The RwLock implementation on linux has a tendency to starve writers.
Since the fs device workload has a roughly evenly distributed number of
reads and writes, using a Mutex should give us better performance.
BUG=b:150264042
TEST=vm.Blogbbench.virtiofs
Change-Id: I85ec4053bf03d19bb21b420c0aa506720e666708
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2103604
Auto-Submit: Chirantan Ekbote <chirantan@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Stephen Barber <smbarber@chromium.org>
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Using the `open_inode` method on an fd for a symlink results in the
kernel returning -ELOOP. Since there are no `*at` methods for extended
attributes, manually read the path for the file and then use the
l{get,set,list,remove}xattr method on the returned path.
BUG=b:136128512
TEST=boot arcvm with virtio-fs and selinux enabled
Change-Id: I2fde57db8a075838a3a877309f6cf89059f19258
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2120763
Auto-Submit: Chirantan Ekbote <chirantan@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Stephen Barber <smbarber@chromium.org>
Reviewed-by: Stephen Barber <smbarber@chromium.org>
`offset_dst` is supposed to be a pointer.
BUG=none
TEST=none
Change-Id: I033501ba5a57a130625e68be88457b15ad1484ef
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2120762
Auto-Submit: Chirantan Ekbote <chirantan@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
The current vhost-net msix irq injection flow is from vhost-kernel
to crosvm vhost-net, then to the KVM for irq injection. It still need
crosvm vhost-net to trigger irq, which is because the set_vring_call
is not directly using the msix irq fd.
To optimize this flow to be from vhost-kernel to KVM directly, need:
1. if the msix is enabled and unmasked, use the misx irq fd for the
vring_call directly so that all the misx injection can directly to
KVM from vhost-kernel.
2. if the msix is disabled or masked, use the indirect vhost_interrupt
fd to let the crosvm to control the irq injection.
BUG=None
TEST=cargo test -p devices
TEST=start crosvm with vhost-net, and run the iperf3 on the network
without any issue
Change-Id: Idb3427f69f23b728305ed63d88973156a03e7c6b
Signed-off-by: Chuanxiao Dong <chuanxiao.dong@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2046452
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Pass the Option of the response socket should be used by the
activate thread, to communicate with its device model.
BUG=None
TEST=cargo test -p devices
Change-Id: I929f4c901468e920116f2a744ec73571d91080e3
Signed-off-by: Chuanxiao Dong <chuanxiao.dong@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2046451
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Add a pair of control socket for vhost net. This pair can be
used for the vhost-net device model to control and
communicate with its activate thread.
BUG=None
TEST=cargo test -p devices
Change-Id: I8bacdb9132787dc499ef00eea1df26ff3b35028d
Signed-off-by: Chuanxiao Dong <chuanxiao.dong@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2046450
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
On devices with only one configuration, skip the code that attempts to
change the device's active configuration, since it must always be the
single available configuration.
This works around an issue observed with some USB devices (e.g. Servo
Micro) where the initial Get Configuration control request fails with
-EPIPE.
BUG=chromium:1061382
BUG=b:151408644
TEST=Attach servo micro, see /dev/ttyUSB[012], screen /dev/ttyUSB0
Change-Id: Ic3333e1d70a0c57b090de64e4d3b7932ce2cf81d
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2108871
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Tested-by: George Engelbrecht <engeg@google.com>
Commit-Queue: George Engelbrecht <engeg@google.com>
* Remove RESOURCE_V2_UNREF
* Add RESOURCE_MAP/RESOURCE_UNMAP to enable resources without guest
storage that don't need to be mapped directly either
BUG=chromium:924405
TEST=compile and test
Change-Id: I10d6cd120d86131fa7ed8917ddad25cdb99ae50c
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2015587
Tested-by: Gurchetan Singh <gurchetansingh@chromium.org>
Auto-Submit: Gurchetan Singh <gurchetansingh@chromium.org>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Commit-Queue: Gurchetan Singh <gurchetansingh@chromium.org>
Rebase of zero-copy virtio-gpu flow:
* Removes guest_memory_type/guest_caching_type in favor of a bitmask
* Removes ALLOCATION_METADATA, since ideally we'd just read from
guest memory to get guest responses
* Renames HOST_COHERENT to HOST_VISIBLE
* Adds a few more feature flags
BUG=chromium:924405
TEST=compile
Change-Id: I0d5a84b66cfa6d09f7e2d07ed8e761e7ba850284
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2013767
Tested-by: Gurchetan Singh <gurchetansingh@chromium.org>
Commit-Queue: Gurchetan Singh <gurchetansingh@chromium.org>
Reviewed-by: Lingfeng Yang <lfy@google.com>
Reviewed-by: Zach Reizner <zachr@chromium.org>
1. Replace --cras-audio, --cras-capture, null-audio options by --ac97
option to create audio devices.
2. "--ac97 backend=BACKEND\
[capture=true,capture_effect=EFFECT]" is comma
separated key=value pairs for setting up Ac97 devices.
It can be given more than once to create multiple devices. Possible
key values are:
backend=(null, cras) - Where to route the audio device.
`null` for /dev/null, and cras for CRAS server.
capture=true - Enable audio capture.
capture_effects - | separated effects to be enabled
for recording. The only supported effect value now is EchoCancellation
or aec.
BUG=b:140866281
TEST=1.crosvm run -r ./vm_rootfs.img -c 4 -m 1024 -s /run --cid 5 --host_ip\
100.115.92.25 --netmask 255.255.255.252 --ac97\
backend=cras,capture=true,capture_effect=aec\
--mac d2:47:f7:c5:9e:53 ./vm_kernel
2. Record with the vm by:
arecord -D hw:0,0 -d5 -fS16_LE -c2 -r48000 /tmp/test.mp3
3. Verify that AEC is enabled within the recording stream by
cras_test_cleint.
Cq-Depend: chromium:2053654
Cq-Depend: chromium:2095644
Cq-Depend: chromium:2038221
Change-Id: Ia9e0e7cda1671a4842ec77a354efaa4a2dc745eb
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2038413
Tested-by: Judy Hsiao <judyhsiao@chromium.org>
Commit-Queue: Judy Hsiao <judyhsiao@chromium.org>
Reviewed-by: Chih-Yang Hsia <paulhsia@chromium.org>
Auto-Submit: Judy Hsiao <judyhsiao@chromium.org>
This is a virtio device that provides a serial console. It has
constructors matching the existing Serial device (new_in_out, new_out,
and new_sink) that take generic io::Read and io::Write streams.
This change just adds the device code; additional changes are required
to add the console device to the command-line parsing and device setup
code.
BUG=chromium:1059924
TEST=boot linux with console=hvc0
Change-Id: I917157d5ecb5160c9b00b499eabe6fb08486776c
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2095534
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
Add new definitions and constants to support fuse minor version 31.
These include the FUSE_SETUPMAPPING and FUSE_REMOVEMAPPING opcodes used
by the virtio-fs driver for implementing DAX support.
BUG=b:147341783
TEST=vm.Virtiofs
Change-Id: Ie59ec1a44e555910f2ee2c5ba7afccb8bd435db9
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2105823
Tested-by: Chirantan Ekbote <chirantan@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Commit-Queue: Chirantan Ekbote <chirantan@chromium.org>
Add support for FOPEN_CACHE_DIR so that the guest can cache directory
entries for longer.
BUG=b:150264964
TEST=vm.Virtiofs
Change-Id: Iade67b54084ed72378afa70af9e9e0f7f0bc03e8
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2105821
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Chirantan Ekbote <chirantan@chromium.org>
rustc now warns about return statements that have an extra set of
parenthesis. Remove such instances so that the code is warning free.
TEST=cargo build completes without warnings
Change-Id: I55148f8aceca8ba90f6bead2b6929e2c843351aa
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2104767
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Tested-by: Dylan Reid <dgreid@chromium.org>
Commit-Queue: Dylan Reid <dgreid@chromium.org>
Specifying O_DIRECT in the 9p device doesn't actually work correctly and
leads to an error. O_DIRECT handling in the fs device works correctly
but also makes it look much worse in disk I/O benchmarks because the
block device gets the benefit of the host cache while the fs device
depends on the performance of the actual storage device.
BUG=none
TEST=`tast run vm.Fio.*`
Change-Id: I738e4032081e331ef956c9d4c33616607e403d86
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2093967
Commit-Queue: Chirantan Ekbote <chirantan@chromium.org>
Tested-by: Chirantan Ekbote <chirantan@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Add an `iter` method to the `Reader` struct so that callers can iterate
over the objects in place rather than having to store them in a separate
collection.
BUG=none
TEST=unit tests
Change-Id: I29671910a4244a8d7786ca2eb241416ae72b8c9f
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2093966
Tested-by: Chirantan Ekbote <chirantan@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Commit-Queue: Chirantan Ekbote <chirantan@chromium.org>
Transfer plane offsets and strides for exported GPU resource over resource
bridge as well as a resource itself.
These metadata will be required by virtio-video decoder and encoder.
BUG=b:120456557
TEST=Start ARCVM on atlas
Change-Id: Iaf539857c0f8525bd5be294521e75ad32cae05e7
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1787032
Reviewed-by: Keiichi Watanabe <keiichiw@chromium.org>
Reviewed-by: David Stevens <stevensd@chromium.org>
Reviewed-by: Chirantan Ekbote <chirantan@chromium.org>
Commit-Queue: Keiichi Watanabe <keiichiw@chromium.org>
Tested-by: Keiichi Watanabe <keiichiw@chromium.org>
The test is flaky and causes continuous build failures.
BUG=chromium:1058881
TEST=Tests run, offending test is listed as ignored.
Exempt-From-Owner-Approval: Sheriff action to silence flakes on builders
Change-Id: I204e6ef548e6f203b0c15b0d01fde3b88660bd44
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2090414
Reviewed-by: Mattias Nissler <mnissler@chromium.org>
Commit-Queue: Mattias Nissler <mnissler@chromium.org>
Tested-by: Mattias Nissler <mnissler@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
VIRTIO_RING_F_EVENT_IDX use Wrapping(u16) for irq suppressing, this
test case to avoid some corner case for Wrapping.
BUG=None
TEST=Run build_test.py
Change-Id: I47d377056fefcc36739bb197e30319deafb0faf4
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2073902
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Xiong Zhang <xiong.y.zhang@intel.corp-partner.google.com>
current one container contains one group only, but one container could
contain multi groups actually. The main gap that current code to
support multi groups is that container will be initialized multi times
when multi groups exist, as each group will initialize container one time.
This patch extracts the code which should run one time only on a
container, so when the first group is added into container, this
container initialize code will run once. The container once initialize
code contains:
a. Set iommu driver type as VfioType1V2
b. Setup Iommu table on each guest memory region
c. create vfio_kvm device, so kernel kvm and vfio is associated.
BUG=chromium:992270
TEST=passthrough two/three vfio devices into guest, these devices belong
to different vfio groups, then check these devices function in guest.
Change-Id: I94c9c86f70f49957a5e5c1dfd2c7d823ad042320
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2078970
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Xiong Zhang <xiong.y.zhang@intel.corp-partner.google.com>
Previously, all input events in CrosVM were required to be linux
input_events, which have a timestamp field that is actually unused by
when we send/receive from the guest which are of type
virtio_input_event. This CL allows CrosVM to understand both types of input
events in a first class manner. It is a follow up on
https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1930405.
This CL also addresses some bugs with window driven input:
1. attach_event_device was being called before the surface was
created, so the devices were never attached.
2. The default touchpad size was not being set to the display window
size.
Additionally, it removes the unused event "filter" feature on event
sources.
Breaking change: from this point forward, CrosVM will treat input events sent
via a socket (e.g. SocketEventSource) to be virtio_input_events.
BUG=None
TEST=builds + manual
Change-Id: I7fec07c582e5a071a6f116975ba70d6e621bb483
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2034046
Reviewed-by: Zach Reizner <zachr@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Noah Gold <nkgold@google.com>
Devices use irqfd to inject interrupts, we listen to them in the main
thread and activate userspace pic/ioapic accordingly.
BUG=chromium:908689
TEST=lanuch linux guest with `--split-irqchip` flag
Change-Id: If30d17ce7ec9e26dba782c89cc1b9b2ff897a70d
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1945798
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Commit-Queue: Zhuocheng Ding <zhuocheng.ding@intel.corp-partner.google.com>
Previous interrupt suppress patch only supply crude interrupt suppress,
VIRTIO_RING_F_EVENT_IDX feature supply a more performant alternative:
1) where the driver specifies how far the device can progress before a
notification is required
2) where the device specifies how far the driver can progress before a
interrrupt is required.
