Java4Shell: RCE in OpenJDK when processing Base64

(calif.io)

539 points | by werrett 5 hours ago ago

106 comments

  • lucasluitjes 3 hours ago ago

    From the section on performing heap grooming over the network ("Remote Heap Feng Shui"):

    > The current attack is sensitive to network conditions, like RTT and jitter. Low, stable round-trip times are preferred for a more reliable attack. In our controlled environment, we performed this attack over the internet through public IP addresses instead of just assuming perfect laboratory conditions. For the current attack path to achieve optimal stability and reliability in exploitation, Calif turned to physics and utilized money to solve the problem, rather than experimenting with TCP/IP stack configuration. In other words, Calif rented another VM in the same region as our target.

    And the "Limitations" section:

    > There is no universal payload. Each chain needs a decoder route, a live writer, a returned reader or another feedback path, a heap layout, and an application action. Every target needs a bespoke profile. The exploit is not a universal point-and-shoot.

    Based on the complexity and constraints, short-term mass-exploitation seems unlikely. It will take some time before there are reliable exploits, and they may require running on infra close to the target.

    But a long tail of unpatched java applications/services at high-value targets getting exploited, seems plausible to me.

    • tru3_power 22 minutes ago ago

      Decoder route pretty low bar given basic auth right? I think the higher bar is the hardware requirements for exploitation

    • luka2233 3 hours ago ago

      still pretty scary for multi-tenant infra

      • hinkley an hour ago ago

        Yes but don’t those usually have nginx or friends in front of them? Which apparently frustrates the block size of data being transmitted.

    • afavour 3 hours ago ago

      Yes, I guess the TL;DR is the chances of a drive by attack are low, but a sufficiently motivated actor with a specific target in mind might be able to exploit it.

      So, patch your systems, governments!

      • hinkley an hour ago ago

        I’m sure they’re all working late on a Tuesday doing just that. /s

    • lowbloodsugar 2 hours ago ago

      >Based on the complexity and constraints, short-term mass-exploitation seems unlikely.

      Don't see how this is a safe assumption in the modern era of LLMs.

      • afavour 2 hours ago ago

        > The current attack is sensitive to network conditions, like RTT and jitter.

        LLMs aren’t magic, they can’t alter physics.

        • makeitdouble 2 hours ago ago

          With only a few major cloud providers, running a VM in physical proximity isn't a herculean task though.

          In particular booting a VM is cheap, having them in all potential regions won't break the bank.

        • Xirdus 2 hours ago ago

          Something like 20% of all internet is colocated in us-east-1.

    • Uptrenda an hour ago ago

      Are any of these new meme exploits actually exploitable? Do you guys remember when we had human exploits that actually worked...

    • jonstewart 3 hours ago ago

      > Based on the complexity and constraints, short-term mass-exploitation seems unlikely. It will take some time before there are reliable exploits

      Mass exploitation, maybe. But LLMs are pretty good at finding exploits now as Calif showed at unprompted a few weeks ago.

      • AdrenalinMd 3 hours ago ago

        Indeed, it can be a SKILL.md that writes the exploit. So in the end it could still be quite mass exploitable, only depends how much context is needed to write the exploit.

  • ievans 5 hours ago ago

    Super impressive work by Calif as usual.

    The exploit conditions are very specific but there are still going to be a lot of hosts impacted. We at Semgrep published some free detection scripts here: https://github.com/semgrep/detection-scripts/tree/main/CVE-2...

