[RISCV] Prevent re-ordering some adds after shifts

Summary:
DAGCombine will normally turn a `(shl (add x, c1), c2)` into `(add (shl x, c2), c1 << c2)`, where `c1` and `c2` are constants. This can be prevented by a callback in TargetLowering.

On RISC-V, materialising the constant `c1 << c2` can be more expensive than materialising `c1`, because materialising the former may take more instructions, and may use a register, where materialising the latter would not.

This patch implements the hook in RISCVTargetLowering to prevent this transform, in the cases where:
- `c1` fits into the immediate field in an `addi` instruction.
- `c1` takes fewer instructions to materialise than `c1 << c2`.

In future, DAGCombine could do the check to see whether `c1` fits into an add immediate, which might simplify more targets hooks than just RISC-V.

Reviewers: asb, luismarques, efriedma

Reviewed By: asb

Subscribers: xbolva00, lebedev.ri, craig.topper, lewis-revill, Jim, hiraditya, rbar, johnrusso, simoncook, apazos, sabuasal, niosHD, kito-cheng, shiva0217, jrtc27, zzheng, edward-jones, rogfer01, MartinMosbeck, brucehoult, the_o, rkruppe, PkmX, jocewei, psnobl, benna, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D62857

llvm-svn: 363736

1 files changed, 9 insertions, 0 deletions

diff --git a/llvm/lib/Target/RISCV/Utils/RISCVMatInt.h b/llvm/lib/Target/RISCV/Utils/RISCVMatInt.h
index bb23d389bdf..b12ae2eade9 100644
--- a/llvm/lib/Target/RISCV/Utils/RISCVMatInt.h
+++ b/llvm/lib/Target/RISCV/Utils/RISCVMatInt.h
@@ -9,6 +9,7 @@
 #ifndef LLVM_LIB_TARGET_RISCV_MATINT_H
 #define LLVM_LIB_TARGET_RISCV_MATINT_H
 
+#include "llvm/ADT/APInt.h"
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/Support/MachineValueType.h"
 #include <cstdint>
@@ -30,6 +31,14 @@ using InstSeq = SmallVector<Inst, 8>;
 // order to allow this helper to be used from both the MC layer and during
 // instruction selection.
 void generateInstSeq(int64_t Val, bool IsRV64, InstSeq &Res);
+
+// Helper to estimate the number of instructions required to materialise the
+// given immediate value into a register. This estimate does not account for
+// `Val` possibly fitting into an immediate, and so may over-estimate.
+//
+// This will attempt to produce instructions to materialise `Val` as an
+// `Size`-bit immediate. `IsRV64` should match the target architecture.
+int getIntMatCost(const APInt &Val, unsigned Size, bool IsRV64);
 } // namespace RISCVMatInt
 } // namespace llvm
 #endif