This patch add this feature into blk.
For gpu and network, this could be added also, but gpu and network
performance don't get better.
BUG=None
TEST=run benchmark for blk in guest
Change-Id: I73fe3f8b72a9e88fd6073890bc6ab2bee891d51d
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2008341
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Xiong Zhang <xiong.y.zhang@intel.corp-partner.google.com>
TODO: Route irqfd to PIC, and use signal to kick vCPU thread when
interrupt is triggered.
BUG=chromium:908689
TEST=Unit tests in file.
Change-Id: I9a87502da57e725d3bb26d746a337d0ba44ef337
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1945797
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Zhuocheng Ding <zhuocheng.ding@intel.corp-partner.google.com>
Virtio spec define some features like VIRTIO_F_RING_INDIRECT_DESC,
VIRTIO_F_RING_EVENT_IDX, VIRTIO_F_RING_PACKED. If one feature is enabled
in guest, device's queue will have corresponding behavior for this
feature. Queue's ack_features() interface let queue know which features
is enabled in guest.
BUG=None
TEST=build_test
Change-Id: I865f93940a1f4db8ca6da8829136127353cc1c6f
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2008340
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Xiong Zhang <xiong.y.zhang@intel.corp-partner.google.com>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Convert playback and capture for the AC97 device to use the zero-copy
ShmStreamSource instead of the old StreamSource.
In the process, rework start_playback and start_capture unit tests so
they rely less on sleep statements.
BUG=chromium:968724
BUG=chromium:1006035
TEST="sox -n -r 48000 -b 16 output.raw synth 5 sine 330 &&
aplay -f dat output.raw" within a VM, check that sine wave is played
accurately.
Change-Id: Ie9cddbc5285a9505872c9951a8a1da01de70eb88
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1749950
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Fletcher Woodruff <fletcherw@chromium.org>
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Reviewed-by: Chih-Yang Hsia <paulhsia@chromium.org>
Add acpi_rsdp_addr in boot_params to allow crosvm to pass
a physicall address of RSDP to the Linux guest kernel, so
that the linux guest kernel can parse the constructed ACPI
tables and enable the ACPI.
Although there is ACPI tables but as we still have "acpi=off"
in command line parameter, there is still no ACPI in guest kernel.
The ACPI construction refers to the implementation of the
Cloud-hypervisor commit:
- arch: x86_64: Generate basic ACPI tables
BUG=chromium:1018674
TEST=None
Change-Id: Ibcb2ae98c43da8ef8c07a07eda9213f61570d14c
Signed-off-by: Chuanxiao Dong <chuanxiao.dong@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2035351
Reviewed-by: Tomasz Jeznach <tjeznach@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Tomasz Jeznach <tjeznach@chromium.org>
Change the behavior of the resize operation on virtio-block devices so
that it causes a disk to become fully allocated (non-sparse) even if it
had previously been sparse.
This means that we could have a disk that was previously sparse and is
now non-sparse, so treat discard requests for sparse disks as a no-op
instead of an error. This is acceptable since discard is a hint and
doing nothing is a valid implementation.
BUG=chromium:858815
TEST=`crosvm disk resize` a sparse disk
Change-Id: I8b79e460e5432cc71bed98172527fe1cd2e726ed
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2070846
Reviewed-by: David Munro <davidmunro@google.com>
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
This more gracefully handles failure of msix_enable; in particular, if
it fails, the self.enabled state is returned to false so that future
device operations won't try to access uninitialized parts of
self.irq_vec.
In addition, the AddMsiRoute response is now checked (previously, it was
just ignored), and errors are bubbled up via MsixError rather than just
printing a message.
BUG=chromium:1041351
TEST=Boot Crostini on nami
Change-Id: I9999f149817bc9f49176487446b52e74fb8be9a9
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2067964
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
If vfio device's bar is mmappable, vcpu could access it directly through
ept without trapping. But msix's table and pba exist on pci bar, they must
be trapped and emulated by crosvm, so these bars mmappable must be
disabled.
BUG=chromium:992270
TEST=pass through a device with msix cap to guest, then test device
function in guest.
Change-Id: If7504a924902c940e00cc759c1ca64a116bbca17
Signed-off-by: Xiong Zhang <xiong.y.zhang@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1987815
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
This change implements MSI routing and injection, so that service_irq
can actually inject an interrupt into guest.
BUG=chromium:908689
TEST=Unit tests in file.
Change-Id: I2db4f00f569db56f5765c707faaa87c64fd3da9f
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1945795
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Zhuocheng Ding <zhuocheng.ding@intel.corp-partner.google.com>
When VFIO device have msix capability, vfio kernel doesn't emulate msix,
so all the msix emulation are handled by crosvm. This include
msix capability register read/write, msix table read/write, msix pba
table read/write.
BUG=chromium:992270
TEST=passthrough a device with msix capabilty to guest, and check device
msix function in guest
Change-Id: Ic39737662a5051ac6b9e29aad227d3d4946190a8
Signed-off-by: Xiong Zhang <xiong.y.zhang@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1987814
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
When msix is enabled, msix will have multi vectors, this patch enable
mutlti vectors in vfio_device.irq_enable() function.
BUG=chromium:992270
TEST=passthrough a device with msix capability to guest, and check
device msix function in guest
Change-Id: I5f8265e7badec8551ff9a974462f08425ee93ab2
Signed-off-by: Xiong Zhang <xiong.y.zhang@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1987813
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
The flag in avail descriptor supplies irq suppress, it could reduce irq
injection from device, so many redundant interrupts could be removed
from guest, then improve guest performance.
BUG=None
TEST=run fio read and fio write in guest
Change-Id: I68789d8ca24d0e84d0b446db65057f4da2fac56f
Signed-off-by: Xiong Zhang <xiong.y.zhang@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2008339
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
For suspend request from VM, will write suspend event and notify
crosvm main process to pause VCPUs.
For resume request, it is not from VM itself but by the resume
command through crosvm socket. Resume request will notify the PM
device to fill its wakeup registers with wakeup event so that
when VCPUs start to run, VM can know there is wakeup from outside.
BUG=chromium:1018674
TEST=cargo test -p devices
Change-Id: I4724ffee10150065a62bf520076c16cbc70b7749
Signed-off-by: Chuanxiao Dong <chuanxiao.dong@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2035169
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Tomasz Jeznach <tjeznach@chromium.org>
Adds a method Reader::collect() to read a collection of DataInit types,
and a method Writer::consume() to write a collection of DataInit types.
BUG=b:147334004
TEST=cargo test -p devices
Change-Id: Ib5947d30b44b74dc6cf0474e5b87778aad6f08a0
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2061516
Reviewed-by: Keiichi Watanabe <keiichiw@chromium.org>
Commit-Queue: Keiichi Watanabe <keiichiw@chromium.org>
Tested-by: Keiichi Watanabe <keiichiw@chromium.org>
Adds a new backend type, gfxstream, that calls out to a C library
implementing the actual rendering.
The purpose is to allow the Cuttlefish and Android Studio Emulator teams
to use crosvm with the current API-forwarding style of rendering
employed in the Android Studio Emulator.
Also, introduces a new key to the --gpu command line interface,
backend=, which selects from different backends. Note that the previous
behavior is now deprecated and will be removed after some time
(when all clients switch over to backend=).
The gfxstream backend itself implements a subset of 3d-related resource
and context creation/transfer/execbuffer commands. Their meaning is
specific to the way in which they are interpreted in the backend
library.
To interface with display, gfx stream backend takes a callback that is
run on guest vsync. The callback is responsible for repainting the
display's contents. gfx stream provides a callback, get_pixels, that can
be used asynchronously. The asyncness is not taken advantage of
currently but will be useful for cases where the client attached to the
VMM might want to update at a different rate than guest vsync.
The guts of the stream backend library is currently defined here:
https://android.googlesource.com/platform/external/qemu/+/refs/heads/emu-master-dev/android-qemu2-glue/emulation/virtio-goldfish-pipe.cpp
The linking of the library is controlled via the feature "gfxstream".
If the feature is turned off, we use a default do-nothing stub impl.
Next steps:
- integrate virtio-gpu-next so as to have host coherent memory for
vulkan
- Figure out low latency command submit/response with SUBMIT_CMD_3DV2
BUG=b:146066070
Change-Id: If647381c15e5459cec85e2325f97e2f0a963b083
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2033305
Tested-by: kokoro <noreply+kokoro@google.com>
Tested-by: Lingfeng Yang <lfy@google.com>
Reviewed-by: Lingfeng Yang <lfy@google.com>
Reviewed-by: Jason Macnak <natsu@google.com>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Auto-Submit: Lingfeng Yang <lfy@google.com>
Commit-Queue: Zach Reizner <zachr@chromium.org>
Current blk interrupt is injected into guest after device handle a
batch of requests. While this patch injects interrupt at the end of each
request. So guest block will get much more interrupts and could handle
request more quickly.
With this patch, the guest fio read test improves 13%, while fio write
doesn't get better.
BUG=none
TEST=run fio_read and fio_write in guest
Change-Id: Ib0bd23e624dfc5d940d6cc124468b898d8ba128e
Signed-off-by: Xiong Zhang <xiong.y.zhang@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2008338
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Add ACPI PM resource emulation code in devices, so that it can support
the ACPI PM requestion from guest OS.
BUG=chromium:1018674
TEST=cargo test -p devices
Change-Id: I7b82b1c3a6f609136e493b55420b947afd1d5cfc
Signed-off-by: Chuanxiao Dong <chuanxiao.dong@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2035168
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Tomasz Jeznach <tjeznach@chromium.org>
Stop using 30 and 31 as device IDs of virtio-wl and
virtio-tpm, as these numbers were reserved for virtio-video devices
in the upstream [1].
Instead, use integers from 63, which is the largest
number we can use for a virtio device ID.
Note that this CL must be merged after kernels with CL:2024135 landed.
[1]: https://github.com/oasis-tcs/virtio-spec/issues/67
BUG=chromium:1031512
TEST=Check if /dev/wl0 exists on ARCVM with CL:2024135
Change-Id: I267c7702d3c28642492f560170a0d1d9d6523c31
Signed-off-by: Keiichi Watanabe <keiichiw@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2025127
Reviewed-by: Chirantan Ekbote <chirantan@chromium.org>
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Commit-Queue: Fergus Dall <sidereal@google.com>
When the user specifies the display size through the gpu argument but
not specifies the size of the single touch device, the display size
will be used as the size of these touch devices.
Use default() to initialize the GpuParameters. Allow initialize the
GpuParameters dynamically in the future.
Change-Id: I9fa04f8ff479732370514fbaeb062d737adba319
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2043072
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Kaiyi Li <kaiyili@google.com>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Loop vfio device config register, then find out the msi and msix
pci capability.
both msi and msix need IrqRequestSocket for adding its routing info
into kvm routing table, but vfio device has one IrqRequestSocket only,
and only msi or msix is enabled at runtime, so Arc is used to let msi
and msix share one device IrqRequestSocket.
BUG=chromium:992270
TEST=pass a device with msix capability to guest, and check device msix
function in guest
Change-Id: I008ccd0e98507dc4d587418fbe00aa23029bdbad
Signed-off-by: Xiong Zhang <xiong.y.zhang@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1987812
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Driver use virtio_wmb() before avail_index updating, here fence() is
compared with it, this guarantee that the descriptor_index is read after
avail_index.
BUG=none
TEST=run fio_read and fio_write in guest
Change-Id: I6a5e1bc7c915effd433b05aca246302c5a3e313c
Signed-off-by: Xiong Zhang <xiong.y.zhang@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2035079
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Vhost-net reset method can stop the activate thread and let it to
return the ownership of eventfds and tap back to the device, so that
it is possible to be activated again.
BUG=None
TEST=launch Crosvm guest with vhost-net. It works fine with iperf test.
TEST=cargo test -p devices
Change-Id: I8f3cb6939595b4dffa0e95bcffbf9fcc8fcf71ba
Signed-off-by: Chuanxiao Dong <chuanxiao.dong@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2009665
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
the event_source needs to be returned through the reset method to allow
the virtio-input device model reactivated again.
