openocd/src/target/arm7_9_common.c

2932 lines
89 KiB
C
Raw Normal View History

/***************************************************************************
* Copyright (C) 2005 by Dominic Rath *
* Dominic.Rath@gmx.de *
* *
* Copyright (C) 2007-2010 Øyvind Harboe *
* oyvind.harboe@zylin.com *
* *
* Copyright (C) 2008 by Spencer Oliver *
* spen@spen-soft.co.uk *
* *
* Copyright (C) 2008 by Hongtao Zheng *
* hontor@126.com *
* *
* Copyright (C) 2009 by David Brownell *
* *
* This program is free software; you can redistribute it and/or modify *
* it under the terms of the GNU General Public License as published by *
* the Free Software Foundation; either version 2 of the License, or *
* (at your option) any later version. *
* *
* This program is distributed in the hope that it will be useful, *
* but WITHOUT ANY WARRANTY; without even the implied warranty of *
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the *
* GNU General Public License for more details. *
* *
* You should have received a copy of the GNU General Public License *
* along with this program. If not, see <http://www.gnu.org/licenses/>. *
***************************************************************************/
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
#include "breakpoints.h"
#include "embeddedice.h"
#include "target_request.h"
#include "etm.h"
#include <helper/time_support.h>
#include "arm_simulator.h"
#include "arm_semihosting.h"
#include "algorithm.h"
#include "register.h"
#include "armv4_5.h"
/**
* @file
* Hold common code supporting the ARM7 and ARM9 core generations.
*
* While the ARM core implementations evolved substantially during these
* two generations, they look quite similar from the JTAG perspective.
* Both have similar debug facilities, based on the same two scan chains
* providing access to the core and to an EmbeddedICE module. Both can
* support similar ETM and ETB modules, for tracing. And both expose
* what could be viewed as "ARM Classic", with multiple processor modes,
* shadowed registers, and support for the Thumb instruction set.
*
* Processor differences include things like presence or absence of MMU
* and cache, pipeline sizes, use of a modified Harvard Architecture
* (with separate instruction and data buses from the CPU), support
* for cpu clock gating during idle, and more.
*/
static int arm7_9_debug_entry(struct target *target);
/**
* Clear watchpoints for an ARM7/9 target.
*
* @param arm7_9 Pointer to the common struct for an ARM7/9 target
* @return JTAG error status after executing queue
*/
static int arm7_9_clear_watchpoints(struct arm7_9_common *arm7_9)
{
LOG_DEBUG("-");
embeddedice_write_reg(&arm7_9->eice_cache->reg_list[EICE_W0_CONTROL_VALUE], 0x0);
embeddedice_write_reg(&arm7_9->eice_cache->reg_list[EICE_W1_CONTROL_VALUE], 0x0);
arm7_9->sw_breakpoint_count = 0;
arm7_9->sw_breakpoints_added = 0;
arm7_9->wp0_used = 0;
arm7_9->wp1_used = arm7_9->wp1_used_default;
arm7_9->wp_available = arm7_9->wp_available_max;
return jtag_execute_queue();
}
/**
* Assign a watchpoint to one of the two available hardware comparators in an
* ARM7 or ARM9 target.
*
* @param arm7_9 Pointer to the common struct for an ARM7/9 target
* @param breakpoint Pointer to the breakpoint to be used as a watchpoint
*/
static void arm7_9_assign_wp(struct arm7_9_common *arm7_9, struct breakpoint *breakpoint)
{
if (!arm7_9->wp0_used) {
arm7_9->wp0_used = 1;
breakpoint->set = 1;
arm7_9->wp_available--;
} else if (!arm7_9->wp1_used) {
arm7_9->wp1_used = 1;
breakpoint->set = 2;
arm7_9->wp_available--;
} else
LOG_ERROR("BUG: no hardware comparator available");
LOG_DEBUG("BPID: %" PRIu32 " (0x%08" TARGET_PRIxADDR ") using hw wp: %d",
breakpoint->unique_id,
breakpoint->address,
breakpoint->set);
}
/**
* Setup an ARM7/9 target's embedded ICE registers for software breakpoints.
*
* @param arm7_9 Pointer to common struct for ARM7/9 targets
* @return Error codes if there is a problem finding a watchpoint or the result
* of executing the JTAG queue
*/
static int arm7_9_set_software_breakpoints(struct arm7_9_common *arm7_9)
{
if (arm7_9->sw_breakpoints_added)
return ERROR_OK;
if (arm7_9->wp_available < 1) {
LOG_WARNING("can't enable sw breakpoints with no watchpoint unit available");
return ERROR_TARGET_RESOURCE_NOT_AVAILABLE;
}
arm7_9->wp_available--;
/* pick a breakpoint unit */
if (!arm7_9->wp0_used) {
arm7_9->sw_breakpoints_added = 1;
arm7_9->wp0_used = 3;
} else if (!arm7_9->wp1_used) {
arm7_9->sw_breakpoints_added = 2;
arm7_9->wp1_used = 3;
} else {
LOG_ERROR("BUG: both watchpoints used, but wp_available >= 1");
return ERROR_FAIL;
}
if (arm7_9->sw_breakpoints_added == 1) {
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W0_DATA_VALUE], arm7_9->arm_bkpt);
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W0_DATA_MASK], 0x0);
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W0_ADDR_MASK], 0xffffffffu);
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W0_CONTROL_MASK], ~EICE_W_CTRL_nOPC & 0xff);
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W0_CONTROL_VALUE], EICE_W_CTRL_ENABLE);
} else if (arm7_9->sw_breakpoints_added == 2) {
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W1_DATA_VALUE], arm7_9->arm_bkpt);
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W1_DATA_MASK], 0x0);
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W1_ADDR_MASK], 0xffffffffu);
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W1_CONTROL_MASK], ~EICE_W_CTRL_nOPC & 0xff);
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W1_CONTROL_VALUE], EICE_W_CTRL_ENABLE);
} else {
LOG_ERROR("BUG: both watchpoints used, but wp_available >= 1");
return ERROR_FAIL;
}
LOG_DEBUG("SW BP using hw wp: %d",
arm7_9->sw_breakpoints_added);
return jtag_execute_queue();
}
/**
* Setup the common pieces for an ARM7/9 target after reset or on startup.
*
* @param target Pointer to an ARM7/9 target to setup
* @return Result of clearing the watchpoints on the target
*/
static int arm7_9_setup(struct target *target)
{
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
return arm7_9_clear_watchpoints(arm7_9);
}
/**
* Set either a hardware or software breakpoint on an ARM7/9 target. The
* breakpoint is set up even if it is already set. Some actions, e.g. reset,
* might have erased the values in Embedded ICE.
*
* @param target Pointer to the target device to set the breakpoints on
* @param breakpoint Pointer to the breakpoint to be set
* @return For hardware breakpoints, this is the result of executing the JTAG
* queue. For software breakpoints, this will be the status of the
* required memory reads and writes
*/
static int arm7_9_set_breakpoint(struct target *target, struct breakpoint *breakpoint)
{
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
int retval = ERROR_OK;
LOG_DEBUG("BPID: %" PRIu32 ", Address: 0x%08" TARGET_PRIxADDR ", Type: %d",
breakpoint->unique_id,
breakpoint->address,
breakpoint->type);
if (target->state != TARGET_HALTED) {
LOG_WARNING("target not halted");
return ERROR_TARGET_NOT_HALTED;
}
if (breakpoint->type == BKPT_HARD) {
/* either an ARM (4 byte) or Thumb (2 byte) breakpoint */
uint32_t mask = (breakpoint->length == 4) ? 0x3u : 0x1u;
/* reassign a hw breakpoint */
if (breakpoint->set == 0)
arm7_9_assign_wp(arm7_9, breakpoint);
if (breakpoint->set == 1) {
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W0_ADDR_VALUE], breakpoint->address);
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W0_ADDR_MASK], mask);
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W0_DATA_MASK], 0xffffffffu);
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W0_CONTROL_MASK], ~EICE_W_CTRL_nOPC & 0xff);
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W0_CONTROL_VALUE], EICE_W_CTRL_ENABLE);
} else if (breakpoint->set == 2) {
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W1_ADDR_VALUE], breakpoint->address);
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W1_ADDR_MASK], mask);
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W1_DATA_MASK], 0xffffffffu);
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W1_CONTROL_MASK], ~EICE_W_CTRL_nOPC & 0xff);
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W1_CONTROL_VALUE], EICE_W_CTRL_ENABLE);
} else {
LOG_ERROR("BUG: no hardware comparator available");
return ERROR_OK;
}
retval = jtag_execute_queue();
} else if (breakpoint->type == BKPT_SOFT) {
/* did we already set this breakpoint? */
if (breakpoint->set)
return ERROR_OK;
if (breakpoint->length == 4) {
uint32_t verify = 0xffffffff;
/* keep the original instruction in target endianness */
retval = target_read_memory(target, breakpoint->address, 4, 1, breakpoint->orig_instr);
if (retval != ERROR_OK)
return retval;
/* write the breakpoint instruction in target
* endianness (arm7_9->arm_bkpt is host endian) */
retval = target_write_u32(target, breakpoint->address, arm7_9->arm_bkpt);
if (retval != ERROR_OK)
return retval;
retval = target_read_u32(target, breakpoint->address, &verify);
if (retval != ERROR_OK)
return retval;
if (verify != arm7_9->arm_bkpt) {
LOG_ERROR("Unable to set 32 bit software breakpoint at address %08" TARGET_PRIxADDR
" - check that memory is read/writable", breakpoint->address);
return ERROR_OK;
}
} else {
uint16_t verify = 0xffff;
/* keep the original instruction in target endianness */
retval = target_read_memory(target, breakpoint->address, 2, 1, breakpoint->orig_instr);
if (retval != ERROR_OK)
return retval;
/* write the breakpoint instruction in target
* endianness (arm7_9->thumb_bkpt is host endian) */
retval = target_write_u16(target, breakpoint->address, arm7_9->thumb_bkpt);
if (retval != ERROR_OK)
return retval;
retval = target_read_u16(target, breakpoint->address, &verify);
if (retval != ERROR_OK)
return retval;
if (verify != arm7_9->thumb_bkpt) {
LOG_ERROR("Unable to set thumb software breakpoint at address %08" TARGET_PRIxADDR
" - check that memory is read/writable", breakpoint->address);
return ERROR_OK;
}
}
retval = arm7_9_set_software_breakpoints(arm7_9);
if (retval != ERROR_OK)
return retval;
arm7_9->sw_breakpoint_count++;
breakpoint->set = 1;
}
return retval;
}
/**
* Unsets an existing breakpoint on an ARM7/9 target. If it is a hardware
* breakpoint, the watchpoint used will be freed and the Embedded ICE registers
* will be updated. Otherwise, the software breakpoint will be restored to its
* original instruction if it hasn't already been modified.
