summaryrefslogtreecommitdiffhomepage
path: root/pkg/sentry/vfs
diff options
context:
space:
mode:
Diffstat (limited to 'pkg/sentry/vfs')
-rw-r--r--pkg/sentry/vfs/BUILD4
-rw-r--r--pkg/sentry/vfs/dentry.go2
-rw-r--r--pkg/sentry/vfs/device.go100
-rw-r--r--pkg/sentry/vfs/file_description.go101
-rw-r--r--pkg/sentry/vfs/file_description_impl_util.go28
-rw-r--r--pkg/sentry/vfs/file_description_impl_util_test.go2
-rw-r--r--pkg/sentry/vfs/filesystem.go21
-rw-r--r--pkg/sentry/vfs/filesystem_type.go55
-rw-r--r--pkg/sentry/vfs/mount.go15
-rw-r--r--pkg/sentry/vfs/mount_test.go3
-rw-r--r--pkg/sentry/vfs/mount_unsafe.go4
-rw-r--r--pkg/sentry/vfs/options.go4
-rw-r--r--pkg/sentry/vfs/pathname.go3
-rw-r--r--pkg/sentry/vfs/permissions.go24
-rw-r--r--pkg/sentry/vfs/resolving_path.go2
-rw-r--r--pkg/sentry/vfs/vfs.go20
16 files changed, 349 insertions, 39 deletions
diff --git a/pkg/sentry/vfs/BUILD b/pkg/sentry/vfs/BUILD
index e3e554b88..35c7be259 100644
--- a/pkg/sentry/vfs/BUILD
+++ b/pkg/sentry/vfs/BUILD
@@ -9,6 +9,7 @@ go_library(
"context.go",
"debug.go",
"dentry.go",
+ "device.go",
"file_description.go",
"file_description_impl_util.go",
"filesystem.go",
@@ -33,7 +34,7 @@ go_library(
"//pkg/sentry/kernel/auth",
"//pkg/sentry/memmap",
"//pkg/sentry/usermem",
- "//pkg/syncutil",
+ "//pkg/sync",
"//pkg/syserror",
"//pkg/waiter",
],
@@ -53,6 +54,7 @@ go_test(
"//pkg/sentry/context/contexttest",
"//pkg/sentry/kernel/auth",
"//pkg/sentry/usermem",
+ "//pkg/sync",
"//pkg/syserror",
],
)
diff --git a/pkg/sentry/vfs/dentry.go b/pkg/sentry/vfs/dentry.go
index 1bc9c4a38..486a76475 100644
--- a/pkg/sentry/vfs/dentry.go
+++ b/pkg/sentry/vfs/dentry.go
@@ -16,9 +16,9 @@ package vfs
import (
"fmt"
- "sync"
"sync/atomic"
+ "gvisor.dev/gvisor/pkg/sync"
"gvisor.dev/gvisor/pkg/syserror"
)
diff --git a/pkg/sentry/vfs/device.go b/pkg/sentry/vfs/device.go
new file mode 100644
index 000000000..cb672e36f
--- /dev/null
+++ b/pkg/sentry/vfs/device.go
@@ -0,0 +1,100 @@
+// Copyright 2019 The gVisor Authors.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+// http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package vfs
+
+import (
+ "fmt"
+
+ "gvisor.dev/gvisor/pkg/sentry/context"
+ "gvisor.dev/gvisor/pkg/syserror"
+)
+
+// DeviceKind indicates whether a device is a block or character device.
+type DeviceKind uint32
+
+const (
+ // BlockDevice indicates a block device.
+ BlockDevice DeviceKind = iota
+
+ // CharDevice indicates a character device.
+ CharDevice
+)
+
+// String implements fmt.Stringer.String.
+func (kind DeviceKind) String() string {
+ switch kind {
+ case BlockDevice:
+ return "block"
+ case CharDevice:
+ return "character"
+ default:
+ return fmt.Sprintf("invalid device kind %d", kind)
+ }
+}
+
+type devTuple struct {
+ kind DeviceKind
+ major uint32
+ minor uint32
+}
+
+// A Device backs device special files.
+type Device interface {
+ // Open returns a FileDescription representing this device.
+ Open(ctx context.Context, mnt *Mount, d *Dentry, opts OpenOptions) (*FileDescription, error)
+}
+
+type registeredDevice struct {
+ dev Device
+ opts RegisterDeviceOptions
+}
+
+// RegisterDeviceOptions contains options to
+// VirtualFilesystem.RegisterDevice().