    They look for these five conditions:

      * HotSpot VM: any OpenJDK build, Oracle JDK, or GraalVM JDK. Only HotSpot generates the stub. OpenJ9 and GCJ are OK. Azul Prime (Zing), which replaces C2 with its own JIT, and any VM name not recognized as HotSpot are [?], never [!].
      * x86-64 JVM on an x86-64 host.
      * Version 17.0.6 or later, not at or above a version listed in FIXED_IN, and no newer than the newest releases checked (see below). Newer versions are [?]: they may carry the fix.
      * The JVM reports avx512_vbmi and avx512bw, and UseAVX >= 3.
      * Base64 intrinsics are on and C2/JVMCI can compile. These turn it off: -XX:-UseBASE64Intrinsics, -XX:DisableIntrinsic=_base64_decodeBlock, -Xint, -XX:TieredStopAtLevel below 4.
    • bikezen 4 hours ago ago

      The write-up itself matches your version assumption, however it also says:

      > According to Oracle, the following Oracle Java SE versions are vulnerable: 8u511b33-perf, 17.0.21.0.3, 21.0.13.0.3, 25.0.5.0.3, 27.0.1.0.3. We thank Oracle for providing this information ahead of publication time.

      So according to Oracle at least, JRE8 versions are affected as well.

    • testtestmetest 2 hours ago ago

      Calif is an OpenAI funded shop..

      • thephyber an hour ago ago

        Oracle owns Java and Oracle is "OpenAI funded" in the sense that Oracle's data center buildouts are being done using OpenAI as a consumer.

        The circular financing continues!

        But seriously, the funding here doesn't matter any more than Project Zero's funding comes from Google. Big companies are trying to improve security for the common software substrates of our industries / governments.

      • milkshakes an hour ago ago

        and?

    • cjbprime 4 hours ago ago

      Wouldn't it be prudent for someone to patch even if they aren't running their code on an AVX512 system right this moment? They could be running on one.. next moment.

      • Banditoz 3 hours ago ago

        Sure, but I think it brings it down from "patch immediately" severity, to just "bump the JDK with next service deploy" severity.

        Or, do running machines often have their CPU changed from under them without much notice?

        • stackskipton 3 hours ago ago

          Depends on the company and deployment method. As Ops person, I generally don't think much about rolling out different Kubernetes Nodes and what SKU I select. So yes, you could be running on one node type today and tomorrow I upgrade the nodes under you.

          • xmodem 3 hours ago ago

            We’re using a second-tier cloud provider’s managed Kubernetes offering, and in investigating our level of exposure to this, I just learned that there is variance within a node pool on whether AVX-512 is available.

        • Sohcahtoa82 3 hours ago ago

          > Or, do running machines often have their CPU changed from under them without much notice?

          I dunno about other cloud providers, but I know AWS can move your EC2 VM from one system to another transparently. There is a server ID you can get from the metadata, but unless you're checking it frequently, you can be moved without ever being aware.

          • PhilipRoman 2 hours ago ago

            Question is if they expose it to the guest. Many (most?) cloud platforms will make the guest see the lowest common denominator, specifically to avoid angry customers when their working setup suddenly breaks.

            In the case of transparent migration, I think it's basically a must. No JIT (or even libc probably) is going to support CPU features changing at runtime.

          • johnbellone an hour ago ago

            You’ll need to restart the guest.

    • bonsai_spool 4 hours ago ago

      This looks like Claudese - may be kind to write such things yourself if you intend for humans to read them. (For example, “ HotSpot VM: any OpenJDK build, Oracle JDK, or GraalVM JDK. Only HotSpot generates the stub. OpenJ9 and GCJ are OK. Azul Prime (Zing), which replaces C2 with its own JIT, and any VM name not recognized as HotSpot are [?], never [!].” — I tried parsing at least five times before realizing you’re asking me to decipher LLM talk)

      • Yokolos 3 hours ago ago

        I still don't understand what's meant with Azul. Is it affected or not? What is it never?

        • kaboberg 2 hours ago ago

          Azul Core definitely is - it uses upstream HotSpot.

          Azul Platform Prime (formerly Zing) may or may not be depending on whether they touched the base64 intrinsic. Since it is closed-source there is no way to verify via code, which is how our research team investigates ecosystem events like this.

  • leca 5 hours ago ago

    What’s that old Chinese proverb? “May you live in interesting times”? It feels like we’re living in interesting times. How long will it take for a bad actor to seriously compromise critical infrastructure? My bet is within six months.

