warnings: fix alignment warnings
[fw/openocd] / src / target / xscale.c
index 6f2d6ee3dfe7ec562b92e2b06a7e07fc4187cb49..e0ce400b46101a5b480467693deccf0757a64db4 100644 (file)
 #include "config.h"
 #endif
 
+#include "breakpoints.h"
 #include "xscale.h"
 #include "target_type.h"
-#include "arm7_9_common.h"
+#include "arm_jtag.h"
 #include "arm_simulator.h"
 #include "arm_disassembler.h"
-#include "time_support.h"
+#include <helper/time_support.h>
+#include "register.h"
 #include "image.h"
+#include "arm_opcodes.h"
+#include "armv4_5.h"
 
 
 /*
 
 
 /* forward declarations */
-static int xscale_resume(struct target_s *, int current,
+static int xscale_resume(struct target *, int current,
        uint32_t address, int handle_breakpoints, int debug_execution);
-static int xscale_debug_entry(target_t *);
-static int xscale_restore_context(target_t *);
-static int xscale_get_reg(reg_t *reg);
-static int xscale_set_reg(reg_t *reg, uint8_t *buf);
-static int xscale_set_breakpoint(struct target_s *, breakpoint_t *);
-static int xscale_set_watchpoint(struct target_s *, watchpoint_t *);
-static int xscale_unset_breakpoint(struct target_s *, breakpoint_t *);
-static int xscale_read_trace(target_t *);
+static int xscale_debug_entry(struct target *);
+static int xscale_restore_banked(struct target *);
+static int xscale_get_reg(struct reg *reg);
+static int xscale_set_reg(struct reg *reg, uint8_t *buf);
+static int xscale_set_breakpoint(struct target *, struct breakpoint *);
+static int xscale_set_watchpoint(struct target *, struct watchpoint *);
+static int xscale_unset_breakpoint(struct target *, struct breakpoint *);
+static int xscale_read_trace(struct target *);
 
 
 /* This XScale "debug handler" is loaded into the processor's
@@ -106,7 +110,7 @@ static char *const xscale_reg_list[] =
        "XSCALE_TXRXCTRL",
 };
 
-static const xscale_reg_t xscale_reg_arch_info[] =
+static const struct xscale_reg xscale_reg_arch_info[] =
 {
        {XSCALE_MAINID, NULL},
        {XSCALE_CACHETYPE, NULL},
@@ -132,10 +136,8 @@ static const xscale_reg_t xscale_reg_arch_info[] =
        {-1, NULL}, /* TXRXCTRL implicit access via JTAG */
 };
 
-static int xscale_reg_arch_type = -1;
-
 /* convenience wrapper to access XScale specific registers */
-static int xscale_set_reg_u32(reg_t *reg, uint32_t value)
+static int xscale_set_reg_u32(struct reg *reg, uint32_t value)
 {
        uint8_t buf[4];
 
@@ -146,8 +148,8 @@ static int xscale_set_reg_u32(reg_t *reg, uint32_t value)
 
 static const char xscale_not[] = "target is not an XScale";
 
-static int xscale_verify_pointer(struct command_context_s *cmd_ctx,
-               struct xscale_common_s *xscale)
+static int xscale_verify_pointer(struct command_context *cmd_ctx,
+               struct xscale_common *xscale)
 {
        if (xscale->common_magic != XSCALE_COMMON_MAGIC) {
                command_print(cmd_ctx, xscale_not);
@@ -156,33 +158,31 @@ static int xscale_verify_pointer(struct command_context_s *cmd_ctx,
        return ERROR_OK;
 }
 
-static int xscale_jtag_set_instr(jtag_tap_t *tap, uint32_t new_instr)
+static int xscale_jtag_set_instr(struct jtag_tap *tap, uint32_t new_instr, tap_state_t end_state)
 {
-       if (tap == NULL)
-               return ERROR_FAIL;
+       assert (tap != NULL);
 
        if (buf_get_u32(tap->cur_instr, 0, tap->ir_length) != new_instr)
        {
-               scan_field_t field;
+               struct scan_field field;
                uint8_t scratch[4];
 
                memset(&field, 0, sizeof field);
-               field.tap = tap;
                field.num_bits = tap->ir_length;
                field.out_value = scratch;
-               buf_set_u32(field.out_value, 0, field.num_bits, new_instr);
+               buf_set_u32(scratch, 0, field.num_bits, new_instr);
 
-               jtag_add_ir_scan(1, &field, jtag_get_end_state());
+               jtag_add_ir_scan(tap, &field, end_state);
        }
 
        return ERROR_OK;
 }
 
-static int xscale_read_dcsr(target_t *target)
+static int xscale_read_dcsr(struct target *target)
 {
-       struct xscale_common_s *xscale = target_to_xscale(target);
+       struct xscale_common *xscale = target_to_xscale(target);
        int retval;
-       scan_field_t fields[3];
+       struct scan_field fields[3];
        uint8_t field0 = 0x0;
        uint8_t field0_check_value = 0x2;
        uint8_t field0_check_mask = 0x7;
@@ -190,31 +190,29 @@ static int xscale_read_dcsr(target_t *target)
        uint8_t field2_check_value = 0x0;
        uint8_t field2_check_mask = 0x1;
 
-       jtag_set_end_state(TAP_DRPAUSE);
-       xscale_jtag_set_instr(target->tap, XSCALE_SELDCSR);
+       xscale_jtag_set_instr(target->tap,
+               XSCALE_SELDCSR << xscale->xscale_variant,
+               TAP_DRPAUSE);
 
        buf_set_u32(&field0, 1, 1, xscale->hold_rst);
        buf_set_u32(&field0, 2, 1, xscale->external_debug_break);
 
        memset(&fields, 0, sizeof fields);
 
-       fields[0].tap = target->tap;
        fields[0].num_bits = 3;
        fields[0].out_value = &field0;
        uint8_t tmp;
        fields[0].in_value = &tmp;
 
-       fields[1].tap = target->tap;
        fields[1].num_bits = 32;
        fields[1].in_value = xscale->reg_cache->reg_list[XSCALE_DCSR].value;
 
-       fields[2].tap = target->tap;
        fields[2].num_bits = 1;
        fields[2].out_value = &field2;
        uint8_t tmp2;
        fields[2].in_value = &tmp2;
 
-       jtag_add_dr_scan(3, fields, jtag_get_end_state());
+       jtag_add_dr_scan(target->tap, 3, fields, TAP_DRPAUSE);
 
        jtag_check_value_mask(fields + 0, &field0_check_value, &field0_check_mask);
        jtag_check_value_mask(fields + 2, &field2_check_value, &field2_check_mask);
@@ -235,9 +233,7 @@ static int xscale_read_dcsr(target_t *target)
        fields[1].out_value = xscale->reg_cache->reg_list[XSCALE_DCSR].value;
        fields[1].in_value = NULL;
 
-       jtag_set_end_state(TAP_IDLE);
-
-       jtag_add_dr_scan(3, fields, jtag_get_end_state());
+       jtag_add_dr_scan(target->tap, 3, fields, TAP_DRPAUSE);
 
        /* DANGER!!! this must be here. It will make sure that the arguments
         * to jtag_set_check_value() does not go out of scope! */
@@ -248,17 +244,18 @@ static int xscale_read_dcsr(target_t *target)
 static void xscale_getbuf(jtag_callback_data_t arg)
 {
        uint8_t *in = (uint8_t *)arg;
-       *((uint32_t *)in) = buf_get_u32(in, 0, 32);
+       *((uint32_t *)arg) = buf_get_u32(in, 0, 32);
 }
 
-static int xscale_receive(target_t *target, uint32_t *buffer, int num_words)
+static int xscale_receive(struct target *target, uint32_t *buffer, int num_words)
 {
        if (num_words == 0)
                return ERROR_INVALID_ARGUMENTS;
 
+       struct xscale_common *xscale = target_to_xscale(target);
        int retval = ERROR_OK;
        tap_state_t path[3];
-       scan_field_t fields[3];
+       struct scan_field fields[3];
        uint8_t *field0 = malloc(num_words * 1);
        uint8_t field0_check_value = 0x2;
        uint8_t field0_check_mask = 0x6;
@@ -275,22 +272,20 @@ static int xscale_receive(target_t *target, uint32_t *buffer, int num_words)
 
        memset(&fields, 0, sizeof fields);
 
-       fields[0].tap = target->tap;
        fields[0].num_bits = 3;
        fields[0].check_value = &field0_check_value;
        fields[0].check_mask = &field0_check_mask;
 
-       fields[1].tap = target->tap;
        fields[1].num_bits = 32;
 
-       fields[2].tap = target->tap;
        fields[2].num_bits = 1;
        fields[2].check_value = &field2_check_value;
        fields[2].check_mask = &field2_check_mask;
 
-       jtag_set_end_state(TAP_IDLE);
-       xscale_jtag_set_instr(target->tap, XSCALE_DBGTX);
-       jtag_add_runtest(1, jtag_get_end_state()); /* ensures that we're in the TAP_IDLE state as the above could be a no-op */
+       xscale_jtag_set_instr(target->tap,
+               XSCALE_DBGTX << xscale->xscale_variant,
+               TAP_IDLE);
+       jtag_add_runtest(1, TAP_IDLE); /* ensures that we're in the TAP_IDLE state as the above could be a no-op */
 
        /* repeat until all words have been collected */
        int attempts = 0;
@@ -306,7 +301,7 @@ static int xscale_receive(target_t *target, uint32_t *buffer, int num_words)
 
                        fields[1].in_value = (uint8_t *)(field1 + i);
 
-                       jtag_add_dr_scan_check(3, fields, jtag_set_end_state(TAP_IDLE));
+                       jtag_add_dr_scan_check(target->tap, 3, fields, TAP_IDLE);
 
                        jtag_add_callback(xscale_getbuf, (jtag_callback_data_t)(field1 + i));
 
@@ -355,23 +350,23 @@ static int xscale_receive(target_t *target, uint32_t *buffer, int num_words)
        return retval;
 }
 
-static int xscale_read_tx(target_t *target, int consume)
+static int xscale_read_tx(struct target *target, int consume)
 {
-       struct xscale_common_s *xscale = target_to_xscale(target);
+       struct xscale_common *xscale = target_to_xscale(target);
        tap_state_t path[3];
        tap_state_t noconsume_path[6];
        int retval;
        struct timeval timeout, now;
-       scan_field_t fields[3];
+       struct scan_field fields[3];
        uint8_t field0_in = 0x0;
        uint8_t field0_check_value = 0x2;
        uint8_t field0_check_mask = 0x6;
        uint8_t field2_check_value = 0x0;
        uint8_t field2_check_mask = 0x1;
 
-       jtag_set_end_state(TAP_IDLE);
-
-       xscale_jtag_set_instr(target->tap, XSCALE_DBGTX);
+       xscale_jtag_set_instr(target->tap,
+               XSCALE_DBGTX << xscale->xscale_variant,
+               TAP_IDLE);
 
        path[0] = TAP_DRSELECT;
        path[1] = TAP_DRCAPTURE;
@@ -386,15 +381,12 @@ static int xscale_read_tx(target_t *target, int consume)
 
        memset(&fields, 0, sizeof fields);
 
-       fields[0].tap = target->tap;
        fields[0].num_bits = 3;
        fields[0].in_value = &field0_in;
 
-       fields[1].tap = target->tap;
        fields[1].num_bits = 32;
        fields[1].in_value = xscale->reg_cache->reg_list[XSCALE_TX].value;
 
-       fields[2].tap = target->tap;
        fields[2].num_bits = 1;
        uint8_t tmp;
        fields[2].in_value = &tmp;
@@ -412,10 +404,10 @@ static int xscale_read_tx(target_t *target, int consume)
                        jtag_add_pathmove(3, path);
                else
                {
-                       jtag_add_pathmove(sizeof(noconsume_path)/sizeof(*noconsume_path), noconsume_path);
+                       jtag_add_pathmove(ARRAY_SIZE(noconsume_path), noconsume_path);
                }
 
-               jtag_add_dr_scan(3, fields, jtag_set_end_state(TAP_IDLE));
+               jtag_add_dr_scan(target->tap, 3, fields, TAP_IDLE);
 
                jtag_check_value_mask(fields + 0, &field0_check_value, &field0_check_mask);
                jtag_check_value_mask(fields + 2, &field2_check_value, &field2_check_mask);
@@ -453,12 +445,12 @@ static int xscale_read_tx(target_t *target, int consume)
        return ERROR_OK;
 }
 
-static int xscale_write_rx(target_t *target)
+static int xscale_write_rx(struct target *target)
 {
-       struct xscale_common_s *xscale = target_to_xscale(target);
+       struct xscale_common *xscale = target_to_xscale(target);
        int retval;
        struct timeval timeout, now;
-       scan_field_t fields[3];
+       struct scan_field fields[3];
        uint8_t field0_out = 0x0;
        uint8_t field0_in = 0x0;
        uint8_t field0_check_value = 0x2;
@@ -467,22 +459,19 @@ static int xscale_write_rx(target_t *target)
        uint8_t field2_check_value = 0x0;
        uint8_t field2_check_mask = 0x1;
 
-       jtag_set_end_state(TAP_IDLE);
-
-       xscale_jtag_set_instr(target->tap, XSCALE_DBGRX);
+       xscale_jtag_set_instr(target->tap,
+               XSCALE_DBGRX << xscale->xscale_variant,
+               TAP_IDLE);
 
        memset(&fields, 0, sizeof fields);
 
-       fields[0].tap = target->tap;
        fields[0].num_bits = 3;
        fields[0].out_value = &field0_out;
        fields[0].in_value = &field0_in;
 
-       fields[1].tap = target->tap;
        fields[1].num_bits = 32;
        fields[1].out_value = xscale->reg_cache->reg_list[XSCALE_RX].value;
 
-       fields[2].tap = target->tap;
        fields[2].num_bits = 1;
        fields[2].out_value = &field2;
        uint8_t tmp;
@@ -495,7 +484,7 @@ static int xscale_write_rx(target_t *target)
        LOG_DEBUG("polling RX");
        for (;;)
        {
-               jtag_add_dr_scan(3, fields, jtag_set_end_state(TAP_IDLE));
+               jtag_add_dr_scan(target->tap, 3, fields, TAP_IDLE);
 
                jtag_check_value_mask(fields + 0, &field0_check_value, &field0_check_mask);
                jtag_check_value_mask(fields + 2, &field2_check_value, &field2_check_mask);
@@ -527,7 +516,7 @@ static int xscale_write_rx(target_t *target)
 
        /* set rx_valid */
        field2 = 0x1;
-       jtag_add_dr_scan(3, fields, jtag_set_end_state(TAP_IDLE));
+       jtag_add_dr_scan(target->tap, 3, fields, TAP_IDLE);
 
        if ((retval = jtag_execute_queue()) != ERROR_OK)
        {
@@ -539,16 +528,17 @@ static int xscale_write_rx(target_t *target)
 }
 
 /* send count elements of size byte to the debug handler */
-static int xscale_send(target_t *target, uint8_t *buffer, int count, int size)
+static int xscale_send(struct target *target, uint8_t *buffer, int count, int size)
 {
+       struct xscale_common *xscale = target_to_xscale(target);
        uint32_t t[3];
        int bits[3];
        int retval;
        int done_count = 0;
 
-       jtag_set_end_state(TAP_IDLE);
-
-       xscale_jtag_set_instr(target->tap, XSCALE_DBGRX);
+       xscale_jtag_set_instr(target->tap,
+               XSCALE_DBGRX << xscale->xscale_variant,
+               TAP_IDLE);
 
        bits[0]=3;
        t[0]=0;
@@ -583,13 +573,13 @@ static int xscale_send(target_t *target, uint8_t *buffer, int count, int size)
                        break;
                default:
                        LOG_ERROR("BUG: size neither 4, 2 nor 1");
-                       exit(-1);
+                       return ERROR_INVALID_ARGUMENTS;
                }
                jtag_add_dr_out(target->tap,
                                3,
                                bits,
                                t,
-                               jtag_set_end_state(TAP_IDLE));
+                               TAP_IDLE);
                buffer += size;
        }
 
@@ -602,19 +592,19 @@ static int xscale_send(target_t *target, uint8_t *buffer, int count, int size)
        return ERROR_OK;
 }
 
-static int xscale_send_u32(target_t *target, uint32_t value)
+static int xscale_send_u32(struct target *target, uint32_t value)
 {
-       struct xscale_common_s *xscale = target_to_xscale(target);
+       struct xscale_common *xscale = target_to_xscale(target);
 
        buf_set_u32(xscale->reg_cache->reg_list[XSCALE_RX].value, 0, 32, value);
        return xscale_write_rx(target);
 }
 
-static int xscale_write_dcsr(target_t *target, int hold_rst, int ext_dbg_brk)
+static int xscale_write_dcsr(struct target *target, int hold_rst, int ext_dbg_brk)
 {
-       struct xscale_common_s *xscale = target_to_xscale(target);
+       struct xscale_common *xscale = target_to_xscale(target);
        int retval;
-       scan_field_t fields[3];
+       struct scan_field fields[3];
        uint8_t field0 = 0x0;
        uint8_t field0_check_value = 0x2;
        uint8_t field0_check_mask = 0x7;
@@ -628,31 +618,29 @@ static int xscale_write_dcsr(target_t *target, int hold_rst, int ext_dbg_brk)
        if (ext_dbg_brk != -1)
                xscale->external_debug_break = ext_dbg_brk;
 
-       jtag_set_end_state(TAP_IDLE);
-       xscale_jtag_set_instr(target->tap, XSCALE_SELDCSR);
+       xscale_jtag_set_instr(target->tap,
+               XSCALE_SELDCSR << xscale->xscale_variant,
+               TAP_IDLE);
 
        buf_set_u32(&field0, 1, 1, xscale->hold_rst);
        buf_set_u32(&field0, 2, 1, xscale->external_debug_break);
 
        memset(&fields, 0, sizeof fields);
 
-       fields[0].tap = target->tap;
        fields[0].num_bits = 3;
        fields[0].out_value = &field0;
        uint8_t tmp;
        fields[0].in_value = &tmp;
 
-       fields[1].tap = target->tap;
        fields[1].num_bits = 32;
        fields[1].out_value = xscale->reg_cache->reg_list[XSCALE_DCSR].value;
 
