To support the development of JEP 475 (JDK-8322295), a G1 configuration has been devised that stresses infrequent paths of G1's write barriers. This configuration [1] is equivalent to running the JVM with the following settings:
-XX:-G1UseAdaptiveIHOP -XX:InitiatingHeapOccupancyPercent=0 -XX:G1SATBBufferSize=2 -XX:G1HeapRegionSize=1048576 -XX:G1UpdateBufferSize=2 -XX:GCCardSizeInBytes=max(ObjectAlignmentInBytes, 128) -XX:G1RSetUpdatingPauseTimePercent=0.

While this configuration is useful to improve test coverage and find low-frequency bugs in the G1 barrier implementation, it causes multiple tests to fail because they are not designed to be run under such extreme conditions. These tests are:

compiler/whitebox/ForceNMethodSweepTest.java
gc/arguments/TestG1HeapRegionSize.java
gc/arguments/TestTargetSurvivorRatioFlag.java
gc/TestFullGCALot.java
java/awt/font/TextLayout/FontLayoutStressTest.java
runtime/cds/appcds/cacheObject/ArchivedIntegerCacheTest.java

Additionally, this configuration causes open/test/hotspot/jtreg/:hotspot_slow_compiler to take more than 4h on macosx-x64-debug when run with -XX:CompileThreshold=100.

This bug report is created primarily for failure tracking purposes, but these tests might be adapted or excluded when the G1 barrier stress configuration is used if this is found to be useful.

[1] https://github.com/robcasloz/jdk/blob/6fb36e509d86edb3651a73c3d69f535d1bd3f21d/src/hotspot/share/gc/g1/g1Arguments.cpp#L247-L278