Fix various lazy blob errors involving dedupe by chainID

cache/manager.go

@@ -143,7 +143,7 @@ func (cm *cacheManager) GetByBlob(ctx context.Context, desc ocispecs.Descriptor,
 	descHandlers := descHandlersOf(opts...)
 	if desc.Digest != "" && (descHandlers == nil || descHandlers[desc.Digest] == nil) {
 		if _, err := cm.ContentStore.Info(ctx, desc.Digest); errors.Is(err, cerrdefs.ErrNotFound) {
-			return nil, NeedsRemoteProviderError([]digest.Digest{desc.Digest})
+			return nil, NeedsRemoteProviderError([]DigestDescriptionPair{{Digest: desc.Digest}})
 		} else if err != nil {
 			return nil, err
 		}
@@ -396,7 +396,7 @@ func (cm *cacheManager) getRecord(ctx context.Context, id string, opts ...RefOpt
 			if isLazy, err := cr.isLazy(ctx); err != nil {
 				return err
 			} else if isLazy && dhs[blob] == nil {
-				missing = append(missing, blob)
+				missing = append(missing, DigestDescriptionPair{Digest: blob, Description: cr.GetDescription()})
 			}
 			return nil
 		}); err != nil {

cache/opts.go

@@ -30,10 +30,19 @@ func descHandlersOf(opts ...RefOption) DescHandlers {
 
 type DescHandlerKey digest.Digest
 
-type NeedsRemoteProviderError []digest.Digest //nolint:errname
+type NeedsRemoteProviderError []DigestDescriptionPair //nolint:errname
+
+type DigestDescriptionPair struct {
+	Digest      digest.Digest
+	Description string
+}
+
+func (d DigestDescriptionPair) String() string {
+	return fmt.Sprintf("%s: %s", d.Digest, d.Description)
+}
 
 func (m NeedsRemoteProviderError) Error() string {
-	return fmt.Sprintf("missing descriptor handlers for lazy blobs %+v", []digest.Digest(m))
+	return fmt.Sprintf("missing descriptor handlers for lazy blobs %+v", []DigestDescriptionPair(m))
 }
 
 type Unlazy session.Group

cache/refs.go

@@ -49,6 +49,7 @@ type Ref interface {
 	RefMetadata
 	Release(context.Context) error
 	IdentityMapping() *idtools.IdentityMapping
+	DescHandlers() DescHandlers
 	DescHandler(digest.Digest) *DescHandler
 }
 
@@ -612,6 +613,13 @@ func (sr *immutableRef) LayerChain() RefList {
 	return l
 }
 
+func (sr *immutableRef) DescHandlers() DescHandlers {
+	// clone to prevent mutation of internal state
+	dhs := make(DescHandlers)
+	maps.Copy(dhs, sr.descHandlers)
+	return dhs
+}
+
 func (sr *immutableRef) DescHandler(dgst digest.Digest) *DescHandler {
 	return sr.descHandlers[dgst]
 }
@@ -640,6 +648,13 @@ func (sr *mutableRef) traceLogFields() logrus.Fields {
 	return m
 }
 
+func (sr *mutableRef) DescHandlers() DescHandlers {
+	// clone to prevent mutation of internal state
+	dhs := make(DescHandlers)
+	maps.Copy(dhs, sr.descHandlers)
+	return dhs
+}
+
 func (sr *mutableRef) DescHandler(dgst digest.Digest) *DescHandler {
 	return sr.descHandlers[dgst]
 }

cache/remotecache/azblob/importer.go

@@ -134,7 +134,7 @@ func (ci *importer) loadManifest(ctx context.Context, name string) (*v1.CacheCha
 	progress.OneOff(ctx, fmt.Sprintf("found %d layers in cache", len(allLayers)))(nil)
 
 	cc := v1.NewCacheChains()
-	if err := v1.ParseConfig(config, allLayers, cc); err != nil {
+	if err := v1.ParseConfig(ctx, config, allLayers, cc); err != nil {
 		return nil, err
 	}
 

cache/remotecache/gha/gha.go

@@ -356,7 +356,7 @@ func (ci *importer) loadScope(ctx context.Context, scope string) (*v1.CacheChain
 	}
 
 	cc := v1.NewCacheChains()
-	if err := v1.ParseConfig(config, allLayers, cc); err != nil {
+	if err := v1.ParseConfig(ctx, config, allLayers, cc); err != nil {
 		return nil, err
 	}
 	return cc, nil

cache/remotecache/import.go

@@ -117,7 +117,7 @@ func (ci *contentCacheImporter) Resolve(ctx context.Context, desc ocispecs.Descr
 	}
 
 	cc := v1.NewCacheChains()
-	if err := v1.Parse(dt, allLayers, cc); err != nil {
+	if err := v1.Parse(ctx, dt, allLayers, cc); err != nil {
 		return nil, err
 	}
 
