Request constructor should propagate origin internal field #1321

wanderview · 2021-10-06T19:29:29Z

Currently, the Request constructor always resets the request origin internal field to "client". See step 12 in:

https://fetch.spec.whatwg.org/#dom-request

This is problematic because a pass-through FetchEvent SW handler will end up using the service worker's origin for things like:

Origin header
Sec-Fetch-Site header
SameSite cookies

It would be safer for sites if we propagated the internal origin field from the original request. We also need to propagate the tainted origin flag as well in order to properly handle redirects.

This was discussed at the virtual F2F (w3c/ServiceWorker#1604). Notes are available.

The text was updated successfully, but these errors were encountered:

annevk · 2021-10-12T15:04:48Z

So given #1327 (comment) do we want to propagate "origin" or "site for cookies"?

Because it seems that otherwise the service worker might still select the wrong cookies.

wanderview · 2021-10-12T15:47:07Z

Well, I'm not sure if "site for cookies" is adequate to populate origin and sec-fetch-site headers.

I guess what you are saying is if we have A frames B1 and then B1 navigates to B2, then SameSite=strict cookies should not be sent for the B2 request. I guess we would need to propagate "site for cookies" as well. I'm not sure how hard that would be implementation-wise, though.

annevk · 2021-10-12T16:25:26Z

Right, if we care enough about the CSRF angle to propagate origin (and top-level?), we also need ancestor information.

wanderview · 2021-10-12T16:33:31Z

Well, at least for my example I don't think it would matter. If A frames B1 and B1 navigates to B2, then B2 has the same ancestor chain. So once service workers are partitioned and set its site for cookies based on the partition key the effect would be the same.

I guess propagating site for cookies would still be relevant for cases where B1 does window.open() to B2. Do we allow frames to create popups like that?

annevk · 2021-10-12T16:41:55Z

Ah right, we discussed that. Though it would mean that "site for cookies" would be a mix between the request authority and the service worker authority, which is somewhat weird.

Frames can create such popups, yes. So that indicates that the weird thing is probably bad.

wanderview · 2021-10-14T18:20:03Z

I don't see any information on request representing "site for cookies" right now. Does that exist? Are SameSite cookies not integrated in fetch spec yet?

Maybe we should fork "site for cookies" propagation out into a separate issue.

This CL contains essentially two changes: 1. The request initiator origin is plumbed through service workers that do `fetch(evt.request)`. In addition to plumbing, this requires changes to how we validate navigation requests in the CorsURLLoaderFactory. 2. Introduces an internal flag to indicate that a request should be treated as a main-frame navigation for SameSite cookie computation. This is necessary to override the service worker's IsolationInfo when a navigation request is passed through. For more detailed information about these changes please see the internal design doc at: https://docs.google.com/document/d/1KZscujuV7bCFEnzJW-0DaCPU-I40RJimQKoCcI0umTQ/edit?usp=sharing In addition, there is some discussion of these features in the following spec issues: whatwg/fetch#1321 whatwg/fetch#1327 Bug: 1115847,1241188 Change-Id: I7e236fa20aeabb705aef40fcf8d5c36da6d2798c

This CL contains essentially two changes: 1. The request initiator origin is plumbed through service workers that do `fetch(evt.request)`. In addition to plumbing, this requires changes to how we validate navigation requests in the CorsURLLoaderFactory. 2. Tracks the original destination of a request passed through a service worker. This is then used in the network service to force SameSite=Lax cookies to treat the request as a main frame navigation where appropriate. For more detailed information about these changes please see the internal design doc at: https://docs.google.com/document/d/1KZscujuV7bCFEnzJW-0DaCPU-I40RJimQKoCcI0umTQ/edit?usp=sharing In addition, there is some discussion of these features in the following spec issues: whatwg/fetch#1321 whatwg/fetch#1327 Bug: 1115847,1241188 Change-Id: I7e236fa20aeabb705aef40fcf8d5c36da6d2798c

This CL contains essentially two changes: 1. The request initiator origin is plumbed through service workers that do `fetch(evt.request)`. In addition to plumbing, this requires changes to how we validate navigation requests in the CorsURLLoaderFactory. 2. Tracks the original destination of a request passed through a service worker. This is then used in the network service to force SameSite=Lax cookies to treat the request as a main frame navigation where appropriate. For more detailed information about these changes please see the internal design doc at: https://docs.google.com/document/d/1KZscujuV7bCFEnzJW-0DaCPU-I40RJimQKoCcI0umTQ/edit?usp=sharing In addition, there is some discussion of these features in the following spec issues: whatwg/fetch#1321 whatwg/fetch#1327 The test includes WPT tests that verify navigation headers and SameSite cookies. Note, chrome has a couple expected failures in the SameSite cookie tests because of the "lax-allowing-unsafe" intervention that is currently enabled. See: https://source.chromium.org/chromium/chromium/src/+/main:third_party/blink/web_tests/TestExpectations;l=4635;drc=e8133cbf2469adb99c6610483ab78bcfb8cc4c76 Bug: 1115847,1241188 Change-Id: I7e236fa20aeabb705aef40fcf8d5c36da6d2798c

This CL contains essentially two changes: 1. The request initiator origin is plumbed through service workers that do `fetch(evt.request)`. In addition to plumbing, this requires changes to how we validate navigation requests in the CorsURLLoaderFactory. 2. Tracks the original destination of a request passed through a service worker. This is then used in the network service to force SameSite=Lax cookies to treat the request as a main frame navigation where appropriate. For more detailed information about these changes please see the internal design doc at: https://docs.google.com/document/d/1KZscujuV7bCFEnzJW-0DaCPU-I40RJimQKoCcI0umTQ/edit?usp=sharing In addition, there is some discussion of these features in the following spec issues: whatwg/fetch#1321 whatwg/fetch#1327 The test includes WPT tests that verify navigation headers and SameSite cookies. Note, chrome has a couple expected failures in the SameSite cookie tests because of the "lax-allowing-unsafe" intervention that is currently enabled. See: https://source.chromium.org/chromium/chromium/src/+/main:third_party/blink/web_tests/TestExpectations;l=4635;drc=e8133cbf2469adb99c6610483ab78bcfb8cc4c76 Bug: 1115847,1241188 Change-Id: I7e236fa20aeabb705aef40fcf8d5c36da6d2798c Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3115917 Reviewed-by: Matt Menke <mmenke@chromium.org> Reviewed-by: Yutaka Hirano <yhirano@chromium.org> Reviewed-by: Nasko Oskov <nasko@chromium.org> Reviewed-by: Łukasz Anforowicz <lukasza@chromium.org> Commit-Queue: Ben Kelly <wanderview@chromium.org> Cr-Commit-Position: refs/heads/main@{#936029}

…rkers." This reverts commit da0a6501cf321579bd46a27ff9fba1bb8ea910bb. Reason for revert: Failure on many bots with the following error message: The service worker navigation preload request was cancelled before 'preloadResponse' settled. If you intend to use 'preloadResponse', use waitUntil() or respondWith() to wait for the promise to settle.", source: (0) Original change's description: > Fetch: Plumb request initiator through passthrough service workers. > > This CL contains essentially two changes: > > 1. The request initiator origin is plumbed through service workers > that do `fetch(evt.request)`. In addition to plumbing, this > requires changes to how we validate navigation requests in the > CorsURLLoaderFactory. > 2. Tracks the original destination of a request passed through a > service worker. This is then used in the network service to force > SameSite=Lax cookies to treat the request as a main frame navigation > where appropriate. > > For more detailed information about these changes please see the > internal design doc at: > > https://docs.google.com/document/d/1KZscujuV7bCFEnzJW-0DaCPU-I40RJimQKoCcI0umTQ/edit?usp=sharing > > In addition, there is some discussion of these features in the following > spec issues: > > whatwg/fetch#1321 > whatwg/fetch#1327 > > The test includes WPT tests that verify navigation headers and SameSite > cookies. Note, chrome has a couple expected failures in the SameSite > cookie tests because of the "lax-allowing-unsafe" intervention that is > currently enabled. See: > > https://source.chromium.org/chromium/chromium/src/+/main:third_party/blink/web_tests/TestExpectations;l=4635;drc=e8133cbf2469adb99c6610483ab78bcfb8cc4c76 > > Bug: 1115847,1241188 > Change-Id: I7e236fa20aeabb705aef40fcf8d5c36da6d2798c > Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3115917 > Reviewed-by: Matt Menke <mmenke@chromium.org> > Reviewed-by: Yutaka Hirano <yhirano@chromium.org> > Reviewed-by: Nasko Oskov <nasko@chromium.org> > Reviewed-by: Łukasz Anforowicz <lukasza@chromium.org> > Commit-Queue: Ben Kelly <wanderview@chromium.org> > Cr-Commit-Position: refs/heads/main@{#936029} Bug: 1115847,1241188 Change-Id: I3044a6d20de172b4a8ab7e39a9f26191580003fa No-Presubmit: true No-Tree-Checks: true No-Try: true

