feat(dot/parachain): implement candidate validation from chainstate (#…

…4067) Co-authored-by: Timothy Wu <[email protected]> test(dot/parachain/backing): make sure parachain candidate reaches quorum (#4115) - written integration test to check everything mentioned above - I re-wrote the logic to convert the attested to the backed candidate. (was not working as expected) - minor changes in dot/parachain/backing/per_relay_parent_state.go to improve the readability. test(dot/parachain/backing): validation fail does not stop the candidate backing subsystem (#4117) address merge conflicts chore(dot/parachain): improve Run method of subsystems (#4113) - I have updated the Run method of the subsystem interface to pass only necessary arguments. - removed context, context cancel func, wait group and OverseerToSubSystem channel from the state struct of all the subsystems(example: candidateBacking{}, AvailabilityStoreSubsystem{}). - improved Run method of some subsystems. - removed return type from RegisterSubsystem method of overseer. - because of these changes, some tests were failing, so I fixed those tests.
ChainSafe · Oct 4, 2024 · c0ef147 · c0ef147
1 parent 566f455
commit c0ef147
Show file tree

Hide file tree

Showing 24 changed files with 633 additions and 354 deletions.
diff --git a/dot/parachain/availability-store/availability_store.go b/dot/parachain/availability-store/availability_store.go
@@ -8,7 +8,6 @@ import (
 	"encoding/binary"
 	"errors"
 	"fmt"
-	"sync"
 	"time"
 
 	"github.com/ChainSafe/gossamer/dot/parachain/chainapi"
@@ -75,12 +74,7 @@ var defaultPruningConfig = pruningConfig{
 
 // AvailabilityStoreSubsystem is the struct that holds subsystem data for the availability store
 type AvailabilityStoreSubsystem struct {
-	ctx    context.Context
-	cancel context.CancelFunc
-	wg     sync.WaitGroup
-
 	subSystemToOverseer    chan<- any
-	OverseerToSubSystem    <-chan any
 	availabilityStore      availabilityStore
 	finalizedBlockNumber   parachaintypes.BlockNumber
 	knownUnfinalizedBlocks knownUnfinalizedBlocks
@@ -91,12 +85,7 @@ type AvailabilityStoreSubsystem struct {
 func NewAvailabilityStoreSubsystem(db database.Database) *AvailabilityStoreSubsystem {
 	availabilityStore := NewAvailabilityStore(db)
 
-	ctx := context.Background()
-	ctx, cancel := context.WithCancel(ctx)
-
 	availabilityStoreSubsystem := &AvailabilityStoreSubsystem{
-		ctx:                    ctx,
-		cancel:                 cancel,
 		pruningConfig:          defaultPruningConfig,
 		availabilityStore:      *availabilityStore,
 		knownUnfinalizedBlocks: *newKnownUnfinalizedBlock(),
@@ -420,89 +409,85 @@ func branchesFromChunks(chunks [][]byte) (branches, error) {
 }
 
 // Run runs the availability store subsystem
-func (av *AvailabilityStoreSubsystem) Run(ctx context.Context, OverseerToSubsystem chan any,
-	SubsystemToOverseer chan any) {
-
-	av.wg.Add(1)
-	go av.processMessages()
+func (av *AvailabilityStoreSubsystem) Run(ctx context.Context, overseerToSubsystem <-chan any) {
+	for {
+		select {
+		case msg := <-overseerToSubsystem:
+			logger.Infof("received message %T, %v", msg, msg)
+			av.processMessage(msg)
+		case <-time.After(av.pruningConfig.pruningInterval):
+			av.pruneAll()
+		case <-ctx.Done():
+			if err := ctx.Err(); err != nil {
+				logger.Errorf("ctx error: %v", err)
+			}
+			return
+		}
+	}
 }
 
