At the moment if the end-user does not specify the gic-version along
with KVM acceleration, v2 is set by default. However most of the
systems now have GICv3 and sometimes they do not support GICv2
compatibility.
This patch keeps the default v2 selection in all cases except
in the KVM accelerated mode when either
- the host does not support GICv2 in-kernel emulation or
- number of VCPUS exceeds 8.
Those cases did not work anyway so we do not break any compatibility.
Now we get v3 selected in such a case.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reported-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Andrew Jones <drjones@redhat.com>
Message-id: 20200311131618.7187-7-eric.auger@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Restructure the finalize_gic_version with switch cases and
clearly separate the following cases:
- KVM mode / in-kernel irqchip
- KVM mode / userspace irqchip
- TCG mode
In KVM mode / in-kernel irqchip , we explictly check whether
the chosen version is supported by the host. If the end-user
explicitly sets v2/v3 and this is not supported by the host,
then the user gets an explicit error message. Note that for
old kernels where the CREATE_DEVICE ioctl doesn't exist then
we will now fail if the user specifically asked for gicv2,
where previously we (probably) would have succeeded.
In KVM mode / userspace irqchip we immediatly output an error
in case the end-user explicitly selected v3. Also we warn the
end-user about the unexpected usage of gic-version=host in
that case as only userspace GICv2 is supported.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Andrew Jones <drjones@redhat.com>
Message-id: 20200311131618.7187-6-eric.auger@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Convert kvm_arm_vgic_probe() so that it returns a
bitmap of supported in-kernel emulation VGIC versions instead
of the max version: at the moment values can be v2 and v3.
This allows to expose the case where the host GICv3 also
supports GICv2 emulation. This will be useful to choose the
default version in KVM accelerated mode.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Andrew Jones <drjones@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20200311131618.7187-5-eric.auger@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
The tests themselves are the same as the ISA device ones.
Only the main() changes as the "tpm-tis-device" device gets
instantiated. Also the base address of the device is not
0xFED40000 anymore but matches the base address of the
ARM virt platform bus.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Stefan Berger <stefanb@linux.ibm.com>
Message-id: 20200305165149.618-11-eric.auger@redhat.com
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
ISA and sysbus TPM-TIS devices will share their tests. Only
the main() will change (instantiation option is different).
Also the base address of the TPM-TIS device is going to be
different. on x86 it is located at 0xFED40000 while on ARM
it can be located at any location, discovered through the
device tree description.
So we put shared test functions in a new object module.
Each test needs to set tpm_tis_base_addr global variable.
Also take benefit of this move to fix "block comments using
a leading */ on a separate line" checkpatch warnings.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Stefan Berger <stefanb@linux.ibm.com>
Message-id: 20200305165149.618-10-eric.auger@redhat.com
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
We plan to use swtpm test functions on ARM for testing the
sysbus TPM-TIS device. However on ARM there is no default machine
type. So we need to explictly pass some machine options on startup.
Let's allow this by adding a new parameter to both swtpm test
functions and update all call sites.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Stefan Berger <stefanb@linux.ibm.com>
Message-id: 20200305165149.618-9-eric.auger@redhat.com
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
Let the TPM TIS SYSBUS device be dynamically instantiable
in ARM virt. A device tree node is dynamically created
(TPM via MMIO).
The TPM Physical Presence interface (PPI) is not supported.
To run with the swtmp TPM emulator, the qemu command line must
be augmented with:
-chardev socket,id=chrtpm,path=swtpm-sock \
-tpmdev emulator,id=tpm0,chardev=chrtpm \
-device tpm-tis-device,tpmdev=tpm0 \
swtpm/libtpms command line example:
swtpm socket --tpm2 -t -d --tpmstate dir=/tmp/tpm \
--ctrl type=unixio,path=swtpm-sock
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Stefan Berger <stefanb@linux.ibm.com>
Tested-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Acked-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Message-id: 20200305165149.618-7-eric.auger@redhat.com
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
As we plan to introduce a SysBus TPM TIS device, let's
make the TPMState a common struct usable by both the
ISADevice and the SysBusDevice. TPMStateISA embeds the
struct and inherits from the ISADevice.
The prototype of functions bound to be used by both
the ISA and SysBus devices is changed to take TPMState
handle.
A bunch of structs also are renamed to be specialized
for the ISA device. Besides those transformations, no
functional change is expected.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Stefan Berger <stefanb@linux.ibm.com>
Tested-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Acked-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Message-id: 20200305165149.618-3-eric.auger@redhat.com
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
Adds the "virtio,pci-iommu" node in the host bridge node and
the RID mapping, excluding the IOMMU RID.
