RISC-V: Fix ICE and codegen error of scalar move in RV32 system.

We need to reset the AVL to 0 or 1 for scalar move for RV32 system,
For any non-zero AVL input, we set that to 1, and zero will keep as zero.

We are using wrong way (by andi with 1) before to achieve that, and it
will cause ICE with const_int, and also wrong behavior, so now we have
two code path, one for const_int and one for non-const_int.

bug.C:144:2: error: unrecognizable insn:
  144 |  }
      |  ^
(insn 684 683 685 26 (set (reg:SI 513)
        (and:SI (const_int 4 [0x4])
            (const_int 1 [0x1]))) "bug.C":115:47 -1
     (nil))

andi a4,a4,1 ===> sgtu a4,a4,zero
vsetlvi tu        vsetvli tu
vlse              vlse

gcc/ChangeLog:

	* config/riscv/riscv-protos.h (gen_avl_for_scalar_move): New function.
	* config/riscv/riscv-v.cc (gen_avl_for_scalar_move): New function.
	* config/riscv/vector.md: Fix scalar move bug.

gcc/testsuite/ChangeLog:

	* gcc.target/riscv/rvv/base/scalar_move-6.c: Adapt test.
	* gcc.target/riscv/rvv/base/scalar_move-9.c: New test.

gcc/config/riscv/riscv-protos.h

@@ -205,6 +205,7 @@ enum vlen_enum
 };
 bool slide1_sew64_helper (int, machine_mode, machine_mode,
 			  machine_mode, rtx *);
+rtx gen_avl_for_scalar_move (rtx);
 }
 
 /* We classify builtin types into two classes:

gcc/config/riscv/riscv-v.cc

@@ -701,4 +701,27 @@ slide1_sew64_helper (int unspec, machine_mode mode, machine_mode demote_mode,
   return true;
 }
 
+rtx
+gen_avl_for_scalar_move (rtx avl)
+{
+  /* AVL for scalar move has different behavior between 0 and large than 0.  */
+  if (CONST_INT_P (avl))
+    {
+      /* So we could just set AVL to 1 for any constant other than 0.  */
+      if (rtx_equal_p (avl, const0_rtx))
+	return const0_rtx;
+      else
+	return const1_rtx;
+    }
+  else
+    {
+      /* For non-constant value, we set any non zero value to 1 by
+	 `sgtu new_avl,input_avl,zero` + `vsetvli`.  */
+      rtx tmp = gen_reg_rtx (Pmode);
+      emit_insn (
+	gen_rtx_SET (tmp, gen_rtx_fmt_ee (GTU, Pmode, avl, const0_rtx)));
+      return tmp;
+    }
+}
+
 } // namespace riscv_vector

gcc/config/riscv/vector.md

@@ -1229,9 +1229,7 @@
           else if (GET_MODE_BITSIZE (<VEL>mode) > GET_MODE_BITSIZE (Pmode))
             {
 	      // Case 2: vmv.s.x (TU) ==> andi vl + vlse.v (TU) in RV32 system.
-	      rtx tmp = gen_reg_rtx (Pmode);
-              emit_insn (gen_rtx_SET (tmp, gen_rtx_AND (Pmode, operands[4], const1_rtx)));
-	      operands[4] = tmp;
+	      operands[4] = riscv_vector::gen_avl_for_scalar_move (operands[4]);
 	      operands[1] = CONSTM1_RTX (<VM>mode);
 	    }
           else
@@ -1292,9 +1290,7 @@
        vlse64.v  */
     if (satisfies_constraint_Wb1 (operands[1]))
       {
-	rtx tmp = gen_reg_rtx (Pmode);
-        emit_insn (gen_rtx_SET (tmp, gen_rtx_AND (Pmode, operands[4], const1_rtx)));
-        operands[4] = tmp;
+	operands[4] = riscv_vector::gen_avl_for_scalar_move (operands[4]);
 	operands[1] = CONSTM1_RTX (<VM>mode);
       }
   }

gcc/testsuite/gcc.target/riscv/rvv/base/scalar_move-6.c

@@ -37,8 +37,6 @@ void foo2 (void *base, void *out, size_t vl)
 /*
 ** foo3:
 ** ...
-** andi\t[a-x0-9]+,\s*[a-x0-9]+,\s*1
-** ...
 ** vlse64.v\tv[0-9]+,0\([a-x0-9]+\),zero
 ** ...
 ** ret
@@ -54,8 +52,6 @@ void foo3 (void *base, void *out, size_t vl)
 /*
 ** foo4:
 ** ...
-** andi\t[a-x0-9]+,\s*[a-x0-9]+,\s*1
-** ...
 ** vlse64.v\tv[0-9]+,0\([a-x0-9]+\),zero
 ** ...
 ** ret
@@ -137,8 +133,6 @@ void foo9 (void *base, void *out, size_t vl)
 /*
 ** foo10:
 ** ...
-** andi\t[a-x0-9]+,\s*[a-x0-9]+,\s*1
-** ...
 ** vmv.v.i\tv[0-9]+,\s*-15
 ** ...
 */
@@ -167,8 +161,6 @@ void foo11 (void *base, void *out, size_t vl)
 /*
 ** foo12:
 ** ...
-** andi\t[a-x0-9]+,\s*[a-x0-9]+,\s*1
-** ...
 ** vmv.v.i\tv[0-9]+,\s*0
 ** ...
 ** ret

gcc/testsuite/gcc.target/riscv/rvv/base/scalar_move-9.c

@@ -0,0 +1,26 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv32gcv -mabi=ilp32d -fno-schedule-insns -fno-schedule-insns2 -O3" } */
+
+#include "riscv_vector.h"
+
+vuint64m2_t f1(vuint64m2_t var_17, uint64_t var_60)
+{
+  vuint64m2_t var_16 = __riscv_vmv_s_x_u64m2_tu(var_17,var_60, 0);
+  return var_16;
+}
+
+vuint64m2_t f2(vuint64m2_t var_17, uint64_t var_60)
+{
+  vuint64m2_t var_16 = __riscv_vmv_s_x_u64m2_tu(var_17,var_60, 4);
+  return var_16;
+}
+
+vuint64m2_t f3(vuint64m2_t var_17, uint64_t var_60, size_t vl)
+{
+  vuint64m2_t var_16 = __riscv_vmv_s_x_u64m2_tu(var_17,var_60, vl);
+  return var_16;
+}
+
+/* { dg-final { scan-assembler-times {vsetivli\s+zero,\s*0,\s*e64,\s*m2,\s*t[au],\s*m[au]} 1 } } */
+/* { dg-final { scan-assembler-times {vsetivli\s+zero,\s*1,\s*e64,\s*m2,\s*t[au],\s*m[au]} 1 } } */
+/* { dg-final { scan-assembler-times {sgtu} 1 } } */