I'm not sure what I was smoking at the time of 5d74ad1a08 "xen: arm:
implement do_multicall_call for both 32 and 64-bit" but it is obviously
insufficient since it doesn't actually wire up the hypercall.
Before doing so we need to make the usual adjustments for ARM and turn the
unsigned longs into xen_ulong_t. There is no difference in the resulting
structure for x86.
There are knock on changes to the trace interface, but again they are nops on
x86.
For 32-bit ARM guests we require that the arguments which they pass to a
hypercall via a multicall do not use the upper bits of xen_ulong_t and kill
them if they violate this. This should ensure that no ABI surprises can be
silently lurking when running on a 32-bit hypervisor waiting to pounce when the
same kernel is run on a 64-bit hypervisor. Killing the guest is harsh but it
will be far easier to relax the restriction if it turns out to cause problems
than to tighten it up if we were lax to begin with.
In the interests of clarity and always using explicitly sized types change the
unsigned int in the hypercall arguments to a uint32_t. There is no actual
change here on any platform.
We should consider backporting this to 4.4.1 in case a guest decides they want
to use a multicall in common code e.g. I suggested such a thing while
reviewing a netback change recently.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Cc: keir@xen.org
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Acked-by: George Dunlap <george.dunlap@eu.citrix.com>
Acked-by: Julien Grall <julien.grall@linaro.org>
This patch does the following:
1. Deprecate the HVMOP_[sg]et_mem_access HVM ops.
2. Remove the enums and structs associated with the HVM ops.
Signed-off-by: Aravindh Puthiyaparambil <aravindp@cisco.com>
Acked-by: Tim Deegan <tim@xen.org>
Remove xc_hvm_[sg]et_mem_access() APIs which have been replaced by
xc_[sg]_mem_access() APIs.
Signed-off-by: Aravindh Puthiyaparambil <aravindp@cisco.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Modify the xen-access test program to use the new mem_access APIs.
Signed-off-by: Aravindh Puthiyaparambil <aravindp@cisco.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
This patch does the following:
1. Add mem_access ops under XENMEM_access_ops.
2. Add enums and structs whose names are generic.
Signed-off-by: Aravindh Puthiyaparambil <aravindp@cisco.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Tim Deegan <tim@xen.org>
Newer versions of clang attempt to parse inline assembler even when
not asked to assemble it. Wrap our not-for-assembly runes as strings
of the form ``.ascii "==>MAGIC RUNES<=="'' so clang doesn't choke on
them.
While we're at it, assemble more of the final output line in the C
file, to make the sed expression shorter.
Reported-by: Julien Grall <julien.grall@linaro.org>
Suggested-by: Jan Beulich <JBeulich@suse.com>
Signed-off-by: Tim Deegan <tim@xen.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
Add a "force-ept" command line option to allow EPT to be enabled when
VMX feature VM_ENTRY_LOAD_GUEST_PAT is not present.
Due to CVE-2013-2212, this feature is required by default as a
prerequisite for using EPT. If you are not using PCI Passthrough, or
trust the guest administrator who would be using passthrough, then the
requirement can be relaxed. This option is particularly useful for
nested virtualization, to allow the L1 hypervisor to use EPT even if
the L0 hypervisor does not provide VM_ENTRY_LOAD_GUEST_PAT.
Signed-off-by: Aravindh Puthiyaparambil <aravindp@cisco.com>
Acked-by: Kevin Tian <kevin.tian@intel.com>
Introducing an extension to XEN_DOMCTL_[gs]et_ext_vcpucontext similar
to the generic MSR save/restore logic recently added for HVM.
This also moves some debug register related declarations/definition to
the header intended for these.
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Tested-by: Aravind Gopalakrishnan<aravind.gopalakrishnan@amd.com>
Reviewed-by: Aravind Gopalakrishnan<aravind.gopalakrishnan@amd.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
The function vgic_irq_rank may return NULL is the IRQ is not in range handled
by the guest. This will result to derefence a NULL pointer which will crash
Xen.
I've checked the rest of the emulation and this is only place where the lock
is taken before the rank is checked.
This is CVE-2014-2986 / XSA-94.
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Reported-by: Thomas Leonard <talex5@gmail.com>
Reviewed-by: Jan Beulich <JBeulich@suse.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Xen is currently installed into /boot, which is a hardcoded path. This
makes it impossible to install two xen.rpm packages from 'make rpmball'
in parallel because rpm complains about conflicting files.
