[Lldb-commits] [PATCH] D117113: [lldb] [llgs] Implement qXfer:siginfo:read

[Pavel Labath via Phabricator via lldb-commits]

labath added inline comments.


================
Comment at: lldb/include/lldb/Host/common/NativeProcessProtocol.h:195-198
+  virtual llvm::Expected<std::unique_ptr<llvm::MemoryBuffer>>
+  GetSiginfo() const {
+    return llvm::make_error<UnimplementedError>();
+  }
----------------
I think it would be more natural/consistent to have this function on a specific NativeThreadProtocol object (and let the user choose whether it wants the siginfo for the current thread, or someone else).


================
Comment at: lldb/source/Plugins/Process/Linux/NativeProcessLinux.cpp:1987
+NativeProcessLinux::GetSiginfo() const {
+  siginfo_t siginfo;
+  Status error = GetSignalInfo(GetCurrentThreadID(), &siginfo);
----------------
I'd be better to outright allocate a memory buffer of the right size (`getNewUninitMemBuffer`), and directly write the data there.


================
Comment at: lldb/test/API/tools/lldb-server/TestLldbGdbServer.py:1398
+        self.reset_test_sequence()
+        # TODO: freebsd doesn't report crashing thread in T
+        self.test_sequence.add_log_lines(
----------------
Didn't D117103 fix that?


================
Comment at: lldb/test/API/tools/lldb-server/TestLldbGdbServer.py:1437
+        pad = ""
+        if sys.maxsize > 2**32:
+            pad = "i"
----------------
you want the address width of the debugged binary, not python, right? We already have some utilities to help with that (grep for `parse_process_info_response`).


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D117113/new/

https://reviews.llvm.org/D117113