about summary refs log tree commit diff
path: root/tvix/castore
diff options
context:
space:
mode:
authorFlorian Klink <flokli@flokli.de>2024-03-20T18·21+0200
committerclbot <clbot@tvl.fyi>2024-03-24T17·39+0000
commitc92ef2df64f4013e72037cefb548f68d158488cc (patch)
treed3d536e062a521fbcdcae8975d285315e47abb1b /tvix/castore
parentbd953d0703e072246fa7d7a79f9a46a8d2006a26 (diff)
feat(tvix/castore/directory): add ClosureValidator r/7772
This can be used to validate a Directory closure (connected DAG of
Directories), and their insertion order.

Directories need to be inserted (via `add`), in an order from the leaves
to the root. During insertion, we validate as much as we can at that
time:

 - individual validation of Directory messages
 - validation of insertion order (no upload of not-yet-known Directories)
 - validation of size fields of referred Directories

Internally it keeps all received Directories (and their sizes) in a HashMap,
keyed by digest.

Once all Directories have been inserted, a drain() function can be
called to get a (deduplicated and) validated list of directories, in
from-leaves-to-root order (to be stored somewhere).

While assembling that list, a check for graph connectivity is performed
too, to ensure there's no separate components being sent (and only one
root).

It adds a test suite for these cases, which is much nicer to test than
where we previously had these checks (only in the gRPC server wrapper).

Followup CLs will move the existing putters to use this.

