crates/oxy-core/src/repo/fsck.rs at main · andymccall/oxygen

◆ main oxygen / crates / oxy-core / src / repo / fsck.rs

19.7 KiB · Rust · last changed The artifact as an object type of its own: oxy-core's artifact module, named by its object hash, checked by fsck (refs #108) 7 days ago source blame history raw

//! Repository integrity check (spec §21.1): `oxy fsck`.
//!
//! Reads everything and changes nothing. Four passes:
//!
//! 1. every stored object: header, length, hash, and that its body decodes
//!    as its type says;
//! 2. the operation log: every operation reachable from the head is
//!    present and decodes, and clocks never run backwards;
//! 3. references: the current view's bookmarks, tags, remote bookmarks,
//!    workspaces and policy, then the dependency closure of everything
//!    they name, with each change's author identity, blobs and chunks,
//!    and each change's signatures or seals;
//! 4. derived data: every bookmark's state validates, every checkpoint
//!    restores exactly the state it names, and reference and tag
//!    attestations carry good signatures.
//!
//! Missing blobs are warnings, not errors, in a partial clone
//! (`filter.max-blob` or a sparse filter), and changes beyond a shallow
//! boundary are expected to be absent.

use super::*;
use std::collections::{HashSet, VecDeque};

#[derive(Clone, Copy, Debug, PartialEq, Eq, PartialOrd, Ord)]
pub enum Severity {
    /// Worth knowing; the repository is still sound.
    Warning,
    /// Data is missing, corrupt or fails verification.
    Error,
}

#[derive(Clone, Debug, PartialEq, Eq)]
pub struct Problem {
    pub severity: Severity,
    /// The object the problem is about, when there is one.
    pub object: Option<ObjectRef>,
    pub what: String,
}

#[derive(Clone, Debug, Default, PartialEq, Eq)]
pub struct FsckReport {
    pub objects: usize,
    pub ops: usize,
    /// Changes checked, working-copy changes included.
    pub changes: usize,
    /// Committed changes authenticated by their own signature or by a seal.
    pub signed: usize,
    pub unsigned: usize,
    pub checkpoints: usize,
    pub attestations: usize,
    pub problems: Vec<Problem>,
}

impl FsckReport {
    pub fn errors(&self) -> usize {
        self.problems
            .iter()
            .filter(|p| p.severity == Severity::Error)
            .count()
    }

    pub fn warnings(&self) -> usize {
        self.problems
            .iter()
            .filter(|p| p.severity == Severity::Warning)
            .count()
    }

    fn error(&mut self, object: Option<ObjectRef>, what: impl Into<String>) {
        self.problems.push(Problem {
            severity: Severity::Error,
            object,
            what: what.into(),
        });
    }

    fn warn(&mut self, object: Option<ObjectRef>, what: impl Into<String>) {
        self.problems.push(Problem {
            severity: Severity::Warning,
            object,
            what: what.into(),
        });
    }
}

/// Whether `body` decodes as an object of type `ty`.
fn decodes(ty: ObjectType, body: &[u8]) -> std::result::Result<(), String> {
    let r = match ty {
        ObjectType::Chunk => Ok(()),
        ObjectType::Blob => Blob::decode(body).map(|_| ()),
        // Structure only: the text blobs are checked with the change graph.
        ObjectType::Change => {
            Change::decode_with(body, &mut |_, _, _| Ok(crate::lines::Lines::default())).map(|_| ())
        }
        ObjectType::Checkpoint => crate::objects::Checkpoint::decode(body).map(|_| ()),
        ObjectType::Op => match Op::decode(body) {
            Ok(_) => Ok(()),
            Err(e) => Op::decode_format1(body).map(|_| ()).map_err(|_| e),
        },
        ObjectType::Attest => Attest::decode(body).map(|_| ()),
        ObjectType::Identity => Identity::decode(body).map(|_| ()),
        ObjectType::Policy => crate::policy::Policy::decode(body).map(|_| ()),
        ObjectType::Receipt => crate::receipt::Receipt::decode(body).map(|_| ()),
        ObjectType::Certificate => crate::cert::Certificate::decode(body).map(|_| ()),
        ObjectType::Grant => crate::cert::Grant::decode(body).map(|_| ()),
        ObjectType::Artifact => crate::artifact::Artifact::decode(body).map(|_| ()),
        ObjectType::State => crate::objects::StateObject::decode(body).map(|_| ()),
    };
    r.map_err(|e| e.to_string())
}

impl Repo {
    /// Check the whole repository; see the module documentation.
    pub fn fsck(&self) -> Result<FsckReport> {
        // The head as it is now, as the operations pass reads it.
        self.forget_head();
        let mut rep = FsckReport::default();
        let bad = {
            let _t = crate::timing::phase("fsck: objects");
            self.fsck_objects(&mut rep)?
        };
        {
            let _t = crate::timing::phase("fsck: operations");
            self.fsck_ops(&mut rep, &bad);
        }
        {
            let _t = crate::timing::phase("fsck: references");
            self.fsck_references(&mut rep, &bad)?;
        }
        {
            let _t = crate::timing::phase("fsck: states and checkpoints");
            self.fsck_derived(&mut rep)?;
        }
        rep.problems.sort_by_key(|p| std::cmp::Reverse(p.severity));
        Ok(rep)
    }