BUG=None
TEST=cargo test -p devices
Change-Id: I07a4add40b1c233e1ed328ccef1a1abd453ea0f7
Signed-off-by: Chuanxiao Dong <chuanxiao.dong@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2032351
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
worker.run might encounter error when running but we still need it
to give the ownerships of the kill eventfd back so that it
can have a second round activate. And Worker structure also contains
several other eventfds which will be needed for a second round activate
so change to put this eventfd into Worker as well.
BUG=None
TEST=launch Crosvm guest with vhost-net and vsock. Both of them work fine.
TEST=cargo test -p devices
Change-Id: I34477cfa3de23d7ab849f741d0ffb098c720a629
Signed-off-by: Chuanxiao Dong <chuanxiao.dong@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2009664
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Tested-by: Dylan Reid <dgreid@chromium.org>
The high 32 bits of writable_bits was set incorrectly when adding 64-bit
memory BARs to PciConfiguration: it would effectively always be all
zeroes (no writable bits) instead of all ones (all writable bits).
The writable_bits field is used to determine which bits to force to 0
when reading the BAR, which is used by the guest to determine the size
of a BAR: write an all-ones value to the BAR, read it back, and the
resulting value has only the writable bits still set. Since PCI BARs
must be a power of two in size, the effective size of the BAR is the
bitwise inverse of the resulting value plus one.
For 64-bit BARs, this process is the same, except that two contiguous
32-bit registers are combined, so for a 4096-byte 64-bit BAR, the
writable_bits field should be 0xFFFFFFFF_FFFFF000; however, with the
previous (buggy) code, it was 0x00000000_FFFFF000.
Add checks to the unit tests to verify that the writable_bits field is
correctly calculated as well.
BUG=None
TEST=cargo test -p devices pci_configuration
TEST=Boot Linux 4.19 kernel in crosvm
Change-Id: Ib97aa5dccf9bf042328c0fc9defe1797fc67bb05
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2033620
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
Transfer TRBs have a flag that indicates that data is transferred within
the TRB itself instead of as a separate buffer. Add support for this
type of transfer in the ScatterGatherBuffer implementation.
This fixes USB support when using Linux 5.1+ as the guest kernel, since
it now uses immediate data transfers.
BUG=chromium:1046564
TEST=`adb root` to connected phone on Linux 5.4 guest kernel
Change-Id: I6c37db422ac8e65d10e1a91807b15e903ad614de
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2026262
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
Add a function to get a resource FD via resource_bridge by extracting code from
the wayland device implementation.
This function will be used by virtio-video devices.
BUG=b:147465619
TEST=Run ARCVM and check that a window is displayed
Change-Id: I7b064c9a04bd082e30dd488d0b14731228e6047d
Signed-off-by: Keiichi Watanabe <keiichiw@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2014520
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Chirantan Ekbote <chirantan@chromium.org>
... which does not require virglrenderer (or any renderer).
This will allow the Cuttlefish team to use minigbm as its gralloc
implementation when both hardware acceleration is available and
unavailable.
Adds a GPU `Backend` trait with all of the existing methods of the
current backend and converts the existing `Backend` into
`Virtio3DBackend` which implements the new trait.
Adds a `Virtio2DBackend` which creates resources with byte vectors on
the host and implements transfers via the old code from
gpu_buffer/src/lib.rs.
Adds a runtime flag to select between 2D and 3D mode with 3D mode as
the default.
Moves the process_resource_bridge() function to the `Frontend` and
instead expose a export_resource() function on the `Backend` to avoid
some code duplication.
BUG=b:123764798
TEST=build + launch cuttlefish w/ 2D mode (minigbm + custom hwcomposer)
TEST=built + launch cuttlefish w/ 2D mode (minigbm + drm_hwcomposer)
TEST=built + launch cuttlefish w/ 3D mode (minigbm + drm_hwcomposer)
Change-Id: Ie5b7a6f80f7e0da72a910644ba42d2f34b246be8
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1993913
Commit-Queue: Jason Macnak <natsu@google.com>
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Reviewed-by: Lingfeng Yang <lfy@google.com>
Set the backend with null fd can reset the vq in vhost, which
can allow the activate fn to run again.
BUG=None
TEST=launch Crosvm guest with vhost-net. It works fine with iperf test.
TEST=cargo test -p devices
Change-Id: Ida952409147fd6fbd1d8f69b3a88a7ef03051d65
Signed-off-by: Chuanxiao Dong <chuanxiao.dong@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2009523
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
The plan is to use shared mem cap for virtio-gpu.
BUG=chromium:924405
TEST=build
Change-Id: Id2829c2cd9883aca19641eff625c65a8db335e7a
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1963334
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Tested-by: Gurchetan Singh <gurchetansingh@chromium.org>
Commit-Queue: Gurchetan Singh <gurchetansingh@chromium.org>
When a non-sparse disk is resized, we should allocate storage for the
newly-expanded space when the disk is grown to maintain the
non-sparseness. To accomplish this, add a call to allocate in the
resize function in the block device.
BUG=chromium:858815
TEST=`crosvm disk resize ...` and verify disk image is fully allocated
Change-Id: If263aa2b5c9da11b8bfc0586e4ac1575f2bd7084
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2015829
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Don't store the BAR value, as it can potentially be updated by the
guest. (This is not supported by our PCI device model just yet, but this
is still the correct thing to do and matches other crosvm PCI devices.)
BUG=None
TEST=Add USB device on nami
Change-Id: Ie42d08429e7ff124178c818877b4cee83003d66f
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1924782
Reviewed-by: Zach Reizner <zachr@chromium.org>
As indicated in the comments, it seems that this warning is unnecessary.
We've added explicit handling for the UNSET case, and left the warning
intact to handle any out of range values.
BUG=chromium:1041054
TEST=builds.
Change-Id: I2d41159f2d4ccbb4d75d2a8f4bab54586ec65442
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1995308
Reviewed-by: Zach Reizner <zachr@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Noah Gold <nkgold@google.com>
Auto-Submit: Noah Gold <nkgold@google.com>
vhost set_owner fn only needs to be called once. Put it in activate
fn will block the vhost devices to be activated again in future.
on_device_sandboxed is a good place to put the set_owner as it only
run once. So put it there.
BUG=None
TEST=launch Crosvm guest with vhost-net and vsock. Both of them can work
TEST=cargo test -p devices
Change-Id: I45308e26b026c9141e4426d8b1bbe1944612a915
Signed-off-by: Chuanxiao Dong <chuanxiao.dong@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1954173
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Activate_vqs is used to do the queue preparation before really
running. The virtio-vhost device might need to do some cleanup
to allow a second round activate in the future. How to do the
cleanup is depending on how the vhost virtio devices.
Just add an interface called cleanup_vqs to allow the vhost virtio
devices to do their own cleanup stuff.
BUG=None
TEST=launch Crosvm guest with vhost-net and vsock. Both of them can work
TEST=cargo test -p devices
Change-Id: I2472e79a8b63c9336f886cde55ffef6a78008ad8
Signed-off-by: Chuanxiao Dong <chuanxiao.dong@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1954172
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Adds a stub display that emulates a display without actually
displaying contents anywhere.
This is needed for transitioning Cuttlefish to always using minigbm
as its gralloc implementation. Cuttlefish currently uses a custom
gralloc and hwcomposer implementation when running without hardware
acceleration. The Cuttlefish team would like to start with removing
our custom gralloc implementation and use minigbm. For this, we need
to add a virtio 2D backend to crosvm. Our hwcomposer implementation
currenlly sends framebuffers from the guest to the host via sockets.
The gpu backend still requires a display so we need a stub display
to use with the 2D backend for the period of time while we are
either still using our hwcomposer implementation or until our
hwcomposer implementation is updated to use the virtio backend for
display.
BUG=b:123764798
BUG=chromium:1033787
TEST=built and launched with Cuttlefish locally
Change-Id: I1a7e259d914a53252200c59589c4142e76c6b96b
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1993947
Reviewed-by: Gurchetan Singh <gurchetansingh@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Jason Macnak <natsu@google.com>
The only device that used user_command was Serial. This change makes
Serial device use a thread to read from its input instead of using
user_command.
BUG=chromium:1033787
TEST=./build_test
run crosvm with stdio serial with and without sandbox
Change-Id: Ia0f2ee83d94ad2fee3f1f4f89aa734b976e33507
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1966435
Tested-by: Zach Reizner <zachr@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Commit-Queue: Zach Reizner <zachr@chromium.org>
Auto-Submit: Zach Reizner <zachr@chromium.org>
Add a control queue for virtio_net, and implement the command to set
available networking offloads.
Set offloads initially when acking features from the guest. We previously set
offloads on unconditionally.
Add TUNSETOFFLOAD to the allowed ioctls for virtio_net.
BUG=chromium:1031413
TEST=boot 5.4 guest, check vmtap offloads enabled with ethtool
TEST=enable ip_forward in guest, check vmtap offloads disabled with ethtool
Change-Id: I4129aa03419798906bd95cf65a6a4ab63069f50b
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1968200
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Commit-Queue: Stephen Barber <smbarber@chromium.org>
Tested-by: Stephen Barber <smbarber@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Guest can specify which socket it wants to connect by passing a
parameter to VIRTWL_IOCTL_NEW_CTX_NAMED.
Even after this CL, only the unnamed wayland socket is used for composition.
Additional sockets are used for IPC purpose (e.g. camera).
BUG=b:146100044
TEST=Camera works
Cq-Depend: chromium:1962108
Change-Id: Ibd8efbae1b2177cc0381d88d151643183c31b519
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1963412
Tested-by: Ryo Hashimoto <hashimoto@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Commit-Queue: Ryo Hashimoto <hashimoto@chromium.org>
PciRoot doesn't support multifunciion which forward pci config r/w
from function > 0 to function 0, so if the vfio device have multifunction
flag, guest will find the existence of all the other functions, actually it
is from the function 0. In order to fix these extra functions, this
patch clear the multifunction flag, so guest won't probe the other
functions.
BUG=chromium:992270
TEST=pass through a device with multifunction into guest
Change-Id: I395636411e0d7d2a3729de16e638d7f6b2dde552
Signed-off-by: Xiong Zhang <xiong.y.zhang@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1954221
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
igd opregion is used by igd driver to get vbt info and exhange info
between bios and driver, but it isn't a standard pci resource, host
bios allocate, reserve its memory, and report the memory base address
through cfg_register 0xFC on native.
As crosvm doesn't have bios, it is hard to allocate and reserve opregion
for guest. Here opregion is faked as mmio memory, and let crosvm
allocate guest memory from mmio space, report its base to cfg_register
0xFC also.
guest driver read cfg_register 0xFC to get opregion base address, then rw
it throgh this address. Read is forwarded to vfio kernel and write is
ignored.
BUG=chromium:992270
TEST=crosvm --vfio /sys/devices/pci0000:00/0000:00:02.0, pass through
host igd into linux guest, the physical local display lightup and show linux desktop.
Change-Id: I1cc3618e99313fc1f88b96dcbc635f090b19340c
Signed-off-by: Xiong Zhang <xiong.y.zhang@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1688689
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Adds bindings to the X11 display window to capture keyboard & mouse
input & send it to the guest via an EventDevice.
Original implementation by zachr@chromium.org.
BUG=chromium:1023975
TEST=None
Change-Id: I33156a8ca0b8c610a2080e3b6891cca2a865734b
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1971121
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Noah Gold <nkgold@google.com>
Reviewed-by: Zach Reizner <zachr@chromium.org>
The reset method will be called when guest virtio driver is
resetting the device. Currently the balloon/Rng/block/net
virtio drivers will re-configure the virt queue during the
reset so they required to be re-activated for using the new
virt queue configurations. To support this, need these device
models to return back the moved ownership of the important
variables so that they can do the re-activate.
BUG=chromium:1030609
TEST=Launch linux guest and follow the reproduce steps in BUG#1030609 to check
if balloon/Rng/block/net driver still complain failure.
Change-Id: I5b40fd303ea334484c590982e3e0874ea4e854ee
Signed-off-by: Chuanxiao Dong <chuanxiao.dong@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1971097
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
It refers to the implementation of the Cloud-hypervisor
commit:
- vm-virtio: Reset underlying device on driver request
If the driver triggers a reset by writing zero into the status register
then reset the underlying device if supported. A device reset also
requires resetting various aspects of the queue.