…rkers." This reverts commit da0a6501cf321579bd46a27ff9fba1bb8ea910bb. Reason for revert: Failure on many bots with the following error message: The service worker navigation preload request was cancelled before 'preloadResponse' settled. If you intend to use 'preloadResponse', use waitUntil() or respondWith() to wait for the promise to settle.", source: (0) Original change's description: > Fetch: Plumb request initiator through passthrough service workers. > > This CL contains essentially two changes: > > 1. The request initiator origin is plumbed through service workers > that do `fetch(evt.request)`. In addition to plumbing, this > requires changes to how we validate navigation requests in the > CorsURLLoaderFactory. > 2. Tracks the original destination of a request passed through a > service worker. This is then used in the network service to force > SameSite=Lax cookies to treat the request as a main frame navigation > where appropriate. > > For more detailed information about these changes please see the > internal design doc at: > > https://docs.google.com/document/d/1KZscujuV7bCFEnzJW-0DaCPU-I40RJimQKoCcI0umTQ/edit?usp=sharing > > In addition, there is some discussion of these features in the following > spec issues: > > whatwg/fetch#1321 > whatwg/fetch#1327 > > The test includes WPT tests that verify navigation headers and SameSite > cookies. Note, chrome has a couple expected failures in the SameSite > cookie tests because of the "lax-allowing-unsafe" intervention that is > currently enabled. See: > > https://source.chromium.org/chromium/chromium/src/+/main:third_party/blink/web_tests/TestExpectations;l=4635;drc=e8133cbf2469adb99c6610483ab78bcfb8cc4c76 > > Bug: 1115847,1241188 > Change-Id: I7e236fa20aeabb705aef40fcf8d5c36da6d2798c > Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3115917 > Reviewed-by: Matt Menke <mmenke@chromium.org> > Reviewed-by: Yutaka Hirano <yhirano@chromium.org> > Reviewed-by: Nasko Oskov <nasko@chromium.org> > Reviewed-by: Łukasz Anforowicz <lukasza@chromium.org> > Commit-Queue: Ben Kelly <wanderview@chromium.org> > Cr-Commit-Position: refs/heads/main@{#936029} Bug: 1115847,1241188 Change-Id: I3044a6d20de172b4a8ab7e39a9f26191580003fa No-Presubmit: true No-Tree-Checks: true No-Try: true Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3251692 Auto-Submit: Alan Screen <awscreen@chromium.org> Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com> Commit-Queue: Alan Screen <awscreen@chromium.org> Owners-Override: Alan Screen <awscreen@chromium.org> Cr-Commit-Position: refs/heads/main@{#936125}

…rkers." This is a reland of da0a6501cf321579bd46a27ff9fba1bb8ea910bb Original change's description: > Fetch: Plumb request initiator through passthrough service workers. > > This CL contains essentially two changes: > > 1. The request initiator origin is plumbed through service workers > that do `fetch(evt.request)`. In addition to plumbing, this > requires changes to how we validate navigation requests in the > CorsURLLoaderFactory. > 2. Tracks the original destination of a request passed through a > service worker. This is then used in the network service to force > SameSite=Lax cookies to treat the request as a main frame navigation > where appropriate. > > For more detailed information about these changes please see the > internal design doc at: > > https://docs.google.com/document/d/1KZscujuV7bCFEnzJW-0DaCPU-I40RJimQKoCcI0umTQ/edit?usp=sharing > > In addition, there is some discussion of these features in the following > spec issues: > > whatwg/fetch#1321 > whatwg/fetch#1327 > > The test includes WPT tests that verify navigation headers and SameSite > cookies. Note, chrome has a couple expected failures in the SameSite > cookie tests because of the "lax-allowing-unsafe" intervention that is > currently enabled. See: > > https://source.chromium.org/chromium/chromium/src/+/main:third_party/blink/web_tests/TestExpectations;l=4635;drc=e8133cbf2469adb99c6610483ab78bcfb8cc4c76 > > Bug: 1115847,1241188 > Change-Id: I7e236fa20aeabb705aef40fcf8d5c36da6d2798c > Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3115917 > Reviewed-by: Matt Menke <mmenke@chromium.org> > Reviewed-by: Yutaka Hirano <yhirano@chromium.org> > Reviewed-by: Nasko Oskov <nasko@chromium.org> > Reviewed-by: Łukasz Anforowicz <lukasza@chromium.org> > Commit-Queue: Ben Kelly <wanderview@chromium.org> > Cr-Commit-Position: refs/heads/main@{#936029} Bug: 1115847,1241188 Change-Id: Ia26acbdd0d7ce6583d9a44f83ed086708657b8bd

…rkers." This is a reland of da0a6501cf321579bd46a27ff9fba1bb8ea910bb This CL also includes a change to mark the two WPT tests as requiring long timeout durations. On my fast build machine with an opt build they take ~5 seconds each to complete and the default timeout is 10 seconds. On slower bots with debug builds its highly likely that these tests would be marked as timing out. This change gives them a 60 second timeout instead. Original change's description: > Fetch: Plumb request initiator through passthrough service workers. > > This CL contains essentially two changes: > > 1. The request initiator origin is plumbed through service workers > that do `fetch(evt.request)`. In addition to plumbing, this > requires changes to how we validate navigation requests in the > CorsURLLoaderFactory. > 2. Tracks the original destination of a request passed through a > service worker. This is then used in the network service to force > SameSite=Lax cookies to treat the request as a main frame navigation > where appropriate. > > For more detailed information about these changes please see the > internal design doc at: > > https://docs.google.com/document/d/1KZscujuV7bCFEnzJW-0DaCPU-I40RJimQKoCcI0umTQ/edit?usp=sharing > > In addition, there is some discussion of these features in the following > spec issues: > > whatwg/fetch#1321 > whatwg/fetch#1327 > > The test includes WPT tests that verify navigation headers and SameSite > cookies. Note, chrome has a couple expected failures in the SameSite > cookie tests because of the "lax-allowing-unsafe" intervention that is > currently enabled. See: > > https://source.chromium.org/chromium/chromium/src/+/main:third_party/blink/web_tests/TestExpectations;l=4635;drc=e8133cbf2469adb99c6610483ab78bcfb8cc4c76 > > Bug: 1115847,1241188 > Change-Id: I7e236fa20aeabb705aef40fcf8d5c36da6d2798c > Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3115917 > Reviewed-by: Matt Menke <mmenke@chromium.org> > Reviewed-by: Yutaka Hirano <yhirano@chromium.org> > Reviewed-by: Nasko Oskov <nasko@chromium.org> > Reviewed-by: Łukasz Anforowicz <lukasza@chromium.org> > Commit-Queue: Ben Kelly <wanderview@chromium.org> > Cr-Commit-Position: refs/heads/main@{#936029} Bug: 1115847,1241188 Change-Id: Ia26acbdd0d7ce6583d9a44f83ed086708657b8bd

