Sync from master (75a3f3dd3a930eab1dc58fa0258a9d1605a0ac11)

23 files changed, 589 insertions, 340 deletions

diff --git a/libcpu/arm/am335x/start_gcc.S b/libcpu/arm/am335x/start_gcc.S
index 6b76fb04a4..7e996e2a80 100644
--- a/libcpu/arm/am335x/start_gcc.S
+++ b/libcpu/arm/am335x/start_gcc.S
@@ -33,6 +33,7 @@
 .globl stack_start
 .globl stack_top
 
+.align 3
 stack_start:
 .rept ISR_Stack_Size
 .long 0
diff --git a/libcpu/arm/arm926/context_gcc.S b/libcpu/arm/arm926/context_gcc.S
index b6b7863679..f69aa9f5ba 100644
--- a/libcpu/arm/arm926/context_gcc.S
+++ b/libcpu/arm/arm926/context_gcc.S
@@ -10,6 +10,7 @@
 
 #define NOINT   0xC0
 
+.text
 ;/*
 ; * rt_base_t rt_hw_interrupt_disable();
 ; */
diff --git a/libcpu/arm/arm926/cpuport.c b/libcpu/arm/arm926/cpuport.c
index 1efb23e896..035e97d543 100644
--- a/libcpu/arm/arm926/cpuport.c
+++ b/libcpu/arm/arm926/cpuport.c
@@ -23,30 +23,30 @@ rt_inline rt_uint32_t cp15_rd(void)
 {
     rt_uint32_t i;
 
-    __asm volatile("mrc p15, 0, %0, c1, c0, 0":"=r" (i));
+    __asm volatile("mrc p15, 0, %0, c1, c0, 0":"=r"(i));
     return i;
 }
 
 rt_inline void cache_enable(rt_uint32_t bit)
 {
     __asm volatile(\
-                         "mrc  p15,0,r0,c1,c0,0\n\t"    \
-                         "orr  r0,r0,%0\n\t"            \
-                         "mcr  p15,0,r0,c1,c0,0"        \
-                         :                              \
-                         :"r" (bit)                     \
-                         :"memory");
+                   "mrc  p15,0,r0,c1,c0,0\n\t"    \
+                   "orr  r0,r0,%0\n\t"            \
+                   "mcr  p15,0,r0,c1,c0,0"        \
+                   :                              \
+                   : "r"(bit)                     \
+                   : "memory");
 }
 
 rt_inline void cache_disable(rt_uint32_t bit)
 {
     __asm volatile(\
-                         "mrc  p15,0,r0,c1,c0,0\n\t"    \
-                         "bic  r0,r0,%0\n\t"            \
-                         "mcr  p15,0,r0,c1,c0,0"        \
-                         :                              \
-                         :"r" (bit)                     \
-                         :"memory");
+                   "mrc  p15,0,r0,c1,c0,0\n\t"    \
+                   "bic  r0,r0,%0\n\t"            \
+                   "mcr  p15,0,r0,c1,c0,0"        \
+                   :                              \
+                   : "r"(bit)                     \
+                   : "memory");
 }
 #endif
 
@@ -152,7 +152,7 @@ void rt_hw_cpu_reset()
     rt_kprintf("Restarting system...\n");
     machine_reset();
 
-    while(1);    /* loop forever and wait for reset to happen */
+    while (1);   /* loop forever and wait for reset to happen */
 
     /* NEVER REACHED */
 }
@@ -206,21 +206,7 @@ int __rt_ffs(int value)
 #elif defined(__GNUC__) || defined(__ICCARM__)
 int __rt_ffs(int value)
 {
-    register rt_uint32_t x;
-
-    if (value == 0)
-        return value;
-
-    __asm
-    (
-        "rsb %[temp], %[val], #0\n"
-        "and %[temp], %[temp], %[val]\n"
-        "clz %[temp], %[temp]\n"
-        "rsb %[temp], %[temp], #32\n"
-        :[temp] "=r"(x)
-        :[val] "r"(value)
-    );
-    return x;
+    return __builtin_ffs(value);
 }
 #endif
 
diff --git a/libcpu/arm/arm926/machine.c b/libcpu/arm/arm926/machine.c
new file mode 100644
index 0000000000..aa26d10ec6
--- /dev/null
+++ b/libcpu/arm/arm926/machine.c
@@ -0,0 +1,27 @@
+/*
+ * Copyright (c) 2006-2018, RT-Thread Development Team
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Change Logs:
+ * Date           Author       Notes
+ * 2018-02-08     RT-Thread    the first version
+ */
+
+#include <rthw.h>
+#include <rtthread.h>
+
+RT_WEAK void machine_reset(void)
+{
+    rt_kprintf("reboot system...\n");
+    rt_hw_interrupt_disable();
+    while (1);
+}
+
+RT_WEAK void machine_shutdown(void)
+{
+    rt_kprintf("shutdown...\n");
+    rt_hw_interrupt_disable();
+    while (1);
+}
+
diff --git a/libcpu/arm/arm926/mmu.c b/libcpu/arm/arm926/mmu.c
index f1e475189b..19d70b9b0b 100644
--- a/libcpu/arm/arm926/mmu.c
+++ b/libcpu/arm/arm926/mmu.c
@@ -140,7 +140,7 @@ void mmu_clean_invalidated_dcache(rt_uint32_t buffer, rt_uint32_t size)
 
     ptr = buffer & ~(CACHE_LINE_SIZE - 1);
 
-    while(ptr < buffer + size)
+    while (ptr < buffer + size)
     {
         __asm volatile { MCR p15, 0, ptr, c7, c14, 1 }
         ptr += CACHE_LINE_SIZE;
@@ -211,18 +211,18 @@ void mmu_setttbase(register rt_uint32_t i)
      * set by page table entry
      */
     value = 0;
-    asm volatile ("mcr p15, 0, %0, c8, c7, 0"::"r"(value));
+    asm volatile("mcr p15, 0, %0, c8, c7, 0"::"r"(value));
 
     value = 0x55555555;
-    asm volatile ("mcr p15, 0, %0, c3, c0, 0"::"r"(value));
+    asm volatile("mcr p15, 0, %0, c3, c0, 0"::"r"(value));
 
-    asm volatile ("mcr p15, 0, %0, c2, c0, 0"::"r"(i));
+    asm volatile("mcr p15, 0, %0, c2, c0, 0"::"r"(i));
 
 }
 
 void mmu_set_domain(register rt_uint32_t i)
 {
-    asm volatile ("mcr p15,0, %0, c3, c0,  0": :"r" (i));
+    asm volatile("mcr p15,0, %0, c3, c0,  0": :"r"(i));
 }
 
 void mmu_enable()
@@ -321,7 +321,7 @@ void mmu_disable_alignfault()
 
 void mmu_clean_invalidated_cache_index(int index)
 {
-    asm volatile ("mcr p15, 0, %0, c7, c14, 2": :"r" (index));
+    asm volatile("mcr p15, 0, %0, c7, c14, 2": :"r"(index));
 }
 
 void mmu_clean_invalidated_dcache(rt_uint32_t buffer, rt_uint32_t size)
@@ -330,9 +330,9 @@ void mmu_clean_invalidated_dcache(rt_uint32_t buffer, rt_uint32_t size)
 
     ptr = buffer & ~(CACHE_LINE_SIZE - 1);
 
