Re:Pre-auth RCE flaw in Java 17 and later (CVE-2026-114938)
by yabones on Tuesday November 17, 2026 @04:09PM
(#50538221)
Some unverified sources have said it's in the Base64 decoder, but strangely only effecting AMD64 machines, so anything on ARM/graviton allegedly isn't impacted. Some sort of bug in some hyper-optimized code I reckon. That might mean any place that parses cookies, headers, urls, JWTs, SAML payloads, certificates, etc, might be exposed. Yikes.
But the facts aren't really out yet, we'll learn more in coming hours and days...
Re:Pre-auth RCE flaw in Java 17 and later (CVE-2026-114938)
by AgentME on Tuesday November 17, 2026 @04:08PM
(#50538205)
Unconfirmed, but people on Twitter suggest this vulnerability has to do with the AVX512 implementation of Base64 decoding, and so this vulnerability can be worked around by running Java with the -XX:UseAVX=2 flag. (https://x.com/sean_a_cassidy/status/2122791344573976817)
Re:Pre-auth RCE flaw in Java 17 and later (CVE-2026-114938)
by elric on Tuesday November 17, 2026 @04:28PM
(#50538444)
There are presumably more bugs lurking in the hand crafted assembly intrinsics in the JVM. The many eyeballs and shallow bugs thing is only true when the eyeballs actually understand the code.
Re:Pre-auth RCE flaw in Java 17 and later (CVE-2026-114938)
by philipkglass on Tuesday November 17, 2026 @05:24PM
(#50539121)
I assume that the advice is referring to situations where you might have a mixture of ARM and AMD64 hosts, like on AWS, so you need to be careful about global flag updates that might not work across all instances.
Re:Pre-auth RCE flaw in Java 17 and later (CVE-2026-114938)
by articulatepang on Tuesday November 17, 2026 @05:47PM
(#50539357)
I think the idea is that you’re already running the appropriate JVM on each architecture, but want to push a global command line flag update to every JVM you run, regardless of architecture, that prevents this exploit.
Until, presumably, there’s fixes and you can test, update etc.
Re:Pre-auth RCE flaw in Java 17 and later (CVE-2026-114938)
by marginalia_nu on Tuesday November 17, 2026 @07:06PM
(#50540084)
I don't know if I'd count paving over the reality of what you're running with an abstraction layer as something simpler. If anything it demonstrably makes addressing issues like these harder.
Re:Pre-auth RCE flaw in Java 17 and later (CVE-2026-114938)
by articulatepang on Tuesday November 17, 2026 @08:01PM
(#50540522)
I think you're right, but many people are running systems at large scale, with lots of hosts and all kinds of abstraction layers and stuff. Now this exploit shows up. What are they to do? It's just a practical thing they can do right now.
First we must stop the bleeding. After that, yes, it's worth figuring out if they should change how things are set up.
I've been in situations like this before. Sometimes the complications and abstractions were my fault, sometimes I inherited them. Either way, as responsible owners of the overall system or business, we've gotta be able to respond fast to issues.
Re:Pre-auth RCE flaw in Java 17 and later (CVE-2026-114938)
by desolate_muffin on Tuesday November 17, 2026 @11:52PM
(#50542027)
Whenever I see code like this, I am always struck by how there are two types of programmers—those like me, and those who can work on whatever the hell that is.
Re:Pre-auth RCE flaw in Java 17 and later (CVE-2026-114938)
by 12_throw_away on Tuesday November 17, 2026 @07:37PM
(#50540329)
I realize that speeding up Java's base64 probably had a measurable impact on a significant proportion of all webservers everywhere. But at the same time, I wonder if it could maybe be done without implementing the entire thing in assembly, by hand? Here's the comment describing how it works:
Re:Pre-auth RCE flaw in Java 17 and later (CVE-2026-114938)
by marginalia_nu on Tuesday November 17, 2026 @04:39PM
(#50538568)
It's only on specific architectures though it kinda falls apart. It would be impressive if the bug was so portable it let you RCE on sub-AVX512 machines and ARM and whatever too. Maybe with the new vector API we'll see such powerful vulnerabilities.
