On 05/14/2012 04:43 PM, mani wrote: > Dear Kim, > > I have a query here .. > > > My point is that it would be better for read to not preempt > write-for-page_reclaim. > And we can identify it by PG_reclaim. You can get the idea. > > I think If there is no page available then no read will proceed. > When read request comes it reclaim the pages (starts the write if > syncable pages ) and get back after reclaiming the pages. > Only then a read request will come to the MMC subsystem. > And i think the reclaim algorithm will reclaim some substantial amount > of pages at a time instead of a single page. > So if we get few pages during the reclamation so there will be no > problem in halting the another write ops for proceeding the reads ? > > Can we think of a scenario when we are reclaiming the pages and write > ops is going on where as a high priority read for the interrupt handler > is pending ? > > Please correct me if i am wrong. For example, System can have lots of order-0 pages but little order-big pages. In this case, for getting big contiguos memory, reclaimer should write out dirty pages while it can handle order-0 page read request. > > Thanks & Regards > Manish -- Kind regards, Minchan Kim -- 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=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>