-    while(ptr < buffer + size)
+    while (ptr < buffer + size)
     {
-        asm volatile ("mcr p15, 0, %0, c7, c14, 1": :"r" (ptr));
+        asm volatile("mcr p15, 0, %0, c7, c14, 1": :"r"(ptr));
 
         ptr += CACHE_LINE_SIZE;
     }
@@ -347,7 +347,7 @@ void mmu_clean_dcache(rt_uint32_t buffer, rt_uint32_t size)
 
     while (ptr < buffer + size)
     {
-        asm volatile ("mcr p15, 0, %0, c7, c10, 1": :"r" (ptr));
+        asm volatile("mcr p15, 0, %0, c7, c10, 1": :"r"(ptr));
 
         ptr += CACHE_LINE_SIZE;
     }
@@ -361,7 +361,7 @@ void mmu_invalidate_dcache(rt_uint32_t buffer, rt_uint32_t size)
 
     while (ptr < buffer + size)
     {
-        asm volatile ("mcr p15, 0, %0, c7, c6, 1": :"r" (ptr));
+        asm volatile("mcr p15, 0, %0, c7, c6, 1": :"r"(ptr));
 
         ptr += CACHE_LINE_SIZE;
     }
@@ -369,19 +369,19 @@ void mmu_invalidate_dcache(rt_uint32_t buffer, rt_uint32_t size)
 
 void mmu_invalidate_tlb()
 {
-    asm volatile ("mcr p15, 0, %0, c8, c7, 0": :"r" (0));
+    asm volatile("mcr p15, 0, %0, c8, c7, 0": :"r"(0));
 
 }
 
 void mmu_invalidate_icache()
 {
-    asm volatile ("mcr p15, 0, %0, c7, c5, 0": :"r" (0));
+    asm volatile("mcr p15, 0, %0, c7, c5, 0": :"r"(0));
 
 }
 
 void mmu_invalidate_dcache_all()
 {
-    asm volatile ("mcr p15, 0, %0, c7, c6, 0": :"r" (0));
+    asm volatile("mcr p15, 0, %0, c7, c6, 0": :"r"(0));
 
 }
 #endif
@@ -389,10 +389,10 @@ void mmu_invalidate_dcache_all()
 /* level1 page table */
 #if defined(__ICCARM__)
 #pragma data_alignment=(16*1024)
-static volatile rt_uint32_t _page_table[4*1024];
+static volatile rt_uint32_t _page_table[4 * 1024];
 #else
-static volatile rt_uint32_t _page_table[4*1024] \
-    __attribute__((aligned(16*1024)));
+static volatile rt_uint32_t _page_table[4 * 1024] \
+__attribute__((aligned(16 * 1024)));
 #endif
 
 void mmu_setmtt(rt_uint32_t vaddrStart, rt_uint32_t vaddrEnd,
@@ -401,11 +401,11 @@ void mmu_setmtt(rt_uint32_t vaddrStart, rt_uint32_t vaddrEnd,
     volatile rt_uint32_t *pTT;
     volatile int nSec;
     int i = 0;
-    pTT=(rt_uint32_t *)_page_table+(vaddrStart>>20);
-    nSec=(vaddrEnd>>20)-(vaddrStart>>20);
-    for(i=0; i<=nSec; i++)
+    pTT = (rt_uint32_t *)_page_table + (vaddrStart >> 20);
+    nSec = (vaddrEnd >> 20) - (vaddrStart >> 20);
+    for (i = 0; i <= nSec; i++)
     {
-        *pTT = attr |(((paddrStart>>20)+i)<<20);
+        *pTT = attr | (((paddrStart >> 20) + i) << 20);
         pTT++;
     }
 }
diff --git a/libcpu/arm/arm926/mmu.h b/libcpu/arm/arm926/mmu.h
index 7b930f5358..36fdbeacd8 100644
--- a/libcpu/arm/arm926/mmu.h
+++ b/libcpu/arm/arm926/mmu.h
@@ -5,6 +5,7 @@
  *
  * Change Logs:
  * Date           Author       Notes
+ * 2018-02-08     RT-Thread    the first version
  */
 
 #ifndef __MMU_H__
@@ -45,5 +46,7 @@ struct mem_desc
 };
 
 void rt_hw_mmu_init(struct mem_desc *mdesc, rt_uint32_t size);
-
+void mmu_clean_invalidated_dcache(rt_uint32_t buffer, rt_uint32_t size);
+void mmu_clean_dcache(rt_uint32_t buffer, rt_uint32_t size);
+void mmu_invalidate_dcache(rt_uint32_t buffer, rt_uint32_t size);
 #endif
diff --git a/libcpu/arm/arm926/stack.c b/libcpu/arm/arm926/stack.c
index fa98feb824..f90bfa8c22 100644
--- a/libcpu/arm/arm926/stack.c
+++ b/libcpu/arm/arm926/stack.c
@@ -38,7 +38,7 @@ rt_uint8_t *rt_hw_stack_init(void *tentry, void *parameter,
 
     stack_addr += sizeof(rt_uint32_t);
     stack_addr  = (rt_uint8_t *)RT_ALIGN_DOWN((rt_uint32_t)stack_addr, 8);
-    stk      = (rt_uint32_t *)stack_addr;
+    stk  = (rt_uint32_t *)stack_addr;
 
     *(--stk) = (rt_uint32_t)tentry;         /* entry point */
     *(--stk) = (rt_uint32_t)texit;          /* lr */
diff --git a/libcpu/arm/arm926/start_gcc.S b/libcpu/arm/arm926/start_gcc.S
index fee8e018cb..35c1f1722e 100644
--- a/libcpu/arm/arm926/start_gcc.S
+++ b/libcpu/arm/arm926/start_gcc.S
@@ -11,295 +11,320 @@
  * 2015-06-04     aozima     Align stack address to 8 byte.
  */
 
-#include "rt_low_level_init.h"
+.equ MODE_USR,        0x10
+.equ MODE_FIQ,        0x11
+.equ MODE_IRQ,        0x12
+.equ MODE_SVC,        0x13
+.equ MODE_ABT,        0x17
+.equ MODE_UND,        0x1B
+.equ MODE_SYS,        0x1F
+.equ MODEMASK,        0x1F
+.equ NOINT,           0xC0
 
-#define S_FRAME_SIZE    (18*4)   //72
+.equ I_BIT,           0x80
+.equ F_BIT,           0x40
 
-@#define S_SPSR          (17*4)   //SPSR
-@#define S_CPSR          (16*4)   //CPSR
-#define S_PC            (15*4)   //R15
-@#define S_LR            (14*4)   //R14
-@#define S_SP            (13*4)   //R13
+.equ UND_STACK_SIZE,  0x00000100
+.equ SVC_STACK_SIZE,  0x00000100
+.equ ABT_STACK_SIZE,  0x00000100
+.equ FIQ_STACK_SIZE,  0x00000100
+.equ IRQ_STACK_SIZE,  0x00000100
+.equ SYS_STACK_SIZE,  0x00000100
 
-@#define S_IP            (12*4)   //R12
-@#define S_FP            (11*4)   //R11
-@#define S_R10           (10*4)
-@#define S_R9            (9*4)
-@#define S_R8            (8*4)
-@#define S_R7            (7*4)
-@#define S_R6            (6*4)
-@#define S_R5            (5*4)
-@#define S_R4            (4*4)
-@#define S_R3            (3*4)
-@#define S_R2            (2*4)
-@#define S_R1            (1*4)
-@#define S_R0            (0*4)
+ /*
+ ***************************************
+ * Interrupt vector table
+ ***************************************
+ */
+.section .vectors
+.code 32
 
