Chrome OS does not currently package memoffset-0.6.5 (only 0.6.4), so
building within cros_sdk will fail to find the expected crate. We don't
need any API requirements past what is provided by 0.6, so just request
that. At the same time, upgrade the cros_asyncv2 memoffset from 0.5 to
0.6 so we don't need two separate versions. (The crosvm ebuild was not
requiring a 0.5 version, so this was technically a bug.)
BUG=None
TEST=cargo build
TEST=emerge-hatch crosvm
Change-Id: I2b6c778cdbf57977e6fffd73867e0c4818c44a61
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3502625
Auto-Submit: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-by: Vineeth Pillai <vineethrp@google.com>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
When Crosvm receives a KVM_SYSTEM_EVENT_S2IDLE event, it means the
guest VM is entering s2idle suspend state, and it's requesting the
host to put the whole system in the same state.
If hypervisor sends the request on behalf of a privileged VM, Crosvm
can initiate s2idle suspend on the host since it's been granted with
system power management privileges.
BUG=b:199383670
TEST=trigger s2idle from ChromeOS guest
Change-Id: I589a7e1d93abb5ab971fd58de94ee75df48c3c6d
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3414635
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Dmytro Maluka <dmaluka@google.com>
Commit-Queue: Allen Webb <allenwebb@google.com>
These may be used by the pVM firmware or bootloader for protected VMs on
Android to indicate the reason why a VM failed to boot, such as a
verification failure, and so we want to have them in the logs for
debugging.
BUG=b:220084045
Change-Id: I449873fb85d590d9e7978a93648e90586ecfff10
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3473452
Reviewed-by: Dmitry Torokhov <dtor@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Andrew Walbran <qwandor@google.com>
Run tests for sys_util_core, poll_token_derive and balloon_control on
windows.
Using dotfiles to disable/serialize test runs of a subset of crates does
not work well with third party crates as it forces us to commit the dot
file to the crate.
The patch modifies and uses the script that runs linux tests.
This patch also allows us to
- build/test child crate even if parent crate has disabled build/test.
- avoid building crosvm if it is not explicitly specified.
RIP short lived .windows_build_test_skip. You allowed us to run noop
kokoro tests.
Test: py .\tools\impl\test_runner.py --arch x86_64
Bug: b:215610772
Change-Id: Icc6d04ffd7c0c33d4f60aeac16fc7d23881c387d
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3459809
Reviewed-by: Dennis Kempin <denniskempin@google.com>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Vikram Auradkar <auradkar@google.com>
This should have been updated with the pKVM ABI change in
https://crrev.com/c/3330204, as protected mode is now selected when
creating the VM, and pVM firmmware is optional.
BUG=b:209794844
Change-Id: I2850b4c2ef523a15b8215df070eb9da066a67cb5
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3420330
Reviewed-by: Keiichi Watanabe <keiichiw@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Andrew Walbran <qwandor@google.com>
It's a simple enum, so it should be Copy.
TEST=tools/dev_container tools/run_tests
TEST=tools/dev_container tools/run_tests --target=vm:aarch64
Change-Id: Ieeacedb10fa6a93d6cdd9878097c66aa936f8e30
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3420329
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Andrew Walbran <qwandor@google.com>
The patch also adds files to skip building and testing crates on
windows. When we run
```
tools/windows/build_test.py --skip_file_name .windows_build_test_skip
```
the build/test succeeds without actually doing anything as build/test
for all crates is skipped by creating '.windows_build_test_skip'.
Bug: 213170957
Test: Ran script on downstream repo
Change-Id: Iebd2cea463ee722be4feaed88229e1fb5e9fd6c5
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3417918
Reviewed-by: Dennis Kempin <denniskempin@google.com>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Vikram Auradkar <auradkar@google.com>
The newly-added KVM_CHECK_EXTENSION call to retrieve
KVM_CAP_ARM_VM_IPA_SIZE was checking for errors by looking for return
values less than 0, but KVM_CHECK_EXTENSION returns 0 for unknown
extensions (e.g. on older kernels before this extension was introduced).
Treat 0 as unsupported as well so the default 40-bit size is returned in
this case.