-       fields[2].tap = target->tap;
        fields[2].num_bits = 1;
        fields[2].out_value = &field2;
        uint8_t tmp2;
        fields[2].in_value = &tmp2;
 
-       jtag_add_dr_scan(3, fields, jtag_get_end_state());
+       jtag_add_dr_scan(target->tap, 3, fields, TAP_IDLE);
 
        jtag_check_value_mask(fields + 0, &field0_check_value, &field0_check_mask);
        jtag_check_value_mask(fields + 2, &field2_check_value, &field2_check_mask);
@@ -681,18 +669,20 @@ static unsigned int parity (unsigned int v)
        return (0x6996 >> v) & 1;
 }
 
-static int xscale_load_ic(target_t *target, uint32_t va, uint32_t buffer[8])
+static int xscale_load_ic(struct target *target, uint32_t va, uint32_t buffer[8])
 {
+       struct xscale_common *xscale = target_to_xscale(target);
        uint8_t packet[4];
        uint8_t cmd;
        int word;
-       scan_field_t fields[2];
+       struct scan_field fields[2];
 
        LOG_DEBUG("loading miniIC at 0x%8.8" PRIx32 "", va);
 
        /* LDIC into IR */
-       jtag_set_end_state(TAP_IDLE);
-       xscale_jtag_set_instr(target->tap, XSCALE_LDIC);
+       xscale_jtag_set_instr(target->tap,
+               XSCALE_LDIC << xscale->xscale_variant,
+               TAP_IDLE);
 
        /* CMD is b011 to load a cacheline into the Mini ICache.
         * Loading into the main ICache is deprecated, and unused.
@@ -705,15 +695,13 @@ static int xscale_load_ic(target_t *target, uint32_t va, uint32_t buffer[8])
 
        memset(&fields, 0, sizeof fields);
 
-       fields[0].tap = target->tap;
        fields[0].num_bits = 6;
        fields[0].out_value = &cmd;
 
-       fields[1].tap = target->tap;
        fields[1].num_bits = 27;
        fields[1].out_value = packet;
 
-       jtag_add_dr_scan(2, fields, jtag_get_end_state());
+       jtag_add_dr_scan(target->tap, 2, fields, TAP_IDLE);
 
        /* rest of packet is a cacheline: 8 instructions, with parity */
        fields[0].num_bits = 32;
@@ -730,20 +718,22 @@ static int xscale_load_ic(target_t *target, uint32_t va, uint32_t buffer[8])
                memcpy(&value, packet, sizeof(uint32_t));
                cmd = parity(value);
 
-               jtag_add_dr_scan(2, fields, jtag_get_end_state());
+               jtag_add_dr_scan(target->tap, 2, fields, TAP_IDLE);
        }
 
        return jtag_execute_queue();
 }
 
-static int xscale_invalidate_ic_line(target_t *target, uint32_t va)
+static int xscale_invalidate_ic_line(struct target *target, uint32_t va)
 {
+       struct xscale_common *xscale = target_to_xscale(target);
        uint8_t packet[4];
        uint8_t cmd;
-       scan_field_t fields[2];
+       struct scan_field fields[2];
 
-       jtag_set_end_state(TAP_IDLE);
-       xscale_jtag_set_instr(target->tap, XSCALE_LDIC);
+       xscale_jtag_set_instr(target->tap,
+               XSCALE_LDIC << xscale->xscale_variant,
+               TAP_IDLE);
 
        /* CMD for invalidate IC line b000, bits [6:4] b000 */
        buf_set_u32(&cmd, 0, 6, 0x0);
@@ -753,22 +743,20 @@ static int xscale_invalidate_ic_line(target_t *target, uint32_t va)
 
        memset(&fields, 0, sizeof fields);
 
-       fields[0].tap = target->tap;
        fields[0].num_bits = 6;
        fields[0].out_value = &cmd;
 
-       fields[1].tap = target->tap;
        fields[1].num_bits = 27;
        fields[1].out_value = packet;
 
-       jtag_add_dr_scan(2, fields, jtag_get_end_state());
+       jtag_add_dr_scan(target->tap, 2, fields, TAP_IDLE);
 
        return ERROR_OK;
 }
 
-static int xscale_update_vectors(target_t *target)
+static int xscale_update_vectors(struct target *target)
 {
-       struct xscale_common_s *xscale = target_to_xscale(target);
+       struct xscale_common *xscale = target_to_xscale(target);
        int i;
        int retval;
 
@@ -830,10 +818,10 @@ static int xscale_update_vectors(target_t *target)
        return ERROR_OK;
 }
 
-static int xscale_arch_state(struct target_s *target)
+static int xscale_arch_state(struct target *target)
 {
-       struct xscale_common_s *xscale = target_to_xscale(target);
-       struct armv4_5_common_s *armv4_5 = &xscale->armv4_5_common;
+       struct xscale_common *xscale = target_to_xscale(target);
+       struct arm *armv4_5 = &xscale->armv4_5_common;
 
        static const char *state[] =
        {
@@ -845,21 +833,14 @@ static int xscale_arch_state(struct target_s *target)
                "", "\n(processor reset)", "\n(trace buffer full)"
        };
 
-       if (armv4_5->common_magic != ARMV4_5_COMMON_MAGIC)
+       if (armv4_5->common_magic != ARM_COMMON_MAGIC)
        {
                LOG_ERROR("BUG: called for a non-ARMv4/5 target");
-               exit(-1);
+               return ERROR_INVALID_ARGUMENTS;
        }
 
-       LOG_USER("target halted in %s state due to %s, current mode: %s\n"
-                       "cpsr: 0x%8.8" PRIx32 " pc: 0x%8.8" PRIx32 "\n"
-                       "MMU: %s, D-Cache: %s, I-Cache: %s"
-                       "%s",
-                        armv4_5_state_strings[armv4_5->core_state],
-                        Jim_Nvp_value2name_simple(nvp_target_debug_reason, target->debug_reason)->name ,
-                        armv4_5_mode_strings[armv4_5_mode_to_number(armv4_5->core_mode)],
-                        buf_get_u32(armv4_5->core_cache->reg_list[ARMV4_5_CPSR].value, 0, 32),
-                        buf_get_u32(armv4_5->core_cache->reg_list[15].value, 0, 32),
+       arm_arch_state(target);
+       LOG_USER("MMU: %s, D-Cache: %s, I-Cache: %s%s",
                         state[xscale->armv4_5_mmu.mmu_enabled],
                         state[xscale->armv4_5_mmu.armv4_5_cache.d_u_cache_enabled],
                         state[xscale->armv4_5_mmu.armv4_5_cache.i_cache_enabled],
@@ -868,7 +849,7 @@ static int xscale_arch_state(struct target_s *target)
        return ERROR_OK;
 }
 
-static int xscale_poll(target_t *target)
+static int xscale_poll(struct target *target)
 {
        int retval = ERROR_OK;
 
@@ -908,13 +889,13 @@ static int xscale_poll(target_t *target)
        return retval;
 }
 
-static int xscale_debug_entry(target_t *target)
+static int xscale_debug_entry(struct target *target)
 {
-       struct xscale_common_s *xscale = target_to_xscale(target);
-       struct armv4_5_common_s *armv4_5 = &xscale->armv4_5_common;
+       struct xscale_common *xscale = target_to_xscale(target);
+       struct arm *armv4_5 = &xscale->armv4_5_common;
        uint32_t pc;
        uint32_t buffer[10];
-       int i;
+       unsigned i;
        int retval;
        uint32_t moe;
 
@@ -934,9 +915,9 @@ static int xscale_debug_entry(target_t *target)
        LOG_DEBUG("r0: 0x%8.8" PRIx32 "", buffer[0]);
 
        /* move pc from buffer to register cache */
-       buf_set_u32(armv4_5->core_cache->reg_list[15].value, 0, 32, buffer[1]);
-       armv4_5->core_cache->reg_list[15].dirty = 1;
-       armv4_5->core_cache->reg_list[15].valid = 1;
+       buf_set_u32(armv4_5->pc->value, 0, 32, buffer[1]);
+       armv4_5->pc->dirty = 1;
+       armv4_5->pc->valid = 1;
        LOG_DEBUG("pc: 0x%8.8" PRIx32 "", buffer[1]);
 
        /* move data from buffer to register cache */
@@ -948,36 +929,24 @@ static int xscale_debug_entry(target_t *target)
                LOG_DEBUG("r%i: 0x%8.8" PRIx32 "", i, buffer[i + 1]);
        }
 
-       buf_set_u32(armv4_5->core_cache->reg_list[ARMV4_5_CPSR].value, 0, 32, buffer[9]);
-       armv4_5->core_cache->reg_list[ARMV4_5_CPSR].dirty = 1;
-       armv4_5->core_cache->reg_list[ARMV4_5_CPSR].valid = 1;
+       arm_set_cpsr(armv4_5, buffer[9]);
        LOG_DEBUG("cpsr: 0x%8.8" PRIx32 "", buffer[9]);
 
-       armv4_5->core_mode = buffer[9] & 0x1f;
-       if (armv4_5_mode_to_number(armv4_5->core_mode) == -1)
+       if (!is_arm_mode(armv4_5->core_mode))
        {
                target->state = TARGET_UNKNOWN;
                LOG_ERROR("cpsr contains invalid mode value - communication failure");
                return ERROR_TARGET_FAILURE;
        }
-       LOG_DEBUG("target entered debug state in %s mode", armv4_5_mode_strings[armv4_5_mode_to_number(armv4_5->core_mode)]);
-
-       if (buffer[9] & 0x20)
-               armv4_5->core_state = ARMV4_5_STATE_THUMB;
-       else
-               armv4_5->core_state = ARMV4_5_STATE_ARM;
-
-
-       if (armv4_5_mode_to_number(armv4_5->core_mode)==-1)
-               return ERROR_FAIL;
+       LOG_DEBUG("target entered debug state in %s mode",
+                        arm_mode_name(armv4_5->core_mode));
 
        /* get banked registers, r8 to r14, and spsr if not in USR/SYS mode */
-       if ((armv4_5->core_mode != ARMV4_5_MODE_USR) && (armv4_5->core_mode != ARMV4_5_MODE_SYS))
-       {
+       if (armv4_5->spsr) {
                xscale_receive(target, buffer, 8);
-               buf_set_u32(ARMV4_5_CORE_REG_MODE(armv4_5->core_cache, armv4_5->core_mode, 16).value, 0, 32, buffer[7]);
-               ARMV4_5_CORE_REG_MODE(armv4_5->core_cache, armv4_5->core_mode, 16).dirty = 0;
-               ARMV4_5_CORE_REG_MODE(armv4_5->core_cache, armv4_5->core_mode, 16).valid = 1;
+               buf_set_u32(armv4_5->spsr->value, 0, 32, buffer[7]);
+               armv4_5->spsr->dirty = false;
+               armv4_5->spsr->valid = true;
        }
        else
        {
@@ -985,20 +954,27 @@ static int xscale_debug_entry(target_t *target)
                xscale_receive(target, buffer, 7);
        }
 
-       /* move data from buffer to register cache */
+       /* move data from buffer to right banked register in cache */
        for (i = 8; i <= 14; i++)
        {
-               buf_set_u32(ARMV4_5_CORE_REG_MODE(armv4_5->core_cache, armv4_5->core_mode, i).value, 0, 32, buffer[i - 8]);
-               ARMV4_5_CORE_REG_MODE(armv4_5->core_cache, armv4_5->core_mode, i).dirty = 0;
-               ARMV4_5_CORE_REG_MODE(armv4_5->core_cache, armv4_5->core_mode, i).valid = 1;
+               struct reg *r = arm_reg_current(armv4_5, i);
+
+               buf_set_u32(r->value, 0, 32, buffer[i - 8]);
+               r->dirty = false;
+               r->valid = true;
        }
 
+       /* mark xscale regs invalid to ensure they are retrieved from the
+        * debug handler if requested  */
+       for (i = 0; i < xscale->reg_cache->num_regs; i++)
+          xscale->reg_cache->reg_list[i].valid = 0;
+
        /* examine debug reason */
        xscale_read_dcsr(target);
        moe = buf_get_u32(xscale->reg_cache->reg_list[XSCALE_DCSR].value, 2, 3);
 
        /* stored PC (for calculating fixup) */
-       pc = buf_get_u32(armv4_5->core_cache->reg_list[15].value, 0, 32);
+       pc = buf_get_u32(armv4_5->pc->value, 0, 32);
 
        switch (moe)
        {
@@ -1045,7 +1021,7 @@ static int xscale_debug_entry(target_t *target)
        }
 
        /* apply PC fixup */
-       buf_set_u32(armv4_5->core_cache->reg_list[15].value, 0, 32, pc);
+       buf_set_u32(armv4_5->pc->value, 0, 32, pc);
 
        /* on the first debug entry, identify cache type */
        if (xscale->armv4_5_mmu.armv4_5_cache.ctype == -1)
@@ -1088,9 +1064,9 @@ static int xscale_debug_entry(target_t *target)
        return ERROR_OK;
 }
 
-static int xscale_halt(target_t *target)
+static int xscale_halt(struct target *target)
 {
-       struct xscale_common_s *xscale = target_to_xscale(target);
+       struct xscale_common *xscale = target_to_xscale(target);
 
        LOG_DEBUG("target->state: %s",
                  target_state_name(target));
@@ -1122,15 +1098,15 @@ static int xscale_halt(target_t *target)
        return ERROR_OK;
 }
 
-static int xscale_enable_single_step(struct target_s *target, uint32_t next_pc)
+static int xscale_enable_single_step(struct target *target, uint32_t next_pc)
 {
-       struct xscale_common_s *xscale = target_to_xscale(target);
-       reg_t *ibcr0 = &xscale->reg_cache->reg_list[XSCALE_IBCR0];
+       struct xscale_common *xscale = target_to_xscale(target);
+       struct reg *ibcr0 = &xscale->reg_cache->reg_list[XSCALE_IBCR0];
        int retval;
 
        if (xscale->ibcr0_used)
        {
-               breakpoint_t *ibcr0_bp = breakpoint_find(target, buf_get_u32(ibcr0->value, 0, 32) & 0xfffffffe);
+               struct breakpoint *ibcr0_bp = breakpoint_find(target, buf_get_u32(ibcr0->value, 0, 32) & 0xfffffffe);
 
                if (ibcr0_bp)
                {
@@ -1149,10 +1125,10 @@ static int xscale_enable_single_step(struct target_s *target, uint32_t next_pc)
        return ERROR_OK;
 }
 
-static int xscale_disable_single_step(struct target_s *target)
+static int xscale_disable_single_step(struct target *target)
 {
-       struct xscale_common_s *xscale = target_to_xscale(target);
-       reg_t *ibcr0 = &xscale->reg_cache->reg_list[XSCALE_IBCR0];
+       struct xscale_common *xscale = target_to_xscale(target);
+       struct reg *ibcr0 = &xscale->reg_cache->reg_list[XSCALE_IBCR0];
        int retval;
 
        if ((retval = xscale_set_reg_u32(ibcr0, 0x0)) != ERROR_OK)
@@ -1161,9 +1137,9 @@ static int xscale_disable_single_step(struct target_s *target)
        return ERROR_OK;
 }
 
-static void xscale_enable_watchpoints(struct target_s *target)
+static void xscale_enable_watchpoints(struct target *target)
 {
-       watchpoint_t *watchpoint = target->watchpoints;
+       struct watchpoint *watchpoint = target->watchpoints;
 
        while (watchpoint)
        {
@@ -1173,9 +1149,9 @@ static void xscale_enable_watchpoints(struct target_s *target)
        }
 }
 
-static void xscale_enable_breakpoints(struct target_s *target)
+static void xscale_enable_breakpoints(struct target *target)
 {
-       breakpoint_t *breakpoint = target->breakpoints;
+       struct breakpoint *breakpoint = target->breakpoints;
 
        /* set any pending breakpoints */
        while (breakpoint)
@@ -1186,12 +1162,12 @@ static void xscale_enable_breakpoints(struct target_s *target)
        }
 }
 
-static int xscale_resume(struct target_s *target, int current,
+static int xscale_resume(struct target *target, int current,
                uint32_t address, int handle_breakpoints, int debug_execution)
 {
-       struct xscale_common_s *xscale = target_to_xscale(target);
-       struct armv4_5_common_s *armv4_5 = &xscale->armv4_5_common;
-       breakpoint_t *breakpoint = target->breakpoints;
+       struct xscale_common *xscale = target_to_xscale(target);
+       struct arm *armv4_5 = &xscale->armv4_5_common;
+       struct breakpoint *breakpoint = target->breakpoints;
        uint32_t current_pc;
        int retval;
        int i;
@@ -1215,23 +1191,26 @@ static int xscale_resume(struct target_s *target, int current,
 
        /* current = 1: continue on current pc, otherwise continue at <address> */
        if (!current)
-               buf_set_u32(armv4_5->core_cache->reg_list[15].value, 0, 32, address);
+               buf_set_u32(armv4_5->pc->value, 0, 32, address);
 
-       current_pc = buf_get_u32(armv4_5->core_cache->reg_list[15].value, 0, 32);
+       current_pc = buf_get_u32(armv4_5->pc->value, 0, 32);
 
        /* if we're at the reset vector, we have to simulate the branch */
        if (current_pc == 0x0)
        {
                arm_simulate_step(target, NULL);
-               current_pc = buf_get_u32(armv4_5->core_cache->reg_list[15].value, 0, 32);
+               current_pc = buf_get_u32(armv4_5->pc->value, 0, 32);
        }
 
        /* the front-end may request us not to handle breakpoints */
        if (handle_breakpoints)
        {
-               if ((breakpoint = breakpoint_find(target, buf_get_u32(armv4_5->core_cache->reg_list[15].value, 0, 32))))
+               breakpoint = breakpoint_find(target,
+                               buf_get_u32(armv4_5->pc->value, 0, 32));
+               if (breakpoint != NULL)
                {
                        uint32_t next_pc;
+                       int saved_trace_buffer_enabled;
 
                        /* there's a breakpoint at the current PC, we have to step over it */
                        LOG_DEBUG("unset breakpoint at 0x%8.8" PRIx32 "", breakpoint->address);
@@ -1249,21 +1228,16 @@ static int xscale_resume(struct target_s *target, int current,
                        xscale_enable_single_step(target, next_pc);
 
                        /* restore banked registers */
-                       xscale_restore_context(target);
+                       retval = xscale_restore_banked(target);
 