-#define MODE_SYS        0x1F
-#define MODE_FIQ        0x11
-#define MODE_IRQ        0x12
-#define MODE_SVC        0x13
-#define MODE_ABT        0x17
-#define MODE_UND        0x1B
-#define MODEMASK        0x1F
+.global system_vectors
+system_vectors:
+    ldr pc, _vector_reset
+    ldr pc, _vector_undef
+    ldr pc, _vector_swi
+    ldr pc, _vector_pabt
+    ldr pc, _vector_dabt
+    ldr pc, _vector_resv
+    ldr pc, _vector_irq
+    ldr pc, _vector_fiq
 
-#define NOINT           0xC0
+_vector_reset:
+    .word reset
+_vector_undef:
+    .word vector_undef
+_vector_swi:
+    .word vector_swi
+_vector_pabt:
+    .word vector_pabt
+_vector_dabt:
+    .word vector_dabt
+_vector_resv:
+    .word vector_resv
+_vector_irq:
+    .word vector_irq
+_vector_fiq:
+    .word vector_fiq
 
-@;----------------------- Stack and Heap Definitions ---------------------------
-    .section .nobss, "w"
+.balignl    16,0xdeadbeef
 
-    .space UND_STK_SIZE
+ /*
+ ***************************************
+ *  Stack and Heap Definitions 
+ ***************************************
+ */
+    .section .data
+    .space UND_STACK_SIZE
     .align 3
-    .global UND_STACK_START
-UND_STACK_START:
+    .global und_stack_start
+und_stack_start:
 
-    .space ABT_STK_SIZE
+    .space ABT_STACK_SIZE
     .align 3
-    .global ABT_STACK_START
-ABT_STACK_START:
+    .global abt_stack_start
+abt_stack_start:
 
-    .space FIQ_STK_SIZE
+    .space FIQ_STACK_SIZE
     .align 3
-    .global FIQ_STACK_START
-FIQ_STACK_START:
+    .global fiq_stack_start
+fiq_stack_start:
 
-    .space IRQ_STK_SIZE
+    .space IRQ_STACK_SIZE
     .align 3
-    .global IRQ_STACK_START
-IRQ_STACK_START:
+    .global irq_stack_start
+irq_stack_start:
 
-    .skip SYS_STK_SIZE
+    .skip SYS_STACK_SIZE
     .align 3
-    .global SYS_STACK_START
-SYS_STACK_START:
+    .global sys_stack_start
+sys_stack_start:
 
-    .space SVC_STK_SIZE
+    .space SVC_STACK_SIZE
     .align 3
-    .global SVC_STACK_START
-SVC_STACK_START:
-
-@;--------------Jump vector table-----------------------------------------------
-    .section .init, "ax"
-    .arm
+    .global svc_stack_start
+svc_stack_start:
 
-    .global start
-start:
-    LDR     PC, vector_reset
-    LDR     PC, vector_undef
-    LDR     PC, vector_swi
-    LDR     PC, vector_pabt
-    LDR     PC, vector_dabt
-    LDR     PC, vector_resv
-    LDR     PC, vector_irq
-    LDR     PC, vector_fiq
+/*
+ ***************************************
+ * Startup Code 
+ ***************************************
+ */
+    .section .text
+    .global reset
+reset:
+    /* Enter svc mode and mask interrupts */
+    mrs r0, cpsr
+    bic r0, r0, #MODEMASK
+    orr r0, r0, #MODE_SVC|NOINT
+    msr cpsr_cxsf, r0
 
-vector_reset:
-    .word  Reset_Handler
-vector_undef:
-    .word  Undef_Handler
-vector_swi:
-    .word  SWI_Handler
-vector_pabt:
-    .word  PAbt_Handler
-vector_dabt:
-    .word  DAbt_Handler
-vector_resv:
-    .word  Resv_Handler
-vector_irq:
-    .word  IRQ_Handler
-vector_fiq:
-    .word  FIQ_Handler
+    /* init cpu  */
+    bl  cpu_init_crit
+    
+    /* todo:copyself to link address */
+    
+    /* Copy vector to the correct address */
+    ldr r0, =system_vectors
+    mrc p15, 0, r2, c1, c0, 0
+    ands r2, r2, #(1 << 13)
+    ldreq r1, =0x00000000
+    ldrne r1, =0xffff0000
+    ldmia r0!, {r2-r8, r10}
+    stmia r1!, {r2-r8, r10}
+    ldmia r0!, {r2-r8, r10}
+    stmia r1!, {r2-r8, r10}
 
-    .balignl     16,0xdeadbeef
+    /* turn off the watchdog */
+    ldr r0, =0x01C20CB8
+    mov     r1, #0x0
+    str     r1, [r0]
 
-@;----------------- Reset Handler ---------------------------------------------
-    .global rt_low_level_init
-    .global main
-    .global Reset_Handler
-Reset_Handler:
-    @; Set the cpu to SVC32 mode
-    MRS     R0, CPSR
-    BIC     R0, R0, #MODEMASK
-    ORR     R0, R0, #MODE_SVC|NOINT
-    MSR     CPSR_cxsf, R0
+    /* mask all IRQs source */
+    ldr r1, =0xffffffff
+    ldr r0, =0x01C20430
+    str r1, [r0], #0x04
+    str r1, [r0]
+    
+    /* Call low level init function */
+    ldr     sp, =svc_stack_start
+    ldr     r0, =rt_low_level_init
+    blx     r0
     
-    @; Set CO-Processor
-    @; little-end锛宒isbale I/D Cache MMU, vector table is 0x00000000
-    MRC     P15, 0, R0, C1, C0, 0   @; Read CP15
-    LDR     R1, =0x00003085         @; set clear bits
-    BIC     R0, R0, R1
-    MCR     P15, 0, R0, C1, C0, 0   @; Write CP15
+    /* init stack */
+    bl stack_setup
+    
+    /* clear bss */
+    mov     r0, #0
+    ldr     r1, =__bss_start
+    ldr     r2, =__bss_end
+
+bss_clear_loop:
+    cmp     r1, r2
+    strlo   r0, [r1], #4
+    blo     bss_clear_loop
+       
+    /* call c++ constructors of global objects */
+    /*
+    ldr     r0, =__ctors_start__
+    ldr     r1, =__ctors_end__
+
+ctor_loop:
+    cmp     r0, r1
+    beq     ctor_end
+    ldr     r2, [r0], #4
+    stmfd   sp!, {r0-r1}
+    mov     lr, pc
+    bx      r2
+    ldmfd   sp!, {r0-r1}
+    b       ctor_loop
+ctor_end:
+    */
+    /* start RT-Thread Kernel */
+    ldr     pc, _rtthread_startup
+_rtthread_startup:
+    .word  rtthread_startup
 
-    @; Call low level init function,
-    @; disable and clear all IRQs, Init MMU, Init interrupt controller, etc.
-    LDR     SP, =SVC_STACK_START
-    LDR     R0, =rt_low_level_init
-    BLX     R0
 
-Setup_Stack:
-    @; Setup Stack for each mode
-    MRS     R0, CPSR
-    BIC     R0, R0, #MODEMASK
 
-    ORR     R1, R0, #MODE_UND|NOINT
-    MSR     CPSR_cxsf, R1            @; Undef mode
-    LDR     SP, =UND_STACK_START
+cpu_init_crit:
+    /* invalidate I/D caches */
+    mov r0, #0
+    mcr p15, 0, r0, c7, c7, 0
+    mcr p15, 0, r0, c8, c7, 0
 
