1998698 - RFP: orientation spoofs and iframes

Status: RESOLVED FIXED

(Core :: CSS Parsing and Computation, defect)

Description

[Thorin [:thorin]]

Attachment: test.png

In Bug 1607032 and Bug 1918202 we limited device orientation spoofs to primary results (based on our screen dimension spoofs) - there is no additional entropy as it's still based on our spoofed values - it was to make our spoof "spec compliant" and benign (non-paradoxical). Note that android uses a different angle value to desktops

This is similar to Bug 1885101 where we had to patch iframes to return the spoofed screen etc dimensions: i.e Match screen and window properties with top window.

Something similar is happening here - we're not adjusting the iframe's expected value based on our spoofs

• On desktop it always seems to report {"mozOrientation":"portrait-primary","orientation.angle":90,"orientation.type":"portrait-primary"} regardless of the iframe's dimensions

STR:

• control
  • disable RFP and have device on desktop in landscape mode
  • run PoC
  • everything should match e.g. {"mozOrientation":"landscape-primary","orientation.angle":0,"orientation.type":"landscape-primary"}
• enable RFP
  • make the window landscape (so RFP spoofs as landscape, which you already are)
  • run PoC
  • document (top window) is fine (landscape), but the three different sizes iframes all report portrait

Comment 1

[Thorin [:thorin]]

Attachment: device_orientation.html

Comment 2

[BugBot [:suhaib / :marco/ :calixte]]

The severity field is not set for this bug.
:boris, could you have a look please?

For more information, please visit BugBot documentation.

Comment 3

[Pier Angelo Vendrame]

I think I found the problem: GetTopInnerSizeForRFP() needs to always be called with the top browsing context, i.e., bc->Top()->GetTopInnerSizeForRFP() instead of bc->GetTopInnerSizeForRFP(), which is what the various orientation calls do.
I could send a patch to change all of them (as well as a test based on Thorin's attachment), but I don't understand whether the name is just a bit unlucky, or if a mechanism for actually setting the actual top size is missing.
Emilio: I've seen that you commented as a reviewer on Bug 1885101 where this was added, perhaps can you answer my question?

Comment 4

[Emilio Cobos Álvarez [:emilio]]

Hmm, agreed that the name is a bit unlucky. What other fields do is adding a slightly more convenient accessor that goes through Top(), e.g. here.

Maybe do that? Maybe combined with a rename of TopInnerSizeForRFP to something else, but I don't have any good suggestion off-hand.

Comment 5

[Pier Angelo Vendrame]

Perfect, thank you!

Comment 6

[Pier Angelo Vendrame]

Attachment: Bug 1998698 - Fix iframe orientation spoofing with RFP. r?emilio

Also, rename the TopInnerSizeForRFP for better clarity.

Comment 7

[Pulsebot]

Pushed by [e-mail]:
https://github.com/mozilla-firefox/firefox/commit/93f728bd3dd9
https://hg.mozilla.org/integration/autoland/rev/6e1cf6623972
Fix iframe orientation spoofing with RFP. r=emilio,layout-reviewers

Comment 8

[Serban Stanca [:SerbanS]]

https://hg.mozilla.org/mozilla-central/rev/6e1cf6623972

Comment 9

[Thorin [:thorin]]

verified fixed on desktop, can't test on android yet, which has different angles