summary refs log tree commit diff
path: root/src
Commit message (Collapse)AuthorAge
* crosvm: add support for bpf policy filesMatt Delco2020-01-22
| | | | | | | | | | | | | | | | | | | | Change adds supports for providing pre-compiled bpf files as the policy file for jailing. In short it's more effient to compile once on the build machine than each time at runtime. Additionally libminijail's support for more efficient bpfs (which use a binary tree instead of a linear search) is currently only available via tools that are based around pre-compiled use. BUG=None TEST=Ran build_test and verified that tests can pass with both bpf and policy files (though the tests might only exercise the jail for the plugin). Change-Id: Idd93e3c802fc79da93850d6bad1db660576bc9ba Signed-off-by: Matt Delco <delco@chromium.org> Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1914416 Tested-by: kokoro <noreply+kokoro@google.com> Reviewed-by: Dylan Reid <dgreid@chromium.org>
* MsgSocket: Don't implement DerefChirantan Ekbote2020-01-17
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The Deref trait is usually only implemented by smart pointers or by trivial wrappers around the underlying type. MsgSocket does not fit into either category because it wraps a `UnixSeqPacket` to provide new functionality. Having it implement can lead to confusing error messages, especially for people who are new to rust and are not familiar with the Deref trait and Deref coercion. For example, calling `sock.send()` on a MsgSocket without first adding `use msg_socket::MsgSender` leads to the compiler complaining about mis-matched types for the `send` method. `UnixSeqPacket::send` expects a `&[u8]` while `MsgSocket::send` expects a `&M`. The compiler also gives no clues that it is implicitly coercing the socket to a `&UnixSeqPacket` and using the `send` method from there. Drop the `Deref` implementation. `MsgSocket` already implements `AsRef<UniqSeqPacket>` so anything that _needs_ to access the underlying `UnixSeqPacket` should just use that. BUG=none TEST=unit tests Change-Id: If02ef7173ae21d85d517e808489ed4d6d09ae90b Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2002997 Reviewed-by: Daniel Verkamp <dverkamp@chromium.org> Reviewed-by: Stephen Barber <smbarber@chromium.org> Tested-by: kokoro <noreply+kokoro@google.com> Commit-Queue: Chirantan Ekbote <chirantan@chromium.org>
* crosvm: log 'run' arguments parsing errors also to syslogDmitry Torokhov2020-01-16
| | | | | | | | | | | | | | | | | | | | We are inconsistent in when we print errors to stdout versus strerr and syslog, with argument parsing errors from "crosvm run" subcommand only logged to stdout whereas other types of errors going to stederr/syslog. Let's make sure argument parsing errors will be logged to stderr/syslog as well, otherwise it is very confusing to analyze crosvm failures when started by VM concierge as it does not dump crosvm stdout/stderr anywhere. BUG=None TEST=Try "crosvm run" with bad arguments when syslog is available, see error messages in syslog. Change-Id: Ie7a284f4a604716ecc655c5a38ec2147b55d63a3 Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2002677 Commit-Queue: Dmitry Torokhov <dtor@chromium.org> Tested-by: Dmitry Torokhov <dtor@chromium.org> Tested-by: kokoro <noreply+kokoro@google.com> Reviewed-by: Dylan Reid <dgreid@chromium.org>
* gpu_display: implement stub displayJason Macnak2020-01-14
| | | | | | | | | | | | | | | | | | | | | | | | | | | | 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>
* Add a new command to dump version info.Yi Sun2020-01-13