BUG=b:215174283
BUG=b:210727578
TEST=Boot crosvm on kevin (arm device with 4.4 host kernel)
Fixes: 891ea3e88a ("aarch64: limit high MMIO addresses to VM phys addr range")
Change-Id: I19a3ae00862882c030c1e776cb3ca10edeffd339
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3399344
Reviewed-by: Dennis Kempin <denniskempin@google.com>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
The high MMIO region of the system allocator was previously set up with
the upper limit of 2**64 - 1, but physical addresses on actual systems
are not that large. Now that pmem allocates from the upper end of the
high MMIO region, the size of the allocator actually matters.
Limit the aarch64 high MMIO allocator to the VM's physical address size
so that addresses at the end of the allocation range are usable and pmem
devices can be created on aarch64.
BUG=b:210727578
TEST=Boot with pmem device on aarch64
Change-Id: I67962c2b03f0f265349ae889c988eb69ef526f4a
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3366858
Reviewed-by: Dmitry Torokhov <dtor@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Andrew Walbran <qwandor@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
Add a function that returns the size of guest physical addresses. This
wraps KVM_CAP_ARM_VM_IPA_SIZE on aarch64 and CPUID to query host
physical address size on x86_64. This replaces the phy_max_address_bits
function from x86_64, and all callers are migrated to use the new Vm
function instead.
BUG=b:210727578
TEST=check on trogdor64 with debug prints
Change-Id: I3107fe6357fcf166b7ad0e2a7c55919705da3b0c
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3364971
Reviewed-by: Dmitry Torokhov <dtor@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Alexandre Courbot <acourbot@chromium.org>
Reviewed-by: Andrew Walbran <qwandor@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
This is useful mostly for development and testing.
BUG=b:209795495
Change-Id: Ie8c4dfade2cbc770daa7d97e22b1574fdf895a19
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3330210
Reviewed-by: Will Deacon <willdeacon@google.com>
Reviewed-by: Chirantan Ekbote
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
This involves two main changes:
* Protected VMs must be created with KVM_VM_TYPE_ARM_PROTECTED.
* pVM firmware is now loaded by IPA rather than memslot ID.
There are also a lot of trivial changes because the ProtectionType enum
was moved from the devices crate to the hypervisor crate.
BUG=b:209794844
TEST=Will tested manually with patched kernel and dummy firmware
Change-Id: I1dd75e20063ca4736f155292ca5f70b94664fdd9
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3330204
Auto-Submit: Andrew Walbran <qwandor@google.com>
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
As before, some are automatically fixed by clippy, some manually
fixed.
BUG=b:192373803
TEST=./tools/presubmit
Change-Id: Ifcab4cf60775ee6bb7d4530af6406a74958432ed
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3283683
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Dennis Kempin <denniskempin@google.com>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Developers may need to manually clean up the common/enumn directory - it
will be left behind if there are build artifacts (Cargo.lock, target
directory, etc.):
rm -rf common/enumn
BUG=b:205344148
TEST=cargo build
TEST=tools/presubmit
TEST=emerge-hatch crosvm # with https://crrev.com/c/3265967
Change-Id: I1af3bdd22f40e87895a78a5cbc8033476058c927
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3278774
Reviewed-by: Dennis Kempin <denniskempin@google.com>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
The value 0 passed to KVM_CREATE_VM has a special meaning for KVM/arm64.
It indicates that the VM is configured with a 40bit IPA space.
However, not all HW support such an IPA space, and the KVM_CREATE_VM
call fails on these systems.
In order to maximize compatibility, we can ask KVM for the maximum
supported IPA size, and use that as an input to KVM_CREATE_VM, at
which point the kernel will instantiate a VM with that IPA size.
This enables crosvm on exotic hardware such as the Apple-M1.
TEST=tools/run_tests
Change-Id: I7ea39ac6e5de6a1389c0c30cdfeb7c970c411589
Signed-off-by: Marc Zyngier <mzyngier@google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3124677
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Chirantan Ekbote <chirantan@chromium.org>
This change contains the results of running
./tools/contib/cargo_refactor.py
This will break the next uprev, and needs to be synchronizized
with the corresponding ebuild changes in https://crrev.com/c/3248925
BUG=b:195126527
TEST=./tools/run_tests
Change-Id: Ied15a1841887bb8f59fba65b912b81acf69beb73
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3248129
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Dennis Kempin <denniskempin@google.com>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
If we need descriptor for things that aren't file or other shared_memory objects, we can create a separate descriptor change, but it shouldn't be the default.