This is done in the virtio-iommu-pci hotplug handler which
gets called only if no firmware is loaded or if -no-acpi is
passed on the command line. As non DT integration is
not yet supported by the kernel we must make sure we
are in DT mode. This limitation will be removed as soon
as the topology description feature gets supported.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Message-Id: <20200214132745.23392-10-eric.auger@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
This patch adds virtio-iommu-pci, which is the pci proxy for
the virtio-iommu device.
Currently non DT integration is not yet supported by the kernel.
So the machine must implement a hotplug handler for the
virtio-iommu-pci device that creates the device tree iommu-map
bindings as documented in kernel documentation:
Documentation/devicetree/bindings/virtio/iommu.txt
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Jean-Philippe Brucker <jean-philippe@linaro.org>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Message-Id: <20200214132745.23392-9-eric.auger@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
This patch implements the endpoint attach/detach to/from
a domain.
Domain and endpoint internal datatypes are introduced.
Both are stored in RB trees. The domain owns a list of
endpoints attached to it. Also helpers to get/put
end points and domains are introduced.
As for the IOMMU memory regions, a callback is called on
PCI bus enumeration that initializes for a given device
on the bus hierarchy an IOMMU memory region. The PCI bus
hierarchy is stored locally in IOMMUPciBus and IOMMUDevice
objects.
At the time of the enumeration, the bus number may not be
computed yet.
So operations that will need to retrieve the IOMMUdevice
and its IOMMU memory region from the bus number and devfn,
once the bus number is garanteed to be frozen, use an array
of IOMMUPciBus, lazily populated.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Message-Id: <20200214132745.23392-4-eric.auger@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Instead of inserting read elements at the head and
then reversing the list, it is simpler to add
each element after the previous one. Introduce
QLIST_RAW_INSERT_AFTER helper and use it in
get_qlist().
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Suggested-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Support QLIST migration using the same principle as QTAILQ:
94869d5c52 ("migration: migrate QTAILQ").
The VMSTATE_QLIST_V macro has the same proto as VMSTATE_QTAILQ_V.
The change mainly resides in QLIST RAW macros: QLIST_RAW_INSERT_HEAD
and QLIST_RAW_REVERSE.
Tests also are provided.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Host kernel within [4.18, 5.3] report an erroneous KVM_MAX_VCPUS=512
for ARM. The actual capability to instantiate more than 256 vcpus
was fixed in 5.4 with the upgrade of the KVM_IRQ_LINE ABI to support
vcpu id encoded on 12 bits instead of 8 and a redistributor consuming
a single KVM IO device instead of 2.
So let's check this capability when attempting to use more than 256
vcpus within any ARM kvm accelerated machine.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Andrew Jones <drjones@redhat.com>
Acked-by: Marc Zyngier <maz@kernel.org>
Message-id: 20191003154640.22451-4-eric.auger@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Host kernels that expose the KVM_CAP_ARM_IRQ_LINE_LAYOUT_2 capability
allow injection of interrupts along with vcpu ids larger than 255.
Let's encode the vpcu id on 12 bits according to the upgraded KVM_IRQ_LINE
ABI when needed.
Given that we have two callsites that need to assemble
the value for kvm_set_irq(), a new helper routine, kvm_arm_set_irq
is introduced.
Without that patch qemu exits with "kvm_set_irq: Invalid argument"
message.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reported-by: Zenghui Yu <yuzenghui@huawei.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Andrew Jones <drjones@redhat.com>
Acked-by: Marc Zyngier <maz@kernel.org>
Message-id: 20191003154640.22451-3-eric.auger@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Introduce support for GTree migration. A custom save/restore
is implemented. Each item is made of a key and a data.
If the key is a pointer to an object, 2 VMSDs are passed into
the GTree VMStateField.
When putting the items, the tree is traversed in sorted order by
g_tree_foreach.
On the get() path, gtrees must be allocated using the proper
key compare, key destroy and value destroy. This must be handled
beforehand, for example in a pre_load method.
Tests are added to test save/dump of structs containing gtrees
including the virtio-iommu domain/mappings scenario.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Message-Id: <20191011121724.433-1-eric.auger@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
uintptr_t fixup for test on 32bit
Currently, when a notifier is attempted to be registered and its
flags are not supported (especially the MAP one) by the IOMMU MR,
we generally abruptly exit in the IOMMU code. The failure could be
handled more nicely in the caller and especially in the VFIO code.
