From afeca99aef3f015258a7255fe5d7d020e604a1f9 Mon Sep 17 00:00:00 2001
From: Ivan Shapovalov <intelfx@intelfx.name>
Date: Sat, 30 Nov 2024 14:35:02 +0400
Subject: [PATCH] restorer: implement same-file tracking and cloning
 ("reflinks")

Note that "cloning" means either true block cloning via `ioctl(FICLONE)`
or any kind of local copy in general via fallback to `io.Copy()`.

TBD:
- fallback to restoring a file normally if it could not be cloned
- track all potential cloning sources for each file (every copy that was
  restored rather than cloned) and try to clone from each copy in turn
  (Why: imagine a set of duplicate files being restored N:M to a set of
   distinct filesystems or subvolumes or datasets, such that some pairs
   can be used as operands to a block cloning operation and some cannot)
- progress reporting (report how much space we have saved, or not)

Non-goals:
- cloning individual blobs via `ioctl(FICLONERANGE)`
  (Why: this is not going to work very well, if at all, due to blobs not
   being aligned to any kind of a fundamental block size, _and_ this
   will impact cloning entire files unless the latter is special-cased,
   which is exactly what is being done here.)
---
 internal/restorer/restorer.go | 39 +++++++++++++++++++++++++++++++++++
 1 file changed, 39 insertions(+)

diff --git a/internal/restorer/restorer.go b/internal/restorer/restorer.go
index acf411f24..bbeef5dd7 100644
--- a/internal/restorer/restorer.go
+++ b/internal/restorer/restorer.go
@@ -316,6 +316,24 @@ func (res *Restorer) restoreHardlinkAt(node *restic.Node, target, path, location
 	return res.restoreNodeMetadataTo(node, path, location)
 }
 
+func (res *Restorer) restoreReflink(node *restic.Node, target, path, location string) error {
+	cloned := true
+	if !res.opts.DryRun {
+		var err error
+		if err = fs.Remove(path); err != nil && !errors.Is(err, os.ErrNotExist) {
+			return errors.Wrap(err, "RemoveNode")
+		}
+		cloned, err = fs.Clone(target, path)
+		if err != nil {
+			return errors.WithStack(err)
+		}
+	}
+
+	res.opts.Progress.AddProgress(location, restoreui.ActionFileRestored, node.Size, node.Size)
+	// reflinked files *do* have separate metadata
+	return res.restoreNodeMetadataTo(node, path, location)
+}
+
 func (res *Restorer) ensureDir(target string) error {
 	if res.opts.DryRun {
 		return nil
@@ -358,6 +376,7 @@ func (res *Restorer) RestoreTo(ctx context.Context, dst string) (uint64, error)
 	}
 
 	idx := NewHardlinkIndex[string]()
+	refIdx := NewReflinkIndex()
 	filerestorer := newFileRestorer(dst, res.repo.LoadBlobsFromPack, res.repo.LookupBlob,
 		res.repo.Connections(), res.opts.Sparse, res.opts.Delete, res.repo.StartWarmup, res.opts.Progress)
 	filerestorer.Error = res.Error
@@ -397,6 +416,16 @@ func (res *Restorer) RestoreTo(ctx context.Context, dst string) (uint64, error)
 				idx.Add(node.Inode, node.DeviceID, location)
 			}
 
+			// do not bother reflinking empty files
+			if res.opts.Reflinks && node.Size > 0 {
+				refOrig, refIsOrig := refIdx.Put(location, node.Content)
+				if !refIsOrig {
+					debug.Log("reflink (deferring): orig=%s, link=%s", refOrig, location)
+					res.opts.Progress.AddFile(node.Size)
+					return nil
+				}
+			}
+
 			buf, err = res.withOverwriteCheck(ctx, node, target, location, false, buf, func(updateMetadataOnly bool, matches *fileState) error {
 				if updateMetadataOnly {
 					res.opts.Progress.AddSkippedFile(location, node.Size)
@@ -453,6 +482,16 @@ func (res *Restorer) RestoreTo(ctx context.Context, dst string) (uint64, error)
 				return err
 			}
 
+			if res.opts.Reflinks && node.Size > 0 {
+				if orig, hasOrig := refIdx.Get(node.Content); hasOrig && orig != location {
+					debug.Log("reflink (restoring): orig=%s, link=%s", orig, location)
+					_, err := res.withOverwriteCheck(ctx, node, target, location, false, nil, func(_ bool, _ *fileState) error {
+						return res.restoreReflink(node, filerestorer.targetPath(orig), target, location)
+					})
+					return err
+				}
+			}
+
 			if _, ok := res.hasRestoredFile(location); ok {
 				return res.restoreNodeMetadataTo(node, target, location)
 			}