-                       /* send resume request (command 0x30 or 0x31)
-                        * clean the trace buffer if it is to be enabled (0x62) */
-                       if (xscale->trace.buffer_enabled)
-                       {
-                               xscale_send_u32(target, 0x62);
-                               xscale_send_u32(target, 0x31);
-                       }
-                       else
-                               xscale_send_u32(target, 0x30);
+                       /* send resume request */
+                       xscale_send_u32(target, 0x30);
 
                        /* send CPSR */
-                       xscale_send_u32(target, buf_get_u32(armv4_5->core_cache->reg_list[ARMV4_5_CPSR].value, 0, 32));
-                       LOG_DEBUG("writing cpsr with value 0x%8.8" PRIx32 "", buf_get_u32(armv4_5->core_cache->reg_list[ARMV4_5_CPSR].value, 0, 32));
+                       xscale_send_u32(target,
+                               buf_get_u32(armv4_5->cpsr->value, 0, 32));
+                       LOG_DEBUG("writing cpsr with value 0x%8.8" PRIx32,
+                               buf_get_u32(armv4_5->cpsr->value, 0, 32));
 
                        for (i = 7; i >= 0; i--)
                        {
@@ -1273,12 +1247,21 @@ static int xscale_resume(struct target_s *target, int current,
                        }
 
                        /* send PC */
-                       xscale_send_u32(target, buf_get_u32(armv4_5->core_cache->reg_list[15].value, 0, 32));
-                       LOG_DEBUG("writing PC with value 0x%8.8" PRIx32 "", buf_get_u32(armv4_5->core_cache->reg_list[15].value, 0, 32));
+                       xscale_send_u32(target,
+                                       buf_get_u32(armv4_5->pc->value, 0, 32));
+                       LOG_DEBUG("writing PC with value 0x%8.8" PRIx32,
+                                       buf_get_u32(armv4_5->pc->value, 0, 32));
+
+                       /* disable trace data collection in xscale_debug_entry() */
+                       saved_trace_buffer_enabled = xscale->trace.buffer_enabled;
+                       xscale->trace.buffer_enabled = 0;
 
                        /* wait for and process debug entry */
                        xscale_debug_entry(target);
 
+                       /* re-enable trace buffer, if enabled previously */
+                       xscale->trace.buffer_enabled = saved_trace_buffer_enabled;
+
                        LOG_DEBUG("disable single-step");
                        xscale_disable_single_step(target);
 
@@ -1292,7 +1275,7 @@ static int xscale_resume(struct target_s *target, int current,
        xscale_enable_watchpoints(target);
 
        /* restore banked registers */
-       xscale_restore_context(target);
+       retval = xscale_restore_banked(target);
 
        /* send resume request (command 0x30 or 0x31)
         * clean the trace buffer if it is to be enabled (0x62) */
@@ -1305,8 +1288,9 @@ static int xscale_resume(struct target_s *target, int current,
                xscale_send_u32(target, 0x30);
 
        /* send CPSR */
-       xscale_send_u32(target, buf_get_u32(armv4_5->core_cache->reg_list[ARMV4_5_CPSR].value, 0, 32));
-       LOG_DEBUG("writing cpsr with value 0x%8.8" PRIx32 "", buf_get_u32(armv4_5->core_cache->reg_list[ARMV4_5_CPSR].value, 0, 32));
+       xscale_send_u32(target, buf_get_u32(armv4_5->cpsr->value, 0, 32));
+       LOG_DEBUG("writing cpsr with value 0x%8.8" PRIx32,
+                       buf_get_u32(armv4_5->cpsr->value, 0, 32));
 
        for (i = 7; i >= 0; i--)
        {
@@ -1316,15 +1300,16 @@ static int xscale_resume(struct target_s *target, int current,
        }
 
        /* send PC */
-       xscale_send_u32(target, buf_get_u32(armv4_5->core_cache->reg_list[15].value, 0, 32));
-       LOG_DEBUG("writing PC with value 0x%8.8" PRIx32 "", buf_get_u32(armv4_5->core_cache->reg_list[15].value, 0, 32));
+       xscale_send_u32(target, buf_get_u32(armv4_5->pc->value, 0, 32));
+       LOG_DEBUG("wrote PC with value 0x%8.8" PRIx32,
+                       buf_get_u32(armv4_5->pc->value, 0, 32));
 
        target->debug_reason = DBG_REASON_NOTHALTED;
 
        if (!debug_execution)
        {
                /* registers are now invalid */
-               armv4_5_invalidate_core_regs(target);
+               register_cache_invalidate(armv4_5->core_cache);
                target->state = TARGET_RUNNING;
                target_call_event_callbacks(target, TARGET_EVENT_RESUMED);
        }
@@ -1339,11 +1324,11 @@ static int xscale_resume(struct target_s *target, int current,
        return ERROR_OK;
 }
 
-static int xscale_step_inner(struct target_s *target, int current,
+static int xscale_step_inner(struct target *target, int current,
                uint32_t address, int handle_breakpoints)
 {
-       struct xscale_common_s *xscale = target_to_xscale(target);
-       struct armv4_5_common_s *armv4_5 = &xscale->armv4_5_common;
+       struct xscale_common *xscale = target_to_xscale(target);
+       struct arm *armv4_5 = &xscale->armv4_5_common;
        uint32_t next_pc;
        int retval;
        int i;
@@ -1354,7 +1339,7 @@ static int xscale_step_inner(struct target_s *target, int current,
        if ((retval = arm_simulate_step(target, &next_pc)) != ERROR_OK)
        {
                uint32_t current_opcode, current_pc;
-               current_pc = buf_get_u32(armv4_5->core_cache->reg_list[15].value, 0, 32);
+               current_pc = buf_get_u32(armv4_5->pc->value, 0, 32);
 
                target_read_u32(target, current_pc, &current_opcode);
                LOG_ERROR("BUG: couldn't calculate PC of next instruction, current opcode was 0x%8.8" PRIx32 "", current_opcode);
@@ -1366,7 +1351,7 @@ static int xscale_step_inner(struct target_s *target, int current,
                return retval;
 
        /* restore banked registers */
-       if ((retval = xscale_restore_context(target)) != ERROR_OK)
+       if ((retval = xscale_restore_banked(target)) != ERROR_OK)
                return retval;
 
        /* send resume request (command 0x30 or 0x31)
@@ -1383,9 +1368,12 @@ static int xscale_step_inner(struct target_s *target, int current,
                        return retval;
 
        /* send CPSR */
-       if ((retval = xscale_send_u32(target, buf_get_u32(armv4_5->core_cache->reg_list[ARMV4_5_CPSR].value, 0, 32))) != ERROR_OK)
+       retval = xscale_send_u32(target,
+                       buf_get_u32(armv4_5->cpsr->value, 0, 32));
+       if (retval != ERROR_OK)
                return retval;
-       LOG_DEBUG("writing cpsr with value 0x%8.8" PRIx32 "", buf_get_u32(armv4_5->core_cache->reg_list[ARMV4_5_CPSR].value, 0, 32));
+       LOG_DEBUG("writing cpsr with value 0x%8.8" PRIx32,
+                       buf_get_u32(armv4_5->cpsr->value, 0, 32));
 
        for (i = 7; i >= 0; i--)
        {
@@ -1396,15 +1384,17 @@ static int xscale_step_inner(struct target_s *target, int current,
        }
 
        /* send PC */
-       if ((retval = xscale_send_u32(target, buf_get_u32(armv4_5->core_cache->reg_list[15].value, 0, 32))) != ERROR_OK)
+       retval = xscale_send_u32(target,
+                       buf_get_u32(armv4_5->pc->value, 0, 32));
+       if (retval != ERROR_OK)
                return retval;
-       LOG_DEBUG("writing PC with value 0x%8.8" PRIx32, buf_get_u32(armv4_5->core_cache->reg_list[15].value, 0, 32));
+       LOG_DEBUG("wrote PC with value 0x%8.8" PRIx32,
+                       buf_get_u32(armv4_5->pc->value, 0, 32));
 
        target_call_event_callbacks(target, TARGET_EVENT_RESUMED);
 
        /* registers are now invalid */
-       if ((retval = armv4_5_invalidate_core_regs(target)) != ERROR_OK)
-               return retval;
+       register_cache_invalidate(armv4_5->core_cache);
 
        /* wait for and process debug entry */
        if ((retval = xscale_debug_entry(target)) != ERROR_OK)
@@ -1419,11 +1409,11 @@ static int xscale_step_inner(struct target_s *target, int current,
        return ERROR_OK;
 }
 
-static int xscale_step(struct target_s *target, int current,
+static int xscale_step(struct target *target, int current,
                uint32_t address, int handle_breakpoints)
 {
-       struct armv4_5_common_s *armv4_5 = target_to_armv4_5(target);
-       breakpoint_t *breakpoint = target->breakpoints;
+       struct arm *armv4_5 = target_to_arm(target);
+       struct breakpoint *breakpoint = NULL;
 
        uint32_t current_pc;
        int retval;
@@ -1436,16 +1426,16 @@ static int xscale_step(struct target_s *target, int current,
 
        /* current = 1: continue on current pc, otherwise continue at <address> */
        if (!current)
-               buf_set_u32(armv4_5->core_cache->reg_list[15].value, 0, 32, address);
+               buf_set_u32(armv4_5->pc->value, 0, 32, address);
 
-       current_pc = buf_get_u32(armv4_5->core_cache->reg_list[15].value, 0, 32);
+       current_pc = buf_get_u32(armv4_5->pc->value, 0, 32);
 
        /* if we're at the reset vector, we have to simulate the step */
        if (current_pc == 0x0)
        {
                if ((retval = arm_simulate_step(target, NULL)) != ERROR_OK)
                        return retval;
-               current_pc = buf_get_u32(armv4_5->core_cache->reg_list[15].value, 0, 32);
+               current_pc = buf_get_u32(armv4_5->pc->value, 0, 32);
 
                target->debug_reason = DBG_REASON_SINGLESTEP;
                target_call_event_callbacks(target, TARGET_EVENT_HALTED);
@@ -1455,11 +1445,13 @@ static int xscale_step(struct target_s *target, int current,
 
        /* the front-end may request us not to handle breakpoints */
        if (handle_breakpoints)
-               if ((breakpoint = breakpoint_find(target, buf_get_u32(armv4_5->core_cache->reg_list[15].value, 0, 32))))
-               {
-                       if ((retval = xscale_unset_breakpoint(target, breakpoint)) != ERROR_OK)
-                               return retval;
-               }
+               breakpoint = breakpoint_find(target,
+                               buf_get_u32(armv4_5->pc->value, 0, 32));
+       if (breakpoint != NULL) {
+               retval = xscale_unset_breakpoint(target, breakpoint);
+               if (retval != ERROR_OK)
+                       return retval;
+       }
 
        retval = xscale_step_inner(target, current, address, handle_breakpoints);
 
@@ -1474,9 +1466,9 @@ static int xscale_step(struct target_s *target, int current,
 
 }
 
-static int xscale_assert_reset(target_t *target)
+static int xscale_assert_reset(struct target *target)
 {
-       struct xscale_common_s *xscale = target_to_xscale(target);
+       struct xscale_common *xscale = target_to_xscale(target);
 
        LOG_DEBUG("target->state: %s",
                  target_state_name(target));
@@ -1484,8 +1476,9 @@ static int xscale_assert_reset(target_t *target)
        /* select DCSR instruction (set endstate to R-T-I to ensure we don't
         * end up in T-L-R, which would reset JTAG
         */
-       jtag_set_end_state(TAP_IDLE);
-       xscale_jtag_set_instr(target->tap, XSCALE_SELDCSR);
+       xscale_jtag_set_instr(target->tap,
+               XSCALE_SELDCSR << xscale->xscale_variant,
+               TAP_IDLE);
 
        /* set Hold reset, Halt mode and Trap Reset */
        buf_set_u32(xscale->reg_cache->reg_list[XSCALE_DCSR].value, 30, 1, 0x1);
@@ -1493,7 +1486,7 @@ static int xscale_assert_reset(target_t *target)
        xscale_write_dcsr(target, 1, 0);
 
        /* select BYPASS, because having DCSR selected caused problems on the PXA27x */
-       xscale_jtag_set_instr(target->tap, 0x7f);
+       xscale_jtag_set_instr(target->tap, ~0, TAP_IDLE);
        jtag_execute_queue();
 
        /* assert reset */
@@ -1515,10 +1508,10 @@ static int xscale_assert_reset(target_t *target)
        return ERROR_OK;
 }
 
-static int xscale_deassert_reset(target_t *target)
+static int xscale_deassert_reset(struct target *target)
 {
-       struct xscale_common_s *xscale = target_to_xscale(target);
-       breakpoint_t *breakpoint = target->breakpoints;
+       struct xscale_common *xscale = target_to_xscale(target);
+       struct breakpoint *breakpoint = target->breakpoints;
 
        LOG_DEBUG("-");
 
@@ -1540,7 +1533,7 @@ static int xscale_deassert_reset(target_t *target)
                breakpoint = breakpoint->next;
        }
 
-       armv4_5_invalidate_core_regs(target);
+       register_cache_invalidate(xscale->armv4_5_common.core_cache);
 
        /* FIXME mark hardware watchpoints got unset too.  Also,
         * at least some of the XScale registers are invalid...
@@ -1563,7 +1556,7 @@ static int xscale_deassert_reset(target_t *target)
                /* wait 300ms; 150 and 100ms were not enough */
                jtag_add_sleep(300*1000);
 
-               jtag_add_runtest(2030, jtag_set_end_state(TAP_IDLE));
+               jtag_add_runtest(2030, TAP_IDLE);
                jtag_execute_queue();
 
                /* set Hold reset, Halt mode and Trap Reset */
@@ -1620,7 +1613,7 @@ static int xscale_deassert_reset(target_t *target)
                if (retval != ERROR_OK)
                        return retval;
 
-               jtag_add_runtest(30, jtag_set_end_state(TAP_IDLE));
+               jtag_add_runtest(30, TAP_IDLE);
 
                jtag_add_sleep(100000);
 
@@ -1649,23 +1642,25 @@ static int xscale_deassert_reset(target_t *target)
        return ERROR_OK;
 }
 
-static int xscale_read_core_reg(struct target_s *target, int num,
-               enum armv4_5_mode mode)
+static int xscale_read_core_reg(struct target *target, struct reg *r,
+               int num, enum arm_mode mode)
 {
+       /** \todo add debug handler support for core register reads */
        LOG_ERROR("not implemented");
        return ERROR_OK;
 }
 
-static int xscale_write_core_reg(struct target_s *target, int num,
-               enum armv4_5_mode mode, uint32_t value)
+static int xscale_write_core_reg(struct target *target, struct reg *r,
+               int num, enum arm_mode mode, uint32_t value)
 {
+       /** \todo add debug handler support for core register writes */
        LOG_ERROR("not implemented");
        return ERROR_OK;
 }
 
-static int xscale_full_context(target_t *target)
+static int xscale_full_context(struct target *target)
 {
-       struct armv4_5_common_s *armv4_5 = target_to_armv4_5(target);
+       struct arm *armv4_5 = target_to_arm(target);
 
        uint32_t *buffer;
 
@@ -1687,50 +1682,56 @@ static int xscale_full_context(target_t *target)
         */
        for (i = 1; i < 7; i++)
        {
-               int valid = 1;
+               enum arm_mode mode = armv4_5_number_to_mode(i);
+               bool valid = true;
+               struct reg *r;
+
+               if (mode == ARM_MODE_USR)
+                       continue;
 
                /* check if there are invalid registers in the current mode
                 */
-               for (j = 0; j <= 16; j++)
+               for (j = 0; valid && j <= 16; j++)
                {
-                       if (ARMV4_5_CORE_REG_MODE(armv4_5->core_cache, armv4_5_number_to_mode(i), j).valid == 0)
-                               valid = 0;
+                       if (!ARMV4_5_CORE_REG_MODE(armv4_5->core_cache,
+                                       mode, j).valid)
+                               valid = false;
                }
+               if (valid)
+                       continue;
 
-               if (!valid)
-               {
-                       uint32_t tmp_cpsr;
+               /* request banked registers */
+               xscale_send_u32(target, 0x0);
 
-                       /* request banked registers */
-                       xscale_send_u32(target, 0x0);
+               /* send CPSR for desired bank mode */
+               xscale_send_u32(target, mode | 0xc0 /* I/F bits */);
 
-                       tmp_cpsr = 0x0;
-                       tmp_cpsr |= armv4_5_number_to_mode(i);
-                       tmp_cpsr |= 0xc0; /* I/F bits */
-
-                       /* send CPSR for desired mode */
-                       xscale_send_u32(target, tmp_cpsr);
+               /* get banked registers:  r8 to r14; and SPSR
+                * except in USR/SYS mode
+                */
+               if (mode != ARM_MODE_SYS) {
+                       /* SPSR */
+                       r = &ARMV4_5_CORE_REG_MODE(armv4_5->core_cache,
+                                       mode, 16);
+
+                       xscale_receive(target, buffer, 8);
+
+                       buf_set_u32(r->value, 0, 32, buffer[7]);
+                       r->dirty = false;
+                       r->valid = true;
+               } else {
+                       xscale_receive(target, buffer, 7);
+               }
 
-                       /* get banked registers, r8 to r14, and spsr if not in USR/SYS mode */
-                       if ((armv4_5_number_to_mode(i) != ARMV4_5_MODE_USR) && (armv4_5_number_to_mode(i) != ARMV4_5_MODE_SYS))
-                       {
-                               xscale_receive(target, buffer, 8);
-                               buf_set_u32(ARMV4_5_CORE_REG_MODE(armv4_5->core_cache, armv4_5->core_mode, 16).value, 0, 32, buffer[7]);
-                               ARMV4_5_CORE_REG_MODE(armv4_5->core_cache, armv4_5_number_to_mode(i), 16).dirty = 0;
-                               ARMV4_5_CORE_REG_MODE(armv4_5->core_cache, armv4_5_number_to_mode(i), 16).valid = 1;
-                       }
-                       else
-                       {
-                               xscale_receive(target, buffer, 7);
-                       }
+               /* move data from buffer to register cache */
+               for (j = 8; j <= 14; j++)
+               {
+                       r = &ARMV4_5_CORE_REG_MODE(armv4_5->core_cache,
+                                       mode, j);
 
