feat(fs/mem): tarfs implementation [currently broken]

This neds to be refactored such that it does traversal of tar files by itself.

Author: fogti

Cargo.lock

@@ -142,6 +142,7 @@ time = { version = "0.3", default-features = false }
 volatile = "0.6"
 zerocopy = { version = "0.8", default-features = false }
 uhyve-interface = "0.1.4"
+tar-no-std = "0.4.2"
 
 [dependencies.smoltcp]
 version = "0.12"

Cargo.toml

@@ -12,6 +12,7 @@ use core::mem::{MaybeUninit, offset_of};
 use align_address::Align;
 use async_lock::{Mutex, RwLock};
 use async_trait::async_trait;
+use hermit_entry::ThinTree;
 
 use crate::errno::Errno;
 use crate::executor::block_on;
@@ -302,6 +303,10 @@ impl RomFile {
 	pub fn new(data: &'static [u8], mode: AccessPermission) -> Self {
 		let microseconds = arch::kernel::systemtime::now_micros();
 		let t = timespec::from_usec(microseconds as i64);
+		Self::new_with_timestamp(data, mode, t)
+	}
+
+	pub fn new_with_timestamp(data: &'static [u8], mode: AccessPermission, t: timespec) -> Self {
 		let attr = FileAttr {
 			st_size: data.len().try_into().unwrap(),
 			st_mode: mode | AccessPermission::S_IFREG,
@@ -372,19 +377,16 @@ impl RamFile {
 	}
 }
 
+type MemDirectoryMap = BTreeMap<String, Box<dyn VfsNode + core::marker::Send + core::marker::Sync>>;
+
 pub struct MemDirectoryInterface {
 	/// Directory entries
-	inner:
-		Arc<RwLock<BTreeMap<String, Box<dyn VfsNode + core::marker::Send + core::marker::Sync>>>>,
+	inner: Arc<RwLock<MemDirectoryMap>>,
 	read_idx: Mutex<usize>,
 }
 
 impl MemDirectoryInterface {
-	pub fn new(
-		inner: Arc<
-			RwLock<BTreeMap<String, Box<dyn VfsNode + core::marker::Send + core::marker::Sync>>>,
-		>,
-	) -> Self {
+	pub fn new(inner: Arc<RwLock<MemDirectoryMap>>) -> Self {
 		Self {
 			inner,
 			read_idx: Mutex::new(0),
@@ -455,11 +457,47 @@ impl ObjectInterface for MemDirectoryInterface {
 
 #[derive(Debug)]
 pub(crate) struct MemDirectory {
-	inner:
-		Arc<RwLock<BTreeMap<String, Box<dyn VfsNode + core::marker::Send + core::marker::Sync>>>>,
+	inner: Arc<RwLock<MemDirectoryMap>>,
 	attr: FileAttr,
 }
 
+fn thin_tree_to_vfs_node(
+	thin_tree: ThinTree<'static>,
+	t: timespec,
+) -> io::Result<Box<dyn VfsNode + Send + Sync>> {
+	Ok(match thin_tree {
+		ThinTree::File { content, metadata } => {
+			let mut st_mode = AccessPermission::S_IRUSR;
+			if metadata.is_exec {
+				st_mode |= AccessPermission::S_IXUSR;
+			}
+			Box::new(RomFile::new_with_timestamp(content, st_mode, t))
+				as Box<dyn VfsNode + Send + Sync>
+		}
+		ThinTree::Directory(d) => Box::new(MemDirectory {
+			inner: Arc::new(RwLock::new(
+				d.into_iter()
+					.map(|(key, value)| {
+						Ok((
+							core::str::from_utf8(key)
+								.expect("unable to interpret Hermit image entry filename as string")
+								.to_owned(),
+							thin_tree_to_vfs_node(value, t)?,
+						))
+					})
+					.collect::<io::Result<_>>()?,
+			)),
+			attr: FileAttr {
+				st_mode: AccessPermission::S_IRUSR | AccessPermission::S_IFDIR,
+				st_atim: t,
+				st_mtim: t,
+				st_ctim: t,
+				..Default::default()
+			},
+		}) as Box<dyn VfsNode + Send + Sync>,
+	})
+}
+
 impl MemDirectory {
 	pub fn new(mode: AccessPermission) -> Self {
 		let microseconds = arch::kernel::systemtime::now_micros();
@@ -477,6 +515,52 @@ impl MemDirectory {
 		}
 	}
 
