUPP like in Linux.

Signed-off-by: Samuel Thibault <samuel.thibault@citrix.com>
17 years ago[Mini-OS] Move _ctype into a module
Keir Fraser [Fri, 23 Nov 2007 16:41:54 +0000 (16:41 +0000)]
[Mini-OS] Move _ctype into a module

Move the _ctype array into its own module, instead of having it
defined in a header.

Signed-off-by: Samuel Thibault <samuel.thibault@citrix.com>
17 years agovt-d: Fix ISA IRQ alias issue
Keir Fraser [Fri, 23 Nov 2007 16:39:45 +0000 (16:39 +0000)]
vt-d: Fix ISA IRQ alias issue

When assign multiple devices to guest which uses PIC, ISA IRQ alias
may occur. This patch splits ISA IRQ and GSI eoi function. In ISA IRQ
eoi function, searches all assigned mirqs and does eoi for the
corresponding mirqs which match the eoi ISA IRQ. Therefore fix ISA IRQ
alias issue.

Signed-off-by: Weidong Han <weidong.han@intel.com>
17 years ago[Mini-OS] Make bind_virq return the port
Keir Fraser [Fri, 23 Nov 2007 16:25:59 +0000 (16:25 +0000)]
[Mini-OS] Make bind_virq return the port

Make bind_virq() return the port bound to the irq, which is useful
e.g. for masking/unmasking it.

Signed-off-by: Samuel Thibault <samuel.thibault@citrix.com>
17 years ago[Mini-OS] Fix x86 arch_switch_thread
Keir Fraser [Fri, 23 Nov 2007 16:23:28 +0000 (16:23 +0000)]
[Mini-OS] Fix x86 arch_switch_thread

Fix x86 arch_switch_thread by making it pure assembly.
There were missing general register clobbers for x86_64, and BP should
theorically be clobbered too, but gcc does not believe that, so the
only simple safe solution is to use pure assembly.

Signed-off-by: Samuel Thibault <samuel.thibault@citrix.com>
17 years ago[Mini-OS] Optimize get_current()
Keir Fraser [Fri, 23 Nov 2007 16:23:03 +0000 (16:23 +0000)]
[Mini-OS] Optimize get_current()

Let gcc perform the computation with SP itself, leading to yet better
code.

Signed-off-by: Samuel Thibault <samuel.thibault@citrix.com>
17 years ago[Mini-OS] Fix stack closures
Keir Fraser [Fri, 23 Nov 2007 16:22:36 +0000 (16:22 +0000)]
[Mini-OS] Fix stack closures

So as to make backtracing tools happy, correctly close x86 stacks for
new threads as well as on callback in the x86_32 case since there is
no unwind marker.

Signed-off-by: Samuel Thibault <samuel.thibault@citrix.com>
17 years ago[Mini-OS] Fix x86 initial stack alignment
Keir Fraser [Fri, 23 Nov 2007 16:22:13 +0000 (16:22 +0000)]
[Mini-OS] Fix x86 initial stack alignment

This fixes the initial stack alignment for x86, which is required for
current to return a fine NULL instead of a random value or possibly
crash during initialization.

Signed-off-by: Grzegorz Milos <gm281@cam.ac.uk>
Signed-off-by: Samuel Thibault <samuel.thibault@citrix.com>
17 years agoxend: Add support for NetBSD.
Keir Fraser [Fri, 23 Nov 2007 11:58:11 +0000 (11:58 +0000)]
xend: Add support for NetBSD.
Signed-off-by: Christoph Egger <Christoph.Egger@amd.com>
17 years agohvm: Clean up VMCS/VMCB construction.
Keir Fraser [Thu, 22 Nov 2007 19:55:42 +0000 (19:55 +0000)]
hvm: Clean up VMCS/VMCB construction.
Signed-off-by: Keir Fraser <keir.fraser@eu.citrix.com>
17 years agox86: emulate I/O port access breakpoints
Keir Fraser [Thu, 22 Nov 2007 19:23:40 +0000 (19:23 +0000)]
x86: emulate I/O port access breakpoints

Emulate the trapping on I/O port accesses when emulating IN/OUT.

