Implement the infrastructure that lets a virtio-mem device map the guest
memory into the device. This is necessary since with virtio-mem zones
memory can be added or removed and the vfio-user device must be
informed.
Fixes: #3025
Signed-off-by: Rob Bradford <robert.bradford@intel.com>
By moving this from the VfioUserPciDevice to DeviceManager the client
can be reused for handling DMA mapping behind an IOMMU.
Signed-off-by: Rob Bradford <robert.bradford@intel.com>
Looking up devices on the port I/O bus is time consuming during the
boot at there is an O(lg n) tree lookup and the overhead from taking a
lock on the bus contents.
Avoid this by adding a fast path uses the hardcoded port address and
size and directs PCI config requests directly to the device.
Command line:
target/release/cloud-hypervisor --kernel ~/src/linux/vmlinux --cmdline "root=/dev/vda1 console=ttyS0" --serial tty --console off --disk path=~/workloads/focal-server-cloudimg-amd64-custom-20210609-0.raw --api-socket /tmp/api
PIO exit: 17913
PCI fast path: 17871
Percentage on fast path: 99.8%
perf before:
marvin:~/src/cloud-hypervisor (main *)$ perf report -g | grep resolve
6.20% 6.20% vcpu0 cloud-hypervisor [.] vm_device:🚌:Bus::resolve
perf after:
marvin:~/src/cloud-hypervisor (2021-09-17-ioapic-fast-path *)$ perf report -g | grep resolve
0.08% 0.08% vcpu0 cloud-hypervisor [.] vm_device:🚌:Bus::resolve
The compromise required to implement this fast path is bringing the
creation of the PciConfigIo device into the DeviceManager::new() so that
it can be used in the VmmOps struct which is created before
DeviceManager::create_devices() is called.
Signed-off-by: Rob Bradford <robert.bradford@intel.com>
This message only occurs sporadically and so it should be included at
info!() level. Enhance the output to also include the BAR number.
Signed-off-by: Rob Bradford <robert.bradford@intel.com>
This resolves an issue with hotplug -> removal -> hotplug of a vfio-user
device as the allocator was not updated with the now unused entries.
Signed-off-by: Rob Bradford <robert.bradford@intel.com>
When mapping the region into the guest ensure that all the fields are
updated correctly as the unmap code path checks that they are set.
Signed-off-by: Rob Bradford <robert.bradford@intel.com>
Taking advantage of the refactored VFIO code implement a new
VfioUserPciDevice that wraps the client for vfio-user and exposes the
BusDevice and PciDevice into the VMM.
Signed-off-by: Rob Bradford <robert.bradford@intel.com>
After the refactoring to split the common VFIO code out for vfio-user
there were some inconsistencies in the error handling. Correct this so
that the error is independent of the transport (hardware vs user) VFIO
and migrate to anyhow/thiserror in the process. Some unused errors from
earlier refactoring have also been removed.
Signed-off-by: Rob Bradford <robert.bradford@intel.com>
Returning a reference is not possible for the vfio-user code as it is
constructed for the function.
Signed-off-by: Rob Bradford <robert.bradford@intel.com>
Rename the wrapper trait and structs since this will be used for more
than reading the PCI config.
Signed-off-by: Rob Bradford <robert.bradford@intel.com>
This allows the code to be used from a different module in the same
crate for vfio-user support.
Signed-off-by: Rob Bradford <robert.bradford@intel.com>
Split data that will need to be common between VfioPciDevice and
VfioUserPciDevice into a common struct. Currently this has no methods
but they will be added soon.
Signed-off-by: Rob Bradford <robert.bradford@intel.com>
By splitting this into a trait with common code extracted then this
will allow extensive reuse of logic in the vfio-user version.
This commit also changed the order of parameters on
::write_config_dword() to place offset first to match the other
functions.
Signed-off-by: Rob Bradford <robert.bradford@intel.com>
With the new beta version, clippy complains about redundant allocation
when using Arc<Box<dyn T>>, and suggests replacing it simply with
Arc<dyn T>.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
The BAR calculation code was incorrect for calculating I/O BARs but also
has misleading comments (mixing bits and bytes, first and least
significant, etc).
