all: Reorder sequences for better rename detection (#6574)

5 files changed, 163 insertions, 10 deletions

diff --git a/lib/db/keyer.go b/lib/db/keyer.go
index d60644e447..88103eb07e 100644
--- a/lib/db/keyer.go
+++ b/lib/db/keyer.go
@@ -62,6 +62,9 @@ const (
 
 	// KeyTypeBlockList <block list hash> = BlockList
 	KeyTypeBlockList = 13
+
+	// KeyTypeBlockListMap <int32 folder ID> <block list hash> <file name> = <nothing>
+	KeyTypeBlockListMap = 14
 )
 
 type keyer interface {
@@ -79,6 +82,8 @@ type keyer interface {
 	// block map key stuff (former BlockMap)
 	GenerateBlockMapKey(key, folder, hash, name []byte) (blockMapKey, error)
 	NameFromBlockMapKey(key []byte) []byte
+	GenerateBlockListMapKey(key, folder, hash, name []byte) (blockListMapKey, error)
+	NameFromBlockListMapKey(key []byte) []byte
 
 	// file need index
 	GenerateNeedFileKey(key, folder, name []byte) (needFileKey, error)
@@ -203,6 +208,29 @@ func (k blockMapKey) WithoutHashAndName() []byte {
 	return k[:keyPrefixLen+keyFolderLen]
 }
 
+type blockListMapKey []byte
+
+func (k defaultKeyer) GenerateBlockListMapKey(key, folder, hash, name []byte) (blockListMapKey, error) {
+	folderID, err := k.folderIdx.ID(folder)
+	if err != nil {
+		return nil, err
+	}
+	key = resize(key, keyPrefixLen+keyFolderLen+keyHashLen+len(name))
+	key[0] = KeyTypeBlockListMap
+	binary.BigEndian.PutUint32(key[keyPrefixLen:], folderID)
+	copy(key[keyPrefixLen+keyFolderLen:], hash)
+	copy(key[keyPrefixLen+keyFolderLen+keyHashLen:], name)
+	return key, nil
+}
+
+func (k defaultKeyer) NameFromBlockListMapKey(key []byte) []byte {
+	return key[keyPrefixLen+keyFolderLen+keyHashLen:]
+}
+
+func (k blockListMapKey) WithoutHashAndName() []byte {
+	return k[:keyPrefixLen+keyFolderLen]
+}
+
 type needFileKey []byte
 
 func (k needFileKey) WithoutName() []byte {
diff --git a/lib/db/lowlevel.go b/lib/db/lowlevel.go
index d7fc8d4c81..985aa29eb1 100644
--- a/lib/db/lowlevel.go
+++ b/lib/db/lowlevel.go
@@ -195,6 +195,7 @@ func (db *Lowlevel) updateLocalFiles(folder []byte, fs []protocol.FileInfo, meta
 		if ok && unchanged(f, ef) {
 			continue
 		}
+		blocksHashSame := ok && bytes.Equal(ef.BlocksHash, f.BlocksHash)
 
 		if ok {
 			if !ef.IsDirectory() && !ef.IsDeleted() && !ef.IsInvalid() {
@@ -207,6 +208,15 @@ func (db *Lowlevel) updateLocalFiles(folder []byte, fs []protocol.FileInfo, meta
 						return err
 					}
 				}
+				if !blocksHashSame {
+					keyBuf, err := db.keyer.GenerateBlockListMapKey(keyBuf, folder, ef.BlocksHash, name)
+					if err != nil {
+						return err
+					}
+					if err = t.Delete(keyBuf); err != nil {
+						return err
+					}
+				}
 			}
 
 			keyBuf, err = db.keyer.GenerateSequenceKey(keyBuf, folder, ef.SequenceNo())
@@ -260,6 +270,15 @@ func (db *Lowlevel) updateLocalFiles(folder []byte, fs []protocol.FileInfo, meta
 					return err
 				}
 			}
+			if !blocksHashSame {
+				keyBuf, err := db.keyer.GenerateBlockListMapKey(keyBuf, folder, f.BlocksHash, name)
+				if err != nil {
+					return err
+				}
+				if err = t.Put(keyBuf, nil); err != nil {
+					return err
+				}
+			}
 		}
 
 		if err := t.Checkpoint(func() error {
@@ -296,7 +315,7 @@ func (db *Lowlevel) dropFolder(folder []byte) error {
 	}
 
