Merge pull request #3418 from bigmagic123/raspi-64-sdio

fix raspi3-64 sdio driver

10 files changed, 456 insertions, 426 deletions

diff --git a/bsp/raspberry-pi/raspi3-64/.config b/bsp/raspberry-pi/raspi3-64/.config
index 8ef205d539..28e075c64a 100644
--- a/bsp/raspberry-pi/raspi3-64/.config
+++ b/bsp/raspberry-pi/raspi3-64/.config
@@ -65,7 +65,7 @@ CONFIG_RT_USING_DEVICE=y
 CONFIG_RT_USING_DEVICE_OPS=y
 # CONFIG_RT_USING_INTERRUPT_INFO is not set
 CONFIG_RT_USING_CONSOLE=y
-CONFIG_RT_CONSOLEBUF_SIZE=128
+CONFIG_RT_CONSOLEBUF_SIZE=512
 CONFIG_RT_CONSOLE_DEVICE_NAME="uart1"
 CONFIG_RT_VER_NUM=0x40003
 CONFIG_ARCH_CPU_64BIT=y
@@ -77,7 +77,7 @@ CONFIG_ARCH_CPU_64BIT=y
 #
 CONFIG_RT_USING_COMPONENTS_INIT=y
 CONFIG_RT_USING_USER_MAIN=y
-CONFIG_RT_MAIN_THREAD_STACK_SIZE=2048
+CONFIG_RT_MAIN_THREAD_STACK_SIZE=4096
 CONFIG_RT_MAIN_THREAD_PRIORITY=10
 
 #
@@ -113,7 +113,23 @@ CONFIG_DFS_FILESYSTEMS_MAX=2
 CONFIG_DFS_FILESYSTEM_TYPES_MAX=2
 CONFIG_DFS_FD_MAX=16
 # CONFIG_RT_USING_DFS_MNTTABLE is not set
-# CONFIG_RT_USING_DFS_ELMFAT is not set
+CONFIG_RT_USING_DFS_ELMFAT=y
+
+#
+# elm-chan's FatFs, Generic FAT Filesystem Module
+#
+CONFIG_RT_DFS_ELM_CODE_PAGE=437
+CONFIG_RT_DFS_ELM_WORD_ACCESS=y
+# CONFIG_RT_DFS_ELM_USE_LFN_0 is not set
+# CONFIG_RT_DFS_ELM_USE_LFN_1 is not set
+# CONFIG_RT_DFS_ELM_USE_LFN_2 is not set
+CONFIG_RT_DFS_ELM_USE_LFN_3=y
+CONFIG_RT_DFS_ELM_USE_LFN=3
+CONFIG_RT_DFS_ELM_MAX_LFN=255
+CONFIG_RT_DFS_ELM_DRIVES=2
+CONFIG_RT_DFS_ELM_MAX_SECTOR_SIZE=512
+# CONFIG_RT_DFS_ELM_USE_ERASE is not set
+CONFIG_RT_DFS_ELM_REENTRANT=y
 CONFIG_RT_USING_DFS_DEVFS=y
 # CONFIG_RT_USING_DFS_ROMFS is not set
 # CONFIG_RT_USING_DFS_RAMFS is not set
@@ -140,7 +156,13 @@ CONFIG_RT_USING_PIN=y
 # CONFIG_RT_USING_MTD_NAND is not set
 # CONFIG_RT_USING_PM is not set
 # CONFIG_RT_USING_RTC is not set
-# CONFIG_RT_USING_SDIO is not set
+CONFIG_RT_USING_SDIO=y
+CONFIG_RT_SDIO_STACK_SIZE=2048
+CONFIG_RT_SDIO_THREAD_PRIORITY=15
+CONFIG_RT_MMCSD_STACK_SIZE=4096
+CONFIG_RT_MMCSD_THREAD_PREORITY=22
+CONFIG_RT_MMCSD_MAX_PARTITION=16
+CONFIG_RT_SDIO_DEBUG=y
 # CONFIG_RT_USING_SPI is not set
 # CONFIG_RT_USING_WDT is not set
 # CONFIG_RT_USING_AUDIO is not set
@@ -448,7 +470,8 @@ CONFIG_BSP_USING_CORETIMER=y
 # CONFIG_BSP_USING_SPI is not set
 # CONFIG_BSP_USING_WDT is not set
 # CONFIG_BSP_USING_RTC is not set
-# CONFIG_BSP_USING_SDIO is not set
+CONFIG_BSP_USING_SDIO=y
+CONFIG_BSP_USING_SDIO0=y
 
 #
 # Board Peripheral Drivers
diff --git a/bsp/raspberry-pi/raspi3-64/applications/mnt.c b/bsp/raspberry-pi/raspi3-64/applications/mnt.c
index 8ec394eb8c..33b466434c 100644
--- a/bsp/raspberry-pi/raspi3-64/applications/mnt.c
+++ b/bsp/raspberry-pi/raspi3-64/applications/mnt.c
@@ -10,7 +10,19 @@
 
 #include <rtthread.h>
 
+#ifdef BSP_USING_SDIO0
+#include <dfs_fs.h>
+
 int mnt_init(void)
 {
+    rt_thread_delay(RT_TICK_PER_SECOND);
+    if (dfs_mount("sd0", "/", "elm", 0, 0) == 0)
+    {
+        rt_kprintf("file system initialization done!\n");
+    }
+
     return 0;
 }
+INIT_ENV_EXPORT(mnt_init);
+#endif
+
diff --git a/bsp/raspberry-pi/raspi3-64/driver/board.c b/bsp/raspberry-pi/raspi3-64/driver/board.c
index a12755657b..cfaa3c9294 100644
--- a/bsp/raspberry-pi/raspi3-64/driver/board.c
+++ b/bsp/raspberry-pi/raspi3-64/driver/board.c
@@ -93,10 +93,11 @@ void idle_wfi(void)
 void rt_hw_board_init(void)
 {
     mmu_init();
-    armv8_map(0, 0, 0x800000, MEM_ATTR_MEMORY);
-    armv8_map(0x3f00B000, 0x3f00B000, 0x1000, MEM_ATTR_IO);//timer
+    armv8_map(0, 0, 0x6400000, MEM_ATTR_MEMORY);
+    armv8_map(0x3f000000, 0x3f000000, 0x200000, MEM_ATTR_IO);//timer
     armv8_map(0x3f200000, 0x3f200000, 0x16000, MEM_ATTR_IO);//uart
     armv8_map(0x40000000, 0x40000000, 0x1000, MEM_ATTR_IO);//core timer
+    armv8_map(0x3F300000, 0x3F300000, 0x1000, MEM_ATTR_IO);//sdio
     mmu_enable();
     
     /* initialize hardware interrupt */
diff --git a/bsp/raspberry-pi/raspi3-64/driver/board.h b/bsp/raspberry-pi/raspi3-64/driver/board.h
index 40fbfd9d84..e42c249201 100644
--- a/bsp/raspberry-pi/raspi3-64/driver/board.h
+++ b/bsp/raspberry-pi/raspi3-64/driver/board.h
@@ -21,7 +21,7 @@ extern unsigned char __bss_start;
 extern unsigned char __bss_end;
 
 #define RT_HW_HEAP_BEGIN    (void*)&__bss_end
-#define RT_HW_HEAP_END      (void*)(RT_HW_HEAP_BEGIN + 4 * 1024 * 1024)
+#define RT_HW_HEAP_END      (void*)(RT_HW_HEAP_BEGIN + 64 * 1024 * 1024)
 
 void rt_hw_board_init(void);
 
diff --git a/bsp/raspberry-pi/raspi3-64/driver/drv_sdio.c b/bsp/raspberry-pi/raspi3-64/driver/drv_sdio.c
index 4f3018c0ef..785f335ebb 100644
--- a/bsp/raspberry-pi/raspi3-64/driver/drv_sdio.c
+++ b/bsp/raspberry-pi/raspi3-64/driver/drv_sdio.c
@@ -9,264 +9,306 @@
  * 2019-07-29     zdzn           first version
  */
 
-#include <rtthread.h>
-#include <rthw.h>
-#include <rtdevice.h>
-#include <string.h>
-
 #include "drv_sdio.h"
-#include "interrupt.h"
-#include "drv_gpio.h"
-
-#include "bcm283x.h"
-#include <drivers/mmcsd_core.h>
-#include "bcm283x.h"
-
+#include "raspi.h"
 
