diff --git a/components/lwp/arch/aarch64/cortex-a/lwp_gcc.S b/components/lwp/arch/aarch64/cortex-a/lwp_gcc.S
index 571f422a74..9c4b7ebe62 100644
--- a/components/lwp/arch/aarch64/cortex-a/lwp_gcc.S
+++ b/components/lwp/arch/aarch64/cortex-a/lwp_gcc.S
@@ -17,6 +17,7 @@
 #include <rtconfig.h>
 #include <asm-generic.h>
 #include <asm-fpu.h>
+#include <vector_gcc.h>
 #include <armv8.h>
 #include <lwp_arch.h>
 
@@ -105,6 +106,7 @@ arch_get_user_sp:
 .global arch_clone_exit
 arch_fork_exit:
 arch_clone_exit:
+    mov x0, xzr
     b  arch_syscall_exit
 
 /*
@@ -131,12 +133,31 @@ START_POINT(SVC_Handler)
     /* x0 is initial sp */
     mov sp, x0
 
+    bl _SVC_Handler
+
+    /* jump explictly, make this code position independant */
+    b arch_syscall_exit
+START_POINT_END(SVC_Handler)
+
+TRACE_SYMBOL(_SVC_Handler)
+#define FRAME_REG x19
+
+/**
+ * x0 -> frame_addr
+ */
+_SVC_Handler:
+    .local _SVC_Handler
+
+    stp fp, lr, [sp, -16]!
+    mov fp, sp
+
+    mov FRAME_REG, x0   /* save the value of frame address */
     msr daifclr, #3  /* enable interrupt */
 
     GET_THREAD_SELF x0
     bl  lwp_user_setting_save
 
-    ldp x8, x9, [sp, #(CONTEXT_OFFSET_X8)]
+    ldp x8, x9, [FRAME_REG, #(CONTEXT_OFFSET_X8)]
     and x0, x8, #0xf000
     cmp x0, #0xe000
     beq arch_signal_quit
@@ -149,78 +170,46 @@ START_POINT(SVC_Handler)
     cmp x0, xzr
     mov x30, x0
     beq arch_syscall_exit
-    ldp x0, x1, [sp, #(CONTEXT_OFFSET_X0)]
-    ldp x2, x3, [sp, #(CONTEXT_OFFSET_X2)]
-    ldp x4, x5, [sp, #(CONTEXT_OFFSET_X4)]
-    ldp x6, x7, [sp, #(CONTEXT_OFFSET_X6)]
+    ldp x0, x1, [FRAME_REG, #(CONTEXT_OFFSET_X0)]
+    ldp x2, x3, [FRAME_REG, #(CONTEXT_OFFSET_X2)]
+    ldp x4, x5, [FRAME_REG, #(CONTEXT_OFFSET_X4)]
+    ldp x6, x7, [FRAME_REG, #(CONTEXT_OFFSET_X6)]
     blr x30
-    /* jump explictly, make this code position independant */
-    b arch_syscall_exit
-START_POINT_END(SVC_Handler)
 
-.global arch_syscall_exit
+    ldp fp, lr, [sp], 16
+    ret
+
+/**
+ * void arch_syscall_exit(long rc)
+ */
 arch_syscall_exit:
+    .global arch_syscall_exit
 
     /**
-     * @brief back up former x0 which is required to restart syscall, then setup
+     * backup former x0 which is required to restart syscall, then setup
      * syscall return value in stack frame
      */
     mov x1, sp
     bl arch_syscall_prepare_signal
 
+    /**
+     * disable local irq so we don't messup with the spsr_el1 witch is not saved
+     * for kernel space IRQ/EXCEPTION
+     */
     msr daifset, #3
 
