diff --git a/tcg/tcg.c b/tcg/tcg.c
@@ -94,6 +94,19 @@ typedef struct QEMU_PACKED {
     DebugFrameFDEHeader fde;
 } DebugFrameHeader;
 
+typedef struct TCGLabelQemuLdst {
+    bool is_ld;             /* qemu_ld: true, qemu_st: false */
+    MemOpIdx oi;
+    TCGType type;           /* result type of a load */
+    TCGReg addrlo_reg;      /* reg index for low word of guest virtual addr */
+    TCGReg addrhi_reg;      /* reg index for high word of guest virtual addr */
+    TCGReg datalo_reg;      /* reg index for low word to be loaded or stored */
+    TCGReg datahi_reg;      /* reg index for high word to be loaded or stored */
+    const tcg_insn_unit *raddr;   /* addr of the next IR of qemu_ld/st IR */
+    tcg_insn_unit *label_ptr[2]; /* label pointers to be updated */
+    QSIMPLEQ_ENTRY(TCGLabelQemuLdst) next;
+} TCGLabelQemuLdst;
+
 static void tcg_register_jit_int(const void *buf, size_t size,
                                  const void *debug_frame,
                                  size_t debug_frame_size)
@@ -793,6 +806,25 @@ static void init_ffi_layouts(void)
 }
 #endif /* CONFIG_TCG_INTERPRETER */
 
+static inline bool arg_slot_reg_p(unsigned arg_slot)
+{
+    /*
+     * Split the sizeof away from the comparison to avoid Werror from
+     * "unsigned < 0 is always false", when iarg_regs is empty.
+     */
+    unsigned nreg = ARRAY_SIZE(tcg_target_call_iarg_regs);
+    return arg_slot < nreg;
+}
+
+static inline int arg_slot_stk_ofs(unsigned arg_slot)
+{
+    unsigned max = TCG_STATIC_CALL_ARGS_SIZE / sizeof(tcg_target_long);
+    unsigned stk_slot = arg_slot - ARRAY_SIZE(tcg_target_call_iarg_regs);
+
+    tcg_debug_assert(stk_slot < max);
+    return TCG_TARGET_CALL_STACK_OFFSET + stk_slot * sizeof(tcg_target_long);
+}
+
 typedef struct TCGCumulativeArgs {
     int arg_idx;                /* tcg_gen_callN args[] */
     int info_in_idx;            /* TCGHelperInfo in[] */
@@ -1032,6 +1064,7 @@ static void init_call_layout(TCGHelperInfo *info)
             }
         }
         assert(ref_base + cum.ref_slot <= max_stk_slots);
+        ref_base += max_reg_slots;
 
         if (ref_base != 0) {
             for (int i = cum.info_in_idx - 1; i >= 0; --i) {
@@ -3218,7 +3251,7 @@ liveness_pass_1(TCGContext *s)
                         case TCG_CALL_ARG_NORMAL:
                         case TCG_CALL_ARG_EXTEND_U:
                         case TCG_CALL_ARG_EXTEND_S:
-                            if (REG_P(loc)) {
+                            if (arg_slot_reg_p(loc->arg_slot)) {
                                 *la_temp_pref(ts) = 0;
                                 break;
                             }
@@ -3245,7 +3278,7 @@ liveness_pass_1(TCGContext *s)
                     case TCG_CALL_ARG_NORMAL:
                     case TCG_CALL_ARG_EXTEND_U:
                     case TCG_CALL_ARG_EXTEND_S:
-                        if (REG_P(loc)) {
+                        if (arg_slot_reg_p(loc->arg_slot)) {
                             tcg_regset_set_reg(*la_temp_pref(ts),
                                 tcg_target_call_iarg_regs[loc->arg_slot]);
                         }
@@ -4803,7 +4836,7 @@ static void load_arg_reg(TCGContext *s, TCGReg reg, TCGTemp *ts,
     }
 }
 
-static void load_arg_stk(TCGContext *s, int stk_slot, TCGTemp *ts,
+static void load_arg_stk(TCGContext *s, unsigned arg_slot, TCGTemp *ts,
                          TCGRegSet allocated_regs)
 {
     /*
@@ -4813,30 +4846,27 @@ static void load_arg_stk(TCGContext *s, int stk_slot, TCGTemp *ts,
      */
     temp_load(s, ts, tcg_target_available_regs[ts->type], allocated_regs, 0);
     tcg_out_st(s, ts->type, ts->reg, TCG_REG_CALL_STACK,
-               TCG_TARGET_CALL_STACK_OFFSET +
-               stk_slot * sizeof(tcg_target_long));
+               arg_slot_stk_ofs(arg_slot));
 }
 
 static void load_arg_normal(TCGContext *s, const TCGCallArgumentLoc *l,
                             TCGTemp *ts, TCGRegSet *allocated_regs)
 {
-    if (REG_P(l)) {
+    if (arg_slot_reg_p(l->arg_slot)) {
         TCGReg reg = tcg_target_call_iarg_regs[l->arg_slot];
         load_arg_reg(s, reg, ts, *allocated_regs);
         tcg_regset_set_reg(*allocated_regs, reg);
     } else {
-        load_arg_stk(s, l->arg_slot - ARRAY_SIZE(tcg_target_call_iarg_regs),
-                     ts, *allocated_regs);
+        load_arg_stk(s, l->arg_slot, ts, *allocated_regs);
     }
 }
 