The reset method of a virtio device might return false if it is failed
to reset the device or it is not implemented. In this case, we don't
reset the queues. Otherwise the queues will also be reset together with
a successful device reset.
BUG=chromium:1030609
TEST=cargo test -p devices
Change-Id: Iad2be38149e423a79d8366dc72e570a1d6eb297c
Signed-off-by: Chuanxiao Dong <chuanxiao.dong@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1971096
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
rust 1.40 has a new warning when there are extra parens, soRemove them.
Change-Id: Ibb565cac41ba8d72879f4cb8b949f2be30e55167
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1978140
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Tested-by: Dylan Reid <dgreid@chromium.org>
Commit-Queue: Dylan Reid <dgreid@chromium.org>
This display isn't typically used when the virt-wl device is
available and it can lead to hung fds during plug/unplug with
external displays and docks.
BUG=chromium:1027379, chromium:1027447
TEST=
Tested five times each on R79-12607.47.0 sarien:
=> without patch --> CPU usage goes to 100% after unplug
=> with patch --> CPU usages remains normal
Change-Id: Iea57f05002acc661f36b180e9e88c37b1f1b9047
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1977100
Reviewed-by: Gurchetan Singh <gurchetansingh@chromium.org>
Tested-by: Gurchetan Singh <gurchetansingh@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Gurchetan Singh <gurchetansingh@chromium.org>
igd stolen memory isn't a standard pci resource, so guest couldn't
access it. Once guest i915 driver see its size is zero, i915 won't use
it. Here sotlen memory's size register PciCfg 0x51 is forced to zero.
BUG=chromium:992270
TEST=crosvm run --vfio=/sys/devices/pci0000:00/0000:00:02.0, pass
through host igd into linux guest. The physical local display lightup and show linux desktop.
Change-Id: I1a0a6edda5d92d32307ea8025ef96677ca62c125
Signed-off-by: Xiong Zhang <xiong.y.zhang@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1688368
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Could be used by virtio-fs and virtio-gpu.
(cherry picked from crrev.com/c/1493014)
[took out PCI portions and modified commit message -- @gsingh]
BUG=chromium:924405
TEST=build and run
Change-Id: I47fd4482aa7c11e08bfb4f6c990221ae7a11a11d
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1963333
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Tested-by: Gurchetan Singh <gurchetansingh@chromium.org>
Commit-Queue: Gurchetan Singh <gurchetansingh@chromium.org>
According to ac97 spec, warm reset is specified to no-op when the device
is running. We should ignore it while the device is recording audio as
well.
BUG=chromium:1026538
TEST=Build
Change-Id: I64be922149c90be9a7a8669ef6d56591e6de632b
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1960058
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Chih-Yang Hsia <paulhsia@chromium.org>
Create a function move_to_next_buffer which handles incrementing civ and
piv properly.
Set the PICB register when that function is called, not when we read a
guest buffer.
BUG=chromium:968724
TEST=playback on-device
Change-Id: Ib384efceeac4be0e056c20591d93fe32b7305db6
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1954207
Reviewed-by: Fletcher Woodruff <fletcherw@chromium.org>
Reviewed-by: Chih-Yang Hsia <paulhsia@chromium.org>
Commit-Queue: Fletcher Woodruff <fletcherw@chromium.org>
Tested-by: Fletcher Woodruff <fletcherw@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
This allows overriding the default logical block size (512 bytes) with
other values, such as 4096 for 4K block size disks.
BUG=chromium:942700
TEST=crosvm run -r vm_rootfs,block_size=4096 vm_kernel
TEST=verify block size with lsblk --output-all
Change-Id: Ia6db05f369a76557a2afb8b48b5cc2b66cf84b01
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1954220
Reviewed-by: Zach Reizner <zachr@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
The PIT implementation has the assumption that addresses are I/O port
numbers, so we should use full address mode. i8042 is also changed to
full address mode to avoid the conflict on port 0x61.
BUG=chromium:908689
TEST=None
Change-Id: Ibbb851e3a46ac7fc71576990a1618196de92e33c
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1945794
Commit-Queue: Zhuocheng Ding <zhuocheng.ding@intel.corp-partner.google.com>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Interrupt polarity is ignored by kvm since 3.15, and the irqfd interface
does not support polarity. We should not consider polarity either.
BUG=chromium:908689
TEST=Unit tests in file. Integration testing is blocked on rest of
split-irqchip being implemented.
Change-Id: Ibae23b81274f867bbf56009854611628dbb4c154
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1945793
Commit-Queue: Zhuocheng Ding <zhuocheng.ding@intel.corp-partner.google.com>
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-by: Stephen Barber <smbarber@chromium.org>
This change enables Cuttlefish to run with a user specified display size on top
of virtio gpu accelerated graphics rendering.
This change makes the width and height an argument/flag and adds the necessary
plumbing to pass this width and height through the gpu backend.
BUG=b:134086390
TEST=built crosvm and booted cuttlefish locally
Change-Id: Idabf7ef083b2377e3ebf3b50dd0296f4bf7e8ddc
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1927872
Tested-by: kokoro <noreply+kokoro@google.com>
Tested-by: Dylan Reid <dgreid@chromium.org>
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Commit-Queue: Jason Macnak <natsu@google.com>
Add support for FS_IOC_{GET,SET}_ENCRYPTION_POLICY. Unfortunately,
since the I/O direction is encoded backwards in the ioctl definitions,
these will only work with on a kernel that's compiled with a patch to
mark them as unrestricted FUSE ioctls.
BUG=b:136127632
TEST=Compile and run the vfs_crypto.c program on a virtio-fs mount
inside a VM
Change-Id: I124c5a943111b453dd44921a079a2baa1036dfd4
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1952570
Reviewed-by: Chirantan Ekbote <chirantan@chromium.org>
Tested-by: Chirantan Ekbote <chirantan@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Chirantan Ekbote <chirantan@chromium.org>
The interrupt_evt/interrupt_resample_evt/queue_evts were move to activate
so the reset method has to return them back (although there is no virtio
device implemented the reset method yet).
Instead of move, another way is just clone these EventFds so that they
don't need needed to be returned.
The advantage of doing this can avoid changing every virtio device reset
method (again, currently there is no virtio device implemented the reset
method yet, but this method is going to used when needs to support the
reset of a virtio device) for returning them, which can simplify the code.
And reset method just needs to take care their own specific resources.
The disadvantage is that, the clone of an EventFd is try_clone which might
return error code (although this shouldn't happen). If in such case, the virtio
device won't be activated.
BUG=None
TEST=launch crosvm guest with virtio devices(rng/balloon/blk/net).
Change-Id: I6e55782c3ecc46bfa878aff24b85a58a7ed66365
Signed-off-by: Chuanxiao Dong <chuanxiao.dong@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1925682
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
if device bar is mappable, map bar's gpa to hpa in EPT, guest vcpu
could access this bar directly through EPT without trapping. This
could improve performance.
vm.add_mmio_memory could help do this, here vfio_pci send
RegisterMmapMemory request through vm_control socket to do this.
BUG=chromium:992270
TEST=none
Change-Id: I3b4274372f7dcd32e18084d55f037b6fe45ed422
Signed-off-by: Xiong Zhang <xiong.y.zhang@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1581147
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Two virtio descriptor_utils tests were using SharedMemory to stand in
for I/o targets with a fixed size; replace these with File to avoid
needing the FileReadWriteVolatile impl for SharedMemory, which isn't
used anywhere else in the crosvm code base.
This slightly changes the behavior under test in the reader_failing_io
test, since it was previously using the SharedMemory seal functionality
to make the region ungrowable; this is an unusual corner case, and (as
mentioned in the comment that was previously at the end of the test) it
is testing implementation details of write() on shared memory on Linux.
Instead, just use a read-only file so that write() to it will fail and
cause the same observable result.
BUG=None
TEST=./build_test.py
Change-Id: I6d62cd70791f1dec625b750ecd01cc51e307f971
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1939783
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Reviewed-by: Chirantan Ekbote <chirantan@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
All virtio devices can use the same generic calculation for number of
MSI-X vectors required: number of queues plus one for configuration
changes. Move this calculation to the VirtioPciDevice implementation
and remove the Option to unconditionally enable MSI-X support for all
PCI virtio devices.
BUG=chromium:854765
TEST=Verify all virtio interrupts in /proc/interrupts are PCI-MSI
Change-Id: I5905ab52840e7617b0b342ec6ca3f75dccd16e4d
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1925169
Reviewed-by: Zide Chen <zide.chen@intel.com>
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
PlaybackError and CaptureError encode the same error cases. Combine the
two into one Error type, AudioError.
BUG=None
TEST=builds
Change-Id: I44259227d67a0284c9a11c4aafd86fafe1006f8b
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1925727
Reviewed-by: Chih-Yang Hsia <paulhsia@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Fletcher Woodruff <fletcherw@chromium.org>
This new trait allows DiskFile implementors to provide the length of the
file directly rather than using SeekFrom::End with seek().
BUG=None
TEST=./build_test
TEST=Boot Termina in crosvm
Change-Id: I9447ebb43dbd5fbb32a3a6b6d2fc969b9406cdbc
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1913961
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Dylan Reid <dgreid@chromium.org>
This eliminates an extra seek per guest write zeroes request.
Additionally, it allows us to stop depending on the file cursor and pass
the offset directly, making multi-queue implementation easier.
BUG=chromium:858815
TEST=Boot Termina in crosvm
Change-Id: I8b15a39752a1b68597a2b1e1fd72382a484a3cb2
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1913521
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Dylan Reid <dgreid@chromium.org>
The setres{u,g}id and the gete{u,g}id system calls on arm use 16 bit
values for uid_t and gid_t. This causes the kernel to interpret a
uid/gid of 65535 as -1, which means do nothing.
Use the 32-bit variants of these system calls instead so that we can use
32-bit values. Normally, libc would take care of hiding these kinds of
implementation details but since we can't use the libc wrappers for
these system calls, we have to do it ourselves.
BUG=b:136128319
TEST=`tast run vm.Virtiofs` on kevin
Change-Id: I6c0fda42c131e059139000828b3a53d4a73f340c
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1923569
Tested-by: Chirantan Ekbote <chirantan@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Chirantan Ekbote <chirantan@chromium.org>
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Having more than one thread watch the interrupt resample event meant
that the threads would race to acknowledge the event: if thread B called
epoll_wait after thread A got a readable event for the resample eventfd
but before thread A called `read()` on the eventfd, then thread B would
also get a readable event for that eventfd. Both threads would then
attempt to acknowledge the event but only one would succeed. This would
leave the other thread blocked on the `read()` call until the resample
event became readable again.
Fix this by having only one worker watch the resample event.
BUG=b:136128319
TEST=`tast run vm.Virtiofs` on kevin
Change-Id: I5cd781a9c79ac718207a944433ea20a967735237
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1923568
Tested-by: Chirantan Ekbote <chirantan@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Chirantan Ekbote <chirantan@chromium.org>
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
From ac97 spec, SR_CELV should be synced with civ == lvi in bus_master.
intel8x0 is not using the bit but some other driver or system might be
using it.
Remove checking if new_sr equals to old_sr since update_sr will do
nothing if the input value equals to the old sr value.
Add unit tests steps to test the expected results.
BUG=chromium:1026538
TEST=Unit tests
Change-Id: I3dc9f42e2beed8e635a65dbefba44677c73fcc5b
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1925917
Tested-by: Chih-Yang Hsia <paulhsia@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Dylan Reid <dgreid@chromium.org>
Reviewed-by: Dylan Reid <dgreid@chromium.org>
In bus_master control register (CR), reset registers (RR) operation is
refactored by:
- Don't call stop_audio(), since the bus master should always be in stop
state while getting this control. (From AC'97 spec: Setting it when the
Run bit is set will cause undefined consequences."). And the driver
will always disable the audio first by setting 0 to CR_RPBM bit.
- While doing the registers reset, clean up sr by using update_sr since
assigning 0 to sr directly won't unset the interrupt bit in global
status register and the driver might go into snd_intel8x0_update()
with a stopped substream.
- Introduce helper function - reset_func_regs()
Add steps in unit tests which
- Start the bus masters with "Interrupt on Completion Enable" (CR_IOCE) bit.
- Verify if the interrupt bit in global status register is set / unset.
