Memory sharing: better handling of ENOMEM while unsharing

If unsharing fails with ENOMEM, we were:
 - leaving the list of gfns backed by the shared page in an inconsistent state
 - cycling forever on the hap page fault handler.
 - Attempting to produce a mem event (which could sleep on a wait queue)
   while holding locks.
 - Not checking, for all callers, that unshare could have indeed failed.

Fix bugs above, and sanitize callers to place a ring event in an unlocked
context, or without requiring to go to sleep on a wait queue.

A note on the rationale for unshare error handling:
 1. Unshare can only fail with ENOMEM. Any other error conditions BUG_ON()
 2. We notify a potential dom0 helper through a mem_event ring. But we
    allow the notification to not go to sleep. If the event ring is full
    of ENOMEM warnings, then the helper will already have been kicked enough.
 3. We cannot "just" go to sleep until the unshare is resolved, because we
    might be buried deep into locks (e.g. something -> copy_to_user ->
    __hvm_copy)
 4. So, we make sure we:
    4.1. return an error
    4.2. do not corrupt memory shared with other guests
    4.3. do not corrupt memory private to the current guest
    4.4. do not corrupt the hypervisor memory sharing meta data
    4.5. let the guest deal with the error, if propagation will reach that far

Signed-off-by: Andres Lagar-Cavilla <andres@lagarcavilla.org>
Acked-by: Tim Deegan <tim@xen.org>
Committed-by: Tim Deegan <tim@xen.org>

x86/mm: Allow to not sleep on mem event ring

Under extreme congestion conditions, generating a mem event may put the vcpu to
sleep on a wait queue if the ring is full. This is generally desirable, although
fairly convoluted to work with, since sleeping on a wait queue requires a
non-atomic context (i.e. no locks held).

Introduce an allow_sleep flag to make this optional. The default API remains
such that all current callers set allow_sleep to true and thus will sleep if
necessary.

The end-use is for cases in which loss of guest mem events is tolerable. One
such consumer to be added later is the unsharing code under ENOMEM conditions.

Signed-off-by: Andres Lagar-Cavilla <andres@lagarcavilla.org>
Signed-off-by: Adin Scannell <adin@scannell.ca>
Acked-by: Tim Deegan <tim@xen.org>
Committed-by: Tim Deegan <tim@xen.org>

Mem event: don't leave zombie domains if there are wait-queued vcpus

Vcpus in wait queues retain a domain reference. Upon domain destruction, we
were not taking care of draining the wait queues.

Signed-off-by: Andres Lagar-Cavilla <andres@lagarcavilla.org>
Acked-by: Tim Deegan <tim@xen.org>
Committed-by: Tim Deegan <tim@xen.org>

vmx: Make it clear that the VMEXIT reason is in hex.

Signed-off-by: Tim Deegan <tim@xen.org>
Acked-by: Keir Fraser <keir@xen.org>
Committed-by: Tim Deegan <tim@xen.org>

x86/mm: Fix deadlock between p2m and event channel locks.

The hvm io emulation code holds the p2m lock for the duration of the
emulation, which may include sending an event to qemu. On a separate path,
map_domain_pirq grabs the event channel and p2m locks in opposite order.

Fix this by ensuring liveness of the ram_gfn used by io emulation, with a
page ref.

Reported-by: "Hao, Xudong" <xudong.hao@intel.com>
Signed-off-by: "Hao, Xudong" <xudong.hao@intel.com>
Signed-off-by: Andres Lagar-Cavilla <andres@lagarcavilla.org>
Acked-by: Tim Deegan <tim@xen.org>
Committed-by: Tim Deegan <tim@xen.org>

autoconf: add missing compression checks for libxc

Move missing checks from tools/libxc/Makefile to configure script.

Signed-off-by: Roger Pau Monne <roger.pau@entel.upc.edu>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Acked-by: Jan Beulich <JBeulich@suse.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>

autoconf: check for uuid.h or uuid/uuid.h with -luuid

Check for uuid/uuid.h and if found check for -luuid usability (Linux
case), if not, check for uuid (NetBSD). One of this tests has to
succeed to be able to build Xen tools.

Signed-off-by: Roger Pau Monne <roger.pau@entel.upc.edu>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>

tools: Add explicit clean rule for SeaBIOS

Since seabios-dir is cloned during build we need to check that it exists before
recursing into it for clean, following the pattern used for qemu-*-dir etc.

