I'm looking for mm code/heuristics/flags where the following occurs: This (anonymous) page was: - previously swapped to a swap device - then later read back in from the swap device Now memory pressure has resulted in a need to reclaim memory so: - this page is discovered to still be clean, i.e. it matches the page still on the swap device, so - the pageframe is thus an obvious candidate for reclaim I'd be grateful for any pointers/education... For example, is such a page always in the swapcache? Is it also in the page cache? Is it always INactive since it was read but never written? What flags are set/unset? What function or code snippet identifies such a page and does this code need to be protected by the swaplock or pagelock or ??? (Sorry if any of these are stupid questions...) Purpose: I'm looking into zcache (and future KVM/memcg tmem backend) changes to exploit a "writethrough" and/or "lazy writeback" cacheing model for pages put into zcache via frontswap, as discussed with Andrea and one or two others at LSF12/MM. Either model provides more flexibility for zcache to more effectively manage persistent pages. Thanks! Dan -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@xxxxxxxxx. For more info on Linux MM, see: http://www.linux-mm.org/ . Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/ Don't email: <a href