Tim Deegan [Fri, 1 Jun 2012 09:20:37 +0000 (10:20 +0100)]
arm: Move hyp-mode entry code out of line.
This code is grottier than the rest of the start-of-day code and
very specific to the software model we're developing on.
Segregate it accordingly, by putting it in its own file.
Signed-off-by: Tim Deegan <tim@xen.org> Acked-by: Ian Campbell <ian.campbell@citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
Implement vcpu_mark_events_pending using the vgic to inject PPI 31, that
we reserve for Xen usage.
In the future the interrupt used for event injection might be dynamic
and could be written into the device tree.
Otherwise it could be an SGI choosen by the guest and passed to Xen
through an hypercall.
Considering that:
- it is easy to determine if an event notification
interrupt has already been EOI'd by the guest just looking at the
evtchn_upcall_pending bit in the shared_info page;
- we can safely assume that there is at most one event notification
interrupt pending at any time in any set of LR registers because we
never inject more than a single event notification interrupt in one vcpu
(see vcpu_mark_events_pending);
we can avoid requesting maintenance interrupts for
VGIC_IRQ_EVTCHN_CALLBACK, provided that we check for event notification
interrupts that need to be cleared in the following places:
- maintenance interrupt entry;
- gic_set_guest_irq;
that is every time we are about to write to an LR.
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Acked-by: Ian Campbell <ian.campbell@citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
Instead of using a softirq to check whether we need to set the VI bit in
the HCR (IRQ injection in the guest), always check the lr_mask on
leave_hypervisor_tail.
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Acked-by: Ian Campbell <ian.campbell@citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
If the vgic needs to inject a virtual irq into the guest, but no free
LR registers are available, add the irq to a list and return.
Whenever an LR register becomes available we add the queued irq to it
and remove it from the list.
We use the gic lock to protect the list and the bitmask.
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Acked-by: Ian Campbell <ian.campbell@citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
Andreas Herrmann [Thu, 31 May 2012 08:15:42 +0000 (10:15 +0200)]
x86/amd: re-enable CPU topology extensions in case BIOS has disabled it
BIOS will switch off the corresponding feature flag on family
15h models 10h-1fh non-desktop CPUs.
The topology extension CPUID leafs are required to detect which
cores belong to the same compute unit. (thread siblings mask is
set accordingly and also correct information about L1i and L2
cache sharing depends on this).
W/o this patch we wouldn't see which cores belong to the same
compute unit and also cache sharing information for L1i and L2
would be incorrect on such systems.
Signed-off-by: Andreas Herrmann <andreas.herrmann3@amd.com> Signed-off-by: Jan Beulich <jbeulich@suse.com> Acked-by: Keir Fraser <keir@xen.org> Committed-by: Jan Beulich <jbeulich@suse.com>
Keir Fraser [Wed, 30 May 2012 08:27:51 +0000 (09:27 +0100)]
xen: Define new struct hvm_trap and cleanup vmx exception
Define new struct hvm_trap to represent information of trap, and
renames hvm_inject_exception to hvm_inject_trap, then define a couple
of wrappers around that function for existing callers.
Boris Ostrovsky [Wed, 30 May 2012 08:26:02 +0000 (09:26 +0100)]
xenpm: Fix reporting of C0 residence times
Idle state residence times as provided by pmstat_get_cx_stat() are not
reported precisely since remote core may be in idle state and
therefore has not updated its statistics at the time local core
collected them. This causes C0 residencies as calculated by xenpm to
sometimes become negative.
Signed-off-by: Boris Ostrovsky <boris.ostrovsky@amd.com> Committed-by: Keir Fraser <keir@xen.org>
Jan Beulich [Wed, 30 May 2012 08:23:33 +0000 (09:23 +0100)]
gnttab: cleanup
- introduce local variables (shortcuts for frequently used
<dom>->grant_table)
- adjust first parameter of mapcount()
- drop lock acquisition from gnttab_get_version()
- remove hard tabs and adjust formatting
Signed-off-by: Jan Beulich <jbeulich@suse.com> Tested-by: Andrew Thomas <andrew.thomas@oracle.com> Committed-by: Keir Fraser <keir@xen.org>
Jan Beulich [Wed, 30 May 2012 08:22:17 +0000 (09:22 +0100)]
gnttab: don't use domain lock for serialization
Instead use the affected domain's grant table lock, at once reducing
the scopes during which locks are being held and hence allowing
significantly better parallelism.
