diff --git a/core/arch/arm/kernel/link_dummies.c b/core/arch/arm/kernel/link_dummies.c
index 373db8d2edd..12b7f9acf37 100644
--- a/core/arch/arm/kernel/link_dummies.c
+++ b/core/arch/arm/kernel/link_dummies.c
@@ -22,9 +22,11 @@ TEE_Result __section(".text.dummy.init_teecore") init_teecore(void)
 	return TEE_SUCCESS;
 }
 
-void __section(".text.dummy.__thread_std_smc_entry")
-__thread_std_smc_entry(struct thread_smc_args *args __unused)
+uint32_t __section(".text.dummy.__thread_std_smc_entry")
+__thread_std_smc_entry(uint32_t a0 __unused, uint32_t a1 __unused,
+		       uint32_t a2 __unused, uint32_t a3 __unused)
 {
+	return 0;
 }
 void __section(".text.dummy.__wq_rpc")
 __wq_rpc(uint32_t func __unused, int id __unused,
diff --git a/core/arch/arm/kernel/thread_optee_smc.c b/core/arch/arm/kernel/thread_optee_smc.c
index f6d0e73bce4..8c068862626 100644
--- a/core/arch/arm/kernel/thread_optee_smc.c
+++ b/core/arch/arm/kernel/thread_optee_smc.c
@@ -153,44 +153,44 @@ static struct mobj *map_cmd_buffer(paddr_t pargi __unused,
 }
 #endif /*CFG_CORE_DYN_SHM*/
 
-static void std_smc_entry(struct thread_smc_args *smc_args)
+static uint32_t std_smc_entry(uint32_t a0, uint32_t a1, uint32_t a2,
+			      uint32_t a3 __unused)
 {
 	paddr_t parg = 0;
 	struct optee_msg_arg *arg = NULL;
 	uint32_t num_params = 0;
 	struct mobj *mobj = NULL;
+	uint32_t rv = 0;
 
-	if (smc_args->a0 != OPTEE_SMC_CALL_WITH_ARG) {
-		EMSG("Unknown SMC 0x%" PRIx64, (uint64_t)smc_args->a0);
+	if (a0 != OPTEE_SMC_CALL_WITH_ARG) {
+		EMSG("Unknown SMC 0x%"PRIx32, a0);
 		DMSG("Expected 0x%x", OPTEE_SMC_CALL_WITH_ARG);
-		smc_args->a0 = OPTEE_SMC_RETURN_EBADCMD;
-		return;
+		return OPTEE_SMC_RETURN_EBADCMD;
 	}
-	parg = (uint64_t)smc_args->a1 << 32 | smc_args->a2;
+	parg = reg_pair_to_64(a1, a2);
 
 	/* Check if this region is in static shared space */
 	if (core_pbuf_is(CORE_MEM_NSEC_SHM, parg,
 			 sizeof(struct optee_msg_arg))) {
 		mobj = get_cmd_buffer(parg, &num_params);
 	} else {
-		if (parg & SMALL_PAGE_MASK) {
-			smc_args->a0 = OPTEE_SMC_RETURN_EBADADDR;
-			return;
-		}
+		if (parg & SMALL_PAGE_MASK)
+			return OPTEE_SMC_RETURN_EBADADDR;
 		mobj = map_cmd_buffer(parg, &num_params);
 	}
 
 	if (!mobj || !ALIGNMENT_IS_OK(parg, struct optee_msg_arg)) {
 		EMSG("Bad arg address 0x%" PRIxPA, parg);
-		smc_args->a0 = OPTEE_SMC_RETURN_EBADADDR;
 		mobj_free(mobj);
-		return;
+		return OPTEE_SMC_RETURN_EBADADDR;
 	}
 
 	arg = mobj_get_va(mobj, 0);
 	assert(arg && mobj_is_nonsec(mobj));
-	smc_args->a0 = tee_entry_std(arg, num_params);
+	rv = tee_entry_std(arg, num_params);
 	mobj_free(mobj);
+
+	return rv;
 }
 
 /*
@@ -199,14 +199,17 @@ static void std_smc_entry(struct thread_smc_args *smc_args)
  * Note: this function is weak just to make it possible to exclude it from
  * the unpaged area.
  */
-void __weak __thread_std_smc_entry(struct thread_smc_args *args)
+uint32_t __weak __thread_std_smc_entry(uint32_t a0, uint32_t a1, uint32_t a2,
+				       uint32_t a3)
 {
+	uint32_t rv = 0;
+
 #ifdef CFG_VIRTUALIZATION
 	virt_on_stdcall();
 #endif
-	std_smc_entry(args);
+	rv = std_smc_entry(a0, a1, a2, a3);
 