+type RegisterDeviceOptions struct {
+ // GroupName is the name shown for this device registration in
+ // /proc/devices. If GroupName is empty, this registration will not be
+ // shown in /proc/devices.
+ GroupName string
+}
+
+// RegisterDevice registers the given Device in vfs with the given major and
+// minor device numbers.
+func (vfs *VirtualFilesystem) RegisterDevice(kind DeviceKind, major, minor uint32, dev Device, opts *RegisterDeviceOptions) error {
+ tup := devTuple{kind, major, minor}
+ vfs.devicesMu.Lock()
+ defer vfs.devicesMu.Unlock()
+ if existing, ok := vfs.devices[tup]; ok {
+ return fmt.Errorf("%s device number (%d, %d) is already registered to device type %T", kind, major, minor, existing.dev)
+ }
+ vfs.devices[tup] = &registeredDevice{
+ dev: dev,
+ opts: *opts,
+ }
+ return nil
+}
+
+// OpenDeviceSpecialFile returns a FileDescription representing the given
+// device.
+func (vfs *VirtualFilesystem) OpenDeviceSpecialFile(ctx context.Context, mnt *Mount, d *Dentry, kind DeviceKind, major, minor uint32, opts *OpenOptions) (*FileDescription, error) {
+ tup := devTuple{kind, major, minor}
+ vfs.devicesMu.RLock()
+ defer vfs.devicesMu.RUnlock()
+ rd, ok := vfs.devices[tup]
+ if !ok {
+ return nil, syserror.ENXIO
+ }
+ return rd.dev.Open(ctx, mnt, d, *opts)
+}
diff --git a/pkg/sentry/vfs/file_description.go b/pkg/sentry/vfs/file_description.go
index 0b053201a..6afe280bc 100644
--- a/pkg/sentry/vfs/file_description.go
+++ b/pkg/sentry/vfs/file_description.go
@@ -61,11 +61,25 @@ type FileDescriptionOptions struct {
// If AllowDirectIO is true, allow O_DIRECT to be set on the file. This is
// usually only the case if O_DIRECT would actually have an effect.
AllowDirectIO bool
+
+ // If UseDentryMetadata is true, calls to FileDescription methods that
+ // interact with file and filesystem metadata (Stat, SetStat, StatFS,
+ // Listxattr, Getxattr, Setxattr, Removexattr) are implemented by calling
+ // the corresponding FilesystemImpl methods instead of the corresponding
+ // FileDescriptionImpl methods.
+ //
+ // UseDentryMetadata is intended for file descriptions that are implemented
+ // outside of individual filesystems, such as pipes, sockets, and device
+ // special files. FileDescriptions for which UseDentryMetadata is true may
+ // embed DentryMetadataFileDescriptionImpl to obtain appropriate
+ // implementations of FileDescriptionImpl methods that should not be
+ // called.
+ UseDentryMetadata bool
}
-// Init must be called before first use of fd. It takes ownership of references
-// on mnt and d held by the caller. statusFlags is the initial file description
-// status flags, which is usually the full set of flags passed to open(2).
+// Init must be called before first use of fd. It takes references on mnt and
+// d. statusFlags is the initial file description status flags, which is
+// usually the full set of flags passed to open(2).
func (fd *FileDescription) Init(impl FileDescriptionImpl, statusFlags uint32, mnt *Mount, d *Dentry, opts *FileDescriptionOptions) {
fd.refs = 1
fd.statusFlags = statusFlags | linux.O_LARGEFILE
@@ -73,6 +87,7 @@ func (fd *FileDescription) Init(impl FileDescriptionImpl, statusFlags uint32, mn
mount: mnt,
dentry: d,
}
+ fd.vd.IncRef()
fd.opts = *opts
fd.impl = impl
}
@@ -140,7 +155,7 @@ func (fd *FileDescription) SetStatusFlags(ctx context.Context, creds *auth.Crede
// sense. However, the check as actually implemented seems to be "O_APPEND
// cannot be changed if the file is marked as append-only".
if (flags^oldFlags)&linux.O_APPEND != 0 {
- stat, err := fd.impl.Stat(ctx, StatOptions{
+ stat, err := fd.Stat(ctx, StatOptions{
// There is no mask bit for stx_attributes.
Mask: 0,
// Linux just reads inode::i_flags directly.
