Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Keep PendingComponents in da_checker during import_block #5845

Merged
merged 5 commits into from
Jun 3, 2024
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
14 changes: 14 additions & 0 deletions beacon_node/beacon_chain/src/beacon_chain.rs
Original file line number Diff line number Diff line change
Expand Up @@ -3328,6 +3328,20 @@ impl<T: BeaconChainTypes> BeaconChain<T> {
"payload_verification_handle",
)
.await??;

// Remove block components from da_checker AFTER completing block import. Then we can assert
// the following invariant:
// > A valid unfinalized block is either in fork-choice or da_checker.
//
// If we remove the block when it becomes available, there's some time window during
// `import_block` where the block is nowhere. Consumers of the da_checker can handle the
// extend time a block may exist in the da_checker.
//
// If `import_block` errors (only errors with internal errors), the pending components will
// be pruned on data_availability_checker maintenance as finality advances.
self.data_availability_checker
.remove_pending_components(block_root);

Ok(AvailabilityProcessingStatus::Imported(block_root))
}

Expand Down
5 changes: 5 additions & 0 deletions beacon_node/beacon_chain/src/data_availability_checker.rs
Original file line number Diff line number Diff line change
Expand Up @@ -164,6 +164,11 @@ impl<T: BeaconChainTypes> DataAvailabilityChecker<T> {
.put_pending_executed_block(executed_block)
}

pub fn remove_pending_components(&self, block_root: Hash256) {
self.availability_cache
.remove_pending_components(block_root)
}

/// Verifies kzg commitments for an RpcBlock, returns a `MaybeAvailableBlock` that may
/// include the fully available block.
///
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -498,6 +498,21 @@ impl<T: BeaconChainTypes> Critical<T> {
}
}

/// Removes and returns the pending_components corresponding to
/// the `block_root` or `None` if it does not exist
Comment on lines +501 to +502
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggested change
/// Removes and returns the pending_components corresponding to
/// the `block_root` or `None` if it does not exist
/// Removes the `pending_components` corresponding to the `block_root`.

pub fn remove_pending_components(&mut self, block_root: Hash256) {
match self.in_memory.pop_entry(&block_root) {
Some { .. } => {}
None => {
// not in memory, is it in the store?
// We don't need to remove the data from the store as we have removed it from
// `store_keys` so we won't go looking for it on disk. The maintenance thread
// will remove it from disk the next time it runs.
self.store_keys.remove(&block_root);
}
}
}

/// Returns the number of pending component entries in memory.
pub fn num_blocks(&self) -> usize {
self.in_memory.len()
Expand Down Expand Up @@ -607,6 +622,11 @@ impl<T: BeaconChainTypes> OverflowLRUCache<T> {
pending_components.merge_blobs(fixed_blobs);

if pending_components.is_available() {
write_lock.put_pending_components(
block_root,
pending_components.clone(),
&self.overflow_store,
)?;
Comment on lines +625 to +629
Copy link
Member

@jimmygchen jimmygchen May 29, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We're cloning pending components for a second time here and I'm not sure if it's necessary.

If the pending component is newly created in this function, it won't immediately become available. So this means the pending components would always be in the overflow_store if it becomes available here right?

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Oh never mind, we do need this as we're mutating a clone!

Copy link
Member

@ethDreamer ethDreamer May 30, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

You don't need to modify pop_pending_components to get_pending_components. Instead you should keep it as is. Because the write lock is held the whole time, there is no difference between:

lock = get_write_lock()
item = cache.remove()
item.mutate()
cache.insert(item)
drop(lock)

and

lock = get_write_lock()
item = cache.get()
item.mutate()
cache.update(item)
drop(lock)

You can just re-insert the item even when the components are complete (just like you did above) and then remove them with the same pop_pending_components method.

// No need to hold the write lock anymore
drop(write_lock);
pending_components.make_available(|diet_block| {
Expand Down Expand Up @@ -646,6 +666,11 @@ impl<T: BeaconChainTypes> OverflowLRUCache<T> {

// Check if we have all components and entire set is consistent.
if pending_components.is_available() {
write_lock.put_pending_components(
block_root,
pending_components.clone(),
&self.overflow_store,
)?;
// No need to hold the write lock anymore
drop(write_lock);
pending_components.make_available(|diet_block| {
Expand All @@ -661,6 +686,10 @@ impl<T: BeaconChainTypes> OverflowLRUCache<T> {
}
}

pub fn remove_pending_components(&self, block_root: Hash256) {
self.critical.write().remove_pending_components(block_root);
}

/// write all in memory objects to disk
pub fn write_all_to_disk(&self) -> Result<(), AvailabilityCheckError> {
let maintenance_lock = self.maintenance_lock.lock();
Expand Down Expand Up @@ -1195,10 +1224,17 @@ mod test {
matches!(availability, Availability::Available(_)),
"block doesn't have blobs, should be available"
);
assert_eq!(
cache.critical.read().in_memory.len(),
1,
"cache should still have block as it hasn't been imported yet"
);
// remove the blob to simulate successful import
cache.remove_pending_components(root);
assert_eq!(
cache.critical.read().in_memory.len(),
0,
"cache should be empty because we don't have blobs"
"cache should be empty now that block has been imported"
);
} else {
assert!(
Expand Down Expand Up @@ -1263,6 +1299,12 @@ mod test {
"block should be available: {:?}",
availability
);
assert!(
cache.critical.read().in_memory.len() == 1,
"cache should still have available block until import"
);
// remove the blob to simulate successful import
cache.remove_pending_components(root);
assert!(
cache.critical.read().in_memory.is_empty(),
"cache should be empty now that all components available"
Expand Down Expand Up @@ -1378,6 +1420,8 @@ mod test {
.expect("should put blob");
if blob_index == expected_blobs - 1 {
assert!(matches!(availability, Availability::Available(_)));
// remove the block from the cache to simulate import
cache.remove_pending_components(roots[0]);
} else {
// the first block should be brought back into memory
assert!(
Expand Down
Loading