diff --git a/pkg/sentry/fsimpl/sys/sys.go b/pkg/sentry/fsimpl/sys/sys.go
index 7fcb2d26b..3c6b41137 100644
--- a/pkg/sentry/fsimpl/sys/sys.go
+++ b/pkg/sentry/fsimpl/sys/sys.go
@@ -34,6 +34,7 @@ import (
 const (
 	// Name is the default filesystem name.
 	Name                     = "sysfs"
+	defaultSysMode           = linux.FileMode(0444)
 	defaultSysDirMode        = linux.FileMode(0755)
 	defaultMaxCachedDentries = uint64(1000)
 )
@@ -43,6 +44,15 @@ const (
 // +stateify savable
 type FilesystemType struct{}
 
+// InternalData contains internal data passed in via
+// vfs.GetFilesystemOptions.InternalData.
+//
+// +stateify savable
+type InternalData struct {
+	// ProductName is the value to be set to devices/virtual/dmi/id/product_name.
+	ProductName string
+}
+
 // filesystem implements vfs.FilesystemImpl.
 //
 // +stateify savable
@@ -96,18 +106,38 @@ func (fsType FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.Virt
 		fsDirChildren["cgroup"] = fs.newDir(ctx, creds, defaultSysDirMode, nil)
 	}
 
-	root := fs.newDir(ctx, creds, defaultSysDirMode, map[string]kernfs.Inode{
-		"block": fs.newDir(ctx, creds, defaultSysDirMode, nil),
-		"bus":   fs.newDir(ctx, creds, defaultSysDirMode, nil),
-		"class": fs.newDir(ctx, creds, defaultSysDirMode, map[string]kernfs.Inode{
-			"power_supply": fs.newDir(ctx, creds, defaultSysDirMode, nil),
+	classSub := map[string]kernfs.Inode{
+		"power_supply": fs.newDir(ctx, creds, defaultSysDirMode, nil),
+	}
+	devicesSub := map[string]kernfs.Inode{
+		"system": fs.newDir(ctx, creds, defaultSysDirMode, map[string]kernfs.Inode{
+			"cpu": cpuDir(ctx, fs, creds),
 		}),
-		"dev": fs.newDir(ctx, creds, defaultSysDirMode, nil),
-		"devices": fs.newDir(ctx, creds, defaultSysDirMode, map[string]kernfs.Inode{
-			"system": fs.newDir(ctx, creds, defaultSysDirMode, map[string]kernfs.Inode{
-				"cpu": cpuDir(ctx, fs, creds),
+	}
+	productName := ""
+	if opts.InternalData != nil {
+		data := opts.InternalData.(*InternalData)
+		productName = data.ProductName
+	}
+	if len(productName) > 0 {
+		log.Debugf("Setting product_name: %q", productName)
+		classSub["dmi"] = fs.newDir(ctx, creds, defaultSysDirMode, map[string]kernfs.Inode{
+			"id": kernfs.NewStaticSymlink(ctx, creds, linux.UNNAMED_MAJOR, fs.devMinor, fs.NextIno(), "../../devices/virtual/dmi/id"),
+		})
+		devicesSub["virtual"] = fs.newDir(ctx, creds, defaultSysDirMode, map[string]kernfs.Inode{
+			"dmi": fs.newDir(ctx, creds, defaultSysDirMode, map[string]kernfs.Inode{
+				"id": fs.newDir(ctx, creds, defaultSysDirMode, map[string]kernfs.Inode{
+					"product_name": fs.newStaticFile(ctx, creds, defaultSysMode, productName+"\n"),
+				}),
 			}),
-		}),
+		})
+	}
+	root := fs.newDir(ctx, creds, defaultSysDirMode, map[string]kernfs.Inode{
+		"block":    fs.newDir(ctx, creds, defaultSysDirMode, nil),
+		"bus":      fs.newDir(ctx, creds, defaultSysDirMode, nil),
+		"class":    fs.newDir(ctx, creds, defaultSysDirMode, classSub),
+		"dev":      fs.newDir(ctx, creds, defaultSysDirMode, nil),
+		"devices":  fs.newDir(ctx, creds, defaultSysDirMode, devicesSub),
 		"firmware": fs.newDir(ctx, creds, defaultSysDirMode, nil),
 		"fs":       fs.newDir(ctx, creds, defaultSysDirMode, fsDirChildren),
 		"kernel":   kernelDir(ctx, fs, creds),
@@ -239,3 +269,15 @@ type implStatFS struct{}
 func (*implStatFS) StatFS(context.Context, *vfs.Filesystem) (linux.Statfs, error) {
 	return vfs.GenericStatFS(linux.SYSFS_MAGIC), nil
 }
+
+// +stateify savable
+type staticFile struct {
+	kernfs.DynamicBytesFile
+	vfs.StaticData
+}
+
+func (fs *filesystem) newStaticFile(ctx context.Context, creds *auth.Credentials, mode linux.FileMode, data string) kernfs.Inode {
+	s := &staticFile{StaticData: vfs.StaticData{Data: data}}
+	s.Init(ctx, creds, linux.UNNAMED_MAJOR, fs.devMinor, fs.NextIno(), s, mode)
+	return s
+}
diff --git a/runsc/boot/controller.go b/runsc/boot/controller.go
index 76e1f596b..22686f711 100644
--- a/runsc/boot/controller.go
+++ b/runsc/boot/controller.go
@@ -432,7 +432,7 @@ func (cm *containerManager) Restore(o *RestoreOpts, _ *struct{}) error {
 
