Hello Andrew, On Tue, Jun 09, 2015 at 02:35:48PM -0700, Andrew Morton wrote: > On Tue, 9 Jun 2015 23:43:51 +0900 Minchan Kim <[email protected]> wrote: > > > For system uses swap heavily and has lots of shared anonymous page, > > it's very trouble to find swap set size per process because currently > > smaps doesn't report proportional set size of swap. > > It ends up that sum of the number of swap for all processes is greater > > than swap device size. > > > > This patch introduces SwapPss field on /proc/<pid>/smaps. > > > > We should be told quite a bit more about the value of this change, > please. Use cases, what problems it solves, etc. Enough to justify > adding new code to the kernel, enough to justify adding yet another > userspace interface which must be maintained for ever.
The goal is same with present pages's PSS. We want to know per-process workingset size for smart memory management by userland platform memory manager. Our products use swap(ex, zram) heavily to maximize memory efficiency. IOW, workingset includes rss + swap. However, without this feature, it's really hard to figure out how many each process consumes availabe memory(rss + zram-swap) if the system has lots of shared anonymous memory(e.g, android). > > > --- a/Documentation/filesystems/proc.txt > > +++ b/Documentation/filesystems/proc.txt > > > > ... > > > > @@ -441,7 +442,7 @@ indicates the amount of memory currently marked as > > referenced or accessed. > > a mapping associated with a file may contain anonymous pages: when > > MAP_PRIVATE > > and a page is modified, the file page is replaced by a private anonymous > > copy. > > "Swap" shows how much would-be-anonymous memory is also used, but out on > > -swap. > > +swap. "SwapPss" shows process' proportional swap share of this mapping. > > > > "VmFlags" field deserves a separate description. This member represents > > the kernel > > flags associated with the particular virtual memory area in two letter > > encoded > > Documentation/filesystems/proc.txt doesn't actually explain what > "proportional share" means. A patient reader will hopefully find the > comment over PSS_SHIFT in fs/proc/task_mmu.c, but that isn't very > user-friendly. Okay, I will try to add more comment about PSS in proc.txt on next spin if you are not against this feature. Thanks. > > -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
