Protecting yourself from POODLE attacks

Topic

New Reply

	TOP STORY Protecting yourself from POODLE attacks
	By Susan Bradley No, this isn’t about Fluffy gone rogue. To keep our online browsing safe, we rely heavily security protocols — the “S” in HTTPS. But a new exploit — POODLE — shows that commonly used security protocols aren’t as secure as we thought; websites and browsers will both need an upgrade.

---

The full text of this column is posted at http://windowssecrets.com/top-story/protecting-yourself-from-poodle-attacks/ (opens in a new window/tab).

Columnists typically cannot reply to comments here, but do incorporate the best tips into future columns.[/td]

[/tr][/tbl]

Reply | Quote

Replies

Probably a web formatting glitch, but in the shortcut workaround to disable SSL in Chrome, the character before ssl should be TWO hyphens, not an N dash, e.g.:

–ssl-version-min=tls1

not –ssl-version-min=tls1

EDIT: The shortcut target addition is OK if you copy and paste it from the newsletter email; just the web page got two hyphens converted to a dash.

Bruce

Reply | Quote

Hi Bruce,
Wondered if the shortcut could point to tls2 ?
My understanding is that it would automatically fall back to tls1.1 or tls1 as required if admins are too lazy or busy to implement the latest ?

Reply | Quote

I tried the test site using up to date chrome (both 32-bit and 64-bit) without the fix described here and it said it was safe.

Reply | Quote

Firefox, SeaMonkey, Pale Moon – Change the value of “0” for “security.tls.version.min” to “1” using about:config.

Chrome – In HKEY_CLASSES_ROOT, edit the http/shell/open/command from:

[indent]”C:Program FilesGoogleChromeApplicationchrome.exe” — “%1″[/indent]

to:

[indent]”C:Program FilesGoogleChromeApplicationchrome.exe” –ssl-version-min=tls1 — “%1″[/indent]

(Using this registry approach protects all of Chrome, no matter how you open it).

Internet Explorer – Tools > Internet Options > Advanced tab, scroll all the way down, deselect/uncheck “Use SSL 3.0”, click Apply > OK.

For a simple test, Poodletest.com displays a poodle dog if your browser still supports SSL 3.0, and a Springfield terrier if it doesn’t. On the other hand, Qualys SSL Labs provides a more detailed analysis of the SSL protocols your browser supports.

Reply | Quote

Qualys still states I’m vulnerable even when using the registry update but Poodletest states I’m safe. Strange; who do I believe?

Reply | Quote

Just tried the Chrome fix for Iron Browser and it works with it too. The path’s different of course. For my WinXP VM:

“C:Program FilesSRWare Ironchrome.exe” –ssl-version-min=tls1

Reply | Quote

For a simple test, Poodletest.com displays a poodle dog if your browser still supports SSL 3.0, and a Springfield terrier if it doesn’t. On the other hand, Qualys SSL Labs provides a more detailed analysis of the SSL protocols your browser supports.

I got the terrier (Not Vulnerable) and on the Qualys site I got “Vulnerable” from the same browser (untouched Chrome). I am not sure one should rely of the simple test…..

Reply | Quote

Hmmm. Not certain what to make of this as I DO NOT USE ANY of the browsers for which fixes are specifically offered and I do not have but one browser–to the extent that same is possible, I have DISABLE IE. Am I to conclude that browsers that are not specifically mentioned, e.g., Opera, are not at risk? I’ve also stopped updating Opera because of its penchant to overwrite all the configuration changes I have made to it, so would I need to be concerned about what specific version of ANY BROWSER I might henceforth use?

Reply | Quote

Isn’t Opera based of Firefox? Try the ‘about:config’ in post number five.

Reply | Quote

Isn’t Opera based of Firefox? Try the ‘about:config’ in post number five.

Opera was originally developed for Linux, while if I am not mistaken, Firefox is based on Netscape Communicator, so any parallels between them are apt to be largely coincidental.

Reply | Quote

Opera was originally developed for Linux

Nope it was first used on Windows.

Just because you don't know where you are going doesn't mean any road will get you there.

Reply | Quote

Isn’t Opera based of Firefox? Try the ‘about:config’ in post number five.

For most of its life, Opera was based on its own proprietary software called Presto. But, over the last couple years it transitioned from Presto to a version of Chromium.

Reply | Quote

When I accessed poodletest.com using Firefox, I received the following warning from my antivirus suite:

“Address:
http://www.poodletest.com/

Trend Micro has confirmed that this website can transmit malicious software or has been involved in online scams or fraud.

