8330846: (jstack) Add stacks of mounted virtual threads

[Iñigo Mediavilla]

Hello,

While there's ongoing work on:

https://github.com/openjdk/jdk/pull/19482

to add the stack trace of mounted virtual threads to the `jcmd <pid>
Thread.print` command, I'm starting to think about how I could do to print
the stack trace for virtual threads from `jstack` but I'm not sure about
what would be the best approach, so I'd be happy to get people's thoughts
on the topic.

Based on what I've seen `jstack` seems to fetch a ThreadList of JavaThreads
and then relies on a bunch of Java code that uses attribute offsets to get
the values of certain attributes of the JavaThread in the VM. With those
attributes the java code can obtain the data that it requires about the
threads, their monitors and also to get the latest frame in the stack to be
able to iterate the stack. If I were to follow the same principle, I'd have
to replicate the code that was added to the VM to read the stack chunks of
virtual threads in Java, knowing that it looks like that the JVM code had
to be written for multiple CPU architectures, I'm a bit concern about how
complex following that approach would be.

I considered following the approach used by `ThreadDumper.java`, that
relies on `ThreadContainers` to get the virtual threads and then iterates
over `Thread.getStackTrace` for the virtual thread. But I don't think that
would work because I don't think the `ThreadContainers` can be accessed
from `Jstack`.

So I'm basically a bit blocked, and I think that I'd need some help from
people that have followed loom more closely and have a clearer view of
things and can probably provide me with hints on how to approach the
problem differently.

Thanks in advance for your help.

Best

Íñigo Mediavilla Saiz
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.org/pipermail/serviceability-dev/attachments/20240604/157a1d54/attachment.htm>