*
* @param target Pointer to ARM7/9 target to unset the breakpoint from
* @param breakpoint Pointer to breakpoint to be unset
* @return For hardware breakpoints, this is the result of executing the JTAG
* queue. For software breakpoints, this will be the status of the
* required memory reads and writes
*/
static int arm7_9_unset_breakpoint(struct target *target, struct breakpoint *breakpoint)
{
int retval = ERROR_OK;
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
LOG_DEBUG("BPID: %" PRIu32 ", Address: 0x%08" TARGET_PRIxADDR,
breakpoint->unique_id,
breakpoint->address);
if (!breakpoint->set) {
LOG_WARNING("breakpoint not set");
return ERROR_OK;
}
if (breakpoint->type == BKPT_HARD) {
LOG_DEBUG("BPID: %" PRIu32 " Releasing hw wp: %d",
breakpoint->unique_id,
breakpoint->set);
if (breakpoint->set == 1) {
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W0_CONTROL_VALUE], 0x0);
arm7_9->wp0_used = 0;
arm7_9->wp_available++;
} else if (breakpoint->set == 2) {
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W1_CONTROL_VALUE], 0x0);
arm7_9->wp1_used = 0;
arm7_9->wp_available++;
}
retval = jtag_execute_queue();
breakpoint->set = 0;
} else {
/* restore original instruction (kept in target endianness) */
if (breakpoint->length == 4) {
uint32_t current_instr;
/* check that user program as not modified breakpoint instruction */
retval = target_read_memory(target,
breakpoint->address, 4, 1, (uint8_t *)&current_instr);
if (retval != ERROR_OK)
return retval;
current_instr = target_buffer_get_u32(target, (uint8_t *)&current_instr);
if (current_instr == arm7_9->arm_bkpt) {
retval = target_write_memory(target,
breakpoint->address, 4, 1, breakpoint->orig_instr);
if (retval != ERROR_OK)
return retval;
}
} else {
uint16_t current_instr;
/* check that user program as not modified breakpoint instruction */
retval = target_read_memory(target,
breakpoint->address, 2, 1, (uint8_t *)&current_instr);
if (retval != ERROR_OK)
return retval;
current_instr = target_buffer_get_u16(target, (uint8_t *)&current_instr);
if (current_instr == arm7_9->thumb_bkpt) {
retval = target_write_memory(target,
breakpoint->address, 2, 1, breakpoint->orig_instr);
if (retval != ERROR_OK)
return retval;
}
}
if (--arm7_9->sw_breakpoint_count == 0) {
/* We have removed the last sw breakpoint, clear the hw breakpoint we used
*to implement it */
if (arm7_9->sw_breakpoints_added == 1)
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[
EICE_W0_CONTROL_VALUE], 0);
else if (arm7_9->sw_breakpoints_added == 2)
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[
EICE_W1_CONTROL_VALUE], 0);
}
breakpoint->set = 0;
}
return retval;
}
/**
* Add a breakpoint to an ARM7/9 target. This makes sure that there are no
* dangling breakpoints and that the desired breakpoint can be added.
*
* @param target Pointer to the target ARM7/9 device to add a breakpoint to
* @param breakpoint Pointer to the breakpoint to be added
* @return An error status if there is a problem adding the breakpoint or the
* result of setting the breakpoint
*/
int arm7_9_add_breakpoint(struct target *target, struct breakpoint *breakpoint)
{
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
if (arm7_9->breakpoint_count == 0) {
/* make sure we don't have any dangling breakpoints. This is vital upon
* GDB connect/disconnect
*/
arm7_9_clear_watchpoints(arm7_9);
}
if ((breakpoint->type == BKPT_HARD) && (arm7_9->wp_available < 1)) {
LOG_INFO("no watchpoint 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");
return ERROR_TARGET_RESOURCE_NOT_AVAILABLE;
}
if (breakpoint->type == BKPT_HARD)
arm7_9_assign_wp(arm7_9, breakpoint);
arm7_9->breakpoint_count++;
return arm7_9_set_breakpoint(target, breakpoint);
}
/**
* Removes a breakpoint from an ARM7/9 target. This will make sure there are no
* dangling breakpoints and updates available watchpoints if it is a hardware
* breakpoint.
*
* @param target Pointer to the target to have a breakpoint removed
* @param breakpoint Pointer to the breakpoint to be removed
* @return Error status if there was a problem unsetting the breakpoint or the
* watchpoints could not be cleared
*/
int arm7_9_remove_breakpoint(struct target *target, struct breakpoint *breakpoint)
{
int retval = ERROR_OK;
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
retval = arm7_9_unset_breakpoint(target, breakpoint);
if (retval != ERROR_OK)
return retval;
if (breakpoint->type == BKPT_HARD)
arm7_9->wp_available++;
arm7_9->breakpoint_count--;
if (arm7_9->breakpoint_count == 0) {
/* make sure we don't have any dangling breakpoints */
retval = arm7_9_clear_watchpoints(arm7_9);
if (retval != ERROR_OK)
return retval;
}
return ERROR_OK;
}
/**
* Sets a watchpoint for an ARM7/9 target in one of the watchpoint units. It is
* considered a bug to call this function when there are no available watchpoint
* units.
*
* @param target Pointer to an ARM7/9 target to set a watchpoint on
* @param watchpoint Pointer to the watchpoint to be set
* @return Error status if watchpoint set fails or the result of executing the
* JTAG queue
*/
static int arm7_9_set_watchpoint(struct target *target, struct watchpoint *watchpoint)
{
int retval = ERROR_OK;
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
int rw_mask = 1;
uint32_t mask;
mask = watchpoint->length - 1;
if (target->state != TARGET_HALTED) {
LOG_WARNING("target not halted");
return ERROR_TARGET_NOT_HALTED;
}
if (watchpoint->rw == WPT_ACCESS)
rw_mask = 0;
else
rw_mask = 1;
if (!arm7_9->wp0_used) {
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W0_ADDR_VALUE],
watchpoint->address);
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W0_ADDR_MASK], mask);
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W0_DATA_MASK],
watchpoint->mask);
if (watchpoint->mask != 0xffffffffu)
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W0_DATA_VALUE],
watchpoint->value);
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W0_CONTROL_MASK],
0xff & ~EICE_W_CTRL_nOPC & ~rw_mask);
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W0_CONTROL_VALUE],
EICE_W_CTRL_ENABLE | EICE_W_CTRL_nOPC | (watchpoint->rw & 1));
retval = jtag_execute_queue();
if (retval != ERROR_OK)
return retval;
watchpoint->set = 1;
arm7_9->wp0_used = 2;
} else if (!arm7_9->wp1_used) {
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W1_ADDR_VALUE],
watchpoint->address);
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W1_ADDR_MASK], mask);
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W1_DATA_MASK],
watchpoint->mask);
if (watchpoint->mask != 0xffffffffu)
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W1_DATA_VALUE],
watchpoint->value);
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W1_CONTROL_MASK],
0xff & ~EICE_W_CTRL_nOPC & ~rw_mask);
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W1_CONTROL_VALUE],
EICE_W_CTRL_ENABLE | EICE_W_CTRL_nOPC | (watchpoint->rw & 1));
retval = jtag_execute_queue();
if (retval != ERROR_OK)
return retval;
watchpoint->set = 2;
arm7_9->wp1_used = 2;
} else {
LOG_ERROR("BUG: no hardware comparator available");
return ERROR_OK;
}
return ERROR_OK;
}
/**
* Unset an existing watchpoint and clear the used watchpoint unit.
*
* @param target Pointer to the target to have the watchpoint removed
* @param watchpoint Pointer to the watchpoint to be removed
* @return Error status while trying to unset the watchpoint or the result of
* executing the JTAG queue
*/
static int arm7_9_unset_watchpoint(struct target *target, struct watchpoint *watchpoint)
{
int retval = ERROR_OK;
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
if (target->state != TARGET_HALTED) {
LOG_WARNING("target not halted");
return ERROR_TARGET_NOT_HALTED;
}
if (!watchpoint->set) {
LOG_WARNING("breakpoint not set");
return ERROR_OK;
}
if (watchpoint->set == 1) {
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W0_CONTROL_VALUE], 0x0);
retval = jtag_execute_queue();
if (retval != ERROR_OK)
return retval;
arm7_9->wp0_used = 0;
} else if (watchpoint->set == 2) {
embeddedice_set_reg(&arm7_9->eice_cache->reg_list[EICE_W1_CONTROL_VALUE], 0x0);
retval = jtag_execute_queue();
if (retval != ERROR_OK)
return retval;
arm7_9->wp1_used = 0;
}
watchpoint->set = 0;
return ERROR_OK;
}
/**
* Add a watchpoint to an ARM7/9 target. If there are no watchpoint units
* available, an error response is returned.
*
* @param target Pointer to the ARM7/9 target to add a watchpoint to
* @param watchpoint Pointer to the watchpoint to be added
* @return Error status while trying to add the watchpoint
*/
int arm7_9_add_watchpoint(struct target *target, struct watchpoint *watchpoint)
{
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
if (arm7_9->wp_available < 1)
return ERROR_TARGET_RESOURCE_NOT_AVAILABLE;
if ((watchpoint->length != 1) && (watchpoint->length != 2) && (watchpoint->length != 4))
return ERROR_TARGET_RESOURCE_NOT_AVAILABLE;
arm7_9->wp_available--;
return ERROR_OK;
}
/**
* Remove a watchpoint from an ARM7/9 target. The watchpoint will be unset and
* the used watchpoint unit will be reopened.
*
* @param target Pointer to the target to remove a watchpoint from
* @param watchpoint Pointer to the watchpoint to be removed
* @return Result of trying to unset the watchpoint
*/
int arm7_9_remove_watchpoint(struct target *target, struct watchpoint *watchpoint)
{
int retval = ERROR_OK;
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
if (watchpoint->set) {
retval = arm7_9_unset_watchpoint(target, watchpoint);
if (retval != ERROR_OK)
return retval;
}
arm7_9->wp_available++;
return ERROR_OK;
}
/**
* Restarts the target by sending a RESTART instruction and moving the JTAG
* state to IDLE. This includes a timeout waiting for DBGACK and SYSCOMP to be
* asserted by the processor.
*
* @param target Pointer to target to issue commands to
* @return Error status if there is a timeout or a problem while executing the
* JTAG queue
*/
int arm7_9_execute_sys_speed(struct target *target)
{
int retval;
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
struct arm_jtag *jtag_info = &arm7_9->jtag_info;
struct reg *dbg_stat = &arm7_9->eice_cache->reg_list[EICE_DBG_STAT];
/* set RESTART instruction */
if (arm7_9->need_bypass_before_restart) {
arm7_9->need_bypass_before_restart = 0;
retval = arm_jtag_set_instr(jtag_info->tap, 0xf, NULL, TAP_IDLE);
if (retval != ERROR_OK)
return retval;
}
retval = arm_jtag_set_instr(jtag_info->tap, 0x4, NULL, TAP_IDLE);
if (retval != ERROR_OK)
return retval;
int64_t then = timeval_ms();
bool timeout;
while (!(timeout = ((timeval_ms()-then) > 1000))) {
/* read debug status register */
embeddedice_read_reg(dbg_stat);
retval = jtag_execute_queue();
if (retval != ERROR_OK)
return retval;
if ((buf_get_u32(dbg_stat->value, EICE_DBG_STATUS_DBGACK, 1))
&& (buf_get_u32(dbg_stat->value, EICE_DBG_STATUS_SYSCOMP, 1)))
break;
if (debug_level >= 3)
alive_sleep(100);
else
keep_alive();
}
if (timeout) {
LOG_ERROR("timeout waiting for SYSCOMP & DBGACK, last DBG_STATUS: %" PRIx32 "",
buf_get_u32(dbg_stat->value, 0, dbg_stat->size));
return ERROR_TARGET_TIMEOUT;
}
return ERROR_OK;
}
/**
* Restarts the target by sending a RESTART instruction and moving the JTAG
* state to IDLE. This validates that DBGACK and SYSCOMP are set without
* waiting until they are.
*
* @param target Pointer to the target to issue commands to
* @return Always ERROR_OK
*/
static int arm7_9_execute_fast_sys_speed(struct target *target)
{
static int set;
static uint8_t check_value[4], check_mask[4];
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
struct arm_jtag *jtag_info = &arm7_9->jtag_info;
struct reg *dbg_stat = &arm7_9->eice_cache->reg_list[EICE_DBG_STAT];
int retval;
/* set RESTART instruction */
if (arm7_9->need_bypass_before_restart) {
arm7_9->need_bypass_before_restart = 0;
retval = arm_jtag_set_instr(jtag_info->tap, 0xf, NULL, TAP_IDLE);
if (retval != ERROR_OK)
return retval;
}
retval = arm_jtag_set_instr(jtag_info->tap, 0x4, NULL, TAP_IDLE);
if (retval != ERROR_OK)
return retval;
if (!set) {
/* check for DBGACK and SYSCOMP set (others don't care) */
/* NB! These are constants that must be available until after next jtag_execute() and
* we evaluate the values upon first execution in lieu of setting up these constants
* during early setup.
* */
buf_set_u32(check_value, 0, 32, 0x9);
buf_set_u32(check_mask, 0, 32, 0x9);
set = 1;
}
/* read debug status register */
embeddedice_read_reg_w_check(dbg_stat, check_value, check_mask);
return ERROR_OK;
}
/**
* Get some data from the ARM7/9 target.