    /// Pass 1. Returns the objects that failed, so later passes report
    /// them once rather than at every reference.
    fn fsck_objects(&self, rep: &mut FsckReport) -> Result<HashSet<ObjectRef>> {
        use rayon::prelude::*;
        let mut bad = HashSet::new();
        // A chunk at a time, read (and verified) and decoded on every core.
        for refs in self.store.list()?.chunks(8192) {
            let checked: Vec<std::result::Result<(), String>> = self
                .store
                .get_many(refs)
                .into_par_iter()
                .map(|got| match got {
                    Ok((ty, body)) => decodes(ty, &body)
                        .map_err(|e| format!("{} object does not decode: {e}", ty.name())),
                    Err(e) => Err(format!("unreadable object: {e}")),
                })
                .collect();
            for (r, c) in refs.iter().zip(checked) {
                rep.objects += 1;
                if let Err(e) = c {
                    rep.error(Some(*r), e);
                    bad.insert(*r);
                }
            }
        }
        Ok(bad)
    }

    /// Pass 2.
    fn fsck_ops(&self, rep: &mut FsckReport, bad: &HashSet<ObjectRef>) {
        let head = match self.read_head() {
            Ok((h, _)) => h,
            Err(RepoError::Msg(m)) if m.starts_with("corrupt") => {
                rep.error(None, ".oxy/ops/head does not hold an object reference");
                return;
            }
            Err(e) => {
                rep.error(None, format!("cannot read .oxy/ops/head: {e}"));
                return;
            }
        };
        let mut seen = HashSet::new();
        let mut queue: VecDeque<(ObjectRef, Option<(ObjectRef, u64)>)> =
            VecDeque::from([(head, None)]);
        while let Some((r, child)) = queue.pop_front() {
            if !seen.insert(r) {
                continue;
            }
            if bad.contains(&r) {
                continue;
            }
            let body = match self.store.get_typed(&r, ObjectType::Op) {
                Ok(b) => b,
                Err(StoreError::NotFound(_)) => {
                    let by = child
                        .map(|(c, _)| format!(", the parent of operation {}", c.short(12)))
                        .unwrap_or_default();
                    rep.error(
                        Some(r),
                        format!("operation missing{by}; the log is cut here"),
                    );
                    continue;
                }
                Err(e) => {
                    rep.error(Some(r), format!("operation unreadable: {e}"));
                    continue;
                }
            };
            rep.ops += 1;
            let (parents, wall_ms) = match Op::decode(&body) {
                Ok(op) => (op.parents, op.hlc.wall_ms),
                Err(_) => match Op::decode_format1(&body) {
                    Ok(op) => (op.parents, op.hlc.wall_ms),
                    Err(e) => {
                        rep.error(Some(r), format!("operation does not decode: {e}"));
                        continue;
                    }
                },
            };
            if let Some((c, child_ms)) = child {
                if wall_ms > child_ms {
                    rep.warn(
                        Some(c),
                        format!(
                            "operation is timed before its parent {} (a clock was wrong)",
                            r.short(12)
                        ),
                    );
                }
            }
            queue.extend(parents.into_iter().map(|p| (p, Some((r, wall_ms)))));
        }
    }

    /// Pass 3.
    fn fsck_references(&self, rep: &mut FsckReport, bad: &HashSet<ObjectRef>) -> Result<()> {
        let view = match self.view() {
            Ok(v) => v,
            Err(e) => {
                rep.error(None, format!("cannot read the current view: {e}"));
                return Ok(());
            }
        };
        let boundary = self.boundary();
        let partial = self.config.max_blob.is_some() || !self.config.sparse_paths.is_empty();

        let mut roots: Vec<(ChangeHash, String)> = Vec::new();
        let mut states: Vec<(ObjectRef, String)> = Vec::new();
        states.extend(
            view.bookmarks
                .iter()
                .map(|(n, s)| (*s, format!("bookmark {n}"))),
        );
        states.extend(view.tags.iter().map(|(n, s)| (*s, format!("tag {n}"))));
        for (remote, bms) in &view.remotes {
            states.extend(bms.iter().map(|(n, s)| (*s, format!("{remote}/{n}"))));
        }
        for (st, what) in states {
            match self.store.get_state(&st) {
                Ok(s) => roots.extend(s.heads.into_iter().map(|h| (h, what.clone()))),
                Err(e) if !bad.contains(&st) => {
                    rep.error(Some(st), format!("state of {what} unusable: {e}"))
                }
                Err(_) => {}
            }
        }
        for (ws, h) in &view.workspaces {
            roots.push((*h, format!("workspace {}", ws.to_hex())));
        }
        if let Some(p) = view.policy {
            match self.store.get_typed(&p, ObjectType::Policy) {
                Ok(bytes) => {
                    if let Ok(policy) = crate::policy::Policy::decode(&bytes) {
                        for id in &policy.identities {
                            if !self.store.has(&id.hash()) {
                                rep.warn(
                                    Some(id.hash()),
                                    "an identity the policy names is not stored",
                                );
                            }
                        }
                    }
                }
                Err(e) if !bad.contains(&p) => {
                    rep.error(Some(p), format!("the policy is unusable: {e}"))
                }
                Err(_) => {}
            }
        }