 	// Set up the restore environment.
 	ctx := k.SupervisorContext()
-	mntr := newContainerMounter(&cm.l.root, cm.l.k, cm.l.mountHints, kernel.VFS2Enabled)
+	mntr := newContainerMounter(&cm.l.root, cm.l.k, cm.l.mountHints, kernel.VFS2Enabled, cm.l.productName)
 	if kernel.VFS2Enabled {
 		ctx, err = mntr.configureRestore(ctx)
 		if err != nil {
diff --git a/runsc/boot/fs.go b/runsc/boot/fs.go
index 3ef746b54..c8bc07085 100644
--- a/runsc/boot/fs.go
+++ b/runsc/boot/fs.go
@@ -622,15 +622,20 @@ type containerMounter struct {
 	k *kernel.Kernel
 
 	hints *podMountHints
+
+	// productName is the value to show in
+	// /sys/devices/virtual/dmi/id/product_name.
+	productName string
 }
 
-func newContainerMounter(info *containerInfo, k *kernel.Kernel, hints *podMountHints, vfs2Enabled bool) *containerMounter {
+func newContainerMounter(info *containerInfo, k *kernel.Kernel, hints *podMountHints, vfs2Enabled bool, productName string) *containerMounter {
 	return &containerMounter{
-		root:   info.spec.Root,
-		mounts: compileMounts(info.spec, info.conf, vfs2Enabled),
-		fds:    fdDispenser{fds: info.goferFDs},
-		k:      k,
-		hints:  hints,
+		root:        info.spec.Root,
+		mounts:      compileMounts(info.spec, info.conf, vfs2Enabled),
+		fds:         fdDispenser{fds: info.goferFDs},
+		k:           k,
+		hints:       hints,
+		productName: productName,
 	}
 }
 
diff --git a/runsc/boot/loader.go b/runsc/boot/loader.go
index d229e290b..eddc01a6f 100644
--- a/runsc/boot/loader.go
+++ b/runsc/boot/loader.go
@@ -144,6 +144,10 @@ type Loader struct {
 	// mountHints provides extra information about mounts for containers that
 	// apply to the entire pod.
 	mountHints *podMountHints
+
+	// productName is the value to show in
+	// /sys/devices/virtual/dmi/id/product_name.
+	productName string
 }
 
 // execID uniquely identifies a sentry process that is executed in a container.
@@ -219,6 +223,9 @@ type Args struct {
 	// TraceFD is the file descriptor to write a Go execution trace to.
 	// Valid if >=0.
 	TraceFD int
+	// ProductName is the value to show in
+	// /sys/devices/virtual/dmi/id/product_name.
+	ProductName string
 }
 
 // make sure stdioFDs are always the same on initial start and on restore
@@ -424,6 +431,7 @@ func New(args Args) (*Loader, error) {
 		mountHints:    mountHints,
 		root:          info,
 		stopProfiling: stopProfiling,
+		productName:   args.ProductName,
 	}
 
 	// We don't care about child signals; some platforms can generate a
@@ -769,7 +777,7 @@ func (l *Loader) createContainerProcess(root bool, cid string, info *containerIn
 	}
 	l.startGoferMonitor(cid, int32(info.goferFDs[0].FD()))
 
