Crash on macosx-aarch64 when merging Leyden premain with mainline

ioi.lam at oracle.com ioi.lam at oracle.com
Mon Sep 22 00:04:27 UTC 2025


On 9/21/25 3:53 AM, Andrew Haley wrote:
> On 19/09/2025 09:54, Andrew Haley wrote:
>> On 18/09/2025 23:28, Dean Long wrote:
>>> If it only happens on macos aarch64 then it could be related to
>>> os::current_thread_enable_wx() state.
>>
>> Yes. That is almost certainly the case.
>
> Assuming that this is a WX problem, I'm working on a PR which is a 
> rewrite of MacOS WX handling, and will fix this problem. I hope to 
> merge into mainline this week. My PR is nearly done You could ignore 
> this crash for a few days.
>
> https://github.com/openjdk/jdk/pull/26562
>
Hi Andrew, do you know why we are seeing the crash in leyden but not in 
mainline?

Unfortunately the crash happens when building the JDK on macos, so it 
will basically introduce major disruption in testing if we merge the 
current mainline into Leyden.

I wonder if we can identify the cause of the current issue (merge leyden 
with fdc11a1569248c9b671b66d547b4616aeb953ecf), and perform a local fix, 
before we wait for the above PR. Maybe we are just missing a WX 
transition in the leyden-specific code for stubs?

Even if the above PR goes into the mainline, it will be difficult to 
merge that to leyden in a single step, as there will be conflicts on 
both the compiler side and CDS side. It will be difficult for one person 
to resolve both sides of the conflicts.

Thanks

- Ioi



More information about the leyden-dev mailing list