Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson.citrix.com>
Committed-by: Ian Jackson <ian.jackson.citrix.com>
It has no callers, the only code which adds consoles in internal to libxl and
uses libxl__device_console_add directly.
Rather than worrying about what the public API should look like in this case
simply remove it, adding new APIs is much easier than fixing broken ones...
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson.citrix.com>
Committed-by: Ian Jackson <ian.jackson.citrix.com>
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson.citrix.com>
Committed-by: Ian Jackson <ian.jackson.citrix.com>
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson.citrix.com>
Committed-by: Ian Jackson <ian.jackson.citrix.com>
It's not immediately clear that "l1" iterates over device types and "l2"
iterates over individual devices. Name things in a way which makes this more
obvious.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson.citrix.com>
Committed-by: Ian Jackson <ian.jackson.citrix.com>
Note that the "wait" parameter added to libxl_device_remove is different to the
wait paramter previously used by similar functions. In the past not-wait meant
forced whereas now in means wait for a graceful shutdown, as opposed to setting
off a graceful shutdown but not waiting.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson.citrix.com>
Committed-by: Ian Jackson <ian.jackson.citrix.com>
Use libxl__device in libxl_devices_destroy and
libxl__device_pci_remove_xenstore.
Doing this allows us to use the common functions for removing devices.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson.citrix.com>
Committed-by: Ian Jackson <ian.jackson.citrix.com>
Currently this appears to only include "/local/domain/<domid>/device/suspend".
Ultimately the caller will nuke the whole guest directory anyway but not having
this function remove things which don't look like devices seems less
surprising.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson.citrix.com>
Committed-by: Ian Jackson <ian.jackson.citrix.com>
I'd like to use the from_string functionality...
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson.citrix.com>
Committed-by: Ian Jackson <ian.jackson.citrix.com>
Most forced users can now simply call libxl__device_destroy directly.
libxl__devices_destroy is something of a special case, it is really
just iterating over an opaque set of xenstore directories and removing
them. Until this can be refactored just do the force-remove case
manually, doing otherwise led to too much entanglement with the other
callers of libxl__device_destroy which do know about specific device
types.
For the time being do the same in libxl__device_pci_remove_xenstore.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson.citrix.com>
Committed-by: Ian Jackson <ian.jackson.citrix.com>
The function libxl__device_destroy currently takes a force parameter however:
* in the forced case we initiate a graceful shutdown and then
immediately nuke the backend directory, quite likely before anyone
got a chance to react.
* the callers all have a "wait" variable and pass in "!wait" as the
force argument which is confusing since not waiting is not really
the same thing as forcing the destroy.
Therefore split the function into libxl__device_remove and
libxl__device_destroy. The former initiates a graceful shutdown which
the latter simply nukes the backend directory.
This makes some of the callers look a bit odd but that should fall out as I
continue to pull this piece of string.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson.citrix.com>
Committed-by: Ian Jackson <ian.jackson.citrix.com>
Currently takes a string.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson.citrix.com>
Committed-by: Ian Jackson <ian.jackson.citrix.com>
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson.citrix.com>
Committed-by: Ian Jackson <ian.jackson.citrix.com>
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson.citrix.com>
Committed-by: Ian Jackson <ian.jackson.citrix.com>
libxl_device_nic_list returns an array of libxl_device_nic and
libxl_device_nic_getinfo retrieves further information.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson.citrix.com>
Committed-by: Ian Jackson <ian.jackson.citrix.com>
Instead store a duplicate of the "device-type" node in the backend dir
and use that instead.
This maintains the invariant that the list function is always "safe".
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson.citrix.com>
Committed-by: Ian Jackson <ian.jackson.citrix.com>
Temporarily retain unsafe behaviour of reading f.e. directory.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson.citrix.com>
Committed-by: Ian Jackson <ian.jackson.citrix.com>
- handle realloc errors
- remove redundancy of libxl__append_disk_list_of_type return value
and ndisks paramter.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson.citrix.com>
Committed-by: Ian Jackson <ian.jackson.citrix.com>
Subsequent patches will endeavour to make reality match this defined
interface.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson.citrix.com>
Committed-by: Ian Jackson <ian.jackson.citrix.com>
Now that IDL type destructors are no longer libxl_*_destroy we can use the more
common libxl_NOUN_VERB structure here.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson.citrix.com>
Committed-by: Ian Jackson <ian.jackson.citrix.com>
Destroy is an overloaded term which would commonly like to be used for actual
destructive operations, such as destroying a domain etc.