 	// Remove all sequences related to the folder
-	k1, err := db.keyer.GenerateSequenceKey(nil, folder, 0)
+	k1, err := db.keyer.GenerateSequenceKey(k0, folder, 0)
 	if err != nil {
 		return err
 	}
@@ -305,7 +324,7 @@ func (db *Lowlevel) dropFolder(folder []byte) error {
 	}
 
 	// Remove all items related to the given folder from the global bucket
-	k2, err := db.keyer.GenerateGlobalVersionKey(nil, folder, nil)
+	k2, err := db.keyer.GenerateGlobalVersionKey(k1, folder, nil)
 	if err != nil {
 		return err
 	}
@@ -314,7 +333,7 @@ func (db *Lowlevel) dropFolder(folder []byte) error {
 	}
 
 	// Remove all needs related to the folder
-	k3, err := db.keyer.GenerateNeedFileKey(nil, folder, nil)
+	k3, err := db.keyer.GenerateNeedFileKey(k2, folder, nil)
 	if err != nil {
 		return err
 	}
@@ -323,7 +342,7 @@ func (db *Lowlevel) dropFolder(folder []byte) error {
 	}
 
 	// Remove the blockmap of the folder
-	k4, err := db.keyer.GenerateBlockMapKey(nil, folder, nil, nil)
+	k4, err := db.keyer.GenerateBlockMapKey(k3, folder, nil, nil)
 	if err != nil {
 		return err
 	}
@@ -331,6 +350,14 @@ func (db *Lowlevel) dropFolder(folder []byte) error {
 		return err
 	}
 
+	k5, err := db.keyer.GenerateBlockListMapKey(k4, folder, nil, nil)
+	if err != nil {
+		return err
+	}
+	if err := t.deleteKeyPrefix(k5.WithoutHashAndName()); err != nil {
+		return err
+	}
+
 	return t.Commit()
 }
 
@@ -383,6 +410,13 @@ func (db *Lowlevel) dropDeviceFolder(device, folder []byte, meta *metadataTracke
 		if err := t.deleteKeyPrefix(key.WithoutHashAndName()); err != nil {
 			return err
 		}
+		key2, err := db.keyer.GenerateBlockListMapKey(key, folder, nil, nil)
+		if err != nil {
+			return err
+		}
+		if err := t.deleteKeyPrefix(key2.WithoutHashAndName()); err != nil {
+			return err
+		}
 	}
 	return t.Commit()
 }
diff --git a/lib/db/schemaupdater.go b/lib/db/schemaupdater.go
index e68e733c24..f973b2f9f3 100644
--- a/lib/db/schemaupdater.go
+++ b/lib/db/schemaupdater.go
@@ -22,9 +22,9 @@ import (
 //   6: v0.14.50
 //   7: v0.14.53
 //   8-9: v1.4.0
-//   10: v1.6.0
+//   10-11: v1.6.0
 const (
-	dbVersion             = 10
+	dbVersion             = 11
 	dbMinSyncthingVersion = "v1.6.0"
 )
 
@@ -85,8 +85,9 @@ func (db *schemaUpdater) updateSchema() error {
 		{5, db.updateSchemaTo5},
 		{6, db.updateSchema5to6},
 		{7, db.updateSchema6to7},
-		{9, db.updateSchemato9},
-		{10, db.updateSchemato10},
+		{9, db.updateSchemaTo9},
+		{10, db.updateSchemaTo10},
+		{11, db.updateSchemaTo11},
 	}
 
 	for _, m := range migrations {
@@ -450,7 +451,7 @@ func (db *schemaUpdater) updateSchema6to7(_ int) error {
 	return t.Commit()
 }
 
-func (db *schemaUpdater) updateSchemato9(prev int) error {
+func (db *schemaUpdater) updateSchemaTo9(prev int) error {
 	// Loads and rewrites all files with blocks, to deduplicate block lists.
 	// Checks for missing or incorrect sequence entries and rewrites those.
 
