Linux expects that PCI bars lie within a compatible bridge window, which
are typically specified via _CRS objects in ACPI. This change adds an
option to restrict mmio allocations (i.e. PCI bars) to within a
specified set of ranges. The specified set of ranges is intersected with
the default mmio allocation ranges generated by the crosvm arch code to
produce the final mmio allocation ranges.
This change is required to remove pci=nocrs from the CrOS guest's kernel
command line flags. Removing that flag is a prerequisite for enabling
virtio-iommu, since without the configuration information from ACPI, the
kernel reserves all IOVAs in iova_reserve_pci_windows.
BUG=b:181736020
TEST=boot manatee w/o pci=nocrs kernel cmdline flag
TEST=tast run trogdor|hatch arc.Boot.vm
Change-Id: I0a096420c5d5ef56dd76021951968e264ce40f42
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3499900
Reviewed-by: Junichi Uekawa <uekawa@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: David Stevens <stevensd@chromium.org>
Add a method to SystemAllocator that returns a vector of all pools and
use it to generate the PCI ranges in build_vm. This also fixes a bug
where x86 used the size as the end address for the high_mmio region.
BUG=b:181736020
TEST=tast run trogdor|hatch arc.Boot.vm
Change-Id: Id4a300ec61c15bc08fb069b80428b22461ca935b
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3516668
Reviewed-by: Junichi Uekawa <uekawa@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: David Stevens <stevensd@chromium.org>
Remove the ramoops region from high_mmio when constructing the system
allocator. This means the aarch64 code no longer needs to manually
adjust high_mmio when determining the pci regions.
BUG=b:181736020
TEST=Check arcvm pstore still works
Change-Id: I81ca398a1984f0efb30c0a4d4b620bd50fe9df85
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3516667
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: David Stevens <stevensd@chromium.org>
Use RangeInclusive when constructing an AddressAllocator. A followup
change will add an iterator constructor, so using RangeInclusive makes
that API cleaner.
BUG=b:181736020
TEST=cargo test
Change-Id: I59c28b9cbd06c86482c0c57704fc2753a7886ac6
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3516666
Reviewed-by: Junichi Uekawa <uekawa@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: David Stevens <stevensd@chromium.org>
Have the arch code return the SystemAllocatorConfig instead of a
SystemAllocator. This will be used to allow the core code to apply
additional restrictions on top of the arch code's restrictions.
BUG=b:181736020
TEST=compiles
Change-Id: I4d9ca277f039586e664648492c8744967dcd2ee5
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3516665
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-by: Junichi Uekawa <uekawa@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: David Stevens <stevensd@chromium.org>
If AddressAllocator::allocate_at() fails, check if there is an existing
overlapping allocation and return it with Error::ExistingAlloc if so;
otherwise, return Error::OutOfSpace as before. This allows the caller to
distinguish between failure due to overlap and failure due to a range
that is not managed by the allocator at all.
BUG=b:188011323
TEST=cargo test -p resources
Change-Id: I1f6cde63a995dc461db221c509f4d0a1f0c7f186
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3498208
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Junichi Uekawa <uekawa@chromium.org>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
This reverts commit d2d66bc0a4.
Reason for revert: It turns out that adding the first page to the pool
of memory managed by the MMIO allocator has undesired consequences
since crosvm will actually use it for MMIO regions. The first page
has special semantics in other code though, and thus we get stray
accesses to this region, with hard-to-predict consequences.
BUG=b:188011323
TEST=cq
Original change's description:
> system_allocator: allow more than one region to be in the pool
>
> Allows crosvm-direct to have 0-0xfff regions to be mapped.
>
> limitations: Only the first regions gets reflected in the
> pool_base/pool_size.
>
> BUG=b:188011323
> BUG=b:184815519
> TEST=build
>
> Change-Id: I9da3cb2b8d5611068f9323d6ebf62f44162838b4
> Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3450017
> Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
> Tested-by: kokoro <noreply+kokoro@google.com>
> Commit-Queue: Junichi Uekawa <uekawa@chromium.org>
Bug: b:188011323
Bug: b:184815519
Change-Id: Ib42b3007662a7a49ad876b83a01f1bb88d09d5f7
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3497136
Reviewed-by: Keiichi Watanabe <keiichiw@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Mattias Nissler <mnissler@chromium.org>
Commit-Queue: Mattias Nissler <mnissler@chromium.org>
Make VfioMsixAllocator a generic resource allocator.
BUG=b:185084350
TEST=boot Linux kernel and check dmesg
Change-Id: I4a301d6829600a90fe0ddef4ccccbd4c8f511208
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3166887
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
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>
io ports is a 16 bit thing, check that it is actually 16 bits and fix
the parameters.
