]> git.sur5r.net Git - openocd/blobdiff - src/target/cortex_swjdp.c
Change tap_state naming to be consistent with SVF documentation.
[openocd] / src / target / cortex_swjdp.c
index 160f4c53f5e81b7d918cf570719c82f6de0ebed9..974ced02284e18663b80a70acbb152fddedd7582 100644 (file)
@@ -2,6 +2,9 @@
  *   Copyright (C) 2006 by Magnus Lundin                                   *
  *   lundin@mlu.mine.nu                                                    *
  *                                                                         *
+ *   Copyright (C) 2008 by Spencer Oliver                                  *
+ *   spen@spen-soft.co.uk                                                  *
+ *                                                                         *
  *   This program is free software; you can redistribute it and/or modify  *
  *   it under the terms of the GNU General Public License as published by  *
  *   the Free Software Foundation; either version 2 of the License, or     *
@@ -21,8 +24,8 @@
  *                                                                         *
  * CoreSight (Light?) SerialWireJtagDebugPort                              *
  *                                                                         *
- * CoreSight™ DAP-Lite TRM, ARM DDI 0316A                                  *
- * Cortex-M3™ TRM, ARM DDI 0337C                                            *
+ * CoreSight(tm) DAP-Lite TRM, ARM DDI 0316A                                *
+ * Cortex-M3(tm) TRM, ARM DDI 0337C                                         *
  *                                                                         *
 ***************************************************************************/
 #ifdef HAVE_CONFIG_H
 #include "cortex_swjdp.h"
 #include "jtag.h"
 #include "log.h"
+#include "time_support.h"
 #include <stdlib.h>
 
 /*
-
-Transaction Mode:
-swjdp->trans_mode = TRANS_MODE_COMPOSITE;
-Uses Overrun checking mode and does not do actual JTAG send/receive or transaction 
-result checking until swjdp_end_transaction()
-This must be done before using or deallocating any return variables.
-
-swjdp->trans_mode == TRANS_MODE_ATOMIC
-All reads and writes to the AHB bus are checked for valid completion, and return values
-are immediatley available.
-
+ * Transaction Mode:
+ * swjdp->trans_mode = TRANS_MODE_COMPOSITE;
+ * Uses Overrun checking mode and does not do actual JTAG send/receive or transaction
+ * result checking until swjdp_end_transaction()
+ * This must be done before using or deallocating any return variables.
+ * swjdp->trans_mode == TRANS_MODE_ATOMIC
+ * All reads and writes to the AHB bus are checked for valid completion, and return values
+ * are immediatley available.
 */
 
 /***************************************************************************
@@ -62,11 +63,11 @@ int swjdp_scan(arm_jtag_t *jtag_info, u8 instr, u8 reg_addr, u8 RnW, u8 *outvalu
 {
        scan_field_t fields[2];
        u8 out_addr_buf;
-       
-       jtag_add_end_state(TAP_RTI);
+
+       jtag_add_end_state(TAP_IDLE);
        arm_jtag_set_instr(jtag_info, instr, NULL);
 
-       fields[0].device = jtag_info->chain_pos;
+       fields[0].tap = jtag_info->tap;
        fields[0].num_bits = 3;
        buf_set_u32(&out_addr_buf, 0, 3, ((reg_addr >> 1) & 0x6) | (RnW & 0x1));
        fields[0].out_value = &out_addr_buf;
@@ -77,7 +78,7 @@ int swjdp_scan(arm_jtag_t *jtag_info, u8 instr, u8 reg_addr, u8 RnW, u8 *outvalu
        fields[0].in_handler = NULL;
        fields[0].in_handler_priv = NULL;
 
-       fields[1].device = jtag_info->chain_pos;
+       fields[1].tap = jtag_info->tap;
        fields[1].num_bits = 32;
        fields[1].out_value = outvalue;
        fields[1].out_mask = NULL;
@@ -98,11 +99,11 @@ int swjdp_scan_u32(arm_jtag_t *jtag_info, u8 instr, u8 reg_addr, u8 RnW, u32 out
        scan_field_t fields[2];
        u8 out_value_buf[4];
        u8 out_addr_buf;
-       
-       jtag_add_end_state(TAP_RTI);
+
+       jtag_add_end_state(TAP_IDLE);
        arm_jtag_set_instr(jtag_info, instr, NULL);
 
-       fields[0].device = jtag_info->chain_pos;
+       fields[0].tap = jtag_info->tap;
        fields[0].num_bits = 3;
        buf_set_u32(&out_addr_buf, 0, 3, ((reg_addr >> 1) & 0x6) | (RnW & 0x1));
        fields[0].out_value = &out_addr_buf;
@@ -113,7 +114,7 @@ int swjdp_scan_u32(arm_jtag_t *jtag_info, u8 instr, u8 reg_addr, u8 RnW, u32 out
        fields[0].in_handler = NULL;
        fields[0].in_handler_priv = NULL;
 
-       fields[1].device = jtag_info->chain_pos;
+       fields[1].tap = jtag_info->tap;
        fields[1].num_bits = 32;
        buf_set_u32(out_value_buf, 0, 32, outvalue);
        fields[1].out_value = out_value_buf;
@@ -137,15 +138,15 @@ int swjdp_scan_u32(arm_jtag_t *jtag_info, u8 instr, u8 reg_addr, u8 RnW, u32 out
        return ERROR_OK;
 }
 
