Files
containerd/core/mount/temp.go
Rodrigo Campos 41953f7ac1 cri: Fix userns with Dockerfile VOLUME mounts that need copy
If a Dockerfile is using a `VOLUME` directive and the directory exists
in the rootfs, like in this example:

	FROM docker.io/library/alpine:latest
	VOLUME [ "/run" ]

The alpine container image already contains a "/run" directory. This
will force the code in WithVolumes() to copy its  content to the new
volume created for the VOLUME directive. This copies the content as well
as the ownership.

However, as we perform the mounts from the host POV without being inside
a userns, the idmap option will just shift the IDs in ways that will
screw up the ownerships when copied. We should only use the idmap option
when running the container inside a userns, so the ownerships are fine
(the userns will do a shift and the idmap another, to make it all seem
as if there was no UID/GID shift in the first place).

This PR does just that, remove the idmap option from mounts so we copy
the files without any ID transformations. It's simpler and easier to
reason about if we just don't mount with the idmap option here: all
files are copied just fine without ID transformations and ID
transformation is applied via the idmap option at mount time when
running the pod.

Also, note that `VOLUME` directives that refer to directories that don't
exist on the rootfs work fine (`VOLUME [ "/rata" ]` for example), as
there is no copy done in that case so the permissions weren't changed.

Signed-off-by: Rodrigo Campos <rodrigoca@microsoft.com>
2025-08-26 11:15:55 +02:00

147 lines
4.4 KiB
Go

/*
Copyright The containerd 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 mount
import (
"context"
"fmt"
"os"
"strings"
"github.com/containerd/log"
)
var tempMountLocation = getTempDir()
// WithTempMount mounts the provided mounts to a temp dir, and pass the temp dir to f.
// The mounts are valid during the call to the f.
// Finally we will unmount and remove the temp dir regardless of the result of f.
//
// NOTE: The volatile option of overlayfs doesn't allow to mount again using the
// same upper / work dirs. Since it's a temp mount, avoid using that option here
// if found.
func WithTempMount(ctx context.Context, mounts []Mount, f func(root string) error) (err error) {
root, uerr := os.MkdirTemp(tempMountLocation, "containerd-mount")
if uerr != nil {
return fmt.Errorf("failed to create temp dir: %w", uerr)
}
// We use Remove here instead of RemoveAll.
// The RemoveAll will delete the temp dir and all children it contains.
// When the Unmount fails, RemoveAll will incorrectly delete data from
// the mounted dir. However, if we use Remove, even though we won't
// successfully delete the temp dir and it may leak, we won't loss data
// from the mounted dir.
// For details, please refer to #1868 #1785.
defer func() {
if uerr = os.Remove(root); uerr != nil {
log.G(ctx).WithError(uerr).WithField("dir", root).Error("failed to remove mount temp dir")
}
}()
// We should do defer first, if not we will not do Unmount when only a part of Mounts are failed.
defer func() {
if uerr = UnmountMounts(mounts, root, 0); uerr != nil {
uerr = fmt.Errorf("failed to unmount %s: %w", root, uerr)
if err == nil {
err = uerr
} else {
err = fmt.Errorf("%s: %w", uerr.Error(), err)
}
}
}()
if uerr = All(RemoveVolatileOption(mounts), root); uerr != nil {
return fmt.Errorf("failed to mount %s: %w", root, uerr)
}
if err := f(root); err != nil {
return fmt.Errorf("mount callback failed on %s: %w", root, err)
}
return nil
}
// RemoveVolatileOption copies and remove the volatile option for overlay
// type, since overlayfs doesn't allow to mount again using the same upper/work
// dirs.
//
// REF: https://docs.kernel.org/filesystems/overlayfs.html#volatile-mount
//
// TODO: Make this logic conditional once the kernel supports reusing
// overlayfs volatile mounts.
func RemoveVolatileOption(mounts []Mount) []Mount {
var out []Mount
for i, m := range mounts {
if m.Type != "overlay" {
continue
}
for j, opt := range m.Options {
if opt == "volatile" {
if out == nil {
out = copyMounts(mounts)
}
out[i].Options = append(out[i].Options[:j], out[i].Options[j+1:]...)
break
}
}
}
if out != nil {
return out
}
return mounts
}
// RemoveIDMapOption copies and removes the uidmap/gidmap options on any of the mounts using it.
func RemoveIDMapOption(mounts []Mount) []Mount {
var out []Mount
for i, m := range mounts {
for j, opt := range m.Options {
if strings.HasPrefix(opt, "uidmap") || strings.HasPrefix(opt, "gidmap") {
if out == nil {
out = copyMounts(mounts)
}
out[i].Options = append(out[i].Options[:j], out[i].Options[j+1:]...)
}
}
}
if out != nil {
return out
}
return mounts
}
// copyMounts creates a copy of the original slice to allow for modification and not altering the original
func copyMounts(in []Mount) []Mount {
out := make([]Mount, len(in))
copy(out, in)
return out
}
// WithReadonlyTempMount mounts the provided mounts to a temp dir as readonly,
// and pass the temp dir to f. The mounts are valid during the call to the f.
// Finally we will unmount and remove the temp dir regardless of the result of f.
func WithReadonlyTempMount(ctx context.Context, mounts []Mount, f func(root string) error) (err error) {
return WithTempMount(ctx, readonlyMounts(mounts), f)
}
func getTempDir() string {
if xdg := os.Getenv("XDG_RUNTIME_DIR"); xdg != "" {
return xdg
}
return os.TempDir()
}