This commit changes the existing libxl code to be sure to grant access
permission to PCI-related I/O memory ranges, while setting up passthrough
of PCI devices specified in the domain's configuration, and to VGA-related
memory ranges, while setting up VGA passthrough (if gfx_passthru = 1 in
the domain's configuration).
As for the latter, the newly-added code does not replace any existing one,
but instead matches the calls to xc_domain_memory_mapping() performed by
QEMU on the path that is executed if gfx passthru is enabled and follows
the registration of a new VGA controller (in register_vga_regions(),
defined in hw/pt-graphics.c). In fact, VGA needs some extra memory
ranges to be mapped with respect to PCI; QEMU expects that access to those
memory ranges is implicitly granted when he calls the hypervisor with the
function xc_domain_memory_mapping(): this commit calls iomem_permission
for it when needed by checking the passthru PCI device's class.
NOTE: the code added by this commit still does not verify if the passthru
of the framebuffer area is being performed for the primary GPU, but
only replicates the behavior of QEMU which is limited to performing
the passthru for all PCI devices of VGA class.
This commit is instrumental to the last one in the series, which will
separate the functions of the iomem_permission and memory_mapping DOMCTLs,
so that requesting an I/O-memory range will not imply that access to such
a range is implicitly granted.
Signed-off-by: Arianna Avanzini <avanzini.arianna@gmail.com>
Acked-by: Ian Campbell <Ian.Campbell@eu.citrix.com>
Cc: Dario Faggioli <dario.faggioli@citrix.com>
Cc: Paolo Valente <paolo.valente@unimore.it>
Cc: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Cc: Julien Grall <julien.grall@citrix.com>
Cc: Jan Beulich <JBeulich@suse.com>
Cc: Ian Jackson <Ian.Jackson@eu.citrix.com>
Cc: Eric Trudeau <etrudeau@broadcom.com>
Cc: Viktor Kleinik <viktor.kleinik@globallogic.com>
Cc: Andrii Tseglytskyi <andrii.tseglytskyi@globallogic.com>
This small helper is responsible for generating Dom0 JSON config
stub and writing Dom0 xenstore entries. This helpers subsumes two calls
to xenstore-write in xencommons script.
Dom0 UUID is intentionally left untouched, so it is always all
zeros. This makes sure that we don't leak Dom0 stubs across rebooting.
Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
[ ijc -- folded in incremental patch for *BSD ]
Originally the code allowed users to attach the same device more than
once. It just stupidly overwrites xenstore entries. This is bogus as
frontend will be very confused.
Introduce a helper function to check if the device to be written to
xenstore already exists. A new error code is also introduced.
The check and add are within one xs transaction.
Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
The configuration is stored in libxl-json format. It will be used as
template to reconstruct domain configuration during runtime.
There's only one write to disk when domain creation finishes. We
therefore have a window that the domain exists but has no JSON config in
disk. We define this state as domain being created or destroyed. Any
other operations that need to access JSON config should bail.
Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Introduce a new format in libxl userdata store called "libxl-json". This
file format contains JSON version of libxl_domain_config, generated by
libxl. Applications are not supposed to access this file directly.
Two internal functions to get and set libxl_domain_configuration
are also introduced. Also introduce a new error code to indicate
abnormal state that libxl-json config file is empty.
Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>
Originally libxl user data store didn't have lock at all. There could be
such race condition as mentioned by Ian Jackson:
Task 1 Task 2
Creating the domain Trying to shut down
actually create domain
observe domid
start domain destruction
delete all userdata
destroy domain
store the userdata
*** forbidden state created: userdata exists but domain doesn't
*** userdata has been leaked
[ would now bomb out ]
This patch adds in proper locking to libxl user data store. The lock is
associated with a specific domain (i.e. a per-domain lock).
As for locking hierachy, we first take CTX lock (which is implemented
with pthread recursive mutex so even if the application has taken it
we're fine), then take the file lock. These locks are released in
reversed order.
A new libxl error code ERROR_LOCK_FAIL is introduced to describe failure
to acquire locks.
Also factor out libxl__userdata_{retrieve,store}, so that other
functions that already hold the lock can call them to manipulate
user data.
Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
This lock is used to protect all userdata files related to a particular
domain, which include but are not limited to domain configuration. A
new "domain-userdata-lock" entry is introduced in libxl registry.
This lock works among different processes and different threads within
the same process.
Locking protocol inspired by Ian Jackson's chiark-utils with-lock-ex. A
file lock is taken with flock(2). If that succeeds that thread fstat the
fd and stat the lock file path. If the device and inode match then the
lock has been successfully acquired. This lock remains acquired until
the lock file gets deleted or released by flock(2). If device and inode
don't match then another thread acquired the lock and deleted the file
in the meantime; lock procedure should restart.
Portability note: this lock utilises flock(2) so a proper implementation
of flock(2) is required -- that is, it should not be implemented with
fcntl(2).
Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
The flag specifies that the hypervisor maps a grant page to guest
physical address == machine address of the page in addition to the
normal grant mapping address.
Frontends are allowed to map the same page multiple times using multiple
grant references. On the backend side it can be difficult to find out
the physical address corresponding to a particular machine address,
especially at the completion of a dma operation. To simplify address
translations, we introduce a second mapping of the grant at physical
address == machine address so that dom0 can issue cache maintenance
operations without having to find the pfn.
Call arch_grant_map_page_identity and arch_grant_unmap_page_identity
from __gnttab_map_grant_ref and __gnttab_unmap_common to introduce the
second mapping if the domain is directly mapped. To do so we also need
to change gnttab_need_iommu_mapping to just be defined as
is_domain_direct_mapped on arm.
Remove arm_smmu_map_page and arm_smmu_unmap_page as they have become
unused.
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Julien Grall <julien.grall@linaro.org>
Introduce two arch specific functions to create a new p2m mapping of
granted pages at pfn == mfn.
We don't an x86 implementation as these functions should never be
compiled on x86 (they are called from an if (0) statement).
Base the implementation of arm_smmu_(un)map_page on
arch_grant_(un)map_page_identity.
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Acked-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Jan Beulich <jbeulich@suse.com>
When the function domain_vgic_init is failing to initialize pending_irqs,
it will free shared_irqs. Few call later, domain_vgic_free will be called
an try to free a second time the same variable. This will result to a double
free.
Remove the free in domain_vgic_init and rely on domain_vgic_free to correctly
release the memory.
Signed-off-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
It's only used by test.c and probably isn't something that should be
exposed for general use (it just calls gettimeofday).
This is also useful for ARM, which doesn't use math.c, where this used
to live.
Signed-off-by: Thomas Leonard <talex5@gmail.com>
Acked-by: Samuel Thibault <samuel.thibault@ens-lyon.org>
This breaks the include cycle hypervisor.h -> hypercall-x86_32.h -> mm.h
-> lib.h -> gntmap.h -> os.h -> hypervisor.h.
Signed-off-by: Thomas Leonard <talex5@gmail.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Samuel Thibault <samuel.thibault@ens-lyon.org>
Stubdomain code as-is doesn't compile on FreeBSD, so disable it for
now in order to have a working build.
Please re-run autoconf after committing.
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Cc: Ian Jackson <Ian.Jackson@eu.citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Autoconf already has appropriate Linux/non-Linux logic for CONFIG_BLKTAP2.
Let LIBXL_BLKTAP inherit directly from CONFIG_BLKTAP2, rather than requiring a
further LIBXL_BLKTAP=n in addition to ./configure --disable-blktap2 to get
libxl to successfully link under Linux.
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
CC: Ian Campbell <Ian.Campbell@citrix.com>
CC: Ian Jackson <Ian.Jackson@eu.citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
caml & C stubdom examples should not enable the tests, they already
provide their own main function.
Signed-off-by: Samuel Thibault <samuel.thibault@ens-lyon.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
As it needs to be, considering that we assign -1 to it (and
that, later, we check for it to be -1), to signify 'all vcpus'.
While at it, fix a coding style nit and improve error reporting.
Signed-off-by: Dario Faggioli <dario.faggioli@citrix.com>
Reviewed-by: Don Slutz <dslutz@verizon.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
OCaml is replacing its own uint32 type with the ISO C99 uint32_t type
in 4.03.0+dev. Ths patch replaces the remaining uses of uint32 with
uint32_t in the ocaml C code for xen.
Signed-off-by: Michael Young <m.a.young@durham.ac.uk>
Acked-by: David Scott <dave.scott@citrix.com>
With the introduction of the XEN_DOMCTL_setvnumainfo
in patch titled: "xen: vnuma topology and subop hypercalls"
we put in the plumbing here to use from the toolstack. The user
is allowed to call this multiple times if they wish so.
It will error out if the nr_vnodes or nr_vcpus is zero.
Signed-off-by: Elena Ufimtseva <ufimtseva@gmail.com>
Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Define xsm_get_vnumainfo hypercall used for domain which
wish to receive vnuma topology. Add xsm hook for
XEN_DOMCTL_setvnumainfo. Also adds basic policies.
Signed-off-by: Elena Ufimtseva <ufimtseva@gmail.com>
Acked-by: Daniel De Graaf <dgdegra@tycho.nsa.gov>
Define interface, structures and hypercalls for toolstack to
build vnuma topology and for guests that wish to retrieve it.
Two subop hypercalls introduced by patch:
XEN_DOMCTL_setvnumainfo to define vNUMA domain topology per domain
and XENMEM_get_vnumainfo to retrieve that topology by guest.