    The upside is that it’ll force legacy institutions and organizations with sub-par security to get good at security and modernize their legacy stacks. That, or get pwned in short order.

    • testtestmetest 2 hours ago ago

      That's what they want you to expect. Either way, the worst possible people on our side will take advantage of it, and possibly cause it for that opportunity.

    • makeitdouble 2 hours ago ago

      We've already seen an attack on transfusion blood storage.

      I wouldn't minimize the impact of the ones on supermaket freezers either. It was technically dumb and somewhat laughable, but we're talking food. The affected chains had stores in very rural areas with elder people living in remote places, these supermakets would be the only purveyor of long conservation food.

      I'd argue we're already there, albeit at small scale yet.

    • thephyber an hour ago ago

      > How long will it take for a bad actor to seriously compromise critical infrastructure?

      That happened when Colonial Pipeline was hacked in 2017. It shut down a large percentage of gas station gas purchases on the US East Coast.

      There were lots of severe vulns since then: shellshock, log4j, the PHP vuln from a few weeks ago.

    • ctoth 4 hours ago ago

      > My bet is within six months

      My bet is it already happened two months ago and there's a whole secret scrum we only vaguely know about happening right now.

      • adrianpike 4 hours ago ago

        Instability is often correlated to urgently performed change, and we've seen a lot of instability around the world lately.

        • pixl97 4 hours ago ago

          A lot of organizations we sitting on internal bugs for years because they thought they were impossible to exploit, and now their red team can break them in hours.

          Most businesses are stuck in a trap now. They know the software is exploitable, but they are organizationally slow in testing and rollout. They have to start moving fast with security and IT organizations that are cut to the bone, yet at the same time contain to much box checking deadweight.

          • jeremyjh 3 hours ago ago

            3/4 of the Fortune 500 will conclude they should hire expensive consultants to create more checkboxes.

        • ChuckSneed 4 hours ago ago

          Instability can also be attributed to AI slop code. That's the simplest explanation.

          • thephyber an hour ago ago

            I don't think AI slop code is the proximal cause for most of this. It is, however, the most common meme for blame.

            My suspicion is (1) company execs are being told to expect company extinction if they don't force faster velocity and (2) we are increasingly building more and more complex systems, so neither AI nor humans can find all of the issues before they cause outages.

      • Analemma_ 4 hours ago ago

        I think this bet is optimistically assuming a far more competent world than the one we actually have. Have you seen the idiot clowns actually in power? The agency that might've been tasked with this hypothetical secret scrum, CISA, has been gutted and left to rot because they thought it was an enemy plot to censor them on social media or whatever. There's no secret room where the smart people are fixing things.

        • scottyah 3 hours ago ago

          There are many secret rooms where many smart people are fixing many things. It's more telling about yourself that you don't know of any.

          • jeremyjh 3 hours ago ago

            There is no secret room where smart people people aligned with the interests of the people of the United States are fixing things. There may not be any smart people left in the Federal government who aren't actively engaged in looting it.

          • salamandars 2 hours ago ago

            Tell us about one of them, then.

            • antonvs 32 minutes ago ago

              He could tell you, but then he'd have to... lie.

    • flockonus 27 minutes ago ago

      In the face of massive hacks happening this week and last - i'd give 4 days.

      ps: wouldn't even be surprised if some of the banks hacks were related to this (branch) of exploit, given how much banks seem to rely on dated Java

    • loloquwowndueo an hour ago ago

      That’s the thing - it’s not a proverb, it’s a curse my guy

  • fsnovask 2 hours ago ago

    >The best hacker today is neither AI nor human. It is the centaur hacker, part human, part AI.

    I think only really smart people are going to be left in Security and Software. I can barely fathom what this group is doing with AI right now (ignoring that their model probably isn't nerfed and mine would be) and I don't think any of this is trainable for the average developer. This is starting to feel like some of the scientific fields where careers start when you get a MS/PhD. Or like the average developer is on the same level of not being able to convert a Word document to PDF.