-	mntr := newContainerMounter(info, l.k, l.mountHints, kernel.VFS2Enabled)
+	mntr := newContainerMounter(info, l.k, l.mountHints, kernel.VFS2Enabled, l.productName)
 	if root {
 		if err := mntr.processHints(info.conf, info.procArgs.Credentials); err != nil {
 			return nil, nil, nil, err
diff --git a/runsc/boot/loader_test.go b/runsc/boot/loader_test.go
index 86334dbba..0f68f0795 100644
--- a/runsc/boot/loader_test.go
+++ b/runsc/boot/loader_test.go
@@ -447,7 +447,7 @@ func TestCreateMountNamespace(t *testing.T) {
 				goferFDs: []*fd.FD{fd.New(sandEnd)},
 			}
 
-			mntr := newContainerMounter(&info, nil, &podMountHints{}, false /* vfs2Enabled */)
+			mntr := newContainerMounter(&info, nil, &podMountHints{}, false /* vfs2Enabled */, "")
 			mns, err := mntr.createMountNamespace(ctx, conf)
 			if err != nil {
 				t.Fatalf("failed to create mount namespace: %v", err)
@@ -487,7 +487,7 @@ func TestCreateMountNamespaceVFS2(t *testing.T) {
 			defer l.Destroy()
 			defer loaderCleanup()
 
-			mntr := newContainerMounter(&l.root, l.k, l.mountHints, true /* vfs2Enabled */)
+			mntr := newContainerMounter(&l.root, l.k, l.mountHints, true /* vfs2Enabled */, "")
 			if err := mntr.processHints(l.root.conf, l.root.procArgs.Credentials); err != nil {
 				t.Fatalf("failed process hints: %v", err)
 			}
@@ -716,7 +716,7 @@ func TestRestoreEnvironment(t *testing.T) {
 				spec:     tc.spec,
 				goferFDs: ioFDs,
 			}
-			mntr := newContainerMounter(&info, nil, &podMountHints{}, conf.VFS2)
+			mntr := newContainerMounter(&info, nil, &podMountHints{}, conf.VFS2, "")
 			actualRenv, err := mntr.createRestoreEnvironment(conf)
 			if !tc.errorExpected && err != nil {
 				t.Fatalf("could not create restore environment for test:%s", tc.name)
diff --git a/runsc/boot/vfs.go b/runsc/boot/vfs.go
index 0029f54e1..fe7304826 100644
--- a/runsc/boot/vfs.go
+++ b/runsc/boot/vfs.go
@@ -500,12 +500,17 @@ func (c *containerMounter) getMountNameAndOptionsVFS2(conf *config.Config, m *mo
 
 	// Find filesystem name and FS specific data field.
 	switch m.mount.Type {
-	case devpts.Name, devtmpfs.Name, proc.Name, sys.Name:
+	case devpts.Name, devtmpfs.Name, proc.Name:
 		// Nothing to do.
 
 	case nonefs:
 		fsName = sys.Name
 
+	case sys.Name:
+		if len(c.productName) > 0 {
+			internalData = &sys.InternalData{ProductName: c.productName}
+		}
+
 	case tmpfs.Name:
 		var err error
 		data, err = parseAndFilterOptions(m.mount.Options, tmpfsAllowedData...)
diff --git a/runsc/cmd/boot.go b/runsc/cmd/boot.go
index 9ded710d3..b1b90b127 100644
--- a/runsc/cmd/boot.go
+++ b/runsc/cmd/boot.go
@@ -16,6 +16,7 @@ package cmd
 
 import (
 	"context"
+	"io/ioutil"
 	"os"
 	"runtime/debug"
 	"strings"
@@ -107,6 +108,10 @@ type Boot struct {
 	// terminates. This flag is set when the command execve's itself because
 	// parent death signal doesn't propagate through execve when uid/gid changes.
 	attached bool
+
+	// productName is the value to show in
+	// /sys/devices/virtual/dmi/id/product_name.
+	productName string
 }
 
 // Name implements subcommands.Command.Name.
@@ -146,6 +151,7 @@ func (b *Boot) SetFlags(f *flag.FlagSet) {
 	f.IntVar(&b.profileMutexFD, "profile-mutex-fd", -1, "file descriptor to write mutex profile to. -1 disables profiling.")
 	f.IntVar(&b.traceFD, "trace-fd", -1, "file descriptor to write Go execution trace to. -1 disables tracing.")
 	f.BoolVar(&b.attached, "attached", false, "if attached is true, kills the sandbox process when the parent process terminates")
+	f.StringVar(&b.productName, "product-name", "", "value to show in /sys/devices/virtual/dmi/id/product_name")
 }
 
