Adapting proxo 4.5 to the latest OpenSSL DLLs

[herbalist]

I've got too many versions of these files. Must be getting them mixed up. I think I mistakenly downloaded the files from your post, #6. Need to clean this mess up before I go any further.

[herbalist]

Finally, I got the correct files together and it appears to be working. I'm still getting certificate errors on certain sites. I don't know if these are within the intended scope of the fix. With Proxomitron bypassed, the site renders properly. On https://firstlook.org/theintercept/ I get

  SSL error.png (Size: 14.22 KB / Downloads: 646)
With Proxomitron filtering, only the text is visible.

[JJoe]

(Jan. 26, 2015 02:07 AM)herbalist Wrote:  On https://firstlook.org/theintercept/ I get

With Proxomitron filtering, only the text is visible.

This is expected due to the self signed cert that the Proxomitron is giving the browser|client. The browser is expecting the cert to say "firstlook.org" but it says "Proxomitron".

It may be possible to set the browser to ignore this mismatch. Which browser are you using?

[herbalist]

Seamonkey 2.0.14, last 98 compatible versions, even with KernelEx.

[herbalist]

I just ran into another strange complication that makes no sense and is probably beyond the scope of this project. I have 2 instances of Proxomitron installed. The first is the standard one we've been working with. Since it appeared to be working, I copied all of the new and modded files to the 2nd instance. This copy, named Proxomitron2.exe is launched using SocksCap 2.40. This enables me to route Proxomitron to the Tor Socks Proxy port, 9050. The unmodified version of Proxomitron worked flawlessly with it. The modified version doesn't work with SocksCap at all. It keeps trying to connect directly. Any ideas?

[JJoe]

It has been years since I used Seamonkey. Check "about:config" for something like "security.mixed_content.block_active_content" and set it to "false".

Quote: The modified version doesn't work with SocksCap at all. It keeps trying to connect directly. Any ideas?

SocksCap was a long time ago.

[herbalist]

That it was. Unfortunately, the more current options don't work on this box for what I had in mind. For now, I'll just use the modified copy for direct connections and the original version with SocksCap and Tor. I'll look at it again another day. The way this day has been, I wouldn't be surprised if I've missed something obvious, again.

[amy]

I have tested the OpenSSL 1.0.1l DLLs I posted (http://prxbx.com/forums/showthread.php?t...6#pid18016 ) with 98SE and patched Proxomitron and can confirm that it works.

Opera 12.16 (the one I got from http://arc.opera.com/pub/opera/win/1216/en/ ) does not seem to send any nonce extension; I've tested 9.21 and 9.26, and they certainly do - a malformed one that new OpenSSL rejects (both slightly different - .21 sends "universal constructed integer" while .26 sends "context-specific constructed integer", but the spec requires "sequence of".) It's not at all a problem if you use Firefox or Chrome or IE, as AFAIK none of them send any nonce extension.

Quote:This copy, named Proxomitron2.exe is launched using SocksCap 2.40. This enables me to route Proxomitron to the Tor Socks Proxy port, 9050. The unmodified version of Proxomitron worked flawlessly with it. The modified version doesn't work with SocksCap at all. It keeps trying to connect directly. Any ideas?

The patch does not change any code that deals with how Proxomitron connects. Make sure you're SocksCap'ing the right file...

[herbalist]

Quote:The patch does not change any code that deals with how Proxomitron connects. Make sure you're SocksCap'ing the right file...

I rechecked everything today. Made a new Proxomitron folder well away from the others with copies of the original and modified proxomitron executables. Each has a new name. Same results. The original connects via Tor when Socksified. The modified attempts to connect directly.
Just to make sure that I haven't messed up the files again, are these correct?
original
MD5 f2867bee7180cdc839f7636fddc1aa74
SHA-256 7adc0296d97e24417000c5cac53c8dfb34a5e6ddedceec168ffe45648803285b
modified
MD5 d675477025d6af758f10ed1b87a366e6
SHA-256 5a6160c7f6eeb28b10de7fc698f115176c8e579e44b4e209b088942f12e33425
I'm at a loss of what else I could be overlooking.

[JJoe]

(Jan. 26, 2015 10:27 PM)herbalist Wrote:  are these correct?
original
MD5 f2867bee7180cdc839f7636fddc1aa74
SHA-256 7adc0296d97e24417000c5cac53c8dfb34a5e6ddedceec168ffe45648803285b
modified
MD5 d675477025d6af758f10ed1b87a366e6
SHA-256 5a6160c7f6eeb28b10de7fc698f115176c8e579e44b4e209b088942f12e33425

"original" is the unpatched N45j. "modified" is the result of the patcher that I posted. It has the RWIN unlock and amy's patches.


Edit: Added details.

[herbalist]

Other than the SocksCap issue, the patched version is working well so far. Nice piece of work.

[ProxRocks]

(Jan. 27, 2015 02:15 AM)JJoe Wrote:  ... "modified" is the result of the patcher that I posted.

was there a publicly posted link that i missed?

[JJoe]

(Jan. 27, 2015 11:31 AM)ProxRocks Wrote:  

(Jan. 27, 2015 02:15 AM)JJoe Wrote:  ... "modified" is the result of the patcher that I posted.

was there a publicly posted link that i missed?

Posted in response to http://prxbx.com/forums/showthread.php?t...4#pid18034 and removed later. It was nothing special.

[amy]

That SocksCap issue is odd, one thing to check is make sure SocksCap is starting up the right .exe - you can temporarily rename the other one so if it tries to run that one it will fail. Other than that I will try to reproduce it when I have some time...

...and is it just me or do I somehow have a negative postcount?

[ProxRocks]

(Jan. 28, 2015 11:16 AM)amy Wrote:  ...and is it just me or do I somehow have a negative postcount?

COOL !!!...