Bug 83325 – Get rid of the useless flag PREEMPT_GEOLOCATION_PERMISSION

• Component: WebCore Misc.
• Status: RESOLVED FIXED

[Benjamin Poulain]

The flag PREEMPT_GEOLOCATION_PERMISSION was added in
http://trac.webkit.org/changeset/63742 around some code of
Geolocation.

I could not find any case where PREEMPT_GEOLOCATION_PERMISSION is not
used. Unless someone does not use this flag, I plan to remove the flag
and the related useless code.

[Benjamin Poulain]

Created attachment 135933 [details]
Patch

[Ryosuke Niwa]

Comment on attachment 135933 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=135933&action=review

> Source/WebCore/ChangeLog:23
> +        (WebCore::Geolocation::positionChanged): The case (!isAllowed()) was there
> +        to support granting the authorization for WTF_USE_PREEMPT_GEOLOCATION_PERMISSION.

You mean when PREEMPT_GEOLOCATION_PERMISSION is disabled?

[Benjamin Poulain]

> > Source/WebCore/ChangeLog:23
> > +        (WebCore::Geolocation::positionChanged): The case (!isAllowed()) was there
> > +        to support granting the authorization for WTF_USE_PREEMPT_GEOLOCATION_PERMISSION.
> 
> You mean when PREEMPT_GEOLOCATION_PERMISSION is disabled?

Yes indeed. I will reword that when landing.

[Benjamin Poulain]

Committed r113505: <http://trac.webkit.org/changeset/113505>

[Alexey Proskuryakov]

Was there a specific reason to use CRASH instead of ASSERT? We usually only do that for out of memory conditions, and use ASSERT.

[Benjamin Poulain]

> Was there a specific reason to use CRASH instead of ASSERT? We usually only do that for out of memory conditions, and use ASSERT.

This particular check comes from a hardening patch I previously did. At the time, Adam suggested it would be better to crash so we learn if that happens I the wild. At the time, I did not aggree because there were too many corner case for granting permission.

Now that the permission system is greatly simplified, I think CRASH() is reasonable.