feat(statediff): refill directly from frozen IAVL

.github/workflows/build-rocksdb.yml

@@ -18,6 +18,42 @@ env:
   DOCKERFILE: Dockerfile.debank-rocksdb
 
 jobs:
+  build-statediff-rewriter:
+    if: github.event_name == 'pull_request'
+    runs-on: ubuntu-latest
+    permissions:
+      contents: read
+    steps:
+      - name: Free disk space
+        uses: jlumbroso/free-disk-space@54081f138730dfa15788a46383842cd2f914a1be # v1.3.1
+        with:
+          tool-cache: true
+          android: true
+          dotnet: true
+          haskell: true
+          large-packages: false
+          swap-storage: false
+      - uses: actions/checkout@v4
+      - uses: docker/setup-buildx-action@v3
+      - name: Build rewriter image without pushing
+        uses: docker/build-push-action@v6
+        with:
+          file: Dockerfile.statediff-rewriter
+          context: .
+          push: false
+          load: true
+          platforms: linux/amd64
+          tags: cronos-statediff-rewriter:pr-${{ github.event.pull_request.number }}
+          build-args: |
+            GIT_COMMIT=${{ github.sha }}
+          provenance: false
+      - name: Verify rewriter runtime and provenance
+        run: |
+          image="cronos-statediff-rewriter:pr-${{ github.event.pull_request.number }}"
+          test "$(docker image inspect --format '{{ index .Config.Labels "org.opencontainers.image.revision" }}' "${image}")" = "${{ github.sha }}"
+          test "$(docker run --rm "${image}" --version)" = "${{ github.sha }}"
+          docker run --rm "${image}" --help
+
   build-rocksdb:
     strategy:
       fail-fast: false
@@ -33,7 +69,7 @@ jobs:
       id-token: write
     steps:
       - name: Free disk space
-        uses: jlumbroso/free-disk-space@main
+        uses: jlumbroso/free-disk-space@54081f138730dfa15788a46383842cd2f914a1be # v1.3.1
         with:
           tool-cache: true
           android: true

CHANGELOG.md

@@ -4,6 +4,9 @@
 
 ### Improvements
 
+* [Chaintable/cronos#6](https://github.com/Chaintable/cronos/pull/6) feat: add state-diff refill from frozen IAVL, bounded changeset batches, and streaming changeset directories.
+* [Chaintable/cronos#6](https://github.com/Chaintable/cronos/pull/6) feat: add bounded production segments for historical state-diff recovery.
+* [Chaintable/cronos#3](https://github.com/Chaintable/cronos/pull/3) feat: add fail-closed tooling for canonical historical state-diff recovery.
 * [#2095](https://github.com/crypto-org-chain/cronos/pull/2095) feat: add hardcoded unblockable address list that is filtered out of both the mempool and consensus blocklists.
 
 ### Bug fixes

Dockerfile.statediff-rewriter

@@ -0,0 +1,35 @@
+FROM golang:1.25-bookworm AS go-builder
+
+RUN apt-get update && \
+    apt-get install -y libgflags-dev libsnappy-dev zlib1g-dev libbz2-dev liblz4-dev libzstd-dev && \
+    rm -rf /var/lib/apt/lists/*
+
+RUN git clone https://github.com/facebook/rocksdb.git /tmp/rocksdb && \
+    cd /tmp/rocksdb && \
+    git checkout v10.5.1 && \
+    PORTABLE=1 DISABLE_WARNING_AS_ERROR=1 make -j$(nproc) install-shared && \
+    ldconfig && \
+    rm -rf /tmp/rocksdb
+
+WORKDIR /code
+COPY . /code/
+ARG GIT_COMMIT
+RUN test -n "${GIT_COMMIT}" && \
+    GOPRIVATE='github.com/Chaintable/*' make build-statediff-rewriter COMMIT="${GIT_COMMIT}"
+
+FROM ubuntu:22.04
+
+ARG GIT_COMMIT
+LABEL org.opencontainers.image.revision="${GIT_COMMIT}"
+
+RUN apt-get update && \
+    apt-get install -y libsnappy1v5 libgflags2.2 zlib1g libbz2-1.0 liblz4-1 libzstd1 ca-certificates && \
+    rm -rf /var/lib/apt/lists/*
+
+COPY --from=go-builder /usr/local/lib/ /usr/local/lib/
+RUN ldconfig
+COPY --from=go-builder /code/build/statediff-rewriter /usr/bin/statediff-rewriter
+
+WORKDIR /work
+ENTRYPOINT ["/usr/bin/statediff-rewriter"]
+CMD ["--help"]