    • eru 2 hours ago ago

      > I think only really smart people are going to be left in Security and Software.

      I'm looking forward to eventually getting there. At the moment, there's so much low hanging fruit that even someone without a clue like me can just point a fuzzer at some random Linux kernel code and still discover problems.

      (Most of the problems I found from that exercise are patched by now, if you use a recent enough kernel.)

      I used Kimi to drive the fuzzer. The model didn't have to do anything smart, just do the tedious work of writing the glue code to make the fuzzing work. AFL++ did the 'smart' parts, and I had the idea to kick it off and oversaw the patches.

    • colonCapitalDee 2 hours ago ago

      I don't know, it feels like just another skill to me. You learn your debugger, you learn the shell, you learn web development... you learn AI.

      We'll see how long the centaur era lasts though. Hopefully a very long time

      • eru 2 hours ago ago

        For comparison for chess, it didn't last that long.

  • drum55 4 hours ago ago

    They aren’t wrong that we should be deleting more support and removing more features. The idea that the linux kernel by default builds support for all sorts of ancient systems nobody has ever heard of is absurd.

    • Gigachad 2 hours ago ago

      The kernel already dynamically loads kernel modules based on hardware it's running on so most of the junk isn't loaded. But for the stuff you can't dynamically load, it's a tradeoff between stripping the kernel down and supporting the most hardware.

      Maybe distros could provide two builds, one for modern hardware only and one for maximum compatibility.

    • eru 2 hours ago ago

      You can use the kernel config to remove these things.

    • AuthAuth an hour ago ago

      Its time to come home to Gentoo

  • ozgrakkurt an hour ago ago

    Not even fuzzing these things are criminal in 2026.

    Is it difficult to just go through Java, nodejs etc. Base64/hex/.. implementations and fuzz them all?

    Does anyone know why this would be missed?

    • knapsack-dev 31 minutes ago ago

      Maybe the fuzzer wasn't running on the CPUs that were affected

      • jooops1 16 minutes ago ago

        Then you have coverage metrics showing that.

  • mzs 5 hours ago ago

    You can also mitigate Java4Shell by starting OpenJDK with either:

    -XX:-UseBASE64Intrinsics to disable hardware-accelerated Base64 encoding and decoding, or

    -XX:UseAVX=2 to disable AVX-512. Apply this flag only on x86-64 hosts. OpenJDK on ARM doesn't recognize it and will fail to start.

    • marginalia_nu 5 hours ago ago

      Might be worth examining whether AVX-512 is even supported on the machine before freaking out, too.

  • exabrial 9 minutes ago ago

    Eclipse Temurin has not released a patch yet....

  • Noe2097 4 hours ago ago

    This attack surface (manually optimized assembly code in the middle of software stacks) looks AWESOME

  • george_max 2 hours ago ago

    This only fires on x86 servers whose CPUs have AVX-512 VBMI. ARM boxes like Graviton and older Xeons never run the buggy code.

    Could this be what triggered the recent outages?

    https://x.com/InfiShark/status/2122786295105913332

  • rbanffy 4 hours ago ago

    Not everywhere... Requires an x86-64 processir with AVX512.

    • formerly_proven 4 hours ago ago

      Pretty much any Intel server and most AMD servers are going to support AVX512 and conversely most VMs running in the cloud will have access to it.

      • rbanffy 4 hours ago ago

        Speak for yourself. All my cloud VMs are ARM, with one exception, one s390x running Linux.

  • bitlad an hour ago ago

    It seems like ARM64 AArch64 instances are not affected.

  • EGreg an hour ago ago

    For anyone who cares, I have a solution "for now":

    https://github.com/Qbix/proxy/blob/main/java4shell.md

    When PHP2shell was announced, before the patches were available, I wrote a proxy to sanitize input. You should have an input-sanitizing proxy in front of your app servers anyway instead of waiting for patches for as-yet-undiscovered CVEs.