 // Name returns the name of the availability store subsystem
 func (*AvailabilityStoreSubsystem) Name() parachaintypes.SubSystemName {
 	return parachaintypes.AvailabilityStore
 }
 
-func (av *AvailabilityStoreSubsystem) processMessages() {
-	for {
-		select {
-		case msg := <-av.OverseerToSubSystem:
-			logger.Infof("received message %T, %v", msg, msg)
-			switch msg := msg.(type) {
-			case QueryAvailableData:
-				err := av.handleQueryAvailableData(msg)
-				if err != nil {
-					logger.Errorf("failed to handle available data: %w", err)
-				}
-			case QueryDataAvailability:
-				err := av.handleQueryDataAvailability(msg)
-				if err != nil {
-					logger.Errorf("failed to handle query data availability: %w", err)
-				}
-			case QueryChunk:
-				err := av.handleQueryChunk(msg)
-				if err != nil {
-					logger.Errorf("failed to handle query chunk: %w", err)
-				}
-			case QueryChunkSize:
-				err := av.handleQueryChunkSize(msg)
-				if err != nil {
-					logger.Errorf("failed to handle query chunk size: %w", err)
-				}
-			case QueryAllChunks:
-				err := av.handleQueryAllChunks(msg)
-				if err != nil {
-					logger.Errorf("failed to handle query all chunks: %w", err)
-				}
-			case QueryChunkAvailability:
-				err := av.handleQueryChunkAvailability(msg)
-				if err != nil {
-					logger.Errorf("failed to handle query chunk availability: %w", err)
-				}
-			case StoreChunk:
-				err := av.handleStoreChunk(msg)
-				if err != nil {
-					logger.Errorf("failed to handle store chunk: %w", err)
-				}
-			case StoreAvailableData:
-				err := av.handleStoreAvailableData(msg)
-				if err != nil {
-					logger.Errorf("failed to handle store available data: %w", err)
-				}
-			case parachaintypes.ActiveLeavesUpdateSignal:
-				err := av.ProcessActiveLeavesUpdateSignal(msg)
-				if err != nil {
-					logger.Errorf("failed to process active leaves update signal: %w", err)
-				}
-			case parachaintypes.BlockFinalizedSignal:
-				err := av.ProcessBlockFinalizedSignal(msg)
-				if err != nil {
-					logger.Errorf("failed to process block finalized signal: %w", err)
-				}
+func (av *AvailabilityStoreSubsystem) processMessage(msg any) {
+	switch msg := msg.(type) {
+	case QueryAvailableData:
+		err := av.handleQueryAvailableData(msg)
+		if err != nil {
+			logger.Errorf("failed to handle available data: %w", err)
+		}
+	case QueryDataAvailability:
+		err := av.handleQueryDataAvailability(msg)
+		if err != nil {
+			logger.Errorf("failed to handle query data availability: %w", err)
+		}
+	case QueryChunk:
+		err := av.handleQueryChunk(msg)
+		if err != nil {
+			logger.Errorf("failed to handle query chunk: %w", err)
+		}
+	case QueryChunkSize:
+		err := av.handleQueryChunkSize(msg)
+		if err != nil {
+			logger.Errorf("failed to handle query chunk size: %w", err)
+		}
+	case QueryAllChunks:
+		err := av.handleQueryAllChunks(msg)
+		if err != nil {
+			logger.Errorf("failed to handle query all chunks: %w", err)
+		}
+	case QueryChunkAvailability:
+		err := av.handleQueryChunkAvailability(msg)
+		if err != nil {
+			logger.Errorf("failed to handle query chunk availability: %w", err)
+		}
+	case StoreChunk:
+		err := av.handleStoreChunk(msg)
+		if err != nil {
+			logger.Errorf("failed to handle store chunk: %w", err)
+		}
+	case StoreAvailableData:
+		err := av.handleStoreAvailableData(msg)
+		if err != nil {
+			logger.Errorf("failed to handle store available data: %w", err)
+		}
+	case parachaintypes.ActiveLeavesUpdateSignal:
+		err := av.ProcessActiveLeavesUpdateSignal(msg)
+		if err != nil {
+			logger.Errorf("failed to process active leaves update signal: %w", err)
+		}
+	case parachaintypes.BlockFinalizedSignal:
+		err := av.ProcessBlockFinalizedSignal(msg)
+		if err != nil {
+			logger.Errorf("failed to process block finalized signal: %w", err)
+		}
 