*
* @param target Pointer to the ARM7/9 target to read data from
* @param size The number of 32bit words to be read
* @param buffer Pointer to the buffer that will hold the data
* @return The result of receiving data from the Embedded ICE unit
*/
int arm7_9_target_request_data(struct target *target, uint32_t size, uint8_t *buffer)
{
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
struct arm_jtag *jtag_info = &arm7_9->jtag_info;
uint32_t *data;
int retval = ERROR_OK;
uint32_t i;
data = malloc(size * (sizeof(uint32_t)));
retval = embeddedice_receive(jtag_info, data, size);
/* return the 32-bit ints in the 8-bit array */
for (i = 0; i < size; i++)
h_u32_to_le(buffer + (i * 4), data[i]);
free(data);
return retval;
}
/**
* Handles requests to an ARM7/9 target. If debug messaging is enabled, the
* target is running and the DCC control register has the W bit high, this will
* execute the request on the target.
*
* @param priv Void pointer expected to be a struct target pointer
* @return ERROR_OK unless there are issues with the JTAG queue or when reading
* from the Embedded ICE unit
*/
static int arm7_9_handle_target_request(void *priv)
{
int retval = ERROR_OK;
struct target *target = priv;
if (!target_was_examined(target))
return ERROR_OK;
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
struct arm_jtag *jtag_info = &arm7_9->jtag_info;
struct reg *dcc_control = &arm7_9->eice_cache->reg_list[EICE_COMMS_CTRL];
if (!target->dbg_msg_enabled)
return ERROR_OK;
if (target->state == TARGET_RUNNING) {
/* read DCC control register */
embeddedice_read_reg(dcc_control);
retval = jtag_execute_queue();
if (retval != ERROR_OK)
return retval;
/* check W bit */
if (buf_get_u32(dcc_control->value, 1, 1) == 1) {
uint32_t request;
retval = embeddedice_receive(jtag_info, &request, 1);
if (retval != ERROR_OK)
return retval;
retval = target_request(target, request);
if (retval != ERROR_OK)
return retval;
}
}
return ERROR_OK;
}
/**
* Polls an ARM7/9 target for its current status. If DBGACK is set, the target
* is manipulated to the right halted state based on its current state. This is
* what happens:
*
* <table>
* <tr><th > State</th><th > Action</th></tr>
* <tr><td > TARGET_RUNNING | TARGET_RESET</td>
* <td > Enters debug mode. If TARGET_RESET, pc may be checked</td></tr>
* <tr><td > TARGET_UNKNOWN</td><td > Warning is logged</td></tr>
* <tr><td > TARGET_DEBUG_RUNNING</td><td > Enters debug mode</td></tr>
* <tr><td > TARGET_HALTED</td><td > Nothing</td></tr>
* </table>
*
* If the target does not end up in the halted state, a warning is produced. If
* DBGACK is cleared, then the target is expected to either be running or
* running in debug.
*
* @param target Pointer to the ARM7/9 target to poll
* @return ERROR_OK or an error status if a command fails
*/
int arm7_9_poll(struct target *target)
{
int retval;
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
struct reg *dbg_stat = &arm7_9->eice_cache->reg_list[EICE_DBG_STAT];
/* read debug status register */
embeddedice_read_reg(dbg_stat);
retval = jtag_execute_queue();
if (retval != ERROR_OK)
return retval;
if (buf_get_u32(dbg_stat->value, EICE_DBG_STATUS_DBGACK, 1)) {
/* LOG_DEBUG("DBGACK set, dbg_state->value: 0x%x", buf_get_u32(dbg_stat->value, 0, *32));*/
if (target->state == TARGET_UNKNOWN) {
/* Starting OpenOCD with target in debug-halt */
target->state = TARGET_RUNNING;
LOG_DEBUG("DBGACK already set during server startup.");
}
if ((target->state == TARGET_RUNNING) || (target->state == TARGET_RESET)) {
target->state = TARGET_HALTED;
retval = arm7_9_debug_entry(target);
if (retval != ERROR_OK)
return retval;
if (arm_semihosting(target, &retval) != 0)
return retval;
retval = target_call_event_callbacks(target, TARGET_EVENT_HALTED);
if (retval != ERROR_OK)
return retval;
}
if (target->state == TARGET_DEBUG_RUNNING) {
target->state = TARGET_HALTED;
retval = arm7_9_debug_entry(target);
if (retval != ERROR_OK)
return retval;
retval = target_call_event_callbacks(target, TARGET_EVENT_DEBUG_HALTED);
if (retval != ERROR_OK)
return retval;
}
if (target->state != TARGET_HALTED)
LOG_WARNING(
"DBGACK set, but the target did not end up in the halted state %d",
target->state);
} else {
if (target->state != TARGET_DEBUG_RUNNING)
target->state = TARGET_RUNNING;
}
return ERROR_OK;
}
/**
* Asserts the reset (SRST) on an ARM7/9 target. Some -S targets (ARM966E-S in
* the STR912 isn't affected, ARM926EJ-S in the LPC3180 and AT91SAM9260 is
* affected) completely stop the JTAG clock while the core is held in reset
* (SRST). It isn't possible to program the halt condition once reset is
* asserted, hence a hook that allows the target to set up its reset-halt
* condition is setup prior to asserting reset.
*
* @param target Pointer to an ARM7/9 target to assert reset on
* @return ERROR_FAIL if the JTAG device does not have SRST, otherwise ERROR_OK
*/
int arm7_9_assert_reset(struct target *target)
{
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
enum reset_types jtag_reset_config = jtag_get_reset_config();
bool use_event = false;
target: improve robustness of reset command Before this change jim_target_reset() checked examined state of a target and failed without calling .assert_reset in particular target layer (and without comprehensible warning to user). Cortex-M target (which refuses access to DP under active SRST): If connection is lost then reset process fails before asserting SRST and connection with MCU is not restored. This resulted in: 1) A lot of Cortex-M MCUs required use of reset button or cycling power after firmware blocked SWD access somehow (sleep, misconfigured clock etc). If firmware blocks SWD access early during initialization, a MCU could become completely inaccessible by SWD. 2) If OpenOCD is (re)started and a MCU is in a broken state unresponsive to SWD, reset command does not work even if it could help to restore communication. Hopefully this scenario is not possible under full JTAG. jim_target_reset() in target.c now does not check examined state and delegates this task to a particular target. All targets have been checked and xx_assert_reset() (or xx_deassert_reset()) procedures were changed to check examined state if needed. Targets except arm11, cortex_a and cortex_m just fail if target is not examined although it may be possible to use at least hw reset. Left as TODO for developers familiar with these targets. cortex_m_assert_reset(): memory access errors are stored instead of immediate returning them to a higher level. Errors from less important reads/writes are ignored. Requested reset always leads to a configured action. arm11_assert_reset() just asserts hw reset in case of not examined target. cortex_a_assert_reset() works as usual in case of not examined target. Change-Id: I84fa869f4f58e2fa83b6ea75de84440d9dc3d929 Signed-off-by: Tomas Vanek <vanekt@fbl.cz> Reviewed-on: http://openocd.zylin.com/2606 Tested-by: jenkins Reviewed-by: Matthias Welwarsky <matthias@welwarsky.de> Reviewed-by: Paul Fertser <fercerpav@gmail.com>
2015-03-15 18:09:15 +00:00
/* TODO: apply hw reset signal in not examined state */
if (!(target_was_examined(target))) {
LOG_WARNING("Reset is not asserted because the target is not examined.");
LOG_WARNING("Use a reset button or power cycle the target.");
return ERROR_TARGET_NOT_EXAMINED;
}
LOG_DEBUG("target->state: %s", target_state_name(target));
if (target_has_event_action(target, TARGET_EVENT_RESET_ASSERT))
use_event = true;
else if (!(jtag_reset_config & RESET_HAS_SRST)) {
LOG_ERROR("%s: how to reset?", target_name(target));
return ERROR_FAIL;
}
/* At this point trst has been asserted/deasserted once. We would
* like to program EmbeddedICE while SRST is asserted, instead of
* depending on SRST to leave that module alone. However, many CPUs
* gate the JTAG clock while SRST is asserted; or JTAG may need
* clock stability guarantees (adaptive clocking might help).
*
* So we assume JTAG access during SRST is off the menu unless it's
* been specifically enabled.
*/
bool srst_asserted = false;
if (!use_event && !(jtag_reset_config & RESET_SRST_PULLS_TRST)
&& (jtag_reset_config & RESET_SRST_NO_GATING)) {
jtag_add_reset(0, 1);
srst_asserted = true;
}
if (target->reset_halt) {
/*
* For targets that don't support communication while SRST is
* asserted, we need to set up the reset vector catch first.
*
* When we use TRST+SRST and that's equivalent to a power-up
* reset, these settings may well be reset anyway; so setting
* them here won't matter.
*/
if (arm7_9->has_vector_catch) {
/* program vector catch register to catch reset */
embeddedice_write_reg(&arm7_9->eice_cache->reg_list[EICE_VEC_CATCH], 0x1);
/* extra runtest added as issues were found with
* certain ARM9 cores (maybe more) - AT91SAM9260
* and STR9
*/
jtag_add_runtest(1, TAP_IDLE);
} else {
/* program watchpoint unit to match on reset vector
* address
*/
embeddedice_write_reg(&arm7_9->eice_cache->reg_list[EICE_W0_ADDR_VALUE], 0x0);
embeddedice_write_reg(&arm7_9->eice_cache->reg_list[EICE_W0_ADDR_MASK], 0x3);
embeddedice_write_reg(&arm7_9->eice_cache->reg_list[EICE_W0_DATA_MASK], 0xffffffff);
embeddedice_write_reg(&arm7_9->eice_cache->reg_list[EICE_W0_CONTROL_VALUE], EICE_W_CTRL_ENABLE);
embeddedice_write_reg(&arm7_9->eice_cache->reg_list[EICE_W0_CONTROL_MASK], ~EICE_W_CTRL_nOPC & 0xff);
}
}
if (use_event)
target_handle_event(target, TARGET_EVENT_RESET_ASSERT);
else {
/* If we use SRST ... we'd like to issue just SRST, but the
* board or chip may be set up so we have to assert TRST as
* well. On some chips that combination is equivalent to a
* power-up reset, and generally clobbers EICE state.
*/
if (jtag_reset_config & RESET_SRST_PULLS_TRST)
jtag_add_reset(1, 1);
else if (!srst_asserted)
jtag_add_reset(0, 1);
jtag_add_sleep(50000);
}
target->state = TARGET_RESET;
register_cache_invalidate(arm7_9->arm.core_cache);
/* REVISIT why isn't standard debug entry logic sufficient?? */
if (target->reset_halt && (!(jtag_reset_config & RESET_SRST_PULLS_TRST) || use_event)) {
/* debug entry was prepared above */
target->debug_reason = DBG_REASON_DBGRQ;
}
return ERROR_OK;
}
/**
* Deassert the reset (SRST) signal on an ARM7/9 target. If SRST pulls TRST
* and the target is being reset into a halt, a warning will be triggered
* because it is not possible to reset into a halted mode in this case. The
* target is halted using the target's functions.
*
* @param target Pointer to the target to have the reset deasserted
* @return ERROR_OK or an error from polling or halting the target
*/
int arm7_9_deassert_reset(struct target *target)
{
int retval = ERROR_OK;
LOG_DEBUG("target->state: %s", target_state_name(target));
/* deassert reset lines */
jtag_add_reset(0, 0);
/* In case polling is disabled, we need to examine the
* target and poll here for this target to work correctly.
*
* Otherwise, e.g. halt will fail afterwards with bogus
* error messages as halt will believe that reset is
* still in effect.
*/
retval = target_examine_one(target);
if (retval != ERROR_OK)
return retval;
retval = target_poll(target);
if (retval != ERROR_OK)
return retval;
enum reset_types jtag_reset_config = jtag_get_reset_config();
if (target->reset_halt && (jtag_reset_config & RESET_SRST_PULLS_TRST) != 0) {
LOG_WARNING(
"srst pulls trst - can not reset into halted mode. Issuing halt after reset.");
retval = target_halt(target);
if (retval != ERROR_OK)
return retval;
}
return retval;
}
/**
* Clears the halt condition for an ARM7/9 target. If it isn't coming out of
* reset and if DBGRQ is used, it is programmed to be deasserted. If the reset
* vector catch was used, it is restored. Otherwise, the control value is
* restored and the watchpoint unit is restored if it was in use.
