podman rmi should only untag image if parent of another

podman rmi was deleting an image even if it was a parent of
another image. This fix just untags the image instead.
This also fixes podman rmi to remove intermediate images of
an image when the image is removed.

Signed-off-by: umohnani8 <umohnani@redhat.com>

Closes: #1055
Approved by: mheon

cmd/podman/images.go

@@ -7,7 +7,8 @@ import (
 	"strings"
 	"time"
 
-	"github.com/containers/storage"
+	"github.com/sirupsen/logrus"
+
 	"github.com/docker/go-units"
 	digest "github.com/opencontainers/go-digest"
 	"github.com/pkg/errors"
@@ -259,12 +260,16 @@ func sortImagesOutput(sortBy string, imagesOutput imagesSorted) imagesSorted {
 }
 
 // getImagesTemplateOutput returns the images information to be printed in human readable format
-func getImagesTemplateOutput(ctx context.Context, runtime *libpod.Runtime, images []*image.Image, opts imagesOptions, storageLayers []storage.Layer) (imagesOutput imagesSorted) {
+func getImagesTemplateOutput(ctx context.Context, runtime *libpod.Runtime, images []*image.Image, opts imagesOptions) (imagesOutput imagesSorted) {
 	for _, img := range images {
 		// If all is false and the image doesn't have a name, check to see if the top layer of the image is a parent
 		// to another image's top layer. If it is, then it is an intermediate image so don't print out if the --all flag
 		// is not set.
-		if !opts.all && len(img.Names()) == 0 && !layerIsLeaf(storageLayers, img.TopLayer()) {
+		isParent, err := img.IsParent()
+		if err != nil {
+			logrus.Errorf("error checking if image is a parent %q: %v", img.ID(), err)
+		}
+		if !opts.all && len(img.Names()) == 0 && isParent {
 			continue
 		}
 		createdTime := img.Created()
@@ -325,17 +330,13 @@ func generateImagesOutput(ctx context.Context, runtime *libpod.Runtime, images [
 		return nil
 	}
 	var out formats.Writer
-	storageLayers, err := runtime.GetLayers()
-	if err != nil {
-		return errors.Wrap(err, "error getting layers from store")
-	}
 
 	switch opts.format {
 	case formats.JSONString:
 		imagesOutput := getImagesJSONOutput(ctx, runtime, images)
 		out = formats.JSONStructArray{Output: imagesToGeneric([]imagesTemplateParams{}, imagesOutput)}
 	default:
-		imagesOutput := getImagesTemplateOutput(ctx, runtime, images, opts, storageLayers)
+		imagesOutput := getImagesTemplateOutput(ctx, runtime, images, opts)
 		out = formats.StdoutTemplateArray{Output: imagesToGeneric(imagesOutput, []imagesJSONParams{}), Template: opts.outputformat, Fields: imagesOutput[0].HeaderMap()}
 	}
 	return formats.Writer(out).Out()
@@ -391,14 +392,3 @@ func CreateFilterFuncs(ctx context.Context, r *libpod.Runtime, c *cli.Context, i
 	}
 	return filterFuncs, nil
 }
-
-// layerIsLeaf goes through the layers in the store and checks if "layer" is
-// the parent of any other layer in store.
-func layerIsLeaf(storageLayers []storage.Layer, layer string) bool {
-	for _, storeLayer := range storageLayers {
-		if storeLayer.Parent == layer {
-			return false
-		}
-	}
-	return true
-}

libpod/image/image.go

@@ -59,6 +59,9 @@ type Runtime struct {
 	SignaturePolicyPath string
 }
 
+// ErrRepoTagNotFound is the error returned when the image id given doesn't match a rep tag in store
+var ErrRepoTagNotFound = errors.New("unable to match user input to any specific repotag")
+
 // NewImageRuntimeFromStore creates an ImageRuntime based on a provided store
 func NewImageRuntimeFromStore(store storage.Store) *Runtime {
 	return &Runtime{
@@ -333,9 +336,39 @@ func (i *Image) TopLayer() string {
 