@@ -154,7 +169,7 @@ func (fd *FileDescription) SetStatusFlags(ctx context.Context, creds *auth.Crede
}
}
if (flags&linux.O_NOATIME != 0) && (oldFlags&linux.O_NOATIME == 0) {
- stat, err := fd.impl.Stat(ctx, StatOptions{
+ stat, err := fd.Stat(ctx, StatOptions{
Mask: linux.STATX_UID,
// Linux's inode_owner_or_capable() just reads inode::i_uid
// directly.
@@ -348,17 +363,47 @@ func (fd *FileDescription) OnClose(ctx context.Context) error {
// Stat returns metadata for the file represented by fd.
func (fd *FileDescription) Stat(ctx context.Context, opts StatOptions) (linux.Statx, error) {
+ if fd.opts.UseDentryMetadata {
+ vfsObj := fd.vd.mount.vfs
+ rp := vfsObj.getResolvingPath(auth.CredentialsFromContext(ctx), &PathOperation{
+ Root: fd.vd,
+ Start: fd.vd,
+ })
+ stat, err := fd.vd.mount.fs.impl.StatAt(ctx, rp, opts)
+ vfsObj.putResolvingPath(rp)
+ return stat, err
+ }
return fd.impl.Stat(ctx, opts)
}
// SetStat updates metadata for the file represented by fd.
func (fd *FileDescription) SetStat(ctx context.Context, opts SetStatOptions) error {
+ if fd.opts.UseDentryMetadata {
+ vfsObj := fd.vd.mount.vfs
+ rp := vfsObj.getResolvingPath(auth.CredentialsFromContext(ctx), &PathOperation{
+ Root: fd.vd,
+ Start: fd.vd,
+ })
+ err := fd.vd.mount.fs.impl.SetStatAt(ctx, rp, opts)
+ vfsObj.putResolvingPath(rp)
+ return err
+ }
return fd.impl.SetStat(ctx, opts)
}
// StatFS returns metadata for the filesystem containing the file represented
// by fd.
func (fd *FileDescription) StatFS(ctx context.Context) (linux.Statfs, error) {
+ if fd.opts.UseDentryMetadata {
+ vfsObj := fd.vd.mount.vfs
+ rp := vfsObj.getResolvingPath(auth.CredentialsFromContext(ctx), &PathOperation{
+ Root: fd.vd,
+ Start: fd.vd,
+ })
+ statfs, err := fd.vd.mount.fs.impl.StatFSAt(ctx, rp)
+ vfsObj.putResolvingPath(rp)
+ return statfs, err
+ }
return fd.impl.StatFS(ctx)
}
@@ -417,6 +462,16 @@ func (fd *FileDescription) Ioctl(ctx context.Context, uio usermem.IO, args arch.
// Listxattr returns all extended attribute names for the file represented by
// fd.
func (fd *FileDescription) Listxattr(ctx context.Context) ([]string, error) {
+ if fd.opts.UseDentryMetadata {
+ vfsObj := fd.vd.mount.vfs
+ rp := vfsObj.getResolvingPath(auth.CredentialsFromContext(ctx), &PathOperation{
+ Root: fd.vd,
+ Start: fd.vd,
+ })
+ names, err := fd.vd.mount.fs.impl.ListxattrAt(ctx, rp)
+ vfsObj.putResolvingPath(rp)
+ return names, err
+ }
names, err := fd.impl.Listxattr(ctx)
if err == syserror.ENOTSUP {
// Linux doesn't actually return ENOTSUP in this case; instead,
@@ -431,18 +486,48 @@ func (fd *FileDescription) Listxattr(ctx context.Context) ([]string, error) {
// Getxattr returns the value associated with the given extended attribute for
// the file represented by fd.
func (fd *FileDescription) Getxattr(ctx context.Context, name string) (string, error) {
+ if fd.opts.UseDentryMetadata {
+ vfsObj := fd.vd.mount.vfs
+ rp := vfsObj.getResolvingPath(auth.CredentialsFromContext(ctx), &PathOperation{
+ Root: fd.vd,
+ Start: fd.vd,
+ })
+ val, err := fd.vd.mount.fs.impl.GetxattrAt(ctx, rp, name)
+ vfsObj.putResolvingPath(rp)
+ return val, err
+ }
return fd.impl.Getxattr(ctx, name)
}
// Setxattr changes the value associated with the given extended attribute for
// the file represented by fd.