-/* scan_inout_check adds one extra inscan for DPAP_READ commands to read variables */ 
+/* scan_inout_check adds one extra inscan for DPAP_READ commands to read variables */
 int scan_inout_check(swjdp_common_t *swjdp, u8 instr, u8 reg_addr, u8 RnW, u8 *outvalue, u8 *invalue)
 {
        swjdp_scan(swjdp->jtag_info, instr, reg_addr, RnW, outvalue, NULL, NULL);
        if ((RnW == DPAP_READ) && (invalue != NULL))
        {
-               swjdp_scan(swjdp->jtag_info, SWJDP_IR_DPACC, 0xC, DPAP_READ, 0, invalue, &swjdp->ack);
+               swjdp_scan(swjdp->jtag_info, SWJDP_IR_DPACC, DP_RDBUFF, DPAP_READ, 0, invalue, &swjdp->ack);
        }
-       
+
        /* In TRANS_MODE_ATOMIC all SWJDP_IR_APACC transactions wait for ack=OK/FAULT and the check CTRL_STAT */
        if ((instr == SWJDP_IR_APACC) && (swjdp->trans_mode == TRANS_MODE_ATOMIC))
        {
@@ -157,13 +158,12 @@ int scan_inout_check(swjdp_common_t *swjdp, u8 instr, u8 reg_addr, u8 RnW, u8 *o
 
 int scan_inout_check_u32(swjdp_common_t *swjdp, u8 instr, u8 reg_addr, u8 RnW, u32 outvalue, u32 *invalue)
 {
-
        swjdp_scan_u32(swjdp->jtag_info, instr, reg_addr, RnW, outvalue, NULL, NULL);
        if ((RnW==DPAP_READ) && (invalue != NULL))
        {
-               swjdp_scan_u32(swjdp->jtag_info, SWJDP_IR_DPACC, 0xC, DPAP_READ, 0, invalue, &swjdp->ack);
+               swjdp_scan_u32(swjdp->jtag_info, SWJDP_IR_DPACC, DP_RDBUFF, DPAP_READ, 0, invalue, &swjdp->ack);
        }
-       
+
        /* In TRANS_MODE_ATOMIC all SWJDP_IR_APACC transactions wait for ack=OK/FAULT and then check CTRL_STAT */
        if ((instr == SWJDP_IR_APACC) && (swjdp->trans_mode == TRANS_MODE_ATOMIC))
        {
@@ -175,42 +175,57 @@ int scan_inout_check_u32(swjdp_common_t *swjdp, u8 instr, u8 reg_addr, u8 RnW, u
 
 int swjdp_transaction_endcheck(swjdp_common_t *swjdp)
 {
-       int waitcount = 0;
+       int retval;
        u32 ctrlstat;
 
+       keep_alive();
+
+       /* Danger!!!! BROKEN!!!! */
        scan_inout_check_u32(swjdp, SWJDP_IR_DPACC, DP_CTRL_STAT, DPAP_READ, 0, &ctrlstat);
+       /* Danger!!!! BROKEN!!!! Why will jtag_execute_queue() fail here????
+       R956 introduced the check on return value here and now Michael Schwingen reports
+       that this code no longer works....
+
+       https://lists.berlios.de/pipermail/openocd-development/2008-September/003107.html
+       */
+       if ((retval=jtag_execute_queue())!=ERROR_OK)
+       {
+               LOG_ERROR("BUG: Why does this fail the first time????");
+       }
+       /* Why??? second time it works??? */
        scan_inout_check_u32(swjdp, SWJDP_IR_DPACC, DP_CTRL_STAT, DPAP_READ, 0, &ctrlstat);
-       jtag_execute_queue();
-       
+       if ((retval=jtag_execute_queue())!=ERROR_OK)
+               return retval;
+
        swjdp->ack = swjdp->ack & 0x7;
-       
+
+       long long then=timeval_ms();
        while (swjdp->ack != 2)
        {
                if (swjdp->ack == 1)
                {
-                       waitcount++;
-                       if (waitcount > 100)
+                       if ((timeval_ms()-then) > 1000)
                        {
-                               WARNING("Timeout waiting for ACK = OK/FAULT in SWJDP transaction");
-
+                               LOG_WARNING("Timeout (1000ms) waiting for ACK = OK/FAULT in SWJDP transaction");
                                return ERROR_JTAG_DEVICE_ERROR;
                        }
                }
                else
                {
-                       WARNING("Invalid ACK in SWJDP transaction");
+                       LOG_WARNING("Invalid ACK in SWJDP transaction");
                        return ERROR_JTAG_DEVICE_ERROR;
                }
 
                scan_inout_check_u32(swjdp, SWJDP_IR_DPACC, DP_CTRL_STAT, DPAP_READ, 0, &ctrlstat);
-               jtag_execute_queue();
+               if ((retval=jtag_execute_queue())!=ERROR_OK)
+                       return retval;
                swjdp->ack = swjdp->ack & 0x7;
        }
 
        /* Check for STICKYERR and STICKYORUN */
        if (ctrlstat & (SSTICKYORUN | SSTICKYERR))
        {
-               DEBUG("swjdp: CTRL/STAT error 0x%x", ctrlstat);
+               LOG_DEBUG("swjdp: CTRL/STAT error 0x%x", ctrlstat);
                /* Check power to debug regions */
                if ((ctrlstat & 0xf0000000) != 0xf0000000)
                {
@@ -219,28 +234,30 @@ int swjdp_transaction_endcheck(swjdp_common_t *swjdp)
                else
                {
                        u32 dcb_dhcsr,nvic_shcsr, nvic_bfar, nvic_cfsr;
-                       
+
                        if (ctrlstat & SSTICKYORUN)
-                               ERROR("SWJ-DP OVERRUN - check clock or reduce jtag speed");
-                       
+                               LOG_ERROR("SWJ-DP OVERRUN - check clock or reduce jtag speed");
+
                        if (ctrlstat & SSTICKYERR)
-                               ERROR("SWJ-DP STICKY ERROR");
-                       
+                               LOG_ERROR("SWJ-DP STICKY ERROR");
+
                        /* Clear Sticky Error Bits */
                        scan_inout_check_u32(swjdp, SWJDP_IR_DPACC, DP_CTRL_STAT, DPAP_WRITE, swjdp->dp_ctrl_stat | SSTICKYORUN | SSTICKYERR, NULL);
                        scan_inout_check_u32(swjdp, SWJDP_IR_DPACC, DP_CTRL_STAT, DPAP_READ, 0, &ctrlstat);
-                       jtag_execute_queue();
+                       if ((retval=jtag_execute_queue())!=ERROR_OK)
+                               return retval;
 
