[Lldb-commits] [PATCH] D24764: Refactor NativeRegisterContextLinux_x86_64 code.

[Pavel Labath via lldb-commits]

labath requested changes to this revision.
labath added a comment.
This revision now requires changes to proceed.

I am going to assume you know more about the exact details of the intel cpu's than me, so I am not going comment on the technical details. The code seems cleaner, so this looks like a good thing to do. Just please try to avoid using the const_cast if it's no absolutely necessary (and it almost never is).


================
Comment at: source/Plugins/Process/Linux/NativeRegisterContextLinux_x86_64.cpp:807
@@ -827,2 +806,3 @@
+  if (m_xstate_type == XStateType::Invalid) {
     if (const_cast<NativeRegisterContextLinux_x86_64 *>(this)->ReadFPR().Fail())
       return false;
----------------
I don't think the `const_cast` here is a good idea. How about we just make this functions non-const ?

================
Comment at: source/Plugins/Process/Linux/NativeRegisterContextLinux_x86_64.cpp:808
@@ -807,3 @@
-  // Check if FXSAVE is enabled.
-  if (!__get_cpuid(1, &rax, &rbx, &rcx, &rdx))
-    return false;
----------------
I remember you #including some header to get the __get_cpuid function. Can that be removed now?


https://reviews.llvm.org/D24764