From 3a2e21fee78f932eba19a4d20d0914176cc9f5c8 Mon Sep 17 00:00:00 2001
From: chrislu <chris.lu@gmail.com>
Date: Tue, 13 Aug 2024 12:01:46 -0700
Subject: [PATCH] fix EC reading on nLargeBlockRows

fix https://github.com/seaweedfs/seaweedfs/issues/5465
---
 weed/Makefile                            |  4 ++--
 weed/storage/erasure_coding/ec_locate.go | 17 +++++++----------
 weed/storage/erasure_coding/ec_test.go   | 12 ++++++++++++
 3 files changed, 21 insertions(+), 12 deletions(-)

diff --git a/weed/Makefile b/weed/Makefile
index 9a0553ad2..7d1a0d2ed 100644
--- a/weed/Makefile
+++ b/weed/Makefile
@@ -26,8 +26,8 @@ debug_server:
 	dlv --listen=:2345 --headless=true --api-version=2 --accept-multiclient exec ./weed -- server -dir=~/tmp/99 -filer -volume.port=8343 -s3 -volume.max=0 -master.volumeSizeLimitMB=1024 -volume.preStopSeconds=1
 
 debug_volume:
-	go build -gcflags="all=-N -l"
-	dlv --listen=:2345 --headless=true --api-version=2 --accept-multiclient exec ./weed -- volume -dir=~/tmp/100 -port 8564 -max=30 -preStopSeconds=2
+	go build -tags=5BytesOffset -gcflags="all=-N -l"
+	dlv --listen=:2345 --headless=true --api-version=2 --accept-multiclient exec ./weed -- volume -dir=/Users/chrislu/tmp/x/volume_data -port 8564 -max=1 -preStopSeconds=2
 
 debug_webdav:
 	go build -gcflags="all=-N -l"
diff --git a/weed/storage/erasure_coding/ec_locate.go b/weed/storage/erasure_coding/ec_locate.go
index be7e085e6..4b092695c 100644
--- a/weed/storage/erasure_coding/ec_locate.go
+++ b/weed/storage/erasure_coding/ec_locate.go
@@ -5,25 +5,22 @@ import (
 )
 
 type Interval struct {
-	BlockIndex          int
+	BlockIndex          int  // the index of the block in either the large blocks or the small blocks
 	InnerBlockOffset    int64
 	Size                types.Size
-	IsLargeBlock        bool
+	IsLargeBlock        bool // whether the block is a large block or a small block
 	LargeBlockRowsCount int
 }
 
 func LocateData(largeBlockLength, smallBlockLength int64, datSize int64, offset int64, size types.Size) (intervals []Interval) {
-	blockIndex, isLargeBlock, innerBlockOffset := locateOffset(largeBlockLength, smallBlockLength, datSize, offset)
-
-	// adding DataShardsCount*smallBlockLength to ensure we can derive the number of large block size from a shard size
-	nLargeBlockRows := int((datSize + DataShardsCount*smallBlockLength) / (largeBlockLength * DataShardsCount))
+	blockIndex, isLargeBlock, nLargeBlockRows, innerBlockOffset := locateOffset(largeBlockLength, smallBlockLength, datSize, offset)
 
 	for size > 0 {
 		interval := Interval{
 			BlockIndex:          blockIndex,
 			InnerBlockOffset:    innerBlockOffset,
 			IsLargeBlock:        isLargeBlock,
-			LargeBlockRowsCount: nLargeBlockRows,
+			LargeBlockRowsCount: int(nLargeBlockRows),
 		}
 
 		blockRemaining := largeBlockLength - innerBlockOffset
@@ -41,7 +38,7 @@ func LocateData(largeBlockLength, smallBlockLength int64, datSize int64, offset
 
 		size -= interval.Size
 		blockIndex += 1
-		if isLargeBlock && blockIndex == nLargeBlockRows*DataShardsCount {
+		if isLargeBlock && blockIndex == interval.LargeBlockRowsCount*DataShardsCount {
 			isLargeBlock = false
 			blockIndex = 0
 		}
@@ -51,9 +48,9 @@ func LocateData(largeBlockLength, smallBlockLength int64, datSize int64, offset
 	return
 }
 
-func locateOffset(largeBlockLength, smallBlockLength int64, datSize int64, offset int64) (blockIndex int, isLargeBlock bool, innerBlockOffset int64) {
+func locateOffset(largeBlockLength, smallBlockLength int64, datSize int64, offset int64) (blockIndex int, isLargeBlock bool, nLargeBlockRows int64, innerBlockOffset int64) {
 	largeRowSize := largeBlockLength * DataShardsCount
-	nLargeBlockRows := datSize / (largeBlockLength * DataShardsCount)
+	nLargeBlockRows = datSize / (largeBlockLength * DataShardsCount)
 
 	// if offset is within the large block area
 	if offset < nLargeBlockRows*largeRowSize {
diff --git a/weed/storage/erasure_coding/ec_test.go b/weed/storage/erasure_coding/ec_test.go
index aa46cf6d4..fc0adbe9f 100644
--- a/weed/storage/erasure_coding/ec_test.go
+++ b/weed/storage/erasure_coding/ec_test.go
@@ -3,6 +3,7 @@ package erasure_coding
 import (
 	"bytes"
 	"fmt"
+	"github.com/stretchr/testify/assert"
 	"math/rand"
 	"os"
 	"testing"
@@ -208,3 +209,14 @@ func (this Interval) sameAs(that Interval) bool {
 		this.BlockIndex == that.BlockIndex &&
 		this.Size == that.Size
 }
+
+func TestLocateData2(t *testing.T) {
+	intervals := LocateData(ErasureCodingLargeBlockSize, ErasureCodingSmallBlockSize, 32205678320, 21479557912, 4194339)
+	assert.Equal(t, intervals, []Interval{
+		{BlockIndex: 4, InnerBlockOffset: 527128, Size: 521448, IsLargeBlock: false, LargeBlockRowsCount: 2},
+		{BlockIndex: 5, InnerBlockOffset: 0, Size: 1048576, IsLargeBlock: false, LargeBlockRowsCount: 2},
+		{BlockIndex: 6, InnerBlockOffset: 0, Size: 1048576, IsLargeBlock: false, LargeBlockRowsCount: 2},
+		{BlockIndex: 7, InnerBlockOffset: 0, Size: 1048576, IsLargeBlock: false, LargeBlockRowsCount: 2},
+		{BlockIndex: 8, InnerBlockOffset: 0, Size: 527163, IsLargeBlock: false, LargeBlockRowsCount: 2},
+	})
+}