github/kata-containers
1
0
Fork 1
mirror of https://github.com/kata-containers/kata-containers.git synced 2025-08-30 14:25:43 +00:00
Code Issues Projects Releases Wiki Activity

runtime-rs: Consolidate the handling of fds passed to QEMU

Browse Source 
File descriptors that are passed to QEMU need some special care.
We want them to be closed when the QEMU process is started. But
at the same time, it is required that the associated rust File
structures, either coming from the` std::fs` or the `tokio::fs`
crates, are still in scope when the QEMU process is forked. This
is currently achieved by keeping File structures in variables
at the outer scope of `start_vm()`. This scheme is currently
duplicated, with similar justifications in the corresponding
comments.

Consolidate all this handling in one place with a more generic
explanation.

Fixes #9281

Signed-off-by: Greg Kurz <groug@kaod.org>
This commit is contained in:
Greg Kurz 2024-03-12 18:01:05 +01:00
parent 6a112cc7a5
commit 1e526a4769
1 changed files with 8 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

17
src/runtime-rs/crates/hypervisor/src/qemu/inner.rs
View File

@ -66,13 +66,11 @@ impl QemuInner {
let mut cmdline = QemuCmdLine::new(&self.id, &self.config)?;
// If there's a Vsock Device in `self.devices` the vhost-vsock file
// descriptor needs to stay open until the qemu process launches.
// This is why we need to store it in a variable at this scope.
let mut _vhost_fd = None;
// We need to keep the vhost-net/tuntap file descriptor open until the QEMU process launches.
// However, we're likely not interested in the specific type of file descriptor itself. We just
// want to ensure any fds associated with network devices remain open within the current scope.
// CAUTION: File descriptors that are passed to QEMU must stay open until the QEMU process
// is started and closed afterwards. This is achieved by collecting them in _fds_for_qemu.
// It is mandatory for _fds_for_qemu to last until the QEMU process is forked. Leave it
// in the outer scope of this function for this to happen. The files in _fds_for_qemu
// should not be used in any way.
let mut _fds_for_qemu: Vec<std::fs::File> = Vec::new();
for device in &mut self.devices {
@ -89,7 +87,7 @@ impl QemuInner {
DeviceType::Vsock(vsock_dev) => {
let fd = vsock_dev.init_config().await?;
cmdline.add_vsock(fd.as_raw_fd(), vsock_dev.config.guest_cid)?;
_vhost_fd = Some(fd);
_fds_for_qemu.push(fd.into_std().await);
}
DeviceType::Block(block_dev) => {
if block_dev.config.path_on_host == self.config.boot_info.initrd {
@ -114,7 +112,8 @@ impl QemuInner {
// we need ensure add_network_device happens in netns.
let _netns_guard = NetnsGuard::new(&netns).context("new netns guard")?;
_fds_for_qemu = cmdline.add_network_device(&network.config, network_info)?;
_fds_for_qemu
.append(&mut cmdline.add_network_device(&network.config, network_info)?);
}
_ => info!(sl!(), "qemu cmdline: unsupported device: {:?}", device),
}