-#include <rtdbg.h>
-
-#ifdef RT_USING_SDIO
-#define CONFIG_MMC_USE_DMA
-#define DMA_ALIGN       (32U)
+static rt_uint32_t sdCommandTable[] = {
+    SD_CMD_INDEX(0),
+    SD_CMD_RESERVED(1),
+    SD_CMD_INDEX(2) | SD_RESP_R2,
+    SD_CMD_INDEX(3) | SD_RESP_R1,
+    SD_CMD_INDEX(4),
+    SD_CMD_RESERVED(5), //SD_CMD_INDEX(5) | SD_RESP_R4,
+    SD_CMD_INDEX(6) | SD_RESP_R1,
+    SD_CMD_INDEX(7) | SD_RESP_R1b,
+    SD_CMD_INDEX(8) | SD_RESP_R1,
+    SD_CMD_INDEX(9) | SD_RESP_R2,
+    SD_CMD_INDEX(10) | SD_RESP_R2,
+    SD_CMD_INDEX(11) | SD_RESP_R1,
+    SD_CMD_INDEX(12) | SD_RESP_R1b | SD_CMD_TYPE_ABORT,
+    SD_CMD_INDEX(13) | SD_RESP_R1,
+    SD_CMD_RESERVED(14),
+    SD_CMD_INDEX(15),
+    SD_CMD_INDEX(16) | SD_RESP_R1,
+    SD_CMD_INDEX(17) | SD_RESP_R1 | SD_DATA_READ,
+    SD_CMD_INDEX(18) | SD_RESP_R1 | SD_DATA_READ | SD_CMD_MULTI_BLOCK | SD_CMD_BLKCNT_EN,
+    SD_CMD_INDEX(19) | SD_RESP_R1 | SD_DATA_READ,
+    SD_CMD_INDEX(20) | SD_RESP_R1b,
+    SD_CMD_RESERVED(21),
+    SD_CMD_RESERVED(22),
+    SD_CMD_INDEX(23) | SD_RESP_R1,
+    SD_CMD_INDEX(24) | SD_RESP_R1 | SD_DATA_WRITE,
+    SD_CMD_INDEX(25) | SD_RESP_R1 | SD_DATA_WRITE | SD_CMD_MULTI_BLOCK | SD_CMD_BLKCNT_EN,
+    SD_CMD_INDEX(26) | SD_RESP_R1 | SD_DATA_WRITE, //add
+    SD_CMD_INDEX(27) | SD_RESP_R1 | SD_DATA_WRITE,
+    SD_CMD_INDEX(28) | SD_RESP_R1b,
+    SD_CMD_INDEX(29) | SD_RESP_R1b,
+    SD_CMD_INDEX(30) | SD_RESP_R1 | SD_DATA_READ,
+    SD_CMD_RESERVED(31),
+    SD_CMD_INDEX(32) | SD_RESP_R1,
+    SD_CMD_INDEX(33) | SD_RESP_R1,
+    SD_CMD_RESERVED(34),
+    SD_CMD_INDEX(35) | SD_RESP_R1, //add
+    SD_CMD_INDEX(36) | SD_RESP_R1, //add
+    SD_CMD_RESERVED(37),
+    SD_CMD_INDEX(38) | SD_RESP_R1b,
+    SD_CMD_INDEX(39) | SD_RESP_R4, //add
+    SD_CMD_INDEX(40) | SD_RESP_R5, //add
+    SD_CMD_INDEX(41) | SD_RESP_R3, //add, mov from harbote
+    SD_CMD_RESERVED(42) | SD_RESP_R1,
+    SD_CMD_RESERVED(43),
+    SD_CMD_RESERVED(44),
+    SD_CMD_RESERVED(45),
+    SD_CMD_RESERVED(46),
+    SD_CMD_RESERVED(47),
+    SD_CMD_RESERVED(48),
+    SD_CMD_RESERVED(49),
+    SD_CMD_RESERVED(50),
+    SD_CMD_INDEX(51) | SD_RESP_R1 | SD_DATA_READ,
+    SD_CMD_RESERVED(52),
+    SD_CMD_RESERVED(53),
+    SD_CMD_RESERVED(54),
+    SD_CMD_INDEX(55) | SD_RESP_R3,
+    SD_CMD_INDEX(56) | SD_RESP_R1 | SD_CMD_ISDATA,
+    SD_CMD_RESERVED(57),
+    SD_CMD_RESERVED(58),
+    SD_CMD_RESERVED(59),
+    SD_CMD_RESERVED(60),
+    SD_CMD_RESERVED(61),
+    SD_CMD_RESERVED(62),
+    SD_CMD_RESERVED(63)
+};
 
-typedef struct EMMCCommand
+static inline rt_uint32_t read32(rt_uint32_t addr)
 {
-    const char* name;
-    unsigned int code;
-    unsigned char resp;
-    unsigned char rca;
-    int delay;
-} EMMCCommand;
+    return (*((volatile unsigned int*)((rt_uint64_t)addr)));
+    //return (*((volatile rt_uint64_t *)(((long))addr)));
+}
 
-static EMMCCommand sdCommandTable[] =
+static inline void write32(rt_uint32_t addr, rt_uint32_t value)
 {
-    {"GO_IDLE_STATE", 0x00000000 | CMD_RSPNS_NO                             , RESP_NO , RCA_NO  ,0},
-    {"ALL_SEND_CID" , 0x02000000 | CMD_RSPNS_136                            , RESP_R2I, RCA_NO  ,0},
-    {"SEND_REL_ADDR", 0x03000000 | CMD_RSPNS_48                             , RESP_R6 , RCA_NO  ,0},
-    {"SET_DSR"      , 0x04000000 | CMD_RSPNS_NO                             , RESP_NO , RCA_NO  ,0},
-    {"SWITCH_FUNC"  , 0x06000000 | CMD_RSPNS_48                             , RESP_R1 , RCA_NO  ,0},
-    {"CARD_SELECT"  , 0x07000000 | CMD_RSPNS_48B                            , RESP_R1b, RCA_YES ,0},
-    {"SEND_IF_COND" , 0x08000000 | CMD_RSPNS_48                             , RESP_R7 , RCA_NO  ,100},
-    {"SEND_CSD"     , 0x09000000 | CMD_RSPNS_136                            , RESP_R2S, RCA_YES ,0},
-    {"SEND_CID"     , 0x0A000000 | CMD_RSPNS_136                            , RESP_R2I, RCA_YES ,0},
-    {"VOLT_SWITCH"  , 0x0B000000 | CMD_RSPNS_48                             , RESP_R1 , RCA_NO  ,0},
-    {"STOP_TRANS"   , 0x0C000000 | CMD_RSPNS_48B                            , RESP_R1b, RCA_NO  ,0},
-    {"SEND_STATUS"  , 0x0D000000 | CMD_RSPNS_48                             , RESP_R1 , RCA_YES ,0},
-    {"GO_INACTIVE"  , 0x0F000000 | CMD_RSPNS_NO                             , RESP_NO , RCA_YES ,0},
-    {"SET_BLOCKLEN" , 0x10000000 | CMD_RSPNS_48                             , RESP_R1 , RCA_NO  ,0},
-    {"READ_SINGLE"  , 0x11000000 | CMD_RSPNS_48 | CMD_IS_DATA   | TM_DAT_DIR_CH, RESP_R1 , RCA_NO  ,0},
-    {"READ_MULTI"   , 0x12000000 | CMD_RSPNS_48 | TM_MULTI_DATA | TM_DAT_DIR_CH, RESP_R1 , RCA_NO  ,0},
-    {"SEND_TUNING"  , 0x13000000 | CMD_RSPNS_48                             , RESP_R1 , RCA_NO  ,0},
-    {"SPEED_CLASS"  , 0x14000000 | CMD_RSPNS_48B                            , RESP_R1b, RCA_NO  ,0},
-    {"SET_BLOCKCNT" , 0x17000000 | CMD_RSPNS_48                             , RESP_R1 , RCA_NO  ,0},
-    {"WRITE_SINGLE" , 0x18000000 | CMD_RSPNS_48 | CMD_IS_DATA   | TM_DAT_DIR_HC, RESP_R1 , RCA_NO  ,0},
-    {"WRITE_MULTI"  , 0x19000000 | CMD_RSPNS_48 | TM_MULTI_DATA | TM_DAT_DIR_HC, RESP_R1 , RCA_NO  ,0},
-    {"PROGRAM_CSD"  , 0x1B000000 | CMD_RSPNS_48                             , RESP_R1 , RCA_NO  ,0},
-    {"SET_WRITE_PR" , 0x1C000000 | CMD_RSPNS_48B                            , RESP_R1b, RCA_NO  ,0},
-    {"CLR_WRITE_PR" , 0x1D000000 | CMD_RSPNS_48B                            , RESP_R1b, RCA_NO  ,0},
-    {"SND_WRITE_PR" , 0x1E000000 | CMD_RSPNS_48                             , RESP_R1 , RCA_NO  ,0},
-    {"ERASE_WR_ST"  , 0x20000000 | CMD_RSPNS_48                             , RESP_R1 , RCA_NO  ,0},
-    {"ERASE_WR_END" , 0x21000000 | CMD_RSPNS_48                             , RESP_R1 , RCA_NO  ,0},
-    {"ERASE"        , 0x26000000 | CMD_RSPNS_48B                            , RESP_R1b, RCA_NO  ,0},
-    {"LOCK_UNLOCK"  , 0x2A000000 | CMD_RSPNS_48                             , RESP_R1 , RCA_NO  ,0},
-    {"APP_CMD"      , 0x37000000 | CMD_RSPNS_NO                             , RESP_NO , RCA_NO  ,100},
-    {"APP_CMD"      , 0x37000000 | CMD_RSPNS_48                             , RESP_R1 , RCA_YES ,0},
-    {"GEN_CMD"      , 0x38000000 | CMD_RSPNS_48                             , RESP_R1 , RCA_NO  ,0},
-
-  // APP commands must be prefixed by an APP_CMD.
-    {"SET_BUS_WIDTH", 0x06000000 | CMD_RSPNS_48                             , RESP_R1 , RCA_NO  ,0},
-    {"SD_STATUS"    , 0x0D000000 | CMD_RSPNS_48                             , RESP_R1 , RCA_YES ,0}, // RCA???
-    {"SEND_NUM_WRBL", 0x16000000 | CMD_RSPNS_48                             , RESP_R1 , RCA_NO  ,0},
-    {"SEND_NUM_ERS" , 0x17000000 | CMD_RSPNS_48                             , RESP_R1 , RCA_NO  ,0},
-    {"SD_SENDOPCOND", 0x29000000 | CMD_RSPNS_48                             , RESP_R3 , RCA_NO  ,1000},
-    {"SET_CLR_DET"  , 0x2A000000 | CMD_RSPNS_48                             , RESP_R1 , RCA_NO  ,0},
-    {"SEND_SCR"     , 0x33000000 | CMD_RSPNS_48 | CMD_IS_DATA | TM_DAT_DIR_CH   , RESP_R1 , RCA_NO  ,0},
-};
-
-static rt_err_t sdhci_setwidth(struct sdhci_t * sdhci, rt_uint32_t width);
-static rt_err_t sdhci_setclock(struct sdhci_t * sdhci, rt_uint32_t clock);
-static rt_err_t sdhci_transfer(struct sdhci_t * sdhci, struct sdhci_cmd_t * cmd, struct sdhci_data_t * dat);
+    (*((volatile unsigned int*)((rt_uint64_t)addr))) = value;
+    //*((volatile rt_uint64_t *)(((long))addr)) = value;
+}
 