Change-Id: Ie88c832924c170a24626e9e3e91d868497b5d7a4
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11220
Tested-by: BuildkiteCI
Reviewed-by: raitobezarius <tvl@lahfa.xyz>
Autosubmit: flokli <flokli@flokli.de>
Diffstat (limited to 'tvix/castore')
-rw-r--r--tvix/castore/src/directoryservice/closure_validator.rs254
-rw-r--r--tvix/castore/src/directoryservice/mod.rs2
2 files changed, 256 insertions, 0 deletions
diff --git a/tvix/castore/src/directoryservice/closure_validator.rs b/tvix/castore/src/directoryservice/closure_validator.rs
new file mode 100644
index 000000000000..ecd9acfca123
--- /dev/null
+++ b/tvix/castore/src/directoryservice/closure_validator.rs
@@ -0,0 +1,254 @@
+use std::collections::{HashMap, HashSet};
+
+use bstr::ByteSlice;
+
+use tracing::{instrument, trace};
+
+use crate::{
+    proto::{self, Directory},
+    B3Digest, Error,
+};
+
+/// This can be used to validate a Directory closure (DAG of connected
+/// Directories), and their insertion order.
+///
+/// Directories need to be inserted (via `add`), in an order from the leaves to
+/// the root. During insertion, We validate as much as we can at that time:
+///
+///  - individual validation of Directory messages
+///  - validation of insertion order (no upload of not-yet-known Directories)
+///  - validation of size fields of referred Directories
+///
+/// Internally it keeps all received Directories (and their sizes) in a HashMap,
+/// keyed by digest.
+///
+/// Once all Directories have been inserted, a finalize function can be
+/// called to get a (deduplicated and) validated list of directories, in
+/// insertion order.
+/// During finalize, a check for graph connectivity is performed too, to ensure
+/// there's no disconnected components, and only one root.
+#[derive(Default)]
+pub struct ClosureValidator {
+    directories_and_sizes: HashMap<B3Digest, (Directory, u64)>,
+
+    /// Keeps track of the last-inserted directory digest. Used to start the
+    /// connectivity check.
+    last_directory_digest: Option<B3Digest>,
+}
+
+impl ClosureValidator {
+    /// Insert a new Directory into the closure.
+    /// Perform individual Directory validation, validation of insertion order
+    // and size fields.
+    #[instrument(level = "trace", skip_all, fields(directory.digest=%directory.digest()), err)]
+    pub fn add(&mut self, directory: proto::Directory) -> Result<(), Error> {
+        let digest = directory.digest();
+
+        if self.directories_and_sizes.contains_key(&digest) {
+            trace!("already seen, skipping");
+            return Ok(());
+        }
+
+        // Do some general validation
+        directory
+            .validate()
+            .map_err(|e| Error::InvalidRequest(e.to_string()))?;
+
+        // Ensure the directory only refers to directories which we already accepted.
+        for dir in &directory.directories {
+            let dir_dgst = B3Digest::try_from(dir.digest.to_owned()).unwrap(); // validated
+
+            // Ensure the digest has already been seen
+            let (_, recorded_dir_size) =
+                self.directories_and_sizes.get(&dir_dgst).ok_or_else(|| {
+                    Error::InvalidRequest(format!(
+                        "'{}' refers to unseen child dir: {}",
+                        dir.name.as_bstr(),
+                        dir_dgst
+                    ))
+                })?;
+
+            // Ensure the size specified in the child node matches our records.
+            if dir.size != *recorded_dir_size {
+                return Err(Error::InvalidRequest(format!(
+                    "'{}' has wrong size, specified {}, recorded {}",
+                    dir.name.as_bstr(),
+                    dir.size,
+                    recorded_dir_size
+                )));
+            }
+        }
+
+        trace!("inserting");
+        let directory_size = directory.size();
+        self.directories_and_sizes
+            .insert(digest.clone(), (directory, directory_size));
+        self.last_directory_digest = Some(digest);
+
+        Ok(())
+    }
+
+    /// Ensure that all inserted Directories are connected, then return a
+    /// (deduplicated) and validated list of directories, in from-leaves-to-root
+    /// order.
+    /// In case no elements have been inserted, returns an empty list.
+    #[instrument(level = "trace", skip_all, err)]
+    pub(crate) fn finalize(mut self) -> Result<Vec<Directory>, Error> {
+        if self.last_directory_digest.is_none() {
+            return Ok(vec![]);
+        }
+        let root_digest = self.last_directory_digest.unwrap();
+
+        // recursively walk all directories reachable from there,
+        // ensure we visited all nodes that were uploaded.
+        // If not, we might have received multiple disconnected graphs.
+
+        // The list of directories we know we need to visit.
+        // Once we're finished working, and (there's no errors), this in reversed order will
+        // be a valid insertion order, and directories_and_sizes will be empty.
+        let mut dirs_to_visit = Vec::with_capacity(self.directories_and_sizes.len());
+        dirs_to_visit.push(
+            self.directories_and_sizes
+                .remove(&root_digest)
+                .expect("root digest not found")
+                .0,
+        );
+        // The set of digests seen while going visiting all directories.
+        let mut seen_dir_digests = HashSet::new();
+
+        // This loop moves gradually to the end of `dirs_to_visit`, while it's being
+        // extended.
+        // The loop stops once it reaches the end.
+        let mut i = 0;
+        while let Some(directory) = dirs_to_visit.get(i).map(|d| d.to_owned()) {
+            // lookup all child directories and put them in the back of dirs_to_visit,
+            // if they're not already there.
+            for child_dir in &directory.directories {
+                let child_digest = B3Digest::try_from(child_dir.digest.to_owned()).unwrap(); // validated