So let's allow memory_region_register_iommu_notifier() to fail as
well as notify_flag_changed() callback.
All sites implementing the callback are updated. This patch does
not yet remove the exit(1) in the amd_iommu code.
in SMMUv3 we turn the warning message into an error message saying
that the assigned device would not work properly.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The container error integer field is currently used to store
the first error potentially encountered during any
vfio_listener_region_add() call. However this fails to propagate
detailed error messages up to the vfio_connect_container caller.
Instead of using an integer, let's use an Error handle.
Messages are slightly reworded to accomodate the propagation.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
An IOVA/ASID invalidation is notified to all IOMMU Memory Regions
through smmuv3_inv_notifiers_iova/smmuv3_notify_iova.
When the notification occurs it is possible that some of the
PCIe devices associated to the notified regions do not have a
valid stream table entry. In that case we output a LOG_GUEST_ERROR
message, for example:
invalid sid=<SID> (L1STD span=0)
"smmuv3_notify_iova error decoding the configuration for iommu mr=<MR>
This is unfortunate as the user gets the impression that there
are some translation decoding errors whereas there are not.
This patch adds a new field in SMMUEventInfo that tells whether
the detection of an invalid STE must lead to an error report.
invalid_ste_allowed is set before doing the invalidations and
kept unset on actual translation.
The other configuration decoding error messages are kept since if the
STE is valid then the rest of the config must be correct.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Message-id: 20190822172350.12008-6-eric.auger@redhat.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Report an error in case we fail to set a trigger action
on any VFIO_PCI_MSIX_IRQ_INDEX subindex. This might be
useful in debugging a device that is not working properly.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reported-by: Coverity (CID 1402196)
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
The code used to assign an interrupt index/subindex to an
eventfd is duplicated many times. Let's introduce an helper that
allows to set/unset the signaling for an ACTION_TRIGGER,
ACTION_MASK or ACTION_UNMASK action.
In the error message, we now use errno in case of any
VFIO_DEVICE_SET_IRQS ioctl failure.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Reviewed-by: Li Qiang <liq3ea@gmail.com>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
The SMMUNotifierNode struct is not necessary and brings extra
complexity so let's remove it. We now directly track the SMMUDevices
which have registered IOMMU MR notifiers.
This is inspired from the same transformation on intel-iommu
done in commit b4a4ba0d68
("intel-iommu: remove IntelIOMMUNotifierNode")
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Message-id: 20190409160219.19026-1-eric.auger@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
As NVDIMM support is looming for ARM and SPAPR, let's
move the acpi_nvdimm_state to the generic machine struct
instead of duplicating the same code in several machines.
It is also renamed into nvdimms_state and becomes a pointer.
nvdimm and nvdimm-persistence become generic machine options.
They become guarded by a nvdimm_supported machine class member.
We also add a description for those options.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Suggested-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20190308182053.5487-3-eric.auger@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Now we have the extended memory map (high IO regions beyond the
scalable RAM) and dynamic IPA range support at KVM/ARM level
we can bump the legacy 255GB initial RAM limit. The actual maximum
RAM size now depends on the physical CPU and host kernel, in
accelerated mode. In TCG mode, it depends on the VCPU
AA64MMFR0.PARANGE.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Message-id: 20190304101339.25970-11-eric.auger@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
We are about to allow the memory map to grow beyond 1TB and
potentially overshoot the VCPU AA64MMFR0.PARANGE.
In aarch64 mode and when highmem is set, let's check the VCPU
PA range is sufficient to address the highest GPA of the memory
map.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Message-id: 20190304101339.25970-10-eric.auger@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
This patch implements the machine class kvm_type() callback.
It returns the number of bits requested to implement the whole GPA
range including the RAM and IO regions located beyond.
The returned value is passed though the KVM_CREATE_VM ioctl and
this allows KVM to set the stage2 tables dynamically.
To compute the highest GPA used in the memory map, kvm_type()
must freeze the memory map by calling virt_set_memmap().
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Message-id: 20190304101339.25970-9-eric.auger@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Up to now the memory map has been static and the high IO region
base has always been 256GiB.
This patch modifies the virt_set_memmap() function, which freezes
the memory map, so that the high IO range base becomes floating,
located after the initial RAM and the device memory.
The function computes
- the base of the device memory,
- the size of the device memory,
- the high IO region base
- the highest GPA used in the memory map.
Entries of the high IO region are assigned a base address. The
device memory is initialized.
