Unverified Commit 0414913b authored by Lemon-miaow's avatar Lemon-miaow
Browse files

fix(api): serialise RedeemPlayerBindCode — concurrent redeem 500s become clean... 

fix(api): serialise RedeemPlayerBindCode — concurrent redeem 500s become clean 400s/idempotent converges

6-way concurrent redeem of one code 500'd on users_username_key (each request
generated a fresh user id but the same uuid-derived username), plus the rarer
two-codes-one-uuid race. Same drift family as VerifyLinkCode, which already
locks its code row and handles the conflict.

- SELECT ... FOR UPDATE the code row: same-code racers serialise; losers exit
  as ErrLinkCodeInvalid (400 invalid_code), no user row is attempted.
- INSERT users ... ON CONFLICT (username) DO NOTHING + re-read by username:
  cross-code racers converge on the winner's row (role checked, staff still
  refused) instead of a unique-violation 500.
- account_links ON CONFLICT (mc_uuid) DO NOTHING for the same race.

Verified live: same-code x6 = 1x200 + 5x400; two-codes x2 = 2x200 same user;
db clean; zero unmapped errors.
parent dcc3b740

AUDIT-2026-09-22.md

+4 −0
Changes for AUDIT-2026-09-22.md: 4 added lines, 0 removed lines.
Original line number Diff line number Diff line
@@ -15,6 +15,7 @@ IPv6-only 接入（`ssh -6 -i ~/.ssh/id_ed25519 root@fdb2:2c26:f4e4:0:21c:42ff:f 
| 5 | 依赖漏洞：pgx v5.7.1（GO-2026-5004，可达 pgrepo.go）、x/net、x/text | govulncheck | 升 pgx v5.9.2 / x/net v0.55.0 / x/text v0.39.0 |

| 16 | **`ConsumeLoginEmailOTP` PG 实现与接口契约漂移**：契约/`fakeRepo` 说「同 VerifyEmailOTP 生命周期（扣尝试/锁定/ErrOTPInvalid/ErrOTPLocked）」，PG 却是单条 UPDATE+`ErrNotFound` → 错码/重放/过期在 login 门、op-login finish、migration confirm 三个入口全部 500；且尝试次数永不累计、`otpMaxAttempts` 锁定失效 | 真机：login 门重放正确码 → **HTTP 500**；修复前错码不扣次。修复后复测：5 次错码 400 且 attempts=5（正确码因锁定也 400、码未消费）、新码可用、重放 400 | `pgrepo.go` 改置为 VerifyEmailOTP 同构事务（FOR UPDATE、先锁后比、mismatch 扣次、match 消费），无 users 写副作用。提交 `52549f7` |

| 17 | **`ListPendingOpLogins` PG 少列**：接口注释承诺「joined to its staff username」，handler 输出 `username`/`created_at`，fake 正确填充；PG SQL 未 JOIN 也未取 `created_at` → 真机 pending 列表 username 为空、created_at 为 `0001-01-01` | 真机 internal `/op-login/pending` 响应 | `pgrepo.go` SQL 改为 JOIN users + 取 created_at。提交见分支 |

| 18 | **绑定码并发兑换 500**：`RedeemPlayerBindCode` 无 `FOR UPDATE`（同文件 `VerifyLinkCode` 有），且裸 INSERT。6 路并发同码兑换 → 3×HTTP 500（`users_username_key` 唯一冲突）+ 1×400 + 2×200；跨码并发同 UUID 同样会撞 | 真机四组并发测试 + API 日志 `unmapped error ... duplicate key` | 同码：码行 `FOR UPDATE`（输家干净地 400 invalid_code）；跨码：`INSERT users ... ON CONFLICT (username) DO NOTHING`+重读、`account_links ON CONFLICT (mc_uuid) DO NOTHING`（两路汇聚同一 user）。复测：同码×6=1×200+5×400、双码×2=2×200 同 ID、DB 干净、日志零 unmapped |



## 待决策台账（未修）
@@ -40,6 +41,9 @@ IPv6-only 接入（`ssh -6 -i ~/.ssh/id_ed25519 root@fdb2:2c26:f4e4:0:21c:42ff:f 
- **毒档案 fail-closed**：穿越/绝对路径/符号链接条目 → `archive entry escapes target` 退出码 1，零写入 ✅