This change adjusts the algorithm to more closely match the version
described in the PCI specification and takes advantage of Rust's binary
literals for ease of reading. Although this is slightly longer by
calculating the 64-bit and 32-bit paths separately I think this is
easier to read.
Signed-off-by: Rob Bradford <robert.bradford@intel.com>
Previously the same function was used to both create and remove regions.
This worked on KVM because it uses size 0 to indicate removal.
MSHV has two calls -- one for creation and one for removal. It also
requires having the size field available because it is not slot based.
Split set_user_memory_region to {create/remove}_user_memory_region. For
KVM they still use set_user_memory_region underneath, but for MSHV they
map to different functions.
This fixes user memory region removal on MSHV.
Signed-off-by: Wei Liu <liuwe@microsoft.com>
Now all crates use edition = "2018" then the majority of the "extern
crate" statements can be removed. Only those for importing macros need
to remain.
Signed-off-by: Rob Bradford <robert.bradford@intel.com>
The logic wasn't quite right, as it wasn't detecting BAR reprogramming
when the upper part of the address was identical. For instance, a BAR
moved from 0x7fc0000000 to 0x7fd0000000 wasn't detected properly.
The logic has been updated and cleaned up to fix this issue, which was
observed when running Windows guests. This fixes the network hotplug
support as well.
Fixes#1797Fixes#1798
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
Simplify snapshot & restore code by using generics to specify helper
functions that take / make a Serialize / Deserialize struct
Signed-off-by: Rob Bradford <robert.bradford@intel.com>
warning: name `IORegion` contains a capitalized acronym
--> pci/src/configuration.rs:320:5
|
320 | IORegion = 0x01,
| ^^^^^^^^ help: consider making the acronym lowercase, except the initial letter (notice the capitalization): `IoRegion`
|
= help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#upper_case_acronyms
Signed-off-by: Rob Bradford <robert.bradford@intel.com>
Now that virtio-mem devices can update VFIO mappings through dedicated
handlers, let's provide them from the DeviceManager.
Important to note these handlers should either be provided to virtio-mem
devices or to the unique virtio-iommu device. This must be mutually
exclusive.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
Instead of letting the VfioPciDevice take the decision on how/when to
perform the DMA mapping/unmapping, we move this to the DeviceManager
instead.
The point is to let the DeviceManager choose which guest memory regions
should be mapped or not. In particular, we don't want the virtio-mem
region to be mapped/unmapped as it will be virtio-mem device
responsibility to do so.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
This commit moves both pci and vmm code from the internal vfio-ioctls
crate to the upstream one from the rust-vmm project.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
In case the VFIO device does not support MSI or MSI-X, the capabilities
should not be parsed, avoiding the exposure of unsupported capabilities.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
Make sure to propagate the error coming from VfioDevice when trying to
enable INTx, MSI or MSI-X interrutps.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
With all the preliminary work done in the previous commits, we can
update the VFIO implementation to support INTx along with MSI and MSI-X.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
In anticipation for supporting the notifier function for the legacy
interrupt source group, we need this function to return an EventFd
instead of a reference to this same EventFd.
The reason is we can't return a reference when there's an Arc<Mutex<>>
involved in the call chain.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
We need to be able to return the barrier from the code that prepares to
activate the virtio device. This triggered by a write to the
configuration fields stored in the PCI BAR. Since bars can be accessed
by both memory mapping and through PCI config I/O several prototypes
must be changed.
Signed-off-by: Rob Bradford <robert.bradford@intel.com>
This can be uses to indicate to the caller that it should wait on the
barrier before returning as there is some asynchronous activity
triggered by the write which requires the KVM exit to block until it's
completed.
This is useful for having vCPU thread wait for the VMM thread to proceed
to activate the virtio devices.
See #1863
Signed-off-by: Rob Bradford <robert.bradford@intel.com>
This is a new clippy check introduced in 1.47 which requires the use of
the matches!() macro for simple match blocks that return a boolean.
