[RISCV][NFC] Move Zawrs/Zacas implementation to RISCVInstrInfoZa.td

[wangpc-pp]

To keep the structure of TableGen files clear.

The definitions are simplified by the way.

[llvmbot]

@llvm/pr-subscribers-backend-risc-v

Author: Wang Pengcheng (wangpc-pp)

Changes

To keep the structure of TableGen files clear.

The definitions are simplified by the way.

---

Full diff: https://github.com/llvm/llvm-project/pull/76940.diff

2 Files Affected:

• (modified) llvm/lib/Target/RISCV/RISCVInstrInfo.td (+1-16)
• (added) llvm/lib/Target/RISCV/RISCVInstrInfoZawrs.td (+26)

diff --git a/llvm/lib/Target/RISCV/RISCVInstrInfo.td b/llvm/lib/Target/RISCV/RISCVInstrInfo.td
index 35e8edf5d2fa72..066dd07a2ee7f0 100644
--- a/llvm/lib/Target/RISCV/RISCVInstrInfo.td
+++ b/llvm/lib/Target/RISCV/RISCVInstrInfo.td
@@ -729,22 +729,6 @@ def UNIMP : RVInstI<0b001, OPC_SYSTEM, (outs), (ins), "unimp", "">,
   let imm12 = 0b110000000000;
 }
 
-let Predicates = [HasStdExtZawrs] in {
-def WRS_NTO : RVInstI<0b000, OPC_SYSTEM, (outs), (ins), "wrs.nto", "">,
-              Sched<[]> {
-  let rs1 = 0;
-  let rd = 0;
-  let imm12 = 0b000000001101;
-}
-
-def WRS_STO : RVInstI<0b000, OPC_SYSTEM, (outs), (ins), "wrs.sto", "">,
-              Sched<[]> {
-  let rs1 = 0;
-  let rd = 0;
-  let imm12 = 0b000000011101;
-}
-} // Predicates = [HasStdExtZawrs]
-
 } // hasSideEffects = 1, mayLoad = 0, mayStore = 0
 
 def CSRRW : CSR_ir<0b001, "csrrw">;
@@ -2095,6 +2079,7 @@ include "RISCVInstrInfoM.td"
 
 // Atomic
 include "RISCVInstrInfoA.td"
+include "RISCVInstrInfoZawrs.td"
 
 // Scalar FP
 include "RISCVInstrInfoF.td"
diff --git a/llvm/lib/Target/RISCV/RISCVInstrInfoZawrs.td b/llvm/lib/Target/RISCV/RISCVInstrInfoZawrs.td
new file mode 100644
index 00000000000000..fac95eec814a46
--- /dev/null
+++ b/llvm/lib/Target/RISCV/RISCVInstrInfoZawrs.td
@@ -0,0 +1,26 @@
+//===-- RISCVInstrInfoZawrs.td --------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+//
+// This file describes the RISC-V instructions from the standard
+// Wait-on-Reservation-Set (Zawrs) extension document.
+//
+//===----------------------------------------------------------------------===//
+
+let hasSideEffects = 1, mayLoad = 0, mayStore = 0 in {
+class WRSInst<bits<12> funct12, string opcodestr>
+    : RVInstI<0b000, OPC_SYSTEM, (outs), (ins), opcodestr, ""> {
+  let rs1 = 0;
+  let rd = 0;
+  let imm12 = funct12;
+}
+}
+
+let Predicates = [HasStdExtZawrs] in {
+def WRS_NTO : WRSInst<0b000000001101, "wrs.nto">, Sched<[]>;
+def WRS_STO : WRSInst<0b000000011101, "wrs.sto">, Sched<[]>;
+} // Predicates = [HasStdExtZawrs]

[asb]

LGTM. I think it's not necessarily the case that every extension should have it's own file, but Zawrs would better belong in RISCVInstrInfoA or RISCVInstrInfoZa.td I think if we felt it didn't deserve its own file.

[topperc]

LGTM. I think it's not necessarily the case that every extension should have it's own file

I agree. Adding a file for every extension doesn't scale well if we keep adding extensions at the rate we are.

[wangpc-pp]

LGTM. I think it's not necessarily the case that every extension should have it's own file

I agree. Adding a file for every extension doesn't scale well if we keep adding extensions at the rate we are.

What about adding it to RISCVInstrInfoZa.td? We will have Zabha (https://github.com/riscv/riscv-zabha) and may have Zalasr (https://github.com/mehnadnerd/riscv-zalasr) later.

And what is the policy? Should we put all (or these simple ones) Zi* extensions into RISCVInstrInfoZi.td for example? We already have RISCVInstrInfoZb.td and RISCVInstrInfoZc.td.

[asb]

LGTM. I think it's not necessarily the case that every extension should have it's own file

I agree. Adding a file for every extension doesn't scale well if we keep adding extensions at the rate we are.

What about adding it to RISCVInstrInfoZa.td? We will have Zabha (https://github.com/riscv/riscv-zabha) and may have Zalasr (https://github.com/mehnadnerd/riscv-zalasr) later.

We could move Zacas there too.

And what is the policy? Should we put all (or these simple ones) Zi* extensions into RISCVInstrInfoZi.td for example? We already have RISCVInstrInfoZb.td and RISCVInstrInfoZc.td.

It would be nice if you could always find a Zfoo extension by opening RISCVInstrInfoZfoo.td for RISCVInstrInfoZf.td Sadly, the subset extensions like Zmmul or Zca mean you'd need to look in the single-letter RISCVInstrInfo td file too. So that has to be a special case.

[topperc]

LGTM