On Mon, 8 Jun 2009, Izik Eidus wrote:
Thanks for the fix,
(I saw it while i wrote the RFC patch for the madvise, but beacuse that i
thought that the RFC fix this (you can see the removel of the second call to
scan_get_next_index()), and we move to madvise, I thought that no patch is
needed for this code, guess I was wrong)
Ah, no, I hadn't noticed that, this patch is from several weeks ago,
before you even posted the madvise() version.
I think myself that we ought to fix the algorithm as it stands now in
mmotm, rather than hiding the fix in amongst later interface changes.
But it's not a big deal, so long as it gets fixed in the end.
By the time Andrew sends KVM to Linus, it shouldn't be the
patches currently in mmotm with more on top:
they should be
re-presented with all trace of /dev/ksm gone by then.
Hugh