1139270 - [EME] Setting a key to output-downscaled causes it to not be playable

Status: RESOLVED FIXED

(Core :: Audio/Video, defect)

Description

[Chris Pearce [:cpearce (Not reading bugmail)]]

When a CDM/GMP sets a keyId to be output-downscaled, we're supposed to still use the CDM to decrypt content for that key. We're not doing that however, we're stopping sending input to the CDM.

I think this is because we only send input to the CDM if the key is in usable state; we should instead send if the key is usable *or* output-downscaled.

JW: This is blocking Adobe, can you take it please?

Comment 1

[Chris Pearce [:cpearce (Not reading bugmail)]]

To be clear, the fail case/STR is:

CDM plays video encrypted with a keyId, during playback sets keyId as "output-downscaled" status, and Gecko refuses to send more input. Gecko should still send input for "output-downscaled" status keyIds.

Comment 2

[JW Wang [:jwwang] [:jw_wang]]

Not sure if EMEMediaDataDecoderProxy::Input() is the right place to look at? It looks like we can return true for CDMCaps::AutoLock::IsKeyUsable() when a key is either kGMPUsable or kGMPOutputDownscaled so that SamplesWaitingForKey won't hold samples from Input().

Comment 3

[Chris Pearce [:cpearce (Not reading bugmail)]]

I suspect that'd do it. You should test it; just temporarily modify set a timer in gmp-clearkey which marks keys as output-downscaled.

Comment 4

[JW Wang [:jwwang] [:jw_wang]]

Attachment: 1139270_testcase.patch

A test case to send "output-downscaled" during playback.

Comment 5

[JW Wang [:jwwang] [:jw_wang]]

test steps:
1. apply 1139270_testcase.patch
2. ./mach run http://people.mozilla.org/~cpearce/mse-clearkey/

You will see video stuck after a while.

Comment 6

[JW Wang [:jwwang] [:jw_wang]]

Attachment: 1139270_handle_output_downscaled-v1.patch

Mark kGMPOutputDownscaled keys "able to decrypt" so we can still decrypt samples with keys that are kGMPOutputDownscaled.

Comment 7

[Chris Pearce [:cpearce (Not reading bugmail)]]

Comment on attachment 8572505 [details] [diff] [review]
1139270_handle_output_downscaled-v1.patch

Review of attachment 8572505 [details] [diff] [review]:
-----------------------------------------------------------------

We should get a testcase somehow so that we don't regress this, but first let's land this!

Comment 8

[Chris Pearce [:cpearce (Not reading bugmail)]]

Must uplift.

Comment 9

[Chris Pearce [:cpearce (Not reading bugmail)]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/9943568857c5

Comment 10

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/9943568857c5

Comment 11

[Chris Pearce [:cpearce (Not reading bugmail)]]

Comment on attachment 8572505 [details] [diff] [review]
1139270_handle_output_downscaled-v1.patch

Approval Request Comment
[Feature/regressing bug #]: EME
[User impact if declined]: EME videos could stop playback mid stream.
[Describe test coverage new/current, TreeHerder]: Local manual testing performed. Currently this only affects Adobe's EME plugin, and it's difficult to write a standalone test.
[Risks and why]: Low, this patch makes us more likely to play, not less likely. 
[String/UUID change made/needed]: None.

Comment 12

[Lawrence Mandel [:lmandel] (use needinfo)]

Comment on attachment 8572505 [details] [diff] [review]
1139270_handle_output_downscaled-v1.patch

This change is isolated to EME and doesn't look to pose a risk to other video playback paths. Beta+ Aurora+

Comment 13

[Chris Pearce [:cpearce (Not reading bugmail)]]

https://hg.mozilla.org/releases/mozilla-aurora/rev/582136af3821

Comment 14

[Chris Pearce [:cpearce (Not reading bugmail)]]

https://hg.mozilla.org/releases/mozilla-beta/rev/a73d01c8d332