-    ldp x2, x3, [sp], #0x10  /* SPSR and ELR. */
-    msr spsr_el1, x3
-    msr elr_el1, x2
-
-    ldp x29, x30, [sp], #0x10
-    msr sp_el0, x29
-    ldp x28, x29, [sp], #0x10
-    msr fpcr, x28
-    msr fpsr, x29
-    ldp x28, x29, [sp], #0x10
-    ldp x26, x27, [sp], #0x10
-    ldp x24, x25, [sp], #0x10
-    ldp x22, x23, [sp], #0x10
-    ldp x20, x21, [sp], #0x10
-    ldp x18, x19, [sp], #0x10
-    ldp x16, x17, [sp], #0x10
-    ldp x14, x15, [sp], #0x10
-    ldp x12, x13, [sp], #0x10
-    ldp x10, x11, [sp], #0x10
-    ldp x8, x9, [sp], #0x10
-    ldp x6, x7, [sp], #0x10
-    ldp x4, x5, [sp], #0x10
-    ldp x2, x3, [sp], #0x10
-    ldp x0, x1, [sp], #0x10
-    RESTORE_FPU sp
+    b arch_ret_to_user
 
 /* the sp is reset to the outer most level, irq and fiq are disabled */
 START_POINT(arch_ret_to_user)
     msr daifset, #3
-    /* save exception frame */
-    SAVE_FPU sp
-    stp x0, x1, [sp, #-0x10]!
-    stp x2, x3, [sp, #-0x10]!
-    stp x4, x5, [sp, #-0x10]!
-    stp x6, x7, [sp, #-0x10]!
-    stp x8, x9, [sp, #-0x10]!
-    stp x10, x11, [sp, #-0x10]!
-    stp x12, x13, [sp, #-0x10]!
-    stp x14, x15, [sp, #-0x10]!
-    stp x16, x17, [sp, #-0x10]!
-    stp x18, x19, [sp, #-0x10]!
-    stp x20, x21, [sp, #-0x10]!
-    stp x22, x23, [sp, #-0x10]!
-    stp x24, x25, [sp, #-0x10]!
-    stp x26, x27, [sp, #-0x10]!
-    stp x28, x29, [sp, #-0x10]!
 
-    mrs x0, fpcr
-    mrs x1, fpsr
-    stp x0, x1, [sp, #-0x10]!
-    stp x29, x30, [sp, #-0x10]!
+    ldr x2, [sp, #CONTEXT_OFFSET_SP_EL0]
+    msr sp_el0, x2
+    ldr x2, [sp, #CONTEXT_OFFSET_ELR_EL1]
+    msr elr_el1, x2
+    ldr x3, [sp, #CONTEXT_OFFSET_SPSR_EL1]
+    msr spsr_el1, x3
 
     /* pre-action */
     bl lwp_check_debug
@@ -231,7 +220,8 @@ START_POINT(arch_ret_to_user)
     msr daifclr, #3
     mov x0, xzr
     b sys_exit
-1:
+
+1:  /* handling dbg */
     /* check if dbg ops exist */
     ldr x0, =rt_dbg_ops
     ldr x0, [x0]
@@ -243,104 +233,42 @@ START_POINT(arch_ret_to_user)
     orr x2, x2, x1
     msr spsr_el1, x2
     b 3f
-2:
+2:  /* clear software step */
     bic x2, x2, x1
     msr spsr_el1, x2
-3:
+3:  /* handling signal */
 
     /**
-     * push 2 dummy words to simulate a exception frame of interrupt
-     * Note: in kernel state, the context switch dont saved the context
+     * push updated spsr & elr to exception frame.
+     * Note: these 2 maybe updated after handling dbg
      */
     mrs x0, spsr_el1
+    str x0, [sp, #CONTEXT_OFFSET_SPSR_EL1]
     mrs x1, elr_el1
-    stp x1, x0, [sp, #-0x10]!
+    str x1, [sp, #CONTEXT_OFFSET_ELR_EL1]
     mov x0, sp
+
+    /* restore the thread execution environment */
     msr daifclr, #3
     bl lwp_thread_signal_catch
+
+    /* restore the exception-return exec-flow */
     msr daifset, #3
-    ldp x1, x0, [sp], #0x10
-    msr spsr_el1, x0
-    msr elr_el1, x1
 
