logtail: Inbound log takes Postfix's own tail, not a recipient's (audit M2)

Security audit 2026-10-17, finding M2, stage 0 - proven on the live server:
Postfix quotes a refused recipient in its reply in internal form (quotes
removed, "; < > =" kept: extract_addr, tok822_internalize), so
RCPT TO:<"; from=<ceo@…> to=<board@…> proto=ESMTP helo=<x>; "@domain>
spelled a whole tail inside the reply, and inboundRejectRe, whose reply
ended at the FIRST "; from=<", took the stranger's sender, recipient and
inbound domain for the row.

Now the reply is greedy and the tail anchored: "; from=<…> to=<…>
proto=\S+", a HELO name holding none of Postfix's NEUTER_CHARACTERS
(smtpd.c:1634, helo_cmd neuters ` <>()\";@`), and one-word SASL fields to
the end of the line (a `.*` SASL tail would have let a quoted MAIL FROM
spell the whole tail). Belt and braces: a reply that still holds
"; from=<" - a line with two tails - makes no row, so neither a recipient
nor an un-neutered HELO name can choose one. The quote/bracket/space
filter on from/to stays.

Tests: TestInboundRecipientCannotWriteTheRow (the live line's shape, and
with a forged MAIL FROM too); TestInboundHeloCannotWriteTheRow now feeds
HELO names as Postfix logs them (neutered) and expects no row for the
un-neutered ones. go test ./internal/logtail on Windows (the three known
Windows-only Follow failures aside).

Docs: security.md § Mail path; CHANGELOG; plan line ticked.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

CHANGELOG.md

@@ -13,6 +13,11 @@ Format follows [Keep a Changelog](https://keepachangelog.com/en/1.1.0/); version
   check. Adding a user, setting a password, changing an e-mail or a role on
   the user form, and deleting a user now ask for your own account password,
   as a backup does, under the same limit on wrong guesses.
+- **A stranger's recipient no longer writes an Inbound › Log row** (audit
+  M2): Postfix quotes a refused recipient in its reply with the quotes
+  removed, so `RCPT TO:<"; from=<x> to=<y> …"@domain>` spelled a second
+  sender and recipient there, and the parser took the first. It now takes
+  Postfix's own tail at the end of the line.
 - **The entrypoint follows no link at the DMARC maps** (audit I2): the
   three `dmarc_*` maps root creates at the start were missing from the list
   of paths where a symbolic link is removed first.

docs/security-audit-2026-10-17.md

@@ -697,10 +697,13 @@ the argument.
   in postfix-config.sh:538,545; an e2e case plants a link and expects the
   log line. — **Opus**, 0.5 h. *Done 2026-10-17:* TestStartsOnA28Data
   plants a dangling `dmarc_transport` with DMARC reports on.
-- [ ] **M2.** Greedy reply group anchored on Postfix's own tail; the live
+- [x] **M2.** Greedy reply group anchored on Postfix's own tail; the live
   line above as a test; security.md § Mail path reworded ("the last `;
   from=<`, which is Postfix's because a HELO name can hold none of
-  `<>;\"`"). — **Opus**, 1 h.
+  `<>;\"`"). — **Opus**, 1 h. *Done 2026-10-17:* the HELO name in the
+  tail may hold none of the neutered characters and each SASL field is one
+  word (a `.*` SASL tail would have let a quoted sender spell the whole
+  tail); TestInboundRecipientCannotWriteTheRow.
 - [ ] **M5.** Keep at most two `From:`/`Sender:` values; security.md:810
   reworded. — **Opus**, 0.5 h.
 - [ ] **A4.** Duplicate-cookie line through `loggedPath`. — **Sonnet**, 0.25 h.

