Inflight I/O: Implement missing traits

This commit implements the get and set inflight fd members of the
VhostUserSlaveReqHandlerMut trait, which is used to pass inflight
I/O queue tracking regions as memfds.

Fixes rust-vmm#14.

Signed-off-by: Harshavardhan Unnibhavi <[email protected]>

src/handler.rs

@@ -6,24 +6,31 @@
 use std::error;
 use std::fs::File;
 use std::io;
-use std::os::unix::io::AsRawFd;
+use std::mem;
+use std::os::unix::io::RawFd;
+use std::os::unix::io::{AsRawFd, FromRawFd};
 use std::sync::Arc;
 use std::thread;
 
+use libc::c_void;
+
 use vhost::vhost_user::message::{
- VhostUserConfigFlags, VhostUserMemoryRegion, VhostUserProtocolFeatures,
+ DescStatePacked, DescStateSplit, QueueRegionPacked, QueueRegionSplit, VhostUserConfigFlags,
+ VhostUserInflight, VhostUserMemoryRegion, VhostUserProtocolFeatures,
  VhostUserSingleMemoryRegion, VhostUserVirtioFeatures, VhostUserVringAddrFlags,
  VhostUserVringState,
 };
 use vhost::vhost_user::{
  Error as VhostUserError, Result as VhostUserResult, SlaveFsCacheReq,
  VhostUserSlaveReqHandlerMut,
 };
+use virtio_bindings::bindings::virtio_net::VIRTIO_F_RING_PACKED;
 use virtio_bindings::bindings::virtio_ring::VIRTIO_RING_F_EVENT_IDX;
 use vm_memory::bitmap::Bitmap;
 use vm_memory::mmap::NewBitmap;
 use vm_memory::{
- FileOffset, GuestAddress, GuestAddressSpace, GuestMemoryMmap, GuestRegionMmap, MmapRegion,
+ Address, FileOffset, GuestAddress, GuestAddressSpace, GuestMemoryMmap, GuestRegionMmap,
+ MmapRegion,
 };
 use vmm_sys_util::epoll::EventSet;
 
@@ -71,6 +78,22 @@ struct AddrMapping {
  gpa_base: u64,
 }
 
+struct InflightFdState {
+ inflight_file: Option<File>,
+ inflight_mapping_addr: Option<GuestAddress>,
+ inflight_mmap_size: usize,
+}
+
+impl InflightFdState {
+ fn new() -> Self {
+ InflightFdState {
+ inflight_file: None,
+ inflight_mapping_addr: None,
+ inflight_mmap_size: 0,
+ }
+ }
+}
+
 pub struct VhostUserHandler<S, V, B>
 where
  S: VhostUserBackend<V, B>,
@@ -90,6 +113,7 @@ where
  atomic_mem: GM<B>,
  vrings: Vec<V>,
  worker_threads: Vec<thread::JoinHandle<VringEpollResult<()>>>,
+ inflight_state: InflightFdState,
 }
 
 impl<S, V, B> VhostUserHandler<S, V, B>
@@ -147,6 +171,7 @@ where
  atomic_mem,
  vrings,
  worker_threads,
+ inflight_state: InflightFdState::new(),
  })
  }
 }
@@ -564,20 +589,206 @@ where
 
  fn get_inflight_fd(
  &mut self,
- _inflight: &vhost::vhost_user::message::VhostUserInflight,
+ inflight: &vhost::vhost_user::message::VhostUserInflight,
  ) -> VhostUserResult<(vhost::vhost_user::message::VhostUserInflight, File)> {
- // Assume the backend hasn't negotiated the inflight feature; it
- // wouldn't be correct for the backend to do so, as we don't (yet)
- // provide a way for it to handle such requests.
- Err(VhostUserError::InvalidOperation)
+ let ret_val = -1;
+ // Total size of the inflight queue region
+ let mut total_mmap_size =
+ self.get_inflight_queue_size(inflight.queue_size) * inflight.num_queues as usize;
+
+ // Create a memfd region to hold the queues for inflight I/O tracking
+ let (mmap_ptr, inflight_fd, dup_inflight_fd) = self.memfd_alloc(total_mmap_size)?;
+
+ self.inflight_state.inflight_file = Some(unsafe { File::from_raw_fd(inflight_fd as i32) });
+
+ if mmap_ptr == ret_val as *mut c_void {
+ total_mmap_size = 0;
+ self.inflight_state.inflight_mapping_addr = None;
+ } else {
+ unsafe { libc::memset(mmap_ptr, 0, total_mmap_size) };
+ self.inflight_state.inflight_mapping_addr = Some(GuestAddress::new(mmap_ptr as u64));
+ }
+ self.inflight_state.inflight_mmap_size = total_mmap_size;
+
+ Ok((
+ VhostUserInflight {
+ mmap_size: total_mmap_size as u64,
+ mmap_offset: 0,
+ num_queues: inflight.num_queues,
+ queue_size: inflight.queue_size,
+ },
+ unsafe { File::from_raw_fd(dup_inflight_fd as RawFd) },
+ ))
  }
 
