Fix problem with open FD when copy up is triggered in overlayfs

Linux kernel before 4.19 doesn't implement a feature that updates
open FD after a file is open for write (and is copied to the upper
layer). Already open FD will continue to read the old file content
until they are reopened. This is especially problematic for gVisor
because it caches open files.

Flag was added to force readonly files to be reopenned when the
same file is open for write. This is only needed if using kernels
prior to 4.19.

Closes #1006

It's difficult to really test this because we never run on tests
on older kernels. I'm adding a test in GKE which uses kernels
with the overlayfs problem for 1.14 and lower.

PiperOrigin-RevId: 275115289

3 files changed, 61 insertions, 2 deletions

diff --git a/pkg/sentry/fs/fsutil/host_file_mapper.go b/pkg/sentry/fs/fsutil/host_file_mapper.go
index e239f12a5..b06a71cc2 100644
--- a/pkg/sentry/fs/fsutil/host_file_mapper.go
+++ b/pkg/sentry/fs/fsutil/host_file_mapper.go
@@ -209,3 +209,29 @@ func (f *HostFileMapper) unmapAndRemoveLocked(chunkStart uint64, m mapping) {
 	}
 	delete(f.mappings, chunkStart)
 }
+
+// RegenerateMappings must be called when the file description mapped by f
+// changes, to replace existing mappings of the previous file description.
+func (f *HostFileMapper) RegenerateMappings(fd int) error {
+	f.mapsMu.Lock()
+	defer f.mapsMu.Unlock()
+
+	for chunkStart, m := range f.mappings {
+		prot := syscall.PROT_READ
+		if m.writable {
+			prot |= syscall.PROT_WRITE
+		}
+		_, _, errno := syscall.Syscall6(
+			syscall.SYS_MMAP,
+			m.addr,
+			chunkSize,
+			uintptr(prot),
+			syscall.MAP_SHARED|syscall.MAP_FIXED,
+			uintptr(fd),
+			uintptr(chunkStart))
+		if errno != 0 {
+			return errno
+		}
+	}
+	return nil
+}
diff --git a/pkg/sentry/fs/fsutil/host_mappable.go b/pkg/sentry/fs/fsutil/host_mappable.go
index 693625ddc..30475f340 100644
--- a/pkg/sentry/fs/fsutil/host_mappable.go
+++ b/pkg/sentry/fs/fsutil/host_mappable.go
@@ -100,13 +100,30 @@ func (h *HostMappable) Translate(ctx context.Context, required, optional memmap.
 }
 
 // InvalidateUnsavable implements memmap.Mappable.InvalidateUnsavable.
-func (h *HostMappable) InvalidateUnsavable(ctx context.Context) error {
+func (h *HostMappable) InvalidateUnsavable(_ context.Context) error {
 	h.mu.Lock()
 	h.mappings.InvalidateAll(memmap.InvalidateOpts{})
 	h.mu.Unlock()
 	return nil
 }
 
+// NotifyChangeFD must be called after the file description represented by
+// CachedFileObject.FD() changes.
+func (h *HostMappable) NotifyChangeFD() error {
+	// Update existing sentry mappings to refer to the new file description.
+	if err := h.hostFileMapper.RegenerateMappings(h.backingFile.FD()); err != nil {
+		return err
+	}
+
+	// Shoot down existing application mappings of the old file description;
+	// they will be remapped with the new file description on demand.
+	h.mu.Lock()
+	defer h.mu.Unlock()
+
+	h.mappings.InvalidateAll(memmap.InvalidateOpts{})
+	return nil
+}
+
 // MapInternal implements platform.File.MapInternal.
 func (h *HostMappable) MapInternal(fr platform.FileRange, at usermem.AccessType) (safemem.BlockSeq, error) {
 	return h.hostFileMapper.MapInternal(fr, h.backingFile.FD(), at.Write)
diff --git a/pkg/sentry/fs/fsutil/inode_cached.go b/pkg/sentry/fs/fsutil/inode_cached.go
index dd80757dc..798920d18 100644
--- a/pkg/sentry/fs/fsutil/inode_cached.go
+++ b/pkg/sentry/fs/fsutil/inode_cached.go
@@ -959,6 +959,23 @@ func (c *CachingInodeOperations) InvalidateUnsavable(ctx context.Context) error
 	return nil
 }
 
+// NotifyChangeFD must be called after the file description represented by
+// CachedFileObject.FD() changes.
+func (c *CachingInodeOperations) NotifyChangeFD() error {
+	// Update existing sentry mappings to refer to the new file description.
+	if err := c.hostFileMapper.RegenerateMappings(c.backingFile.FD()); err != nil {
+		return err
+	}
+
+	// Shoot down existing application mappings of the old file description;
+	// they will be remapped with the new file description on demand.
+	c.mapsMu.Lock()
+	defer c.mapsMu.Unlock()
+
+	c.mappings.InvalidateAll(memmap.InvalidateOpts{})
+	return nil
+}
+
 // Evict implements pgalloc.EvictableMemoryUser.Evict.
 func (c *CachingInodeOperations) Evict(ctx context.Context, er pgalloc.EvictableRange) {
 	c.mapsMu.Lock()
@@ -1027,7 +1044,6 @@ func (c *CachingInodeOperations) DecRef(fr platform.FileRange) {
 	}
 	c.refs.MergeAdjacent(fr)
 	c.dataMu.Unlock()
-
 }
 
 // MapInternal implements platform.File.MapInternal. This is used when we