*
* @param target Pointer to the ARM7/9 target to have halt cleared
* @return Always ERROR_OK
*/
static int arm7_9_clear_halt(struct target *target)
{
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
struct reg *dbg_ctrl = &arm7_9->eice_cache->reg_list[EICE_DBG_CTRL];
/* we used DBGRQ only if we didn't come out of reset */
if (!arm7_9->debug_entry_from_reset && arm7_9->use_dbgrq) {
/* program EmbeddedICE Debug Control Register to deassert DBGRQ
*/
buf_set_u32(dbg_ctrl->value, EICE_DBG_CONTROL_DBGRQ, 1, 0);
embeddedice_store_reg(dbg_ctrl);
} else {
if (arm7_9->debug_entry_from_reset && arm7_9->has_vector_catch) {
/* if we came out of reset, and vector catch is supported, we used
* vector catch to enter debug state
* restore the register in that case
*/
embeddedice_store_reg(&arm7_9->eice_cache->reg_list[EICE_VEC_CATCH]);
} else {
/* restore registers if watchpoint unit 0 was in use
*/
if (arm7_9->wp0_used) {
if (arm7_9->debug_entry_from_reset)
embeddedice_store_reg(&arm7_9->eice_cache->reg_list[
EICE_W0_ADDR_VALUE]);
embeddedice_store_reg(&arm7_9->eice_cache->reg_list[
EICE_W0_ADDR_MASK]);
embeddedice_store_reg(&arm7_9->eice_cache->reg_list[
EICE_W0_DATA_MASK]);
embeddedice_store_reg(&arm7_9->eice_cache->reg_list[
EICE_W0_CONTROL_MASK]);
}
/* control value always has to be restored, as it was either disabled,
* or enabled with possibly different bits
*/
embeddedice_store_reg(&arm7_9->eice_cache->reg_list[EICE_W0_CONTROL_VALUE]);
}
}
return ERROR_OK;
}
/**
* Issue a software reset and halt to an ARM7/9 target. The target is halted
* and then there is a wait until the processor shows the halt. This wait can
* timeout and results in an error being returned. The software reset involves
* clearing the halt, updating the debug control register, changing to ARM mode,
* reset of the program counter, and reset of all of the registers.
*
* @param target Pointer to the ARM7/9 target to be reset and halted by software
* @return Error status if any of the commands fail, otherwise ERROR_OK
*/
int arm7_9_soft_reset_halt(struct target *target)
{
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
struct arm *arm = &arm7_9->arm;
struct reg *dbg_stat = &arm7_9->eice_cache->reg_list[EICE_DBG_STAT];
struct reg *dbg_ctrl = &arm7_9->eice_cache->reg_list[EICE_DBG_CTRL];
int i;
int retval;
/* FIX!!! replace some of this code with tcl commands
*
* halt # the halt command is synchronous
* armv4_5 core_state arm
*
*/
retval = target_halt(target);
if (retval != ERROR_OK)
return retval;
long long then = timeval_ms();
int timeout;
while (!(timeout = ((timeval_ms()-then) > 1000))) {
if (buf_get_u32(dbg_stat->value, EICE_DBG_STATUS_DBGACK, 1) != 0)
break;
embeddedice_read_reg(dbg_stat);
retval = jtag_execute_queue();
if (retval != ERROR_OK)
return retval;
if (debug_level >= 3)
alive_sleep(100);
else
keep_alive();
}
if (timeout) {
LOG_ERROR("Failed to halt CPU after 1 sec");
return ERROR_TARGET_TIMEOUT;
}
target->state = TARGET_HALTED;
/* program EmbeddedICE Debug Control Register to assert DBGACK and INTDIS
* ensure that DBGRQ is cleared
*/
buf_set_u32(dbg_ctrl->value, EICE_DBG_CONTROL_DBGACK, 1, 1);
buf_set_u32(dbg_ctrl->value, EICE_DBG_CONTROL_DBGRQ, 1, 0);
buf_set_u32(dbg_ctrl->value, EICE_DBG_CONTROL_INTDIS, 1, 1);
embeddedice_store_reg(dbg_ctrl);
retval = arm7_9_clear_halt(target);
if (retval != ERROR_OK)
return retval;
/* if the target is in Thumb state, change to ARM state */
if (buf_get_u32(dbg_stat->value, EICE_DBG_STATUS_ITBIT, 1)) {
uint32_t r0_thumb, pc_thumb;
LOG_DEBUG("target entered debug from Thumb state, changing to ARM");
/* Entered debug from Thumb mode */
arm->core_state = ARM_STATE_THUMB;
arm7_9->change_to_arm(target, &r0_thumb, &pc_thumb);
}
/* REVISIT likewise for bit 5 -- switch Jazelle-to-ARM */
/* all register content is now invalid */
register_cache_invalidate(arm->core_cache);
/* SVC, ARM state, IRQ and FIQ disabled */
uint32_t cpsr;
cpsr = buf_get_u32(arm->cpsr->value, 0, 32);
cpsr &= ~0xff;
cpsr |= 0xd3;
arm_set_cpsr(arm, cpsr);
arm->cpsr->dirty = true;
/* start fetching from 0x0 */
buf_set_u32(arm->pc->value, 0, 32, 0x0);
arm->pc->dirty = true;
arm->pc->valid = true;
/* reset registers */
for (i = 0; i <= 14; i++) {
struct reg *r = arm_reg_current(arm, i);
buf_set_u32(r->value, 0, 32, 0xffffffff);
r->dirty = true;
r->valid = true;
}
retval = target_call_event_callbacks(target, TARGET_EVENT_HALTED);
if (retval != ERROR_OK)
return retval;
return ERROR_OK;
}
/**
* Halt an ARM7/9 target. This is accomplished by either asserting the DBGRQ
* line or by programming a watchpoint to trigger on any address. It is
* considered a bug to call this function while the target is in the
* TARGET_RESET state.
*
* @param target Pointer to the ARM7/9 target to be halted
* @return Always ERROR_OK
*/
int arm7_9_halt(struct target *target)
{
if (target->state == TARGET_RESET) {
LOG_ERROR(
"BUG: arm7/9 does not support halt during reset. This is handled in arm7_9_assert_reset()");
return ERROR_OK;
}
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
struct reg *dbg_ctrl = &arm7_9->eice_cache->reg_list[EICE_DBG_CTRL];
LOG_DEBUG("target->state: %s",
target_state_name(target));
if (target->state == TARGET_HALTED) {
LOG_DEBUG("target was already halted");
return ERROR_OK;
}
if (target->state == TARGET_UNKNOWN)
LOG_WARNING("target was in unknown state when halt was requested");
if (arm7_9->use_dbgrq) {
/* program EmbeddedICE Debug Control Register to assert DBGRQ
*/
if (arm7_9->set_special_dbgrq)
arm7_9->set_special_dbgrq(target);
else {
buf_set_u32(dbg_ctrl->value, EICE_DBG_CONTROL_DBGRQ, 1, 1);
embeddedice_store_reg(dbg_ctrl);
}
} else {
/* program watchpoint unit to match on any address
*/
embeddedice_write_reg(&arm7_9->eice_cache->reg_list[EICE_W0_ADDR_MASK], 0xffffffff);
embeddedice_write_reg(&arm7_9->eice_cache->reg_list[EICE_W0_DATA_MASK], 0xffffffff);
embeddedice_write_reg(&arm7_9->eice_cache->reg_list[EICE_W0_CONTROL_VALUE],
EICE_W_CTRL_ENABLE);
embeddedice_write_reg(&arm7_9->eice_cache->reg_list[EICE_W0_CONTROL_MASK],
~EICE_W_CTRL_nOPC & 0xff);
}
target->debug_reason = DBG_REASON_DBGRQ;
return ERROR_OK;
}
/**
* Handle an ARM7/9 target's entry into debug mode. The halt is cleared on the
* ARM. The JTAG queue is then executed and the reason for debug entry is
* examined. Once done, the target is verified to be halted and the processor
* is forced into ARM mode. The core registers are saved for the current core
* mode and the program counter (register 15) is updated as needed. The core
* registers and CPSR and SPSR are saved for restoration later.
*
* @param target Pointer to target that is entering debug mode
* @return Error code if anything fails, otherwise ERROR_OK
*/
static int arm7_9_debug_entry(struct target *target)
{
int i;
uint32_t context[16];
uint32_t *context_p[16];
uint32_t r0_thumb, pc_thumb;
uint32_t cpsr, cpsr_mask = 0;
int retval;
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
struct arm *arm = &arm7_9->arm;
struct reg *dbg_stat = &arm7_9->eice_cache->reg_list[EICE_DBG_STAT];
struct reg *dbg_ctrl = &arm7_9->eice_cache->reg_list[EICE_DBG_CTRL];
#ifdef _DEBUG_ARM7_9_
LOG_DEBUG("-");
#endif
/* program EmbeddedICE Debug Control Register to assert DBGACK and INTDIS
* ensure that DBGRQ is cleared
*/
buf_set_u32(dbg_ctrl->value, EICE_DBG_CONTROL_DBGACK, 1, 1);
buf_set_u32(dbg_ctrl->value, EICE_DBG_CONTROL_DBGRQ, 1, 0);
buf_set_u32(dbg_ctrl->value, EICE_DBG_CONTROL_INTDIS, 1, 1);
embeddedice_store_reg(dbg_ctrl);
retval = arm7_9_clear_halt(target);
if (retval != ERROR_OK)
return retval;
retval = jtag_execute_queue();
if (retval != ERROR_OK)
return retval;
retval = arm7_9->examine_debug_reason(target);
if (retval != ERROR_OK)
return retval;
if (target->state != TARGET_HALTED) {
LOG_WARNING("target not halted");
return ERROR_TARGET_NOT_HALTED;
}
/* if the target is in Thumb state, change to ARM state */
if (buf_get_u32(dbg_stat->value, EICE_DBG_STATUS_ITBIT, 1)) {
LOG_DEBUG("target entered debug from Thumb state");
/* Entered debug from Thumb mode */
arm->core_state = ARM_STATE_THUMB;
cpsr_mask = 1 << 5;
arm7_9->change_to_arm(target, &r0_thumb, &pc_thumb);
LOG_DEBUG("r0_thumb: 0x%8.8" PRIx32
", pc_thumb: 0x%8.8" PRIx32, r0_thumb, pc_thumb);
} else if (buf_get_u32(dbg_stat->value, 5, 1)) {
/* \todo Get some vaguely correct handling of Jazelle, if
* anyone ever uses it and full info becomes available.
* See ARM9EJS TRM B.7.1 for how to switch J->ARM; and
* B.7.3 for the reverse. That'd be the bare minimum...
*/
LOG_DEBUG("target entered debug from Jazelle state");
arm->core_state = ARM_STATE_JAZELLE;
cpsr_mask = 1 << 24;
LOG_ERROR("Jazelle debug entry -- BROKEN!");
} else {
LOG_DEBUG("target entered debug from ARM state");
/* Entered debug from ARM mode */
arm->core_state = ARM_STATE_ARM;
}
for (i = 0; i < 16; i++)
context_p[i] = &context[i];
/* save core registers (r0 - r15 of current core mode) */
arm7_9->read_core_regs(target, 0xffff, context_p);
arm7_9->read_xpsr(target, &cpsr, 0);
retval = jtag_execute_queue();
if (retval != ERROR_OK)
return retval;
/* Sync our CPSR copy with J or T bits EICE reported, but
* which we then erased by putting the core into ARM mode.
