995745 - [OMTC] New Google maps turn black after loading

Status: VERIFIED FIXED

(Core :: Graphics: CanvasWebGL, defect)

Description

[Jim Jeffery not reading bug-mail 1/2/11]

Was pointed out to me this morning that the 'New Google Maps' is fading to 'black' after initial page load. 

STR: 
Enable OMTC 
Load Google Maps and opt into the 'new maps' if your not already
Once the page loads, it turns black

Turning 'off' OMTC fixes the blackness

This is a regression as this was 'fixed' once by bug 975824

Comment 1

[Jim Jeffery not reading bug-mail 1/2/11]

Tested on Win7 x64 with latest hourly tinderbox build cset:

https://hg.mozilla.org/mozilla-central/rev/ebdf2740dc3e

Comment 2

[Jim Jeffery not reading bug-mail 1/2/11]

This may be a slight variation of bug 975824, as there was a re-vamped test added in that bug to catch the issue with compression. 

Either the test is not working, or this is a different case.

Comment 3

[Alice0775 White]

Regression window(m-i):
Good:
https://hg.mozilla.org/integration/mozilla-inbound/rev/5f286e259379
Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) Gecko/20100101 Firefox/31.0 ID:20140401004332
Bad:
https://hg.mozilla.org/integration/mozilla-inbound/rev/d1f8ac35bdd4
Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) Gecko/20100101 Firefox/31.0 ID:20140401005432
Pushlog:
http://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=5f286e259379&tochange=d1f8ac35bdd4

Regressed by: Bug 990338

Comment 4

[Milan Sreckovic [:milan] (needinfo for best results)]

Matt, do you think it could be a regression from the Moz2D work in bug 990338?  If so, what's the timeline you could take a look at this?  It appears to be the same as bug 994969, and e10s is identifying it as their biggest pain in the ... dog-fooding.

Comment 5

[Matt Woodrow (:mattwoodrow)]

This is because the Moz2D code path in CopyableCanvasLayer that I switched us over to does crazy things. It's taking a snapshot of the destination, and abusing copy-on-write behaviour to write directly into the destination. This works on some Moz2D backends, so we got away with it for a while.

roc is going to fix this as part of his current work on clarifying the mutability of SourceSurfaces.

Comment 6

[Benjamin Kerensa [:bkerensa]]

Going to track this

Comment 10

[Matt Woodrow (:mattwoodrow)]

Attachment: Fix CopyableCanvasLayer

Bug 997758 seems to be stuck for the moment so I've extracted the bit that's needed to fix this bug.

Patch is by roc, r=me

Comment 11

[Matt Woodrow (:mattwoodrow)]

https://hg.mozilla.org/integration/mozilla-inbound/rev/2468eeb92c8d

Comment 12

[Ed Morley [:emorley]]

https://hg.mozilla.org/mozilla-central/rev/2468eeb92c8d

Comment 14

[Paul Silaghi, QA [:pauly]]

Reprouced Nightly 2014-04-28, Win 7 x64.
Verified fixed 32.0a1 (2014-05-06).

Comment 15

[Sylvestre Ledru [:Sylvestre]]

Robert, the bug is marked as affecting 31 but we didn't receive any uplift request. So, is the bug exist in 31 or the uplift request is missing? Thanks

Comment 16

[Robert O'Callahan (:roc) (email my personal email if necessary)]

This only affects OMTC so we don't need an uplift to 31 since OMTC isn't on by default there.

Comment 17

[Sylvestre Ledru [:Sylvestre]]

Untracking. Cf comment #16