-    ORR     R1, R0, #MODE_ABT|NOINT
-    MSR     CPSR_cxsf, R1            @; Abort mode
-    LDR     SP, =ABT_STACK_START
+    /* disable MMU stuff and caches */
+    mrc p15, 0, r0, c1, c0, 0
+    bic r0, r0, #0x00002300
+    bic r0, r0, #0x00000087
+    orr r0, r0, #0x00000002
+    orr r0, r0, #0x00001000
+    mcr p15, 0, r0, c1, c0, 0
 
-    ORR     R1, R0, #MODE_IRQ|NOINT
-    MSR     CPSR_cxsf, R1            @; IRQ mode
-    LDR     SP, =IRQ_STACK_START
+    bx lr
+ 
+stack_setup:
+    /* Setup Stack for each mode */
+    mrs     r0, cpsr
+    bic     r0, r0, #MODEMASK
 
-    ORR     R1, R0, #MODE_FIQ|NOINT
-    MSR     CPSR_cxsf, R1            @; FIQ mode
-    LDR     SP, =FIQ_STACK_START
+    orr     r1, r0, #MODE_UND|NOINT
+    msr     cpsr_cxsf, r1
+    ldr     sp, =und_stack_start
 
-    ORR     R1, R0, #MODE_SYS|NOINT
-    MSR     CPSR_cxsf,R1             @; SYS/User mode
-    LDR     SP, =SYS_STACK_START
+    orr     r1, r0, #MODE_ABT|NOINT
+    msr     cpsr_cxsf, r1
+    ldr     sp, =abt_stack_start
 
-    ORR     R1, R0, #MODE_SVC|NOINT
-    MSR     CPSR_cxsf, R1            @; SVC mode
-    LDR     SP, =SVC_STACK_START
+    orr     r1, r0, #MODE_IRQ|NOINT
+    msr     cpsr_cxsf, r1
+    ldr     sp, =irq_stack_start
 
-    @; clear .bss
-    MOV     R0, #0                   @; get a zero
-    LDR     R1, =__bss_start__       @; bss start
-    LDR     R2, =__bss_end__         @; bss end
+    orr     r1, r0, #MODE_FIQ|NOINT
+    msr     cpsr_cxsf, r1
+    ldr     sp, =fiq_stack_start
 
-bss_clear_loop:
-    CMP     R1, R2                   @; check if data to clear
-    STRLO   R0, [R1], #4             @; clear 4 bytes
-    BLO     bss_clear_loop           @; loop until done
+    orr     r1, r0, #MODE_SYS|NOINT
+    msr     cpsr_cxsf,r1
+    ldr     sp, =sys_stack_start
 
-    @; call C++ constructors of global objects
-    LDR     R0, =__ctors_start__
-    LDR     R1, =__ctors_end__
+    orr     r1, r0, #MODE_SVC|NOINT
+    msr     cpsr_cxsf, r1
+    ldr     sp, =svc_stack_start
 
-ctor_loop:
-    CMP     R0, R1
-    BEQ     ctor_end
-    LDR     R2, [R0], #4
-    STMFD   SP!, {R0-R1}
-    MOV     LR, PC
-    BX      R2
-    LDMFD   SP!, {R0-R1}
-    B       ctor_loop
-ctor_end:
+    bx      lr
+ 
+/*
+ ***************************************
+ * exception handlers 
+ ***************************************
+ */
+    /* Interrupt */
+vector_fiq:
+    stmfd   sp!,{r0-r7,lr}
+    bl      rt_hw_trap_fiq
+    ldmfd   sp!,{r0-r7,lr}
+    subs    pc, lr, #4
 
-    @; Enter the C code
-    LDR     R0, =rtthread_startup
-    BLX     R0
+vector_irq:
+    stmfd   sp!, {r0-r12,lr}
 
-@;----------------- Exception Handler -----------------------------------------
-    .global rt_hw_trap_udef
-    .global rt_hw_trap_swi
-    .global rt_hw_trap_pabt
-    .global rt_hw_trap_dabt
-    .global rt_hw_trap_resv
-    .global rt_hw_trap_irq
-    .global rt_hw_trap_fiq
+    bl      rt_interrupt_enter
+    bl      rt_hw_trap_irq
+    bl      rt_interrupt_leave
 
-    .global rt_interrupt_enter
-    .global rt_interrupt_leave
-    .global rt_thread_switch_interrupt_flag
-    .global rt_interrupt_from_thread
-    .global rt_interrupt_to_thread
-    
-    .align  5
-Undef_Handler:
-    SUB     SP, SP, #S_FRAME_SIZE
-    STMIA   SP, {R0 - R12}          @; Calling R0-R12
-    ADD     R8, SP, #S_PC
-    STMDB   R8, {SP, LR}            @; Calling SP, LR
-    STR     LR, [R8, #0]            @; Save calling PC
-    MRS     R6, SPSR
-    STR     R6, [R8, #4]            @; Save CPSR
-    STR     R0, [R8, #8]            @; Save SPSR
-    MOV     R0, SP
-    BL      rt_hw_trap_udef
-    
-    .align  5
-SWI_Handler:
-    BL      rt_hw_trap_swi
-    
-    .align  5
-PAbt_Handler:
-    BL      rt_hw_trap_pabt
+    ldr     r0, =rt_thread_switch_interrupt_flag
+    ldr     r1, [r0]
+    cmp     r1, #1
+    beq     rt_hw_context_switch_interrupt_do
 
-    .align  5
-DAbt_Handler:
-    SUB     SP, SP, #S_FRAME_SIZE
-    STMIA   SP, {R0 - R12}          @; Calling R0-R12
-    ADD     R8, SP, #S_PC
-    STMDB   R8, {SP, LR}            @; Calling SP, LR
-    STR     LR, [R8, #0]            @; Save calling PC
-    MRS     R6, SPSR
-    STR     R6, [R8, #4]            @; Save CPSR
-    STR     R0, [R8, #8]            @; Save SPSR
-    MOV     R0, SP
-    BL      rt_hw_trap_dabt
+    ldmfd   sp!, {r0-r12,lr}
+    subs    pc,  lr, #4
 
-    .align  5
-Resv_Handler:
-    BL      rt_hw_trap_resv
+rt_hw_context_switch_interrupt_do:
+    mov     r1,  #0         
+    str     r1,  [r0]
 
-    .align  5
-FIQ_Handler:
-    STMFD   SP!, {R0-R7,LR}
-    BL      rt_hw_trap_fiq
-    LDMFD   SP!, {R0-R7,LR}
-    SUBS    PC, LR, #4
+    mov     r1, sp          
+    add     sp, sp, #4*4
+    ldmfd   sp!, {r4-r12,lr}
+    mrs     r0,  spsr       
+    sub     r2,  lr, #4     
 
-    .align  5
-IRQ_Handler:
-    STMFD   SP!, {R0-R12,LR}
-    BL      rt_interrupt_enter
-    BL      rt_hw_trap_irq
-    BL      rt_interrupt_leave
+    msr     cpsr_c, #I_BIT|F_BIT|MODE_SVC
 
-    @; If rt_thread_switch_interrupt_flag set,
-    @; jump to rt_hw_context_switch_interrupt_do and don't return
-    LDR     R0, =rt_thread_switch_interrupt_flag
-    LDR     R1, [R0]
-    CMP     R1, #1
-    BEQ     rt_hw_context_switch_interrupt_do
+    stmfd   sp!, {r2}       
+    stmfd   sp!, {r4-r12,lr}
+    ldmfd   r1,  {r1-r4}    
+    stmfd   sp!, {r1-r4}    
+    stmfd   sp!, {r0}       
 