        // Working-copy changes are never signed; they are checked but not tallied.
        let working: HashSet<ChangeHash> = view.workspaces.values().copied().collect();
        let mut seen: HashSet<ChangeHash> = HashSet::new();
        let mut blobs_seen: HashSet<ObjectRef> = HashSet::new();
        let mut stack = roots;
        while let Some((h, needed_by)) = stack.pop() {
            if !seen.insert(h) {
                continue;
            }
            // The skeleton: the text is checked as the blobs it is stored
            // in, not decoded (the objects pass checked every object).
            let (c, texts) = match self.store.get_change_lazy(&h) {
                Ok(c) => c,
                Err(StoreError::NotFound(_)) if boundary.contains(&h) => continue,
                Err(StoreError::NotFound(_)) => {
                    rep.error(Some(h), format!("change missing, needed by {needed_by}"));
                    continue;
                }
                Err(e) => {
                    if !bad.contains(&h) {
                        rep.error(Some(h), format!("change unusable: {e}"));
                    }
                    continue;
                }
            };
            rep.changes += 1;
            let label = format!("change {}", h.short(12));
            stack.extend(c.deps.iter().map(|d| (*d, label.clone())));
            if !c.author.is_zero() && !self.store.has(&c.author) {
                rep.warn(
                    Some(h),
                    format!("author identity {} is not stored", c.author.short(12)),
                );
            }
            for t in texts {
                if let crate::objects::TextRef::Blob(b, _, _) = t {
                    if blobs_seen.insert(b) {
                        // A change cannot be read without its text: never partial.
                        self.fsck_blob(rep, b, &label, false, bad);
                    }
                }
            }
            for e in &c.edits {
                let Edit::Replace { old, new, .. } = e else {
                    continue;
                };
                for b in [old, new].into_iter().flatten() {
                    if !blobs_seen.insert(*b) {
                        continue;
                    }
                    self.fsck_blob(rep, *b, &label, partial, bad);
                }
            }
            if working.contains(&h) {
                continue;
            }
            match self.authenticate_change(&h) {
                Ok(SignatureStatus::Valid { .. }) => rep.signed += 1,
                Ok(SignatureStatus::Unsigned) => rep.unsigned += 1,
                Ok(SignatureStatus::Invalid(why)) => {
                    rep.error(Some(h), format!("bad signature: {why}"))
                }
                Err(e) => rep.error(Some(h), format!("cannot check signatures: {e}")),
            }
        }
        Ok(())
    }

    fn fsck_blob(
        &self,
        rep: &mut FsckReport,
        b: ObjectRef,
        used_by: &str,
        partial: bool,
        bad: &HashSet<ObjectRef>,
    ) {
        match self.store.get_typed(&b, ObjectType::Blob) {
            Ok(bytes) => {
                let Ok(blob) = Blob::decode(&bytes) else {
                    return;
                };
                for ch in blob.chunks {
                    if !self.store.has(&ch) {
                        rep.error(
                            Some(ch),
                            format!("chunk missing from blob {} ({used_by})", b.short(12)),
                        );
                    }
                }
            }
            Err(StoreError::NotFound(_)) if partial => rep.warn(
                Some(b),
                format!("blob not fetched ({used_by}); expected in a partial clone"),
            ),
            Err(StoreError::NotFound(_)) => rep.error(Some(b), format!("blob missing ({used_by})")),
            Err(e) if !bad.contains(&b) => {
                rep.error(Some(b), format!("blob unusable ({used_by}): {e}"))
            }
            Err(_) => {}
        }
    }

    /// Whether `heads` form a valid state: its history loads and every file
    /// graph builds. Graphs are built a batch at a time and dropped, so the
    /// check needs the memory of a batch, not of the whole tree.
    fn check_state(&self, heads: &[ChangeHash]) -> Result<()> {
        let mut st = self.load_state_history(heads)?;
        let nodes = st.unbuilt_nodes();
        for batch in nodes.chunks(2048) {
            self.load_graphs_of(&mut st, batch)?;
            st.drop_graphs(batch);
        }
        Ok(())
    }