BUG=chromium:1026538
TEST=Unit tests
Change-Id: Ie90ca4c82cc3c867992ecaeb61ef4b3e9dd0d079
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1925916
Reviewed-by: Chih-Yang Hsia <paulhsia@chromium.org>
Tested-by: Chih-Yang Hsia <paulhsia@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Dylan Reid <dgreid@chromium.org>
Extend the --disk option and other related options to allow a particular
disk to have the sparse operations (virtio-blk's discard command)
enabled or disabled.
By default, the sparse flag will be enabled for virtio-blk devices,
matching current behavior.
BUG=chromium:858815
TEST=Run `crosvm with --rwdisk file.img,sparse=false` and try to discard
Change-Id: Ib72c949711fbe869a3f444d7f929a80d0e039f72
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1906750
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Zach Reizner <zachr@chromium.org>
At vfio device initialization, it will probe all the defined regions,
most device don't have vga region, then when crosvm probe vga region info,
kernel vfio will return -EINVAL. So when error happens, it should continue
getting the next region info, instead of exit.
BUG=None
TEST=passthrough non graphic device into guest
Change-Id: If4d2f723df45f58ccd733e03c854cdcef3530bdb
Signed-off-by: Xiong Zhang <xiong.y.zhang@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1918486
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Re-enable MSI-X for virtio-blk and virtio-net now that the underlying
issue causing hangs at startup has been fixed (CL:1917495).
BUG=chromium:1019986
TEST=Boot Termina on nami
This reverts commit 85858f580e.
Change-Id: I5a5e197243a16aee2b2aaf3145a1180749b097b2
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1918261
Reviewed-by: Zide Chen <zide.chen@intel.com>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
The queue_evts() and interrupt_evt() functions were public, but nothing
was calling them. Remove them to clean up the unused code.
BUG=None
TEST=./build_test
Change-Id: Id36e78343869746c733bba04383ab93c9d377601
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1898270
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Add handling of the virtio device MSI-X configuration change vector by
using the signal function that was previously factored out.
BUG=chromium:854765
TEST=./build_test
TEST=trigger disk config change with `crosvm disk resize ...`
Change-Id: I462c23e10d152f896586bb70b95634a53088d480
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1898269
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Zide Chen <zide.chen@intel.corp-partner.google.com>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Factor out the common creation of struct Interrupt.
No functional change.
BUG=chromium:854765
TEST=./build_test
Change-Id: Idf8804771ba1af5181818f643e15e1b42918258a
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1898268
Tested-by: kokoro <noreply+kokoro@google.com>
This consolidates the status byte manipulation in process_one_request()
instead of requiring both that function and execute_request() to deal
with it.
The tests are modified to run the full process_one_request() function
instead of just execute_request() to exercise the full descriptor
parsing logic, and they are adapted to read the status of the request
from the status byte in the buffer from the descriptor since
process_one_request() returns successfully as long as the descriptor
parsing succeeded, even if the requested I/O failed.
BUG=None
TEST=./build_test
Change-Id: I17affabc2d3c30c810643ce260152cf34893b772
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1918479
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Reviewed-by: Chirantan Ekbote <chirantan@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
The msix entries might be changed by guest during msix maksed. The
current implementation won't update the MSIX route table in this case
which can cause KVM still inject the IRQ according to the old routing.
To fix this, we should update the msix route regardless the msix mask
status.
BUG=chromium:1023692
TEST=cargo test -p devices
Change-Id: Ifa356b3834ff454ecfca1dbdd97a7ca940d1f2b6
Signed-off-by: Chuanxiao Dong <chuanxiao.dong@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1911721
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-by: Zide Chen <zide.chen@intel.corp-partner.google.com>
Tested-by: kokoro <noreply+kokoro@google.com>
Previously, PciConfiguration::get_bar_addr would only correctly return
the value of a 32-bit memory region; implement support for the other
valid BAR types as well.
BUG=None
TEST=cargo test -p devices
Change-Id: I221187dfb96b31d7fead73eccf605a0886021d8b
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1880164
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Performance-wise this about breaks even, but greatly simplifies the
virtio-net handling for processing received frames.
BUG=chromium:753630
TEST=crostini.NetworkPerf
Change-Id: Ie7b576020ecfe2a6cc41b7f72bd7143795a9a457
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1906996
Tested-by: kokoro <noreply+kokoro@google.com>
Tested-by: Stephen Barber <smbarber@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Commit-Queue: Stephen Barber <smbarber@chromium.org>
Don't assume the file system is running as the root user when changing
credentials. Instead keep track of the thread euid/egid and use those
when restoring thread credentials.
BUG=b:136128319
TEST=`tast run vm.VirtioFs`
Change-Id: I37d59def99cd71de68aa7f94941031a86df54329
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1890584
Tested-by: Chirantan Ekbote <chirantan@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Chirantan Ekbote <chirantan@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Add _exact/_all variants of the FileReadWriteAtVolatile functions on
descriptor Reader/Writer, and use them in the block device to replace
the short read/short write error cases. This ensures all data is
read/written even if the underlying implementation (in particular,
qcow2) does not transfer the full amount of data in one
read_vectored_at_volatile/write_vectored_at_volatile call.
BUG=chromium:1023422
TEST=`mkfs.btrfs /dev/vdb` with a qcow2 disk
Change-Id: Ia37a333947f6f63faf3d4a06cfcc297309d5aff6
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1907443
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Use 64bit flag in vfio device's bar to get correct mmio allocator.
BUG=chromium:992270
TEST=none
Change-Id: I8f3dab48eb6dc0b92071803aa3526cadda8034c7
Signed-off-by: Xiong Zhang <xiong.y.zhang@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1581143
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Since unified allocator is used to allocate mmio, this patch remove the
device memory name, and rename device to mmio.
BUG=chromium:992270
TEST=this patch doesn't change function, run build_test
Change-Id: I234b0db4b3c5de8cfee372ace5212a980564d0c7
Signed-off-by: Xiong Zhang <xiong.y.zhang@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1895234
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Current mmio and device two allocators exist, the purpose to define
two allocator is:
Accessing to gpa from mmio allocator cause vm exit, while gpa from
device allocator doesn't cause vm exit.
Whether vm exits exist or not, dependency on whether
vm->add_device_memory() is called with gpa from allocator or not.Even
if gpa is from mmio alloator, and vm->add_device_memory() is called
with this gpa, accessing this gpa won't cause vm exit. So mmio allocator
and device allocator couldn't guarantee the original purpose.
This patch unify mmio allocator and device allocator into one mmio
allocator.
BUG=chromium:992270
TEST=this patch doesn't change function, so just run build_test
Change-Id: If87d5c2838eb122ef627fa45c394b1b3ccfafeb0
Signed-off-by: Xiong Zhang <xiong.y.zhang@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1895233
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
This allows the caller to grab a buffer without committing to using it,
which can be used in the case where two resources (a virtio buffer plus
some other resource) need to be acquired simultaneously.
BUG=None
TEST=build_test.py
Change-Id: Icb61de99db807648ff02c41f95b3128ecce41501
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1904638
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Tested-by: Stephen Barber <smbarber@chromium.org>
Commit-Queue: Stephen Barber <smbarber@chromium.org>
Convert the virtio wayland device to use the descriptor_util
Reader/Writer helpers to simplify the code and allow support
of arbitrary descriptor layouts.
BUG=chromium:966258
TEST=./build_test.py
Change-Id: Ic854b76d378be261db4f21cba475bd0abc4af80e
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1815418
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
The virtio-blk configuration space has a `seg_max` field that lets the
device inform the driver of the maximum number of segments allowed
within a single request. The Linux virtio block driver assumes that if
the corresponding feature (VIRTIO_BLK_F_SEG_MAX) is not advertised, then
only one segment can be used.
Add a segment limit based on sysconf(_SC_IOV_MAX) to allow the Linux
block stack to make use of multiple segments in a single request, which
will get translated into a single readv/writev call in the crosvm block
device.
BUG=None
TEST=strace crosvm virtio-blk process and note preadv with iov_cnt > 1
Change-Id: Ia14ebebb85daa21e2d43437bb74886f32e6e8187
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1876806
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
The HANDLE_KILLPRIV feature tells the kernel that the file system will
take care of clearing the setuid and setgid bits when a file is written
to by someone other than the owner.
However, this doesn't work when writeback caching is enabled as the
write may be buffered and flushed later, which would prevent the bits
from being cleared on write.
Remove the HANDLE_KILLPRIV feature when writeback caching is enabled.
BUG=b:136128319
TEST=`tast run vm.VirtioFs`
Change-Id: Icef98e878603cc428f83db37857d69bc6da4486c
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1890582
Tested-by: Chirantan Ekbote <chirantan@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Chirantan Ekbote <chirantan@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Add a fuzzer for the virtio-fs server, which is responsible for decoding
a byte stream into FUSE messages.
BUG=none
TEST=run it with cros_fuzz
Change-Id: Ic7695f2106d3f81e6cf09b98ffedc51831238f1e
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1865272
Tested-by: Chirantan Ekbote <chirantan@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Chirantan Ekbote <chirantan@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Add some context for debugging failures so it is possible to determine
which register read is failing.
BUG=None
TEST=./build_test.py
Change-Id: I6084971bc6dbd1f7b5d46e6c5d7ba017bb32edc6
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1893637
Reviewed-by: Zach Reizner <zachr@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
This will be used for configuration interrupts as well.
No functional change.
BUG=chromium:854765
TEST=./build_test.py
Change-Id: Iacccfd0a93a5c90783033a8e37598c2683704351
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1898267
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Zide Chen <zide.chen@intel.corp-partner.google.com>
Reviewed-by: Zach Reizner <zachr@chromium.org>
The virtio specification allows the driver to configure a queue's MSI-X
vector to the magic NO_VECTOR value (0xffff); in this case, if MSI-X is
enabled, no interrupt should be delivered (neither MSI-X nor INTx).
BUG=chromium:854765
TEST=./build_test.py
Change-Id: Icb5e82bf9a57ded60fc8c022c4d8630b5ab70dcf
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1898266
Reviewed-by: Zide Chen <zide.chen@intel.corp-partner.google.com>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
The first interrupt_status.fetch_or() operation already sets the
appropriate bit; calling fetch_or() again with the same value is
unnecessary.
In addition, if the interrupt_status field has any bit set (not just the
USED_RING bit), then the interrupt is already pending and we don't need
to trigger it again.
BUG=chromium:854765
TEST=./build_test.py
Change-Id: Iba7fb9b934d062db801f8ba0e743618f9db580ee
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1898045
Reviewed-by: Zide Chen <zide.chen@intel.corp-partner.google.com>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
The virtio specification says that the device must have all queue and
configuration change events unmapped upon reset. The queue MSI-X vector
configuration was already initialized to VIRTIO_MSI_NO_VECTOR (0xffff),
but the device configuration change notification vector was initialized
to 0. Move the constant to the virtio module so it can be used to
initialize the config vector to the correct value.
BUG=chromium:854765
TEST=./build_test.py
Change-Id: Ife1117e54196a898782238a2b81e69b20ac79784
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1898044
Reviewed-by: Zide Chen <zide.chen@intel.corp-partner.google.com>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Add a new virtio-fs device for sharing files between the host and the
guest. This change adds all the device infrastructure necessary for the
driver probe to succeed but doesn't currently handle the actual fuse
protocol. Additionally, shared memory support is not currently
implemented. The device is not hooked up to the command line.
Testing this device requires a kernel with the virtio-fs patches. To
test with a standard crostini setup, use
https://user.git.corp.google.com/chirantan/virtiofs/+/refs/heads/chromeos-5.1
which is the 5.1 kernel with the virtio-fs and chromium-specific
virtio-{gpu,wl} patches applied.
BUG=b:136128319
TEST=`tast run vm.VirtioFs`
Change-Id: I09dcefafaf0d2a7e13d54df11384dfcee3b85ba6
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1705654
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-by: Stephen Barber <smbarber@chromium.org>
struct stat64 uses different types on 32-bit platforms like arm; cast to
the types used there to allow compilation on both x86-64 and arm.
In addition, a 64-bit offset was being passed to libc::ftruncate, but
this API takes a 32-bit off_t on 32-bit platforms; switch to ftruncate64
to allow the full range of offsets.
