Files
seaweedfs/weed/s3api/s3_sse_s3.go
T
Chris Lu 47b491b53c mount: version open file handles by filer log position (#10403)
* filer: stamp a log position on lookup and remote-cache responses

Metadata events are logged after their store write and stamped with the
filer clock. Reading that clock before serving an entry therefore gives
a timestamp with a causal guarantee: every event at or below it is
reflected in the returned entry. Clients caching filer state can use it
as the entry's version to order the response against subscription
events, including events committed before the call but delivered after
it.

* mount: version open file handles by filer log position

A subscription event refreshing an open handle did a second lookup; a
transient failure left the handle pinned to its old entry with no
retry, since the subscription cursor had already advanced. The deeper
problem is ordering: the handle is a cache written by three unordered
channels — the async invalidation worker, local mutation acks, and
open-time lookups — and overwriting cached state safely requires
knowing which write is newer.

The filer log timestamp is that order, and it now travels with every
value instead of being derived out of band. Events carry it natively;
lookup and remote-cache responses carry the log position stamped before
the serving read; mutation acks carry it in their returned event; and
the local store pairs each read with a version cursor advanced under
the same lock as the store write. Each handle records the version its
entry reflects, and one rule replaces the per-site reasoning: state at
or below the handle's version is old news and must not be installed.

The invalidation itself applies the event's own entry — no lookup, so
no transient-failure window — except under a cached parent, where the
store entry is the ordered merge of the event and anything applied
since, and its version outranks the event's. An uncached parent
receives no store writes, so a hit there would be a stale leftover
masking the event. A vacated path (delete, rename away) keeps the last
entry so unlinked-but-open reads still work. Directory builds version
the completed directory at the listing snapshot and re-invalidate
buffered events at that version, since their mid-build refresh ran
against an incomplete store.

The tests replay every race this replaces machinery for: rollback of a
newer local flush (queued, cached, and read-through), stale leftovers
under uncached parents, the build window including abort, handles
opened after an event was queued, events landing mid-lookup, and
undelivered events at remote-cache time across a filer failover.

* filer: serialize the log position fence with mutations, stamp mutation acks

The fence stamped before an unlocked entry read could precede state the
read returned: a mutation writes storage first and assigns its event
timestamp only at notify time, so a lookup racing that window handed
the mount an entry newer than its fence, and the event's later delivery
looked like fresh news — destroying dirty pages for a change the handle
already had. The mutation handlers already hold an exclusive per-path
lock across read, write, and notify; the lookup and remote-cache reads
now take it shared around the stamp and the read, making the fence
exact: everything at or below it is in the entry, nothing above it is.

A no-change update returns success without an event, leaving the mount
nothing to fence with even though the response confirms current state.
Create and update acks now carry a log position stamped under the same
lock, and the mount falls back to it whenever the ack has no event.

Also regenerate the VT marshalers, which the earlier generation missed:
without them a VT round-trip silently zeroed every log position.

* java: sync filer.proto

* mount: scope store versions to what they vouch for; atomic handle install

The store's version cursor claimed too much. Advanced by local mutation
acks and directory listing snapshots, it inflated the version of store
reads for unrelated paths whose events the subscription still owed, and
those events were then fenced out permanently. The cursor now tracks
subscription progress only — events arrive in log order, so everything
at or below it has been delivered for every path — and a completed
listing records its snapshot as a per-directory floor instead of a
global claim. Local acks never touch it: they version their own handle
directly. Buffered build events advance the cursor at delivery, since
their store write may never happen (abort) while their invalidation is
already queued; their read-through directory pairs no store read with
it, and rename fragments are applied first.

Concurrent first opens raced: a slower opener's older lookup could
overwrite the newer entry a faster opener had installed, while the
monotonic version kept the newer timestamp — an old entry fenced at a
new version, immune to every correcting event. Entry and version are
now installed as one decision under the handle map lock, and an install
that does not outrank the handle's version is dropped.

The remote-cache commit also escaped the fence: it wrote storage and
notified without the path lock, so a lookup's shared-locked fence and
read could land between the two and hand out the cached state
under-versioned. The commit now re-reads and writes under the exclusive
path lock, and backs off entirely when the entry changed during the
download — the concurrent writer supersedes the cached content.

* mount: floors gate store applies; installs respect handle users; renames join the fence

A directory floor certifies the listing state as of its snapshot, but a
delayed event at or below the floor was still applied to the store —
rolling the content back to pre-snapshot state while the floor kept
claiming the snapshot version, so the correcting events were fenced out
of every future read. Events are now gated against the affected
directory's floor, each half of a rename independently.

Fences are lower bounds: a listing or lookup can include a mutation
whose event has not been delivered yet, and that event later passes
every gate carrying state the handle already holds. Such a re-delivery
now advances the version without destroying dirty pages or reinstalling
the entry — invalidating local writes over a no-op was the real damage
in every remaining under-fence window, including the unlocked listing
snapshot, which no per-path lock can serialize.

The concurrent-open install moved from the map lock to the handle lock
every reader, writer, and invalidation synchronizes on, and rejects
what cannot improve the handle: dirty state (local writes would be
lost), unversioned lookup responses (they cannot outrank anything, and
two zero-version opens must not overwrite each other), and anything not
strictly newer. New handles are still fully initialized before the map
exposes them.

Renames committed metadata and emitted events with no path lock, so a
lookup could read the renamed state under a fence preceding its events.
Both rename handlers now hold the source and destination locks, ordered
by path, across commit and notification; descendants of a renamed
directory are not individually locked and rely on the no-op re-delivery
handling above.

* mount: per-entry store versions replace the cursor and directory floors

The store's aggregate versions — a global subscription cursor and
per-directory listing floors — were versions at coarser granularity
than the values they described, and every over-claiming bug in this
series traced to that gap: an aggregate vouching for state its source
never saw. Each store entry now carries the filer log position of the
write that produced it — the event that applied it, or the listing
snapshot that inserted it, recorded in the store's key-value space
under the same lock as the entry write. The store becomes what the
handle already is: a last-writer-wins register with one rule, install
only what outranks the current claim.

The cursor, the floors, their advancement rules, the pairing ordering
constraint, and the floor gating all collapse into that rule. Applies
are gated per entry, each half of a rename independently; an
unversioned local write clears the claim its content no longer proves;
version records lingering after a bulk folder wipe cannot fence a
recreate, since a claim only blocks while its entry exists. Listing
inserts are stamped at build completion, before the buffered replay so
newer replayed events override the stamp.

Filer side, the fence dance every versioned read must perform is now a
single choke point, fencedFindEntry, so a future read RPC gets the
lock-serialized stamp by construction rather than by convention.

* mount: judge no-op re-deliveries against an immutable base, not the live entry

The equal-state skip compared the incoming event to the live handle
entry, but local writes mutate the live entry — size, timestamps,
chunks — so a delayed event re-delivering the base the handle was
opened with no longer matched, and the installer destroyed the dirty
pages and rolled the entry back over nothing new. The handle now keeps
an immutable snapshot of the filer state it last installed or
acknowledged, refreshed at every install and mutation ack (flush acks
snapshot the request entry before the id mapping mutates it), and the
no-op judgment runs against that base: an event carrying the base
brings nothing, whatever the live entry has diverged to since.

* mount: tombstones for versioned deletes, absence floors, copy enrollment

Four gaps in the per-entry version protocol, all the same shape: a
versioned fact with nothing carrying its version.