    /// Pass 4.
    fn fsck_derived(&self, rep: &mut FsckReport) -> Result<()> {
        let Ok(view) = self.view() else { return Ok(()) };
        let shallow = self.is_shallow();
        if !shallow {
            let mut checked: HashSet<ObjectRef> = HashSet::new();
            for (name, st) in &view.bookmarks {
                if !checked.insert(*st) {
                    continue;
                }
                let Ok(s) = self.store.get_state(st) else {
                    continue;
                };
                if let Err(e) = self.check_state(&s.heads) {
                    rep.error(
                        Some(*st),
                        format!("bookmark {name} does not form a valid state: {e}"),
                    );
                }
            }
        }
        for r in self.checkpoint_refs()? {
            rep.checkpoints += 1;
            let cp = match self.store.get_typed(&r, ObjectType::Checkpoint) {
                Ok(b) => match crate::objects::Checkpoint::decode(&b) {
                    Ok(cp) => cp,
                    Err(_) => continue,
                },
                Err(StoreError::NotFound(_)) => {
                    rep.warn(
                        Some(r),
                        "checkpoint pointer names a missing checkpoint; it will be rewritten",
                    );
                    continue;
                }
                Err(_) => continue,
            };
            let payload = match self.read_blob(&cp.payload) {
                Ok(p) => p,
                Err(e) => {
                    rep.error(Some(r), format!("checkpoint payload unreadable: {e}"));
                    continue;
                }
            };
            if let Err(e) = self.state_from_checkpoint(&payload) {
                rep.error(Some(r), format!("checkpoint payload does not decode: {e}"));
                continue;
            }
            // The payload must be what the named changes produce, in its own version.
            let version = crate::cbor::decode(&payload).ok().and_then(|v| {
                crate::objects::map(&v, "payload").ok().and_then(|m| {
                    m.get(&0)
                        .and_then(|x| crate::objects::uint(x, "version").ok())
                })
            });
            let expected = |full: &State| {
                if version == Some(crate::state::CHECKPOINT_INDEX_VERSION) {
                    full.checkpoint_index(&full.checkpoint_file_graphs(), Repo::blob_ref)
                } else {
                    full.checkpoint_payload()
                }
            };
            if !shallow {
                match self.load_state_full(&cp.heads) {
                    Ok(full) if expected(&full) != payload => {
                        rep.error(Some(r), "checkpoint disagrees with the changes it names")
                    }
                    Ok(_) => {}
                    Err(e) => rep.error(
                        Some(r),
                        format!("checkpoint names changes that do not form a state: {e}"),
                    ),
                }
            }
        }
        let mut attests = Vec::new();
        for name in view.bookmarks.keys() {
            if let Ok(Some((r, _))) = self.latest_ref_attest(name) {
                attests.push((r, format!("bookmark {name}")));
            }
        }
        for name in view.tags.keys() {
            if let Ok(Some((r, _))) = self.tag_attest(name) {
                attests.push((r, format!("tag {name}")));
            }
        }
        for (r, what) in attests {
            rep.attestations += 1;
            let sigs = self.store.signatures(&r)?;
            if sigs.is_empty() {
                rep.warn(Some(r), format!("attestation for {what} is unsigned"));
            }
            for sig in sigs {
                match self.store.get_identity(&sig.key) {
                    Ok(id) => {
                        if let Err(e) = verify_signature(&id, &sig, Domain::Attest, &r) {
                            rep.error(
                                Some(r),
                                format!("attestation for {what} has a bad signature: {e}"),
                            );
                        }
                    }
                    Err(_) => rep.warn(
                        Some(r),
                        format!(
                            "attestation for {what} is signed by {}, whose identity is not stored",
                            sig.key.short(12)
                        ),
                    ),
                }
            }
        }
        Ok(())
    }
}

`/`	Search
`g` `c`	Code
`g` `h`	History
`g` `b`	Bookmarks
`g` `t`	Trust
`y`	Permalink to this state (files and trees)
`?`	This sheet

