[aarch64-port-dev ] FW: [Mach5] mach5-one-phh-JDK-8151779-20200427-2151-10554367: FAILED

Thomas Stüfe thomas.stuefe at gmail.com
Tue Apr 28 16:29:39 UTC 2020


On a related note, I wonder whether it would be possible to fake a larger
page size than the system uses. Well, I don't wonder since we do this on
AIX, for complicated reasons.

It requires a bit of work. So, not only spoofing os::vm_page_size(), but
also fixing the places where the native page size shines thru, e.g. making
sure os::reserve_memory() returns always os::vm_page_size() aligned memory.
I wonder whether it would be worth the work, that way one could simulate a
larger page size with a switch and test for errors like this.

..Thomas


On Tue, Apr 28, 2020 at 12:50 PM Andrew Haley <aph at redhat.com> wrote:

> On 4/28/20 11:11 AM, Nick Gasson wrote:
> >>
> >> My nightly buildbot started failing recently on aarch64.
> >> One issue is the error message of [cds] prevents configure from bootjdk
> determining.
> >>
> https://hg.openjdk.java.net/jdk/jdk/file/1b8f9e72b22b/make/autoconf/boot-jdk.m4#l77
> >>
> >> ./bin/java --version
> >> [0.006s][error][cds] Unable to map CDS archive --
> os::vm_allocation_granularity() expected: 65536 actual: 4096
> >> openjdk 14.0.1 2020-04-14
> >> OpenJDK Runtime Environment AdoptOpenJDK (build 14.0.1+7)
> >> OpenJDK 64-Bit Server VM AdoptOpenJDK (build 14.0.1+7, mixed mode)
> >>
> >> Are you aware of this issue? I can’t see [cds] line on my linux/x86_64
> host.
> >> Maybe it’s aarch64-only.  CC aarch-port-dev.
> >
> > Was your boot JDK built on a machine configured with a different page
> > size to your current machine? Looks like the CDS archive was dumped on a
> > machine with 64k pages but you're running with 4k pages. There's a JBS
> > issue for this:
> >
> > https://bugs.openjdk.java.net/browse/JDK-8236847
>
> The thread seems to have died here:
>
>
> http://mail.openjdk.java.net/pipermail/hotspot-runtime-dev/2020-February/038207.html
>
> We really need to get this fixed. If anyone reading this has machines with
> both 4k and 64k pages, please do the experiment and we'll make a suitable
> patch. Everything here has 64k pages.
>
> --
> Andrew Haley  (he/him)
> Java Platform Lead Engineer
> Red Hat UK Ltd. <https://www.redhat.com>
> https://keybase.io/andrewhaley
> EAC8 43EB D3EF DB98 CC77 2FAD A5CD 6035 332F A671
>
>


More information about the hotspot-dev mailing list