Add support for handling PMU interrupts for PV(H) guests.
VPMU for the interrupted VCPU is unloaded until the guest issues XENPMU_flush
hypercall. This allows the guest to access PMU MSR values that are stored in
VPMU context which is shared between hypervisor and domain, thus avoiding
traps to hypervisor.
Since the interrupt handler may now force VPMU context save (i.e. set
VPMU_CONTEXT_SAVE flag) we need to make changes to amd_vpmu_save() which
until now expected this flag to be set only when the counters were stopped.
Signed-off-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Acked-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Acked-by: Kevin Tian <kevin.tian@intel.com>
Reviewed-by: Dietmar Hahn <dietmar.hahn@ts.fujitsu.com>
CentOS7 complains that 'ret' might be unused, and indeed this is the case for
`xl psr-hwinfo --cat`.
The logic for selecting which information to print was rather awkward.
Introduce a new 'all' which default to true, and is cleared if specific
options are selected. This allows for a far more clear logic when choosing
whether to print information or not.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
CC: Ian Campbell <Ian.Campbell@citrix.com>
CC: Ian Jackson <Ian.Jackson@eu.citrix.com>
CC: Wei Liu <wei.liu2@citrix.com>
CC: Chao Peng <chao.p.peng@linux.intel.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Tested-by: Chao Peng <chao.p.peng@linux.intel.com>
A number of fields of PMU control MSRs are defined as Reserved. AMD
documentation requires that such fields are preserved when the register
is written by software.
Add checks to amd_vpmu_do_wrmsr() to make sure that guests don't attempt
to modify those bits.
Signed-off-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Reviewed-by: Aravind Gopalakrishnan <Aravind.Gopalakrishnan@amd.com>
No need to compute those masks on every MSR access.
Also, when checking MSR_P6_EVNTSELx registers make sure that bit 21
(which is a reserved bit) is not set.
Signed-off-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Acked-by: Kevin Tian <kevin.tian@intel.com>
Reviewed-by: Dietmar Hahn <dietmar.hahn@ts.fujitsu.com>
Intercept accesses to PMU MSRs and process them in VPMU module. If vpmu ops
for VCPU are not initialized (which is the case, for example, for PV guests that
are not "VPMU-enlightened") access to MSRs will return failure.
Dump VPMU state for all domains (HVM and PV) when requested.
Signed-off-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Kevin Tian <kevin.tian@intel.com>
Reviewed-by: Dietmar Hahn <dietmar.hahn@ts.fujitsu.com>
Tested-by: Dietmar Hahn <dietmar.hahn@ts.fujitsu.com>
Hypervisor cannot easily inject faults into PV guests from arch-specific VPMU
read/write MSR handlers (unlike it is in the case of HVM guests).
With this patch vpmu_do_msr() will return an error code to indicate whether an
error was encountered during MSR processing (instead of stating that the access
was to a VPMU register). The caller will then decide how to deal with the error.
As part of this patch we also check for validity of certain MSR accesses right
when we determine which register is being written, as opposed to postponing this
until later.
Signed-off-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Acked-by: Kevin Tian <kevin.tian@intel.com>
Reviewed-by: Dietmar Hahn <dietmar.hahn@ts.fujitsu.com>
Save VPMU state during context switch for both HVM and PV(H) guests.
A subsequent patch ("x86/VPMU: NMI-based VPMU support") will make it possible
for vpmu_switch_to() to call vmx_vmcs_try_enter()->vcpu_pause() which needs
is_running to be correctly set/cleared. To prepare for that, call context_saved()
before vpmu_switch_to() is executed. (Note that while this change could have
been dalayed until that later patch, the changes are harmless to existing code
and so we do it here)
Signed-off-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Dietmar Hahn <dietmar.hahn@ts.fujitsu.com>
Add runtime interface for setting PMU mode and flags. Three main modes are
provided:
* XENPMU_MODE_OFF: PMU is not virtualized
* XENPMU_MODE_SELF: Guests can access PMU MSRs and receive PMU interrupts.
* XENPMU_MODE_HV: Same as XENPMU_MODE_SELF for non-proviledged guests, dom0
can profile itself and the hypervisor.