This reverts commit 533c5c8258.
Reason for revert: this causes conflicts with other platforms.
Original change's description:
> vm_memory: Add from_desciptor() in MemoryMappingBuilder
>
> MemoryMappingBuilder had `from_file()` and `from_shared_memory`, which
> are almost the same. So, this commit adds `from_descriptor()` to replace
> both of the two.
>
> BUG=b:194137301
> TEST=build
>
> Change-Id: Ia13f5e8e0f95a5c32e47dc9b3be13b7a7fa510bf
> Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3159881
> Tested-by: kokoro <noreply+kokoro@google.com>
> Commit-Queue: Keiichi Watanabe <keiichiw@chromium.org>
> Reviewed-by: Chirantan Ekbote <chirantan@chromium.org>
Bug: b:194137301
Change-Id: Ie203ee3eb3dcddd41c5e55b6980dc6292eb24f85
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3183183
Auto-Submit: Udam Saini <udam@google.com>
Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
Commit-Queue: Udam Saini <udam@google.com>
Reviewed-by: Noah Gold <nkgold@google.com>
Reviewed-by: Keiichi Watanabe <keiichiw@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
MemoryMappingBuilder had `from_file()` and `from_shared_memory`, which
are almost the same. So, this commit adds `from_descriptor()` to replace
both of the two.
BUG=b:194137301
TEST=build
Change-Id: Ia13f5e8e0f95a5c32e47dc9b3be13b7a7fa510bf
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3159881
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Keiichi Watanabe <keiichiw@chromium.org>
Reviewed-by: Chirantan Ekbote <chirantan@chromium.org>
On all architectures, KVM_CREATE_VM takes an argument known
as the 'machine type identifier'. This machine type is
architecture dependent, and the documentation helpfully says:
You probably want to use 0 as machine type.
So let's do that.
Change-Id: I8a8a0f7b78e32012c5ab841097c05a02fe0532ff
Signed-off-by: Marc Zyngier <mzyngier@google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3124676
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Andrew Walbran <qwandor@google.com>
Reviewed-by: Chirantan Ekbote <chirantan@chromium.org>
Reviewed-by: Keiichi Watanabe <keiichiw@chromium.org>
Reviewed-by: Dennis Kempin <denniskempin@google.com>
Tree-wide cleanup of new clippy warning in Rust 1.54 that warns about
needless borrows:
error: this expression borrows a reference (`&...`) that is
immediately dereferenced by the compiler
https://rust-lang.github.io/rust-clippy/master/index.html#needless_borrow
BUG=b:197251702
TEST=bin/clippy # with rust-toolchain = 1.54.0
Change-Id: Ib702ec524d4623d264a00ec11dbc2150c411a67b
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3108321
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Allen Webb <allenwebb@google.com>
Reviewed-by: Tomasz Jeznach <tjeznach@chromium.org>
Reviewed-by: Chirantan Ekbote <chirantan@chromium.org>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
The conversion code for KVM <-> hypervisor representation of the Local
APIC state used the unsafe mem::transmute() function to view an array of
i8 as u8 instead for use with the Rust endian conversion functions.
Casting between integer types of the same size with `as` is defined in
Rust as a "no-op" (the bitwise representation is preserved), just like
in C, so transmuting at the slice level is not needed. These can instead
be written as simple loops to avoid the unsafe code.
To ensure this does not regress code quality, I have compared the code
generated for the x86-64 release build. The kvm_lapic_state to
LapicState conversion compiles to identical code, and the reverse
compiles to slightly different code (the compiler decides to emit a loop
instead of unrolling the 64-element copy), but the conversion of each
element still compiles down to a pair of MOV instructions.
The corresponding unit test has also been updated to avoid transmute, as
it was unnecessary there - the individual array element can be cast with
the `as` operator rather than transmuting the whole array.
