Some block devices (ZFS volume) may require BLKDISCARD and BLKZEROOUT
ioctls for discard and write_zeroes operations respectively.
There is no good way to probe whether fallocate is supported on a block
device. Arguably, punch_hole and write_zeroes are rare. Instead of
having a complex scheme for the IO uring backend, we force it to always
use ioctls. The code can be changed if the synchronized ioctls become a
performance issue.
Changes:
- Detect block devices at construction time
- Use BLKDISCARD ioctl for punch_hole (discard) on block devices
- Use BLKZEROOUT ioctl for write_zeroes on block devices
- Add BLKDISCARD/BLKZEROOUT to VirtioBlock seccomp whitelist
- Keep fallocate() path for regular files (no behavior change)
- Consolidate some helper functions to the new sparse module
Signed-off-by: Wei Liu <liuwe@microsoft.com>
submit_batch_requests pushed each BatchRequest into the io_uring SQ in
turn and used `?` to bail on the first push failure.
Leaving the initial SQEs visible to the kernel — but submitter.submit()
was never called, and every other call site in this file gates submit()
behind a preceding sq.push() that now also fails on the full ring.
This could allow a guest to DoS it's own queue or worse if the buffer is
freed early.
Signed-off-by: Dylan Reid <dgreid@fb.com>
Keep the underlying io_uring submission queue push error in raw async
I/O paths instead of replacing it with a generic full-queue message.
Signed-off-by: Keith Adler <kadler@cloudflare.com>
Update open_raw to construct RawDisk instead of choosing between
RawFileDisk, RawFileDiskSync and RawFileDiskAio. The backend decision
is now made inside RawDisk::create_async_io.
Remove the DiskFile wrapper structs from raw_sync.rs, raw_async.rs
and raw_async_aio.rs. Only the AsyncIo worker structs remain in those
files. Reduce their module visibility to pub(crate).
Signed-off-by: Anatol Belski <anbelski@linux.microsoft.com>
The new_ prefix in Rust conventionally denotes constructors that return
Self (e.g. Vec::new(), File::new()). AsyncDiskFile::new_async_io does
not return Self. It is a factory method that constructs and returns a
Box<dyn AsyncIo> worker bound to the disk file descriptor and
metadata. The create_ prefix communicates this: the caller receives
a freshly constructed object of a different type.
This rename touches every format backend in block plus two external
callers in virtio-devices and performance-metrics. Every change is a
mechanical s/new_async_io/create_async_io/ substitution. No functional
change.
Ref: #7877 (task 3.2.8)
Signed-off-by: Anatol Belski <anbelski@linux.microsoft.com>
Change RawFileAsync::new() from std::io::Result to BlockResult,
aligning it with RawFileAsyncAio::new(). Each fallible call inside
the constructor now maps to BlockErrorKind::Io explicitly.
FixedVhdAsync::new() follows the same change since its only
fallible operation is constructing a RawFileAsync. The intermediate
DiskFileError::NewAsyncIo wrapping in both new_async_io() call
sites is no longer needed and is removed.
Signed-off-by: Anatol Belski <anbelski@linux.microsoft.com>
Block devices (LVM volumes, loop devices, RBD, etc.) cannot be resized
via ftruncate - they are resized externally. When vm.resize-disk is
called for a block device backend, verify the device size matches the
requested size instead of attempting ftruncate.
This enables the resize-disk API to work with block device backends by
validating the externally-resized device matches the expected size.
Signed-off-by: Vincent Thomas <vincent@v-thomas.com>
Update device_manager.rs to construct DiskBackend::Next
instead of DiskBackend::Legacy for the io_uring raw backend.
Signed-off-by: Muminul Islam <muislam@microsoft.com>
Add disk_file::AsyncDiskFile trait implementation for
RawFileDisk with try_clone() and new_async_io() methods.
try_clone() duplicates the underlying file descriptor and
wraps it in a new RawFileDisk. new_async_io() creates a
RawFileAsync (io_uring) backend, wrapping errors in
BlockError instead of DiskFileError.