Also remove usage of "buildmakevars2shellvars" except when used to poopulate
the environment for qemu-xen-traditional's xen-setup script, which is the only
user.

Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>

autoconf: check for Python.h header and -lpython* lib

Check that the package usually called python-dev is present.

Signed-off-by: Roger Pau Monne <roger.pau@entel.upc.edu>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>

xl: Implement sched-credit schedule parameter command-line interface

Add features to the sched-credit interface to allow querying and
displaying scheduler parameters.

Signed-off-by: George Dunlap <george.dunlap@eu.citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>

xl: Refactor sched_domain_output to have a callback for pool information

Allow a scheduler to provide a callback to display pool-wide information,
providing a default.  This is in preparation for displaying pool-wide
scheduler parameters on this line.

Signed-off-by: George Dunlap <george.dunlap@eu.citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>

tapdisk2: initialize 'name' parameter of BLKTAP2_IOCTL_CREATE_DEVICE

This keeps the kernel driver from printing garbage.

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>

xl: display max_cpu_ids for xl info

Expose `max_cpu_id' in stdout from `xl info'

Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>

Merge

gitignore: Add cscope files

Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>

arm: add generated files to .gitignore and .hgignore

Signed-off-by: David Vrabel <david.vrabel@citrix.com>
Committed-by: Ian Campbell <ian.campbell@citrix.com>

arm: introduce more hypercalls

Implement xen_version, event_channel_op, memory_op sysctl and physdev_op
hypercalls.

Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Committed-by: Ian Campbell <ian.campbell@citrix.com>

arm: handle dom0_max_vcpus=0 case properly

Also use xzalloc_array.

Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Committed-by: Ian Campbell <ian.campbell@citrix.com>

arm: fix inflight_irqs list priority ordering

Lower priority integers mean higher priority.
Also when we are about to insert the lowest priority IRQ so far, add it
at the end.

Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Committed-by: Ian Campbell <ian.campbell@citrix.com>

arm: rename link to inflight

The link field in pending_irq has a confusing name so rename it to
inflight and comment its behaviour.

Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Committed-by: Ian Campbell <ian.campbell@citrix.com>

tools/hotplug/Linux/init.d/: create /var/run/xen if not exists

Signed-off-by: Fabio Fantoni <fabio.fantoni@heliman.it>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>

tools/python: Fix indenting in 25030:6ced0ed954d6

Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>

tools/python: Py_INCREF(Py_None) when returing Py_None

Signed-off-by: Marek Marczykowski <marmarek@invisiblethingslab.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>

xl block-detach: allow other than numeric-decimal specification of the disk

... to be in sync with block-attach.  And rename the function.

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>

libxl: always set a default value for VFB.opengl

Fixes an assertion failure.

Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Reported-by: Fantu <fantonifabio@tiscali.it>
Tested-by: Fantu <fantonifabio@tiscali.it>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>

tools, build: Fix libaio Makefile to put files in $(DESTDIR)

Signed-off-by: George Dunlap <george.dunlap@eu.citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>

Merge

arm: use r12 to pass the hypercall number

** This is a guest visible ABI change which requires an updated guest kernel **

Use r12 to pass the hypercall number and r0-r4 for the hypercall
arguments.

Use the ISS to pass an hypervisor specific tag.

Remove passing unused registers to arm_hypercall_table: we don't have 6
arguments hypercalls and we never use 64 bit values as hypercall
arguments, 64 bit values are only contained within structs passed as
arguments.

Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
[ use #ifndef NDEBUG, fix coding style, expand calling convention comment
  slightly and added a big fat note about ABI change - ijc ]
Committed-by: Ian Campbell <ian.campbell@citrix.com>

xenstat: Use local domain names

The domain name stored in /local/domain/$domid/name is simpler to
access and is the only domain name modified by "xl rename". Use this
domain name in libxenstat's reporting.

Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>

libxl: fix libxl_get_cpu_topology

Fix upper bound of for loop, thus preventing memory corruption.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>

libxl: publish number of paged pages.

This is information is currently not harvested from libxc.

Signed-off-by: Andres Lagar-Cavilla <andres@lagarcavilla.org>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>

Update SEABIOS_UPSTREAM_TAG

Switch to recently released 1.6.3.2 from upstream. This incorporates all the
changesets which we previously applied to 1.6.3.1.

Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>

Merge

autoconf: add check for curses library

Check for a curses compatible library (curses or ncurses basically).
One of those is needed to compile Xen tools (gtraceview and xentop).

Modify Makefiles/sources to use configure output (fetch CURSES_LIBS
from tools/Tools.mk and header to include from tools/config.h)

Signed-off-by: Roger Pau Monne <roger.pau@entel.upc.edu>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>

arm: Put the CPU into a sleep state in the idle loop

The ARM doesn't actually say that the WFI instruction will return
immediately if CPSR.I is clear and an instruction is pending, but
this seems to match up with how linux uses it.

Signed-off-by: Tim Deegan <tim@xen.org>
Committed-by: Ian Campbell <ian.campbell@citrix.com>

arm: Actually free __init/__initdata ranges on boot

Signed-off-by: Tim Deegan <tim@xen.org>
Committed-by: Ian Campbell <ian.campbell@citrix.com>

arm: new hook for late MMU setup on secondary CPUs

The boot CPU turns on W^X in setup_pagetables().  Do the same for other
CPUs after they boot.  If we go to per-CPU pagetables, this is where
that will happen.

Signed-off-by: Tim Deegan <tim@xen.org>
Committed-by: Ian Campbell <ian.campbell@citrix.com>

arm: Shutdown and reboot

Reboot runes grabbed from linux's SP810 reset function.
Doesn't seem to work on the model, though.

Signed-off-by: Tim Deegan <tim@xen.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Committed-by: Ian Campbell <ian.campbell@citrix.com>

arm: SMP CPU shutdown

For completeness, also implelent the CPU shutdown path.

Signed-off-by: Tim Deegan <tim@xen.org>
Committed-by: Ian Campbell <ian.campbell@citrix.com>

arm: Boot secondary CPUs into C

Secondary CPUs come up directly onto the stack of the appropriate idle
vcpu; the boot CPU starts on a statically allocated stack and switches
over to the idle vcpu's one once the idle domain has been built.

Signed-off-by: Tim Deegan <tim@xen.org>
Committed-by: Ian Campbell <ian.campbell@citrix.com>

arm: start plumbing in SMP bringup in C

Still a noop, but no longer just a dummy symbol.

Signed-off-by: Tim Deegan <tim@xen.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Committed-by: Ian Campbell <ian.campbell@citrix.com>

arm: per-cpu areas

Signed-off-by: Tim Deegan <tim@xen.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Committed-by: Ian Campbell <ian.campbell@citrix.com>

arm: More SMP bringup

Bring non-boot CPUs up as far as running on the relocated pagetables,
one at a time, before the non-relocated copy of Xen gets reused for
general memory pools.

Don't yet bring them up into C; that will happen later when stacks are
allocated.

Signed-off-by: Tim Deegan <tim@xen.org>
Committed-by: Ian Campbell <ian.campbell@citrix.com>

arm: Add a comment explaining the GICD writes in the GICC init function

Signed-off-by: Tim Deegan <tim@xen.org>
Committed-by: Ian Campbell <ian.campbell@citrix.com>

arm: implement udelay()

Signed-off-by: Tim Deegan <tim@xen.org>
Committed-by: Ian Campbell <ian.campbell@citrix.com>

x86/mm: drop gfn ref before calling handle_mmio()

Otherwise we get a deadlock between the p2m lock and the event lock,
which handle_mmio() acquires.

Signed-off-by: Tim Deegan <tim@xen.org>
Acked-by: Andres Lagar-Cavilla <andres@lagarcavilla.org>

low-mem virq: Parentheses around ternary operator in check_low_mem_virq()

Signed-off-by: Keir Fraser <keir@xen.org>

Tools: After a helper maps a ring, yank it from the guest physmap

This limits the ability of the guest to play around with its own rings, and DoS
itself.

Signed-off-by: Andres Lagar-Cavilla <andres@lagarcavilla.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Committed-by: Tim Deegan <tim@xen.org>

x86/mm: Fix mem event error message typos

Signed-off-by: Andres Lagar-Cavilla <andres@lagarcavilla.org>
Acked-by: Tim Deegan <tim@xen.org>
Committed-by: Tim Deegan <tim@xen.org>

x86/mm: Clean up mem event structures on domain destruction

Otherwise we wind up with zombie domains, still holding onto refs to the mem
event ring pages.