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | With command "version", we can dump version information via "crosvm version". By default it would dump the version specified in Cargo.toml like following: "crosvm package version 0.1.0". And if the environment varable "PKG_VERSION=123456" is given during building crosvm, it would dump followed by a package version, like following: "crosvm package version 0.1.0-123456". Sometimes, we need to know which exact version does a crosvm binary come from for developing and testing. It is useful if the git sha is built-in a crosvm binary. BUG=none TEST=PKG_VERSION=xxxxxx Cargo build && crosvm version TEST=PKG_VERSION=xxxxx emerge-eve crosvm && crosvm version TEST=./bin/fmt TEST=./bin/clippy TEST=./build_test.py --x86_64-sysroot /build/eve no errors reported and all behavor is as expected. v2: Refine version info. Change-Id: I89686dbe6ab2888d8a6ce5752a37241b4c00160d Signed-off-by: Yi Sun <yi.sun@intel.corp-partner.google.com> Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1989256 Reviewed-by: Zach Reizner <zachr@chromium.org> Tested-by: kokoro <noreply+kokoro@google.com>
* devices: remove user_command from proxy deviceZach Reizner2020-01-09
| | | | | | | | | | | | | | | | | | 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>
* crosvm: align pmem region size to 2MiBStephen Barber2020-01-07
| | | | | | | | | | | | | | | | | Linux commit 7ea6216049ff9cf250a6722cd766d99c8d1424e5 "mm/sparsemem: prepare for sub-section ranges" added validation of memory region sizes for hotplugging. This requires alignment of the region to 2MiB, which can be done with a MemoryMappingArena that will pad the end of the region with read-only pages. BUG=chromium:1031408 TEST=crostini.Sanity.artifact with 5.4 guest kernel Change-Id: I526f23a5ef32edd3268cd23f010e2bc20f9c305a Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1979257 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>
* devices: virtio: wl: Support multiple socketsRyo Hashimoto2020-01-07
| | | | | | | | | | | | | | | | | | | 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>
* disk: move qcow into disk crateDaniel Verkamp2020-01-06
| | | | | | | | | | | | | | | | | | | Move qcow from being its own crate into a module of the disk crate, similar to the composite disk module. This will allow use of qcow from disk and vice versa without introducing a circular crate dependency. BUG=None TEST=./build_test.py TEST=USE='asan fuzzer' emerge-nami crosvm Change-Id: I77512bbe4b94faff1b5460f9796ee56505135580 Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1972477 Reviewed-by: Cody Schuffelen <schuffelen@google.com> Reviewed-by: Zach Reizner <zachr@chromium.org> Tested-by: kokoro <noreply+kokoro@google.com> Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
* plugin: mount read-only /proc into the jailDmitry Torokhov2019-12-31
| | | | | | | | | | | | | | | | | | Breakpad requires /proc, so let's mount it (as read only) into plugin jail. We can't initiate it from concierge as concierge can only pass bind mounts to crosvm, whereas here we need to mount or own copy of /proc that reflects out own namespace(s). BUG=None TEST=Run Plugin VM, check /proc of the plugin process. Cq-Depend: 1982695 Change-Id: I5d673e3c8733a61afc5986f9bf110b05c45bbdce Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1982820 Reviewed-by: Chirantan Ekbote <chirantan@chromium.org> Tested-by: kokoro <noreply+kokoro@google.com> Tested-by: Dmitry Torokhov <dtor@chromium.org> Commit-Queue: Dmitry Torokhov <dtor@chromium.org>
* Cast refernces of libc::RLIMIT_NOFILE to i32 for newer libc crate.Allen Webb2019-12-26
| | | | | | | | | | | | BUG=chromium:1021609 TEST=emerge-sand crosvm Change-Id: I9658d3816cc846be2b6320f2f03036375de67cbe Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1979532 Tested-by: Allen Webb <allenwebb@google.com> Tested-by: kokoro <noreply+kokoro@google.com> Reviewed-by: Chirantan Ekbote <chirantan@chromium.org> Commit-Queue: Allen Webb <allenwebb@google.com>
* Fix parsing arguments that require a value at the end of command lineDmitry Torokhov2019-12-22