BUG=None
TEST=cargo test -p hypervisor
Change-Id: I7e792b5507235e5234afe114a1ca744931e047d5
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2947934
Reviewed-by: Chirantan Ekbote <chirantan@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
ARM supports paravirtual clocks or "stolen time" accounting by
allocating a page of memory in the guest memory map for a pvtime
structure and then sharing that address with the guest and updating it
in the host to tell the guest about how much time has been stolen from
the guest because the guest hasn't been running. Read all about it
here[1]. Add support to crosvm to allocate this page (for now a 64K size
chunk of memory that all VCPUs share access to) and then set the address
of the page in each VCPU when starting up.
[1] https://www.kernel.org/doc/html/latest/virt/kvm/arm/pvtime.html#stolen-time
BUG=chromium:1130828, b:169094241
TEST=emerge-trogdor crosvm && cros deploy $DUT crosvm && \
tast run $DUT crostini.Basic.buster_stable;
dmesg | grep pv # in the guest shows "steal time"
Change-Id: Ie3497bb22fb0e38eeff8ebac14b4213824bb6ca6
Disallow-Recycled-Builds: test-failures
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2871394
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Stephen Boyd <swboyd@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Enable support for runtime verification of number
of irqchip kernel emulated inputs, up to 120 pins.
KVM implementation supporting extended input pins shall
report KVM_CHECK_EXTENSION/KVM_CAP_IOAPIC_NUM_PINS value.
BUG=b:179648314
TEST=On systems with 24/120 pin IOAPIC kvm emulation.
Change-Id: I80063216310e427d664e3eaca3aba27e8a972cde
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2893366
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Tomasz Jeznach <tjeznach@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
This CL addresses some minor issues with the existing interface:
1. from_descriptor is too generic for some platforms that require
special handling for file/File backed mappings.
2. Nearly all call sites pass either File or SharedMemory. Now
we just have from_ methods for those types to preserve type
information.
3. Other platforms require additional fields in MemoryMapping, so a
tuple struct no longer makes sense.
4. The mmap syscall error message was misleading as we use it for more
than just the mmap syscall.
BUG=None
TEST=builds
Change-Id: I74c41bad52bb81880a11231cd18f47e233548a24
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2815614
Reviewed-by: Udam Saini <udam@google.com>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Noah Gold <nkgold@google.com>
Allowing each region to have a separate backing FD will make it possible
to build GuestMemory from the vhost `SET_MEM_TABLE` message that
transmits the memory regions for virtio queues in vhost-user devices.
Change-Id: I6f9bc6136915da9d873ea896823e3b8f426ca69d
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2795282
Reviewed-by: Gurchetan Singh <gurchetansingh@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Dylan Reid <dgreid@chromium.org>
Some judgement calls were made about unnecessary wrapping. Usually they
would get resolved by removing the wrapping or returning a convenient
error, but the ones that returned results for consistency with other
functions were added to the allow list.
The error handling in the usb code had a lot of unit error types which
is now a clippy lint. This was resolved by either removing the result
entirely or returning a convenient error.
The field_reassign_with_default lint is faulty and was added to the list
of supressions. This affected virtio-wayland code.
BUG=b:179277332
TEST=cargo clippy with rustc 1.50+
Change-Id: Ie812cdeaf7c42f4f2b47b1dc87f05a7c87a60f8f
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2757510
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Reviewed-by: Gurchetan Singh <gurchetansingh@chromium.org>
Auto-Submit: Zach Reizner <zachr@chromium.org>
Commit-Queue: Dennis Kempin <denniskempin@google.com>
For some PCI device, its MMIO bar size may not be page size aligned.
When setting user memory region for such bar with not aligned size, KVM
will report failure back and failed to map that bar. As current crosvm
can continue run with this failure, the performance will be hurt as each
time when guest is accessing this bar, it will trap to hypervisor.
To resolve this, extend the size to be page size aligned when setting
user memory region in KVM and do DMA map. This should be safe to extend
because the mmap actually rounds up the mmap size to be page aligned.
BUG=None
TEST=boot vm with a passthrough device whose bar has unaligned size
Change-Id: Ic816984ec503edf7f12da4893b78d996ebf93976
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2717448
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Dylan Reid <dgreid@chromium.org>
More recent Intel IO-APICs can support more than 24 interrupt
lines. This change enables variable size of IO-APIC lines for
user level IO-APIC emulation code (split-irqchip).