 // Remove an image; container removal for the image must be done
 // outside the context of images
+// TODO: the force param does nothing as of now. Need to move container
+// handling logic here eventually.
 func (i *Image) Remove(force bool) error {
-	_, err := i.imageruntime.store.DeleteImage(i.ID(), true)
+	parent, err := i.GetParent()
-	return err
+	if err != nil {
+		return err
+	}
+	if _, err := i.imageruntime.store.DeleteImage(i.ID(), true); err != nil {
+		return err
+	}
+	for parent != nil {
+		nextParent, err := parent.GetParent()
+		if err != nil {
+			return err
+		}
+		children, err := parent.GetChildren()
+		if err != nil {
+			return err
+		}
+		// Do not remove if image is a base image and is not untagged, or if
+		// the image has more children.
+		if (nextParent == nil && len(parent.Names()) > 0) || len(children) > 0 {
+			return nil
+		}
+		id := parent.ID()
+		if _, err := i.imageruntime.store.DeleteImage(id, true); err != nil {
+			logrus.Debugf("unable to remove intermediate image %q: %v", id, err)
+		} else {
+			fmt.Println(id)
+		}
+		parent = nextParent
+	}
+	return nil
 }
 
 // Decompose an Image
@@ -902,7 +935,7 @@ func (i *Image) MatchRepoTag(input string) (string, error) {
 		}
 	}
 	if maxCount == 0 {
-		return "", errors.Errorf("unable to match user input to any specific repotag")
+		return "", ErrRepoTagNotFound
 	}
 	if len(results[maxCount]) > 1 {
 		return "", errors.Errorf("user input matched multiple repotags for the image")
@@ -916,6 +949,68 @@ func splitString(input string) string {
 	return split[len(split)-1]
 }
 
+// IsParent goes through the layers in the store and checks if i.TopLayer is
+// the parent of any other layer in store. Double check that image with that
+// layer exists as well.
+func (i *Image) IsParent() (bool, error) {
+	children, err := i.GetChildren()
+	if err != nil {
+		return false, err
+	}
+	return len(children) > 0, nil
+}
+
+// GetParent returns the image ID of the parent. Return nil if a parent is not found.
+func (i *Image) GetParent() (*Image, error) {
+	images, err := i.imageruntime.GetImages()
+	if err != nil {
+		return nil, err
+	}
+	layer, err := i.imageruntime.store.Layer(i.TopLayer())
+	if err != nil {
+		return nil, err
+	}
+	for _, img := range images {
+		if img.TopLayer() == layer.Parent {
+			return img, nil
+		}
+	}
+	return nil, nil
+}
+
+// GetChildren returns a list of the imageIDs that depend on the image
+func (i *Image) GetChildren() ([]string, error) {
+	var children []string
+	images, err := i.imageruntime.GetImages()
+	if err != nil {
+		return nil, err
+	}
+	layers, err := i.imageruntime.store.Layers()
+	if err != nil {
+		return nil, err
+	}
+
+	for _, layer := range layers {
+		if layer.Parent == i.TopLayer() {
+			if imageID := getImageOfTopLayer(images, layer.ID); len(imageID) > 0 {
+				children = append(children, imageID...)
+			}
+		}
+	}
+	return children, nil
+}
+
+// getImageOfTopLayer returns the image ID where layer is the top layer of the image
+func getImageOfTopLayer(images []*Image, layer string) []string {
+	var matches []string
+	for _, img := range images {
+		if img.TopLayer() == layer {
+			matches = append(matches, img.ID())
+		}
+	}
+	return matches
+}
+
 // InputIsID returns a bool if the user input for an image
 // is the image's partial or full id
 func (i *Image) InputIsID() bool {

libpod/runtime_img.go

@@ -75,7 +75,7 @@ type CopyOptions struct {
 
 // RemoveImage deletes an image from local storage
 // Images being used by running containers can only be removed if force=true
-func (r *Runtime) RemoveImage(ctx context.Context, image *image.Image, force bool) (string, error) {
+func (r *Runtime) RemoveImage(ctx context.Context, img *image.Image, force bool) (string, error) {
 	r.lock.Lock()
 	defer r.lock.Unlock()
 