Signed-off-by: Muminul Islam <muislam@microsoft.com>
Add empty disk_file::DiskFile impl for RawFileDisk. This
marker supertrait requires DiskSize + Geometry + Sync, all
of which are now satisfied.
Signed-off-by: Muminul Islam <muislam@microsoft.com>
Add disk_file::Resizable trait implementation for
RawFileDisk. Calls file.set_len(size) and wraps the I/O
error in BlockError on failure.
Signed-off-by: Muminul Islam <muislam@microsoft.com>
Add disk_file::SparseCapable trait implementation for
RawFileDisk. Delegates to probe_sparse_support() to detect
whether the underlying file supports hole-punching.
Signed-off-by: Muminul Islam <muislam@microsoft.com>
Add disk_file::Geometry trait implementation for RawFileDisk.
Probes disk topology from the file, falling back to defaults
on failure. Takes &self instead of &mut self and uses
unwrap_or_else for cleaner error handling.
Signed-off-by: Muminul Islam <muislam@microsoft.com>
Add disk_file::DiskFd trait implementation for RawFileDisk.
Delegates to file.as_raw_fd() via BorrowedDiskFd, taking
&self instead of &mut self.
Signed-off-by: Muminul Islam <muislam@microsoft.com>
Add disk_file::PhysicalSize trait implementation for
RawFileDisk. Returns metadata().len() wrapped in BlockError
on failure, consistent with the DiskSize impl.
Signed-off-by: Muminul Islam <muislam@microsoft.com>
Add disk_file::DiskSize trait implementation for RawFileDisk
using BlockError and BlockResult. Uses metadata().len()
instead of seek(SeekFrom::End(0)), taking &self instead of
&mut self.
Add BlockError, BlockErrorKind, BlockResult, and disk_file
imports needed by this and subsequent trait impls.
Signed-off-by: Muminul Islam <muislam@microsoft.com>
Add #[derive(Debug)] to RawFileDisk. This is required by the
new disk_file traits which have Send + Debug bounds.
Signed-off-by: Muminul Islam <muislam@microsoft.com>
Use query_device_size() instead of seek(End(0)) and metadata().len()
to correctly handle block device and regular file handles.
Signed-off-by: Anatol Belski <anbelski@linux.microsoft.com>
raw_sync, raw_async, and raw_async_aio each defined
FALLOC_FL_PUNCH_HOLE, FALLOC_FL_KEEP_SIZE, and FALLOC_FL_ZERO_RANGE as
local constants in their punch_hole() and write_zeroes()
implementations. These are available from the libc crate directly.
Signed-off-by: Anatol Belski <anbelski@linux.microsoft.com>
O_DIRECT requires buffer addresses to be aligned to the backend
device's logical block size. The existing bounce buffer logic in
execute_async() hardcodes SECTOR_SIZE (512) for the alignment check
and bounce buffer allocation. This is insufficient for devices with
a 4096-byte logical block size, where misaligned buffers cause
-EINVAL from the host kernel.
Add an alignment() method to the AsyncIo trait that returns the
backend's logical block size, defaulting to SECTOR_SIZE. The three
raw I/O backends (io_uring, AIO, synchronous) probe the device
topology via DiskTopology::probe() at creation time and return the
actual logical block size. All image format backends would simply
use the default value of 512 bytes since their underlying are
not block devices.
execute_async() now queries disk_image.alignment() instead of using
the hardcoded SECTOR_SIZE
Fixes: #7720
Signed-off-by: Saravanan D <saravanand@crusoe.ai>
Implement punch_hole() and write_zeroes() for raw file backends using
io_uring and fallocate.
punch_hole() uses FALLOC_FL_PUNCH_HOLE to deallocate storage.
write_zeroes() uses FALLOC_FL_ZERO_RANGE to write zeros efficiently.
Both use FALLOC_FL_KEEP_SIZE to maintain file size.
Signed-off-by: Anatol Belski <anbelski@linux.microsoft.com>
Add capability query to DiskFile trait to check backend
support for sparse operations (punch hole, write zeroes,
discard). Only advertise VIRTIO_BLK_F_DISCARD and
VIRTIO_BLK_F_WRITE_ZEROES when the backend supports these
operations.