| | | | | | | | | | | | | | | | | | | | | | | We may be in state of waiting for the value for the parameter, and run out of the parameters. In this case we should try to parse the parameter as if it does not have a value and see if that succeeds. This makes sure that crosvm run ... --plugin-mount fails with error that --plugin-mount option needs a value instead of succeeding. BUG=None TEST=cargo test Change-Id: I9f3f1f3c7e6e2ca88efed1eeea5a52dd4aae70ef Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1975097 Reviewed-by: Dmitry Torokhov <dtor@chromium.org> Tested-by: Dmitry Torokhov <dtor@chromium.org> Tested-by: kokoro <noreply+kokoro@google.com> Commit-Queue: Dmitry Torokhov <dtor@chromium.org>
* When parsing --parameter <value> argument do not clobber parser errorsDmitry Torokhov2019-12-22
| | | | | | | | | | | | | | | | | | | | | | When parsing arguments in form of "--parameter <value>", because it is yet unclear whether the parameter needs a value or not, we first try to parse it together with the next argument, and if that fails, try to parse it standalone. If both fail, then we should take the error from the first one, as it is more informative. This way, doing crosvm run --plugin ... --plugin-mount bin --plugin-mount lib will fail with error that "bin is not absolute path" and not that value is missing for "--plugin-mount". BUG=None TEST=cargo test Change-Id: Icc3bb0fcab4fb90e38eff3d116679091b598120d Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1975096 Commit-Queue: Dmitry Torokhov <dtor@chromium.org> Reviewed-by: Dylan Reid <dgreid@chromium.org> Tested-by: kokoro <noreply+kokoro@google.com>
* crosvm: fix removing comments when processing -file optionsDmitry Torokhov2019-12-22
| | | | | | | | | | | | | | | | | As it turns out trim_end_matching() does not do what I expected (namely trim the string when it encounters matching substring) but rather expects to remove exactly matching suffix, so it actually fails to remove comments. Let's switch to using splitn() that does what I want. BUG=b:144454617 TEST=Run Plugin VM Change-Id: Ibe8cd14d9d3c14ff99dd9abfa8f286e452d96913 Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1975095 Tested-by: kokoro <noreply+kokoro@google.com> Tested-by: Dmitry Torokhov <dtor@chromium.org> Commit-Queue: Dmitry Torokhov <dtor@chromium.org> Reviewed-by: Dylan Reid <dgreid@chromium.org>
* crosvm: treat FailEntry as fatal to a vcpuStephen Barber2019-12-20
| | | | | | | | | | | | | | | | | FailEntry indicates an arch-specific failure to enter a VM. Treat this as fatal to the vcpu. Pass the u64 hardware failure reason from the kvm_run struct up to the client. BUG=chromium:1036009 TEST=crosvm on hatch nested VM dies immediately instead of infinite looping Change-Id: Iecb279b5b08ae1edc085717dce65e3ca46cbd30e Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1977221 Tested-by: kokoro <noreply+kokoro@google.com> Tested-by: Stephen Barber <smbarber@chromium.org> Reviewed-by: Stephen Barber <smbarber@chromium.org> Commit-Queue: Stephen Barber <smbarber@chromium.org>
* crosvm: allow shorthand notation in plugin-mount and plugin-gid-mapDmitry Torokhov2019-12-18
| | | | | | | | | | | | | | | | | | | | | | | | Let's allow shorthand notation in the form of: --plugin-mount=<src>[:[<dst>][:<writable>]] --plugin-gid-map=<inner>[:[<outer>][:<count>]] so that we can invoke crosvm as crosvm ... --plugin-mount=/bin --plugin-mount=/dev/log::true \ --plugin-gid-map=123 --plugin-gid-map=567::5 as repeating the data for both src and destination muddies the waters and is prone to errors. BUG=None TEST=cargo test Change-Id: I6f0a075ea3b27d4ec3dcf88698069930c158e759 Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1967786 Reviewed-by: Zach Reizner <zachr@chromium.org> Tested-by: kokoro <noreply+kokoro@google.com> Commit-Queue: Dmitry Torokhov <dtor@chromium.org>
