[llvm] [RISCV][GISel] Share an atomic load isel pattern GISel RV64 and SDAG RV32. NFC (PR #161721)

[via llvm-commits]

llvmbot wrote:


<!--LLVM PR SUMMARY COMMENT-->

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

Author: Craig Topper (topperc)

<details>
<summary>Changes</summary>

Use stricter type for RV64 only patterns.

Stores are different because atomic_store doesn't differentiate truncating and non-truncating stores.

---
Full diff: https://github.com/llvm/llvm-project/pull/161721.diff


2 Files Affected:

- (modified) llvm/lib/Target/RISCV/RISCVGISel.td (+1-1) 
- (modified) llvm/lib/Target/RISCV/RISCVInstrInfoA.td (+4-5) 


``````````diff
diff --git a/llvm/lib/Target/RISCV/RISCVGISel.td b/llvm/lib/Target/RISCV/RISCVGISel.td
index 19d5aff023d53..af1ceb6bcda4e 100644
--- a/llvm/lib/Target/RISCV/RISCVGISel.td
+++ b/llvm/lib/Target/RISCV/RISCVGISel.td
@@ -118,7 +118,7 @@ let Predicates = [HasAtomicLdSt] in {
 }
 
 let Predicates = [HasAtomicLdSt, IsRV64] in {
-  def : LdPat<atomic_load_nonext_32, LW, i32>;
+  // Load pattern is in RISCVInstrInfoA.td and shared with RV32.
   def : StPat<atomic_store_32, SW, GPR, i32>;
 }
 
diff --git a/llvm/lib/Target/RISCV/RISCVInstrInfoA.td b/llvm/lib/Target/RISCV/RISCVInstrInfoA.td
index 99992d196b43d..25accd93eaa03 100644
--- a/llvm/lib/Target/RISCV/RISCVInstrInfoA.td
+++ b/llvm/lib/Target/RISCV/RISCVInstrInfoA.td
@@ -174,15 +174,14 @@ let Predicates = [HasAtomicLdSt] in {
   def : StPat<relaxed_store<atomic_store_8>,  SB, GPR, XLenVT>;
   def : StPat<relaxed_store<atomic_store_16>, SH, GPR, XLenVT>;
   def : StPat<relaxed_store<atomic_store_32>, SW, GPR, XLenVT>;
-}
 
-let Predicates = [HasAtomicLdSt, IsRV32] in {
-  def : LdPat<relaxed_load<atomic_load_nonext_32>, LW>;
+  // Used by GISel for RV32 and RV64.
+  def : LdPat<relaxed_load<atomic_load_nonext_32>, LW, i32>;
 }
 
 let Predicates = [HasAtomicLdSt, IsRV64] in {
-  def : LdPat<relaxed_load<atomic_load_asext_32>, LW>;
-  def : LdPat<relaxed_load<atomic_load_zext_32>, LWU>;
+  def : LdPat<relaxed_load<atomic_load_asext_32>, LW, i64>;
+  def : LdPat<relaxed_load<atomic_load_zext_32>, LWU, i64>;
   def : LdPat<relaxed_load<atomic_load_nonext_64>, LD, i64>;
   def : StPat<relaxed_store<atomic_store_64>, SD, GPR, i64>;
 }

``````````

</details>


https://github.com/llvm/llvm-project/pull/161721