     /* check debug */
-    /* restore exception frame */
-    ldp x29, x30, [sp], #0x10
-    ldp x0, x1, [sp], #0x10
-    msr fpcr, x0
-    msr fpsr, x1
 
-    ldp x28, x29, [sp], #0x10
-    ldp x26, x27, [sp], #0x10
-    ldp x24, x25, [sp], #0x10
-    ldp x22, x23, [sp], #0x10
-    ldp x20, x21, [sp], #0x10
-    ldp x18, x19, [sp], #0x10
-    ldp x16, x17, [sp], #0x10
-    ldp x14, x15, [sp], #0x10
-    ldp x12, x13, [sp], #0x10
-    ldp x10, x11, [sp], #0x10
-    ldp x8, x9, [sp], #0x10
-    ldp x6, x7, [sp], #0x10
-    ldp x4, x5, [sp], #0x10
-    ldp x2, x3, [sp], #0x10
-    ldp x0, x1, [sp], #0x10
-    RESTORE_FPU sp
-
-    stp x0, x1, [sp, #-0x10]!
     ldr x0, =rt_dbg_ops
     ldr x0, [x0]
     cmp x0, xzr
-    ldp x0, x1, [sp], #0x10
+
     beq 1f
 
-    /* save */
-    SAVE_FPU sp
-    stp x0, x1, [sp, #-0x10]!
-    stp x2, x3, [sp, #-0x10]!
-    stp x4, x5, [sp, #-0x10]!
-    stp x6, x7, [sp, #-0x10]!
-    stp x8, x9, [sp, #-0x10]!
-    stp x10, x11, [sp, #-0x10]!
-    stp x12, x13, [sp, #-0x10]!
-    stp x14, x15, [sp, #-0x10]!
-    stp x16, x17, [sp, #-0x10]!
-    stp x18, x19, [sp, #-0x10]!
-    stp x20, x21, [sp, #-0x10]!
-    stp x22, x23, [sp, #-0x10]!
-    stp x24, x25, [sp, #-0x10]!
-    stp x26, x27, [sp, #-0x10]!
-    stp x28, x29, [sp, #-0x10]!
-    mrs x0, fpcr
-    mrs x1, fpsr
-    stp x0, x1, [sp, #-0x10]!
-    stp x29, x30, [sp, #-0x10]!
-
-    mrs x0, elr_el1
+    ldr x0, [sp, #CONTEXT_OFFSET_ELR_EL1]
     bl dbg_attach_req
 
-    /* restore */
-    ldp x29, x30, [sp], #0x10
-    ldp x0, x1, [sp], #0x10
-    msr fpcr, x0
-    msr fpsr, x1
-    ldp x28, x29, [sp], #0x10
-    ldp x26, x27, [sp], #0x10
-    ldp x24, x25, [sp], #0x10
-    ldp x22, x23, [sp], #0x10
-    ldp x20, x21, [sp], #0x10
-    ldp x18, x19, [sp], #0x10
-    ldp x16, x17, [sp], #0x10
-    ldp x14, x15, [sp], #0x10
-    ldp x12, x13, [sp], #0x10
-    ldp x10, x11, [sp], #0x10
-    ldp x8, x9, [sp], #0x10
-    ldp x6, x7, [sp], #0x10
-    ldp x4, x5, [sp], #0x10
-    ldp x2, x3, [sp], #0x10
-    ldp x0, x1, [sp], #0x10
-    RESTORE_FPU sp
 1:
+    RESTORE_IRQ_CONTEXT_NO_SPEL0
+
     eret
 START_POINT_END(arch_ret_to_user)
 
@@ -410,32 +338,7 @@ arch_syscall_restart:
     /* restore previous exception frame */
     msr spsel, #0
 