Please close this page.” Is Susan Bradley aware of this? Is it truly a valid threat? What do you make of it?

Reply | Quote

Poodletest.com[/url]

Known Issues

Make sure you clear your cache between tests.

The test requires that you are able to connect to an SSLv3 only site. There are some false positives/false negatives that you may experience. For example, if your connection is slow, the connection to the test site will time out and your browser may wrongly show up as not vulnerable.

Browser Specific Issues:

Firefox

Firefox is picky as to what ciphers it accepts. The test site supports a wide range of ciphers to allow Firefox to connect.

Safari

Apple stated that the Safari update released on Oct 17th no longer allows block ciphers via SSLv3. The test site (on purpose) only supports block ciphers as they are vulnerable to POODLE. However, my testing so far shows that Safari will still connect to the test site using ciphers like AES256. Safari should show up as not-vulnerable if it only supports stream ciphers over SSLv3.

Android

I am getting some reports of inconsistent and wrong results with Android. Haven’t quite been able to reproduce some of the reported issues.

More Information:

[*]To report problems: jullrich -at- sans -dot- edu (please add “poodletest” to subject and note your IP address)
in case you don’t know your IP address, use myipaddress.com[/url]

[*]Internet Storm Center (check for updates)

[*]Google Advisory

[*]OpenSSL Advisory

[*]Microsoft Advisory

Acknowledgements:

Thanks Andreas for suggesting a javascript trick to avoid image caching.

Reply | Quote

.
DisableSSLv3.com[/url]

A community-powered step-by-step tutorial on disabling the security protocol you now love to hate.

Reply | Quote

blogs.opera.com/security > Security changes in Opera 25; the poodle attacks[/url]

Reply | Quote

If you are using Firefox or Pale Moon, try the following. Install the Pale Moon Commander add-on. Then select Options, click the Security button, click the SSL tab, and where it says, lowest supported protocol, click on the drop-down box, and select TLS 1.0, and finally the OK button to close the window. That’s it!

Reply | Quote

Yep. Wikipedia: Opera browser

Opera began in 1994 as a research project at Telenor, the largest Norwegian telecommunications company. In 1995, it branched out into a separate company named Opera Software ASA.[12] Opera was first released publicly with version 2.0 in 1996,[13] which only ran on Microsoft Windows.[14] In an attempt to capitalize on the emerging market for Internet-connected handheld devices, a project to port Opera to mobile device platforms was started in 1998.[14] Opera 4.0, released in 2000,[13] included a new cross-platform core that facilitated creation of editions of Opera for multiple operating systems and platforms.[15]

Reply | Quote

On the 15th, Opera developers remotely changed the settings in Opera 12.17 and 25, disabling SSL and enabling TLS 1, 1.1 and 1.2. However with 12.17, those latter two TLS settings caused multiple random crashes. The fix for now has been to disable 1.1 and 1.2. An Opera developer wrote that they won’t be re-remoting those for now. Bug reports have been filed. The links can be found at
– http://blogs.opera.com/security/
– https://forums.opera.com/discussion/…ntil-yesterday (Oct 21 – rseiler)

Reply | Quote

On the 15th, Opera developers remotely changed the settings in Opera 12.17 and 25, disabling SSL and enabling TLS 1, 1.1 and 1.2. However with 12.17, those latter two TLS settings caused multiple random crashes. The fix for now has been to disable 1.1 and 1.2. An Opera developer wrote that they won’t be re-remoting those for now. Bug reports have been filed. The links can be found at
– http://blogs.opera.com/security/
– https://forums.opera.com/discussion/…ntil-yesterday (Oct 21 – rseiler)

I Finally understand what you are saying!!! Thank You for repeating it this way. The trouble was I have had TLS 1.1 and 1.2 set to “on” for months because of concerns even prior to HeartBleed but never realized that they were the cause of Opera 12.17’s frequent crashes. I believe it has stopped crashing now and I am much relieved.

Best Regards

Crysta

--------------------------------------

1. Tower Totals: 2xSSD ~512GB, 2xHHD 20 TB, Memory 32GB

SSDs: 6xOS Partitions, 2xW8.1 Main & Test, 2x10.0 Test, Pro, x64

CPU i7 2600 K, SandyBridge/CougarPoint, 4 cores, 8 Threads, 3.4 GHz
Graphics Radeon RX 580, RX 580 ONLY Over Clocked
More perishable

2xMonitors Asus DVI, Sony 55" UHD TV HDMI

1. NUC 5i7 2cores, 4 Thread, Memory 8GB, 3.1 GHz, M2SSD 140GB
1xOS W8.1 Pro, NAS Dependent, Same Sony above.

-----------------

Reply | Quote

Shame on me, I forgot Opera. And I have it installed even.