Signed-off-by: Rob Bradford <robert.bradford@intel.com>
By looking at Linux kernel boot time, we identified that a lot of time
was spent registering and unregistering IRQ fds to KVM. This is not
efficient and certainly not a wrong behavior from the Linux kernel,
but rather a problem with the Cloud-Hypervisor's implementation of
MSI-X.
The way to fix this issue is by ensuring the initial conditions are
correct, which means the entire MSI-X vector table must be disabled
and masked. Additionally, each vector must be individually masked.
With these correct conditions, Linux won't start masking interrupt
vectors, and later unmask them since they will be seen as masked from
the beginning. This means the OS will simply have to unmask them when
needed, avoiding the extra operation.
Another aspect of this patch is to prevent Cloud-Hypervisor from
enabling (by registering IRQ fd) all vectors when either the global
'mask' or 'enable' bits are set. Instead, we can simply let the mask()
and unmask() operations take care of it if needed.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
Misspellings were identified by https://github.com/marketplace/actions/check-spelling
* Initial corrections suggested by Google Sheets
* Additional corrections by Google Chrome auto-suggest
* Some manual corrections
Signed-off-by: Josh Soref <jsoref@users.noreply.github.com>
This removes the dependency of the pci crate on the devices crate which
now only contains the device implementations themselves.
Signed-off-by: Rob Bradford <robert.bradford@intel.com>
There will be some cases where the implementation of the snapshot()
function from the Snapshottable trait will require to modify some
internal data, therefore we make this possible by updating the trait
definition with snapshot(&mut self).
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
This simplies some of the handling for PCI BARs particularly with
respect to snapshot and restore. No attempt has been made to handle the
64-bit bar handling in a different manner to that which was used before.
Fixes: #1153
Signed-off-by: Rob Bradford <robert.bradford@intel.com>
In this commit we saved the BDF of a PCI device and set it to "devid"
in GSI routing entry, because this field is mandatory for GICv3-ITS.
Signed-off-by: Michael Zhao <michael.zhao@arm.com>
The type is now hypervisor::Vm. Switch from KVM specific name vm_fd to a
generic name just like 8186a8eee6 ("vmm: interrupt: Rename vm_fd").
No functional change.
Signed-off-by: Wei Liu <liuwe@microsoft.com>
Start moving the vmm, arch and pci crates to being hypervisor agnostic
by using the hypervisor trait and abstractions. This is not a complete
switch and there are still some remaining KVM dependencies.
Signed-off-by: Muminul Islam <muislam@microsoft.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>
Currently, not every feature of the cloud-hypervisor is enabled
on AArch64, which means that on AArch64 machines, the
`run_unit_tests.sh` needs to be tailored and some unit test cases
should be run on x86_64 only.
Also this commit fixes the typo and unifies `Arm64` and `AArch64`
in the AArch64 document.
Signed-off-by: Henry Wang <Henry.Wang@arm.com>
cloud-hypervisor: 763.978581807s: ERROR:pci/src/vfio.rs:651 -- failed to remove all guest memory regions from iommu table
when poweroff a vm with vfio device, clh will finally remove all guest memory region from iommu table
with the method unset_dma_map, not method setup_dma_map.
Signed-off-by: LiYa'nan <oliverliyn@gmail.com>
There is a much stronger PCI dependency from vfio_pci.rs than a VFIO one
from pci/src/vfio.rs. It seems more natural to have the PCI specific
VFIO implementation in the PCI crate rather than the other way around.
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>
In order to let the PciBus user choose where a device should be placed
on the bus, a new function get_device_id() is introduced. This will be
helpful in the context of snapshot/restore as the caller will be able to
place the PCI devices on the same slot they were placed before the
snapshot was taken.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
The PCI configuration from each PCI device is modified at runtime as we
can expect the guest OS to write to some PCI capability structure, or
move the BAR to a different location in the guest address space.