-    LDMFD   SP!, {R0-R12,LR}
-    SUBS    PC, LR, #4
+    ldr     r4,  =rt_interrupt_from_thread
+    ldr     r5,  [r4]
+    str     sp,  [r5]       
 
-@;------ void rt_hw_context_switch_interrupt_do(rt_base_t flag) -----------------
-rt_hw_context_switch_interrupt_do:
-    MOV     R1,  #0                 @; Clear flag
-    STR     R1,  [R0]               @; Save to flag variable
+    ldr     r6,  =rt_interrupt_to_thread
+    ldr     r6,  [r6]
+    ldr     sp,  [r6]       
 
-    LDMFD   SP!, {R0-R12,LR}        @; Reload saved registers
-    STMFD   SP, {R0-R2}             @; Save R0-R2
-    SUB     R1,  SP, #4*3           @; Save old task's SP to R1
-    SUB     R2,  LR, #4             @; Save old task's PC to R2
+    ldmfd   sp!, {r4}       
+    msr     spsr_cxsf, r4
 
-    MRS     R0,  SPSR               @; Get CPSR of interrupt thread
+    ldmfd   sp!, {r0-r12,lr,pc}^ 
 
-    MSR     CPSR_c, #MODE_SVC|NOINT @; Switch to SVC mode and no interrupt
+    /* Exception */
+.macro push_svc_reg
+    sub     sp, sp, #17 * 4
+    stmia   sp, {r0 - r12} 
+    mov     r0, sp
+    mrs     r6, spsr       
+    str     lr, [r0, #15*4]
+    str     r6, [r0, #16*4]
+    str     sp, [r0, #13*4]
+    str     lr, [r0, #14*4]
+.endm
 
-    STMFD   SP!, {R2}               @; Push old task's PC
-    STMFD   SP!, {R3-R12,LR}        @; Push old task's LR,R12-R3
-    LDMFD   R1, {R1-R3}
-    STMFD   SP!, {R1-R3}            @; Push old task's R2-R0
-    STMFD   SP!, {R0}               @; Push old task's CPSR
+vector_swi:
+    push_svc_reg
+    bl      rt_hw_trap_swi
+    b       .
 
-    LDR     R4,  =rt_interrupt_from_thread
-    LDR     R5,  [R4]               @; R5 = stack ptr in old tasks's TCB
-    STR     SP,  [R5]               @; Store SP in preempted tasks's TCB
+vector_undef:
+    push_svc_reg
+    bl      rt_hw_trap_udef
+    b       .
 
-    LDR     R6,  =rt_interrupt_to_thread
-    LDR     R6,  [R6]               @; R6 = stack ptr in new tasks's TCB
-    LDR     SP,  [R6]               @; Get new task's stack pointer
+vector_pabt:
+    push_svc_reg
+    bl      rt_hw_trap_pabt
+    b       .
 
-    LDMFD   SP!, {R4}               @; Pop new task's SPSR
-    MSR     SPSR_cxsf, R4
+vector_dabt:
+    push_svc_reg
+    bl      rt_hw_trap_dabt
+    b       .
 
-    LDMFD   SP!, {R0-R12,LR,PC}^    @; pop new task's R0-R12,LR & PC SPSR 2 CPSR
+vector_resv:
+    push_svc_reg
+    bl      rt_hw_trap_resv
+    b       .
diff --git a/libcpu/arm/arm926/trap.c b/libcpu/arm/arm926/trap.c
index 054cc2741a..9e7ba48037 100644
--- a/libcpu/arm/arm926/trap.c
+++ b/libcpu/arm/arm926/trap.c
@@ -41,14 +41,18 @@ struct rt_hw_register
     rt_uint32_t cpsr;
     rt_uint32_t ORIG_r0;
 };
-
+static rt_err_t (*rt_exception_hook)(void *context) = RT_NULL;
+void rt_hw_exception_install(rt_err_t (*exception_handle)(void *context))
+{
+    rt_exception_hook = exception_handle;
+}
 /**
  * this function will show registers of CPU
  *
  * @param regs the registers point
  */
 
-void rt_hw_show_register (struct rt_hw_register *regs)
+void rt_hw_show_register(struct rt_hw_register *regs)
 {
     rt_kprintf("Execption:\n");
     rt_kprintf("r00:0x%08x r01:0x%08x r02:0x%08x r03:0x%08x\n",
@@ -74,6 +78,13 @@ void rt_hw_show_register (struct rt_hw_register *regs)
  */
 void rt_hw_trap_udef(struct rt_hw_register *regs)
 {
+    if (rt_exception_hook != RT_NULL)
+    {
+        rt_err_t result;
+
+        result = rt_exception_hook(regs);
+        if (result == RT_EOK) return;
+    }
     rt_hw_show_register(regs);
 
     rt_kprintf("undefined instruction\n");
@@ -96,6 +107,13 @@ void rt_hw_trap_udef(struct rt_hw_register *regs)
  */
 void rt_hw_trap_swi(struct rt_hw_register *regs)
 {
+    if (rt_exception_hook != RT_NULL)
+    {
+        rt_err_t result;
+
+        result = rt_exception_hook(regs);
+        if (result == RT_EOK) return;
+    }
     rt_hw_show_register(regs);
 
     rt_kprintf("software interrupt\n");
@@ -112,6 +130,13 @@ void rt_hw_trap_swi(struct rt_hw_register *regs)
  */
 void rt_hw_trap_pabt(struct rt_hw_register *regs)
 {
+    if (rt_exception_hook != RT_NULL)
+    {
+        rt_err_t result;
+
+        result = rt_exception_hook(regs);
+        if (result == RT_EOK) return;
+    }
     rt_hw_show_register(regs);
 
     rt_kprintf("prefetch abort\n");
@@ -133,6 +158,13 @@ void rt_hw_trap_pabt(struct rt_hw_register *regs)
  */
 void rt_hw_trap_dabt(struct rt_hw_register *regs)
 {
+    if (rt_exception_hook != RT_NULL)
+    {
+        rt_err_t result;
+
+        result = rt_exception_hook(regs);
+        if (result == RT_EOK) return;
+    }
     rt_hw_show_register(regs);
 
     rt_kprintf("data abort\n");
@@ -153,55 +185,26 @@ void rt_hw_trap_dabt(struct rt_hw_register *regs)
  */
 void rt_hw_trap_resv(struct rt_hw_register *regs)
 {
+    if (rt_exception_hook != RT_NULL)
+    {
+        rt_err_t result;
+
+        result = rt_exception_hook(regs);
+        if (result == RT_EOK) return;
+    }
     rt_kprintf("not used\n");
     rt_hw_show_register(regs);
     rt_hw_cpu_shutdown();
 }
 
-extern struct rt_irq_desc irq_desc[];
-extern rt_uint32_t rt_hw_interrupt_get_active(rt_uint32_t fiq_irq);
-extern void rt_hw_interrupt_ack(rt_uint32_t fiq_irq, rt_uint32_t id);
-    
-void rt_hw_trap_irq()
-{
-    rt_isr_handler_t isr_func;
-    rt_uint32_t irq;
-    void *param;
-
-    /* get irq number */
-    irq = rt_hw_interrupt_get_active(INT_IRQ);
+extern void rt_interrupt_dispatch(rt_uint32_t fiq_irq);
 
-    /* get interrupt service routine */
-    isr_func = irq_desc[irq].handler;
-    param = irq_desc[irq].param;
-
-    /* turn to interrupt service routine */
-    isr_func(irq, param);
-
-    rt_hw_interrupt_ack(INT_IRQ, irq);
-#ifdef RT_USING_INTERRUPT_INFO
-    irq_desc[irq].counter ++;
-#endif
+void rt_hw_trap_irq(void)
+{
+    rt_interrupt_dispatch(INT_IRQ);
 }
 