BTW the firefox folks recommend you do not use about:config

Reply | Quote

Uh oh, too late.

Reply | Quote

I tried bot ways for Chrome and neither worked. The dog page show okay but the other one shows it didn’t work. Don’t know what to do now.

Reply | Quote

Several websites test whether your currently open browser supports SSL 3.0. For a simple test, Poodletest.com displays a poodle dog if your browser still supports SSL 3.0, and a Springfield terrier if it doesn’t. On the other hand, Qualys SSL Labs (site) provides a more detailed analysis of the SSL protocols your browser supports.

When I try this, the first says I am not vulnerable, but the second says I am. Just a bit confusing!

Reply | Quote

I have had the same experience as others in regard to Chrome. Qualys states that the vunerability still exists whilst Poodletest indicates that it doesn’t. I hope that we can get some guidance on this issue soon.
Thanks

Reply | Quote

My version of FF is not accepted by the addon and Chrome passes w/o change.

security.enable_ssl3;false

worked in FF 10.0.0.2

However there is this line:

services.sync.prefs.sync.security.enable_ssl3;true

which I have not yet changed.

Just because you don't know where you are going doesn't mean any road will get you there.

Reply | Quote

Hi, just wanted to add, for Windows 8 users (yes we do exist) if you want to change the shortcut on the taskbar you have to right click on the icon then right click on the application name in the pop-up menu then click properties – see http://www.nextofwindows.com/how-to-customize-pinned-taskbar-shortcuts/

Reply | Quote

Right-click the link you use to launch Chrome and select Properties.

Now append one blank space and then this string:

[indent]–ssl-version-min=tls1[/indent]

to the end of the text already in the Target: block.

For example, if it was:

[indent]”C:Program Files (x86)GoogleChromeApplicationchrome.exe”[/indent]

it would now be:

[indent]”C:Program Files (x86)GoogleChromeApplicationchrome.exe” –ssl-version-min=tls1[/indent]

Now go back to https://www.ssllabs.com/ssltest/viewMyClient.html and test.

Reply | Quote

Right-click the link you use to launch Chrome and select Properties.

Now append one blank space and then this string:

[indent]–ssl-version-min=tls1[/indent]

to the end of the text already in the Target: block.

For example, if it was:

[indent]”C:Program Files (x86)GoogleChromeApplicationchrome.exe”[/indent]

it would now be:

[indent]”C:Program Files (x86)GoogleChromeApplicationchrome.exe” –ssl-version-min=tls1[/indent]

Now go back to https://www.ssllabs.com/ssltest/viewMyClient.html and test.

I use Chromium and Firefox in Linux, and the same revisions work there, too. The desktop shortcuts look different, but the same added parameter for minimum SSL security level works with the Chromium launcher shortcut on the Ubuntu desktop. But it is important that the Unity Sidebar Launcher Chromium Icon be unlocked, and the revised desktop launcher be repinned to the Sidebar Launcher for all normal launches of Chromium in Ubuntu to be protected.

Firefox 33 for Linux takes the same add-on (SSL Version Control 0.2) as does the Windows version.

For Windows users, I would STRONGLY suggest unpinning any Taskbar launchers for either Firefox or Chrome, and repinning these after making the necessary revisions for minimum SSL security levels. The same might apply to any Windows 8/8.1 users who may have pinned the browsers to the Modern/Metro Start Screen. Unpin, revise the shortcut, and repin after testing.

Since the POODLE Test site did not detect these differences, but the SSL Labs site did, I would recommend relying on the latter, not the former.

-- rc primak

Reply | Quote

Aha!!! That’s worked and the mystery resolved.

Thanks bobprimak

Reply | Quote

Right-click the link you use to launch Chrome and select Properties.

Now append one blank space and then this string:

[indent]–ssl-version-min=tls1[/indent]

to the end of the text already in the Target: block.

For example, if it was:

[indent]”C:Program Files (x86)GoogleChromeApplicationchrome.exe”[/indent]

it would now be:

[indent]”C:Program Files (x86)GoogleChromeApplicationchrome.exe” –ssl-version-min=tls1[/indent]

Now go back to https://www.ssllabs.com/ssltest/viewMyClient.html and test.

