On Thu, Apr 12, 2012 at 11:02 AM, Shane G <[email protected]> wrote:
> No, by lazy in this context I meant that freed memory (pages) are not > immediately moved to the free list. This even extends to task termination. > If memory pressure ramps up sufficiently, kswapd will get kicked to balance > out the trees. Could take a while - like forever. > > In addition to what Rob mentioned, there is de-dup and compressed swap cache > out in the wild already. How does the mug end user figure out what's what ?. Hmm... when I add things up from smaps, my number is higher than what "free" says. I see VMA's like 3fffd5d3000-3fffd5d7000 r--p 0017c000 5e:01 98592 /lib64/libc-2.11.3.so Size: 16 kB Rss: 8 kB Pss: 6 kB Shared_Clean: 4 kB Shared_Dirty: 0 kB Private_Clean: 4 kB Private_Dirty: 0 kB Referenced: 0 kB Anonymous: 8 kB AnonHugePages: 0 kB Swap: 8 kB KernelPageSize: 4 kB MMUPageSize: 4 kB Locked: 0 kB With 22 processes having this mapped, I would count it as 22 times 8 kB while it really is just 8 kB on swap? And how come part of this is private when it's read-only? Rob ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to [email protected] 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/
