Files
s3/weed/iam/integration/session_revocation.go
zooqueen dd74d3e6d6 Replace AGPL MinIO fork with Apache-2.0 SeaweedFS fork; rebrand CLI weed->s3
hanzoai/s3 was an AGPL-3.0 MinIO fork whose network-copyleft blocks embedding
and reselling as a managed service. Replace it wholesale with a fork of
SeaweedFS (github.com/seaweedfs/seaweedfs, Apache-2.0, Go) which ships a
built-in S3 gateway + Filer.

- License: AGPL-3.0 -> Apache-2.0 (upstream LICENSE kept verbatim, incl.
  "Copyright 2025 Chris Lu"); NOTICE added crediting SeaweedFS. No AGPL/BUSL/
  SSPL/GPL or enterprise ("ee") files remain.
- Module: go.mod path set to github.com/hanzoai/s3 with a self-replace
  (replace github.com/seaweedfs/seaweedfs => ./) so the ~1800 internal Go files
  keep their upstream import paths with zero churn. Internal identifiers and
  import paths are intentionally NOT renamed.
- CLI rebrand: the binary is built as `s3` (was `weed`); root usage, per-command
  help/usage/flag strings, shell autocomplete program name, and the product name
  in CLI output now read "Hanzo S3" / `s3`. Users never see `weed`.
  (e.g. `s3 server`, `s3 filer`, `s3 master`, `s3 volume`, `s3 s3`, `s3 mount`.)
- Build: `make` (or `go build -o s3 ./weed`) builds the `s3` binary green;
  `cmd/weed-db` and `cmd/weed-sql` also build; `go vet ./weed/command/` clean.
- README/LLM.md rewritten for Hanzo S3; AGENTS.md/CLAUDE.md symlinks preserved.
- Dropped a 33MB prebuilt Mach-O `stress` binary upstream had committed; `.gitignore`
  updated to ignore the `s3` build output.
2026-06-18 23:41:25 -07:00