Makefile

@@ -93,26 +93,32 @@ build_tags_comma_sep := $(subst $(whitespace),$(comma),$(build_tags))
 ldflags += -X github.com/cosmos/cosmos-sdk/version.Name=cronos \
 	-X github.com/cosmos/cosmos-sdk/version.AppName=cronosd \
 	-X github.com/cosmos/cosmos-sdk/version.Version=$(VERSION) \
-	-X github.com/cosmos/cosmos-sdk/version.Commit=$(COMMIT) \
-	-X github.com/cosmos/cosmos-sdk/version.BuildTags=$(build_tags_comma_sep)
+	-X github.com/cosmos/cosmos-sdk/version.Commit=$(COMMIT)
 
-BUILD_FLAGS := -tags "$(build_tags)" -ldflags '$(ldflags)'
+BUILD_FLAGS := -tags "$(build_tags)" -ldflags '$(ldflags) -X github.com/cosmos/cosmos-sdk/version.BuildTags=$(build_tags_comma_sep)'
+statediff_rewriter_build_tags := $(strip $(filter-out rocksdb grocksdb_clean_link,$(build_tags)) rocksdb grocksdb_clean_link)
+statediff_rewriter_build_tags_comma_sep := $(subst $(whitespace),$(comma),$(statediff_rewriter_build_tags))
+STATEDIFF_REWRITER_BUILD_FLAGS := -tags "$(statediff_rewriter_build_tags)" -ldflags '$(ldflags) -X github.com/cosmos/cosmos-sdk/version.BuildTags=$(statediff_rewriter_build_tags_comma_sep)'
 # check for nostrip option
 ifeq (,$(findstring nostrip,$(COSMOS_BUILD_OPTIONS)))
   BUILD_FLAGS += -trimpath
+  STATEDIFF_REWRITER_BUILD_FLAGS += -trimpath
 endif
 
 all: build
 build: check-network print-ledger go.sum
 	@go build -mod=readonly $(BUILD_FLAGS) -o $(BUILDDIR)/cronosd ./cmd/cronosd
 
+build-statediff-rewriter: check-network go.sum
+	@CGO_ENABLED=1 go build -mod=readonly $(STATEDIFF_REWRITER_BUILD_FLAGS) -o $(BUILDDIR)/statediff-rewriter ./cmd/statediff-rewriter
+
 install: check-network print-ledger go.sum
 	@go install -mod=readonly $(BUILD_FLAGS) ./cmd/cronosd
 
 test:
 	@go test -tags=objstore -v -mod=readonly $(PACKAGES) -coverprofile=$(COVERAGE) -covermode=atomic
 
-.PHONY: all clean build install test
+.PHONY: all clean build build-statediff-rewriter install test
 
 clean:
 	rm -rf $(BUILDDIR)/

cmd/cronosd/cmd/rollback_to_version.go

@@ -5,11 +5,11 @@ import (
 	"strconv"
 
 	dbm "github.com/cosmos/cosmos-db"
-	"github.com/cosmos/cosmos-sdk/server"
-	"github.com/spf13/cobra"
-
 	"github.com/crypto-org-chain/cronos/app"
 	"github.com/crypto-org-chain/cronos/cmd/cronosd/opendb"
+	"github.com/spf13/cobra"
+
+	"github.com/cosmos/cosmos-sdk/server"
 )
 
