Warp/server/internal/keyserver/keyserver.go

// Package keyserver holds what end-to-end encryption needs a homeserver to
// keep for its users' devices: their identity keys, their one-time and
// fallback keys, the user's cross-signing keys and the signatures between
// all of those, the server-side backup of room keys, and the device list
// stream that tells other devices when any of it changed.
//
// None of it is something the server can use. Every key here is public, or
// encrypted to a key the server never sees; what the server provides is a
// place to publish them, the guarantee that a one-time key is handed out
// once, and the bookkeeping of who has to be told about a change. That is
// the whole of the design, and the reason the package is mostly checks:
// the server cannot tell a good key from a bad one, but it can refuse a
// key uploaded for somebody else's device, a cross-signing key that its
// master key did not sign, and a signature that does not verify.
//
// The transport of encrypted messages between devices — to-device
// messaging — is the syncapi's, because a device only ever receives one in
// a /sync. So is the reporting of device list changes, for the same
// reason: whose keys changed is one of the streams of a sync response.
// What is here is the writing of that stream.
//
// Phase 7 of PLAN.md.
package keyserver

import (
	"context"
	"errors"
	"fmt"
	"log/slog"
	"time"

	"github.com/dariuszjaniak/warp/server/internal/federation"
	"github.com/dariuszjaniak/warp/server/internal/matrix/id"
	"github.com/dariuszjaniak/warp/server/internal/storage"
)

// Errors callers have to tell apart. Each is wrapped with the detail of
// what was wrong, so a handler can answer with the right errcode and still
// tell the client which key it was.
var (
	// ErrInvalidParam is a request that is well-formed JSON but not a
	// valid one: keys for another user's device, a key ID without an
	// algorithm, a cross-signing key with the wrong usage.
	ErrInvalidParam = errors.New("invalid parameter")

	// ErrMissingParam is a request that leaves out something it needs,
	// such as a self-signing key with no master key to have signed it.
	ErrMissingParam = errors.New("missing parameter")

	// ErrInvalidSignature is a signature that does not verify.
	ErrInvalidSignature = errors.New("invalid signature")

	// ErrKeyConflict is a one-time key uploaded under an ID that already
	// holds a different one.
	ErrKeyConflict = storage.ErrKeyConflict

	// ErrNoSuchBackup is a key backup version that does not exist.
	ErrNoSuchBackup = storage.ErrNoSuchBackup

	// ErrNoSuchBackupKey is a room key that is not in the backup.
	ErrNoSuchBackupKey = storage.ErrNoSuchBackupKey
)

// WrongBackupVersionError is a write to a key backup version that is not
// the user's current one. The spec answers it with the current version, so
// that a client whose backup was replaced from another device finds out
// which one to use.
type WrongBackupVersionError struct {
	Current string
}

func (e *WrongBackupVersionError) Error() string {
	return fmt.Sprintf("backup version is not the current one (%s)", e.Current)
}

// Rooms says who shares a room with whom, which is who has to be told when
// a user's keys change. *roomserver.API satisfies it.
type Rooms interface {
	UsersSharingRooms(ctx context.Context, userID string) ([]string, error)
}

// Notifier wakes the long polls of users. *syncapi.Notifier satisfies it.
type Notifier interface {
	NotifyUsers(userIDs []string)
}

// Devices is the part of the userapi the keyserver reads: a user's devices,
// for their display names and to tell a device that still exists from one
// that does not. *userapi.API satisfies it.
type Devices interface {
	Devices(ctx context.Context, localpart string) ([]storage.Device, error)
}

// Remote asks other servers about their users' keys. *federation.Client
// satisfies it; it refuses every request while federation is off, which is
// what keeps a key query from reaching out with the switch down.
type Remote interface {
	QueryUserKeys(ctx context.Context, destination string, req federation.UserKeysQueryRequest) (federation.UserKeysQueryResponse, error)
	ClaimUserKeys(ctx context.Context, destination string, req federation.UserKeysClaimRequest) (federation.UserKeysClaimResponse, error)
	UserDevices(ctx context.Context, destination, userID string) (federation.UserDevicesResponse, error)
}

