jasonmolenda wrote: Thanks @DavidSpickett for looking into these. Let me looked into the tagged ptr watchpoint issue right now, that sounds like a bug. One of the changes I made in the second landing of my patch was to the TestStepOverWatchpoint C file, which had three 1-byte globals, one which it would put a `read` watchpoint on, and second that it would put a `write` watchpoint on. lldb will send these as separate watchpoint requests (Z2,addr,1). If the variables were allocated in consecutive bytes, which seems likely, and we use a doubleword resolution watchpoint, then the remote stub would need to realize that we're watching for 'read' accesses on one byte, and 'write' accesses on a different byte within that doubleword, and use a single hardware watchpoint register with read+write access enabled.
I don't know if lldb-server would handle this correctly, but debugserver wouldn't. And it's not what the test case intends to be testing. When I re-landed my patch, I changed the variables to `long` so they're more likely to be the size of the hardware watchpoint granule and aligned to that alignment. I wonder why it was still failing, i'll have to look more. https://github.com/llvm/llvm-project/pull/66308 _______________________________________________ lldb-commits mailing list lldb-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