Use BOOT_DIR to define the install path to make it possible to install
xen.gz below --prefix=.
Signed-off-by: Olaf Hering <olaf@aepfle.de>
Reviewed-by: Ian Campbell <ian.campbell@citrix.com>
For rem being zero (where rem is the remainder of a division by
BITS_PER_LONG), shifts by "BITS_PER_LONG - rem" degenerate to undefined
constructs.
An alternative would be to remove these implicitly unused functions.
Coverity ID 1192174 (__bitmap_shift_right)
Coverity ID 1192175 (__bitmap_shift_left)
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
The function early_uart_init contains specific initialization for the
Arndale Board 5250.
Usually U-boot as already setup the UART correctly (ie. clock, baud rate...)
so we don't have to do again.
Futhermore, this code won't work on new platform such as the Arndale Octa.
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Cc: HyonYoung Choi <commani@gmail.com>
Cc: Meng Xu <xumengpanda@gmail.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
xencommons creates the _hardcoded_ /var/run/xen already, there is no
need to create and package this directory during make install|rpmball.
Without this change installing the resulting xen.rpm will fail on
systems where /var/run is a symlink. rpm complains that '/var/run' (the
symlink) is already owned by some other system rpm package.
Using XEN_RUN_DIR instead of /var/run/xen tree-wide may be done in a
followup patch.
Signed-off-by: Olaf Hering <olaf@aepfle.de>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
pygrub stores the temporary kernel+initrd files in /var/run/pygrub, or
any other directory specified with --output-directory=<dir>. If the
default dir is missing pygrub will create it.
Without this change installing the resulting xen.rpm will fail on
systems where /var/run is a symlink. rpm complains that '/var/run' (the
symlink) is already owned by some other system rpm package.
Signed-off-by: Olaf Hering <olaf@aepfle.de>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
/var/lock/subsys is used only in tools/hotplug/Linux/init.d/xendomains.
The start() function already does a "mkdir /var/lock/subsys", so its not
required to create this directory during make install|rpmball.
Without this change installing the resulting xen.rpm will fail on
systems where /var/lock is a symlink. rpm complains that '/var/lock'
(the symlink) is already owned by some other system rpm package.
Signed-off-by: Olaf Hering <olaf@aepfle.de>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
xenstored already does "mkdir /var/run/xenstored" on startup, so its not
required to create this directory during make install|rpmball.
Without this change installing the resulting xen.rpm will fail on
systems where /var/run is a symlink. rpm complains that '/var/run' (the
symlink) is already owned by some other system rpm package.
Signed-off-by: Olaf Hering <olaf@aepfle.de>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Because there is not currently a method for the vTPM Manager to obtain a
build hash of a vTPM, use the hash of the vTPM's XSM label as a
substitute. This allows the vTPM Manager to distinguish between vTPMs
intended to be paired with a hardware domain kernel (which cannot use
pv-grub) and vTPMs which are paired with a pv-grub domain and therefore
contain reliable measurements of the guest kernel in PCRs 4 and 5.
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
This adds an optional argument --vtpm-label=<label> to the pv-grub
command line. If specified, a vtpm device must be connected to the
pv-grub domain and the backend of this device must have the given XSM
label (which may start with a * to indicate a wildcard). Verifying the
label of the vTPM before sending measurements prevents a disaggregated
control domain that has access to xenstore but not to the guest domains
from causing the measurements performed by pv-grub to be discarded,
allowing the forgery of arbitrary kernel measurements in the TPM.
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Cc: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Acked-by: Samuel Thibault <samuel.thibault@ens-lyon.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
This allows the client of a vTPM to request a quote from the physical
TPM which includes PCRs from both the physical and virtual TPMs, signed
by an AIK from the physical TPM. This quote can be used to provide
evidence of the complete launch environment of a virtual machine.
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
For a vTPM to be useful for remote attestation, proof that the vTPM's EK
was generated and held within a secure vTPM implementation is necessary.
This patch adds an ordinal to the vTPM which will request a quote
providing this evidence from the TPM Manager; it only functions during
the first startup of a given vTPM in order to provide proof that the EK
was freshly generated (and not a key whose private part is available
elsewhere).