Also allow 8-byte breakpoints on x86-64 (and on i686 if the hardware
supports them), and tighten the condition for loading debug registers
during context switch.

Signed-off-by: Jan Beulich <jbeulich@novell.com>
Signed-off-by: Keir Fraser <keir.fraser@eu.citrix.com>
17 years agox86: single step after instruction emulation
Keir Fraser [Thu, 22 Nov 2007 18:28:47 +0000 (18:28 +0000)]
x86: single step after instruction emulation

Inject single step trap after emulating instructions if guest's
EFLAGS.TF is set.

Signed-off-by: Jan Beulich <jbeulich@novell.com>
Signed-off-by: Keir Fraser <keir.fraser@eu.citrix.com>
17 years agoacm: Fix an exit label.
Keir Fraser [Thu, 22 Nov 2007 17:44:51 +0000 (17:44 +0000)]
acm: Fix an exit label.

Jump to a different error exit label if error occurs early in the
policy update process.

Signed-off-by: Stefan Berger <stefanb@us.ibm.com>
17 years agonetfront/back: Specify interface allowing multicast address add/remove
Keir Fraser [Thu, 22 Nov 2007 17:43:29 +0000 (17:43 +0000)]
netfront/back: Specify interface allowing multicast address add/remove
requests to be passed from the frontend to the backend.

Signed-off-by: David Edmondson <dme@sun.com>
Signed-off-by: Keir Fraser <keir.fraser@eu.citrix.com>
17 years ago[SVM] handle MC threshold registers for Barcelona
Keir Fraser [Thu, 22 Nov 2007 15:10:47 +0000 (15:10 +0000)]
[SVM] handle MC threshold registers for Barcelona
Signed-off-by: Christoph Egger <Christoph.Egger@amd.com>
17 years agox86: make set_task_gate() x86-32-only since x86-64 doesn't have task gates.
Keir Fraser [Thu, 22 Nov 2007 14:16:25 +0000 (14:16 +0000)]
x86: make set_task_gate() x86-32-only since x86-64 doesn't have task gates.
Signed-off-by: Jan Beulich <jbeulich@novell.com>
17 years agox86-64: access only 4 bytes a the reset vector location on smpboot.
Keir Fraser [Thu, 22 Nov 2007 14:12:47 +0000 (14:12 +0000)]
x86-64: access only 4 bytes a the reset vector location on smpboot.
Signed-off-by: Jan Beulich <jbeulich@novell.com>
17 years agoHeader dependency fix.
Keir Fraser [Thu, 22 Nov 2007 14:04:44 +0000 (14:04 +0000)]
Header dependency fix.
Signed-off-by: Jan Beulich <jbeulich@novell.com>
17 years agoMerge with ia64 tree.
Keir Fraser [Thu, 22 Nov 2007 10:44:54 +0000 (10:44 +0000)]
Merge with ia64 tree.

17 years agolibxc: Minor clean up of xc_core, and fix for -fstrict-overflow.
Keir Fraser [Thu, 22 Nov 2007 10:40:45 +0000 (10:40 +0000)]
libxc: Minor clean up of xc_core, and fix for -fstrict-overflow.
Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
17 years agohvm: Clean up CPUID 0x80000001 emulation. Filter out RDTSCP feature
Keir Fraser [Thu, 22 Nov 2007 10:33:10 +0000 (10:33 +0000)]
hvm: Clean up CPUID 0x80000001 emulation. Filter out RDTSCP feature
and emulate with #UD on AMD.
Signed-off-by: Keir Fraser <keir.fraser@eu.citrix.com>
17 years agomerge with xen-unstable.hg (staging)
Alex Williamson [Wed, 21 Nov 2007 16:12:06 +0000 (09:12 -0700)]
merge with xen-unstable.hg (staging)

17 years agox86: rmb() can be weakened according to new Intel spec.
Keir Fraser [Wed, 21 Nov 2007 14:36:07 +0000 (14:36 +0000)]
x86: rmb() can be weakened according to new Intel spec.