-    ldp x2, x3, [sp], #0x10
-    msr elr_el1, x2
-    msr spsr_el1, x3
-
-    ldp x29, x30, [sp], #0x10
-
-    ldp x28, x29, [sp], #0x10
-    msr fpcr, x28
-    msr fpsr, x29
-
-    ldp x28, x29, [sp], #0x10
-    ldp x26, x27, [sp], #0x10
-    ldp x24, x25, [sp], #0x10
-    ldp x22, x23, [sp], #0x10
-    ldp x20, x21, [sp], #0x10
-    ldp x18, x19, [sp], #0x10
-    ldp x16, x17, [sp], #0x10
-    ldp x14, x15, [sp], #0x10
-    ldp x12, x13, [sp], #0x10
-    ldp x10, x11, [sp], #0x10
-    ldp x8, x9, [sp], #0x10
-    ldp x6, x7, [sp], #0x10
-    ldp x4, x5, [sp], #0x10
-    ldp x2, x3, [sp], #0x10
-    ldp x0, x1, [sp], #0x10
-    RESTORE_FPU sp
+    RESTORE_IRQ_CONTEXT_NO_SPEL0
 
     msr spsel, #1
 
@@ -443,8 +346,8 @@ arch_syscall_restart:
 
 arch_signal_quit:
 
-    /* drop current exception frame */
-    add sp, sp, #CONTEXT_SIZE
+    /* drop current exception frame & sigreturn */
+    add sp, sp, #(CONTEXT_SIZE + 0x10)
     mov x1, sp
     mrs x0, sp_el0
     bl  arch_signal_ucontext_restore
@@ -460,35 +363,12 @@ arch_signal_quit:
     /* restore previous exception frame */
     msr spsel, #0
 
-    ldp x2, x3, [sp], #0x10
-    msr elr_el1, x2
-    msr spsr_el1, x3
-
-    ldp x29, x30, [sp], #0x10
-
-    ldp x28, x29, [sp], #0x10
-    msr fpcr, x28
-    msr fpsr, x29
-
-    ldp x28, x29, [sp], #0x10
-    ldp x26, x27, [sp], #0x10
-    ldp x24, x25, [sp], #0x10
-    ldp x22, x23, [sp], #0x10
-    ldp x20, x21, [sp], #0x10
-    ldp x18, x19, [sp], #0x10
-    ldp x16, x17, [sp], #0x10
-    ldp x14, x15, [sp], #0x10
-    ldp x12, x13, [sp], #0x10
-    ldp x10, x11, [sp], #0x10
-    ldp x8, x9, [sp], #0x10
-    ldp x6, x7, [sp], #0x10
-    ldp x4, x5, [sp], #0x10
-    ldp x2, x3, [sp], #0x10
-    ldp x0, x1, [sp], #0x10
-    RESTORE_FPU sp
+    RESTORE_IRQ_CONTEXT_NO_SPEL0
 
     msr spsel, #1
 
+    SAVE_IRQ_CONTEXT
+
     b arch_ret_to_user
 
 /**
diff --git a/libcpu/aarch64/common/include/vector_gcc.h b/libcpu/aarch64/common/include/vector_gcc.h
index 31c532316e..63df0c59e3 100644
--- a/libcpu/aarch64/common/include/vector_gcc.h
+++ b/libcpu/aarch64/common/include/vector_gcc.h
@@ -54,6 +54,52 @@
 #include "../up/context_gcc.h"
 #endif
 