*/
arm_set_cpsr(arm, cpsr | cpsr_mask);
if (!is_arm_mode(arm->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",
arm_mode_name(arm->core_mode));
if (arm->core_state == ARM_STATE_THUMB) {
LOG_DEBUG("thumb state, applying fixups");
context[0] = r0_thumb;
context[15] = pc_thumb;
} else if (arm->core_state == ARM_STATE_ARM) {
/* adjust value stored by STM */
context[15] -= 3 * 4;
}
if ((target->debug_reason != DBG_REASON_DBGRQ) || (!arm7_9->use_dbgrq))
context[15] -= 3 * ((arm->core_state == ARM_STATE_ARM) ? 4 : 2);
else
context[15] -= arm7_9->dbgreq_adjust_pc *
((arm->core_state == ARM_STATE_ARM) ? 4 : 2);
for (i = 0; i <= 15; i++) {
struct reg *r = arm_reg_current(arm, i);
LOG_DEBUG("r%i: 0x%8.8" PRIx32 "", i, context[i]);
buf_set_u32(r->value, 0, 32, context[i]);
/* r0 and r15 (pc) have to be restored later */
r->dirty = (i == 0) || (i == 15);
r->valid = true;
}
LOG_DEBUG("entered debug state at PC 0x%" PRIx32 "", context[15]);
/* exceptions other than USR & SYS have a saved program status register */
if (arm->spsr) {
uint32_t spsr;
arm7_9->read_xpsr(target, &spsr, 1);
retval = jtag_execute_queue();
if (retval != ERROR_OK)
return retval;
buf_set_u32(arm->spsr->value, 0, 32, spsr);
arm->spsr->dirty = false;
arm->spsr->valid = true;
}
retval = jtag_execute_queue();
if (retval != ERROR_OK)
return retval;
if (arm7_9->post_debug_entry) {
retval = arm7_9->post_debug_entry(target);
if (retval != ERROR_OK)
return retval;
}
return ERROR_OK;
}
/**
* Validate the full context for an ARM7/9 target in all processor modes. If
* there are any invalid registers for the target, they will all be read. This
* includes the PSR.
*
* @param target Pointer to the ARM7/9 target to capture the full context from
* @return Error if the target is not halted, has an invalid core mode, or if
* the JTAG queue fails to execute
*/
static int arm7_9_full_context(struct target *target)
{
int i;
int retval;
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
struct arm *arm = &arm7_9->arm;
struct {
uint32_t value;
uint8_t *reg_p;
} read_cache[6 * (16 + 1)];
int read_cache_idx = 0;
LOG_DEBUG("-");
if (target->state != TARGET_HALTED) {
LOG_WARNING("target not halted");
return ERROR_TARGET_NOT_HALTED;
}
if (!is_arm_mode(arm->core_mode)) {
LOG_ERROR("not a valid arm core mode - communication failure?");
return ERROR_FAIL;
}
/* iterate through processor modes (User, FIQ, IRQ, SVC, ABT, UND)
* SYS shares registers with User, so we don't touch SYS
*/
for (i = 0; i < 6; i++) {
uint32_t mask = 0;
uint32_t *reg_p[16];
int j;
bool valid = true;
/* check if there are invalid registers in the current mode
*/
for (j = 0; j <= 16; j++) {
if (!ARMV4_5_CORE_REG_MODE(arm->core_cache, armv4_5_number_to_mode(i), j).valid)
valid = false;
}
if (!valid) {
uint32_t tmp_cpsr;
/* change processor mode (and mask T bit) */
tmp_cpsr = buf_get_u32(arm->cpsr->value, 0, 8)
& 0xe0;
tmp_cpsr |= armv4_5_number_to_mode(i);
tmp_cpsr &= ~0x20;
arm7_9->write_xpsr_im8(target, tmp_cpsr & 0xff, 0, 0);
for (j = 0; j < 15; j++) {
if (!ARMV4_5_CORE_REG_MODE(arm->core_cache,
armv4_5_number_to_mode(i), j).valid) {
read_cache[read_cache_idx].reg_p = ARMV4_5_CORE_REG_MODE(
arm->core_cache,
armv4_5_number_to_mode(i),
j).value;
reg_p[j] = &read_cache[read_cache_idx].value;
read_cache_idx++;
mask |= 1 << j;
ARMV4_5_CORE_REG_MODE(arm->core_cache,
armv4_5_number_to_mode(i),
j).valid = true;
ARMV4_5_CORE_REG_MODE(arm->core_cache,
armv4_5_number_to_mode(i),
j).dirty = false;
}
}
/* if only the PSR is invalid, mask is all zeroes */
if (mask)
arm7_9->read_core_regs(target, mask, reg_p);
/* check if the PSR has to be read */
if (!ARMV4_5_CORE_REG_MODE(arm->core_cache, armv4_5_number_to_mode(i),
16).valid) {
read_cache[read_cache_idx].reg_p = ARMV4_5_CORE_REG_MODE(arm->core_cache,
armv4_5_number_to_mode(i), 16).value;
arm7_9->read_xpsr(target, &read_cache[read_cache_idx].value, 1);
read_cache_idx++;
ARMV4_5_CORE_REG_MODE(arm->core_cache, armv4_5_number_to_mode(i),
16).valid = true;
ARMV4_5_CORE_REG_MODE(arm->core_cache, armv4_5_number_to_mode(i),
16).dirty = false;
}
}
}
/* restore processor mode (mask T bit) */
arm7_9->write_xpsr_im8(target,
buf_get_u32(arm->cpsr->value, 0, 8) & ~0x20, 0, 0);
retval = jtag_execute_queue();
if (retval != ERROR_OK)
return retval;
/*
* FIXME: regs in cache should be tagged as 'valid' only now,
* not before the jtag_execute_queue()
*/
while (read_cache_idx) {
read_cache_idx--;
buf_set_u32(read_cache[read_cache_idx].reg_p, 0, 32, read_cache[read_cache_idx].value);
}
return ERROR_OK;
}
/**
* Restore the processor context on an ARM7/9 target. The full processor
* context is analyzed to see if any of the registers are dirty on this end, but
* have a valid new value. If this is the case, the processor is changed to the
* appropriate mode and the new register values are written out to the
* processor. If there happens to be a dirty register with an invalid value, an
* error will be logged.
*
* @param target Pointer to the ARM7/9 target to have its context restored
* @return Error status if the target is not halted or the core mode in the
* armv4_5 struct is invalid.
*/
static int arm7_9_restore_context(struct target *target)
{
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
struct arm *arm = &arm7_9->arm;
struct reg *reg;
enum arm_mode current_mode = arm->core_mode;
int i, j;
bool dirty;
int mode_change;
LOG_DEBUG("-");
if (target->state != TARGET_HALTED) {
LOG_WARNING("target not halted");
return ERROR_TARGET_NOT_HALTED;
}
if (arm7_9->pre_restore_context)
arm7_9->pre_restore_context(target);
if (!is_arm_mode(arm->core_mode)) {
LOG_ERROR("not a valid arm core mode - communication failure?");
return ERROR_FAIL;
}
/* iterate through processor modes (User, FIQ, IRQ, SVC, ABT, UND)
* SYS shares registers with User, so we don't touch SYS
*/
for (i = 0; i < 6; i++) {
LOG_DEBUG("examining %s mode",
arm_mode_name(arm->core_mode));
dirty = false;
mode_change = 0;
/* check if there are dirty registers in the current mode
*/
for (j = 0; j <= 16; j++) {
reg = &ARMV4_5_CORE_REG_MODE(arm->core_cache, armv4_5_number_to_mode(i), j);
if (reg->dirty) {
if (reg->valid) {
dirty = true;
LOG_DEBUG("examining dirty reg: %s", reg->name);
struct arm_reg *reg_arch_info;
reg_arch_info = reg->arch_info;
if ((reg_arch_info->mode != ARM_MODE_ANY)
&& (reg_arch_info->mode != current_mode)
&& !((reg_arch_info->mode == ARM_MODE_USR)
&& (arm->core_mode == ARM_MODE_SYS))
&& !((reg_arch_info->mode == ARM_MODE_SYS)
&& (arm->core_mode == ARM_MODE_USR))) {
mode_change = 1;
LOG_DEBUG("require mode change");
}
} else
LOG_ERROR("BUG: dirty register '%s', but no valid data",
reg->name);
}
}
if (dirty) {
uint32_t mask = 0x0;
int num_regs = 0;
uint32_t regs[16];
if (mode_change) {
uint32_t tmp_cpsr;
/* change processor mode (mask T bit) */
tmp_cpsr = buf_get_u32(arm->cpsr->value,
0, 8) & 0xe0;
tmp_cpsr |= armv4_5_number_to_mode(i);
tmp_cpsr &= ~0x20;
arm7_9->write_xpsr_im8(target, tmp_cpsr & 0xff, 0, 0);
current_mode = armv4_5_number_to_mode(i);
}
for (j = 0; j <= 14; j++) {
reg = &ARMV4_5_CORE_REG_MODE(arm->core_cache,
armv4_5_number_to_mode(i),
j);
if (reg->dirty) {
regs[j] = buf_get_u32(reg->value, 0, 32);
mask |= 1 << j;
num_regs++;
reg->dirty = false;
reg->valid = true;
LOG_DEBUG("writing register %i mode %s "
"with value 0x%8.8" PRIx32, j,
arm_mode_name(arm->core_mode),
regs[j]);
}
}
if (mask)
arm7_9->write_core_regs(target, mask, regs);
reg =
&ARMV4_5_CORE_REG_MODE(arm->core_cache, armv4_5_number_to_mode(
i), 16);
struct arm_reg *reg_arch_info;
reg_arch_info = reg->arch_info;
if ((reg->dirty) && (reg_arch_info->mode != ARM_MODE_ANY)) {
LOG_DEBUG("writing SPSR of mode %i with value 0x%8.8" PRIx32 "",
i,
buf_get_u32(reg->value, 0, 32));
arm7_9->write_xpsr(target, buf_get_u32(reg->value, 0, 32), 1);
}
}
}
if (!arm->cpsr->dirty && (arm->core_mode != current_mode)) {
/* restore processor mode (mask T bit) */
uint32_t tmp_cpsr;
tmp_cpsr = buf_get_u32(arm->cpsr->value, 0, 8) & 0xE0;
tmp_cpsr |= armv4_5_number_to_mode(i);
tmp_cpsr &= ~0x20;
LOG_DEBUG("writing lower 8 bit of cpsr with value 0x%2.2x", (unsigned)(tmp_cpsr));
arm7_9->write_xpsr_im8(target, tmp_cpsr & 0xff, 0, 0);
} else if (arm->cpsr->dirty) {
/* CPSR has been changed, full restore necessary (mask T bit) */
LOG_DEBUG("writing cpsr with value 0x%8.8" PRIx32,
buf_get_u32(arm->cpsr->value, 0, 32));
arm7_9->write_xpsr(target,
buf_get_u32(arm->cpsr->value, 0, 32)
& ~0x20, 0);
arm->cpsr->dirty = false;
arm->cpsr->valid = true;
}
/* restore PC */
LOG_DEBUG("writing PC with value 0x%8.8" PRIx32,
buf_get_u32(arm->pc->value, 0, 32));
arm7_9->write_pc(target, buf_get_u32(arm->pc->value, 0, 32));
arm->pc->dirty = false;
return ERROR_OK;
}
/**
* Restart the core of an ARM7/9 target. A RESTART command is sent to the
* instruction register and the JTAG state is set to TAP_IDLE causing a core
* restart.
*
* @param target Pointer to the ARM7/9 target to be restarted
* @return Result of executing the JTAG queue
*/
static int arm7_9_restart_core(struct target *target)
{
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
struct arm_jtag *jtag_info = &arm7_9->jtag_info;
int retval;
/* set RESTART instruction */
if (arm7_9->need_bypass_before_restart) {
arm7_9->need_bypass_before_restart = 0;
retval = arm_jtag_set_instr(jtag_info->tap, 0xf, NULL, TAP_IDLE);
if (retval != ERROR_OK)
return retval;
}
retval = arm_jtag_set_instr(jtag_info->tap, 0x4, NULL, TAP_IDLE);
if (retval != ERROR_OK)
return retval;
jtag_add_runtest(1, TAP_IDLE);
return jtag_execute_queue();
}
/**
* Enable the watchpoints on an ARM7/9 target. The target's watchpoints are
* iterated through and are set on the target if they aren't already set.
*
* @param target Pointer to the ARM7/9 target to enable watchpoints on
*/
static void arm7_9_enable_watchpoints(struct target *target)
{
struct watchpoint *watchpoint = target->watchpoints;
while (watchpoint) {
if (watchpoint->set == 0)
arm7_9_set_watchpoint(target, watchpoint);
watchpoint = watchpoint->next;
}
}
/**
* Enable the breakpoints on an ARM7/9 target. The target's breakpoints are
* iterated through and are set on the target.