A deletion is a fact about a path with no entry left to hold it —
clearing the record let a delayed older event resurrect the deleted
path, permanently, since the deletion's own redelivery is
dedup-suppressed. Versioned deletes now leave a tombstone record that
fences without an entry; renames tombstone their source the same way.
Plain records still only block while their entry exists, so records
lingering after a bulk folder wipe cannot fence a recreate.

A completed listing proves absences as well as presences: a name it
omitted was deleted as of the snapshot, and a delayed create below the
snapshot re-creates it. The snapshot is kept per directory strictly as
an absence fence, consulted only when a path has neither an entry nor
a version record — present entries carry their own versions and never
touch it, which is what separates this from the over-claiming floor it
replaces.

A rebuild against a pre-upgrade filer returns no snapshot; stamping
now clears the children's records in that case, so a reinserted entry
cannot reactivate the stale claim its previous incarnation left
behind and reject valid events below it.

Server-side copies installed the copied entry without enrolling in the
base protocol, so the copy's own event differed from the stale
pre-copy base and destroyed writes made to the destination after the
copy. The install now refreshes the base and takes its version from
the fenced readback.

* mount: deletion facts outlive the cache's knowledge of the entry

A versioned delete of a path the store held no entry for recorded
nothing, so a delayed older event recreated the path — permanently,
with the deletion's redelivery dedup-suppressed. The tombstone is now
written whenever a versioned event vacates a path: the deletion is a
fact about the path, not about what this cache happened to hold.

For an absent entry, the listing's absence floor now speaks whatever
older record remains: a tombstone at one position does not exhaust
what is known about the path when a newer snapshot has confirmed the
name still absent, and an event between the two was slipping past
both.

A committed copy whose readback failed installed a synthesized base
with local timestamps; the copy's real event legitimately differs from
it, and was read as foreign state — destroying writes made to the
destination after the copy. The handle now marks that its own event is
en route and adopts that event's state as the base without touching
the live entry or the dirty pages; the adoption is one-shot, so a
genuinely foreign event still invalidates.

* mount: authoritative acks cancel pending event adoption; tombstones scoped and pruned

The copy-event adoption flag could outlive its purpose: a flush after
the failed readback installs a newer base and advances the version, the
copy's own event is then version gated without consuming the flag, and
the next genuinely foreign event was silently adopted — base advanced,
live entry and dirty pages untouched — leaving the mount to later
overwrite that remote change. Every local acknowledgment now installs
its base through one helper that also cancels any pending adoption: the
ack supersedes the mutation the adoption was waiting for.

Tombstones were written for every versioned delete under the mount and
survived directory eviction by design, growing LevelDB with historical
deletions on delete-heavy mounts. They are now scoped to directories
whose cached state the fence actually protects — an uncached parent
never serves from the store nor applies the resurrecting insert — and a
completed listing prunes the direct-child tombstones its absence floor
supersedes, leaving only those above the snapshot. The store gains a
key-prefix visitor for the sweep.

* mount: acked saves install their value; trailer snapshots; direct-child prune range

A version must never advance without its value. saveEntry stamped any
open handle with the acknowledgment's version, but a handle opened
while the save was in flight holds the pre-mutation entry — stamping it
fenced out the events carrying the state it lacked, permanently, with
the local apply performing no invalidation and the redelivery
deduplicated. The acknowledged entry is now installed together with its
version, through the same guarded install the racing-open path uses:
under the handle lock, only when it outranks the handle, never over
dirty local writes.

Empty listings return no in-band snapshot — a snapshot-only response
would be read as an entry by older consumers — so directories that end
empty gained no absence floor and their tombstones were never pruned.
The filer now sends the snapshot in the stream trailer, which older
clients ignore, and the client reads it when no in-band snapshot
arrived. Empty directories get real floors, their tombstones prune,
and their buffered replays gain the snapshot filter instead of the
replay-all fallback.

Version records now encode the parent directory and name separated by
a NUL, making a directory's direct children one contiguous key range:
the tombstone prune scans exactly them under the cache lock, instead
of walking every descendant record — the whole store, for root.

* mount: fix dirty-page loss, uid/gid base, download race, copy adopt, leak; dedup

Correctness fixes from the versioned-invalidation review:

- A foreign delete/rename-away of a file held open with unflushed local
  writes destroyed the dirty pages unconditionally. A process may keep
  writing to an unlinked-but-open file and those writes were already
  acknowledged; preserve the pages when the handle is dirty.
- downloadRemoteEntry stored the handle's base with filer-side uid/gid
  while every candidate it is later compared against is in local form,
  so under a non-identity UidGidMapper an unchanged re-delivery looked
  foreign and force-destroyed dirty pages. Map the base to local.
- downloadRemoteEntry wrote the entry/base/version triple under only the
  handle's shared lock, so two concurrent reads of the same remote-only
  file could tear it. Serialize the install with a dedicated mutex
  (invalidation is already excluded by the exclusive handle lock).
- A committed server-side copy whose readback failed adopted the FIRST
  event past the version gate as its base; a foreign write delivered
  first was silently swallowed. Adopt only an event whose content
  matches the synthesized base — the copy's own event — and install any
  other normally.
- The deferred-create path relied on AcquireFileHandle installing the
  passed entry on a pre-existing handle, which the version rework
  dropped. Restore that install in the compat wrapper; the versioned
  open path keeps its gated install.

Growth and hot-path cost:

- Per-entry version records and tombstones leaked when a directory was
  evicted or read-through without a rebuild. An uncached directory
  gates its own inserts, so its records fence nothing; clear a
  directory's child version records when it is wiped for eviction.
- FindEntry paid for the version KvGet on every lookup/getattr cache hit
  and threw it away. FindEntry now reads only the entry; the hot
  lookupEntry cache-hit path skips the version entirely.

Cleanups:

- Extract ackVersionTsNs over the shared response interface, replacing
  the metadata-event-else-log-ts snippet copy-pasted at four ack sites.
- Extract acquireRenamePathLocks, replacing the verbatim sorted
  two-path lock fence in both rename handlers.

* mount: no resurrection on foreign delete, version no-event acks, gate downloads, tighten copy adopt

Follow-ups to the review patches:

- Preserving dirty pages on a foreign delete let the next flush pass the
  isDeleted guard and CreateEntry, resurrecting the remotely-unlinked
  name. Mark the handle deleted in the vacate branch: the open fd can
  still read its buffered writes, but a flush no longer recreates the
  file.
- A no-event acknowledgment (log fence only) synthesized a metadata
  event with TsNs 0, so the cache stored the entry unversioned and an
  older subscriber event rolled it back. Stamp the synthesized event
  with the ack's log position at all four ack sites.
- downloadRemoteEntry serialized its install but did not check the
  version, so an older response arriving last overwrote the entry/base
  while the monotonic version kept the newer value, fencing corrections
  out. Install only when the response is at least as new as the handle.
- sameEntryContent compared only size and chunks, so a foreign chmod
  with unchanged content was adopted as the copy's own event. Compare
  everything except server-assigned timestamps, so a metadata-only
  foreign change installs instead.

* mount: trim comments to the non-obvious why

The versioning work accumulated multi-line comment blocks restating what
the code says. Keep the constraint a reader cannot derive — why a fence
is exact, why a version must not advance without its value, why an
uncached parent's records fence nothing — and drop the rest.

