Since the release of Firefox 3 I’ve gotten this question over and over again:
Why does Firefox 3 make it so hard to use a self-signed certificates?
It was pointed out to me, rightfully so, that this tends to affect the free software community more than others because it’s easier for technical people to set up a self-signed certs. True, but we did it for a good reason.
Johnathan has a post up that talks about SSL in Firefox 3, a post that I hope will put some of those questions to rest. He talks about why we did it, what the change was and how it’s actually an improvement and brings better security to the web. It’s worth a read.
-
Anyone currently using a self-signed certificate should go get a certificate from StartCom instead. It costs nothing, and most non-IE browsers accept it. IE users will continue to see a warning, but they get warned for self-signed certificates anyway.
-
I’m quite simply not believing this positively effects security without data, period. It is too contrary to what I know about human nature.
Craft a version of Firefox that randomly claims a cert is invalid, even when it is fine and lets see what some real data looks like when you install it with the permission of your friendly neighborhood librarian (remember, no actual security change, we have just modified Firefox to claim to be insecure when things are in-fact fine). I’d bet 95% of people will still go to gmail or BB&T with a bad cert error.
–Rob
-
I like the way it handles self-signed certs better than the way it used to. Instead of getting a dialog every time you visit the site, you now only have to override it once, and it’ll never bother you again unless the cert changes. You can actually tell it now “I explicitly trust this certificate” and it believes you. You couldn’t do that on Firefox 2, especially if there was a hostname mismatch on the cert (say you’re cheap and only have one IP and are hosting several domains on it, and more than one of them uses SSL). That gives me piece of mind, knowing that Firefox will tell me if the cert changes.
-
Trackback from journal zen - par Vincent on August 6, 2008 at 11:52 am
-
I guess none of the developer use a Linksys WRT54G router. Because this thing, each time it reboots, it change the certificate. Worse. If you made the exception permanent, you are double screwed as I had to *guess* which certificate to remove from the list to be able to at least access my wireless router.
-
Hub: I have a Linksys WRT54G (two of them, actually), and I never noticed that. Of course, the fact that I immediately wiped the default firmware and re-flashed them with OpenWRT after getting them probably has something to do with that. Linksys makes awesome router hardware, but their firmware is usually lacking a bit. Putting third-party firmware on them gives you a kick-ass router without lots of strange problems like randomly-changing SSL certs and randomly-dropped connections.
-
I don’t mind seeing a warning as in Firefox 2. I do find it excessive when I have to click half a dozen of times to dismiss the warning. Moreover it is the same warning no matter what the reason is to declare the certificate as “invalid”. Very numbing and uneducating.
-
Explaining why you did it doesn’t make it better. Its simply broken by design. SSL is about privacy as much as about authentication and 99.9999% of all warnings are false positives for perfectly legitimate sites. Which makes it stupid. And even if a warning is ok, the huge warning sign with the 4 click procedure to get rid of it is borderline idiotic. Fix it please.
-
“If you can’t prove who you are talking to then you can’t prove that the connection is secure”
That is true, you cannot prove that the connection is secure. But if you are connected to the correct website and there is no man-in-the-middle attack (which will be the case almost all the time) SSL is still better than simply using http because other people cannot listen to your communication.
So even not-verified SSL has its uses. Identity may be a huge part of SSL but it sure as hell isn’t everything.
And even if that was not the case nobody argues against a warning. But no 500*500 red pixel warning which forces you to do a 4-click marathon to get rid of it. How about a good old popup window with:The encryption of this site isn’t authenticated by a verified authority so the identity cannot be proven.
* Bring me out of here
* Ignore once
* Ignore always for this site.Thats how potentially dangerous actions like executing downloaded files, looking at attachements and allowing programs access to the firewall are handled across a multitude of products. I (and everyone else) wants to be in control of the computer not the other way round.
You may be correct in principle but if almost everyone who is using your product doesn’t like how something is done and a change is easy perhaps it would be good to acknowledge this and simply change it.
-
Ever use a Cisco Guard? Arbor Networks Peak Netflow? Linksys WRT54G w/ DD-WRT? Juniper Web Interface? Most hardware based devices which have an SSL web interface will use an unsigned certificate. Well I admire the principle behind it and the guts to do it, etc. etc. it makes my life a lot more difficult now that i have to go through multiple screens just to get around the warning. As long as the communication between my browser and the hardware device is encrypted, I’m ok. I don’t care if it was signed by Thawte or Verisign, etc. that’s good for banks and commerce sites, but if you’re working with various hardware, it becomes a real hinderance. I’m just looking for a way to disable it, or bring back the FireFox 2.0 behavior, or something. Either an option in the browser, or even an extension/addon. It’s damned annoying.
-
Hi. The new self signed ssl thing sucks. With all due respect. Here’s why – I’m currently developing a website to track local reptile and amphibian populations in Shasta County – with the worldwide amphibian decline, we’ve already lost a few and some others are almost gone – scattered in small isolated populations that aren’t reproducing. No one in the science community seems to be tracking the populations until it is too late.
So I’m putting together a site explicitly for the purpose of allowing hobbyists to upload their field data, and we can track populations. This needs a login, and login should be secure. All I’m using SSL for is registration, user prefs, and login.
I ain’t paying for a SSL cert, I’m already spending hosting fees plus many hours of database/php development, and I’m not making a dead dime.
Because I want the site to be safer for my users providing an SSL login, FireFox has the audacity to tell users that I’m not a legitimate site because legitimate sites won’t ask them to accept a self signed cert?
