Implement basic umounting for vfs2.

This is required to test filesystems with a non-trivial implementation of
FilesystemImpl.Release(). Propagation isn't handled yet, and umount isn't yet
plumbed out to VirtualFilesystem.UmountAt(), but otherwise the implementation
of umount is believed to be correct.

- Move entering mountTable.seq writer critical sections to callers of
  mountTable.{insert,remove}Seqed. This is required since umount(2) must ensure
  that no new references are taken on the candidate mount after checking that
  it isn't busy, which is only possible by entering a vfs.mountTable.seq writer
  critical section before the check and remaining in it until after
  VFS.umountRecursiveLocked() is complete. (Linux does the same thing:
  fs/namespace.c:do_umount() => lock_mount_hash(),
  fs/pnode.c:propagate_mount_busy(), umount_tree(), unlock_mount_hash().)

- It's not possible for dentry deletion to umount while only holding
  VFS.mountMu for reading, but it's also very unappealing to hold VFS.mountMu
  exclusively around e.g. gofer unlink RPCs. Introduce dentry.mu to avoid these
  problems. This means that VFS.mountMu is never acquired for reading, so
  change it to a sync.Mutex.

PiperOrigin-RevId: 282444343

2 files changed, 23 insertions, 3 deletions

diff --git a/pkg/sentry/fsimpl/memfs/BUILD b/pkg/sentry/fsimpl/memfs/BUILD
index 952b20c51..bc5c0b591 100644
--- a/pkg/sentry/fsimpl/memfs/BUILD
+++ b/pkg/sentry/fsimpl/memfs/BUILD
@@ -1,9 +1,10 @@
 load("//tools/go_stateify:defs.bzl", "go_library")
 load("@io_bazel_rules_go//go:def.bzl", "go_test")
-load("//tools/go_generics:defs.bzl", "go_template_instance")
 
 package(licenses = ["notice"])
 
+load("//tools/go_generics:defs.bzl", "go_template_instance")
+
 go_template_instance(
     name = "dentry_list",
     out = "dentry_list.go",
@@ -48,6 +49,7 @@ go_test(
     deps = [
         ":memfs",
         "//pkg/abi/linux",
+        "//pkg/refs",
         "//pkg/sentry/context",
         "//pkg/sentry/context/contexttest",
         "//pkg/sentry/fs",
diff --git a/pkg/sentry/fsimpl/memfs/benchmark_test.go b/pkg/sentry/fsimpl/memfs/benchmark_test.go
index a94b17db6..23a846c08 100644
--- a/pkg/sentry/fsimpl/memfs/benchmark_test.go
+++ b/pkg/sentry/fsimpl/memfs/benchmark_test.go
@@ -21,6 +21,7 @@ import (
 	"testing"
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/refs"
 	"gvisor.dev/gvisor/pkg/sentry/context"
 	"gvisor.dev/gvisor/pkg/sentry/context/contexttest"
 	"gvisor.dev/gvisor/pkg/sentry/fs"
@@ -160,6 +161,8 @@ func BenchmarkVFS1TmpfsStat(b *testing.B) {
 					b.Fatalf("stat(%q) failed: %v", filePath, err)
 				}
 			}
+			// Don't include deferred cleanup in benchmark time.
+			b.StopTimer()
 		})
 	}
 }
@@ -177,6 +180,7 @@ func BenchmarkVFS2MemfsStat(b *testing.B) {
 			if err != nil {
 				b.Fatalf("failed to create tmpfs root mount: %v", err)
 			}
+			defer mntns.DecRef(vfsObj)
 
 			var filePathBuilder strings.Builder
 			filePathBuilder.WriteByte('/')
@@ -186,7 +190,6 @@ func BenchmarkVFS2MemfsStat(b *testing.B) {
 			defer root.DecRef()
 			vd := root
 			vd.IncRef()
-			defer vd.DecRef()
 			for i := depth; i > 0; i-- {
 				name := fmt.Sprintf("%d", i)
 				pop := vfs.PathOperation{
@@ -219,6 +222,8 @@ func BenchmarkVFS2MemfsStat(b *testing.B) {
 				Flags: linux.O_RDWR | linux.O_CREAT | linux.O_EXCL,
 				Mode:  0644,
 			})
+			vd.DecRef()
+			vd = vfs.VirtualDentry{}
 			if err != nil {
 				b.Fatalf("failed to create file %q: %v", filename, err)
 			}
@@ -243,6 +248,8 @@ func BenchmarkVFS2MemfsStat(b *testing.B) {
 					b.Fatalf("got wrong permissions (%0o)", stat.Mode)
 				}
 			}
+			// Don't include deferred cleanup in benchmark time.
+			b.StopTimer()
 		})
 	}
 }
@@ -343,6 +350,8 @@ func BenchmarkVFS1TmpfsMountStat(b *testing.B) {
 					b.Fatalf("stat(%q) failed: %v", filePath, err)
 				}
 			}
+			// Don't include deferred cleanup in benchmark time.
+			b.StopTimer()
 		})
 	}
 }
@@ -360,6 +369,7 @@ func BenchmarkVFS2MemfsMountStat(b *testing.B) {
 			if err != nil {
 				b.Fatalf("failed to create tmpfs root mount: %v", err)
 			}
+			defer mntns.DecRef(vfsObj)
 
 			var filePathBuilder strings.Builder
 			filePathBuilder.WriteByte('/')
@@ -395,7 +405,6 @@ func BenchmarkVFS2MemfsMountStat(b *testing.B) {
 			if err != nil {
 				b.Fatalf("failed to walk to mount root: %v", err)
 			}
-			defer vd.DecRef()
 			for i := depth; i > 0; i-- {
 				name := fmt.Sprintf("%d", i)
 				pop := vfs.PathOperation{
@@ -435,6 +444,7 @@ func BenchmarkVFS2MemfsMountStat(b *testing.B) {
 				Flags: linux.O_RDWR | linux.O_CREAT | linux.O_EXCL,
 				Mode:  0644,
 			})
+			vd.DecRef()
 			if err != nil {
 				b.Fatalf("failed to create file %q: %v", filename, err)
 			}
@@ -459,6 +469,14 @@ func BenchmarkVFS2MemfsMountStat(b *testing.B) {
 					b.Fatalf("got wrong permissions (%0o)", stat.Mode)
 				}
 			}
+			// Don't include deferred cleanup in benchmark time.
+			b.StopTimer()
 		})
 	}
 }
+
+func init() {
+	// Turn off reference leak checking for a fair comparison between vfs1 and
+	// vfs2.
+	refs.SetLeakMode(refs.NoLeakChecking)
+}