Dispose isn't a great term but it does the job.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson.citrix.com>
Committed-by: Ian Jackson <ian.jackson.citrix.com>
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Ian Jackson <ian.jackson.citrix.com>
Committed-by: Ian Jackson <ian.jackson.citrix.com>
Fix FPU switching uncovered with Hyper-V: Do FPU switching always on
the n1 vmcb rather doing it on the current active vmcb. The FPU state
between n1 and n2 vmcb is already synced in the vmrun and vmexit
emulation.
This fixes the situation where the n2 vmcb had the #NM intercept bit
set but the virtual vmcb did not. That means it was believed the l1
guest can handle an #NM intercept but was actually wrong.
The result were three #GP's with error code 0x11 in the l1 guest.
Signed-off-by: Christoph Egger <Christoph.Egger@amd.com>
Committed-by: Keir Fraser <keir@xen.org>
pvhvm running with more than 32 vcpus and pv_irq/pv_time enabled
need vcpu placement to work, or else it will softlockup.
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@oracle.com>
Committed-by: Keir Fraser <keir@xen.org>
In order to have two different DSDT tables for rombios and SeaBIOS,
this patch introduce a new parameter to acpi_build_tables() which
contain the DSDT table to load.
Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>
Committed-by: Keir Fraser <keir@xen.org>
The ACPI PIIX4 device in QEMU upstream as not the same behavior to
handle PCI hotplug. This patch introduce the necessary change to the
DSDT ACPI table to behave as expceted by the new QEMU.
To switch to this new DSDT table version, there is a new option
--dm-version to mk_dsdt.
Change are inspired by SeaBIOS DSDT source code.
There is few things missing with the new QEMU:
- QEMU provide the plugged/unplugged status only per slot (and not
per func like qemu-xen-traditionnal.
- I did not include the _STA ACPI method that give the status of a
device (present, functionning properly) because qemu-xen does not
handle it.
- I did not include the _RMV method that say if the device can be
removed,
because the IO port of QEMU that give this status always return
true. In
SeaBIOS table, they have a specific _RMV method for VGA, ISA that
return
false. But I'm not sure that we can do the same in Xen.
So, the only way to remove a device is from outside of the guest (like
with xl pci-detatch), and can not be initiated from inside.
Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>
Committed-by: Keir Fraser <keir@xen.org>
Use __attribute__((format(printf,..))) for the function _stmt to
prevent any mistake.
Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>
Committed-by: Keir Fraser <keir@xen.org>
This patch move the IO port range reservation from the dsdt.asl to
mk_dsdt. This IO port range need to be generated by mk_dsdt, because
qemu-xen use different port. The IO port for qemu-xen will be added in
a later patch.
Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>
Committed-by: Keir Fraser <keir@xen.org>
With this new option, there is no need to compile mk_dsdt for each
DSDT table that we want. The Makefile is a bit reorganize to handle
this new option and to prepare more change in a coming patch.
Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>
asserted
PV on HVM guests can loose level interrupts coming from emulated
devices if they have been remapped onto event channels. The reason is
that we are missing the code to inject a pirq again in the guest when
the guest EOIs it, if it corresponds to an emulated level interrupt
and the interrupt is still asserted.
Fix this issue and also return error when the guest tries to get the
irq_status of a non-existing pirq.
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Committed-by: Keir Fraser <keir@xen.org>
Alter the 'reserved' uint8_t to be used a 'flag'. We use only for one
flag: BLKIF_DISCARD_SECURE.
That flag can only be set if the backend has set 'discard-secure' to
one. If backend has not set 'discard-secure' to one, that flag will
have no effect.
Acked-by: Jan Beulich <JBeulich@suse.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Committed-by: Keir Fraser <keir@xen.org>
We flesh out details on what is expected of 'feature-flush' and
what are some of the extra parameters that the frontend can read
from the backend. Those extra parameters are: : discard-aligment,
and discard-granularity.
Acked-by: Jan Beulich <JBeulich@suse.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Committed-by: Keir Fraser <keir@xen.org>
Just a simple sed s/trim/discard/. We are ignoring the comments
which are incorrect.
The reason for the name change is that TRIM is specific to ATA
while the operation can be done on top of SCSI interfaces too.
Hence the rename to something more generic.
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Committed-by: Keir Fraser <keir@xen.org>
Fedora 16 grub2 configuration file can have lines like
set default="${saved_entry}"
and a string containing an integer is expected
Signed-off-by: Michael Young <m.a.young@durham.ac.uk>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>
The grub2 configuration file in Fedora 16 can have one or more
menuentrys in a submenu, with configuration of the form
submenu "Xen 4.1" {
menuentry ... {
...
