diff --git a/src/hotspot/cpu/riscv/c2_MacroAssembler_riscv.cpp b/src/hotspot/cpu/riscv/c2_MacroAssembler_riscv.cpp
index 34a61177774ea..8ebdb8a61b520 100644
--- a/src/hotspot/cpu/riscv/c2_MacroAssembler_riscv.cpp
+++ b/src/hotspot/cpu/riscv/c2_MacroAssembler_riscv.cpp
@@ -1382,15 +1382,183 @@ void C2_MacroAssembler::string_indexof_linearscan(Register haystack, Register ne
   bind(DONE);
 }
 
+// Compare longwords
+void C2_MacroAssembler::string_compare_long_same_encoding(Register result, Register str1, Register str2,
+                                                  const bool isLL, Register cnt1, Register cnt2,
+                                                  Register tmp1, Register tmp2, Register tmp3,
+                                                  const int STUB_THRESHOLD, Label *STUB, Label *SHORT_STRING, Label *DONE) {
+  Label TAIL_CHECK, TAIL, NEXT_WORD, DIFFERENCE;
+
+  const int base_offset = isLL ? arrayOopDesc::base_offset_in_bytes(T_BYTE)
+                               : arrayOopDesc::base_offset_in_bytes(T_CHAR);
+  assert((base_offset % (UseCompactObjectHeaders ? 4 :
+                        (UseCompressedClassPointers ? 8 : 4))) == 0, "Must be");
+
+  const int minCharsInWord = isLL ? wordSize : wordSize / 2;
+
+  // load first parts of strings and finish initialization while loading
+  beq(str1, str2, *DONE);
+  // Alignment
+  if (AvoidUnalignedAccesses && (base_offset % 8) != 0) {
+    lwu(tmp1, Address(str1));
+    lwu(tmp2, Address(str2));
+    bne(tmp1, tmp2, DIFFERENCE);
+    addi(str1, str1, 4);
+    addi(str2, str2, 4);
+    subi(cnt2, cnt2, minCharsInWord / 2);
+
+    // A very short string
+    mv(t0, minCharsInWord);
+    ble(cnt2, t0, *SHORT_STRING);
+  }
+#ifdef ASSERT
+  if (AvoidUnalignedAccesses) {
+    Label align_ok;
+    orr(t0, str1, str2);
+    andi(t0, t0, 0x7);
+    beqz(t0, align_ok);
+    stop("bad alignment");
+    bind(align_ok);
+  }
+#endif
+  // load 8 bytes once to compare
+  ld(tmp1, Address(str1));
+  ld(tmp2, Address(str2));
+  mv(t0, STUB_THRESHOLD);
+  bge(cnt2, t0, *STUB);
+  subi(cnt2, cnt2, minCharsInWord);
+  beqz(cnt2, TAIL_CHECK);
+  // convert cnt2 from characters to bytes
+  if (!isLL) {
+    slli(cnt2, cnt2, 1);
+  }
+  add(str2, str2, cnt2);
+  add(str1, str1, cnt2);
+  sub(cnt2, zr, cnt2);
+  addi(cnt2, cnt2, 8);
+  bne(tmp1, tmp2, DIFFERENCE);
+  bgez(cnt2, TAIL);
+
+  // main loop
+  bind(NEXT_WORD);
+    // 8-byte aligned loads when AvoidUnalignedAccesses is enabled
+    add(t0, str1, cnt2);
+    ld(tmp1, Address(t0));
+    add(t0, str2, cnt2);
+    ld(tmp2, Address(t0));
+    addi(cnt2, cnt2, 8);
+    bne(tmp1, tmp2, DIFFERENCE);
+    bltz(cnt2, NEXT_WORD);
+
+  bind(TAIL);
+  load_long_misaligned(tmp1, Address(str1), tmp3, isLL ? 1 : 2);
+  load_long_misaligned(tmp2, Address(str2), tmp3, isLL ? 1 : 2);
+
+  bind(TAIL_CHECK);
+  beq(tmp1, tmp2, *DONE);
+
+  // Find the first different characters in the longwords and
+  // compute their difference.
+  bind(DIFFERENCE);
+  xorr(tmp3, tmp1, tmp2);
+  // count bits of trailing zero chars
+  ctzc_bits(result, tmp3, isLL);
+  srl(tmp1, tmp1, result);
+  srl(tmp2, tmp2, result);
