[11u] RFR 8250928: JFR: Improve hash algorithm for stack traces

[Jaroslav Bachorík]

On Tue, Sep 1, 2020 at 12:03 PM Aleksey Shipilev <shade at redhat.com> wrote:
>
> On 9/1/20 10:52 AM, Jaroslav Bachorík wrote:
> > I would like to request a backport of this fix improving the hashing
> > function and reducing the chance of conflict in storing JFR stacktrace
> > data.
> >
> > JIRA: https://bugs.openjdk.java.net/browse/JDK-8250928
> > Webrev: http://cr.openjdk.java.net/~jbachorik/8250928/11u/
>
> I have the same concern here as for 8u backport [1].
>
> There are two places where the hash code is computed. While 11u backport matches what mainline does,
> and this is acceptable, this raises a question if mainline really needs to have those two hash codes
> computed the same way. And if so, does backporting the only one of the computations break anything?

Hm, this makes me wonder why the hash computation was not updated in
both places in the mainline. With what we have now there the hash for
the same stack trace object would be different depending on whether it
is calculated in 'record_thread' or 'record_safe' method. This does
not sound right to me.
Perhaps, the mainline should be fixed to use the 'improved'
calculation in both places first?

-JB-

>
> --
> Thanks,
> -Aleksey
>
> [1] https://mail.openjdk.java.net/pipermail/jdk8u-dev/2020-September/012603.html
>