-                       /* move data from buffer to register cache */
-                       for (j = 8; j <= 14; j++)
-                       {
-                               buf_set_u32(ARMV4_5_CORE_REG_MODE(armv4_5->core_cache, armv4_5_number_to_mode(i), j).value, 0, 32, buffer[j - 8]);
-                               ARMV4_5_CORE_REG_MODE(armv4_5->core_cache, armv4_5_number_to_mode(i), j).dirty = 0;
-                               ARMV4_5_CORE_REG_MODE(armv4_5->core_cache, armv4_5_number_to_mode(i), j).valid = 1;
-                       }
+                       buf_set_u32(r->value, 0, 32, buffer[j - 8]);
+                       r->dirty = false;
+                       r->valid = true;
                }
        }
 
@@ -1739,9 +1740,9 @@ static int xscale_full_context(target_t *target)
        return ERROR_OK;
 }
 
-static int xscale_restore_context(target_t *target)
+static int xscale_restore_banked(struct target *target)
 {
-       struct armv4_5_common_s *armv4_5 = target_to_armv4_5(target);
+       struct arm *armv4_5 = target_to_arm(target);
 
        int i, j;
 
@@ -1752,64 +1753,70 @@ static int xscale_restore_context(target_t *target)
        }
 
        /* iterate through processor modes (FIQ, IRQ, SVC, ABT, UND and SYS)
-       * we can't enter User mode on an XScale (unpredictable),
-       * but User shares registers with SYS
-       */
+        * and check if any banked registers need to be written.  Ignore
+        * USR mode (number 0) in favor of SYS; we can't enter User mode on
+        * an XScale (unpredictable), but they share all registers.
+        */
        for (i = 1; i < 7; i++)
        {
-               int dirty = 0;
+               enum arm_mode mode = armv4_5_number_to_mode(i);
+               struct reg *r;
 
-               /* check if there are invalid registers in the current mode
-               */
+               if (mode == ARM_MODE_USR)
+                       continue;
+
+               /* check if there are dirty registers in this mode */
                for (j = 8; j <= 14; j++)
                {
-                       if (ARMV4_5_CORE_REG_MODE(armv4_5->core_cache, armv4_5_number_to_mode(i), j).dirty == 1)
-                               dirty = 1;
+                       if (ARMV4_5_CORE_REG_MODE(armv4_5->core_cache,
+                                       mode, j).dirty)
+                               goto dirty;
                }
 
                /* if not USR/SYS, check if the SPSR needs to be written */
-               if ((armv4_5_number_to_mode(i) != ARMV4_5_MODE_USR) && (armv4_5_number_to_mode(i) != ARMV4_5_MODE_SYS))
+               if (mode != ARM_MODE_SYS)
                {
-                       if (ARMV4_5_CORE_REG_MODE(armv4_5->core_cache, armv4_5_number_to_mode(i), 16).dirty == 1)
-                               dirty = 1;
+                       if (ARMV4_5_CORE_REG_MODE(armv4_5->core_cache,
+                                       mode, 16).dirty)
+                               goto dirty;
                }
 
-               if (dirty)
-               {
-                       uint32_t tmp_cpsr;
+               /* there's nothing to flush for this mode */
+               continue;
 
-                       /* send banked registers */
-                       xscale_send_u32(target, 0x1);
+dirty:
+               /* command 0x1:  "send banked registers" */
+               xscale_send_u32(target, 0x1);
 
-                       tmp_cpsr = 0x0;
-                       tmp_cpsr |= armv4_5_number_to_mode(i);
-                       tmp_cpsr |= 0xc0; /* I/F bits */
+               /* send CPSR for desired mode */
+               xscale_send_u32(target, mode | 0xc0 /* I/F bits */);
 
-                       /* send CPSR for desired mode */
-                       xscale_send_u32(target, tmp_cpsr);
-
-                       /* send banked registers, r8 to r14, and spsr if not in USR/SYS mode */
-                       for (j = 8; j <= 14; j++)
-                       {
-                               xscale_send_u32(target, buf_get_u32(ARMV4_5_CORE_REG_MODE(armv4_5->core_cache, armv4_5->core_mode, j).value, 0, 32));
-                               ARMV4_5_CORE_REG_MODE(armv4_5->core_cache, armv4_5_number_to_mode(i), j).dirty = 0;
-                       }
+               /* send r8 to r14/lr ... only FIQ needs more than r13..r14,
+                * but this protocol doesn't understand that nuance.
+                */
+               for (j = 8; j <= 14; j++) {
+                       r = &ARMV4_5_CORE_REG_MODE(armv4_5->core_cache,
+                                       mode, j);
+                       xscale_send_u32(target, buf_get_u32(r->value, 0, 32));
+                       r->dirty = false;
+               }
 
-                       if ((armv4_5_number_to_mode(i) != ARMV4_5_MODE_USR) && (armv4_5_number_to_mode(i) != ARMV4_5_MODE_SYS))
-                       {
-                               xscale_send_u32(target, buf_get_u32(ARMV4_5_CORE_REG_MODE(armv4_5->core_cache, armv4_5->core_mode, 16).value, 0, 32));
-                               ARMV4_5_CORE_REG_MODE(armv4_5->core_cache, armv4_5_number_to_mode(i), 16).dirty = 0;
-                       }
+               /* send spsr if not in USR/SYS mode */
+               if (mode != ARM_MODE_SYS) {
+                       r = &ARMV4_5_CORE_REG_MODE(armv4_5->core_cache,
+                                       mode, 16);
+                       xscale_send_u32(target, buf_get_u32(r->value, 0, 32));
+                       r->dirty = false;
                }
        }
 
        return ERROR_OK;
 }
 
-static int xscale_read_memory(struct target_s *target, uint32_t address,
+static int xscale_read_memory(struct target *target, uint32_t address,
                uint32_t size, uint32_t count, uint8_t *buffer)
 {
-       struct xscale_common_s *xscale = target_to_xscale(target);
+       struct xscale_common *xscale = target_to_xscale(target);
        uint32_t *buf32;
        uint32_t i;
        int retval;
@@ -1863,8 +1870,8 @@ static int xscale_read_memory(struct target_s *target, uint32_t address,
                                *buffer++ = buf32[i] & 0xff;
                                break;
                        default:
-                               LOG_ERROR("should never get here");
-                               exit(-1);
+                               LOG_ERROR("invalid read size");
+                               return ERROR_INVALID_ARGUMENTS;
                }
        }
 
@@ -1885,10 +1892,25 @@ static int xscale_read_memory(struct target_s *target, uint32_t address,
        return ERROR_OK;
 }
 
-static int xscale_write_memory(struct target_s *target, uint32_t address,
+static int xscale_read_phys_memory(struct target *target, uint32_t address,
                uint32_t size, uint32_t count, uint8_t *buffer)
 {
-       struct xscale_common_s *xscale = target_to_xscale(target);
+       struct xscale_common *xscale = target_to_xscale(target);
+
+       /* with MMU inactive, there are only physical addresses */
+       if (!xscale->armv4_5_mmu.mmu_enabled)
+               return xscale_read_memory(target, address, size, count, buffer);
+
+       /** \todo: provide a non-stub implementation of this routine. */
+       LOG_ERROR("%s: %s is not implemented.  Disable MMU?",
+                       target_name(target), __func__);
+       return ERROR_FAIL;
+}
+
+static int xscale_write_memory(struct target *target, uint32_t address,
+               uint32_t size, uint32_t count, uint8_t *buffer)
+{
+       struct xscale_common *xscale = target_to_xscale(target);
        int retval;
 
        LOG_DEBUG("address: 0x%8.8" PRIx32 ", size: 0x%8.8" PRIx32 ", count: 0x%8.8" PRIx32, address, size, count);
@@ -1957,37 +1979,61 @@ static int xscale_write_memory(struct target_s *target, uint32_t address,
                if ((retval = xscale_send_u32(target, 0x60)) != ERROR_OK)
                        return retval;
 
+               LOG_ERROR("data abort writing memory");
                return ERROR_TARGET_DATA_ABORT;
        }
 
        return ERROR_OK;
 }
 
-static int xscale_bulk_write_memory(target_t *target, uint32_t address,
+static int xscale_write_phys_memory(struct target *target, uint32_t address,
+               uint32_t size, uint32_t count, uint8_t *buffer)
+{
+       struct xscale_common *xscale = target_to_xscale(target);
+
+       /* with MMU inactive, there are only physical addresses */
+       if (!xscale->armv4_5_mmu.mmu_enabled)
+               return xscale_read_memory(target, address, size, count, buffer);
+
+       /** \todo: provide a non-stub implementation of this routine. */
+       LOG_ERROR("%s: %s is not implemented.  Disable MMU?",
+                       target_name(target), __func__);
+       return ERROR_FAIL;
+}
+
+static int xscale_bulk_write_memory(struct target *target, uint32_t address,
                uint32_t count, uint8_t *buffer)
 {
        return xscale_write_memory(target, address, 4, count, buffer);
 }
 
-static uint32_t xscale_get_ttb(target_t *target)
+static int xscale_get_ttb(struct target *target, uint32_t *result)
 {
-       struct xscale_common_s *xscale = target_to_xscale(target);
+       struct xscale_common *xscale = target_to_xscale(target);
        uint32_t ttb;
+       int retval;
 
-       xscale_get_reg(&xscale->reg_cache->reg_list[XSCALE_TTB]);
+       retval = xscale_get_reg(&xscale->reg_cache->reg_list[XSCALE_TTB]);
+       if (retval != ERROR_OK)
+               return retval;
        ttb = buf_get_u32(xscale->reg_cache->reg_list[XSCALE_TTB].value, 0, 32);
 
-       return ttb;
+       *result = ttb;
+
+       return ERROR_OK;
 }
 
-static void xscale_disable_mmu_caches(target_t *target, int mmu,
+static int xscale_disable_mmu_caches(struct target *target, int mmu,
                int d_u_cache, int i_cache)
 {
-       struct xscale_common_s *xscale = target_to_xscale(target);
+       struct xscale_common *xscale = target_to_xscale(target);
        uint32_t cp15_control;
+       int retval;
 
        /* read cp15 control register */
-       xscale_get_reg(&xscale->reg_cache->reg_list[XSCALE_CTRL]);
+       retval = xscale_get_reg(&xscale->reg_cache->reg_list[XSCALE_CTRL]);
+       if (retval !=ERROR_OK)
+               return retval;
        cp15_control = buf_get_u32(xscale->reg_cache->reg_list[XSCALE_CTRL].value, 0, 32);
 
        if (mmu)
@@ -1996,11 +2042,17 @@ static void xscale_disable_mmu_caches(target_t *target, int mmu,
        if (d_u_cache)
        {
                /* clean DCache */
-               xscale_send_u32(target, 0x50);
-               xscale_send_u32(target, xscale->cache_clean_address);
+               retval = xscale_send_u32(target, 0x50);
+               if (retval !=ERROR_OK)
+                       return retval;
+               retval = xscale_send_u32(target, xscale->cache_clean_address);
+               if (retval !=ERROR_OK)
+                       return retval;
 
                /* invalidate DCache */
-               xscale_send_u32(target, 0x51);
+               retval = xscale_send_u32(target, 0x51);
+               if (retval !=ERROR_OK)
+                       return retval;
 
                cp15_control &= ~0x4U;
        }
@@ -2008,25 +2060,33 @@ static void xscale_disable_mmu_caches(target_t *target, int mmu,
        if (i_cache)
        {
                /* invalidate ICache */
-               xscale_send_u32(target, 0x52);
+               retval = xscale_send_u32(target, 0x52);
+               if (retval !=ERROR_OK)
+                       return retval;
                cp15_control &= ~0x1000U;
        }
 
        /* write new cp15 control register */
-       xscale_set_reg_u32(&xscale->reg_cache->reg_list[XSCALE_CTRL], cp15_control);
+       retval = xscale_set_reg_u32(&xscale->reg_cache->reg_list[XSCALE_CTRL], cp15_control);
+       if (retval !=ERROR_OK)
+               return retval;
 
        /* execute cpwait to ensure outstanding operations complete */
-       xscale_send_u32(target, 0x53);
+       retval = xscale_send_u32(target, 0x53);
+       return retval;
 }
 
-static void xscale_enable_mmu_caches(target_t *target, int mmu,
+static int xscale_enable_mmu_caches(struct target *target, int mmu,
                int d_u_cache, int i_cache)
 {
-       struct xscale_common_s *xscale = target_to_xscale(target);
+       struct xscale_common *xscale = target_to_xscale(target);
        uint32_t cp15_control;
+       int retval;
 
        /* read cp15 control register */
-       xscale_get_reg(&xscale->reg_cache->reg_list[XSCALE_CTRL]);
+       retval = xscale_get_reg(&xscale->reg_cache->reg_list[XSCALE_CTRL]);
+       if (retval !=ERROR_OK)
+               return retval;
        cp15_control = buf_get_u32(xscale->reg_cache->reg_list[XSCALE_CTRL].value, 0, 32);
 
        if (mmu)
@@ -2039,17 +2099,20 @@ static void xscale_enable_mmu_caches(target_t *target, int mmu,
                cp15_control |= 0x1000U;
 
        /* write new cp15 control register */
-       xscale_set_reg_u32(&xscale->reg_cache->reg_list[XSCALE_CTRL], cp15_control);
+       retval = xscale_set_reg_u32(&xscale->reg_cache->reg_list[XSCALE_CTRL], cp15_control);
+       if (retval !=ERROR_OK)
+               return retval;
 
        /* execute cpwait to ensure outstanding operations complete */
-       xscale_send_u32(target, 0x53);
+       retval = xscale_send_u32(target, 0x53);
+       return retval;
 }
 
-static int xscale_set_breakpoint(struct target_s *target,
-               breakpoint_t *breakpoint)
+static int xscale_set_breakpoint(struct target *target,
+               struct breakpoint *breakpoint)
 {
        int retval;
-       struct xscale_common_s *xscale = target_to_xscale(target);
+       struct xscale_common *xscale = target_to_xscale(target);
 
        if (target->state != TARGET_HALTED)
        {
@@ -2079,9 +2142,9 @@ static int xscale_set_breakpoint(struct target_s *target,
                        breakpoint->set = 2;    /* breakpoint set on second breakpoint register */
                }
                else
-               {
+               {       /* bug: availability previously verified in xscale_add_breakpoint() */
                        LOG_ERROR("BUG: no hardware comparator available");
-                       return ERROR_OK;
+                       return ERROR_TARGET_RESOURCE_NOT_AVAILABLE;
                }
        }
        else if (breakpoint->type == BKPT_SOFT)
@@ -2093,7 +2156,7 @@ static int xscale_set_breakpoint(struct target_s *target,
                        {
                                return retval;
                        }
-                       /* write the original instruction in target endianness (arm7_9->arm_bkpt is host endian) */
+                       /* write the bkpt instruction in target endianness (arm7_9->arm_bkpt is host endian) */
                        if ((retval = target_write_u32(target, breakpoint->address, xscale->arm_bkpt)) != ERROR_OK)
                        {
                                return retval;
@@ -2106,38 +2169,37 @@ static int xscale_set_breakpoint(struct target_s *target,
                        {
                                return retval;
                        }
-                       /* write the original instruction in target endianness (arm7_9->arm_bkpt is host endian) */
-                       if ((retval = target_write_u32(target, breakpoint->address, xscale->thumb_bkpt)) != ERROR_OK)
+                       /* write the bkpt instruction in target endianness (arm7_9->arm_bkpt is host endian) */
+                       if ((retval = target_write_u16(target, breakpoint->address, xscale->thumb_bkpt)) != ERROR_OK)
                        {
                                return retval;
                        }
                }
                breakpoint->set = 1;
+
+               xscale_send_u32(target, 0x50);   /* clean dcache */
+               xscale_send_u32(target, xscale->cache_clean_address);
+               xscale_send_u32(target, 0x51);   /* invalidate dcache */
+               xscale_send_u32(target, 0x52);   /* invalidate icache and flush fetch buffers */
        }
 
        return ERROR_OK;
 }
 
-static int xscale_add_breakpoint(struct target_s *target,
-               breakpoint_t *breakpoint)
+static int xscale_add_breakpoint(struct target *target,
+               struct breakpoint *breakpoint)
 {
-       struct xscale_common_s *xscale = target_to_xscale(target);
-
-       if (target->state != TARGET_HALTED)
-       {
-               LOG_WARNING("target not halted");
-               return ERROR_TARGET_NOT_HALTED;
-       }
+       struct xscale_common *xscale = target_to_xscale(target);
 
        if ((breakpoint->type == BKPT_HARD) && (xscale->ibcr_available < 1))
        {
-               LOG_INFO("no breakpoint unit available for hardware breakpoint");
+               LOG_ERROR("no breakpoint unit available for hardware breakpoint");
                return ERROR_TARGET_RESOURCE_NOT_AVAILABLE;
        }
 
        if ((breakpoint->length != 2) && (breakpoint->length != 4))
        {
-               LOG_INFO("only breakpoints of two (Thumb) or four (ARM) bytes length supported");
+               LOG_ERROR("only breakpoints of two (Thumb) or four (ARM) bytes length supported");
                return ERROR_TARGET_RESOURCE_NOT_AVAILABLE;
        }
 
@@ -2146,14 +2208,14 @@ static int xscale_add_breakpoint(struct target_s *target,
                xscale->ibcr_available--;
        }
 
-       return ERROR_OK;
+       return xscale_set_breakpoint(target, breakpoint);
 }
 
-static int xscale_unset_breakpoint(struct target_s *target,
-               breakpoint_t *breakpoint)
+static int xscale_unset_breakpoint(struct target *target,
+               struct breakpoint *breakpoint)
 {
        int retval;
-       struct xscale_common_s *xscale = target_to_xscale(target);
+       struct xscale_common *xscale = target_to_xscale(target);
 
        if (target->state != TARGET_HALTED)
        {
@@ -2199,18 +2261,23 @@ static int xscale_unset_breakpoint(struct target_s *target,
                        }
                }
                breakpoint->set = 0;
+
+               xscale_send_u32(target, 0x50);   /* clean dcache */
+               xscale_send_u32(target, xscale->cache_clean_address);
+               xscale_send_u32(target, 0x51);   /* invalidate dcache */
+               xscale_send_u32(target, 0x52);   /* invalidate icache and flush fetch buffers */
        }
 
        return ERROR_OK;
 }
 
-static int xscale_remove_breakpoint(struct target_s *target, breakpoint_t *breakpoint)
+static int xscale_remove_breakpoint(struct target *target, struct breakpoint *breakpoint)
 {
-       struct xscale_common_s *xscale = target_to_xscale(target);
+       struct xscale_common *xscale = target_to_xscale(target);
 
        if (target->state != TARGET_HALTED)
        {
-               LOG_WARNING("target not halted");
+               LOG_ERROR("target not halted");
                return ERROR_TARGET_NOT_HALTED;
        }
 
