On Tue, Mar 04, 2025 at 06:01:04PM +0100, Heinz Mauelshagen wrote: > > As Mikulas shared earlier: > > The Android people concluded that loop is too slow and rather than using > > loop they want to map a file using a table with dm-linear targets over the > > image of the host filesystem. So, they are already doing what dm-loop is > > doing. Which again is completely handwavy. What is the workload. What are the issues with the workload? Where is the time spent making it slower. e.g. the extent mapping code in this patch is a lot less efficient than the btree based extent lookup in XFS. If it is faster it is not looking up the extents but something else, and you need to figure out what it is.
