diff --git a/arm-gen.c b/arm-gen.c
index a2974d5a..b925fe33 100644
--- a/arm-gen.c
+++ b/arm-gen.c
@@ -2315,7 +2315,12 @@ ST_FUNC void gen_vla_sp_restore(int addr) {
 
 /* Subtract from the stack pointer, and push the resulting value onto the stack */
 ST_FUNC void gen_vla_alloc(CType *type, int align) {
-    int r = intr(gv(RC_INT));
+    int r;
+#if defined(CONFIG_TCC_BCHECK)
+    if (tcc_state->do_bounds_check)
+        vpushv(vtop);
+#endif
+    r = intr(gv(RC_INT));
     o(0xE04D0000|(r<<12)|r); /* sub r, sp, r */
 #ifdef TCC_ARM_EABI
     if (align < 8)
@@ -2328,6 +2333,18 @@ ST_FUNC void gen_vla_alloc(CType *type, int align) {
         tcc_error("alignment is not a power of 2: %i", align);
     o(stuff_const(0xE3C0D000|(r<<16), align - 1)); /* bic sp, r, #align-1 */
     vpop();
+#if defined(CONFIG_TCC_BCHECK)
+    if (tcc_state->do_bounds_check) {
+        vpushi(0);
+        vtop->r = TREG_R0;
+        o(0xe1a0000d | (vtop->r << 12)); // mov r0,sp
+        vswap();
+        vpush_global_sym(&func_old_type, TOK___bound_new_region);
+        vrott(3);
+        gfunc_call(2);
+        func_bound_add_epilog = 1;
+    }
+#endif
 }
 
 /* end of ARM code generator */
diff --git a/arm64-gen.c b/arm64-gen.c
index 8edb9e80..34659c5b 100644
--- a/arm64-gen.c
+++ b/arm64-gen.c
@@ -2049,11 +2049,28 @@ ST_FUNC void gen_vla_sp_restore(int addr) {
 }
 
 ST_FUNC void gen_vla_alloc(CType *type, int align) {
-    uint32_t r = intr(gv(RC_INT));
+    uint32_t r;
+#if defined(CONFIG_TCC_BCHECK)
+    if (tcc_state->do_bounds_check)
+        vpushv(vtop);
+#endif
+    r = intr(gv(RC_INT));
     o(0x91003c00 | r | r << 5); // add x(r),x(r),#15
     o(0x927cec00 | r | r << 5); // bic x(r),x(r),#15
     o(0xcb2063ff | r << 16); // sub sp,sp,x(r)
     vpop();
+#if defined(CONFIG_TCC_BCHECK)
+    if (tcc_state->do_bounds_check) {
+        vpushi(0);
+        vtop->r = TREG_R(0);
+        o(0x910003e0 | vtop->r); // mov r0,sp
+        vswap();
+        vpush_global_sym(&func_old_type, TOK___bound_new_region);
+        vrott(3);
+        gfunc_call(2);
+        func_bound_add_epilog = 1;
+    }
+#endif
 }
 
 /* end of A64 code generator */
diff --git a/riscv64-gen.c b/riscv64-gen.c
index 4d7f366b..37677027 100644
--- a/riscv64-gen.c
+++ b/riscv64-gen.c
@@ -1355,10 +1355,27 @@ ST_FUNC void gen_vla_sp_restore(int addr)
 
 ST_FUNC void gen_vla_alloc(CType *type, int align)
 {
-    int rr = ireg(gv(RC_INT));
+    int rr;
+#if defined(CONFIG_TCC_BCHECK)
+    if (tcc_state->do_bounds_check)
+        vpushv(vtop);
+#endif
+    rr = ireg(gv(RC_INT));
     EI(0x13, 0, rr, rr, 15);   // addi RR, RR, 15
     EI(0x13, 7, rr, rr, -16);  // andi, RR, RR, -16
     ER(0x33, 0, 2, 2, rr, 0x20); // sub sp, sp, rr
     vpop();
+#if defined(CONFIG_TCC_BCHECK)
+    if (tcc_state->do_bounds_check) {
+        vpushi(0);
+        vtop->r = TREG_R(0);
+        o(0x00010513); /* mv a0,sp */
+        vswap();
+        vpush_global_sym(&func_old_type, TOK___bound_new_region);
+        vrott(3);
+        gfunc_call(2);
+        func_bound_add_epilog = 1;
+    }
+#endif
 }
 #endif
diff --git a/tcctok.h b/tcctok.h
index 61e17485..7c2c2c14 100644
--- a/tcctok.h
+++ b/tcctok.h
@@ -306,6 +306,7 @@
      DEF(TOK___bound_local_new, "__bound_local_new")
      DEF(TOK___bound_local_delete, "__bound_local_delete")
      DEF(TOK___bound_setjmp, "__bound_setjmp")
+     DEF(TOK___bound_new_region, "__bound_new_region")
 # ifdef TCC_TARGET_PE
 #  ifdef TCC_TARGET_X86_64
      DEF(TOK___bound_alloca_nr, "__bound_alloca_nr")
diff --git a/tests/Makefile b/tests/Makefile
index afdcdcfb..dfda50c1 100644
--- a/tests/Makefile
+++ b/tests/Makefile
@@ -179,7 +179,7 @@ memtest:
 
 # memory and bound check auto test
 BOUNDS_OK  = 1 4 8 10 14 16
-BOUNDS_FAIL= 2 5 6 7 9 11 12 13 15 17
+BOUNDS_FAIL= 2 5 6 7 9 11 12 13 15 17 18
 
 btest: boundtest.c
 	@echo ------------ $@ ------------
diff --git a/tests/boundtest.c b/tests/boundtest.c
index 50f3d30e..0d833950 100644
--- a/tests/boundtest.c
+++ b/tests/boundtest.c
@@ -231,6 +231,16 @@ int test17()
     return 0;
 }
 
+int test18(void)
+{
+    int i, sum = 0, n = TAB_SIZE;
+    int tab[n];
+    for(i=0;i<TAB_SIZE+1;i++) {
+        sum += tab[i];
+    }
+    return sum;
+}
+
 int (*table_test[])(void) = {
     test1,
     test2,
@@ -248,7 +258,8 @@ int (*table_test[])(void) = {
     test14,
     test15,
     test16,
-    test17
+    test17,
+    test18
 };
 
 int main(int argc, char **argv)