Signed-off-by: Anatol Belski <anbelski@linux.microsoft.com>
Add punch_hole() and write_zeroes() methods to the AsyncIo trait
with stub implementations for all backends. These will be used to
support DISCARD and WRITE_ZEROES operations.
Signed-off-by: Anatol Belski <anbelski@linux.microsoft.com>
Implement the batch submission function for raw disk, default it is
enabled. After parsing the requests this method is
called for better IO latency and bandwidth.
Signed-off-by: Bo Chen <bchen@crusoe.ai>
Signed-off-by: Muminul Islam <muislam@microsoft.com>
This was caught by the nightly compiler during cargo fuzz build.
error: lifetime flowing from input to output with different syntax can be confusing
--> /home/runner/work/cloud-hypervisor/cloud-hypervisor/hypervisor/src/arch/x86/emulator/mod.rs:493:26
|
493 | pub fn new(platform: &mut dyn PlatformEmulator<CpuState = T>) -> Emulator<T> {
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ ----------- the lifetime gets resolved as `'_`
| |
| this lifetime flows to the output
|
= note: `-D mismatched-lifetime-syntaxes` implied by `-D warnings`
= help: to override `-D warnings` add `#[allow(mismatched_lifetime_syntaxes)]`
help: one option is to remove the lifetime for references and use the anonymous lifetime for paths
|
493 | pub fn new(platform: &mut dyn PlatformEmulator<CpuState = T>) -> Emulator<'_, T> {
Signed-off-by: Jinank Jain <jinankjain@microsoft.com>
The ASYNC flag forces requests to go to worker threads. Worker threads
are expensive. Let the kernel decide what to do.
With this change, I no longer see an excessive amount of io_uring worker
threads.
Quote from the manual for io_uring_sqe_set_flags(3):
```
IOSQE_ASYNC
Normal operation for io_uring is to try and issue an sqe
as non-blocking first, and if that fails, execute it in an
async manner. To support more efficient overlapped
operation of requests that the application knows/assumes
will always (or most of the time) block, the application
can ask for an sqe to be issued async from the start. Note
that this flag immediately causes the SQE to be offloaded
to an async helper thread with no initial non-blocking
attempt. This may be less efficient and should not be
used liberally or without understanding the performance
and efficiency tradeoffs.
```
Signed-off-by: Wei Liu <liuwe@microsoft.com>
Instead of silently ignoring the error, return an error to the callers.
This in practice should never happen, because the submission queue size
(ring depth) is the same as the virtio queue size. Virtio queue won't
push more requests than there are submission queue entries.
Signed-off-by: Wei Liu <liuwe@microsoft.com>
Historically the Cloud Hypervisor coding style has been to ensure that
all imports are ordered and placed in a single group. Unfortunately
cargo fmt has no support for ensuring that all imports are in a single
group so if whitespace lines were added as part of the import statements
then they would only be odered correctly in the group.
By adopting "group_imports="StdExternalCrate" we can enforce a style
where imports are placed in at most three groups for std, external
crates and the crate itself. Choosing a style enforceable by the tooling
reduces the reviewer burden.
Signed-off-by: Rob Bradford <rbradford@rivosinc.com>
It doens't make sense to limit this to the async_io module and import
across modules so place in default create module.
Signed-off-by: Rob Bradford <rbradford@rivosinc.com>
warning: this argument is a mutable reference, but not used mutably
--> block/src/async_io.rs:68:28
|
68 | fn query_block_size(f: &mut File, block_size_type: BlockSize) -> std::io::Result<u64> {
| ^^^^^^^^^ help: consider changing to: `&File`
|
= help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#needless_pass_by_ref_mut
Signed-off-by: Rob Bradford <rbradford@rivosinc.com>
This commit merges crates `qcow`, `vhdx` and `block_util` into the
crate `block`, which can allow `qcow` to use functions from `block_util`
without introducing a circular crate dependency.
This commit is based on crosvm implementation:
f2eecc4152
Signed-off-by: Yu Li <liyu.yukiteru@bytedance.com>