-void rt_hw_trap_fiq()
+void rt_hw_trap_fiq(void)
 {
-    rt_isr_handler_t isr_func;
-    rt_uint32_t irq;
-    void *param;
-
-    /* get irq number */
-    irq = rt_hw_interrupt_get_active(INT_FIQ);
-
-    /* get interrupt service routine */
-    isr_func = irq_desc[irq].handler;
-    param = irq_desc[irq].param;
-
-    /* turn to interrupt service routine */
-    isr_func(irq, param);
-
-    rt_hw_interrupt_ack(INT_FIQ, irq);
-#ifdef RT_USING_INTERRUPT_INFO
-    irq_desc[irq].counter ++;
-#endif
+    rt_interrupt_dispatch(INT_FIQ);
 }
diff --git a/libcpu/arm/cortex-a/cache.c b/libcpu/arm/cortex-a/cache.c
index 6d28d9e9f9..30af86baa1 100644
--- a/libcpu/arm/cortex-a/cache.c
+++ b/libcpu/arm/cortex-a/cache.c
@@ -30,6 +30,7 @@ void rt_hw_cpu_icache_invalidate(void *addr, int size)
     rt_uint32_t start_addr = (rt_uint32_t)addr;
     rt_uint32_t end_addr = (rt_uint32_t) addr + size + line_size - 1;
 
+    asm volatile ("dmb":::"memory");
     start_addr &= ~(line_size-1);
     end_addr &= ~(line_size-1);
     while (start_addr < end_addr)
@@ -37,6 +38,7 @@ void rt_hw_cpu_icache_invalidate(void *addr, int size)
         asm volatile ("mcr p15, 0, %0, c7, c5, 1" :: "r"(start_addr));  /* icimvau */
         start_addr += line_size;
     }
+    asm volatile ("dsb\n\tisb":::"memory");
 }
 
 void rt_hw_cpu_dcache_invalidate(void *addr, int size)
@@ -45,6 +47,7 @@ void rt_hw_cpu_dcache_invalidate(void *addr, int size)
     rt_uint32_t start_addr = (rt_uint32_t)addr;
     rt_uint32_t end_addr = (rt_uint32_t) addr + size + line_size - 1;
 
+    asm volatile ("dmb":::"memory");
     start_addr &= ~(line_size-1);
     end_addr &= ~(line_size-1);
     while (start_addr < end_addr)
@@ -52,6 +55,7 @@ void rt_hw_cpu_dcache_invalidate(void *addr, int size)
         asm volatile ("mcr p15, 0, %0, c7, c6, 1" :: "r"(start_addr));  /* dcimvac */
         start_addr += line_size;
     }
+    asm volatile ("dsb":::"memory");
 }
 
 void rt_hw_cpu_dcache_clean(void *addr, int size)
@@ -60,6 +64,7 @@ void rt_hw_cpu_dcache_clean(void *addr, int size)
     rt_uint32_t start_addr = (rt_uint32_t)addr;
     rt_uint32_t end_addr = (rt_uint32_t) addr + size + line_size - 1;
 
+    asm volatile ("dmb":::"memory");
     start_addr &= ~(line_size-1);
     end_addr &= ~(line_size-1);
     while (start_addr < end_addr)
@@ -67,6 +72,7 @@ void rt_hw_cpu_dcache_clean(void *addr, int size)
         asm volatile ("mcr p15, 0, %0, c7, c10, 1" :: "r"(start_addr)); /* dccmvac */
         start_addr += line_size;
     }
+    asm volatile ("dsb":::"memory");
 }
 
 void rt_hw_cpu_icache_ops(int ops, void *addr, int size)
diff --git a/libcpu/arm/cortex-a/context_gcc.S b/libcpu/arm/cortex-a/context_gcc.S
index dff7a1de68..768c5138e9 100644
--- a/libcpu/arm/cortex-a/context_gcc.S
+++ b/libcpu/arm/cortex-a/context_gcc.S
@@ -76,6 +76,18 @@ rt_hw_context_switch:
     stmfd sp, {r13, r14}^   @ push usr_sp usr_lr
     sub sp, #8
 #endif
+#ifdef RT_USING_FPU
+    /* fpu context */
+    vmrs r6, fpexc
+    tst  r6, #(1<<30)
+    beq 1f
+    vstmdb sp!, {d0-d15}
+    vstmdb sp!, {d16-d31}
+    vmrs r5, fpscr
+    stmfd sp!, {r5}
+1:
+    stmfd sp!, {r6}
+#endif
 
     str sp, [r0]            @ store sp in preempted tasks TCB
     ldr sp, [r1]            @ get new task stack pointer
@@ -147,6 +159,18 @@ rt_hw_context_switch_exit:
     mov     sp, r0
 #endif
 #endif
+#ifdef RT_USING_FPU
+/* fpu context */
+    ldmfd sp!, {r6}
+    vmsr fpexc, r6
+    tst  r6, #(1<<30)
+    beq 1f
+    ldmfd sp!, {r5}
+    vmsr fpscr, r5
+    vldmia sp!, {d16-d31}
+    vldmia sp!, {d0-d15}
+1:
+#endif
 
 #ifdef RT_USING_LWP
     ldmfd   sp, {r13, r14}^ /* usr_sp, usr_lr */
diff --git a/libcpu/arm/cortex-a/cpu.c b/libcpu/arm/cortex-a/cpu.c
index 6b129f89ee..7c6bf58a4b 100644
--- a/libcpu/arm/cortex-a/cpu.c
+++ b/libcpu/arm/cortex-a/cpu.c
@@ -14,6 +14,7 @@
 #include <board.h>
 
 #ifdef RT_USING_SMP
+
 int rt_hw_cpu_id(void)
 {
     int cpu_id;
@@ -25,6 +26,11 @@ int rt_hw_cpu_id(void)
     return cpu_id;
 };
 
+void rt_hw_spin_lock_init(rt_hw_spinlock_t *lock)
+{
+    lock->slock = 0;
+}
+
 void rt_hw_spin_lock(rt_hw_spinlock_t *lock)
 {
     unsigned long tmp;
diff --git a/libcpu/arm/cortex-a/stack.c b/libcpu/arm/cortex-a/stack.c
index d76b8a428a..07d795ab5b 100644
--- a/libcpu/arm/cortex-a/stack.c
+++ b/libcpu/arm/cortex-a/stack.c
@@ -60,6 +60,9 @@ rt_uint8_t *rt_hw_stack_init(void *tentry, void *parameter,
     *(--stk) = 0;       /* user lr */
     *(--stk) = 0;       /* user sp*/
 #endif
+#ifdef RT_USING_FPU
+    *(--stk) = 0;       /* not use fpu*/
+#endif
 
     /* return task's current stack address */
     return (rt_uint8_t *)stk;
diff --git a/libcpu/arm/cortex-a/start_gcc.S b/libcpu/arm/cortex-a/start_gcc.S
index 014846f3aa..268cc7ef56 100644
--- a/libcpu/arm/cortex-a/start_gcc.S
+++ b/libcpu/arm/cortex-a/start_gcc.S
@@ -22,7 +22,11 @@
 .equ I_Bit,           0x80            @ when I bit is set, IRQ is disabled
 .equ F_Bit,           0x40            @ when F bit is set, FIQ is disabled
 