@@ -90,50 +90,58 @@ func (r *Runtime) RemoveImage(ctx context.Context, image *image.Image, force boo
 	}
 	imageCtrs := []*Container{}
 	for _, ctr := range ctrs {
-		if ctr.config.RootfsImageID == image.ID() {
+		if ctr.config.RootfsImageID == img.ID() {
 			imageCtrs = append(imageCtrs, ctr)
 		}
 	}
-	if len(imageCtrs) > 0 && len(image.Names()) <= 1 {
+	if len(imageCtrs) > 0 && len(img.Names()) <= 1 {
 		if force {
 			for _, ctr := range imageCtrs {
 				if err := r.removeContainer(ctx, ctr, true); err != nil {
-					return "", errors.Wrapf(err, "error removing image %s: container %s using image could not be removed", image.ID(), ctr.ID())
+					return "", errors.Wrapf(err, "error removing image %s: container %s using image could not be removed", img.ID(), ctr.ID())
 				}
 			}
 		} else {
-			return "", fmt.Errorf("could not remove image %s as it is being used by %d containers", image.ID(), len(imageCtrs))
+			return "", fmt.Errorf("could not remove image %s as it is being used by %d containers", img.ID(), len(imageCtrs))
 		}
 	}
 
-	if len(image.Names()) > 1 && !image.InputIsID() {
+	hasChildren, err := img.IsParent()
+	if err != nil {
+		return "", err
+	}
+
+	if (len(img.Names()) > 1 && !img.InputIsID()) || hasChildren {
 		// If the image has multiple reponames, we do not technically delete
 		// the image. we figure out which repotag the user is trying to refer
 		// to and untag it.
-		repoName, err := image.MatchRepoTag(image.InputName)
+		repoName, err := img.MatchRepoTag(img.InputName)
+		if hasChildren && err == image.ErrRepoTagNotFound {
+			return "", errors.Errorf("unable to delete %q (cannot be forced) - image has dependent child images", img.ID())
+		}
 		if err != nil {
 			return "", err
 		}
-		if err := image.UntagImage(repoName); err != nil {
+		if err := img.UntagImage(repoName); err != nil {
 			return "", err
 		}
 		return fmt.Sprintf("Untagged: %s", repoName), nil
-	} else if len(image.Names()) > 1 && image.InputIsID() && !force {
+	} else if len(img.Names()) > 1 && img.InputIsID() && !force {
 		// If the user requests to delete an image by ID and the image has multiple
 		// reponames and no force is applied, we error out.
-		return "", fmt.Errorf("unable to delete %s (must force) - image is referred to in multiple tags", image.ID())
+		return "", fmt.Errorf("unable to delete %s (must force) - image is referred to in multiple tags", img.ID())
 	}
-	err = image.Remove(force)
+	err = img.Remove(force)
 	if err != nil && errors.Cause(err) == storage.ErrImageUsedByContainer {
-		if errStorage := r.rmStorageContainers(force, image); errStorage == nil {
+		if errStorage := r.rmStorageContainers(force, img); errStorage == nil {
 			// Containers associated with the image should be deleted now,
 			// let's try removing the image again.
-			err = image.Remove(force)
+			err = img.Remove(force)
 		} else {
 			err = errStorage
 		}
 	}
-	return image.ID(), err
+	return img.ID(), err
 }
 
 // Remove containers that are in storage rather than Podman.