Signed-off-by: Andres Lagar-Cavilla <andres@lagarcavilla.org>
Acked-by: Tim Deegan <tim@xen.org>
Committed-by: Tim Deegan <tim@xen.org>

Tools: libxc side for setting up the mem sharing ring

This ring is used to report failed allocations in the unshare path.

Signed-off-by: Andres Lagar-Cavilla <andres@lagarcavilla.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Committed-by: Tim Deegan <tim@xen.org>

x86/mm: wire up sharing ring

Now that we have an interface close to finalizing, do the necessary plumbing to
set up a ring for reporting failed allocations in the unshare path.

Signed-off-by: Andres Lagar-Cavilla <andres@lagarcavilla.org>
Acked-by: Tim Deegan <tim@xen.org>
Committed-by: Tim Deegan <tim@xen.org>

Use a reserved pfn in the guest address space to store mem event rings

This solves a long-standing issue in which the pages backing these rings were
pages belonging to dom0 user-space processes. Thus, if the process would die
unexpectedly, Xen would keep posting events to a page now belonging to some
other process.

We update all API-consumers in tree (xenpaging and xen-access).

This is an API/ABI change, so please speak up if it breaks your accumptions.

The patch touches tools, hypervisor x86/hvm bits, and hypervisor x86/mm bits.

Signed-off-by: Andres Lagar-Cavilla <andres@lagarcavilla.org>
Acked-by: Tim Deegan <tim@xen.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Committed-by: Tim Deegan <tim@xen.org>

x86/hvm: refactor calls to prepare and tear down a helper ring

These are currently used for the rings connecting Xen with qemu. Refactor them
so the same code can be later used for mem event rings.

Signed-off-by: Andres Lagar-Cavilla <andres@lagarcavilla.org>
Acked-by: Tim Deegan <tim@xen.org>
Committed-by: Tim Deegan <tim@xen.org>

Tools: Remove shared page from mem_event/access/paging interfaces

Don't use the superfluous shared page, return the event channel directly as
part of the domctl struct, instead.

In-tree consumers (xenpaging, xen-access) updated. This is an ABI/API change,
so please voice any concerns.

Known pending issues:
- pager could die and its ring page could be used by some other process, yet
  Xen retains the mapping to it.
- use a saner interface for the paging_load buffer.

This change also affects the x86/mm bits in the hypervisor that process the
mem_event setup domctl.

Signed-off-by: Andres Lagar-Cavilla <andres@lagarcavilla.org>
Acked-by: Tim Deegan <tim@xen.org>
Acked-by: Olaf Hering <olaf@aepfle.de>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Committed-by: Tim Deegan <tim@xen.org>

x86/cpuidle: improve data arrangement

.. to reduce the amount of holes (wasted space).

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Keir Fraser <keir@xen.org>

XENPF_set_processor_pminfo XEN_PM_CX overflows states array

Calling XENPF_set_processor_pminfo with XEN_PM_CX could cause states
array in "struct acpi_processor_power" to exceed its limit.

The array used to be reset (by function cpuidle_init_cpu()) for each
hypercall. The patch puts it back that way and adds an assertion to
make it clear in case that happens again.

Signed-off-by: Eric Chanudet <eric.chanudet@eu.citrix.com>
- convert assertion to printk() & bail
- eliminate struct acpi_processor_cx's valid member (not read anymore)
- further adjustments to one-time-only vs each-time operations in
  cpuidle_init_cpu()
- don't use ACPI_STATE_Cn as array index anymore

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Keir Fraser <keir@xen.org>

oprofile: don't pass around redundant, easily derived arguments

Passing both a struct vcpu pointer and the corresponding struct domain
one is simply pointless, especially when intermediate functions just
forward it without themselves making use of the already obtained value.

Also constify a few function parameters.

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Keir Fraser <keir@xen.org>

ia64: fix build (once more)

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

Revert 24987:42afd4d1669b

Signed-off-by: Keir Fraser <keir@xen.org>

hvm: Print support/enable status of HAP superpage mappings.

Signed-off-by: Keir Fraser <keir@xen.org>

NMI: Command line parameter for watchdog timeout

Introduce a command parameter to set the watchtog timeout.  Manually
specifying "watchdog_timeout=<seconds>" on the command line will also
turn the watchdog on.  For consistency, move opt_watchdog into nmi.c
along with opt_watchdog_timeout.

Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Committed-by: Keir Fraser <keir@xen.org>

svm: Fake out the Bus Unit Config MSR on revF AMD CPUs

Win2k8 x64 reads this MSR on revF chips, where it wasn't publically
available; it uses a magic constant in %rdi as a password, which we
don't have in rdmsr_safe().  Since we'll ignore the later writes, just
use a plausible value here (the reset value from rev10h chips) if the
real CPU didn't provide one.

Signed-off-by: George Dunlap <george.dunlap@eu.citrix.com>
Committed-by: Keir Fraser <keir@xen.org>

x86: Add more specific logging for get_page_from_l1e failures

This patch should clarify which of the two possible reasons cause the
messages
of the form "(XEN) mm.c:908:d1 Error getting mfn 29b2b (pfn
5555555555555555)
from L1 entry 0000000029b2b067 for l1e_owner=1, pg_owner=1".

Signed-off-by: George Dunlap <george.dunlap@eu.citrix.com>
Committed-by: Keir Fraser <keir@xen.org>

x86: Reduce severity of hvm_op warnings

Signed-off-by: George Dunlap <george.dunlap@eu.citrix.com>
Committed-by: Keir Fraser <keir@xen.org>

x86: Make sure log-dirty is turned off before trying to dismantle it

Signed-off-by: George Dunlap <george.dunlap@eu.citrix.com>
Committed-by: Keir Fraser <keir@xen.org>

hvm: Add command line option to enable ASID support -- on by default

Signed-off-by: George Dunlap <george.dunlap@eu.citrix.com>
Signed-off-by: Keir Fraser <keir@xen.org>

Low mem virq incremental adjustments

Consider tmem before firing the virq.

Add .gitignore rune.

Signed-off-by: Andres Lagar-Cavilla <andres@lagarcavilla.org>
Acked-by: Dan Magenheimer <dan.magenheimer@oracle.com>
Committed-by: Keir Fraser <keir@xen.org>

provide a single, common implementation for get_order_from_{bytes,pages}()

All three per-architecture implementations were identical, and I cannot
see how future architectures would need any sort of customization here
(the only per-architecture aspect here is the actual PAGE_SHIFT value).

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Keir Fraser <keir@xen.org>

passthrough: fix domain ref leak after 24888:71159fb049f2

As pointed out by Keir.

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Keir Fraser <keir@xen.org>

Revert 24973:50a70b652b43 "x86: Use deep C states for off-lined CPUs"

applied already

Signed-off-by: Keir Fraser <keir@xen.org>

DOCS: Initial document regarding Xen's command line parameters

Still a work in progress, but submitted as a start.

Changes since v1:
  - Include all up to date information from the wiki.

Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Committed-by: Keir Fraser <keir@xen.org>

build: Don't remove user-supplied xen-version during make clean

xen/Makefile is designed to allow the user to supply a file named
xen/xen-include to change the format of xen version strings.
Unfortunately, "make clean" removes xen/xen*, which will remove this
file.

Make the clean process more targeted.

Signed-off-by: George Dunlap <george.dunlap@eu.citrix.com>
Committed-by: Keir Fraser <keir@xen.org>

build: Add xenversion as a top-level target

We already have xenversion as a target in the xen subdirectory;
simply extend it so that "make xenversion" does what you'd expect.

Signed-off-by: George Dunlap <george.dunlap@eu.citrix.com>
Committed-by: Keir Fraser <keir@xen.org>

RTC: Add alarm support

Use a timer to emulate update cycle. The timer runs only when AIE is
set.

Signed-off-by: Yang Zhang <yang.z.zhang@Intel.com>
Committed-by: Keir Fraser <keir@xen.org>

RTC: Add RTC update-ended interrupt support

Use a timer to emulate update cycle. When update cycle ended and UIE
is set, raise an interrupt. The timer runs only when AF is cleared.

Signed-off-by: Yang Zhang <yang.z.zhang@Intel.com>
Committed-by: Keir Fraser <keir@xen.org>

RTC: Add UIP(update in progress) check logic

The UIP(update in progress) is set when RTC is in updating. And the
update cycle begins 244us later after UIP is set.

Signed-off-by: Yang Zhang <yang.z.zhang@Intel.com>
Committed-by: Keir Fraser <keir@xen.org>

RTC: fix 12-hour mode

Hours in 12-hour mode are in the 1-12 range, not 0-11.

Signed-off-by: Yang Zhang <yang.z.zhang@Intel.com>
Committed-by: Keir Fraser <keir@xen.org>

RTC: use offset to caculate rtc clock

There has no need to run two periodic timer to update RTC time.
Use the offset instead.