* mount: distinguish rename from delete, tighten the download and adopt gates

- A rename emits a nil old-path invalidation just like an unlink, so the
  vacate branch marked the handle deleted and later writes through the
  already-open descriptor were skipped instead of persisted. Carry the
  delete/rename distinction on the invalidation and mark only an actual
  delete.
- The remote-download install accepted an unversioned response
  regardless of the handle's version, so during a rolling upgrade a
  delayed response could install stale content under a newer version.
  Require the response to be at least as new, with one exception: a
  handle still lacking local chunks takes the content anyway — it cannot
  read without it — but does not claim the response's log position.
- Copy-event adoption returned without installing, so a foreign touch
  arriving before the copy's own event lost its timestamps. Content is
  unchanged either way, so the dirty pages stay valid; a clean handle now
  takes the entry, while a dirty one keeps its diverged version.

* mount: one directory floor instead of a record per child; agree on TTL

Review feedback:

- Build completion wrote one KV record per direct child inside the cache
  write lock, so a large directory stalled every other cache operation
  for O(children) store writes. The directory's listing snapshot already
  covers every child it saw; make that floor the version for any child
  without a record of its own, and a child earns a record only when a
  later event touches it. One map write per build replaces the per-child
  writes, with the same fencing.
- The presence probe read the store directly and so counted a
  TTL-expired entry as present, judging the path by a record describing
  content that has logically vanished. It now applies the same expiry
  the read path does, and an expired path falls back to its directory
  floor.
- Preserve ErrNotFound identity when the commit-time re-read finds the
  object deleted, so callers still surface a 404.
- Assert the rename-away source fence timestamp in the invalidation test.

Also record the tombstone ceiling: distinct deleted names in a cached
directory accumulate until it is rebuilt or evicted, which prunes
everything at or below the new snapshot.

* mount: pin the fence's clock domain instead of letting skew decide

A log-position fence is stamped by one filer's clock under that filer's
in-process lock, so comparing it to an event another filer logged is
comparing two unrelated clocks. The two error directions are not equally
costly: applying an event the fence already covered is a re-apply the
base-equality check absorbs, while skipping one it does not cover leaves
the handle holding exactly the state the event was meant to correct,
with the subscription cursor already past it — the unhealable staleness
this whole PR exists to remove.

So refuse to guess. Fences now carry the signature of the filer that
stamped them, and a handle records it alongside the position. An event
is only fenced out when the filer that logged it is the one that stamped
the fence — the logging filer appends its own signature, so its presence
identifies the clock domain. Events from any other filer are applied.
Positions taken from events keep comparing as before; the subscription
already delivers those in order.

The invalidation callback takes a struct now: it carries the path,
entry, position, delete/rename distinction, and signatures, and was
about to need a fifth positional parameter.

* mount: follow a foreign rename; key page invalidation on content, not equality

- A rename's old-path invalidation now carries the destination, and the
  handle follows the file there: an open fd tracks the inode, and leaving
  it on the old path made its next flush recreate that name instead of
  updating the renamed file.
- Dirty pages overlay content, so only a content change invalidates them.
  Keying that on exact equality meant any timestamp-only event destroyed
  them, which the copy-adoption marker existed to paper over — a foreign
  touch could consume the marker and leave the copy's own event to drop
  the post-copy writes. Comparing content instead makes the marker
  unnecessary, so it is gone: a metadata-only event keeps the overlay,
  and a dirty handle keeps its diverged entry unless foreign content
  supersedes it.
- A remote download response that is merely older is now refused even
  when the handle still lacks chunks; only an unversioned one is taken
  (and claims no position), since an older response's content predates
  what the handle reflects.
- A refused or unversioned download no longer publishes to the metadata
  cache, where a zero-position event would clear the entry's version and
  let an older subscriber event roll the cache back.

* mount: page invalidation keys on content alone; unversioned writes claim no position

- sameEntryContent compared everything but timestamps, so a foreign
  chmod, chown, or xattr change counted as a content change and
  destroyed the dirty-page overlay. It was strict only to serve the
  copy-adoption marker, which is gone; its one caller now asks the
  question it actually needs — did the bytes change — so metadata-only
  events leave the overlay alone.
- A rename over an existing file destroys that file, but its open handle
  was left live and still pointed at the name the renamed source now
  occupies, so its flush could overwrite it. MovePath already reports the
  displaced inode; mark that handle deleted.
- An acknowledgment was refused whenever its position was numerically
  lower, even when a different filer stamped the fence it lost to. Two
  known, differing signatures mean unrelated clocks, so the comparison no
  longer applies there; unknown signatures still compare as before.
- A local write with no log position behind it now records that
  explicitly instead of deleting its version record. Absence means the
  directory listing covers the path, which is why the snapshot floor
  applies; local content the listing never saw must not inherit it, or
  the events that would correct it are fenced out.

* mount: widen the existing lookup functions instead of forking WithVersion twins

The versioning work grew a parallel function for every accessor that
needed to return a log position — lookupEntryWithVersion beside
lookupEntry, maybeLoadEntryWithVersion beside maybeLoadEntry,
FindEntryWithVersion beside FindEntry, AcquireFileHandleWithVersion
beside AcquireFileHandle, advanceEntryVersion beside
advanceEntryVersionTsNs, plus a getPbEntryWithVersion wrapper and an
InsertListedEntriesForTest hook. Two names for one operation is two
places to keep in step, and the split let callers pick the one that
happened to compile.

Each pair is now the single original name carrying the position, with
callers that do not want it discarding it. filer_pb.GetEntry returns the
fence its response already carried rather than a mount-side wrapper
re-issuing the lookup, and InsertEntry takes the position its content
reflects rather than a test-only twin that inserted without one.

The one behavioural knot the merge exposed: AcquireFileHandle had been
installing the entry on a pre-existing handle only in its unversioned
form, which conflated 'the caller is authoritative' with 'the lookup had
no version'. Deferred create is the only caller that means the former,
so it now installs explicitly and the map function just acquires.
2026-07-23 17:44:02 -07:00