387335ff Andy McCall 1 month ago ↶	1	//! Repository integrity check (spec §21.1): `oxy fsck`.
	2	//!
	3	//! Reads everything and changes nothing. Four passes:
	4	//!
	5	//! 1. every stored object: header, length, hash, and that its body decodes
	6	//! as its type says;
	7	//! 2. the operation log: every operation reachable from the head is
	8	//! present and decodes, and clocks never run backwards;
	9	//! 3. references: the current view's bookmarks, tags, remote bookmarks,
	10	//! workspaces and policy, then the dependency closure of everything
	11	//! they name, with each change's author identity, blobs and chunks,
	12	//! and each change's signatures or seals;
	13	//! 4. derived data: every bookmark's state validates, every checkpoint
	14	//! restores exactly the state it names, and reference and tag
	15	//! attestations carry good signatures.
	16	//!
	17	//! Missing blobs are warnings, not errors, in a partial clone
	18	//! (`filter.max-blob` or a sparse filter), and changes beyond a shallow
	19	//! boundary are expected to be absent.
	20
	21	use super::*;
	22	use std::collections::{HashSet, VecDeque};
	23
	24	#[derive(Clone, Copy, Debug, PartialEq, Eq, PartialOrd, Ord)]
	25	pub enum Severity {
	26	/// Worth knowing; the repository is still sound.
	27	Warning,
	28	/// Data is missing, corrupt or fails verification.
	29	Error,
	30	}
	31
	32	#[derive(Clone, Debug, PartialEq, Eq)]
	33	pub struct Problem {
	34	pub severity: Severity,
	35	/// The object the problem is about, when there is one.
	36	pub object: Option<ObjectRef>,
	37	pub what: String,
	38	}
	39
	40	#[derive(Clone, Debug, Default, PartialEq, Eq)]
	41	pub struct FsckReport {
	42	pub objects: usize,
	43	pub ops: usize,
	44	/// Changes checked, working-copy changes included.
	45	pub changes: usize,
	46	/// Committed changes authenticated by their own signature or by a seal.
	47	pub signed: usize,
	48	pub unsigned: usize,
	49	pub checkpoints: usize,
	50	pub attestations: usize,
	51	pub problems: Vec<Problem>,
	52	}
	53
	54	impl FsckReport {
	55	pub fn errors(&self) -> usize {
	56	self.problems
	57	.iter()
	58	.filter(\|p\| p.severity == Severity::Error)
	59	.count()
	60	}
	61
	62	pub fn warnings(&self) -> usize {
	63	self.problems
	64	.iter()
65	.filter(\|p\| p.severity == Severity::Warning)
66	.count()
67	}
68
69	fn error(&mut self, object: Option<ObjectRef>, what: impl Into<String>) {
70	self.problems.push(Problem {
71	severity: Severity::Error,
72	object,
73	what: what.into(),
74	});
75	}
76
77	fn warn(&mut self, object: Option<ObjectRef>, what: impl Into<String>) {
78	self.problems.push(Problem {
79	severity: Severity::Warning,
80	object,
81	what: what.into(),
82	});
83	}
84	}
85
86	/// Whether `body` decodes as an object of type `ty`.
87	fn decodes(ty: ObjectType, body: &[u8]) -> std::result::Result<(), String> {
88	let r = match ty {
89	ObjectType::Chunk => Ok(()),
90	ObjectType::Blob => Blob::decode(body).map(\|_\| ()),
6f8aeb99 Andy McCall 1 month ago ↶	91	// Structure only: the text blobs are checked with the change graph.
6f8aeb99 Andy McCall 1 month ago ↶	92	ObjectType::Change => {
0f93f49c Andy McCall 1 month ago ↶	93	Change::decode_with(body, &mut \|_, _, _\| Ok(crate::lines::Lines::default())).map(\|_\| ())
6f8aeb99 Andy McCall 1 month ago ↶	94	}
387335ff Andy McCall 1 month ago ↶	95	ObjectType::Checkpoint => crate::objects::Checkpoint::decode(body).map(\|_\| ()),
	96	ObjectType::Op => match Op::decode(body) {
	97	Ok(_) => Ok(()),
	98	Err(e) => Op::decode_format1(body).map(\|_\| ()).map_err(\|_\| e),
	99	},
	100	ObjectType::Attest => Attest::decode(body).map(\|_\| ()),
	101	ObjectType::Identity => Identity::decode(body).map(\|_\| ()),
	102	ObjectType::Policy => crate::policy::Policy::decode(body).map(\|_\| ()),
fc1a466a Andy McCall 1 month ago ↶	103	ObjectType::Receipt => crate::receipt::Receipt::decode(body).map(\|_\| ()),
02e69188 Andy McCall 19 days ago ↶	104	ObjectType::Certificate => crate::cert::Certificate::decode(body).map(\|_\| ()),
a2387d54 Andy McCall 18 days ago ↶	105	ObjectType::Grant => crate::cert::Grant::decode(body).map(\|_\| ()),
6c3dbe20 Andy McCall 7 days ago ↶	106	ObjectType::Artifact => crate::artifact::Artifact::decode(body).map(\|_\| ()),
387335ff Andy McCall 1 month ago ↶	107	ObjectType::State => crate::objects::StateObject::decode(body).map(\|_\| ()),
	108	};
	109	r.map_err(\|e\| e.to_string())
	110	}
	111
	112	impl Repo {
	113	/// Check the whole repository; see the module documentation.
