Seems like this test is flaky. I will mark it as flaky. 

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/71b89422121ee304d169b8512539b04103523a12

commit 71b89422121ee304d169b8512539b04103523a12
Author: Luna Lu <loonybear@chromium.org>
Date: Tue Nov 14 17:49:25 2017

mark test message-port-start-and-close-different-microtask.html as flaky

TBR: alexclarke
Bug:  784909 
Change-Id: I2439160310386e88d2b91555bb605e5b8329f1c6
Reviewed-on: https://chromium-review.googlesource.com/768850
Reviewed-by: Luna Lu <loonybear@chromium.org>
Commit-Queue: Luna Lu <loonybear@chromium.org>
Cr-Commit-Position: refs/heads/master@{#516349}
[modify] https://crrev.com/71b89422121ee304d169b8512539b04103523a12/third_party/WebKit/LayoutTests/TestExpectations

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/5ef638ebf438e3786a221f691e3cb17be36b26a8

commit 5ef638ebf438e3786a221f691e3cb17be36b26a8
Author: Bret Sepulveda <bsep@chromium.org>
Date: Mon Jan 29 23:13:09 2018

Mark message-port-start-and-close-different-microtask.html as flaky.

This was done previously, but the virtual/mouseevent_fractional version
was missed. Also marking them flaky only on Win 7, as per the original
report.

TBR=alexclarke@chromium.org

Bug:  784909 
Change-Id: Iaa7ad21c750acedf179a9da0dcf686c65bf7300f
Reviewed-on: https://chromium-review.googlesource.com/891979
Reviewed-by: Bret Sepulveda <bsep@chromium.org>
Commit-Queue: Bret Sepulveda <bsep@chromium.org>
Cr-Commit-Position: refs/heads/master@{#532648}
[modify] https://crrev.com/5ef638ebf438e3786a221f691e3cb17be36b26a8/third_party/WebKit/LayoutTests/TestExpectations

I think this test is just incorrect. Nothing in the spec guarantees that the message needs to be delivered before the close is is run. Starting the message port merely "enables" the ports message queue, making it one of the task sources for the event loop. But then there are no guarantees as to what order the event loop processes tasks from that task source relative to the task posted by setTimeout() that does the close().

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/86285aba8f31e98d79a322d1c48356c2f738d809

commit 86285aba8f31e98d79a322d1c48356c2f738d809
Author: Marijn Kruisselbrink <mek@chromium.org>
Date: Fri Feb 02 22:44:16 2018

Remove message-port-start-and-close-different-microtask.html

This test makes assumptions about ordering of different task sources
that are not guaranteed by either the spec or the implementation.

Bug:  784909 
Change-Id: I6c5d25fd5c684594ca272a84a664692c90caff6e
Reviewed-on: https://chromium-review.googlesource.com/896034
Commit-Queue: Marijn Kruisselbrink <mek@chromium.org>
Reviewed-by: Alex Clarke <alexclarke@chromium.org>
Cr-Commit-Position: refs/heads/master@{#534179}
[modify] https://crrev.com/86285aba8f31e98d79a322d1c48356c2f738d809/third_party/WebKit/LayoutTests/FlagExpectations/disable-blink-features=RootLayerScrolling
[modify] https://crrev.com/86285aba8f31e98d79a322d1c48356c2f738d809/third_party/WebKit/LayoutTests/TestExpectations
[delete] https://crrev.com/827b0b6c5a688b37e330a18a1da1229b16752e18/third_party/WebKit/LayoutTests/fast/events/message-port-start-and-close-different-microtask-expected.txt
[delete] https://crrev.com/827b0b6c5a688b37e330a18a1da1229b16752e18/third_party/WebKit/LayoutTests/fast/events/message-port-start-and-close-different-microtask.html