Resource Load Statistics: Check both origins and cookieHostNames for domain matches in data removal

• Component: WebKit2
• Status: RESOLVED FIXED

[John Wilander]

WebsiteDataRecords either have origins, cookieHostNames, or both depending on their WebsiteDataTypes. We need to check both for a match during data record removal.

[Radar WebKit Bug Importer]

<rdar://problem/31573091>

[John Wilander]

Created attachment 306890 [details]
Patch

[Build Bot]

Comment on attachment 306890 [details]
Patch

Attachment 306890 [details] did not pass mac-wk2-ews (mac-wk2):
Output: http://webkit-queues.webkit.org/results/3522314

New failing tests:
webrtc/negotiatedneeded-event-addStream.html

[Build Bot]

Created attachment 306892 [details]
Archive of layout-test-results from ews106 for mac-elcapitan-wk2

The attached test failures were seen while running run-webkit-tests on the mac-wk2-ews.
Bot: ews106  Port: mac-elcapitan-wk2  Platform: Mac OS X 10.11.6

[youenn fablet]

(In reply to Build Bot from comment #3)
> Comment on attachment 306890 [details]
> Patch
> 
> Attachment 306890 [details] did not pass mac-wk2-ews (mac-wk2):
> Output: http://webkit-queues.webkit.org/results/3522314
> 
> New failing tests:
> webrtc/negotiatedneeded-event-addStream.html

Crash is unrelated to this patch.
This is a new regression tracked in bug 170754

[John Wilander]

Testing requires https://bugs.webkit.org/show_bug.cgi?id=170798.

[Brent Fulgham]

Comment on attachment 306890 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=306890&action=review

I think this looks good, but we need a WK2 reviewer.

> Source/WebKit2/UIProcess/WebsiteData/WebsiteDataRecord.cpp:120
> +bool WebsiteDataRecord::validForTopPrivatelyControlledDomain(const String& topPrivatelyControlledDomain)

It seems like this could be 'const'.

> Source/WebKit2/UIProcess/WebsiteData/WebsiteDataStore.cpp:501
> +    fetchData(dataTypes, fetchOptions, [topPrivatelyControlledDomains, completionHandler, this](auto existingDataRecords) {

Since ebsiteDataStore::fetchData does a WTFMove on the records it passes to this completion handler, it seems like "auto existingDataRecords" should really be "auto&& existingDataRecords".

> Source/WebKit2/UIProcess/WebsiteData/WebsiteDataStore.cpp:507
> +                    matchingDataRecords.append(dataRecord);

Maybe this could be WTFMove'd into matchindDataRecords? Might need to make "auto& dataRecord" -> "auto&& dataRecord"? Not sure if that works.

> Source/WebKit2/UIProcess/WebsiteData/WebsiteDataStore.cpp:508
> +                    domainsWithMatchingDataRecords.append(topPrivatelyControlledDomain);

Maybe this could be WTFMove'd, too?

[John Wilander]

Created attachment 307325 [details]
Patch

[John Wilander]

Thanks, Brent! I could fix all the things you mentioned except for the last WTFMove of topPrivatelyControlledDomain. I suspect the compiler gets upset because I'm moving an element in the set I'm iterating over but if you can spot a mistake I've made I'm happy to try again I had to introduce move semantics for that vector all the way from the call in WebResourceLoadStatisticsStore::removeDataRecords().

[Alex Christensen]

Comment on attachment 307325 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=307325&action=review

r=me

> Source/WebKit2/UIProcess/WebsiteData/WebsiteDataRecord.cpp:132
> +    for (auto& dataRecordOriginData : origins) {

Can this be const auto&?

> Source/WebKit2/UIProcess/WebsiteData/WebsiteDataStore.cpp:507
> +                if (dataRecord.validForTopPrivatelyControlledDomain(topPrivatelyControlledDomain)) {
> +                    matchingDataRecords.append(WTFMove(dataRecord));

Maybe matchesTopPrivatelyControlledDomain would be a better name than validForTopPrivatelyControlledDomain.

[John Wilander]

Created attachment 307441 [details]
Patch for landing

[WebKit Commit Bot]

Comment on attachment 307441 [details]
Patch for landing

Clearing flags on attachment: 307441

Committed r215491: <http://trac.webkit.org/changeset/215491>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.