Panic in ffs_blkfree()

[Terry Lambert]

"Alan L. Cox" wrote:
> Please take a look at vm_object_page_clean().  If end == 0, it relies on
> object->size to determine the extent of the pages for cleaning.  In
> light of the above, particularly, the positive indices, we should be
> worried that msync is skipping pages.

If we are going to worry about that, we should worry about it
syncing more pages than you asked it to, as well (see comments
in msync() code).  Face it, msync() is broken.

-- Terry