Signed-off-by: Yang Zhang <yang.z.zhang@Intel.com>
Committed-by: Keir Fraser <keir@xen.org>

x86: Use deep C states for off-lined CPUs

Currently when a core is taken off-line it is placed in C1 state
(unless MONITOR/MWAIT is used). This patch allows a core to go to
deeper C states resulting in significantly higher power savings.

Signed-off-by: Boris Ostrovsky <boris.ostrovsky@amd.com>
Committed-by: Keir Fraser <keir@xen.org>

Add gtags target for xen/Makefile. Also update .hgignore.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>

x86/xenoprof: fix 32-bit guest stack handling after c/s 24537:3c0a533d3af0

32-bit guests don't have 64-bit precudrure return addresses - both
elements of struct frame_head_32bit should be 32 bits wide, not just
the frame link pointer.

Further, consolidate the whole handling here (also in the native size
guest case) to properly use guest handles and guest memory accessors.

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Keir Fraser <keir@xen.org>

x86/cpuidle: deny access to the I/O port used for EM_SYSIO

Nothing, not even Dom0, should fiddle with this.

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Keir Fraser <keir@xen.org>

x86/cpuidle: restrict scope of mwait_ptr in acpi_dead_idle()

... just to make sure it doesn't get used improperly (resulting from
the discussion around what became c/s 24968:8964c223836c).

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Keir Fraser <keir@xen.org>

x86: Use deep C states for off-lined CPUs

Currently when a core is taken off-line it is placed in C1 state (unless
MONITOR/MWAIT is used). This patch allows a core to go to deeper C states
resulting in significantly higher power savings.

Signed-off-by: Boris Ostrovsky <boris.ostrovsky@amd.com>
Committed-by: Jan Beulich <jbeulich@suse.com>

libxl: only set b_info->u.hvm.bios if type if HVM

This fixes a PV guest breakage causes by 24932:6eac4a1ee646 "libxl: Select BIOS
using libxl_domain_build_info_setdefaults".

I have also included a sanity check on b_info->type.

Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Committed-by: Ian Campbell <ian.campbell@citrix.com>

QEMU_TAG update

add new bootloader xenpvnetboot for pv guest

`xenpvnetboot' supports getting boot images from following locations:

 - http://host/path
 - https://host/path
 - ftp://host/path
 - file:///path
 - tftp://host/path
 - nfs:host:/path
 - /path
 - /path/file.iso
 - /path/filesystem.img
 - /dev/sda1
 - nfs+iso:host:/path/file.iso
 - nfs+iso:host:/path/filesystem.img

To use it, make `xenpvnetboot' as bootloader for PV guest::

    bootloader = '/usr/bin/xenpvnetboot'

To get boot images from various locations, set the right bootloader
arguments, e.g.:

    bootloarder_args = ['--location=http://192.168.0.1/fedora/x86_64']
    bootloarder_args = ['--location=ftp://192.168.0.1/fedora/x86_64']
    bootloarder_args = ['--location=file:///fedora/x86_64']
    bootloarder_args = ['--location=tftp://192.168.0.1/fedora/x86_64']
    bootloarder_args = ['--location=/fedora/x86_64']
    bootloarder_args = ['--location=/fedora/Fedora-16-x86_64-DVD.iso']
    bootloarder_args = ['--location=nfs:192.168.0.1:/fedora/x86_64']
    bootloarder_args = ['--location=nfs+iso:192.168.0.1:/fedora/Fedora-16-x86_64-DVD.iso']

You can use `kernel' and `ramdisk' to specify the relative path of
boot kernel and ramdisk. `xenpvnetboot' will join them with the
location to find the boot kernel and ramdisk, e.g.:

    kernel = 'images/pxeboot/vmlinuz'
    ramdisk = 'images/pxeboot/initrd.img'
    bootloarder_args = ['--location=http://192.168.0.1/fedora/x86_64']

    kernel = 'fedora/x86_64/images/pxeboot/vmlinuz'
    ramdisk = 'fedora/x86_64/images/pxeboot/initrd.img'
    bootloarder_args = ['--location=http://192.168.0.1/']

You can also omit the `--location' option and specify the full URL for
`kernel' and `ramdisk' directly, e.g.:

    kernel = 'http://192.168.0.1/fedora/x86_64/images/pxeboot/vmlinuz'
    ramdisk = 'http://192.168.0.1/fedora/x86_64/images/pxeboot/initrd.img'

