On Fri, Oct 28, 2016 at 04:16:01PM +0000, Michael Callahan wrote: > Doesn't the block layer already do a reasonable job of merging adjacent > discards? It certainly didn't with my fast NVMe device. But even if it did we have all the information available here, so it's still cheaper to do it in the submitter. One thing we should try is to move it into __blkdev_issue_discard in the future, though. Note that I've also started to look into merging disjoint discard range into a single request for ranged trim/unmap/deallocate and that also seems to work nicely, but it will need some large tree-wide changes. -- To unsubscribe from this list: send the line "unsubscribe linux-xfs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