	114	pub fn fsck(&self) -> Result<FsckReport> {
767ac6cd Andy McCall 18 days ago	115	// The head as it is now, as the operations pass reads it.
767ac6cd Andy McCall 18 days ago	116	self.forget_head();
387335ff Andy McCall 1 month ago ↶	117	let mut rep = FsckReport::default();
d77e1702 Andy McCall 1 month ago ↶	118	let bad = {
	119	let _t = crate::timing::phase("fsck: objects");
	120	self.fsck_objects(&mut rep)?
	121	};
	122	{
	123	let _t = crate::timing::phase("fsck: operations");
	124	self.fsck_ops(&mut rep, &bad);
	125	}
	126	{
	127	let _t = crate::timing::phase("fsck: references");
	128	self.fsck_references(&mut rep, &bad)?;
	129	}
	130	{
	131	let _t = crate::timing::phase("fsck: states and checkpoints");
	132	self.fsck_derived(&mut rep)?;
	133	}
387335ff Andy McCall 1 month ago ↶	134	rep.problems.sort_by_key(\|p\| std::cmp::Reverse(p.severity));
	135	Ok(rep)
	136	}
	137
	138	/// Pass 1. Returns the objects that failed, so later passes report
	139	/// them once rather than at every reference.
	140	fn fsck_objects(&self, rep: &mut FsckReport) -> Result<HashSet<ObjectRef>> {
acadb5ed Andy McCall 1 month ago ↶	141	use rayon::prelude::*;
387335ff Andy McCall 1 month ago ↶	142	let mut bad = HashSet::new();
acadb5ed Andy McCall 1 month ago ↶	143	// A chunk at a time, read (and verified) and decoded on every core.
	144	for refs in self.store.list()?.chunks(8192) {
	145	let checked: Vec<std::result::Result<(), String>> = self
	146	.store
	147	.get_many(refs)
	148	.into_par_iter()
	149	.map(\|got\| match got {
	150	Ok((ty, body)) => decodes(ty, &body)
	151	.map_err(\|e\| format!("{} object does not decode: {e}", ty.name())),
	152	Err(e) => Err(format!("unreadable object: {e}")),
	153	})
	154	.collect();
	155	for (r, c) in refs.iter().zip(checked) {
	156	rep.objects += 1;
	157	if let Err(e) = c {
	158	rep.error(Some(*r), e);
	159	bad.insert(*r);
387335ff Andy McCall 1 month ago ↶	160	}
	161	}
	162	}
	163	Ok(bad)
	164	}
	165
	166	/// Pass 2.
	167	fn fsck_ops(&self, rep: &mut FsckReport, bad: &HashSet<ObjectRef>) {
dd21fd82 Andy McCall 1 month ago ↶	168	let head = match self.read_head() {
	169	Ok((h, _)) => h,
	170	Err(RepoError::Msg(m)) if m.starts_with("corrupt") => {
	171	rep.error(None, ".oxy/ops/head does not hold an object reference");
	172	return;
	173	}
387335ff Andy McCall 1 month ago ↶	174	Err(e) => {
	175	rep.error(None, format!("cannot read .oxy/ops/head: {e}"));
	176	return;
	177	}
	178	};
	179	let mut seen = HashSet::new();
	180	let mut queue: VecDeque<(ObjectRef, Option<(ObjectRef, u64)>)> =
	181	VecDeque::from([(head, None)]);
	182	while let Some((r, child)) = queue.pop_front() {
	183	if !seen.insert(r) {
	184	continue;
	185	}
	186	if bad.contains(&r) {
	187	continue;
	188	}
	189	let body = match self.store.get_typed(&r, ObjectType::Op) {
	190	Ok(b) => b,
	191	Err(StoreError::NotFound(_)) => {
	192	let by = child
	193	.map(\|(c, _)\| format!(", the parent of operation {}", c.short(12)))
	194	.unwrap_or_default();
	195	rep.error(
	196	Some(r),
	197	format!("operation missing{by}; the log is cut here"),
	198	);
	199	continue;
	200	}
	201	Err(e) => {
	202	rep.error(Some(r), format!("operation unreadable: {e}"));
	203	continue;
	204	}
	205	};
	206	rep.ops += 1;
	207	let (parents, wall_ms) = match Op::decode(&body) {
	208	Ok(op) => (op.parents, op.hlc.wall_ms),
	209	Err(_) => match Op::decode_format1(&body) {
	210	Ok(op) => (op.parents, op.hlc.wall_ms),
	211	Err(e) => {
	212	rep.error(Some(r), format!("operation does not decode: {e}"));
	213	continue;
	214	}
	215	},
	216	};
	217	if let Some((c, child_ms)) = child {
	218	if wall_ms > child_ms {
	219	rep.warn(
	220	Some(c),
	221	format!(
	222	"operation is timed before its parent {} (a clock was wrong)",
	223	r.short(12)
	224	),
	225	);
	226	}
	227	}
	228	queue.extend(parents.into_iter().map(\|p\| (p, Some((r, wall_ms)))));
	229	}
	230	}
	231
	232	/// Pass 3.
	233	fn fsck_references(&self, rep: &mut FsckReport, bad: &HashSet<ObjectRef>) -> Result<()> {
	234	let view = match self.view() {
	235	Ok(v) => v,
	236	Err(e) => {
	237	rep.error(None, format!("cannot read the current view: {e}"));
238	return Ok(());
239	}
240	};
241	let boundary = self.boundary();
242	let partial = self.config.max_blob.is_some() \|\| !self.config.sparse_paths.is_empty();
243
244	let mut roots: Vec<(ChangeHash, String)> = Vec::new();
245	let mut states: Vec<(ObjectRef, String)> = Vec::new();
246	states.extend(