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
When sending commands to a vTPM, commands with the VTPM_TAG_REQ2 tag are
passed directly to the TPM Manager since they are used in the management
interface to the TPM Manager. The VTPM_TAG_REQ tag is translated to
TPM_TAG_RQU_COMMAND to allow access to the physical TPM for certain
ordinals (PCRRead, Extend, and GetRandom).
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Infineon 1.2 TPMs fail TPM_Seal commands with TPM_BAD_PARAMETER when
PCRS are specified by a TPM_PCR_INFO structure. Using a
TPM_PCR_INFO_LONG structure to specify PCRs succeeds, so update to use
that. This also requires changes to use TPM_STORED_DATA12 for the
result.
Signed-off-by: Jason Andryuk <andryuk@aero.org>
Acked-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
The TPM locality must be specified when using TPM_PCR_INFO_LONG for
TPM_Seal. Store the locality so it can be provided later.
Signed-off-by: Jason Andryuk <andryuk@aero.org>
Acked-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
The manage-vtpmmgr.pl script is an example client for interacting with
the TPM Manager; it is intended to run in a management domain with a
vTPM (which may be dom0). It is used to create and manage vTPMs and
vTPM groups.
The calc.pl script is an example manager of a vTPM group. It signs
the configuration lists used by the TPM Manager with a locally held
private key.
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
This is a complete rewrite of the disk format and key hierarchy for the
TPM Manager. The new format supports multiple groups of vTPMs which
define the permitted configurations where a given vTPM's keys are
available, allowing upgrades of critical components while retaining the
secrecy of cryptographic keys.
New features of the TPM Manager are explained in the README and in the
definitions of the management commands in vtpm_manager.h.
New features for vTPMs:
1. The size of the state blob for a vTPM is expanded from 52 to 64
bytes in order to support future vTPMs using SHA-2/3 instead of SHA-1.
2. vTPMs can obtain a quote from the physical TPM with certain
resettable PCRs set to include information about the vTPM. This can be
used by a vTPM to provide evidence of its integrity, including the
secrecy of its EK, and for deep quotes.
Some additional changes made by this rewrite that may impact existing
users:
1. The value of WELLKNOWN_OWNER_AUTH was incorrect for the physical TPM;
the convention is to use all zero bits for well-known authentication
values, not all one bits.
2. Randomly generating the owner auth value for the physical TPM is no
longer supported, as it prevents later creation or certification of
AIKs (which the old manager did not support).
3. The vTPM Manager needs to be provisioned with a PCR composite and an
upgrade authority's public key before it will save data across boots.
The current implementation still has some limitations:
* 5 valid system PCR selections per group
* The vTPM Manager's disk can use at most 2MB of space
* The vTPM domain's build hash is always set to null/zero
Most of the code relating to upgrade and rollback protection is
currently stubbed out, but future versions can add:
* Support for using the TPM's monotonic counter to prevent rollback
of vTPM data by taking and restoring disk snapshots
* Masking the master disk encryption key using a value stored in the
TPM's NVRAM so that revocation of old data is possible without
relying on all previously authorized software stacks to respect the
monotonic counter's value
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Debug and performance registers are not properly switched by Xen.
Trap them and inject an undefined instruction, except for those registers
which might be unconditionally accessed which we implement as RAZ/WI.
This is CVE-2014-2915 / XSA-93.
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
On Cortex-A15, CP15 c15 contains registers to retrieve data from L1/L2 RAM.
Exposing this registers to guest may result to leak data from Xen and/or
another guest.
By default trap every registers and inject an undefined instruction.
This is CVE-2014-2915 / XSA-93.
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Some cp15 c9/c10/c11 encodings are used for:
- cache control
- TCM control
- branch predictor control
All of them are implementation defined. For now inject an undefined exception
if the guest wants try to access it.
This is CVE-2014-2915 / XSA-93.
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
A guest is allowed to use invalidate cache by set/way instruction (i.e DCISW)
without any restriction. As the cache is shared with Xen, the guest invalidate
an address being in used by Xen. This may lead a Xen crash because the memory
state is invalid.
Set the bit HCR.SWIO to upgrade invalidate cache by set/way instruction to an
invalidate and clean.
This is CVE-2014-2915 / XSA-93.