Signed-off-by: Jan Beulich <jbeulich@suse.com> Tested-by: Andrew Thomas <andrew.thomas@oracle.com> Committed-by: Keir Fraser <keir@xen.org>
Jan Beulich [Wed, 30 May 2012 08:21:04 +0000 (09:21 +0100)]
x86: fix nested HVM initialization
- no need for calling nestedhvm_setup() explicitly (can be a normal
init-call, and can be __init)
- calling _xmalloc() for multi-page, page-aligned memory regions is
inefficient - use alloc_xenheap_pages() instead
- albeit an allocation error is unlikely here, add error handling
nevertheless (and have nestedhvm_vcpu_initialise() bail if an error
occurred during setup)
- nestedhvm_enabled() must no access d->arch.hvm_domain without first
checking that 'd' actually represents a HVM domain
Signed-off-by: Jan Beulich <JBeulich@suse.com> Committed-by: Keir Fraser <keir@xen.org>
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Acked-by: Ian Jackson <Ian.Jackson@eu.citrix.com> Acked-by: Ian Campbell <Ian.Campbell@citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
libxl: move e820_names, e820_sanitize, libxl__e820_alloc to libxl_x86.c
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Acked-by: Ian Jackson <Ian.Jackson@eu.citrix.com> Acked-by: Ian Campbell <Ian.Campbell@citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Acked-by: Ian Jackson <Ian.Jackson@eu.citrix.com> Acked-by: Ian Campbell <Ian.Campbell@citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
Introduce an arch specific internal domain creation function.
The X86 version of libxl__arch_domain_create is going to be introduced
in the next few patches, to make it easier to spot all the code motions.
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Acked-by: Ian Jackson <Ian.Jackson@eu.citrix.com> Acked-by: Ian Campbell <Ian.Campbell@citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
libxl_cpuid_destroy has been renamed to libxl_cpuid_dispose; also cpuid
functions are only available on x86, so move them to libxl_cpuid.c.
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Acked-by: Ian Campbell <ian.campbell@citrix.com> Acked-by: Ian Jackson <ian.jackson@eu.citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Acked-by: Ian Campbell <ian.campbell@citrix.com> Acked-by: Ian Jackson <ian.jackson@eu.citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Acked-by: Ian Campbell <Ian.campbell@citrix.com> Acked-by: Ian Jackson <ian.jackson@eu.citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
Introduce an empty implementation of the arch specific ARM functions in
xc_dom_arm.c.
Provide empty implementations of xc_domain_save and xc_domain_restore
when CONFIG_MIGRATE is not set.
Move xc_hvm_build.c to xc_hvm_build_x86.c because the implementation is
x86 specific, introduce xc_hvm_build_arm.c with empty stubs.
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Acked-by: Ian Jackson <Ian.Jackson@eu.citrix.com> Acked-by: Ian Campbell <Ian.Campbell@citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
Ian Campbell [Tue, 29 May 2012 15:44:06 +0000 (16:44 +0100)]
libxl: remove lockdir and config dir from the API
These are only used by xl.
Rename _libxl_paths.h -> _paths.h, these are not actually "libxl" paths but
rather are part of the Xen build time configuration. It is fine for xl to also
consume them.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com> Acked-by: Ian Jackson <ian.jackson@eu.citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
xl: destroy the disk on successful removal during block-detach
main_blockdetach needs to call libxl_device_disk_destroy so that all the
disk related entries are properly removed from xenstore.
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Acked-by: Ian Jackson <ian.jackson@eu.citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Acked-by: Ian Jackson <ian.jackson@eu.citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
- Spawn a QEMU instance at boot time to handle disk local attach
requests.
- Implement libxl__device_disk_local_attach for QDISKs in terms of a
regular disk attach with the frontend and backend running in the same
domain.
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Acked-by: Ian Campbell <ian.campbell@citrix.com> Acked-by: Ian Jackson <ian.jackson@eu.citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
Introduce libxl__alloc_vdev: find a spare virtual block device in the
domain passed as argument.
