krata/hypha/src/image/mod.rs

322 lines
11 KiB
Rust
Raw Normal View History

2024-01-18 08:02:21 +00:00
pub mod cache;
2024-01-18 18:16:59 +00:00
pub mod fetch;
pub mod name;
2024-01-18 08:02:21 +00:00
2024-01-17 22:29:05 +00:00
use crate::error::{HyphaError, Result};
2024-01-18 08:02:21 +00:00
use crate::image::cache::ImageCache;
2024-01-18 18:16:59 +00:00
use crate::image::fetch::RegistryClient;
use crate::image::name::ImageName;
2024-01-17 22:29:05 +00:00
use backhand::{FilesystemWriter, NodeHeader};
2024-01-20 10:41:49 +00:00
use flate2::read::GzDecoder;
2024-01-17 22:29:05 +00:00
use log::{debug, trace};
2024-01-20 10:41:49 +00:00
use oci_spec::image::{Descriptor, ImageConfiguration, ImageManifest, MediaType};
2024-01-17 22:29:05 +00:00
use std::fs;
use std::fs::File;
2024-01-20 10:41:49 +00:00
use std::io::{copy, BufReader, Seek, SeekFrom, Write};
2024-01-17 22:29:05 +00:00
use std::os::unix::fs::{FileTypeExt, MetadataExt, PermissionsExt};
2024-01-20 10:41:49 +00:00
use std::path::{Path, PathBuf};
use tar::Entry;
2024-01-17 22:29:05 +00:00
use uuid::Uuid;
use walkdir::WalkDir;
pub const IMAGE_SQUASHFS_VERSION: u64 = 1;
2024-01-18 08:02:21 +00:00
pub struct ImageInfo {
2024-01-18 22:52:35 +00:00
pub image_squashfs: PathBuf,
2024-01-18 08:02:21 +00:00
pub manifest: ImageManifest,
2024-01-18 08:15:36 +00:00
pub config: ImageConfiguration,
2024-01-18 08:02:21 +00:00
}
impl ImageInfo {
2024-01-18 08:15:36 +00:00
fn new(
squashfs: PathBuf,
manifest: ImageManifest,
config: ImageConfiguration,
) -> Result<ImageInfo> {
Ok(ImageInfo {
2024-01-18 22:52:35 +00:00
image_squashfs: squashfs,
2024-01-18 08:15:36 +00:00
manifest,
config,
})
2024-01-18 08:02:21 +00:00
}
}
pub struct ImageCompiler<'a> {
cache: &'a ImageCache,
}
2024-01-17 22:29:05 +00:00
2024-01-18 08:02:21 +00:00
impl ImageCompiler<'_> {
pub fn new(cache: &ImageCache) -> Result<ImageCompiler> {
Ok(ImageCompiler { cache })
2024-01-17 22:29:05 +00:00
}
2024-01-18 08:02:21 +00:00
pub fn compile(&self, image: &ImageName) -> Result<ImageInfo> {
2024-01-17 22:29:05 +00:00
debug!("ImageCompiler compile image={image}");
let mut tmp_dir = std::env::temp_dir().clone();
tmp_dir.push(format!("hypha-compile-{}", Uuid::new_v4()));
2024-01-20 10:41:49 +00:00
2024-01-17 22:29:05 +00:00
let mut image_dir = tmp_dir.clone();
image_dir.push("image");
fs::create_dir_all(&image_dir)?;
2024-01-20 10:41:49 +00:00
let mut layer_dir = tmp_dir.clone();
layer_dir.push("layer");
fs::create_dir_all(&layer_dir)?;
2024-01-17 22:29:05 +00:00
let mut squash_file = tmp_dir.clone();
squash_file.push("image.squashfs");
2024-01-20 10:41:49 +00:00
let info = self.download_and_compile(image, &layer_dir, &image_dir, &squash_file)?;
fs::remove_dir_all(&tmp_dir)?;
2024-01-18 08:02:21 +00:00
Ok(info)
2024-01-17 22:29:05 +00:00
}
2024-01-18 08:02:21 +00:00
fn download_and_compile(
&self,
image: &ImageName,
2024-01-20 10:41:49 +00:00
layer_dir: &Path,
2024-01-18 08:02:21 +00:00
image_dir: &PathBuf,
squash_file: &PathBuf,
) -> Result<ImageInfo> {
2024-01-17 22:29:05 +00:00
debug!(
2024-01-20 10:41:49 +00:00
"ImageCompiler download manifest image={image}, image_dir={}",
2024-01-17 22:29:05 +00:00
image_dir.to_str().unwrap()
);
2024-01-18 18:16:59 +00:00
let mut client = RegistryClient::new(image.registry_url()?)?;