-                       DEBUG("swjdp: status 0x%x", ctrlstat);
-                       
-                       /* Can we find out the reason for the error ?? */                       
+                       LOG_DEBUG("swjdp: status 0x%x", ctrlstat);
+
+                       /* Can we find out the reason for the error ?? */
                        ahbap_read_system_atomic_u32(swjdp, DCB_DHCSR, &dcb_dhcsr);
                        ahbap_read_system_atomic_u32(swjdp, NVIC_SHCSR, &nvic_shcsr);
                        ahbap_read_system_atomic_u32(swjdp, NVIC_CFSR, &nvic_cfsr);
                        ahbap_read_system_atomic_u32(swjdp, NVIC_BFAR, &nvic_bfar);
-                       ERROR("dcb_dhcsr 0x%x, nvic_shcsr 0x%x, nvic_cfsr 0x%x, nvic_bfar 0x%x", dcb_dhcsr, nvic_shcsr, nvic_cfsr, nvic_bfar);
+                       LOG_ERROR("dcb_dhcsr 0x%x, nvic_shcsr 0x%x, nvic_cfsr 0x%x, nvic_bfar 0x%x", dcb_dhcsr, nvic_shcsr, nvic_cfsr, nvic_bfar);
                }
-               jtag_execute_queue();
+               if ((retval=jtag_execute_queue())!=ERROR_OK)
+                       return retval;
                return ERROR_JTAG_DEVICE_ERROR;
        }
 
@@ -255,17 +272,12 @@ int swjdp_transaction_endcheck(swjdp_common_t *swjdp)
 
 int swjdp_write_dpacc(swjdp_common_t *swjdp, u32 value, u8 reg_addr)
 {
-       u8 out_value_buf[4];
-       
-       buf_set_u32(out_value_buf, 0, 32, value);
-       return scan_inout_check(swjdp, SWJDP_IR_DPACC, reg_addr, DPAP_WRITE, out_value_buf, NULL);
+       return scan_inout_check_u32(swjdp, SWJDP_IR_DPACC, reg_addr, DPAP_WRITE, value, NULL);
 }
 
 int swjdp_read_dpacc(swjdp_common_t *swjdp, u32 *value, u8 reg_addr)
 {
-       scan_inout_check_u32(swjdp, SWJDP_IR_DPACC, reg_addr, DPAP_READ, 0, value);
-
-    return ERROR_OK;
+       return scan_inout_check_u32(swjdp, SWJDP_IR_DPACC, reg_addr, DPAP_READ, 0, value);
 }
 
 int swjdp_bankselect_apacc(swjdp_common_t *swjdp,u32 reg_addr)
@@ -300,7 +312,7 @@ int ahbap_read_reg(swjdp_common_t *swjdp, u32 reg_addr, u8 *in_value_buf)
 int ahbap_write_reg_u32(swjdp_common_t *swjdp, u32 reg_addr, u32 value)
 {
        u8 out_value_buf[4];
-       
+
        buf_set_u32(out_value_buf, 0, 32, value);
        swjdp_bankselect_apacc(swjdp, reg_addr);
        scan_inout_check(swjdp, SWJDP_IR_APACC, reg_addr, DPAP_WRITE, out_value_buf, NULL);
@@ -327,19 +339,19 @@ int ahbap_setup_accessport(swjdp_common_t *swjdp, u32 csw, u32 tar)
        csw = csw | CSW_DBGSWENABLE | CSW_MASTER_DEBUG | CSW_HPROT;
        if (csw != swjdp->ap_csw_value)
        {
-               //DEBUG("swjdp : Set CSW %x",csw);
-               ahbap_write_reg_u32(swjdp, AHBAP_CSW, csw ); 
+               /* LOG_DEBUG("swjdp : Set CSW %x",csw); */
+               ahbap_write_reg_u32(swjdp, AHBAP_CSW, csw );
                swjdp->ap_csw_value = csw;
        }
        if (tar != swjdp->ap_tar_value)
        {
-               //DEBUG("swjdp : Set TAR %x",tar);
+               /* LOG_DEBUG("swjdp : Set TAR %x",tar); */
                ahbap_write_reg_u32(swjdp, AHBAP_TAR, tar );
                swjdp->ap_tar_value = tar;
        }
        if (csw & CSW_ADDRINC_MASK)
-       {       
-               /* Do not cache TAR value when autoincrementing */      
+       {
+               /* Do not cache TAR value when autoincrementing */
                swjdp->ap_tar_value = -1;
        }
        return ERROR_OK;
@@ -359,14 +371,14 @@ int ahbap_read_system_u32(swjdp_common_t *swjdp, u32 address, u32 *value)
 
        ahbap_setup_accessport(swjdp, CSW_32BIT | CSW_ADDRINC_OFF, address & 0xFFFFFFF0);
        ahbap_read_reg_u32(swjdp, AHBAP_BD0 | (address & 0xC), value );
-       
+
        return ERROR_OK;
 }
 
 int ahbap_read_system_atomic_u32(swjdp_common_t *swjdp, u32 address, u32 *value)
 {
        ahbap_read_system_u32(swjdp, address, value);
-       
+
        return swjdp_transaction_endcheck(swjdp);
 }
 