Re:Pre-auth RCE flaw in Java 17 and later (CVE-2026-114938)
by iLoveOncall on Tuesday November 17, 2026 @04:55PM
(#50538774)
What a horribly written and misleading article. They just make things up to make themselves look good, for example "Memory corruption is the most common class of vulnerability on the Internet". No the fuck it's not? SQL injections is BY FAR the most common class of vulnerability.
This is a much much narrower issue than PHP2Shell.
Re:Pre-auth RCE flaw in Java 17 and later (CVE-2026-114938)
by tptacek on Tuesday November 17, 2026 @04:55PM
(#50538788)
I like the idea that you're just going to sort of shout down Calif on a point like this. Have you looked at their roster lately?
If you any part of your identity is based on the idea that firms like this aren't finding insane vulnerabilities right now, I think after they find the RCE in OpenJDK's core Base64 processing you're going to have to consider sucking it up, taking the L, and adjusting your priors.
Re:Pre-auth RCE flaw in Java 17 and later (CVE-2026-114938)
by tptacek on Tuesday November 17, 2026 @05:49PM
(#50539373)
On a simple count like that you could probably beat SQLI on "Javascript prototype pollution", too. But those are 5 classes of vulnerabilities with wildly different impacts, and platform-level SQLI is actually pretty rare.
Re:Pre-auth RCE flaw in Java 17 and later (CVE-2026-114938)
by tptacek on Tuesday November 17, 2026 @06:59PM
(#50540019)
Not so much that I'm above criticism, I'm clearly not, so much as how funny it is to suggest "coworker of Thai Duong, Tavis Ormandy, Michal Zalewski, Dion Blazakis, and Stefan Esser" is a dunk. Shit, now I kind of do want to be a Calif.io employee.
Re:Pre-auth RCE flaw in Java 17 and later (CVE-2026-114938)
by ranger_danger on Tuesday November 17, 2026 @07:13PM
(#50540147)
I have seen their comments across this site for a very long time. Unfortunately I do not respect or put any stock in what they say, because of how unbelievably rude and dogmatic they are, and always have been.
"As a rule, strong feelings about issues do not emerge from deep understanding."
Re:Pre-auth RCE flaw in Java 17 and later (CVE-2026-114938)
by tptacek on Tuesday November 17, 2026 @07:52PM
(#50540445)
Well that hurts my feelings but also you're right: nobody should respect or put any stock into what I say because of my identity or status (whatever that might be).
Re:Pre-auth RCE flaw in Java 17 and later (CVE-2026-114938)
by cjbprime on Tuesday November 17, 2026 @06:32PM
(#50539741)
Is "the most common class of vulnerability" not supposed to be measured by a simple count? It's the blog post's own plain language and it doesn't say anything about impact.
Re:Pre-auth RCE flaw in Java 17 and later (CVE-2026-114938)
by deathanatos on Tuesday November 17, 2026 @06:42PM
(#50539842)
> Is "the most common class of vulnerability" not supposed to be measured by a simple count?
Trivially yes? It'd be a rate of some ilk at least. Upthread, to their credit, was a "score" of undefined nature.
But also, more upthread is talking about "memory corruption", which is going to include far more than just CWE-787. CWEs are a bit weird, and things I'd include in a loose colloquail definition like that would include stack based overflows, heap based overflows (which are for some reason different in CWE land), "classic buffer overflows" (CWE-120, again, their wording), out of bounds writes, OoB reads, null derefs, and use-after-free. All separate CWEs.
Re:Pre-auth RCE flaw in Java 17 and later (CVE-2026-114938)
by tptacek on Tuesday November 17, 2026 @06:46PM
(#50539885)
It's just a category error comparison. SQLI is a web pest vulnerability; memory corruption RCE is a vulnerability research vulnerability. They're different disciplines. Web SQLI vulnerabilities are usually application-code-specific (not just "application specific", but with a specific origin in a mistake made in application code). Big ticket vulnerability research publications are generally platform vulnerabilities; one specific coding mistake impacting thousands or millions of installs.
It's very weird to look at a memory corruption RCE and say SQLI is "more common". On typical web targets, memory corruption RCE is basically unheard of (and SQLI is not the most common sev:hi vulnerability; SSRF probably outranks it).
These are just vulnerability researchers talking like vulnerability researchers.