258 lines
8.2 KiB
Go

package integration
import (
"context"
"crypto/sha1"
"encoding/hex"
"encoding/json"
"errors"
"fmt"
"io"
"strings"
"sync"
"time"
"github.com/seaweedfs/seaweedfs/weed/glog"
"github.com/seaweedfs/seaweedfs/weed/pb"
"github.com/seaweedfs/seaweedfs/weed/pb/filer_pb"
"google.golang.org/grpc"
)
// RevocationEntry is one row in the revocation list. ExpiresAt is set to the
// session's natural expiration so the store can garbage-collect entries that
// can no longer be reused.
type RevocationEntry struct {
JTI string `json:"jti"`
RevokedAt time.Time `json:"revokedAt"`
ExpiresAt time.Time `json:"expiresAt"`
Reason string `json:"reason,omitempty"`
}
// SessionRevocationStore is the per-deployment blocklist of revoked sessions.
// Implementations must be safe for concurrent use; the IsRevoked path is hot
// (checked on every signed request) and must not block on slow IO.
type SessionRevocationStore interface {
Revoke(ctx context.Context, filerAddress string, entry *RevocationEntry) error
IsRevoked(ctx context.Context, filerAddress string, jti string) (bool, error)
Purge(ctx context.Context, filerAddress string, before time.Time) (int, error)
}
// MemorySessionRevocationStore keeps the blocklist in process memory. Suitable
// for single-node deployments and tests; for HA, swap in the filer-backed
// implementation so revocations propagate across `weed` instances.
type MemorySessionRevocationStore struct {
mu sync.RWMutex
entries map[string]*RevocationEntry
}
// NewMemorySessionRevocationStore returns an empty in-memory blocklist.
func NewMemorySessionRevocationStore() *MemorySessionRevocationStore {
return &MemorySessionRevocationStore{entries: make(map[string]*RevocationEntry)}
}
func (m *MemorySessionRevocationStore) Revoke(ctx context.Context, _ string, entry *RevocationEntry) error {
if entry == nil || entry.JTI == "" {
return fmt.Errorf("entry with JTI is required")
}
m.mu.Lock()
defer m.mu.Unlock()
cp := *entry
if cp.RevokedAt.IsZero() {
cp.RevokedAt = time.Now().UTC()
}
m.entries[entry.JTI] = &cp
return nil
}
func (m *MemorySessionRevocationStore) IsRevoked(ctx context.Context, _ string, jti string) (bool, error) {
m.mu.RLock()
defer m.mu.RUnlock()
_, ok := m.entries[jti]
return ok, nil
}
func (m *MemorySessionRevocationStore) Purge(ctx context.Context, _ string, before time.Time) (int, error) {
m.mu.Lock()
defer m.mu.Unlock()
count := 0
for k, v := range m.entries {
if !v.ExpiresAt.IsZero() && v.ExpiresAt.Before(before) {
delete(m.entries, k)
count++
}
}
return count, nil
}
// FilerSessionRevocationStore persists entries under a filer directory.
// Files are named after the JTI so IsRevoked is a single LookupDirectoryEntry
// call — O(1) on a name-indexed filer. Purge enumerates the directory; that
// cost is operator-controlled (cron-driven) so the hot path stays cheap.
type FilerSessionRevocationStore struct {
grpcDialOption grpc.DialOption
basePath string
filerAddressProvider func() string
}
// NewFilerSessionRevocationStore returns a filer-backed blocklist store.
// Default basePath `/etc/iam/revoked-sessions` aligns with the other IAM
// directories and is safe to back up alongside roles + providers.
func NewFilerSessionRevocationStore(config map[string]interface{}, filerAddressProvider func() string) *FilerSessionRevocationStore {
store := &FilerSessionRevocationStore{
basePath: "/etc/iam/revoked-sessions",
filerAddressProvider: filerAddressProvider,
}
if config != nil {
if bp, ok := config["basePath"].(string); ok && bp != "" {
store.basePath = strings.TrimSuffix(bp, "/")
}
}
glog.V(2).Infof("Initialized FilerSessionRevocationStore with basePath %s", store.basePath)
return store
}
func (f *FilerSessionRevocationStore) resolveFilerAddress(filerAddress string) string {
if filerAddress != "" {
return filerAddress
}
if f.filerAddressProvider != nil {
return f.filerAddressProvider()
}
return ""
}
// fileName hashes the JTI before using it as a filename. RevokeSession is
// an exported API that accepts an arbitrary string; even though every
// SeaweedFS-issued session id is a safe random token, an external caller
// could pass "../../etc/passwd" or similar. Hashing produces a fixed-width
// hex name that's both filesystem-safe and stable, so the lookup-on-revoke
// path still finds the right entry.
func (f *FilerSessionRevocationStore) fileName(jti string) string {
sum := sha1.Sum([]byte(jti))
return hex.EncodeToString(sum[:]) + ".json"
}
func (f *FilerSessionRevocationStore) Revoke(ctx context.Context, filerAddress string, entry *RevocationEntry) error {
filerAddress = f.resolveFilerAddress(filerAddress)
if filerAddress == "" {
return fmt.Errorf("filer address is required")
}
if entry == nil || entry.JTI == "" {
return fmt.Errorf("entry with JTI is required")
}
if entry.RevokedAt.IsZero() {
entry.RevokedAt = time.Now().UTC()
}
data, err := json.MarshalIndent(entry, "", " ")
if err != nil {
return fmt.Errorf("marshal revocation entry: %v", err)
}
return f.withFilerClient(filerAddress, func(client filer_pb.SeaweedFilerClient) error {
_, err := client.CreateEntry(ctx, &filer_pb.CreateEntryRequest{
Directory: f.basePath,
Entry: &filer_pb.Entry{
Name: f.fileName(entry.JTI),
IsDirectory: false,
Attributes: &filer_pb.FuseAttributes{
Mtime: time.Now().Unix(),
Crtime: time.Now().Unix(),
FileMode: uint32(0o600),
},
Content: data,
},
})
if err != nil {
return fmt.Errorf("revoke %s: %v", entry.JTI, err)
}
return nil
})
}
func (f *FilerSessionRevocationStore) IsRevoked(ctx context.Context, filerAddress string, jti string) (bool, error) {
filerAddress = f.resolveFilerAddress(filerAddress)
if filerAddress == "" {
return false, fmt.Errorf("filer address is required")
}
revoked := false
err := f.withFilerClient(filerAddress, func(client filer_pb.SeaweedFilerClient) error {
resp, err := client.LookupDirectoryEntry(ctx, &filer_pb.LookupDirectoryEntryRequest{
Directory: f.basePath,
Name: f.fileName(jti),
})
if err != nil {
if strings.Contains(err.Error(), "not found") || strings.Contains(err.Error(), "no such") {
return nil
}
return err
}
revoked = resp.Entry != nil
return nil
})
return revoked, err
}
func (f *FilerSessionRevocationStore) Purge(ctx context.Context, filerAddress string, before time.Time) (int, error) {
filerAddress = f.resolveFilerAddress(filerAddress)
if filerAddress == "" {
return 0, fmt.Errorf("filer address is required")
}
count := 0
err := f.withFilerClient(filerAddress, func(client filer_pb.SeaweedFilerClient) error {
// Stream-paginate the directory: ListEntriesRequest has no built-in
// "page until done" semantics, so we use StartFromFileName to walk
// the directory in chunks and avoid the previous hardcoded 10k cap.
const pageSize = 1000
startFrom := ""
for {
stream, err := client.ListEntries(ctx, &filer_pb.ListEntriesRequest{
Directory: f.basePath,
Limit: pageSize,
StartFromFileName: startFrom,
InclusiveStartFrom: false,
})
if err != nil {
return fmt.Errorf("list revocation entries: %w", err)
}
lastName := ""
pageCount := 0
for {
resp, recvErr := stream.Recv()
if recvErr != nil {
if errors.Is(recvErr, io.EOF) {
break
}
return fmt.Errorf("recv revocation entry: %w", recvErr)
}
if resp.Entry == nil || resp.Entry.IsDirectory {
continue
}
lastName = resp.Entry.Name
pageCount++
var entry RevocationEntry
if err := json.Unmarshal(resp.Entry.Content, &entry); err != nil {
continue
}
if entry.ExpiresAt.IsZero() || entry.ExpiresAt.After(before) {
continue
}
if _, err := client.DeleteEntry(ctx, &filer_pb.DeleteEntryRequest{
Directory: f.basePath,
Name: resp.Entry.Name,
IsDeleteData: true,
}); err == nil {
count++
}
}
if pageCount < pageSize {
return nil
}
startFrom = lastName
}
})
return count, err
}
func (f *FilerSessionRevocationStore) withFilerClient(filerAddress string, fn func(filer_pb.SeaweedFilerClient) error) error {
return pb.WithGrpcFilerClient(false, 0, pb.ServerAddress(filerAddress), f.grpcDialOption, fn)
}