This CL contains essentially two changes: 1. The request initiator origin is plumbed through service workers that do `fetch(evt.request)`. In addition to plumbing, this requires changes to how we validate navigation requests in the CorsURLLoaderFactory. 2. Tracks the original destination of a request passed through a service worker. This is then used in the network service to force SameSite=Lax cookies to treat the request as a main frame navigation where appropriate. For more detailed information about these changes please see the internal design doc at: https://docs.google.com/document/d/1KZscujuV7bCFEnzJW-0DaCPU-I40RJimQKoCcI0umTQ/edit?usp=sharing In addition, there is some discussion of these features in the following spec issues: whatwg/fetch#1321 whatwg/fetch#1327 The test includes WPT tests that verify navigation headers and SameSite cookies. Note, chrome has a couple expected failures in the SameSite cookie tests because of the "lax-allowing-unsafe" intervention that is currently enabled. See: https://source.chromium.org/chromium/chromium/src/+/main:third_party/blink/web_tests/TestExpectations;l=4635;drc=e8133cbf2469adb99c6610483ab78bcfb8cc4c76 Bug: 1115847,1241188 Change-Id: I7e236fa20aeabb705aef40fcf8d5c36da6d2798c Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3115917 Reviewed-by: Matt Menke <mmenke@chromium.org> Reviewed-by: Yutaka Hirano <yhirano@chromium.org> Reviewed-by: Nasko Oskov <nasko@chromium.org> Reviewed-by: Łukasz Anforowicz <lukasza@chromium.org> Commit-Queue: Ben Kelly <wanderview@chromium.org> Cr-Commit-Position: refs/heads/main@{#936029}

…rkers." This reverts commit da0a650. Reason for revert: Failure on many bots with the following error message: The service worker navigation preload request was cancelled before 'preloadResponse' settled. If you intend to use 'preloadResponse', use waitUntil() or respondWith() to wait for the promise to settle.", source: (0) Original change's description: > Fetch: Plumb request initiator through passthrough service workers. > > This CL contains essentially two changes: > > 1. The request initiator origin is plumbed through service workers > that do `fetch(evt.request)`. In addition to plumbing, this > requires changes to how we validate navigation requests in the > CorsURLLoaderFactory. > 2. Tracks the original destination of a request passed through a > service worker. This is then used in the network service to force > SameSite=Lax cookies to treat the request as a main frame navigation > where appropriate. > > For more detailed information about these changes please see the > internal design doc at: > > https://docs.google.com/document/d/1KZscujuV7bCFEnzJW-0DaCPU-I40RJimQKoCcI0umTQ/edit?usp=sharing > > In addition, there is some discussion of these features in the following > spec issues: > > whatwg/fetch#1321 > whatwg/fetch#1327 > > The test includes WPT tests that verify navigation headers and SameSite > cookies. Note, chrome has a couple expected failures in the SameSite > cookie tests because of the "lax-allowing-unsafe" intervention that is > currently enabled. See: > > https://source.chromium.org/chromium/chromium/src/+/main:third_party/blink/web_tests/TestExpectations;l=4635;drc=e8133cbf2469adb99c6610483ab78bcfb8cc4c76 > > Bug: 1115847,1241188 > Change-Id: I7e236fa20aeabb705aef40fcf8d5c36da6d2798c > Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3115917 > Reviewed-by: Matt Menke <mmenke@chromium.org> > Reviewed-by: Yutaka Hirano <yhirano@chromium.org> > Reviewed-by: Nasko Oskov <nasko@chromium.org> > Reviewed-by: Łukasz Anforowicz <lukasza@chromium.org> > Commit-Queue: Ben Kelly <wanderview@chromium.org> > Cr-Commit-Position: refs/heads/main@{#936029} Bug: 1115847,1241188 Change-Id: I3044a6d20de172b4a8ab7e39a9f26191580003fa No-Presubmit: true No-Tree-Checks: true No-Try: true Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3251692 Auto-Submit: Alan Screen <awscreen@chromium.org> Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com> Commit-Queue: Alan Screen <awscreen@chromium.org> Owners-Override: Alan Screen <awscreen@chromium.org> Cr-Commit-Position: refs/heads/main@{#936125}

…rkers." This is a reland of da0a6501cf321579bd46a27ff9fba1bb8ea910bb This CL also includes a change to mark the two WPT tests as requiring long timeout durations. On my fast build machine with an opt build they take ~5 seconds each to complete and the default timeout is 10 seconds. On slower bots with debug builds its highly likely that these tests would be marked as timing out. This change gives them a 60 second timeout instead. Original change's description: > Fetch: Plumb request initiator through passthrough service workers. > > This CL contains essentially two changes: > > 1. The request initiator origin is plumbed through service workers > that do `fetch(evt.request)`. In addition to plumbing, this > requires changes to how we validate navigation requests in the > CorsURLLoaderFactory. > 2. Tracks the original destination of a request passed through a > service worker. This is then used in the network service to force > SameSite=Lax cookies to treat the request as a main frame navigation > where appropriate. > > For more detailed information about these changes please see the > internal design doc at: > > https://docs.google.com/document/d/1KZscujuV7bCFEnzJW-0DaCPU-I40RJimQKoCcI0umTQ/edit?usp=sharing > > In addition, there is some discussion of these features in the following > spec issues: > > whatwg/fetch#1321 > whatwg/fetch#1327 > > The test includes WPT tests that verify navigation headers and SameSite > cookies. Note, chrome has a couple expected failures in the SameSite > cookie tests because of the "lax-allowing-unsafe" intervention that is > currently enabled. See: > > https://source.chromium.org/chromium/chromium/src/+/main:third_party/blink/web_tests/TestExpectations;l=4635;drc=e8133cbf2469adb99c6610483ab78bcfb8cc4c76 > > Bug: 1115847,1241188 > Change-Id: I7e236fa20aeabb705aef40fcf8d5c36da6d2798c > Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3115917 > Reviewed-by: Matt Menke <mmenke@chromium.org> > Reviewed-by: Yutaka Hirano <yhirano@chromium.org> > Reviewed-by: Nasko Oskov <nasko@chromium.org> > Reviewed-by: Łukasz Anforowicz <lukasza@chromium.org> > Commit-Queue: Ben Kelly <wanderview@chromium.org> > Cr-Commit-Position: refs/heads/main@{#936029} Bug: 1115847,1241188 Change-Id: Ia26acbdd0d7ce6583d9a44f83ed086708657b8bd Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3251368 Reviewed-by: Matt Menke <mmenke@chromium.org> Reviewed-by: Yutaka Hirano <yhirano@chromium.org> Reviewed-by: Nasko Oskov <nasko@chromium.org> Reviewed-by: Łukasz Anforowicz <lukasza@chromium.org> Auto-Submit: Ben Kelly <wanderview@chromium.org> Commit-Queue: Ben Kelly <wanderview@chromium.org> Cr-Commit-Position: refs/heads/main@{#936560}