For all the reasons why such configuration might differ from the initial
configuration, we must store the registers values to be able to restore
them with the right values whenever a PCI device is being restored.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
In order to restore devices relying on MSI-X, the MsixConfig structure
must be restored with the correct values. Additionally, the KVM routes
must be restored so that interrupts can be delivered through KVM the way
they were configured before the snapshot was taken.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
When reporting the BAR size it is necessary to return a value that is
encoded such that all the bits are set that represent the mask of the
natural alignment of the field.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
Signed-off-by: Rob Bradford <robert.bradford@intel.com>
Add support for specifying the PCI revision in the PCI configuration and
populate this with the value of 1 for virtio-pci devices.
The virtio-pci specification is slightly ambiguous only saying that
transitional (i.e. devices that support legacy and virtio 1.0) should
set this to 0. In practice it seems that software expects the revision
to be set to 1 for modern only devices.
Signed-off-by: Rob Bradford <robert.bradford@intel.com>
By using a Vec to hold the list of devices on the PciBus, there's a
problem when we use unplug. Indeed, the vector of devices gets reduced
and if the unplugged device was not the last one from the list, every
other device after this one is shifted on the bus.
To solve this problem, a HashMap is used. This allows to keep track of
the exact place where each device stands on the bus.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
The point of this new method is to let the caller decide when the
implementation of the PciDevice should free the BARs previously
allocated through the other method allocate_bars().
This provides a way to perform proper cleanup for any PCI device.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
Upon removal of a PCI device, make sure we don't hold onto the device ID
as it could be reused for another device later.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
In order to handle the case where devices are very often plugged and
unplugged from a VM, we need to handle the PCI device ID allocation
better.
Any PCI device could be removed, which means we cannot simply rely on
the vector size to give the next available PCI device ID.
That's why this patch stores in memory the information about the 32
slots availability. Based on this information, whenever a new slot is
needed, the code can correctly provide an available ID, or simply return
an error because all slots are taken.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
Simple function relying on the retain() method from std::Vec, allowing
to remove every occurence of the same device.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
Now that the BusDevice devices are stored as Weak references by the IO
and MMIO buses, there's no need to use Weak references from the PciBus
anymore.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
We want to prevent from losing interrupts while they are masked. The
way they can be lost is due to the internals of how they are connected
through KVM. An eventfd is registered to a specific GSI, and then a
route is associated with this same GSI.
The current code adds/removes a route whenever a mask/unmask action
happens. Problem with this approach, KVM will consume the eventfd but
it won't be able to find an associated route and eventually it won't
be able to deliver the interrupt.
That's why this patch introduces a different way of masking/unmasking
the interrupts, simply by registering/unregistering the eventfd with the
GSI. This way, when the vector is masked, the eventfd is going to be
written but nothing will happen because KVM won't consume the event.
Whenever the unmask happens, the eventfd will be registered with a
specific GSI, and if there's some pending events, KVM will trigger them,
based on the route associated with the GSI.
Suggested-by: Liu Jiang <gerry@linux.alibaba.com>
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
We should not assume the offset produced by ECAM is identical to the
CONFIG_ADDRESS register of legacy PCI port io enumeration.
Signed-off-by: Qiu Wenbo <qiuwenbo@phytium.com.cn>
In order to anticipate the need to support more features related to the
access of a device's PCI config space, this commits changes the self
reference in the function read_config_register() to be mutable.
This also brings some more flexibility for any implementation of the
PciDevice trait.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
Both InterruptDelivery and InterruptParameters can be removed from the
pci crate as they are not used anymore.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
There's no need for assign_irq() or assign_msix() functions from the
PciDevice trait, as we can see it's never used anywhere in the codebase.
That's why it's better to remove these methods from the trait, and
slightly adapt the existing code.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
Now that KVM specific interrupts are handled through InterruptManager
trait implementation, the pci crate does not need to rely on kvm_ioctls
and kvm_bindings crates.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
Based on all the previous changes, we can at this point replace the
entire interrupt management with the implementation of InterruptManager
and InterruptSourceGroup traits.