*
* @param target Pointer to the ARM7/9 target to enable breakpoints on
*/
static void arm7_9_enable_breakpoints(struct target *target)
{
struct breakpoint *breakpoint = target->breakpoints;
/* set any pending breakpoints */
while (breakpoint) {
arm7_9_set_breakpoint(target, breakpoint);
breakpoint = breakpoint->next;
}
}
int arm7_9_resume(struct target *target,
int current,
target_addr_t address,
int handle_breakpoints,
int debug_execution)
{
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
struct arm *arm = &arm7_9->arm;
struct reg *dbg_ctrl = &arm7_9->eice_cache->reg_list[EICE_DBG_CTRL];
int err, retval = ERROR_OK;
LOG_DEBUG("-");
if (target->state != TARGET_HALTED) {
LOG_WARNING("target not halted");
return ERROR_TARGET_NOT_HALTED;
}
if (!debug_execution)
target_free_all_working_areas(target);
/* current = 1: continue on current pc, otherwise continue at <address> */
if (!current)
buf_set_u32(arm->pc->value, 0, 32, address);
uint32_t current_pc;
current_pc = buf_get_u32(arm->pc->value, 0, 32);
/* the front-end may request us not to handle breakpoints */
if (handle_breakpoints) {
struct breakpoint *breakpoint;
breakpoint = breakpoint_find(target,
buf_get_u32(arm->pc->value, 0, 32));
if (breakpoint != NULL) {
LOG_DEBUG("unset breakpoint at 0x%8.8" TARGET_PRIxADDR " (id: %" PRIu32,
breakpoint->address,
breakpoint->unique_id);
retval = arm7_9_unset_breakpoint(target, breakpoint);
if (retval != ERROR_OK)
return retval;
/* calculate PC of next instruction */
uint32_t next_pc;
retval = arm_simulate_step(target, &next_pc);
if (retval != ERROR_OK) {
uint32_t current_opcode;
target_read_u32(target, current_pc, &current_opcode);
LOG_ERROR(
"Couldn't calculate PC of next instruction, current opcode was 0x%8.8" PRIx32 "",
current_opcode);
return retval;
}
LOG_DEBUG("enable single-step");
arm7_9->enable_single_step(target, next_pc);
target->debug_reason = DBG_REASON_SINGLESTEP;
retval = arm7_9_restore_context(target);
if (retval != ERROR_OK)
return retval;
if (arm->core_state == ARM_STATE_ARM)
arm7_9->branch_resume(target);
else if (arm->core_state == ARM_STATE_THUMB)
arm7_9->branch_resume_thumb(target);
else {
LOG_ERROR("unhandled core state");
return ERROR_FAIL;
}
buf_set_u32(dbg_ctrl->value, EICE_DBG_CONTROL_DBGACK, 1, 0);
embeddedice_write_reg(dbg_ctrl,
buf_get_u32(dbg_ctrl->value, 0, dbg_ctrl->size));
err = arm7_9_execute_sys_speed(target);
LOG_DEBUG("disable single-step");
arm7_9->disable_single_step(target);
if (err != ERROR_OK) {
retval = arm7_9_set_breakpoint(target, breakpoint);
if (retval != ERROR_OK)
return retval;
target->state = TARGET_UNKNOWN;
return err;
}
retval = arm7_9_debug_entry(target);
if (retval != ERROR_OK)
return retval;
LOG_DEBUG("new PC after step: 0x%8.8" PRIx32,
buf_get_u32(arm->pc->value, 0, 32));
LOG_DEBUG("set breakpoint at 0x%8.8" TARGET_PRIxADDR "", breakpoint->address);
retval = arm7_9_set_breakpoint(target, breakpoint);
if (retval != ERROR_OK)
return retval;
}
}
/* enable any pending breakpoints and watchpoints */
arm7_9_enable_breakpoints(target);
arm7_9_enable_watchpoints(target);
retval = arm7_9_restore_context(target);
if (retval != ERROR_OK)
return retval;
if (arm->core_state == ARM_STATE_ARM)
arm7_9->branch_resume(target);
else if (arm->core_state == ARM_STATE_THUMB)
arm7_9->branch_resume_thumb(target);
else {
LOG_ERROR("unhandled core state");
return ERROR_FAIL;
}
/* deassert DBGACK and INTDIS */
buf_set_u32(dbg_ctrl->value, EICE_DBG_CONTROL_DBGACK, 1, 0);
/* INTDIS only when we really resume, not during debug execution */
if (!debug_execution)
buf_set_u32(dbg_ctrl->value, EICE_DBG_CONTROL_INTDIS, 1, 0);
embeddedice_write_reg(dbg_ctrl, buf_get_u32(dbg_ctrl->value, 0, dbg_ctrl->size));
retval = arm7_9_restart_core(target);
if (retval != ERROR_OK)
return retval;
target->debug_reason = DBG_REASON_NOTHALTED;
if (!debug_execution) {
/* registers are now invalid */
register_cache_invalidate(arm->core_cache);
target->state = TARGET_RUNNING;
retval = target_call_event_callbacks(target, TARGET_EVENT_RESUMED);
if (retval != ERROR_OK)
return retval;
} else {
target->state = TARGET_DEBUG_RUNNING;
retval = target_call_event_callbacks(target, TARGET_EVENT_DEBUG_RESUMED);
if (retval != ERROR_OK)
return retval;
}
LOG_DEBUG("target resumed");
return ERROR_OK;
}
void arm7_9_enable_eice_step(struct target *target, uint32_t next_pc)
{
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
struct arm *arm = &arm7_9->arm;
uint32_t current_pc;
current_pc = buf_get_u32(arm->pc->value, 0, 32);
if (next_pc != current_pc) {
/* setup an inverse breakpoint on the current PC
* - comparator 1 matches the current address
* - rangeout from comparator 1 is connected to comparator 0 rangein
* - comparator 0 matches any address, as long as rangein is low */
embeddedice_write_reg(&arm7_9->eice_cache->reg_list[EICE_W0_ADDR_MASK], 0xffffffff);
embeddedice_write_reg(&arm7_9->eice_cache->reg_list[EICE_W0_DATA_MASK], 0xffffffff);
embeddedice_write_reg(&arm7_9->eice_cache->reg_list[EICE_W0_CONTROL_VALUE],
EICE_W_CTRL_ENABLE);
embeddedice_write_reg(&arm7_9->eice_cache->reg_list[EICE_W0_CONTROL_MASK],
~(EICE_W_CTRL_RANGE | EICE_W_CTRL_nOPC) & 0xff);
embeddedice_write_reg(&arm7_9->eice_cache->reg_list[EICE_W1_ADDR_VALUE],
current_pc);
embeddedice_write_reg(&arm7_9->eice_cache->reg_list[EICE_W1_ADDR_MASK], 0);
embeddedice_write_reg(&arm7_9->eice_cache->reg_list[EICE_W1_DATA_MASK], 0xffffffff);
embeddedice_write_reg(&arm7_9->eice_cache->reg_list[EICE_W1_CONTROL_VALUE], 0x0);
embeddedice_write_reg(&arm7_9->eice_cache->reg_list[EICE_W1_CONTROL_MASK],
~EICE_W_CTRL_nOPC & 0xff);
} else {
embeddedice_write_reg(&arm7_9->eice_cache->reg_list[EICE_W0_ADDR_MASK], 0xffffffff);
embeddedice_write_reg(&arm7_9->eice_cache->reg_list[EICE_W0_DATA_MASK], 0xffffffff);
embeddedice_write_reg(&arm7_9->eice_cache->reg_list[EICE_W0_CONTROL_VALUE], 0x0);
embeddedice_write_reg(&arm7_9->eice_cache->reg_list[EICE_W0_CONTROL_MASK], 0xff);
embeddedice_write_reg(&arm7_9->eice_cache->reg_list[EICE_W1_ADDR_VALUE], next_pc);
embeddedice_write_reg(&arm7_9->eice_cache->reg_list[EICE_W1_ADDR_MASK], 0);
embeddedice_write_reg(&arm7_9->eice_cache->reg_list[EICE_W1_DATA_MASK], 0xffffffff);
embeddedice_write_reg(&arm7_9->eice_cache->reg_list[EICE_W1_CONTROL_VALUE],
EICE_W_CTRL_ENABLE);
embeddedice_write_reg(&arm7_9->eice_cache->reg_list[EICE_W1_CONTROL_MASK],
~EICE_W_CTRL_nOPC & 0xff);
}
}
void arm7_9_disable_eice_step(struct target *target)
{
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
embeddedice_store_reg(&arm7_9->eice_cache->reg_list[EICE_W0_ADDR_MASK]);
embeddedice_store_reg(&arm7_9->eice_cache->reg_list[EICE_W0_DATA_MASK]);
embeddedice_store_reg(&arm7_9->eice_cache->reg_list[EICE_W0_CONTROL_VALUE]);
embeddedice_store_reg(&arm7_9->eice_cache->reg_list[EICE_W0_CONTROL_MASK]);
embeddedice_store_reg(&arm7_9->eice_cache->reg_list[EICE_W1_ADDR_VALUE]);
embeddedice_store_reg(&arm7_9->eice_cache->reg_list[EICE_W1_ADDR_MASK]);
embeddedice_store_reg(&arm7_9->eice_cache->reg_list[EICE_W1_DATA_MASK]);
embeddedice_store_reg(&arm7_9->eice_cache->reg_list[EICE_W1_CONTROL_MASK]);
embeddedice_store_reg(&arm7_9->eice_cache->reg_list[EICE_W1_CONTROL_VALUE]);
}
int arm7_9_step(struct target *target, int current, target_addr_t address, int handle_breakpoints)
{
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
struct arm *arm = &arm7_9->arm;
struct breakpoint *breakpoint = NULL;
int err, retval;
if (target->state != TARGET_HALTED) {
LOG_WARNING("target not halted");
return ERROR_TARGET_NOT_HALTED;
}
/* current = 1: continue on current pc, otherwise continue at <address> */
if (!current)
buf_set_u32(arm->pc->value, 0, 32, address);
uint32_t current_pc = buf_get_u32(arm->pc->value, 0, 32);
/* the front-end may request us not to handle breakpoints */
if (handle_breakpoints)
breakpoint = breakpoint_find(target, current_pc);
if (breakpoint != NULL) {
retval = arm7_9_unset_breakpoint(target, breakpoint);
if (retval != ERROR_OK)
return retval;
}
target->debug_reason = DBG_REASON_SINGLESTEP;
/* calculate PC of next instruction */
uint32_t next_pc;
retval = arm_simulate_step(target, &next_pc);
if (retval != ERROR_OK) {
uint32_t current_opcode;
target_read_u32(target, current_pc, &current_opcode);
LOG_ERROR(
"Couldn't calculate PC of next instruction, current opcode was 0x%8.8" PRIx32 "",
current_opcode);
return retval;
}
retval = arm7_9_restore_context(target);
if (retval != ERROR_OK)
return retval;
arm7_9->enable_single_step(target, next_pc);
if (arm->core_state == ARM_STATE_ARM)
arm7_9->branch_resume(target);
else if (arm->core_state == ARM_STATE_THUMB)
arm7_9->branch_resume_thumb(target);
else {
LOG_ERROR("unhandled core state");
return ERROR_FAIL;
}
retval = target_call_event_callbacks(target, TARGET_EVENT_RESUMED);
if (retval != ERROR_OK)
return retval;
err = arm7_9_execute_sys_speed(target);
arm7_9->disable_single_step(target);
/* registers are now invalid */
register_cache_invalidate(arm->core_cache);
if (err != ERROR_OK)
target->state = TARGET_UNKNOWN;
else {
retval = arm7_9_debug_entry(target);
if (retval != ERROR_OK)
return retval;
retval = target_call_event_callbacks(target, TARGET_EVENT_HALTED);
if (retval != ERROR_OK)
return retval;
LOG_DEBUG("target stepped");
}
if (breakpoint) {
retval = arm7_9_set_breakpoint(target, breakpoint);
if (retval != ERROR_OK)
return retval;
}
return err;
}
static int arm7_9_read_core_reg(struct target *target, struct reg *r,
int num, enum arm_mode mode)
{
uint32_t *reg_p[16];
int retval;
struct arm_reg *areg = r->arch_info;
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
struct arm *arm = &arm7_9->arm;
if (!is_arm_mode(arm->core_mode))
return ERROR_FAIL;
if ((num < 0) || (num > 16))
return ERROR_COMMAND_SYNTAX_ERROR;
if ((mode != ARM_MODE_ANY) && (mode != arm->core_mode)
&& (areg->mode != ARM_MODE_ANY)) {
uint32_t tmp_cpsr;
/* change processor mode (mask T bit) */
tmp_cpsr = buf_get_u32(arm->cpsr->value, 0, 8) & 0xE0;
tmp_cpsr |= mode;
tmp_cpsr &= ~0x20;
arm7_9->write_xpsr_im8(target, tmp_cpsr & 0xff, 0, 0);
}
uint32_t value = 0;
if ((num >= 0) && (num <= 15)) {
/* read a normal core register */
reg_p[num] = &value;
arm7_9->read_core_regs(target, 1 << num, reg_p);
} else {
/* read a program status register
* if the register mode is MODE_ANY, we read the cpsr, otherwise a spsr
*/
arm7_9->read_xpsr(target, &value, areg->mode != ARM_MODE_ANY);
}
retval = jtag_execute_queue();
if (retval != ERROR_OK)
