See design doc: https://docs.google.com/document/d/1zcfWSsgwTY8WYYlxd0G8QTR6_b2j38uMfjjXBOuMZnM/edit#

for more details.

Hm. Some rough prototyping and xnu spelunking seems to imply that task_info[TASK_VM_INFO].compressed + task_info[TASK_VM_INFO].internal should give the same results as VmSwap + RssAnon on Linux. I'm also observing that this the computed number appears to match [within 1%] [resident + swapped ] Anonymous memory + dirtied COW memory.

Confirmed that this does not count reusable memory, which is separated out into task_info[TASK_VM_INFO].reusable

Long analysis: 
https://docs.google.com/document/d/1vltgFPqylHqpxkyyCM9taVPWNOTJkzu_GjuqdEwYofM/edit#heading=h.pxyyb8lojcy

summary:
Use task_info(TASK_VM_INFO).phys_footprint, subtract out IOSurfaces and SharedMemory regions.

More details:
We also need to hook this into the new memory-uma infrastructure. e.g. Which component performs the post-processing? [subtracting out data from other MDPs]. Do we need to make a new MDP to provide phys_footprint?

We probably want to throw together a quick outline of how this will work.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/c874a28819750c3da98cbc8abc8ace690bfa9e1e

commit c874a28819750c3da98cbc8abc8ace690bfa9e1e
Author: erikchen <erikchen@chromium.org>
Date: Wed Apr 26 00:21:53 2017

Add a field platform_private_footprint_ to ProcessMemoryTotals.

The field holds platform-specific data that will be used by the Memory-Infra
service to compute the private memory footprint.

This CL populates the field on macOS, and leaves TODOs to populate the field for
other OSes.

BUG=707021

Review-Url: https://codereview.chromium.org/2838803003
Cr-Commit-Position: refs/heads/master@{#467177}

[modify] https://crrev.com/c874a28819750c3da98cbc8abc8ace690bfa9e1e/base/process/process_metrics.h
[modify] https://crrev.com/c874a28819750c3da98cbc8abc8ace690bfa9e1e/base/process/process_metrics_mac.cc
[modify] https://crrev.com/c874a28819750c3da98cbc8abc8ace690bfa9e1e/base/trace_event/process_memory_totals.cc
[modify] https://crrev.com/c874a28819750c3da98cbc8abc8ace690bfa9e1e/base/trace_event/process_memory_totals.h
[modify] https://crrev.com/c874a28819750c3da98cbc8abc8ace690bfa9e1e/chrome/browser/memory_details_mac.cc
[modify] https://crrev.com/c874a28819750c3da98cbc8abc8ace690bfa9e1e/components/tracing/common/process_metrics_memory_dump_provider.cc

Is this task related to  crbug.com/742042 ?

Yes. We actually need to subtract out resident, shared memory on each process.