Add more check to safely truncate unnecessary data.

Signed-off-by: Kohei Tokunaga <[email protected]>

Author: ktock

crfs.go

@@ -1060,9 +1060,18 @@ func (h *nodeHandle) Read(ctx context.Context, req *fuse.ReadRequest, resp *fuse
 		if err != nil {
 			return err
 		}
-		if nr == 0 {
-			// Truncate unncessary data
-			chunkData = chunkData[req.Offset:]
+		if ce.ChunkOffset < req.Offset && req.Offset < ce.ChunkOffset+ce.ChunkSize {
+			
+			// Data between req.Offset and ce.ChunkOffset is not required, so we truncate it
+			// If we've already read first nr bytes we also don't need to read it again. It
+			// is possible to occur if we failed to copy full chunkData into resp.Data in a
+			// last loop and we are reading same chunk again.
+			off := req.Offset - ce.ChunkOffset + int64(nr)
+			size := int64(len(chunkData))
+			if off > size {
+				off = size
+			}
+			chunkData = chunkData[off:]
 		}
 		n := copy(resp.Data[nr:], chunkData)
 		nr += n

stargz/stargz.go

@@ -367,6 +367,9 @@ func (r *Reader) ChunkEntryForOffset(name string, offset int64) (e *TOCEntry, ok
 	}
 	ents := r.chunks[name]
 	if len(ents) < 2 {
+		if offset >= e.ChunkSize {
+			return nil, false
+		}
 		return e, true
 	}
 	i := sort.Search(len(ents), func(i int) bool {