 // RollbackToVersionCmd force-rolls the application multistore back to an explicit

cmd/statediff-rewriter/archive.go

@@ -0,0 +1,98 @@
+package main
+
+import (
+	"fmt"
+	"os"
+	"path/filepath"
+	"strings"
+
+	dbm "github.com/cosmos/cosmos-db"
+	"github.com/cosmos/iavl"
+	iavldb "github.com/cosmos/iavl/db"
+	"github.com/crypto-org-chain/cronos/cmd/cronosd/opendb"
+
+	"cosmossdk.io/log"
+	"cosmossdk.io/store/metrics"
+	"cosmossdk.io/store/rootmulti"
+	storetypes "cosmossdk.io/store/types"
+	"cosmossdk.io/store/wrapper"
+)
+
+type archiveReader struct {
+	home  string
+	db    dbm.DB
+	store *rootmulti.Store
+}
+
+func openArchive(home string) (*archiveReader, error) {
+	if !rocksDBBuild {
+		return nil, fmt.Errorf("statediff-rewriter requires build tags rocksdb and grocksdb_clean_link")
+	}
+	databasePath := filepath.Join(home, "data", "application.db")
+	if _, err := requireReadOnlyFilesystem(databasePath, "archive database"); err != nil {
+		return nil, err
+	}
+	db, err := opendb.OpenReadOnlyDB(home, dbm.RocksDBBackend)
+	if err != nil {
+		return nil, err
+	}
+	return &archiveReader{home: home, db: db, store: rootmulti.NewStore(db, log.NewNopLogger(), metrics.NewNoOpMetrics())}, nil
+}
+
+func (a *archiveReader) Close() error { return a.db.Close() }
+
+func (a *archiveReader) evmIAVLDB() iavldb.DB {
+	return wrapper.NewDBWrapper(dbm.NewPrefixDB(a.db, []byte(evmIAVLPrefix)))
+}
+
+func (a *archiveReader) evmStateChangeSource(cacheSize int) keyRangeStateChangeSource {
+	return iavl.NewImmutableTree(a.evmIAVLDB(), cacheSize, true, log.NewNopLogger())
+}
+
+func (a *archiveReader) evmLegacyLatestVersion() (int64, error) {
+	return legacyLatestRootVersion(a.evmIAVLDB())
+}
+
+func (a *archiveReader) commitInfo(version int64) (*storetypes.CommitInfo, error) {
+	info, err := a.store.GetCommitInfo(version)
+	if err != nil {
+		return nil, fmt.Errorf("read commit info %d: %w", version, err)
+	}
+	if info.Version != version {
+		return nil, fmt.Errorf("commit info %d has version %d", version, info.Version)
+	}
+	return info, nil
+}
+
+func (a *archiveReader) identity() (archiveIdentity, error) {
+	databaseIdentity, err := readRocksDBIdentity(a.home)
+	if err != nil {
+		return archiveIdentity{}, err
+	}
+	latest := rootmulti.GetLatestVersion(a.db)
+	info, err := a.commitInfo(latest)
+	if err != nil {
+		return archiveIdentity{}, err
+	}
+	return archiveIdentity{
+		Home: a.home, DatabaseIdentity: databaseIdentity,
+		LatestVersion: latest, FinalCommitHash: commonHash(info.Hash()),
+	}, nil
+}
+
+func readRocksDBIdentity(home string) (string, error) {
+	path := filepath.Join(home, "data", "application.db", "IDENTITY")
+	body, err := os.ReadFile(path)
+	if err != nil {
+		return "", fmt.Errorf("read RocksDB identity %s: %w", path, err)
+	}
+	fields := strings.Fields(string(body))
+	if len(fields) != 1 {
+		return "", fmt.Errorf("RocksDB identity %s must contain exactly one value", path)
+	}
+	return fields[0], nil
+}
+
+func commonHash(body []byte) string {
+	return fmt.Sprintf("0x%x", body)
+}

cmd/statediff-rewriter/archive_test.go

@@ -0,0 +1,24 @@
+package main
+
+import (
+	"os"
+	"path/filepath"
+	"testing"
+
+	"github.com/stretchr/testify/require"
+)
+
+func TestReadRocksDBIdentity(t *testing.T) {
+	home := t.TempDir()
+	dir := filepath.Join(home, "data", "application.db")
+	require.NoError(t, os.MkdirAll(dir, 0o755))
+	require.NoError(t, os.WriteFile(filepath.Join(dir, "IDENTITY"), []byte("db-123\n"), 0o600))
+
+	identity, err := readRocksDBIdentity(home)
+	require.NoError(t, err)
+	require.Equal(t, "db-123", identity)
+
+	require.NoError(t, os.WriteFile(filepath.Join(dir, "IDENTITY"), []byte("db-123\ndb-456\n"), 0o600))
+	_, err = readRocksDBIdentity(home)
+	require.ErrorContains(t, err, "exactly one value")
+}