924 lines
33 KiB
Go

package s3api
import (
"bytes"
"context"
"crypto/aes"
"crypto/cipher"
"crypto/rand"
"crypto/sha256"
"encoding/base64"
"encoding/hex"
"encoding/json"
"errors"
"fmt"
"io"
mathrand "math/rand"
"net/http"
"os"
"strings"
"sync"
"time"
"github.com/seaweedfs/seaweedfs/weed/glog"
"github.com/seaweedfs/seaweedfs/weed/pb"
"github.com/seaweedfs/seaweedfs/weed/pb/filer_pb"
"github.com/seaweedfs/seaweedfs/weed/s3api/s3_constants"
"github.com/seaweedfs/seaweedfs/weed/util"
"github.com/seaweedfs/seaweedfs/weed/wdclient"
"golang.org/x/crypto/hkdf"
"google.golang.org/grpc"
)
// SSE-S3 uses AES-256 encryption with server-managed keys
const (
SSES3Algorithm = s3_constants.SSEAlgorithmAES256
SSES3KeySize = 32 // 256 bits
)
// SSES3Key represents a server-managed encryption key for SSE-S3
type SSES3Key struct {
Key []byte
KeyID string
Algorithm string
IV []byte // Initialization Vector for this key
KeyCommitment []byte // HMAC-SHA256 commitment binding key to IV+algorithm
}
// IsSSES3RequestInternal checks if the request specifies SSE-S3 encryption
func IsSSES3RequestInternal(r *http.Request) bool {
sseHeader := r.Header.Get(s3_constants.AmzServerSideEncryption)
result := sseHeader == SSES3Algorithm
// Debug: log header detection for SSE-S3 requests
if result {
glog.V(4).Infof("SSE-S3 detection: method=%s, header=%q, expected=%q, result=%t, copySource=%q", r.Method, sseHeader, SSES3Algorithm, result, r.Header.Get("X-Amz-Copy-Source"))
}
return result
}
// IsSSES3EncryptedInternal checks if the object metadata indicates SSE-S3 encryption
// An object is considered SSE-S3 encrypted only if it has BOTH the encryption header
// AND the actual encryption key metadata. This prevents false positives when an object
// has leftover headers from a previous encryption state (e.g., after being decrypted
// during a copy operation). Fixes GitHub issue #7562.
func IsSSES3EncryptedInternal(metadata map[string][]byte) bool {
// Check for SSE-S3 algorithm header
sseAlgorithm, hasHeader := metadata[s3_constants.AmzServerSideEncryption]
if !hasHeader || string(sseAlgorithm) != SSES3Algorithm {
return false
}
// Must also have the actual encryption key to be considered encrypted
// Without the key, the object cannot be decrypted and should be treated as unencrypted
_, hasKey := metadata[s3_constants.SeaweedFSSSES3Key]
return hasKey
}
// GenerateSSES3Key generates a new SSE-S3 encryption key
func GenerateSSES3Key() (*SSES3Key, error) {
key := make([]byte, SSES3KeySize)
if _, err := io.ReadFull(rand.Reader, key); err != nil {
return nil, fmt.Errorf("failed to generate SSE-S3 key: %w", err)
}
// Generate a key ID for tracking
keyID := fmt.Sprintf("sse-s3-key-%d", mathrand.Int63())
return &SSES3Key{
Key: key,
KeyID: keyID,
Algorithm: SSES3Algorithm,
}, nil
}
// CreateSSES3EncryptedReader creates an encrypted reader for SSE-S3
// Returns the encrypted reader and the IV for metadata storage
func CreateSSES3EncryptedReader(reader io.Reader, key *SSES3Key) (io.Reader, []byte, error) {
// Create AES cipher
block, err := aes.NewCipher(key.Key)
if err != nil {
return nil, nil, fmt.Errorf("create AES cipher: %w", err)
}
// Generate random IV
iv := make([]byte, aes.BlockSize)
if _, err := io.ReadFull(rand.Reader, iv); err != nil {
return nil, nil, fmt.Errorf("generate IV: %w", err)
}
// Create CTR mode cipher
stream := cipher.NewCTR(block, iv)
// Return encrypted reader and IV separately for metadata storage
encryptedReader := &cipher.StreamReader{S: stream, R: reader}
return encryptedReader, iv, nil
}
// CreateSSES3DecryptedReader creates a decrypted reader for SSE-S3 using IV from metadata
func CreateSSES3DecryptedReader(reader io.Reader, key *SSES3Key, iv []byte) (io.Reader, error) {
// IV comes from object metadata, which is mutable. Validate before passing
// to cipher.NewCTR so a tampered length produces an error rather than the
// crypto/cipher panic the documentation specifies.
if err := ValidateIV(iv, "SSE-S3 IV"); err != nil {
return nil, err
}
// Verify key commitment before decryption if one exists in metadata
if err := VerifyKeyCommitment(key.Key, iv, key.Algorithm, key.KeyCommitment); err != nil {
return nil, err
}
// Create AES cipher
block, err := aes.NewCipher(key.Key)
if err != nil {
return nil, fmt.Errorf("create AES cipher: %w", err)
}
// Create CTR mode cipher with the provided IV
stream := cipher.NewCTR(block, iv)
decryptReader := &cipher.StreamReader{S: stream, R: reader}
// Wrap with closer if the underlying reader implements io.Closer
if closer, ok := reader.(io.Closer); ok {
return &decryptReaderCloser{
Reader: decryptReader,
underlyingCloser: closer,
}, nil
}
return decryptReader, nil
}
// SerializeSSES3Metadata serializes SSE-S3 metadata for storage using envelope encryption
func SerializeSSES3Metadata(key *SSES3Key) ([]byte, error) {
if err := ValidateSSES3Key(key); err != nil {
return nil, err
}
// Encrypt the DEK using the global key manager's super key
keyManager := GetSSES3KeyManager()
encryptedDEK, nonce, err := keyManager.encryptKeyWithSuperKey(key.Key)
if err != nil {
return nil, fmt.Errorf("failed to encrypt DEK: %w", err)
}
metadata := map[string]string{
"algorithm": key.Algorithm,
"keyId": key.KeyID,
"encryptedDEK": base64.StdEncoding.EncodeToString(encryptedDEK),
"nonce": base64.StdEncoding.EncodeToString(nonce),
}
// Include IV if present (needed for chunk-level decryption)
if key.IV != nil {
metadata["iv"] = base64.StdEncoding.EncodeToString(key.IV)
// Compute and store key commitment binding key ↔ IV + algorithm
commitment := ComputeKeyCommitment(key.Key, key.IV, key.Algorithm)
metadata["keyCommitment"] = base64.StdEncoding.EncodeToString(commitment)
}
// Use JSON for proper serialization
data, err := json.Marshal(metadata)
if err != nil {
return nil, fmt.Errorf("marshal SSE-S3 metadata: %w", err)
}
return data, nil
}
// DeserializeSSES3Metadata deserializes SSE-S3 metadata from storage and decrypts the DEK
func DeserializeSSES3Metadata(data []byte, keyManager *SSES3KeyManager) (*SSES3Key, error) {
if len(data) == 0 {
return nil, fmt.Errorf("empty SSE-S3 metadata")
}
// Parse the JSON metadata
var metadata map[string]string
if err := json.Unmarshal(data, &metadata); err != nil {
return nil, fmt.Errorf("failed to parse SSE-S3 metadata: %w", err)
}
keyID, exists := metadata["keyId"]
if !exists {
return nil, fmt.Errorf("keyId not found in SSE-S3 metadata")
}
algorithm, exists := metadata["algorithm"]
if !exists {
algorithm = s3_constants.SSEAlgorithmAES256 // Default algorithm
}
// Decode the encrypted DEK and nonce