+#ifdef RT_USING_FPU
+.equ UND_Stack_Size,     0x00000400
+#else
 .equ UND_Stack_Size,     0x00000000
+#endif
 .equ SVC_Stack_Size,     0x00000400
 .equ ABT_Stack_Size,     0x00000000
 .equ RT_FIQ_STACK_PGSZ,  0x00000000
@@ -37,6 +41,7 @@
 .globl stack_start
 .globl stack_top
 
+.align 3
 stack_start:
 .rept ISR_Stack_Size
 .byte 0
@@ -50,6 +55,11 @@ _reset:
     /* set the cpu to SVC32 mode and disable interrupt */
     cps #Mode_SVC
 
+#ifdef RT_USING_FPU
+    mov r4, #0xfffffff
+    mcr p15, 0, r4, c1, c0, 2
+#endif
+
     /* disable the data alignment check */
     mrc p15, 0, r1, c1, c0, 0
     bic r1, #(1<<1)
@@ -176,6 +186,19 @@ vector_irq:
     stmfd   r0, {r13, r14}^ /* usr_sp, usr_lr */
     sub   r0, #8
 #endif
+#ifdef RT_USING_FPU
+    /* fpu context */
+    vmrs r6, fpexc
+    tst  r6, #(1<<30)
+    beq 1f
+    vstmdb r0!, {d0-d15}
+    vstmdb r0!, {d16-d31}
+    vmrs r5, fpscr
+    stmfd r0!, {r5}
+1:
+    stmfd r0!, {r6}
+#endif
+
     /* now irq stack is clean */
     /* r0 is task svc_sp */
     /* backup r0 -> r8 */
@@ -234,6 +257,18 @@ rt_hw_context_switch_interrupt_do:
     stmfd sp, {r13, r14}^  @push usr_sp, usr_lr
     sub sp, #8
 #endif
+#ifdef RT_USING_FPU
+    /* fpu context */
+    vmrs r6, fpexc
+    tst  r6, #(1<<30)
+    beq 1f
+    vstmdb sp!, {d0-d15}
+    vstmdb sp!, {d16-d31}
+    vmrs r5, fpscr
+    stmfd sp!, {r5}
+1:
+    stmfd sp!, {r6}
+#endif
 
     ldr     r4,  =rt_interrupt_from_thread
     ldr     r5,  [r4]
@@ -243,6 +278,19 @@ rt_hw_context_switch_interrupt_do:
     ldr     r6,  [r6]
     ldr     sp,  [r6]       @ get new task's stack pointer
 
+#ifdef RT_USING_FPU
+/* fpu context */
+    ldmfd sp!, {r6}
+    vmsr fpexc, r6
+    tst  r6, #(1<<30)
+    beq 1f
+    ldmfd sp!, {r5}
+    vmsr fpscr, r5
+    vldmia sp!, {d16-d31}
+    vldmia sp!, {d0-d15}
+1:
+#endif
+
 #ifdef RT_USING_LWP
     ldmfd sp, {r13, r14}^  @pop usr_sp, usr_lr
     add sp, #8
@@ -278,7 +326,14 @@ vector_swi:
     .globl  vector_undef
 vector_undef:
     push_svc_reg
+    cps #Mode_UND
     bl      rt_hw_trap_undef
+#ifdef RT_USING_FPU
+    ldr     lr, [sp, #15*4]
+    ldmia   sp, {r0 - r12}
+    add     sp, sp, #17 * 4
+    movs    pc, lr
+#endif
     b       .
 
     .align  5
@@ -315,6 +370,12 @@ set_secondary_cpu_boot_address:
 
 .global secondary_cpu_start
 secondary_cpu_start:
+
+#ifdef RT_USING_FPU
+    mov r4, #0xfffffff
+    mcr p15, 0, r4, c1, c0, 2
+#endif
+
     mrc p15, 0, r1, c1, c0, 1
     mov r0, #(1<<6)
     orr r1, r0
@@ -324,6 +385,11 @@ secondary_cpu_start:
     bic r0, #(1<<13)
     mcr p15, 0, r0, c1, c0, 0
 
+#ifdef RT_USING_FPU
+    cps #Mode_UND
+    ldr sp, =und_stack_2_limit
+#endif
+
     cps #Mode_IRQ
     ldr sp, =irq_stack_2_limit
 
@@ -349,3 +415,8 @@ irq_stack_2:
     .space (1 << 10)
 irq_stack_2_limit:
 
+#ifdef RT_USING_FPU
+und_stack_2:
+    .space (1 << 10)
+und_stack_2_limit:
+#endif
diff --git a/libcpu/arm/cortex-a/trap.c b/libcpu/arm/cortex-a/trap.c
index a25f6ac862..dc232a46ee 100644
--- a/libcpu/arm/cortex-a/trap.c
+++ b/libcpu/arm/cortex-a/trap.c
@@ -45,6 +45,39 @@ void rt_hw_show_register(struct rt_hw_exp_stack *regs)
  */
 void rt_hw_trap_undef(struct rt_hw_exp_stack *regs)
 {
+#ifdef RT_USING_FPU
+    {
+        uint32_t ins;
+        uint32_t addr;
+
+        if (regs->cpsr & (1 << 5))
+        {
+            /* thumb mode */
+            addr = regs->pc - 2;
+            ins = (uint32_t)*(uint16_t*)addr;
+            if ((ins & (3 << 11)) != 0)
+            {
+                /* 32 bit ins */
+                ins <<= 16;
+                ins += *(uint16_t*)(addr + 2);
+            }
+        }
+        else
+        {
+            addr = regs->pc - 4;
+            ins = *(uint32_t*)addr;
+        }
+        if ((ins & 0xe00) == 0xa00)
+        {
+            /* float ins */
+            uint32_t val = (1U << 30);
+
+            asm volatile ("vmsr fpexc, %0"::"r"(val):"memory");
+            regs->pc = addr;
+            return;
+        }
+    }
+#endif
     rt_kprintf("undefined instruction:\n");
     rt_hw_show_register(regs);
 #ifdef RT_USING_FINSH
diff --git a/libcpu/arm/cortex-m0/cpuport.c b/libcpu/arm/cortex-m0/cpuport.c
index 5d36fa11fe..19aec42230 100644
--- a/libcpu/arm/cortex-m0/cpuport.c
+++ b/libcpu/arm/cortex-m0/cpuport.c
@@ -119,7 +119,7 @@ void rt_hw_hard_fault_exception(struct exception_stack_frame *contex)
 #define SCB_HFSR        (*(volatile const unsigned *)0xE000ED2C) /* HardFault Status Register */
 #define SCB_MMAR        (*(volatile const unsigned *)0xE000ED34) /* MemManage Fault Address register */
 #define SCB_BFAR        (*(volatile const unsigned *)0xE000ED38) /* Bus Fault Address Register */
-#define SCB_AIRCR       (*(volatile unsigned long *)0xE000ED00)  /* Reset control Address Register */
+#define SCB_AIRCR       (*(volatile unsigned long *)0xE000ED0C)  /* Reset control Address Register */
 #define SCB_RESET_VALUE 0x05FA0004                               /* Reset value, write to SCB_AIRCR can reset cpu */
 
 #define SCB_CFSR_MFSR   (*(volatile const unsigned char*)0xE000ED28)  /* Memory-management Fault Status Register */
diff --git a/libcpu/arm/cortex-m3/cpuport.c b/libcpu/arm/cortex-m3/cpuport.c
index a386747750..d17d4fdf2e 100644
--- a/libcpu/arm/cortex-m3/cpuport.c
+++ b/libcpu/arm/cortex-m3/cpuport.c
@@ -11,6 +11,7 @@
  * 2012-12-23   aozima      stack addr align to 8byte.
  * 2012-12-29   Bernard     Add exception hook.
  * 2013-07-09   aozima      enhancement hard fault exception handler.
+ * 2019-07-03   yangjie     add __rt_ffs() for armclang.
  */
 