BUG=None
TEST=read intel SDM, run crosvm test
Change-Id: I50b6d5593b0699317ac2f852836208a46240714b
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3470601
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Commit-Queue: Junichi Uekawa <uekawa@chromium.org>
Some tests had unrealistic values, fix them in case I want to add some
validation in the future.
BUG=None
TEST=None
Change-Id: I126c83b4ac91442f87aae8be9e84565a7a3d98a2
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3446980
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Noah Gold <nkgold@google.com>
Commit-Queue: Junichi Uekawa <uekawa@chromium.org>
Allows crosvm-direct to have 0-0xfff regions to be mapped.
limitations: Only the first regions gets reflected in the
pool_base/pool_size.
BUG=b:188011323
BUG=b:184815519
TEST=build
Change-Id: I9da3cb2b8d5611068f9323d6ebf62f44162838b4
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3450017
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Junichi Uekawa <uekawa@chromium.org>
In preparation to allow multiple low memory regions to be passed on, I wanted to
make initialization simpler.
Introduce `MemRegion` struct instead of tuple to help me understand it is a base
and size.
BUG=b:188011323
TEST=build
Change-Id: Ie8b54354a25c478d5ad0a0185b7e07d28840dd87
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3439666
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Junichi Uekawa <uekawa@chromium.org>
While reading code, some of the comments need fixing.
BUG=None
TEST=None
Change-Id: Iff9cd639f5e14512680475bcf9240c07551690a5
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3418634
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Junichi Uekawa <uekawa@chromium.org>
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>
This is a generalization of the existing --direct-mmio option with the
following improvements:
* Memory is mapped directly into the guest namespace rather than
handled via MMIO exits.
* The --file-backed-mapping option can be specified multiple times to
set up mappings backed by different files.
* Command line option format is extensible, allowing to add further
parameters in the future as needed.
* O_SYNC mapping is configurable from the command line, making this
useful for mapping both MMIO regions and regular system memory into
the VM.
One limitation that this new implementation brings is that mappings
need to be page-aligned (which isn't the case for --direct-mmio). I'm
not aware of existing users that require fine-granular file-backed
MMIO mappings though, so this is hopefully OK.
The existing --direct-mmio option code is left untouched, but will be
removed later after converting over existing usage.
BUG=b:184815519, b:188011323
TEST=unit tests for command line parsing, manual test to specify --file-backed-mapping and observe that the backing file content is accessible from the guest.
Change-Id: I86aee74d2032caaf9130569203d5749994a4413d
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3328602
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-by: Junichi Uekawa <uekawa@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Mattias Nissler <mnissler@chromium.org>
When allocating PCI addresses, prefer to allocate at the device
granularity (e.g. 00:01.0, 00:02.0, 00:03.0, etc.), but when these are
all allocated, start using functions (e.g. 00:01.1, 00:01.2, 00:01.3).
This enables us to add more PCI devices (31 x 8 instead of just 31)
while maintaining the existing PCI address allocation order for setups
that do not require many devices.
BUG=b:210795995
TEST=boot Linux with 32+ virtio-pci devices
Change-Id: Ic06d3c9939dd78061a51e49b4f7ec457c81c7621
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3343182
Reviewed-by: Dmitry Torokhov <dtor@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
Rather than calling get_high_mmio_base_size() twice, once in the Arch
create_system_allocator() callback and once in build_vm(), it would be
cleaner to retrieve the MMIO regions from the existing system allocator.
Add a couple of helper functions to AddressAllocator so we can get the
original region back.
This also helps in the next patch, which will pass additional
information to get_high_mmio_base_size() that is messy to calculate
twice.
BUG=b:210727578
TEST=Boot crosvm on trogdor64
Change-Id: Ia5c7d7c7ff7edb406a9b4af482b277204f865474
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3373450
Reviewed-by: Dmitry Torokhov <dtor@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
Device gets irq_num from system_allocator when device is added, this
irq_num should return to system_allocator when device is hotplug removed.
this commit adds a ReleseOneIrq interface into VmIrqRequest, it release
one gsi into system_allocator. So vfio-pci device could call it at device
close function.
For msi and msix interrupt vectors, they have irq tube already and could
call ReleaseOneIrq easily.
For legacy INTx, the gsi for vfio-pci's INTx is gotten from host, and this
gsi maybe share with other device, so the commit doesn't release this gsi
at vfio-pci device's remove, otherwise the gsi shared device will be broken.