* crosvm: plugin-mount-file and plugin-gid-map-file optionsDmitry Torokhov2019-12-17
| | | | | | | | | | | | | | List of bind-mounts and gid maps can be quite long, so let's allow listing them in text files, when convenient. BUG=b:144454617 TEST=Run Plugin VM Change-Id: I1218dab5a7e87b9f1ba44de6828da890fddb99fe Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1967785 Reviewed-by: Zach Reizner <zachr@chromium.org> Tested-by: kokoro <noreply+kokoro@google.com> Commit-Queue: Dmitry Torokhov <dtor@chromium.org>
* crosvm: factor our parsing bind-mount and gid map optionsDmitry Torokhov2019-12-17
| | | | | | | | | | | | | | In order to support specifying list of bind-mounts and gid maps options via files, factor out the parsing code. BUG=b:144454617 TEST=Run Plugin VM Change-Id: I772116cd77827ef722559e9fce450235fdbf6478 Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1967784 Tested-by: kokoro <noreply+kokoro@google.com> Reviewed-by: Zach Reizner <zachr@chromium.org> Commit-Queue: Dmitry Torokhov <dtor@chromium.org>
* Makes gpu renderer flags configurable via command lineJason Macnak2019-12-11
| | | | | | | | | | | | BUG=b:134086390 TEST=built crosvm and booted cuttlefish locally with gpu Change-Id: I4d816ddb52a2eadd06088d204d95118289a3f587 Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1927873 Reviewed-by: Zach Reizner <zachr@chromium.org> Tested-by: Jason Macnak <natsu@google.com> Tested-by: kokoro <noreply+kokoro@google.com> Commit-Queue: Jason Macnak <natsu@google.com>
* utilize EventDevices to make virtio-input devicesZach Reizner2019-12-11
| | | | | | | | | | | | BUG=chromium:1023975 TEST=./build_test Change-Id: I10267e535d4d1dae90b2b5f30db046c388791a16 Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1930409 Reviewed-by: Zach Reizner <zachr@chromium.org> Commit-Queue: Zach Reizner <zachr@chromium.org> Tested-by: Zach Reizner <zachr@chromium.org> Tested-by: kokoro <noreply+kokoro@google.com>
* devices: block: add block_size option for disksDaniel Verkamp2019-12-10
| | | | | | | | | | | | | | | 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>
* Remove render node forward codeLepton Wu2019-12-10
| | | | | | | | | | | | | We don't use it for ARCVM now. BUG=None TEST=`emerge-eve crosvm` Change-Id: I14343a2eb24c24a0216950b7c4eeb56ac482973e Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1958569 Reviewed-by: Zach Reizner <zachr@chromium.org> Reviewed-by: Dylan Reid <dgreid@chromium.org> Tested-by: kokoro <noreply+kokoro@google.com> Commit-Queue: Lepton Wu <lepton@chromium.org> Auto-Submit: Lepton Wu <lepton@chromium.org>
* Add runnable vcpuDylan Reid2019-12-10
| | | | | | | | | | | | | Add a new type `RunnableVcpu` for a vcpu that is bound to a thread. This adds type safety to ensure that vcpus are only ever run on one thread because RunnableVcpu can't `Send`. It also ensures multiple vcpus can't run on the same thread. Change-Id: Ia50dc127bc7a4ea4ce3ca99ef1062edbcaa912d0 Signed-off-by: Dylan Reid <dgreid@chromium.org> Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1898909 Tested-by: kokoro <noreply+kokoro@google.com> Reviewed-by: Zach Reizner <zachr@chromium.org>
* Support configurable screen sizesJason Macnak2019-12-10
| | | | | | | | | | | | | | | | | | 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>
* linux.rs: Don't use /proc/sys/fs/file-maxChirantan Ekbote2019-12-09
