[PATCH] D72700: [DSE] Add first version of MemorySSA-backed DSE (Bottom up walk).

[Alina Sbirlea via Phabricator via llvm-commits]

asbirlea added a comment.

Some more minor comments, but I think this is a reasonable first version to check in.



================
Comment at: llvm/lib/Transforms/Scalar/DeadStoreElimination.cpp:1383
+//    upwards.
+// 2. Check that there no reads between DomAccess and the StartDef by checking
+//    all uses starting at DomAccess and walking until we see StartDef.
----------------
s/there/there are


================
Comment at: llvm/lib/Transforms/Scalar/DeadStoreElimination.cpp:1523
+  bool isCompleteOverwrite(MemoryLocation DefLoc, Instruction *UseInst) const {
+    if (!UseInst->mayWriteToMemory())
+      return false;
----------------
Can you add a comment here, along the lines:
`UseInst has a MemoryDef associated in MemorySSA. It's possible for a MemoryDef to not write to memory, e.g. a volatile load is modeled as a MemoryDef.`


================
Comment at: llvm/lib/Transforms/Scalar/DeadStoreElimination.cpp:1692
+
+      DeadInst->eraseFromParent();
+    }
----------------
The MemDep variant of DSE also attempts to keep debug info. Does this also make sense here?
```
    // Try to preserve debug information attached to the dead instruction.
    salvageDebugInfo(*DeadInst);
```


================
Comment at: llvm/lib/Transforms/Scalar/DeadStoreElimination.cpp:1769
+    // Walk MemorySSA upward to find MemoryDefs that might be killed by SI.
+    while ((Next = State.getDomMemoryDef(StartDef, Current, SILoc,
+                                         DefVisibleToCaller, ScanLimit))) {
----------------
Nit: Move declaration of Next inside while condition?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D72700/new/

https://reviews.llvm.org/D72700