@@ -390,7 +402,7 @@ int ahbap_write_system_u32(swjdp_common_t *swjdp, u32 address, u32 value)
 int ahbap_write_system_atomic_u32(swjdp_common_t *swjdp, u32 address, u32 value)
 {
        ahbap_write_system_u32(swjdp, address, value);
-       
+
        return swjdp_transaction_endcheck(swjdp);
 }
 
@@ -401,35 +413,54 @@ int ahbap_write_system_atomic_u32(swjdp_common_t *swjdp, u32 address, u32 value)
 * Write a buffer in target order (little endian)                             *
 *                                                                            *
 *****************************************************************************/
-int ahbap_write_buf(swjdp_common_t *swjdp, u8 *buffer, int count, u32 address)
+int ahbap_write_buf_u32(swjdp_common_t *swjdp, u8 *buffer, int count, u32 address)
 {
        u32 outvalue;
        int wcount, blocksize, writecount, errorcount = 0, retval = ERROR_OK;
+       u32 adr = address;
+       u8* pBuffer = buffer;
 
        swjdp->trans_mode = TRANS_MODE_COMPOSITE;
 
-       while ((address & 0x3) && (count > 0))
+       count >>= 2;
+       wcount = count;
+
+       /* if we have an unaligned access - reorder data */
+       if (adr & 0x3u)
        {
-               ahbap_setup_accessport(swjdp, CSW_8BIT | CSW_ADDRINC_SINGLE, address);
-               outvalue = (*buffer++) << 8 * (address & 0x3);
-               ahbap_write_reg_u32(swjdp, AHBAP_DRW, outvalue );
-               swjdp_transaction_endcheck(swjdp);
-               count--;
-               address++;
+               for (writecount = 0; writecount < count; writecount++)
+               {
+                       int i;
+                       outvalue = *((u32*)pBuffer);
+
+                       for (i = 0; i < 4; i++ )
+                       {
+                               *((u8*)pBuffer + (adr & 0x3)) = outvalue;
+                               outvalue >>= 8;
+                               adr++;
+                       }
+                       pBuffer += 4;
+               }
        }
-       wcount = count >> 2;
-       count = count - 4 * wcount;
+
        while (wcount > 0)
        {
                /* Adjust to write blocks within 4K aligned boundaries */
                blocksize = (0x1000 - (0xFFF & address)) >> 2;
                if (wcount < blocksize)
                        blocksize = wcount;
+
+               /* handle unaligned data at 4k boundary */
+               if (blocksize == 0)
+                       blocksize = 1;
+
                ahbap_setup_accessport(swjdp, CSW_32BIT | CSW_ADDRINC_SINGLE, address);
-               for (writecount=0; writecount<blocksize; writecount++)
+
+               for (writecount = 0; writecount < blocksize; writecount++)
                {
                        ahbap_write_reg(swjdp, AHBAP_DRW, buffer + 4 * writecount );
                }
+
                if (swjdp_transaction_endcheck(swjdp) == ERROR_OK)
                {
                        wcount = wcount - blocksize;
@@ -440,21 +471,83 @@ int ahbap_write_buf(swjdp_common_t *swjdp, u8 *buffer, int count, u32 address)
                {
                        errorcount++;
                }
+
                if (errorcount > 1)
                {
-                       WARNING("Block write error address 0x%x, wcount 0x%x", address, wcount);
+                       LOG_WARNING("Block write error address 0x%x, wcount 0x%x", address, wcount);
                        return ERROR_JTAG_DEVICE_ERROR;
                }
        }
-       
-       while (count > 0)
+
+       return retval;
+}
+
+int ahbap_write_buf_packed_u16(swjdp_common_t *swjdp, u8 *buffer, int count, u32 address)
+{
+       u32 outvalue;
+       int retval = ERROR_OK;
+       int wcount, blocksize, writecount, i;
+
+       swjdp->trans_mode = TRANS_MODE_COMPOSITE;
+
+       wcount = count >> 1;
+
+       while (wcount > 0)
        {
-               ahbap_setup_accessport(swjdp, CSW_8BIT | CSW_ADDRINC_SINGLE, address);
-               outvalue = (*buffer++) << 8 * (address & 0x3);
-               ahbap_write_reg_u32(swjdp, AHBAP_DRW, outvalue );
-               retval = swjdp_transaction_endcheck(swjdp);
-               count--;
-               address++;
+               int nbytes;
+
+               /* Adjust to read within 4K block boundaries */
+               blocksize = (0x1000 - (0xFFF & address)) >> 1;
+
+               if (wcount < blocksize)
+                       blocksize = wcount;
+
+               /* handle unaligned data at 4k boundary */
+               if (blocksize == 0)
+                       blocksize = 1;
+
+               ahbap_setup_accessport(swjdp, CSW_16BIT | CSW_ADDRINC_PACKED, address);
+               writecount = blocksize;
+
+               do
+               {
+                       nbytes = MIN((writecount << 1), 4);
+
+                       if (nbytes < 4 )
+                       {
+                               if (ahbap_write_buf_u16(swjdp, buffer, nbytes, address) != ERROR_OK)
+                               {
+                                       LOG_WARNING("Block read error address 0x%x, count 0x%x", address, count);
+                                       return ERROR_JTAG_DEVICE_ERROR;
+                               }
+
+                               address += nbytes >> 1;
+                       }
+                       else
+                       {
+                               outvalue = *((u32*)buffer);
+
+                               for (i = 0; i < nbytes; i++ )
+                               {
+                                       *((u8*)buffer + (address & 0x3)) = outvalue;
+                                       outvalue >>= 8;
+                                       address++;
+                               }
+
+                               outvalue = *((u32*)buffer);
+                               ahbap_write_reg_u32(swjdp, AHBAP_DRW, outvalue);
+                               if (swjdp_transaction_endcheck(swjdp) != ERROR_OK)
+                               {
+                                       LOG_WARNING("Block read error address 0x%x, count 0x%x", address, count);
+                                       return ERROR_JTAG_DEVICE_ERROR;
+                               }
+                       }
+
+                       buffer += nbytes >> 1;
+                       writecount -= nbytes >> 1;
+
+               } while (writecount);
+               wcount -= blocksize;
        }
 
