[PATCH] D66677: [GlobalISel] Introduce a G_DYN_STACKALLOC opcode to represent dynamic allocas

[Matt Arsenault via Phabricator via llvm-commits]

arsenm added inline comments.


================
Comment at: llvm/include/llvm/Target/GenericOpcodes.td:195
+  let InOperandList = (ins type1:$size, i32imm:$align);
+  let hasSideEffects = 0;
+}
----------------
aemerson wrote:
> arsenm wrote:
> > Should this be considered a side effect?
> I don't think so? G_FRAME_INDEX doesn't have side effects, so I don't think this should be counted as having it either. I do intend to use this to implement stack checking calls but that's a target dependent thing.
The DAG equivalent has a chain. I think you could observe different behavior if you reordered a small and a large request


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D66677/new/

https://reviews.llvm.org/D66677