Note that PMU modes are different from what can be provided at Xen's boot line
with 'vpmu' argument. An 'off' (or '0') value is equivalent to XENPMU_MODE_OFF.
Any other value, on the other hand, will cause VPMU mode to be set to
XENPMU_MODE_SELF during boot.
For feature flags only Intel's BTS is currently supported.
Mode and flags are set via HYPERVISOR_xenpmu_op hypercall.
Signed-off-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Acked-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Acked-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Dietmar Hahn <dietmar.hahn@ts.fujitsu.com>
Acked-by: Kevin Tian <kevin.tian@intel.com>
Add pmu.h header files, move various macros and structures that will be
shared between hypervisor and PV guests to it.
Move MSR banks out of architectural PMU structures to allow for larger sizes
in the future. The banks are allocated immediately after the context and
PMU structures store offsets to them.
While making these updates, also:
* Remove unused vpmu_domain() macro from vpmu.h
* Convert msraddr_to_bitpos() into an inline and make it a little faster by
realizing that all Intel's PMU-related MSRs are in the lower MSR range.
Signed-off-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Acked-by: Kevin Tian <kevin.tian@intel.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
* Modify the GICv3 driver to recognize a such device. I wasn't able
to find a register which tell if GICv2 is supported on GICv3. The only
way to find it seems to check if the DT node provides GICC and GICV.
* Disable access to ICC_SRE_EL1 to guest using vGICv2
* The LR is slightly different for vGICv2. The interrupt is always
injected with group0.
* Add a comment explaining why Group1 is used for vGICv3.
Signed-off-by: Julien Grall <julien.grall@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
A platform may have a GIC compatible with previous version of the
device.
This is allow to virtualize an unmodified OS on new hardware if the GIC
is compatible with older version.
When a guest is created, the vGIC will emulate same version as the
hardware. Although, the user can specify in the configuration file the
preferred version (currently only GICv2 and GICv3 are supported).
Signed-off-by: Julien Grall <julien.grall@citrix.com>
Cc: Ian Jackson <ian.jackson@eu.citrix.com>
Cc: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Julien has a patch "arm: Allow the user to specify the GIC version"
which adds an anonymous struct to libxl_domain_build_info:
@ -480,6 +486,11 @@ libxl_domain_build_info = Struct("domain_build_info",[
])),
("invalid", None),
], keyvar_init_val = "LIBXL_DOMAIN_TYPE_INVALID")),
+
+
+ ("arch_arm", Struct(None, [("gic_version", libxl_gic_version),
+ ])),
+
], dir=DIR_IN
)
(libxl_gic_version is an enum). This is not currently supported by the
ocaml genwrap.py. Add a simple pass which handles simple anonymous
unions as top level members of a struct type, but not more deeply
nested since that would be a much more complex change and is not
currently required.
With Juliens patch applied the relevant resulting change to the .mli
is:
--- tools/ocaml/libs/xl/_libxl_BACKUP_types.mli.in 2015-07-08 11:22:35.000000000 +0100
+++ tools/ocaml/libs/xl/_libxl_types.mli.in 2015-07-08 12:25:56.000000000 +0100
@@ -469,6 +477,10 @@ module Domain_build_info : sig
type type__union = Hvm of type_hvm | Pv of type_pv | Invalid
+ type arch_arm__anon = {
+ gic_version : gic_version;
+ }
+
type t =
{
max_vcpus : int;
@@ -510,6 +522,7 @@ module Domain_build_info : sig
ramdisk : string option;
device_tree : string option;
xl_type : type__union;
+ arch_arm : arch_arm__anon;
}
val default : ctx -> ?xl_type:domain_type -> unit -> t
end
The .ml differs similarly. Without Julien's patch there is no change.
gen_struct is refactored slightly to take the indent level as an
argument, since it is now used at a different level.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Cc: Julien Grall <julien.grall@citrix.com>
Cc: Dave Scott <Dave.Scott@citrix.com>
Cc: Rob Hoes <Rob.Hoes@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Acked-by: Rob Hoes <rob.hoes@citrix.com>
Acked-by: David Scott <dave.scott@citrix.com>
Usage:
hdtype=ide|ahci (default=ide)
If hdtype=ahci adds ich9 disk controller in ahci mode and uses it with
upstream qemu to emulate disks instead of ide.