Both Intel and AMD agree that, from a programmer's viewpoint:
 Loads cannot be reordered relative to other loads.
 Stores cannot be reordered relative to other stores.

Intel64 Architecture Memory Ordering White Paper
<http://developer.intel.com/products/processor/manuals/318147.pdf>

AMD64 Architecture Programmer's Manual, Volume 2: System Programming
<http://www.amd.com/us-en/assets/content_type/\
 white_papers_and_tech_docs/24593.pdf>

Signed-off-by: Keir Fraser <keir.fraser@eu.citrix.com>
17 years agox86: Re-factor and clean up system.h.
Keir Fraser [Wed, 21 Nov 2007 14:27:38 +0000 (14:27 +0000)]
x86: Re-factor and clean up system.h.
Signed-off-by: Keir Fraser <keir.fraser@eu.citrix.com>
17 years agoi386: Remove unnecessary, and broken, address check I added to seg
Keir Fraser [Wed, 21 Nov 2007 12:00:20 +0000 (12:00 +0000)]
i386: Remove unnecessary, and broken, address check I added to seg
fixup code. Oops.
Signed-off-by: Keir Fraser <keir.fraser@eu.citrix.com>
17 years agoi386: adjustment to segment fixup code.
Keir Fraser [Wed, 21 Nov 2007 11:49:41 +0000 (11:49 +0000)]
i386: adjustment to segment fixup code.

Clean up and support more instructions.

Signed-off-by: Jan Beulich <jbeulich@novell.com>
17 years agox86: Make IDT/GDT/LDT updates safe.
Keir Fraser [Wed, 21 Nov 2007 11:38:51 +0000 (11:38 +0000)]
x86: Make IDT/GDT/LDT updates safe.

This involves either determining that the entry will not be
read/written while the update takes place, or atomically making the
entry 'present', or doing the entire write atomically, as appropriate.

This issue raised, and original patch provided, by Jan Beulich.

Signed-off-by: Keir Fraser <keir.fraser@eu.citrix.com>
17 years agoioemu: Add e100 NIC support. Req'd for w2k3/IA64.
Keir Fraser [Wed, 21 Nov 2007 09:52:04 +0000 (09:52 +0000)]
ioemu: Add e100 NIC support. Req'd for w2k3/IA64.
Signed-off-by: Zhang Xin <xing.z.zhang@intel.com>
17 years agohvm: Do not crash guest if it does an unaligned access to an HPET
Keir Fraser [Wed, 21 Nov 2007 09:49:09 +0000 (09:49 +0000)]
hvm: Do not crash guest if it does an unaligned access to an HPET
register. Some CrashMe test apaprently does this and we don't want
that to crash the domain it runs in.
From: Dexuan Cui <dexuan.cui@intel.com>
Signed-off-by: Keir Fraser <keir.fraser@eu.citrix.com>
17 years agoRemove unused smpboot.h header file.
Keir Fraser [Wed, 21 Nov 2007 09:45:22 +0000 (09:45 +0000)]
Remove unused smpboot.h header file.
Signed-off-by: Keir Fraser <keir.fraser@eu.citrix.com>
17 years agomerge with xen-unstable.hg (staging)
Alex Williamson [Tue, 20 Nov 2007 18:53:44 +0000 (11:53 -0700)]
merge with xen-unstable.hg (staging)

17 years agoxen: Allow granting of foreign access to iomem pages, and with
Keir Fraser [Tue, 20 Nov 2007 17:26:48 +0000 (17:26 +0000)]
xen: Allow granting of foreign access to iomem pages, and with
arbitrary cache attributes.
Signed-off-by: Kieran Mansley <kmansley@solarflare.com>
Signed-off-by: Keir Fraser <keir.fraser@eu.citrix.com>
17 years ago[IA64] vti save-restore: save.h clean up
Alex Williamson [Tue, 20 Nov 2007 16:28:15 +0000 (09:28 -0700)]
[IA64] vti save-restore: save.h clean up

Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
17 years ago[IA64] vti save-restore: save/restore opt_feature status
Alex Williamson [Tue, 20 Nov 2007 16:21:27 +0000 (09:21 -0700)]
[IA64] vti save-restore: save/restore opt_feature status

Make hvm domain save/restore support opt_feature.

Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
17 years ago[IA64] vti save-restore: preparation opt_feature support
Alex Williamson [Tue, 20 Nov 2007 16:14:43 +0000 (09:14 -0700)]
[IA64] vti save-restore: preparation opt_feature support

Pass struct domain* to domain_opt_feature().
This patch is preparation patch for HVM domain save/restore opt_feature.

Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
17 years ago[IA64] vti save-restore: fix opt_feature hypercall
Alex Williamson [Tue, 20 Nov 2007 16:11:15 +0000 (09:11 -0700)]
[IA64] vti save-restore: fix opt_feature hypercall

- Fix XEN_IA64_OPTF_IDENT_MAP_REG[457] definitions.
- SMP fix. writing to domain->arch.opt_feature is racy.

Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
17 years ago[IA64] vti save-restore: fix vacpi_save()
Alex Williamson [Tue, 20 Nov 2007 15:46:24 +0000 (08:46 -0700)]
[IA64] vti save-restore: fix vacpi_save()

Always returned 0.

Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
17 years agox86, 32-on-64: Improve checking in vcpu_destroy_pagetables(). It *is*
Keir Fraser [Tue, 20 Nov 2007 15:34:25 +0000 (15:34 +0000)]
x86, 32-on-64: Improve checking in vcpu_destroy_pagetables(). It *is*
possible for 64-bit guest to have matching guest_table and
guest_table_user.
Original patch by John Levon <levon@movementarian.org>
Signed-off-by: Keir Fraser <keir.fraser@eu.citrix.com>
17 years agoxenapi: Extension to get vif total i/o stats.
Keir Fraser [Tue, 20 Nov 2007 15:18:09 +0000 (15:18 +0000)]
xenapi: Extension to get vif total i/o stats.
From: Stefan de Konink <skinkie@xs4all.nl>
Signed-off-by: Keir Fraser <keir.fraser@eu.citrix.com>
17 years agoACM: Test sharing as part of the authorization check.
Keir Fraser [Tue, 20 Nov 2007 15:13:29 +0000 (15:13 +0000)]
ACM: Test sharing as part of the authorization check.

Protected the policy from being changed while the hooks are being
called.

Signed-off-by: Stefan Berger <stefanb@us.ibm.com>
17 years agosvm: Fix __update_guest_eip() to clear interrupt shadow.
Keir Fraser [Tue, 20 Nov 2007 15:05:36 +0000 (15:05 +0000)]
svm: Fix __update_guest_eip() to clear interrupt shadow.
Get rid of assertions about return value of get_instruction_length()
-- instead test in __update_guest_eip() and crash the domain.
Cache value of 'current' in svm_do_hlt().

The mismanagement of the interrupt shadow was found by Christoph
Egger of AMD.

Signed-off-by: Keir Fraser <keir@xensource.com>
17 years agoxen: Domain0 builder should set start_info magic string to
Keir Fraser [Tue, 20 Nov 2007 14:35:36 +0000 (14:35 +0000)]
xen: Domain0 builder should set start_info magic string to
'xen-3.0-...' just like user-space builder for domUs.
Signed-off-by: Keir Fraser <keir@xensource.com>
17 years agox86: Move log-dirty-mode assertion to correct place.
Keir Fraser [Sat, 17 Nov 2007 13:11:25 +0000 (13:11 +0000)]
x86: Move log-dirty-mode assertion to correct place.
Signed-off-by: Keir Fraser <keir@xensource.com>
17 years agoLog dirty radix tree code cleanup. Also do not deference non-existent
Keir Fraser [Fri, 16 Nov 2007 20:06:15 +0000 (20:06 +0000)]
Log dirty radix tree code cleanup. Also do not deference non-existent
pointer in paging_new_log_dirty_*() functions if allocation fails.
Signed-off-by: Keir Fraser <keir.fraser@citrix.com>
17 years agohvm: RTC emulation - RTC_UIE/RTC_SET fix
Keir Fraser [Fri, 16 Nov 2007 19:07:46 +0000 (19:07 +0000)]
hvm: RTC emulation - RTC_UIE/RTC_SET fix