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Acked-by: Ian Jackson <ian.jackson@eu.citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
Introduce a blkdev_start in xl.conf and a corresponding string in
libxl_domain_build_info.
Add a blkdev_start parameter to libxl__device_disk_local_attach: it is
going to be used in a following patch.
blkdev_start specifies the first block device to be used for temporary
block device allocations by the toolstack.
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Acked-by: Ian Jackson <ian.jackson@eu.citrix.com> Acked-by: Ian Campbell <ian.campbell@citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
Introduce libxl__device_disk_add that takes an additional
xs_transaction_t paramter.
Implement libxl_device_disk_add using libxl__device_disk_add.
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Acked-by: Ian Jackson <ian.jackson@eu.citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Acked-by: Ian Jackson <ian.jackson@eu.citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
libxl: add a transaction parameter to libxl__device_generic_add
Add a xs_transaction_t parameter to libxl__device_generic_add, if it is
XBT_NULL, allocate a proper one.
Update all the callers.
This patch contains a large number of unchecked xenstore operations, we
might want to fix this in the future.
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Acked-by: Ian Jackson <ian.jackson@eu.citrix.com> Acked-by: Ian Campbell <ian.campbell@citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
libxl: libxl__device_disk_local_attach return a new libxl_device_disk
Introduce a new libxl_device_disk* parameter to
libxl__device_disk_local_attach, the parameter is allocated by the
caller. libxl__device_disk_local_attach is going to fill the new disk
with informations about the new locally attached disk. The new
libxl_device_disk should be passed to libxl__device_disk_local_detach
afterwards.
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Acked-by: Ian Jackson <ian.jackson@eu.citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
libxl: make libxl_device_disk_local_attach/detach internal functions
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Acked-by: Ian Campbell <ian.campbell@citrix.com> Acked-by: Ian Jackson <ian.jackson@eu.citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
Christoph Egger [Tue, 29 May 2012 10:12:45 +0000 (11:12 +0100)]
xenstore: fix crash on platforms with no gntdev driver implementation.
Fix pointer checks introduced in changeset 24757:aae516b78fce.
Signed-off-by: Christoph Egger <Christoph.Egger@amd.com> Acked-by: Ian Campbell <ian.campbell@citrix.com> Acked-by: Ian Jackson <ian.jackson@eu.citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
Ian Campbell [Tue, 29 May 2012 09:31:39 +0000 (10:31 +0100)]
xl: remove all local "ctx" variables
xl has a global "ctx" variable, so there should be no need to pass a
ctx to any function.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com> Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
[ ijc -- dropped xl_fork hunks on commit, this aspect was already done in
xl: track child processes for the benefit of libxl ] Committed-by: Ian Campbell <ian.campbell@citrix.com>
Ian Campbell [Tue, 29 May 2012 09:31:38 +0000 (10:31 +0100)]
libxl: do not overwrite user supplied config when running bootloader
Currently when running the bootloader libxl will update b_info->u.pv.kernel,
.ramdisk, .cmdline and .bootloader. This can expose internal details, such
as temporary paths in /var/run/xen/bootloader.*/ to the user. This is
problematic because it means that the user cannot re-use the struct as is.
This does not effect xl in Xen 4.2+ since it always reparses the guest config
and reinitialises the build info, however it did cause issues with reboot in
4.1 (reported by Dmitry Morozhnikov) and may cause issues for other users of
libxl.
Instead make the libxl bootloader infrastructure provide output to its caller
which is slurped into the internal libxl__domain_build_state datastructure. If
no bootloader is configured then the bootloader instead propagates the user
supplied b_info config.
In order to simplify this push the error handling for the case where there is
no bootdisk down into libxl__bootloader_run. In principal there is no reason
why it shouldn't be possible to do a pure netboot guest with a suitable
bootloader, but I don't fix that here.
This change allow us to make the libxl_file_reference an internal API, and
eventually we might be able to get rid of it.
Also removes the publix libxl_run_bootloader interface, neither xl nor libvirt
use it.