| | | | | | | | | | | | | | | | | | | | | | | | | | Don't use /proc/sys/fs/file-max when setting the max open file limit for the virtio-fs device. This will fail when the value is larger than the hard limit set for the crosvm process, unless it also has CAP_SYS_ADMIN in the initial namespace. Instead, just use the hard limit as returned by `prlimit64`. Increasing the soft limit up to the hard limit is allowed even for completely unprivileged processes. It is up to the process that spawned crosvm to ensure that the hard limit is high enough that the virtio-fs server will not run out of fds. BUG=b:142344095 TEST=Start a termina VM with a virtio-fs device after applying CL:1939193 Change-Id: I4fb4c33ffe6378ed3109fddcb0fc2bf3da850252 Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1957767 Tested-by: Chirantan Ekbote <chirantan@chromium.org> Tested-by: kokoro <noreply+kokoro@google.com> Auto-Submit: Chirantan Ekbote <chirantan@chromium.org> Commit-Queue: Stephen Barber <smbarber@chromium.org> Reviewed-by: Daniel Verkamp <dverkamp@chromium.org> Reviewed-by: Stephen Barber <smbarber@chromium.org>
* Adds a flag_or_value command line argument optionJason Macnak2019-12-09
| | | | | | | | | | | | | | | | | | | | | | ... to allow for specifying either '--gpu' to get default params or '--gpu=width=800,height=600,egl,surfaceless' to get custom params. Note: this introduces some potential ambiguity if the flag_or_value argument is the last arg specified before a positional argument. For example, 'crosvm run --gpu something' could either be parsed as a single value ('--gpu=something') or a flag and a positional argument ('--gpu' and 'something'). This change currently expects that the functor passed to set_arguments() will return an error on '--gpu=something' in order to resolve the disambiguity. BUG=b:134086390 TEST=built crosvm and booted cuttlefish locally Change-Id: Ib9be604af82583a39d7bd0a403e2168ef7350389 Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1929603 Tested-by: kokoro <noreply+kokoro@google.com> Commit-Queue: Jason Macnak <natsu@google.com> Reviewed-by: Zach Reizner <zachr@chromium.org>
* linux: check memory size calculation overflowDaniel Verkamp2019-12-06
| | | | | | | | | | | | | | | | | | | | | | | | | | On systems where usize is 32 bits wide (e.g. 32-bit arm), the calculation of memory_size in bytes based on the -m configuration option in megabytes would silently overflow when the resulting value was wider than 32 bits. Change the shift that converts megabytes to bytes into a checked_mul so that a suitable error is produced if the size overflows. Additionally, change the cfg.memory type to u64 instead of usize; this is representing a size in megabytes, so its maximum value isn't related to the size of an object in memory anyway, and this avoids the need for a cast in the memory_size calculation. Requesting a memory size larger than the crosvm process can map will still result in an error at a later stage in guest startup. BUG=chromium:1028747 TEST=`crosvm run -m $((5 * 1024)) ...` on kevin Change-Id: I8fef7070bab4dafff70ed54738b26d0bb7632150 Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1937551 Reviewed-by: Stephen Barber <smbarber@chromium.org> Tested-by: kokoro <noreply+kokoro@google.com> Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
* vfio: Implement bar mappableXiong Zhang2019-12-06
| | | | | | | | | | | | | | | | | | 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>
* devices: virtio: enable MSI-X for all devicesDaniel Verkamp2019-12-06
| | | | | | | | | | | | | | | | | | 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>
* linux.rs: Set open file limits for all devicesChirantan Ekbote2019-11-28
| | | | | | | | | | | | | | | | | | | | | | | | | The virtio-fs device opens a lot of fds and needs to have a pretty high open file limit. In order for this to be successful on Chrome OS, the parent process (concierge) also needs to have a very high open file limit. This unfortunately has the side-effect of raising the open file limit for _all_ crosvm devices. Deal with this by setting the open file limit back down to a reasonable value (1024) for all devices that use `create_base_minijail`. This was the value of the open file limit for concierge before the increase. BUG=b:142344095 TEST=Check /proc/<pid>/limits for various crosvm device processes and see that they have the proper value for "Max open files". Change-Id: I87e3fe62fe22e68bff5ba5b60d85d39060a111dd Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1939527 Tested-by: Chirantan Ekbote <chirantan@chromium.org> Tested-by: kokoro <noreply+kokoro@google.com> Auto-Submit: Chirantan Ekbote <chirantan@chromium.org> Commit-Queue: Nicolas Boichat <drinkcat@chromium.org> Reviewed-by: Daniel Verkamp <dverkamp@chromium.org> Reviewed-by: Stephen Barber <smbarber@chromium.org>
