SIGSEGV caused by compiled Java code

[Volker Simonis]

Hi Reto,

Thanks a lot for the reproducer!

This seems to be a C1 issue and it looks like it was fixed in JDK 21
(i.e. I could easily reproduce it with JDK 17 & 20 but not with 21).
If we're lucky it was a simple fix which can easily be downported.

Best regards,
Volker


On Mon, Oct 16, 2023 at 2:57 PM Reto Merz <reto.merz at abacus.ch> wrote:
>
> Hello,
>
> Thanks Douglas Simon, David Griffiths and Jesper Wilhelmsson for your time, feedback and suggestions.
>
> I was able to create a simple reproducer [1] which cause the crash dump [2]
> very often after a few seconds. Sometimes 2-3 attempts are needed.
>
> The problem is still only reproducible with Linux OS, not Windows.
> I suspect that the problem is related to the used ReentrantLock.
>
> As I expected, the problem is not related to Adoptium/Temurin build.
> It is also reproducible with latest Oracle 17 LTS build ([3]).
>
> So I have reported it via bugreport.java.com (internal review ID: 9076126)
> I think "adoptium-support" ([4]) would not be the right place for that sort of bug report.
>
> If I receive a public JDK issues link, I will post it here.
>
> Regards
>
> Thanks
> Reto
>
> [1]
> https://github.com/retomerz/pasetbin/blob/main/Reproducer.java
>
> [2]
> https://github.com/retomerz/pasetbin/blob/main/hs_err_pid113346.log
>
> [3]
> IMPLEMENTOR="Oracle Corporation"
> JAVA_RUNTIME_VERSION="17.0.8+9-LTS-211"
> JAVA_VERSION="17.0.8"
> JAVA_VERSION_DATE="2023-07-18"
> LIBC="gnu"
> MODULES="java.base [...]"
> OS_ARCH="x86_64"
> OS_NAME="Linux"
> SOURCE=".:git:3495b01a06c1 open:git:66f469c5cf06"
>
> [4]
> https://github.com/adoptium/adoptium-support/issues
>
>
> On 15 Oct 2023, at 00:41, Jesper Wilhelmsson <mailto:jesper.wilhelmsson at oracle.com> wrote:
>
> Hi Reto,
>
> If you think you have found a bug then please do report it through bugreport.java.com. Provide as much relevant info as you possibly can. A reproducer is much appreciated. You can find more information about what info to provide here: https://openjdk.org/guide/#filing-an-issue
>
> Thanks,
> /Jesper
>
>
> On 14 Oct 2023, at 19:12, Reto Merz <mailto:reto.merz at abacus.ch> wrote:
>
> Hello,
>
> We can always reproduce a JVM crash SIGSEGV caused by compiled Java code.
> At the moment we can it only reproduce on Linux system (Ubuntu 22.04.3 LTS), and not on Windows.
>
> First we have seen it with version
>    OpenJDK Runtime Environment Temurin-17.0.4.1+1 (17.0.4.1+1) (build 17.0.4.1+1)
>
> After that we updatet to the latest 17 LTS Temurin release
>    OpenJDK Runtime Environment Temurin-17.0.8.1+1 (17.0.8.1+1) (build 17.0.8.1+1)
>
> but we can still reproduce it.
>
> Here is the content of the full hs_err_pid*.log:
>    https://github.com/retomerz/pasetbin/blob/main/hs_err_pid215731.log
>
> The Java code of the mentioned method (ch.abacus.abareport.Logger.logOldImpl(Level,String,Throwable))
> is below [1].
>
> Is this a known problem or should we report this via bugreport.java.com?
> In case we should report it, what exact info should be provided?
>
> Thanks
>
> Regards
> Reto
>
> [1]
> private static void logOldImpl(Level level, String message, Throwable throwable){
>    if (LOG_TO_SYSTEM_OUT){
>        System.out.println(/*NlsIgnore*/"LOG: " + level + " -> " + message);
>    } else {
>        try {
>            if (ULCParams.hasULCParams()){
>                message = ((AVSystem)((WeakReference)ULCParams.currentULCParams().properties().get(AVSystem.PARAM_STRING)).get()).getReport().getFile().getURL().toString() + " - " + message;
>            }
>        } catch (Exception e){
>            message = /*NlsIgnore*/"no system - " + message;
>        }
>        getLogger().log(level, message, throwable);
>    }
>
>    if (throwable != null){
>        throwable.printStackTrace();
>    }
> }
>