This bug affects the following JTREG test:

    com/sun/jdi/MethodExitReturnValuesTest.java

This test started failing when I synced up with main/baseline
after last week's c2_baseline push:

    20050824132849.sgoldman.c2_merge

Below is the entry from my nightly testing analysis report:

New JDI_REGRESSION failures (from 2005.08.25)
* com/sun/jdi/MethodExitReturnValuesTest.java
        This test failed due to unexpected method return values on
        Win-AMD64 Server VM (machine mark1), Solaris AMD64 Server
        VM (machine vm-v20z-3) and Linux AMD64 Server VM (machine
        wowamd):

            failure: sin: expected 0.8084964038195901, got 5.28560558E-315
            failure: byte: expected 89, got 0
            failure: char: expected x, got \u0000
            failure: int: expected 1, got 1079194419
            failure: long: expected 9223372036854775807, got 1079194419
            failure: short: expected 8, got 13107
            failure: boolean: expected false, got true
            failure: Expected 44, but got 37

Steve G. has been investigating the failure and asked me
to file the bug on his behalf.