  fn set_inflight_fd(
  &mut self,
- _inflight: &vhost::vhost_user::message::VhostUserInflight,
- _file: File,
+ inflight: &vhost::vhost_user::message::VhostUserInflight,
+ file: File,
  ) -> VhostUserResult<()> {
- Err(VhostUserError::InvalidOperation)
+ let ret_val = -1;
+
+ // Need to unmap any previously mmaped regions as closing the
+ // associated file doesn't unmap it automatically
+ if let Some(inflight_addr) = self.inflight_state.inflight_mapping_addr {
+ unsafe {
+ libc::munmap(
+ inflight_addr.raw_value() as *mut c_void,
+ self.inflight_state.inflight_mmap_size,
+ )
+ };
+ }
+
+ let inflight_fd = file.as_raw_fd();
+
+ self.inflight_state.inflight_file = Some(file);
+ let mmap_size = inflight.mmap_size;
+ let mmap_offset = inflight.mmap_offset;
+
+ let mmap_ptr = unsafe {
+ libc::mmap(
+ std::ptr::null_mut::<c_void>(),
+ mmap_size as usize,
+ libc::PROT_READ | libc::PROT_WRITE,
+ libc::MAP_SHARED,
+ inflight_fd,
+ mmap_offset as i64,
+ )
+ };
+
+ if mmap_ptr == ret_val as *mut c_void {
+ self.inflight_state.inflight_mapping_addr = None;
+ self.inflight_state.inflight_mmap_size = 0;
+ } else {
+ self.inflight_state.inflight_mapping_addr = Some(GuestAddress::new(mmap_ptr as u64));
+ self.inflight_state.inflight_mmap_size = mmap_size as usize;
+ }
+
+ self.set_inflight_region_desc_num(inflight.num_queues, inflight.queue_size);
+
+ Ok(())
+ }
+}
+
+impl<S, V, B> VhostUserHandler<S, V, B>
+where
+ S: VhostUserBackend<V, B>,
+ V: VringT<GM<B>>,
+ B: NewBitmap + Clone,
+{
+ fn get_inflight_queue_size(&mut self, queue_size: u16) -> usize {
+ let queue_region_size;
+ let descr_state_size;
+ let virtio_features = self.get_features().unwrap();
+
+ if virtio_features & (1 << VIRTIO_F_RING_PACKED) == 0 {
+ // Use descriptor and queue states for split virtqueues
+ queue_region_size = mem::size_of::<QueueRegionSplit>();
+ descr_state_size = mem::size_of::<DescStateSplit>();
+ } else {
+ // Use descriptor and queue states for packed virtqueues
+ queue_region_size = mem::size_of::<QueueRegionPacked>();
+ descr_state_size = mem::size_of::<DescStatePacked>();
+ }
+ queue_region_size + descr_state_size * queue_size as usize
+ }
+
+ fn memfd_alloc(&self, mmap_size: usize) -> VhostUserResult<(*mut c_void, i64, i64)> {
+ let mut ret_val;
+ let inflight_file;
+ let dup_inflight_file;
+
+ ret_val = unsafe {
+ libc::syscall(
+ libc::SYS_memfd_create,
+ &std::ffi::CString::new("inflight-region").unwrap(),
+ libc::MFD_ALLOW_SEALING,
+ )
+ };
+
+ if ret_val == -1 {
+ return Err(VhostUserError::MemFdCreateError);
+ }
+
+ inflight_file = ret_val;
+
+ ret_val = unsafe { libc::ftruncate(inflight_file as RawFd, mmap_size as i64) } as i64;
+
+ if ret_val == -1 {
+ return Err(VhostUserError::FileTrucateError);
+ }
+
+ ret_val = unsafe {
+ libc::fcntl(
+ inflight_file as RawFd,
+ libc::F_ADD_SEALS,
+ libc::F_SEAL_GROW | libc::F_SEAL_SHRINK | libc::F_SEAL_SEAL,
+ )
+ } as i64;
+
+ if ret_val == -1 {
+ return Err(VhostUserError::MemFdSealError);
+ }
+
+ ret_val = unsafe { libc::dup(inflight_file as RawFd).into() };
+
+ if ret_val == -1 {
+ return Err(VhostUserError::MemFdCreateError);
+ }
+
+ dup_inflight_file = ret_val;
+
+ Ok((
+ unsafe {
+ libc::mmap(
+ std::ptr::null_mut(),
+ mmap_size,
+ libc::PROT_READ | libc::PROT_WRITE,
+ libc::MAP_SHARED,
+ inflight_file as RawFd,
+ 0,
+ )
+ },
+ inflight_file,
+ dup_inflight_file,
+ ))
+ }
+
+ fn set_desc_num_packed(&mut self, inflight_region: u64, num_queues: u16, queue_size: u16) {
+ let raw_ptr = inflight_region as *mut QueueRegionPacked;
+
+ for i in 0..num_queues {
+ unsafe {
+ let queue_region = raw_ptr.offset(i as isize).as_mut().unwrap();
+ queue_region.desc_num = queue_size;
+ }
+ }
+ }
+
+ fn set_desc_num_split(&mut self, inflight_region: u64, num_queues: u16, queue_size: u16) {
+ let raw_ptr = inflight_region as *mut QueueRegionSplit;
+
+ for i in 0..num_queues {
+ unsafe {
+ let queue_region = raw_ptr.offset(i as isize).as_mut().unwrap();
+ queue_region.desc_num = queue_size;
+ }
+ }
+ }
+
+ fn set_inflight_region_desc_num(&mut self, num_queues: u16, queue_size: u16) {
+ let inflight_region = self
+ .inflight_state
+ .inflight_mapping_addr
+ .unwrap()
+ .raw_value();
+
+ let virtio_features = self.get_features().unwrap();
+
+ match virtio_features & (1 << VIRTIO_F_RING_PACKED) {
+ 0 => self.set_desc_num_split(inflight_region, num_queues, queue_size),
+ _ => self.set_desc_num_packed(inflight_region, num_queues, queue_size),
+ };
  }
 }