mount: reuse the listed entry's path instead of rebuilding it (#10817)

readdir built dirPath.Child(name) for every child while entry.FullPath was
already that exact string, from NewFullPath in the meta cache store or from
FromPbEntry on the read-through path. One allocation per entry, and on a wide
tree with long paths that is most of what a listing allocates.

BenchmarkReadDirectory/kernel_readdirplus over 200k entries: 2,039,656 ->
1,839,318 allocs/op, 174.5 -> 167.8 MB/op, 152.6 -> 135.1 ms/op.

weed/mount/weedfs_dir_read.go

@@ -3,6 +3,7 @@ package mount
 import (
 	"context"
 	"errors"
+	"strings"
 	"sync"
 	"time"
 
@@ -219,7 +220,11 @@ func (wfs *WFS) doReadDirectory(input *fuse.ReadIn, out DirEntrySink, isPlusMode
 	processEachEntryFn := func(entry *filer.Entry, index int64) bool {
 		dirEntry.Name = entry.Name()
 		dirEntry.Mode = toSyscallMode(entry.Mode)
-		childPath := dirPath.Child(dirEntry.Name)
+		// Rebuild only for a sanitized name: that is the one a LOOKUP carries.
+		childPath := entry.FullPath
+		if !strings.HasSuffix(string(childPath), dirEntry.Name) {
+			childPath = dirPath.Child(dirEntry.Name)
+		}
 		var inode uint64
 		if takesLookupRef {
 			inode = wfs.inodeToPath.Lookup(childPath, entry.Crtime.Unix(), entry.IsDirectory(), len(entry.HardLinkId) > 0, entry.Inode, false)

weed/mount/weedfs_dir_read_pagination_test.go

@@ -336,3 +336,24 @@ func TestReadDirDirectTrimsConsumedEntries(t *testing.T) {
 		t.Errorf("handle held %d entries at peak, want well under the %d in the directory", peak, total)
 	}
 }
+
+// The table has to key on the sanitized name, not the stored bytes.
+func TestReadDirPlusSanitizedName(t *testing.T) {
+	dir := util.FullPath("/images")
+	const rawName = "bad\xffname.jpg"
+	wfs := newPagingWFS(t, dir, []string{"good.jpg", rawName}, 0)
+	dirInode, _ := wfs.inodeToPath.GetInode(dir)
+
+	sink := &benchSink{plus: true, takesRef: true, sinkLimit: 16}
+	if got := walkOnce(t, wfs, dirInode, sink, false); got != 4 {
+		t.Fatalf("listed %d entries, want 4 (. .. and two children)", got)
+	}
+
+	sanitized := dir.Child(util.SanitizeUTF8Name(rawName))
+	if !wfs.inodeToPath.HasPath(sanitized) {
+		t.Errorf("%s not in the inode table", sanitized)
+	}
+	if wfs.inodeToPath.HasPath(dir.Child(rawName)) {
+		t.Errorf("%s keyed on the unsanitized name", dir.Child(rawName))
+	}
+}