@@ -499,7 +500,7 @@ func (db *schemaUpdater) updateSchemato9(prev int) error {
 	return t.Commit()
 }
 
-func (db *schemaUpdater) updateSchemato10(_ int) error {
+func (db *schemaUpdater) updateSchemaTo10(_ int) error {
 	t, err := db.newReadWriteTransaction()
 	if err != nil {
 		return err
@@ -568,3 +569,44 @@ func (db *schemaUpdater) updateSchemato10(_ int) error {
 
 	return t.Commit()
 }
+
+func (db *schemaUpdater) updateSchemaTo11(_ int) error {
+	// Populates block list map for every folder.
+
+	t, err := db.newReadWriteTransaction()
+	if err != nil {
+		return err
+	}
+	defer t.close()
+
+	var dk []byte
+	for _, folderStr := range db.ListFolders() {
+		folder := []byte(folderStr)
+		var putErr error
+		err := t.withHave(folder, protocol.LocalDeviceID[:], nil, true, func(fi FileIntf) bool {
+			f := fi.(FileInfoTruncated)
+			if f.IsDirectory() || f.IsDeleted() || f.IsInvalid() || f.BlocksHash == nil {
+				return true
+			}
+
+			name := []byte(f.FileName())
+			dk, putErr = db.keyer.GenerateBlockListMapKey(dk, folder, f.BlocksHash, name)
+			if putErr != nil {
+				return false
+			}
+
+			if putErr = t.Put(dk, nil); putErr != nil {
+				return false
+			}
+			putErr = t.Checkpoint()
+			return putErr == nil
+		})
+		if putErr != nil {
+			return putErr
+		}
+		if err != nil {
+			return err
+		}
+	}
+	return t.Commit()
+}
diff --git a/lib/db/set.go b/lib/db/set.go
index 69276b0e66..365a9ecf6f 100644
--- a/lib/db/set.go
+++ b/lib/db/set.go
@@ -365,6 +365,13 @@ func (s *Snapshot) RemoteNeedFolderFiles(device protocol.DeviceID, page, perpage
 	return files
 }
 
+func (s *Snapshot) WithBlocksHash(hash []byte, fn Iterator) {
+	l.Debugf(`%s WithBlocksHash("%x")`, s.folder, hash)
+	if err := s.t.withBlocksHash([]byte(s.folder), hash, nativeFileIterator(fn)); err != nil && !backend.IsClosed(err) {
+		panic(err)
+	}
+}
+
 func (s *FileSet) Sequence(device protocol.DeviceID) int64 {
 	return s.meta.Sequence(device)
 }
diff --git a/lib/db/transactions.go b/lib/db/transactions.go
index 1c40eada65..36459d19e0 100644
--- a/lib/db/transactions.go
+++ b/lib/db/transactions.go
@@ -9,6 +9,7 @@ package db
 import (
 	"bytes"
 	"errors"
+	"github.com/syncthing/syncthing/lib/osutil"
 
 	"github.com/syncthing/syncthing/lib/db/backend"
 	"github.com/syncthing/syncthing/lib/protocol"
@@ -304,6 +305,47 @@ func (t *readOnlyTransaction) withGlobal(folder, prefix []byte, truncate bool, f
 	return dbi.Error()
 }
 
+func (t *readOnlyTransaction) withBlocksHash(folder, hash []byte, iterator Iterator) error {
+	key, err := t.keyer.GenerateBlockListMapKey(nil, folder, hash, nil)
+	if err != nil {
+		return err
+	}
+
+	iter, err := t.NewPrefixIterator(key)
+	if err != nil {
+		return err
+	}
+	defer iter.Release()
+
+	for iter.Next() {
+		file := string(t.keyer.NameFromBlockListMapKey(iter.Key()))
+		f, ok, err := t.getFile(folder, protocol.LocalDeviceID[:], []byte(osutil.NormalizedFilename(file)))
+		if err != nil {
+			return err
+		}
+		if !ok {
+			continue
+		}
+		f.Name = osutil.NativeFilename(f.Name)
+
+		if !bytes.Equal(f.BlocksHash, hash) {
+			l.Warnf("Mismatching block map list hashes: got %x expected %x", f.BlocksHash, hash)
+			continue
+		}
+
+		if f.IsDeleted() || f.IsInvalid() || f.IsDirectory() || f.IsSymlink() {
+			l.Warnf("Found something of unexpected type in block list map: %s", f)
+			continue
+		}
+
+		if !iterator(f) {
+			break
+		}
+	}
+
+	return iter.Error()
+}
+
 func (t *readOnlyTransaction) availability(folder, file []byte) ([]protocol.DeviceID, error) {
 	vl, err := t.getGlobalVersions(nil, folder, file)
 	if backend.IsNotFound(err) {