Serialize a SecurityOrigin using effectivePort() rather than port().

Serialize a SecurityOrigin using effectivePort() rather than port().

Even though the URL spec and blink::SecurityOrigin normalize origins with
default ports to (scheme, host, 0/null), url::Origin instead normalizes them
to (scheme, host, effective_port). This fixes passing such origins over mojo
by always serializing the effective_port.

This fixes a crash when trying to use BroadcastChannel from a page with
a default port origin.

BUG=161070
Committed: https://crrev.com/41e640498f1d4d98ee775f65cea06ff1977c142b
Cr-Commit-Position: refs/heads/master@{#404420}

[Marijn Kruisselbrink, 2016-07-07 23:04:43]

Description was changed from

==========
Serialize a SecurityOrigin using effectivePort() rather than port().

Even though the URL spec and blink::SecurityOrigin normalize origins with
default ports to (scheme, host, 0/null), url::Origin instead normalizes them
to (scheme, host, effective_port). This fixes passing such origins over mojo
by always serializing the effective_port.

BUG=161070
==========

to

==========
Serialize a SecurityOrigin for mojo using effectivePort() rather than port().

Even though the URL spec and blink::SecurityOrigin normalize origins with
default ports to (scheme, host, 0/null), url::Origin instead normalizes them
to (scheme, host, effective_port). This fixes passing such origins over mojo
by always serializing the effective_port.

BUG=161070
==========

[Marijn Kruisselbrink, 2016-07-07 23:04:49]

Description was changed from

==========
Serialize a SecurityOrigin for mojo using effectivePort() rather than port().

Even though the URL spec and blink::SecurityOrigin normalize origins with
default ports to (scheme, host, 0/null), url::Origin instead normalizes them
to (scheme, host, effective_port). This fixes passing such origins over mojo
by always serializing the effective_port.

BUG=161070
==========

to

==========
Serialize a SecurityOrigin using effectivePort() rather than port().

Even though the URL spec and blink::SecurityOrigin normalize origins with
default ports to (scheme, host, 0/null), url::Origin instead normalizes them
to (scheme, host, effective_port). This fixes passing such origins over mojo
by always serializing the effective_port.

BUG=161070
==========

[Marijn Kruisselbrink, 2016-07-07 23:07:52]

Patchset #1 (id:1) has been deleted

[Marijn Kruisselbrink, 2016-07-07 23:09:23]

Description was changed from

==========
Serialize a SecurityOrigin using effectivePort() rather than port().

Even though the URL spec and blink::SecurityOrigin normalize origins with
default ports to (scheme, host, 0/null), url::Origin instead normalizes them
to (scheme, host, effective_port). This fixes passing such origins over mojo
by always serializing the effective_port.

BUG=161070
==========

to

==========
Serialize a SecurityOrigin using effectivePort() rather than port().

Even though the URL spec and blink::SecurityOrigin normalize origins with
default ports to (scheme, host, 0/null), url::Origin instead normalizes them
to (scheme, host, effective_port). This fixes passing such origins over mojo
by always serializing the effective_port.

This fixes a crash when trying to use BroadcastChannel from a page with
a default port origin.

BUG=161070
==========

[Marijn Kruisselbrink, 2016-07-07 23:11:22]

Patchset #1 (id:20001) has been deleted

[Marijn Kruisselbrink, 2016-07-07 23:13:43]

mek@chromium.org changed reviewers:
+ dcheng@chromium.org

[Marijn Kruisselbrink, 2016-07-07 23:13:43]

This also shows the shortcoming of testing struct traits by only
serializing/deserializing with the same struct traits... Ideally we'd have tests
that serialized something with blink specific struct traits, deserialized it
with non-blink struct traits, did the whole thing in reverse, and then compared
the output. Not sure how to best do that though...

[dcheng, 2016-07-08 06:03:23]

On 2016/07/07 23:13:43, Marijn Kruisselbrink wrote:
> This also shows the shortcoming of testing struct traits by only
> serializing/deserializing with the same struct traits... Ideally we'd have
tests
> that serialized something with blink specific struct traits, deserialized it
> with non-blink struct traits, did the whole thing in reverse, and then
compared
> the output. Not sure how to best do that though...

Doh. LGTM.

Fundamentally, this seems like it would have been fairly tricky to catch. And
there's always this weird semantic mismatch between url::Origin and
SecurityOrigin, since url::Origin doesn't have the distinction between effective
port / port, nor does it have the nullable concept (yet: mkwst@ is adding this
atm)

[Marijn Kruisselbrink, 2016-07-08 16:09:05]

The CQ bit was checked by mek@chromium.org

[commit-bot: I haz the power, 2016-07-08 16:09:35]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-07-08 17:02:35]

Committed patchset #1 (id:40001)

[commit-bot: I haz the power, 2016-07-08 17:02:39]

CQ bit was unchecked.

[commit-bot: I haz the power, 2016-07-08 17:05:36]

Description was changed from

==========
Serialize a SecurityOrigin using effectivePort() rather than port().

Even though the URL spec and blink::SecurityOrigin normalize origins with
default ports to (scheme, host, 0/null), url::Origin instead normalizes them
to (scheme, host, effective_port). This fixes passing such origins over mojo
by always serializing the effective_port.

This fixes a crash when trying to use BroadcastChannel from a page with
a default port origin.

BUG=161070
==========

to

==========
Serialize a SecurityOrigin using effectivePort() rather than port().

Even though the URL spec and blink::SecurityOrigin normalize origins with
default ports to (scheme, host, 0/null), url::Origin instead normalizes them
to (scheme, host, effective_port). This fixes passing such origins over mojo
by always serializing the effective_port.

This fixes a crash when trying to use BroadcastChannel from a page with
a default port origin.

BUG=161070

Committed: https://crrev.com/41e640498f1d4d98ee775f65cea06ff1977c142b
Cr-Commit-Position: refs/heads/master@{#404420}
==========

[commit-bot: I haz the power, 2016-07-08 17:05:37]

Patchset 1 (id:??) landed as
https://crrev.com/41e640498f1d4d98ee775f65cea06ff1977c142b
Cr-Commit-Position: refs/heads/master@{#404420}