        return retval;
@@ -464,9 +557,12 @@ int ahbap_write_buf_u16(swjdp_common_t *swjdp, u8 *buffer, int count, u32 addres
 {
        u32 outvalue;
        int retval = ERROR_OK;
-       
+
+       if (count >= 4)
+               return ahbap_write_buf_packed_u16(swjdp, buffer, count, address);
+
        swjdp->trans_mode = TRANS_MODE_COMPOSITE;
-       
+
        while (count > 0)
        {
                ahbap_setup_accessport(swjdp, CSW_16BIT | CSW_ADDRINC_SINGLE, address);
@@ -481,38 +577,128 @@ int ahbap_write_buf_u16(swjdp_common_t *swjdp, u8 *buffer, int count, u32 addres
        return retval;
 }
 
-/*****************************************************************************
-*                                                                            *
-* ahbap_read_buf(swjdp_common_t *swjdp, u8 *buffer, int count, u32 address)  *
-*                                                                            *
-* Read block fast in target order (little endian) into a buffer       *
-*                                                                            *
-*****************************************************************************/
-int ahbap_read_buf(swjdp_common_t *swjdp, u8 *buffer, int count, u32 address)
+int ahbap_write_buf_packed_u8(swjdp_common_t *swjdp, u8 *buffer, int count, u32 address)
 {
-       u32 invalue;
-       int wcount, blocksize, readcount, errorcount = 0, retval = ERROR_OK;
+       u32 outvalue;
+       int retval = ERROR_OK;
+       int wcount, blocksize, writecount, i;
 
        swjdp->trans_mode = TRANS_MODE_COMPOSITE;
 
-       while ((address & 0x3) && (count > 0))
+       wcount = count;
+
+       while (wcount > 0)
+       {
+               int nbytes;
+
+               /* Adjust to read within 4K block boundaries */
+               blocksize = (0x1000 - (0xFFF & address));
+
+               if (wcount < blocksize)
+                       blocksize = wcount;
+
+               ahbap_setup_accessport(swjdp, CSW_8BIT | CSW_ADDRINC_PACKED, address);
+               writecount = blocksize;
+
+               do
+               {
+                       nbytes = MIN(writecount, 4);
+
+                       if (nbytes < 4 )
+                       {
+                               if (ahbap_write_buf_u8(swjdp, buffer, nbytes, address) != ERROR_OK)
+                               {
+                                       LOG_WARNING("Block read error address 0x%x, count 0x%x", address, count);
+                                       return ERROR_JTAG_DEVICE_ERROR;
+                               }
+
+                               address += nbytes;
+                       }
+                       else
+                       {
+                               outvalue = *((u32*)buffer);
+
+                               for (i = 0; i < nbytes; i++ )
+                               {
+                                       *((u8*)buffer + (address & 0x3)) = outvalue;
+                                       outvalue >>= 8;
+                                       address++;
+                               }
+
+                               outvalue = *((u32*)buffer);
+                               ahbap_write_reg_u32(swjdp, AHBAP_DRW, outvalue);
+                               if (swjdp_transaction_endcheck(swjdp) != ERROR_OK)
+                               {
+                                       LOG_WARNING("Block read error address 0x%x, count 0x%x", address, count);
+                                       return ERROR_JTAG_DEVICE_ERROR;
+                               }
+                       }
+
+                       buffer += nbytes;
+                       writecount -= nbytes;
+
+               } while (writecount);
+               wcount -= blocksize;
+       }
+
+       return retval;
+}
+
+int ahbap_write_buf_u8(swjdp_common_t *swjdp, u8 *buffer, int count, u32 address)
+{
+       u32 outvalue;
+       int retval = ERROR_OK;
+
+       if (count >= 4)
+               return ahbap_write_buf_packed_u8(swjdp, buffer, count, address);
+
+       swjdp->trans_mode = TRANS_MODE_COMPOSITE;
+
+       while (count > 0)
        {
                ahbap_setup_accessport(swjdp, CSW_8BIT | CSW_ADDRINC_SINGLE, address);
-               ahbap_read_reg_u32(swjdp, AHBAP_DRW, &invalue);
-               swjdp_transaction_endcheck(swjdp);
-               *buffer++ = (invalue >> 8 * (address & 0x3)) & 0xFF;
+               outvalue = *((u8*)buffer) << 8 * (address & 0x3);
+               ahbap_write_reg_u32(swjdp, AHBAP_DRW, outvalue );
+               retval = swjdp_transaction_endcheck(swjdp);
                count--;
                address++;
+               buffer++;
        }
-       wcount = count >> 2;
-       count = count - 4 * wcount;
+
+       return retval;
+}
+
+/*********************************************************************************
+*                                                                                *
+* ahbap_read_buf_u32(swjdp_common_t *swjdp, u8 *buffer, int count, u32 address)  *
+*                                                                                *
+* Read block fast in target order (little endian) into a buffer                  *
+*                                                                                *
+**********************************************************************************/
+int ahbap_read_buf_u32(swjdp_common_t *swjdp, u8 *buffer, int count, u32 address)
+{
+       int wcount, blocksize, readcount, errorcount = 0, retval = ERROR_OK;
+       u32 adr = address;
+       u8* pBuffer = buffer;
+
+       swjdp->trans_mode = TRANS_MODE_COMPOSITE;
+
+       count >>= 2;
+       wcount = count;
+
        while (wcount > 0)
        {
                /* Adjust to read within 4K block boundaries */
                blocksize = (0x1000 - (0xFFF & address)) >> 2;
                if (wcount < blocksize)
                        blocksize = wcount;
+
+               /* handle unaligned data at 4k boundary */
+               if (blocksize == 0)
+                       blocksize = 1;
+
                ahbap_setup_accessport(swjdp, CSW_32BIT | CSW_ADDRINC_SINGLE, address);
+
                /* Scan out first read */
                swjdp_scan(swjdp->jtag_info, SWJDP_IR_APACC, AHBAP_DRW, DPAP_READ, 0, NULL, NULL);
                for (readcount = 0; readcount < blocksize - 1; readcount++)
@@ -520,104 +706,218 @@ int ahbap_read_buf(swjdp_common_t *swjdp, u8 *buffer, int count, u32 address)
                        /* Scan out read instruction and scan in previous value */
                        swjdp_scan(swjdp->jtag_info, SWJDP_IR_APACC, AHBAP_DRW, DPAP_READ, 0, buffer + 4 * readcount, &swjdp->ack);
                }
+
                /* Scan in last value */
-               swjdp_scan(swjdp->jtag_info, SWJDP_IR_DPACC, 0xC, DPAP_READ, 0, buffer + 4 * readcount, &swjdp->ack);
+               swjdp_scan(swjdp->jtag_info, SWJDP_IR_DPACC, DP_RDBUFF, DPAP_READ, 0, buffer + 4 * readcount, &swjdp->ack);
                if (swjdp_transaction_endcheck(swjdp) == ERROR_OK)
                {
                        wcount = wcount - blocksize;
                        address += 4 * blocksize;
-                       buffer += 4 * blocksize; 
+                       buffer += 4 * blocksize;
                }
                else
                {
                        errorcount++;
                }
+
                if (errorcount > 1)
                {
-                       WARNING("Block read error address 0x%x, count 0x%x", address, count);
+                       LOG_WARNING("Block read error address 0x%x, count 0x%x", address, count);
                        return ERROR_JTAG_DEVICE_ERROR;
                }
        }
 