I am proposing this for 4.2 due to the API change.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com> Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
[ ijc -- reduced log message in libxl__build_pv from INFO to DEBUG ] Committed-by: Ian Campbell <ian.campbell@citrix.com>
Ian Jackson [Tue, 29 May 2012 09:31:37 +0000 (10:31 +0100)]
xl: track child processes for the benefit of libxl
Each time xl forks, it needs to record the pid, so that its exit
status can be preserved if it happens that libxl's event loop reaped
it. Consequently we also have a new wrapper for waitpid, which in
that case returns the previously-reaped status.
When we get a console ready callback from libxl, check to see if we
have spawned but not reaped a previous console client, and if so reap
it now. (This is necessary to prevent improper use of the xlchild
struct, but has the happy consequence of checking the exit status of
the first console client in the pygrub case.)
Refactor the two calls to libxl_ctx_alloc into a new function
xl_ctx_alloc which also sets the child reaped handler callback.
All of this has the effect of suppressing a message
unknown child [nnnn] unexpected exited status zero
which would sometimes (depending on a race) appear with `xl create -c'
and pygrub.
Signed-off-by: Ian Jackson <ian.jackson@eu.citrix.com> Reported-by: Ian Campbell <ian.campbell@citrix.com> Acked-by: Ian Campbell <ian.campbell@citrix.com> Cc: Roger Pau Monne <roger.pau@citrix.com>
[ ijc -- corrected return codes in xl_reaped_callback to match documented
convention. ] Committed-by: Ian Campbell <ian.campbell@citrix.com>
George Dunlap [Tue, 29 May 2012 09:31:36 +0000 (10:31 +0100)]
libxl: When checking BDF of existing slots, function should be decimal, not hex
Signed-off-by: George Dunlap <george.dunlap@eu.citrix.com> Spotted-by: Konrad Wilk <konrad.wilk@oracle.com> Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
xen: domain_pirq_to_emuirq return IRQ_UNBOUND by default
domain_pirq_to_emuirq should return IRQ_UNBOUND rather than 0 on
missing entries. Add a default parameter to pirq_field, so that
callers can set any default return value they want; use IRQ_UNBOUND in
domain_pirq_to_emuirq.
This patch fixes a regression introduced by 23573: save/restore
failing on upstream QEMU with PV on HVM guests.
Jan Beulich [Wed, 23 May 2012 15:20:18 +0000 (17:20 +0200)]
amd iommu: improve parse_event_log_entry()
- message table should be static (no need to set it up each time the
function gets executed)
- don't bail on out-of-range event code values
- use message table also to print the kind of otherwise unhandled
event codes
Signed-off-by: Jan Beulich <jbeulich@suse.com> Acked-by: Wei Wang <wei.wang2@amd.com>
Roger Pau Monne [Wed, 23 May 2012 10:06:49 +0000 (11:06 +0100)]
libxl: reorder libxl_device unplug functions
This is a reorder of functions, no functional change. This is needed
because in future patches much code is added to libxl_device and it
needs to follow the usual ao operation scheme (prototypes, functions
and callbacks in order they should be called)
Signed-off-by: Roger Pau Monne <roger.pau@citrix.com> Acked-by: Ian Jackson <ian.jackson@eu.citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
Roger Pau Monne [Wed, 23 May 2012 10:06:48 +0000 (11:06 +0100)]
libxl: add libxl__xs_path_cleanup
Add a function which behaves like "xenstore-rm -t", and which will be
used to clean xenstore after unplug since we will be no longer
executing xen-hotplug-cleanup script, that used to do that for us.
Signed-off-by: Roger Pau Monne <roger.pau@citrix.com> Acked-by: Ian Jackson <ian.jackson@eu.citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
Roger Pau Monne [Wed, 23 May 2012 10:06:47 +0000 (11:06 +0100)]
libxl: fix libxl__xs_directory usage of transaction
libxl__xs_directory takes a transaction parameter, but completely
ignores it, passing XBT_NULL unconditionally to xs_directory.
Signed-off-by: Roger Pau Monne <roger.pau@citrix.com> Acked-by: Ian Jackson <ian.jackson@eu.citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
Roger Pau Monne [Wed, 23 May 2012 10:06:47 +0000 (11:06 +0100)]
libxl: pass env vars to libxl__exec
Add another parameter to libxl__exec call that contains the
environment variables to use when performing the execvp call.