BUG=b:185084350
TEST=Boot a guest and hotplug vfio pci device repeatedly
Change-Id: Ibcca226b4b5b2092b5bc94bef8219eea82979086
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2955580
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
For bridge's memroy and prefetch memory window, current the limit is
smaller than the base, this means pcie rp doesn't support this window,
but OS still allocates 2M memroy window and 2M prefetch memory window
for it, this has two problems:
a. kernel allocates mmio, but crosvm's system allocator doesn't capture
this allocation and crosvm still use system allocator to allocate mmio
for hotplug in device, so it may have conflict between kernel and crosvm.
b. the 2MB window is too small, as the added device behind this bridge
is allocated mmio from this window by kernel, if device required mmio
is large than 2MB, the device couldn't work.
This commit preallocates 8MB memory window and 64MB prefetch memory window
for one pcie rp, and let kernel allocate mmio for its children devices
from this window, kernel also maintains the allocation and free of this
window, crosvm system allocator doesn't care about which device own the
segment of this window.
BUG=b:185084350
TEST=Boot a guest and check pcie rp bridge window's base, limit and size
Change-Id: I566a2d81b458da1d2d5f07b689fa898453586adc
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3319706
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
Each vfio pci device has a pci address, in order to support vfio
pci device hotplug repeatedly, its pci address should be returned
to system, so it could be hotplug in the next time.
BUG=b:185084350
TEST=Boot a guest and hotplug vfio pci device repeatedly
Change-Id: I0daf5bae9695c4ac87ac09581b80b69048538c0e
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2955578
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
When pci bridige is created, caller should specify its primary bus
number and secondary bus number. And supposing kernel won't modify
them as the current pci device topology is simple enough, once
guest modify them, warn message will be printed.
The secondary bus number shouldn't be assigned to any other device,
so this patch loop the bus number to find a free bus number.
BUG=b:185084350
TEST=Boot a vm with passthrough device and check its function
Change-Id: Iae72a0e0401a6e75c62582456b92792a1a36211a
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2955569
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Original reason for revert: Causes crosvm to not start on AMD based devices b/204222491
Fixed by clamping the maximum physical address to 64 TiB, which is the
largest physical address supported by x86-64 Linux with only 4-level
page tables enabled.
In addition, the high MMIO region size calculation has been split into a
separate function, get_high_mmio_size(), and an off-by-one error in the
size calculation was fixed.
Original change's description:
> Devices: Allocate non pci bar mmio at top of HighMmio
>
> Some devices like virtio-pmem, intel host bridge need private mmio, these private
> mmio isn't exposed to Vm through pci bar, and crosvm use mmio allocation interface
> to allocate them, so guest OS pci bus driver couldn't detect them, and once guest
> OS reallocate pci mmio, these private mmmio could be assigned to pci device bar.
> then resource conflict happens.
>
> This patch allocates these non pci bar mmio at the top of HighMmio, since HighMmio
> is large enough and OS allocates pci mmio from low to high. So these non pci bar
> mmio won't conflict with pci bar.
>
> BUG=b:199442120
> BUG=b:185084350
> TEST=crosvm run --pmem-device xxxx
>
> Change-Id: I615d7901d595dd46ba6362f88b71d96baa9c4c8f
> Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3184549
> Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
> Tested-by: kokoro <noreply+kokoro@google.com>
> Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
This is a reland of https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3184549
BUG=b:199442120
BUG=b:185084350
TEST=Crostini starts on grunt (AMD device)
TEST=crosvm run --pmem-device rootfs -p 'root=/dev/pmem0' kernel
Change-Id: I7c93f7132dbfbc3f38ecbfa0504b59214a1eb6bf
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3251788
Reviewed-by: Xiong Zhang <xiong.y.zhang@intel.corp-partner.google.com>
Reviewed-by: Dennis Kempin <denniskempin@google.com>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@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>
This reverts commit 9557588786.
Reason for revert: Causes crosvm to not start on AMD based devices b/204222491
Original change's description:
> Devices: Allocate non pci bar mmio at top of HighMmio
>
> Some devices like virtio-pmem, intel host bridge need private mmio, these private
> mmio isn't exposed to Vm through pci bar, and crosvm use mmio allocation interface
> to allocate them, so guest OS pci bus driver couldn't detect them, and once guest
> OS reallocate pci mmio, these private mmmio could be assigned to pci device bar.
> then resource conflict happens.
>
> This patch allocates these non pci bar mmio at the top of HighMmio, since HighMmio
> is large enough and OS allocates pci mmio from low to high. So these non pci bar
> mmio won't conflict with pci bar.