It doesn't support cdroms which still using ide (cdroms will use
"-device ide-cd" as new qemu parameter)
Ahci requires new qemu parameter but for now other emulated disks cases
remains with old ones (I did it in other patch, not needed by this one)
I did it as libxl parameter disabled by default to avoid possible
problems:
- with save/restore/migration (restoring with ahci a domU that was with
ide instead)
- windows < 8 without pv drivers (a registry key change is needed for
AHCI<->IDE change FWIK to avoid possible blue screen)
- windows XP or older that many not support ahci by default.
Setting AHCI with libxl parameter and default to disabled seems the best
solution.
AHCI increase hvm domUs boot performance. On linux hvm domU I saw up to
only 20% of the previous total boot time, whereas boot time decrease a
lot on W7 domUs for most of boots I have done. Small difference in boot
time compared to ide mode on W8 and newer (probably other xen
improvements or fixes are needed not ahci related)
Signed-off-by: Fabio Fantoni <fabio.fantoni@m2r.biz>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
[ ijc -- adjust name of LIBXL_HAVE #define as discussed on list,
fixup pod syntax in xl.cfg.pod.5 ]
Add document to introduce basic concepts and terms in PSR family
technologies and the xl interfaces.
Signed-off-by: Chao Peng <chao.p.peng@linux.intel.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
This is the xc/xl changes to support Intel Cache Allocation
Technology(CAT).
'xl psr-hwinfo' is updated to show CAT info and two new commands
for CAT are introduced:
- xl psr-cat-cbm-set [-s socket] <domain> <cbm>
Set cache capacity bitmasks(CBM) for a domain.
- xl psr-cat-show <domain>
Show CAT domain information.
Examples:
[root@vmm-psr vmm]# xl psr-hwinfo --cat
Cache Allocation Technology (CAT):
Socket ID : 0
L3 Cache : 12288KB
Maximum COS : 15
CBM length : 12
Default CBM : 0xfff
[root@vmm-psr vmm]# xl psr-cat-cbm-set 0 0xff
[root@vmm-psr vmm]# xl psr-cat-show
Socket ID : 0
L3 Cache : 12288KB
Default CBM : 0xfff
ID NAME CBM
0 Domain-0 0xff
Signed-off-by: Chao Peng <chao.p.peng@linux.intel.com>
Reviewed-by: Dario Faggioli <dario.faggioli@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Add libxl_socket_bitmap_alloc() to allow allocating a socket specific
libxl_bitmap (as it is for cpu/node bitmap).
Internal function libxl__count_physical_sockets() is introduced together
to get the socket count when the size of bitmap is not specified.
Signed-off-by: Chao Peng <chao.p.peng@linux.intel.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
By naming, placing and bit shift convention, it could be taken as
implied that MEM_ACCESS_EMULATE and MEM_ACCESS_EMULATE_NOWRITE are
mem_access event specific flags (instead of being generally
applicable as vm_event flags). This patch renames them to
VM_EVENT_FLAG_EMULATE and VM_EVENT_FLAG_EMULATE_NOWRITE
respectively, and uses bit shifts following the rest of the
VM_EVENT_FLAG_ constants.
Signed-off-by: Razvan Cojocaru <rcojocaru@bitdefender.com>
Acked-by: Tamas K Lengyel <tlengyel@novetta.com>
Instead switch to an explicit -include $(XEN_ROOT)/tools/config.h to
pickup config.h.
Most places already do this, fixup the rest.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
As with other in-tree users avoid -L + -l.
This avoids any confusion with versions of these libraries already
installed on the system and the possibility of linking against them by
mistake.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
This will reflect that we effectively emulate the same version as the
hardware GIC for the guest.
Signed-off-by: Julien Grall <julien.grall@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Unlikely that it may seem localtime_r could fail, which would result in a
null pointer dereference. In this case, it shoud log the errno, (instead of
the date/time), and and continue its logging, as this is still useful.
Signed-off-by: Jennifer Herbert <jennifer.herbert@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Various xl sub-commands take additional parameters containing = as
additional config fragments.
The handling of these config fragments has a number of bugs:
1. Use of a static 1024-byte buffer. (If truncation would occur,
with semi-trusted input, a security risk arises due to quotes
being lost.)
