From f1c76a82861f59e4fd03c7696a669553ae9d2f96 Mon Sep 17 00:00:00 2001
From: Michael Eischer <michael.eischer@fau.de>
Date: Wed, 5 Jun 2024 22:01:55 +0200
Subject: [PATCH] restore: fix corrupted sparse files

---
 internal/restorer/filerestorer.go  |  6 +++++
 internal/restorer/restorer_test.go | 38 ++++++++++++++++++++++++++++++
 2 files changed, 44 insertions(+)

diff --git a/internal/restorer/filerestorer.go b/internal/restorer/filerestorer.go
index 1e67debeb..f57d58598 100644
--- a/internal/restorer/filerestorer.go
+++ b/internal/restorer/filerestorer.go
@@ -153,6 +153,12 @@ func (r *fileRestorer) restoreFiles(ctx context.Context) error {
 			// in addition, a short chunk will never match r.zeroChunk which would prevent sparseness for short files
 			file.sparse = r.sparse
 		}
+		if file.state != nil {
+			// The restorer currently cannot punch new holes into an existing files.
+			// Thus sections that contained data but should be sparse after restoring
+			// the snapshot would still contain the old data resulting in a corrupt restore.
+			file.sparse = false
+		}
 
 		if err != nil {
 			// repository index is messed up, can't do anything
diff --git a/internal/restorer/restorer_test.go b/internal/restorer/restorer_test.go
index 0d43ffa9d..25ce668db 100644
--- a/internal/restorer/restorer_test.go
+++ b/internal/restorer/restorer_test.go
@@ -895,6 +895,44 @@ func TestRestorerSparseFiles(t *testing.T) {
 		len(zeros), blocks, 100*sparsity)
 }
 
+func TestRestorerSparseOverwrite(t *testing.T) {
+	baseSnapshot := Snapshot{
+		Nodes: map[string]Node{
+			"foo": File{Data: "content: new\n"},
+		},
+	}
+	var zero [14]byte
+	sparseSnapshot := Snapshot{
+		Nodes: map[string]Node{
+			"foo": File{Data: string(zero[:])},
+		},
+	}
+
+	repo := repository.TestRepository(t)
+	tempdir := filepath.Join(rtest.TempDir(t), "target")
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+
+	// base snapshot
+	sn, id := saveSnapshot(t, repo, baseSnapshot, noopGetGenericAttributes)
+	t.Logf("base snapshot saved as %v", id.Str())
+
+	res := NewRestorer(repo, sn, Options{Sparse: true})
+	err := res.RestoreTo(ctx, tempdir)
+	rtest.OK(t, err)
+
+	// sparse snapshot
+	sn, id = saveSnapshot(t, repo, sparseSnapshot, noopGetGenericAttributes)
+	t.Logf("base snapshot saved as %v", id.Str())
+
+	res = NewRestorer(repo, sn, Options{Sparse: true, Overwrite: OverwriteAlways})
+	err = res.RestoreTo(ctx, tempdir)
+	rtest.OK(t, err)
+	files, err := res.VerifyFiles(ctx, tempdir)
+	rtest.OK(t, err)
+	rtest.Equals(t, 1, files, "unexpected number of verified files")
+}
+
 func TestRestorerOverwriteBehavior(t *testing.T) {
 	baseTime := time.Now()
 	baseSnapshot := Snapshot{