I remember reading about a javascript sandbox escape method years ago that involved bunches of jump commands written in such a way that any bit flipped would jump to another jump command directed at your desired code.
The attack relied on cosmic rays flipping bits of memory, but was accelerated by using a light bulb placed near the memory chips to induce heat based errors.
I've often thought about these types of attacks, and wondered how advanced they've become.
Disclaimer: I'm extremely paranoid about computer security to the point where I almost don't care about it anymore and assume that any machine I use is compromised.
As an aside: Is it not required to put a date of publication inside scientific papers? I've always thought that documentation was paramount for papers such as these, and as such the date would be one of the primary pieces of information present, but I've come across many papers (including this one) for which I had to find publication information on third party websites (for this particular paper: Researchgate)
The attack relied on cosmic rays flipping bits of memory, but was accelerated by using a light bulb placed near the memory chips to induce heat based errors.
edit: found it! https://www.cs.princeton.edu/~appel/papers/memerr.pdf
I've often thought about these types of attacks, and wondered how advanced they've become.
Disclaimer: I'm extremely paranoid about computer security to the point where I almost don't care about it anymore and assume that any machine I use is compromised.