It looks like there are still issues with the swap_burst_read
feature. I am definitely seeing corruption with it on (after
I removed the break statement and fixed the logic in one of the
other tests). I have to run exhaustive tests to determine
if the corruption goes away after turning it off first, and
if so I'll then be able to focus on why it is occuring with
the feature turned on.
I'm not sure what I am missing, it should be ok to read multiple
missing swap-backed pages for a VM object with the tests I'm making.
-Matt