encryptedDEKStr, exists := metadata["encryptedDEK"]
if !exists {
return nil, fmt.Errorf("encryptedDEK not found in SSE-S3 metadata")
}
encryptedDEK, err := base64.StdEncoding.DecodeString(encryptedDEKStr)
if err != nil {
return nil, fmt.Errorf("failed to decode encrypted DEK: %w", err)
}
nonceStr, exists := metadata["nonce"]
if !exists {
return nil, fmt.Errorf("nonce not found in SSE-S3 metadata")
}
nonce, err := base64.StdEncoding.DecodeString(nonceStr)
if err != nil {
return nil, fmt.Errorf("failed to decode nonce: %w", err)
}
// Decrypt the DEK using the key manager
if keyManager == nil {
return nil, fmt.Errorf("key manager is required for SSE-S3 key retrieval")
}
dekBytes, err := keyManager.decryptKeyWithSuperKey(encryptedDEK, nonce)
if err != nil {
return nil, fmt.Errorf("failed to decrypt DEK: %w", err)
}
// Reconstruct the key
key := &SSES3Key{
Key: dekBytes,
KeyID: keyID,
Algorithm: algorithm,
}
// Restore IV if present in metadata (for chunk-level decryption)
if ivStr, exists := metadata["iv"]; exists {
iv, err := base64.StdEncoding.DecodeString(ivStr)
if err != nil {
return nil, fmt.Errorf("failed to decode IV: %w", err)
}
key.IV = iv
}
// Restore key commitment if present (for tamper detection)
if commitStr, exists := metadata["keyCommitment"]; exists {
commitment, err := base64.StdEncoding.DecodeString(commitStr)
if err != nil {
return nil, fmt.Errorf("failed to decode key commitment: %w", err)
}
key.KeyCommitment = commitment
}
return key, nil
}
// SSES3KeyManager manages SSE-S3 encryption keys using envelope encryption
// Instead of storing keys in memory, it uses a super key (KEK) to encrypt/decrypt DEKs
type SSES3KeyManager struct {
mu sync.RWMutex
superKey []byte // 256-bit master key (KEK - Key Encryption Key)
filerClient filer_pb.FilerClient // Filer client for KEK persistence
kekPath string // Path in filer where KEK is stored (e.g., /etc/s3/sse_kek)
kekPassphrase string // If set, KEK is encrypted at rest using a key derived from this passphrase
}
const (
// KEK storage layout on the filer. The migration code paths
// updateKEKContent / generateAndSaveSuperKeyToFiler rely on the directory
// + filename split; defaultKEKPath is the joined form kept for the
// existing reader code.
SSES3KEKDirectory = "/etc/s3"
SSES3KEKParentDir = "/etc"
SSES3KEKDirName = "s3"
SSES3KEKFileName = "sse_kek"
// Legacy KEK path on the filer (backward compatibility)
defaultKEKPath = SSES3KEKDirectory + "/" + SSES3KEKFileName
// security.toml keys (also settable via env vars WEED_S3_SSE_KEK / WEED_S3_SSE_KEY):
//
// s3.sse.kek: hex-encoded 256-bit key, same format as /etc/s3/sse_kek.
// Drop-in replacement for the filer-stored KEK. If /etc/s3/sse_kek also
// exists, the values must match or the server refuses to start.
//
// s3.sse.key: any secret string; a 256-bit key is derived via HKDF-SHA256.
// Cannot be used while /etc/s3/sse_kek exists — the filer file must be
// deleted first (to avoid silently orphaning old data).
sseS3KEKConfigKey = "s3.sse.kek"
sseS3KeyConfigKey = "s3.sse.key"
sseS3KEKPassphraseConfigKey = "s3.sse.kek.passphrase"
)
// legacyKEKWrappingSalt is the fixed salt the original implementation used
// for HKDF derivation. It is retained for backward compatibility — KEKs
// wrapped before per-installation salts shipped (the v1 format below) are
// still unwrappable. New writes always use a random salt.
var legacyKEKWrappingSalt = []byte("seaweedfs-sse-s3-kek-wrapping-v1")
// kekWrappedV2Magic identifies the new on-disk format that prefixes the
// wrapped KEK with a random salt. Seeing this magic at byte 0 of the
// decoded payload tells unwrapKEK to read the per-installation salt
// instead of falling back to legacyKEKWrappingSalt.
var kekWrappedV2Magic = []byte{0x53, 0x57, 0x76, 0x32} // "SWv2"
// kekRandomSaltSize is the per-installation salt length in bytes for HKDF.
// 32 bytes matches the SHA-256 output and is the standard recommendation.
const kekRandomSaltSize = 32
// NewSSES3KeyManager creates a new SSE-S3 key manager with envelope encryption.
// If kekPassphrase is non-empty, the KEK is encrypted at rest using a key derived from it.
func NewSSES3KeyManager(kekPassphrase ...string) *SSES3KeyManager {
km := &SSES3KeyManager{
kekPath: defaultKEKPath,
}
if len(kekPassphrase) > 0 {
km.kekPassphrase = kekPassphrase[0]
}
return km
}
// deriveWrappingKey derives a 256-bit AES key from the configured passphrase
// using HKDF-SHA256 with the supplied salt. Per-installation random salts
// land in the v2 format; the legacy fixed salt is still accepted for KEKs
// that were wrapped before random salts shipped.
func (km *SSES3KeyManager) deriveWrappingKey(salt []byte) ([]byte, error) {
if km.kekPassphrase == "" {
return nil, fmt.Errorf("no KEK passphrase configured")
}
hkdfReader := hkdf.New(sha256.New, []byte(km.kekPassphrase), salt, []byte("kek-wrapping"))
wrappingKey := make([]byte, SSES3KeySize)
if _, err := io.ReadFull(hkdfReader, wrappingKey); err != nil {
return nil, fmt.Errorf("HKDF derive wrapping key: %w", err)
}
return wrappingKey, nil
}
// wrapKEK encrypts the KEK using AES-GCM with a freshly-derived wrapping
// key. Output is base64(magic || salt || nonce || ciphertext+tag) — the
// random salt is the defence against rainbow-table precomputation against a
// shared passphrase, and storing it next to the ciphertext means the
// installation can rotate the passphrase without having to migrate the salt
// separately.
func (km *SSES3KeyManager) wrapKEK(kek []byte) ([]byte, error) {
salt := make([]byte, kekRandomSaltSize)
if _, err := io.ReadFull(rand.Reader, salt); err != nil {
return nil, fmt.Errorf("generate KEK salt: %w", err)
}
wrappingKey, err := km.deriveWrappingKey(salt)
if err != nil {
return nil, err
}
block, err := aes.NewCipher(wrappingKey)
if err != nil {
return nil, err
}
gcm, err := cipher.NewGCM(block)
if err != nil {
return nil, err
}
nonce := make([]byte, gcm.NonceSize())
if _, err := io.ReadFull(rand.Reader, nonce); err != nil {
return nil, err
}
header := make([]byte, 0, len(kekWrappedV2Magic)+len(salt))
header = append(header, kekWrappedV2Magic...)
header = append(header, salt...)
sealed := gcm.Seal(append(header, nonce...), nonce, kek, nil) // magic || salt || nonce || ciphertext+tag
return []byte(base64.StdEncoding.EncodeToString(sealed)), nil
}
// unwrapKEK decrypts a wrapped KEK produced by wrapKEK. Two on-disk formats
// are accepted:
//
// v2 (preferred): magic("SWv2") || salt || nonce || ciphertext+tag — the
// salt is read from the payload before HKDF runs.
// v1 (legacy): nonce || ciphertext+tag — falls back to the fixed
