RFR (XS) 8215356: Disable x86_32 Shenandoah build to avoid hotspot/tier1 failures (original) (raw)

Magnus Ihse Bursie magnus.ihse.bursie at oracle.com
Thu Dec 13 15:36:13 UTC 2018

(Also, for build changes there are no hard and fast rules about reviews like in Hotspot. A single review is typically okay.)

/Magnus

13 dec. 2018 kl. 16:34 skrev Magnus Ihse Bursie <magnus.ihse.bursie at oracle.com>:

Yes. /Magnus

13 dec. 2018 kl. 16:02 skrev Aleksey Shipilev <shade at redhat.com>:

Thanks! I would like to consider it trivial. Build system devs, are you good with this? -Aleksey

On 12/13/18 3:30 PM, Roman Kennke wrote: ok by me. Thanks! Roman

Am 13.12.18 um 14:58 schrieb Aleksey Shipilev: Bug; https://bugs.openjdk.java.net/browse/JDK-8215356

There are lots of hotspot/tier1 tests on x8632 caused by Shenandoah. This mode is experimental, and we can disable it without prejudice to make tests clean. Once x8632 support is fixed, we can re-enable it back. Fix: diff -r 748cab25d902 -r f9d158363d15 make/autoconf/hotspot.m4 --- a/make/autoconf/hotspot.m4 Thu Dec 13 14:39:16 2018 +0100 +++ b/make/autoconf/hotspot.m4 Thu Dec 13 14:48:30 2018 +0100 @@ -327,7 +327,7 @@ # Only enable Shenandoah on supported arches ACMSGCHECKING([if shenandoah can be built]) - if test "x$OPENJDKTARGETCPUARCH" = "xx86" || test "x$OPENJDKTARGETCPU" = "xaarch64" ; then + if test "x$OPENJDKTARGETCPU" = "xx8664" || test "x$OPENJDKTARGETCPU" = "xaarch64" ; then ACMSGRESULT([yes]) else DISABLEDJVMFEATURES="$DISABLEDJVMFEATURES shenandoahgc" Testing: x8664 tier1gcshenandoah, x8632 hotspot/tier1. Thanks, -Aleksey

More information about the build-dev mailing list