+  if (isLL) {
+    zext(tmp1, tmp1, 8);
+    zext(tmp2, tmp2, 8);
+  } else {
+    zext(tmp1, tmp1, 16);
+    zext(tmp2, tmp2, 16);
+  }
+  sub(result, tmp1, tmp2);
+
+  j(*DONE);
+}
+
+// Compare longwords
+void C2_MacroAssembler::string_compare_long_different_encoding(Register result, Register str1, Register str2,
+                                               bool isLU, Register cnt1, Register cnt2,
+                                               Register tmp1, Register tmp2, Register tmp3,
+                                               const int STUB_THRESHOLD, Label *STUB, Label *DONE) {
+  Label TAIL, NEXT_WORD, DIFFERENCE;
+
+  const int base_offset = arrayOopDesc::base_offset_in_bytes(T_CHAR);
+  assert((base_offset % (UseCompactObjectHeaders ? 4 :
+                          (UseCompressedClassPointers ? 8 : 4))) == 0, "Must be");
+
+  Register strL = isLU ? str1 : str2;
+  Register strU = isLU ? str2 : str1;
+  Register tmpL = tmp1, tmpU = tmp2;
+
+  // load first parts of strings and finish initialization while loading
+  mv(t0, STUB_THRESHOLD);
+  bge(cnt2, t0, *STUB);
+  lwu(tmpL, Address(strL));
+  load_long_misaligned(tmpU, Address(strU), tmp3, (base_offset % 8) != 0 ? 4 : 8);
+  subi(cnt2, cnt2, 4);
+  add(strL, strL, cnt2);
+  sub(cnt1, zr, cnt2);
+  slli(cnt2, cnt2, 1);
+  add(strU, strU, cnt2);
+  inflate_lo32(tmp3, tmpL);
+  mv(tmpL, tmp3);
+  sub(cnt2, zr, cnt2);
+  addi(cnt1, cnt1, 4);
+  addi(cnt2, cnt2, 8);
+  bne(tmpL, tmpU, DIFFERENCE);
+  bgez(cnt2, TAIL);
+
+  // main loop
+  bind(NEXT_WORD);
+    add(t0, strL, cnt1);
+    lwu(tmpL, Address(t0));
+    add(t0, strU, cnt2);
+    load_long_misaligned(tmpU, Address(t0), tmp3, (base_offset % 8) != 0 ? 4 : 8);
+    addi(cnt1, cnt1, 4);
+    inflate_lo32(tmp3, tmpL);
+    mv(tmpL, tmp3);
+    addi(cnt2, cnt2, 8);
+    bne(tmpL, tmpU, DIFFERENCE);
+    bltz(cnt2, NEXT_WORD);
+
+  bind(TAIL);
+  load_int_misaligned(tmpL, Address(strL), tmp3, false);
+  load_long_misaligned(tmpU, Address(strU), tmp3, 2);
+  inflate_lo32(tmp3, tmpL);
+  mv(tmpL, tmp3);
+
+  beq(tmpL, tmpU, *DONE);
+
+  // Find the first different characters in the longwords and
+  // compute their difference.
+  bind(DIFFERENCE);
+  xorr(tmp3, tmpL, tmpU);
+  // count bits of trailing zero chars
+  ctzc_bits(result, tmp3);
+  srl(tmpL, tmpL, result);
+  srl(tmpU, tmpU, result);
+  zext(tmpL, tmpL, 16);
+  zext(tmpU, tmpU, 16);
+  if (isLU) {
+    sub(result, tmpL, tmpU);
+  } else {
+    sub(result, tmpU, tmpL);
+  }
+
+  j(*DONE);
+}
+
 // Compare strings.
 void C2_MacroAssembler::string_compare(Register str1, Register str2,
                                        Register cnt1, Register cnt2, Register result,
                                        Register tmp1, Register tmp2, Register tmp3,
                                        int ae)
 {
-  Label DONE, SHORT_LOOP, SHORT_STRING, SHORT_LAST, TAIL, STUB,
-        DIFFERENCE, NEXT_WORD, SHORT_LOOP_TAIL, SHORT_LAST2, SHORT_LAST_INIT,
-        SHORT_LOOP_START, TAIL_CHECK, L;
+  Label DONE, SHORT_LOOP, SHORT_STRING, SHORT_LAST, STUB,
+        SHORT_LOOP_TAIL, SHORT_LAST2, SHORT_LAST_INIT,
+        SHORT_LOOP_START, L;
 