I came across a reader comment in Scott Helme’s blog about Poodle that if you are using a 64-bit version of Windows, the modification you should make to the shortcut for Chrome, in the Target field, after …chrome.exe” should be /ssl-version-min=tls1; i.e. using “/” rather than “–” as is the case for the 32-bit Windows version of Chrome. reference: https://scotthelme.co.uk/sslv3-goes-to-the-dogs-poodle-kills-off-protocol/, and find the comment by contributor “rootberg.” I am running Windows 7 Pro 64-bit, and this fix works. I tried the several test programs before and after making the change. I too, get a false reassurance from http://www.poodletest.com (the Terrier) before making the modification. Just don’t trust that test. Use https://zmap.io/sslv3, or the ssllabs one to confirm your modification.

Reply | Quote

I came across a reader comment in Scott Helme’s blog about Poodle that if you are using a 64-bit version of Windows, the modification you should make to the shortcut for Chrome, in the Target field, after …chrome.exe” should be /ssl-version-min=tls1; i.e. using “/” rather than “–” as is the case for the 32-bit Windows version of Chrome. reference: https://scotthelme.co.uk/sslv3-goes-to-the-dogs-poodle-kills-off-protocol/, and find the comment by contributor “rootberg.”

The blog’s author said 15 days ago in response to that comment that he would test the claim and update his instructions if necessary, but has not done so.

–ssl-version-min=tls1 works fine for me on 64-bit Chrome (as checked on all three test sites).

Bruce

Reply | Quote

I applied the fix to Chrome, Firefox, and Internet Explorer 11 on my Win7 Pro PC. Poodle Test for Chrome & Firefox said “Not vulnerable.” Test for IE said “Vulnerable.” Went to Settings, etc., and sure enough, “Use SSL 3.0” was checked. I unchecked it, closed IE and reopened it.

Poodle Test said IE was “Not vulnerable.”

I closed IE, and a few minutes later, came back to IE, it displayed the home page for a few seconds, and then: “Internet Explorer has stopped working A problem caused the program to stop working correctly. Windows will close the program and notify you if a solution is available.”

Several re-tries of IE have had the same result. Interestingly, AOL, which is based on IE, tests “Not vulnerable” and is not crashing at the moment.

Reply | Quote

I applied the patch to Chrome and FF, but since I almost never use IE I left IE unpatched. In the last two days I’ve come across at least half a dozen websites that don’t work properly with SSL3.0 disabled, and they DID work with the unpatched IE. I double-checked by running an unpatched Chrome and sure enough the websites worked fine with SSL3.0 enabled. I’d bet there’s a good chance there were other sites that didn’t work but I didn’t notice the missing content.

Is there any way to protect ourselves from attack without silently emasculating websites??

Reply | Quote

I applied the patch to Chrome and FF, but since I almost never use IE I left IE unpatched. In the last two days I’ve come across at least half a dozen websites that don’t work properly with SSL3.0 disabled, and they DID work with the unpatched IE. I double-checked by running an unpatched Chrome and sure enough the websites worked fine with SSL3.0 enabled. I’d bet there’s a good chance there were other sites that didn’t work but I didn’t notice the missing content.

Is there any way to protect ourselves from attack without silently emasculating websites??

Tell those web sites that you won’t be visiting them any more until they stop forcing you to connect in an insecure manner. (I don’t think there’s any real alternative.)

Some web servers have dropped support of SSL 3.0 to prevent the POODLE attack, such as CloudFlare,[12] and Wikimedia.[13]
POODLE Prevention at Wikipedia

Fortunately a University of Michigan study shows that very few sites rely on SSL 3.0. Servers that support SSLv3 (along with other newer versions of TLS) are likely vulnerable to the attack. The report shows that this is 3 percent of the top 1 million domains on Alexa. Some still allow this technology since they wanted to allow users of old browsers to still access the site.
SSL 3.0 POODLE Vulnerability

Servers administrators don’t want to lock out the critical IE6 market, so they also support SSLv3. And we all suffer.
Attack of the week: POODLE

Critical? IE6 is 3% of the market (less than 1% in the U.S.) and users deserve to be forcibly moved on. (It’s 13 years old!)

Google and Facebook dropped support for IE6 4.5 years ago.

Persuade 3% of web sites they should abandon 3% of their potential users? Or just avoid those sites until they get the message?

Bruce

Reply | Quote

(Again, the fix for this exploit has to happen on both ends of Internet connections — the client and the server.)

Again, I think that statement is confusing (OK, wrong.)