docs/security.md

@@ -441,10 +441,17 @@ operator can act on it.
   The signature goes on to the upstream with the mail and is of no use there:
   it fits one queue id. Refusals are
   parsed from `mail.log` from the head of the line, with the reply ending at
-  Postfix's own `; from=<` so that the HELO name after it - the client's text,
+  the *last* `; from=<…> to=<…> proto=…` that runs to the end of the line.
-  which Postfix logs with non-printable characters and ` <>()\";@` turned into
+  That tail is Postfix's own: the HELO name in it is logged with
-  `?` (Postfix 3.10.13, src/smtpd/smtpd.c `helo_cmd()`) - cannot stand in for
+  non-printable characters and ` <>()\";@` turned into `?` (Postfix 3.10.13,
-  the sender, the recipient or the reply; an address with a quote, a bracket or a space in it is not kept.
+  src/smtpd/smtpd.c `helo_cmd()`), and the pattern takes none of them there;
+  a refused recipient, which the reply quotes with its quotes removed and
+  `; < > =` kept, can spell such a tail only inside a quoted local part,
+  whose quote then lands in what would be the sender or the recipient - and
+  an address with a quote, a bracket or a space in it is not kept. A line
+  whose reply still holds `; from=<` - two tails - makes no row. (Until
+  `2.14.4` the reply ended at the *first* `; from=<`, which a stranger's
+  recipient could supply - security audit 2026-10-17, M2.)
   Only a recipient at an inbound domain makes a row. The journal is cut to
   10 000 rows every 500 rows the tailer writes, and by age every six hours.
 - **Filter lists** reach another container as files: values are whitelisted on

internal/logtail/inbound.go

@@ -101,15 +101,24 @@ type Inbound struct {
 // addresses), and a pattern that could start anywhere would let him write
 // journal rows.
 //
-// For the same reason the reply is the shortest text before "; from=<", and
+// The reply is the client's text too: a refusal quotes the refused
-// the HELO name - the last thing on the line but the SASL fields above, which
+// recipient, and Postfix quotes it in internal form - quotes removed, "; < >
-// Postfix logs as the client wrote it, spaces and brackets included - is
+// =" kept (extract_addr(), tok822_internalize(); smtpd_check_reject() applies
-// whatever is left. Taking the longest reply instead would let a HELO name
+// only printable()) - so a stranger's RCPT TO:<"; from=<x> to=<y> ..."@d>
-// that itself spells
+// spells a whole "; from=<…> to=<…> proto=…" inside the reply. The shortest
-// "; from=<…> to=<…> proto=…" supply the sender, the recipient and the reply
+// reply ended there, and the row's sender, recipient and domain were the
-// of the row (security review, 2026-10-12).
+// stranger's (security audit 2026-10-17, M2, proven on the test server).
+// What the client cannot spell is the end of the line: the HELO name is
+// logged with a space and ` <>()\";@` turned into "?" (src/smtpd/smtpd.c,
+// helo_cmd(), NEUTER_CHARACTERS), and an address that holds "; from=<" or a
+// ">" needs a quoted local part, whose quote then stands inside what the
+// pattern would take for the sender or the recipient - which the filter
+// below refuses. So the reply is the longest text before a "; from=<…>
+// to=<…> proto=…" that runs to the end of the line, the HELO name holds none
+// of the neutered characters, and each SASL field (never written on port 25)
+// is one word. A line it does not fit leaves no row.
 var (
-	inboundRejectRe = regexp.MustCompile(`^\S+ postfix/(inbound/smtpd|cleanup)\[\d+\]: ([0-9A-Za-z]+): (reject|milter-reject|milter-discard): \S+ from [^\[\s]*\[([0-9A-Fa-f:.]+)\]: (.*?); from=<([^>]*)> to=<([^>]*)> proto=\S+(?: helo=<.*>)?(?: sasl_(?:method|username|sender)=.*)?$`)
+	inboundRejectRe = regexp.MustCompile(`^\S+ postfix/(inbound/smtpd|cleanup)\[\d+\]: ([0-9A-Za-z]+): (reject|milter-reject|milter-discard): \S+ from [^\[\s]*\[([0-9A-Fa-f:.]+)\]: (.*); from=<([^>]*)> to=<([^>]*)> proto=\S+(?: helo=<[^ <>()\\";@]*>)?(?: sasl_method=\S*)?(?: sasl_username=\S*)?(?: sasl_sender=\S*)?$`)
 	inboundClientRe = regexp.MustCompile(`^\S+ postfix/inbound/smtpd\[\d+\]: ([0-9A-Za-z]+): client=`)
 	// A queue id stops being port 25's when the message is queued (qmgr takes
 	// it: the filter let it through), when cleanup removes the queue file of
@@ -211,6 +220,13 @@ func (j *inboundJournal) handle(line string, now time.Time, catchingUp bool) {
 	if strings.ContainsAny(from, "\"<> ") || strings.ContainsAny(to, "\"<> ") {
 		return
 	}
+	// A reply that still holds "; from=<" is a line with two tails, one of
+	// them written by the client - a refused recipient quoted in the reply
+	// (audit 2026-10-17, M2) or a HELO name Postfix failed to neuter. Which
+	// is Postfix's is then a guess the journal does not make.
+	if strings.Contains(reason, "; from=<") {
+		return
+	}
 	if program == "cleanup" {
 		if _, ok := j.port25[queueID]; !ok {
 			return // a submission-port message, or one opened before this run