247	view.bookmarks
248	.iter()
249	.map(\|(n, s)\| (*s, format!("bookmark {n}"))),
250	);
251	states.extend(view.tags.iter().map(\|(n, s)\| (*s, format!("tag {n}"))));
252	for (remote, bms) in &view.remotes {
253	states.extend(bms.iter().map(\|(n, s)\| (*s, format!("{remote}/{n}"))));
254	}
255	for (st, what) in states {
256	match self.store.get_state(&st) {
257	Ok(s) => roots.extend(s.heads.into_iter().map(\|h\| (h, what.clone()))),
258	Err(e) if !bad.contains(&st) => {
259	rep.error(Some(st), format!("state of {what} unusable: {e}"))
260	}
261	Err(_) => {}
262	}
263	}
264	for (ws, h) in &view.workspaces {
265	roots.push((*h, format!("workspace {}", ws.to_hex())));
266	}
267	if let Some(p) = view.policy {
268	match self.store.get_typed(&p, ObjectType::Policy) {
269	Ok(bytes) => {
270	if let Ok(policy) = crate::policy::Policy::decode(&bytes) {
271	for id in &policy.identities {
272	if !self.store.has(&id.hash()) {
273	rep.warn(
274	Some(id.hash()),
275	"an identity the policy names is not stored",
276	);
277	}
278	}
279	}
280	}
281	Err(e) if !bad.contains(&p) => {
282	rep.error(Some(p), format!("the policy is unusable: {e}"))
283	}
284	Err(_) => {}
285	}
286	}
287
288	// Working-copy changes are never signed; they are checked but not tallied.
289	let working: HashSet<ChangeHash> = view.workspaces.values().copied().collect();
290	let mut seen: HashSet<ChangeHash> = HashSet::new();
291	let mut blobs_seen: HashSet<ObjectRef> = HashSet::new();
292	let mut stack = roots;
293	while let Some((h, needed_by)) = stack.pop() {
294	if !seen.insert(h) {
295	continue;
296	}
d77e1702 Andy McCall 1 month ago ↶	297	// The skeleton: the text is checked as the blobs it is stored
	298	// in, not decoded (the objects pass checked every object).
	299	let (c, texts) = match self.store.get_change_lazy(&h) {
387335ff Andy McCall 1 month ago ↶	300	Ok(c) => c,
	301	Err(StoreError::NotFound(_)) if boundary.contains(&h) => continue,
	302	Err(StoreError::NotFound(_)) => {
	303	rep.error(Some(h), format!("change missing, needed by {needed_by}"));
	304	continue;
	305	}
	306	Err(e) => {
	307	if !bad.contains(&h) {
	308	rep.error(Some(h), format!("change unusable: {e}"));
	309	}
	310	continue;
	311	}
	312	};
	313	rep.changes += 1;
	314	let label = format!("change {}", h.short(12));
	315	stack.extend(c.deps.iter().map(\|d\| (*d, label.clone())));
	316	if !c.author.is_zero() && !self.store.has(&c.author) {
	317	rep.warn(
	318	Some(h),
	319	format!("author identity {} is not stored", c.author.short(12)),
	320	);
d77e1702 Andy McCall 1 month ago ↶	321	}
	322	for t in texts {
	323	if let crate::objects::TextRef::Blob(b, _, _) = t {
	324	if blobs_seen.insert(b) {
	325	// A change cannot be read without its text: never partial.
	326	self.fsck_blob(rep, b, &label, false, bad);
	327	}
	328	}
387335ff Andy McCall 1 month ago ↶	329	}
	330	for e in &c.edits {
	331	let Edit::Replace { old, new, .. } = e else {
	332	continue;
	333	};
	334	for b in [old, new].into_iter().flatten() {
	335	if !blobs_seen.insert(*b) {
	336	continue;
	337	}
	338	self.fsck_blob(rep, *b, &label, partial, bad);
	339	}
	340	}
	341	if working.contains(&h) {
	342	continue;
	343	}
	344	match self.authenticate_change(&h) {
	345	Ok(SignatureStatus::Valid { .. }) => rep.signed += 1,
	346	Ok(SignatureStatus::Unsigned) => rep.unsigned += 1,
	347	Ok(SignatureStatus::Invalid(why)) => {
	348	rep.error(Some(h), format!("bad signature: {why}"))
	349	}
	350	Err(e) => rep.error(Some(h), format!("cannot check signatures: {e}")),
	351	}
	352	}
	353	Ok(())
	354	}
	355
	356	fn fsck_blob(
	357	&self,
	358	rep: &mut FsckReport,
	359	b: ObjectRef,
	360	used_by: &str,
	361	partial: bool,
	362	bad: &HashSet<ObjectRef>,
	363	) {
	364	match self.store.get_typed(&b, ObjectType::Blob) {
	365	Ok(bytes) => {
	366	let Ok(blob) = Blob::decode(&bytes) else {
	367	return;
	368	};
	369	for ch in blob.chunks {
	370	if !self.store.has(&ch) {
	371	rep.error(
	372	Some(ch),
	373	format!("chunk missing from blob {} ({used_by})", b.short(12)),
	374	);
	375	}
	376	}
	377	}
	378	Err(StoreError::NotFound(_)) if partial => rep.warn(
	379	Some(b),
	380	format!("blob not fetched ({used_by}); expected in a partial clone"),
	381	),
	382	Err(StoreError::NotFound(_)) => rep.error(Some(b), format!("blob missing ({used_by})")),
	383	Err(e) if !bad.contains(&b) => {
	384	rep.error(Some(b), format!("blob unusable ({used_by}): {e}"))
	385	}
	386	Err(_) => {}
	387	}
	388	}
	389