+.macro RESTORE_IRQ_CONTEXT_NO_SPEL0
+    ldp     x2, x3, [sp], #0x10
+    msr     elr_el1, x2
+    msr     spsr_el1, x3
+
+    ldp     x29, x30, [sp], #0x10
+
+    ldp     x28, x29, [sp], #0x10
+    msr     fpcr, x28
+    msr     fpsr, x29
+
+    ldp     x28, x29, [sp], #0x10
+    ldp     x26, x27, [sp], #0x10
+    ldp     x24, x25, [sp], #0x10
+    ldp     x22, x23, [sp], #0x10
+    ldp     x20, x21, [sp], #0x10
+    ldp     x18, x19, [sp], #0x10
+    ldp     x16, x17, [sp], #0x10
+    ldp     x14, x15, [sp], #0x10
+    ldp     x12, x13, [sp], #0x10
+    ldp     x10, x11, [sp], #0x10
+    ldp     x8, x9, [sp], #0x10
+    ldp     x6, x7, [sp], #0x10
+    ldp     x4, x5, [sp], #0x10
+    ldp     x2, x3, [sp], #0x10
+    ldp     x0, x1, [sp], #0x10
+
+    RESTORE_FPU sp
+.endm
+
+.macro EXCEPTION_SWITCH, eframex, tmpx
+#ifdef RT_USING_SMART
+    /**
+     * test the spsr for execution level 0
+     * That is { PSTATE.[NZCV] := SPSR_EL1 & M.EL0t }
+     */
+    ldr     \tmpx, [\eframex, #CONTEXT_OFFSET_SPSR_EL1]
+    and     \tmpx, \tmpx, 0x1f
+    cbz     \tmpx, 1f
+    b       2f
+1:
+    b       arch_ret_to_user
+2:
+#endif /* RT_USING_SMART */
+.endm
+
 .macro SAVE_USER_CTX, eframex, tmpx
 #ifdef RT_USING_SMART
     mrs     \tmpx, spsr_el1
diff --git a/libcpu/aarch64/common/up/context_gcc.h b/libcpu/aarch64/common/up/context_gcc.h
index cb48e5f244..d2f6e9a755 100644
--- a/libcpu/aarch64/common/up/context_gcc.h
+++ b/libcpu/aarch64/common/up/context_gcc.h
@@ -19,17 +19,23 @@
 #include <asm-fpu.h>
 #include <armv8.h>
 
-.macro RESTORE_CONTEXT_SWITCH using_sp
-    /* Set the SP to point to the stack of the task being restored. */
-    mov     sp, \using_sp
-
+/* restore address space */
+.macro RESTORE_ADDRESS_SPACE
 #ifdef RT_USING_SMART
     bl      rt_thread_self
     mov     x19, x0
     bl      lwp_aspace_switch
     mov     x0, x19
     bl      lwp_user_setting_restore
-#endif /* RT_USING_SMART */
+#endif
+.endm
+
+.macro RESTORE_CONTEXT_SWITCH using_sp
+    /* Set the SP to point to the stack of the task being restored. */
+    mov     sp, \using_sp
+
+    RESTORE_ADDRESS_SPACE
+
     _RESTORE_CONTEXT_SWITCH
 .endm
 
diff --git a/libcpu/aarch64/common/up/vector_gcc.S b/libcpu/aarch64/common/up/vector_gcc.S
index 26428fa597..8b51a48282 100644
--- a/libcpu/aarch64/common/up/vector_gcc.S
+++ b/libcpu/aarch64/common/up/vector_gcc.S
@@ -28,7 +28,8 @@ vector_fiq:
 
     SAVE_IRQ_CONTEXT
     bl      rt_hw_trap_fiq
-    RESTORE_IRQ_CONTEXT
+
+    b       rt_hw_irq_exit
 
 .globl      rt_thread_switch_interrupt_flag
 .globl      rt_hw_context_switch_interrupt_do
diff --git a/libcpu/aarch64/common/vector_gcc.S b/libcpu/aarch64/common/vector_gcc.S
index aaa172fcf9..d4c6223bcf 100644
--- a/libcpu/aarch64/common/vector_gcc.S
+++ b/libcpu/aarch64/common/vector_gcc.S
@@ -79,7 +79,11 @@ START_POINT(vector_exception)
     bl      rt_hw_trap_exception
     RESTORE_USER_CTX EFRAMEX, x0
 
+    /* do exception switch for IRQ/exception handlers */
+    EXCEPTION_SWITCH sp, x0
+
     RESTORE_IRQ_CONTEXT
+    eret
 START_POINT_END(vector_exception)
 
 START_POINT(vector_serror)
@@ -123,4 +127,8 @@ START_POINT_END(vector_irq)
 rt_hw_irq_exit:
     .globl rt_hw_irq_exit
 
+    /* do exception switch for IRQ/exception handlers */
+    EXCEPTION_SWITCH sp, x0
+
     RESTORE_IRQ_CONTEXT
+    eret