DavidSpickett wrote: > I do wonder if we can reliably instruction step over every instruction on > every target (atomics instructions/sequences seem to be a sore point).
https://github.com/llvm/llvm-project/issues/34572 / https://github.com/llvm/llvm-project/issues/24318 / https://github.com/llvm/llvm-project/issues/60259 If it was just single instructions, we realise we can't step it accurately, disable all software watchpoints, and tell the user what happened. The problem with sequences is we don't know that we're in them. Very few people have complained about atomic sequences, but then again, most people don't instruction single step a lot of code. I can't think of another "sequence" thing we don't handle, but I feel like there can't be more than 2/3 things. That combined with v8.1 having single instruction atomics, mitigates it further. Let me do some research to see if it really is a handful of sequence types. https://github.com/llvm/llvm-project/pull/151195 _______________________________________________ lldb-commits mailing list lldb-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
