STR (1) Run firefox -no-remote -chrome chrome://global/content/test-ipcbrowser.xul (2) Load http://people.mozilla.com/~cjones/jsuri.html Two frames with the text "frame1" and "frame2" should appear. Instead, the frames are blank and this assertion appears twice on stderr ###!!! ASSERTION: Trying to initialize PSM/NSS in a non-chrome process!: 'Error', file /home/cjones/mozilla/mozilla-central/security/manager/ssl/src/nsNSSComponent.cpp, line 307 This works in fennec, for reasons unknown. I recall jdm wrestling with this for data: URIs a while back. data: URIs indeed do work in <browser remote>.

"totally guaranteed safe fix" per IRC discussion. Not sure I did this in the right place. I see call in this file to mStreamChannel->Open(), which would presumably need a similar fix if it pops up.

Comment on attachment 496403 [details] [diff] [review] Don't run the URI classifier on javascript: URIs I think it would be safer to just mask this out of aLoadFlags at the top of nsJSChannel::SetLoadFlags. That would make sure that if someone calls GetLoadFlags then SetLoadFlags on us in various places the right things still happen.

Carrying over r+. Requesting approval because this blocks bug 615386 currently, and would require ugly workarounds to unblock.