This patch fixes a bug in the RTC code which appears to originate in
having written the emuated device following an incorrect
specification. VMware has (or had, at least, at the time we were still
testing on both VMWare and Xen) the same issue. In the current code,
when RTC_SET is set, RTC_UIE is cleared. This does not match the
behavior of real hardware, where the case is simply that no update
ended interrupts are sent as long as RTC_SET is set, but the UE ints
will resume as soon as RTC_SET is cleared and the clock update is
done. This little patch fixes this issue. In practicality, this means
OS/2 can now set the time without having the clock stop. I don't know
if any other guests have been affected by this issue.

From: Trolle Selander <trolle.selander@gmail.com>
Signed-off-by: Keir Fraser <keir.fraser@citrix.com>
17 years agohvm: make dirty logging stop requiring physical pages of order > 0
Keir Fraser [Fri, 16 Nov 2007 18:33:24 +0000 (18:33 +0000)]
hvm: make dirty logging stop requiring physical pages of order > 0

This patch re-implements the (x86) hypervisor dirty page log with a
simple four-level radix tree whose nodes are all single pages, thus
making migration require only order-0 pages (where before it required
at least an order-5 page).

Unlike the p2m radix tree implementation, the interior nodes of this
tree are NOT page table nodes.  I chose a lazy-allocation and -mapping
approach because most pages are not marked dirty while dirty-logging is
enabled.  There are doubtless situations (the 'stream' benchmark, for
example) where a more complex p2m-like approach is faster, but I'm not
sure they're worth the effort.

Signed-off-by: Dave Lively <dlively@virtualiron.com>
17 years agox86: Clean up some files of inline asm, and fix
Keir Fraser [Fri, 16 Nov 2007 17:59:34 +0000 (17:59 +0000)]
x86: Clean up some files of inline asm, and fix
__copy_{to,from}_user_ll() for gcc 3.4 (asm output constraints all
need to be '=&').
Based on a patch by Jan Beulich.
Signed-off-by: Keir Fraser <keir.fraser@citrix.com>
17 years agohvm passthru: Uses of XC_PAGE_MASK should be XC_PAGE_SIZE-1.
Keir Fraser [Fri, 16 Nov 2007 17:09:53 +0000 (17:09 +0000)]
hvm passthru: Uses of XC_PAGE_MASK should be XC_PAGE_SIZE-1.
Signed-off-by: Weidong Han <weidong.han@intel.com>
17 years agoxentrace: make xentrace and xenmon work on Solaris and *BSD.
Keir Fraser [Fri, 16 Nov 2007 17:08:05 +0000 (17:08 +0000)]
xentrace: make xentrace and xenmon work on Solaris and *BSD.
 - Use getopt() to get rid of argp dependency which does not exist
   on Solaris and *BSD. Done by Tariq Magdon-Ismail.
 - Minor modifications by me (Christoph) to make it also work on *BSD.
 - Tested on Linux by me (Christoph). No functional change on Linux.
 - Tariq ok'd BSD modifications for Solaris
 - Tariq ok'd submission by me :)

Signed-off-by: Tariq Magdon-Ismail <tariqmi@sun.com>
Signed-off-by: Christoph Egger <Christoph.Egger@amd.com>
17 years agoxentrace: Fix TypeError handling of xentrace_format.
Keir Fraser [Fri, 16 Nov 2007 17:05:20 +0000 (17:05 +0000)]
xentrace: Fix TypeError handling of xentrace_format.
Signed-off-by: Yosuke Iwamatsu <y-iwamatsu@ab.jp.nec.com>
17 years agoqemu: more verbose on failures.
Keir Fraser [Fri, 16 Nov 2007 17:03:46 +0000 (17:03 +0000)]
qemu: more verbose on failures.
Signed-off-by: Christoph Egger <Christoph.Egger@amd.com>
17 years agovt-d: Clean up VT-d code
Keir Fraser [Fri, 16 Nov 2007 17:01:50 +0000 (17:01 +0000)]
vt-d: Clean up VT-d code
Signed-off-by: Weidong Han <weidong.han@intel.com>
17 years agopvfb: PVFB SDL backend chokes on bogus screen updates
Keir Fraser [Fri, 16 Nov 2007 16:53:43 +0000 (16:53 +0000)]
pvfb: PVFB SDL backend chokes on bogus screen updates