func (fd *FileDescription) Setxattr(ctx context.Context, opts SetxattrOptions) error {
+ if fd.opts.UseDentryMetadata {
+ vfsObj := fd.vd.mount.vfs
+ rp := vfsObj.getResolvingPath(auth.CredentialsFromContext(ctx), &PathOperation{
+ Root: fd.vd,
+ Start: fd.vd,
+ })
+ err := fd.vd.mount.fs.impl.SetxattrAt(ctx, rp, opts)
+ vfsObj.putResolvingPath(rp)
+ return err
+ }
return fd.impl.Setxattr(ctx, opts)
}
// Removexattr removes the given extended attribute from the file represented
// by fd.
func (fd *FileDescription) Removexattr(ctx context.Context, name string) error {
+ if fd.opts.UseDentryMetadata {
+ vfsObj := fd.vd.mount.vfs
+ rp := vfsObj.getResolvingPath(auth.CredentialsFromContext(ctx), &PathOperation{
+ Root: fd.vd,
+ Start: fd.vd,
+ })
+ err := fd.vd.mount.fs.impl.RemovexattrAt(ctx, rp, name)
+ vfsObj.putResolvingPath(rp)
+ return err
+ }
return fd.impl.Removexattr(ctx, name)
}
@@ -464,7 +549,7 @@ func (fd *FileDescription) MappedName(ctx context.Context) string {
// DeviceID implements memmap.MappingIdentity.DeviceID.
func (fd *FileDescription) DeviceID() uint64 {
- stat, err := fd.impl.Stat(context.Background(), StatOptions{
+ stat, err := fd.Stat(context.Background(), StatOptions{
// There is no STATX_DEV; we assume that Stat will return it if it's
// available regardless of mask.
Mask: 0,
@@ -480,7 +565,7 @@ func (fd *FileDescription) DeviceID() uint64 {
// InodeID implements memmap.MappingIdentity.InodeID.
func (fd *FileDescription) InodeID() uint64 {
- stat, err := fd.impl.Stat(context.Background(), StatOptions{
+ stat, err := fd.Stat(context.Background(), StatOptions{
Mask: linux.STATX_INO,
// fs/proc/task_mmu.c:show_map_vma() just reads inode::i_ino directly.
Sync: linux.AT_STATX_DONT_SYNC,
@@ -493,5 +578,5 @@ func (fd *FileDescription) InodeID() uint64 {
// Msync implements memmap.MappingIdentity.Msync.
func (fd *FileDescription) Msync(ctx context.Context, mr memmap.MappableRange) error {
- return fd.impl.Sync(ctx)
+ return fd.Sync(ctx)
}
diff --git a/pkg/sentry/vfs/file_description_impl_util.go b/pkg/sentry/vfs/file_description_impl_util.go
index 3df49991c..c00b3c84b 100644
--- a/pkg/sentry/vfs/file_description_impl_util.go
+++ b/pkg/sentry/vfs/file_description_impl_util.go
@@ -17,13 +17,13 @@ package vfs
import (
"bytes"
"io"
- "sync"
"gvisor.dev/gvisor/pkg/abi/linux"
"gvisor.dev/gvisor/pkg/sentry/arch"
"gvisor.dev/gvisor/pkg/sentry/context"
"gvisor.dev/gvisor/pkg/sentry/memmap"
"gvisor.dev/gvisor/pkg/sentry/usermem"
+ "gvisor.dev/gvisor/pkg/sync"
"gvisor.dev/gvisor/pkg/syserror"
"gvisor.dev/gvisor/pkg/waiter"
)
@@ -177,6 +177,21 @@ func (DirectoryFileDescriptionDefaultImpl) Write(ctx context.Context, src userme
return 0, syserror.EISDIR
}
+// DentryMetadataFileDescriptionImpl may be embedded by implementations of
+// FileDescriptionImpl for which FileDescriptionOptions.UseDentryMetadata is
+// true to obtain implementations of Stat and SetStat that panic.
+type DentryMetadataFileDescriptionImpl struct{}
+
+// Stat implements FileDescriptionImpl.Stat.
+func (DentryMetadataFileDescriptionImpl) Stat(ctx context.Context, opts StatOptions) (linux.Statx, error) {
+ panic("illegal call to DentryMetadataFileDescriptionImpl.Stat")
+}
+
+// SetStat implements FileDescriptionImpl.SetStat.
