Re: Why is Shmem included in Cached in /proc/meminfo?

[Vlastimil Babka]

On 8/30/21 16:41, Matthew Wilcox wrote:
> On Mon, Aug 30, 2021 at 07:34:38AM -0700, Randy Dunlap wrote:
>> [add linux-mm mailing list]
>> 
>> On 8/30/21 12:44 AM, Mikko Rantalainen wrote:
>> > It's not immediately obvious from fs/proc/meminfo.c function
>> > meminfo_proc_show() but the output of Cached: field seems to always
>> > include all of Shmem: field, too.
>> > 
>> > Is this intentional? Usually cache is something that can be discarded if
>> > needed but shared memory (e.g. used to contain files in tmpfs) cannot be
>> > discarded without a data-loss. As such, I'd argue that it shouldn't be
>> > included in the Cached: output.
> 
> That's a reasonable position to take.
> 
> Another point of view is that everything in tmpfs is part of the page
> cache and can be written out to swap, so keeping it as part of Cached
> is not misleading.

Yeah, but with that view, anonymous memory can be also written out to swap. But
it's non-intuitive that something called "Cached" will contain something that
(if not dirty) can't be just dropped. I've had to point this Shmem oddity out a
number of times to someone, so I would say that it would be much better if it
was not part of Cached.

However, if we change it now, we might create even larger confusion. People
looking at the output for the first time (and IIRC also the 'free' command uses
it) on a new kernel wouldn't be misled anymore. But people working with both old
and new kernels will now have to take in account that it changed at some
point... not good.

> I can see it both ways, and personally, I'd lean towards clarifying
> the documentation about how shmem is accounted rather than changing
> how the memory usage is reported.
>