test/e2e/rmi_test.go

@@ -103,4 +103,100 @@ var _ = Describe("Podman rmi", func() {
 		resultForce.WaitWithDefaultTimeout()
 		Expect(resultForce.ExitCode()).To(Equal(0))
 	})
+
+	It("podman rmi image that is a parent of another image", func() {
+		session := podmanTest.Podman([]string{"rmi", "-fa"})
+		session.WaitWithDefaultTimeout()
+		Expect(session.ExitCode()).To(Equal(0))
+
+		session = podmanTest.Podman([]string{"run", "--name", "c_test", ALPINE, "true"})
+		session.WaitWithDefaultTimeout()
+		Expect(session.ExitCode()).To(Equal(0))
+
+		session = podmanTest.Podman([]string{"commit", "-q", "c_test", "test"})
+		session.WaitWithDefaultTimeout()
+		Expect(session.ExitCode()).To(Equal(0))
+
+		session = podmanTest.Podman([]string{"rm", "c_test"})
+		session.WaitWithDefaultTimeout()
+		Expect(session.ExitCode()).To(Equal(0))
+
+		session = podmanTest.Podman([]string{"rmi", ALPINE})
+		session.WaitWithDefaultTimeout()
+		Expect(session.ExitCode()).To(Equal(0))
+
+		session = podmanTest.Podman([]string{"images", "-q"})
+		session.WaitWithDefaultTimeout()
+		Expect(session.ExitCode()).To(Equal(0))
+		Expect(len(session.OutputToStringArray())).To(Equal(1))
+
+		session = podmanTest.Podman([]string{"images", "-q", "-a"})
+		session.WaitWithDefaultTimeout()
+		Expect(session.ExitCode()).To(Equal(0))
+		Expect(len(session.OutputToStringArray())).To(Equal(2))
+		untaggedImg := session.OutputToStringArray()[1]
+
+		session = podmanTest.Podman([]string{"rmi", "-f", untaggedImg})
+		session.WaitWithDefaultTimeout()
+		Expect(session.ExitCode()).To(Not(Equal(0)))
+	})
+
+	It("podman rmi with cached images", func() {
+		session := podmanTest.Podman([]string{"rmi", "-fa"})
+		session.WaitWithDefaultTimeout()
+		Expect(session.ExitCode()).To(Equal(0))
+
+		dockerfile := `FROM docker.io/library/alpine:latest
+		RUN mkdir hello
+		RUN touch test.txt
+		ENV foo=bar
+		`
+		podmanTest.BuildImage(dockerfile, "test", "true")
+
+		dockerfile = `FROM docker.io/library/alpine:latest
+		RUN mkdir hello
+		RUN touch test.txt
+		RUN mkdir blah
+		ENV foo=bar
+		`
+		podmanTest.BuildImage(dockerfile, "test2", "true")
+
+		session = podmanTest.Podman([]string{"images", "-q", "-a"})
+		session.WaitWithDefaultTimeout()
+		Expect(session.ExitCode()).To(Equal(0))
+		numOfImages := len(session.OutputToStringArray())
+
+		session = podmanTest.Podman([]string{"rmi", "test2"})
+		session.WaitWithDefaultTimeout()
+		Expect(session.ExitCode()).To(Equal(0))
+
+		session = podmanTest.Podman([]string{"images", "-q", "-a"})
+		session.WaitWithDefaultTimeout()
+		Expect(session.ExitCode()).To(Equal(0))
+		Expect(numOfImages - len(session.OutputToStringArray())).To(Equal(2))
+
+		session = podmanTest.Podman([]string{"rmi", "test"})
+		session.WaitWithDefaultTimeout()
+		Expect(session.ExitCode()).To(Equal(0))
+
+		session = podmanTest.Podman([]string{"images", "-q", "-a"})
+		session.WaitWithDefaultTimeout()
+		Expect(session.ExitCode()).To(Equal(0))
+		Expect(len(session.OutputToStringArray())).To(Equal(1))
+
+		podmanTest.BuildImage(dockerfile, "test3", "true")
+
+		session = podmanTest.Podman([]string{"rmi", ALPINE})
+		session.WaitWithDefaultTimeout()
+		Expect(session.ExitCode()).To(Equal(0))
+
+		session = podmanTest.Podman([]string{"rmi", "test3"})
+		session.WaitWithDefaultTimeout()
+		Expect(session.ExitCode()).To(Equal(0))
+
+		session = podmanTest.Podman([]string{"images", "-q", "-a"})
+		session.WaitWithDefaultTimeout()
+		Expect(session.ExitCode()).To(Equal(0))
+		Expect(len(session.OutputToString())).To(Equal(0))
+	})
 })