// Outbound queues EDUs to other servers. *federationapi.Queue satisfies it.
type Outbound interface {
	Running() bool
	QueueEDU(ctx context.Context, destinations []string, edu federation.EDU)
}

// API is the keyserver component.
type API struct {
	serverName string
	store      storage.KeyStore
	rooms      Rooms
	notifier   Notifier
	devices    Devices
	remote     Remote
	outbound   Outbound
	log        *slog.Logger

	// The background work of the remote key cache: fetching stale copies
	// again, and dropping the copies of users who no longer share a room
	// with anybody here.
	background *background
	resyncs    *workQueue
	sweeps     *workQueue
}

// Options configure the component.
type Options struct {
	// ServerName is the Matrix server name.
	ServerName string

	// Store is the key half of storage.
	Store storage.KeyStore

	// Rooms says who shares a room with a user. Nil means nobody does,
	// which is only useful in a test.
	Rooms Rooms

	// Notifier wakes the syncs a key change concerns. Nil wakes nobody.
	Notifier Notifier

	// Devices is the userapi.
	Devices Devices

	// Remote asks other servers for their users' keys. Nil fails every
	// remote user in a query or claim as unreachable.
	Remote Remote

	// Outbound sends device list updates to other servers. Nil, or a
	// queue that is not running, sends nothing.
	Outbound Outbound

	Logger *slog.Logger
}

// New builds the component.
func New(opts Options) (*API, error) {
	switch {
	case opts.ServerName == "":
		return nil, errors.New("keyserver: ServerName must not be empty")
	case opts.Store == nil:
		return nil, errors.New("keyserver: Store must not be nil")
	case opts.Devices == nil:
		return nil, errors.New("keyserver: Devices must not be nil")
	}
	if opts.Logger == nil {
		opts.Logger = slog.Default()
	}
	k := &API{
		serverName: opts.ServerName,
		store:      opts.Store,
		rooms:      opts.Rooms,
		notifier:   opts.Notifier,
		devices:    opts.Devices,
		remote:     opts.Remote,
		outbound:   opts.Outbound,
		log:        opts.Logger,
		background: newBackground(),
	}
	k.resyncs = newWorkQueue(k, k.resync)
	k.sweeps = newWorkQueue(k, k.sweep)
	return k, nil
}

// SetOutbound connects the federation queue, which is built after the
// keyserver's own dependencies. Nil disconnects it.
func (k *API) SetOutbound(o Outbound) { k.outbound = o }

// isLocal reports whether a user belongs to this server.
func (k *API) isLocal(userID string) bool {
	server, err := id.ServerNameOf(userID)
	return err == nil && server == k.serverName
}

// serverOf returns the server of a user, or "" for a malformed user ID.
func serverOf(userID string) string {
	server, err := id.ServerNameOf(userID)
	if err != nil {
		return ""
	}
	return server
}

// localpartOf returns the localpart of a local user.
func (k *API) localpartOf(userID string) (string, bool) {
	localpart, server, err := id.SplitUserID(userID)
	if err != nil || server != k.serverName {
		return "", false
	}
	return localpart, true
}

// defaultRemoteTimeout is how long a key query or claim waits for other
// servers when the client did not say. A client asking for keys is about
// to send a message, and a server that has not answered in ten seconds is
// better reported as a failure than waited on.
const defaultRemoteTimeout = 10 * time.Second

// maxRemoteTimeout caps what a client may ask for.
const maxRemoteTimeout = time.Minute

// remoteTimeout clamps a client's timeout.
func remoteTimeout(d time.Duration) time.Duration {
	switch {
	case d <= 0:
		return defaultRemoteTimeout
	case d > maxRemoteTimeout:
		return maxRemoteTimeout
	}
	return d
}

func invalidParamf(format string, args ...any) error {
	return fmt.Errorf("%w: %s", ErrInvalidParam, fmt.Sprintf(format, args...))
}

func missingParamf(format string, args ...any) error {
	return fmt.Errorf("%w: %s", ErrMissingParam, fmt.Sprintf(format, args...))
}

func invalidSignaturef(format string, args ...any) error {
	return fmt.Errorf("%w: %s", ErrInvalidSignature, fmt.Sprintf(format, args...))
}