The highest GPA used in the memory map will be used at VM creation
to choose the requested IPA size.
Setting all the existing highmem IO regions beyond the RAM
allows to have a single contiguous RAM region (initial RAM and
possible hotpluggable device memory). That way we do not need
to do invasive changes in the EDK2 FW to support a dynamic
RAM base.
Still the user cannot request an initial RAM size greater than 255GB.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Message-id: 20190304101339.25970-8-eric.auger@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
The machine RAM attributes will need to be analyzed during the
configure_accelerator() process. especially kvm_type() arm64
machine callback will use them to know how many IPA/GPA bits are
needed to model the whole RAM range. So let's assign those machine
state fields before calling configure_accelerator.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Message-id: 20190304101339.25970-7-eric.auger@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Add the kvm_arm_get_max_vm_ipa_size() helper that returns the
number of bits in the IPA address space supported by KVM.
This capability needs to be known to create the VM with a
specific IPA max size (kvm_type passed along KVM_CREATE_VM ioctl.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Message-id: 20190304101339.25970-6-eric.auger@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
In the prospect to introduce an extended memory map supporting more
RAM, let's split the memory map array into two parts:
- the former a15memmap, renamed base_memmap, contains regions below
and including the RAM. MemMapEntries initialized in this array
have a static size and base address.
- extended_memmap, only initialized with entries located after the
RAM. MemMapEntries initialized in this array only get their size
initialized. Their base address is dynamically computed depending
on the the top of the RAM, with same alignment as their size.
Eventually base_memmap entries are copied into the extended_memmap
array. Using two separate arrays however clarifies which entries
are statically allocated and those which are dynamically allocated.
This new split will allow to grow the RAM size without changing the
description of the high IO entries.
We introduce a new virt_set_memmap() helper function which
"freezes" the memory map. We call it in machvirt_init as
memory attributes of the machine are not yet set when
virt_instance_init() gets called.
The memory map is unchanged (the top of the initial RAM still is
256GiB). Then come the high IO regions with same layout as before.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Message-id: 20190304101339.25970-4-eric.auger@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
We introduce the vfio_init_container_type() helper.
It computes the highest usable iommu type and then
set the container and the iommu type.
Its usage in vfio_connect_container() makes the code
ready for addition of new iommu types.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Commit af7d64ede0 (hw/arm/sysbus-fdt: Allow device matching with DT
compatible value) introduced a match_fn callback which gets called
for each registered combo to check whether a sysbus device can be
dynamically instantiated. However the callback gets called even if
the device type does not match the binding combo typename field.
This causes an assert when passing "-device ramfb" to the qemu
command line as vfio_platform_match() gets called on a non
vfio-platform device.
To fix this regression, let's change the add_fdt_node() logic so
that we first check the type and if the match_fn callback is defined,
then we also call it.
Binding combos only requesting a type check do not define the
match_fn callback.
Fixes: af7d64ede0 (hw/arm/sysbus-fdt: Allow device matching with
DT compatible value)
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reported-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Alex Williamson <alex.williamson@redhat.com>
Tested-by: Geert Uytterhoeven <geert+renesas@glider.be>
Message-id: 20181106184212.29377-1-eric.auger@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Up to now we have relied on the device type to identify a device tree
node creation function. Since we would like the vfio-platform device to
be instantiable with different compatible strings we introduce the
capability to specialize the node creation depending on actual
compatible value.
NodeCreationPair is renamed into BindingEntry. The struct is enhanced
with compat and match_fn() fields. We introduce a new matching function
adapted to the vfio-platform generic device.
Soon, the AMD XGBE can be instantiated with either manner, i.e.:
-device vfio-amd-xgbe,host=e0900000.xgmac
or using the new option line:
-device vfio-platform,host=e0900000.xgmac
Signed-off-by: Eric Auger <eric.auger@redhat.com>
[geert: Match using compatible values in sysfs instead of user-supplied
manufacturer/model options, reword]
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Tested-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Up to now the vfio-platform device has been abstract and could not be
instantiated. The integration of a new vfio platform device required
creating a dummy derived device which only set the compatible string.
Following the few vfio-platform device integrations we have seen the
actual requested adaptation happens on device tree node creation
(sysbus-fdt).
Hence remove the abstract setting, and read the list of compatible
values from sysfs if not set by a derived device.
Update the amd-xgbe and calxeda-xgmac drivers to fill in the number of
compatible values, as there can now be more than one.