    PHP < 8 doesn't have patches coming. It's either the proxy or be pwned. Some our honeypot machines running PHP 5 got infiltrated with full chained kernel-level exploits. The ones with the proxy didn't. Which one do you want to be?

    https://github.com/Qbix/proxy -- now updated to help mitigate java4shell

    Having said that, my proxy approach wouldn't fully protect ALL Java apps in this case, because:

    A Java program that opens a TLS connection to a malicious server can be compromised, even if it exposes no service of its own.

    The problem is that you can't intercept and "heal" TLS connections with a proxy, unless you also spoof the TLS certificates. However, since most Java apps don't pin certs, a workaround is using /etc/hosts and a CA cert added in cacerts. Then the proxy approach works even on outbound traffic. And you don't have to wait for the patches.

      1. Add maliciousserver.com -> 127.0.0.1 (or proxy IP) in /etc/hosts
      2. Java app resolves the hostname, connects to your proxy on 443
      3. Proxy presents a cert for maliciousserver.com signed by your local CA
      4. JVM checks: right hostname, trusted CA? handshake succeeds
      5. Proxy connects to the real server (using its own resolver that bypasses /etc/hosts, or the real IP directly)
      6. Normalizes Base64 in the response (assuming Calif relies on a wacky base64)
      7. Relays to Java client (safe)
    
    so what you'd do is:

      1. Generate proxy CA cert
      2. Install in JVM truststore (keytool)
      3. /etc/hosts entries for known hostnames
      4. forward_proxy config block (port, ca_cert, ca_key)
      5. You have three options for routing: 
        HTTPS_PROXY env, 
        JVM flags (-Dhttps.proxyHost), 
        or iptables REDIRECT (covers code that ignores proxy vars)
      6. Start qbixproxy
      7. Verify with curl
    
    The usual disclaimers... lock down the executable, removing any ability to read or write files, only pass traffic through like a proxy. Investigate the repo yourself, with your own Claude or ChatGPT etc. The repo even has links to pre-fill the prompts to do just that.

    I'm putting this out under MIT license to help people, but under this license it disclaims all warranties, including implied warranty for merchantibility etc.

  • spidermonkey23 4 hours ago ago

    The string of A's reminds me of the ssh exploit recently.

  • secondcoming 4 hours ago ago

    Why would anyone write the scalar part of a SIMD'd function in assembly code!?

    • magackame 4 hours ago ago

      pure love for the game

  • dboreham 4 hours ago ago

    "vibe-uncoding" -- have actually been doing quite a bit of that in my older projects.

  • ethagnawl 4 minutes ago ago

    That Vibe Uncoding section didn't go where I thought it was going. Instead of suggesting people stop vibe coding and potentially introducing more/new vulns, they suggest vibing the removal of features and hoping the model or the tests it generated or whatever catch the edge cases. It really does feel like we're circling the drain here, folks.

    Also, (S)FTP is dead? Since when?

  • mushufasa 4 hours ago ago

    where is the dancing girl in the exploit videos???

    • 3dedb728-3f77 4 hours ago ago

      Yes, this is pretty much required now after the last.

      • cpach 3 hours ago ago

        I’m probably boring, but IMHO a plain video is easier to digest (:

  • titzer 3 hours ago ago

    Encode binary data in text because some dumb thing can't process binary. Text back to binary slow, so better make use of ridiculously wide vector width thing to speed up dumb decoding step we should have never become reliant on in hot paths. Derp, memory safety bug. Derp, die because performance "optimization".

    Our incredibly high tower of Jenga blocks is falling down under its own weight.

    So many promotions because of optimizing cosmically unnecessary work, though.

    Derp, derp, this is humans making computers.

    As a person who does these kinds of optimizations, it hits in the feels.

  • bullen 2 hours ago ago

    This is the code I use since 20 years back:

    http://move.rupy.se/file/base64.txt

    If this part becomes the bottleneck; so I need to write a GPU accelerated version I'll eat my own hat.

