Fix procfs bugs in vfs2.

- Support writing on proc/[pid]/{uid,gid}map
- Return EIO for writing to static files.

Updates #2923.

PiperOrigin-RevId: 318188503

5 files changed, 68 insertions, 8 deletions

diff --git a/pkg/sentry/fsimpl/kernfs/dynamic_bytes_file.go b/pkg/sentry/fsimpl/kernfs/dynamic_bytes_file.go
index c1215b70a..6886b0876 100644
--- a/pkg/sentry/fsimpl/kernfs/dynamic_bytes_file.go
+++ b/pkg/sentry/fsimpl/kernfs/dynamic_bytes_file.go
@@ -101,12 +101,12 @@ func (fd *DynamicBytesFD) Seek(ctx context.Context, offset int64, whence int32)
 	return fd.DynamicBytesFileDescriptionImpl.Seek(ctx, offset, whence)
 }
 
-// Read implmenets vfs.FileDescriptionImpl.Read.
+// Read implements vfs.FileDescriptionImpl.Read.
 func (fd *DynamicBytesFD) Read(ctx context.Context, dst usermem.IOSequence, opts vfs.ReadOptions) (int64, error) {
 	return fd.DynamicBytesFileDescriptionImpl.Read(ctx, dst, opts)
 }
 
-// PRead implmenets vfs.FileDescriptionImpl.PRead.
+// PRead implements vfs.FileDescriptionImpl.PRead.
 func (fd *DynamicBytesFD) PRead(ctx context.Context, dst usermem.IOSequence, offset int64, opts vfs.ReadOptions) (int64, error) {
 	return fd.DynamicBytesFileDescriptionImpl.PRead(ctx, dst, offset, opts)
 }
diff --git a/pkg/sentry/fsimpl/kernfs/inode_impl_util.go b/pkg/sentry/fsimpl/kernfs/inode_impl_util.go
index 650bd7b88..53aec4918 100644
--- a/pkg/sentry/fsimpl/kernfs/inode_impl_util.go
+++ b/pkg/sentry/fsimpl/kernfs/inode_impl_util.go
@@ -293,6 +293,8 @@ func (a *InodeAttrs) SetStat(ctx context.Context, fs *vfs.Filesystem, creds *aut
 	// inode numbers are immutable after node creation.
 
 	// TODO(gvisor.dev/issue/1193): Implement other stat fields like timestamps.
+	// Also, STATX_SIZE will need some special handling, because read-only static
+	// files should return EIO for truncate operations.
 
 	return nil
 }
diff --git a/pkg/sentry/fsimpl/proc/task_files.go b/pkg/sentry/fsimpl/proc/task_files.go
index ba4405026..286c23f01 100644
--- a/pkg/sentry/fsimpl/proc/task_files.go
+++ b/pkg/sentry/fsimpl/proc/task_files.go
@@ -35,6 +35,10 @@ import (
 	"gvisor.dev/gvisor/pkg/usermem"
 )
 
+// "There is an (arbitrary) limit on the number of lines in the file. As at
+// Linux 3.18, the limit is five lines." - user_namespaces(7)
+const maxIDMapLines = 5
+
 // mm gets the kernel task's MemoryManager. No additional reference is taken on
 // mm here. This is safe because MemoryManager.destroy is required to leave the
 // MemoryManager in a state where it's still usable as a DynamicBytesSource.
@@ -283,7 +287,8 @@ func (d *commData) Generate(ctx context.Context, buf *bytes.Buffer) error {
 	return nil
 }
 
-// idMapData implements vfs.DynamicBytesSource for /proc/[pid]/{gid_map|uid_map}.
+// idMapData implements vfs.WritableDynamicBytesSource for
+// /proc/[pid]/{gid_map|uid_map}.
 //
 // +stateify savable
 type idMapData struct {
@@ -309,6 +314,59 @@ func (d *idMapData) Generate(ctx context.Context, buf *bytes.Buffer) error {
 	return nil
 }
 
