[PATCH] D11399: invariant.group and invariant.group.barrier docs

Nick Lewycky via llvm-commits llvm-commits at lists.llvm.org
Tue Sep 8 13:50:10 PDT 2015


nlewycky added inline comments.

================
Comment at: docs/LangRef.rst:4427
@@ +4426,3 @@
+
+The ``invariant.group`` metadata may be atached to ``load``/``store`` instruction.
+The existence of the ``invariant.group`` metadata on the instruction tells 
----------------
"atached" --> "attached"

================
Comment at: docs/LangRef.rst:4430
@@ +4429,3 @@
+the optimizer that every ``load`` and ``store`` to the same pointer operand 
+within the same invariant group, can be assume to load or store the same bit 
+value (but see the ``llvm.invariant.group.barrier`` intrinsic which affects 
----------------
"can be assume" --> "can be assumed"

================
Comment at: docs/LangRef.rst:4430
@@ +4429,3 @@
+the optimizer that every ``load`` and ``store`` to the same pointer operand 
+within the same invariant group, can be assume to load or store the same bit 
+value (but see the ``llvm.invariant.group.barrier`` intrinsic which affects 
----------------
nlewycky wrote:
> "can be assume" --> "can be assumed"
What does it mean to be assumed to store the same bit value? In the case of loads I think it means that the load must return the same value. Is the store stating that the store must be storing the same value, that is the value operand of the two stores is bit equivalent?

================
Comment at: docs/LangRef.rst:4580-4581
@@ -4537,2 +4579,4 @@
 
+
+
 Objective-C Garbage Collection Module Flags Metadata
----------------
This should either be removed or committed independently.

================
Comment at: docs/LangRef.rst:11448
@@ +11447,3 @@
+
+Returns another pointer that aliases it's argument but which is considered different 
+for the purposes of ``load``/``store`` ``invariant.group`` metadata.
----------------
"it's" --> "its"

================
Comment at: docs/LangRef.rst:11457
@@ -11376,2 +11456,3 @@
 
+
 '``llvm.var.annotation``' Intrinsic
----------------
Either revert this newline or commit it independently if correct.


http://reviews.llvm.org/D11399





More information about the llvm-commits mailing list