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Reported-by: Thomas Leonard <tal36@cam.ac.uk>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
In Xen we only handle save/restore for coprocessor 10 and 11 (NEON). Other
coprocessors (0-9, 12-13) are currently exposed to the guest and may lead
to data shared between guest.
Disable access to all coprocessor except 10 and 11 by setting correctly
HCTPR.
This is CVE-2014-2915 / XSA-93.
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Currently Xen panics if it's unable to handle a coprocessor/sysreg instruction.
Replace this behavior by inject an undefined instruction to the faulty guest
and log if Xen is in debug mode.
This is CVE-2014-2915 / XSA-93.
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
c/s 3d90d6e6 "x86/EPT: split super pages upon mismatching memory types"
accidentally introduced an error path where the epte domain page would be
unmapped twice if splitting the superpage failed.
Only unmap the page if the loop is to be continued. When breaking from the
loop, the page will be unmapped by the subsequent code.
Coverity-ID: 1203047
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
This adds a hypervisor command line option "hardware_dom=" which takes a
domain ID. When the domain with this ID is created, it will be used
as the hardware domain.
This is intended to be used when domain 0 is a dedicated stub domain for
domain building, allowing the hardware domain to be de-privileged and
act only as a driver domain.
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
When a domain is saved each VCPU's TSC value needs to be preserved. To get it we
use hvm_get_guest_tsc(). This routine (either itself or via get_s_time() which
it may call) calculates VCPU's TSC based on current host's TSC value (by doing a
rdtscll()). Since this is performed for each VCPU separately we end up with
un-synchronized TSCs.
Similarly, during a restore each VCPU is assigned its TSC based on host's current
tick, causing virtual TSCs to diverge further.
With this, we can easily get into situation where a guest may see time going
backwards.
Instead of reading new TSC value for each VCPU when saving/restoring it we should
use the same value across all VCPUs.
Reported-by: Philippe Coquard <philippe.coquard@mpsa.com>
Signed-off-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
TSC ratio enabling logic is inverted: we want to use it when we
are running in native tsc mode, i.e. when d->arch.vtsc is zero.
Also, since now svm_set_tsc_offset()'s calculations depend
on vtsc's value, we need to call hvm_funcs.set_tsc_offset() after
vtsc changes in tsc_set_info().
In addition, with TSC ratio enabled, svm_set_tsc_offset() will
need to do rdtsc. With that we may end up having TSCs on guest's
processors out of sync. d->arch.hvm_domain.sync_tsc which is set
by the boot processor can now be used by APs as reference TSC
value instead of host's current TSC.
Signed-off-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
We should be able to continue using native RDTSC(P) execution on
HVM/PVH guests after migration if host and guest frequencies are
equal (this includes the case when the frequencies are made equal
by TSC scaling feature).
This also allows us to revert main part of commit 4aab59a3 (svm: Do not
intercept RDTSC(P) when TSC scaling is supported by hardware) which
was wrong: while RDTSC intercepts were disabled domain's vtsc could
still be set, leading to inconsistent view of guest's TSC.
Signed-off-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
Error checks exist in the respective code path that expect negative
values to indicate errors, yet negative values can't be communicated
through size_t. Thus ssize_t needs to be introduced (also on ARM for
consistency, even if the code in question isn't currently being used
on there).
The bug is theoretical only in so far as all the involved code is
effectively dead. Reflect this by excluding that code from non-debug
builds.
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Christoph Egger <chegger@amazon.de>
Considering that
- the workaround is expensive (iterating through the entire P2M space
of a domain),
- the planned elimination of the expensiveness (by propagating the type
change step by step to the individual P2M leaves) wouldn't address
the IOMMU side of things (as for it to obey to the changed
permissions the adjustments must be pushed down immediately through
the entire tree)
- the proper solution (PHYSDEVOP_msix_prepare) should by now be
implemented by all security conscious Dom0 kernels
remove the workaround, killing eventual guests that would be known to
become a security risk instead.
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Kevin Tian <kevin.tian@intel.com>
This requires setting the hardware_domain variable earlier in
domain_create so that functions called from it (primarily in
arch_domain_create) observe the correct results when they call
is_hardware_domain.
Signed-off-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Ian Campbell
Aravindh Puthiyaparambil
Tim Deegan
Jan Beulich
Olaf Hering
Julien Grall
Daniel De Graaf
Jason Andryuk
Wei Liu
Andrew Cooper
Boris Ostrovsky