+
+                // In case the digest is neither already visited nor already in dirs_to_visit,
+                // add it to the end of it.
+                // We don't need to check for the hash we're currently
+                // visiting, as we can't self-reference.
+                if !seen_dir_digests.contains(&child_digest) {
+                    dirs_to_visit.push(
+                        self.directories_and_sizes
+                            .remove(&child_digest)
+                            .expect("child digest not found")
+                            .0,
+                    );
+                    seen_dir_digests.insert(child_digest);
+                }
+            }
+
+            i += 1;
+        }
+
+        // check directories_and_sizes is empty.
+        if !self.directories_and_sizes.is_empty() {
+            if cfg!(debug_assertions) {
+                return Err(Error::InvalidRequest(format!(
+                    "found {} disconnected nodes: {:?}",
+                    self.directories_and_sizes.len(),
+                    self.directories_and_sizes
+                )));
+            } else {
+                return Err(Error::InvalidRequest(format!(
+                    "found {} disconnected nodes",
+                    self.directories_and_sizes.len()
+                )));
+            }
+        }
+
+        // Reverse to have correct insertion order.
+        dirs_to_visit.reverse();
+
+        Ok(dirs_to_visit)
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use crate::{
+        fixtures::{DIRECTORY_A, DIRECTORY_B, DIRECTORY_C},
+        proto::{self, Directory},
+    };
+    use lazy_static::lazy_static;
+    use test_case::test_case;
+
+    lazy_static! {
+        pub static ref BROKEN_DIRECTORY : Directory = Directory {
+            symlinks: vec![proto::SymlinkNode {
+                name: "".into(), // invalid name!
+                target: "doesntmatter".into(),
+            }],
+            ..Default::default()
+        };
+
+        pub static ref BROKEN_PARENT_DIRECTORY: Directory = Directory {
+            directories: vec![proto::DirectoryNode {
+                name: "foo".into(),
+                digest: DIRECTORY_A.digest().into(),
+                size: DIRECTORY_A.size() + 42, // wrong!
+            }],
+            ..Default::default()
+        };
+    }
+
+    use super::ClosureValidator;
+
+    /// Uploading an empty directory should succeed.
+    #[test_case(vec![&DIRECTORY_A], false, Some(vec![&DIRECTORY_A]); "empty directory")]
+    /// Uploading A, then B (referring to A) should succeed.
+    #[test_case(vec![&DIRECTORY_A, &DIRECTORY_B], false, Some(vec![&DIRECTORY_A, &DIRECTORY_B]); "simple closure")]
+    /// Uploading A, then A, then C (referring to A twice) should succeed.
+    /// We pretend to be a dumb client not deduping directories.
+    #[test_case(vec![&DIRECTORY_A, &DIRECTORY_A, &DIRECTORY_C], false, Some(vec![&DIRECTORY_A, &DIRECTORY_C]); "same child")]
+    /// Uploading A, then C (referring to A twice) should succeed.
+    #[test_case(vec![&DIRECTORY_A, &DIRECTORY_C], false, Some(vec![&DIRECTORY_A, &DIRECTORY_C]); "same child dedup")]
+    /// Uploading A, then C (referring to A twice), then B (itself referring to A) should fail during close,
+    /// as B itself would be left unconnected.
+    #[test_case(vec![&DIRECTORY_A, &DIRECTORY_C, &DIRECTORY_B], false, None; "unconnected node")]
+    /// Uploading B (referring to A) should fail immediately, because A was never uploaded.
+    #[test_case(vec![&DIRECTORY_B], true, None; "dangling pointer")]
+    /// Uploading a directory failing validation should fail immediately.
+    #[test_case(vec![&BROKEN_DIRECTORY], true, None; "failing validation")]
+    /// Uploading a directory which refers to another Directory with a wrong size should fail.
+    #[test_case(vec![&DIRECTORY_A, &BROKEN_PARENT_DIRECTORY], true, None; "wrong size in parent")]
+    fn test_uploads(
+        directories_to_upload: Vec<&Directory>,
+        exp_fail_upload_last: bool,
+        exp_finalize: Option<Vec<&Directory>>, // Some(_) if finalize successful, None if not.
+    ) {
+        let mut dcv = ClosureValidator::default();
+        let len_directories_to_upload = directories_to_upload.len();
+
+        for (i, d) in directories_to_upload.iter().enumerate() {
+            let resp = dcv.add((*d).clone());
+            if i == len_directories_to_upload - 1 && exp_fail_upload_last {
+                assert!(resp.is_err(), "expect last put to fail");
+
+                // We don't really care anymore what finalize() would return, as
+                // the add() failed.
+                return;
+            } else {
+                assert!(resp.is_ok(), "expect put to succeed");
+            }
+        }
+
+        // everything was uploaded successfully. Test drain().
+        let resp = dcv.finalize();
+
+        match exp_finalize {
+            Some(exp_drain) => {
+                assert_eq!(
+                    Vec::from_iter(exp_drain.into_iter().map(|e| e.to_owned())),
+                    resp.expect("drain should succeed")
+                );
+            }
+            None => {
+                resp.expect_err("drain should fail");
+            }
+        }
+    }
+}
diff --git a/tvix/castore/src/directoryservice/mod.rs b/tvix/castore/src/directoryservice/mod.rs
index ca8e60e87304..b6d1c2fcab13 100644
--- a/tvix/castore/src/directoryservice/mod.rs
+++ b/tvix/castore/src/directoryservice/mod.rs
@@ -2,6 +2,7 @@ use crate::{proto, B3Digest, Error};
 use futures::stream::BoxStream;
 use tonic::async_trait;
 
+mod closure_validator;
 mod from_addr;
 mod grpc;
 mod memory;
@@ -9,6 +10,7 @@ mod sled;
 mod traverse;
 mod utils;
 
+pub use self::closure_validator::ClosureValidator;
 pub use self::from_addr::from_addr;
 pub use self::grpc::GRPCDirectoryService;
 pub use self::memory::MemoryDirectoryService;