-       while (count > 0)
+       /* if we have an unaligned access - reorder data */
+       if (adr & 0x3u)
        {
-               ahbap_setup_accessport(swjdp, CSW_8BIT | CSW_ADDRINC_SINGLE, address);
-               ahbap_read_reg_u32(swjdp, AHBAP_DRW, &invalue );
-               retval = swjdp_transaction_endcheck(swjdp);
-               *buffer++ = (invalue >> 8 * (address & 0x3)) & 0xFF;
-               count--;
-               address++;
+               for (readcount = 0; readcount < count; readcount++)
+               {
+                       int i;
+                       u32 data = *((u32*)pBuffer);
+
+                       for (i = 0; i < 4; i++ )
+                       {
+                               *((u8*)pBuffer) = (data >> 8 * (adr & 0x3));
+                               pBuffer++;
+                               adr++;
+                       }
+               }
        }
 
        return retval;
 }
 
-int ahbap_read_buf_u16(swjdp_common_t *swjdp, u8 *buffer, int count, u32 address)
+int ahbap_read_buf_packed_u16(swjdp_common_t *swjdp, u8 *buffer, int count, u32 address)
 {
        u32 invalue;
        int retval = ERROR_OK;
-       
+       int wcount, blocksize, readcount, i;
+
        swjdp->trans_mode = TRANS_MODE_COMPOSITE;
-       
-       while (count > 0)
+
+       wcount = count >> 1;
+
+       while (wcount > 0)
        {
-               ahbap_setup_accessport(swjdp, CSW_16BIT | CSW_ADDRINC_SINGLE, address);
-               ahbap_read_reg_u32(swjdp, AHBAP_DRW, &invalue );
-               retval = swjdp_transaction_endcheck(swjdp);
-               *((u16*)buffer) = (invalue >> 8 * (address & 0x3));
-               count -= 2;
-               address += 2;
-               buffer += 2;
+               int nbytes;
+
+               /* Adjust to read within 4K block boundaries */
+               blocksize = (0x1000 - (0xFFF & address)) >> 1;
+               if (wcount < blocksize)
+                       blocksize = wcount;
+
+               ahbap_setup_accessport(swjdp, CSW_16BIT | CSW_ADDRINC_PACKED, address);
+
+               /* handle unaligned data at 4k boundary */
+               if (blocksize == 0)
+                       blocksize = 1;
+               readcount = blocksize;
+
+               do
+               {
+                       ahbap_read_reg_u32(swjdp, AHBAP_DRW, &invalue );
+                       if (swjdp_transaction_endcheck(swjdp) != ERROR_OK)
+                       {
+                               LOG_WARNING("Block read error address 0x%x, count 0x%x", address, count);
+                               return ERROR_JTAG_DEVICE_ERROR;
+                       }
+
+                       nbytes = MIN((readcount << 1), 4);
+
+                       for (i = 0; i < nbytes; i++ )
+                       {
+                               *((u8*)buffer) = (invalue >> 8 * (address & 0x3));
+                               buffer++;
+                               address++;
+                       }
+
+                       readcount -= (nbytes >> 1);
+               } while (readcount);
+               wcount -= blocksize;
        }
 
