1317759 - Remove use of LegacyIsCallerChromeOrNativeCode usage in Event::GetScreenCoords

Status: RESOLVED FIXED

(Core :: DOM: Events, defect, P2)

Description

[Boris Zbarsky [:bzbarsky]]

This was added in bug 418986 but is pretty bizarre, because some of the callers can be pretty randomly either during script execution or not.

I think the right thing to do is to remove this check and instead have the check in the places where we really care: MouseEvent::ScreenX/Y and Touch::ScreenX/Y.  Note that the latter is sort of inconsistent right now depending on whether page script was on stack when Touch::InitializePoints got called, so may well be leaking fingerprinting information when we don't want it to be.

Comment 1

[Boris Zbarsky [:bzbarsky]]

The callers of the XPCOM event screenX/Y getters are:

1) nsXULPopupListener::LaunchPopup
2) ChromeTooltipListener::MouseMove
3) nsXULTooltipListener::MouseMove
4) nsXULTooltipListener::CheckTreeBodyMove
5) nsBaseDragService::InvokeDragSessionWithImage
6) nsBaseDragService::InvokeDragSessionWithSelection

and I think those all want unsanitized screen coords.

Comment 2

[Boris Zbarsky [:bzbarsky]]

Attachment: Remove LegacyIsCallerChromeOrNativeCode use in Event::GetScreenCoords in favor or caller type checks at the entrypoints where we really care about pretending that our screen is client-area sized

Comment 3

[Olli Pettay [:smaug][[e-mail]]]

Comment on attachment 8811023 [details] [diff] [review]
Remove LegacyIsCallerChromeOrNativeCode use in Event::GetScreenCoords in favor or caller type checks at the entrypoints where we really care about pretending that our screen is client-area sized

Ah, the touch part if because of Touch::InitializePoints.

ResistFingerprinting() behavior is still odd, but this patch doesn't make it any worse.

Comment 4

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/3ea1cc459783
Remove LegacyIsCallerChromeOrNativeCode use in Event::GetScreenCoords in favor or caller type checks at the entrypoints where we really care about pretending that our screen is client-area sized.  r=smaug

Comment 5

[Wes Kocher (:KWierso) (Not reading bugmail; email directly if needed)]

https://hg.mozilla.org/mozilla-central/rev/3ea1cc459783