Re: [RFC PATCH] proc: do not include shmem and driver pages in /proc/meminfo::Cached

2016-02-28 Thread Konstantin Khlebnikov
On Mon, Feb 29, 2016 at 3:03 AM, Hugh Dickins wrote: > On Fri, 19 Feb 2016, Andrew Morton wrote: >> On Fri, 19 Feb 2016 09:40:45 +0300 Konstantin Khlebnikov >> wrote: >> >> > >> What are your thoughts on this? >> > > >> > > My thoughts are NAK. A misleading stat is not so bad as a >> > > mislea

Re: [RFC PATCH] proc: do not include shmem and driver pages in /proc/meminfo::Cached

2016-02-28 Thread Hugh Dickins
On Fri, 19 Feb 2016, Andrew Morton wrote: > On Fri, 19 Feb 2016 09:40:45 +0300 Konstantin Khlebnikov > wrote: > > > >> What are your thoughts on this? > > > > > > My thoughts are NAK. A misleading stat is not so bad as a > > > misleading stat whose meaning we change in some random kernel. > > >

Re: [RFC PATCH] proc: do not include shmem and driver pages in /proc/meminfo::Cached

2016-02-28 Thread Hugh Dickins
On Fri, 19 Feb 2016, Konstantin Khlebnikov wrote: > On Fri, Feb 19, 2016 at 1:57 AM, Hugh Dickins wrote: > > On Thu, 18 Feb 2016, Johannes Weiner wrote: > > > >> Even before we added MemAvailable, users knew that page cache is > >> easily convertible to free memory on pressure, and estimated their

Re: [RFC PATCH] proc: do not include shmem and driver pages in /proc/meminfo::Cached

2016-02-19 Thread Andrew Morton
On Fri, 19 Feb 2016 09:40:45 +0300 Konstantin Khlebnikov wrote: > >> What are your thoughts on this? > > > > My thoughts are NAK. A misleading stat is not so bad as a > > misleading stat whose meaning we change in some random kernel. > > > > By all means improve Documentation/filesystems/proc.t

Re: [RFC PATCH] proc: do not include shmem and driver pages in /proc/meminfo::Cached

2016-02-18 Thread Konstantin Khlebnikov
On Fri, Feb 19, 2016 at 9:40 AM, Konstantin Khlebnikov wrote: > On Fri, Feb 19, 2016 at 1:57 AM, Hugh Dickins wrote: >> On Thu, 18 Feb 2016, Johannes Weiner wrote: >> >>> Even before we added MemAvailable, users knew that page cache is >>> easily convertible to free memory on pressure, and estima

Re: [RFC PATCH] proc: do not include shmem and driver pages in /proc/meminfo::Cached

2016-02-18 Thread Konstantin Khlebnikov
On Fri, Feb 19, 2016 at 1:57 AM, Hugh Dickins wrote: > On Thu, 18 Feb 2016, Johannes Weiner wrote: > >> Even before we added MemAvailable, users knew that page cache is >> easily convertible to free memory on pressure, and estimated their >> "available" memory by looking at the sum of MemFree, Cac

Re: [RFC PATCH] proc: do not include shmem and driver pages in /proc/meminfo::Cached

2016-02-18 Thread Hugh Dickins
On Thu, 18 Feb 2016, Johannes Weiner wrote: > Even before we added MemAvailable, users knew that page cache is > easily convertible to free memory on pressure, and estimated their > "available" memory by looking at the sum of MemFree, Cached, Buffers. > However, "Cached" is calculated using NR_FIL

Re: [RFC PATCH] proc: do not include shmem and driver pages in /proc/meminfo::Cached

2016-02-18 Thread Rik van Riel
On Thu, 2016-02-18 at 15:36 -0500, Johannes Weiner wrote: >  > The semantics of Cached including shmem and kernel pages have been > this way forever, dictated by the single-LRU implementation rather > They may have been that way forever, but they have also been confusing to users forever, so ...

[RFC PATCH] proc: do not include shmem and driver pages in /proc/meminfo::Cached

2016-02-18 Thread Johannes Weiner
Even before we added MemAvailable, users knew that page cache is easily convertible to free memory on pressure, and estimated their "available" memory by looking at the sum of MemFree, Cached, Buffers. However, "Cached" is calculated using NR_FILE_PAGES, which includes shmem and random driver pages