        return retval;
 }
 
-int ahbap_block_read_u32(swjdp_common_t *swjdp, u32 *buffer, int count, u32 address)
+int ahbap_read_buf_u16(swjdp_common_t *swjdp, u8 *buffer, int count, u32 address)
 {
-       int readcount, errorcount = 0;
-       u32 blocksize;
-       
+       u32 invalue, i;
+       int retval = ERROR_OK;
+
+       if (count >= 4)
+               return ahbap_read_buf_packed_u16(swjdp, buffer, count, address);
+
        swjdp->trans_mode = TRANS_MODE_COMPOSITE;
-       
+
        while (count > 0)
        {
-               /* Adjust to read within 4K block boundaries */
-               blocksize = (0x1000 - (0xFFF & address)) >> 2;
-               if (count < blocksize)
-                       blocksize = count;
-               ahbap_setup_accessport(swjdp, CSW_32BIT | CSW_ADDRINC_SINGLE, address);
-               for (readcount = 0; readcount < blocksize; readcount++)
-               {
-                       ahbap_read_reg_u32(swjdp, AHBAP_DRW, buffer + readcount );
-               }
-               if (swjdp_transaction_endcheck(swjdp) == ERROR_OK)
+               ahbap_setup_accessport(swjdp, CSW_16BIT | CSW_ADDRINC_SINGLE, address);
+               ahbap_read_reg_u32(swjdp, AHBAP_DRW, &invalue );
+               retval = swjdp_transaction_endcheck(swjdp);
+               if (address & 0x1)
                {
-                       count = count - blocksize;
-                       address = address + 4 * blocksize;
-                       buffer = buffer + blocksize;
+                       for (i = 0; i < 2; i++ )
+                       {
+                               *((u8*)buffer) = (invalue >> 8 * (address & 0x3));
+                               buffer++;
+                               address++;
+                       }
                }
                else
                {
-                       errorcount++;
+                       *((u16*)buffer) = (invalue >> 8 * (address & 0x3));
+                       address += 2;
+                       buffer += 2;
                }
-               if (errorcount > 1)
+               count -= 2;
+       }
+
+       return retval;
+}
+
+int ahbap_read_buf_packed_u8(swjdp_common_t *swjdp, u8 *buffer, int count, u32 address)
+{
+       u32 invalue;
+       int retval = ERROR_OK;
+       int wcount, blocksize, readcount, i;
+
+       swjdp->trans_mode = TRANS_MODE_COMPOSITE;
+
+       wcount = count;
+
+       while (wcount > 0)
+       {
+               int nbytes;
+
+               /* Adjust to read within 4K block boundaries */
+               blocksize = (0x1000 - (0xFFF & address));
+
+               if (wcount < blocksize)
+                       blocksize = wcount;
+
+               ahbap_setup_accessport(swjdp, CSW_8BIT | CSW_ADDRINC_PACKED, address);
+               readcount = blocksize;
+
+               do
                {
-                       WARNING("Block read error address 0x%x, count 0x%x", address, count);
-                       return ERROR_JTAG_DEVICE_ERROR;
-               }
+                       ahbap_read_reg_u32(swjdp, AHBAP_DRW, &invalue );
+                       if (swjdp_transaction_endcheck(swjdp) != ERROR_OK)
+                       {
+                               LOG_WARNING("Block read error address 0x%x, count 0x%x", address, count);
+                               return ERROR_JTAG_DEVICE_ERROR;
+                       }
+
+                       nbytes = MIN(readcount, 4);
+
+                       for (i = 0; i < nbytes; i++ )
+                       {
+                               *((u8*)buffer) = (invalue >> 8 * (address & 0x3));
+                               buffer++;
+                               address++;
+                       }
+
+                       readcount -= nbytes;
+               } while (readcount);
+               wcount -= blocksize;
        }
 
-       return ERROR_OK;
+       return retval;
+}
+
+int ahbap_read_buf_u8(swjdp_common_t *swjdp, u8 *buffer, int count, u32 address)
+{
+       u32 invalue;
+       int retval = ERROR_OK;
+
+       if (count >= 4)
+               return ahbap_read_buf_packed_u8(swjdp, buffer, count, address);
+
+       swjdp->trans_mode = TRANS_MODE_COMPOSITE;
+
+       while (count > 0)
+       {
+               ahbap_setup_accessport(swjdp, CSW_8BIT | CSW_ADDRINC_SINGLE, address);
+               ahbap_read_reg_u32(swjdp, AHBAP_DRW, &invalue );
+               retval = swjdp_transaction_endcheck(swjdp);
+               *((u8*)buffer) = (invalue >> 8 * (address & 0x3));
+               count--;
+               address++;
+               buffer++;
+       }
+
+       return retval;
 }
 
 int ahbap_read_coreregister_u32(swjdp_common_t *swjdp, u32 *value, int regnum)
 {
        int retval;
        u32 dcrdr;
-       
+
+       /* because the DCB_DCRDR is used for the emulated dcc channel
+        * we gave to save/restore the DCB_DCRDR when used */
+
        ahbap_read_system_atomic_u32(swjdp, DCB_DCRDR, &dcrdr);
-       
+
        swjdp->trans_mode = TRANS_MODE_COMPOSITE;
 
