Critical crypto bug exposes Yahoo Mail passwords Russian-roulette style

OpenSSL defect still exposing sensitive data even after patch is released.

Read the whole story

 

[jrose]

myforwik[/url]":2btkpzak]

IMO SSL/TLS is now completely broken. The number of potential certificates that have been exploited and that could now be used for man in the middle attacks could be in the millions..... the list of black listed certificates will be in the millions and/or the number of blacklisted sub certficate authorities is probably going to be 10,000+. Vendors already hate just including one or two items on the blacklist, let alone this number of items....

I've had my suspicions that the NSA/other TLA has had something like this up their sleeve for some time. If it wasn't something like this, they've probably just bought one of the major CAs.

IMHO, the whole idea of a CA is broken anyway - you are trusting a third party who may or may not be trustworthy. The only way to verify certs properly IMHO is to do it out of band via snail mail or some other method.

 

[jrose]

kliu0x52[/url]":3u403zj7]

ExcessPhase[/url]":3u403zj7]ArrayBoundscheck?
Why does this sound like C programming?
Why am I not coding in C anymore since 1996?

And what would you like to use to write low-level security libraries that will be used in a wide variety of scenarios and processes? Write this in a managed language, and now you have the problem of every program requiring TLS also requiring whatever runtimes, libraries, etc. that your managed language uses.

There is a reason people still use C, just as there is still a reason for people to understand assembly. Just because C isn't what you'd use to write most software doesn't mean that there aren't places where C is by far the best option available.

I dunno. Maybe ADA? Pascal? One of the myriad of other languages which isn't quite as sucky for performance as managed code, but at least does proper type checking and array bounds checking?

 

[jrose]

Luridis[/url]"762o81h]
Pascal is two years older than C and Ada appeared 8 years later. None of them are managed code in their original form, which tells me you're not entirely familiar with what that actually means.

I never said they were managed code (though there was an alternative thread of discussion suggesting managed code: I did not). I said they do array bounds checking (they also enforce type safety). Which despite not being managed code is a hell of a lot more safety than C offers - without the overhead of managed code. The bounds checking would have caught this bug.

Pascal got some traction as an OS language and Ada got very little. C took over the world of kernel, driver and low API programming and allows you to do everything you do on electronics today.

Err... no shit? That doesn't mean we can't learn from the mistakes of the past.

There is a reason for that. Why don't you try finding out why C beat out the other languages in OS development before throwing disparaging statements at it before learning the when and why? I'll even give you a hint: #ifdef.

I'm not the one saying there is no need for C or other low level languages. MY viewpoint is that for security critical stuff like this, you write it in a language which makes it harder to shoot yourself in the foot than C does.

You limit the use of C like you limit the use of assembler: to things that are performance critical (and only then if not mission-critical) or REQUIRE low level hacking that is difficult or impossible in other "safer" languages.

Because: programmers are human. Humans fuck up. Fuck ups in security critical software like this are extremely bad. There is a reason that C was not used for military systems, and Ada was required.

ADA was specifically designed for use in problems like this, in mission critical code where bugs have a significant impact.

Just because you have a hammer (C) does not mean that every problem is a nail.

 

[jrose]

Erm.

All the code in OpenSSL should be considered "mission critical".

Having had a look at some of the comments and code examples from people who have worked with it, I won't be trusting OpenSSL for anything I can avoid using it in.

It's absolutely disgustingly bad code. I can not believe the state of it, considering its purpose.

 

[jrose]

Otus[/url]":2lf8ykow]

jrose[/url]":2lf8ykow]Erm.

All the code in OpenSSL should be considered "mission critical".

Having had a look at some of the comments and code examples from people who have worked with it, I won't be trusting OpenSSL for anything I can avoid using it in.

It's absolutely disgustingly bad code. I can not believe the state of it, considering its purpose.

Just make sure to check you aren't moving to even worse code...

Have you had a look at the OpenSSL code? I'm not sure that is possible...