Chuck Silvers <chuq@chuq.com> writes:

> yea, I was wondering if that would start to cause problems someday.

The reason ffs was ok is because it flushed that partial page by
itself.

> calling pgo_flush() with a non-page-aligned address should really be
> an error, since pgo_flush() needs to round differently depending on
> what it's doing.  for PGO_CLEANIT we want to include partial pages
> on the ends, for PGO_FREE we want to exclude them, and it doesn't
> much matter for PGO_DEACTIVATE.  the caller ought to do the appropriate
> rounding for the context so that pgo_flush() doesn't have to mess with it.

Hm, better to hunt them?

enami.