| 1 | package release |
| 2 | |
| 3 | import ( |
| 4 | "context" |
| 5 | "crypto/ed25519" |
| 6 | "encoding/base64" |
| 7 | "encoding/json" |
| 8 | "errors" |
| 9 | "fmt" |
| 10 | "os" |
| 11 | "regexp" |
| 12 | "time" |
| 13 | ) |
| 14 | |
| 15 | // SystemPointer is the platform's own channel pointer: what a ChannelPointer |
| 16 | // is for an app, for the thing that does the adopting. It lives under |
| 17 | // control/, not apps/, because the substrate is not an app (the |
| 18 | // console-as-an-app spec's dividing line), and it is ROOT-KEY-ONLY: a scoped |
| 19 | // grant's rungs name app channels, and a key that can roll the platform is a |
| 20 | // bigger key than any app promoter should hold. Spec §3. |
| 21 | type SystemPointer struct { |
| 22 | ActivationID string `json:"activation_id,omitempty"` |
| 23 | CancellationID string `json:"cancellation_id,omitempty"` |
| 24 | Schema int `json:"schema"` |
| 25 | Component string `json:"component"` |
| 26 | Channel string `json:"channel"` |
| 27 | Current SystemRelease `json:"current"` |
| 28 | History []SystemRelease `json:"history,omitempty"` |
| 29 | Spread string `json:"spread"` |
| 30 | PromotedAt time.Time `json:"promoted_at"` // of THIS write; the box's replay guard reads only this |
| 31 | Audit []Promotion `json:"audit,omitempty"` |
| 32 | } |
| 33 | |
| 34 | // SystemRelease is everything a box needs to install one version. It travels |
| 35 | // whole into History so a rollback re-installs what was there, not a version |
| 36 | // string with someone else's artifacts. |
| 37 | type SystemRelease struct { |
| 38 | ConsoleArtifacts map[string]SystemArtifact `json:"console_artifacts,omitempty"` |
| 39 | ReleaseTag string `json:"release_tag,omitempty"` |
| 40 | SeedsProtocol int `json:"seeds_protocol,omitempty"` |
| 41 | MovesProtocol int `json:"moves_protocol,omitempty"` |
| 42 | Version string `json:"version"` |
| 43 | Artifacts map[string]SystemArtifact `json:"artifacts"` // keyed by "<goos>-<goarch>" |
| 44 | Source string `json:"source,omitempty"` |
| 45 | ReleasedAt time.Time `json:"released_at"` // audit only; never compared |
| 46 | } |
| 47 | |
| 48 | // SystemArtifact is one mirrored binary. Its sha in the signed pointer IS |
| 49 | // the artifact's signature: boxes need no .sig object. |
| 50 | type SystemArtifact struct { |
| 51 | Key string `json:"key"` |
| 52 | SHA256 string `json:"sha256"` |
| 53 | } |
| 54 | |
| 55 | // SystemComponentRouter is the one component today: the substrate binary |
| 56 | // (edge + activator + adoption, one process). A split adds a name here. |
| 57 | const SystemComponentRouter = "router" |
| 58 | |
| 59 | // ErrNoSystemHistory is RollbackSystem's answer when the pointer has never |
| 60 | // moved: there is nothing older to go back to. |
| 61 | var ErrNoSystemHistory = errors.New("system pointer: nothing to roll back to") |
| 62 | |
| 63 | // ErrBadSystemPointer marks every way an object on the wire can fail to be |
| 64 | // this component and channel's signed pointer: a torn wrapper, a grant, a |
| 65 | // signature that does not verify, a payload that does not validate. Callers |
| 66 | // that must tell "this is not authority" from "the bucket is unreachable" — |
| 67 | // the box's self-update refuses on the first and defers on the second — test |
| 68 | // with errors.Is instead of reading the message. |
| 69 | var ErrBadSystemPointer = errors.New("system pointer does not verify") |