-static inline rt_uint32_t read32(rt_uint32_t addr)
+rt_err_t sd_int(struct sdhci_pdata_t * pdat, unsigned int mask)
 {
-    return( *((volatile rt_uint32_t *)(addr)) );
+    unsigned int r;
+    unsigned int m = mask | INT_ERROR_MASK;
+    int cnt = 1000000;
+    while (!(read32(pdat->virt + EMMC_INTERRUPT) & (m | INT_ERROR_MASK)) && cnt--)
+        DELAY_MICROS(1);
+    r = read32(pdat->virt + EMMC_INTERRUPT);
+    if (cnt <= 0 || (r & INT_CMD_TIMEOUT) || (r & INT_DATA_TIMEOUT))
+    {
+        write32(pdat->virt + EMMC_INTERRUPT, r);
+        //qemu maybe can not use sdcard
+        //rt_kprintf("send cmd/data timeout wait for %x int: %x, status: %x\n",mask, r, read32(pdat->virt + EMMC_STATUS));
+        //return -RT_ETIMEOUT;
+    }
+    else if (r & INT_ERROR_MASK)
+    {
+        write32(pdat->virt + EMMC_INTERRUPT, r);
+        rt_kprintf("send cmd/data error %x -> %x\n",r, read32(pdat->virt + EMMC_INTERRUPT));
+        return -RT_ERROR;
+    }
+    write32(pdat->virt + EMMC_INTERRUPT, mask);
+    return RT_EOK;
 }
 
-static inline void write32(rt_uint32_t addr, rt_uint32_t value)
+rt_err_t sd_status(struct sdhci_pdata_t * pdat, unsigned int mask)
 {
-    *((volatile rt_uint32_t *)(addr)) = value;
+    int cnt = 500000;
+    while ((read32(pdat->virt + EMMC_STATUS) & mask) && !(read32(pdat->virt + EMMC_INTERRUPT) & INT_ERROR_MASK) && cnt--)
+        DELAY_MICROS(1);
+    if (cnt <= 0)
+    {
+        return -RT_ETIMEOUT;
+    }
+    else if (read32(pdat->virt + EMMC_INTERRUPT) & INT_ERROR_MASK)
+    {
+        return  -RT_ERROR;
+    } 
+
+    return RT_EOK;
 }
 
 static rt_err_t raspi_transfer_command(struct sdhci_pdata_t * pdat, struct sdhci_cmd_t * cmd)
 {
     rt_uint32_t cmdidx;
-    rt_uint32_t status;
     rt_err_t ret = RT_EOK;
+    ret = sd_status(pdat, SR_CMD_INHIBIT);
+    if (ret)
+    {
+        rt_kprintf("ERROR: EMMC busy %d\n", ret);
+        return ret;
+    }
 
-    if(read32(pdat->virt + EMMC_STATUS) & SR_CMD_INHIBIT)
-        write32(pdat->virt + EMMC_CMDTM, 0x0);
-
-    EMMCCommand* cmdtab = &sdCommandTable[cmd->cmdidx];
-
-    cmdidx = cmdtab->code;
-
+    cmdidx = sdCommandTable[cmd->cmdidx];
+    if (cmdidx == 0xFFFFFFFF)
+        return -RT_EINVAL;
+    if (cmd->datarw == DATA_READ)
+        cmdidx |= SD_DATA_READ;
+    if (cmd->datarw == DATA_WRITE)
+        cmdidx |= SD_DATA_WRITE;
+    mmcsd_dbg("transfer cmd %x(%d) %x %x\n", cmdidx, cmd->cmdidx, cmd->cmdarg, read32(pdat->virt + EMMC_INTERRUPT));
+    write32(pdat->virt + EMMC_INTERRUPT,read32(pdat->virt + EMMC_INTERRUPT));
     write32(pdat->virt + EMMC_ARG1, cmd->cmdarg);
     write32(pdat->virt + EMMC_CMDTM, cmdidx);
-
-    do {
-        status = read32(pdat->virt + EMMC_STATUS);
-    } while(!(status & SR_CMD_INHIBIT));
-
-    if(cmd->resptype & RESP_MASK)
+    if (cmd->cmdidx == SD_APP_OP_COND)
+        DELAY_MICROS(1000);
+    else if ((cmd->cmdidx == SD_SEND_IF_COND) || (cmd->cmdidx == APP_CMD))
+        DELAY_MICROS(100);
+    ret = sd_int(pdat, INT_CMD_DONE);
+    if (ret)
     {
-        cmd->response[0] = read32(pdat->virt + EMMC_RESP0);
-        if(cmd->resptype & RESP_R2)
+        return ret;
+    }
+    if (cmd->resptype & RESP_MASK)
+    {
+
+        if (cmd->resptype & RESP_R2)
         {
-            cmd->response[1] = read32(pdat->virt + EMMC_RESP1);
-            cmd->response[2] = read32(pdat->virt + EMMC_RESP2);
-            cmd->response[3] = read32(pdat->virt + EMMC_RESP3);
+            rt_uint32_t resp[4];
+            resp[0] = read32(pdat->virt + EMMC_RESP0);
+            resp[1] = read32(pdat->virt + EMMC_RESP1);
+            resp[2] = read32(pdat->virt + EMMC_RESP2);
+            resp[3] = read32(pdat->virt + EMMC_RESP3);
+            if (cmd->resptype == RESP_R2)
+            {
+                cmd->response[0] = resp[3]<<8 |((resp[2]>>24)&0xff);
+                cmd->response[1] = resp[2]<<8 |((resp[1]>>24)&0xff);
+                cmd->response[2] = resp[1]<<8 |((resp[0]>>24)&0xff);
+                cmd->response[3] = resp[0]<<8 ;
+            }
+            else
+            {
+                cmd->response[0] = resp[0];
+                cmd->response[1] = resp[1];
+                cmd->response[2] = resp[2];
+                cmd->response[3] = resp[3];
+            }
         }
+        else
+            cmd->response[0] = read32(pdat->virt + EMMC_RESP0);
     }
-
+    mmcsd_dbg("response: %x: %x %x %x %x (%x, %x)\n", cmd->resptype, cmd->response[0], cmd->response[1], cmd->response[2], cmd->response[3], read32(pdat->virt + EMMC_STATUS),read32(pdat->virt + EMMC_INTERRUPT));
     return ret;
 }
 