69816d68 Andy McCall 1 month ago ↶	390	/// Whether `heads` form a valid state: its history loads and every file
	391	/// graph builds. Graphs are built a batch at a time and dropped, so the
	392	/// check needs the memory of a batch, not of the whole tree.
	393	fn check_state(&self, heads: &[ChangeHash]) -> Result<()> {
	394	let mut st = self.load_state_history(heads)?;
	395	let nodes = st.unbuilt_nodes();
	396	for batch in nodes.chunks(2048) {
	397	self.load_graphs_of(&mut st, batch)?;
	398	st.drop_graphs(batch);
	399	}
	400	Ok(())
	401	}
	402
387335ff Andy McCall 1 month ago ↶	403	/// Pass 4.
	404	fn fsck_derived(&self, rep: &mut FsckReport) -> Result<()> {
	405	let Ok(view) = self.view() else { return Ok(()) };
	406	let shallow = self.is_shallow();
	407	if !shallow {
69816d68 Andy McCall 1 month ago ↶	408	let mut checked: HashSet<ObjectRef> = HashSet::new();
387335ff Andy McCall 1 month ago ↶	409	for (name, st) in &view.bookmarks {
69816d68 Andy McCall 1 month ago ↶	410	if !checked.insert(*st) {
	411	continue;
	412	}
387335ff Andy McCall 1 month ago ↶	413	let Ok(s) = self.store.get_state(st) else {
	414	continue;
	415	};
69816d68 Andy McCall 1 month ago ↶	416	if let Err(e) = self.check_state(&s.heads) {
387335ff Andy McCall 1 month ago ↶	417	rep.error(
	418	Some(*st),
	419	format!("bookmark {name} does not form a valid state: {e}"),
	420	);
	421	}
	422	}
	423	}
	424	for r in self.checkpoint_refs()? {
	425	rep.checkpoints += 1;
	426	let cp = match self.store.get_typed(&r, ObjectType::Checkpoint) {
	427	Ok(b) => match crate::objects::Checkpoint::decode(&b) {
	428	Ok(cp) => cp,
	429	Err(_) => continue,
	430	},
	431	Err(StoreError::NotFound(_)) => {
	432	rep.warn(
	433	Some(r),
	434	"checkpoint pointer names a missing checkpoint; it will be rewritten",
	435	);
	436	continue;
	437	}
	438	Err(_) => continue,
	439	};
	440	let payload = match self.read_blob(&cp.payload) {
	441	Ok(p) => p,
	442	Err(e) => {
	443	rep.error(Some(r), format!("checkpoint payload unreadable: {e}"));
	444	continue;
	445	}
	446	};
afaba5f1 Andy McCall 1 month ago ↶	447	if let Err(e) = self.state_from_checkpoint(&payload) {
afaba5f1 Andy McCall 1 month ago ↶	448	rep.error(Some(r), format!("checkpoint payload does not decode: {e}"));
387335ff Andy McCall 1 month ago ↶	449	continue;
387335ff Andy McCall 1 month ago ↶	450	}
afaba5f1 Andy McCall 1 month ago ↶	451	// The payload must be what the named changes produce, in its own version.
	452	let version = crate::cbor::decode(&payload).ok().and_then(\|v\| {
	453	crate::objects::map(&v, "payload").ok().and_then(\|m\| {
	454	m.get(&0)
	455	.and_then(\|x\| crate::objects::uint(x, "version").ok())
	456	})
	457	});
	458	let expected = \|full: &State\| {
	459	if version == Some(crate::state::CHECKPOINT_INDEX_VERSION) {
	460	full.checkpoint_index(&full.checkpoint_file_graphs(), Repo::blob_ref)
	461	} else {
	462	full.checkpoint_payload()
	463	}
	464	};
387335ff Andy McCall 1 month ago ↶	465	if !shallow {
387335ff Andy McCall 1 month ago ↶	466	match self.load_state_full(&cp.heads) {
afaba5f1 Andy McCall 1 month ago ↶	467	Ok(full) if expected(&full) != payload => {
387335ff Andy McCall 1 month ago ↶	468	rep.error(Some(r), "checkpoint disagrees with the changes it names")
	469	}
	470	Ok(_) => {}
	471	Err(e) => rep.error(
	472	Some(r),
	473	format!("checkpoint names changes that do not form a state: {e}"),
	474	),
	475	}
	476	}
	477	}
	478	let mut attests = Vec::new();
	479	for name in view.bookmarks.keys() {
	480	if let Ok(Some((r, _))) = self.latest_ref_attest(name) {
	481	attests.push((r, format!("bookmark {name}")));
	482	}
	483	}
	484	for name in view.tags.keys() {
	485	if let Ok(Some((r, _))) = self.tag_attest(name) {
	486	attests.push((r, format!("tag {name}")));
	487	}
	488	}
	489	for (r, what) in attests {
	490	rep.attestations += 1;
	491	let sigs = self.store.signatures(&r)?;
	492	if sigs.is_empty() {
	493	rep.warn(Some(r), format!("attestation for {what} is unsigned"));
	494	}
	495	for sig in sigs {
	496	match self.store.get_identity(&sig.key) {
	497	Ok(id) => {
	498	if let Err(e) = verify_signature(&id, &sig, Domain::Attest, &r) {
	499	rep.error(
	500	Some(r),
	501	format!("attestation for {what} has a bad signature: {e}"),
	502	);
	503	}
	504	}
	505	Err(_) => rep.warn(
	506	Some(r),
	507	format!(
	508	"attestation for {what} is signed by {}, whose identity is not stored",
	509	sig.key.short(12)
	510	),
	511	),
	512	}
	513	}
	514	}
	515	Ok(())
	516	}
	517	}

Keyboard