let manifest = client.get_manifest(&image.name, &image.reference)?;
2024-01-18 08:02:21 +00:00
let manifest_serialized = serde_json::to_string(&manifest)?;
let cache_key = format!(
"manifest\n{}squashfs-version\n{}\n",
manifest_serialized, IMAGE_SQUASHFS_VERSION
);
let cache_digest = sha256::digest(cache_key);
if let Some(cached) = self.cache.recall(&cache_digest)? {
2024-01-18 08:02:21 +00:00
return Ok(cached);
}
2024-01-18 08:15:36 +00:00
2024-01-20 10:41:49 +00:00
debug!(
"ImageCompiler download config digest={} size={}",
manifest.config().digest(),
manifest.config().size(),
);
2024-01-18 18:16:59 +00:00
let config_bytes = client.get_blob(&image.name, manifest.config())?;
2024-01-18 08:15:36 +00:00
let config: ImageConfiguration = serde_json::from_slice(&config_bytes)?;
2024-01-20 10:41:49 +00:00
let mut layers: Vec<PathBuf> = Vec::new();
2024-01-17 22:29:05 +00:00
for layer in manifest.layers() {
2024-01-20 10:41:49 +00:00
let layer_path = self.download_layer(image, layer, layer_dir, &mut client)?;
layers.push(layer_path);
}
for layer in layers {
let mut file = File::open(&layer)?;
self.process_whiteout_entries(&file, image_dir)?;
file.seek(SeekFrom::Start(0))?;
self.process_write_entries(&file, image_dir)?;
drop(file);
fs::remove_file(&layer)?;
}
self.squash(image_dir, squash_file)?;
let info = ImageInfo::new(squash_file.clone(), manifest.clone(), config)?;
self.cache.store(&cache_digest, &info)
}
fn process_whiteout_entries(&self, file: &File, image_dir: &PathBuf) -> Result<()> {
let mut archive = tar::Archive::new(file);
for entry in archive.entries()? {
let entry = entry?;
let dst = self.check_safe_entry(&entry, image_dir)?;
let Some(name) = dst.file_name() else {
return Err(HyphaError::new("unable to get file name"));
};
let Some(name) = name.to_str() else {
return Err(HyphaError::new("unable to get file name as string"));
};
if !name.starts_with(".wh.") {
continue;
}
let mut dst = dst.clone();
dst.pop();
let opaque = name == ".wh..wh..opq";
if !opaque {
dst.push(name);
self.check_safe_path(&dst, image_dir)?;
}
if opaque {
for entry in fs::read_dir(dst)? {
let entry = entry?;
let path = entry.path();
if path.is_file() {
fs::remove_file(&path)?;
} else {
fs::remove_dir_all(&path)?;
2024-01-17 22:29:05 +00:00
}
}
2024-01-20 10:41:49 +00:00
} else if dst.is_file() {
fs::remove_file(&dst)?;
} else {
fs::remove_dir(&dst)?;
}
}
Ok(())
}
fn process_write_entries(&self, file: &File, image_dir: &PathBuf) -> Result<()> {
let mut archive = tar::Archive::new(file);
for entry in archive.entries()? {
let mut entry = entry?;
let dst = self.check_safe_entry(&entry, image_dir)?;
let Some(name) = dst.file_name() else {
return Err(HyphaError::new("unable to get file name"));
};
let Some(name) = name.to_str() else {
return Err(HyphaError::new("unable to get file name as string"));
};
if name.starts_with(".wh.") {
continue;
}
entry.unpack(dst)?;
}
Ok(())
}
fn check_safe_entry(&self, entry: &Entry<&File>, image_dir: &PathBuf) -> Result<PathBuf> {
let mut dst = image_dir.clone();
dst.push(entry.path()?);
self.check_safe_path(&dst, image_dir)?;
Ok(dst)
}
fn check_safe_path(&self, dst: &PathBuf, image_dir: &PathBuf) -> Result<()> {
let resolved = path_clean::clean(dst);
if !resolved.starts_with(image_dir) {
return Err(HyphaError::new("layer attempts to work outside image dir"));
}
Ok(())
}
fn download_layer(