 static rt_err_t read_bytes(struct sdhci_pdata_t * pdat, rt_uint32_t * buf, rt_uint32_t blkcount, rt_uint32_t blksize)
 {
-    rt_uint32_t * tmp = buf;
-    rt_uint32_t count = blkcount * blksize;
-    rt_uint32_t status, err;
-
-//    status = read32(pdat->virt + PL180_STATUS);
-//    err = status & (PL180_STAT_DAT_CRC_FAIL | PL180_STAT_DAT_TIME_OUT | PL180_STAT_RX_OVERRUN);
-//    while((!err) && (count >= sizeof(rt_uint32_t)))
-//    {
-//        if(status & PL180_STAT_RX_FIFO_AVL)
-//        {
-//            *(tmp) = read32(pdat->virt + PL180_FIFO);
-//            tmp++;
-//            count -= sizeof(rt_uint32_t);
-//        }
-//        status = read32(pdat->virt + PL180_STATUS);
-//        err = status & (PL180_STAT_DAT_CRC_FAIL | PL180_STAT_DAT_TIME_OUT | PL180_STAT_RX_OVERRUN);
-//    }
-//
-//    err = status & (PL180_STAT_DAT_CRC_FAIL | PL180_STAT_DAT_TIME_OUT | PL180_STAT_DAT_BLK_END | PL180_STAT_RX_OVERRUN);
-//    while(!err)
-//    {
-//        status = read32(pdat->virt + PL180_STATUS);
-//        err = status & (PL180_STAT_DAT_CRC_FAIL | PL180_STAT_DAT_TIME_OUT | PL180_STAT_DAT_BLK_END | PL180_STAT_RX_OVERRUN);
-//    }
-//
-//    if(status & PL180_STAT_DAT_TIME_OUT)
-//        return -RT_ERROR;
-//    else if (status & PL180_STAT_DAT_CRC_FAIL)
-//        return -RT_ERROR;
-//    else if (status & PL180_STAT_RX_OVERRUN)
-//        return -RT_ERROR;
-//    write32(pdat->virt + PL180_CLEAR, 0x1DC007FF);
-//
-//    if(count)
-//        return -RT_ERROR;
-
+    int c = 0;
+    rt_err_t ret;
+    int d;
+    while (c < blkcount)
+    {
+        if ((ret = sd_int(pdat, INT_READ_RDY)))
+        {
+            rt_kprintf("timeout happens when reading block %d\n",c);
+            return ret;
+        }
+        for (d=0; d < blksize / 4; d++)
+            if (read32(pdat->virt + EMMC_STATUS) & SR_READ_AVAILABLE)
+                buf[d] = read32(pdat->virt + EMMC_DATA);
+        c++;
+        buf += blksize / 4;
+    }
     return RT_EOK;
 }
 
 static rt_err_t write_bytes(struct sdhci_pdata_t * pdat, rt_uint32_t * buf, rt_uint32_t blkcount, rt_uint32_t blksize)
 {
-    rt_uint32_t * tmp = buf;
-    rt_uint32_t count = blkcount * blksize;
-    rt_uint32_t status, err;
-    int i;
-
-//    status = read32(pdat->virt + PL180_STATUS);
-//    err = status & (PL180_STAT_DAT_CRC_FAIL | PL180_STAT_DAT_TIME_OUT);
-//    while(!err && count)
-//    {
-//        if(status & PL180_STAT_TX_FIFO_HALF)
-//        {
-//            if(count >= 8 * sizeof(rt_uint32_t))
-//            {
-//                for(i = 0; i < 8; i++)
-//                    write32(pdat->virt + PL180_FIFO, *(tmp + i));
-//                tmp += 8;
-//                count -= 8 * sizeof(rt_uint32_t);
-//            }
-//            else
-//            {
-//                while(count >= sizeof(rt_uint32_t))
-//                {
-//                    write32(pdat->virt + PL180_FIFO, *tmp);
-//                    tmp++;
-//                    count -= sizeof(rt_uint32_t);
-//                }
-//            }
-//        }
-//        status = read32(pdat->virt + PL180_STATUS);
-//        err = status & (PL180_STAT_DAT_CRC_FAIL | PL180_STAT_DAT_TIME_OUT);
-//    }
-//
-//    err = status & (PL180_STAT_DAT_CRC_FAIL | PL180_STAT_DAT_TIME_OUT | PL180_STAT_DAT_BLK_END);
-//    while(!err)
-//    {
-//        status = read32(pdat->virt + PL180_STATUS);
-//        err = status & (PL180_STAT_DAT_CRC_FAIL | PL180_STAT_DAT_TIME_OUT | PL180_STAT_DAT_BLK_END);
-//    }
-//
-//    if(status & PL180_STAT_DAT_TIME_OUT)
-//        return -RT_ERROR;
-//    else if (status & PL180_STAT_DAT_CRC_FAIL)
-//        return -RT_ERROR;
-//    write32(pdat->virt + PL180_CLEAR, 0x1DC007FF);
-//
-//    if(count)
-//        return -RT_ERROR;
+    int c = 0;
+    rt_err_t ret;
+    int d;
+    while (c < blkcount)
+    {
+        if ((ret = sd_int(pdat, INT_WRITE_RDY)))
+        {
+            return ret;
+        }
+        for (d=0; d < blksize / 4; d++)
+            write32(pdat->virt + EMMC_DATA, buf[d]);
+        c++;
+        buf += blksize / 4;
+    }
+    if ((ret = sd_int(pdat, INT_DATA_DONE)))
+    {
+        return ret;
+    }
     return RT_EOK;
 }
 
 static rt_err_t raspi_transfer_data(struct sdhci_pdata_t * pdat, struct sdhci_cmd_t * cmd, struct sdhci_data_t * dat)
 {
     rt_uint32_t dlen = (rt_uint32_t)(dat->blkcnt * dat->blksz);
-    rt_uint32_t blksz_bits = dat->blksz - 1;
-    rt_err_t ret = -RT_ERROR;
-
-    write32(pdat->virt + EMMC_BLKSIZECNT, dlen);
+    rt_err_t ret = sd_status(pdat, SR_DAT_INHIBIT);
+    if (ret)
+    {
+        rt_kprintf("ERROR: EMMC busy\n");
+        return ret;
+    }
+    if (dat->blkcnt > 1)
+    {
+        struct sdhci_cmd_t newcmd;
+        newcmd.cmdidx = SET_BLOCK_COUNT;
+        newcmd.cmdarg = dat->blkcnt;
+        newcmd.resptype = RESP_R1;
+        ret = raspi_transfer_command(pdat, &newcmd);
+        if (ret) return ret;
+    }
 
-    if(dat->flag & DATA_DIR_READ)
+    if(dlen < 512)
     {
-        write32(pdat->virt + EMMC_STATUS, SR_READ_TRANSFER);
+        write32(pdat->virt + EMMC_BLKSIZECNT, dlen | 1 << 16);
+    }
+    else
+    {
+        write32(pdat->virt + EMMC_BLKSIZECNT, 512 | (dat->blkcnt) << 16);
+    }
+    if (dat->flag & DATA_DIR_READ)
+    {
+        cmd->datarw = DATA_READ;
         ret = raspi_transfer_command(pdat, cmd);
-        if (ret < 0) return ret;
+        if (ret) return ret;
+        mmcsd_dbg("read_block %d, %d\n", dat->blkcnt, dat->blksz );
         ret = read_bytes(pdat, (rt_uint32_t *)dat->buf, dat->blkcnt, dat->blksz);
     }
-    else if(dat->flag & DATA_DIR_WRITE)
+    else if (dat->flag & DATA_DIR_WRITE)
     {
+        cmd->datarw = DATA_WRITE;
         ret = raspi_transfer_command(pdat, cmd);
-        if (ret < 0) return ret;
-        write32(pdat->virt + EMMC_STATUS, SR_WRITE_TRANSFER);
+        if (ret) return ret;
+        mmcsd_dbg("write_block %d, %d", dat->blkcnt, dat->blksz );
         ret = write_bytes(pdat, (rt_uint32_t *)dat->buf, dat->blkcnt, dat->blksz);
     }
-
     return ret;
 }
 
