FWIW, I'm planning to discuss <https://github.com/w3c/webappsec/issues/668#issuecomment-2709995028> a syntax change at next week's WebAppSec meeting, that can help us avoid these compat issues.
On Tue, Feb 25, 2025 at 7:54 PM Yoav Weiss (@Shopify) < yoavwe...@chromium.org> wrote: > > > On Tue, Feb 25, 2025 at 6:08 PM Mike Taylor <miketa...@chromium.org> > wrote: > >> >> On 2/24/25 4:24 PM, Yoav Weiss (@Shopify) wrote: >> >> >> >> On Mon, Feb 24, 2025 at 7:18 PM Mike Taylor <miketa...@chromium.org> >> wrote: >> >>> On 2/21/25 8:33 AM, Yoav Weiss (@Shopify) wrote: >>> >>> >>> On Thursday, February 20, 2025 at 1:56:59 PM UTC+1 Yoav Weiss wrote: >>> >>> >>> On Thursday, February 20, 2025 at 11:47:00 AM UTC+1 Yoav Weiss wrote: >>> >>> Contact emailsyoavwe...@chromium.org >>> >>> Explainerhttps://github.com/w3c/webappsec-subresource-inte >>> grity/pull/129#:~:text=for%20some%20assets.-,require%2Dsr >>> i%2Dfor%20CSP%20directive,-Subresource%2DIntegrity%20 >>> >>> Specificationhttps://github.com/w3c/webappsec-subresource- >>> integrity/pull/129 >>> >>> >>> >>> The feature and PR were discussed >>> <https://github.com/w3c/webappsec/blob/main/meetings/2025/2025-02-19-minutes.md#reviving-require-sri-for> >>> at the WebAppSec WG call. >>> >>> No objection beyond questions on whether we'd need to expand this to >>> cover stylesheets as well. We'd be able to do that in the future (as a >>> separate intent) if needed. >>> >>> Summary >>> >>> The `require-sri-for` directive gives developers the ability to assert >>> that every resource of a given type needs to be integrity checked. If a >>> resource of that type is attempted to be loaded without integrity metadata, >>> that attempt will fail and trigger a CSP violation report. This intent >>> covers the "script" value of this directive. >>> >>> >>> Blink componentBlink>SecurityFeature>ContentSecurityPolicy >>> <https://issues.chromium.org/issues?q=customfield1222907:%22Blink%3ESecurityFeature%3EContentSecurityPolicy%22> >>> >>> TAG reviewhttps://github.com/w3ctag/design-reviews/issues/1048 >>> >>> TAG review statusPending - No response just yet >>> >>> >>> >>> Risks >>> >>> >>> Interoperability and Compatibility >>> >>> On the compatibility front: >>> >>> This directive was already implemented in the past, and there are some >>> developer >>> docs >>> <https://udn.realityripple.com/docs/Web/HTTP/Headers/Content-Security-Policy/require-sri-for> >>> that still describe it. The current PR and implementation did not diverge >>> from the past implementation. >>> >>> >>> If developers deployed the feature in the past and are now relying on it >>> *not >>> really working*, that may result in surprising breakage. The >>> HTTPArchive shows *0.0011% of page responses* (178 out of 15760519) >>> have an existing `require-sri-for` directive. That's an upper bound - only >>> those that enforce scripts, and have no integrity attributes on some >>> scripts may get broken. >>> >>> >>> Doing some more HA digging I found that it's 153 sites, which is not >>> significantly different. >>> I downloaded their URLs >>> <https://docs.google.com/spreadsheets/d/1NlFHLytc8lQcdP5FXXDltKEPQVE0e8oyjw9k1S-9KPI/edit?usp=sharing> >>> and >>> started going to these sites with the feature enabled. >>> Of those 153, 22 had any blocked assets, 9 had broken functionality or >>> layout and 1 had missing ads. >>> Non-visiblity broken but blocked sites mostly had their analytics >>> blocked. >>> >>> Extrapolating that data brings us to 0.000158% for any blocked assets, >>> and 0.000065% for broken functionality. >>> >>> I'm planning to reach out to the broken sites and make them aware of >>> this change. Many of them seem to be coming from a single provider (similar >>> site and breakage). >>> >>> >>> I also found ~3500 sites that have the `require-sri-for` string in their >>> response bodies (and hence may have it applied). >>> I put together a script that so far scanned ~1800 of them and found no >>> blocked assets. So, it seems like the risk is very low on that front. >>> >>> Thanks, I appreciate you digging in to understand the possible risks. My >>> understanding of the compat risk goes something like (please let me know if >>> I'm missing something): >>> >>> 1. This feature never really shipped, but was implemented behind a flag. >>> >>> 2. Early adopter developers (or menu framework authors?) added >>> require-sri-for for some scripts that they wanted to lock down >>> >> Tiny correction: they added it to the document's CSP, not specific >> scripts. >> >>> (to prevent 3rd-party attackers from modifying them, etc). >>> >>> 3. Now, you actually ship the feature. >>> >>> That means the risks are: >>> >>> a) Some CDN was compromised