…rkers." This is a reland of da0a650 This CL also includes a change to mark the two WPT tests as requiring long timeout durations. On my fast build machine with an opt build they take ~5 seconds each to complete and the default timeout is 10 seconds. On slower bots with debug builds its highly likely that these tests would be marked as timing out. This change gives them a 60 second timeout instead. Original change's description: > Fetch: Plumb request initiator through passthrough service workers. > > This CL contains essentially two changes: > > 1. The request initiator origin is plumbed through service workers > that do `fetch(evt.request)`. In addition to plumbing, this > requires changes to how we validate navigation requests in the > CorsURLLoaderFactory. > 2. Tracks the original destination of a request passed through a > service worker. This is then used in the network service to force > SameSite=Lax cookies to treat the request as a main frame navigation > where appropriate. > > For more detailed information about these changes please see the > internal design doc at: > > https://docs.google.com/document/d/1KZscujuV7bCFEnzJW-0DaCPU-I40RJimQKoCcI0umTQ/edit?usp=sharing > > In addition, there is some discussion of these features in the following > spec issues: > > whatwg/fetch#1321 > whatwg/fetch#1327 > > The test includes WPT tests that verify navigation headers and SameSite > cookies. Note, chrome has a couple expected failures in the SameSite > cookie tests because of the "lax-allowing-unsafe" intervention that is > currently enabled. See: > > https://source.chromium.org/chromium/chromium/src/+/main:third_party/blink/web_tests/TestExpectations;l=4635;drc=e8133cbf2469adb99c6610483ab78bcfb8cc4c76 > > Bug: 1115847,1241188 > Change-Id: I7e236fa20aeabb705aef40fcf8d5c36da6d2798c > Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3115917 > Reviewed-by: Matt Menke <mmenke@chromium.org> > Reviewed-by: Yutaka Hirano <yhirano@chromium.org> > Reviewed-by: Nasko Oskov <nasko@chromium.org> > Reviewed-by: Łukasz Anforowicz <lukasza@chromium.org> > Commit-Queue: Ben Kelly <wanderview@chromium.org> > Cr-Commit-Position: refs/heads/main@{#936029} Bug: 1115847,1241188 Change-Id: Ia26acbdd0d7ce6583d9a44f83ed086708657b8bd Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3251368 Reviewed-by: Matt Menke <mmenke@chromium.org> Reviewed-by: Yutaka Hirano <yhirano@chromium.org> Reviewed-by: Nasko Oskov <nasko@chromium.org> Reviewed-by: Łukasz Anforowicz <lukasza@chromium.org> Auto-Submit: Ben Kelly <wanderview@chromium.org> Commit-Queue: Ben Kelly <wanderview@chromium.org> Cr-Commit-Position: refs/heads/main@{#936560}

…assthrough service workers., a=testonly Automatic update from web-platform-tests Fetch: Plumb request initiator through passthrough service workers. This CL contains essentially two changes: 1. The request initiator origin is plumbed through service workers that do `fetch(evt.request)`. In addition to plumbing, this requires changes to how we validate navigation requests in the CorsURLLoaderFactory. 2. Tracks the original destination of a request passed through a service worker. This is then used in the network service to force SameSite=Lax cookies to treat the request as a main frame navigation where appropriate. For more detailed information about these changes please see the internal design doc at: https://docs.google.com/document/d/1KZscujuV7bCFEnzJW-0DaCPU-I40RJimQKoCcI0umTQ/edit?usp=sharing In addition, there is some discussion of these features in the following spec issues: whatwg/fetch#1321 whatwg/fetch#1327 The test includes WPT tests that verify navigation headers and SameSite cookies. Note, chrome has a couple expected failures in the SameSite cookie tests because of the "lax-allowing-unsafe" intervention that is currently enabled. See: https://source.chromium.org/chromium/chromium/src/+/main:third_party/blink/web_tests/TestExpectations;l=4635;drc=e8133cbf2469adb99c6610483ab78bcfb8cc4c76 Bug: 1115847,1241188 Change-Id: I7e236fa20aeabb705aef40fcf8d5c36da6d2798c Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3115917 Reviewed-by: Matt Menke <mmenke@chromium.org> Reviewed-by: Yutaka Hirano <yhirano@chromium.org> Reviewed-by: Nasko Oskov <nasko@chromium.org> Reviewed-by: Łukasz Anforowicz <lukasza@chromium.org> Commit-Queue: Ben Kelly <wanderview@chromium.org> Cr-Commit-Position: refs/heads/main@{#936029} -- wpt-commits: cd0951f4268af579e9e26913fbefea1684c67f2d wpt-pr: 31378

…hrough passthrough service workers.", a=testonly Automatic update from web-platform-tests Revert "Fetch: Plumb request initiator through passthrough service workers." This reverts commit da0a6501cf321579bd46a27ff9fba1bb8ea910bb. Reason for revert: Failure on many bots with the following error message: The service worker navigation preload request was cancelled before 'preloadResponse' settled. If you intend to use 'preloadResponse', use waitUntil() or respondWith() to wait for the promise to settle.", source: (0) Original change's description: > Fetch: Plumb request initiator through passthrough service workers. > > This CL contains essentially two changes: > > 1. The request initiator origin is plumbed through service workers > that do `fetch(evt.request)`. In addition to plumbing, this > requires changes to how we validate navigation requests in the > CorsURLLoaderFactory. > 2. Tracks the original destination of a request passed through a > service worker. This is then used in the network service to force > SameSite=Lax cookies to treat the request as a main frame navigation > where appropriate. > > For more detailed information about these changes please see the > internal design doc at: > > https://docs.google.com/document/d/1KZscujuV7bCFEnzJW-0DaCPU-I40RJimQKoCcI0umTQ/edit?usp=sharing > > In addition, there is some discussion of these features in the following > spec issues: > > whatwg/fetch#1321 > whatwg/fetch#1327 > > The test includes WPT tests that verify navigation headers and SameSite > cookies. Note, chrome has a couple expected failures in the SameSite > cookie tests because of the "lax-allowing-unsafe" intervention that is > currently enabled. See: > > https://source.chromium.org/chromium/chromium/src/+/main:third_party/blink/web_tests/TestExpectations;l=4635;drc=e8133cbf2469adb99c6610483ab78bcfb8cc4c76 > > Bug: 1115847,1241188 > Change-Id: I7e236fa20aeabb705aef40fcf8d5c36da6d2798c > Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3115917 > Reviewed-by: Matt Menke <mmenke@chromium.org> > Reviewed-by: Yutaka Hirano <yhirano@chromium.org> > Reviewed-by: Nasko Oskov <nasko@chromium.org> > Reviewed-by: Łukasz Anforowicz <lukasza@chromium.org> > Commit-Queue: Ben Kelly <wanderview@chromium.org> > Cr-Commit-Position: refs/heads/main@{#936029} Bug: 1115847,1241188 Change-Id: I3044a6d20de172b4a8ab7e39a9f26191580003fa No-Presubmit: true No-Tree-Checks: true No-Try: true Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3251692 Auto-Submit: Alan Screen <awscreen@chromium.org> Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com> Commit-Queue: Alan Screen <awscreen@chromium.org> Owners-Override: Alan Screen <awscreen@chromium.org> Cr-Commit-Position: refs/heads/main@{#936125} -- wpt-commits: 85ebe754c6a2f8a706d582c71433301ed7901a18 wpt-pr: 31425

…assthrough service workers., a=testonly Automatic update from web-platform-tests Fetch: Plumb request initiator through passthrough service workers. This CL contains essentially two changes: 1. The request initiator origin is plumbed through service workers that do `fetch(evt.request)`. In addition to plumbing, this requires changes to how we validate navigation requests in the CorsURLLoaderFactory. 2. Tracks the original destination of a request passed through a service worker. This is then used in the network service to force SameSite=Lax cookies to treat the request as a main frame navigation where appropriate. For more detailed information about these changes please see the internal design doc at: https://docs.google.com/document/d/1KZscujuV7bCFEnzJW-0DaCPU-I40RJimQKoCcI0umTQ/edit?usp=sharing In addition, there is some discussion of these features in the following spec issues: whatwg/fetch#1321 whatwg/fetch#1327 The test includes WPT tests that verify navigation headers and SameSite cookies. Note, chrome has a couple expected failures in the SameSite cookie tests because of the "lax-allowing-unsafe" intervention that is currently enabled. See: https://source.chromium.org/chromium/chromium/src/+/main:third_party/blink/web_tests/TestExpectations;l=4635;drc=e8133cbf2469adb99c6610483ab78bcfb8cc4c76 Bug: 1115847,1241188 Change-Id: I7e236fa20aeabb705aef40fcf8d5c36da6d2798c Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3115917 Reviewed-by: Matt Menke <mmenke@chromium.org> Reviewed-by: Yutaka Hirano <yhirano@chromium.org> Reviewed-by: Nasko Oskov <nasko@chromium.org> Reviewed-by: Łukasz Anforowicz <lukasza@chromium.org> Commit-Queue: Ben Kelly <wanderview@chromium.org> Cr-Commit-Position: refs/heads/main@{#936029} -- wpt-commits: cd0951f4268af579e9e26913fbefea1684c67f2d wpt-pr: 31378