+static rt_err_t sdhci_transfer(struct sdhci_t * sdhci, struct sdhci_cmd_t * cmd, struct sdhci_data_t * dat)
+{
+    struct sdhci_pdata_t * pdat = (struct sdhci_pdata_t *)sdhci->priv;
+    if (!dat)
+        return raspi_transfer_command(pdat, cmd);
+
+    return raspi_transfer_data(pdat, cmd, dat);
+}
+
 static void mmc_request_send(struct rt_mmcsd_host *host, struct rt_mmcsd_req *req)
 {
     struct sdhci_t *sdhci = (struct sdhci_t *)host->private_data;
     struct sdhci_cmd_t cmd;
     struct sdhci_cmd_t stop;
     struct sdhci_data_t dat;
-
     rt_memset(&cmd, 0, sizeof(struct sdhci_cmd_t));
     rt_memset(&stop, 0, sizeof(struct sdhci_cmd_t));
     rt_memset(&dat, 0, sizeof(struct sdhci_data_t));
 
     cmd.cmdidx = req->cmd->cmd_code;
-    EMMCCommand* cmdtab = &sdCommandTable[cmd.cmdidx];
     cmd.cmdarg = req->cmd->arg;
-    cmd.resptype = cmdtab->resp;
-
-    if(req->data)
+    cmd.resptype =resp_type(req->cmd);
+    if (req->data)
     {
         dat.buf = (rt_uint8_t *)req->data->buf;
         dat.flag = req->data->flags;
@@ -288,43 +330,14 @@ static void mmc_request_send(struct rt_mmcsd_host *host, struct rt_mmcsd_req *re
     if (req->stop)
     {
         stop.cmdidx = req->stop->cmd_code;
-        cmdtab = &sdCommandTable[cmd.cmdidx];
         stop.cmdarg = req->stop->arg;
-        cmd.resptype = cmdtab->resp;
-
+        cmd.resptype =resp_type(req->stop);
         req->stop->err = sdhci_transfer(sdhci, &stop, RT_NULL);
     }
 
     mmcsd_req_complete(host);
 }
 
-
-static rt_err_t sdhci_transfer(struct sdhci_t * sdhci, struct sdhci_cmd_t * cmd, struct sdhci_data_t * dat)
-{
-    struct sdhci_pdata_t * pdat = (struct sdhci_pdata_t *)sdhci->priv;
-
-    if(!dat)
-        return raspi_transfer_command(pdat, cmd);
-
-    return raspi_transfer_data(pdat, cmd, dat);
-}
-
-
-//#ifdef CONFIG_MMC_USE_DMA
-//#ifdef BSP_USING_SDIO0
-////ALIGN(32) static rt_uint8_t dma_buffer[64 * 1024];
-//static rt_uint8_t dma_buffer[64 * 1024];
-//#endif
-//#endif
-
-
-static void mmc_set_iocfg(struct rt_mmcsd_host *host, struct rt_mmcsd_io_cfg *io_cfg)
-{
-    struct sdhci_t * sdhci = (struct sdhci_t *)host->private_data;
-    sdhci_setclock(sdhci, io_cfg->clock);
-    sdhci_setwidth(sdhci, io_cfg->bus_width);
-}
-
 rt_int32_t mmc_card_status(struct rt_mmcsd_host *host)
 {
     return 0;
@@ -344,15 +357,17 @@ static rt_err_t sdhci_setwidth(struct sdhci_t * sdhci, rt_uint32_t width)
 {
     rt_uint32_t temp = 0;
     struct sdhci_pdata_t * pdat = (struct sdhci_pdata_t *)sdhci->priv;
-    temp = read32((pdat->virt + EMMC_CONTROL0));
-    temp |= C0_HCTL_HS_EN;
-    temp |= C0_HCTL_DWITDH;   // always use 4 data lines:
-    write32((pdat->virt + EMMC_CONTROL0), temp);
-
+    if (width == MMCSD_BUS_WIDTH_4)
+    {
+        temp = read32((pdat->virt + EMMC_CONTROL0));
+        temp |= C0_HCTL_HS_EN;
+        temp |= C0_HCTL_DWITDH;   // always use 4 data lines:
+        write32((pdat->virt + EMMC_CONTROL0), temp);
+    }
     return RT_EOK;
 }
 
-static rt_uint32_t sdhci_getdivider( rt_uint32_t sdHostVer, rt_uint32_t freq )
+static rt_uint32_t sdhci_getdivider(rt_uint32_t sdHostVer, rt_uint32_t freq)
 {
     rt_uint32_t divisor;
     rt_uint32_t closest = 41666666 / freq;
@@ -366,7 +381,8 @@ static rt_uint32_t sdhci_getdivider( rt_uint32_t sdHostVer, rt_uint32_t freq )
     else
         divisor = (1 << shiftcount);
 
-    if (divisor <= 2) {
+    if (divisor <= 2)
+    {
         divisor = 2;
         shiftcount = 0;
     }
@@ -384,56 +400,55 @@ static rt_err_t sdhci_setclock(struct sdhci_t * sdhci, rt_uint32_t clock)
     rt_uint32_t temp = 0;
     rt_uint32_t sdHostVer = 0;
     int count = 100000;
-    struct sdhci_pdata_t * pdat = (struct sdhci_pdata_t *)sdhci->priv;
-
-    temp = read32(pdat->virt + EMMC_STATUS);
-    while((temp & (SR_CMD_INHIBIT | SR_DAT_INHIBIT))&&(--count))
-        bcm283x_clo_delayMicros(1);
+    struct sdhci_pdata_t * pdat = (struct sdhci_pdata_t *)(sdhci->priv);
 
-    if( count <= 0 )
+    while ((read32(pdat->virt + EMMC_STATUS) & (SR_CMD_INHIBIT | SR_DAT_INHIBIT)) && (--count))
+        DELAY_MICROS(1);
+    if (count <= 0)
     {
-        rt_kprintf("EMMC: Set clock: timeout waiting for inhibit flags. Status %08x.\n", temp);
+        rt_kprintf("EMMC: Set clock: timeout waiting for inhibit flags. Status %08x.\n",read32(pdat->virt + EMMC_STATUS));
         return RT_ERROR;
     }
 
     // Switch clock off.
     temp = read32((pdat->virt + EMMC_CONTROL1));
-    temp |= ~C1_CLK_EN;
+    temp &= ~C1_CLK_EN;
     write32((pdat->virt + EMMC_CONTROL1),temp);
-
-    bcm283x_clo_delayMicros(10);
-
+    DELAY_MICROS(10);
     // Request the new clock setting and enable the clock
     temp = read32(pdat->virt + EMMC_SLOTISR_VER);
     sdHostVer = (temp & HOST_SPEC_NUM) >> HOST_SPEC_NUM_SHIFT;
-
     int cdiv = sdhci_getdivider(sdHostVer, clock);
     temp = read32((pdat->virt + EMMC_CONTROL1));
     temp = (temp & 0xffff003f) | cdiv;
     write32((pdat->virt + EMMC_CONTROL1),temp);
-    bcm283x_clo_delayMicros(10);
+    DELAY_MICROS(10);
 
     // Enable the clock.
     temp = read32(pdat->virt + EMMC_CONTROL1);
     temp |= C1_CLK_EN;
     write32((pdat->virt + EMMC_CONTROL1),temp);
-    bcm283x_clo_delayMicros(10);
-
+    DELAY_MICROS(10);
     // Wait for clock to be stable.
     count = 10000;
-    temp = read32(pdat->virt + EMMC_CONTROL1);
-    while( !(temp & C1_CLK_STABLE) && count-- )
-        bcm283x_clo_delayMicros(10);
-
-    if( count <= 0 )
+    while (!(read32(pdat->virt + EMMC_CONTROL1) & C1_CLK_STABLE) && count--)
+        DELAY_MICROS(10);
+    if (count <= 0)
     {
-        rt_kprintf("EMMC: ERROR: failed to get stable clock.\n");
+        rt_kprintf("EMMC: ERROR: failed to get stable clock %d.\n", clock);
         return RT_ERROR;
     }
-
+    mmcsd_dbg("set stable clock %d.\n", clock);
     return RT_EOK;
 }
 