From the original report by the Google Security Team who discovered this vulnerability:

Recommendations

The attack described above requires an SSL 3.0 connection to be established, so disabling the SSL 3.0 protocol in the client or in the server (or both) will completely avoid it.

Bruce

Reply | Quote

I applied the patch to Chrome and FF, but since I almost never use IE I left IE unpatched. In the last two days I’ve come across at least half a dozen websites that don’t work properly with SSL3.0 disabled, and they DID work with the unpatched IE. I double-checked by running an unpatched Chrome and sure enough the websites worked fine with SSL3.0 enabled. I’d bet there’s a good chance there were other sites that didn’t work but I didn’t notice the missing content.

Is there any way to protect ourselves from attack without silently emasculating websites??

Did you use either or both of the test sites (in the article) to test Chrome? If so what were the results? There are a number of contributors on this thread who have patched but get conflicting results from these 2 sites.

An alternative way to test would be to visit a website that still runs SSL3, and see if they “work properly” or not. I haven’t had any issues visiting any websites, but that could simply be because I haven’t visited any. Perhaps you could suggest a couple that you have had difficulty with using your patched Chrome.

Given the inconclusive results from the test sites, we need another way to test if Chrome has been patched or not.

Thanks

Reply | Quote

Did you use either or both of the test sites (in the article) to test Chrome? If so what were the results? There are a number of contributors on this thread who have patched but get conflicting results from these 2 sites. [/quote]
I used both. Both said I was protected.

Perhaps you could suggest a couple that you have had difficulty with using your patched Chrome.

Sorry, that was days ago. I don’t remember what they were.

Reply | Quote

Firefox, SeaMonkey, Pale Moon – Change the value of “0” for “security.tls.version.min” to “1” using about:config.

Chrome – In HKEY_CLASSES_ROOT, edit the http/shell/open/command from:
[indent]”C:Program FilesGoogleChromeApplicationchrome.exe” — “%1″[/indent]
to:
[indent]”C:Program FilesGoogleChromeApplicationchrome.exe” –ssl-version-min=tls1 — “%1″[/indent]
(Using this registry approach protects all of Chrome, no matter how you open it).

Internet Explorer – Tools > Internet Options > Advanced tab, scroll all the way down, deselect/uncheck “Use SSL 3.0”, click Apply > OK.

For a simple test, Poodletest.com displays a poodle dog if your browser still supports SSL 3.0, and a Springfield terrier if it doesn’t. On the other hand, Qualys SSL Labs provides a more detailed analysis of the SSL protocols your browser supports.

In Mozilla-SeaMonkey there is a simpler way to get rid of SSL and retain TLS 1.0, TLS 1.1, and TLS 1.2
The settings are in SeaMonkey’s Edit | Preferences | Privacy & Security | SSL.
Once there, uncheck SSL, and at that same location, check that the three TLS versions are checked.

http://forums.mozillazine.org/viewtopic.php?f=40&t=2883851&p=13845671#p13845671%5D%5B/url%5D

After unchecking SSL, I tested the results at Poodletest.com which said I was “Not vulnerable,” and Qualys SSL Labs (https://www.ssllabs.com/ssltest/viewMyClient.html) said that “Your user agent is not vulnerable.”

R.N. (Roger) Folsom

Reply | Quote

Microsoft has released a Fix It to disable the feature which was the subject of the POODLE attack. The Fix It, a program which implements changes in the registry, makes the process simpler than the alternatives.

Reply | Quote

Software setup:

[*]Windows 7 Ultimate 64-bit (with all updates)

[*]Google Chrome Version 38.0.2125.111 m (64-bit)

[*]Google Chrome shortcut Target:

[indent]”C:Program Files (x86)GoogleChromeApplicationchrome.exe” —ssl-version-min=tls1[/indent]

Test results:

[*]https://www.ssllabs.com/ssltest/viewMyClient.html

Your user agent is not vulnerable.

[*]http://www.poodletest.com/

Not vulnerable!

[*]https://zmap.io/sslv3

Good News! Your browser does not support SSLv3.

Reply | Quote

I just discovered a simple way to get step-by-step instructions on disabling SSL v.3 in Mozilla Firefox. Click on Help and then Firefox Help. In the “Search Mozilla Firefox” box, type “Poodle,” and then click on “Does POODLE SSLv3 effect Mozilla Firefox?” Step-by-step instructions will then appear on your screen.

Be sure to type “Poodle,” not “POODLE,” in the search box.

Reply | Quote