+func (DentryMetadataFileDescriptionImpl) SetStat(ctx context.Context, opts SetStatOptions) error {
+ panic("illegal call to DentryMetadataFileDescriptionImpl.SetStat")
+}
+
// DynamicBytesFileDescriptionImpl may be embedded by implementations of
// FileDescriptionImpl that represent read-only regular files whose contents
// are backed by a bytes.Buffer that is regenerated when necessary, consistent
@@ -199,6 +214,17 @@ type DynamicBytesSource interface {
Generate(ctx context.Context, buf *bytes.Buffer) error
}
+// StaticData implements DynamicBytesSource over a static string.
+type StaticData struct {
+ Data string
+}
+
+// Generate implements DynamicBytesSource.
+func (s *StaticData) Generate(ctx context.Context, buf *bytes.Buffer) error {
+ buf.WriteString(s.Data)
+ return nil
+}
+
// SetDataSource must be called exactly once on fd before first use.
func (fd *DynamicBytesFileDescriptionImpl) SetDataSource(data DynamicBytesSource) {
fd.data = data
diff --git a/pkg/sentry/vfs/file_description_impl_util_test.go b/pkg/sentry/vfs/file_description_impl_util_test.go
index 678be07fe..9ed58512f 100644
--- a/pkg/sentry/vfs/file_description_impl_util_test.go
+++ b/pkg/sentry/vfs/file_description_impl_util_test.go
@@ -89,7 +89,7 @@ func TestGenCountFD(t *testing.T) {
creds := auth.CredentialsFromContext(ctx)
vfsObj := New() // vfs.New()
- vfsObj.MustRegisterFilesystemType("testfs", FDTestFilesystemType{})
+ vfsObj.MustRegisterFilesystemType("testfs", FDTestFilesystemType{}, &RegisterFilesystemTypeOptions{})
mntns, err := vfsObj.NewMountNamespace(ctx, creds, "", "testfs", &GetFilesystemOptions{})
if err != nil {
t.Fatalf("failed to create testfs root mount: %v", err)
diff --git a/pkg/sentry/vfs/filesystem.go b/pkg/sentry/vfs/filesystem.go
index 89bd58864..ea78f555b 100644
--- a/pkg/sentry/vfs/filesystem.go
+++ b/pkg/sentry/vfs/filesystem.go
@@ -418,17 +418,38 @@ type FilesystemImpl interface {
UnlinkAt(ctx context.Context, rp *ResolvingPath) error
// ListxattrAt returns all extended attribute names for the file at rp.
+ //
+ // Errors:
+ //
+ // - If extended attributes are not supported by the filesystem,
+ // ListxattrAt returns nil. (See FileDescription.Listxattr for an
+ // explanation.)
ListxattrAt(ctx context.Context, rp *ResolvingPath) ([]string, error)
// GetxattrAt returns the value associated with the given extended
// attribute for the file at rp.
+ //
+ // Errors:
+ //
+ // - If extended attributes are not supported by the filesystem, GetxattrAt
+ // returns ENOTSUP.
GetxattrAt(ctx context.Context, rp *ResolvingPath, name string) (string, error)
// SetxattrAt changes the value associated with the given extended
// attribute for the file at rp.
+ //
+ // Errors:
+ //
+ // - If extended attributes are not supported by the filesystem, SetxattrAt
+ // returns ENOTSUP.
SetxattrAt(ctx context.Context, rp *ResolvingPath, opts SetxattrOptions) error
// RemovexattrAt removes the given extended attribute from the file at rp.
+ //
+ // Errors:
+ //
+ // - If extended attributes are not supported by the filesystem,
+ // RemovexattrAt returns ENOTSUP.
RemovexattrAt(ctx context.Context, rp *ResolvingPath, name string) error
// PrependPath prepends a path from vd to vd.Mount().Root() to b.
diff --git a/pkg/sentry/vfs/filesystem_type.go b/pkg/sentry/vfs/filesystem_type.go
index c335e206d..023301780 100644
--- a/pkg/sentry/vfs/filesystem_type.go
+++ b/pkg/sentry/vfs/filesystem_type.go
@@ -15,6 +15,7 @@
package vfs
import (
+ "bytes"
"fmt"
"gvisor.dev/gvisor/pkg/sentry/context"
@@ -43,28 +44,70 @@ type GetFilesystemOptions struct {
InternalData interface{}
}
+type registeredFilesystemType struct {
+ fsType FilesystemType
+ opts RegisterFilesystemTypeOptions
+}
+
+// RegisterFilesystemTypeOptions contains options to
+// VirtualFilesystem.RegisterFilesystem().