…hrough passthrough service workers.", a=testonly Automatic update from web-platform-tests Revert "Fetch: Plumb request initiator through passthrough service workers." This reverts commit da0a6501cf321579bd46a27ff9fba1bb8ea910bb. Reason for revert: Failure on many bots with the following error message: The service worker navigation preload request was cancelled before 'preloadResponse' settled. If you intend to use 'preloadResponse', use waitUntil() or respondWith() to wait for the promise to settle.", source: (0) Original change's description: > Fetch: Plumb request initiator through passthrough service workers. > > This CL contains essentially two changes: > > 1. The request initiator origin is plumbed through service workers > that do `fetch(evt.request)`. In addition to plumbing, this > requires changes to how we validate navigation requests in the > CorsURLLoaderFactory. > 2. Tracks the original destination of a request passed through a > service worker. This is then used in the network service to force > SameSite=Lax cookies to treat the request as a main frame navigation > where appropriate. > > For more detailed information about these changes please see the > internal design doc at: > > https://docs.google.com/document/d/1KZscujuV7bCFEnzJW-0DaCPU-I40RJimQKoCcI0umTQ/edit?usp=sharing > > In addition, there is some discussion of these features in the following > spec issues: > > whatwg/fetch#1321 > whatwg/fetch#1327 > > The test includes WPT tests that verify navigation headers and SameSite > cookies. Note, chrome has a couple expected failures in the SameSite > cookie tests because of the "lax-allowing-unsafe" intervention that is > currently enabled. See: > > https://source.chromium.org/chromium/chromium/src/+/main:third_party/blink/web_tests/TestExpectations;l=4635;drc=e8133cbf2469adb99c6610483ab78bcfb8cc4c76 > > Bug: 1115847,1241188 > Change-Id: I7e236fa20aeabb705aef40fcf8d5c36da6d2798c > Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3115917 > Reviewed-by: Matt Menke <mmenke@chromium.org> > Reviewed-by: Yutaka Hirano <yhirano@chromium.org> > Reviewed-by: Nasko Oskov <nasko@chromium.org> > Reviewed-by: Łukasz Anforowicz <lukasza@chromium.org> > Commit-Queue: Ben Kelly <wanderview@chromium.org> > Cr-Commit-Position: refs/heads/main@{#936029} Bug: 1115847,1241188 Change-Id: I3044a6d20de172b4a8ab7e39a9f26191580003fa No-Presubmit: true No-Tree-Checks: true No-Try: true Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3251692 Auto-Submit: Alan Screen <awscreen@chromium.org> Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com> Commit-Queue: Alan Screen <awscreen@chromium.org> Owners-Override: Alan Screen <awscreen@chromium.org> Cr-Commit-Position: refs/heads/main@{#936125} -- wpt-commits: 85ebe754c6a2f8a706d582c71433301ed7901a18 wpt-pr: 31425

…hrough passthrough service workers.", a=testonly Automatic update from web-platform-tests Reland "Fetch: Plumb request initiator through passthrough service workers." This is a reland of da0a6501cf321579bd46a27ff9fba1bb8ea910bb This CL also includes a change to mark the two WPT tests as requiring long timeout durations. On my fast build machine with an opt build they take ~5 seconds each to complete and the default timeout is 10 seconds. On slower bots with debug builds its highly likely that these tests would be marked as timing out. This change gives them a 60 second timeout instead. Original change's description: > Fetch: Plumb request initiator through passthrough service workers. > > This CL contains essentially two changes: > > 1. The request initiator origin is plumbed through service workers > that do `fetch(evt.request)`. In addition to plumbing, this > requires changes to how we validate navigation requests in the > CorsURLLoaderFactory. > 2. Tracks the original destination of a request passed through a > service worker. This is then used in the network service to force > SameSite=Lax cookies to treat the request as a main frame navigation > where appropriate. > > For more detailed information about these changes please see the > internal design doc at: > > https://docs.google.com/document/d/1KZscujuV7bCFEnzJW-0DaCPU-I40RJimQKoCcI0umTQ/edit?usp=sharing > > In addition, there is some discussion of these features in the following > spec issues: > > whatwg/fetch#1321 > whatwg/fetch#1327 > > The test includes WPT tests that verify navigation headers and SameSite > cookies. Note, chrome has a couple expected failures in the SameSite > cookie tests because of the "lax-allowing-unsafe" intervention that is > currently enabled. See: > > https://source.chromium.org/chromium/chromium/src/+/main:third_party/blink/web_tests/TestExpectations;l=4635;drc=e8133cbf2469adb99c6610483ab78bcfb8cc4c76 > > Bug: 1115847,1241188 > Change-Id: I7e236fa20aeabb705aef40fcf8d5c36da6d2798c > Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3115917 > Reviewed-by: Matt Menke <mmenke@chromium.org> > Reviewed-by: Yutaka Hirano <yhirano@chromium.org> > Reviewed-by: Nasko Oskov <nasko@chromium.org> > Reviewed-by: Łukasz Anforowicz <lukasza@chromium.org> > Commit-Queue: Ben Kelly <wanderview@chromium.org> > Cr-Commit-Position: refs/heads/main@{#936029} Bug: 1115847,1241188 Change-Id: Ia26acbdd0d7ce6583d9a44f83ed086708657b8bd Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3251368 Reviewed-by: Matt Menke <mmenke@chromium.org> Reviewed-by: Yutaka Hirano <yhirano@chromium.org> Reviewed-by: Nasko Oskov <nasko@chromium.org> Reviewed-by: Łukasz Anforowicz <lukasza@chromium.org> Auto-Submit: Ben Kelly <wanderview@chromium.org> Commit-Queue: Ben Kelly <wanderview@chromium.org> Cr-Commit-Position: refs/heads/main@{#936560} -- wpt-commits: ef9899c9503c50c675ba1c374a04a82600d23aaf wpt-pr: 31430

…st constructor. (Fixes whatwg#1321)

…st constructor. (Fixes whatwg#1321, whatwg#1335)

This CL contains essentially two changes: 1. The request initiator origin is plumbed through service workers that do `fetch(evt.request)`. In addition to plumbing, this requires changes to how we validate navigation requests in the CorsURLLoaderFactory. 2. Tracks the original destination of a request passed through a service worker. This is then used in the network service to force SameSite=Lax cookies to treat the request as a main frame navigation where appropriate. For more detailed information about these changes please see the internal design doc at: https://docs.google.com/document/d/1KZscujuV7bCFEnzJW-0DaCPU-I40RJimQKoCcI0umTQ/edit?usp=sharing In addition, there is some discussion of these features in the following spec issues: whatwg/fetch#1321 whatwg/fetch#1327 The test includes WPT tests that verify navigation headers and SameSite cookies. Note, chrome has a couple expected failures in the SameSite cookie tests because of the "lax-allowing-unsafe" intervention that is currently enabled. See: https://source.chromium.org/chromium/chromium/src/+/main:third_party/blink/web_tests/TestExpectations;l=4635;drc=e8133cbf2469adb99c6610483ab78bcfb8cc4c76 Bug: 1115847,1241188 Change-Id: I7e236fa20aeabb705aef40fcf8d5c36da6d2798c Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3115917 Reviewed-by: Matt Menke <mmenke@chromium.org> Reviewed-by: Yutaka Hirano <yhirano@chromium.org> Reviewed-by: Nasko Oskov <nasko@chromium.org> Reviewed-by: Łukasz Anforowicz <lukasza@chromium.org> Commit-Queue: Ben Kelly <wanderview@chromium.org> Cr-Commit-Position: refs/heads/main@{#936029}

