On 12/10/2015 05:58 AM, Alan Altmark wrote: > On Wednesday, 12/09/2015 at 08:15 GMT, Barton Robinson > <bar...@velocitysoftware.com> wrote: >> My request to get this fixed was rejected by ibm several years ago. The >> problem is that even though linux doesn't have anything on the vdisk, >> z/vm still has to back it. I asked for a diagnose as we do with real >> storage - a way for Linux to tell z/vm the page no longer needs >> backing. So after the swapoff, can you detach the vdisk? That's the >> only way with current technology to tell CP to free it up. > > The Linux community didn't like the memory state changes we asked for with > CMM, so I don't see why they would assent to the change you describe. A > disk is just another address space that would need the same information. > And there's no such thing as "I'm done" with swapping.
Newer Linux swap code does a discard (TRIM) on the full swap disk if you do a swapon or on a bunch of pages if a cluster of that swap disk becomes free. This depends on the parameters of swapon. This was probably added to improve SSD handling. So if diag 0x250 would gain a discard capability, something like that would become possible. Christian ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 ---------------------------------------------------------------------- For more information on Linux on System z, visit http://wiki.linuxvm.org/