-	if (args->a0 == OPTEE_SMC_RETURN_OK) {
+	if (rv == OPTEE_SMC_RETURN_OK) {
 		struct thread_ctx *thr = threads + thread_get_id();
 
 		tee_fs_rpc_cache_clear(&thr->tsd);
@@ -217,6 +220,8 @@ void __weak __thread_std_smc_entry(struct thread_smc_args *args)
 			thr->rpc_mobj = NULL;
 		}
 	}
+
+	return rv;
 }
 
 bool thread_disable_prealloc_rpc_cache(uint64_t *cookie)
diff --git a/core/arch/arm/kernel/thread_optee_smc_a32.S b/core/arch/arm/kernel/thread_optee_smc_a32.S
index cb7fbfafa65..af6e8073ea0 100644
--- a/core/arch/arm/kernel/thread_optee_smc_a32.S
+++ b/core/arch/arm/kernel/thread_optee_smc_a32.S
@@ -161,17 +161,8 @@ KEEP_PAGER thread_vector_table
 FUNC thread_std_smc_entry , :
 UNWIND(	.fnstart)
 UNWIND(	.cantunwind)
-	/* Pass r0-r7 in a struct thread_smc_args */
-	push	{r0-r7}
-	mov	r0, sp
 	bl	__thread_std_smc_entry
-	/*
-	 * Load the returned r0-r3 into preserved registers and skip the
-	 * "returned" r4-r7 since they will not be returned to normal
-	 * world.
-	 */
-	pop	{r4-r7}
-	add	sp, #(4 * 4)
+	mov	r4, r0	/* Save return value for later */
 
 	/* Disable interrupts before switching to temporary stack */
 	cpsid	aif
@@ -182,9 +173,9 @@ UNWIND(	.cantunwind)
 
 	ldr	r0, =TEESMC_OPTEED_RETURN_CALL_DONE
 	mov	r1, r4
-	mov	r2, r5
-	mov	r3, r6
-	mov	r4, r7
+	mov	r2, #0
+	mov	r3, #0
+	mov	r4, #0
 	smc	#0
 	b	.	/* SMC should not return */
 UNWIND(	.fnend)
diff --git a/core/arch/arm/kernel/thread_optee_smc_a64.S b/core/arch/arm/kernel/thread_optee_smc_a64.S
index 49aa05b36a2..0019baf7d5c 100644
--- a/core/arch/arm/kernel/thread_optee_smc_a64.S
+++ b/core/arch/arm/kernel/thread_optee_smc_a64.S
@@ -133,21 +133,8 @@ END_FUNC thread_vector_table
 KEEP_PAGER thread_vector_table
 
 FUNC thread_std_smc_entry , :
-	/* pass x0-x7 in a struct thread_smc_args */
-	sub	sp, sp, #THREAD_SMC_ARGS_SIZE
-	store_xregs sp, THREAD_SMC_ARGS_X0, 0, 7
-	mov	x0, sp
-
-	/* Call the registered handler */
 	bl	__thread_std_smc_entry
-
-	/*
-	 * Load the returned x0-x3 into preserved registers and skip the
-	 * "returned" x4-x7 since they will not be returned to normal
-	 * world.
-	 */
-	load_xregs sp, THREAD_SMC_ARGS_X0, 20, 23
-	add	sp, sp, #THREAD_SMC_ARGS_SIZE
+	mov	w20, w0	/* Save return value for later */
 
 	/* Mask all maskable exceptions before switching to temporary stack */
 	msr	daifset, #DAIFBIT_ALL
@@ -157,10 +144,10 @@ FUNC thread_std_smc_entry , :
 	bl	thread_state_free
 
 	ldr	x0, =TEESMC_OPTEED_RETURN_CALL_DONE
-	mov	x1, x20
-	mov	x2, x21
-	mov	x3, x22
-	mov	x4, x23
+	mov	w1, w20
+	mov	x2, #0
+	mov	x3, #0
+	mov	x4, #0
 	smc	#0
 	b	.	/* SMC should not return */
 END_FUNC thread_std_smc_entry
diff --git a/core/arch/arm/kernel/thread_private.h b/core/arch/arm/kernel/thread_private.h
index cf9153ae210..1fbd13fe77e 100644
--- a/core/arch/arm/kernel/thread_private.h
+++ b/core/arch/arm/kernel/thread_private.h
@@ -147,8 +147,16 @@ void thread_excp_vect_workaround(void);
 void thread_excp_vect_workaround_a15(void);
 void thread_excp_vect_end(void);
 
-/* Handles a stdcall, r0-r7 holds the parameters */
-void thread_std_smc_entry(void);
+/*
+ * Assembly function as the first function in a thread.  Handles a stdcall,
+ * a0-a3 holds the parameters. Hands over to __thread_std_smc_entry() when
+ * everything is set up and does some post processing once
+ * __thread_std_smc_entry() returns.
+ */
+void thread_std_smc_entry(uint32_t a0, uint32_t a1, uint32_t a2, uint32_t a3);
+uint32_t __thread_std_smc_entry(uint32_t a0, uint32_t a1, uint32_t a2,
+				uint32_t a3);
+
 
 /*
  * Resumes execution of currently active thread by restoring context and
@@ -204,8 +212,6 @@ void thread_set_fiq_sp(vaddr_t sp);
 /* Checks stack canaries */
 void thread_check_canaries(void);
 
-void __thread_std_smc_entry(struct thread_smc_args *args);
-
 void thread_alloc_and_run(uint32_t a0, uint32_t a1, uint32_t a2, uint32_t a3);
 void thread_resume_from_rpc(uint32_t a0, uint32_t a1, uint32_t a2, uint32_t a3,
 			    uint32_t a4, uint32_t a5);