Re:Pre-auth RCE flaw in Java 17 and later (CVE-2026-114938)
by cjbprime on Tuesday November 17, 2026 @07:39PM
(#50540337)
> For example, one of many different ways of looking at this issue: perhaps by "...vulnerabilities on the Internet" they could have meant in reference to reported CVEs, [..]
The CWE count is by reference to reported CVEs. If they meant that, they were wrong.
Re:Pre-auth RCE flaw in Java 17 and later (CVE-2026-114938)
by thephyber on Tuesday November 17, 2026 @07:01PM
(#50540032)
I would argue the snapshot of 2025 CWE rankings isn't super helpful to resolve the argument here. It is an objective ranking, but it's missing a lot of context of how an OOB write is observed, reported and fixed.
If an OOB write happens, it is likely to be investigated and solved quickly because the most likely symptoms are going to be memory corruption or a crash. These are super high impact symptoms.
If an XSS or SQLi is present, there is not necessarily any symptom until some security researcher probes the code.
Also, what most people are missing here is that the article didn't need to restrict the memory corruption "to the internet". If you include all of the embedded devices that are not internet enabled and that run on C code, it's much more likely to be a true statement.
Also, the pedantry in this thread is boring. If people actually bothered to understand the intent of what was being written rather than the imperfect communication, this thread wouldn't need to exist.
Re:Pre-auth RCE flaw in Java 17 and later (CVE-2026-114938)
by TZubiri on Tuesday November 17, 2026 @04:30PM
(#50538472)
Looks unrelated, this is being disclosed now but it was discovered a couple of weeks ago and is just now being made public at the same time.
Also I haven't looked into it, but I'm willing to be it's some free after-use memory corruption thing that has a CVSS Attack Complexity of high, the 'RCE' is speculative, and private companies push social media headlines to fit the AI cybersecurity narrative.
Not saying the doom isn't right, but sometimes it's more like a Wednesday than doomsday. Just patch it and move on. Thanks to all involved though in disclosing and patching though.
Re:Pre-auth RCE flaw in Java 17 and later (CVE-2026-114938)
by yabones on Tuesday November 17, 2026 @04:40PM
(#50538586)
Before you keep your security team up all night, it looks like proxies save the day here. Might not be a complete fix, obviously a real patch should be applied as soon as possible.
> Request streaming. Some current grooming profiles need partial request bodies to reach the application. A proxy that waits for the complete body removes their release timing. nginx buffers request bodies by default.
Re:Pre-auth RCE flaw in Java 17 and later (CVE-2026-114938)
by hinkley on Tuesday November 17, 2026 @09:10PM
(#50540993)
The biggest thing I have worked on recently had TWO proxies, but the second was due to process per CPU spiciness and a JVM wouldn't benefit in the least from that. Which still leaves Conway's Law, segregating rules and policies between two teams who manage things on their own basis. Always annoying to punch new headers through on such systems though.
Re:Pre-auth RCE flaw in Java 17 and later (CVE-2026-114938)
by tptacek on Tuesday November 17, 2026 @05:34PM
(#50539246)
The Log4j comparison is always a little frustrating, because it sets the bar somewhere in the neighborhood of "apocalyptic". If you're someone who believes our response to serious vulnerabilities is inadequate, you can see how Log4Shell gives a lot of cover to companies.
It is wild to find a generalizable (if platform-specific) memory corruption vulnerability in Java core Base64 handling; the tornado sirens should clearly be going off. But yes, it's not as bad as Log4Shell.
Re:Pre-auth RCE flaw in Java 17 and later (CVE-2026-114938)
by iLoveOncall on Tuesday November 17, 2026 @04:53PM
(#50538757)
What a horribly written and misleading article. They just make things up to make themselves look good, for example "Memory corruption is the most common class of vulnerability on the Internet". No the fuck it's not? SQL injections is BY FAR the most common class of vulnerability.
https://github.com/corretto/corretto-25/blob/25.0.5.1.2/CHAN...
But the facts aren't really out yet, we'll learn more in coming hours and days...
`grep -ow avx512vbmi /proc/cpuinfo | sort -u`
`-XX:-UseBASE64Intrinsics -XX:UseAVX=2`
I recommend -XX:-UseBASE64Intrinsics instead.
Until, presumably, there’s fixes and you can test, update etc.
First we must stop the bleeding. After that, yes, it's worth figuring out if they should change how things are set up.