   const int STUB_THRESHOLD = 64 + 8;
   bool isLL = ae == StrIntrinsicNode::LL;
@@ -1409,14 +1577,6 @@ void C2_MacroAssembler::string_compare(Register str1, Register str2,
   load_chr_insn str1_load_chr = str1_isL ? (load_chr_insn)&MacroAssembler::lbu : (load_chr_insn)&MacroAssembler::lhu;
   load_chr_insn str2_load_chr = str2_isL ? (load_chr_insn)&MacroAssembler::lbu : (load_chr_insn)&MacroAssembler::lhu;
 
-  int base_offset1 = arrayOopDesc::base_offset_in_bytes(T_BYTE);
-  int base_offset2 = arrayOopDesc::base_offset_in_bytes(T_CHAR);
-
-  assert((base_offset1 % (UseCompactObjectHeaders ? 4 :
-                          (UseCompressedClassPointers ? 8 : 4))) == 0, "Must be");
-  assert((base_offset2 % (UseCompactObjectHeaders ? 4 :
-                          (UseCompressedClassPointers ? 8 : 4))) == 0, "Must be");
-
   BLOCK_COMMENT("string_compare {");
 
   // Bizarrely, the counts are passed in bytes, regardless of whether they
@@ -1434,154 +1594,23 @@ void C2_MacroAssembler::string_compare(Register str1, Register str2,
   mv(cnt2, cnt1);
   bind(L);
 
-  // Load 4 bytes once to compare for alignment before main loop. Note that this
-  // is only possible for LL/UU case. We need to resort to load_long_misaligned
-  // for both LU and UL cases.
-  if (str1_isL == str2_isL) { // LL or UU
-    beq(str1, str2, DONE);
-    int base_offset = isLL ? base_offset1 : base_offset2;
-    if (AvoidUnalignedAccesses && (base_offset % 8) != 0) {
-      mv(t0, minCharsInWord / 2);
-      ble(cnt2, t0, SHORT_STRING);
-      lwu(tmp1, Address(str1));
-      lwu(tmp2, Address(str2));
-      bne(tmp1, tmp2, DIFFERENCE);
-      addi(str1, str1, 4);
-      addi(str2, str2, 4);
-      subi(cnt2, cnt2, minCharsInWord / 2);
-    }
-  }
-
   // A very short string
   mv(t0, minCharsInWord);
   ble(cnt2, t0, SHORT_STRING);
 