| 70 | |
| 71 | const ( |
| 72 | systemSpreadMax = 24 * time.Hour |
| 73 | systemSpreadDefault = 5 * time.Minute |
| 74 | ) |
| 75 | |
| 76 | var ( |
| 77 | systemComponentRe = regexp.MustCompile(`^[a-z][a-z0-9-]{0,31}$`) |
| 78 | systemPlatformRe = regexp.MustCompile(`^[a-z0-9]+-[a-z0-9]+$`) |
| 79 | systemSHA256Re = regexp.MustCompile(`^[0-9a-f]{64}$`) |
| 80 | ) |
| 81 | |
| 82 | // SystemPointerKey is control/system/<component>/<channel>.json. |
| 83 | func SystemPointerKey(component, channel string) string { |
| 84 | return "control/system/" + component + "/" + channel + ".json" |
| 85 | } |
| 86 | |
| 87 | // SystemArtifactKey is bootstrap/router/<version>/carlos-<platform>: under |
| 88 | // bootstrap/ because the store door already means "the platform's own |
| 89 | // binary" by that name, per-version because canary and stable boxes must be |
| 90 | // able to run different binaries at once. |
| 91 | func SystemArtifactKey(version, platform string) string { |
| 92 | return "bootstrap/router/" + version + "/carlos-" + platform |
| 93 | } |
| 94 | |
| 95 | // SpreadDuration is Spread parsed, or the default when it is unparseable — |
| 96 | // validation refuses the unparseable case on the way in, so this only |
| 97 | // defaults for a hand-edited or pre-field pointer. |
| 98 | func (p *SystemPointer) SpreadDuration() time.Duration { |
| 99 | d, err := time.ParseDuration(p.Spread) |
| 100 | if err != nil || d < 0 || d > systemSpreadMax { |
| 101 | return systemSpreadDefault |
| 102 | } |
| 103 | return d |
| 104 | } |
| 105 | |
| 106 | func validateSystemRelease(r SystemRelease) error { |
| 107 | if !ValidVersion(r.Version) || r.Version == "dev" { |
| 108 | return fmt.Errorf("version %q is not a valid platform version", r.Version) |
| 109 | } |
| 110 | for platform, artifact := range r.ConsoleArtifacts { |
| 111 | if !systemPlatformRe.MatchString(platform) || artifact.Key != "bootstrap/console/"+r.Version+"/carlos-"+platform || !systemSHA256Re.MatchString(artifact.SHA256) { |
| 112 | return ErrBadSystemPointer |
| 113 | } |
| 114 | } |
| 115 | if len(r.Artifacts) == 0 { |
| 116 | return fmt.Errorf("version %s names no artifacts", r.Version) |
| 117 | } |
| 118 | for plat, a := range r.Artifacts { |
| 119 | if !systemPlatformRe.MatchString(plat) { |
| 120 | return fmt.Errorf("platform %q is not <goos>-<goarch>", plat) |
| 121 | } |
| 122 | if a.Key != SystemArtifactKey(r.Version, plat) { |
| 123 | return fmt.Errorf("artifact key %q is not %s", a.Key, SystemArtifactKey(r.Version, plat)) |
| 124 | } |
| 125 | if !systemSHA256Re.MatchString(a.SHA256) { |
| 126 | return fmt.Errorf("artifact %s: sha256 %q is not 64 hex characters", a.Key, a.SHA256) |
| 127 | } |
| 128 | } |
| 129 | return nil |
| 130 | } |
| 131 | |
| 132 | func validateSystemPointer(p *SystemPointer, wantComponent, wantChannel string) error { |
| 133 | if p.Schema != 1 { |
| 134 | return fmt.Errorf("system pointer schema %d unsupported", p.Schema) |
| 135 | } |
| 136 | if p.Component != wantComponent || p.Channel != wantChannel { |
| 137 | return fmt.Errorf("system pointer is for %s/%s, not %s/%s", p.Component, p.Channel, wantComponent, wantChannel) |
| 138 | } |
| 139 | if !systemComponentRe.MatchString(p.Component) || !ValidChannel(p.Channel) { |
| 140 | return fmt.Errorf("system pointer %s/%s: bad component or channel", p.Component, p.Channel) |
| 141 | } |
| 142 | if p.CancellationID != "" && p.Current.Version == "" { |