I've been in situations like this before. Sometimes the complications and abstractions were my fault, sometimes I inherited them. Either way, as responsible owners of the overall system or business, we've gotta be able to respond fast to issues.
If it’s Java, Node, etc the binary dependencies are platform specific and 100% if your code is not unless you try very hard.
https://github.com/elide-dev/madura
Just a compiler, but a runner isn't impossible.
I suspect it's not the only issue they're going to find there.
https://github.com/openjdk/jdk/blob/5177a70b385c16f26ede191f...
lol
This is a much much narrower issue than PHP2Shell.
If you any part of your identity is based on the idea that firms like this aren't finding insane vulnerabilities right now, I think after they find the RCE in OpenJDK's core Base64 processing you're going to have to consider sucking it up, taking the L, and adjusting your priors.
==
Top 10 CWEs (2025 Ranking)
Rank ID Weakness Name Score
1 CWE-79 Cross-site Scripting (XSS) 60.38
2 CWE-89 SQL Injection 28.72
3 CWE-352 Cross-Site Request Forgery (CSRF) 13.64
4 CWE-862 Missing Authorization 13.28
5 CWE-787 Out-of-bounds Write 12.68
He's pretty famous in cybersecurity and on HN as a poster.
More that you aren't just the average HN NPC and that your comments are usually high SNR and well informed.
"As a rule, strong feelings about issues do not emerge from deep understanding."
Trivially yes? It'd be a rate of some ilk at least. Upthread, to their credit, was a "score" of undefined nature.
But also, more upthread is talking about "memory corruption", which is going to include far more than just CWE-787. CWEs are a bit weird, and things I'd include in a loose colloquail definition like that would include stack based overflows, heap based overflows (which are for some reason different in CWE land), "classic buffer overflows" (CWE-120, again, their wording), out of bounds writes, OoB reads, null derefs, and use-after-free. All separate CWEs.
It's very weird to look at a memory corruption RCE and say SQLI is "more common". On typical web targets, memory corruption RCE is basically unheard of (and SQLI is not the most common sev:hi vulnerability; SSRF probably outranks it).
These are just vulnerability researchers talking like vulnerability researchers.
The CWE count is by reference to reported CVEs. If they meant that, they were wrong.
If an OOB write happens, it is likely to be investigated and solved quickly because the most likely symptoms are going to be memory corruption or a crash. These are super high impact symptoms.
If an XSS or SQLi is present, there is not necessarily any symptom until some security researcher probes the code.
Also, what most people are missing here is that the article didn't need to restrict the memory corruption "to the internet". If you include all of the embedded devices that are not internet enabled and that run on C code, it's much more likely to be a true statement.
Also, the pedantry in this thread is boring. If people actually bothered to understand the intent of what was being written rather than the imperfect communication, this thread wouldn't need to exist.
Yeah, aren't they the ones that jumped the gun on the PHP vuln and published their weird anime PoC video before the embargo expired?
No one's questioning their technical abilities.
Their communication skills are dogshit, however.
lol
Also I haven't looked into it, but I'm willing to be it's some free after-use memory corruption thing that has a CVSS Attack Complexity of high, the 'RCE' is speculative, and private companies push social media headlines to fit the AI cybersecurity narrative.
Not saying the doom isn't right, but sometimes it's more like a Wednesday than doomsday. Just patch it and move on. Thanks to all involved though in disclosing and patching though.
It's a heap out-of-bounds write in Java’s AVX-512 Base64 decoder.
> Request streaming. Some current grooming profiles need partial request bodies to reach the application. A proxy that waits for the complete body removes their release timing. nginx buffers request bodies by default.
https://calif.io/java4shell/#limitations
Genuine question. Everything I see is Intel all the time. I know that people at home switched to AMD, but they don't run enterprise java.
I also have been told that apparently EPYC was hyped for a while and some hosters give you that. But do corporations run the java applications there?
The name makes it sound like Log4J or the PHP thing, but from what I can tell right now, it kinda isn't all that relevant?
I never did it for any ECS stuff as that was more work to get working from top to bottom.
It is wild to find a generalizable (if platform-specific) memory corruption vulnerability in Java core Base64 handling; the tornado sirens should clearly be going off. But yes, it's not as bad as Log4Shell.