at some point, and now some sketchy scripts >>> will fail to load. Seems like that's security positive, even if it >>> surprises users or developers. >>> >>> b) Perhaps more likely, a page was redesigned and they updated their >>> analytics provider but didn't remember to add hashes. Now some things don't >>> work. >>> >> I suspect it's >> c) they added the header but never actually tested with the feature >> enabled, as it never shipped. >> >> It seems like they added the CSP header, but never added an "integrity" >> attribute to many/most of their scripts. >> >> >>> From your sheet (which is great, thanks), it seems like largest impact >>> is busted menus. Is this a single library? Or common pattern? >>> >>> It seems like a common provider. 6 out of the 9 sites with issues are >> Canadian health/edu related sites. >> >> Breaking health/edu-releated sites is not good... >> > Indeed, although I'm not sure how load-bearing they are.. > >> When broken, is it cosmetic, or are the links in the menu still >> accessible? (I see you're gathering contact info - let me know if you need >> help with that.) >> > It seems like the desktop view is functional, but the mobile view's > hamburger menu is not working (at least in some cases). > >> Assuming we can sort out the menu breakage somehow, I think for the rest >> - the best we can do is roll it out and be ready to killswitch if needed. >> >> >>> >>> >>> >>> >>> *Gecko*: No signal (https://github.com/mozilla/st >>> andards-positions/issues/1173) >>> >>> *WebKit*: No signal (https://github.com/WebKit/sta >>> ndards-positions/issues/458) >>> >>> *Web developers*: Shopify is interested in this. I suspect PCIv4 >>> <https://docs.google.com/document/d/1RcUpbpWPxXTyW0Qwczs9GCTLPD3-LcbbhL4ooBUevTM/edit?tab=t.0> >>> would >>> make some developers interested in making sure their documents' scripts >>> have complete integrity checks. >>> >>> *Other signals*: >>> >>> WebView application risks >>> >>> Does this intent deprecate or change behavior of existing APIs, such >>> that it has potentially high risk for Android WebView-based applications? >>> >>> None >>> >>> >>> Debuggability >>> >>> None >>> >>> >>> Will this feature be supported on all six Blink platforms (Windows, Mac, >>> Linux, ChromeOS, Android, and Android WebView)?Yes >>> >>> Is this feature fully tested by web-platform-tests >>> <https://chromium.googlesource.com/chromium/src/+/main/docs/testing/web_platform_tests.md> >>> ?Yes >>> >>> https://wpt.fyi/results/content-security-policy/tentative/ >>> require-sri-for?label=experimental&label=master&aligned >>> <https://chromium-review.googlesource.com/c/chromium/src/+/5877633> >>> >>> >>> Flag name on about://flagsNone >>> >>> Finch feature nameCSPRequireSRIFor >>> >>> Requires code in //chrome?False >>> >>> Estimated milestonesShipping on desktop135DevTrial on desktop134Shipping >>> on Android135DevTrial on Android134Shipping on WebView135 >>> >>> Anticipated spec changes >>> >>> Open questions about a feature may be a source of future web compat or >>> interop issues. Please list open issues (e.g. links to known github issues >>> in the project for the feature specification) whose resolution may >>> introduce web compat/interop risk (e.g., changing to naming or structure of >>> the API in a non-backward-compatible way). >>> >>> >>> None >>> >>> Link to entry on the Chrome Platform Statushttps://chromestatus.com/ >>> feature/5090023365672960?gate=5186570942152704 >>> >>> Links to previous Intent discussionsIntent to Prototype: https://groups. >>> google.com/a/chromium.org/d/msgid/blink-dev/CAOmohSJUygAmobR9dRkDr% >>> 3DBWQ1h5hv2Lj3WUFN31QZF360A47A%40mail.gmail.com >>> >>> >>> >>> >>> This intent message was generated by Chrome Platform Status >>> <https://chromestatus.com/>. >>> >>> -- >>> You received this message because you are subscribed to the Google >>> Groups "blink-dev" group. >>> To unsubscribe from this group and stop receiving emails from it, send >>> an email to blink-dev+unsubscr...@chromium.org. >>> To view this discussion visit >>> https://groups.google.com/a/chromium.org/d/msgid/blink-dev/8d3107ca-61cc-47f6-badd-8bc6a1f30145n%40chromium.org >>> <https://groups.google.com/a/chromium.org/d/msgid/blink-dev/8d3107ca-61cc-47f6-badd-8bc6a1f30145n%40chromium.org?utm_medium=email&utm_source=footer> >>> . >>> >>> -- You received this message because you are subscribed to the Google Groups "blink-dev" group. To unsubscribe from this group and stop receiving emails from it, send an email to blink-dev+unsubscr...@chromium.org. To view this discussion visit https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAOmohS%2BS-%2Bx1VAmt%3Dg31Xz0BRcZ-nFhsJowkSshzA5H7ee2TiQ%40mail.gmail.com.