Bogus screen update requests from buggy or malicous frontend make SDL
crash.  The VNC backend silently ignores them.  Catch and log them.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
17 years agoxend: host.get_log() close fd
Keir Fraser [Fri, 16 Nov 2007 16:43:57 +0000 (16:43 +0000)]
xend: host.get_log() close fd
Signed-off-by: Stefan Berger <stefanb@us.ibm.com>
17 years agovmx: wbinvd optimization for pass-through domain.
Keir Fraser [Fri, 16 Nov 2007 16:36:38 +0000 (16:36 +0000)]
vmx: wbinvd optimization for pass-through domain.

Optimise wbinvd exit emulation for pass-through domains to avoid
"always wbinvd" when a VCPU is migrated. Instead, do host wbinvd on
all host CPUs when wbinvd exit.

Signed-off-by Yaozu (Eddie) Dong <eddie.dong@intel.com>

17 years agox86, hvm: Small code cleanups.
Keir Fraser [Fri, 16 Nov 2007 16:22:00 +0000 (16:22 +0000)]
x86, hvm: Small code cleanups.
Based on patch from Xin Li.
Signed-off-by: Keir Fraser <keir.fraser@citrix.com>
17 years agoxend, xenapi: Strip characters from logs that XMLRPC layer cannot handle.
Keir Fraser [Fri, 16 Nov 2007 14:45:35 +0000 (14:45 +0000)]
xend, xenapi: Strip characters from logs that XMLRPC layer cannot handle.
From: Ian Jackson <ian.jackson@citrix.com>
Signed-off-by: Keir Fraser <keir.fraser@citrix.com>
17 years agox86, hvm: Allow stdvga acceleration to work with 32-bit x86.
Keir Fraser [Fri, 16 Nov 2007 14:40:22 +0000 (14:40 +0000)]
x86, hvm: Allow stdvga acceleration to work with 32-bit x86.
Signed-off-by: Keir Fraser <keir.fraser@citrix.com>
17 years agoxend: readDomTxn() and friends use dompath, not vmpath.
Keir Fraser [Thu, 15 Nov 2007 22:14:13 +0000 (22:14 +0000)]
xend: readDomTxn() and friends use dompath, not vmpath.
Signed-off-by: Keir Fraser <keir.fraser@citrix.com>
17 years agoFix header dependencies list, broken by c/s 16133.
Keir Fraser [Tue, 13 Nov 2007 20:13:50 +0000 (20:13 +0000)]
Fix header dependencies list, broken by c/s 16133.
Signed-off-by: Jan Beulich <jbeulich@novell.com>
17 years agox86: Do not read/write EFER MSR if it doesn't exist.
Keir Fraser [Tue, 13 Nov 2007 20:08:39 +0000 (20:08 +0000)]
x86: Do not read/write EFER MSR if it doesn't exist.
Signed-off-by: Keir Fraser <keir.fraser@citrix.com>
17 years agoFix non-portable grep usage. Solaris doesn't have -q option.
Keir Fraser [Tue, 13 Nov 2007 19:26:55 +0000 (19:26 +0000)]
Fix non-portable grep usage. Solaris doesn't have -q option.
Signed-off-by: John Levon <john.levon@sun.com>
17 years agox86: GDT reserved page should be a full page.
Keir Fraser [Tue, 13 Nov 2007 19:21:00 +0000 (19:21 +0000)]
x86: GDT reserved page should be a full page.
Signed-off-by: Keir Fraser <keir.fraser@citrix.com>
17 years agovmx: Restore correct host SYSENTER parameters on vmexit.
Keir Fraser [Tue, 13 Nov 2007 19:05:27 +0000 (19:05 +0000)]
vmx: Restore correct host SYSENTER parameters on vmexit.
Also simplify vmx_set_host_env().
HOST_GDT_BASE does not have to change when we shift CPU.
Signed-off-by: Keir Fraser <keir.fraser@citrix.com>
17 years agovmx: Set HOST_FS and HOST_GS to unusable on vmexit.
Keir Fraser [Tue, 13 Nov 2007 18:30:47 +0000 (18:30 +0000)]
vmx: Set HOST_FS and HOST_GS to unusable on vmexit.
Signed-off-by: Keir Fraser <keir@xensource.com>
17 years agotools: Avoid -nostdlib argument to $(LD).
Keir Fraser [Tue, 13 Nov 2007 17:57:22 +0000 (17:57 +0000)]
tools: Avoid -nostdlib argument to $(LD).
This interacts badly with GNU ld on Solaris.
Signed-off-by: Keir Fraser <keir@xensource.com>
17 years agoRevert 16067:9f9f9b68cd08a03fc8cfad9f5ab702e50b6b6463.
Keir Fraser [Tue, 13 Nov 2007 17:28:44 +0000 (17:28 +0000)]
Revert 16067:9f9f9b68cd08a03fc8cfad9f5ab702e50b6b6463.
Signed-off-by: Keir Fraser <keir@xensource.com>
17 years agolibxc: Consistently print errno on write() error in domain_save code.
Keir Fraser [Sun, 11 Nov 2007 18:28:57 +0000 (18:28 +0000)]
libxc: Consistently print errno on write() error in domain_save code.
Signed-off-by: Keir Fraser <keir@xensource.com>
17 years agolibxc: Consolidate read()/write() syscall wrappers to read/write an
Keir Fraser [Sun, 11 Nov 2007 18:22:33 +0000 (18:22 +0000)]
libxc: Consolidate read()/write() syscall wrappers to read/write an
exact number of bytes. The consolidated versions are more watertight
than the various versions previously distributed around the library
source code.
Signed-off-by: Keir Fraser <keir@xensource.com>
17 years agosolaris, xend: Fix FMRI for Solaris SMF.
Keir Fraser [Sat, 10 Nov 2007 10:55:27 +0000 (10:55 +0000)]
solaris, xend: Fix FMRI for Solaris SMF.

