Re: [Jikesrvm-regression] [AIX] regression Failed 56 tests

[David P G. <gr...@us...>]

From the symptoms my guess is that I broke something in JNI yesterday and 
we are using a stale processor register when we come back from a native 
method (for example if a thread switches to a different processor when it 
comes back into RVM via a JNI call).  The logic to deal with this is 
platform-specific, so I think its only broken on PPC.

I'm looking into this now.

--dave