+static void mmc_set_iocfg(struct rt_mmcsd_host *host, struct rt_mmcsd_io_cfg *io_cfg)
+{
+    struct sdhci_t * sdhci = (struct sdhci_t *)host->private_data;
+    sdhci_setclock(sdhci, io_cfg->clock);
+    sdhci_setwidth(sdhci, io_cfg->bus_width);
+}
+
 static const struct rt_mmcsd_host_ops ops =
 {
     mmc_request_send,
@@ -444,27 +459,67 @@ static const struct rt_mmcsd_host_ops ops =
 
 static void sdmmc_gpio_init()
 {
-    int pin;
+//    int pin;
+//    bcm283x_gpio_fsel(47,BCM283X_GPIO_FSEL_INPT);
+//    bcm283x_gpio_set_pud(47, BCM283X_GPIO_PUD_UP);
+//    bcm283x_peri_set_bits(BCM283X_GPIO_BASE + BCM283X_GPIO_GPHEN1, 1<<15, 1<<15);
+//    for (pin = 53; pin >= 48; pin--)
+//    {
+//        bcm283x_gpio_fsel(pin, BCM283X_GPIO_FSEL_ALT3);
+//        bcm283x_gpio_set_pud(pin, BCM283X_GPIO_PUD_UP);
+//    }
+}
 
-    for (pin = BCM_GPIO_PIN_48; pin <= BCM_GPIO_PIN_53; pin++)
+static rt_err_t reset_emmc(struct sdhci_pdata_t * pdat){
+    rt_uint32_t temp;
+    int cnt;
+    write32((pdat->virt + EMMC_CONTROL0),0);
+    temp = read32((pdat->virt + EMMC_CONTROL1));
+    temp |= C1_SRST_HC;
+    write32((pdat->virt + EMMC_CONTROL1),temp);
+    cnt = 10000;
+    do
     {
-        bcm283x_gpio_set_pud(pin, BCM283X_GPIO_PUD_UP);
-        bcm283x_gpio_fsel(pin, BCM283X_GPIO_FSEL_ALT3);
+        DELAY_MICROS(10);
     }
-    bcm283x_gpio_set_pud(pin, BCM283X_GPIO_PUD_UP);
-    bcm283x_gpio_fsel(pin, BCM283X_GPIO_FSEL_INPT);
+    while ((read32((pdat->virt + EMMC_CONTROL1)) & C1_SRST_HC) && cnt--);
+    if (cnt <= 0)
+    {
+        rt_kprintf("ERROR: failed to reset EMMC\n");
+        return RT_ERROR;
+    }
+    temp = read32((pdat->virt + EMMC_CONTROL1));
+    temp |= C1_CLK_INTLEN | C1_TOUNIT_MAX;
+    write32((pdat->virt + EMMC_CONTROL1),temp);
+    DELAY_MICROS(10);
+    return RT_EOK;
 }
 
+#ifdef RT_MMCSD_DBG
+void dump_registers(struct sdhci_pdata_t * pdat){
+    rt_kprintf("EMMC registers:");
+    int i = EMMC_ARG2;
+    for (; i <= EMMC_CONTROL2; i += 4)
+        rt_kprintf("\t%x:%x\n", i, read32(pdat->virt + i));
+    rt_kprintf("\t%x:%x\n", 0x50, read32(pdat->virt + 0x50));
+    rt_kprintf("\t%x:%x\n", 0x70, read32(pdat->virt + 0x70));
+    rt_kprintf("\t%x:%x\n", 0x74, read32(pdat->virt + 0x74));
+    rt_kprintf("\t%x:%x\n", 0x80, read32(pdat->virt + 0x80));
+    rt_kprintf("\t%x:%x\n", 0x84, read32(pdat->virt + 0x84));
+    rt_kprintf("\t%x:%x\n", 0x88, read32(pdat->virt + 0x88));
+    rt_kprintf("\t%x:%x\n", 0x8c, read32(pdat->virt + 0x8c));
+    rt_kprintf("\t%x:%x\n", 0x90, read32(pdat->virt + 0x90));
+    rt_kprintf("\t%x:%x\n", 0xf0, read32(pdat->virt + 0xf0));
+    rt_kprintf("\t%x:%x\n", 0xfc, read32(pdat->virt + 0xfc));
+}
+#endif
+
 int raspi_sdmmc_init(void)
 {
     rt_uint32_t virt;
-    rt_uint32_t id;
     struct rt_mmcsd_host * host = RT_NULL;
     struct sdhci_pdata_t * pdat = RT_NULL;
     struct sdhci_t * sdhci = RT_NULL;
-
-    rt_kprintf("raspi_sdmmc_init start\n");
-
 #ifdef BSP_USING_SDIO0
     host = mmcsd_alloc_host();
     if (!host)
@@ -481,21 +536,19 @@ int raspi_sdmmc_init(void)
     }
     rt_memset(sdhci, 0, sizeof(struct sdhci_t));
 
-    rt_kprintf(">> sdmmc_gpio_init\n");
     sdmmc_gpio_init();
-    rt_kprintf("<< sdmmc_gpio_init\n");
-
     virt = MMC0_BASE_ADDR;
 
     pdat = (struct sdhci_pdata_t *)rt_malloc(sizeof(struct sdhci_pdata_t));
     RT_ASSERT(pdat != RT_NULL);
 
     pdat->virt = (rt_uint32_t)virt;
+    reset_emmc(pdat);
 
     sdhci->name = "sd0";
     sdhci->voltages = VDD_33_34;
     sdhci->width = MMCSD_BUSWIDTH_4;
-    sdhci->clock = 26 * 1000 * 1000;
+    sdhci->clock = 200 * 1000 * 1000;
     sdhci->removeable = RT_TRUE;
 
     sdhci->detect = sdhci_detect;
@@ -503,7 +556,6 @@ int raspi_sdmmc_init(void)
     sdhci->setclock = sdhci_setclock;
     sdhci->transfer = sdhci_transfer;
     sdhci->priv = pdat;
-    //write32(pdat->virt + PL180_POWER, 0xbf);
 
     host->ops = &ops;
     host->freq_min = 400000;
@@ -517,16 +569,20 @@ int raspi_sdmmc_init(void)
 
     host->private_data = sdhci;
 
+    write32((pdat->virt + EMMC_IRPT_EN),0xffffffff);
+    write32((pdat->virt + EMMC_IRPT_MASK),0xffffffff);
+#ifdef RT_MMCSD_DBG
+    dump_registers(pdat);
+#endif
     mmcsd_change(host);
-
+#endif
     return RT_EOK;
 
 err:
-    if(host)  rt_free(host);
-    if(sdhci) rt_free(sdhci);
+    if (host)  rt_free(host);
+    if (sdhci) rt_free(sdhci);
 
     return -RT_EIO;
-#endif
 }
-INIT_APP_EXPORT(raspi_sdmmc_init);
-#endif
+
+INIT_DEVICE_EXPORT(raspi_sdmmc_init);
diff --git a/bsp/raspberry-pi/raspi3-64/driver/drv_sdio.h b/bsp/raspberry-pi/raspi3-64/driver/drv_sdio.h
index 8ab10aebb6..5c052de90a 100644
--- a/bsp/raspberry-pi/raspi3-64/driver/drv_sdio.h
+++ b/bsp/raspberry-pi/raspi3-64/driver/drv_sdio.h
@@ -13,50 +13,12 @@
 #define __DRV_SDIO_H__
 
 #include <rtthread.h>
+#include <rtdevice.h>
+#include <drivers/mmcsd_core.h>
 
-#define MMC0_BASE_ADDR     0x20300000
-
-struct raspi_mmc
-{
-    volatile rt_uint32_t arg2_reg;            /* (0x000) */
-    volatile rt_uint32_t blksizecnt_reg;      /* (0x004) */
-    volatile rt_uint32_t arg1_reg;            /* (0x008) */
-    volatile rt_uint32_t cmdtm_reg;           /* (0x00C) */
-    volatile rt_uint32_t resp0_reg;            /* (0x010) */
-    volatile rt_uint32_t resp1_reg;            /* (0x014) */
-    volatile rt_uint32_t resp2_reg;            /* (0x018) */
-    volatile rt_uint32_t resp3_reg;            /* (0x01C) */
-    volatile rt_uint32_t data_reg;           /* (0x020) */
-    volatile rt_uint32_t status_reg;           /* (0x024) */
-    volatile rt_uint32_t control0_reg;           /* (0x028) */
-    volatile rt_uint32_t control1_reg;           /* (0x02C) */
-    volatile rt_uint32_t interrupt_reg;            /* (0x030) */
-    volatile rt_uint32_t irpt_mask_reg;            /* (0x034) */
-    volatile rt_uint32_t irpt_en_reg;            /* (0x038) */
-    volatile rt_uint32_t control2_reg;            /* (0x03C) */
-    volatile rt_uint32_t reserved1[4];            /* (0x040) */
-    volatile rt_uint32_t force_irpt_reg;            /* (0x050) */
-    volatile rt_uint32_t reserved2[7];            /* (0x054) */
-    volatile rt_uint32_t boot_timeout_reg;            /* (0x070) */
-    volatile rt_uint32_t deg_sel_reg;            /* (0x074) */
-    volatile rt_uint32_t reserved3[2];            /* (0x078) */
-    volatile rt_uint32_t exrdfifo_cfg_reg;      /* (0x080) */
-    volatile rt_uint32_t exrdfifo_cn_reg;       /* (0x084) */
-    volatile rt_uint32_t tune_step_reg;        /* (0x088) */
-    volatile rt_uint32_t tune_step_std_reg;    /* (0x08C) */
-    volatile rt_uint32_t tune_step_ddr_reg;       /* (0x090) */
-    volatile rt_uint32_t reserved4[23];            /* (0x094) */
-    volatile rt_uint32_t spi_int_reg;            /* (0x0f0) */
-    volatile rt_uint32_t reserved5[2];            /* (0x0f4) */
-    volatile rt_uint32_t slotisr_ver_reg;            /* (0x0fC) */
-};
-
-typedef struct raspi_mmc *raspi_mmc_t;
-
-#define MMC0    ((tina_mmc_t)MMC0_BASE_ADDR)
+#include "board.h"
 
