When it comes to memory sampling on Windows, I believe the recent changes were added by stanisc@. Can you please take a look?

FWIW I've just tried to repro the issue on my work desktop and both
55.0.2883.87 m (64-bit) and 57.0.2949.0 canary (64-bit) shows the Memory
column just fine.

Yes, wez@ has already discussed this issue with me.
I couldn't repro this issue on any of my computers either.

Without a repro I am not sure how this may be actionable (short of reverting the change). But the previous implementation caused occasional OOM crashes so it isn't much better either. I searched chrome bugs for this issue to see if it was reported by other users and didn't find anything.

Looking at the TaskManagerImpl and the sampling code, I can't immediately
see how it's even possible for Private and Shared to show values but Memory
not; there's a working-set based calculation used under Windows, but that
would have to be producing exactly -1 for every process to cause them to
display the N/A string.

I think I saw something like this recently too. Perhaps it requires a longer running time.

There is also this issue on Windows: Issue 668718. It could be related. In particular, see comment #2.

Regarding my comment #3, I realized I discussed this issue with another person recently, not wez@, so perhaps this is more widespread than I thought. 

wez@, do you have a repro?

On Windows the content for Memory column is now produced by SharedSampler, which returns Private Working Set and a bunch of other system metrics for all processes in one system call.

I have been seeing this issue repro reliably for me on my Windows 10
laptop, yes; I'll check again when I'm home.

There was an OS update recently, so perhaps something changed in
NtQuerySystemProcessInformation to break the SharedSampler impl?

I am installing all recent updates on my Windows 10 test laptop. I check if this could be related to an update.

stanisc@ and I added some logging and used my laptop to repro this.

The SharedSampler assumes that all processes have at least one thread, whereas on my laptop the compressed memory process has no threads.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/8c4cfd4e82b4465305d60d3830bcbae50076f004

commit 8c4cfd4e82b4465305d60d3830bcbae50076f004
Author: wez <wez@chromium.org>
Date: Tue Dec 13 22:57:30 2016

Fix SharedSampler to cope correctly with zero-thread processes.

SharedSampler previously assumed that all processes have at least one
thread, which appears no longer to be the case for the Windows operating
system's "System and Compressed Memory" process.

BUG= 673175 

Review-Url: https://codereview.chromium.org/2566423002
Cr-Commit-Position: refs/heads/master@{#438322}

[modify] https://crrev.com/8c4cfd4e82b4465305d60d3830bcbae50076f004/chrome/browser/task_manager/sampling/shared_sampler_win.cc

[Automated comment] Request affecting a post-stable build (M55), manual review required.

Your change meets the bar and is auto-approved for M56 (branch: 2924)

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/962230afbdb5e107d5b5f65c94fd520783f25ab6

commit 962230afbdb5e107d5b5f65c94fd520783f25ab6
Author: Wez <wez@chromium.org>
Date: Thu Dec 15 23:41:55 2016

Fix SharedSampler to cope correctly with zero-thread processes.

SharedSampler previously assumed that all processes have at least one
thread, which appears no longer to be the case for the Windows operating
system's "System and Compressed Memory" process.

BUG= 673175 

Review-Url: https://codereview.chromium.org/2566423002
Cr-Commit-Position: refs/heads/master@{#438322}
(cherry picked from commit 8c4cfd4e82b4465305d60d3830bcbae50076f004)

Review-Url: https://codereview.chromium.org/2584743002 .
Cr-Commit-Position: refs/branch-heads/2924@{#520}
Cr-Branched-From: 3a87aecc31cd1ffe751dd72c04e5a96a1fc8108a-refs/heads/master@{#433059}

[modify] https://crrev.com/962230afbdb5e107d5b5f65c94fd520783f25ab6/chrome/browser/task_manager/sampling/shared_sampler_win.cc

Tested on Windows 10 using chrome M56 #56.0.2924.51 and M55 #55.0.2883.75 and followed steps in comment #0 , didn't find any difference.

Attached screenshots for reference.

@wez-- Could you please check the attached screenshots and let us know if the issue is fixed or if we have missed out anything in steps to verify the issue.

Thanks!

Deleted: 673175_M55.PNG 297 KB

	673175_M55.PNG 297 KB View Download

Deleted: 673175_M56.PNG 289 KB

	673175_M56.PNG 289 KB View Download

Re #16: To verify this you need a device on which one or more processes have zero threads active; we were only able to repro that on some laptop devices.

I will check Chrome Canary's behaviour on my laptop and update this bug.

Compared Chrome Stable and Canary on my laptop and confirmed that Canary resolves the issue.

We are not planning any further M55 stable releases.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/a9abedb1e941cc2512f65bb3f66d5446033682d3

commit a9abedb1e941cc2512f65bb3f66d5446033682d3
Author: wez <wez@chromium.org>
Date: Sun Feb 05 02:21:23 2017

Add unit-test for SharedSampler handling of zero-thread processes.

BUG= 683849 ,  673175 

Review-Url: https://codereview.chromium.org/2657393003
Cr-Commit-Position: refs/heads/master@{#448166}

[modify] https://crrev.com/a9abedb1e941cc2512f65bb3f66d5446033682d3/chrome/browser/BUILD.gn
[modify] https://crrev.com/a9abedb1e941cc2512f65bb3f66d5446033682d3/chrome/browser/task_manager/sampling/shared_sampler.h
[modify] https://crrev.com/a9abedb1e941cc2512f65bb3f66d5446033682d3/chrome/browser/task_manager/sampling/shared_sampler_win.cc
[add] https://crrev.com/a9abedb1e941cc2512f65bb3f66d5446033682d3/chrome/browser/task_manager/sampling/shared_sampler_win_defines.h
[rename] https://crrev.com/a9abedb1e941cc2512f65bb3f66d5446033682d3/chrome/browser/task_manager/sampling/shared_sampler_win_unittest.cc
[modify] https://crrev.com/a9abedb1e941cc2512f65bb3f66d5446033682d3/chrome/test/BUILD.gn