With all due respect, screw you! Handle it the way Opera does – just as secure, less of a PITA.
SSL is intended to provide a secure public/private key encryption.
Many PUBLIC websites have legitimate use for such encryption that do not need the hassle and extortion fees of certificate authorties.You want to warn my users that you do not recommend credit card / personal info be submitted, fine – I don’t want or ask for that stuff anyway. I just want to know when and where the frogs are hopping and the snakes are crawling. But to tell my users a legitimate public site wouldn’t ask them to accept a self signed cert? That’s a flat out presumptuous lie. Plenty do.
I’m sorry for the tone of my rant, but if I wanted a “big brother” giving the choice of not using SSL (less secure for my users), paying a fee I have no need to pay as a site that does not ask for sensitive information, or having my users jump through hoops – I’m sorry, but I find that highly offensive.
As a Linux user exclusively, I’ve been an advocate of mozilla for year – since very early on in the pre 1.0 days. This however really ticks me off.
Take care,
Michael A. Peters
PS – my Linksys wireless router also uses a self signed cert, another legitimate use of SSL that doesn’t need a certificate authority signature …
-
So, what make the SSL certificates too expensive? Why it can’t bee cheap as a domain name? Why each domain can’t have a SSL for free?
-
Wow. This has turned into a major pain for me!
I understand the security desire, but even making changes in about:config in 3.0.9 I’m stuck.
In my case I just want endpoint-to-endpoint encryption to my home system for a couple of pages I have there to prevent eavesdropping along the way. There HAS to be a better way…
-
I’m sick of people forcing their “naderers” (“Once their naderers go, they do the damnedest things”; ‘Deus Irae’ Philip K Dick), down my throat, then telling it’s for my “safety”. Let me count the ways …
Firefox & Epiphany (self signed certs rejected), Ubuntu Jaunty (boot to root login disabled, su to root doesn’t work because Gtk+ GUI is broken after an su/sudo [because of a Gnome re-write ]), 3Ware 3DM2 raid management software requiring root access and said Gtk+ GUI for installation, and https (for “security”, even on localhost) with a self-signed cert for access.
So a one hour RAID controller install turns into an 18-hour ordeal.
(I’m skipping a couple of hardware problems. 6hr h/w, 12hr s/w)
To quote the CAPTCHA below “not kindness”. How Ironic.I suggest all software has 3 configuration levels.
Experts: All naderers disabled.
PowerUsers: Only the most important naderers enabled.
Novices: More naderers enabled;
Know Nothings: As many naderers as you can think of.From the depths of my ennui,
boblogicP.S. I also hate the cluttered, annoying “smart bar” with no off switch.
To turn it off, you have to install a plugin/extension, which opens more security holes. It’s called “cutting the string in two, then tying the ends together to make it longer”. -
This whole mess has made me start pushing users to move to Google Chrome or SRWare Iron over Firefox. Please fix it. It’s ridiculous, I have to go through an ordeal just to help someone through a self-signed page that has no intentions whatsoever of preventing man-in-the-middle. And no, there’s not a single damn soul who would care whether or not it takes 4 steps or 1 to go to the page; they’re either going to go or not, it doesn’t matter how many jumping jacks you ask them to do. Take it from Google, they’ve mastered ease of use and security in one swipe and they’re doing it in one step.
I don’t want to offend, but this is absolutely ridiculous to deal with. There at least should be a way to disable via about:config. Didn’t anyone think about the possibility that what they were doing was just about as effective as UAC? (annoys users, still makes it completely plausible to get exploited – although to the power user, UAC actually has value. Not only that, UAC can be disabled.)
Something must be done about this. This is a big deal.
-
Why can we not disable this? At least for a specific subset of hosts…
I understand the goal, but some people want a browser that “just works” and for all of our internal machines and network devices with self-signed certificates Firefox no longer “just works”.
Please fix this. soon.
-
For people who spend a significant part of their day working with hardware devices Firefox is no longer even an option. You have “improved” yourself out of the hunt…I suppose I could run IE7 under Wine…it actually sucks (as opposed to being poorly implemented) but I can connect to my hardware with it.
Better yet, where is the source code? I could get my guys to remove that “improvement” and recompile.
-
OH MY FIN GAWD…. I am an IT tech. I have 25 routers around the country. MOST are the same model of nergear, FVS318. I am FINALLY getting around to storing these in my favorites directory. I stored one, then 5 days later I connected to another one, planning on storing it, but OH MY GAWD, I can’t even connect to it, because IT is using the SAME self signed certificate that the previous Netgear device is using, and I added an exemption, and NOW I get a failure that I can’t even click around. I had to go find the previous certificate that I had added and exception for, delete it, then add this exception, which I will have to delete once I connect to the next one….
Just take this pistol and shoot me why don’t you. -
“SSL does not provide any security without identity”
You, being a developer, should know full well that SSL without identity does provide security against sniffing, which is the reason why routers commonly use SSL, and why every login form should use SSL (people use same password on multiple sites). Sniffing is a lot easier than man in the middle, and is extremely common.
You’re using your knowledge to knowingly mislead and misinform less technically inclined users with half truths or worse, less than half truths; that is in my opinion extremely shameful.
Furthermore, phishing is not affected by this warning due to fact that phishers are for most part NOT using self signed or expired certificates.


27 comments
Comments feed for this article
Trackback link: http://www.0xdeadbeef.com/weblog/2008/08/firefox-3-ssl-and-self-signed-certs/trackback/