Reported version and supported IO-APIC registes matching ICH10
implementation of IO-APIC device.
BUG=b:181795297
TEST=boot and allocate irq from upper range.
Change-Id: I56480befb39c4c268266f04e4a93105402248772
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2733579
Tested-by: Tomasz Jeznach <tjeznach@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Tomasz Jeznach <tjeznach@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Most users will want to keep the default `/dev/kvm` path. However, in
certain environments, namely Borg, the KVM device node may be located
elsewhere.
This is the first of a set of related changes that will make hard-coded
device paths configurable.
BUG=None
TEST=./ci/builder --vm ./run_tests
Change-Id: I6087879c535be3779e20eff1f8fb5080f80cf020
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2736520
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Commit-Queue: Dylan Reid <dgreid@chromium.org>
The args may be interpreted as pointers for some capabilities, so the
caller must ensure that any such pointers are allocated appropriately.
BUG=b:181564686
TEST=cargo test
Change-Id: I244f4d9417e588a6be5681f4718bb9ad7b262c3e
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2726709
Auto-Submit: Andrew Walbran <qwandor@google.com>
Commit-Queue: Dylan Reid <dgreid@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Dylan Reid <dgreid@chromium.org>
- Add an address space region for the protected KVM firmware.
- Query firmware size, mmap something that size and create a memslot.
BUG=b:163789172
TEST=cargo test
Change-Id: I054cf5d763c980d073c17bce70e85a781816b64d
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2623942
Auto-Submit: Andrew Walbran <qwandor@google.com>
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Commit-Queue: Andrew Walbran <qwandor@google.com>
Tested-by: kokoro <noreply+kokoro@google.com>
Support virtio-fs's DAX (direct memory access) operation which allows the guest
to directly access file pages.
Specifically, FUSE_SETUP_MAPPING and FUSE_REMOVE_MAPPING operations are
supported.
This option can be used by specifing `dax` option when mount a file system in
the guest.
The DAX optoin improved file I/O performance in most cases.
In Fio tests, both of read and write score were improved by 1.3-14x depending on
test cases.
In Blogbench tests, which create many small files, DAX improved the write score
by 1.5x while the read score was reduced to ~25% (20391 -> 4593).
Here is an excerpt of results:
Fio
* seq_read: 10.2x (143528 -> 1464911)
* seq_write: 3.3x (61253 -> 896791)
* rand_read: 11.6x (138753 -> 1612739)
* rand_write: 14.6x (61253 -> 896791)
* surfing_read: 1.3x (98473 -> 127907)
* surfing_write: 1.3x (83309 -> 108089)
Blogbench
* read: 0.23x (20391 -> 4593)
* write: 1.50x (248 -> 373)
BUG=b:147341783
TEST=Run vm.{Blogbench, Fio} with CL:2291856
Change-Id: I4a47c601412ed32d926de6304337e1594252d258
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2108315
Tested-by: kokoro <noreply+kokoro@google.com>
Tested-by: Keiichi Watanabe <keiichiw@chromium.org>
Commit-Queue: Keiichi Watanabe <keiichiw@chromium.org>
Reviewed-by: Chirantan Ekbote <chirantan@chromium.org>
Call KVM_REG_ARM_PSCI_VERSION to see the PSCI version and use the value
when creating a device-tree node.
Also stop setting PSCI constants which are ignored by kernel.
BUG=chromium:1141902
TEST=run crosvm on krane and check /proc/device-tree/psci/compatible
Change-Id: I4b8ed7620d7f0e2205b84b5f08cfaa5ae8d94339
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2529289
Tested-by: Keiichi Watanabe <keiichiw@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Keiichi Watanabe <keiichiw@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Regenerate bindings.rs for kvm and add comments about how to generate it.
As a result, manually-added hack related to zero-sized arrays' alignment
was removed, as the bug had been fixed:
https://github.com/rust-lang/rust-bindgen/issues/684
BUG=none
TEST=build_test
Change-Id: I257975ce3cd4667b39381ddafd8b08d9e91de655
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2532546
Tested-by: Keiichi Watanabe <keiichiw@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Commit-Queue: Keiichi Watanabe <keiichiw@chromium.org>
New Vcpu methods get the vcpu ID for a vcpu and the current thread.
