================
@@ -6,11 +6,7 @@
) lldb::SBWatchpointOptions::SetWatchpointTypeRead;
%feature("docstring", "Gets whether the watchpoint should stop on read
accesses."
) lldb::SBWatchpointOptions::GetWatchpointTypeRead;
-%feature("docstring", "Sets whether the watchpoint should stop on write
accesses."
+%feature("docstring", "Sets whether the watchpoint should stop on write
accesses. eWatchpointWriteTypeOnModify is the most commonly useful mode, where
lldb will stop when the watched value has changed. eWatchpointWriteTypeAlways
will stop on any write to the watched region, and on some targets there can
false watchpoint stops where memory near the watched region was written, and
lldb cannot detect that it is a spurious stop."
----------------
jasonmolenda wrote:
Yeah I'll remove that text. It's going to be part of my large watchpoints
enhancement -- on stubs that support it on AArch64, we'll use a masking
watchpoint, so you can watch a 48 byte object but we'll do it by watching 64
bytes of memory. for a straight-up "write" watchpoint, we can't tell whether
the 48 bytes were written to with the same value, or the 16 bytes we aren't
intending to watch were written to. Also with AArch64 in SVE Streaming Mode,
the address matching is at lower accuracy, so a write near a watchpoint can
match the watchpoint, even if it didn't actually access the memory range.
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