>
> BUG=b:199442120
> BUG=b:185084350
> TEST=crosvm run --pmem-device xxxx
>
> Change-Id: I615d7901d595dd46ba6362f88b71d96baa9c4c8f
> Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3184549
> Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
> Tested-by: kokoro <noreply+kokoro@google.com>
> Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
Bug: b:199442120
Bug: b:185084350
Change-Id: Ida1369bf0b70782633b1073ceef50e1bab108f55
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3248906
Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
Commit-Queue: Dennis Kempin <denniskempin@google.com>
Some devices like virtio-pmem, intel host bridge need private mmio, these private
mmio isn't exposed to Vm through pci bar, and crosvm use mmio allocation interface
to allocate them, so guest OS pci bus driver couldn't detect them, and once guest
OS reallocate pci mmio, these private mmmio could be assigned to pci device bar.
then resource conflict happens.
This patch allocates these non pci bar mmio at the top of HighMmio, since HighMmio
is large enough and OS allocates pci mmio from low to high. So these non pci bar
mmio won't conflict with pci bar.
BUG=b:199442120
BUG=b:185084350
TEST=crosvm run --pmem-device xxxx
Change-Id: I615d7901d595dd46ba6362f88b71d96baa9c4c8f
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3184549
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
DT platform bus abstraction requires its own MMIO address range.
Let it allocate from well defined address range.
BUG=b:185504618
TEST=trogdor64-manatee SDHCI and GENIQUP device passthrough boots/works
Change-Id: I357e909433a565968a7f081c727a3b1e420e7729
Signed-off-by: Tomasz Nowicki <tn@semihalf.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2961212
Commit-Queue: Micah Morton <mortonm@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Pci BusNumber is allocated by guest kernel, the BusNumber should be 0
for all the integrated pci devices and vfio-pci device, but pci bridge
and vfio-pcie device may have BusNumber > 0, so caller should know its
device BusNumber and pass it into allocate_pci() and get the desired
PciAddress.
BUG=b:185084350
TEST=None
Change-Id: I3cb18212e6c168c047f655a5f425abdeccbaae55
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2954678
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
Use resources allocator to assign or reserve PCI device address.
For pass-through devices it will enable 1:1 mapping to the host BDF.
Transition to address_allocator for pci address and irq allocations.
BUG=None
TEST=build_test && tast run vm.*
Change-Id: I854da9645a305b7b24acb3dd6d851c3486ed23f7
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2210848
Tested-by: Tomasz Jeznach <tjeznach@chromium.org>
Commit-Queue: Tomasz Jeznach <tjeznach@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
rutabaga_gralloc is a cross-platform, Rust-based buffer
manager.
The rationale for this change is:
1) For the {cross-domain, wayland} context type, we need to
have a good story for the crucial "wl-dmabuf" feature. As
minigbm has been thoroughly tested on ChromeOS and currently
powers the "wl-dmabuf" feature, it only makes sense for us to
have a path to minigbm for the cross-domain prototype. This
will be used by Sommelier.
2) While minigbm allocation works well on Chromebooks, it is
not sufficient for cross-platform purposes. For their Virtual
Graphics Interface (VGI) initiative, Android graphics
virtualization experts have expressed their desire for a Vulkan
based allocator. This will to go alongside cros_gralloc in
minigbm, which is considered by many to be the ""world's
premiere gralloc implementation".
3) Android graphics virtualization experts have expressed their
desire for vkMapMemory(..) to be used when crosvm is in
multi-process mode. Currently, only dma-buf mmap() is supported
for zero-copy blobs in multi-process mode. dma-buf mmap() is not
guaranteed to work on Nvidia (a "must have" for Cuttlefish) or
any other driver for that matter (we *make* it work for ChromeOS).
Possibly only solution: vkMapMemory ;-)
With these goals in mind, here's a summary of the revelant changes:
* Renamed the {gpu_allocator.rs, GpuMemoryAllocator trait} to be
{gralloc.rs, Gralloc trait}.
* Moved all GPU allocation out of the resources crate and into
the rutabaga_gfx crate. This will allow the resources crate to
be focused on managing resources for virtual machines.
* Moved the gpu_buffer crate into the gralloc module in the
rutabaga_gfx crate. The same functionality is now under
"minigbm.rs", "minigbm_bindings.rs" and "rendernode.rs"
* Added an optional dependency on vulkano.rs. vulkano.rs is a safe
Rust wrapper around the Vulkan api [a]. It's emphasis on type
safety makes a good fit for crosvm, though there are other high
quality crates out there (gfx-rs, ash.rs). Though development
has slowed down, it should satisfy goals (2) and (3) quite easily.
