8 ms·
An attempt to backdoor the kernel (2003)
- gnu8 13y agoThat story takes me back. With today's increased paranoia, one wonders if the Bitkeeper fiasco was just a smokescreen to obscure malicious additions to the kernel.
- luckydude 13y agoI'm not sure what you are trying to say here (I'm lm, the BK guy). It's not clear if you are suggesting that BK was helping add backdoors or if all the fuss over the BK license was so they could move to something less paranoid so it would be easier to add backdoors or $something_else.
- gnu8 13y agoIt recently became known that the forces that would do evil encourage and exploit situations like the one that occurred with Bitkeeper. I don't suggest that Bitkeeper was complicit with any such scheme though.
- luckydude 13y agoNot sure what to say to this. First of all, BK had nothing to do with this exploit other than to find it. The article was vague on the details if I remember correctly. We found the problem because the code we wrote to export the BK tree to the CVS tree validated the results. The code managed by BK was fine, the code in the CVS tree had been hacked. It was the validation step that caught it. The statement "forces that would do evil encourage and exploit situations like the one that occurred with Bitkeeper" just makes no sense (to me at least, if someone else can parse it please share).
- dTal 13y agoSeems clear enough to me. The forces that would do evil [if they had the chance] = NSA etc. Thus, "The NSA encourage and exploit situations like the one that occurred with Bitkeeper".
- luckydude 13y agoI am dismayed and disgusted with the NSA as much as the next guy, maybe more than the next guy, but suggesting they had anything to do with stick a root backdoor in the kernel is a stretch. I guess it is possible but seems pretty unlikely. Saying it had anything to do with BK just means you didn't read the original article. BitKeeper found the problem that was in the CVS tree. BitKeeper itself was never compromised.
- dTal 13y agoI was not defending the GP, merely clarifying its syntax.
- stinos 13y agoMaybe use (2003) in the title. I just started reading the text, and went all like erm, wait, what? CVS? BitKeeper? :]
- ThomPete 13y agoYou are right. Done.
- toddkaufmann 13y agothe comments on that page are not related to this attempt specifically. who made the attempt?
- luckydude 13y agoIt was a breakin on kernel.bkbits.net - I don't think we ever figured out who did it. http://lwn.net/Articles/57137/ http://lwn.net/Articles/57137/
- dan1234 13y agoThere's a bit more detail on the actual exploit here: http://lwn.net/Articles/57552/ http://lwn.net/Articles/57552/
- gabriel34 13y agoHistory repeats itself. Interesting how this has been happening since back then and we are still surprised by it. I don't have the numbers, but without Android and Ubuntu it would seem much less appealing then nowadays. I wonder if any of these actually are in place and working, and what is keeping anyone from trying again. On a side note, is open source peer-reviewed software really more secure than closed source? It is a choice between trusting a single entity to make safe programs and not purposefully put backdoors on the code and trusting lots of entities to make safe code, not purposefully backdoor it and trusting the reviewer's abilities.
- StavrosK 13y ago> It is a choice between trusting a single entity to make safe programs and not purposefully put backdoors on the code and trusting lots of entities to make safe code, not purposefully backdoor it and trusting the reviewer's abilities. It is a choice between trusting the reviewer and having the code open to scrutiny, or trusting the entity and not having the code open to scrutiny. Same trust surface, much less opportunity for review.
- gabriel34 13y agoWhat I meant is that there is a trade-off and the choice is highly subjective. Opensource is not always better. Having lots of entities involved lessens the trustworthiness of the program. Being open source enhances it because of community scrutiny. The trustworthiness of a software is function of the trustworthiness of every entity involved in the making of said software or at least in the making of the most vulnerable ones. Surely the reviewers pull the trustworthiness up, but having many entities, some unknown to you, involved can dissolve their effects. Being fruit of bad reviewers or bad programmers, some mistake or malicious code may pass by unnoticed. Which one would you trust more: 1. Open Source software writen by a single trustworthy person and reviewed by many 2. Open Source software writen by many people and reviewed by many