+type RegisterFilesystemTypeOptions struct {
+ // If AllowUserMount is true, allow calls to VirtualFilesystem.MountAt()
+ // for which MountOptions.InternalMount == false to use this filesystem
+ // type.
+ AllowUserMount bool
+
+ // If AllowUserList is true, make this filesystem type visible in
+ // /proc/filesystems.
+ AllowUserList bool
+
+ // If RequiresDevice is true, indicate that mounting this filesystem
+ // requires a block device as the mount source in /proc/filesystems.
+ RequiresDevice bool
+}
+
// RegisterFilesystemType registers the given FilesystemType in vfs with the
// given name.
-func (vfs *VirtualFilesystem) RegisterFilesystemType(name string, fsType FilesystemType) error {
+func (vfs *VirtualFilesystem) RegisterFilesystemType(name string, fsType FilesystemType, opts *RegisterFilesystemTypeOptions) error {
vfs.fsTypesMu.Lock()
defer vfs.fsTypesMu.Unlock()
if existing, ok := vfs.fsTypes[name]; ok {
- return fmt.Errorf("name %q is already registered to filesystem type %T", name, existing)
+ return fmt.Errorf("name %q is already registered to filesystem type %T", name, existing.fsType)
+ }
+ vfs.fsTypes[name] = &registeredFilesystemType{
+ fsType: fsType,
+ opts: *opts,
}
- vfs.fsTypes[name] = fsType
return nil
}
// MustRegisterFilesystemType is equivalent to RegisterFilesystemType but
// panics on failure.
-func (vfs *VirtualFilesystem) MustRegisterFilesystemType(name string, fsType FilesystemType) {
- if err := vfs.RegisterFilesystemType(name, fsType); err != nil {
+func (vfs *VirtualFilesystem) MustRegisterFilesystemType(name string, fsType FilesystemType, opts *RegisterFilesystemTypeOptions) {
+ if err := vfs.RegisterFilesystemType(name, fsType, opts); err != nil {
panic(fmt.Sprintf("failed to register filesystem type %T: %v", fsType, err))
}
}
-func (vfs *VirtualFilesystem) getFilesystemType(name string) FilesystemType {
+func (vfs *VirtualFilesystem) getFilesystemType(name string) *registeredFilesystemType {
vfs.fsTypesMu.RLock()
defer vfs.fsTypesMu.RUnlock()
return vfs.fsTypes[name]
}
+
+// GenerateProcFilesystems emits the contents of /proc/filesystems for vfs to
+// buf.
+func (vfs *VirtualFilesystem) GenerateProcFilesystems(buf *bytes.Buffer) {
+ vfs.fsTypesMu.RLock()
+ defer vfs.fsTypesMu.RUnlock()
+ for name, rft := range vfs.fsTypes {
+ if !rft.opts.AllowUserList {
+ continue
+ }
+ var nodev string
+ if !rft.opts.RequiresDevice {
+ nodev = "nodev"
+ }
+ fmt.Fprintf(buf, "%s\t%s\n", nodev, name)
+ }
+}
diff --git a/pkg/sentry/vfs/mount.go b/pkg/sentry/vfs/mount.go
index ec23ab0dd..00177b371 100644
--- a/pkg/sentry/vfs/mount.go
+++ b/pkg/sentry/vfs/mount.go
@@ -112,11 +112,11 @@ type MountNamespace struct {
// configured by the given arguments. A reference is taken on the returned
// MountNamespace.
func (vfs *VirtualFilesystem) NewMountNamespace(ctx context.Context, creds *auth.Credentials, source, fsTypeName string, opts *GetFilesystemOptions) (*MountNamespace, error) {
- fsType := vfs.getFilesystemType(fsTypeName)
- if fsType == nil {
+ rft := vfs.getFilesystemType(fsTypeName)
+ if rft == nil {
return nil, syserror.ENODEV
}
- fs, root, err := fsType.GetFilesystem(ctx, vfs, creds, source, *opts)
+ fs, root, err := rft.fsType.GetFilesystem(ctx, vfs, creds, source, *opts)
if err != nil {
return nil, err
}
@@ -136,11 +136,14 @@ func (vfs *VirtualFilesystem) NewMountNamespace(ctx context.Context, creds *auth
// MountAt creates and mounts a Filesystem configured by the given arguments.