* linux.rs: Remove references to chronosChirantan Ekbote2019-11-26
| | | | | | | | | | | | | | | | | | | | | | | Even when run on Chrome OS, the crosvm process does not have enough privilege to add the chronos user/group to the {u,g}idmap of the 9p device process. This was never cleaned up because we don't use the 9p device in crostini VMs (seneschal spawns 9s servers in a separate process tree). Remove all references to the chronos user/group and just do what the other devices do: use the crosvm user/group if it exists or fall back to the current euid/egid. BUG=chromium:1028442 TEST=Add `--shared-dir` to the command line flags of a termina VM and see that it starts properly Change-Id: Iad4927d37c35709aee6e15f79b316eb88483458f Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1935581 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>
* io_jail: Replace rlim_t with rlim64_tChirantan Ekbote2019-11-22
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | rlim_t is defined as an unsigned long but importantly, it is defined as what the _kernel_ thinks is an unsigned long. This means that when you have a 32-bit userspace and a 64-bit kernel (like we do for arm64 chromebooks), rlim_t is 64 bits. This isn't really a problem for C and C++ code because they use the headers from the kernel where rlim_t is properly sized but it doesn't really work for rust. The libc crate defines rlim_t as an alias for ::std::os::raw::c_ulong, which leads to the rust compiler thinking that it has a 32 bit width. Hilarity ensues when you attempt to cross the rust -> C FFI barrier with these conflicting definitions. The rust compiler thinks the parameters can fit in 32 bit registers so it puts the `cur` parameter in r2 and the `max` parameter in r3. On the other hand, the C code knows that the parameters are 64-bit values and combines r2/r3 to create the 64-bit `cur` value and uses the first 8 bytes on the stack as the `max` value. This leads to a `cur` value that is way too large and a nonsensical `max` value that depends on whatever happened to be on the stack at the time. Fix this by changing the library bindings to u64 and the Minijail::set_rlimit parameters to rlim64_t. Once we add a method to minijail that accepts rlim64_t's we can switch the library bindings to use that as well. BUG=b:136128319 TEST=`tast run vm.Virtiofs` on kevin Change-Id: I8f58923c4768ecfe827d2a5d73c72dc778fe419c Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1916560 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>
* Enable seccomp for virtio-fsChirantan Ekbote2019-11-19
| | | | | | | | | | | | | | Add x86_64 and arm seccomp policy files for the virtio-fs device. BUG=b:136128319 TEST=Run a vm with a sandboxed virtio-fs device Change-Id: I8ea7c5d3e90696077f0c2b1f942cb286a0b36cf4 Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1916559 Reviewed-by: Dylan Reid <dgreid@chromium.org> Reviewed-by: Stephen Barber <smbarber@chromium.org> Tested-by: kokoro <noreply+kokoro@google.com> Commit-Queue: Chirantan Ekbote <chirantan@chromium.org>
* Add fs device to --shared-dirChirantan Ekbote2019-11-19
| | | | | | | | | | | | | | Expand the `--shared-dir` option to allow callers to select between 9p and virtio-fs for sharing directories. BUG=b:136128319 TEST=start a VM with a virtio-fs based shared directory Change-Id: Ie8afc1965b693805dd6000f0157786317aab060d Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1705656 Reviewed-by: Daniel Verkamp <dverkamp@chromium.org> Tested-by: kokoro <noreply+kokoro@google.com> Commit-Queue: Chirantan Ekbote <chirantan@chromium.org>
