703890 - crash involving feDisplacement map in SVG

Status: RESOLVED WORKSFORME

(Core :: Graphics, defect)

Description

[david.dailey]

User Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:8.0) Gecko/20100101 Firefox/8.0
Build ID: 20111104165243

Steps to reproduce:

https://crash-stats.mozilla.com/report/index/bp-4e4981d8-1994-4198-b69d-215282111119
for web page at 
http://srufaculty.sru.edu/david.dailey/svg/newstuff/filterDisplacementMap4.svg 


Actual results:

Each time I connect to that page, FF8 crashes (hard) -- not just a hang.

The page works in Opera and IE/ASV


Expected results:

Image should be distorted as per animated reflected radial gradient.

Comment 1

[Robert Longson [:longsonr]]

If you turn hardware acceleration off Options->Options->Advanced and look for the Use Hardware acceleration check box does it still crash?

It doesn't crash for me on Trunk on Vista but I've not got a hardware accelerated graphics driver and it looks like you do.

Comment 2

[Boris Zbarsky [:bzbarsky]]

Looks like a null deref in the cairo d2d backend?

Comment 3

[Marcia Knous [:marcia]]

Can repo using Build identifier: Mozilla/5.0 (Windows NT 6.1; Win64; x64; rv:11.0a1) Gecko/20111119 Firefox/11.0a1.

Does not crash when I pref off hardware acceleration.

Comment 4

[david.dailey]

Not sure if this is the same bug or not:
http://cs.sru.edu/~svg/Chapter05/G05xx47.svg
I am pretty sure it worked in Firefox at the time the book was written three years ago.
It works as it should in IE11, in Opera12. Chrome does something totally silly, but it seems impossible to report SVG bugs for Chrome these days.
Let me know if I should open a new bug.

Comment 5

[Tom Klein]

According to mozregression this was last bad on 2012-5-22 and first good on 2012-5-23 (so I'm guessing was fixed by bug 756454).  It no longer crashes, but now asserts, which should be fixed by bug 1212954.

After that, I think it still won't be expected to work since it uses feImage to reference an svg fragment (bug 455986), and maybe because the outer svg of that fragment has a percentage width and height (bug 1097856)...