Signed-off-by: Roger Pau Monne <roger.pau@citrix.com> Acked-by: Ian Jackson <ian.jackson@eu.citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
Jan Beulich [Tue, 22 May 2012 14:30:11 +0000 (16:30 +0200)]
x86: don't hold off NMI delivery when MCE is masked
Likely through copy'n'paste, all three instances of guest MCE
processing jumped to the wrong place (where NMI processing code
correctly jumps to) when MCE-s are temporarily masked (due to one
currently being processed by the guest). A nested, unmasked NMI should
get delivered immediately, however.
Signed-off-by: Jan Beulich <jbeulich@suse.com> Acked-by: Keir Fraser <keir@xen.org>
Andrew Cooper [Tue, 22 May 2012 14:28:45 +0000 (16:28 +0200)]
x86: prevent call to xfree() in dump_irqs() while in an irq context
Because of c/s 24707:96987c324a4f, dump_irqs() can now be called in an
irq context when a bug condition is encountered. If this is the case,
ignore the call to xsm_show_irq_ssid() and the subsequent call to
xfree().
This prevents an assertion failure in xfree(), and should allow all the
debug information to be dumped, before failing with a BUG() because of
the underlying race condition we are attempting to reproduce.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Rather than using the non-obvious conditional around an xfree() that
would be passed NULL only in the inverse case (which could easily get
removed by a future change on the basis that calling xfree(NULL) is
benign), switch the order of checks in xfree() itself and only suppress
the call to XSM that could potentially call xmalloc().
Signed-off-by: Jan Beulich <jbeulich@suse.com> Acked-by: Keir Fraser <keir@xen.org> Committed-by: Jan Beulich <jbeulich@suse.com>
Ian Jackson [Tue, 22 May 2012 10:55:01 +0000 (11:55 +0100)]
libxl: events: improve debugging output for xs watches
* Add debugging output for register/deregister.
* Make the debugging printfs consistent about the order in which they
print the information.
* Where we touch the code, change LIBXL__LOG to LOG.
Signed-off-by: Ian Jackson <ian.jackson@eu.citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
Jan Beulich [Mon, 21 May 2012 10:03:32 +0000 (12:03 +0200)]
x86: don't call generic_identify() redundantly
Right before calling struct cpu_dev's ->c_identify, if non-NULL,
identify_cpu() calls generic_identify(). Hence there's no point for
->c_identify to point to generic_identify, nor for the handler to call
that function. After removing all pointless uses, the function isn't
being used outside the file that's defininig it anymore, and hence can
become static.
Signed-off-by: Jan Beulich <jbeulich@suse.com> Acked-by: Keir Fraser <keir@xen.org>
Ian Campbell [Fri, 18 May 2012 12:40:00 +0000 (13:40 +0100)]
libxl: initialise ao when starting pvqemu for stubdomain
libxl__spawn_local_dm requires the ao to be initialised.
Without this starting an HVM guest with a stub qemu hits the
assert(ao->magic == LIBXL__AO_MAGIC);
in the STATE_AO_GC call from libxl__spawn_local_dm.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com> Acked-by: Ian Jackson <Ian.Jackson@eu.citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
Ian Campbell [Fri, 18 May 2012 12:39:59 +0000 (13:39 +0100)]
libxl: avoid double free of b_info->u.pv.bootloader
b_info is a user provided struct and therefore the content must come from
malloc and not gc such that libxl_domain_build_info_dispose can free it. This
was broken by 25340:373f24c87dee.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com> Acked-by: Ian Jackson <Ian.Jackson@eu.citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
xl remus acts as a frontend to enable remus for a given domain.
* At the moment, only memory checkpointing and blackhole replication is
supported. Support for disk checkpointing and network buffering will
be added in future.
* Replication is done over ssh connection currently (like live migration
with xl). Future versions will have an option to use simple tcp socket
based replication channel (for both Remus & live migration).
Signed-off-by: Shriram Rajagopalan <rshriram@cs.ubc.ca> Acked-by: Ian Campbell <ian.campbell@citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
* Add libxl callback functions for Remus checkpoint suspend, postflush
(aka resume) and checkpoint commit callbacks.
