Remove O(n) lookup on unlink/rename

Currently, the path tracking in the gofer involves an O(n) lookup of
child fidRefs. This causes a significant overhead on unlinks in
directories with lots of child fidRefs (<4k).

In this transition, pathNode moves from sync.Map to normal synchronized
maps. There is a small chance of contention in walk, but the lock is
held for a very short time (and sync.Map also had a chance of requiring
locking).

OTOH, sync.Map makes it very difficult to add a fidRef reverse map.

PiperOrigin-RevId: 254489952

1 files changed, 3 insertions, 3 deletions

diff --git a/pkg/p9/handlers.go b/pkg/p9/handlers.go
index 9e622227b..999b4f684 100644
--- a/pkg/p9/handlers.go
+++ b/pkg/p9/handlers.go
@@ -224,7 +224,7 @@ func (t *Tattach) handle(cs *connState) message {
 		file:     sf,
 		refs:     1,
 		mode:     attr.Mode.FileType(),
-		pathNode: &cs.server.pathTree,
+		pathNode: cs.server.pathTree,
 	}
 	defer root.DecRef()
 
@@ -552,8 +552,8 @@ func (t *Tunlinkat) handle(cs *connState) message {
 		// since we always acquire deeper in the hierarchy, we know
 		// that we are free of lock cycles.
 		childPathNode := ref.pathNode.pathNodeFor(t.Name)
-		childPathNode.mu.Lock()
-		defer childPathNode.mu.Unlock()
+		childPathNode.opMu.Lock()
+		defer childPathNode.opMu.Unlock()
 
 		// Do the unlink.
 		err = ref.file.UnlinkAt(t.Name, t.Flags)