
Loading

Loading
We use strictly necessary cookies to run this site, and analytics cookies to understand how it's used. See our Privacy Policy for details.
A GitHub-reviewed security advisory, not necessarily under active exploitation the way our CISA KEV advisories are, but a real, disclosed vulnerability in an open-source dependency.
Affected packages
Vulnerable: = 1.6.7
Vulnerable: < 1.6.7-1
Patched in 1.6.7-1
Details
Confirmed at HEAD. server.go: cmdXCLIENT = []byte("XCLIENT") is 7 bytes (line 78), so the dispatch match cmdXCLIENT.match(cmd) only guarantees len(input) >= 7. The handler then slices input[8:] (line 586):
case sc.XClientOn && cmdXCLIENT.match(cmd):
if toks := bytes.Split(input[8:], []byte{' '}); len(toks) > 0 { // line 586
A client sending the bare command XCLIENT (exactly 7 bytes) hits input[8:7] and panics runtime error: slice bounds out of range [8:7]. handleClient runs in the per-connection goroutine (server.go:391) which has no recover(); the only recover() calls are in the backend mail-processing workers (a different goroutine never reached during command parsing), so the unrecovered panic crashes the whole SMTP daemon.
Connect to the SMTP port, then send XCLIENT\r\n (a bare LF also works; no HELO needed). Precondition: xclient_on: true (off by default, but a documented nginx-fronted deployment; go-guerrilla does not restrict who may send XCLIENT). Attacker model: any unauthenticated client on the SMTP port.
EXECUTED with a negative control (agent-run, driving the real server.handleClient with XClientOn=true over the suite's mock net.Conn). The bare XCLIENT panics with the exact OOB; the negative control XCLIENT with a trailing space (8 bytes) does not panic and returns 250. The upstream TestXClient tests XCLIENT ADDR=... and XCLIENT c but never the bare 7-byte form (the missing regression case). I independently re-confirmed at HEAD that cmdXCLIENT is 7 bytes (server.go:78), that the handler slices input[8:] (server.go:586), and that handleClient (server.go:391) has no recover. NOT verified: did not stand up a standalone daemon on a real socket to observe OS-level process exit (used the mock net.Conn plus the traced absence of recover; crash-to-exit is standard Go runtime behavior).
Guard the slice against the exact-verb-length case: require len(input) > len(cmdXCLIENT)+1 (or slice at the verb length and trim) before input[8:]. A recover in the per-connection handleClient loop would also contain any residual command-parse panic to the single connection.
The rfc5321 recursive-descent parser (MAIL/RCPT/HELO/EHLO/address) is bounds-safe (next()/peek() return 0 at EOF), and the line/size/recipient caps are enforced. XCLIENT is the only off-by-one; every other command arm slices exactly at its verb length.
Five-source enumeration (repo advisories, OSV, NVD, the global GitHub DB, no vendor CNA) shows the PROXY-header IP-spoof bug (CVE-2025-31135, fixed v1.6.7) is the only published advisory; this XCLIENT panic is undisclosed and not contested.
Dedup gate clean: no phires/go-guerrilla blocking verdict in COVERED-INDEX, DISCLOSURE-INDEX or SCOUT-EXCLUDE.
AI assistance was used while investigating. The panic is executed-verified with a negative control (a bare XCLIENT crashes the real handleClient while XCLIENT with a trailing space returns 250), and the reporter confirmed at HEAD that cmdXCLIENT is 7 bytes while the handler slices input[8:] on a per-connection goroutine with no recover.
EPSS, exploit probability
Low0.45%
estimated chance of real-world exploitation in the next 30 days, higher than 39.7% of every CVE FIRST.org scores
Refreshed 11/20/2026, via FIRST.org's EPSS model, not CVSS, this measures likelihood of exploitation, not how severe it would be.
More go advisories
Talk to us about this
Find out whether this dependency (or others like it) is actually reachable in your codebase.
Get help patching, or building dependency-scanning into your pipeline going forward.
A secure code review or supply chain review finds this before an advisory does.