}
}
(this example occurs when the xen hypervisor is installed on the
guest)
Ignore the submenu line and the corresponding }
Signed-off-by: Michael Young <m.a.young@durham.ac.uk>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
The grub2 configuration file in Fedora 16 can have GPT partition
references like (hd0,gpt2) so remove the "gpt" string where necessary
Signed-off-by: Michael Young <m.a.young@durham.ac.uk>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Fedora 16 puts grub configuration files in /boot/grub2/grub.cfg so
pygrub should look there as well
Signed-off-by: Michael Young <m.a.young@durham.ac.uk>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
On Fedora 16 the first GPT partition is a boot partition for grub2 with
the grub2 configuration in the second partition.
Check all GPT partitions for grub configuration, not just the first.
[ Also remove now-inaccurate comment. -iwj ]
Signed-off-by: Michael Young <m.a.young@durham.ac.uk>
Tested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Ian Jackson <ian.jackson@eu.citrix.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>
Otherwise we erroneously fall through the LIBXL_DOMAIN_TYPE_PV case into the
"default: abort()".
(I'm sure we fixed this once already...)
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Tested-by: Dario Faggioli <dario.faggioli@citrix.com>
Committed-by: Ian Jackson <ian.jackson@eu.citrix.com>
Move HVM io fields into a structure.
On MMIO instruction failure print out some more bytes.
Signed-off-by: Christoph Egger <Christoph.Egger@amd.com>
Committed-by: Keir Fraser <keir@xen.org>
Sandy bridge introduces new MSR to get cc7/pc2 residency (core C-state
7/package C-state 2). Print the cc7/pc2 residency when on sandy bridge
platform.
Signed-off-by: Yang Zhang <yang.z.zhang@intel.com>
Committed-by: Keir Fraser <keir@xen.org>
hpet_fsb_cap_lookup(), if it doesn't find any FSB capable timer,
leaves hpet_events allocated, while hpet_events->cpumask may not have
been, As we're pretty generous with these one-time allocations already
(in that hpet_events doesn't get freed when no usable counters were
found, even if in that case only the first array entry [or none at
all] may get used), simply make the cpumask allocation in the legacy
case independent of whether hpet_events was NULL before.
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Tested-by: Christoph Egger <Christoph.Egger@amd.com>
Acked-by: Christoph Egger <Christoph.Egger@amd.com>
Committed-by: Keir Fraser <keir@xen.org>
microcode_fits() must return distinct values for the success and
no-fit-but-no-error cases, so the caller can react accordingly. Make
it return 1 in the success case, and adjust its single caller.
Also remove an impossible code path - install_equiv_cpu_table(), which
gets called prior to microcode_fits(), never leaves equiv_cpu_table
being NULL without also returning an error.
Note that this is still awaiting testing on a system where the
regression was actually observed (which also requires a new enough
microcode_ctl package). Note also that this will need to be
backported to 4.0 and 4.1 (or the broken c/s that got backported
there reverted).
Signed-off-by: Jan Beulich <jbeulich@novell.com>
Committed-by: Keir Fraser <keir@xen.org>
Hyper-V starts a root domain which effectively an l2 guest.
Hyper-V passes its devices through to the root domain and
let it do the MMIO accesses. The emulation is done by
Xen (host) and Hyper-V forwards the interrupts to the l2 guest.
Signed-off-by: Christoph Egger <Christoph.Egger@amd.com>
Acked-by: Tim Deegan <tim@xen.org>
Committed-by: Tim Deegan <tim@xen.org>
Generally there was a NR_CPUS-bits wide array in these functions and
another (through a cpumask_t) on their callers' stacks, which may get
a little large for big NR_CPUS. As the functions can fail anyway, do
the allocation in there.
For the x86/MCA case this require a little code restructuring: By using
different CPU mask accessors it was possible to avoid allocating a mask
in the broadcast case. Also, this was the only user that failed to
check the return value of the conversion function (which could have led
to undefined behvior).
Also constify the input parameters of the two functions.
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Keir Fraser <keir@xen.org>
Ian Campbell
Christoph Egger
Zhenzhong Duan
Anthony PERARD
Stefano Stabellini
Konrad Rzeszutek Wilk
Michael Young
Yang Zhang
Jan Beulich
Tim Deegan
Jan Beulich