&self,
image: &ImageName,
layer: &Descriptor,
layer_dir: &Path,
client: &mut RegistryClient,
) -> Result<PathBuf> {
debug!(
"ImageCompiler download layer digest={} size={}",
layer.digest(),
layer.size()
);
let mut layer_path = layer_dir.to_path_buf();
layer_path.push(layer.digest());
let mut tmp_path = layer_dir.to_path_buf();
tmp_path.push(format!("{}.tmp", layer.digest()));
{
let mut file = File::create(&layer_path)?;
let size = client.write_blob(&image.name, layer, &mut file)?;
if layer.size() as u64 != size {
return Err(HyphaError::new(
"downloaded layer size differs from size in manifest",
));
2024-01-17 22:29:05 +00:00
}
}
2024-01-20 10:41:49 +00:00
let compressed = match layer.media_type() {
MediaType::ImageLayer => false,
MediaType::ImageLayerGzip => {
let reader = File::open(&layer_path)?;
let mut decoder = GzDecoder::new(&reader);
let mut writer = File::create(&tmp_path)?;
copy(&mut decoder, &mut writer)?;
writer.flush()?;
true
}
MediaType::ImageLayerZstd => {
let reader = File::open(&layer_path)?;
let mut decoder = zstd::Decoder::new(&reader)?;
let mut writer = File::create(&tmp_path)?;
copy(&mut decoder, &mut writer)?;
writer.flush()?;
true
}
_ => return Err(HyphaError::new("found layer with unknown media type")),
};
if compressed {
fs::rename(tmp_path, &layer_path)?;
}
Ok(layer_path)
2024-01-17 22:29:05 +00:00
}
2024-01-18 08:02:21 +00:00
fn squash(&self, image_dir: &PathBuf, squash_file: &PathBuf) -> Result<()> {
2024-01-17 22:29:05 +00:00
let mut writer = FilesystemWriter::default();
let walk = WalkDir::new(image_dir).follow_links(false);
for entry in walk {
let entry = entry?;
let rel = entry
.path()
.strip_prefix(image_dir)?
.to_str()
.ok_or_else(|| HyphaError::new("failed to strip prefix of tmpdir"))?;
let rel = format!("/{}", rel);
trace!("ImageCompiler squash write {}", rel);
let typ = entry.file_type();
let metadata = fs::symlink_metadata(entry.path())?;
let uid = metadata.uid();
let gid = metadata.gid();
let mode = metadata.permissions().mode();
let mtime = metadata.mtime();
if rel == "/" {
writer.set_root_uid(uid);
writer.set_root_gid(gid);
writer.set_root_mode(mode as u16);
continue;
}
let header = NodeHeader {
permissions: mode as u16,
uid,
gid,
mtime: mtime as u32,
};
if typ.is_symlink() {
let symlink = fs::read_link(entry.path())?;
let symlink = symlink
.to_str()
.ok_or_else(|| HyphaError::new("failed to read symlink"))?;
writer.push_symlink(symlink, rel, header)?;
} else if typ.is_dir() {
writer.push_dir(rel, header)?;
} else if typ.is_file() {
let reader = BufReader::new(File::open(entry.path())?);
writer.push_file(reader, rel, header)?;
} else if typ.is_block_device() {
let device = metadata.dev();
writer.push_block_device(device as u32, rel, header)?;
} else if typ.is_char_device() {
let device = metadata.dev();
writer.push_char_device(device as u32, rel, header)?;
} else {
return Err(HyphaError::new("invalid file type"));
}
}
fs::remove_dir_all(image_dir)?;
let squash_file_path = squash_file
.to_str()
.ok_or_else(|| HyphaError::new("failed to convert squashfs string"))?;
let mut out = File::create(squash_file)?;
2024-01-18 08:02:21 +00:00
trace!("ImageCompiler squash generate: {}", squash_file_path);
2024-01-17 22:29:05 +00:00
writer.write(&mut out)?;
2024-01-18 08:02:21 +00:00
Ok(())
2024-01-17 22:29:05 +00:00
}
}