Comment 76 for bug 575160

Looking at this and how OpenSUSE fixed the issue (for "error_code 158 request_code 148 minor_code 7", although the "error_code 172 request_code 152 minor_code 7" error also doesn't seem to come up anymore), I've made a slight change to the debian/rule to use the in-tree cairo, which is exactly what was done in OpenSUSE to resolve the bug.

So far I haven't seen other side-effects to it aside from seamonkey to not crashing when tabs are closed, but this could benefit some more testing.

For those who want to test this in *Maverick*, make sure you can crash seamonkey 2.0.8+build1+nobinonly-0ubuntu1 by doing a search in Google Image search for something, opening a few tabs with the results, then closing all but one tab before closing seamonkey. If you can reproduce the bug this way, then you should be able to test the seamonkey 2.0.8+build1+nobinonly-0ubuntu2~mtrudel1 in my PPA (https://edge.launchpad.net/~mathieu-tl/+archive/ppa, pending build) and be able to do the same process without seamonkey crashing.