[RISCV] Only try LUI+SH*ADD+ADDI for int materialization if LUI+ADDI+SH*ADD failed.

There's an assert in LUI+SH*ADD+ADDI materialization that makes sure the
lower 12 bits aren't zero since that case should have been handled as
LUI+ADDI+SH*ADD. But nothing prevented the LUI+SH*ADD+ADDI checks from
running after the earlier code handled it.

The sequence would be the same length or longer so it wouldn't replace
the earlier sequence, but the assert happened before that was checked.

The vector holding the sequence also wasn't reset before the second
check so that guaranteed the sequence would never be found to be
shorter.

This patch fixes this by only trying the second expansion when the
earlier fails.

Fixes PR54812.

Reviewed By: benshi001

Differential Revision: https://reviews.llvm.org/D123406

(cherry picked from commit 7004643)

Author: topperc

llvm/lib/Target/RISCV/MCTargetDesc/RISCVMatInt.cpp

@@ -302,32 +302,34 @@ InstSeq generateInstSeq(int64_t Val, const FeatureBitset &ActiveFeatures) {
       TmpSeq.push_back(RISCVMatInt::Inst(Opc, 0));
       if (TmpSeq.size() < Res.size())
         Res = TmpSeq;
-    }
-    // Try to use LUI+SH*ADD+ADDI.
-    int64_t Hi52 = ((uint64_t)Val + 0x800ull) & ~0xfffull;
-    int64_t Lo12 = SignExtend64<12>(Val);
-    Div = 0;
-    if (isInt<32>(Hi52 / 3) && (Hi52 % 3) == 0) {
-      Div = 3;
-      Opc = RISCV::SH1ADD;
-    } else if (isInt<32>(Hi52 / 5) && (Hi52 % 5) == 0) {
-      Div = 5;
-      Opc = RISCV::SH2ADD;
-    } else if (isInt<32>(Hi52 / 9) && (Hi52 % 9) == 0) {
-      Div = 9;
-      Opc = RISCV::SH3ADD;
-    }
-    // Build the new instruction sequence.
-    if (Div > 0) {
-      // For Val that has zero Lo12 (implies Val equals to Hi52) should has
-      // already been processed to LUI+SH*ADD by previous optimization.
-      assert(Lo12 != 0 &&
-             "unexpected instruction sequence for immediate materialisation");
-      generateInstSeqImpl(Hi52 / Div, ActiveFeatures, TmpSeq);
-      TmpSeq.push_back(RISCVMatInt::Inst(Opc, 0));
-      TmpSeq.push_back(RISCVMatInt::Inst(RISCV::ADDI, Lo12));
-      if (TmpSeq.size() < Res.size())
-        Res = TmpSeq;
+    } else {
+      // Try to use LUI+SH*ADD+ADDI.
+      int64_t Hi52 = ((uint64_t)Val + 0x800ull) & ~0xfffull;
+      int64_t Lo12 = SignExtend64<12>(Val);
+      Div = 0;
+      if (isInt<32>(Hi52 / 3) && (Hi52 % 3) == 0) {
+        Div = 3;
+        Opc = RISCV::SH1ADD;
+      } else if (isInt<32>(Hi52 / 5) && (Hi52 % 5) == 0) {
+        Div = 5;
+        Opc = RISCV::SH2ADD;
+      } else if (isInt<32>(Hi52 / 9) && (Hi52 % 9) == 0) {
+        Div = 9;
+        Opc = RISCV::SH3ADD;
+      }
+      // Build the new instruction sequence.
+      if (Div > 0) {
+        // For Val that has zero Lo12 (implies Val equals to Hi52) should has
+        // already been processed to LUI+SH*ADD by previous optimization.
+        assert(Lo12 != 0 &&
+               "unexpected instruction sequence for immediate materialisation");
+        assert(TmpSeq.empty() && "Expected empty TmpSeq");
+        generateInstSeqImpl(Hi52 / Div, ActiveFeatures, TmpSeq);
+        TmpSeq.push_back(RISCVMatInt::Inst(Opc, 0));
+        TmpSeq.push_back(RISCVMatInt::Inst(RISCV::ADDI, Lo12));
+        if (TmpSeq.size() < Res.size())
+          Res = TmpSeq;
+      }
     }
   }
 

llvm/test/CodeGen/RISCV/imm.ll

@@ -2401,3 +2401,39 @@ define i64 @li_rori_3() {
 ; RV64IZBS-NEXT:    ret
   ret i64 -2281701377
 }
+
+; This used to assert when compiled with Zba.
+define i64 @PR54812() {
+; RV32I-LABEL: PR54812:
+; RV32I:       # %bb.0:
+; RV32I-NEXT:    lui a0, 521599
+; RV32I-NEXT:    li a1, -1
+; RV32I-NEXT:    ret
+;
+; RV64I-LABEL: PR54812:
+; RV64I:       # %bb.0:
+; RV64I-NEXT:    lui a0, 1048447
+; RV64I-NEXT:    addiw a0, a0, 1407
+; RV64I-NEXT:    slli a0, a0, 12
+; RV64I-NEXT:    ret
+;
+; RV64IZBA-LABEL: PR54812:
+; RV64IZBA:       # %bb.0:
+; RV64IZBA-NEXT:    lui a0, 872917
+; RV64IZBA-NEXT:    sh1add a0, a0, a0
+; RV64IZBA-NEXT:    ret
+;
+; RV64IZBB-LABEL: PR54812:
+; RV64IZBB:       # %bb.0:
+; RV64IZBB-NEXT:    lui a0, 1048447
+; RV64IZBB-NEXT:    addiw a0, a0, 1407
+; RV64IZBB-NEXT:    slli a0, a0, 12
+; RV64IZBB-NEXT:    ret
+;
+; RV64IZBS-LABEL: PR54812:
+; RV64IZBS:       # %bb.0:
+; RV64IZBS-NEXT:    lui a0, 1045887
+; RV64IZBS-NEXT:    bclri a0, a0, 31
+; RV64IZBS-NEXT:    ret
+  ret i64 -2158497792;
+}