// legacyKEKWrappingSalt; rewrapping into v2 happens via the migration
// path in loadSuperKeyFromFiler.
//
// The returned `isV2` flag tells the caller which format was on disk, so
// the migration path can rewrap legacy entries without re-decoding the
// base64 payload a second time.
func (km *SSES3KeyManager) unwrapKEK(wrapped []byte) (kek []byte, isV2 bool, err error) {
raw, err := base64.StdEncoding.DecodeString(string(wrapped))
if err != nil {
return nil, false, fmt.Errorf("base64 decode wrapped KEK: %w", err)
}
salt := legacyKEKWrappingSalt
payload := raw
if len(raw) > len(kekWrappedV2Magic)+kekRandomSaltSize && bytes.Equal(raw[:len(kekWrappedV2Magic)], kekWrappedV2Magic) {
salt = raw[len(kekWrappedV2Magic) : len(kekWrappedV2Magic)+kekRandomSaltSize]
payload = raw[len(kekWrappedV2Magic)+kekRandomSaltSize:]
isV2 = true
}
wrappingKey, err := km.deriveWrappingKey(salt)
if err != nil {
return nil, false, err
}
block, err := aes.NewCipher(wrappingKey)
if err != nil {
return nil, false, err
}
gcm, err := cipher.NewGCM(block)
if err != nil {
return nil, false, err
}
if len(payload) < gcm.NonceSize() {
return nil, false, fmt.Errorf("wrapped KEK too short")
}
nonce := payload[:gcm.NonceSize()]
ciphertext := payload[gcm.NonceSize():]
out, err := gcm.Open(nil, nonce, ciphertext, nil)
if err != nil {
return nil, false, err
}
return out, isV2, nil
}
// deriveKeyFromSecret derives a 256-bit key from an arbitrary secret string
// using HKDF-SHA256. The derivation is deterministic: the same secret always
// produces the same key.
func deriveKeyFromSecret(secret string) ([]byte, error) {
hkdfReader := hkdf.New(sha256.New, []byte(secret), nil, []byte("seaweedfs-sse-s3-kek"))
key := make([]byte, SSES3KeySize)
if _, err := io.ReadFull(hkdfReader, key); err != nil {
return nil, fmt.Errorf("failed to derive key: %w", err)
}
return key, nil
}
// loadFilerKEK tries to load the KEK from /etc/s3/sse_kek on the filer.
// Returns the key bytes on success, nil if the file does not exist or filer
// is not configured, or an error on transient failures (retries internally).
func (km *SSES3KeyManager) loadFilerKEK() ([]byte, error) {
if km.filerClient == nil {
return nil, nil // no filer configured
}
var lastErr error
for i := 0; i < 10; i++ {
err := km.loadSuperKeyFromFiler()
if err == nil {
// loadSuperKeyFromFiler sets km.superKey; grab a copy
key := make([]byte, len(km.superKey))
copy(key, km.superKey)
km.superKey = nil // will be set by caller
return key, nil
}
if errors.Is(err, filer_pb.ErrNotFound) {
return nil, nil // file does not exist
}
lastErr = err
glog.Warningf("SSE-S3 KeyManager: failed to load KEK (attempt %d/10): %v", i+1, err)
time.Sleep(2 * time.Second)
}
return nil, fmt.Errorf("failed to load KEK from %s after 10 attempts: %w", km.kekPath, lastErr)
}
// InitializeWithFiler initializes the key manager with a filer client.
//
// Key source priority (via security.toml or WEED_ env vars):
// 1. s3.sse.kek (env: WEED_S3_SSE_KEK) — hex-encoded, same format as /etc/s3/sse_kek.
// If the filer file also exists, they must match.
// 2. s3.sse.key (env: WEED_S3_SSE_KEY) — any string; 256-bit key derived via HKDF.
// Refused if /etc/s3/sse_kek exists — delete the filer file first.
// 3. Existing /etc/s3/sse_kek on the filer (backward compat).
// 4. SSE-S3 disabled (fail on first encrypt/decrypt attempt).
func (km *SSES3KeyManager) InitializeWithFiler(filerClient filer_pb.FilerClient) error {
// Set filerClient under lock, then release — the rest may do slow I/O
// (filer retries with sleep) and must not block encrypt/decrypt callers.
km.mu.Lock()
km.filerClient = filerClient
km.mu.Unlock()
v := util.GetViper()
cfgKEK := v.GetString(sseS3KEKConfigKey) // hex-encoded, drop-in for filer file
cfgKey := v.GetString(sseS3KeyConfigKey) // any string, HKDF-derived
if cfgKEK != "" && cfgKey != "" {
return fmt.Errorf("only one of %s and %s may be set, not both", sseS3KEKConfigKey, sseS3KeyConfigKey)
}
var resolvedKey []byte
switch {
// --- Case 1: s3.sse.kek (hex, same format as filer file) ---
case cfgKEK != "":
key, err := hex.DecodeString(cfgKEK)
if err != nil {
return fmt.Errorf("invalid %s: must be hex-encoded: %w", sseS3KEKConfigKey, err)
}
if len(key) != SSES3KeySize {
return fmt.Errorf("invalid %s: must be %d bytes (%d hex chars), got %d bytes",
sseS3KEKConfigKey, SSES3KeySize, SSES3KeySize*2, len(key))
}
// Best-effort consistency check: if the filer file exists, warn on
// mismatch. A temporarily unreachable filer must not block startup
// when the operator has explicitly provided a KEK.
filerKey, err := km.loadFilerKEK()
if err != nil {
glog.Warningf("SSE-S3 KeyManager: could not reach filer to verify %s against %s: %v (proceeding with configured KEK)",
sseS3KEKConfigKey, km.kekPath, err)
} else if filerKey != nil && !bytes.Equal(filerKey, key) {
return fmt.Errorf("%s does not match existing %s — "+
"use the same key value as the filer file, or migrate existing data to the new key. "+
"See the Server-Side-Encryption wiki for migration steps",
sseS3KEKConfigKey, km.kekPath)
}
resolvedKey = key
glog.V(0).Infof("SSE-S3 KeyManager: Loaded KEK from %s config", sseS3KEKConfigKey)
// --- Case 2: s3.sse.key (any string, HKDF-derived) ---
case cfgKey != "":
// If the filer still has a legacy KEK file, the operator must migrate
// existing data first — using a derived key would silently orphan
// objects encrypted with the old KEK.
filerKey, err := km.loadFilerKEK()
if err != nil {
glog.Warningf("SSE-S3 KeyManager: could not reach filer to check for legacy %s: %v (proceeding with configured key)",
km.kekPath, err)
} else if filerKey != nil {
return fmt.Errorf("%s cannot be used while %s exists on the filer — "+
"existing objects are encrypted with the filer KEK. "+
"Migrate to %s first (copy the filer KEK value) or follow the key-rotation steps in the Server-Side-Encryption wiki",
sseS3KeyConfigKey, km.kekPath, sseS3KEKConfigKey)
}
derived, err := deriveKeyFromSecret(cfgKey)
if err != nil {
return err
}
resolvedKey = derived
glog.V(0).Infof("SSE-S3 KeyManager: Derived KEK from %s config", sseS3KeyConfigKey)
// --- Case 3: Load existing filer KEK (backward compatibility) ---
default:
filerKey, err := km.loadFilerKEK()
if err != nil {
return err
}
if filerKey != nil {
resolvedKey = filerKey
glog.V(1).Infof("SSE-S3 KeyManager: Loaded KEK from filer %s", km.kekPath)
glog.V(0).Infof("SSE-S3 KeyManager: Consider setting %s in security.toml instead of storing KEK on filer", sseS3KEKConfigKey)
} else {
// --- Case 4: Nothing configured — SSE-S3 disabled ---