@@ -2225,22 +2292,20 @@ static int xscale_remove_breakpoint(struct target_s *target, breakpoint_t *break
        return ERROR_OK;
 }
 
-static int xscale_set_watchpoint(struct target_s *target,
-               watchpoint_t *watchpoint)
+static int xscale_set_watchpoint(struct target *target,
+               struct watchpoint *watchpoint)
 {
-       struct xscale_common_s *xscale = target_to_xscale(target);
-       uint8_t enable = 0;
-       reg_t *dbcon = &xscale->reg_cache->reg_list[XSCALE_DBCON];
+       struct xscale_common *xscale = target_to_xscale(target);
+       uint32_t enable = 0;
+       struct reg *dbcon = &xscale->reg_cache->reg_list[XSCALE_DBCON];
        uint32_t dbcon_value = buf_get_u32(dbcon->value, 0, 32);
 
        if (target->state != TARGET_HALTED)
        {
-               LOG_WARNING("target not halted");
+               LOG_ERROR("target not halted");
                return ERROR_TARGET_NOT_HALTED;
        }
 
-       xscale_get_reg(dbcon);
-
        switch (watchpoint->rw)
        {
                case WPT_READ:
@@ -2256,6 +2321,24 @@ static int xscale_set_watchpoint(struct target_s *target,
                        LOG_ERROR("BUG: watchpoint->rw neither read, write nor access");
        }
 
+       /* For watchpoint across more than one word, both DBR registers must
+          be enlisted, with the second used as a mask. */
+       if (watchpoint->length > 4)
+       {
+          if (xscale->dbr0_used || xscale->dbr1_used) 
+          {
+                 LOG_ERROR("BUG: sufficient hardware comparators unavailable");
+                 return ERROR_TARGET_RESOURCE_NOT_AVAILABLE;
+          }
+
+          /* Write mask value to DBR1, based on the length argument.
+               * Address bits ignored by the comparator are those set in mask. */
+          xscale_set_reg_u32(&xscale->reg_cache->reg_list[XSCALE_DBR1],
+                                                 watchpoint->length - 1);
+          xscale->dbr1_used = 1;
+          enable |= 0x100;                     /* DBCON[M] */
+       }
+
        if (!xscale->dbr0_used)
        {
                xscale_set_reg_u32(&xscale->reg_cache->reg_list[XSCALE_DBR0], watchpoint->address);
@@ -2275,43 +2358,65 @@ static int xscale_set_watchpoint(struct target_s *target,
        else
        {
                LOG_ERROR("BUG: no hardware comparator available");
-               return ERROR_OK;
+               return ERROR_TARGET_RESOURCE_NOT_AVAILABLE;
        }
 
        return ERROR_OK;
 }
 
-static int xscale_add_watchpoint(struct target_s *target,
-               watchpoint_t *watchpoint)
+static int xscale_add_watchpoint(struct target *target,
+               struct watchpoint *watchpoint)
 {
-       struct xscale_common_s *xscale = target_to_xscale(target);
+       struct xscale_common *xscale = target_to_xscale(target);
 
-       if (target->state != TARGET_HALTED)
+       if (xscale->dbr_available < 1)
        {
-               LOG_WARNING("target not halted");
-               return ERROR_TARGET_NOT_HALTED;
+          LOG_ERROR("no more watchpoint registers available");
+          return ERROR_TARGET_RESOURCE_NOT_AVAILABLE;
        }
 
-       if (xscale->dbr_available < 1)
+       if (watchpoint->value)
+          LOG_WARNING("xscale does not support value, mask arguments; ignoring");
+
+       /* check that length is a power of two */
+       for (uint32_t len = watchpoint->length; len != 1; len /= 2)
        {
-               return ERROR_TARGET_RESOURCE_NOT_AVAILABLE;
+          if (len % 2)
+          {
+                 LOG_ERROR("xscale requires that watchpoint length is a power of two");
+                 return ERROR_COMMAND_ARGUMENT_INVALID;
+          }
        }
 
-       if ((watchpoint->length != 1) && (watchpoint->length != 2) && (watchpoint->length != 4))
+       if (watchpoint->length == 4) /* single word watchpoint */
        {
-               return ERROR_TARGET_RESOURCE_NOT_AVAILABLE;
+          xscale->dbr_available--; /* one DBR reg used */
+          return ERROR_OK;
        }
 
-       xscale->dbr_available--;
-
+       /* watchpoints across multiple words require both DBR registers */
+       if (xscale->dbr_available < 2)
+       {
+          LOG_ERROR("insufficient watchpoint registers available");
+          return ERROR_TARGET_RESOURCE_NOT_AVAILABLE;
+       }
+       
+       if (watchpoint->length > watchpoint->address)
+       {
+          LOG_ERROR("xscale does not support watchpoints with length "
+                                "greater than address");
+          return ERROR_COMMAND_ARGUMENT_INVALID;
+       }
+          
+       xscale->dbr_available = 0;
        return ERROR_OK;
 }
 
-static int xscale_unset_watchpoint(struct target_s *target,
-               watchpoint_t *watchpoint)
+static int xscale_unset_watchpoint(struct target *target,
+               struct watchpoint *watchpoint)
 {
-       struct xscale_common_s *xscale = target_to_xscale(target);
-       reg_t *dbcon = &xscale->reg_cache->reg_list[XSCALE_DBCON];
+       struct xscale_common *xscale = target_to_xscale(target);
+       struct reg *dbcon = &xscale->reg_cache->reg_list[XSCALE_DBCON];
        uint32_t dbcon_value = buf_get_u32(dbcon->value, 0, 32);
 
        if (target->state != TARGET_HALTED)
@@ -2328,7 +2433,14 @@ static int xscale_unset_watchpoint(struct target_s *target,
 
        if (watchpoint->set == 1)
        {
-               dbcon_value &= ~0x3;
+          if (watchpoint->length > 4)
+          {
+                 dbcon_value &= ~0x103; /* clear DBCON[M] as well */
+                 xscale->dbr1_used = 0; /* DBR1 was used for mask */
+          }
+          else
+                 dbcon_value &= ~0x3;
+
                xscale_set_reg_u32(dbcon, dbcon_value);
                xscale->dbr0_used = 0;
        }
@@ -2343,13 +2455,13 @@ static int xscale_unset_watchpoint(struct target_s *target,
        return ERROR_OK;
 }
 
-static int xscale_remove_watchpoint(struct target_s *target, watchpoint_t *watchpoint)
+static int xscale_remove_watchpoint(struct target *target, struct watchpoint *watchpoint)
 {
-       struct xscale_common_s *xscale = target_to_xscale(target);
+       struct xscale_common *xscale = target_to_xscale(target);
 
        if (target->state != TARGET_HALTED)
        {
-               LOG_WARNING("target not halted");
+               LOG_ERROR("target not halted");
                return ERROR_TARGET_NOT_HALTED;
        }
 
@@ -2358,16 +2470,19 @@ static int xscale_remove_watchpoint(struct target_s *target, watchpoint_t *watch
                xscale_unset_watchpoint(target, watchpoint);
        }
 
+       if (watchpoint->length > 4)
+          xscale->dbr_available++;     /* both DBR regs now available */
+       
        xscale->dbr_available++;
 
        return ERROR_OK;
 }
 
-static int xscale_get_reg(reg_t *reg)
+static int xscale_get_reg(struct reg *reg)
 {
-       xscale_reg_t *arch_info = reg->arch_info;
-       target_t *target = arch_info->target;
-       struct xscale_common_s *xscale = target_to_xscale(target);
+       struct xscale_reg *arch_info = reg->arch_info;
+       struct target *target = arch_info->target;
+       struct xscale_common *xscale = target_to_xscale(target);
 
        /* DCSR, TX and RX are accessible via JTAG */
        if (strcmp(reg->name, "XSCALE_DCSR") == 0)
@@ -2408,11 +2523,11 @@ static int xscale_get_reg(reg_t *reg)
        return ERROR_OK;
 }
 
-static int xscale_set_reg(reg_t *reg, uint8_t* buf)
+static int xscale_set_reg(struct reg *reg, uint8_t* buf)
 {
-       xscale_reg_t *arch_info = reg->arch_info;
-       target_t *target = arch_info->target;
-       struct xscale_common_s *xscale = target_to_xscale(target);
+       struct xscale_reg *arch_info = reg->arch_info;
+       struct target *target = arch_info->target;
+       struct xscale_common *xscale = target_to_xscale(target);
        uint32_t value = buf_get_u32(buf, 0, 32);
 
        /* DCSR, TX and RX are accessible via JTAG */
@@ -2452,11 +2567,11 @@ static int xscale_set_reg(reg_t *reg, uint8_t* buf)
        return ERROR_OK;
 }
 
-static int xscale_write_dcsr_sw(target_t *target, uint32_t value)
+static int xscale_write_dcsr_sw(struct target *target, uint32_t value)
 {
-       struct xscale_common_s *xscale = target_to_xscale(target);
-       reg_t *dcsr = &xscale->reg_cache->reg_list[XSCALE_DCSR];
-       xscale_reg_t *dcsr_arch_info = dcsr->arch_info;
+       struct xscale_common *xscale = target_to_xscale(target);
+       struct reg *dcsr = &xscale->reg_cache->reg_list[XSCALE_DCSR];
+       struct xscale_reg *dcsr_arch_info = dcsr->arch_info;
 
        /* send CP write request (command 0x41) */
        xscale_send_u32(target, 0x41);
@@ -2471,11 +2586,11 @@ static int xscale_write_dcsr_sw(target_t *target, uint32_t value)
        return ERROR_OK;
 }
 
-static int xscale_read_trace(target_t *target)
+static int xscale_read_trace(struct target *target)
 {
-       struct xscale_common_s *xscale = target_to_xscale(target);
-       struct armv4_5_common_s *armv4_5 = &xscale->armv4_5_common;
-       xscale_trace_data_t **trace_data_p;
+       struct xscale_common *xscale = target_to_xscale(target);
+       struct arm *armv4_5 = &xscale->armv4_5_common;
+       struct xscale_trace_data **trace_data_p;
 
        /* 258 words from debug handler
         * 256 trace buffer entries
@@ -2484,6 +2599,7 @@ static int xscale_read_trace(target_t *target)
        uint32_t trace_buffer[258];
        int is_address[256];
        int i, j;
+       unsigned int num_checkpoints = 0;
 
        if (target->state != TARGET_HALTED)
        {
@@ -2500,23 +2616,27 @@ static int xscale_read_trace(target_t *target)
        /* parse buffer backwards to identify address entries */
        for (i = 255; i >= 0; i--)
        {
+               /* also count number of checkpointed entries */
+               if ((trace_buffer[i] & 0xe0) == 0xc0)
+                       num_checkpoints++;
+
                is_address[i] = 0;
                if (((trace_buffer[i] & 0xf0) == 0x90) ||
                        ((trace_buffer[i] & 0xf0) == 0xd0))
                {
-                       if (i >= 3)
+                       if (i > 0)
                                is_address[--i] = 1;
-                       if (i >= 2)
+                       if (i > 0)
                                is_address[--i] = 1;
-                       if (i >= 1)
+                       if (i > 0)
                                is_address[--i] = 1;
-                       if (i >= 0)
+                       if (i > 0)
                                is_address[--i] = 1;
                }
        }
 
 
-       /* search first non-zero entry */
+       /* search first non-zero entry that is not part of an address */
        for (j = 0; (j < 256) && (trace_buffer[j] == 0) && (!is_address[j]); j++)
                ;
 
@@ -2526,16 +2646,34 @@ static int xscale_read_trace(target_t *target)
                return ERROR_XSCALE_NO_TRACE_DATA;
        }
 
+       /* account for possible partial address at buffer start (wrap mode only) */
+       if (is_address[0])
+       {       /* first entry is address; complete set of 4? */
+               i = 1;
+               while (i < 4)
+                       if (!is_address[i++])
+                               break;
+               if (i < 4)
+                       j += i;   /* partial address; can't use it */
+       }
+
+       /* if first valid entry is indirect branch, can't use that either (no address) */
+       if (((trace_buffer[j] & 0xf0) == 0x90) || ((trace_buffer[j] & 0xf0) == 0xd0))
+               j++;
+
+       /* walk linked list to terminating entry */
        for (trace_data_p = &xscale->trace.data; *trace_data_p; trace_data_p = &(*trace_data_p)->next)
                ;
 
-       *trace_data_p = malloc(sizeof(xscale_trace_data_t));
+       *trace_data_p = malloc(sizeof(struct xscale_trace_data));
        (*trace_data_p)->next = NULL;
        (*trace_data_p)->chkpt0 = trace_buffer[256];
        (*trace_data_p)->chkpt1 = trace_buffer[257];
-       (*trace_data_p)->last_instruction = buf_get_u32(armv4_5->core_cache->reg_list[15].value, 0, 32);
-       (*trace_data_p)->entries = malloc(sizeof(xscale_trace_entry_t) * (256 - j));
+       (*trace_data_p)->last_instruction =
+                       buf_get_u32(armv4_5->pc->value, 0, 32);
+       (*trace_data_p)->entries = malloc(sizeof(struct xscale_trace_entry) * (256 - j));
        (*trace_data_p)->depth = 256 - j;
+       (*trace_data_p)->num_checkpoints = num_checkpoints;
 
        for (i = j; i < 256; i++)
        {
@@ -2549,13 +2687,13 @@ static int xscale_read_trace(target_t *target)
        return ERROR_OK;
 }
 
-static int xscale_read_instruction(target_t *target,
-               arm_instruction_t *instruction)
+static int xscale_read_instruction(struct target *target, uint32_t pc,
+                                                                  struct arm_instruction *instruction)
 {
-       struct xscale_common_s *xscale = target_to_xscale(target);
+       struct xscale_common *const xscale = target_to_xscale(target);
        int i;
        int section = -1;
-       uint32_t size_read;
+       size_t size_read;
        uint32_t opcode;
        int retval;
 
@@ -2565,8 +2703,8 @@ static int xscale_read_instruction(target_t *target,
        /* search for the section the current instruction belongs to */
        for (i = 0; i < xscale->trace.image->num_sections; i++)
        {
-               if ((xscale->trace.image->sections[i].base_address <= xscale->trace.current_pc) &&
-                       (xscale->trace.image->sections[i].base_address + xscale->trace.image->sections[i].size > xscale->trace.current_pc))
+               if ((xscale->trace.image->sections[i].base_address <= pc) &&
+                       (xscale->trace.image->sections[i].base_address + xscale->trace.image->sections[i].size > pc))
                {
                        section = i;
                        break;
@@ -2579,31 +2717,31 @@ static int xscale_read_instruction(target_t *target,
                return ERROR_TRACE_INSTRUCTION_UNAVAILABLE;
        }
 
-       if (xscale->trace.core_state == ARMV4_5_STATE_ARM)
+       if (xscale->trace.core_state == ARM_STATE_ARM)
        {
                uint8_t buf[4];
                if ((retval = image_read_section(xscale->trace.image, section,
-                       xscale->trace.current_pc - xscale->trace.image->sections[section].base_address,
+                       pc - xscale->trace.image->sections[section].base_address,
                        4, buf, &size_read)) != ERROR_OK)
                {
                        LOG_ERROR("error while reading instruction: %i", retval);
                        return ERROR_TRACE_INSTRUCTION_UNAVAILABLE;
                }
                opcode = target_buffer_get_u32(target, buf);
-               arm_evaluate_opcode(opcode, xscale->trace.current_pc, instruction);
+               arm_evaluate_opcode(opcode, pc, instruction);
        }
-       else if (xscale->trace.core_state == ARMV4_5_STATE_THUMB)
+       else if (xscale->trace.core_state == ARM_STATE_THUMB)
        {
                uint8_t buf[2];
                if ((retval = image_read_section(xscale->trace.image, section,
-                       xscale->trace.current_pc - xscale->trace.image->sections[section].base_address,
+                       pc - xscale->trace.image->sections[section].base_address,
                        2, buf, &size_read)) != ERROR_OK)
                {
                        LOG_ERROR("error while reading instruction: %i", retval);
                        return ERROR_TRACE_INSTRUCTION_UNAVAILABLE;
                }
                opcode = target_buffer_get_u16(target, buf);
-               thumb_evaluate_opcode(opcode, xscale->trace.current_pc, instruction);
+               thumb_evaluate_opcode(opcode, pc, instruction);
        }
        else
        {
@@ -2614,232 +2752,259 @@ static int xscale_read_instruction(target_t *target,
        return ERROR_OK;
 }
 
