Neil, Andrew, The following patches replace the stripe-queue patches currently in -mm. Following your suggestion, Neil, I gathered blktrace data on the number of reads generated by sequential write stimulus. It turns out that reduced pre-reading is not the cause of the performance increase, but rather increased write merging. The data, in patch #1, shows a 74% decrease in the number of dispatched writes. I can only assume that this is the explanation for the 65% throughput improvement, because the occurrence of reads actually increased with these patches applied. This take also fixes observed data corruption while running i/o to a synching array (it was wrong to move the flags parameter from r5dev to r5_queue_dev as things could get out of sync... reverted). Next step is to test reshape under this new queuing model. Regards, Dan - To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
