1942306 - Add CSP support for "Create a Trusted Type Policy" for Workers

Status: RESOLVED FIXED

(Core :: DOM: Security, task)

Description

[Frédéric Wang (:fredw)]

Follow-up of bug 1901492.

See

https://searchfox.org/mozilla-central/rev/86c208f86f35d53dc824f18f8e540fe5b0663870/dom/security/trusted-types/TrustedTypePolicyFactory.cpp#61-70
https://w3c.github.io/trusted-types/dist/spec/#should-block-create-policy
https://github.com/w3c/trusted-types/issues/574

The rough idea will be to somehow expose trusted-types directives via CSPInfo, so they can be accessed from the worker thread.

In bug 1901492, this was done for require-trusted-types-for but the data was much simpler (essentially a three states enum).

Comment 1

[Frédéric Wang (:fredw)]

Attachment: Bug 1942306 - Report CSP violations for "Create a Trusted Type Policy" for Workers. r=smaug

This patch rewrites ShouldTrustedTypePolicyCreationBeBlockedByCSP so
that CSP violation reports are performed in a static function
ReportPolicyCreationViolations, separate from the calculation of the
PolicyCreation result.

For requests of policy creation originating from a worker, this allows
to implement a LogPolicyCreationViolationsRunnable in order to report
violations. In follow-up patches, we will implement the calculation of
the PolicyCreation result from the CSPInfo.

Comment 2

[Frédéric Wang (:fredw)]

Attachment: Bug 1942306 - Make ShouldTrustedTypePolicyCreationBeBlockedByCSP work for Workers. r=smaug,tschuster

This is done by exposing on CSPInfo the raw array of tt-expressions
associated to a CSPInfo::ContentSecurityPolicy [1]. Then we use a static
version of nsCSPDirective::ShouldCreateViolationForNewTrustedTypesPolicy
that relies on this raw array instead of an actual nsCSPDirective.

[1] https://w3c.github.io/trusted-types/dist/spec/#tt-expression

Comment 3

[Frédéric Wang (:fredw)]

Attachment: Bug 1942306 - More CSP tests for Trusted Types policy creation. r=smaug

Comment 4

[Frédéric Wang (:fredw)]

https://treeherder.mozilla.org/jobs?repo=try&revision=ea14ea88027e5b7a21ce22498046303ab62db02e

Comment 5

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/951a7cba63f5
More CSP tests for Trusted Types policy creation. r=smaug
https://hg.mozilla.org/integration/autoland/rev/24a402ea04d1
Report CSP violations for "Create a Trusted Type Policy" for Workers. r=smaug
https://hg.mozilla.org/integration/autoland/rev/0a18e08c5a3f
Make ShouldTrustedTypePolicyCreationBeBlockedByCSP work for Workers. r=tschuster

Comment 6

[Web Platform Test Sync Bot [:wpt-sync] (Matrix: #interop:mozilla.org)]

Created web-platform-tests PR https://github.com/web-platform-tests/wpt/pull/51629 for changes under testing/web-platform/tests

Comment 7

[Serban Stanca [:SerbanS]]

https://hg.mozilla.org/mozilla-central/rev/951a7cba63f5
https://hg.mozilla.org/mozilla-central/rev/24a402ea04d1
https://hg.mozilla.org/mozilla-central/rev/0a18e08c5a3f

Comment 8

[Web Platform Test Sync Bot [:wpt-sync] (Matrix: #interop:mozilla.org)]

Upstream PR merged by moz-wptsync-bot

Comment 9

[Tom Schuster (MoCo) [Out of Office]]

Apparently this breaks profiles after upgrading Firefox: bug 1958259. (My fault not noticing this!) I think we should just backout https://hg.mozilla.org/integration/autoland/rev/0a18e08c5a3f in Nightly and Beta. I think I have a potential solution that doesn't require changing the serialization format that I am going to investigate afterwards.

Comment 10

[Tom Schuster (MoCo) [Out of Office]]

Attachment: Bug 1942306 - Backout 0a18e08c5a3fc4f99f71c3621c111a7e9be01a40 for causing profile upgrade errors. r?smaug

Comment 11

[Olli Pettay [:smaug][[e-mail]]]

Since the backout is only part of what landed here, could it be done in a separate bug. Otherwise tracking what is where gets a bit tricky.

Comment 12

[Phabricator Automation]

Comment on attachment 9476931 [details]
Bug 1942306 - Backout 0a18e08c5a3fc4f99f71c3621c111a7e9be01a40 for causing profile upgrade errors. r?smaug

Revision D244336 was moved to bug 1958259. Setting attachment 9476931 [details] to obsolete.