Slashdot Mirror
Firefox Extension HTTPS Everywhere Does What It Sounds Like
climenole writes "HTTPS Everywhere is a Firefox extension produced as a collaboration between The Tor Project and the Electronic Frontier Foundation. It encrypts your communications with a number of major websites. Many sites on the web offer some limited support for encryption over HTTPS, but make it difficult to use. For instance, they may default to unencrypted HTTP, or fill encrypted pages with links that go back to the unencrypted site. The HTTPS Everywhere extension fixes these problems by rewriting all requests to these sites to HTTPS."
noscript has a means of doing this on a per-site basis. Wildcards are accepted.
Then again, if you don't trust the NoSript author after the controversy, this might be a good alternative. I figure NoScript is under more scrutiny than any other extension and the author learned his lesson.
Geez. What kind of poorly written site would do something like quietly defaulting to unencrypted HTTP on a HTTPS request.
https://www.slashdot.org/
Oh wow, this is awesome. I've used greasemonkey scripts with facebook but it's pretty ugly, seems to load the http page before the https page. This sounds perfect. Here's the link https://www.eff.org/files/https-everywhere-latest.xpi which is missing from TFS.
...except not "everywhere", just major sites.
Shouldn't that be https://www.eff.org/https-everywhere ?
It can't work unless these sites already have an https version. If they redirect all 443 traffic to 80 like /., then it does nothing. It might work for facebook since it has a couple pages that allow https, but I'm sure things like their photo servers are probably http only.
Maybe a link to the addon would be useful in the story?
Firefox itself does not really make it easy for the users or for admins to use https everywhere.
I just made a small site, it's for a business, that runs everything through https, I redirect http to https completely. Firefox 3.6.3 on Windows had no problem running the site. IE on windows couldn't open the encrypted pages, Firefox 3.5 on any GNU/Linux distro couldn't open them either, to fix this, I had to add this to /etc/conf.d/ssl.conf : SSLInsecureRenegotiation on
That fixed the IE and FF3.5 on Linux problem.
Here is the description of this flag from apache mod_ssl directive description page:
SSLInsecureRenegotiation Directive
Description: Option to enable support for insecure renegotiation
Syntax: SSLInsecureRenegotiation flag
Default: SSLInsecureRenegotiation off
Context: server config, virtual host
Status: Extension
Module: mod_ssl
Compatibility: Available in httpd 2.2.15 and later, if using OpenSSL 0.9.8m or later
As originally specified, all versions of the SSL and TLS protocols (up to and including TLS/1.2) were vulnerable to a Man-in-the-Middle attack (CVE-2009-3555) during a renegotiation. This vulnerability allowed an attacker to "prefix" a chosen plaintext to the HTTP request as seen by the web server. A protocol extension was developed which fixed this vulnerability if supported by both client and server.
If mod_ssl is linked against OpenSSL version 0.9.8m or later, by default renegotiation is only supported with clients supporting the new protocol extension. If this directive is enabled, renegotiation will be allowed with old (unpatched) clients, albeit insecurely.
Security warning
If this directive is enabled, SSL connections will be vulnerable to the Man-in-the-Middle prefix attack as described in CVE-2009-3555.
Example
SSLInsecureRenegotiation on
The SSL_SECURE_RENEG environment variable can be used from an SSI or CGI script to determine whether secure renegotiation is supported for a given SSL connection.
I wonder if there are other ways of making this work with my other directives:
SSLEngine on
SSLCipherSuite HIGH:MEDIUM:!aNULL:+MD5
SSLVerifyClient none - I am thinking about switching it to 'require' right now, but will have to test all browsers with it again, but have to do it I think.
Oh, and getting it all to run together with apache httpd with mod_ssl + mod_jk + apache tomcat is quite a hassle.
But most unfortunate thing about FF is how it treats the self-signed certificates. It shows it as an SSL ERROR, to which exceptions must be made for the user to be able to enter the site. Can FF developers think about this fact for like longer than a second? It is not an error to run a site with a self-signed certificate, it is a configuration choice and it provides an important role to the site: encrypted traffic for login and for the data transferred to and from the client.
Why is FF showing this to the users as an error? This is not an error, this is by design and it is a special case of usage. Who is not frustrated by the browser treating self signed certificates as if they are some sort of a disease? They provide an important role - a way to secure communications between the server and the browser.
Can this be looked at, because I am SURE this prevents various sites from using encrypted traffic in the first place and it is a BAD thing, not a good one. All traffic needs to be encrypted, but especially user name/password traffic shouldn't be sent around in plain text.
Name it what it is: an exceptional case of using security to encrypt traffic, a case where the site may not necessarily be what it wants to be seen as, but at least the traffic is actually encrypted. It's terrible if someone comes to your site just to see: SSL ERROR on it, OF-COURSE admins don't want THAT message to be shown on their sites, why do you think so few sites do security properly?
You can't handle the truth.
I don't care about ads on his site.
I care about being forced to update NoScript every few days, each time being forced to load his site. I've got another extension, a Flash downloader that does the same thing. They're both either the world's worst programmers, or they're intentionally releasing updates just to drive traffic to their homepages.
It's also incredibly irritating to get interrupted almost every time I go to restart Firefox!
Please help metamoderate.
Unfortunately. No https for slashdot.org - why not Slashdot? Comments on politically orientated stories from "sensitive" countries does not deserve to be encrypted? You should know better Slashdot
It is not an error to run a site with a self-signed certificate
A man in the middle could insert his own self-signed certificate, decrypting the traffic from your site and reencrypting it with his own key pair, and users would be none the wiser. One workaround is to start your own CA, sign its root certificate with PGP, and distribute that cert to your users to install. But then that starts to depend on the PGP web of trust, which in turn depends on air travel to get keys signed.
1. For classic shared hosting solutions using name based hosting, I can almost guarantee if you hit https:///, you're going to hit someone else's virtual host. Many cheap hosting providers w/ limited public IPs will load up domain names on a single IP/Port, but still provide secure hosting to one domain name (on the same port) for shopping cart checkout under a different domain name. Using such a plugin in this use case would not work so well. Then again, would most "smaller sites" really be worthy of encryption in the first place?
2. Not every site is designed w/ the same content root in http vs https. Switching from http to https may completely break if the file structures under the two virtual hosts (potentially entirely separate in Apache) aren't identical (i.e. pointing to the same directory). I'm not touting that this is a best practice, but would be completely feasable if you wanted to keep specific content from being accessed via http and didn't want to bother with mod_rewrite or equivalent.
To the poster above who says there's little CPU penalty for SSL, SSL may not be taxing on the client, but hundreds or thousands of sessions on a server (especially one hosting an app, DB, and Apache) may be another story. Why is someone's assumed paranoid that someone will see that they're reading about cars or home theater equipment on a forum worth requiring a service owner to scale his hardware to the next level to maintain acceptable performance (assuming this phenomenon is multiplied hundred-fold)?
$ man woman *
-bash:
How ridiculous is it, that people get their bank's identity vouched for by a third party they have never met and don't know anything about, when the bank could just put up a fingerprint sign in their lobby and on their paper statements? And people say using a CA is more secure, and less vulnerable to MitM? Really?!?
As copyright owner of this comment, I authorize everyone to defeat any technological measure which limits access to it.