Update the FMRI for SMF to reflect reality (again).

Signed-off-by: John Levon <john.levon@sun.com>
17 years agoxend: Fix appending policy module to end of grub's config file
Keir Fraser [Sat, 10 Nov 2007 10:54:32 +0000 (10:54 +0000)]
xend: Fix appending policy module to end of grub's config file

This patch fixes the case where a module line is supposed to be added
to the very end of the file but the file does not end in with a new
line. Also fixes a problem that in some cases the module line would
not be properly be removed.

Signed-off-by: Stefan Berger <stefanb@us.ibm.com>
17 years agovmx: Enable WBINVD intercepts to avoid real WBINVD for non-vtd guests.
Keir Fraser [Fri, 9 Nov 2007 12:59:58 +0000 (12:59 +0000)]
vmx: Enable WBINVD intercepts to avoid real WBINVD for non-vtd guests.
Signed-off-by: Keir Fraser <keir@xensource.com>
17 years agosvm: Intercept and disallow MONITOR/MWAIT.
Keir Fraser [Fri, 9 Nov 2007 12:08:37 +0000 (12:08 +0000)]
svm: Intercept and disallow MONITOR/MWAIT.
Signed-off-by: Keir Fraser <keir@xensource.com>
17 years agox86, hvm: Better handling of INVD/WBINVD on VMX and SVM.
Keir Fraser [Fri, 9 Nov 2007 12:05:27 +0000 (12:05 +0000)]
x86, hvm: Better handling of INVD/WBINVD on VMX and SVM.
Also better handling of MONITOR/MWAIT on VMX.
Signed-off-by: Keir Fraser <keir@xensource.com>
17 years agoMerge with ia64 tree.
Keir Fraser [Thu, 8 Nov 2007 18:15:36 +0000 (18:15 +0000)]
Merge with ia64 tree.