   // Compare longwords
-  // load first parts of strings and finish initialization while loading
   {
     if (str1_isL == str2_isL) { // LL or UU
-#ifdef ASSERT
-      if (AvoidUnalignedAccesses) {
-        Label align_ok;
-        orr(t0, str1, str2);
-        andi(t0, t0, 0x7);
-        beqz(t0, align_ok);
-        stop("bad alignment");
-        bind(align_ok);
-      }
-#endif
-      // load 8 bytes once to compare
-      ld(tmp1, Address(str1));
-      ld(tmp2, Address(str2));
-      mv(t0, STUB_THRESHOLD);
-      bge(cnt2, t0, STUB);
-      subi(cnt2, cnt2, minCharsInWord);
-      beqz(cnt2, TAIL_CHECK);
-      // convert cnt2 from characters to bytes
-      if (!str1_isL) {
-        slli(cnt2, cnt2, 1);
-      }
-      add(str2, str2, cnt2);
-      add(str1, str1, cnt2);
-      sub(cnt2, zr, cnt2);
-    } else if (isLU) { // LU case
-      mv(t0, STUB_THRESHOLD);
-      bge(cnt2, t0, STUB);
-      lwu(tmp1, Address(str1));
-      load_long_misaligned(tmp2, Address(str2), tmp3, (base_offset2 % 8) != 0 ? 4 : 8);
-      subi(cnt2, cnt2, 4);
-      add(str1, str1, cnt2);
-      sub(cnt1, zr, cnt2);
-      slli(cnt2, cnt2, 1);
-      add(str2, str2, cnt2);
-      inflate_lo32(tmp3, tmp1);
-      mv(tmp1, tmp3);
-      sub(cnt2, zr, cnt2);
-      addi(cnt1, cnt1, 4);
-    } else { // UL case
-      mv(t0, STUB_THRESHOLD);
-      bge(cnt2, t0, STUB);
-      load_long_misaligned(tmp1, Address(str1), tmp3, (base_offset2 % 8) != 0 ? 4 : 8);
-      lwu(tmp2, Address(str2));
-      subi(cnt2, cnt2, 4);
-      slli(t0, cnt2, 1);
-      sub(cnt1, zr, t0);
-      add(str1, str1, t0);
-      add(str2, str2, cnt2);
-      inflate_lo32(tmp3, tmp2);
-      mv(tmp2, tmp3);
-      sub(cnt2, zr, cnt2);
-      addi(cnt1, cnt1, 8);
-    }
-    addi(cnt2, cnt2, isUL ? 4 : 8);
-    bne(tmp1, tmp2, DIFFERENCE);
-    bgez(cnt2, TAIL);
-
-    // main loop
-    bind(NEXT_WORD);
-    if (str1_isL == str2_isL) { // LL or UU
-      // 8-byte aligned loads when AvoidUnalignedAccesses is enabled
-      add(t0, str1, cnt2);
-      ld(tmp1, Address(t0));
-      add(t0, str2, cnt2);
-      ld(tmp2, Address(t0));
-      addi(cnt2, cnt2, 8);
-    } else if (isLU) { // LU case
-      add(t0, str1, cnt1);
-      lwu(tmp1, Address(t0));
-      add(t0, str2, cnt2);
-      load_long_misaligned(tmp2, Address(t0), tmp3, (base_offset2 % 8) != 0 ? 4 : 8);
-      addi(cnt1, cnt1, 4);
-      inflate_lo32(tmp3, tmp1);
-      mv(tmp1, tmp3);
-      addi(cnt2, cnt2, 8);
-    } else { // UL case
-      add(t0, str2, cnt2);
-      lwu(tmp2, Address(t0));
-      add(t0, str1, cnt1);
-      load_long_misaligned(tmp1, Address(t0), tmp3, (base_offset2 % 8) != 0 ? 4 : 8);
-      inflate_lo32(tmp3, tmp2);
-      mv(tmp2, tmp3);
-      addi(cnt1, cnt1, 8);
-      addi(cnt2, cnt2, 4);
+      string_compare_long_same_encoding(result,
+                                str1, str2, isLL,
+                                cnt1, cnt2, tmp1, tmp2, tmp3,
+                                STUB_THRESHOLD, &STUB, &SHORT_STRING, &DONE);
+    } else { // LU or UL
+      string_compare_long_different_encoding(result,
+                                str1, str2, isLU,
+                                cnt1, cnt2, tmp1, tmp2, tmp3,
+                                STUB_THRESHOLD, &STUB, &DONE);
     }
-    bne(tmp1, tmp2, DIFFERENCE);
-    bltz(cnt2, NEXT_WORD);
-    bind(TAIL);
-    if (str1_isL == str2_isL) { // LL or UU
-      load_long_misaligned(tmp1, Address(str1), tmp3, isLL ? 1 : 2);
-      load_long_misaligned(tmp2, Address(str2), tmp3, isLL ? 1 : 2);
-    } else if (isLU) { // LU case
-      load_int_misaligned(tmp1, Address(str1), tmp3, false);
-      load_long_misaligned(tmp2, Address(str2), tmp3, 2);
-      inflate_lo32(tmp3, tmp1);
-      mv(tmp1, tmp3);
-    } else { // UL case
-      load_int_misaligned(tmp2, Address(str2), tmp3, false);
-      load_long_misaligned(tmp1, Address(str1), tmp3, 2);
-      inflate_lo32(tmp3, tmp2);
-      mv(tmp2, tmp3);
-    }
-    bind(TAIL_CHECK);
-    beq(tmp1, tmp2, DONE);
-
-    // Find the first different characters in the longwords and
-    // compute their difference.
-    bind(DIFFERENCE);
-    xorr(tmp3, tmp1, tmp2);
-    // count bits of trailing zero chars
-    ctzc_bits(result, tmp3, isLL);
-    srl(tmp1, tmp1, result);
-    srl(tmp2, tmp2, result);
-    if (isLL) {
-      zext(tmp1, tmp1, 8);
-      zext(tmp2, tmp2, 8);
-    } else {
-      zext(tmp1, tmp1, 16);
-      zext(tmp2, tmp2, 16);
-    }
-    sub(result, tmp1, tmp2);
-    j(DONE);
   }
 