+	/// Create a read-only memory tree from a tar image
+	///
+	/// This ignores top-level files in the image
+	pub fn try_from_image(image: &'static hermit_entry::tar_parser::Bytes) -> io::Result<Self> {
+		let microseconds = arch::kernel::systemtime::now_micros();
+		let t = timespec::from_usec(microseconds as i64);
+
+		// This is not perfectly memory-efficient, but we expect this
+		// to be invoked usually once per kernel boot, so this cost should
+		// be acceptable, given that it reduces code duplication and
+		// makes implementation way easier.
+		let thin_tree = ThinTree::try_from_image(image).map_err(|e| {
+			error!("unable to parse tar image: {e:?}");
+			Errno::Inval
+		})?;
+
+		let tree = match thin_tree {
+			ThinTree::Directory(d) => d
+				.into_iter()
+				.map(|(key, value)| {
+					Ok((
+						core::str::from_utf8(key)
+							.expect("unable to interpret Hermit image entry filename as string")
+							.to_owned(),
+						thin_tree_to_vfs_node(value, t)?,
+					))
+				})
+				.collect::<io::Result<_>>()?,
+			ThinTree::File { .. } => {
+				error!("root of image isn't a directory");
+				return Err(Errno::Inval);
+			}
+		};
+
+		Ok(Self {
+			inner: Arc::new(RwLock::new(tree)),
+			attr: FileAttr {
+				st_mode: AccessPermission::S_IRUSR | AccessPermission::S_IFDIR,
+				st_atim: t,
+				st_mtim: t,
+				st_ctim: t,
+				..Default::default()
+			},
+		})
+	}
+
 	async fn async_traverse_open(
 		&self,
 		components: &mut Vec<&str>,

src/fs/mem.rs

@@ -16,6 +16,7 @@ use hermit_sync::{InterruptSpinMutex, OnceCell};
 use mem::MemDirectory;
 use num_enum::{IntoPrimitive, TryFromPrimitive};
 
+use crate::env::fdt;
 use crate::errno::Errno;
 use crate::executor::block_on;
 use crate::fd::{AccessPermission, ObjectInterface, OpenOption, insert_object, remove_object};
@@ -340,18 +341,51 @@ pub(crate) fn init() {
 	const VERSION: &str = env!("CARGO_PKG_VERSION");
 	const UTC_BUILT_TIME: &str = build_time::build_time_utc!();
 
-	FILESYSTEM.set(Filesystem::new()).unwrap();
-	FILESYSTEM
-		.get()
-		.unwrap()
+	let mut root_filesystem = Filesystem::new();
+
+	// Handle optional Hermit Image specified in FDT.
+	let mut tar_image: Option<&'static [u8]> = None;
+	if let Some(fdt) = fdt() {
+		// per FDT specification, /chosen always exists.
+		let chosen = fdt.find_node("/chosen").unwrap();
+		if let Some(fdt::node::NodeProperty {
+			value: image_reg, ..
+		}) = chosen.property("image_reg")
+		{
+			let cell_sizes = fdt.root().cell_sizes();
+			let split_point = cell_sizes.address_cells * 4;
+			let end_point = split_point + cell_sizes.size_cells * 4;
+			if image_reg.len() == end_point {
+				let (addr, len) = image_reg.split_at(split_point);
+				if addr.len() == core::mem::size_of::<*const u8>()
+					&& len.len() == core::mem::size_of::<usize>()
+				{
+					// TODO: endian-ness?
+					let addr = usize::from_ne_bytes(addr.try_into().unwrap());
+					let len = usize::from_ne_bytes(len.try_into().unwrap());
+					// technically, the following is UB, because the kernel might be contained within...
+					tar_image =
+						Some(unsafe { core::slice::from_raw_parts(addr as *const u8, len) });
+				}
+			}
+		}
+	}
+	if let Some(tar_image) = tar_image {
+		root_filesystem.root =
+			MemDirectory::try_from_image(hermit_entry::tar_parser::Bytes::new(tar_image))
+				.expect("Unable to parse Hermit Image");
+	}
+
+	root_filesystem
 		.mkdir("/tmp", AccessPermission::from_bits(0o777).unwrap())
 		.expect("Unable to create /tmp");
-	FILESYSTEM
-		.get()
-		.unwrap()
+
+	root_filesystem
 		.mkdir("/proc", AccessPermission::from_bits(0o777).unwrap())
 		.expect("Unable to create /proc");
 
+	FILESYSTEM.set(root_filesystem).unwrap();
+
 	if let Ok(mut file) = File::create("/proc/version") {
 		if write!(file, "HermitOS version {VERSION} # UTC {UTC_BUILT_TIME}").is_err() {
 			error!("Unable to write in /proc/version");