While chrome_app_unittests is the last thing in the log, I wonder if chrome.dll is really the thing that's taking so long. I think ninja emits each line as it starts.

Any idea how to check this without an access to a Windows machine?

I just made a local build with the same args.gn used by that bot. It worked fine. Maybe linking took a long time for some targets, but everything succeeded. Maybe the timeout needs to be extended?

That would be quite strange. Dr. Memory doesn't require any additional instrumentation, and its compile time shouldn't be much greater than that of a regular build (it may disable certain optimizations, but if we're hitting the 20-minute timeout on a DrM bot now, we'll be hitting it on the regular build in a couple of weeks).

Looking at https://build.chromium.org/p/chromium.memory.full/builders/Chromium%20Windows%20Builder%20%28DrMemory%29/builds/415/steps/compile/logs/stdio:

[16127/16146] LINK(DLL) chrome.dll chrome.dll.lib chrome.dll.pdb
[16128/16146] LINK initialexe/chrome.exe initialexe/chrome.exe.pdb
[16129/16146] STAMP obj/chrome/reorder_imports.inputdeps.stamp
[16130/16146] LIB obj/chrome/app/test_support.lib
[16131/16146] ACTION //chrome:reorder_imports(//build/toolchain/win:x86)
[16132/16146] STAMP obj/chrome/reorder_imports.stamp
[16133/16146] STAMP obj/chrome/chrome.stamp
[16134/16146] STAMP obj/chrome_elf/chrome_elf_unittests.inputdeps.stamp
[16135/16146] LIB obj/chrome/test/test_support.lib

I suspect there's the following build dependency chain:

chrome.dll -> initialexe/chrome.exe -> obj/chrome/reorder_imports.inputdeps.stamp -> obj/chrome/chrome.stamp

Therefore chrome.dll has already been linked, and it's more likely that one of the test executables is failing to link.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/ad1aac65be3a92db6e76bf62336569ee65246666

commit ad1aac65be3a92db6e76bf62336569ee65246666
Author: glider <glider@chromium.org>
Date: Thu Aug 25 13:45:20 2016

Temporarily remove chrome_app_unittests from the Dr. Memory builder

BUG= 640185 
TBR=dpranke@chromium.org
NOTRY=true

Review-Url: https://codereview.chromium.org/2278113002
Cr-Commit-Position: refs/heads/master@{#414420}

[modify] https://crrev.com/ad1aac65be3a92db6e76bf62336569ee65246666/testing/buildbot/chromium.full.json

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/48586ee56e601e60095f30723060d4bd888ce6fe

commit 48586ee56e601e60095f30723060d4bd888ce6fe
Author: glider <glider@chromium.org>
Date: Fri Aug 26 07:59:56 2016

Remove chrome_app_unittests from chromium_builder_dbg_drmemory_win target

This should probably fix the compile step on the Dr. Memory builder.

BUG= 640185 
TBR=dpranke@chromium.org,cpu@chromium.org
NOTRY=true

Review-Url: https://codereview.chromium.org/2286623002
Cr-Commit-Position: refs/heads/master@{#414670}

[modify] https://crrev.com/48586ee56e601e60095f30723060d4bd888ce6fe/BUILD.gn

Looks lime chrome_app_unittests were innocent (or they are not the only problem).

https://build.chromium.org/p/chromium.memory.full/builders/Chromium%20Windows%20Builder%20%28DrMemory%29/builds/430/steps/compile/logs/stdio :

[2440/2451] LINK initialexe/chrome.exe initialexe/chrome.exe.pdb
[2441/2451] STAMP obj/chrome/reorder_imports.inputdeps.stamp
[2442/2451] STAMP obj/chrome/test/browser_tests_js_webui.stamp
[2443/2451] ACTION //chrome:reorder_imports(//build/toolchain/win:x86)
[2444/2451] STAMP obj/chrome/reorder_imports.stamp
[2445/2451] STAMP obj/chrome/chrome.stamp
[2446/2451] STAMP obj/chrome_elf/chrome_elf_unittests.inputdeps.stamp
[2447/2451] CXX obj/chrome_elf/chrome_elf_unittests/blacklist_test.obj

command timed out: 1200 seconds without output, attempting to kill

remoteFailed: [Failure instance: Traceback (failure with no frames): <class 'twisted.internet.error.ConnectionLost'>: Connection to the other side was lost in a non-clean fashion.
]

Not sure how easy it is, but maybe someone can login to the bot while it's stuck in the 1200 second window, see if link.exe is running, and try to see what it's linking.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/cb7b52a1f850f93570bd514a04f80ad89bf317c0

commit cb7b52a1f850f93570bd514a04f80ad89bf317c0
Author: tommycli <tommycli@chromium.org>
Date: Fri Sep 16 18:15:51 2016

Revert of Remove chrome_app_unittests from chromium_builder_dbg_drmemory_win target (patchset #1 id:1 of https://codereview.chromium.org/2286623002/ )

Reason for revert:
Reverting, because Dr. Memory test bots need that executable. It didn't break until a clobber landed and deleted the executable generated in the past.

Also glider@ reports that chrome_app_unittests was probably innocent (in the bug report) anyways.

First breaking build: https://build.chromium.org/p/chromium.memory.full/builders/Windows%20Unit%20%28DrMemory%20full%29%20%285%29/builds/130

Original issue's description:
> Remove chrome_app_unittests from chromium_builder_dbg_drmemory_win target
>
> This should probably fix the compile step on the Dr. Memory builder.
>
> BUG= 640185 
> TBR=dpranke@chromium.org,cpu@chromium.org
> NOTRY=true
>
> Committed: https://crrev.com/48586ee56e601e60095f30723060d4bd888ce6fe
> Cr-Commit-Position: refs/heads/master@{#414670}

TBR=cpu@chromium.org,dpranke@chromium.org,glider@chromium.org
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG= 640185 

Review-Url: https://codereview.chromium.org/2341353002
Cr-Commit-Position: refs/heads/master@{#419219}

[modify] https://crrev.com/cb7b52a1f850f93570bd514a04f80ad89bf317c0/BUILD.gn

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/8baafc72ea7c28702b2090b34c187ef3bff6af02

commit 8baafc72ea7c28702b2090b34c187ef3bff6af02
Author: tommycli <tommycli@chromium.org>
Date: Fri Sep 16 18:27:05 2016

Revert of Temporarily remove chrome_app_unittests from the Dr. Memory builder (patchset #1 id:1 of https://codereview.chromium.org/2278113002/ )

Reason for revert:
Reverting, because Dr. Memory test bots need that executable. It didn't break until a clobber landed and deleted the executable generated in the past.

Also glider@ reports that chrome_app_unittests was probably innocent (in the bug report) anyways.

First breaking build: https://build.chromium.org/p/chromium.memory.full/builders/Windows%20Unit%20%28DrMemory%20full%29%20%285%29/builds/130

Original issue's description:
> Temporarily remove chrome_app_unittests from the Dr. Memory builder
>
> BUG= 640185 
> TBR=dpranke@chromium.org
> NOTRY=true
>
> Committed: https://crrev.com/ad1aac65be3a92db6e76bf62336569ee65246666
> Cr-Commit-Position: refs/heads/master@{#414420}

TBR=dpranke@chromium.org,glider@chromium.org
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG= 640185 

Review-Url: https://codereview.chromium.org/2347223002
Cr-Commit-Position: refs/heads/master@{#419220}

[modify] https://crrev.com/8baafc72ea7c28702b2090b34c187ef3bff6af02/testing/buildbot/chromium.full.json

We've shut down the DrMemory builders, so this is now moot.