   bind(STUB);
@@ -2638,7 +2667,7 @@ void C2_MacroAssembler::string_compare_v(Register str1, Register str2, Register
 
   int minCharsInWord = encLL ? wordSize : wordSize / 2;
 
-  BLOCK_COMMENT("string_compare {");
+  BLOCK_COMMENT("string_compare_v {");
 
   // for Latin strings, 1 byte for 1 character
   // for UTF16 strings, 2 bytes for 1 character
@@ -2698,6 +2727,8 @@ void C2_MacroAssembler::string_compare_v(Register str1, Register str2, Register
   sub(result, tmp1, tmp2);
 
   bind(DONE);
+
+  BLOCK_COMMENT("} string_compare_v");
 }
 
 void C2_MacroAssembler::byte_array_inflate_v(Register src, Register dst, Register len, Register tmp) {
diff --git a/src/hotspot/cpu/riscv/c2_MacroAssembler_riscv.hpp b/src/hotspot/cpu/riscv/c2_MacroAssembler_riscv.hpp
index c79c360d2eb67..72f3c1460f3a0 100644
--- a/src/hotspot/cpu/riscv/c2_MacroAssembler_riscv.hpp
+++ b/src/hotspot/cpu/riscv/c2_MacroAssembler_riscv.hpp
@@ -39,6 +39,15 @@
                        VectorRegister vrs,
                        bool is_latin, Label& DONE, Assembler::LMUL lmul);
 
+  void string_compare_long_same_encoding(Register result, Register str1, Register str2,
+                                  const bool isLL, Register cnt1, Register cnt2,
+                                  Register tmp1, Register tmp2, Register tmp3,
+                                  const int STUB_THRESHOLD, Label *STUB, Label *SHORT_STRING, Label *DONE);
+  void string_compare_long_different_encoding(Register result, Register str1, Register str2,
+                                  bool isLU, Register cnt1, Register cnt2,
+                                  Register tmp1, Register tmp2, Register tmp3,
+                                  const int STUB_THRESHOLD, Label *STUB, Label *DONE);
+
  public:
   // Code used by cmpFastLock and cmpFastUnlock mach instructions in .ad file.
   void fast_lock(Register object, Register box,
diff --git a/src/hotspot/cpu/riscv/stubGenerator_riscv.cpp b/src/hotspot/cpu/riscv/stubGenerator_riscv.cpp
index 01ed7fcac9f87..249b871bd00bc 100644
--- a/src/hotspot/cpu/riscv/stubGenerator_riscv.cpp
+++ b/src/hotspot/cpu/riscv/stubGenerator_riscv.cpp
@@ -2431,23 +2431,6 @@ class StubGenerator: public StubCodeGenerator {
     return start;
   }
 
