BIND 9.20.30 fixes four High flaws, one may allow remote code execution
EditorialBy TrustList Editorial
ISC's October release closes a GSS-TSIG replay that corrupts memory and three resolver crashes. No 9.18 build is listed, so 9.18 users must move to 9.20.30.
- United States
- Cybersecurity
- Network Security
- Patch Management
- +2 more
About BIND 9.20.30 fixes four High flaws, one may allow remote code execution
BIND 9.20.30 fixes four High flaws, one may allow remote code execution
21 October 2026: ISC has released BIND 9.20.30 and 9.21.27 to fix four High severity vulnerabilities in named, the most serious of which, CVE-2026-94300 (CVSS 8.1), lets an attacker replay signed GSS-TSIG messages to corrupt memory and possibly run code. The other three let a remote attacker crash or exhaust a resolver. ISC knows of no active exploits and offers no workarounds.
Not yet independently verified. ISC's pages list no fixed release on the 9.18 branch; we have not found whether one is planned. We will update this when it can be confirmed, and remove this note.
Fixed versions
- BIND 9.20: 9.20.30
- BIND 9.21 (development): 9.21.27
- Supported Preview Edition: 9.20.30-S1
Every flaw also reaches back into the 9.18 line and earlier, but ISC lists no fixed 9.18 release, so servers still on 9.18 need to move to 9.20.30.
The GSS-TSIG flaw
GSS-TSIG is the Kerberos-based signing that Active Directory environments use for secure dynamic DNS updates, so it is most common where BIND serves zones alongside Windows domains. An attacker who has captured one GSS-TSIG signed UDP message can replay it while the key is still valid, and concurrent replays corrupt memory in named, crashing it. ISC adds: "By also sending crafted queries to a zone the attacker controls, the attacker may potentially achieve remote code execution." The flaw was found in ISC's own testing and independently reported by Dave Aitel of OpenAI. Affected: 9.11.0 to 9.18.50, 9.20.0 to 9.20.29 and 9.21.0 to 9.21.26.
Three ways to take a resolver down
| CVE | CVSS | What happens |
|---|---|---|
| CVE-2026-19630 | 7.5 | Certain inbound messages trigger unlimited extra processing, so many of them exhaust CPU and memory |
| CVE-2026-96888 | 7.5 | named can stop with an assertion failure when a redirect zone is used with qname-minimization strict |
| CVE-2026-97313 | 7.5 | With recursion allowed and two RPZ zones, one recursive-only yes and one recursive-only no, a recursive query followed by a non-recursive one crashes named |
The last two need specific configurations, so an administrator can check whether they apply by reading named.conf. The resource exhaustion flaw needs no special setup and affects resolvers broadly; ISC found it in its own testing.
ISC gave its support customers early notice on 14 October and published on 21 October. Distribution packages follow on each vendor's schedule, so servers on a long-term-support Linux release depend on that vendor's backport.
Sources
- ISC: CVE-2026-94300, Concurrent GSS-TSIG messages corrupt memory, 21 October 2026
- ISC: CVE-2026-19630, Amplified resource exhaustion in inbound message processing, 21 October 2026
- ISC: CVE-2026-96888, 21 October 2026
- ISC: CVE-2026-97313, 21 October 2026
- Canadian Centre for Cyber Security: ISC BIND security advisory (AV26-1072), 21 October 2026
Categories & features
- United States
- Cybersecurity
- Network Security
- Patch Management
- Vulnerability Management
- Linux
TrustList Weekly
The week in software and IT, in one email
The news that matters to buyers, new rankings and our own research. Every Thursday, free, and easy to leave.
More on TrustList
Everything here links back to the same verified catalogue. Pick your next stop.
- More United StatesThe ranking for this subject
- CompaniesAgencies, consultancies and IT service providers, ranked by verified reviews.
- ProductsSoftware and SaaS with pricing, features, integrations and alternatives.
- AwardsAnnual recognition decided by verified reviews and an independent jury.
- LaunchesNew products and releases, voted up by the community every day.
- AI ModelsBenchmark scores and community ratings for every major model.
- RequestsBuyers describe what they need; vendors respond directly.
- PeopleReviewers, authors and makers with public profiles.
- ComparePut up to four listings side by side before you shortlist.