- 混沌：PG 掉线（healthz 仍 200、恢复后连接池自愈）、API pod 击杀（~2s 中断）、整机重启（32s 回归、会话/CRD/停止态全保留）✅

- `felis update` 报告（k3s/velocity 有更新、私有仓库 404 优雅处理）✅

- 账户全套真机 E2E：绑定码新玩家注册（幂等/并发见 #18）、邮箱验证（onboarding 门）、邮箱 OTP 登录（错码扣次/5 次锁定后正确码也 400、重放 400、staff 账号 403 拒绝）、Passkey 注册+discoverable 登录+邮箱优先登录（虚拟认证器；Chrome 要求 `Page.bringToFront` 才能过 focus 检查）、登出吊销会话（旧 cookie 401）

- op-login 全状态机（start→status→approve→finish；早 finish 不烧码、错码扣次且请求保留、重放/重复批准/非管理员批准/未知 handle 全部按契约返回）✅

- 并发：OTP 风暴 8 路 = 1×202 + 7×429 且仅铸 1 码；绑定码并发（同码/双码）见 #18 ✅



## 复现入口速查

internal/api/pgrepo.go

+20 −5
Changes for internal/api/pgrepo.go: 20 added lines, 5 removed lines.
Original line number Diff line number Diff line
@@ -146,13 +146,17 @@ func (p *PGRepo) RedeemPlayerBindCode(ctx context.Context, newUserID, code strin 


	var mcUUID, authSource string

	switch err := tx.QueryRowContext(ctx,

		`SELECT mc_uuid, auth_source FROM account_link_codes WHERE code = $1 AND expires_at > $2`,

		`SELECT mc_uuid, auth_source FROM account_link_codes WHERE code = $1 AND expires_at > $2

		 FOR UPDATE`,

		code, now).Scan(&mcUUID, &authSource); {

	case errors.Is(err, sql.ErrNoRows):

		return "", "", "", ErrLinkCodeInvalid

	case err != nil:

		return "", "", "", err

	}

	// The lock above serialises redeemers of ONE code; the ON CONFLICT arms below

	// cover the rarer cross-code race (two live codes for the same UUID redeemed

	// together), where both transactions reach the inserts before either commits.



	// Create-or-fetch keyed on the verified UUID. An already-linked role='user' player

	// is fetched (idempotent "log in via the game"); a role='admin' STAFF account is
@@ -165,16 +169,27 @@ func (p *PGRepo) RedeemPlayerBindCode(ctx context.Context, newUserID, code strin 
		mcUUID).Scan(&userID, &existingRole); {

	case errors.Is(err, sql.ErrNoRows):

		if _, err := tx.ExecContext(ctx,

			`INSERT INTO users (id, username, role) VALUES ($1, $2, 'user')`,

			`INSERT INTO users (id, username, role) VALUES ($1, $2, 'user')

			 ON CONFLICT (username) DO NOTHING`,

			newUserID, mcUUID); err != nil {

			return "", "", "", fmt.Errorf("create player: %w", err)

		}

		// Re-read by username so a cross-code race converges on the winner's row

		// (our id was discarded by DO NOTHING) instead of a bare 500.

		var role string

		if err := tx.QueryRowContext(ctx,

			`SELECT id, role::text FROM users WHERE username = $1`, mcUUID).Scan(&userID, &role); err != nil {

			return "", "", "", fmt.Errorf("create player: %w", err)

		}

		if role != "user" {

			return "", "", "", ErrPlayerBindForbidden

		}

		if _, err := tx.ExecContext(ctx,

			`INSERT INTO account_links (user_id, mc_uuid, auth_source) VALUES ($1, $2, $3)`,

			newUserID, mcUUID, authSource); err != nil {

			`INSERT INTO account_links (user_id, mc_uuid, auth_source) VALUES ($1, $2, $3)

			 ON CONFLICT (mc_uuid) DO NOTHING`,

			userID, mcUUID, authSource); err != nil {

			return "", "", "", fmt.Errorf("write account link: %w", err)

		}

		userID = newUserID

	case err != nil:

		return "", "", "", err

	default: