back to article Mozilla to cordon off plugin crashes with Firefox 3.6.4 beta

Mozilla released a public beta of Firefox 3.6.4 yesterday, which also debuted “Lorentz” - a project that the open source outfit is hoping will stablise browsing for Windows and Linux users when Adobe’s Flash, Apple’s Quicktime and Microsoft’s Silverlight plugins crash. The browser maker currently claims that 400 million people …

COMMENTS

This topic is closed for new posts.
Silver badge

I cured Adobe, Microsoft and Apple headaches ages ago. And google, for that matter.

It was easy. I banned 'em from my personal network.

It hasn't affected my "computing experience" at all, near as I can tell.

Try it; you might like it.

0
1

At last, someone thinking beyond their own nose.

“If a crash in one of these plugins happens, Firefox will continue to run and users will be able to submit a crash report before reloading the page to try again.”

Oh,so reloading a page doesn't constitute a crash anymore. Please, exactly what everyone complains about with IE(insert version number).

"The plugins will run in their own processes, which should in theory isolate them from a Firefox session if Quicktime, Silverlight or Flash go titsup when a user is browsing the web with the 3.6.4 beta."

Oh, really, so the IE(insert no7 or 8 here), model of tabbed browsing is so fucked up that Mozilla need to emulate it. Purely on the basis that each plugin run on a sepereate process, just like, oh yes don't mention the M$ or IE word. Or perhaps I have that wrong, perhaps it's Linux that's so fucked that the M$ model needs to be followed.

Or , perhaps, there are a few more enlightened people out there, who do actually realise, it doesn't matter how brave your product is, some fuckwit will try to break it.

0
8
Silver badge

Plugin crashes are easy to solve

Just load the shared object then fork() off its execution into a separate child process. Ok , Winblows still can't do fork() even after 20 years but there's no reason they couldn't have done this years ago on the unix releases of the browser.

0
1
Pint

Hello....

...what's all this testing business? You'd think they were trying to get something working properly! Tch! Publish and be damned! ( That's what the big boys do! )

1
0
Go

Do This Right: URL-based Sandboxes

I suggest Mozilla use a sandbox for each www host. That would imply that www.rbs.co.uk would run in a different sandbox than www.greatPr0n.ru.

If www.greatPr0n.ru injects majware into the browser, the other session with RBS would not be affected. I don't know whether this is technically possible with Windows, but certainly it is possible with Linux Security Modules (LSM).

http://en.wikipedia.org/wiki/Linux_Security_Modules

0
0

about time

even non techie friends are moving away from Firefox because it just keeps crashing. If I could get the add ins I use on another browser I too would move.

0
1

I partially agree with Neal

Why separate processes, with the burden of inter-process messaging? Excuse my ignorance of C, but can't they just put try-catch blocks around each call to a plug-in? and if the plug-in has an unhandled exception, just terminate *that* plug-in, not the whole page?

I've a banner ad does something nasty, why should the whole page crash?

0
0
Stop

@try/catch

This would probably work with Java, but with C++ you can do something like this:

try{

char* uninitPtr;

uninitPtr[6263]='X';

}

catch(...)

{}

And this will probably overwrite some totally different memory without you realizing it, before some totally different code reads that memory.

0
0
Silver badge
WTF?

@voodo trucker

"Why separate processes, with the burden of inter-process messaging"

Its simpler than controlling competing threads. You don't have to worry about locking for a start.

"Excuse my ignorance of C, but can't they just put try-catch blocks around each call to a plug-in? and if the plug-in has an unhandled exception,"

Err, C/C++ isn't some fluffy high level language where anything nasty can be caught by an exception. If something in a process corrupts memory or attempts to write where it shouldn't then the OS will terminate the process, end of. Sure , in unix you can catch the segmentation fault and bus error signals but generally there's little point as theres no clean way of recovering from them anyway since you have no idea of whats been corrupted.

0
0
This topic is closed for new posts.

Forums