Signed-off-by: Elena Ufimtseva <ufimtseva@gmail.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Not only did the EVNTSEL ones get defined twice, we can also easily
abstract out the numbers previously attached to them.
While at it also remove an unused Geode-related define.
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Acked-by: Kevin Tian <kevint.tian@intel.com>
This way we don't need to take the desc->lock in order to access
desc->status in many of the gic and vgic functions.
Using *_bit manipulation functions on desc->status is safe on arm64:
status is an unsigned int but is the first field of a struct that
contains pointers, therefore the alignement of the struct is at least 8
bytes.
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Julien Grall <julien.grall@linaro.org>
Currently we read ipriority from vgic_vcpu_inject_irq without taking the
rank lock. Fix that by taking the rank lock and reading ipriority at the
beginning of the function.
As vgic_vcpu_inject_irq is called from the irq.c upon receiving an
interrupt, we need to change the implementation of vgic_lock/unlock_rank
to spin_lock_irqsave to make it safe in irq context.
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Acked-by: Julien Grall <julien.grall@linaro.org>
evtchn_upcall_pending is already set by common code at vcpu creation,
therefore on ARM we also need to call vgic_vcpu_inject_irq for it.
Currently we do that from vgic_enable_irqs as a workaround.
Do this properly by introducing an appropriate arch specific hook:
arch_evtchn_inject. arch_evtchn_inject is called by map_vcpu_info to
inject the evtchn irq into the guest. On ARM is implemented by calling
vgic_vcpu_inject_irq.
On x86 guests typically don't call VCPUOP_register_vcpu_info on vcpu0,
therefore avoiding the issue. However theoretically they could call
VCPUOP_register_vcpu_info on vcpu0 and in that case Xen would need to
inject the event channel notification into the guest if the guest is
HVM. So implement arch_evtchn_inject on x86 by calling
hvm_assert_evtchn_irq.
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Julien Grall <julien.grall@linaro.org>
Migrate physical irqs to the same physical cpu that is running the vcpu
expected to receive the irqs. That is done when enabling irqs, when the
guest writes to GICD_ITARGETSR and when Xen migrates a vcpu to a
different pcpu.
In case of virq migration, if the virq is inflight and in a GICH_LR
register already, delay migrating the corresponding physical irq until
the virq is EOIed by the guest and the MIGRATING flag has been cleared.
This way we make sure that the pcpu running the old vcpu gets
interrupted with a new irq of the same kind, clearing the GICH_LR sooner.
Introduce a new arch specific function, arch_move_irqs, that is empty on
x86 and implements the vgic irq migration code on ARM.
arch_move_irqs is going to be called by from sched.c.
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Julien Grall <julien.grall@linaro.org>
Use vgic_get_target_vcpu to retrieve the target vcpu from do_IRQ.
Remove in-code comments about missing implementation of SGI delivery to
vcpus other than 0.
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Acked-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
We need to take special care when migrating irqs that are already
inflight from one vcpu to another. See "The effect of changes to an
GICD_ITARGETSR", part of chapter 4.3.12 of the ARM Generic Interrupt
Controller Architecture Specification, IHI 0048B.
The main issue from the Xen point of view is that the lr_pending and
inflight lists are per-vcpu. The lock we take to protect them is also
per-vcpu.
In order to avoid issues, if the irq is still lr_pending, we can
immediately move it to the new vcpu for injection.
Otherwise if it is in a GICH_LR register, set a new flag
GIC_IRQ_GUEST_MIGRATING. If GIC_IRQ_GUEST_MIGRATING is set we'll change
the affinity of the physical irq when clearing the LR (in a following
commit). Therefore if the irq is inflight in an LR when the guest writes
to itarget, we wait until the LR is cleared before changing physical
irq affinity. This guarantees that the old vcpu is going to be
interrupted and clear the LR, either because it has been descheduled
after EOIing the interrupt or because it is interrupted by the second
physical irq coming.
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Acked-by: Julien Grall <julien.grall@linaro.org>
vgic_enable_irqs should enable irq delivery to the vcpu specified by
GICD_ITARGETSR, rather than the vcpu that wrote to GICD_ISENABLER.
Similarly vgic_disable_irqs should use the target vcpu specified by
itarget to disable irqs.
itargets can be set to a mask but vgic_get_target_vcpu always returns
the lower vcpu in the mask.
Correctly initialize itargets for SPIs.
Ignore bits in GICD_ITARGETSR corresponding to invalid vcpus.
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Acked-by: Julien Grall <julien.grall@linaro.org>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Arianna Avanzini
Wei Liu
Stefano Stabellini
Julien Grall
Vijaya Kumar K
Karim Raslan
Thomas Leonard
Tamas K Lengyel
Roger Pau Monne
Oleksandr Dmytryshyn
Andrew Cooper
Samuel Thibault
Dario Faggioli
M A Young
Ian Campbell
Jan Beulich
Elena Ufimtseva