…rkers." This reverts commit da0a6501cf321579bd46a27ff9fba1bb8ea910bb. Reason for revert: Failure on many bots with the following error message: The service worker navigation preload request was cancelled before 'preloadResponse' settled. If you intend to use 'preloadResponse', use waitUntil() or respondWith() to wait for the promise to settle.", source: (0) Original change's description: > Fetch: Plumb request initiator through passthrough service workers. > > This CL contains essentially two changes: > > 1. The request initiator origin is plumbed through service workers > that do `fetch(evt.request)`. In addition to plumbing, this > requires changes to how we validate navigation requests in the > CorsURLLoaderFactory. > 2. Tracks the original destination of a request passed through a > service worker. This is then used in the network service to force > SameSite=Lax cookies to treat the request as a main frame navigation > where appropriate. > > For more detailed information about these changes please see the > internal design doc at: > > https://docs.google.com/document/d/1KZscujuV7bCFEnzJW-0DaCPU-I40RJimQKoCcI0umTQ/edit?usp=sharing > > In addition, there is some discussion of these features in the following > spec issues: > > whatwg/fetch#1321 > whatwg/fetch#1327 > > The test includes WPT tests that verify navigation headers and SameSite > cookies. Note, chrome has a couple expected failures in the SameSite > cookie tests because of the "lax-allowing-unsafe" intervention that is > currently enabled. See: > > https://source.chromium.org/chromium/chromium/src/+/main:third_party/blink/web_tests/TestExpectations;l=4635;drc=e8133cbf2469adb99c6610483ab78bcfb8cc4c76 > > Bug: 1115847,1241188 > Change-Id: I7e236fa20aeabb705aef40fcf8d5c36da6d2798c > Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3115917 > Reviewed-by: Matt Menke <mmenke@chromium.org> > Reviewed-by: Yutaka Hirano <yhirano@chromium.org> > Reviewed-by: Nasko Oskov <nasko@chromium.org> > Reviewed-by: Łukasz Anforowicz <lukasza@chromium.org> > Commit-Queue: Ben Kelly <wanderview@chromium.org> > Cr-Commit-Position: refs/heads/main@{#936029} Bug: 1115847,1241188 Change-Id: I3044a6d20de172b4a8ab7e39a9f26191580003fa No-Presubmit: true No-Tree-Checks: true No-Try: true Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3251692 Auto-Submit: Alan Screen <awscreen@chromium.org> Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com> Commit-Queue: Alan Screen <awscreen@chromium.org> Owners-Override: Alan Screen <awscreen@chromium.org> Cr-Commit-Position: refs/heads/main@{#936125}

…rkers." This is a reland of da0a6501cf321579bd46a27ff9fba1bb8ea910bb This CL also includes a change to mark the two WPT tests as requiring long timeout durations. On my fast build machine with an opt build they take ~5 seconds each to complete and the default timeout is 10 seconds. On slower bots with debug builds its highly likely that these tests would be marked as timing out. This change gives them a 60 second timeout instead. Original change's description: > Fetch: Plumb request initiator through passthrough service workers. > > This CL contains essentially two changes: > > 1. The request initiator origin is plumbed through service workers > that do `fetch(evt.request)`. In addition to plumbing, this > requires changes to how we validate navigation requests in the > CorsURLLoaderFactory. > 2. Tracks the original destination of a request passed through a > service worker. This is then used in the network service to force > SameSite=Lax cookies to treat the request as a main frame navigation > where appropriate. > > For more detailed information about these changes please see the > internal design doc at: > > https://docs.google.com/document/d/1KZscujuV7bCFEnzJW-0DaCPU-I40RJimQKoCcI0umTQ/edit?usp=sharing > > In addition, there is some discussion of these features in the following > spec issues: > > whatwg/fetch#1321 > whatwg/fetch#1327 > > The test includes WPT tests that verify navigation headers and SameSite > cookies. Note, chrome has a couple expected failures in the SameSite > cookie tests because of the "lax-allowing-unsafe" intervention that is > currently enabled. See: > > https://source.chromium.org/chromium/chromium/src/+/main:third_party/blink/web_tests/TestExpectations;l=4635;drc=e8133cbf2469adb99c6610483ab78bcfb8cc4c76 > > Bug: 1115847,1241188 > Change-Id: I7e236fa20aeabb705aef40fcf8d5c36da6d2798c > Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3115917 > Reviewed-by: Matt Menke <mmenke@chromium.org> > Reviewed-by: Yutaka Hirano <yhirano@chromium.org> > Reviewed-by: Nasko Oskov <nasko@chromium.org> > Reviewed-by: Łukasz Anforowicz <lukasza@chromium.org> > Commit-Queue: Ben Kelly <wanderview@chromium.org> > Cr-Commit-Position: refs/heads/main@{#936029} Bug: 1115847,1241188 Change-Id: Ia26acbdd0d7ce6583d9a44f83ed086708657b8bd Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3251368 Reviewed-by: Matt Menke <mmenke@chromium.org> Reviewed-by: Yutaka Hirano <yhirano@chromium.org> Reviewed-by: Nasko Oskov <nasko@chromium.org> Reviewed-by: Łukasz Anforowicz <lukasza@chromium.org> Auto-Submit: Ben Kelly <wanderview@chromium.org> Commit-Queue: Ben Kelly <wanderview@chromium.org> Cr-Commit-Position: refs/heads/main@{#936560}

…hatwg#1321, whatwg#1335)

Manual backprort of patch for "CVE-2022-0111: Inappropriate implementation in Navigation." and "CVE-2022-0117: Policy bypass in Service Workers", originally reviewed on https://chromium-review.googlesource.com/c/chromium/src/+/3251368: Reland "Fetch: Plumb request initiator through passthrough service workers." This is a reland of da0a6501cf321579bd46a27ff9fba1bb8ea910bb This CL also includes a change to mark the two WPT tests as requiring long timeout durations. On my fast build machine with an opt build they take ~5 seconds each to complete and the default timeout is 10 seconds. On slower bots with debug builds its highly likely that these tests would be marked as timing out. This change gives them a 60 second timeout instead. Original change's description: > Fetch: Plumb request initiator through passthrough service workers. > > This CL contains essentially two changes: > > 1. The request initiator origin is plumbed through service workers > that do `fetch(evt.request)`. In addition to plumbing, this > requires changes to how we validate navigation requests in the > CorsURLLoaderFactory. > 2. Tracks the original destination of a request passed through a > service worker. This is then used in the network service to force > SameSite=Lax cookies to treat the request as a main frame navigation > where appropriate. > > For more detailed information about these changes please see the > internal design doc at: > > https://docs.google.com/document/d/1KZscujuV7bCFEnzJW-0DaCPU-I40RJimQKoCcI0umTQ/edit?usp=sharing > > In addition, there is some discussion of these features in the following > spec issues: > > whatwg/fetch#1321 > whatwg/fetch#1327 > > The test includes WPT tests that verify navigation headers and SameSite > cookies. Note, chrome has a couple expected failures in the SameSite > cookie tests because of the "lax-allowing-unsafe" intervention that is > currently enabled. See: > > https://source.chromium.org/chromium/chromium/src/+/main:third_party/blink/web_tests/TestExpectations;l=4635;drc=e8133cbf2469adb99c6610483ab78bcfb8cc4c76 > > Bug: 1115847,1241188 > Change-Id: I7e236fa20aeabb705aef40fcf8d5c36da6d2798c > Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3115917 > Reviewed-by: Matt Menke <mmenke@chromium.org> > Reviewed-by: Yutaka Hirano <yhirano@chromium.org> > Reviewed-by: Nasko Oskov <nasko@chromium.org> > Reviewed-by: Łukasz Anforowicz <lukasza@chromium.org> > Commit-Queue: Ben Kelly <wanderview@chromium.org> > Cr-Commit-Position: refs/heads/main@{#936029} Bug: 1115847,1241188 Change-Id: Ia26acbdd0d7ce6583d9a44f83ed086708657b8bd Reviewed-by: Matt Menke <mmenke@chromium.org> Reviewed-by: Yutaka Hirano <yhirano@chromium.org> Reviewed-by: Nasko Oskov <nasko@chromium.org> Reviewed-by: Łukasz Anforowicz <lukasza@chromium.org> Auto-Submit: Ben Kelly <wanderview@chromium.org> Commit-Queue: Ben Kelly <wanderview@chromium.org> Cr-Commit-Position: refs/heads/main@{#936560} Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>