| 143 | return nil |
| 144 | } |
| 145 | if err := validateSystemRelease(p.Current); err != nil { |
| 146 | return fmt.Errorf("system pointer %s/%s: %w", p.Component, p.Channel, err) |
| 147 | } |
| 148 | if d, err := time.ParseDuration(p.Spread); err != nil || d < 0 || d > systemSpreadMax { |
| 149 | return fmt.Errorf("system pointer %s/%s: spread %q must be a duration between 0 and %s", p.Component, p.Channel, p.Spread, systemSpreadMax) |
| 150 | } |
| 151 | if p.PromotedAt.IsZero() { |
| 152 | return fmt.Errorf("system pointer %s/%s: promoted_at unset", p.Component, p.Channel) |
| 153 | } |
| 154 | if len(p.History) > historyMax { |
| 155 | return fmt.Errorf("system pointer %s/%s: history longer than %d", p.Component, p.Channel, historyMax) |
| 156 | } |
| 157 | for _, h := range p.History { |
| 158 | if err := validateSystemRelease(h); err != nil { |
| 159 | return fmt.Errorf("system pointer %s/%s: history: %w", p.Component, p.Channel, err) |
| 160 | } |
| 161 | } |
| 162 | if len(p.Audit) > auditMax { |
| 163 | return fmt.Errorf("system pointer %s/%s: audit longer than %d", p.Component, p.Channel, auditMax) |
| 164 | } |
| 165 | return nil |
| 166 | } |
| 167 | |
| 168 | // signSystemPointer mirrors signPointer with two differences: the signer |
| 169 | // must be the root key, and the wrapper never carries a grant. |
| 170 | func signSystemPointer(sg Signer, p *SystemPointer) ([]byte, error) { |
| 171 | if err := sg.usable(); err != nil { |
| 172 | return nil, err |
| 173 | } |
| 174 | if sg.scope != nil { |
| 175 | return nil, errors.New("system pointer: signed by a scoped key — the platform takes the root key only") |
| 176 | } |
| 177 | p.Schema = 1 |
| 178 | p.PromotedAt = time.Now().UTC().Truncate(time.Second) |
| 179 | if err := validateSystemPointer(p, p.Component, p.Channel); err != nil { |
| 180 | return nil, err |
| 181 | } |
| 182 | payload, err := json.MarshalIndent(p, "", " ") |
| 183 | if err != nil { |
| 184 | return nil, err |
| 185 | } |
| 186 | return signRootPayload(sg, payload) |
| 187 | } |
| 188 | |
| 189 | // verifySystemPointer is the single parse site for system pointers, and so |
| 190 | // the one place ErrBadSystemPointer is attached: every refusal below reaches |
| 191 | // the caller wrapped, whatever its own wording. |
| 192 | func verifySystemPointer(pub ed25519.PublicKey, wrapper []byte, wantComponent, wantChannel string) (*SystemPointer, error) { |
| 193 | p, err := parseSystemPointer(pub, wrapper, wantComponent, wantChannel) |
| 194 | if err != nil { |
| 195 | return nil, fmt.Errorf("%w: %v", ErrBadSystemPointer, err) |
| 196 | } |
| 197 | return p, nil |
| 198 | } |
| 199 | |
| 200 | // parseSystemPointer does the work. A grant on the wire is refused before |
| 201 | // the signature is looked at (the production flag's rule, production.go). |
| 202 | func parseSystemPointer(pub ed25519.PublicKey, wrapper []byte, wantComponent, wantChannel string) (*SystemPointer, error) { |
| 203 | var sw signedWrapper |
| 204 | if err := json.Unmarshal(wrapper, &sw); err != nil { |
| 205 | return nil, fmt.Errorf("system pointer %s/%s: %w", wantComponent, wantChannel, err) |
| 206 | } |
| 207 | if sw.Grant != nil { |
| 208 | return nil, fmt.Errorf("system pointer %s/%s: signed by a scoped key — the platform takes the root key only", wantComponent, wantChannel) |
| 209 | } |
| 210 | sig, err := base64.StdEncoding.DecodeString(sw.Sig) |
| 211 | if err != nil { |
