3 ms·
As far as I understand, performance is not a valid concern. For example, from dvyukov's KTSAN wiki [1]: Given a sufficiently expressive atomic API and a goo
by typical182 7y ago
As far as I understand, performance is not a valid concern.
For example, from dvyukov's KTSAN wiki [1]:
Given a sufficiently expressive atomic API and a good
implementation, you pay only for what you really need (if
you pay just a bit less, generated code becomes incorrect).
So performance is not an argument here.
[1] https://github.com/google/ktsan/wiki/READ_ONCE-and-WRITE_ONCE#performance-considerations https://github.com/google/ktsan/wiki/READ_ONCE-and-WRITE_ONC...
- eloff 7y agoYou're talking about the sanitizer performance. We were talking about the kernel performance in the real world. Fixing these race conditions will have some effect.
- rrss 7y agoNo, that page is about kernel performance, not sanitizer performance. The idea is that if you omit READ_ONCE/WRITE_ONCE when they are needed, it doesn't matter if the code is faster because it is wrong.