glog.V(0).Infof("SSE-S3 KeyManager: No KEK configured. SSE-S3 encryption is disabled. "+
"Set %s or %s in security.toml to enable it.", sseS3KEKConfigKey, sseS3KeyConfigKey)
}
}
// Only hold the lock to write the final state.
km.mu.Lock()
km.superKey = resolvedKey
km.mu.Unlock()
return nil
}
// loadSuperKeyFromFiler loads the KEK from the filer
func (km *SSES3KeyManager) loadSuperKeyFromFiler() error {
if km.filerClient == nil {
return fmt.Errorf("filer client not initialized")
}
// Get the entry from filer
entry, _, _, err := filer_pb.GetEntry(context.Background(), km.filerClient, util.FullPath(km.kekPath))
if err != nil {
return fmt.Errorf("failed to get KEK entry from filer: %w", err)
}
// Read the content
if len(entry.Content) == 0 {
return fmt.Errorf("KEK entry is empty")
}
var key []byte
if km.kekPassphrase != "" {
// Try to unwrap encrypted KEK first
var wasV2 bool
key, wasV2, err = km.unwrapKEK(entry.Content)
if err == nil {
// Successful unwrap: if the payload was the legacy fixed-salt
// format, opportunistically rewrap it under a fresh per-installation
// salt so the next restart picks up the stronger format. The
// version flag comes straight out of unwrapKEK, avoiding a second
// base64 decode pass over the same content.
if !wasV2 {
if rewrapped, wrapErr := km.wrapKEK(key); wrapErr != nil {
glog.Warningf("SSE-S3 KeyManager: failed to rewrap legacy fixed-salt KEK to v2: %v", wrapErr)
} else if updErr := km.updateKEKContent(rewrapped); updErr != nil {
glog.Warningf("SSE-S3 KeyManager: failed to persist v2-rewrapped KEK: %v", updErr)
} else {
glog.V(1).Infof("SSE-S3 KeyManager: migrated KEK from fixed-salt v1 to per-installation salt v2")
}
}
} else {
// Fall back: maybe this is a legacy plaintext hex KEK — try to decode and re-wrap
legacyKey, hexErr := hex.DecodeString(string(entry.Content))
if hexErr != nil || len(legacyKey) != SSES3KeySize {
return fmt.Errorf("failed to unwrap KEK: %w", err)
}
glog.Warningf("SSE-S3 KeyManager: migrating plaintext KEK to encrypted storage")
key = legacyKey
// Re-save in encrypted form. Both failure modes used to be swallowed,
// which left the KEK on disk in plaintext while startup proceeded —
// an operator setting a passphrase saw a silent no-op and no signal
// that the migration had failed. Log loudly so the next restart
// makes the unmigrated state obvious; we still load the in-memory
// key so the server stays up.
wrapped, wrapErr := km.wrapKEK(key)
if wrapErr != nil {
glog.Errorf("SSE-S3 KeyManager: failed to wrap legacy KEK during migration; KEK remains plaintext on filer: %v", wrapErr)
} else if updErr := km.updateKEKContent(wrapped); updErr != nil {
glog.Errorf("SSE-S3 KeyManager: failed to persist wrapped KEK during migration; KEK remains plaintext on filer: %v", updErr)
}
}
} else {
// Legacy plaintext hex mode
glog.Warningf("SSE-S3 KeyManager: KEK stored in plaintext — set a KEK passphrase for encrypted storage")
key, err = hex.DecodeString(string(entry.Content))
if err != nil {
return fmt.Errorf("failed to decode KEK: %w", err)
}
}
if len(key) != SSES3KeySize {
return fmt.Errorf("invalid KEK size: expected %d bytes, got %d", SSES3KeySize, len(key))
}
km.superKey = key
return nil
}
// updateKEKContent overwrites the existing KEK file content in the filer.
// Used by the plaintext→encrypted migration path and by the v1→v2 salt
// rewrap; both run after a successful read of the current KEK, so the
// entry is guaranteed to exist. MkFile uses CreateEntry which fails with
// ErrEntryAlreadyExists when the file is already there — we need
// UpdateEntry instead so the migration actually persists.
//
// Splits km.kekPath at the last "/" so an operator-overridden path is
// honoured. Defaults match defaultKEKPath when km.kekPath is unset.
func (km *SSES3KeyManager) updateKEKContent(content []byte) error {
dir, name := splitKEKPath(km.kekPath)
ctx := context.Background()
return km.filerClient.WithFilerClient(false, func(client filer_pb.SeaweedFilerClient) error {
resp, err := client.LookupDirectoryEntry(ctx, &filer_pb.LookupDirectoryEntryRequest{
Directory: dir,
Name: name,
})
if err != nil {
return fmt.Errorf("lookup KEK entry: %w", err)
}
entry := resp.Entry
if entry == nil {
return fmt.Errorf("KEK entry not found at %s/%s", dir, name)
}
entry.Content = content
if entry.Attributes == nil {
entry.Attributes = &filer_pb.FuseAttributes{}
}
entry.Attributes.FileMode = 0600
entry.Attributes.FileSize = uint64(len(content))
entry.Attributes.Mtime = time.Now().Unix()
return filer_pb.UpdateEntry(ctx, client, &filer_pb.UpdateEntryRequest{
Directory: dir,
Entry: entry,
})
})
}
// splitKEKPath splits an absolute KEK file path into (directory, name).
// Falls back to the default location if the path is empty or has no slash.
func splitKEKPath(p string) (dir, name string) {
if p == "" {
return SSES3KEKDirectory, SSES3KEKFileName
}
idx := strings.LastIndex(p, "/")
if idx <= 0 {
return SSES3KEKDirectory, SSES3KEKFileName
}
return p[:idx], p[idx+1:]
}
// GetOrCreateKey gets an existing key or creates a new one
// With envelope encryption, we always generate a new DEK since we don't store them
func (km *SSES3KeyManager) GetOrCreateKey(keyID string) (*SSES3Key, error) {
// Always generate a new key - we use envelope encryption so no need to cache DEKs
return GenerateSSES3Key()
}
// encryptKeyWithSuperKey encrypts a DEK using the super key (KEK) with AES-GCM
func (km *SSES3KeyManager) encryptKeyWithSuperKey(dek []byte) ([]byte, []byte, error) {
km.mu.RLock()
defer km.mu.RUnlock()
if len(km.superKey) == 0 {
return nil, nil, fmt.Errorf("SSE-S3 encryption is not configured — set %s or %s in security.toml", sseS3KEKConfigKey, sseS3KeyConfigKey)
}
block, err := aes.NewCipher(km.superKey)
if err != nil {
return nil, nil, fmt.Errorf("failed to create cipher: %w", err)
}
gcm, err := cipher.NewGCM(block)
if err != nil {
return nil, nil, fmt.Errorf("failed to create GCM: %w", err)
}
// Generate random nonce
nonce := make([]byte, gcm.NonceSize())
if _, err := io.ReadFull(rand.Reader, nonce); err != nil {
return nil, nil, fmt.Errorf("failed to generate nonce: %w", err)
}
// Encrypt the DEK
encryptedDEK := gcm.Seal(nil, nonce, dek, nil)
return encryptedDEK, nonce, nil
}
// decryptKeyWithSuperKey decrypts a DEK using the super key (KEK) with AES-GCM
func (km *SSES3KeyManager) decryptKeyWithSuperKey(encryptedDEK, nonce []byte) ([]byte, error) {
km.mu.RLock()
defer km.mu.RUnlock()
if len(km.superKey) == 0 {
return nil, fmt.Errorf("SSE-S3 decryption is not configured — set %s or %s in security.toml", sseS3KEKConfigKey, sseS3KeyConfigKey)
