Bug 217389 – [Cocoa] The WebContent development XPC service is not allowed to connect to the XPC endpoint in the Networking process

• Status: RESOLVED
• Resolution: FIXED
• Component: WebKit Misc.

[Per Arne Vollan]

The WebContent development XPC service is not allowed to connect to the XPC endpoint in the Networking process, because this requires a private entitlement. The WebContent development XPC service is not allowed to have private entitlements. The WebContent process needs to connect to the XPC endpoint in the Networking process in order to receive the Launch Services database XPC object.

[Per Arne Vollan]

Created attachment 410646 [details]
Patch

[Per Arne Vollan]

Created attachment 410652 [details]
Patch

[Per Arne Vollan]

<rdar://problem/68585497>

[Per Arne Vollan]

Created attachment 410670 [details]
Patch

[Brent Fulgham]

Comment on attachment 410670 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=410670&action=review

r=me

> Source/WebKit/Shared/Cocoa/XPCEndpoint.mm:33
> +#import <wtf/text/WTFString.h>

Should most of these things be inside PLATFORM(MAC), since they only seem to be used in that build case?

[Per Arne Vollan]

(In reply to Brent Fulgham from comment #5)
> Comment on attachment 410670 [details]
> Patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=410670&action=review
> 
> r=me
> 
> > Source/WebKit/Shared/Cocoa/XPCEndpoint.mm:33
> > +#import <wtf/text/WTFString.h>
> 
> Should most of these things be inside PLATFORM(MAC), since they only seem to
> be used in that build case?

Will fix. Thanks for reviewing!

[Per Arne Vollan]

Created attachment 410682 [details]
Patch

[Per Arne Vollan]

Created attachment 410684 [details]
Patch

[Geoffrey Garen]

Seems like this behavior was always broken in WebContent.Development? If so, and we didn't notice, that's pretty serious. Can we add a regression test for this?

[EWS]

Committed r268072: <https://trac.webkit.org/changeset/268072>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 410684 [details].

[Per Arne Vollan]

(In reply to Geoffrey Garen from comment #9)
> Seems like this behavior was always broken in WebContent.Development? If so,
> and we didn't notice, that's pretty serious. Can we add a regression test
> for this?

Yes, this has been broken for WebContent.Development all along. I can look into creating a test for this.

Thanks for reviewing!

[Per Arne Vollan]

(In reply to Per Arne Vollan from comment #11)
> (In reply to Geoffrey Garen from comment #9)
> > Seems like this behavior was always broken in WebContent.Development? If so,
> > and we didn't notice, that's pretty serious. Can we add a regression test
> > for this?
> 
> Yes, this has been broken for WebContent.Development all along. I can look
> into creating a test for this.
> 
> Thanks for reviewing!

More specifically, it has not been an issue with WebContent in Debug mode, but when loading the system installed version of WebContent.Development.