-static int xscale_branch_address(xscale_trace_data_t *trace_data,
-               int i, uint32_t *target)
+/* Extract address encoded into trace data. 
+ * Write result to address referenced by argument 'target', or 0 if incomplete.  */
+static inline void xscale_branch_address(struct xscale_trace_data *trace_data,
+                                        int i, uint32_t *target)
 {
        /* if there are less than four entries prior to the indirect branch message
         * we can't extract the address */
        if (i < 4)
-       {
-               return -1;
-       }
-
-       *target = (trace_data->entries[i-1].data) | (trace_data->entries[i-2].data << 8) |
-                               (trace_data->entries[i-3].data << 16) | (trace_data->entries[i-4].data << 24);
+               *target = 0;
+       else
+               *target = (trace_data->entries[i-1].data) | (trace_data->entries[i-2].data << 8) |
+                       (trace_data->entries[i-3].data << 16) | (trace_data->entries[i-4].data << 24);
+}
+
+static inline void xscale_display_instruction(struct target *target, uint32_t pc,
+                                                                                         struct arm_instruction *instruction,
+                                                                                         struct command_context *cmd_ctx)
+{
+   int retval = xscale_read_instruction(target, pc, instruction);
+   if (retval == ERROR_OK)
+         command_print(cmd_ctx, "%s", instruction->text);
+   else
+         command_print(cmd_ctx, "0x%8.8" PRIx32 "\t<not found in image>", pc);
+}
+
+static int xscale_analyze_trace(struct target *target, struct command_context *cmd_ctx)
+{
+   struct xscale_common *xscale = target_to_xscale(target);
+   struct xscale_trace_data *trace_data = xscale->trace.data;
+   int i, retval;
+   uint32_t breakpoint_pc;
+   struct arm_instruction instruction;
+   uint32_t current_pc = 0;  /* initialized when address determined */
+       
+   if (!xscale->trace.image)
+         LOG_WARNING("No trace image loaded; use 'xscale trace_image'");
+
+   /* loop for each trace buffer that was loaded from target */
+   while (trace_data)
+   {
+         int chkpt = 0;  /* incremented as checkpointed entries found */
+         int j;
+
+         /* FIXME: set this to correct mode when trace buffer is first enabled */
+         xscale->trace.core_state = ARM_STATE_ARM;
+
+         /* loop for each entry in this trace buffer */
+         for (i = 0; i < trace_data->depth; i++)
+         {
+                int exception = 0;
+                uint32_t chkpt_reg = 0x0;
+                uint32_t branch_target = 0;
+                int count;
+
+                /* trace entry type is upper nybble of 'message byte' */
+                int trace_msg_type = (trace_data->entries[i].data & 0xf0) >> 4;
+
+                /* Target addresses of indirect branches are written into buffer
+                 * before the message byte representing the branch. Skip past it */
+                if (trace_data->entries[i].type == XSCALE_TRACE_ADDRESS)
+                       continue;
+
+                switch (trace_msg_type)
+                {
+                       case 0:         /* Exceptions */
+                       case 1:
+                       case 2:
+                       case 3:
+                       case 4:
+                       case 5:
+                       case 6:
+                       case 7:
+                          exception = (trace_data->entries[i].data & 0x70) >> 4;
+
+                          /* FIXME: vector table may be at ffff0000 */
+                          branch_target = (trace_data->entries[i].data & 0xf0) >> 2;
+                          break;
+
+                       case 8:         /* Direct Branch */
+                          break;
+
+                       case 9:         /* Indirect Branch */
+                          xscale_branch_address(trace_data, i, &branch_target);
+                          break;
+
+                       case 13:           /* Checkpointed Indirect Branch */
+                          xscale_branch_address(trace_data, i, &branch_target);
+                          if ((trace_data->num_checkpoints == 2) && (chkpt == 0))
+                                 chkpt_reg = trace_data->chkpt1; /* 2 chkpts, this is oldest */
+                          else
+                                 chkpt_reg = trace_data->chkpt0; /* 1 chkpt, or 2 and newest */
+
+                          chkpt++;
+                          break;
+
+                       case 12:           /* Checkpointed Direct Branch */
+                          if ((trace_data->num_checkpoints == 2) && (chkpt == 0))
+                                 chkpt_reg = trace_data->chkpt1; /* 2 chkpts, this is oldest */
+                          else
+                                 chkpt_reg = trace_data->chkpt0; /* 1 chkpt, or 2 and newest */
+
+                          /* if no current_pc, checkpoint will be starting point */
+                          if (current_pc == 0)
+                                 branch_target = chkpt_reg;
+
+                          chkpt++;
+                          break;
+
+                       case 15:        /* Roll-over */
+                          break;
+
+                       default:        /* Reserved */
+                          LOG_WARNING("trace is suspect: invalid trace message byte");
+                          continue;
+
+                }
+
+                /* If we don't have the current_pc yet, but we did get the branch target
+                 * (either from the trace buffer on indirect branch, or from a checkpoint reg),
+                 * then we can start displaying instructions at the next iteration, with
+                 * branch_target as the starting point.
+                 */
+                if (current_pc == 0)
+                {
+                       current_pc = branch_target; /* remains 0 unless branch_target obtained */
+                       continue;
+                }
+
+                /* We have current_pc.  Read and display the instructions from the image.
+                 * First, display count instructions (lower nybble of message byte). */
+                count = trace_data->entries[i].data & 0x0f;
+                for (j = 0; j < count; j++)
+                {
+                       xscale_display_instruction(target, current_pc, &instruction, cmd_ctx);
+                       current_pc += xscale->trace.core_state == ARM_STATE_ARM ? 4 : 2;
+                }
+
+                /* An additional instruction is implicitly added to count for
+                 * rollover and some exceptions: undef, swi, prefetch abort. */
+                if ((trace_msg_type == 15) || (exception > 0 && exception < 4))
+                {
+                       xscale_display_instruction(target, current_pc, &instruction, cmd_ctx);
+                       current_pc += xscale->trace.core_state == ARM_STATE_ARM ? 4 : 2;
+                }
+
+                if (trace_msg_type == 15) /* rollover */
+                       continue;
+
+                if (exception)
+                {
+                       command_print(cmd_ctx, "--- exception %i ---", exception);
+                       continue;
+                }
+                       
+                /* not exception or rollover; next instruction is a branch and is
+                 * not included in the count */
+                xscale_display_instruction(target, current_pc, &instruction, cmd_ctx);
+
+                /* for direct branches, extract branch destination from instruction */
+                if ((trace_msg_type == 8) || (trace_msg_type == 12))
+                {
+                       retval = xscale_read_instruction(target, current_pc, &instruction);
+                       if (retval == ERROR_OK)
+                          current_pc = instruction.info.b_bl_bx_blx.target_address;
+                       else
+                          current_pc = 0;      /* branch destination unknown */
 
-       return 0;
-}
+                       /* direct branch w/ checkpoint; can also get from checkpoint reg */
+                       if (trace_msg_type == 12)
+                       {
+                          if (current_pc == 0)
+                                 current_pc = chkpt_reg;
+                          else if (current_pc != chkpt_reg)  /* sanity check */
+                                 LOG_WARNING("trace is suspect: checkpoint register "
+                                                         "inconsistent with adddress from image");
+                       }
 
-static int xscale_analyze_trace(target_t *target, command_context_t *cmd_ctx)
-{
-       struct xscale_common_s *xscale = target_to_xscale(target);
-       int next_pc_ok = 0;
-       uint32_t next_pc = 0x0;
-       xscale_trace_data_t *trace_data = xscale->trace.data;
-       int retval;
+                       if (current_pc == 0)
+                          command_print(cmd_ctx, "address unknown");
 
-       while (trace_data)
-       {
-               int i, chkpt;
-               int rollover;
-               int branch;
-               int exception;
-               xscale->trace.core_state = ARMV4_5_STATE_ARM;
+                       continue;
+                }
 
-               chkpt = 0;
-               rollover = 0;
+                /* indirect branch; the branch destination was read from trace buffer */
+                if ((trace_msg_type == 9) || (trace_msg_type == 13))
+                {
+                       current_pc = branch_target;
 
-               for (i = 0; i < trace_data->depth; i++)
-               {
-                       next_pc_ok = 0;
-                       branch = 0;
-                       exception = 0;
+                       /* sanity check (checkpoint reg is redundant) */
+                       if ((trace_msg_type == 13) && (chkpt_reg != branch_target))
+                          LOG_WARNING("trace is suspect: checkpoint register "
+                                                  "inconsistent with address from trace buffer");
+                }
 
-                       if (trace_data->entries[i].type == XSCALE_TRACE_ADDRESS)
-                               continue;
+         } /* END: for (i = 0; i < trace_data->depth; i++) */
 
-                       switch ((trace_data->entries[i].data & 0xf0) >> 4)
-                       {
-                               case 0:         /* Exceptions */
-                               case 1:
-                               case 2:
-                               case 3:
-                               case 4:
-                               case 5:
-                               case 6:
-                               case 7:
-                                       exception = (trace_data->entries[i].data & 0x70) >> 4;
-                                       next_pc_ok = 1;
-                                       next_pc = (trace_data->entries[i].data & 0xf0) >> 2;
-                                       command_print(cmd_ctx, "--- exception %i ---", (trace_data->entries[i].data & 0xf0) >> 4);
-                                       break;
-                               case 8:         /* Direct Branch */
-                                       branch = 1;
-                                       break;
-                               case 9:         /* Indirect Branch */
-                                       branch = 1;
-                                       if (xscale_branch_address(trace_data, i, &next_pc) == 0)
-                                       {
-                                               next_pc_ok = 1;
-                                       }
-                                       break;
-                               case 13:        /* Checkpointed Indirect Branch */
-                                       if (xscale_branch_address(trace_data, i, &next_pc) == 0)
-                                       {
-                                               next_pc_ok = 1;
-                                               if (((chkpt == 0) && (next_pc != trace_data->chkpt0))
-                                                       || ((chkpt == 1) && (next_pc != trace_data->chkpt1)))
-                                                       LOG_WARNING("checkpointed indirect branch target address doesn't match checkpoint");
-                                       }
-                                       /* explicit fall-through */
-                               case 12:        /* Checkpointed Direct Branch */
-                                       branch = 1;
-                                       if (chkpt == 0)
-                                       {
-                                               next_pc_ok = 1;
-                                               next_pc = trace_data->chkpt0;
-                                               chkpt++;
-                                       }
-                                       else if (chkpt == 1)
-                                       {
-                                               next_pc_ok = 1;
-                                               next_pc = trace_data->chkpt0;
-                                               chkpt++;
-                                       }
-                                       else
-                                       {
-                                               LOG_WARNING("more than two checkpointed branches encountered");
-                                       }
-                                       break;
-                               case 15:        /* Roll-over */
-                                       rollover++;
-                                       continue;
-                               default:        /* Reserved */
-                                       command_print(cmd_ctx, "--- reserved trace message ---");
-                                       LOG_ERROR("BUG: trace message %i is reserved", (trace_data->entries[i].data & 0xf0) >> 4);
-                                       return ERROR_OK;
-                       }
+         breakpoint_pc = trace_data->last_instruction; /* used below */
+         trace_data = trace_data->next;
 
-                       if (xscale->trace.pc_ok)
-                       {
-                               int executed = (trace_data->entries[i].data & 0xf) + rollover * 16;
-                               arm_instruction_t instruction;
+   } /* END: while (trace_data) */
 
-                               if ((exception == 6) || (exception == 7))
-                               {
-                                       /* IRQ or FIQ exception, no instruction executed */
-                                       executed -= 1;
-                               }
+   /* Finally... display all instructions up to the value of the pc when the
+       * debug break occurred (saved when trace data was collected from target).
+       * This is necessary because the trace only records execution branches and 16
+       * consecutive instructions (rollovers), so last few typically missed.
+       */
+   if (current_pc == 0)
+         return ERROR_OK;   /* current_pc was never found */
 
-                               while (executed-- >= 0)
-                               {
-                                       if ((retval = xscale_read_instruction(target, &instruction)) != ERROR_OK)
-                                       {
-                                               /* can't continue tracing with no image available */
-                                               if (retval == ERROR_TRACE_IMAGE_UNAVAILABLE)
-                                               {
-                                                       return retval;
-                                               }
-                                               else if (retval == ERROR_TRACE_INSTRUCTION_UNAVAILABLE)
-                                               {
-                                                       /* TODO: handle incomplete images */
-                                               }
-                                       }
-
-                                       /* a precise abort on a load to the PC is included in the incremental
-                                        * word count, other instructions causing data aborts are not included
-                                        */
-                                       if ((executed == 0) && (exception == 4)
-                                               && ((instruction.type >= ARM_LDR) && (instruction.type <= ARM_LDM)))
-                                       {
-                                               if ((instruction.type == ARM_LDM)
-                                                       && ((instruction.info.load_store_multiple.register_list & 0x8000) == 0))
-                                               {
-                                                       executed--;
-                                               }
-                                               else if (((instruction.type >= ARM_LDR) && (instruction.type <= ARM_LDRSH))
-                                                       && (instruction.info.load_store.Rd != 15))
-                                               {
-                                                       executed--;
-                                               }
-                                       }
-
-                                       /* only the last instruction executed
-                                        * (the one that caused the control flow change)
-                                        * could be a taken branch
-                                        */
-                                       if (((executed == -1) && (branch == 1)) &&
-                                               (((instruction.type == ARM_B) ||
-                                                       (instruction.type == ARM_BL) ||
-                                                       (instruction.type == ARM_BLX)) &&
-                                                       (instruction.info.b_bl_bx_blx.target_address != 0xffffffff)))
-                                       {
-                                               xscale->trace.current_pc = instruction.info.b_bl_bx_blx.target_address;
-                                       }
-                                       else
-                                       {
-                                               xscale->trace.current_pc += (xscale->trace.core_state == ARMV4_5_STATE_ARM) ? 4 : 2;
-                                       }
-                                       command_print(cmd_ctx, "%s", instruction.text);
-                               }
+   /* how many instructions remaining? */
+   int gap_count = (breakpoint_pc - current_pc) /
+         (xscale->trace.core_state == ARM_STATE_ARM ? 4 : 2);
 
-                               rollover = 0;
-                       }
+   /* should never be negative or over 16, but verify */
+   if (gap_count < 0 || gap_count > 16)
+   {
+         LOG_WARNING("trace is suspect: excessive gap at end of trace");
+         return ERROR_OK;  /* bail; large number or negative value no good */
+   }
 
-                       if (next_pc_ok)
-                       {
-                               xscale->trace.current_pc = next_pc;
-                               xscale->trace.pc_ok = 1;
-                       }
-               }
+   /* display remaining instructions */
+   for (i = 0; i < gap_count; i++)
+   {
+         xscale_display_instruction(target, current_pc, &instruction, cmd_ctx);
+         current_pc += xscale->trace.core_state == ARM_STATE_ARM ? 4 : 2;
+   }
 
-               for (; xscale->trace.current_pc < trace_data->last_instruction; xscale->trace.current_pc += (xscale->trace.core_state == ARMV4_5_STATE_ARM) ? 4 : 2)
-               {
-                       arm_instruction_t instruction;
-                       if ((retval = xscale_read_instruction(target, &instruction)) != ERROR_OK)
-                       {
-                               /* can't continue tracing with no image available */
-                               if (retval == ERROR_TRACE_IMAGE_UNAVAILABLE)
-                               {
-                                       return retval;
-                               }
-                               else if (retval == ERROR_TRACE_INSTRUCTION_UNAVAILABLE)
-                               {
-                                       /* TODO: handle incomplete images */
-                               }
-                       }
-                       command_print(cmd_ctx, "%s", instruction.text);
-               }
-
-               trace_data = trace_data->next;
-       }
-
-       return ERROR_OK;
+   return ERROR_OK;
 }
 
-static void xscale_build_reg_cache(target_t *target)
+static const struct reg_arch_type xscale_reg_type = {
+       .get = xscale_get_reg,
+       .set = xscale_set_reg,
+};
+
+static void xscale_build_reg_cache(struct target *target)
 {
-       struct xscale_common_s *xscale = target_to_xscale(target);
-       struct armv4_5_common_s *armv4_5 = &xscale->armv4_5_common;
-       reg_cache_t **cache_p = register_get_last_cache_p(&target->reg_cache);
-       xscale_reg_t *arch_info = malloc(sizeof(xscale_reg_arch_info));
+       struct xscale_common *xscale = target_to_xscale(target);
+       struct arm *armv4_5 = &xscale->armv4_5_common;
+       struct reg_cache **cache_p = register_get_last_cache_p(&target->reg_cache);
+       struct xscale_reg *arch_info = malloc(sizeof(xscale_reg_arch_info));
        int i;
-       int num_regs = sizeof(xscale_reg_arch_info) / sizeof(xscale_reg_t);
-
-       (*cache_p) = armv4_5_build_reg_cache(target, armv4_5);
-       armv4_5->core_cache = (*cache_p);
+       int num_regs = ARRAY_SIZE(xscale_reg_arch_info);
 
-       /* register a register arch-type for XScale dbg registers only once */
-       if (xscale_reg_arch_type == -1)
-               xscale_reg_arch_type = register_reg_arch_type(xscale_get_reg, xscale_set_reg);
+       (*cache_p) = arm_build_reg_cache(target, armv4_5);
 
-       (*cache_p)->next = malloc(sizeof(reg_cache_t));
+       (*cache_p)->next = malloc(sizeof(struct reg_cache));
        cache_p = &(*cache_p)->next;
 
        /* fill in values for the xscale reg cache */
        (*cache_p)->name = "XScale registers";
        (*cache_p)->next = NULL;
-       (*cache_p)->reg_list = malloc(num_regs * sizeof(reg_t));
+       (*cache_p)->reg_list = malloc(num_regs * sizeof(struct reg));
        (*cache_p)->num_regs = num_regs;
 
        for (i = 0; i < num_regs; i++)
@@ -2849,10 +3014,8 @@ static void xscale_build_reg_cache(target_t *target)
                (*cache_p)->reg_list[i].dirty = 0;
                (*cache_p)->reg_list[i].valid = 0;
                (*cache_p)->reg_list[i].size = 32;
-               (*cache_p)->reg_list[i].bitfield_desc = NULL;
-               (*cache_p)->reg_list[i].num_bitfields = 0;
                (*cache_p)->reg_list[i].arch_info = &arch_info[i];
-               (*cache_p)->reg_list[i].arch_type = xscale_reg_arch_type;
+               (*cache_p)->reg_list[i].type = &xscale_reg_type;
                arch_info[i] = xscale_reg_arch_info[i];
                arch_info[i].target = target;
        }
@@ -2860,26 +3023,26 @@ static void xscale_build_reg_cache(target_t *target)
        xscale->reg_cache = (*cache_p);
 }
 
-static int xscale_init_target(struct command_context_s *cmd_ctx,
-               struct target_s *target)
+static int xscale_init_target(struct command_context *cmd_ctx,
+               struct target *target)
 {
        xscale_build_reg_cache(target);
        return ERROR_OK;
 }
 
-static int xscale_init_arch_info(target_t *target,
-               xscale_common_t *xscale, jtag_tap_t *tap, const char *variant)
+static int xscale_init_arch_info(struct target *target,
+               struct xscale_common *xscale, struct jtag_tap *tap, const char *variant)
 {
-       armv4_5_common_t *armv4_5;
+       struct arm *armv4_5;
        uint32_t high_reset_branch, low_reset_branch;
        int i;
 
        armv4_5 = &xscale->armv4_5_common;
 
-       /* store architecture specfic data (none so far) */
+       /* store architecture specfic data */
        xscale->common_magic = XSCALE_COMMON_MAGIC;
 