BUG=b:136128319
TEST=emerge-kevin crosvm
TEST=emerge-nami crosvm
Change-Id: I382aef8509ca723efcf5024b22e140265636dc10
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1899218
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Reviewed-by: Chirantan Ekbote <chirantan@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
This removes the unnecessary copy on the tx path. On nami, this increases
tx throughput by ~60%.
BUG=chromium:753630
TEST=crostini.NetworkPerf
Cq-Depend: chromium:1873142
Change-Id: I58be5a7acef5d118d34b3c42ffb5a34e90070bf4
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1881419
Tested-by: kokoro <noreply+kokoro@google.com>
Tested-by: Stephen Barber <smbarber@chromium.org>
Commit-Queue: Stephen Barber <smbarber@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-by: Chirantan Ekbote <chirantan@chromium.org>
This allows the conversion of (part of) a descriptor chain into an iovec
suitable for use with sys_util functions like send_with_fds().
BUG=None
TEST=./build_test.py
Change-Id: I4e3f7d9c1175c1173661b0661d3fa15d1da72d1a
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1815417
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Zach Reizner <zachr@chromium.org>
This call was changed to not return a Result in "7f64f50
descriptor_utils: check for size overflow in new()".
BUG=b:136128319
TEST=build and run pjdfstests
Change-Id: Ibdc786b26ff35977723ba61c51e8cdf1b631edc8
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1890581
Auto-Submit: Chirantan Ekbote <chirantan@chromium.org>
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Chirantan Ekbote <chirantan@chromium.org>
Add a "passthrough" file system implementation that just forwards it's
requests to the appropriate system call.
BUG=b:136128319
TEST=`tast run vm.VirtioFs`
Change-Id: I802c91dd0af8cdd8b9e761d9f04f874ae41ec033
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1758103
Tested-by: Chirantan Ekbote <chirantan@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Chirantan Ekbote <chirantan@chromium.org>
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Temporarily turn off MSI-X support in the block and net devices since it
seems this is responsible for some test flakiness that manifests as
timeouts/hangs in the ProxyDevice read handler, e.g.:
[devices/src/proxy.rs:238] failed read from child device process
virtio-pci (virtio-block): failed to receive request or response:
Resource temporarily unavailable (os error 11)
This is a minimally-invasive change to disable MSI-X without a full
revert of the relevant patches by just changing the relevant devices so
that they no longer request MSI-X vectors.
BUG=chromium:1019986
BUG=chromium:854765
TEST=check /proc/interrupts inside crosvm does not contain "PCI-MSI"
Change-Id: Ib37b503e609e2b9e22265370bcfe5804f04057ef
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1891643
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Reviewed-by: Zide Chen <zide.chen@intel.corp-partner.google.com>
The `next_avail` field is a Wrapping<u16> but we pull out the underlying
u16 when calculating the descriptor index address offset in Queue::pop
and only convert the result to a u64 after applying all the operations.
This can cause a u16 overflow if the queue size is the max
allowed (2^15). Instead, convert to a u64 immediately after calculating
the index so that the rest of the operations are carried out as u64s and
will not overflow.
BUG=chromium:1018319
TEST=`cros_fuzz reproduce` and unit tests
Change-Id: I49743e239e2a407498d862c5137930f3f0cdf72a
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1884404
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
Auto-Submit: Chirantan Ekbote <chirantan@chromium.org>
Use the "at" variants of the read/write functions in the block device.
This reduces the number of syscalls on the host per I/O to one
(pread64/pwrite64) rather than two (lseek + read/write).
The CompositeDiskFile implementation is also updated in this commit,
since it's both a producer and consumer of DiskFile, and it isn't
trivial to update it in a separate commit without breaking compilation.
BUG=None
TEST=Start Crostini on kevin, banon, and nami
Change-Id: I031e7e87cd6c99504db8c56b1725ea51c1e27a53
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1845948
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Stephen Barber <smbarber@chromium.org>
When vfio device msi is enabled, use VmIrqRequest->AllocateOneMsi() to
allocate one gsi for a msi vector, and link gsi with irqfd through
vm->register_irqfd, use VmIrqRequest->AddMsiRoute() to add msi routing
info into kvm route table.
BUG=chromium:992270
TEST=none
Change-Id: I5e2d2347e5e26f0ef6e12554dae4b12934b65e82
Signed-off-by: Xiong Zhang <xiong.y.zhang@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1581146
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Replace the copy_{to,from} calls for VolatileSlice with
ptr::copy_nonoverlapping. The copy_{to,from} implementations were doing
a volatile read/write per byte, which is significantly slower than just
using a memcpy.
Using copy_nonoverlapping should be safe here as that's how this was
implemented before the refactor.
BUG=chromium:1014999
TEST=unit tests
Change-Id: Iad29e76056ff3064a5fe7e816b517b4ac75eaaef
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1866894
Commit-Queue: Chirantan Ekbote <chirantan@chromium.org>
Tested-by: Chirantan Ekbote <chirantan@chromium.org>
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
When hw reports it could support INTX, this patch enable it by passing
irqfd into vfio kernel.
Then once hw intx happens, the vfio kernel irq handler receives and
handles it, the handler will trigger irqfd and kvm injects the interrupt
into guest.
BUG=chromium:992270
TEST=None
Change-Id: I8b200174a91183b7324b0044fde13b44c751d4d7
Signed-off-by: Xiong Zhang <xiong.y.zhang@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1813457
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
The multikey module provides a BTreeMap implementation that can use one
of 2 different kinds of keys to look up a value. This is needed by the
virtio-fs server since it needs to be able to look up keys either by u64
or by a (ino_t, dev_t) pair.
BUG=b:136127316
TEST=`tast run vm.VirtioFs`
Change-Id: I3a22331e7a15b2316c31ac803bf2813a14bf948f
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1837025
Auto-Submit: Chirantan Ekbote <chirantan@chromium.org>
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Tested-by: Chirantan Ekbote <chirantan@chromium.org>
Commit-Queue: Chirantan Ekbote <chirantan@chromium.org>
Add a `Server` type that links the FUSE protocol with the virtio
transport. It parses messages sent on the virtio queue and then calls
the appropriate method of the `Filesystem` trait.
BUG=b:136128319
TEST=`tast run vm.VirtioFs`
Change-Id: I7d6fb521f6c620efe1bdb4fa0fa8fb8c42a82f45
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1757242
Auto-Submit: Chirantan Ekbote <chirantan@chromium.org>
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Commit-Queue: Chirantan Ekbote <chirantan@chromium.org>
Tested-by: Chirantan Ekbote <chirantan@chromium.org>
Add the `Filesystem` trait, which is the main interface between the
transport and the actual file system implementation.
BUG=b:136128319
TEST=`tast run vm.VirtioFs`
Change-Id: Ic8bc9e231652020501e10ad0be810a9f66e90b8e
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1757241
Auto-Submit: Chirantan Ekbote <chirantan@chromium.org>
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Tested-by: Chirantan Ekbote <chirantan@chromium.org>
Commit-Queue: Chirantan Ekbote <chirantan@chromium.org>
The code to inject interrupt to the guest can be generic to all
virtio devices. This patch:
- move those guest interrupt related fields out of Worker structure and
put in a separate file, making the worker code cleaner.
- remove redandant functions across virtio devices: signal_used_queue(),
signal_config_changed(), etc.
BUG=chromium:854765
TEST=sanity test on eve and Linux
TEST=cargo test -p devices
Change-Id: I8e9f760f2057f192fdc74d16a59fea2e6b08c194
Signed-off-by: Zide Chen <zide.chen@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1869553
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
- signal_used_queue(): trigger MSI-X interrupts to the guest if MSI-X is
enabled, otherwise trigger INTx interrupts
- enable MSI-X on vhost-net: allocate one vhost_interrupt for every
MSI-X vector.
Performance wise, fio random R/W test on eve pixelbook:
INTx MSI-X delta
fio write 8.13MiB/s 9.79MiB/s +1.66MiB/s (+20%)
fio read 24.35MiB/s 29.3MiB/s +4.95MiB/s (+20%)
For networking performance (TCP stream), test results on eve pixelbook:
INTx MSI-X delta
iperf3 5.93Gbits/s 6.57Gbits/s +0.64Gbits/s (+10.7%)
iperf3 -R 5.68Gbits/s 7.37Gbits/s +1.30Gbits/s (+22.8%)
iperf test results on VM launched from Ubuntu host (client sends only):
INTx MSI-X delta
virtio-net 9.53Gbits/s 11.4 Gbits/s +1.87Gbits/s (+19.5%)
vhost 28.34Gbits/s 44.43Gbits/s +16.09Gbits/s (+56.7%)
BUG=chromium:854765
TEST=cargo test -p devices
TEST=tested virtio-net and block on Linux VM and eve pixelbook
Change-Id: Ic4952a094327e6b977f446def8209ea2f796878c
Signed-off-by: Xiong Zhang <xiong.y.zhang@intel.corp-partner.google.com>
Signed-off-by: Zide Chen <zide.chen@intel.corp-partner.google.com>
Signed-off-by: Sainath Grandhi <sainath.grandhi@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1828340
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
Move the check for length overflow that was in available_bytes() into
Reader::new() and Writer::new(). This simplifies callers, since they
can assume that once a valid Reader or Writer has been constructed,
available_bytes() cannot fail. Since we are walking the descriptor
chain during new() anyway, this extra check should be essentially free.
BUG=None
TEST=cargo test -p devices descriptor_utils
Change-Id: Ibeb1defd3728e7b71356650094b0885f3419ed47
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1873142
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Stephen Barber <smbarber@chromium.org>
Allocate per device VmMsi msg_socket for communication between virtio
devices and main VM process, which owns the KVM fd and issues ioctl to
KVM for KVM_IRQFD and KVM_SET_GSI_ROUTING.
BUG=chromium:854765
TEST=None
Change-Id: Ie1c81534912eaab7fbf05b5edef7dca343db301c
Signed-off-by: Xiong Zhang <xiong.y.zhang@intel.corp-partner.google.com>
Signed-off-by: Zide Chen <zide.chen@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1828339
Tested-by: Daniel Verkamp <dverkamp@chromium.org>
Commit-Queue: Stephen Barber <smbarber@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
- add a new field "vector" to struct Queue, which represents the entry
number to the MSI-X Table. This can be used to find out the desired irqfd
to inject MSI-X interrupts to the guest.
- enable MSI-X when MSI-X Enable bit of the Message Control word is
being set: allocate irqfd per MSI-X vector; register the irqfd to KVM;
update GSI routing to KVM.
- update GSI routing if the Message Data or Message Addr of individual
MSI-X table Entry is being changed in run time.
BUG=chromium:854765
TEST=cargo test -p devices
Change-Id: I81533999ab6cd9ec5f111b256caf34077a4a7d1a
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
Signed-off-by: Xiong Zhang <xiong.y.zhang@intel.corp-partner.google.com>
Signed-off-by: Zide Chen <zide.chen@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1828338
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: Daniel Verkamp <dverkamp@chromium.org>
Commit-Queue: Stephen Barber <smbarber@chromium.org>
The MsixConfig struct is responsible for all the operations of MSI-X
Capability Structure and MSI-X Table.
A msix_config object is created for each virtio device.
BUG=chromium:854765
TEST=cargo test -p devices
Change-Id: Ide7c34d335d49a201f20b0a4307bcda97d1d61b7
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
Signed-off-by: Xiong Zhang <xiong.y.zhang@intel.corp-partner.google.com>
Signed-off-by: Zide Chen <zide.chen@intel.corp-partner.google.com>
Signed-off-by: Sainath Grandhi <sainath.grandhi@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1828337
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: Daniel Verkamp <dverkamp@chromium.org>
Commit-Queue: Stephen Barber <smbarber@chromium.org>
The MSI-X feature is ported from Cloud-hypervisor commit 69e27288a2e.
(https://github.com/intel/cloud-hypervisor.git)
In this commit:
- add a new "msix" module to the pci crate.
- implement the MSI-X Capability Structure.
- implement per virtio device msix_vectors() function which represents the
supported MSI-X vector for this device.
BUG=chromium:854765
TEST=launch Crosvm on eve and Linux
TEST=cargo test -p devices
TEST=./bin/clippy
TEST=./build_test.py --x86_64-sysroot /build/eve
Change-Id: I5498b15a3bf115e34764e6610407b3ba204dae7f
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
Signed-off-by: Zide Chen <zide.chen@intel.corp-partner.google.com>
Signed-off-by: Sainath Grandhi <sainath.grandhi@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1873356
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Stephen Barber <smbarber@chromium.org>
The consume function in both the read and write methods should consume
all the VolatileSlices that are given to it rather than just the first
one. The previous implementation was not wrong, just inefficient. This
should fix that.