return retval;
r->valid = true;
r->dirty = false;
buf_set_u32(r->value, 0, 32, value);
if ((mode != ARM_MODE_ANY) && (mode != arm->core_mode)
&& (areg->mode != ARM_MODE_ANY)) {
/* restore processor mode (mask T bit) */
arm7_9->write_xpsr_im8(target,
buf_get_u32(arm->cpsr->value, 0, 8) & ~0x20, 0, 0);
}
return ERROR_OK;
}
static int arm7_9_write_core_reg(struct target *target, struct reg *r,
int num, enum arm_mode mode, uint8_t *value)
{
uint32_t reg[16];
struct arm_reg *areg = r->arch_info;
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
struct arm *arm = &arm7_9->arm;
if (!is_arm_mode(arm->core_mode))
return ERROR_FAIL;
if ((num < 0) || (num > 16))
return ERROR_COMMAND_SYNTAX_ERROR;
if ((mode != ARM_MODE_ANY) && (mode != arm->core_mode)
&& (areg->mode != ARM_MODE_ANY)) {
uint32_t tmp_cpsr;
/* change processor mode (mask T bit) */
tmp_cpsr = buf_get_u32(arm->cpsr->value, 0, 8) & 0xE0;
tmp_cpsr |= mode;
tmp_cpsr &= ~0x20;
arm7_9->write_xpsr_im8(target, tmp_cpsr & 0xff, 0, 0);
}
if ((num >= 0) && (num <= 15)) {
/* write a normal core register */
reg[num] = buf_get_u32(value, 0, 32);
arm7_9->write_core_regs(target, 1 << num, reg);
} else {
/* write a program status register
* if the register mode is MODE_ANY, we write the cpsr, otherwise a spsr
*/
int spsr = (areg->mode != ARM_MODE_ANY);
uint32_t t = buf_get_u32(value, 0, 32);
/* if we're writing the CPSR, mask the T bit */
if (!spsr)
t &= ~0x20;
arm7_9->write_xpsr(target, t, spsr);
}
r->valid = true;
r->dirty = false;
if ((mode != ARM_MODE_ANY) && (mode != arm->core_mode)
&& (areg->mode != ARM_MODE_ANY)) {
/* restore processor mode (mask T bit) */
arm7_9->write_xpsr_im8(target,
buf_get_u32(arm->cpsr->value, 0, 8) & ~0x20, 0, 0);
}
return jtag_execute_queue();
}
int arm7_9_read_memory(struct target *target,
target_addr_t address,
uint32_t size,
uint32_t count,
uint8_t *buffer)
{
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
struct arm *arm = &arm7_9->arm;
uint32_t reg[16];
uint32_t num_accesses = 0;
int thisrun_accesses;
int i;
uint32_t cpsr;
int retval;
int last_reg = 0;
LOG_DEBUG("address: 0x%8.8" TARGET_PRIxADDR ", size: 0x%8.8" PRIx32 ", count: 0x%8.8" PRIx32 "",
address, size, count);
if (target->state != TARGET_HALTED) {
LOG_WARNING("target not halted");
return ERROR_TARGET_NOT_HALTED;
}
/* sanitize arguments */
if (((size != 4) && (size != 2) && (size != 1)) || (count == 0) || !(buffer))
return ERROR_COMMAND_SYNTAX_ERROR;
if (((size == 4) && (address & 0x3u)) || ((size == 2) && (address & 0x1u)))
return ERROR_TARGET_UNALIGNED_ACCESS;
/* load the base register with the address of the first word */
reg[0] = address;
arm7_9->write_core_regs(target, 0x1, reg);
int j = 0;
switch (size) {
case 4:
while (num_accesses < count) {
uint32_t reg_list;
thisrun_accesses =
((count - num_accesses) >= 14) ? 14 : (count - num_accesses);
reg_list = (0xffff >> (15 - thisrun_accesses)) & 0xfffe;
if (last_reg <= thisrun_accesses)
last_reg = thisrun_accesses;
arm7_9->load_word_regs(target, reg_list);
/* fast memory reads are only safe when the target is running
* from a sufficiently high clock (32 kHz is usually too slow)
*/
if (arm7_9->fast_memory_access)
retval = arm7_9_execute_fast_sys_speed(target);
else
retval = arm7_9_execute_sys_speed(target);
if (retval != ERROR_OK)
return retval;
arm7_9->read_core_regs_target_buffer(target, reg_list, buffer, 4);
/* advance buffer, count number of accesses */
buffer += thisrun_accesses * 4;
num_accesses += thisrun_accesses;
if ((j++%1024) == 0)
keep_alive();
}
break;
case 2:
while (num_accesses < count) {
uint32_t reg_list;
thisrun_accesses =
((count - num_accesses) >= 14) ? 14 : (count - num_accesses);
reg_list = (0xffff >> (15 - thisrun_accesses)) & 0xfffe;
for (i = 1; i <= thisrun_accesses; i++) {
if (i > last_reg)
last_reg = i;
arm7_9->load_hword_reg(target, i);
/* fast memory reads are only safe when the target is running
* from a sufficiently high clock (32 kHz is usually too slow)
*/
if (arm7_9->fast_memory_access)
retval = arm7_9_execute_fast_sys_speed(target);
else
retval = arm7_9_execute_sys_speed(target);
if (retval != ERROR_OK)
return retval;
}
arm7_9->read_core_regs_target_buffer(target, reg_list, buffer, 2);
/* advance buffer, count number of accesses */
buffer += thisrun_accesses * 2;
num_accesses += thisrun_accesses;
if ((j++%1024) == 0)
keep_alive();
}
break;
case 1:
while (num_accesses < count) {
uint32_t reg_list;
thisrun_accesses =
((count - num_accesses) >= 14) ? 14 : (count - num_accesses);
reg_list = (0xffff >> (15 - thisrun_accesses)) & 0xfffe;
for (i = 1; i <= thisrun_accesses; i++) {
if (i > last_reg)
last_reg = i;
arm7_9->load_byte_reg(target, i);
/* fast memory reads are only safe when the target is running
* from a sufficiently high clock (32 kHz is usually too slow)
*/
if (arm7_9->fast_memory_access)
retval = arm7_9_execute_fast_sys_speed(target);
else
retval = arm7_9_execute_sys_speed(target);
if (retval != ERROR_OK)
return retval;
}
arm7_9->read_core_regs_target_buffer(target, reg_list, buffer, 1);
/* advance buffer, count number of accesses */
buffer += thisrun_accesses * 1;
num_accesses += thisrun_accesses;
if ((j++%1024) == 0)
keep_alive();
}
break;
}
if (!is_arm_mode(arm->core_mode))
return ERROR_FAIL;
for (i = 0; i <= last_reg; i++) {
struct reg *r = arm_reg_current(arm, i);
r->dirty = r->valid;
}
arm7_9->read_xpsr(target, &cpsr, 0);
retval = jtag_execute_queue();
if (retval != ERROR_OK) {
LOG_ERROR("JTAG error while reading cpsr");
return ERROR_TARGET_DATA_ABORT;
}
if (((cpsr & 0x1f) == ARM_MODE_ABT) && (arm->core_mode != ARM_MODE_ABT)) {
LOG_WARNING(
"memory read caused data abort "
"(address: 0x%8.8" TARGET_PRIxADDR ", size: 0x%" PRIx32 ", count: 0x%" PRIx32 ")",
address,
size,
count);
arm7_9->write_xpsr_im8(target,
buf_get_u32(arm->cpsr->value, 0, 8)
& ~0x20, 0, 0);
return ERROR_TARGET_DATA_ABORT;
}
return ERROR_OK;
}
int arm7_9_write_memory(struct target *target,
target_addr_t address,
uint32_t size,
uint32_t count,
const uint8_t *buffer)
{
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
struct arm *arm = &arm7_9->arm;
struct reg *dbg_ctrl = &arm7_9->eice_cache->reg_list[EICE_DBG_CTRL];
uint32_t reg[16];
uint32_t num_accesses = 0;
int thisrun_accesses;
int i;
uint32_t cpsr;
int retval;
int last_reg = 0;
#ifdef _DEBUG_ARM7_9_
LOG_DEBUG("address: 0x%8.8x, size: 0x%8.8x, count: 0x%8.8x", address, size, count);
#endif
if (target->state != TARGET_HALTED) {
LOG_WARNING("target not halted");
return ERROR_TARGET_NOT_HALTED;
}
/* sanitize arguments */
if (((size != 4) && (size != 2) && (size != 1)) || (count == 0) || !(buffer))
return ERROR_COMMAND_SYNTAX_ERROR;
if (((size == 4) && (address & 0x3u)) || ((size == 2) && (address & 0x1u)))
return ERROR_TARGET_UNALIGNED_ACCESS;
/* load the base register with the address of the first word */
reg[0] = address;
arm7_9->write_core_regs(target, 0x1, reg);
/* Clear DBGACK, to make sure memory fetches work as expected */
buf_set_u32(dbg_ctrl->value, EICE_DBG_CONTROL_DBGACK, 1, 0);
embeddedice_store_reg(dbg_ctrl);
switch (size) {
case 4:
while (num_accesses < count) {
uint32_t reg_list;
thisrun_accesses =
((count - num_accesses) >= 14) ? 14 : (count - num_accesses);
reg_list = (0xffff >> (15 - thisrun_accesses)) & 0xfffe;
for (i = 1; i <= thisrun_accesses; i++) {
if (i > last_reg)
last_reg = i;
reg[i] = target_buffer_get_u32(target, buffer);
buffer += 4;
}
arm7_9->write_core_regs(target, reg_list, reg);
arm7_9->store_word_regs(target, reg_list);
/* fast memory writes are only safe when the target is running
* from a sufficiently high clock (32 kHz is usually too slow)
*/
if (arm7_9->fast_memory_access)
retval = arm7_9_execute_fast_sys_speed(target);
else {
retval = arm7_9_execute_sys_speed(target);
target: slow targets could cause GDB to time out This second half of the patch is proposed to clean up some GDB keep alive issues on arm7_9 targets that start up with very slow clocks. If an attempt is made to write to key registers on the processor with a slow jtag speed, GDB timeout warnings appear on the console (at least mine) when "reset halt" or "reset init" commands are issued from the gdb client: *** BEFORE PATCH *** (gdb) monitor reset init fast memory access is disabled 2 kHz keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1026). Workaround: increase "set remotetimeout" in GDB JTAG tap: at91sam9g20.cpu tap/device found: 0x0792603f (mfg: 0x01f, part: 0x7926, ver: 0x0) target state: halted target halted in ARM state due to breakpoint, current mode: Supervisor cpsr: 0x000000d3 pc: 0x00000000 MMU: disabled, D-Cache: disabled, I-Cache: disabled keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1027). Workaround: increase "set remotetimeout" in GDB keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1006). Workaround: increase "set remotetimeout" in GDB keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1006). Workaround: increase "set remotetimeout" in GDB keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1006). Workaround: increase "set remotetimeout" in GDB keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1004). Workaround: increase "set remotetimeout" in GDB RCLK - adaptive dcc downloads are enabled fast memory access is enabled NAND flash device 'NAND 256MiB 3,3V 8-bit' found (gdb) I added additional keep alive steps in areas that troubleshooting revealed were causing problems. I only did this however for non-fast write memory accesses. I don't think most people would be using fast memory accesses to write to memory when the jtag and system clocks are slow anyway. If you disagree with my feeling, think there is a more elegant way to handle the problem, or think the patch will cause other unforeseen problems with other targets, let me know. As you can see below, the patch does eliminate the problem on my development station and I suspect that it will benefit others. *** AFTER PATCH *** (gdb) monitor reset init fast memory access is disabled 2 kHz JTAG tap: at91sam9g20.cpu tap/device found: 0x0792603f (mfg: 0x01f, part: 0x7926, ver: 0x0) target state: halted target halted in ARM state due to breakpoint, current mode: Supervisor cpsr: 0x000000d3 pc: 0x00000000 MMU: disabled, D-Cache: disabled, I-Cache: disabled RCLK - adaptive dcc downloads are enabled fast memory access is enabled NAND flash device 'NAND 256MiB 3,3V 8-bit' found (gdb) Gary Carlson Gary Carlson, MSEE Principal Engineer Carlson-Minot Inc.