+func (d *idMapData) Write(ctx context.Context, src usermem.IOSequence, offset int64) (int64, error) {
+	// "In addition, the number of bytes written to the file must be less than
+	// the system page size, and the write must be performed at the start of
+	// the file ..." - user_namespaces(7)
+	srclen := src.NumBytes()
+	if srclen >= usermem.PageSize || offset != 0 {
+		return 0, syserror.EINVAL
+	}
+	b := make([]byte, srclen)
+	if _, err := src.CopyIn(ctx, b); err != nil {
+		return 0, err
+	}
+
+	// Truncate from the first NULL byte.
+	var nul int64
+	nul = int64(bytes.IndexByte(b, 0))
+	if nul == -1 {
+		nul = srclen
+	}
+	b = b[:nul]
+	// Remove the last \n.
+	if nul >= 1 && b[nul-1] == '\n' {
+		b = b[:nul-1]
+	}
+	lines := bytes.SplitN(b, []byte("\n"), maxIDMapLines+1)
+	if len(lines) > maxIDMapLines {
+		return 0, syserror.EINVAL
+	}
+
+	entries := make([]auth.IDMapEntry, len(lines))
+	for i, l := range lines {
+		var e auth.IDMapEntry
+		_, err := fmt.Sscan(string(l), &e.FirstID, &e.FirstParentID, &e.Length)
+		if err != nil {
+			return 0, syserror.EINVAL
+		}
+		entries[i] = e
+	}
+	var err error
+	if d.gids {
+		err = d.task.UserNamespace().SetGIDMap(ctx, entries)
+	} else {
+		err = d.task.UserNamespace().SetUIDMap(ctx, entries)
+	}
+	if err != nil {
+		return 0, err
+	}
+
+	// On success, Linux's kernel/user_namespace.c:map_write() always returns
+	// count, even if fewer bytes were used.
+	return int64(srclen), nil
+}
+
 // mapsData implements vfs.DynamicBytesSource for /proc/[pid]/maps.
 //
 // +stateify savable
diff --git a/pkg/sentry/vfs/file_description_impl_util.go b/pkg/sentry/vfs/file_description_impl_util.go
index 1e66997ce..3fec0d6d6 100644
--- a/pkg/sentry/vfs/file_description_impl_util.go
+++ b/pkg/sentry/vfs/file_description_impl_util.go
@@ -327,7 +327,7 @@ func (fd *DynamicBytesFileDescriptionImpl) pwriteLocked(ctx context.Context, src
 
 	writable, ok := fd.data.(WritableDynamicBytesSource)
 	if !ok {
-		return 0, syserror.EINVAL
+		return 0, syserror.EIO
 	}
 	n, err := writable.Write(ctx, src, offset)
 	if err != nil {
diff --git a/pkg/sentry/vfs/file_description_impl_util_test.go b/pkg/sentry/vfs/file_description_impl_util_test.go
index 5061f6ac9..3b7e1c273 100644
--- a/pkg/sentry/vfs/file_description_impl_util_test.go
+++ b/pkg/sentry/vfs/file_description_impl_util_test.go
@@ -155,11 +155,11 @@ func TestGenCountFD(t *testing.T) {
 	}
 
 	// Write and PWrite fails.
-	if _, err := fd.Write(ctx, ioseq, WriteOptions{}); err != syserror.EINVAL {
-		t.Errorf("Write: got err %v, wanted %v", err, syserror.EINVAL)
+	if _, err := fd.Write(ctx, ioseq, WriteOptions{}); err != syserror.EIO {
+		t.Errorf("Write: got err %v, wanted %v", err, syserror.EIO)
 	}
-	if _, err := fd.PWrite(ctx, ioseq, 0, WriteOptions{}); err != syserror.EINVAL {
-		t.Errorf("Write: got err %v, wanted %v", err, syserror.EINVAL)
+	if _, err := fd.PWrite(ctx, ioseq, 0, WriteOptions{}); err != syserror.EIO {
+		t.Errorf("Write: got err %v, wanted %v", err, syserror.EIO)
 	}
 }