Also add a test to make sure that this doesn't regress in the future.
BUG=none
TEST=unit tests
Change-Id: I02ec22269cdd6cdc329dd62367b99352a4dc1245
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1865271
Tested-by: Chirantan Ekbote <chirantan@chromium.org>
Commit-Queue: Chirantan Ekbote <chirantan@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-by: Stephen Barber <smbarber@chromium.org>
It's quite costly to inject virtual interrupt to the guest, especially
in INTx case.
To reduce the number of interrupts, in process_rx(), we don't have to
inject interrupt on every frame, but wait until process_rx() finishes
processing all frames.
On eve, iperf3 gets ~15% improvement, "iperf3 -R" gets ~30% improvement.
BUG=chromium:854765
TEST=iperf3 on eve and Linux
Change-Id: Ie0560d8f42235d2371addb6de34c5f93d11a405f
Signed-off-by: Zide Chen <zide.chen@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1865021
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Commit-Queue: Stephen Barber <smbarber@chromium.org>
Tested-by: Stephen Barber <smbarber@chromium.org>
The audio_streams interface now supports specifying a sample format.
Update call sites to indicate that the desired format is S16LE.
BUG=chromium:1010667
TEST=aplay within vm
Cq-Depend: chromium:1856646
Change-Id: Ib69ff9b39196905f0f429eaf771f6f92901bfc71
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1856586
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Tested-by: Fletcher Woodruff <fletcherw@chromium.org>
Commit-Queue: Fletcher Woodruff <fletcherw@chromium.org>
Rather than using `use ::vhost::...` to disambiguate the imports, remove
the conflicting `use virtio_sys::vhost` and add `virtio_sys::` to each
location that used `vhost::...` previously.
The `use ::vhost::...` syntax confuses rustfmt when run directly on
these two files, causing it to rewrite the imports into something that
doesn't actually compile.
BUG=None
TEST=rustfmt --check devices/src/virtio/vhost/net.rs
TEST=rustfmt --check devices/src/virtio/vhost/vsock.rs
Change-Id: I8483f5327a1e2b3ae4887f0b3cef20a917d7410e
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1865370
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Reviewed-by: Stephen Barber <smbarber@chromium.org>
This just removes a few extraneous blank lines that the new rustfmt
doesn't like.
BUG=None
TEST=bin/fmt --check
Change-Id: I4482f873bdfe19f2f73f86cfdd99d6cce873593c
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1863000
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Drop the dependency on libusb and reimplement the host USB backend using
usb_sys to wrap the Linux usbdevfs ioctls.
This allows sandboxing to work without any dependency on libusb patches,
and it gives us the flexibility to modify and update the USB backend
without depending on an external third-party library.
BUG=chromium:987833
TEST=`adb logcat` on nami with Nexus 5 attached
TEST=deploy app to phone with Android Studio
TEST=Run EdgeTPU USB accelerator demo (including DFU mode transition)
Cq-Depend: chromium:1773695
Change-Id: I4321c2b6142caac15f48f197795a37d59d268831
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1783601
Reviewed-by: Zach Reizner <zachr@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
crosvm doesn't support MSI/MSI-x, but kvmgt vgpu support MSI only
through cfg msi capability. This is a simple msi implementation, it
detects msi capability and track msi control, data and address info, then
call vfio kernel to enable / disable msi interrupt.
Currently it supports one vetor per MSI. It could extend to multi vetors and
MSI-x.
BUG=chromium:992270
TEST=none
Change-Id: I04fc95f23a07f9698237c014d9f909d011f447ef
Signed-off-by: Xiong Zhang <xiong.y.zhang@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1581142
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
Fix boxed_local, const_static_lifetime, useless_format, and
redundant_closure clippy warnings in the VFIO code.
This fixes all clippy warnings except a single instance of
let_and_return in VfioPciDevice::keep_fds(), since that code is modified
in an upcoming patch.
BUG=None
TEST=./build_test.py
TEST=bin/clippy
Change-Id: I548adbc6b92448fc0db82ed72214d73b0eabaf5c
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1822697
Reviewed-by: Xiong Zhang <xiong.y.zhang@intel.corp-partner.google.com>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Tested-by: Xiong Zhang <xiong.y.zhang@intel.corp-partner.google.com>
Tested-by: kokoro <noreply+kokoro@google.com>
Add the constants and struct definitions from the kernel fuse interface.
These bindings are manually generated from `include/uapi/linux/fuse.h`
in the kernel repo.
BUG=b:136128319
TEST=none; these aren't used anywhere yet
Change-Id: I03d11bc55eca6b8269f1e63a1187ef458ee16f28
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1705655
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Commit-Queue: Chirantan Ekbote <chirantan@chromium.org>
Refactor the Reader and Writer implementations for DescriptorChains.
This has several changes:
* Change the DescriptorChainConsumer to keep a
VecDeque<VolatileSlice> instead of an iterator. This delegates the
fiddly business of sub-slicing chunks of memory to the VolatileSlice
implementation.
* Read in the entire DescriptorChain once when the Reader or Writer is
first constructed. This allows us to validate the DescriptorChain
in the beginning rather than having to deal with an invalid
DescriptorChain in the middle of the device operating on it.
Combined with the check that enforces the ordering of read/write
descriptors in a previous change we can be sure that the entire
descriptor chain that we have copied in is valid.
* Add a new `split_at` method so that we can split the Reader/Writer
into multiple pieces, each responsible for reading/writing a
separate part of the DescriptorChain. This is particularly useful
for implementing zero-copy data transfer as we sometimes need to
write the data first and then update an earlier part of the buffer
with the number of bytes written.
* Stop caching the available bytes in the DescriptorChain. The
previous implementation iterated over the remaining descriptors in
the chain and then only updated the cached value. If a mis-behaving
guest then changed one of the later descriptors, the cached value
would no longer be valid.
* Check for integer overflow when calculating the number of bytes
available in the chain. A guest could fill a chain with five 1GB
descriptors and cause an integer overflow on a 32-bit machine.
This would previously crash the device process since we compile with
integer overflow checks enabled but it would be better to return an
error instead.
* Clean up the Read/Write impls. Having 2 different functions called
`read`, with different behavior is just confusing. Consolidate on
the Read/Write traits from `std::io`.
* Change the `read_to` and `write_from` functions to be generic over
types that implement `FileReadWriteVolatile` since we are not
allowed to assume that it's safe to call read or write on something
just because it implements `AsRawFd`. Also add `*at` variants that
read or write to a particular offset rather than the kernel offset.
* Change the callback passed to the `consume` function of
`DescriptorChainConsumer` to take a `&[VolatileSlice]` instead.
This way we can use the `*vectored` versions of some methods to
reduce the number of I/O syscalls we need to make.
* Change the `Result` types that are returned. Functions that perform
I/O return an `io::Result`. Functions that only work on guest
memory return a `guest_memory::Result`. This makes it easier to
inter-operate with the functions from `std::io`.
* Change some u64/u32 parameters to usize to avoid having to convert
back and forth between the two in various places.
BUG=b:136128319
TEST=unit tests
Change-Id: I15102f7b4035d66b5ce0891df42b656411e8279f
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1757240
Auto-Submit: Chirantan Ekbote <chirantan@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
Write accessess cannot fail (in the CommandResult sense) and the result
did not carry any data, so remove the response from the Write command.
This should improve the speed of write requests for sandboxed devices.
For example, with the sandboxed serial device, boot time with a release
build of crosvm on my workstation goes from 1.7 seconds to 1.2 seconds,
measured by timing a boot with a missing init so that the kernel panics
and shuts down immediately.
BUG=None
TEST=time crosvm run -p init=bogus vm_kernel
Change-Id: I125bb831235ca741ae1cc6c86a02a5d863d1a211
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1853970
Reviewed-by: Zach Reizner <zachr@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
This change plumbs the jail throughout the arch specific device creation
process. It also adds a custom callback support for the ProxyDevice so
that the main process can interrupt the child serial process when it has
incoming bytes.
TEST=crosvm run
BUG=None
Change-Id: I6af7d2cb0acbba9bf42eaeeb294cee2bce4a1f36
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1752589
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Tested-by: Zach Reizner <zachr@chromium.org>
Commit-Queue: Zach Reizner <zachr@chromium.org>
- Send only one event while re-sampling.
- Don't sent event if the new sr is identical to the old one
This can reduce the rate to trigger the issue.
BUG=chromium:937977
TEST=Build and run lots of aplay and arecord in guest vm
Change-Id: Ibd21f363076c977ae256079e2615094b7ed2408b
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1840752
Tested-by: Chih-Yang Hsia <paulhsia@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Commit-Queue: Chih-Yang Hsia <paulhsia@chromium.org>
Add error handling for adding/removing the tapfd to epoll.
We only remove the tap fd from the poll context if the tap is
readable, i.e. it would busy loop, so don't assume it's removed
from the poll context when there's a deferred rx frame.
BUG=chromium:1010742
TEST=arcvm network works
Change-Id: I84aab2dbe7ea31d724f04d3b3fb0a6916f232300
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1842399
Tested-by: kokoro <noreply+kokoro@google.com>
Tested-by: Stephen Barber <smbarber@chromium.org>
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Reviewed-by: Chirantan Ekbote <chirantan@chromium.org>
Commit-Queue: Stephen Barber <smbarber@chromium.org>
Do not write to the irq_evt eventfd when interrupt_high is false; the
value written to the eventfd is ignored, so despite the '0' in the call,
this would re-trigger the interrupt even when it should not have been
asserted. Since ac97 is a PCI device, its interrupt is level triggered,
and re-asserting it on EOI is handled by the irq_resample_thread code.
BUG=None
TEST=`aplay /dev/urandom` from Crostini on nami
Change-Id: I6ad8e40b818e0495ad58b6902d88dd61103aed9d
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1838762
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Reviewed-by: Chih-Yang Hsia <paulhsia@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
The virtio spec requires that all read-only descriptors appear in the
chain before any write-only descriptors. Enforce this in the
`checked_new` function by adding a new `required_flags` parameter. The
`next_descriptor` function will set this to `VIRTQ_DESC_F_WRITE` if the
current descriptor is write-only. This ensures that once we see a
write-only descriptor, all following descriptors must be write-only.
BUG=b:136127316
TEST=none
Change-Id: Id8f942a4236a20f62f35439f3648dbec17e14c00
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1757239
Auto-Submit: Chirantan Ekbote <chirantan@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Commit-Queue: Chirantan Ekbote <chirantan@chromium.org>
If the guest is unable to return rx queue buffers to the device, we should
temporarily stop polling for reads on the tap fd. Otherwise, we'll spin and
burn CPU needlessly.
BUG=chromium:1010742
TEST=repro from b/141940546
Change-Id: Iac004e870779a8dd39004f44b44e17a2b45bcfa1
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1836914
Tested-by: Stephen Barber <smbarber@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Commit-Queue: Stephen Barber <smbarber@chromium.org>
Convert the virtio pmem device to use the descriptor_utils Reader/Writer
helpers to simplify the code and allow support of arbitrary descriptor
layouts.
BUG=chromium:966258
TEST=./build_test.py
Change-Id: I9ccbdf2833980e4c44e19975f9091f9aea56c94b
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1811713
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Convert the virtio input device to use the descriptor_utils
Reader/Writer helpers to simplify the code and allow support of
arbitrary descriptor layouts.
BUG=chromium:966258
TEST=./build_test.py
Change-Id: Ia9272496dc59b29ea9cde9f6454099c881242d4c
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1811712
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Switching the devices to the new interface reduces code duplication and
will ease fuzzing the devices as they now have a common input and output
interface for descriptors.
BUG=chromium:966258
TEST=vm.CrostiniStartEverything
Change-Id: I823c04dfc24e017433f8e8ab167bbd5dfafd338b
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1647371
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
For each guest memory region, setup the corresponding gpa to hva map
in the kernel vfio iommu table. Then the kernel vfio driver could
get the hpa through gpa. Device could use this gpa for dma also.