@@ -238,7 +238,7 @@ func (ci *contentCacheImporter) importInlineCache(ctx context.Context, dt []byte
 					return errors.WithStack(err)
 				}
 				cc := v1.NewCacheChains()
-				if err := v1.ParseConfig(config, layers, cc); err != nil {
+				if err := v1.ParseConfig(ctx, config, layers, cc); err != nil {
 					return err
 				}
 				mu.Lock()

cache/remotecache/inline/inline.go

@@ -76,7 +76,7 @@ func (ce *exporter) ExportForLayers(ctx context.Context, layers []digest.Digest)
 	}
 
 	cc := v1.NewCacheChains()
-	if err := v1.ParseConfig(*config, descs2, cc); err != nil {
+	if err := v1.ParseConfig(ctx, *config, descs2, cc); err != nil {
 		return nil, err
 	}
 

cache/remotecache/s3/s3.go

@@ -365,7 +365,7 @@ func (i *importer) load(ctx context.Context) (*v1.CacheChains, error) {
 	}
 
 	cc := v1.NewCacheChains()
-	if err := v1.ParseConfig(config, allLayers, cc); err != nil {
+	if err := v1.ParseConfig(ctx, config, allLayers, cc); err != nil {
 		return nil, err
 	}
 	return cc, nil

cache/remotecache/v1/chains.go

@@ -7,8 +7,10 @@ import (
 	"time"
 
 	"github.com/containerd/containerd/content"
+	cerrdefs "github.com/containerd/errdefs"
 	"github.com/moby/buildkit/session"
 	"github.com/moby/buildkit/solver"
+	"github.com/moby/buildkit/util/bklog"
 	digest "github.com/opencontainers/go-digest"
 	ocispecs "github.com/opencontainers/image-spec/specs-go/v1"
 	"github.com/pkg/errors"
@@ -215,7 +217,26 @@ func (c *item) removeLink(src *item) bool {
 	return found
 }
 
-func (c *item) AddResult(_ digest.Digest, _ int, createdAt time.Time, result *solver.Remote) {
+func (c *item) AddResult(ctx context.Context, _ digest.Digest, _ int, createdAt time.Time, result *solver.Remote) {
+	if result == nil {
+		return
+	}
+
+	if len(result.Descriptors) == 0 {
+		bklog.G(ctx).Warnf("no descriptors for item %s result, skipping", c.dgst)
+		return
+	}
+
+	if result.Provider != nil {
+		for _, desc := range result.Descriptors {
+			_, err := result.Provider.Info(ctx, desc.Digest)
+			if err != nil && !cerrdefs.IsNotFound(err) {
+				bklog.G(ctx).Warnf("failed to get info for item %s descriptor %s, skipping item result: %v", c.dgst, desc.Digest, err)
+				return
+			}
+		}
+	}
+
 	c.resultTime = createdAt
 	c.result = result
 }
@@ -305,7 +326,7 @@ func (c *item) walkAllResults(fn func(i *item) error, visited map[*item]struct{}
 type nopRecord struct {
 }
 
-func (c *nopRecord) AddResult(_ digest.Digest, _ int, createdAt time.Time, result *solver.Remote) {
+func (c *nopRecord) AddResult(_ context.Context, _ digest.Digest, _ int, createdAt time.Time, result *solver.Remote) {
 }
 
 func (c *nopRecord) LinkFrom(rec solver.CacheExporterRecord, index int, selector string) {

cache/remotecache/v1/chains_test.go

@@ -29,7 +29,7 @@ func TestSimpleMarshal(t *testing.T) {
 				Digest: dgst("d1"),
 			}},
 		}
-		baz.AddResult("", 0, time.Now(), r0)
+		baz.AddResult(context.TODO(), "", 0, time.Now(), r0)
 	}
 
 	addRecords()
@@ -76,7 +76,7 @@ func TestSimpleMarshal(t *testing.T) {
 	require.NoError(t, err)
 
 	newChains := NewCacheChains()
-	err = Parse(dt, descPairs, newChains)
+	err = Parse(context.TODO(), dt, descPairs, newChains)
 	require.NoError(t, err)
 
 	cfg3, _, err := cc.Marshal(context.TODO())

cache/remotecache/v1/parse.go

@@ -1,6 +1,7 @@
 package cacheimport
 
 import (
+	"context"
 	"encoding/json"
 
