My own experience is that RAII has been very useful in low-level work when used to capture cleanup/finalization. Results in fewer if-statements and SLOC.
Issues of real-time performance and new vs malloc vs in-place/nothrow new also seem like orthogonal concerns here. The larger problem is that yahoo who's using the wrong algorithm and/or maintaining driver code w/o benefit of code review ;)
My (admittedly short) experience writing kernel drivers with C++ was that RAII could be great help, but there's no way you can rely on any library you did not write yourself (and one that you wrote with the express intent to use for low-level work at that).
C++ code just tends to assume it's fine to do whatever crazy stuff it assumes needs to be done, like allocating 4 bytes at a time using new.
C code is usually more careful about that, and quite a few packages offer control of their memory allocation in C. (Yes, C++ does allow you to overload new and delete, and have allocator traits, and more. Come back when you've actually implemented this for a library you did not write, and we'll discuss war stories if I can remember them. It was 2001, it was ugly back then, and I'd be surprised if it is better looking now).
Issues of real-time performance and new vs malloc vs in-place/nothrow new also seem like orthogonal concerns here. The larger problem is that yahoo who's using the wrong algorithm and/or maintaining driver code w/o benefit of code review ;)