kernel: allow to access Task.netns without taking Task.mu

This allows to avoind unnecessary lock-ordering dependencies on task.mu.

5 files changed, 19 insertions, 12 deletions

diff --git a/pkg/sentry/inet/BUILD b/pkg/sentry/inet/BUILD
index 5bba9de0b..2363cec5f 100644
--- a/pkg/sentry/inet/BUILD
+++ b/pkg/sentry/inet/BUILD
@@ -1,13 +1,26 @@
 load("//tools:defs.bzl", "go_library")
+load("//tools/go_generics:defs.bzl", "go_template_instance")
 
 package(
     default_visibility = ["//:sandbox"],
     licenses = ["notice"],
 )
 
+go_template_instance(
+    name = "atomicptr_netns",
+    out = "atomicptr_netns_unsafe.go",
+    package = "inet",
+    prefix = "Namespace",
+    template = "//pkg/sync/atomicptr:generic_atomicptr",
+    types = {
+        "Value": "Namespace",
+    },
+)
+
 go_library(
     name = "inet",
     srcs = [
+        "atomicptr_netns_unsafe.go",
         "context.go",
         "inet.go",
         "namespace.go",
diff --git a/pkg/sentry/kernel/task.go b/pkg/sentry/kernel/task.go
index 9a95bf44c..2eebd1b47 100644
--- a/pkg/sentry/kernel/task.go
+++ b/pkg/sentry/kernel/task.go
@@ -513,7 +513,7 @@ type Task struct {
 	// netns is the task's network namespace. netns is never nil.
 	//
 	// netns is protected by mu.
-	netns *inet.Namespace
+	netns inet.NamespaceAtomicPtr
 
 	// If rseqPreempted is true, before the next call to p.Switch(),
 	// interrupt rseq critical regions as defined by rseqAddr and
diff --git a/pkg/sentry/kernel/task_clone.go b/pkg/sentry/kernel/task_clone.go
index 26a981f36..a6d8fb163 100644
--- a/pkg/sentry/kernel/task_clone.go
+++ b/pkg/sentry/kernel/task_clone.go
@@ -444,7 +444,7 @@ func (t *Task) Unshare(flags int32) error {
 			t.mu.Unlock()
 			return linuxerr.EPERM
 		}
-		t.netns = inet.NewNamespace(t.netns)
+		t.netns.Store(inet.NewNamespace(t.netns.Load()))
 	}
 	if flags&linux.CLONE_NEWUTS != 0 {
 		if !haveCapSysAdmin {
diff --git a/pkg/sentry/kernel/task_net.go b/pkg/sentry/kernel/task_net.go
index f7711232c..e31e2b2e8 100644
--- a/pkg/sentry/kernel/task_net.go
+++ b/pkg/sentry/kernel/task_net.go
@@ -20,9 +20,7 @@ import (
 
 // IsNetworkNamespaced returns true if t is in a non-root network namespace.
 func (t *Task) IsNetworkNamespaced() bool {
-	t.mu.Lock()
-	defer t.mu.Unlock()
-	return !t.netns.IsRoot()
+	return !t.netns.Load().IsRoot()
 }
 
 // NetworkContext returns the network stack used by the task. NetworkContext
@@ -31,14 +29,10 @@ func (t *Task) IsNetworkNamespaced() bool {
 // TODO(gvisor.dev/issue/1833): Migrate callers of this method to
 // NetworkNamespace().
 func (t *Task) NetworkContext() inet.Stack {
-	t.mu.Lock()
-	defer t.mu.Unlock()
-	return t.netns.Stack()
+	return t.netns.Load().Stack()
 }
 
 // NetworkNamespace returns the network namespace observed by the task.
 func (t *Task) NetworkNamespace() *inet.Namespace {
-	t.mu.Lock()
-	defer t.mu.Unlock()
-	return t.netns
+	return t.netns.Load()
 }
diff --git a/pkg/sentry/kernel/task_start.go b/pkg/sentry/kernel/task_start.go
index 217c6f531..4919dea7c 100644
--- a/pkg/sentry/kernel/task_start.go
+++ b/pkg/sentry/kernel/task_start.go
@@ -140,7 +140,6 @@ func (ts *TaskSet) newTask(cfg *TaskConfig) (*Task, error) {
 		allowedCPUMask:     cfg.AllowedCPUMask.Copy(),
 		ioUsage:            &usage.IO{},
 		niceness:           cfg.Niceness,
-		netns:              cfg.NetworkNamespace,
 		utsns:              cfg.UTSNamespace,
 		ipcns:              cfg.IPCNamespace,
 		abstractSockets:    cfg.AbstractSocketNamespace,
@@ -152,6 +151,7 @@ func (ts *TaskSet) newTask(cfg *TaskConfig) (*Task, error) {
 		containerID:        cfg.ContainerID,
 		cgroups:            make(map[Cgroup]struct{}),
 	}
+	t.netns.Store(cfg.NetworkNamespace)
 	t.creds.Store(cfg.Credentials)
 	t.endStopCond.L = &t.tg.signalHandlers.mu
 	t.ptraceTracer.Store((*Task)(nil))