 	"github.com/moby/buildkit/solver"
@@ -9,27 +10,27 @@ import (
 	"github.com/pkg/errors"
 )
 
-func Parse(configJSON []byte, provider DescriptorProvider, t solver.CacheExporterTarget) error {
+func Parse(ctx context.Context, configJSON []byte, provider DescriptorProvider, t solver.CacheExporterTarget) error {
 	var config CacheConfig
 	if err := json.Unmarshal(configJSON, &config); err != nil {
 		return errors.WithStack(err)
 	}
 
-	return ParseConfig(config, provider, t)
+	return ParseConfig(ctx, config, provider, t)
 }
 
-func ParseConfig(config CacheConfig, provider DescriptorProvider, t solver.CacheExporterTarget) error {
+func ParseConfig(ctx context.Context, config CacheConfig, provider DescriptorProvider, t solver.CacheExporterTarget) error {
 	cache := map[int]solver.CacheExporterRecord{}
 
 	for i := range config.Records {
-		if _, err := parseRecord(config, i, provider, t, cache); err != nil {
+		if _, err := parseRecord(ctx, config, i, provider, t, cache); err != nil {
 			return err
 		}
 	}
 	return nil
 }
 
-func parseRecord(cc CacheConfig, idx int, provider DescriptorProvider, t solver.CacheExporterTarget, cache map[int]solver.CacheExporterRecord) (solver.CacheExporterRecord, error) {
+func parseRecord(ctx context.Context, cc CacheConfig, idx int, provider DescriptorProvider, t solver.CacheExporterTarget, cache map[int]solver.CacheExporterRecord) (solver.CacheExporterRecord, error) {
 	if r, ok := cache[idx]; ok {
 		if r == nil {
 			return nil, errors.Errorf("invalid looping record")
@@ -46,7 +47,7 @@ func parseRecord(cc CacheConfig, idx int, provider DescriptorProvider, t solver.
 	cache[idx] = nil
 	for i, inputs := range rec.Inputs {
 		for _, inp := range inputs {
-			src, err := parseRecord(cc, inp.LinkIndex, provider, t, cache)
+			src, err := parseRecord(ctx, cc, inp.LinkIndex, provider, t, cache)
 			if err != nil {
 				return nil, err
 			}
@@ -61,7 +62,7 @@ func parseRecord(cc CacheConfig, idx int, provider DescriptorProvider, t solver.
 			return nil, err
 		}
 		if remote != nil {
-			r.AddResult("", 0, res.CreatedAt, remote)
+			r.AddResult(ctx, "", 0, res.CreatedAt, remote)
 		}
 	}
 
@@ -86,7 +87,7 @@ func parseRecord(cc CacheConfig, idx int, provider DescriptorProvider, t solver.
 		}
 		if remote != nil {
 			remote.Provider = mp
-			r.AddResult("", 0, res.CreatedAt, remote)
+			r.AddResult(ctx, "", 0, res.CreatedAt, remote)
 		}
 	}
 

cache/remotecache/v1/utils.go

@@ -5,7 +5,6 @@ import (
 	"fmt"
 	"sort"
 
-	cerrdefs "github.com/containerd/errdefs"
 	"github.com/moby/buildkit/solver"
 	"github.com/moby/buildkit/util/bklog"
 	digest "github.com/opencontainers/go-digest"
@@ -275,20 +274,6 @@ type marshalState struct {
 }
 
 func marshalRemote(ctx context.Context, r *solver.Remote, state *marshalState) string {
-	if len(r.Descriptors) == 0 {
-		return ""
-	}
-
-	if r.Provider != nil {
-		for _, d := range r.Descriptors {
-			if _, err := r.Provider.Info(ctx, d.Digest); err != nil {
-				if !cerrdefs.IsNotImplemented(err) {
-					return ""
-				}
-			}
-		}
-	}
-
 	var parentID string
 	if len(r.Descriptors) > 1 {
 		r2 := &solver.Remote{