        /* ahbap_write_system_u32(swjdp, DCB_DCRSR, regnum); */
@@ -627,7 +927,7 @@ int ahbap_read_coreregister_u32(swjdp_common_t *swjdp, u32 *value, int regnum)
        /* ahbap_read_system_u32(swjdp, DCB_DCRDR, value); */
        ahbap_setup_accessport(swjdp, CSW_32BIT | CSW_ADDRINC_OFF, DCB_DCRDR & 0xFFFFFFF0);
        ahbap_read_reg_u32(swjdp, AHBAP_BD0 | (DCB_DCRDR & 0xC), value );
-       
+
        retval = swjdp_transaction_endcheck(swjdp);
        ahbap_write_system_atomic_u32(swjdp, DCB_DCRDR, dcrdr);
        return retval;
@@ -637,11 +937,14 @@ int ahbap_write_coreregister_u32(swjdp_common_t *swjdp, u32 value, int regnum)
 {
        int retval;
        u32 dcrdr;
-       
+
+       /* because the DCB_DCRDR is used for the emulated dcc channel
+        * we gave to save/restore the DCB_DCRDR when used */
+
        ahbap_read_system_atomic_u32(swjdp, DCB_DCRDR, &dcrdr);
-       
+
        swjdp->trans_mode = TRANS_MODE_COMPOSITE;
-       
+
        /* ahbap_write_system_u32(swjdp, DCB_DCRDR, core_regs[i]); */
        ahbap_setup_accessport(swjdp, CSW_32BIT | CSW_ADDRINC_OFF, DCB_DCRDR & 0xFFFFFFF0);
        ahbap_write_reg_u32(swjdp, AHBAP_BD0 | (DCB_DCRDR & 0xC), value );
@@ -649,7 +952,7 @@ int ahbap_write_coreregister_u32(swjdp_common_t *swjdp, u32 value, int regnum)
        /* ahbap_write_system_u32(swjdp, DCB_DCRSR, i | DCRSR_WnR       ); */
        ahbap_setup_accessport(swjdp, CSW_32BIT | CSW_ADDRINC_OFF, DCB_DCRSR & 0xFFFFFFF0);
        ahbap_write_reg_u32(swjdp, AHBAP_BD0 | (DCB_DCRSR & 0xC), regnum | DCRSR_WnR );
-       
+
        retval = swjdp_transaction_endcheck(swjdp);
        ahbap_write_system_atomic_u32(swjdp, DCB_DCRDR, dcrdr);
        return retval;
@@ -660,49 +963,53 @@ int ahbap_debugport_init(swjdp_common_t *swjdp)
        u32 idreg, romaddr, dummy;
        u32 ctrlstat;
        int cnt = 0;
-       
-       DEBUG(" ");
-       
+       int retval;
+
+       LOG_DEBUG(" ");
+
        swjdp->ap_csw_value = -1;
        swjdp->ap_tar_value = -1;
        swjdp->trans_mode = TRANS_MODE_ATOMIC;
        swjdp_read_dpacc(swjdp, &dummy, DP_CTRL_STAT);
        swjdp_write_dpacc(swjdp, SSTICKYERR, DP_CTRL_STAT);
        swjdp_read_dpacc(swjdp, &dummy, DP_CTRL_STAT);
-       
+
        swjdp->dp_ctrl_stat = CDBGPWRUPREQ | CSYSPWRUPREQ;
 
        swjdp_write_dpacc(swjdp, swjdp->dp_ctrl_stat, DP_CTRL_STAT);
        swjdp_read_dpacc(swjdp, &ctrlstat, DP_CTRL_STAT);
-       jtag_execute_queue();
+       if ((retval=jtag_execute_queue())!=ERROR_OK)
+               return retval;
 
        /* Check that we have debug power domains activated */
        while (!(ctrlstat & CDBGPWRUPACK) && (cnt++ < 10))
        {
-               DEBUG("swjdp: wait CDBGPWRUPACK");
+               LOG_DEBUG("swjdp: wait CDBGPWRUPACK");
                swjdp_read_dpacc(swjdp, &ctrlstat, DP_CTRL_STAT);
-               jtag_execute_queue();
-               usleep(10000);
+               if ((retval=jtag_execute_queue())!=ERROR_OK)
+                       return retval;
+               alive_sleep(10);
        }
 
        while (!(ctrlstat & CSYSPWRUPACK) && (cnt++ < 10))
        {
-               DEBUG("swjdp: wait CSYSPWRUPACK");
+               LOG_DEBUG("swjdp: wait CSYSPWRUPACK");
                swjdp_read_dpacc(swjdp, &ctrlstat, DP_CTRL_STAT);
-               jtag_execute_queue();
-               usleep(10000);
+               if ((retval=jtag_execute_queue())!=ERROR_OK)
+                       return retval;
+               alive_sleep(10);
        }
 
        swjdp_read_dpacc(swjdp, &dummy, DP_CTRL_STAT);
        /* With debug power on we can activate OVERRUN checking */
        swjdp->dp_ctrl_stat = CDBGPWRUPREQ | CSYSPWRUPREQ | CORUNDETECT;
-       swjdp_write_dpacc(swjdp, swjdp->dp_ctrl_stat , DP_CTRL_STAT);
+       swjdp_write_dpacc(swjdp, swjdp->dp_ctrl_stat, DP_CTRL_STAT);
        swjdp_read_dpacc(swjdp, &dummy, DP_CTRL_STAT);
-       
+
        ahbap_read_reg_u32(swjdp, 0xFC, &idreg);
        ahbap_read_reg_u32(swjdp, 0xF8, &romaddr);
-       
-       DEBUG("AHB-AP ID Register 0x%x, Debug ROM Address 0x%x", idreg, romaddr);       
-       
+
+       LOG_DEBUG("AHB-AP ID Register 0x%x, Debug ROM Address 0x%x", idreg, romaddr);
+
        return ERROR_OK;
 }