}
block, err := aes.NewCipher(km.superKey)
if err != nil {
return nil, fmt.Errorf("failed to create cipher: %w", err)
}
gcm, err := cipher.NewGCM(block)
if err != nil {
return nil, fmt.Errorf("failed to create GCM: %w", err)
}
if len(nonce) != gcm.NonceSize() {
return nil, fmt.Errorf("invalid nonce size: expected %d, got %d", gcm.NonceSize(), len(nonce))
}
// Decrypt the DEK
dek, err := gcm.Open(nil, nonce, encryptedDEK, nil)
if err != nil {
return nil, fmt.Errorf("failed to decrypt DEK: %w", err)
}
return dek, nil
}
// StoreKey is now a no-op since we use envelope encryption and don't cache DEKs
// The encrypted DEK is stored in the object metadata, not in the key manager
func (km *SSES3KeyManager) StoreKey(key *SSES3Key) {
// No-op: With envelope encryption, we don't need to store keys in memory
// The DEK is encrypted with the super key and stored in object metadata
}
// GetMasterKey returns a derived key from the master KEK for STS signing
// This uses HKDF to isolate the STS security domain from the SSE-S3 domain
func (km *SSES3KeyManager) GetMasterKey() []byte {
km.mu.RLock()
defer km.mu.RUnlock()
if len(km.superKey) == 0 {
return nil
}
// Derive a separate key for STS to isolate security domains
// We use the KEK as the secret, and "seaweedfs-sts-signing-key" as the info
hkdfReader := hkdf.New(sha256.New, km.superKey, nil, []byte("seaweedfs-sts-signing-key"))
derived := make([]byte, 32) // 256-bit derived key
if _, err := io.ReadFull(hkdfReader, derived); err != nil {
glog.Errorf("Failed to derive STS key: %v", err)
return nil
}
return derived
}
// SSES3KEKPassphraseEnv is the legacy environment variable from which the
// global SSE-S3 key manager picks up its KEK-wrapping passphrase. The Viper
// config key sseS3KEKPassphraseConfigKey ("s3.sse.kek.passphrase") is the
// preferred way to set it — same precedence as s3.sse.kek and s3.sse.key —
// but the env var is honoured as a fallback so deployments that wired only
// the env keep working.
const SSES3KEKPassphraseEnv = "WEED_S3_SSE_KEK_PASSPHRASE"
// Global SSE-S3 key manager instance
var globalSSES3KeyManager = NewSSES3KeyManager()
// SetKEKPassphrase configures the KEK-wrapping passphrase. Must be called
// before InitializeWithFiler — the load path reads the passphrase to decide
// whether to attempt unwrap or fall back to plaintext-hex parsing.
func (km *SSES3KeyManager) SetKEKPassphrase(passphrase string) {
km.mu.Lock()
defer km.mu.Unlock()
km.kekPassphrase = passphrase
}
// GetSSES3KeyManager returns the global SSE-S3 key manager
func GetSSES3KeyManager() *SSES3KeyManager {
return globalSSES3KeyManager
}
// KeyManagerFilerClient wraps wdclient.FilerClient to satisfy filer_pb.FilerClient interface
type KeyManagerFilerClient struct {
*wdclient.FilerClient
grpcDialOption grpc.DialOption
}
func (k *KeyManagerFilerClient) AdjustedUrl(location *filer_pb.Location) string {
return location.Url
}
func (k *KeyManagerFilerClient) WithFilerClient(streamingMode bool, fn func(filer_pb.SeaweedFilerClient) error) error {
filerAddress := k.GetCurrentFiler()
if filerAddress == "" {
return fmt.Errorf("no filer available")
}
return pb.WithGrpcFilerClient(streamingMode, 0, filerAddress, k.grpcDialOption, fn)
}
// InitializeGlobalSSES3KeyManager initializes the global key manager with
// filer access. The KEK-wrapping passphrase is sourced from the Viper
// config key s3.sse.kek.passphrase (matching the s3.sse.kek and
// s3.sse.key conventions, settable via security.toml or
// WEED_S3_SSE_KEK_PASSPHRASE env), with a fallback to the bare
// SSES3KEKPassphraseEnv lookup for deployments wired before the Viper key
// existed. If neither is set the KEK falls back to plaintext at-rest
// storage (with a startup warning).
func InitializeGlobalSSES3KeyManager(filerClient *wdclient.FilerClient, grpcDialOption grpc.DialOption) error {
passphrase := util.GetViper().GetString(sseS3KEKPassphraseConfigKey)
if passphrase == "" {
passphrase = os.Getenv(SSES3KEKPassphraseEnv)
}
if passphrase != "" {
globalSSES3KeyManager.SetKEKPassphrase(passphrase)
} else {
glog.Warningf("SSE-S3 KeyManager: neither %s nor %s is set; the KEK will be stored on the filer in plaintext. Set one to enable encrypted-at-rest KEK storage.", sseS3KEKPassphraseConfigKey, SSES3KEKPassphraseEnv)
}
wrapper := &KeyManagerFilerClient{
FilerClient: filerClient,
grpcDialOption: grpcDialOption,
}
return globalSSES3KeyManager.InitializeWithFiler(wrapper)
}
// GetSSES3IV extracts the IV for single-part SSE-S3 objects
// Priority: 1) object-level metadata (for inline/small files), 2) first chunk metadata
func GetSSES3IV(entry *filer_pb.Entry, sseS3Key *SSES3Key, keyManager *SSES3KeyManager) ([]byte, error) {
// First check if IV is in the object-level key (for small/inline files)
if len(sseS3Key.IV) > 0 {
return sseS3Key.IV, nil
}
// Fallback: Get IV from first chunk's metadata (for chunked files)
if len(entry.GetChunks()) > 0 {
chunk := entry.GetChunks()[0]
if len(chunk.GetSseMetadata()) > 0 {
chunkKey, err := DeserializeSSES3Metadata(chunk.GetSseMetadata(), keyManager)
if err != nil {
return nil, fmt.Errorf("failed to deserialize chunk SSE-S3 metadata: %w", err)
}
if len(chunkKey.IV) > 0 {
return chunkKey.IV, nil
}
}
}
return nil, fmt.Errorf("SSE-S3 IV not found in object or chunk metadata")
}
// CreateSSES3EncryptedReaderWithBaseIV creates an encrypted reader using a base IV for multipart upload consistency.
// The returned IV is the offset-derived IV, calculated from the input baseIV and offset.
func CreateSSES3EncryptedReaderWithBaseIV(reader io.Reader, key *SSES3Key, baseIV []byte, offset int64) (io.Reader, []byte /* derivedIV */, error) {
// Validate key to prevent panics and security issues
if key == nil {
return nil, nil, fmt.Errorf("SSES3Key is nil")
}
if key.Key == nil || len(key.Key) != SSES3KeySize {
return nil, nil, fmt.Errorf("invalid SSES3Key: must be %d bytes, got %d", SSES3KeySize, len(key.Key))
}
if err := ValidateSSES3Key(key); err != nil {
return nil, nil, err
}
block, err := aes.NewCipher(key.Key)
if err != nil {
return nil, nil, fmt.Errorf("create AES cipher: %w", err)
}
// Calculate the proper IV with offset to ensure unique IV per chunk/part
// This prevents the severe security vulnerability of IV reuse in CTR mode
// Skip is not used here because we're encrypting from the start (not reading a range)
iv, _ := calculateIVWithOffset(baseIV, offset)
stream := cipher.NewCTR(block, iv)
encryptedReader := &cipher.StreamReader{S: stream, R: reader}
return encryptedReader, iv, nil
}