I see little value in these transformations. Did we catch any real issue here? All this type casting and replacement is just churn.
I certainly don't want such churn in libbpf, and I'd leave BPF selftests as is as well. int vs u64 can have subtle and non-obvious implications for BPF code generation (for no-alu32 variants especially), and I think BPF CI already exposed some of those already.
I think we can live without -Wsign-compare just fine.
I was convinced by [1] that this needed to be done not just for current version of the code but for future code being more robust initially.I have already done all the work and I can follow daniel's suggestion for the next version.Otherwise,This means then that the TODO comment to add that compiler warning in the makefile needs to be removed.
Also I wanted to ask since the CI bot had success with my patch.What does this [2] mean exactly.
Thank you for your time reviewing and helping.
Regards, Mehdi
[1]:https://github.com/kernel-patches/bpf/commit/495d2d8133fd1407519170a5238f455... [2]:https://github.com/kernel-patches/bpf/actions/runs/18006172526