By using KvmInterruptManager from the DeviceManager, we can provide both
VirtioPciDevice and VfioPciDevice a way to pick the kind of
InterruptSourceGroup they want to create. Because they choose the type
of interrupt to be MSI/MSI-X, they will be given a MsiInterruptGroup.
Both MsixConfig and MsiConfig are responsible for the update of the GSI
routes, which is why, by passing the MsiInterruptGroup to them, they can
still perform the GSI route management without knowing implementation
details. That's where the InterruptSourceGroup is powerful, as it
provides a generic way to manage interrupt, no matter the type of
interrupt and no matter which hypervisor might be in use.
Once the full replacement has been achieved, both SystemAllocator and
KVM specific dependencies can be removed.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
Callbacks are not the most Rust idiomatic way of programming. The right
way is to use a Trait to provide multiple implementation of the same
interface.
Additionally, a Trait will allow for multiple functions to be defined
while using callbacks means that a new callback must be introduced for
each new function we want to add.
For these two reasons, the current commit modifies the existing
VirtioInterrupt callback into a Trait of the same name.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
At this point, both MSI and MSI-X handle the KVM GSI routing update,
which means the vfio crate does not have to deal with it anymore.
Therefore, several functions can be removed from the vfio-pci code, as
they are not needed anymore.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
Now that MsiConfig has access to both KVM VmFd and the list of GSI
routes, the update of the KVM GSI routes can be directly done from
MsiConfig instead of specifically from the vfio-pci implementation.
By moving the KVM GSI routes update at the MsiConfig level, any PCI
device such as vfio-pci, virtio-pci, or any other emulated PCI device
can benefit from it, without having to implement it on their own.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
The same way we have MsixConfig in charge of managing whatever relates
to MSI-X vectors, we need a MsiConfig structure to manage MSI vectors.
The MsiCap structure is still needed as a low level API, but it is now
part of the MsiConfig which oversees anything related to MSI.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
In order to factorize one step further, we let MsixConfig perform the
interrupt enabling/disabling. This is done by registering/unregistering
the KVM irq_fds of all GSI routes related to this device.
And now that MsixConfig is in charge of the irq_fds, vfio-pci must rely
on it to retrieve them and provide them to the vfio driver.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
Now that MsixConfig has access to the irq_fd descriptors associated with
each vector, it can directly write to it anytime it needs to trigger an
interrupt.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
Now that MsixConfig has access to both KVM VmFd and the list of GSI
routes, the update of the KVM GSI routes can be directly done from
MsixConfig instead of specifically from the vfio-pci implementation.
By moving the KVM GSI routes update at the MsixConfig level, both
vfio-pci and virtio-pci (or any other emulated PCI device) can benefit
from it, without having to implement it on their own.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
Because MsixConfig will be responsible for updating KVM GSI routes at
some point, it is necessary that it can access the list of routes
contained by gsi_msi_routes.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
Because MsixConfig will be responsible for updating the KVM GSI routes
at some point, it must have access to the VmFd to invoke the KVM ioctl
KVM_SET_GSI_ROUTING.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
The point here is to let MsixConfig take care of the GSI allocation,
which means the SystemAllocator must be passed from the vmm crate all
the way down to the pci crate.
Once this is done, the GSI allocation and irq_fd creation is performed
by MsixConfig directly.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
In order to anticipate the need for both msi.rs and msix.rs to rely on
some KVM utils and InterruptRoute structure to handle the update of the
KVM GSI routes, this commit adds these utilities directly to the pci
crate. So far, these were exclusively used by the vfio crate, which is
why there were located there.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
The current code was always considering 0xffffffff being written to the
register as a sign it was expecting to get the size, hence the BAR
reprogramming detection was stating this case was not a reprogramming
case.
Problem is, if the value 0xffffffff is directed at a 64bits BAR, this
might be the high or low part of a 64bits address which is meant to be
the new address of the BAR, which means we would miss the detection of
the BAR being reprogrammed here.
This commit improves the code using finer granularity checks in order to
detect this corner case correctly.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