-
-#define BIT(x)              (1<<(x))
+#define MMC0_BASE_ADDR     0x3F300000
 
 /* Struct for Intrrrupt Information */
 #define SDXC_CmdDone       BIT(0)
@@ -78,25 +40,6 @@ typedef struct raspi_mmc *raspi_mmc_t;
 #define SDXC_DENDErr       BIT(22)
 #define SDXC_ACMDErr       BIT(24)
 
-
-/*
-    SD CMD reg
-REG[0-5] : Cmd ID
-REG[6]   : Has response
-REG[7]   : Long response
-REG[8]   : Check response CRC
-REG[9]   : Has data
-REG[10]  : Write
-REG[11]  : Steam mode
-REG[12]  : Auto stop
-REG[13]  : Wait previous over
-REG[14]  : About cmd
-REG[15]  : Send initialization
-REG[21]  : Update clock
-REG[31]  : Load cmd
-*/
-
-
 #define SDXC_BLKCNT_EN          BIT(1)
 #define SDXC_AUTO_CMD12_EN      BIT(2)
 #define SDXC_AUTO_CMD23_EN      BIT(3)
@@ -112,82 +55,20 @@ REG[31]  : Load cmd
 #define SDXC_CMD_RESUME         BIT(23)
 #define SDXC_CMD_ABORT          BIT(23)|BIT(22)
 
-//#define SDXC_CHECK_CRC_CMD      BIT(19)
-//
-//#define SDXC_RESPONSE_CMD        BIT(6)
-//#define SDXC_LONG_RESPONSE_CMD   BIT(7)
-//#define SDXC_CHECK_CRC_CMD       BIT(8)
-//#define SDXC_HAS_DATA_CMD        BIT(9)
-//#define SDXC_WRITE_CMD           BIT(10)
-//#define SDXC_STEAM_CMD           BIT(11)
-//#define SDXC_AUTO_STOP_CMD       BIT(12)
-//#define SDXC_WAIT_OVER_CMD       BIT(13)
-//#define SDXC_ABOUT_CMD           BIT(14)
-//#define SDXC_SEND_INIT_CMD       BIT(15)
-//#define SDXC_UPDATE_CLOCK_CMD    BIT(21)
-//#define SDXC_LOAD_CMD            BIT(31)
-
-/* 
-    SD status reg 
-REG[0]   : FIFO_RX_LEVEL
-REG[1]   : FIFO_TX_LEVEL
-REG[2]   : FIFO_EMPTY
-REG[3]   : FIFO_FULL
-REG[4-7] : FSM_STA
-REG[8]   : CARD_PRESENT
-REG[9]   : CARD_BUSY
-REG[10]  : FSM_BUSY
-REG[11-16]: RESP_IDX 
-REG[17-21]: FIFO_LEVEL
-REG[31]   : DMA_REQ
-*/
-
 #define SDXC_CMD_INHIBIT    BIT(0)
 #define SDXC_DAT_INHIBIT    BIT(1)
 #define SDXC_DAT_ACTIVE     BIT(2)
 #define SDXC_WRITE_TRANSFER     BIT(8)
 #define SDXC_READ_TRANSFER     BIT(9)
 
-//
-//
-//#define SDXC_FIFO_RX_LEVEL    BIT(0)
-//#define SDXC_FIFO_TX_LEVEL    BIT(1)
-//#define SDXC_FIFO_EMPTY       BIT(2)
-//#define SDXC_FIFO_FULL        BIT(3)
-//#define SDXC_CARD_PRESENT     BIT(8)
-//#define SDXC_CARD_BUSY        BIT(9)
-//#define SDXC_FSM_BUSY         BIT(10)
-//#define SDXC_DMA_REQ          BIT(31)
-
-struct mmc_des_v4p1
-{
-    rt_uint32_t                         : 1,
-                    dic                 : 1, /* disable interrupt on completion                                    */
-                    last_des            : 1, /* 1-this data buffer is the last buffer                              */
-                    first_des           : 1, /* 1-data buffer is the first buffer,0-data buffer contained in the next descriptor is 1st buffer        */
-                    des_chain           : 1, /* 1-the 2nd address in the descriptor is the next descriptor address */
-                    end_of_ring         : 1, /* 1-last descriptor flag when using dual data buffer in descriptor   */
-                                        : 24,
-                    card_err_sum        : 1, /* transfer error flag                                                */
-                    own                 : 1; /* des owner:1-idma owns it, 0-host owns it                           */
-
-#define SDXC_DES_NUM_SHIFT 12  /* smhc2!! */
-#define SDXC_DES_BUFFER_MAX_LEN (1 << SDXC_DES_NUM_SHIFT)
-    rt_uint32_t data_buf1_sz     : 16,
-                data_buf2_sz     : 16;
-    rt_uint32_t buf_addr_ptr1;
-    rt_uint32_t buf_addr_ptr2;
-};
-
 struct sdhci_cmd_t
 {
     rt_uint32_t cmdidx;
     rt_uint32_t cmdarg;
-    //const char* name;
-   // rt_uint32_t code;
     rt_uint32_t resptype;
-  //  rt_uint8_t rca;
-  //  rt_uint32_t delay;
+    rt_uint32_t datarw;
+#define DATA_READ 1
+#define DATA_WRITE 2
     rt_uint32_t response[4];
 };
 
@@ -240,17 +121,6 @@ struct sdhci_pdata_t
 #define TM_BLKCNT_EN     0x00000002
 #define TM_MULTI_DATA    (CMD_IS_DATA|TM_MULTI_BLOCK|TM_BLKCNT_EN)
 
-// Response types.
-// Note that on the PI, the index and CRC are dropped, leaving 32 bits in RESP0.
-#define RESP_NO    0     // No response
-//#define RESP_R1    1     // 48  RESP0    contains card status
-#define RESP_R1b  11     // 48  RESP0    contains card status, data line indicates busy
-#define RESP_R2I   2     // 136 RESP0..3 contains 128 bit CID shifted down by 8 bits as no CRC
-#define RESP_R2S  12     // 136 RESP0..3 contains 128 bit CSD shifted down by 8 bits as no CRC
-//#define RESP_R3    3     // 48  RESP0    contains OCR register
-//#define RESP_R6    6     // 48  RESP0    contains RCA and status bits 23,22,19,12:0
-//#define RESP_R7    7     // 48  RESP0    contains voltage acceptance and check pattern
-
 #define RCA_NO     1
 #define RCA_YES    2
 
@@ -341,5 +211,43 @@ struct sdhci_pdata_t
 #define SR_DAT_INHIBIT       0x00000002
 #define SR_CMD_INHIBIT       0x00000001
 
+#define CONFIG_MMC_USE_DMA
+#define DMA_ALIGN       (32U)
 