Manual icherry-pick of patch for "CVE-2022-0111: Inappropriate implementation in Navigation." and "CVE-2022-0117: Policy bypass in Service Workers", originally reviewed on https://chromium-review.googlesource.com/c/chromium/src/+/3251368: Reland "Fetch: Plumb request initiator through passthrough service workers." This is a reland of da0a6501cf321579bd46a27ff9fba1bb8ea910bb This CL also includes a change to mark the two WPT tests as requiring long timeout durations. On my fast build machine with an opt build they take ~5 seconds each to complete and the default timeout is 10 seconds. On slower bots with debug builds its highly likely that these tests would be marked as timing out. This change gives them a 60 second timeout instead. Original change's description: > Fetch: Plumb request initiator through passthrough service workers. > > This CL contains essentially two changes: > > 1. The request initiator origin is plumbed through service workers > that do `fetch(evt.request)`. In addition to plumbing, this > requires changes to how we validate navigation requests in the > CorsURLLoaderFactory. > 2. Tracks the original destination of a request passed through a > service worker. This is then used in the network service to force > SameSite=Lax cookies to treat the request as a main frame navigation > where appropriate. > > For more detailed information about these changes please see the > internal design doc at: > > https://docs.google.com/document/d/1KZscujuV7bCFEnzJW-0DaCPU-I40RJimQKoCcI0umTQ/edit?usp=sharing > > In addition, there is some discussion of these features in the following > spec issues: > > whatwg/fetch#1321 > whatwg/fetch#1327 > > The test includes WPT tests that verify navigation headers and SameSite > cookies. Note, chrome has a couple expected failures in the SameSite > cookie tests because of the "lax-allowing-unsafe" intervention that is > currently enabled. See: > > https://source.chromium.org/chromium/chromium/src/+/main:third_party/blink/web_tests/TestExpectations;l=4635;drc=e8133cbf2469adb99c6610483ab78bcfb8cc4c76 > > Bug: 1115847,1241188 > Change-Id: I7e236fa20aeabb705aef40fcf8d5c36da6d2798c > Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3115917 > Reviewed-by: Matt Menke <mmenke@chromium.org> > Reviewed-by: Yutaka Hirano <yhirano@chromium.org> > Reviewed-by: Nasko Oskov <nasko@chromium.org> > Reviewed-by: Łukasz Anforowicz <lukasza@chromium.org> > Commit-Queue: Ben Kelly <wanderview@chromium.org> > Cr-Commit-Position: refs/heads/main@{#936029} Bug: 1115847,1241188 Change-Id: Ia26acbdd0d7ce6583d9a44f83ed086708657b8bd Reviewed-by: Matt Menke <mmenke@chromium.org> Reviewed-by: Yutaka Hirano <yhirano@chromium.org> Reviewed-by: Nasko Oskov <nasko@chromium.org> Reviewed-by: Łukasz Anforowicz <lukasza@chromium.org> Auto-Submit: Ben Kelly <wanderview@chromium.org> Commit-Queue: Ben Kelly <wanderview@chromium.org> Cr-Commit-Position: refs/heads/main@{#936560} Reviewed-by: Michal Klocek <michal.klocek@qt.io>

Tests: WPT service-workers/service-worker/navigation-headers.https.html and service-workers/service-worker/same-site-cookies.https.html. Fixes whatwg#1321 and fixes whatwg#1335. Co-authored-by: Anne van Kesteren <annevk@annevk.nl>

Manual backport of patch for "CVE-2022-0111: Inappropriate implementation in Navigation." and "CVE-2022-0117: Policy bypass in Service Workers", originally reviewed on https://chromium-review.googlesource.com/c/chromium/src/+/3251368: Reland "Fetch: Plumb request initiator through passthrough service workers." This is a reland of da0a6501cf321579bd46a27ff9fba1bb8ea910bb This CL also includes a change to mark the two WPT tests as requiring long timeout durations. On my fast build machine with an opt build they take ~5 seconds each to complete and the default timeout is 10 seconds. On slower bots with debug builds its highly likely that these tests would be marked as timing out. This change gives them a 60 second timeout instead. Original change's description: > Fetch: Plumb request initiator through passthrough service workers. > > This CL contains essentially two changes: > > 1. The request initiator origin is plumbed through service workers >that do `fetch(evt.request)`. In addition to plumbing, this >requires changes to how we validate navigation requests in the >CorsURLLoaderFactory. > 2. Tracks the original destination of a request passed through a >service worker. This is then used in the network service to force >SameSite=Lax cookies to treat the request as a main frame navigation >where appropriate. > > For more detailed information about these changes please see the > internal design doc at: > > https://docs.google.com/document/d/1KZscujuV7bCFEnzJW-0DaCPU-I40RJimQKoCcI0umTQ/edit?usp=sharing > > In addition, there is some discussion of these features in the following > spec issues: > > whatwg/fetch#1321 > whatwg/fetch#1327 > > The test includes WPT tests that verify navigation headers and SameSite > cookies. Note, chrome has a couple expected failures in the SameSite > cookie tests because of the "lax-allowing-unsafe" intervention that is > currently enabled. See: > > https://source.chromium.org/chromium/chromium/src/+/main:third_party/blink/web_tests/TestExpectations;l=4635;drc=e8133cbf2469adb99c6610483ab78bcfb8cc4c76 > > Bug: 1115847,1241188 > Change-Id: I7e236fa20aeabb705aef40fcf8d5c36da6d2798c > Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3115917 > Reviewed-by: Matt Menke <mmenke@chromium.org> > Reviewed-by: Yutaka Hirano <yhirano@chromium.org> > Reviewed-by: Nasko Oskov <nasko@chromium.org> > Reviewed-by: Łukasz Anforowicz <lukasza@chromium.org> > Commit-Queue: Ben Kelly <wanderview@chromium.org> > Cr-Commit-Position: refs/heads/main@{#936029} Bug: 1115847,1241188 Change-Id: Ia26acbdd0d7ce6583d9a44f83ed086708657b8bd Reviewed-by: Matt Menke <mmenke@chromium.org> Reviewed-by: Michal Klocek <michal.klocek@qt.io>

This CL contains essentially two changes: 1. The request initiator origin is plumbed through service workers that do `fetch(evt.request)`. In addition to plumbing, this requires changes to how we validate navigation requests in the CorsURLLoaderFactory. 2. Tracks the original destination of a request passed through a service worker. This is then used in the network service to force SameSite=Lax cookies to treat the request as a main frame navigation where appropriate. For more detailed information about these changes please see the internal design doc at: https://docs.google.com/document/d/1KZscujuV7bCFEnzJW-0DaCPU-I40RJimQKoCcI0umTQ/edit?usp=sharing In addition, there is some discussion of these features in the following spec issues: whatwg/fetch#1321 whatwg/fetch#1327 The test includes WPT tests that verify navigation headers and SameSite cookies. Note, chrome has a couple expected failures in the SameSite cookie tests because of the "lax-allowing-unsafe" intervention that is currently enabled. See: https://source.chromium.org/chromium/chromium/src/+/main:third_party/blink/web_tests/TestExpectations;l=4635;drc=e8133cbf2469adb99c6610483ab78bcfb8cc4c76 Bug: 1115847,1241188 Change-Id: I7e236fa20aeabb705aef40fcf8d5c36da6d2798c Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3115917 Reviewed-by: Matt Menke <mmenke@chromium.org> Reviewed-by: Yutaka Hirano <yhirano@chromium.org> Reviewed-by: Nasko Oskov <nasko@chromium.org> Reviewed-by: Łukasz Anforowicz <lukasza@chromium.org> Commit-Queue: Ben Kelly <wanderview@chromium.org> Cr-Commit-Position: refs/heads/main@{#936029} NOKEYCHECK=True GitOrigin-RevId: da0a6501cf321579bd46a27ff9fba1bb8ea910bb