 // Execute implements subcommands.Command.Execute.  It starts a sandbox in a
@@ -161,6 +167,16 @@ func (b *Boot) Execute(_ context.Context, f *flag.FlagSet, args ...interface{})
 	// Set traceback level
 	debug.SetTraceback(conf.Traceback)
 
+	if len(b.productName) == 0 {
+		// Do this before chroot takes effect, otherwise we can't read /sys.
+		if product, err := ioutil.ReadFile("/sys/devices/virtual/dmi/id/product_name"); err != nil {
+			log.Warningf("Not setting product_name: %v", err)
+		} else {
+			b.productName = strings.TrimSpace(string(product))
+			log.Infof("Setting product_name: %q", b.productName)
+		}
+	}
+
 	if b.attached {
 		// Ensure this process is killed after parent process terminates when
 		// attached mode is enabled. In the unfortunate event that the parent
@@ -177,7 +193,7 @@ func (b *Boot) Execute(_ context.Context, f *flag.FlagSet, args ...interface{})
 
 		if !b.applyCaps && !conf.Rootless {
 			// Remove --apply-caps arg to call myself. It has already been done.
-			args := prepareArgs(b.attached, "setup-root")
+			args := b.prepareArgs("setup-root")
 
 			// Note that we've already read the spec from the spec FD, and
 			// we will read it again after the exec call. This works
@@ -217,7 +233,7 @@ func (b *Boot) Execute(_ context.Context, f *flag.FlagSet, args ...interface{})
 
 		// Remove --apply-caps and --setup-root arg to call myself. Both have
 		// already been done.
-		args := prepareArgs(b.attached, "setup-root", "apply-caps")
+		args := b.prepareArgs("setup-root", "apply-caps")
 
 		// Note that we've already read the spec from the spec FD, and
 		// we will read it again after the exec call. This works
@@ -271,6 +287,7 @@ func (b *Boot) Execute(_ context.Context, f *flag.FlagSet, args ...interface{})
 		ProfileHeapFD:  b.profileHeapFD,
 		ProfileMutexFD: b.profileMutexFD,
 		TraceFD:        b.traceFD,
+		ProductName:    b.productName,
 	}
 	l, err := boot.New(bootArgs)
 	if err != nil {
@@ -308,7 +325,7 @@ func (b *Boot) Execute(_ context.Context, f *flag.FlagSet, args ...interface{})
 	return subcommands.ExitSuccess
 }
 
-func prepareArgs(attached bool, exclude ...string) []string {
+func (b *Boot) prepareArgs(exclude ...string) []string {
 	var args []string
 	for _, arg := range os.Args {
 		for _, excl := range exclude {
@@ -317,10 +334,17 @@ func prepareArgs(attached bool, exclude ...string) []string {
 			}
 		}
 		args = append(args, arg)
-		if attached && arg == "boot" {
-			// Strategicaly place "--attached" after the command. This is needed
-			// to ensure the new process is killed when the parent process terminates.
-			args = append(args, "--attached")
+		// Strategically add parameters after the command and before the container
+		// ID at the end.
+		if arg == "boot" {
+			if b.attached {
+				// This is needed to ensure the new process is killed when the parent
+				// process terminates.
+				args = append(args, "--attached")
+			}
+			if len(b.productName) > 0 {
+				args = append(args, "--product-name", b.productName)
+			}
 		}
 	skip:
 	}
diff --git a/test/e2e/integration_test.go b/test/e2e/integration_test.go
index 1003e0209..206398856 100644
--- a/test/e2e/integration_test.go
+++ b/test/e2e/integration_test.go
@@ -798,3 +798,23 @@ func TestDeleteInterface(t *testing.T) {
 		t.Fatalf("loopback interface is removed")
 	}
 }
+
+func TestProductName(t *testing.T) {
+	want, err := ioutil.ReadFile("/sys/devices/virtual/dmi/id/product_name")
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	ctx := context.Background()
+	d := dockerutil.MakeContainer(ctx, t)
+	defer d.CleanUp(ctx)
+
+	opts := dockerutil.RunOpts{Image: "basic/alpine"}
+	got, err := d.Run(ctx, opts, "cat", "/sys/devices/virtual/dmi/id/product_name")
+	if err != nil {
+		t.Fatalf("docker run failed: %v", err)
+	}
+	if string(want) != got {
+		t.Errorf("invalid product name, want: %q, got: %q", want, got)
+	}
+}