A new VcpuX86_64 method injects NMIs. request_interrupt_window now
takes a bool parameter, so the request window flag can be turned off.
These methods are to support the userspace irqchip. The thread-local
vcpu ID will help route MMIO requests to the appropriate APIC. Clearing
the interrupt window request is needed because KVM won't clear the
flag if the kernel APIC is disabled.
BUG=chromium:1077058
TEST=boots with KVM kernel irqchip, KVM split irqchip
Change-Id: I2a08988e71663f0d2a63985f3239568c7a9befea
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2419673
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Colin Downs-Razouk <colindr@google.com>
For hypervisors that do not handle ioevents in-kernel, we need to make
some minor adjustments to the hypervisor abstraction api:
- register/unregister_ioevents needs &mut self instead of &self
- move handle_io_events from Vcpu to Vm, because the ioevents are
registered on the Vm not the Vcpu.
BUG=chromium:1077058
TEST=cargo test -p hypervisor
Change-Id: Iddd61502a716143e117c454e64a9c9bd2578148d
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2405299
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Colin Downs-Razouk <colindr@google.com>
This involves removing RawFd trait implementations from
Event and Timer, and resolving the echoing dependencies from
there.
Ultimately, this results mostly in changes across ioctl
(new thin layer in base), kvm, msg_on_socket, and a few other
areas. As usual, most changes are negligible.
BUG=b:162363783
TEST=./build_test
Change-Id: I47dec94666bc3430fed627b4767919c9cb4fce6f
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2462330
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Michael Hoyle <mikehoyle@google.com>
Move MemSlotOrd out of the kvm hypervisor implementation so other
implementations can share the same struct.
BUG=chromium:1077058
TEST=cargo test -p hypervisor
Change-Id: I9d992699defab53e32517bb40e29192e31c45983
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2405301
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Colin Downs-Razouk <colindr@google.com>
This allows code to create references to traits such as `&dyn Vcpu` for
`&dyn IrqChip`. This also allows keeping such traits inside of opaque
`Box` like wrappers.
To achieve this, trait methods referencing `Self` have an additonal
`where` clause that restricts them to sized types. Associated types are
removed and replaced with their trait equivalents (as parameters) or an
opaque Box (as return values).
To work around certain cases where a concrete type is needed, such as
for KVM based `IrqChip` impls, the `downcast_rs` trait is used to allow
`dynamic_cast` style casting.
The binary size impact of this change is small. For comparison, here is
the size change with this commit:
armv7a -0.49% (-9 kiB)
aarch64 -0.17% (-3 kiB)
x86_64 +1.77% (+40 KiB)
BUG=None
TEST=build_test
Cq-Depend: chromium:2466660
Change-Id: I1a408734832d2971ea2979c1bc64b2ffe051b02d
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2439298
Reviewed-by: Zach Reizner <zachr@chromium.org>
Tested-by: Zach Reizner <zachr@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Zach Reizner <zachr@chromium.org>
Note the CL size is large entirely due to the rename,
the changes are mostly negligible.
Also making a few small additional changes in sys_util
areas that don't need much attention in base. This includes
typedefing and adding specific imports for areas that don't
require significant interface changes.
BUG=b:162363783
TEST=./build_test
Change-Id: I4a2c9c4cdce7565806ed338e241c6b8c82c855c7
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2415180
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Michael Hoyle <mikehoyle@google.com>
This removes RunnableVcpu as an associated type on Vcpu traits, instead
using the new VcpuRunHandle type to enforce the same guarantees. This
simplifies implementations of Vcpu becuase only one impl is needed with
all the methods provided.
The other advantage to removing RunnableVcpu was that a Vcpu could be
try_cloned, defeating the purpose of consuming the Vcpu in the first
place with to_runnable. The type system and ownership model gave a false
sense of compile time safety. The usage of VcpuRunHandle is more
consistent with the fact that it is a runtime check that the programmer
is doing the safe thing. It's also consistent with all hypervisor impls.
BUG=None
TEST=./build_test
Change-Id: I14bbc5d939744cb92fb9cb3152edb82603cadd11
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2442096
Reviewed-by: Zach Reizner <zachr@chromium.org>
Tested-by: Zach Reizner <zachr@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Zach Reizner <zachr@chromium.org>