-       /* we don't really *need* variant info ... */
+       /* we don't really *need* a variant param ... */
        if (variant) {
                int ir_length = 0;
 
@@ -2892,6 +3055,8 @@ static int xscale_init_arch_info(target_t *target,
                                || strcmp(variant, "ixp45x") == 0
                                || strcmp(variant, "ixp46x") == 0)
                        ir_length = 7;
+               else if (strcmp(variant, "pxa3xx") == 0)
+                       ir_length = 11;
                else
                        LOG_WARNING("%s: unrecognized variant %s",
                                tap->dotted_name, variant);
@@ -2903,6 +3068,12 @@ static int xscale_init_arch_info(target_t *target,
                }
        }
 
+       /* PXA3xx shifts the JTAG instructions */
+       if (tap->ir_length == 11)
+               xscale->xscale_variant = XSCALE_PXA3XX;
+       else
+               xscale->xscale_variant = XSCALE_IXP4XX_PXA2XX;
+
        /* the debug handler isn't installed (and thus not running) at this time */
        xscale->handler_address = 0xfe000800;
 
@@ -2941,6 +3112,9 @@ static int xscale_init_arch_info(target_t *target,
        xscale->dbr0_used = 0;
        xscale->dbr1_used = 0;
 
+       LOG_INFO("%s: hardware has 2 breakpoints and 2 watchpoints",
+                       target_name(target));
+
        xscale->arm_bkpt = ARMV5_BKPT(0x0);
        xscale->thumb_bkpt = ARMV5_T_BKPT(0x0) & 0xffff;
 
@@ -2958,7 +3132,7 @@ static int xscale_init_arch_info(target_t *target,
        armv4_5->write_core_reg = xscale_write_core_reg;
        armv4_5->full_context = xscale_full_context;
 
-       armv4_5_init_arch_info(target, armv4_5);
+       arm_init_arch_info(target, armv4_5);
 
        xscale->armv4_5_mmu.armv4_5_cache.ctype = -1;
        xscale->armv4_5_mmu.get_ttb = xscale_get_ttb;
@@ -2972,9 +3146,9 @@ static int xscale_init_arch_info(target_t *target,
        return ERROR_OK;
 }
 
-static int xscale_target_create(struct target_s *target, Jim_Interp *interp)
+static int xscale_target_create(struct target *target, Jim_Interp *interp)
 {
-       xscale_common_t *xscale;
+       struct xscale_common *xscale;
 
        if (sizeof xscale_debug_handler - 1 > 0x800) {
                LOG_ERROR("debug_handler.bin: larger than 2kb");
@@ -2991,29 +3165,29 @@ static int xscale_target_create(struct target_s *target, Jim_Interp *interp)
 
 COMMAND_HANDLER(xscale_handle_debug_handler_command)
 {
-       target_t *target = NULL;
-       xscale_common_t *xscale;
+       struct target *target = NULL;
+       struct xscale_common *xscale;
        int retval;
        uint32_t handler_address;
 
-       if (argc < 2)
+       if (CMD_ARGC < 2)
        {
                LOG_ERROR("'xscale debug_handler <target#> <address>' command takes two required operands");
                return ERROR_OK;
        }
 
-       if ((target = get_target(args[0])) == NULL)
+       if ((target = get_target(CMD_ARGV[0])) == NULL)
        {
-               LOG_ERROR("target '%s' not defined", args[0]);
+               LOG_ERROR("target '%s' not defined", CMD_ARGV[0]);
                return ERROR_FAIL;
        }
 
        xscale = target_to_xscale(target);
-       retval = xscale_verify_pointer(cmd_ctx, xscale);
+       retval = xscale_verify_pointer(CMD_CTX, xscale);
        if (retval != ERROR_OK)
                return retval;
 
-       COMMAND_PARSE_NUMBER(u32, args[1], handler_address);
+       COMMAND_PARSE_NUMBER(u32, CMD_ARGV[1], handler_address);
 
        if (((handler_address >= 0x800) && (handler_address <= 0x1fef800)) ||
                ((handler_address >= 0xfe000800) && (handler_address <= 0xfffff800)))
@@ -3031,28 +3205,28 @@ COMMAND_HANDLER(xscale_handle_debug_handler_command)
 
 COMMAND_HANDLER(xscale_handle_cache_clean_address_command)
 {
-       target_t *target = NULL;
-       xscale_common_t *xscale;
+       struct target *target = NULL;
+       struct xscale_common *xscale;
        int retval;
        uint32_t cache_clean_address;
 
-       if (argc < 2)
+       if (CMD_ARGC < 2)
        {
                return ERROR_COMMAND_SYNTAX_ERROR;
        }
 
-       target = get_target(args[0]);
+       target = get_target(CMD_ARGV[0]);
        if (target == NULL)
        {
-               LOG_ERROR("target '%s' not defined", args[0]);
+               LOG_ERROR("target '%s' not defined", CMD_ARGV[0]);
                return ERROR_FAIL;
        }
        xscale = target_to_xscale(target);
-       retval = xscale_verify_pointer(cmd_ctx, xscale);
+       retval = xscale_verify_pointer(CMD_CTX, xscale);
        if (retval != ERROR_OK)
                return retval;
 
-       COMMAND_PARSE_NUMBER(u32, args[1], cache_clean_address);
+       COMMAND_PARSE_NUMBER(u32, CMD_ARGV[1], cache_clean_address);
 
        if (cache_clean_address & 0xffff)
        {
@@ -3068,43 +3242,40 @@ COMMAND_HANDLER(xscale_handle_cache_clean_address_command)
 
 COMMAND_HANDLER(xscale_handle_cache_info_command)
 {
-       target_t *target = get_current_target(cmd_ctx);
-       struct xscale_common_s *xscale = target_to_xscale(target);
+       struct target *target = get_current_target(CMD_CTX);
+       struct xscale_common *xscale = target_to_xscale(target);
        int retval;
 
-       retval = xscale_verify_pointer(cmd_ctx, xscale);
+       retval = xscale_verify_pointer(CMD_CTX, xscale);
        if (retval != ERROR_OK)
                return retval;
 
-       return armv4_5_handle_cache_info_command(cmd_ctx, &xscale->armv4_5_mmu.armv4_5_cache);
+       return armv4_5_handle_cache_info_command(CMD_CTX, &xscale->armv4_5_mmu.armv4_5_cache);
 }
 
-static int xscale_virt2phys(struct target_s *target,
+static int xscale_virt2phys(struct target *target,
                uint32_t virtual, uint32_t *physical)
 {
-       struct xscale_common_s *xscale = target_to_xscale(target);
-       int type;
+       struct xscale_common *xscale = target_to_xscale(target);
        uint32_t cb;
-       int domain;
-       uint32_t ap;
 
        if (xscale->common_magic != XSCALE_COMMON_MAGIC) {
                LOG_ERROR(xscale_not);
                return ERROR_TARGET_INVALID;
        }
 
-       uint32_t ret = armv4_5_mmu_translate_va(target, &xscale->armv4_5_mmu, virtual, &type, &cb, &domain, &ap);
-       if (type == -1)
-       {
-               return ret;
-       }
+       uint32_t ret;
+       int retval = armv4_5_mmu_translate_va(target, &xscale->armv4_5_mmu,
+                       virtual, &cb, &ret);
+       if (retval != ERROR_OK)
+               return retval;
        *physical = ret;
        return ERROR_OK;
 }
 
-static int xscale_mmu(struct target_s *target, int *enabled)
+static int xscale_mmu(struct target *target, int *enabled)
 {
-       struct xscale_common_s *xscale = target_to_xscale(target);
+       struct xscale_common *xscale = target_to_xscale(target);
 
        if (target->state != TARGET_HALTED)
        {
@@ -3117,114 +3288,104 @@ static int xscale_mmu(struct target_s *target, int *enabled)
 
 COMMAND_HANDLER(xscale_handle_mmu_command)
 {
-       target_t *target = get_current_target(cmd_ctx);
-       struct xscale_common_s *xscale = target_to_xscale(target);
+       struct target *target = get_current_target(CMD_CTX);
+       struct xscale_common *xscale = target_to_xscale(target);
        int retval;
 
-       retval = xscale_verify_pointer(cmd_ctx, xscale);
+       retval = xscale_verify_pointer(CMD_CTX, xscale);
        if (retval != ERROR_OK)
                return retval;
 
        if (target->state != TARGET_HALTED)
        {
-               command_print(cmd_ctx, "target must be stopped for \"%s\" command", cmd);
+               command_print(CMD_CTX, "target must be stopped for \"%s\" command", CMD_NAME);
                return ERROR_OK;
        }
 
-       if (argc >= 1)
+       if (CMD_ARGC >= 1)
        {
-               if (strcmp("enable", args[0]) == 0)
-               {
+               bool enable;
+               COMMAND_PARSE_ENABLE(CMD_ARGV[0], enable);
+               if (enable)
                        xscale_enable_mmu_caches(target, 1, 0, 0);
-                       xscale->armv4_5_mmu.mmu_enabled = 1;
-               }
-               else if (strcmp("disable", args[0]) == 0)
-               {
+               else
                        xscale_disable_mmu_caches(target, 1, 0, 0);
-                       xscale->armv4_5_mmu.mmu_enabled = 0;
-               }
+               xscale->armv4_5_mmu.mmu_enabled = enable;
        }
 
-       command_print(cmd_ctx, "mmu %s", (xscale->armv4_5_mmu.mmu_enabled) ? "enabled" : "disabled");
+       command_print(CMD_CTX, "mmu %s", (xscale->armv4_5_mmu.mmu_enabled) ? "enabled" : "disabled");
 
        return ERROR_OK;
 }
 
 COMMAND_HANDLER(xscale_handle_idcache_command)
 {
-       target_t *target = get_current_target(cmd_ctx);
-       struct xscale_common_s *xscale = target_to_xscale(target);
-       int icache = 0, dcache = 0;
-       int retval;
+       struct target *target = get_current_target(CMD_CTX);
+       struct xscale_common *xscale = target_to_xscale(target);
 
-       retval = xscale_verify_pointer(cmd_ctx, xscale);
+       int retval = xscale_verify_pointer(CMD_CTX, xscale);
        if (retval != ERROR_OK)
                return retval;
 
        if (target->state != TARGET_HALTED)
        {
-               command_print(cmd_ctx, "target must be stopped for \"%s\" command", cmd);
+               command_print(CMD_CTX, "target must be stopped for \"%s\" command", CMD_NAME);
                return ERROR_OK;
        }
 
-       if (strcmp(cmd, "icache") == 0)
-               icache = 1;
-       else if (strcmp(cmd, "dcache") == 0)
-               dcache = 1;
-
-       if (argc >= 1)
+       bool icache = false;
+       if (strcmp(CMD_NAME, "icache") == 0)
+               icache = true;
+       if (CMD_ARGC >= 1)
        {
-               if (strcmp("enable", args[0]) == 0)
-               {
-                       xscale_enable_mmu_caches(target, 0, dcache, icache);
-
-                       if (icache)
-                               xscale->armv4_5_mmu.armv4_5_cache.i_cache_enabled = 1;
-                       else if (dcache)
-                               xscale->armv4_5_mmu.armv4_5_cache.d_u_cache_enabled = 1;
-               }
-               else if (strcmp("disable", args[0]) == 0)
-               {
-                       xscale_disable_mmu_caches(target, 0, dcache, icache);
-
-                       if (icache)
-                               xscale->armv4_5_mmu.armv4_5_cache.i_cache_enabled = 0;
-                       else if (dcache)
-                               xscale->armv4_5_mmu.armv4_5_cache.d_u_cache_enabled = 0;
+               bool enable;
+               COMMAND_PARSE_ENABLE(CMD_ARGV[0], enable);
+               if (icache) {
+                       xscale->armv4_5_mmu.armv4_5_cache.i_cache_enabled = enable;
+                       if (enable)
+                               xscale_enable_mmu_caches(target, 0, 0, 1);
+                       else
+                               xscale_disable_mmu_caches(target, 0, 0, 1);
+               } else {
+                       xscale->armv4_5_mmu.armv4_5_cache.d_u_cache_enabled = enable;
+                       if (enable)
+                               xscale_enable_mmu_caches(target, 0, 1, 0);
+                       else
+                               xscale_disable_mmu_caches(target, 0, 1, 0);
                }
        }
 
-       if (icache)
-               command_print(cmd_ctx, "icache %s", (xscale->armv4_5_mmu.armv4_5_cache.i_cache_enabled) ? "enabled" : "disabled");
-
-       if (dcache)
-               command_print(cmd_ctx, "dcache %s", (xscale->armv4_5_mmu.armv4_5_cache.d_u_cache_enabled) ? "enabled" : "disabled");
+       bool enabled = icache ?
+                                       xscale->armv4_5_mmu.armv4_5_cache.i_cache_enabled :
+                                       xscale->armv4_5_mmu.armv4_5_cache.d_u_cache_enabled;
+       const char *msg = enabled ? "enabled" : "disabled";
+       command_print(CMD_CTX, "%s %s", CMD_NAME, msg);
 
        return ERROR_OK;
 }
 
 COMMAND_HANDLER(xscale_handle_vector_catch_command)
 {
-       target_t *target = get_current_target(cmd_ctx);
-       struct xscale_common_s *xscale = target_to_xscale(target);
+       struct target *target = get_current_target(CMD_CTX);
+       struct xscale_common *xscale = target_to_xscale(target);
        int retval;
 
-       retval = xscale_verify_pointer(cmd_ctx, xscale);
+       retval = xscale_verify_pointer(CMD_CTX, xscale);
        if (retval != ERROR_OK)
                return retval;
 
-       if (argc < 1)
+       if (CMD_ARGC < 1)
        {
-               command_print(cmd_ctx, "usage: xscale vector_catch [mask]");
+               command_print(CMD_CTX, "usage: xscale vector_catch [mask]");
        }
        else
        {
-               COMMAND_PARSE_NUMBER(u8, args[0], xscale->vector_catch);
+               COMMAND_PARSE_NUMBER(u8, CMD_ARGV[0], xscale->vector_catch);
                buf_set_u32(xscale->reg_cache->reg_list[XSCALE_DCSR].value, 16, 8, xscale->vector_catch);
                xscale_write_dcsr(target, -1, -1);
        }
 
-       command_print(cmd_ctx, "vector catch mask: 0x%2.2x", xscale->vector_catch);
+       command_print(CMD_CTX, "vector catch mask: 0x%2.2x", xscale->vector_catch);
 
        return ERROR_OK;
 }
@@ -3232,47 +3393,47 @@ COMMAND_HANDLER(xscale_handle_vector_catch_command)
 
 COMMAND_HANDLER(xscale_handle_vector_table_command)
 {
-       target_t *target = get_current_target(cmd_ctx);
-       struct xscale_common_s *xscale = target_to_xscale(target);
+       struct target *target = get_current_target(CMD_CTX);
+       struct xscale_common *xscale = target_to_xscale(target);
        int err = 0;
        int retval;
 
-       retval = xscale_verify_pointer(cmd_ctx, xscale);
+       retval = xscale_verify_pointer(CMD_CTX, xscale);
        if (retval != ERROR_OK)
                return retval;
 
-       if (argc == 0) /* print current settings */
+       if (CMD_ARGC == 0) /* print current settings */
        {
                int idx;
 
-               command_print(cmd_ctx, "active user-set static vectors:");
+               command_print(CMD_CTX, "active user-set static vectors:");
                for (idx = 1; idx < 8; idx++)
                        if (xscale->static_low_vectors_set & (1 << idx))
-                               command_print(cmd_ctx, "low  %d: 0x%" PRIx32, idx, xscale->static_low_vectors[idx]);
+                               command_print(CMD_CTX, "low  %d: 0x%" PRIx32, idx, xscale->static_low_vectors[idx]);
                for (idx = 1; idx < 8; idx++)
                        if (xscale->static_high_vectors_set & (1 << idx))
-                               command_print(cmd_ctx, "high %d: 0x%" PRIx32, idx, xscale->static_high_vectors[idx]);
+                               command_print(CMD_CTX, "high %d: 0x%" PRIx32, idx, xscale->static_high_vectors[idx]);
                return ERROR_OK;
        }
 
-       if (argc != 3)
+       if (CMD_ARGC != 3)
                err = 1;
        else
        {
                int idx;
-               COMMAND_PARSE_NUMBER(int, args[1], idx);
+               COMMAND_PARSE_NUMBER(int, CMD_ARGV[1], idx);
                uint32_t vec;
-               COMMAND_PARSE_NUMBER(u32, args[2], vec);
+               COMMAND_PARSE_NUMBER(u32, CMD_ARGV[2], vec);
 
                if (idx < 1 || idx >= 8)
                        err = 1;
 
-               if (!err && strcmp(args[0], "low") == 0)
+               if (!err && strcmp(CMD_ARGV[0], "low") == 0)
                {
                        xscale->static_low_vectors_set |= (1<<idx);
                        xscale->static_low_vectors[idx] = vec;
                }
-               else if (!err && (strcmp(args[0], "high") == 0))
+               else if (!err && (strcmp(CMD_ARGV[0], "high") == 0))
                {
                        xscale->static_high_vectors_set |= (1<<idx);
                        xscale->static_high_vectors[idx] = vec;
@@ -3282,7 +3443,7 @@ COMMAND_HANDLER(xscale_handle_vector_table_command)
        }
 
        if (err)
-               command_print(cmd_ctx, "usage: xscale vector_table <high|low> <index> <code>");
+               command_print(CMD_CTX, "usage: xscale vector_table <high|low> <index> <code>");
 
        return ERROR_OK;
 }
@@ -3290,25 +3451,24 @@ COMMAND_HANDLER(xscale_handle_vector_table_command)
 
 COMMAND_HANDLER(xscale_handle_trace_buffer_command)
 {
-       target_t *target = get_current_target(cmd_ctx);
-       struct xscale_common_s *xscale = target_to_xscale(target);
-       struct armv4_5_common_s *armv4_5 = &xscale->armv4_5_common;
+       struct target *target = get_current_target(CMD_CTX);
+       struct xscale_common *xscale = target_to_xscale(target);
        uint32_t dcsr_value;
        int retval;
 
-       retval = xscale_verify_pointer(cmd_ctx, xscale);
+       retval = xscale_verify_pointer(CMD_CTX, xscale);
        if (retval != ERROR_OK)
                return retval;
 
        if (target->state != TARGET_HALTED)
        {
-               command_print(cmd_ctx, "target must be stopped for \"%s\" command", cmd);
+               command_print(CMD_CTX, "target must be stopped for \"%s\" command", CMD_NAME);
                return ERROR_OK;
        }
 
