RISC-V build currently could seldom crash at 0x0 when JFR is enabled. The reason is unknown because there is very little useful information in the hs_err_pid log: the pc is 0x0 and the ra is 0x0, too. Note that it is also reproducible after checking out to [the RISC-V initial load patch](https://github.com/openjdk/riscv-port/commit/7148c39832fe522bad91df77ef4b5c9371193c35) on the HiFive Unleashed board. Still addressing the issue yet.

Approach to reproduce the issue:
modify the line `@run main/othervm jdk.jfr.jcmd.TestJcmdStartStopDefault` in `test/jdk/jdk/jfr/jcmd/TestJcmdStartStopDefault.java` to

```
@run main/othervm jdk.jfr.jcmd.TestJcmdStartStopDefault
@run main/othervm jdk.jfr.jcmd.TestJcmdStartStopDefault
@run main/othervm jdk.jfr.jcmd.TestJcmdStartStopDefault
@run main/othervm jdk.jfr.jcmd.TestJcmdStartStopDefault
...
```
to violently jtreg this test in a do...while shell loop could help to reproduce this issue.

An interesting thing is that C910 could not reproduce this after a one-week while-true long run.