-			default:
-				if msg != nil {
-					// this error shouldn't happen, so we'll panic to catch it during development
-					panic(fmt.Sprintf("%s: %T", parachaintypes.ErrUnknownOverseerMessage.Error(), msg))
-				}
-			}
-		case <-av.ctx.Done():
-			if err := av.ctx.Err(); err != nil {
-				logger.Errorf("ctx error: %v", err)
-			}
-			av.wg.Done()
-			return
-		case <-time.After(av.pruningConfig.pruningInterval):
-			av.pruneAll()
+	default:
+		if msg != nil {
+			// this error shouldn't happen, so we'll panic to catch it during development
+			panic(fmt.Sprintf("%s: %T", parachaintypes.ErrUnknownOverseerMessage.Error(), msg))
 		}
 	}
 }
@@ -928,8 +913,7 @@ func (av *AvailabilityStoreSubsystem) processPruneKey(key []byte) error {
 }
 
 func (av *AvailabilityStoreSubsystem) Stop() {
-	av.cancel()
-	av.wg.Wait()
+	logger.Infof("Stopping availability store subsystem")
 }
 
 func (av *AvailabilityStoreSubsystem) deleteUnfinalizedHeight(writer database.Writer,

diff --git a/dot/parachain/availability-store/availability_store_test.go b/dot/parachain/availability-store/availability_store_test.go
@@ -968,7 +968,7 @@ func newTestHarness(t *testing.T) *testHarness {
 
 	require.NoError(t, err)
 
-	availabilityStore.OverseerToSubSystem = harness.overseer.RegisterSubsystem(availabilityStore)
+	harness.overseer.RegisterSubsystem(availabilityStore)
 
 	return harness
 }
@@ -1115,18 +1115,20 @@ func buildAvailableDataBranchesRoot(t *testing.T, numValidators uint32, availabl
 }
 
 func newTestOverseer() *testOverseer {
-	ctx := context.Background()
+	ctx, cancel := context.WithCancel(context.Background())
 
 	return &testOverseer{
 		ctx:                  ctx,
+		cancel:               cancel,
 		subsystems:           make(map[parachaintypes.Subsystem]chan any),
 		SubsystemsToOverseer: make(chan any),
 	}
 }
 
 type testOverseer struct {
-	ctx context.Context
-	wg  sync.WaitGroup
+	ctx    context.Context
+	cancel context.CancelFunc
+	wg     sync.WaitGroup
 
 	subsystems           map[parachaintypes.Subsystem]chan any
 	SubsystemsToOverseer chan any
@@ -1148,7 +1150,7 @@ func (to *testOverseer) Start() error {
 	for subsystem, overseerToSubSystem := range to.subsystems {
 		to.wg.Add(1)
 		go func(sub parachaintypes.Subsystem, overseerToSubSystem chan any) {
-			sub.Run(to.ctx, overseerToSubSystem, to.SubsystemsToOverseer)
+			sub.Run(to.ctx, overseerToSubSystem)
 			logger.Infof("subsystem %v stopped", sub)
 			to.wg.Done()
 		}(subsystem, overseerToSubSystem)

diff --git a/dot/parachain/availability-store/mock_runtime_instance_test.go b/dot/parachain/availability-store/mock_runtime_instance_test.go
diff --git a/dot/parachain/backing/candidate_backing.go b/dot/parachain/backing/candidate_backing.go
@@ -48,11 +48,7 @@ var (
 