* devices: block: add option to control sparsenessDaniel Verkamp2019-11-18
| | | | | | | | | | | | | | | | | | 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>
* main: remove EPOLLHUP epoll item from host kernel synchronouslyZide Chen2019-11-17
| | | | | | | | | | | | | | | | | | | control_sockets.swap_remove() could cause host kernel to invoke ep_remove() to remove the epoll item. But it's called from the task work, and it could be deferred after next poll_ctx.wait() which could unexpectedly pick up epoll events from the already closed fd. BUG=chromium:1019986 TEST=launch Crosvm guest from heavy loaded Linux host Change-Id: I474a7a47a484e3acfae4383d61601e1553bd674f Signed-off-by: Zide Chen <zide.chen@intel.com> Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1917495 Reviewed-by: Zach Reizner <zachr@chromium.org> Reviewed-by: Daniel Verkamp <dverkamp@chromium.org> Tested-by: kokoro <noreply+kokoro@google.com>
* resource: Rename device memory to mmioXiong Zhang2019-11-10
| | | | | | | | | | | | | | 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>
* Resource: Unify mmio allocatorXiong Zhang2019-11-10
| | | | | | | | | | | | | | | | | | | | | | | | | | 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>
* Remove unnecessary static str lifetimesDaniel Verkamp2019-11-08
| | | | | | | | | | | | | | | | | | Const and static variables imply a static lifetime, so declaring it isn't necessary. Fixes clippy warnings of the form: Statics have by default a `'static` lifetime BUG=None TEST=bin/clippy Change-Id: I99e63b435b6b2fd5fbce9ef617e89ec114587ad7 Signed-off-by: Daniel Verkamp <dverkamp@chromium.org> Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1896087 Reviewed-by: Zach Reizner <zachr@chromium.org> Tested-by: kokoro <noreply+kokoro@google.com>
* crosvm: add plugin API for async writesMatt Delco2019-11-06
| | | | | | | | | | | | | | | | | | | | | | | A plugin might care to be immediately notified when a write is made to a port, but it doesn't care to have the VM stopped while the plugin calls back to resume the VM. Unfortunately this means that multiple messages can be queued up in the pipe and read() together by the plugin API. Protobuf's parsing function doesn't report how many bytes it read, so I've resorted to having crosvm prefix every message with a length and then have the plugin lib parse this number. Impact on performance has not been measured. BUG=b:143294496 TEST=Local build and run of build_test. Verified that new unit test was executed, exercised the case where multiple msgs are received together, and completed successfully. Change-Id: If6ef463e7b4d2e688e649f832a764fa644bf2d36 Signed-off-by: Matt Delco <delco@chromium.org> Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1896376 Tested-by: kokoro <noreply+kokoro@google.com> Reviewed-by: Stephen Barber <smbarber@chromium.org>
* crosvm: pre-cache answers to plugin get callsMatt Delco2019-10-31
| | | | | | | | | | | | | | | | This change tries to improve the performance of a plugin-based VM by adding a hint API that allows crosvm to proactively push cpu state to the plugin when certain ports for hypercalls are accessed by the VM. BUG=None TEST=build and run. See performance increase significantly. Change-Id: I71af24ebc034095ffea42eedb9ffda0afc719cd6 Signed-off-by: Matt Delco <delco@chromium.org> Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1873005 Tested-by: kokoro <noreply+kokoro@google.com> Reviewed-by: Zach Reizner <zachr@chromium.org>
* crosvm: Linux VM use immediate exitMatt Delco2019-10-30