+#define SD_CMD_INDEX(a)        ((a) << 24)
+#define SD_CMD_RESERVED(a)  0xffffffff
+#define SD_CMD_INDEX(a)        ((a) << 24)
+#define SD_CMD_TYPE_NORMAL    0x0
+#define SD_CMD_TYPE_SUSPEND    (1 << 22)
+#define SD_CMD_TYPE_RESUME    (2 << 22)
+#define SD_CMD_TYPE_ABORT    (3 << 22)
+#define SD_CMD_TYPE_MASK    (3 << 22)
+#define SD_CMD_ISDATA        (1 << 21)
+#define SD_CMD_IXCHK_EN        (1 << 20)
+#define SD_CMD_CRCCHK_EN    (1 << 19)
+#define SD_CMD_RSPNS_TYPE_NONE    0            // For no response
+#define SD_CMD_RSPNS_TYPE_136    (1 << 16)        // For response R2 (with CRC), R3,4 (no CRC)
+#define SD_CMD_RSPNS_TYPE_48    (2 << 16)        // For responses R1, R5, R6, R7 (with CRC)
+#define SD_CMD_RSPNS_TYPE_48B    (3 << 16)        // For responses R1b, R5b (with CRC)
+#define SD_CMD_RSPNS_TYPE_MASK  (3 << 16)
+#define SD_CMD_MULTI_BLOCK    (1 << 5)
+#define SD_CMD_DAT_DIR_HC    0
+#define SD_CMD_DAT_DIR_CH    (1 << 4)
+#define SD_CMD_AUTO_CMD_EN_NONE    0
+#define SD_CMD_AUTO_CMD_EN_CMD12    (1 << 2)
+#define SD_CMD_AUTO_CMD_EN_CMD23    (2 << 2)
+#define SD_CMD_BLKCNT_EN        (1 << 1)
+#define SD_CMD_DMA          1
+#define SD_RESP_NONE        SD_CMD_RSPNS_TYPE_NONE
+#define SD_RESP_R1          (SD_CMD_RSPNS_TYPE_48)           // | SD_CMD_CRCCHK_EN)
+#define SD_RESP_R1b         (SD_CMD_RSPNS_TYPE_48B)         // | SD_CMD_CRCCHK_EN)
+#define SD_RESP_R2          (SD_CMD_RSPNS_TYPE_136)           //| SD_CMD_CRCCHK_EN)
+#define SD_RESP_R3          SD_CMD_RSPNS_TYPE_48
+#define SD_RESP_R4          SD_CMD_RSPNS_TYPE_136
+#define SD_RESP_R5          (SD_CMD_RSPNS_TYPE_48 | SD_CMD_CRCCHK_EN)
+#define SD_RESP_R5b         (SD_CMD_RSPNS_TYPE_48B | SD_CMD_CRCCHK_EN)
+#define SD_RESP_R6          (SD_CMD_RSPNS_TYPE_48 | SD_CMD_CRCCHK_EN)
+#define SD_RESP_R7          (SD_CMD_RSPNS_TYPE_48 | SD_CMD_CRCCHK_EN)
+#define SD_DATA_READ        (SD_CMD_ISDATA | SD_CMD_DAT_DIR_CH)
+#define SD_DATA_WRITE       (SD_CMD_ISDATA | SD_CMD_DAT_DIR_HC)
 #endif
diff --git a/bsp/raspberry-pi/raspi3-64/driver/raspi.h b/bsp/raspberry-pi/raspi3-64/driver/raspi.h
index 0d55df2a11..001d1b7962 100644
--- a/bsp/raspberry-pi/raspi3-64/driver/raspi.h
+++ b/bsp/raspberry-pi/raspi3-64/driver/raspi.h
@@ -67,4 +67,10 @@ enum rpi_pin_name
     RPI_GPIO_PIN_NUM,
 };
 
+#define DELAY_MICROS(micros) \
+    do{ \
+ rt_uint32_t compare = STIMER_CLO + micros * 25; \
+ while (STIMER_CLO < compare); \
+    } while (0) \
+
 #endif
diff --git a/bsp/raspberry-pi/raspi3-64/qemu-64.bat b/bsp/raspberry-pi/raspi3-64/qemu-64.bat
index c009905e23..db14a102df 100644..100755
--- a/bsp/raspberry-pi/raspi3-64/qemu-64.bat
+++ b/bsp/raspberry-pi/raspi3-64/qemu-64.bat
@@ -1 +1,4 @@
-qemu-system-aarch64 -M raspi3 -kernel kernel8.img -serial null -serial stdio
\ No newline at end of file
+if [ ! -f "sd.bin" ]; then
+dd if=/dev/zero of=sd.bin bs=1024 count=65536
+fi
+qemu-system-aarch64 -M raspi3 -kernel kernel8.img -serial null -serial stdio -sd sd.bin -nographic -monitor pty
\ No newline at end of file
diff --git a/bsp/raspberry-pi/raspi3-64/rtconfig.h b/bsp/raspberry-pi/raspi3-64/rtconfig.h
index 18210fc276..0e569b52ef 100644
--- a/bsp/raspberry-pi/raspi3-64/rtconfig.h
+++ b/bsp/raspberry-pi/raspi3-64/rtconfig.h
@@ -43,7 +43,7 @@
 #define RT_USING_DEVICE
 #define RT_USING_DEVICE_OPS
 #define RT_USING_CONSOLE
-#define RT_CONSOLEBUF_SIZE 128
+#define RT_CONSOLEBUF_SIZE 512
 #define RT_CONSOLE_DEVICE_NAME "uart1"
 #define RT_VER_NUM 0x40003
 #define ARCH_CPU_64BIT
@@ -52,7 +52,7 @@
 
 #define RT_USING_COMPONENTS_INIT
 #define RT_USING_USER_MAIN
-#define RT_MAIN_THREAD_STACK_SIZE 2048
+#define RT_MAIN_THREAD_STACK_SIZE 4096
 #define RT_MAIN_THREAD_PRIORITY 10
 
 /* C++ features */
@@ -81,6 +81,18 @@
 #define DFS_FILESYSTEMS_MAX 2
 #define DFS_FILESYSTEM_TYPES_MAX 2
 #define DFS_FD_MAX 16
+#define RT_USING_DFS_ELMFAT
+
+/* elm-chan's FatFs, Generic FAT Filesystem Module */
+
+#define RT_DFS_ELM_CODE_PAGE 437
+#define RT_DFS_ELM_WORD_ACCESS
+#define RT_DFS_ELM_USE_LFN_3
+#define RT_DFS_ELM_USE_LFN 3
+#define RT_DFS_ELM_MAX_LFN 255
+#define RT_DFS_ELM_DRIVES 2
+#define RT_DFS_ELM_MAX_SECTOR_SIZE 512
+#define RT_DFS_ELM_REENTRANT
 #define RT_USING_DFS_DEVFS
 
 /* Device Drivers */
@@ -90,6 +102,13 @@
 #define RT_USING_SERIAL
 #define RT_SERIAL_RB_BUFSZ 64
 #define RT_USING_PIN
+#define RT_USING_SDIO
+#define RT_SDIO_STACK_SIZE 2048
+#define RT_SDIO_THREAD_PRIORITY 15
+#define RT_MMCSD_STACK_SIZE 4096
+#define RT_MMCSD_THREAD_PREORITY 22
+#define RT_MMCSD_MAX_PARTITION 16
+#define RT_SDIO_DEBUG
 
 /* Using USB */
 
@@ -174,6 +193,8 @@
 #define RT_USING_UART1
 #define BSP_USING_PIN
 #define BSP_USING_CORETIMER
+#define BSP_USING_SDIO
+#define BSP_USING_SDIO0
 
 /* Board Peripheral Drivers */
 
diff --git a/components/drivers/sdio/mmcsd_core.c b/components/drivers/sdio/mmcsd_core.c
index 8131a33133..99bc6ce744 100644
--- a/components/drivers/sdio/mmcsd_core.c
+++ b/components/drivers/sdio/mmcsd_core.c
@@ -610,7 +610,7 @@ RTM_EXPORT(mmcsd_wait_cd_changed);
 
 void mmcsd_change(struct rt_mmcsd_host *host)
 {
-    rt_mb_send(&mmcsd_detect_mb, (rt_uint32_t)host);
+    rt_mb_send(&mmcsd_detect_mb, (rt_ubase_t)host);
 }
 
 void mmcsd_detect(void *param)
@@ -649,7 +649,7 @@ void mmcsd_detect(void *param)
                     if (init_sd(host, ocr))
                         mmcsd_power_off(host);
                     mmcsd_host_unlock(host);
-                    rt_mb_send(&mmcsd_hotpluge_mb, (rt_uint32_t)host);
+                    rt_mb_send(&mmcsd_hotpluge_mb, (rt_ubase_t)host);
                     continue;
                 }
                 
@@ -662,7 +662,7 @@ void mmcsd_detect(void *param)
                     if (init_mmc(host, ocr))
                         mmcsd_power_off(host);
                     mmcsd_host_unlock(host);
-                    rt_mb_send(&mmcsd_hotpluge_mb, (rt_uint32_t)host);
+                    rt_mb_send(&mmcsd_hotpluge_mb, (rt_ubase_t)host);
                     continue;
                 }
                 mmcsd_host_unlock(host);
@@ -683,7 +683,7 @@ void mmcsd_detect(void *param)
             		host->card = RT_NULL;
             	}
             	mmcsd_host_unlock(host);
-            	rt_mb_send(&mmcsd_hotpluge_mb, (rt_uint32_t)host);
+            	rt_mb_send(&mmcsd_hotpluge_mb, (rt_ubase_t)host);
             }
         }
     }