[PATCH] R600/SI: Remove apparently dead code in legalizeOperands
Tom Stellard
tom at stellard.net
Tue Sep 23 11:01:11 PDT 2014
On Fri, Sep 19, 2014 at 11:36:47PM +0000, Matt Arsenault wrote:
> No tests hit this, and I don't see any way a GlobalAddress
> node would survive beyond lowering on SI. It it would,
> this should probably be inserted by selection.
>
LGTM.
> http://reviews.llvm.org/D5421
>
> Files:
> lib/Target/R600/SIISelLowering.cpp
> Index: lib/Target/R600/SIISelLowering.cpp
> ===================================================================
> --- lib/Target/R600/SIISelLowering.cpp
> +++ lib/Target/R600/SIISelLowering.cpp
> @@ -1737,14 +1737,6 @@
> ensureSRegLimit(DAG, Ops[i], RegClass, ScalarSlotUsed);
> }
> continue;
> - } else {
> - // If it's not a VSrc or SSrc operand check if we have a GlobalAddress.
> - // These will be lowered to immediates, so we will need to insert a MOV.
> - if (isa<GlobalAddressSDNode>(Ops[i])) {
> - SDNode *Node = DAG.getMachineNode(AMDGPU::V_MOV_B32_e32, SDLoc(),
> - Operand.getValueType(), Operand);
> - Ops[i] = SDValue(Node, 0);
> - }
> }
>
> if (i == 1 && DescRev && fitsRegClass(DAG, Ops[0], RegClass)) {
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
More information about the llvm-commits
mailing list