Aarch64 and Cortex A53 - broken since 21.0.7

[Duft Markus]

Hey,

Thanks for the quick answer. Unfortunately, I'm getting the very same behavior with 25+23 (which is the linked build). So I assume it might be something else not yet fixed. I would have loved to be able to build the aarch64 JDK locally, however that seems to be virtually impossible for me (I do have some experience, but I fail to figure out the build of the binutils in the devkit makefiles...

There are only 3 potential aarch64 hotspot commits I would like to look into, however only one of them seems really impactful enough..:


  *
990859cc327: [REDO] Implement C2 VectorizedHashCode on AArch64
  *
21b76f3f07a: AArch64: VM crashes with SIGILL when prctl is disallowed

The first one being the more dramatic one IMHO 🙂 It is super hard to pinpoint my fingers on something, because the behavior is so erratic that not even the problems stay the same from run to run. One of my applications even gets a NoClassDefFound on the main class, as it seems it fails to evaluate the classpath properly (or find classes in jars, ...). If I rearrange the classpath it finds the class and fails differently.

Open for any ideas and/or advice and also willing to try out some - BUT my actual A53 is a real anti-beast. There is no way I could build something directly on that board - it is simply too weak.

Best, Markus



________________________________
From: porters-dev <porters-dev-retn at openjdk.org> on behalf of Andrew Haley <aph-open at littlepinkcloud.com>
Sent: Monday, 26 May 2025 10:25
To: porters-dev at openjdk.org <porters-dev at openjdk.org>
Subject: Re: Aarch64 and Cortex A53 - broken since 21.0.7

On 5/26/25 06:18, Duft Markus wrote:
> Good Morning,
>
> As documented in much detail in the following ticket, 21.0.7 broke Aarch64 support on the Cortex A53 CPU.
>
> https://github.com/adoptium/adoptium-support/issues/1295
>
> There is a reproducer in the Ticket, basically the steps are:

If this is `8353237: [AArch64] Incorrect result of VectorizedHashCode
intrinsic on Cortex-A53`, it's fixed in JDK head. Download it
[here](https://adoptium.net/en-GB/temurin/nightly/?version=25)

--
Andrew Haley  (he/him)
Java Platform Lead Engineer
Red Hat UK Ltd. <https://www.redhat.com>
https://eu01.z.antigena.com/l/0sXdfbZmwaKD_wl11rlRJ40KC4NgfkkdkTFF4Bu04M4B6D0grJ1SK_SuD0iXihVB6JGwXPxxREE~dRxdvw3dZ4b6QPaq8NAOa-g9_sUPxtVLNeqGTtQNAR-XOdFL~Rnj5WYLlEPf9K_QVLORNLLti1JCI6CUvDQCK_GF0dr4N~c33-VSMQ~iC9SgC7ba
EAC8 43EB D3EF DB98 CC77 2FAD A5CD 6035 332F A671


SSI Schäfer Software Development GmbH | Friesachstraße 15 | 8114 Friesach | Austria
Registered Office: Friesach | Commercial Register: 617462h | VAT no. ATU80145424
Commercial Court: Landesgericht für Zivilrechtssachen Graz
Managing Directors: Manfred Muckenhumer, Barbara Schlosser

Unsere Hinweise zum Umgang mit personenbezogenen Daten finden Sie hier<https://www.ssi-schaefer.com/de-at/datenschutz>.
You can find our information on the handling of personal data here<https://www.ssi-schaefer.com/en-de/privacy>.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.org/pipermail/porters-dev/attachments/20250526/2a473f1e/attachment-0001.htm>