877700 - Invalid background copy on Git branching site

Status: RESOLVED FIXED

(Core :: Graphics: Layers, defect)

Description

[Bas Schouten (:bas.schouten)]

Attachment: Bind the source region to the area of the paren layer surface.

In the demo mentioned below we get a container layer with a visible region larger than its parent. This causes a problem in the D3D10 layer manager where CopySubresourceRegion tries to copy invalid data. This patch fixes that.

Comment 1

[Robert Kaiser]

Given that this blocks one of our long-going top-5 crashers and might fix a substantial portion of that, I'd like to track it somewhere - if only to get things actually moving here.

Comment 2

[Robert Kaiser]

Oh, and from what I remember, this crash signature also significantly regressed in 25, even though it has been quite high even before.

Comment 3

[Bas Schouten (:bas.schouten)]

I believe Joe has landed this patch yesterday.

Comment 4

[Joe Drew (not getting mail)]

I tried to, but Inbound was closed. :(

Comment 5

[Bas Schouten (:bas.schouten)]

(In reply to Joe Drew (:JOEDREW! \o/) from comment #4)
> I tried to, but Inbound was closed. :(

Ah, yes, same here last 4 times I tried to land it. I'll try again tomorrow.

Comment 6

[Bas Schouten (:bas.schouten)]

https://hg.mozilla.org/integration/mozilla-inbound/rev/de0950887682

Comment 7

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/de0950887682

Comment 8

[Benjamin Smedberg]

Is this safe for uplift to older branches as well? 23 is too far gone, but if it helps I'd love to have this for 24.

Comment 9

[Bas Schouten (:bas.schouten)]

(In reply to Benjamin Smedberg  [:bsmedberg] from comment #8)
> Is this safe for uplift to older branches as well? 23 is too far gone, but
> if it helps I'd love to have this for 24.

I strongly believe it is, but having it sit on nightly for 1 or 2 days might still be a good idea to make sure I didn't overlook some side-effect.

Comment 10

[Lukas Blakk [:lsblakk] use ?needinfo]

We could look at an uplift nomination if this is low risk enough, ideally before FF24 gets to Beta.

Comment 11

[Robert Kaiser]

As Scoobidiver says in bug 805406 that this patch seems to fix 2/3 of the crashes with this PushClipsToDT signature, I think we should try to get this uplifted to 24.

Comment 12

[bhavana bajaj [:bajaj]]

needsinfo on Bas to help with uplift here.

Comment 13

[Bas Schouten (:bas.schouten)]

If this is making the crash numbers go down -and- it seems stable on nightly so far, we should definitely uplift this!

Comment 14

[Bas Schouten (:bas.schouten)]

Comment on attachment 756013 [details] [diff] [review]
Bind the source region to the area of the paren layer surface.

[Approval Request Comment]
Bug caused by (feature/regressing bug #): Windows Update
User impact if declined: Crashes on websites with certain layer trees
Testing completed (on m-c, etc.): On m-c
Risk to taking this patch (and alternatives if risky): Almost none
String or IDL/UUID changes made by this patch: None

Comment 15

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/releases/mozilla-aurora/rev/a0f4fd0aa9b4

Comment 16

[u279076]

Does this need QA? If so, can you please advise on a test?

Comment 17

[Bas Schouten (:bas.schouten)]

I doubt this is still relevant. Sorry!