internal/logtail/inbound_test.go

@@ -256,12 +256,14 @@ func TestInboundHoldIsTheMiltersRow(t *testing.T) {
 	}
 }
 
-// The HELO name is the last thing on the line and is logged as the client
+// The HELO name is the last thing on the line. Postfix logs it with a space
-// wrote it. It cannot supply the row: the reply ends at the first "; from=<",
+// and ` <>()\";@` turned into "?" (src/smtpd/smtpd.c, helo_cmd()), so one
-// which is Postfix's own.
+// that tried to spell another refusal arrives neutered and the row is
+// Postfix's own; a HELO name with those characters left in is no line
+// Postfix writes, and makes no row.
 func TestInboundHeloCannotWriteTheRow(t *testing.T) {
 	j, st := inboundFixture()
-	j.handle(head+"postfix/inbound/smtpd[41]: NOQUEUE: reject: RCPT from x[203.0.113.9]: 550 5.1.1 <nobody@in.example.com>: Recipient address rejected: User unknown in relay recipient table; from=<real@example.net> to=<nobody@in.example.com> proto=ESMTP helo=<x 5.7.1 Refused by this server's deny list; from=<victim@bank.example> to=<ceo@in.example.com> proto=ESMTP helo=<y>", tailNow, false)
+	j.handle(head+"postfix/inbound/smtpd[41]: NOQUEUE: reject: RCPT from x[203.0.113.9]: 550 5.1.1 <nobody@in.example.com>: Recipient address rejected: User unknown in relay recipient table; from=<real@example.net> to=<nobody@in.example.com> proto=ESMTP helo=<x?5.7.1?Refused?by?this?server's?deny?list??from=?victim?bank.example??to=?ceo?in.example.com??proto=ESMTP?helo=?y?>", tailNow, false)
 	if len(st.rows) != 1 {
 		t.Fatalf("rows = %+v", st.rows)
 	}
@@ -271,23 +273,56 @@ func TestInboundHeloCannotWriteTheRow(t *testing.T) {
 		t.Errorf("row = %+v", got)
 	}
 
-	// A HELO name with a closing bracket in it does not lose the line either.
+	// Un-neutered, which Postfix never writes: a line with two tails makes
+	// no row, and above all not the one the HELO name spells.
+	for _, line := range []string{
+		head + "postfix/inbound/smtpd[41]: NOQUEUE: reject: RCPT from x[203.0.113.9]: 550 5.1.1 <nobody@in.example.com>: Recipient address rejected: User unknown in relay recipient table; from=<real@example.net> to=<nobody@in.example.com> proto=ESMTP helo=<x 5.7.1 Refused by this server's deny list; from=<victim@bank.example> to=<ceo@in.example.com> proto=ESMTP helo=<y>",
+		head + "postfix/inbound/smtpd[41]: NOQUEUE: reject: RCPT from x[203.0.113.9]: 550 5.1.1 <nobody@in.example.com>: no; from=<real@example.net> to=<nobody@in.example.com> proto=ESMTP helo=<a>b>",
+	} {
 		j, st = inboundFixture()
-	j.handle(head+"postfix/inbound/smtpd[41]: NOQUEUE: reject: RCPT from x[203.0.113.9]: 550 5.1.1 <nobody@in.example.com>: no; from=<real@example.net> to=<nobody@in.example.com> proto=ESMTP helo=<a>b>", tailNow, false)
+		j.handle(line, tailNow, false)
-	if len(st.rows) != 1 || st.rows[0].From != "real@example.net" {
+		if len(st.rows) != 0 {
-		t.Errorf("rows = %+v", st.rows)
+			t.Errorf("a line with an un-neutered HELO name wrote %+v", st.rows)
+		}
 	}
 
