From 08e4b56a8abaaadb0701979e4bd857fb8c50c776 Mon Sep 17 00:00:00 2001
From: Chris Lu <chris.lu@gmail.com>
Date: Sun, 26 Jan 2020 13:01:11 -0800
Subject: [PATCH] mount: able to handle large git clone

---
 weed/filesys/dirty_page.go               |  7 +----
 weed/filesys/dirty_page_interval.go      | 34 +++++++++++---------
 weed/filesys/dirty_page_interval_test.go | 40 ++++++++++++++++++++++++
 3 files changed, 60 insertions(+), 21 deletions(-)
 create mode 100644 weed/filesys/dirty_page_interval_test.go

diff --git a/weed/filesys/dirty_page.go b/weed/filesys/dirty_page.go
index f1532a6a0..7a41e371e 100644
--- a/weed/filesys/dirty_page.go
+++ b/weed/filesys/dirty_page.go
@@ -44,12 +44,7 @@ func (pages *ContinuousDirtyPages) AddPage(ctx context.Context, offset int64, da
 		return pages.flushAndSave(ctx, offset, data)
 	}
 
-	hasOverlap := pages.intervals.AddInterval(data, offset)
-	if hasOverlap {
-		chunks, err = pages.saveExistingPagesToStorage(ctx)
-		pages.intervals.AddInterval(data, offset)
-		return
-	}
+	pages.intervals.AddInterval(data, offset)
 
 	var chunk *filer_pb.FileChunk
 	var hasSavedData bool
diff --git a/weed/filesys/dirty_page_interval.go b/weed/filesys/dirty_page_interval.go
index c64196cdf..77fab75ef 100644
--- a/weed/filesys/dirty_page_interval.go
+++ b/weed/filesys/dirty_page_interval.go
@@ -65,7 +65,10 @@ func (c *ContinuousIntervals) TotalSize() (total int64) {
 	return
 }
 
-func (c *ContinuousIntervals) AddInterval(data []byte, offset int64) (hasOverlap bool) {
+func (c *ContinuousIntervals) AddInterval(data []byte, offset int64) {
+
+	// TODO AddInterval needs to handle all possible out of order writes
+
 	interval := &IntervalNode{Data: data, Offset: offset, Size: int64(len(data))}
 
 	var prevList, nextList *IntervalLinkedList
@@ -75,6 +78,10 @@ func (c *ContinuousIntervals) AddInterval(data []byte, offset int64) (hasOverlap
 			nextList = list
 			break
 		}
+		if list.Head.Offset < interval.Offset+interval.Size && interval.Offset+interval.Size <= list.Head.Offset+list.Size() {
+			glog.V(0).Infof("unexpected [%d,%d) overlaps [%d,%d)", interval.Offset, interval.Offset+interval.Size, list.Head.Offset, list.Head.Offset+list.Size())
+			break
+		}
 	}
 
 	for _, list := range c.lists {
@@ -84,20 +91,17 @@ func (c *ContinuousIntervals) AddInterval(data []byte, offset int64) (hasOverlap
 			break
 		}
 		if list.Head.Offset <= offset && offset < list.Head.Offset+list.Size() {
-			if list.Tail.Offset <= offset {
-				dataStartIndex := list.Tail.Offset + list.Tail.Size - offset
-				glog.V(4).Infof("overlap data new [0,%d) same=%v", dataStartIndex, bytes.Compare(interval.Data[0:dataStartIndex], list.Tail.Data[len(list.Tail.Data)-int(dataStartIndex):]))
-				interval.Data = interval.Data[dataStartIndex:]
-				interval.Size -= dataStartIndex
-				interval.Offset = offset + dataStartIndex
-				glog.V(4).Infof("overlapping append as [%d,%d) dataSize=%d", interval.Offset, interval.Offset+interval.Size, len(interval.Data))
-				list.addNodeToTail(interval)
-				prevList = list
-				break
-			}
-			glog.V(4).Infof("overlapped! interval is [%d,%d) dataSize=%d", interval.Offset, interval.Offset+interval.Size, len(interval.Data))
-			hasOverlap = true
-			return
+
+			// the new interval overwrites the old tail
+			dataStartIndex := list.Tail.Offset + list.Tail.Size - offset
+			glog.V(4).Infof("overlap data new [0,%d) same=%v", dataStartIndex, bytes.Compare(interval.Data[0:dataStartIndex], list.Tail.Data[len(list.Tail.Data)-int(dataStartIndex):]))
+			list.Tail.Data = list.Tail.Data[:len(list.Tail.Data)-int(dataStartIndex)]
+			list.Tail.Size -= dataStartIndex
+			glog.V(4).Infof("overlapping append as [%d,%d) dataSize=%d", interval.Offset, interval.Offset+interval.Size, len(interval.Data))
+
+			list.addNodeToTail(interval)
+			prevList = list
+			break
 		}
 	}
 
diff --git a/weed/filesys/dirty_page_interval_test.go b/weed/filesys/dirty_page_interval_test.go
new file mode 100644
index 000000000..4f62f90c9
--- /dev/null
+++ b/weed/filesys/dirty_page_interval_test.go
@@ -0,0 +1,40 @@
+package filesys
+
+import (
+	"bytes"
+	"testing"
+)
+
+func TestContinuousIntervals_AddInterval(t *testing.T) {
+
+	c := &ContinuousIntervals{}
+
+	// 25, 25, 25
+	c.AddInterval(getBytes(25, 3), 0)
+	//  _,  _, 23, 23, 23, 23
+	c.AddInterval(getBytes(23, 4), 2)
+
+	expectedData(t, c, 0, 25, 25, 23, 23, 23, 23)
+}
+
+func expectedData(t *testing.T, c *ContinuousIntervals, offset int, data ...byte) {
+	start, stop := int64(offset), int64(offset+len(data))
+	for _, list := range c.lists {
+		nodeStart, nodeStop := max(start, list.Head.Offset), min(stop, list.Head.Offset+list.Size())
+		if nodeStart < nodeStop {
+			buf := make([]byte, nodeStop-nodeStart)
+			list.ReadData(buf, nodeStart, nodeStop)
+			if bytes.Compare(buf, data[nodeStart-start:nodeStop-start]) != 0 {
+				t.Errorf("expected %v actual %v", data[nodeStart-start:nodeStop-start], buf)
+			}
+		}
+	}
+}
+
+func getBytes(content byte, length int) []byte {
+	data := make([]byte, length)
+	for i := 0; i < length; i++ {
+		data[i] = content
+	}
+	return data
+}