>> Could the same algorithm used for ZFS deduplication be used for memory >> page deduplication? That may be an alternative to page out under >> memory shortage. > >We have memory page deduplication -- it's called copy-on-write, which we >use for multiply referenced private anonymous pages. Doing "dedup" on >arbitrary pages of memory would require removing all writable >translations to those pages before checking them, and the check would be >invalid as soon as the page was written to.
And clearly we don't have a checksum for each page and that is what dedup uses. I think it would possibly require N^2 page comparisons. Hugely expensive. Casper _______________________________________________ on-discuss mailing list on-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/on-discuss