2. Mishandling of the return value from snprintf, so that if
truncation occurs, the to-write pointer is updated with the
wanted-to-write length, resulting in stack corruption. (This is
XSA-137.)
3. Clone-and-hack of the code for constructing the appended
config file.
These are fixed here, by introducing a new function
`string_realloc_append' and using it everywhere. The `extra_info'
buffers are replaced by pointers, which start off NULL and are
explicitly freed on all return paths.
The separate variable which will become dom_info.extra_config is
abolished (which involves moving the clearing of dom_info).
Additional bugs I observe, not fixed here:
4. The functions which now call string_realloc_append use ad-hoc
error returns, with multiple calls to `return'. This currently
necessitates multiple new calls to `free'.
5. Many of the paths in xl call exit(-rc) where rc is a libxl status
code. This is a ridiculous exit status `convention'.
6. The loops for handling extra config data are clone-and-hacks.
7. Once the extra config buffer is accumulated, it must be combined
with the appropriate main config file. The code to do this
combining is clone-and-hacked too.
Signed-off-by: Ian Jackson <Ian.Jackson@eu.citrix.com>
Tested-by: Ian Jackson <Ian.Jackson@eu.citrix.com>
Acked-by: Ian Campbell <ian,campbell@citrix.com>
On a busy host, QEMU may take more than 10s to load and start.
This is likely due to a bug in Linux where the I/O subsystem sometime
produce high latency under load and result in QEMU taking a long time to
load every single dynamic libraries.
Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Since XSA-131 qemu-xen now restricts access to PCI cfg by default. In
order to allow local configuration of the existing libxl_device_pci
"permissive" flag needs to be plumbed through via the new QMP property
added by the XSA-131 patches.
Versions of QEMU prior to XSA-131 did not support this permissive
property, so we only pass it if it is true. Older versions only
supported permissive mode.
qemu-xen-traditional already supports the permissive mode setting via
xenstore.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Cc: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Cc: Anthony PERARD <anthony.perard@citrix.com>
Acked-by: George Dunlap <george.dunlap@eu.citrix.com>
Acked-by: Anthony PERARD <anthony.perard@citrix.com>
Acked-by: Wei Liu <wei.liu2@citrix.com>
The blocktap library isn't really necessary; all the necessary functionality
is available via the tap-ctl binary.
To use:
script=block-tap,vdev=[whatever],target=vhd:/path/to/file.vhd
script=block-tap,vdev=[whatever],target=aio:/path/to/file.raw
Signed-off-by: George Dunlap <george.dunlap@eu.citrix.com>
Acked-by: Wei Liu <wei.liu2@citrix.com>
They are no longer needed, having been replaced by a daemon for
driverdomains which will run scripts as necessary.
Worse yet, they seem to be broken for script-based block devices, such
as block-iscsi. This wouldn't matter so much if they were never run
by default; but if you run block-attach without having created a
domain, then the appropriate node to disable running udev scripts will
not have been written yet, and the attach will silently fail.
Rather than try to sort out that issue, just remove them entirely.
Signed-off-by: George Dunlap <george.dunlap@eu.citrix.com>
Acked-by: Wei Liu <wei.liu2@citrix.com>
There are a lot of paths through
libxl__device_disk_local_initiate_attach(), but they all really boil
down to one thing: Can we just access the file directly, or do we need
to attach it?
The requirements for direct access are fairly simple:
* Is this local (as opposed to a driver domain)?
* Is this a raw format (as opposed to cooked)?
* Does this have no scripts associated with it?
If it meets all those requirements, we can access it directly;
otherwise we need to attach it.
This fixes a bug where bootloader execution fails for disks with
hotplug scripts.
This should fix a theoretical bug when using a qdisk backend in a
driver domain. (Not tested.)
Based on a patch by Roger Pau Monne <roger.pau@citrix.com>.
Signed-off-by: George Dunlap <george.dunlap@eu.citrix.com>
Acked-by: Roger Pau Monné <roger.pau@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Boris Ostrovsky
Andrew Cooper
Jan Beulich
Julien Grall
Ian Campbell
Fabio Fantoni
Chao Peng
Dario Faggioli
Razvan Cojocaru
Jennifer Herbert
Ian Jackson
Anthony PERARD
George Dunlap