Pointer Tainting Still Pointless (but we all see the point of tainting)

摘要

In this paper, we argued several things. First, that the rebuttal by Dalton, Kannan, and Kozyrakis misrepresents our views and the views expressed in our Eurosys'09 paper. This is quite unfortunate, because casual readers may think that we carelessly dismissed an important security technique. We did nothing of the sort. Second, that some of the solutions in the rebuttal are expressly discussed in our paper. Proposing them as a rebuttal suggests that we were not aware of them, which is not the case. Third, that we agree that pointer tainting for malware analysis is problematic, but that our criticism goes beyond false negatives due to evasion by the malware, and expressly includes false positives also. Fourth, that implicit information flow is problematic for detecting memory corruption also.