Note that sysbus-fdt does not support the instantiation of the
vfio-platform device yet.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
[geert: Rebase, set user_creatable=true, use compatible values in sysfs
instead of user-supplied manufacturer/model options, reword]
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Tested-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
At the point smmu_find_add_as() gets called, the bus number might
not be computed. Let's change the name of IOMMU memory region and
just use the devfn and an incrementing index.
The name only is used for debug.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Message-id: 20180921070138.10114-2-eric.auger@redhat.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
[PMM: changed 'uint' to 'unsigned int']
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
smmu_iommu_mr() aims at returning the IOMMUMemoryRegion corresponding
to a given sid. The function extracts both the PCIe bus number and
the devfn to return this data. Current computation of devfn is wrong
as it only returns the PCIe function instead of slot | function.
Fixes 32cfd7f39e ("hw/arm/smmuv3: Cache/invalidate config data")
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Message-id: 1530775623-32399-1-git-send-email-eric.auger@redhat.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
When running dtc on the guest /proc/device-tree we get the
following warning: Warning (unit_address_vs_reg): Node /memory
has a reg or ranges property, but no unit name".
Let's fix that by adding the unit address to the node name. We also
don't create the /memory node anymore in create_fdt(). We directly
create it in load_dtb. /chosen still needs to be created in create_fdt
as the uart needs it. In case the user provided his own dtb, we nop
all memory nodes found in root and create new one(s).
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Message-id: 1530044492-24921-4-git-send-email-eric.auger@redhat.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
When running dtc on the guest /proc/device-tree we get the
following warnings: "Warning (unit_address_vs_reg): Node <name>
has a reg or ranges property, but no unit name", with name:
/intc, /intc/its, /intc/v2m.
Nodes should have a name in the form <name>[@<unit-address>] where
unit-address is the primary address used to access the device, listed
in the node's reg property. This fix seems to make dtc happy.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1530044492-24921-3-git-send-email-eric.auger@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
When an IOMMUMemoryRegion is in front of a virtio device,
address_space_cache_init does not set cache->ptr as the memory
region is not RAM. However when the device performs an access,
we end up in glue() which performs the translation and then uses
MAP_RAM. This latter uses the unset ptr and returns a wrong value
which leads to a SIGSEV in address_space_lduw_internal_cached_slow,
for instance.
In slow path cache->ptr is NULL and MAP_RAM must redirect to
qemu_map_ram_ptr((mr)->ram_block, ofs).
As MAP_RAM, IS_DIRECT and INVALIDATE are the same in _cached_slow
and non cached mode, let's remove those macros.
This fixes the use cases featuring vIOMMU (Intel and ARM SMMU)
which lead to a SIGSEV.
Fixes: 48564041a7 (exec: reintroduce MemoryRegion caching)
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Message-Id: <1528895946-28677-1-git-send-email-eric.auger@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Let's cache config data to avoid fetching and parsing STE/CD
structures on each translation. We invalidate them on data structure
invalidation commands.
We put in place a per-smmu mutex to protect the config cache. This
will be useful too to protect the IOTLB cache. The caches can be
accessed without BQL, ie. in IO dataplane. The same kind of mutex was
put in place in the intel viommu.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1529653501-15358-3-git-send-email-eric.auger@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
virt 3.0 now allows up to 512 vcpus whereas for earlier machine
types, max_cpus was set to 255 and any attempt to start the
machine with vcpus > 255 was rejected at a very early stage,
in vl.c/main level.
512 is the max supported by KVM. Anyway the actual vcpu count
that can be achieved depends on other parameters such as the
acceleration mode, the vgic version, the host kernel version.
Those are discovered later on.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Andrew Jones <drjones@redhat.com>
Message-id: 1529072910-16156-12-git-send-email-eric.auger@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
This patch defines a new ECAM region located after the 256GB limit.
The virt machine state is augmented with a new highmem_ecam field
which guards the usage of this new ECAM region instead of the legacy
16MB one. With the highmem ECAM region, up to 256 PCIe buses can be
used.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
Reviewed-by: Andrew Jones <drjones@redhat.com>
Message-id: 1529072910-16156-9-git-send-email-eric.auger@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
With a VGICv3 KVM device, if the number of vcpus exceeds the
capacity of the legacy redistributor region (123 redistributors),
we now attempt to register a second redistributor region. Up to
512 redistributors can fit in this latter on top of the 123 allowed
by the legacy redistributor region.
Registering this second redistributor region is possible if the
host kernel supports the following VGICv3 KVM device group/attribute:
KVM_DEV_ARM_VGIC_GRP_ADDR/KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION.