BUG=chromium:992270
TEST=none
Change-Id: I04008d68ab2ed182a789d6ee8c97a0ed9e1e4756
Signed-off-by: Xiong Zhang <xiong.y.zhang@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1581141
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
Create VFIO device and VFIO PCI device in create_devices() function, and
intergrate it into PciRootBridge, so guest could see this vfio device.
Add a vfio config parameter, this config point to passthrough or mdev
device sysfs path.
For passthrough case, first user unbind host device from its driver,
then bind host device to vfio-pci. Like:
echo 0000:00:02.0 > /sys/bus/pci/devices/0000:00:02.0/driver/unbind
ech0 8086 1912 > /sys/bus/pci/drivers/vfio-pci/new_id
Finally pass the sysfs to crosvm through
--vfio=/sys/bus/pci/devices/0000:00:02.0
For mdev case, user create a mdev device through
echo $UUID > mdev_type/create, then pass this mdev device to crosvm like
--vfio=/sys/bus/pci/devices/0000:00:02.0/$UUID
BUG=chromium:992270
TEST=none
Change-Id: I0f59d6e93f62f9ab0727ad3a867d204f4ff6ad2d
Signed-off-by: Xiong Zhang <xiong.y.zhang@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1581140
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
In the same spirit as write_all() for the standard io::Write::write()
function, add a write_zeroes_all() function with a default
implementation that calls write_zeroes() in a loop until the requested
length is met. This will allow write_zeroes implementations that don't
necessarily fulfill the entire requested length.
BUG=None
TEST=cargo test -p sys_util write_zeroes
Change-Id: I0fc3a4b3fe8904946e253ab8a2687555b12657be
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1811466
Reviewed-by: Zach Reizner <zachr@chromium.org>
Reviewed-by: Cody Schuffelen <schuffelen@google.com>
Tested-by: kokoro <noreply+kokoro@google.com>
Alloc::PciBar {..} is used as a key in the AddressAllocator's
hashmap, so inform the device about the pci bus/dev numbers.
BUG=chromium:924405
TEST=compile
Change-Id: Ib9d94e516269c1dc9a375c2ceb9775cf5a421156
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1811585
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Tested-by: Gurchetan Singh <gurchetansingh@chromium.org>
Commit-Queue: Gurchetan Singh <gurchetansingh@chromium.org>
According to kernel Documents/vfio.txt and
Documents/vfio-mediated-device.txt,user pass host assigned
device or mdev to crosvm through --vfio parameter, vfio module
open this device and get this device's information.
Implement PciDevice trait on this device, then vfio_pci
module could trap guest pci cfg r/w and mmio r/w,
and transfer this operation into kernel vfio.
Currently the relationship of vfio container:group:device are
1:1:1, in the future it could extend to 1Ⓜ️n.
BUG=chromium:992270
TEST=none
Change-Id: I8006ef65022d56197eaeb464811a59db2ce54b9a
Signed-off-by: Xiong Zhang <xiong.y.zhang@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1580458
Reviewed-by: Zach Reizner <zachr@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
Switch to using Reader/Writer which allows buffers to be passed from
the guest as scatter gathers instead of requiring a single contiguous
buffer.
BUG=chromium:993452
TEST=apitrace replay
Change-Id: Ibe212cfa60eae16d70db248a2a619d272c13f540
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1775365
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Tested-by: David Riley <davidriley@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: David Riley <davidriley@chromium.org>
Make sure all devices join any threads they spawn before returning from
the drop() handler after signaling the exit event.
BUG=chromium:992494
TEST=crosvm exits without errors
Change-Id: I6bc91c32a08f568b041765044caa9aff6f7cf4a9
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1802156
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
The new constructors are shorter and omit the bare `None` in the `anon`
call sites which gave no clues to the reader what the effect of that
`None` was. This should improve readability.
TEST=./build_test
BUG=None
Change-Id: I2e34e7df9a4ccc5da50edf4e963a6a42e3d84b22
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1797188
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Commit-Queue: Zach Reizner <zachr@chromium.org>
Tested-by: Zach Reizner <zachr@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Rather than having a get_canceller() function on UsbTransfer, make the
submit function return the canceller. This makes it clear that the
transfer can't be cancelled before it is submitted.
BUG=None
TEST=None
Change-Id: Ice36c3096a1f8a5aafe93b5d5e27eb371183c19f
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1783599
Reviewed-by: Zach Reizner <zachr@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
This supports virtio disks that depend on multiple file descriptors. All
of the file descriptors are passed to the jail when relevant.
Bug: b/133432409
Change-Id: Idf2e24cd2984c0d12a47a523c13d24c1ba8d173e
Signed-off-by: Cody Schuffelen <schuffelen@google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1691761
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Use the standardized from_le_bytes() functions rather than the byteorder
crate.
BUG=None
TEST=./build_test
Change-Id: I07a062bf63c5d3ae1e25f403713bf9a1677e8cba
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1761155
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Use the stabilized standard to_le_bytes() and from_le_bytes() functions
rather than the byteorder crate.
BUG=None
TEST=./build_test
TEST=cargo test -p devices virtio_pci_common_config
Change-Id: I4106a41484760b9e7e586de07135874238bcadb0
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1761154
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Fix the last instance of this clippy warning:
warning: passing a unit value to a function
... and remove this warning from the "To be resolved" list in
bin/clippy.
BUG=None
TEST=bin/clippy passes without warnings
Change-Id: Ic1d558e935366d80eeadb96bf1ff951ce50edd5b
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1766623
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Reviewed-by: Tomasz Jeznach <tjeznach@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
By registering the notify address with Datamatch::AnyLength, KVM is able
to take advantage of KVM_FAST_MMIO_BUS to accelerate eventfd handling.
Seems this doesn't violate the virtio spec because "writing the 16-bit
virtqueue index" refers to the implementation of front-end driver, for
example, Linux's vp_notify() function. While how to handle the VQ index
write in VMM is not covererd by virtio spec. Here Crosvm ensures that
every VQ has dedicated notify address and KVM implements the notification
by eventfd, which should be fine with the spec.
On eve Pixelbook (R77 host kernel 4.4.185), in 5000 samples, on average
the MMIO write vmexit takes 0.96us with fast_mmio enabled, while it takes
3.36us without fast_mmio.
without fast_mmio:
232812.822491: kvm_exit: reason EPT_MISCONFIG rip 0xffffffff986f18ef info 0 0
232812.822492: kvm_emulate_insn: 0:ffffffff986f18ef:66 89 3e (prot64)
232812.822493: vcpu_match_mmio: gva 0xffffb0f4803a1004 gpa 0xe000f004 Write GPA
232812.822493: kvm_mmio: mmio write len 2 gpa 0xe000f004 val 0x1
232812.822495: kvm_entry: vcpu 1
with fast_mmio:
230585.034396: kvm_exit: reason EPT_MISCONFIG rip 0xffffffff9a6f18ef info 0 0
230585.034397: kvm_fast_mmio: fast mmio at gpa 0xe000f004
230585.034397: kvm_entry: vcpu 1
BUG=chromium:993488
TEST=Boot Crostini on eve and run iperf benchmark
TEST=Analysis kernel trace for vmexit handling time
Change-Id: Id1dac22b37490f7026b6c119c85ca9d104a8a3f4
Signed-off-by: Zide Chen <zide.chen@intel.corp-partner.google.com>
Signed-off-by: Sainath Grandhi <sainath.grandhi@intel.corp-partner.google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1762282
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
We always advertise VIRTIO_GPU_F_VIRGL and don't activate the
worker thread if Renderer::init fails. We're unlikely to
encounter an platform where we can initialize a GBM device, but
can't initialize virglrenderer.
Since our virtio-gpu implementation depends on virglrenderer, we can
pipe 2D hypercalls to virglrenderer (QEMU does this too, when built
with the --enable-virglrenderer option).
Also remove virgl_renderer_resource_info since it's unlikely to work
with non-Mesa drivers.
BUG=chromium:906811
TEST=kmscube renders correctly (though there's a prior bug in closing
the rendering window -- see chromium:991830)
Change-Id: I7851cd0837fd226f523f81c5b4a3389dc85f3e4f
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1743219
Tested-by: Gurchetan Singh <gurchetansingh@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Gurchetan Singh <gurchetansingh@chromium.org>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Auto-Submit: Gurchetan Singh <gurchetansingh@chromium.org>
Now that we're not creating EGL images anymore, we can remove
EGL logic.
BUG=chromium:906811
TEST=freecad works without any local Mesa patches
Change-Id: I09db1c828ae1a331eaeae7c66653a49fe42a04bf
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1725451
Tested-by: Gurchetan Singh <gurchetansingh@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Gurchetan Singh <gurchetansingh@chromium.org>
Auto-Submit: Gurchetan Singh <gurchetansingh@chromium.org>
Reviewed-by: Zach Reizner <zachr@chromium.org>
With YUV support + modifier support coming up, it makes sense to
move GBM allocation inside virglrenderer so we can upstream
our use cases.
In addition, this allows us to use gbm_bo_map(..) for the freecad
issue, which would otherwise be resolved through local patches in
our graphics drivers.
BUG=chromium:906811
TEST=freecad works without Mesa patches
Change-Id: I61db5c58a5bc5a79fda3cec8ad6c322fae6acc9e
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1725450
Tested-by: Gurchetan Singh <gurchetansingh@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Gurchetan Singh <gurchetansingh@chromium.org>
Reviewed-by: Gurchetan Singh <gurchetansingh@chromium.org>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Auto-Submit: Gurchetan Singh <gurchetansingh@chromium.org>
Crosvm's AC97 device had code that was duplicated between playback and
capture stream creation. Abstract that code out so it can be shared.
BUG=chromium:968724
TEST=aplay /dev/urandom within container
Change-Id: If2fb50a0655656726dd9c6255bc84493e91c04e3
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1749948
Tested-by: Fletcher Woodruff <fletcherw@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Chih-Yang Hsia <paulhsia@chromium.org>
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Commit-Queue: Fletcher Woodruff <fletcherw@chromium.org>
Add a new virtio configuration copying function to replace all of the
slightly varying read_config() and write_config() implementations in our
virtio devices. This replaces a lot of tricky bounds-checking code with
a single central implementation, simplifying the devices to a single
call to copy_config() in most cases.
The balloon device is also changed to represent its config space as a
DataInit struct to match most other devices and remove several unwrap()
calls.
BUG=None
TEST=./build_test
TEST=Boot vm_kernel+vm_rootfs in crosvm
TEST=Start Crostini on nami
Change-Id: Ia49bd6dbe609d17455b9562086bc0b24f327be3f
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1749562
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
We don't always shut down the worker threads cleanly, which can lead to a race
when crosvm is exiting. Worker threads that attempt logging to stderr may fail
an expect(), panic, and then panic again trying to write to stderr causing
SIGILL.
Work around this issue for now by using libc's exit, which won't run any
rust-specific cleanup.
BUG=chromium:978319,chromium:992494
TEST=crosvm shuts down without SIGILL/core dumps
Change-Id: I8a99ce8a34220afdf503402d44721a9bea5ec460
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1746830
Tested-by: kokoro <noreply+kokoro@google.com>
Tested-by: Stephen Barber <smbarber@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
Rewrite the virtio block device to use the descriptor Reader/Writer
interfaces - this greatly simplifes the block device code.
This also lets the block device handle arbitrary descriptor layouts,
since the descriptor reader/writer handles that transparently for us.
BUG=chromium:990546
TEST=./build_test
TEST=Boot crosvm with vm_kernel+vm_rootfs on workstation
TEST=Boot full Crostini environment on nami
Change-Id: Ie9a2ba70a6c7ed0ae731660fd991fb88242e275f
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1721371
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Allow use of this helper function in other virtio devices that want to
write virtio descriptor chains as part of their tests.
BUG=chromium:990546
TEST=./build_test
Change-Id: Ib986646dc36b6406c88f20950586e1c665adf167
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1732851
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
This will allow streaming data between a FileReadWriteVolatile and the
descriptor chain Reader/Writer types.
BUG=chromium:990546
TEST=./build_test
Change-Id: Idc97ce99dd1cc340444298f705df4f12e339095d
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1721370
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Zach Reizner <zachr@chromium.org>
This allows moving the read/write cursor around within a chain of
descriptors through the standard io::Seek interface.
BUG=chromium:990546
TEST=./build_test
Change-Id: I26ed368d3c7592188241a343dfeb922f3423d935
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1721369
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Zach Reizner <zachr@chromium.org>