 // CandidateBacking represents the state of the subsystem responsible for managing candidate backing.
 type CandidateBacking struct {
-	ctx    context.Context
-	cancel context.CancelFunc
-
 	SubSystemToOverseer chan<- any
-	OverseerToSubSystem <-chan any
 	// State tracked for all relay-parents backing work is ongoing for. This includes
 	// all active leaves.
 	//
@@ -199,18 +195,15 @@ type StatementMessage struct {
 
 // New creates a new CandidateBacking instance and initialises it with the provided overseer channel.
 func New(overseerChan chan<- any) *CandidateBacking {
-	ctx, cancel := context.WithCancel(context.Background())
 	return &CandidateBacking{
-		ctx:                 ctx,
-		cancel:              cancel,
 		SubSystemToOverseer: overseerChan,
 		perRelayParent:      map[common.Hash]*perRelayParentState{},
 		perCandidate:        map[parachaintypes.CandidateHash]*perCandidateState{},
 		perLeaf:             map[common.Hash]*activeLeafState{},
 	}
 }
 
-func (cb *CandidateBacking) Run(ctx context.Context, overseerToSubSystem chan any, subSystemToOverseer chan any) {
+func (cb *CandidateBacking) Run(ctx context.Context, overseerToSubSystem <-chan any) {
 	chRelayParentAndCommand := make(chan relayParentAndCommand, 1)
 
 	for {
@@ -219,26 +212,24 @@ func (cb *CandidateBacking) Run(ctx context.Context, overseerToSubSystem chan an
 			if err := cb.processValidatedCandidateCommand(rpAndCmd, chRelayParentAndCommand); err != nil {
 				logger.Errorf("processing validated candidated command: %s", err.Error())
 			}
-		case msg, ok := <-cb.OverseerToSubSystem:
+		case msg, ok := <-overseerToSubSystem:
 			if !ok {
 				return
 			}
 			if err := cb.processMessage(msg, chRelayParentAndCommand); err != nil {
 				logger.Errorf("processing message: %s", err.Error())
 			}
-		case <-cb.ctx.Done():
+		case <-ctx.Done():
 			close(chRelayParentAndCommand)
-			if err := cb.ctx.Err(); err != nil {
+			if err := ctx.Err(); err != nil {
 				logger.Errorf("ctx error: %s\n", err)
 			}
 			return
 		}
 	}
 }
 
-func (cb *CandidateBacking) Stop() {
-	cb.cancel()
-}
+func (cb *CandidateBacking) Stop() {}
 
 func (*CandidateBacking) Name() parachaintypes.SubSystemName {
 	return parachaintypes.CandidateBacking

diff --git a/dot/parachain/backing/get_backable_candidates.go b/dot/parachain/backing/get_backable_candidates.go
@@ -9,7 +9,7 @@ import (
 
 // handleGetBackableCandidatesMessage send back the backable candidates via the response channel
 func (cb *CandidateBacking) handleGetBackableCandidatesMessage(requestedCandidates GetBackableCandidatesMessage) {
-	var backedCandidates []*parachaintypes.BackedCandidate
+	backedCandidates := make([]*parachaintypes.BackedCandidate, 0, len(requestedCandidates.Candidates))
 
 	for _, candidate := range requestedCandidates.Candidates {
 		rpState, ok := cb.perRelayParent[candidate.CandidateRelayParent]
@@ -35,7 +35,11 @@ func (cb *CandidateBacking) handleGetBackableCandidatesMessage(requestedCandidat
 			continue
 		}
 
-		backed := attested.toBackedCandidate(&rpState.tableContext)
+		backed, err := attested.toBackedCandidate(&rpState.tableContext)
+		if err != nil {
+			logger.Debugf("converting attested candidate to backed candidate: %w", err)
+			continue
+		}
 		backedCandidates = append(backedCandidates, backed)
 	}