In case the host kernel does not support the registration of several
redistributor regions and the requested number of vcpus exceeds the
capacity of the legacy redistributor region, the GICv3 device
initialization fails with a proper error message and qemu exits.
At the moment the max number of vcpus still is capped by the
virt machine class max_cpus.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Andrew Jones <drjones@redhat.com>
Message-id: 1529072910-16156-8-git-send-email-eric.auger@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
This patch allows the creation of a GICv3 node with 1 or 2
redistributor regions depending on the number of smu_cpus.
The second redistributor region is located just after the
existing RAM region, at 256GB and contains up to up to 512 vcpus.
Please refer to kernel documentation for further node details:
Documentation/devicetree/bindings/interrupt-controller/arm,gic-v3.txt
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Andrew Jones <drjones@redhat.com>
Message-id: 1529072910-16156-6-git-send-email-eric.auger@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
To prepare for multiple redistributor regions, we introduce
an array of uint32_t properties that stores the redistributor
count of each redistributor region.
Non accelerated VGICv3 only supports a single redistributor region.
The capacity of all redist regions is checked against the number of
vcpus.
Machvirt is updated to set those properties, ie. a single
redistributor region with count set to the number of vcpus
capped by 123.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Andrew Jones <drjones@redhat.com>
Message-id: 1529072910-16156-4-git-send-email-eric.auger@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
for KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION attribute, the attribute
data pointed to by kvm_device_attr.addr is a OR of the
redistributor region address and other fields such as the index
of the redistributor region and the number of redistributors the
region can contain.
The existing machine init done notifier framework sets the address
field to the actual address of the device and does not allow to OR
this value with other fields.
This patch extends the KVMDevice struct with a new kda_addr_ormask
member. Its value is passed at registration time and OR'ed with the
resolved address on kvm_arm_set_device_addr().
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1529072910-16156-3-git-send-email-eric.auger@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Coverity complains about use of uninitialized Evt struct.
The EVT_SET_TYPE and similar setters use deposit32() on fields
in the struct, so they read the uninitialized existing values.
In cases where we don't set all the fields in the event struct
we'll end up leaking random uninitialized data from QEMU's
stack into the guest.
Initializing the struct with "Evt evt = {};" ought to satisfy
Coverity and fix the data leak.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reported-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-id: 1526493784-25328-2-git-send-email-eric.auger@redhat.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
We set up the infrastructure to enumerate all the PCI devices
attached to the SMMU and create an associated IOMMU memory
region and address space.
Those info are stored in SMMUDevice objects. The devices are
grouped according to the PCIBus they belong to. A hash table
indexed by the PCIBus pointer is used. Also an array indexed by
the bus number allows to find the list of SMMUDevices.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Prem Mallappa <prem.mallappa@broadcom.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1524665762-31355-3-git-send-email-eric.auger@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Commit 567b5b309a ("vfio/pci: Relax DMA map errors for MMIO regions")
added an error message if a passed memory section address or size
is not aligned to the page size and thus cannot be DMA mapped.
This patch fixes the trace by printing the region name and the
memory region section offset within the address space (instead of
offset_within_region).
We also turn the error_report into a trace event. Indeed, In some
cases, the traces can be confusing to non expert end-users and
let think the use case does not work (whereas it works as before).
This is the case where a BAR is successively mapped at different
GPAs and its sections are not compatible with dma map. The listener
is called several times and traces are issued for each intermediate
mapping. The end-user cannot easily match those GPAs against the
final GPA output by lscpi. So let's keep those information to
informed users. In mid term, the plan is to advise the user about
BAR relocation relevance.
Fixes: 567b5b309a ("vfio/pci: Relax DMA map errors for MMIO regions")
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Voiding the ITS caches is not supposed to happen via
individual register writes. So we introduced a dedicated
ITS KVM device ioctl to perform a cold reset of the ITS:
KVM_DEV_ARM_VGIC_GRP_CTRL/KVM_DEV_ARM_ITS_CTRL_RESET. Let's
use this latter if the kernel supports it.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1511883692-11511-5-git-send-email-eric.auger@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
At the moment the ITS is not properly reset and this causes
various bugs on save/restore. We implement a minimalist reset
through individual register writes but for kernel versions
before v4.15 this fails voiding the vITS cache. We cannot
claim we have a comprehensive reset (hence the error message)
but that's better than nothing.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1511883692-11511-3-git-send-email-eric.auger@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Eric Auger