SIGSEGV in C2 Node::unique_ctrl_out on 11.0.4

[Vitaly Davidovich]

On Wed, Sep 18, 2019 at 4:58 PM Aleksey Shipilev <shade at redhat.com> wrote:

> On 9/18/19 10:53 PM, Vitaly Davidovich wrote:
> > Thanks.  This happens sporadically on our build farm, so it's likely
> going to be very difficult to
> > install enough fastdebug builds in there (nevermind the impact to the
> farm).  I guess if nobody has
> > an inkling, we'll need to see how to proceed.
>
> Download the fastdebug build, take one machine out of the pool (or even
> try locally), and then
> "while true; ant build; done" for a day, waiting for hs_errs/replays to
> fall out? That's how many
> intermittent issues are cornered.

Yeah, the essence is simple enough :) I’ve failed to repro locally after
many attempts, and playing around with JIT settings to try triggering JIT
compilations sooner.  This particular codebase gets built a sufficient # of
times, but the segfault has only been seen 3 times.

Anyway, thanks.  It may be I need to spend more time on this if nobody has
seen something like this.

>
>
> --
> Thanks,
> -Aleksey
>
> --
Sent from my phone