17 years agoFix i386 Xen build.
Keir Fraser [Thu, 8 Nov 2007 18:14:00 +0000 (18:14 +0000)]
Fix i386 Xen build.
Signed-off-by: Keir Fraser <keir@xensource.com>
17 years agoFix xend startup after 16344:0ebac857649591.
Keir Fraser [Thu, 8 Nov 2007 16:27:03 +0000 (16:27 +0000)]
Fix xend startup after 16344:0ebac857649591.
Signed-off-by: Keir Fraser <keir.fraser@eu.citrix.com>
17 years ago[IA64] Add paddr_to_pfn for upstream changes
Alex Williamson [Thu, 8 Nov 2007 16:05:26 +0000 (09:05 -0700)]
[IA64] Add paddr_to_pfn for upstream changes

Signed-off-by: Alex Williamson <alex.williamson@hp.com>
17 years agomerge with xen-unstable.hg (staging)
Alex Williamson [Thu, 8 Nov 2007 16:37:06 +0000 (09:37 -0700)]
merge with xen-unstable.hg (staging)

17 years agoxend: Allow for non-existent xen-python-path binary.
Keir Fraser [Thu, 8 Nov 2007 15:32:14 +0000 (15:32 +0000)]
xend: Allow for non-existent xen-python-path binary.
On Solaris, xen-python-path is unnecessary.

Signed-off-by: John Levon <john.levon@sun.com>
17 years agohvm: Clean up buf_ioreq handling.
Keir Fraser [Thu, 8 Nov 2007 14:50:01 +0000 (14:50 +0000)]
hvm: Clean up buf_ioreq handling.
Also, disable stdvga caching on hvm save/restore, as the shadow vga
state is not preserved.
Signed-off-by: Keir Fraser <keir@xensource.com>
17 years agox86, vt-d: Clean up utils code.
Keir Fraser [Thu, 8 Nov 2007 13:14:03 +0000 (13:14 +0000)]
x86, vt-d: Clean up utils code.
Signed-off-by: Weidong Han <weidong.han@intel.com>
17 years agox86, hvm: Clean up periodic timer code a little. This leads naturally
Keir Fraser [Thu, 8 Nov 2007 10:33:18 +0000 (10:33 +0000)]
x86, hvm: Clean up periodic timer code a little. This leads naturally
to a no-missed-tick-accounting mode which is a combination of ticks
delivered 'off beat' immediately upon re-scheduling when ticks are
missed, then reverting to delivering ticks 'on beat' as usual.
Signed-off-by: Keir Fraser <keir@xensource.com>
17 years agoxm: Fix long options of xm sched-credit
Keir Fraser [Thu, 8 Nov 2007 10:23:22 +0000 (10:23 +0000)]
xm: Fix long options of xm sched-credit

Long options of xm sched-credit command are ignored (--domain,
--weight, --cap). This patch fixes it.

Signed-off-by: Masaki Kanno <kanno.masaki@jp.fujitsu.com>
17 years agox86, hvm: Fix boot of HVM guests with pass-thru devices. Cache
Keir Fraser [Thu, 8 Nov 2007 10:21:47 +0000 (10:21 +0000)]
x86, hvm: Fix boot of HVM guests with pass-thru devices. Cache
attributes should not be propagated for Xen-heap pages.
Signed-off-by: Keir Fraser <keir@xensource.com>
17 years agoReplace is_xen_heap_frame() with is_xen_heap_{page,mfn}(). Callers are
Keir Fraser [Thu, 8 Nov 2007 10:19:18 +0000 (10:19 +0000)]
Replace is_xen_heap_frame() with is_xen_heap_{page,mfn}(). Callers are
changed to use whichever is more convenient.
Signed-off-by: Keir Fraser <keir@xensource.com>
17 years ago[IA64] Fix breakage from upstream and remove duplicate code
Alex Williamson [Wed, 7 Nov 2007 22:11:44 +0000 (15:11 -0700)]
[IA64] Fix breakage from upstream and remove duplicate code

- Get rid of PAGE_SIZE usage in arch-ia64.h when building tools.
- Now that more of xen_init_fv is identical to x86 code, share the
  code rather than duplicate it.
- Change a few things that came over from the ia64 tree to use the
  XC_* macros as already done upstream.

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