-static void load_arg_ref(TCGContext *s, int arg_slot, TCGReg ref_base,
+static void load_arg_ref(TCGContext *s, unsigned arg_slot, TCGReg ref_base,
                          intptr_t ref_off, TCGRegSet *allocated_regs)
 {
     TCGReg reg;
-    int stk_slot = arg_slot - ARRAY_SIZE(tcg_target_call_iarg_regs);
 
-    if (stk_slot < 0) {
+    if (arg_slot_reg_p(arg_slot)) {
         reg = tcg_target_call_iarg_regs[arg_slot];
         tcg_reg_free(s, reg, *allocated_regs);
         tcg_out_addi_ptr(s, reg, ref_base, ref_off);
@@ -4846,8 +4876,7 @@ static void load_arg_ref(TCGContext *s, int arg_slot, TCGReg ref_base,
                             *allocated_regs, 0, false);
         tcg_out_addi_ptr(s, reg, ref_base, ref_off);
         tcg_out_st(s, TCG_TYPE_PTR, reg, TCG_REG_CALL_STACK,
-                   TCG_TARGET_CALL_STACK_OFFSET
-                   + stk_slot * sizeof(tcg_target_long));
+                   arg_slot_stk_ofs(arg_slot));
     }
 }
 
@@ -4877,8 +4906,7 @@ static void tcg_reg_alloc_call(TCGContext *s, TCGOp *op)
         case TCG_CALL_ARG_BY_REF:
             load_arg_stk(s, loc->ref_slot, ts, allocated_regs);
             load_arg_ref(s, loc->arg_slot, TCG_REG_CALL_STACK,
-                         TCG_TARGET_CALL_STACK_OFFSET
-                         + loc->ref_slot * sizeof(tcg_target_long),
+                         arg_slot_stk_ofs(loc->ref_slot),
                          &allocated_regs);
             break;
         case TCG_CALL_ARG_BY_REF_N:

diff --git a/tests/tcg/ppc64/Makefile.target b/tests/tcg/ppc64/Makefile.target
@@ -20,7 +20,7 @@ PPC64_TESTS += mtfsf
 PPC64_TESTS += mffsce
 
 ifneq ($(CROSS_CC_HAS_POWER10),)
-PPC64_TESTS += byte_reverse sha512-vector
+PPC64_TESTS += byte_reverse sha512-vector vector
 endif
 byte_reverse: CFLAGS += -mcpu=power10
 run-byte_reverse: QEMU_OPTS+=-cpu POWER10
@@ -31,6 +31,9 @@ sha512-vector: sha512.c
 
 run-sha512-vector: QEMU_OPTS+=-cpu POWER10
 
+vector: CFLAGS += -mcpu=power10 -I$(SRC_PATH)/include
+run-vector: QEMU_OPTS += -cpu POWER10
+
 PPC64_TESTS += signal_save_restore_xer
 PPC64_TESTS += xxspltw
 

diff --git a/tests/tcg/ppc64/vector.c b/tests/tcg/ppc64/vector.c
@@ -0,0 +1,51 @@
+#include <assert.h>
+#include <stdint.h>
+#include "qemu/compiler.h"
+
+int main(void)
+{
+    unsigned int result_wi;
+    vector unsigned char vbc_bi_src = { 0xFF, 0xFF, 0, 0xFF, 0xFF, 0xFF,
+                                        0xFF, 0xFF, 0xFF, 0xFF, 0, 0, 0,
+                                        0, 0xFF, 0xFF};
+    vector unsigned short vbc_hi_src = { 0xFFFF, 0, 0, 0xFFFF,
+                                         0, 0, 0xFFFF, 0xFFFF};
+    vector unsigned int vbc_wi_src = {0, 0, 0xFFFFFFFF, 0xFFFFFFFF};
+    vector unsigned long long vbc_di_src = {0xFFFFFFFFFFFFFFFF, 0};
+    vector __uint128_t vbc_qi_src;
+
+    asm("vextractbm %0, %1" : "=r" (result_wi) : "v" (vbc_bi_src));
+#if HOST_BIG_ENDIAN
+    assert(result_wi == 0b1101111111000011);
+#else
+    assert(result_wi == 0b1100001111111011);
+#endif
+
+    asm("vextracthm %0, %1" : "=r" (result_wi) : "v" (vbc_hi_src));
+#if HOST_BIG_ENDIAN
+    assert(result_wi == 0b10010011);
+#else
+    assert(result_wi == 0b11001001);
+#endif
+
+    asm("vextractwm %0, %1" : "=r" (result_wi) : "v" (vbc_wi_src));
+#if HOST_BIG_ENDIAN
+    assert(result_wi == 0b0011);
+#else
+    assert(result_wi == 0b1100);
+#endif
+
+    asm("vextractdm %0, %1" : "=r" (result_wi) : "v" (vbc_di_src));
+#if HOST_BIG_ENDIAN
+    assert(result_wi == 0b10);
+#else
+    assert(result_wi == 0b01);
+#endif
+
+    vbc_qi_src[0] = 0x1;
+    vbc_qi_src[0] = vbc_qi_src[0] << 127;
+    asm("vextractqm %0, %1" : "=r" (result_wi) : "v" (vbc_qi_src));
+    assert(result_wi == 0b1);
+
+    return 0;
+}