The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/cdd1d51b7aa6fb0dd7b4db3b125ccfc6e0ddbdbe

commit cdd1d51b7aa6fb0dd7b4db3b125ccfc6e0ddbdbe
Author: sebsg <sebsg@chromium.org>
Date: Thu May 31 16:15:13 2018

Disable http/tests/security/cross-origin-session-storage-allowed.html.

On Linux.

Tbr: tkent@chromium.org
Bug: 848281
Change-Id: I8cb6adf25f14c1494e743758547b67547cc27fbe
Reviewed-on: https://chromium-review.googlesource.com/1080076
Reviewed-by: Sebastien Seguin-Gagnon <sebsg@chromium.org>
Commit-Queue: Sebastien Seguin-Gagnon <sebsg@chromium.org>
Cr-Commit-Position: refs/heads/master@{#563263}
[modify] https://crrev.com/cdd1d51b7aa6fb0dd7b4db3b125ccfc6e0ddbdbe/third_party/WebKit/LayoutTests/TestExpectations

In both of those it's only failing in site_per_process_webkit_layout_tests, not in the regular webkit_layout_tests target afaict? So shouldn't the test expectations have been added to FlagExpectations/site-per-process rather than the top-level TestExpectations?

And for future reference, the test is failing with:crash log for renderer (pid <unknown>):
STDOUT: #CRASHED - renderer
STDERR: [4003:4023:0531/024840.916169:ERROR:render_process_host_impl.cc(4361)] Terminating render process for bad Mojo message: Received bad user message: Access denied for sessionStorage request
STDERR: [4003:4023:0531/024840.916220:ERROR:bad_message.cc(25)] Terminating renderer for bad IPC message, reason 123

Or in other words the check in SessionStorageNamespaceImplMojo::OpenArea somehow thinks the renderer that's trying to load session storage for a particular origin shouldn't have access to it.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/290447db50d51c439cfda76431b1b48b20464207

commit 290447db50d51c439cfda76431b1b48b20464207
Author: sebsg <sebsg@chromium.org>
Date: Fri Jun 01 14:49:23 2018

Disable http/tests/security/cross-origin-session-storage-allowed.html.

For site-per-process

Tbr: tkent@chromium.org
Bug: 848281
Change-Id: Iaa823cd3ff3e1a5e7003a6b2a944349c0ed88785
Reviewed-on: https://chromium-review.googlesource.com/1082000
Reviewed-by: Sebastien Seguin-Gagnon <sebsg@chromium.org>
Commit-Queue: Sebastien Seguin-Gagnon <sebsg@chromium.org>
Cr-Commit-Position: refs/heads/master@{#563627}
[modify] https://crrev.com/290447db50d51c439cfda76431b1b48b20464207/third_party/WebKit/LayoutTests/FlagExpectations/site-per-process
[modify] https://crrev.com/290447db50d51c439cfda76431b1b48b20464207/third_party/WebKit/LayoutTests/TestExpectations

As mek@ notes, this is probably related to issue 848651 and  issue 716490 .  Specifically, there's a chance this layout test might give repro steps for the renderer kills seen in issue 848651.  Will likely be necessary to fix that to reland Mojo session storage in  issue 716490 .

Thanks!

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/6cd49e6360f8ad453421eb5624dff5ecb04c1037

commit 6cd49e6360f8ad453421eb5624dff5ecb04c1037
Author: Daniel Murphy <dmurph@chromium.org>
Date: Sat Jun 16 04:42:39 2018

[SessionStorageS13N] Save process id per bound namespace

Previously, the process id was just the last-bound namespace process.
With site-isolation turned on, multiple processes bind to the same
namespace, so when 'old' ones opened up their storage area, they would
check the wrong process id and fail.

This stores the process id as the 'Context' on the binding set.

Bug: 848281
Change-Id: I5d4a082494f618c361f0ce845a641679330841fe
Reviewed-on: https://chromium-review.googlesource.com/1103553
Reviewed-by: Marijn Kruisselbrink <mek@chromium.org>
Commit-Queue: Marijn Kruisselbrink <mek@chromium.org>
Cr-Commit-Position: refs/heads/master@{#567871}
[modify] https://crrev.com/6cd49e6360f8ad453421eb5624dff5ecb04c1037/content/browser/dom_storage/session_storage_namespace_impl_mojo.cc
[modify] https://crrev.com/6cd49e6360f8ad453421eb5624dff5ecb04c1037/content/browser/dom_storage/session_storage_namespace_impl_mojo.h
[modify] https://crrev.com/6cd49e6360f8ad453421eb5624dff5ecb04c1037/content/browser/dom_storage/session_storage_namespace_impl_mojo_unittest.cc