The assert:

 assert(i > 0, "we should never remove all regions");

in HeapRegionSeq::shrink_by() is incorrect. If we shrink the heap to only contain one region we hit this assert. The reason is that we first decrement the value of i and then check the assert just before we should exit the loop.

So i == 0 actually means that we have one region left.

Here's a reproducer, SysGC is only does System.gc():

$ java -XX:+UseG1GC -XX:G1HeapRegionSize=8m -cp ~/tests/misc-tests/bin SysGC

# To suppress the following error report, specify this argument
# after -XX: or in .hotspotrc: SuppressErrorAt=/heapRegionSeq.cpp:239
#
# A fatal error has been detected by the Java Runtime Environment:
#
# Internal Error (/Users/brutisso/repos/hs-gc-g1-mmap/src/share/vm/gc_implementation/g1/heapRegionSeq.cpp:239), pid=46207, tid=16643
# assert(i > 0) failed: we should never remove all regions
#
# JRE version: Java(TM) SE Runtime Environment (8.0-b79) (build 1.8.0-ea-b79)
# Java VM: Java HotSpot(TM) 64-Bit Server VM (25.0-b27-internal-jvmg mixed mode bsd-amd64 compressed oops)
# Failed to write core dump. Core dumps have been disabled. To enable core dumping, try "ulimit -c unlimited" before starting Java again
#
# An error report file with more information is saved as:
# /Users/brutisso/repos/hs-gc-g1-mmap/hs_err_pid46207.log
#
# If you would like to submit a bug report, please visit:
# http://bugreport.sun.com/bugreport/crash.jsp
#
Current thread is 16643
Dumping core ...
Abort trap: 6