| 212 | return nil, fmt.Errorf("system pointer %s/%s: bad signature encoding", wantComponent, wantChannel) |
| 213 | } |
| 214 | if len(pub) != ed25519.PublicKeySize || !ed25519.Verify(pub, []byte(sw.Payload), sig) { |
| 215 | return nil, fmt.Errorf("system pointer %s/%s: signature does not verify", wantComponent, wantChannel) |
| 216 | } |
| 217 | payload, err := base64.StdEncoding.DecodeString(sw.Payload) |
| 218 | if err != nil { |
| 219 | return nil, fmt.Errorf("system pointer %s/%s: bad payload encoding", wantComponent, wantChannel) |
| 220 | } |
| 221 | var p SystemPointer |
| 222 | if err := json.Unmarshal(payload, &p); err != nil { |
| 223 | return nil, fmt.Errorf("system pointer %s/%s: %w", wantComponent, wantChannel, err) |
| 224 | } |
| 225 | if err := validateSystemPointer(&p, wantComponent, wantChannel); err != nil { |
| 226 | return nil, err |
| 227 | } |
| 228 | return &p, nil |
| 229 | } |
| 230 | |
| 231 | // GetSystemPointer reads and verifies one channel's pointer. A missing key |
| 232 | // is os.ErrNotExist, which is the normal state of every deployment before |
| 233 | // its first `carlos system update`. |
| 234 | func GetSystemPointer(ctx context.Context, s Store, pub ed25519.PublicKey, component, channel string) (*SystemPointer, string, error) { |
| 235 | body, etag, err := s.Get(ctx, SystemPointerKey(component, channel)) |
| 236 | if err != nil { |
| 237 | return nil, "", err |
| 238 | } |
| 239 | p, err := verifySystemPointer(pub, body, component, channel) |
| 240 | if err != nil { |
| 241 | return nil, "", err |
| 242 | } |
| 243 | if p.CancellationID != "" && p.Current.Version == "" { |
| 244 | return nil, etag, os.ErrNotExist |
| 245 | } |
| 246 | return p, etag, nil |
| 247 | } |
| 248 | |
| 249 | // SystemPromoteOptions: who did it, and how long boxes spread their starts. |
| 250 | type SystemPromoteOptions struct { |
| 251 | Actor string |
| 252 | Spread time.Duration // 0 = hotfix, every box at once; unset in the CLI means the default |
| 253 | } |
| 254 | |
| 255 | // PromoteSystem writes the pointer for one channel, CAS-guarded. The |
| 256 | // outgoing release goes whole into History unless it is the same version |
| 257 | // (a re-promote of the running version is a no-op for boxes and must not |
| 258 | // push that version into its own history). |
| 259 | func PromoteSystem(ctx context.Context, s Store, sg Signer, component, channel string, rel SystemRelease, o SystemPromoteOptions) (*SystemPointer, error) { |
| 260 | prepared, err := PrepareSystemPromotion(ctx, s, sg, component, channel, rel, o, "") |
| 261 | if err != nil { |
| 262 | return nil, err |
| 263 | } |
| 264 | return PublishSystemPublication(ctx, s, sg.verifyPub(), prepared) |
| 265 | } |
| 266 | |
| 267 | func PrepareSystemPromotion(ctx context.Context, s Store, sg Signer, component, channel string, rel SystemRelease, o SystemPromoteOptions, id string) (SystemPublication, error) { |
| 268 | key := SystemPointerKey(component, channel) |
| 269 | var old *SystemPointer |
| 270 | body, etag, err := s.Get(ctx, key) |
| 271 | switch { |
| 272 | case err == nil: |
| 273 | if old, err = verifySystemPointer(sg.verifyPub(), body, component, channel); err != nil { |
| 274 | return SystemPublication{}, err |
| 275 | } |
| 276 | case errors.Is(err, os.ErrNotExist): |
| 277 | etag = "" |
| 278 | default: |
| 279 | return SystemPublication{}, err |
| 280 | } |
| 281 | if old != nil && old.CancellationID != "" && old.Current.Version == "" { |