| | | | | | | | | | | | Use immediate exit when avaialable to improve performance. BUG=None TEST=Local build and test. Change-Id: I5a4fb3dc310f3b2969113f6d57290099f84a0187 Signed-off-by: Matt Delco <delco@chromium.org> Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1847861 Reviewed-by: Zach Reizner <zachr@chromium.org>
* plugin: plugin VM use immediate exitMatt Delco2019-10-30
| | | | | | | | | | | | | If the immediate exit feature is available then we use use it to improve performance. BUG=None TEST=Local build and test. Change-Id: I368ae7a711de72955777cd434450789e9a10616c Signed-off-by: Matt Delco <delco@chromium.org> Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1847860 Reviewed-by: Zach Reizner <zachr@chromium.org>
* vfio: Add vfio msi routing information into kvmXiong Zhang2019-10-29
| | | | | | | | | | | | | | | | 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>
* devices: implement MSI control socketXiong Zhang2019-10-24
| | | | | | | | | | | | | | | | | 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>
* plugin: only pause on EINTRMatt Delco2019-10-23
| | | | | | | | | | | | | | | | | | | | | | | | | In the case of 1) an IO exit & callout to plugin, then 2) a pause request by another thread, the vcpu thread will eagerly check for a pause request and might cause another callout to the plugin for the pause. We haven't yet run KVM again for it to emulate the completion of the IO. It's probably less risky to call back into KVM again and let it finish the emulation before we callout to the plugin to make other potential state changes to the VM. This change also reduces the overhead of the non-pause case by not checking for a pause request on each VM exit. The tradeoff is that a pause request might take longer, but these are (or should be) relatively rare so it's better to slow these down (and be more conserative/sane about the state of the VM when pause is reported) in favor of making the non-pause VM exits faster. BUG=None TEST=Local build and run of "build_test". Change-Id: I38609eccd9a2196835f99de5ea84a586928fab30 Signed-off-by: Matt Delco <delco@chromium.org> Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1863725 Reviewed-by: Zach Reizner <zachr@chromium.org>
* plugin: rename signal callMatt Delco2019-10-23
| | | | | | | | | | | | | | There was a rename of the signal handler function for rt signals. This renames the plugin use to be consistent with the non-plugin case. BUG=None. TEST=Local compile. Change-Id: I5bf2e19754183022dbfe31fd1fdeceb83e07c18a Signed-off-by: Matt Delco <delco@chromium.org> Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1846680 Reviewed-by: Zach Reizner <zachr@chromium.org>
* crosvm: defer IPC on set callsMatt Delco2019-10-23
| | | | | | | | | | | | | | If a plugin makes a set call on vcpu registers then we can improve performance by deferring the IPC and instead conbining the request with the next resume call. BUG=None TEST=build and run. Change-Id: I4eb54a3f6eb30c98971aa2f099e3ea5899767eed Signed-off-by: Matt Delco <delco@chromium.org> Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1825262 Reviewed-by: Zach Reizner <zachr@chromium.org>
* crosvm: delete resume response from protobufMatt Delco2019-10-23
| | | | | | | | | | | | | | | The resume response was deleted in a earlier change. This change removes the message type from the proto file and adjusts crosvm so it can still build after this removal. BUG=None TEST=build and run. Change-Id: I27d36a51b1e7eb59258d23da38199b86d7cb3659 Signed-off-by: Matt Delco <delco@chromium.org> Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1825260 Reviewed-by: Zach Reizner <zachr@chromium.org> Reviewed-by: Dylan Reid <dgreid@chromium.org>
* devices: vfio: fix clippy warningsDaniel Verkamp2019-10-17
| | | | | | | | | | | | | | | | | | | | | 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>
generated by cgit-pink 1.4.1 (git 2.36.1) at 2024-05-28 14:53:10 +0000