* suspend callback is a stub that just bounces off
libxl__domain_suspend_common_callback - which suspends the domain and
saves the devices model state to a file.
* resume callback currently just resumes the domain (and the device model).
* commit callback just writes out the saved device model state to the
network and sleeps for the checkpoint interval.
* Introduce a new public API, libxl_domain_remus_start (currently a stub)
that sets up the network and disk buffer and initiates continuous
checkpointing.
* Future patches will augment these callbacks/functions with more functionalities
like issuing network buffer plug/unplug commands, disk checkpoint commands, etc.
Signed-off-by: Shriram Rajagopalan <rshriram@cs.ubc.ca> Acked-by: Ian Campbell <ian.campbell@citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
The guest is "suspended" via libxl_domain_suspend when taking a snapshot.
So call libxl_domain_resume instead of libxl_domain_unpause, when taking
a checkpoint of the domain (using xl save -c).
Signed-off-by: Shriram Rajagopalan <rshriram@cs.ubc.ca> Acked-by: Ian Campbell <ian.campbell@citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
libxl: refactor migrate_domain and generalize migrate_receive
Refactor some tasks like establishing the migration channel,
initial migration protocol exchange into separate functions,
to facilitate re-use, when remus support is introduced. Also,
make migrate_receive generic (instead of resorting to stdin and
stdout as the file descriptors for communication).
Signed-off-by: Shriram Rajagopalan <rshriram@cs.ubc.ca> Acked-by: Ian Campbell <ian.campbell@citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
Christoph Egger [Thu, 17 May 2012 13:45:43 +0000 (14:45 +0100)]
libxl: fix build on platforms where openpty's parameters are not const.
Such as NetBSD. Fixes this build error:
libxl_aoutils.c: In function 'libxl__openptys':
libxl_aoutils.c:281:13: error: passing argument 4 of 'openpty' discards
qualifiers from pointer target type
/usr/include/util.h:92:6: note: expected 'struct termios *' but argument
is of type 'const struct termios *'
libxl_aoutils.c:281:13: error: passing argument 5 of 'openpty' discards
qualifiers from pointer target type
/usr/include/util.h:92:6: note: expected 'struct winsize *' but argument
is of type 'const struct winsize *'
Signed-off-by: Christoph Egger <Christoph.Egger@amd.com> Acked-by: Roger Pau Monne <roger.pau@citrix.com> Acked-by: Ian Campbell <ian.campbell@citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
* In libxl_primary_console_exec, if libxl__domain_type fails, do
not abort, but instead log and return an error. This can occur
if the domid is invalid.
* In xl's autoconnect_console, rename the ctx formal parameter so
that when postfork creates a new ctx and puts it in the global ctx,
we don't end up using the old one (via the formal parameter which
has shadowed the global).
* In xl's autoconnect_console, pass the domid from the event
to libxl_primary_console_exec, rather than using the global domid
(which has not yet been set, since it is only set at completion
of the ao).
This causes xl create -c to once more work with pygrub.
Signed-off-by: Ian Jackson <ian.jackson@eu.citrix.com> Acked-by: Ian Campbell <ian.campbell@citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
Ian Jackson [Thu, 17 May 2012 13:45:41 +0000 (14:45 +0100)]
libxl: Do not use-after-free on ao progress reporting
We need to call libxl__free_all after egc_run_callbacks since some of
the callbacks might be ao progress reports allocated from the egc's
gc.
Fixes a segfault in egc_run_callbacks.
Signed-off-by: Ian Jackson <ian.jackson@eu.citrix.com> Acked-by: Ian Campbell <ian.campbell@citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
Ian Jackson [Thu, 17 May 2012 13:45:40 +0000 (14:45 +0100)]
libxl: events: STATE_AO_GC checks ao is still valid
This will catch earlier the mistake where an ao is completed while it
still has events registered: when the event callback happens for the
long-gone ao, we will crash before attempting to execute any of the
operation-specific code.
Signed-off-by: Ian Jackson <ian.jackson@eu.citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
grant-tables: Use get_page_from_gfn() instead of get_gfn()/put_gfn.
This requires some careful re-engineering of __get_paged_frame and its callers.