 #include <rtthread.h>
@@ -382,6 +383,24 @@ __asm int __rt_ffs(int value)
 exit
     BX      lr
 }
+#elif defined(__CLANG_ARM)
+int __rt_ffs(int value)
+{
+    __asm volatile(
+        "CMP     r0, #0x00            \n"
+        "BEQ     exit                 \n"
+
+        "RBIT    r0, r0               \n"
+        "CLZ     r0, r0               \n"
+        "ADDS    r0, r0, #0x01        \n"
+
+        "exit:                        \n"
+
+        : "=r"(value)
+        : "r"(value)
+    );
+    return value;
+}
 #elif defined(__IAR_SYSTEMS_ICC__)
 int __rt_ffs(int value)
 {
diff --git a/libcpu/arm/cortex-m4/cpuport.c b/libcpu/arm/cortex-m4/cpuport.c
index 18f027895b..f556ba0c26 100644
--- a/libcpu/arm/cortex-m4/cpuport.c
+++ b/libcpu/arm/cortex-m4/cpuport.c
@@ -14,6 +14,7 @@
  * 2012-12-29     Bernard      Add exception hook.
  * 2013-06-23     aozima       support lazy stack optimized.
  * 2018-07-24     aozima       enhancement hard fault exception handler.
+ * 2019-07-03     yangjie      add __rt_ffs() for armclang.
  */
 
 #include <rtthread.h>
@@ -453,7 +454,7 @@ RT_WEAK void rt_hw_cpu_reset(void)
  * @return return the index of the first bit set. If value is 0, then this function
  * shall return 0.
  */
-#if defined(__CC_ARM) || defined(__CLANG_ARM)
+#if defined(__CC_ARM) 
 __asm int __rt_ffs(int value)
 {
     CMP     r0, #0x00
@@ -466,6 +467,24 @@ __asm int __rt_ffs(int value)
 exit
     BX      lr
 }
+#elif defined(__CLANG_ARM)
+int __rt_ffs(int value)
+{
+    __asm volatile(
+        "CMP     r0, #0x00            \n"
+        "BEQ     exit                 \n"
+
+        "RBIT    r0, r0               \n"
+        "CLZ     r0, r0               \n"
+        "ADDS    r0, r0, #0x01        \n"
+
+        "exit:                        \n"
+
+        : "=r"(value)
+        : "r"(value)
+    );
+    return value;
+}
 #elif defined(__IAR_SYSTEMS_ICC__)
 int __rt_ffs(int value)
 {
diff --git a/libcpu/arm/cortex-m7/cpu_cache.c b/libcpu/arm/cortex-m7/cpu_cache.c
index 4a1a9c2201..350b819c00 100644
--- a/libcpu/arm/cortex-m7/cpu_cache.c
+++ b/libcpu/arm/cortex-m7/cpu_cache.c
@@ -72,15 +72,15 @@ void rt_hw_cpu_dcache_ops(int ops, void* addr, int size)
 
     if (ops & (RT_HW_CACHE_FLUSH | RT_HW_CACHE_INVALIDATE))
     {
-        SCB_CleanInvalidateDCache_by_Addr((rt_uint32_t *)startAddr, size_byte);
+        SCB_CleanInvalidateDCache_by_Addr((uint32_t *)startAddr, size_byte);
     }
     else if (ops & RT_HW_CACHE_FLUSH)
     {
-        SCB_CleanDCache_by_Addr((rt_uint32_t *)startAddr, size_byte);
+        SCB_CleanDCache_by_Addr((uint32_t *)startAddr, size_byte);
     }
     else if (ops & RT_HW_CACHE_INVALIDATE)
     {
-        SCB_InvalidateDCache_by_Addr((rt_uint32_t *)startAddr, size_byte);
+        SCB_InvalidateDCache_by_Addr((uint32_t *)startAddr, size_byte);
     }
     else
     {
diff --git a/libcpu/arm/cortex-m7/cpuport.c b/libcpu/arm/cortex-m7/cpuport.c
index 18f027895b..c22242f140 100644
--- a/libcpu/arm/cortex-m7/cpuport.c
+++ b/libcpu/arm/cortex-m7/cpuport.c
@@ -14,6 +14,7 @@
  * 2012-12-29     Bernard      Add exception hook.
  * 2013-06-23     aozima       support lazy stack optimized.
  * 2018-07-24     aozima       enhancement hard fault exception handler.
+ * 2019-07-03     yangjie      add __rt_ffs() for armclang.
  */
 
 #include <rtthread.h>
@@ -453,7 +454,7 @@ RT_WEAK void rt_hw_cpu_reset(void)
  * @return return the index of the first bit set. If value is 0, then this function
  * shall return 0.
  */
-#if defined(__CC_ARM) || defined(__CLANG_ARM)
+#if defined(__CC_ARM)
 __asm int __rt_ffs(int value)
 {
     CMP     r0, #0x00
@@ -466,6 +467,24 @@ __asm int __rt_ffs(int value)
 exit
     BX      lr
 }
+#elif defined(__CLANG_ARM)
+int __rt_ffs(int value)
+{
+    __asm volatile(
+        "CMP     r0, #0x00            \n"
+        "BEQ     exit                 \n"
+
+        "RBIT    r0, r0               \n"
+        "CLZ     r0, r0               \n"
+        "ADDS    r0, r0, #0x01        \n"
+
+        "exit:                        \n"
+
+        : "=r"(value)
+        : "r"(value)
+    );
+    return value;
+}
 #elif defined(__IAR_SYSTEMS_ICC__)
 int __rt_ffs(int value)
 {
diff --git a/libcpu/arm/cortex-r4/start_gcc.S b/libcpu/arm/cortex-r4/start_gcc.S
index ab77b98781..a347503ec0 100644
--- a/libcpu/arm/cortex-r4/start_gcc.S
+++ b/libcpu/arm/cortex-r4/start_gcc.S
@@ -36,6 +36,7 @@
 .globl stack_start
 .globl stack_top
 
+.align 3
 stack_start:
 .rept (UND_Stack_Size + SVC_Stack_Size + ABT_Stack_Size + FIQ_Stack_Size + IRQ_Stack_Size)
 .byte 0
diff --git a/libcpu/arm/realview-a8-vmm/start_gcc.S b/libcpu/arm/realview-a8-vmm/start_gcc.S
index 2ec7fe6c86..879cf992a8 100644
--- a/libcpu/arm/realview-a8-vmm/start_gcc.S
+++ b/libcpu/arm/realview-a8-vmm/start_gcc.S
@@ -47,6 +47,7 @@
 .globl stack_start
 .globl stack_top
 
+.align 3
 stack_start:
 .rept ISR_Stack_Size
 .byte 0
diff --git a/libcpu/arm/zynq7000/start_gcc.S b/libcpu/arm/zynq7000/start_gcc.S
index 6904c45587..18056f0ad9 100644
--- a/libcpu/arm/zynq7000/start_gcc.S
+++ b/libcpu/arm/zynq7000/start_gcc.S
@@ -33,6 +33,7 @@
 .globl stack_start
 .globl stack_top
 
+.align 3
 .bss
 stack_start:
 .rept ISR_Stack_Size