-       if ((argc >= 1) && (strcmp("enable", args[0]) == 0))
+       if ((CMD_ARGC >= 1) && (strcmp("enable", CMD_ARGV[0]) == 0))
        {
-               xscale_trace_data_t *td, *next_td;
+               struct xscale_trace_data *td, *next_td;
                xscale->trace.buffer_enabled = 1;
 
                /* free old trace data */
@@ -3324,37 +3484,24 @@ COMMAND_HANDLER(xscale_handle_trace_buffer_command)
                }
                xscale->trace.data = NULL;
        }
-       else if ((argc >= 1) && (strcmp("disable", args[0]) == 0))
+       else if ((CMD_ARGC >= 1) && (strcmp("disable", CMD_ARGV[0]) == 0))
        {
                xscale->trace.buffer_enabled = 0;
        }
 
-       if ((argc >= 2) && (strcmp("fill", args[1]) == 0))
+       if ((CMD_ARGC >= 2) && (strcmp("fill", CMD_ARGV[1]) == 0))
        {
                uint32_t fill = 1;
-               if (argc >= 3)
-                       COMMAND_PARSE_NUMBER(u32, args[2], fill);
+               if (CMD_ARGC >= 3)
+                       COMMAND_PARSE_NUMBER(u32, CMD_ARGV[2], fill);
                xscale->trace.buffer_fill = fill;
        }
-       else if ((argc >= 2) && (strcmp("wrap", args[1]) == 0))
+       else if ((CMD_ARGC >= 2) && (strcmp("wrap", CMD_ARGV[1]) == 0))
        {
                xscale->trace.buffer_fill = -1;
        }
 
-       if (xscale->trace.buffer_enabled)
-       {
-               /* if we enable the trace buffer in fill-once
-                * mode we know the address of the first instruction */
-               xscale->trace.pc_ok = 1;
-               xscale->trace.current_pc = buf_get_u32(armv4_5->core_cache->reg_list[15].value, 0, 32);
-       }
-       else
-       {
-               /* otherwise the address is unknown, and we have no known good PC */
-               xscale->trace.pc_ok = 0;
-       }
-
-       command_print(cmd_ctx, "trace buffer %s (%s)",
+       command_print(CMD_CTX, "trace buffer %s (%s)",
                (xscale->trace.buffer_enabled) ? "enabled" : "disabled",
                (xscale->trace.buffer_fill > 0) ? "fill" : "wrap");
 
@@ -3369,17 +3516,17 @@ COMMAND_HANDLER(xscale_handle_trace_buffer_command)
 
 COMMAND_HANDLER(xscale_handle_trace_image_command)
 {
-       target_t *target = get_current_target(cmd_ctx);
-       struct xscale_common_s *xscale = target_to_xscale(target);
+       struct target *target = get_current_target(CMD_CTX);
+       struct xscale_common *xscale = target_to_xscale(target);
        int retval;
 
-       if (argc < 1)
+       if (CMD_ARGC < 1)
        {
-               command_print(cmd_ctx, "usage: xscale trace_image <file> [base address] [type]");
+               command_print(CMD_CTX, "usage: xscale trace_image <file> [base address] [type]");
                return ERROR_OK;
        }
 
-       retval = xscale_verify_pointer(cmd_ctx, xscale);
+       retval = xscale_verify_pointer(CMD_CTX, xscale);
        if (retval != ERROR_OK)
                return retval;
 
@@ -3387,25 +3534,25 @@ COMMAND_HANDLER(xscale_handle_trace_image_command)
        {
                image_close(xscale->trace.image);
                free(xscale->trace.image);
-               command_print(cmd_ctx, "previously loaded image found and closed");
+               command_print(CMD_CTX, "previously loaded image found and closed");
        }
 
-       xscale->trace.image = malloc(sizeof(image_t));
+       xscale->trace.image = malloc(sizeof(struct image));
        xscale->trace.image->base_address_set = 0;
        xscale->trace.image->start_address_set = 0;
 
        /* a base address isn't always necessary, default to 0x0 (i.e. don't relocate) */
-       if (argc >= 2)
+       if (CMD_ARGC >= 2)
        {
                xscale->trace.image->base_address_set = 1;
-               COMMAND_PARSE_NUMBER(int, args[1], xscale->trace.image->base_address);
+               COMMAND_PARSE_NUMBER(llong, CMD_ARGV[1], xscale->trace.image->base_address);
        }
        else
        {
                xscale->trace.image->base_address_set = 0;
        }
 
-       if (image_open(xscale->trace.image, args[0], (argc >= 3) ? args[2] : NULL) != ERROR_OK)
+       if (image_open(xscale->trace.image, CMD_ARGV[0], (CMD_ARGC >= 3) ? CMD_ARGV[2] : NULL) != ERROR_OK)
        {
                free(xscale->trace.image);
                xscale->trace.image = NULL;
@@ -3417,25 +3564,25 @@ COMMAND_HANDLER(xscale_handle_trace_image_command)
 
 COMMAND_HANDLER(xscale_handle_dump_trace_command)
 {
-       target_t *target = get_current_target(cmd_ctx);
-       struct xscale_common_s *xscale = target_to_xscale(target);
-       xscale_trace_data_t *trace_data;
-       fileio_t file;
+       struct target *target = get_current_target(CMD_CTX);
+       struct xscale_common *xscale = target_to_xscale(target);
+       struct xscale_trace_data *trace_data;
+       struct fileio file;
        int retval;
 
-       retval = xscale_verify_pointer(cmd_ctx, xscale);
+       retval = xscale_verify_pointer(CMD_CTX, xscale);
        if (retval != ERROR_OK)
                return retval;
 
        if (target->state != TARGET_HALTED)
        {
-               command_print(cmd_ctx, "target must be stopped for \"%s\" command", cmd);
+               command_print(CMD_CTX, "target must be stopped for \"%s\" command", CMD_NAME);
                return ERROR_OK;
        }
 
-       if (argc < 1)
+       if (CMD_ARGC < 1)
        {
-               command_print(cmd_ctx, "usage: xscale dump_trace <file>");
+               command_print(CMD_CTX, "usage: xscale dump_trace <file>");
                return ERROR_OK;
        }
 
@@ -3443,11 +3590,11 @@ COMMAND_HANDLER(xscale_handle_dump_trace_command)
 
        if (!trace_data)
        {
-               command_print(cmd_ctx, "no trace data collected");
+               command_print(CMD_CTX, "no trace data collected");
                return ERROR_OK;
        }
 
-       if (fileio_open(&file, args[0], FILEIO_WRITE, FILEIO_BINARY) != ERROR_OK)
+       if (fileio_open(&file, CMD_ARGV[0], FILEIO_WRITE, FILEIO_BINARY) != ERROR_OK)
        {
                return ERROR_OK;
        }
@@ -3474,39 +3621,39 @@ COMMAND_HANDLER(xscale_handle_dump_trace_command)
 
 COMMAND_HANDLER(xscale_handle_analyze_trace_buffer_command)
 {
-       target_t *target = get_current_target(cmd_ctx);
-       struct xscale_common_s *xscale = target_to_xscale(target);
+       struct target *target = get_current_target(CMD_CTX);
+       struct xscale_common *xscale = target_to_xscale(target);
        int retval;
 
-       retval = xscale_verify_pointer(cmd_ctx, xscale);
+       retval = xscale_verify_pointer(CMD_CTX, xscale);
        if (retval != ERROR_OK)
                return retval;
 
-       xscale_analyze_trace(target, cmd_ctx);
+       xscale_analyze_trace(target, CMD_CTX);
 
        return ERROR_OK;
 }
 
 COMMAND_HANDLER(xscale_handle_cp15)
 {
-       target_t *target = get_current_target(cmd_ctx);
-       struct xscale_common_s *xscale = target_to_xscale(target);
+       struct target *target = get_current_target(CMD_CTX);
+       struct xscale_common *xscale = target_to_xscale(target);
        int retval;
 
-       retval = xscale_verify_pointer(cmd_ctx, xscale);
+       retval = xscale_verify_pointer(CMD_CTX, xscale);
        if (retval != ERROR_OK)
                return retval;
 
        if (target->state != TARGET_HALTED)
        {
-               command_print(cmd_ctx, "target must be stopped for \"%s\" command", cmd);
+               command_print(CMD_CTX, "target must be stopped for \"%s\" command", CMD_NAME);
                return ERROR_OK;
        }
        uint32_t reg_no = 0;
-       reg_t *reg = NULL;
-       if (argc > 0)
+       struct reg *reg = NULL;
+       if (CMD_ARGC > 0)
        {
-               COMMAND_PARSE_NUMBER(u32, args[0], reg_no);
+               COMMAND_PARSE_NUMBER(u32, CMD_ARGV[0], reg_no);
                /*translate from xscale cp15 register no to openocd register*/
                switch (reg_no)
                {
@@ -3535,25 +3682,25 @@ COMMAND_HANDLER(xscale_handle_cp15)
                        reg_no = XSCALE_CPACCESS;
                        break;
                default:
-                       command_print(cmd_ctx, "invalid register number");
+                       command_print(CMD_CTX, "invalid register number");
                        return ERROR_INVALID_ARGUMENTS;
                }
                reg = &xscale->reg_cache->reg_list[reg_no];
 
        }
-       if (argc == 1)
+       if (CMD_ARGC == 1)
        {
                uint32_t value;
 
                /* read cp15 control register */
                xscale_get_reg(reg);
                value = buf_get_u32(reg->value, 0, 32);
-               command_print(cmd_ctx, "%s (/%i): 0x%" PRIx32 "", reg->name, (int)(reg->size), value);
+               command_print(CMD_CTX, "%s (/%i): 0x%" PRIx32 "", reg->name, (int)(reg->size), value);
        }
-       else if (argc == 2)
+       else if (CMD_ARGC == 2)
        {
                uint32_t value;
-               COMMAND_PARSE_NUMBER(u32, args[1], value);
+               COMMAND_PARSE_NUMBER(u32, CMD_ARGV[1], value);
 
                /* send CP write request (command 0x41) */
                xscale_send_u32(target, 0x41);
@@ -3569,44 +3716,130 @@ COMMAND_HANDLER(xscale_handle_cp15)
        }
        else
        {
-               command_print(cmd_ctx, "usage: cp15 [register]<, [value]>");
+               command_print(CMD_CTX, "usage: cp15 [register]<, [value]>");
        }
 
        return ERROR_OK;
 }
 
-static int xscale_register_commands(struct command_context_s *cmd_ctx)
-{
-       command_t *xscale_cmd;
-
-       xscale_cmd = register_command(cmd_ctx, NULL, "xscale", NULL, COMMAND_ANY, "xscale specific commands");
-
-       register_command(cmd_ctx, xscale_cmd, "debug_handler", xscale_handle_debug_handler_command, COMMAND_ANY, "'xscale debug_handler <target#> <address>' command takes two required operands");
-       register_command(cmd_ctx, xscale_cmd, "cache_clean_address", xscale_handle_cache_clean_address_command, COMMAND_ANY, NULL);
-
-       register_command(cmd_ctx, xscale_cmd, "cache_info", xscale_handle_cache_info_command, COMMAND_EXEC, NULL);
-       register_command(cmd_ctx, xscale_cmd, "mmu", xscale_handle_mmu_command, COMMAND_EXEC, "['enable'|'disable'] the MMU");
-       register_command(cmd_ctx, xscale_cmd, "icache", xscale_handle_idcache_command, COMMAND_EXEC, "['enable'|'disable'] the ICache");
-       register_command(cmd_ctx, xscale_cmd, "dcache", xscale_handle_idcache_command, COMMAND_EXEC, "['enable'|'disable'] the DCache");
-
-       register_command(cmd_ctx, xscale_cmd, "vector_catch", xscale_handle_vector_catch_command, COMMAND_EXEC, "<mask> of vectors that should be catched");
-       register_command(cmd_ctx, xscale_cmd, "vector_table", xscale_handle_vector_table_command, COMMAND_EXEC, "<high|low> <index> <code> set static code for exception handler entry");
-
-       register_command(cmd_ctx, xscale_cmd, "trace_buffer", xscale_handle_trace_buffer_command, COMMAND_EXEC, "<enable | disable> ['fill' [n]|'wrap']");
-
-       register_command(cmd_ctx, xscale_cmd, "dump_trace", xscale_handle_dump_trace_command, COMMAND_EXEC, "dump content of trace buffer to <file>");
-       register_command(cmd_ctx, xscale_cmd, "analyze_trace", xscale_handle_analyze_trace_buffer_command, COMMAND_EXEC, "analyze content of trace buffer");
-       register_command(cmd_ctx, xscale_cmd, "trace_image", xscale_handle_trace_image_command,
-               COMMAND_EXEC, "load image from <file> [base address]");
-
-       register_command(cmd_ctx, xscale_cmd, "cp15", xscale_handle_cp15, COMMAND_EXEC, "access coproc 15 <register> [value]");
-
-       armv4_5_register_commands(cmd_ctx);
-
-       return ERROR_OK;
-}
+static const struct command_registration xscale_exec_command_handlers[] = {
+       {
+               .name = "cache_info",
+               .handler = xscale_handle_cache_info_command,
+               .mode = COMMAND_EXEC,
+               .help = "display information about CPU caches",
+       },
+       {
+               .name = "mmu",
+               .handler = xscale_handle_mmu_command,
+               .mode = COMMAND_EXEC,
+               .help = "enable or disable the MMU",
+               .usage = "['enable'|'disable']",
+       },
+       {
+               .name = "icache",
+               .handler = xscale_handle_idcache_command,
+               .mode = COMMAND_EXEC,
+               .help = "display ICache state, optionally enabling or "
+                       "disabling it",
+               .usage = "['enable'|'disable']",
+       },
+       {
+               .name = "dcache",
+               .handler = xscale_handle_idcache_command,
+               .mode = COMMAND_EXEC,
+               .help = "display DCache state, optionally enabling or "
+                       "disabling it",
+               .usage = "['enable'|'disable']",
+       },
+       {
+               .name = "vector_catch",
+               .handler = xscale_handle_vector_catch_command,
+               .mode = COMMAND_EXEC,
+               .help = "set or display 8-bit mask of vectors "
+                       "that should trigger debug entry",
+               .usage = "[mask]",
+       },
+       {
+               .name = "vector_table",
+               .handler = xscale_handle_vector_table_command,
+               .mode = COMMAND_EXEC,
+               .help = "set vector table entry in mini-ICache, "
+                       "or display current tables",
+               .usage = "[('high'|'low') index code]",
+       },
+       {
+               .name = "trace_buffer",
+               .handler = xscale_handle_trace_buffer_command,
+               .mode = COMMAND_EXEC,
+               .help = "display trace buffer status, enable or disable "
+                       "tracing, and optionally reconfigure trace mode",
+               .usage = "['enable'|'disable' ['fill' number|'wrap']]",
+       },
+       {
+               .name = "dump_trace",
+               .handler = xscale_handle_dump_trace_command,
+               .mode = COMMAND_EXEC,
+               .help = "dump content of trace buffer to file",
+               .usage = "filename",
+       },
+       {
+               .name = "analyze_trace",
+               .handler = xscale_handle_analyze_trace_buffer_command,
+               .mode = COMMAND_EXEC,
+               .help = "analyze content of trace buffer",
+               .usage = "",
+       },
+       {
+               .name = "trace_image",
+               .handler = xscale_handle_trace_image_command,
+               .mode = COMMAND_EXEC,
+               .help = "load image from file to address (default 0)",
+               .usage = "filename [offset [filetype]]",
+       },
+       {
+               .name = "cp15",
+               .handler = xscale_handle_cp15,
+               .mode = COMMAND_EXEC,
+               .help = "Read or write coprocessor 15 register.",
+               .usage = "register [value]",
+       },
+       COMMAND_REGISTRATION_DONE
+};
+static const struct command_registration xscale_any_command_handlers[] = {
+       {
+               .name = "debug_handler",
+               .handler = xscale_handle_debug_handler_command,
+               .mode = COMMAND_ANY,
+               .help = "Change address used for debug handler.",
+               .usage = "target address",
+       },
+       {
+               .name = "cache_clean_address",
+               .handler = xscale_handle_cache_clean_address_command,
+               .mode = COMMAND_ANY,
+               .help = "Change address used for cleaning data cache.",
+               .usage = "address",
+       },
+       {
+               .chain = xscale_exec_command_handlers,
+       },
+       COMMAND_REGISTRATION_DONE
+};
+static const struct command_registration xscale_command_handlers[] = {
+       {
+               .chain = arm_command_handlers,
+       },
+       {
+               .name = "xscale",
+               .mode = COMMAND_ANY,
+               .help = "xscale command group",
+               .chain = xscale_any_command_handlers,
+       },
+       COMMAND_REGISTRATION_DONE
+};
 
-target_type_t xscale_target =
+struct target_type xscale_target =
 {
        .name = "xscale",
 
@@ -3623,13 +3856,17 @@ target_type_t xscale_target =
        .deassert_reset = xscale_deassert_reset,
        .soft_reset_halt = NULL,
 
-       .get_gdb_reg_list = armv4_5_get_gdb_reg_list,
+       /* REVISIT on some cores, allow exporting iwmmxt registers ... */
+       .get_gdb_reg_list = arm_get_gdb_reg_list,
 
        .read_memory = xscale_read_memory,
+       .read_phys_memory = xscale_read_phys_memory,
        .write_memory = xscale_write_memory,
+       .write_phys_memory = xscale_write_phys_memory,
        .bulk_write_memory = xscale_bulk_write_memory,
-       .checksum_memory = arm7_9_checksum_memory,
-       .blank_check_memory = arm7_9_blank_check_memory,
+
+       .checksum_memory = arm_checksum_memory,
+       .blank_check_memory = arm_blank_check_memory,
 
        .run_algorithm = armv4_5_run_algorithm,
 
@@ -3638,7 +3875,7 @@ target_type_t xscale_target =
        .add_watchpoint = xscale_add_watchpoint,
        .remove_watchpoint = xscale_remove_watchpoint,
 
-       .register_commands = xscale_register_commands,
+       .commands = xscale_command_handlers,
        .target_create = xscale_target_create,
        .init_target = xscale_init_target,