| 1 | package console |
| 2 | |
| 3 | // Verifying an untrusted GeoIP database in a child process (code review |
| 4 | // R9). A .mmdb from a feed or an upload is parsed by maxminddb, which |
| 5 | // trusts its metadata: a crafted file can panic it (node_count arithmetic) |
| 6 | // or make its tree walk run for 2^128 steps inside a single call. Neither |
| 7 | // may take the console down or hold the publish slot past its deadline, so |
| 8 | // the console never parses such a file itself: it runs its own binary with |
| 9 | // a hidden verb, kills it at the deadline, and reads back a small report. |
| 10 | |
| 11 | import ( |
| 12 | "bytes" |
| 13 | "context" |
| 14 | "encoding/json" |
| 15 | "errors" |
| 16 | "fmt" |
| 17 | "io" |
| 18 | "net" |
| 19 | "os" |
| 20 | "os/exec" |
| 21 | "time" |
| 22 | |
| 23 | "github.com/carlosframework/platform/internal/geo" |
| 24 | ) |
| 25 | |
| 26 | // GeoipVerifyVerb is cmd/console's hidden verb for the verifier child. |
| 27 | const GeoipVerifyVerb = "__geoip-verify" |
| 28 | |
| 29 | // geoipVerifyResult is the child's report. |
| 30 | type geoipVerifyResult struct { |
| 31 | Source string `json:"source"` |
| 32 | BuildEpoch int64 `json:"build_epoch"` |
| 33 | NodeCount uint64 `json:"node_count"` |
| 34 | Countries int `json:"countries"` // how many probe addresses answered a country |
| 35 | } |
| 36 | |
| 37 | // RunGeoipVerify is the child: args are the file, then the probe |
| 38 | // addresses. It writes the report to w and returns the exit code (0 ok, 2 |
| 39 | // refused). A panic or a hang here kills only this process. |
| 40 | func RunGeoipVerify(args []string, w io.Writer) int { |
| 41 | if len(args) < 1 { |
| 42 | return 2 |
| 43 | } |
| 44 | // Measured: DB-IP City Lite (127 MB) verifies within 256 MiB of |
| 45 | // private memory; 512 MiB leaves room for GeoLite2 and growth. |
| 46 | if err := limitVerifierMemory(geoipVerifyDataLimit); err != nil { |
| 47 | return 2 // without the bound this process must not parse anything |
| 48 | } |
| 49 | db, err := geo.OpenMMDBVerified(context.Background(), args[0]) |
| 50 | if errors.Is(err, geo.ErrSearchTreeTooLarge) { |
| 51 | return geoipVerifyExitTree |
| 52 | } |
| 53 | if err != nil { |
| 54 | return 2 |
| 55 | } |
| 56 | defer db.Close() |
| 57 | // Every record through the production schema, here where a pathological |
| 58 | // one can only kill this process (code review R11-1). |
| 59 | if err := db.CheckRecords(context.Background()); err != nil { |
| 60 | return 2 |
| 61 | } |
| 62 | res := geoipVerifyResult{Source: db.Source(), BuildEpoch: db.BuildEpoch().Unix(), NodeCount: db.NodeCount()} |
| 63 | for _, a := range args[1:] { |
| 64 | if r, ok := db.Lookup(net.ParseIP(a)); ok && r.Country != "" { |
| 65 | res.Countries++ |
| 66 | } |
| 67 | } |
| 68 | if err := json.NewEncoder(w).Encode(res); err != nil { |
| 69 | return 2 |
| 70 | } |
| 71 | return 0 |
| 72 | } |
| 73 | |
| 74 | // geoipVerifyDataLimit is the child's private-memory cap (code review R10-1). |
| 75 | const geoipVerifyDataLimit = 512 << 20 |
| 76 | |
| 77 | // geoipVerifyTimeout bounds the child; a var so a test need not wait. |
| 78 | var geoipVerifyTimeout = 5 * time.Minute |
| 79 | |
| 80 | // geoipVerifyExec starts the child; tests point it at the test binary. |
| 81 | var geoipVerifyExec = func(ctx context.Context, args ...string) *exec.Cmd { |
| 82 | self, err := os.Executable() |
| 83 | if err != nil { |
| 84 | self = os.Args[0] |
| 85 | } |
| 86 | return exec.CommandContext(ctx, self, append([]string{GeoipVerifyVerb}, args...)...) |
| 87 | } |
| 88 | |
| 89 | var ( |
| 90 | errGeoipVerifyTimeout = errors.New("verification took too long") |
| 91 | errGeoipVerifyTree = errors.New("malformed search tree") |
| 92 | ) |
| 93 | |
| 94 | // geoipVerifyExitTree is the child's exit code for a search tree with far |
| 95 | // more networks than nodes. |
| 96 | const geoipVerifyExitTree = 3 |
| 97 | |
| 98 | // verifyInChild runs the child and returns its report. A crash, a refusal |
| 99 | // or a malformed report is an invalid database; the deadline is its own |
| 100 | // category; the caller's own cancellation is returned as it is. |
| 101 | func verifyInChild(ctx context.Context, path string, probes []string) (geoipVerifyResult, error) { |
| 102 | vctx, cancel := context.WithTimeout(ctx, geoipVerifyTimeout) |
| 103 | defer cancel() |
| 104 | cmd := geoipVerifyExec(vctx, append([]string{path}, probes...)...) |
| 105 | var out bytes.Buffer |
| 106 | cmd.Stdout = &cappedBuffer{buf: &out, max: 4 << 10} |
| 107 | cmd.Stderr = io.Discard |
| 108 | cmd.WaitDelay = 5 * time.Second |
| 109 | err := cmd.Run() |
| 110 | switch { |
| 111 | case ctx.Err() != nil: |
| 112 | return geoipVerifyResult{}, ctx.Err() |
| 113 | case vctx.Err() != nil: |
| 114 | return geoipVerifyResult{}, errGeoipVerifyTimeout |
| 115 | case err != nil: |
| 116 | var ee *exec.ExitError |
| 117 | if errors.As(err, &ee) && ee.ExitCode() == geoipVerifyExitTree { |
| 118 | return geoipVerifyResult{}, errGeoipVerifyTree |
| 119 | } |
| 120 | return geoipVerifyResult{}, fmt.Errorf("verifier: %w", err) |
| 121 | } |
| 122 | var res geoipVerifyResult |
| 123 | if err := json.Unmarshal(out.Bytes(), &res); err != nil { |
| 124 | return geoipVerifyResult{}, fmt.Errorf("verifier report: %w", err) |
| 125 | } |
| 126 | return res, nil |
| 127 | } |
| 128 | |
| 129 | type cappedBuffer struct { |
| 130 | buf *bytes.Buffer |
| 131 | max int |
| 132 | } |
| 133 | |
| 134 | func (c *cappedBuffer) Write(p []byte) (int, error) { |
| 135 | if c.buf.Len()+len(p) > c.max { |
| 136 | return 0, errors.New("verifier report too large") |
| 137 | } |
| 138 | return c.buf.Write(p) |
| 139 | } |
| 140 | |
| 141 | // VerifyGeoIPFile checks a database in the verifier child; nil means it is |
| 142 | // safe to open in this process with geo.OpenMMDBTrusted. cmd/console uses |
| 143 | // it for the CONSOLE_GEOIP_DB override, so even an operator's broken file |
| 144 | // cannot crash the console at startup. |
| 145 | func VerifyGeoIPFile(ctx context.Context, path string) error { |
| 146 | _, err := verifyInChild(ctx, path, nil) |
| 147 | return err |
| 148 | } |
| 149 | |
amadan