func (vfs *VirtualFilesystem) MountAt(ctx context.Context, creds *auth.Credentials, source string, target *PathOperation, fsTypeName string, opts *MountOptions) error {
- fsType := vfs.getFilesystemType(fsTypeName)
- if fsType == nil {
+ rft := vfs.getFilesystemType(fsTypeName)
+ if rft == nil {
return syserror.ENODEV
}
- fs, root, err := fsType.GetFilesystem(ctx, vfs, creds, source, opts.GetFilesystemOptions)
+ if !opts.InternalMount && !rft.opts.AllowUserMount {
+ return syserror.ENODEV
+ }
+ fs, root, err := rft.fsType.GetFilesystem(ctx, vfs, creds, source, opts.GetFilesystemOptions)
if err != nil {
return err
}
diff --git a/pkg/sentry/vfs/mount_test.go b/pkg/sentry/vfs/mount_test.go
index adff0b94b..3b933468d 100644
--- a/pkg/sentry/vfs/mount_test.go
+++ b/pkg/sentry/vfs/mount_test.go
@@ -17,8 +17,9 @@ package vfs
import (
"fmt"
"runtime"
- "sync"
"testing"
+
+ "gvisor.dev/gvisor/pkg/sync"
)
func TestMountTableLookupEmpty(t *testing.T) {
diff --git a/pkg/sentry/vfs/mount_unsafe.go b/pkg/sentry/vfs/mount_unsafe.go
index ab13fa461..bd90d36c4 100644
--- a/pkg/sentry/vfs/mount_unsafe.go
+++ b/pkg/sentry/vfs/mount_unsafe.go
@@ -26,7 +26,7 @@ import (
"sync/atomic"
"unsafe"
- "gvisor.dev/gvisor/pkg/syncutil"
+ "gvisor.dev/gvisor/pkg/sync"
)
// mountKey represents the location at which a Mount is mounted. It is
@@ -75,7 +75,7 @@ type mountTable struct {
// intrinsics and inline assembly, limiting the performance of this
// approach.)
- seq syncutil.SeqCount
+ seq sync.SeqCount
seed uint32 // for hashing keys
// size holds both length (number of elements) and capacity (number of
diff --git a/pkg/sentry/vfs/options.go b/pkg/sentry/vfs/options.go
index 87d2b0d1c..b7774bf28 100644
--- a/pkg/sentry/vfs/options.go
+++ b/pkg/sentry/vfs/options.go
@@ -50,6 +50,10 @@ type MknodOptions struct {
type MountOptions struct {
// GetFilesystemOptions contains options to FilesystemType.GetFilesystem().
GetFilesystemOptions GetFilesystemOptions
+
+ // If InternalMount is true, allow the use of filesystem types for which
+ // RegisterFilesystemTypeOptions.AllowUserMount == false.
+ InternalMount bool
}
// OpenOptions contains options to VirtualFilesystem.OpenAt() and
diff --git a/pkg/sentry/vfs/pathname.go b/pkg/sentry/vfs/pathname.go
index 8e155654f..cf80df90e 100644
--- a/pkg/sentry/vfs/pathname.go
+++ b/pkg/sentry/vfs/pathname.go
@@ -15,10 +15,9 @@
package vfs
import (
- "sync"
-
"gvisor.dev/gvisor/pkg/fspath"
"gvisor.dev/gvisor/pkg/sentry/context"
+ "gvisor.dev/gvisor/pkg/sync"
"gvisor.dev/gvisor/pkg/syserror"
)
diff --git a/pkg/sentry/vfs/permissions.go b/pkg/sentry/vfs/permissions.go
index f1edb0680..d279d05ca 100644
--- a/pkg/sentry/vfs/permissions.go
+++ b/pkg/sentry/vfs/permissions.go
@@ -30,6 +30,26 @@ const (
MayExec = 1
)
+// OnlyRead returns true if access _only_ allows read.
+func (a AccessTypes) OnlyRead() bool {
+ return a == MayRead
+}
+
+// MayRead returns true if access allows read.
+func (a AccessTypes) MayRead() bool {
+ return a&MayRead != 0
+}
+
+// MayWrite returns true if access allows write.
+func (a AccessTypes) MayWrite() bool {
+ return a&MayWrite != 0
+}
+
+// MayExec returns true if access allows exec.
+func (a AccessTypes) MayExec() bool {
+ return a&MayExec != 0
+}
+
// GenericCheckPermissions checks that creds has the given access rights on a
// file with the given permissions, UID, and GID, subject to the rules of
// fs/namei.c:generic_permission(). isDir is true if the file is a directory.
@@ -53,7 +73,7 @@ func GenericCheckPermissions(creds *auth.Credentials, ats AccessTypes, isDir boo
}
// CAP_DAC_READ_SEARCH allows the caller to read and search arbitrary
// directories, and read arbitrary non-directory files.
- if (isDir && (ats&MayWrite == 0)) || ats == MayRead {
+ if (isDir && !ats.MayWrite()) || ats.OnlyRead() {
if creds.HasCapability(linux.CAP_DAC_READ_SEARCH) {
return nil
}
@@ -61,7 +81,7 @@ func GenericCheckPermissions(creds *auth.Credentials, ats AccessTypes, isDir boo
// CAP_DAC_OVERRIDE allows arbitrary access to directories, read/write
// access to non-directory files, and execute access to non-directory files
// for which at least one execute bit is set.
- if isDir || (ats&MayExec == 0) || (mode&0111 != 0) {
+ if isDir || !ats.MayExec() || (mode&0111 != 0) {
if creds.HasCapability(linux.CAP_DAC_OVERRIDE) {
return nil
}
diff --git a/pkg/sentry/vfs/resolving_path.go b/pkg/sentry/vfs/resolving_path.go
index f0641d314..8a0b382f6 100644
--- a/pkg/sentry/vfs/resolving_path.go
+++ b/pkg/sentry/vfs/resolving_path.go
@@ -16,11 +16,11 @@ package vfs
import (
"fmt"
- "sync"
"gvisor.dev/gvisor/pkg/abi/linux"
"gvisor.dev/gvisor/pkg/fspath"
"gvisor.dev/gvisor/pkg/sentry/kernel/auth"
+ "gvisor.dev/gvisor/pkg/sync"
"gvisor.dev/gvisor/pkg/syserror"
)
diff --git a/pkg/sentry/vfs/vfs.go b/pkg/sentry/vfs/vfs.go
index 3e4df8558..1f21b0b31 100644
--- a/pkg/sentry/vfs/vfs.go
+++ b/pkg/sentry/vfs/vfs.go
@@ -29,12 +29,12 @@ package vfs
import (
"fmt"
- "sync"
"gvisor.dev/gvisor/pkg/abi/linux"
"gvisor.dev/gvisor/pkg/fspath"
"gvisor.dev/gvisor/pkg/sentry/context"
"gvisor.dev/gvisor/pkg/sentry/kernel/auth"
+ "gvisor.dev/gvisor/pkg/sync"
"gvisor.dev/gvisor/pkg/syserror"
)
@@ -75,23 +75,29 @@ type VirtualFilesystem struct {
// mountpoints is analogous to Linux's mountpoint_hashtable.
mountpoints map[*Dentry]map[*Mount]struct{}
+ // devices contains all registered Devices. devices is protected by
+ // devicesMu.
+ devicesMu sync.RWMutex
+ devices map[devTuple]*registeredDevice
+
+ // fsTypes contains all registered FilesystemTypes. fsTypes is protected by
+ // fsTypesMu.
+ fsTypesMu sync.RWMutex
+ fsTypes map[string]*registeredFilesystemType
+
// filesystems contains all Filesystems. filesystems is protected by
// filesystemsMu.
filesystemsMu sync.Mutex
filesystems map[*Filesystem]struct{}
-
- // fsTypes contains all FilesystemTypes that are usable in the
- // VirtualFilesystem. fsTypes is protected by fsTypesMu.
- fsTypesMu sync.RWMutex
- fsTypes map[string]FilesystemType
}
// New returns a new VirtualFilesystem with no mounts or FilesystemTypes.
func New() *VirtualFilesystem {
vfs := &VirtualFilesystem{
mountpoints: make(map[*Dentry]map[*Mount]struct{}),
+ devices: make(map[devTuple]*registeredDevice),
+ fsTypes: make(map[string]*registeredFilesystemType),
filesystems: make(map[*Filesystem]struct{}),
- fsTypes: make(map[string]FilesystemType),
}
vfs.mounts.Init()
return vfs
generated by cgit v1.2.3 (git 2.39.1) at 2025-04-14 19:02:21 +0000