* Added a system_gralloc implementation based on memfd. This can be
used when minigbm or Vulkano features are not used, to replicate the
highly useful "wl-shm" feature in Sommelier. Astute observers will
note this can also enable seamless Wayland windowing without GPU
features for Android too. Some minor changes to the base crate were
needed.
* Cut down on the amount of DrmFormats to the subset needed by
Sommelier and cros_gralloc.
* Moved checked arithmetic into it's own file.
* Internally renamed to "wl-dmabuf" feature to be the "minigbm"
feature. This is because "wl-dmabuf" has a dependency on minigbm.
* Small rutabaga_gfx cleanups
[a] https://github.com/vulkano-rs/vulkano/blob/master/DESIGN.md
BUG=b:146066070, b:173630595, b:150239451
TEST=launch virtual machine with 2D mode
TEST=launch virtual machine with 3D mode
TEST=run sommelier with "wl-dmabuf" and "wl-shm"
Change-Id: I693a39cef64cd98e56d843d3c60caa7983d4d6e1
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2626487
Tested-by: Gurchetan Singh <gurchetansingh@chromium.org>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Commit-Queue: Gurchetan Singh <gurchetansingh@chromium.org>
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>
This will allow allocating an MMIO region that can be in the high (>4G)
address space, but prefers to be in the low region if possible.
BUG=b:169430664
TEST=Boot Crostini with GPU
Change-Id: If90c0d90c95b0dcb3cb451eb7f573eede5ea3b2e
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2436370
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-by: Zach Reizner <zachr@chromium.org>
For now, this crate simply re-exports all of sys_util, but it will
be updated to provide new interfaces when needed. This is the
first step to making crosvm not directly depend on sys_util, so
that we can make the interface changes we need without fear of
negatively affecting (i.e. completely breaking) other usages
within chromeos.
BUG=b:162363783
TEST=./build_test
Change-Id: I7d0aa3d8a1f66af1c7fee8fd649723ef17027150
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2325168
Tested-by: Michael Hoyle <mikehoyle@google.com>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Michael Hoyle <mikehoyle@google.com>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Support for address_allocator by-alloc release.
Address regions coalescing is performed at the
release time.
BUG=None
TEST=cargo test -p resources && tast test vm.*
Change-Id: Ibd39dac923d1b2f8b6a711d2a9fcbb662fc95bdc
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2209171
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Commit-Queue: Tomasz Jeznach <tjeznach@chromium.org>
Support for address range allocation at specific
location and size. Allocation is successful only
if requested range is available and free.
Change to address_allocator allocation algorithm
from contiguous memory allocation to first-fit like
implementation to allow memory release/reuse in future.
BUG=None
TEST=cargo test -p resources && tast test vm.*
Change-Id: I58218f5a2c6a215152904cc1cf0748e842fa7374
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2203297
Tested-by: Tomasz Jeznach <tjeznach@chromium.org>
Commit-Queue: Tomasz Jeznach <tjeznach@chromium.org>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Simple refactor of PCI device addressing to use
PciAddress type providing bus:device.function number.
BUG=None
TEST=build_test & tast run crostini.Sanity
Change-Id: I7755ad6b31aa8c882475cd8212630e1cc86ef49e
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2172766
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Tomasz Jeznach <tjeznach@chromium.org>
Adds support for pstore on ARCVM on x86_64.
The backend file of the buffer will be passed via argument of the crosvm.
BUG=b:144962428
TEST=kernel crash on eve-arcvm, check /sys/fs/pstore/console-ramoops-0
Launch crostini manually on eve-arcvm
Change-Id: I29492ac7a9067aa2ae23eb03fbb942ab7dd3aa8d
Signed-off-by: Kansho Nishida <kansho@google.com>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1973391
Reviewed-by: Keiichi Watanabe <keiichiw@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Kansho Nishida <kansho@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>
The guest kernel will allocate from PCI range of the virtgpu
device, and send physical addresses via a hypercall.
Right now, only support buffers that can be mmap'ed. We could
add optimizations for GBM buffers later if needed.
BUG=chromium:924405
TEST=compile
Change-Id: I094de96a2c35bcd2e18c8a6a2d8cdc39bb392e36
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1626794
Tested-by: kokoro <noreply+kokoro@google.com>
Tested-by: Gurchetan Singh <gurchetansingh@chromium.org>
Commit-Queue: Gurchetan Singh <gurchetansingh@chromium.org>
Reviewed-by: Zach Reizner <zachr@chromium.org>