If only `--location' is specified and `kernel' and `ramdisk' are not
specified, `xenpvnetboot' will search the following places for boot
images from the location::

    ('images/xen/vmlinuz', 'images/xen/initrd.img'), # Fedora <= 10 and OL = 5
    ('boot/i386/vmlinuz-xen', 'boot/i386/initrd-xen'), # openSUSE >= 10.2 and SLES >= 10
    ('boot/x86_64/vmlinuz-xen', 'boot/x86_64/initrd-xen'), # openSUSE >= 10.2 and SLES >= 10
    ('current/images/netboot/xen/vmlinuz', 'current/images/netboot/xen/initrd.gz'), # Debian
    ('images/pxeboot/vmlinuz', 'images/pxeboot/initrd.img'), # Fedora >=10 and OL >= 6
    ('isolinux/vmlinuz', 'isolinux/initrd.img'), # Fedora >= 10 and OL >= 6

`xenpvnetboot' requires python module `urlgrabber'
http://urlgrabber.baseurl.org/.

Signed-off-by: Zhigang Wang <zhigang.x.wang@oracle.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>

xenstore: Portability: do not build init-xenstore-domain if stubdom disabled

Build stubxenstore only when building stubdomains.
Fixes build failure on platforms w/o the relevant ioctl.

Signed-off-by: Christoph Egger <Christoph.Egger@amd.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>

libxl: Implement libxl_sched_credit_param_[gs]et

Implement functions to set credit scheduler global parameters.

Signed-off-by: George Dunlap <george.dunlap@eu.citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>

arm: compile libxc

Introduce an empty implementation of the arch specific ARM functions in
xc_core_arm.c and xc_core_arm.h; define barriers on ARM.

Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Committed-by: Ian Jackson <ian.jackson.citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>

arm: add stub hvm/save.h

From: Ian Campbell <ian.campbell@citrix.com>

Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Committed-by: Ian Jackson <ian.jackson.citrix.com>
Acked-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>

tools: fix python version checking issue

Even if python version is 2.4.3 which is newer than the required
version 2.3, the configure script still raises a version issue.  I
tested my patch with python 2.6.6 and 2.4.3. It will fix a syntax
error like the following.

 checking for python version >= 2.3 ... Traceback (most recent call last):
   File "<string>", line 1, in ?
 TypeError: 'str' object is not callable
 no
 configure: error: Python 2.4.3 is too old, minimum required version is 2.3

Signed-off-by: Yongjie Ren <yongjie.ren@intel.com>
Acked-by: Roger Pau Monne <roger.pau@entel.upc.edu>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>

build: autoconf: fix libdir detection

If user specifies a libdir it is used, if no libdir is specified
configure checks if $exec_prefix/lib64 is a directory and uses that,
if not lib is used.

$prefix is set by passing the command line option or by default when
calling AC_OUTPUT, but since AC_OUTPUT is called at the end, this is
not really helpful, so we have to set $exec_prefix manually to the
correct value, either $prefix if different than NONE or
$ac_default_prefix.

Signed-off-by: Roger Pau Monne <roger.pau@entel.upc.edu>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>

docs: xen-headers: Remove hard tabs

Remove hard tabs from the Perl script.

Signed-off-by: Ian Jackson <Ian.Jackson@eu.citrix.com>
Cc: Ian Campbell <Ian.Campbell@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Committed-by: Ian Jackson <Ian.Jackson@eu.citrix.com>

docs: xen-headers: Annotate typedefs

Parse (some) typedef statements.  (Make type names links to the
underlying struct if there is one, rather than to just the typedef.)

Also, exclude the new arch-arm headers.

Signed-off-by: Ian Jackson <Ian.Jackson@eu.citrix.com>
Cc: Ian Campbell <Ian.Campbell@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Committed-by: Ian Jackson <Ian.Jackson@eu.citrix.com>

Merge

lowmemd: Fix build when xs.h has not been previously installed.

Signed-off-by: Keir Fraser <keir@xen.org>

build: autoconf: remove (yet another) brctl leftover

Signed-off-by: Roger Pau Monne <roger.pau@entel.upc.edu>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>

Lowmemd: Simple demo code to show use of VIRQ_ENOMEM

Signed-off-by: Andres Lagar-Cavilla <andres@lagarcavilla.org>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>