    • layer8 an hour ago ago

      Your decoder implementation accepts all sorts of non-Base64 strings and decodes them to whatever. Among other things, it also accepts MIME Base64 (with line breaks and whitespace) and base64url strings, but decodes them to the wrong values. This is a classic example of insufficient input validation that can lead to parser mismatch vulnerabilities or worse.

    • CAP_NET_ADMIN 2 hours ago ago

      Vectorized base64 can reach speeds in the proximity of memory copy.

      I've easily reached a point where it was worth it to go AVX route in my "for funsies" game that I'm making.

      • CAP_NET_ADMIN 2 hours ago ago

        Just checked, the scalar version reaches around 1-2GB/s when decoding, optimized it gets to 4GB/s, AVX-512 VBMI version reaches 40-55GB/s.

        • bullen 2 hours ago ago

          Interesting, what game are you making?

          • CAP_NET_ADMIN 2 hours ago ago

            Large scale tower/base defense with time rewind mechanics and user-generated structure templates.

            • bullen 22 minutes ago ago

              So where do you need 40GB/s of base64?

  • DarkmSparks 4 hours ago ago

    me using java for so long I have my own util.Base64 class used everywhere in pure java.

    now panicking in case some LLM imported this without me noticing.

    edit: it did. :(

  • LastTrain 3 hours ago ago

    Seems like security through obscurity might be a thing after all.

  • cpach 4 hours ago ago

    .

    • ryanschaefer 4 hours ago ago

      This should be the main thread as this is the primary source

      • cpach 3 hours ago ago

        Fair!

  • brightbed an hour ago ago

    > To speed up Base64 decoding…

    But. Just. Why. Maybe just KISS.

  • jeffbee 4 hours ago ago

    If I owned an entire language platform with a JIT, it would not contain assembly in its standard library, it would simply be improved until it generated something at least that fast from high-level source code on that target CPU.

    This is something I really like about LLVM. Their libc isn't spilling over with hand-written assembly. It's written in straightforward C++ under the belief that the compiler does, or will do, a good job.

  • spullara 4 hours ago ago

    please stop using native code that hasn't been proven

  • 9029 4 hours ago ago

    So who leaked it?

  • Sigmazero60 5 hours ago ago

    We set UseAVX=2 on all our jdk hosts last year for unrelated thermal issues, turns out that saved us here

  • aaronvg 3 hours ago ago

    Makes me wonder whether we would have the same vulnerability if an LLM had written this decoder. It's tricky to write safe hardware accelerated code.

  • eyalitki 5 hours ago ago

    The fact that mainstream modules have such trivial bugs in logic that can be fully covered by testing is staggering. It is about time software developers will have proper test coverage for their code, at least for basic "unit" modules. Maybe we can even name a testing philosophy to match it.

    • Noe2097 4 hours ago ago

      Not sure this is as trivial as it seems. The bug triggers on a specific instruction set, on a precise input length, with a specific character at the end, and when on the hot path, and requires low level memory access checking to be detected.

      • spockz 3 hours ago ago

        For the finished jre yes. But one would hope these kind of checks are also run on a hand rolled decoder function. Fuzzing that one and finding logic bugs is a lot easier

    • ajasmin 4 hours ago ago

      Finding these kind of bugs would have required fuzzing and some kind of bound checking canaries. Not sure all bound checkers work on the hand-optimized assembly.

      Or as it turns out an LLM! Not clear whether their agent spawned any classic tools...

  • nikanj 4 hours ago ago

    Just as with CVE-2026-98432, Debian is immediately patched at embargo drop, Ubuntu still isn't patched.

    Why do we even pay for Ubuntu?

    • cpach 4 hours ago ago

      Are you sure? I can’t see any Debian advisory for this vuln.

      https://www.debian.org/security/

    • mitchweaver 4 hours ago ago

      Uhh, Debian is NOT patched as of this writing.

      Fortunately I run name-brand Oracle(R) Java(tm), which did publish updated packages.