-  // code for comparing 16 bytes of strings with same encoding
-  void compare_string_16_bytes_same(Label &DIFF1, Label &DIFF2) {
-    const Register result = x10, str1 = x11, cnt1 = x12, str2 = x13, tmp1 = x28, tmp2 = x29, tmp4 = x7, tmp5 = x31;
-    __ ld(tmp5, Address(str1));
-    __ addi(str1, str1, 8);
-    __ xorr(tmp4, tmp1, tmp2);
-    __ ld(cnt1, Address(str2));
-    __ addi(str2, str2, 8);
-    __ bnez(tmp4, DIFF1);
-    __ ld(tmp1, Address(str1));
-    __ addi(str1, str1, 8);
-    __ xorr(tmp4, tmp5, cnt1);
-    __ ld(tmp2, Address(str2));
-    __ addi(str2, str2, 8);
-    __ bnez(tmp4, DIFF2);
-  }
-
   // code for comparing 8 characters of strings with Latin1 and Utf16 encoding
   void compare_string_8_x_LU(Register tmpL, Register tmpU,
                              Register strL, Register strU, Label& DIFF) {
@@ -2702,7 +2685,20 @@ class StubGenerator: public StubCodeGenerator {
     __ push_reg(spilled_regs, sp);
     __ bltz(cnt2, TAIL);
     __ bind(SMALL_LOOP);
-      compare_string_16_bytes_same(DIFF, DIFF2);
+      // compare 16 bytes of strings with same encoding
+      __ ld(tmp5, Address(str1));
+      __ addi(str1, str1, 8);
+      __ xorr(tmp4, tmp1, tmp2);
+      __ ld(cnt1, Address(str2));
+      __ addi(str2, str2, 8);
+      __ bnez(tmp4, DIFF);
+      __ ld(tmp1, Address(str1));
+      __ addi(str1, str1, 8);
+      __ xorr(tmp4, tmp5, cnt1);
+      __ ld(tmp2, Address(str2));
+      __ addi(str2, str2, 8);
+      __ bnez(tmp4, DIFF2);
+
       __ subi(cnt2, cnt2, isLL ? 16 : 8);
       __ bgez(cnt2, SMALL_LOOP);
     __ bind(TAIL);
diff --git a/test/hotspot/jtreg/compiler/intrinsics/string/TestStringIntrinsics.java b/test/hotspot/jtreg/compiler/intrinsics/string/TestStringIntrinsics.java
index 062866d657f96..f82a4899d97c1 100644
--- a/test/hotspot/jtreg/compiler/intrinsics/string/TestStringIntrinsics.java
+++ b/test/hotspot/jtreg/compiler/intrinsics/string/TestStringIntrinsics.java
@@ -160,6 +160,7 @@ private void checkIntrinsics(Operation op, Method m, String latin1, String utf16
 
                 // Different lengths
                 invokeAndCheck(m, 1, "ABCD", "ABC");
+                invokeAndCheck(m, '\uff21' - 'A', "ABCEFGHIJKLMNOPQRSTUVWXY\uff21Z", "ABCEFGHIJKLMNOPQRSTUVWXYAZ");
                 invokeAndCheck(m, -1, "\uff21\uff22\uff23", "\uff21\uff22\uff23\uff24");
                 invokeAndCheck(m, 1, "ABC\uff24", "ABC");
                 invokeAndCheck(m, 3, "ABC\uff24\uff25\uff26", "ABC");