| 282 | old = nil |
| 283 | } |
| 284 | next := &SystemPointer{ActivationID: id, Component: component, Channel: channel, Current: rel, Spread: o.Spread.String()} |
| 285 | if rel.ReleasedAt.IsZero() { |
| 286 | next.Current.ReleasedAt = time.Now().UTC().Truncate(time.Second) |
| 287 | } |
| 288 | from := "" |
| 289 | if old != nil { |
| 290 | next.History, next.Audit = old.History, old.Audit |
| 291 | from = old.Current.Version |
| 292 | if old.Current.Version != rel.Version { |
| 293 | next.History = prependCapped(old.Current, old.History, historyMax) |
| 294 | } |
| 295 | } |
| 296 | next.Audit = prependCapped(Promotion{Version: rel.Version, From: from, To: channel, Actor: o.Actor, At: time.Now().UTC()}, next.Audit, auditMax) |
| 297 | wrapper, err := signSystemPointer(sg, next) |
| 298 | if err != nil { |
| 299 | return SystemPublication{}, err |
| 300 | } |
| 301 | return SystemPublication{Component: component, Channel: channel, ID: id, ExpectedETag: etag, Before: body, After: wrapper}, nil |
| 302 | } |
| 303 | |
| 304 | // RollbackSystem POPS History[0] rather than swapping with it: History[0] |
| 305 | // becomes Current and is removed from History, and the release that was |
| 306 | // Current is not pushed back on. So a second rollback goes one release |
| 307 | // further back, not forward again to where it started — the verb means "go |
| 308 | // back another one", and repeating it walks the history down. |
| 309 | // |
| 310 | // It writes a NEW pointer — new PromotedAt — so every box's replay guard |
| 311 | // sees a promote, not a replay. |
| 312 | func RollbackSystem(ctx context.Context, s Store, sg Signer, component, channel, actor string, expectedETag ...string) (*SystemPointer, error) { |
| 313 | prepared, err := PrepareSystemRollback(ctx, s, sg, component, channel, actor, "", expectedETag...) |
| 314 | if err != nil { |
| 315 | return nil, err |
| 316 | } |
| 317 | return PublishSystemPublication(ctx, s, sg.verifyPub(), prepared) |
| 318 | } |
| 319 | |
| 320 | func PrepareSystemRollback(ctx context.Context, s Store, sg Signer, component, channel, actor, id string, expectedETag ...string) (SystemPublication, error) { |
| 321 | key := SystemPointerKey(component, channel) |
| 322 | body, etag, err := s.Get(ctx, key) |
| 323 | if err != nil { |
| 324 | return SystemPublication{}, err |
| 325 | } |
| 326 | if len(expectedETag) > 1 || (len(expectedETag) == 1 && etag != expectedETag[0]) { |
| 327 | return SystemPublication{}, ErrCASConflict |
| 328 | } |
| 329 | old, err := verifySystemPointer(sg.verifyPub(), body, component, channel) |
| 330 | if err != nil { |
| 331 | return SystemPublication{}, err |
| 332 | } |
| 333 | if len(old.History) == 0 { |
| 334 | return SystemPublication{}, ErrNoSystemHistory |
| 335 | } |
| 336 | next := &SystemPointer{ActivationID: id, |
| 337 | Component: component, Channel: channel, |
| 338 | Current: old.History[0], History: old.History[1:], Spread: old.Spread, |
| 339 | Audit: prependCapped(Promotion{Version: old.History[0].Version, From: old.Current.Version, To: channel, Actor: actor, At: time.Now().UTC()}, old.Audit, auditMax), |
| 340 | } |
| 341 | wrapper, err := signSystemPointer(sg, next) |
| 342 | if err != nil { |
| 343 | return SystemPublication{}, err |
| 344 | } |
| 345 | return SystemPublication{Component: component, Channel: channel, ID: id, ExpectedETag: etag, Before: body, After: wrapper}, nil |
| 346 | } |
| 347 | |
amadan