[PATCH] D83337: [MSAN] Instrument libatomic load/store calls

Gui Andrade via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Tue Jul 7 12:49:11 PDT 2020


guiand marked 3 inline comments as done.
guiand added inline comments.


================
Comment at: compiler-rt/test/msan/libatomic.c:37
+#endif
+}
----------------
One thing that turned out a little strange is that because I have to insert instructions *after* the atomic load, including the origin update, the msan reporter decides that the origin is one line below the call. Is there anything I can do about this?


================
Comment at: llvm/lib/Transforms/Instrumentation/MemorySanitizer.cpp:1872
+    OrderingTable[(int)AtomicOrderingCABI::seq_cst] =
+        (int)AtomicOrderingCABI::seq_cst;
+
----------------
Is there a more elegant way to do this?


================
Comment at: llvm/lib/Transforms/Instrumentation/MemorySanitizer.cpp:3544
+    IRBuilder<> NextIRB(CB.getNextNode());
+    Align AlignOne = assumeAligned(1);
+    auto SrcShadowOriginPair =
----------------
Is it valid to assume any alignment other than 1 here? I can't think of a way, but I'd like to make sure.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D83337/new/

https://reviews.llvm.org/D83337





More information about the llvm-commits mailing list