…rkers." This reverts commit da0a6501cf321579bd46a27ff9fba1bb8ea910bb. Reason for revert: Failure on many bots with the following error message: The service worker navigation preload request was cancelled before 'preloadResponse' settled. If you intend to use 'preloadResponse', use waitUntil() or respondWith() to wait for the promise to settle.", source: (0) Original change's description: > Fetch: Plumb request initiator through passthrough service workers. > > This CL contains essentially two changes: > > 1. The request initiator origin is plumbed through service workers > that do `fetch(evt.request)`. In addition to plumbing, this > requires changes to how we validate navigation requests in the > CorsURLLoaderFactory. > 2. Tracks the original destination of a request passed through a > service worker. This is then used in the network service to force > SameSite=Lax cookies to treat the request as a main frame navigation > where appropriate. > > For more detailed information about these changes please see the > internal design doc at: > > https://docs.google.com/document/d/1KZscujuV7bCFEnzJW-0DaCPU-I40RJimQKoCcI0umTQ/edit?usp=sharing > > In addition, there is some discussion of these features in the following > spec issues: > > whatwg/fetch#1321 > whatwg/fetch#1327 > > The test includes WPT tests that verify navigation headers and SameSite > cookies. Note, chrome has a couple expected failures in the SameSite > cookie tests because of the "lax-allowing-unsafe" intervention that is > currently enabled. See: > > https://source.chromium.org/chromium/chromium/src/+/main:third_party/blink/web_tests/TestExpectations;l=4635;drc=e8133cbf2469adb99c6610483ab78bcfb8cc4c76 > > Bug: 1115847,1241188 > Change-Id: I7e236fa20aeabb705aef40fcf8d5c36da6d2798c > Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3115917 > Reviewed-by: Matt Menke <mmenke@chromium.org> > Reviewed-by: Yutaka Hirano <yhirano@chromium.org> > Reviewed-by: Nasko Oskov <nasko@chromium.org> > Reviewed-by: Łukasz Anforowicz <lukasza@chromium.org> > Commit-Queue: Ben Kelly <wanderview@chromium.org> > Cr-Commit-Position: refs/heads/main@{#936029} Bug: 1115847,1241188 Change-Id: I3044a6d20de172b4a8ab7e39a9f26191580003fa No-Presubmit: true No-Tree-Checks: true No-Try: true Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3251692 Auto-Submit: Alan Screen <awscreen@chromium.org> Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com> Commit-Queue: Alan Screen <awscreen@chromium.org> Owners-Override: Alan Screen <awscreen@chromium.org> Cr-Commit-Position: refs/heads/main@{#936125} NOKEYCHECK=True GitOrigin-RevId: a6601b2cf2bb7c0a0ffa3c795a0dbc730ef81d1a

…rkers." This is a reland of da0a6501cf321579bd46a27ff9fba1bb8ea910bb This CL also includes a change to mark the two WPT tests as requiring long timeout durations. On my fast build machine with an opt build they take ~5 seconds each to complete and the default timeout is 10 seconds. On slower bots with debug builds its highly likely that these tests would be marked as timing out. This change gives them a 60 second timeout instead. Original change's description: > Fetch: Plumb request initiator through passthrough service workers. > > This CL contains essentially two changes: > > 1. The request initiator origin is plumbed through service workers > that do `fetch(evt.request)`. In addition to plumbing, this > requires changes to how we validate navigation requests in the > CorsURLLoaderFactory. > 2. Tracks the original destination of a request passed through a > service worker. This is then used in the network service to force > SameSite=Lax cookies to treat the request as a main frame navigation > where appropriate. > > For more detailed information about these changes please see the > internal design doc at: > > https://docs.google.com/document/d/1KZscujuV7bCFEnzJW-0DaCPU-I40RJimQKoCcI0umTQ/edit?usp=sharing > > In addition, there is some discussion of these features in the following > spec issues: > > whatwg/fetch#1321 > whatwg/fetch#1327 > > The test includes WPT tests that verify navigation headers and SameSite > cookies. Note, chrome has a couple expected failures in the SameSite > cookie tests because of the "lax-allowing-unsafe" intervention that is > currently enabled. See: > > https://source.chromium.org/chromium/chromium/src/+/main:third_party/blink/web_tests/TestExpectations;l=4635;drc=e8133cbf2469adb99c6610483ab78bcfb8cc4c76 > > Bug: 1115847,1241188 > Change-Id: I7e236fa20aeabb705aef40fcf8d5c36da6d2798c > Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3115917 > Reviewed-by: Matt Menke <mmenke@chromium.org> > Reviewed-by: Yutaka Hirano <yhirano@chromium.org> > Reviewed-by: Nasko Oskov <nasko@chromium.org> > Reviewed-by: Łukasz Anforowicz <lukasza@chromium.org> > Commit-Queue: Ben Kelly <wanderview@chromium.org> > Cr-Commit-Position: refs/heads/main@{#936029} Bug: 1115847,1241188 Change-Id: Ia26acbdd0d7ce6583d9a44f83ed086708657b8bd Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3251368 Reviewed-by: Matt Menke <mmenke@chromium.org> Reviewed-by: Yutaka Hirano <yhirano@chromium.org> Reviewed-by: Nasko Oskov <nasko@chromium.org> Reviewed-by: Łukasz Anforowicz <lukasza@chromium.org> Auto-Submit: Ben Kelly <wanderview@chromium.org> Commit-Queue: Ben Kelly <wanderview@chromium.org> Cr-Commit-Position: refs/heads/main@{#936560} NOKEYCHECK=True GitOrigin-RevId: 2d916566085e4f09bca93021f2b1650ea6237077

annevk added topic: same-origin policy topic: service workers labels Oct 7, 2021

wanderview mentioned this issue Oct 8, 2021

Request needs to propagate concept of "main frame navigation" #1327

Open

This was referenced Oct 14, 2021

track "site for cookies" on requests and propagate in passthrough service workers #1332

Open

propagate request tainted origin flag and/or URL list through service workers #1335

Closed

chromium-wpt-export-bot mentioned this issue Oct 25, 2021

Fetch: Plumb request initiator through passthrough service workers. web-platform-tests/wpt#31378

Merged

chromium-wpt-export-bot mentioned this issue Oct 28, 2021

Revert "Fetch: Plumb request initiator through passthrough service workers." web-platform-tests/wpt#31425

Merged

chromium-wpt-export-bot mentioned this issue Oct 28, 2021

Reland "Fetch: Plumb request initiator through passthrough service workers." web-platform-tests/wpt#31430

Merged

wanderview added a commit to wanderview/fetch that referenced this issue Oct 29, 2021

Propagate origin through the request constructor. (Fixes whatwg#1321)

4b9ada3

wanderview added a commit to wanderview/fetch that referenced this issue Nov 11, 2021

Propagate origin, tainted origin flag, and URL list through the reque…

05ce682

…st constructor. (Fixes whatwg#1321)

wanderview added a commit to wanderview/fetch that referenced this issue Nov 11, 2021

Propagate origin, tainted origin flag, and URL list through the reque…

400f558

…st constructor. (Fixes whatwg#1321, whatwg#1335)

wanderview added a commit to wanderview/fetch that referenced this issue Nov 22, 2021

Propagate origin and URL list through the request constructor. (Fixes w…

b449bd8

…hatwg#1321, whatwg#1335)

wanderview added a commit to wanderview/fetch that referenced this issue Nov 24, 2021

Propagate origin and URL list through the request constructor. (Fixes w…

76ba6a2

…hatwg#1321, whatwg#1335)

annevk closed this as completed in 8e4553b Dec 1, 2021

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Request constructor should propagate origin internal field #1321

Request constructor should propagate origin internal field #1321

wanderview commented Oct 6, 2021

annevk commented Oct 12, 2021

wanderview commented Oct 12, 2021

annevk commented Oct 12, 2021

wanderview commented Oct 12, 2021

annevk commented Oct 12, 2021

wanderview commented Oct 14, 2021

Request constructor should propagate origin internal field #1321

Request constructor should propagate origin internal field #1321

Comments

wanderview commented Oct 6, 2021

annevk commented Oct 12, 2021

wanderview commented Oct 12, 2021

annevk commented Oct 12, 2021

wanderview commented Oct 12, 2021

annevk commented Oct 12, 2021

wanderview commented Oct 14, 2021