handle gaps in blockgroups

2 files changed, 53 insertions, 5 deletions

diff --git a/lib/btrfsprogs/btrfsinspect/scanforextents/csums.go b/lib/btrfsprogs/btrfsinspect/scanforextents/csums.go
index 5ec87d9..c1a1597 100644
--- a/lib/btrfsprogs/btrfsinspect/scanforextents/csums.go
+++ b/lib/btrfsprogs/btrfsinspect/scanforextents/csums.go
@@ -22,6 +22,7 @@ import (
 	"git.lukeshu.com/btrfs-progs-ng/lib/btrfs/btrfssum"
 	"git.lukeshu.com/btrfs-progs-ng/lib/btrfs/btrfsvol"
 	"git.lukeshu.com/btrfs-progs-ng/lib/btrfsprogs/btrfsutil"
+	"git.lukeshu.com/btrfs-progs-ng/lib/diskio"
 )
 
 const CSumBlockSize = 4 * 1024
@@ -81,6 +82,54 @@ func (run SumRun[Addr]) Walk(ctx context.Context, fn func(Addr, ShortSum) error)
 	return nil
 }
 
+// SumRunWithGaps ////////////////////////////////////////////////////
+
+type SumRunWithGaps[Addr btrfsvol.IntAddr[Addr]] []SumRun[Addr]
+
+func (sg SumRunWithGaps[Addr]) end() Addr {
+	return sg[len(sg)-1].Addr.Add(sg[len(sg)-1].Size())
+}
+
+func (sg SumRunWithGaps[Addr]) Size() btrfsvol.AddrDelta {
+	if len(sg) == 0 {
+		return 0
+	}
+	return sg.end().Sub(sg[0].Addr)
+}
+
+func (sg SumRunWithGaps[Addr]) NumSums() int {
+	return int(sg.Size() / CSumBlockSize)
+}
+
+func (sg SumRunWithGaps[Addr]) SumForAddr(addr Addr) (ShortSum, error) {
+	if len(sg) == 0 {
+		return "", io.EOF
+	}
+	if addr < sg[0].Addr || addr >= sg.end() {
+		return "", io.EOF
+	}
+	for _, run := range sg {
+		if run.Addr > addr {
+			return "", diskio.ErrWildcard
+		}
+		if run.Addr.Add(run.Size()) <= addr {
+			continue
+		}
+		off := int((addr-run.Addr)/CSumBlockSize) * run.ChecksumSize
+		return ShortSum(run.Sums[off : off+run.ChecksumSize]), nil
+	}
+	return "", diskio.ErrWildcard
+}
+
+// Get implements diskio.Sequence[int, ShortSum]
+func (sg SumRunWithGaps[Addr]) Get(sumIdx int64) (ShortSum, error) {
+	if len(sg) == 0 {
+		return "", io.EOF
+	}
+	addr := sg[0].Addr.Add(btrfsvol.AddrDelta(sumIdx) * CSumBlockSize)
+	return sg.SumForAddr(addr)
+}
+
 // AllSums ///////////////////////////////////////////////////////////
 
 type AllSums struct {
diff --git a/lib/btrfsprogs/btrfsinspect/scanforextents/scan.go b/lib/btrfsprogs/btrfsinspect/scanforextents/scan.go
index 64e0871..ca3036a 100644
--- a/lib/btrfsprogs/btrfsinspect/scanforextents/scan.go
+++ b/lib/btrfsprogs/btrfsinspect/scanforextents/scan.go
@@ -20,7 +20,7 @@ import (
 
 func ScanForExtents(ctx context.Context, fs *btrfs.FS, blockgroups map[btrfsvol.LogicalAddr]BlockGroup, sums AllSums) error {
 	dlog.Info(ctx, "Pairing up blockgroups and sums...")
-	bgSums := make(map[btrfsvol.LogicalAddr][]SumRun[btrfsvol.LogicalAddr])
+	bgSums := make(map[btrfsvol.LogicalAddr]SumRunWithGaps[btrfsvol.LogicalAddr])
 	for i, bgLAddr := range maps.SortedKeys(blockgroups) {
 		blockgroup := blockgroups[bgLAddr]
 		for laddr := blockgroup.LAddr; laddr < blockgroup.LAddr.Add(blockgroup.Size); {
@@ -51,10 +51,9 @@ func ScanForExtents(ctx context.Context, fs *btrfs.FS, blockgroups map[btrfsvol.
 	bgMatches := make(map[btrfsvol.LogicalAddr][]btrfsvol.QualifiedPhysicalAddr)
 	for i, bgLAddr := range maps.SortedKeys(blockgroups) {
 		bgRuns := bgSums[bgLAddr]
-		if len(bgRuns) != 1 {
-			// TODO(lukeshu): We aught to handle this rather than erroring and skipping
-			// it.
-			dlog.Errorf(ctx, "blockgroup laddr=%v has holes (%v runs)", bgLAddr, len(bgRuns))
+		if len(bgRuns) == 0 {
+			dlog.Errorf(ctx, "... (%v/%v) blockgroup[laddr=%v] can't be matched because it has 0 runs",
+				i+1, len(bgSums), bgLAddr)
 			continue
 		}
 		bgRun := bgRuns[0]