Functions that previously returned gfn's to be put now return pages to be put.
Tested with Win7 + Citrix PV drivers guest, using speedtest for networking
(yes!) plus the loginVSI framework to constantly hit disk.
Signed-off-by: Andres Lagar-Cavilla <andres@lagarcavilla.org>
Tim Deegan [Thu, 17 May 2012 09:24:54 +0000 (10:24 +0100)]
x86/mm: Introduce get_page_from_gfn().
This new function does a p2m lookup under the read lock, falling back
to the write lock only if it needs to make a change. If the GFN is
backed by RAM, it takes a refcount on the underlying page.
The following patches will convert many paths that currently use
get_gfn/put_gfn to use the new interface. That will avoid serializing
p2m accesses in the common case where no updates are needed (i.e. no
page-sharing, VM paging or other p2m trickery).
George Dunlap [Tue, 15 May 2012 15:28:15 +0000 (16:28 +0100)]
libxl: Introduce pci_assignable_add and pci_assignable_remove
Introduce libxl helper functions to prepare devices to be passed
through to guests. This is meant to replace of all the manual sysfs
commands which are currently required.
pci_assignable_add accepts a BDF for a device and will:
* Unbind a device from its current driver, if any
* If "rebind" is set, it will store the path of the driver from which we
unplugged it in /libxl/pciback/$BDF/driver_path
* If create a slot for it in pciback if one doesn't yet exist
* Bind the device to pciback
At this point it will show up in pci_assignable_list, and is ready to
be passed through to a guest.
pci_assignable_remove accepts a BDF for a device and will:
* Unbind the device from pciback
* Remove the slot from pciback
* If "rebind" is set, and /libx/pciback/$BDF/driver_path exists, it
will attempt to rebind the device to its original driver.
Both functions are idempotent: if the desired end state has already
been reached, they return SUCCESS.
NB that "$BDF" in this case uses '-' instead of ':' and '.', because
':' and '.' are illegal characters in xenstore paths.
Signed-off-by: George Dunlap <george.dunlap@eu.citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
George Dunlap [Tue, 15 May 2012 15:28:15 +0000 (16:28 +0100)]
libxl: Rename pci_list_assignable to pci_assignable_list
...to prepare for a consistent "pci_assignable_*" naming scheme.
Also move the man page entry into the PCI PASS-THROUGH section, rather
than the XEN HOST section.
No functional changes.
Signed-off-by: George Dunlap <george.dunlap@eu.citrix.com> Acked-by: Ian Campbell <ian.campbell@citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
George Dunlap [Tue, 15 May 2012 15:28:14 +0000 (16:28 +0100)]
libxl: Make a helper function write a BDF to a sysfs path
This functionality will be used several times in subsequent patches.
Signed-off-by: George Dunlap <george.dunlap@eu.citrix.com> Acked-by: Ian Campbell <ian.campbell@citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
George Dunlap [Tue, 15 May 2012 15:28:13 +0000 (16:28 +0100)]
xl: Make clear distinction between "filename" and "data source"
Many places in xl there's a variable or element named "filename" which
does not contain a filename, but the source of the data for reporting
purposes. Worse, there are variables which are sometimes actually
used or interpreted as a filename depending on what other variables
are set. This makes it difficult to tell when a string is purely
cosmetic, and when another bit of code may actually attempt to call
"open" with the string.
This patch makes a consistent distinction between "filename" (which
always refers to the name of an actual file, and may be interpreted as
such at some point) and "source" (which may be a filename, or may be
another data source such as a migration stream or saved data).
This does add some variables and reshuffle where assignments happen;
most notably, the "restore_filename" element of struct domain_create
is now only set when restoring from a file.
But at a high level, there should be no funcitonal changes.
Signed-off-by: George Dunlap <george.dunlap@eu.citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
George Dunlap [Tue, 15 May 2012 15:28:11 +0000 (16:28 +0100)]
libxl: Warn that /usr/bin/pygrub is deprecated
Signed-off-by: George Dunlap <george.dunlap@eu.citrix.com> Acked-by: Ian Campbell <ian.campbell@citrix.com> Committed-by: Ian Campbell <ian.campbell@citrix.com>
projects / people / vhanquez / xen-unstable.git / log