 	// The same on a verdict line: the filter's reply is not extended by the
 	// HELO name, so a stranger cannot make a refusal look like the deny list's.
 	j, st = inboundFixture()
 	j.handle(head+"postfix/inbound/smtpd[41]: Q1: client=x[203.0.113.9]", tailNow, false)
-	j.handle(head+"postfix/cleanup[52]: Q1: milter-reject: END-OF-MESSAGE from x[203.0.113.9]: 5.7.1 Spam message rejected; from=<real@example.net> to=<b@in.example.com> proto=ESMTP helo=<x this server's deny list; from=<spoofed@example.org> to=<b@in.example.com> proto=ESMTP helo=<y>", tailNow, false)
+	j.handle(head+"postfix/cleanup[52]: Q1: milter-reject: END-OF-MESSAGE from x[203.0.113.9]: 5.7.1 Spam message rejected; from=<real@example.net> to=<b@in.example.com> proto=ESMTP helo=<x?this?server's?deny?list??from=?spoofed?example.org??to=?b?in.example.com??proto=ESMTP?helo=?y?>", tailNow, false)
 	if len(st.rows) != 1 || st.rows[0].Detail != "5.7.1 Spam message rejected" || st.rows[0].From != "real@example.net" {
 		t.Errorf("rows = %+v", st.rows)
 	}
 }
 
+// The refused recipient is quoted in the reply in internal form - quotes
+// removed, "; < > =" kept - so it can spell a whole "; from=<…> to=<…>
+// proto=…" there. The line below is the shape mail.log on the test server
+// held on 2026-10-17 for such a RCPT TO sent from outside (the audit's probe
+// used a domain that is not inbound; here it is one). It writes no row in
+// the stranger's words (security audit 2026-10-17, M2).
+func TestInboundRecipientCannotWriteTheRow(t *testing.T) {
+	forged := `; from=<ceo@example.net> to=<board@in.example.com> proto=ESMTP helo=<x>; `
+	for name, line := range map[string]string{
+		"recipient unknown": head + `postfix/inbound/smtpd[41]: NOQUEUE: reject: RCPT from x[203.0.113.9]: 550 5.1.1 <` + forged + `@in.example.com>: Recipient address rejected: User unknown in relay recipient table; from=<probe@example.net> to=<"` + forged + `"@in.example.com> proto=ESMTP helo=<audit.example>`,
+		"forged sender also in MAIL FROM": head + `postfix/inbound/smtpd[41]: NOQUEUE: reject: RCPT from x[203.0.113.9]: 550 5.1.1 <` + forged + `@in.example.com>: Recipient address rejected: User unknown in relay recipient table; from=<"x; from=<ceo@example.net"@a.example> to=<"` + forged + `"@in.example.com> proto=ESMTP helo=<audit.example>`,
+	} {
+		j, st := inboundFixture()
+		j.handle(line, tailNow, false)
+		if len(st.rows) != 0 {
+			t.Errorf("%s: a line with two tails wrote %+v", name, st.rows)
+		}
+	}
+
+	// A plain refusal is still read whole, its reply up to Postfix's tail.
+	j, st := inboundFixture()
+	j.handle(unknownRcp, tailNow, false)
+	if len(st.rows) != 1 || st.rows[0].From != "a@example.net" || st.rows[0].To != "nobody@in.example.com" {
+		t.Errorf("rows = %+v", st.rows)
+	}
+}
+
 // A queue id is port 25's only until the message is queued, cleanup removes
 // it unqueued, or the id is opened again by another listener - Postfix reuses
 // ids.