1216200 - inset box-shadow renders incorrectly

Status: VERIFIED FIXED

(Core :: Graphics, defect)

Description

[Daniel Holbert [:dholbert]]

Attachment: testcase 1

STR:
 1. Load attached testcase.

EXPECTED RESULTS:
 Shadow at lower-right of bordered box.

ACTUAL RESULTS:
 Weird disjoint grid of shadows.

Firefox 41 (release) & chrome each give EXPECTED RESULTS.
Firefox Nightly 44.0a1 (2015-10-19) gives ACTUAL RESULTS.

This seems similar to bug 1213545, but the patch there (attachment 8674497 [details] [diff] [review]) doesn't fix the bug. (though it does change the rendering a bit)  But feel free to dupe if we want to extend/adjust that bug's patch to fix this case, too.

Comment 1

[Daniel Holbert [:dholbert]]

Looks like this went from drawing the correct thing to drawing *nothing* with this regression range:

https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=11dc79e232110ba6de5179e46dfbda77b52a88c3&tochange=4313752f69956ae248bd4e7ff3913c8dd4252698

(So, Bug 1188075)

Then a few days ago, this went back to drawing something (hooray! but it draws the wrong thing, per this bug).  That change was likely a result of bug 1211363.

Comment 2

[Daniel Holbert [:dholbert]]

Attachment: screenshot of nightly (top) vs. release (bottom)

Comment 3

[Alice0775 White]

And also affect to a complete theme, See http://forums.mozillazine.org/viewtopic.php?f=23&t=2968697

Comment 4

[u52928]

I think I've got this problem on my system, but it's also related to the theme I'm using.

Please see http://forums.mozillazine.org/viewtopic.php?f=23&t=2968697

I've filed Bug 1216220, but now it may be a dupe.

Comment 5

[u52928]

Thanks Alice.

Comment 7

[Mason Chang [Inactive] [:mchang]]

Attachment: Account for negative spread values in min inset box shadows

This accounts for negative spread values when calculating the min inset blur. If we have a blur radius and a negative spread value, we would create a min inset blur size that is too small to actually blur correctly, and copy over incorrect parts to the final destination. This patch ensures that even if we have a negative spread radius, we still create a surface big enough to do the blur.

Comment 8

[Pulsebot]

https://hg.mozilla.org/integration/mozilla-inbound/rev/eddc6be3bd52

Comment 9

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/eddc6be3bd52

Comment 10

[Markus Stange [:mstange]]

We also need a test here.

Comment 11

[Mason Chang [Inactive] [:mchang]]

Attachment: Reftest with negative spread value

Comment 12

[Markus Stange [:mstange]]

Comment on attachment 8678313 [details] [diff] [review]
Reftest with negative spread value

You can just use about:blank as the reference, so that you don't need to add an empty -ref.html file.

Comment 13

[Paul Silaghi, QA [:pauly]]

Verified fixed FF 44.0a1 (2015-10-25) Win 7, Ubuntu 14.04, OS X 10.10.5

Comment 14

[Mason Chang [Inactive] [:mchang]]

Successful try for reftest - https://treeherder.mozilla.org/#/jobs?repo=try&revision=3a1c98660200

Comment 15

[Pulsebot]

https://hg.mozilla.org/integration/mozilla-inbound/rev/0289fda8ff6b

Comment 16

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

https://hg.mozilla.org/mozilla-central/rev/0289fda8ff6b

Comment 17

[Matty Balaam]

This may be related, let me know if not and I will file a new bug, but inset box-shadows are now displaying completely differently in Firefox than other browsers from version 44. I am on Windows 7.


I have a test case online: http://codepen.io/MattyBalaam/pen/ZQgdmP

Comment 18

[Matty Balaam]

Aplogies, I have found this issue mentioned already. https://bugzilla.mozilla.org/show_bug.cgi?id=1227216