2010-05-19 03:59:00 +00:00
/*
* if memory writes are made when the clock is running slow
* (i.e. 32 kHz) which is necessary in some scripts to reconfigure
* processor operations after a "reset halt" or "reset init",
* need to immediately stroke the keep alive or will end up with
* gdb "keep alive not sent error message" problem.
*/
keep_alive();
}
if (retval != ERROR_OK)
return retval;
num_accesses += thisrun_accesses;
}
break;
case 2:
while (num_accesses < count) {
uint32_t reg_list;
thisrun_accesses =
((count - num_accesses) >= 14) ? 14 : (count - num_accesses);
reg_list = (0xffff >> (15 - thisrun_accesses)) & 0xfffe;
for (i = 1; i <= thisrun_accesses; i++) {
if (i > last_reg)
last_reg = i;
reg[i] = target_buffer_get_u16(target, buffer) & 0xffff;
buffer += 2;
}
arm7_9->write_core_regs(target, reg_list, reg);
for (i = 1; i <= thisrun_accesses; i++) {
arm7_9->store_hword_reg(target, i);
/* fast memory writes are only safe when the target is running
* from a sufficiently high clock (32 kHz is usually too slow)
*/
if (arm7_9->fast_memory_access)
retval = arm7_9_execute_fast_sys_speed(target);
else {
retval = arm7_9_execute_sys_speed(target);
target: slow targets could cause GDB to time out This second half of the patch is proposed to clean up some GDB keep alive issues on arm7_9 targets that start up with very slow clocks. If an attempt is made to write to key registers on the processor with a slow jtag speed, GDB timeout warnings appear on the console (at least mine) when "reset halt" or "reset init" commands are issued from the gdb client: *** BEFORE PATCH *** (gdb) monitor reset init fast memory access is disabled 2 kHz keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1026). Workaround: increase "set remotetimeout" in GDB JTAG tap: at91sam9g20.cpu tap/device found: 0x0792603f (mfg: 0x01f, part: 0x7926, ver: 0x0) target state: halted target halted in ARM state due to breakpoint, current mode: Supervisor cpsr: 0x000000d3 pc: 0x00000000 MMU: disabled, D-Cache: disabled, I-Cache: disabled keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1027). Workaround: increase "set remotetimeout" in GDB keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1006). Workaround: increase "set remotetimeout" in GDB keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1006). Workaround: increase "set remotetimeout" in GDB keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1006). Workaround: increase "set remotetimeout" in GDB keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1004). Workaround: increase "set remotetimeout" in GDB RCLK - adaptive dcc downloads are enabled fast memory access is enabled NAND flash device 'NAND 256MiB 3,3V 8-bit' found (gdb) I added additional keep alive steps in areas that troubleshooting revealed were causing problems. I only did this however for non-fast write memory accesses. I don't think most people would be using fast memory accesses to write to memory when the jtag and system clocks are slow anyway. If you disagree with my feeling, think there is a more elegant way to handle the problem, or think the patch will cause other unforeseen problems with other targets, let me know. As you can see below, the patch does eliminate the problem on my development station and I suspect that it will benefit others. *** AFTER PATCH *** (gdb) monitor reset init fast memory access is disabled 2 kHz JTAG tap: at91sam9g20.cpu tap/device found: 0x0792603f (mfg: 0x01f, part: 0x7926, ver: 0x0) target state: halted target halted in ARM state due to breakpoint, current mode: Supervisor cpsr: 0x000000d3 pc: 0x00000000 MMU: disabled, D-Cache: disabled, I-Cache: disabled RCLK - adaptive dcc downloads are enabled fast memory access is enabled NAND flash device 'NAND 256MiB 3,3V 8-bit' found (gdb) Gary Carlson Gary Carlson, MSEE Principal Engineer Carlson-Minot Inc.
2010-05-19 03:59:00 +00:00
/*
* if memory writes are made when the clock is running slow
* (i.e. 32 kHz) which is necessary in some scripts to reconfigure
* processor operations after a "reset halt" or "reset init",
* need to immediately stroke the keep alive or will end up with
* gdb "keep alive not sent error message" problem.
*/
target: slow targets could cause GDB to time out This second half of the patch is proposed to clean up some GDB keep alive issues on arm7_9 targets that start up with very slow clocks. If an attempt is made to write to key registers on the processor with a slow jtag speed, GDB timeout warnings appear on the console (at least mine) when "reset halt" or "reset init" commands are issued from the gdb client: *** BEFORE PATCH *** (gdb) monitor reset init fast memory access is disabled 2 kHz keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1026). Workaround: increase "set remotetimeout" in GDB JTAG tap: at91sam9g20.cpu tap/device found: 0x0792603f (mfg: 0x01f, part: 0x7926, ver: 0x0) target state: halted target halted in ARM state due to breakpoint, current mode: Supervisor cpsr: 0x000000d3 pc: 0x00000000 MMU: disabled, D-Cache: disabled, I-Cache: disabled keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1027). Workaround: increase "set remotetimeout" in GDB keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1006). Workaround: increase "set remotetimeout" in GDB keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1006). Workaround: increase "set remotetimeout" in GDB keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1006). Workaround: increase "set remotetimeout" in GDB keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1004). Workaround: increase "set remotetimeout" in GDB RCLK - adaptive dcc downloads are enabled fast memory access is enabled NAND flash device 'NAND 256MiB 3,3V 8-bit' found (gdb) I added additional keep alive steps in areas that troubleshooting revealed were causing problems. I only did this however for non-fast write memory accesses. I don't think most people would be using fast memory accesses to write to memory when the jtag and system clocks are slow anyway. If you disagree with my feeling, think there is a more elegant way to handle the problem, or think the patch will cause other unforeseen problems with other targets, let me know. As you can see below, the patch does eliminate the problem on my development station and I suspect that it will benefit others. *** AFTER PATCH *** (gdb) monitor reset init fast memory access is disabled 2 kHz JTAG tap: at91sam9g20.cpu tap/device found: 0x0792603f (mfg: 0x01f, part: 0x7926, ver: 0x0) target state: halted target halted in ARM state due to breakpoint, current mode: Supervisor cpsr: 0x000000d3 pc: 0x00000000 MMU: disabled, D-Cache: disabled, I-Cache: disabled RCLK - adaptive dcc downloads are enabled fast memory access is enabled NAND flash device 'NAND 256MiB 3,3V 8-bit' found (gdb) Gary Carlson Gary Carlson, MSEE Principal Engineer Carlson-Minot Inc.
2010-05-19 03:59:00 +00:00
keep_alive();
target: slow targets could cause GDB to time out This second half of the patch is proposed to clean up some GDB keep alive issues on arm7_9 targets that start up with very slow clocks. If an attempt is made to write to key registers on the processor with a slow jtag speed, GDB timeout warnings appear on the console (at least mine) when "reset halt" or "reset init" commands are issued from the gdb client: *** BEFORE PATCH *** (gdb) monitor reset init fast memory access is disabled 2 kHz keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1026). Workaround: increase "set remotetimeout" in GDB JTAG tap: at91sam9g20.cpu tap/device found: 0x0792603f (mfg: 0x01f, part: 0x7926, ver: 0x0) target state: halted target halted in ARM state due to breakpoint, current mode: Supervisor cpsr: 0x000000d3 pc: 0x00000000 MMU: disabled, D-Cache: disabled, I-Cache: disabled keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1027). Workaround: increase "set remotetimeout" in GDB keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1006). Workaround: increase "set remotetimeout" in GDB keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1006). Workaround: increase "set remotetimeout" in GDB keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1006). Workaround: increase "set remotetimeout" in GDB keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1004). Workaround: increase "set remotetimeout" in GDB RCLK - adaptive dcc downloads are enabled fast memory access is enabled NAND flash device 'NAND 256MiB 3,3V 8-bit' found (gdb) I added additional keep alive steps in areas that troubleshooting revealed were causing problems. I only did this however for non-fast write memory accesses. I don't think most people would be using fast memory accesses to write to memory when the jtag and system clocks are slow anyway. If you disagree with my feeling, think there is a more elegant way to handle the problem, or think the patch will cause other unforeseen problems with other targets, let me know. As you can see below, the patch does eliminate the problem on my development station and I suspect that it will benefit others. *** AFTER PATCH *** (gdb) monitor reset init fast memory access is disabled 2 kHz JTAG tap: at91sam9g20.cpu tap/device found: 0x0792603f (mfg: 0x01f, part: 0x7926, ver: 0x0) target state: halted target halted in ARM state due to breakpoint, current mode: Supervisor cpsr: 0x000000d3 pc: 0x00000000 MMU: disabled, D-Cache: disabled, I-Cache: disabled RCLK - adaptive dcc downloads are enabled fast memory access is enabled NAND flash device 'NAND 256MiB 3,3V 8-bit' found (gdb) Gary Carlson Gary Carlson, MSEE Principal Engineer Carlson-Minot Inc.
2010-05-19 03:59:00 +00:00
}
if (retval != ERROR_OK)
return retval;
}
num_accesses += thisrun_accesses;
}
break;
case 1:
while (num_accesses < count) {
uint32_t reg_list;
thisrun_accesses =
((count - num_accesses) >= 14) ? 14 : (count - num_accesses);
reg_list = (0xffff >> (15 - thisrun_accesses)) & 0xfffe;
for (i = 1; i <= thisrun_accesses; i++) {
if (i > last_reg)
last_reg = i;
reg[i] = *buffer++ & 0xff;
}
arm7_9->write_core_regs(target, reg_list, reg);
for (i = 1; i <= thisrun_accesses; i++) {
arm7_9->store_byte_reg(target, i);
/* fast memory writes are only safe when the target is running
* from a sufficiently high clock (32 kHz is usually too slow)
*/
if (arm7_9->fast_memory_access)
retval = arm7_9_execute_fast_sys_speed(target);
else {
retval = arm7_9_execute_sys_speed(target);
target: slow targets could cause GDB to time out This second half of the patch is proposed to clean up some GDB keep alive issues on arm7_9 targets that start up with very slow clocks. If an attempt is made to write to key registers on the processor with a slow jtag speed, GDB timeout warnings appear on the console (at least mine) when "reset halt" or "reset init" commands are issued from the gdb client: *** BEFORE PATCH *** (gdb) monitor reset init fast memory access is disabled 2 kHz keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1026). Workaround: increase "set remotetimeout" in GDB JTAG tap: at91sam9g20.cpu tap/device found: 0x0792603f (mfg: 0x01f, part: 0x7926, ver: 0x0) target state: halted target halted in ARM state due to breakpoint, current mode: Supervisor cpsr: 0x000000d3 pc: 0x00000000 MMU: disabled, D-Cache: disabled, I-Cache: disabled keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1027). Workaround: increase "set remotetimeout" in GDB keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1006). Workaround: increase "set remotetimeout" in GDB keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1006). Workaround: increase "set remotetimeout" in GDB keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1006). Workaround: increase "set remotetimeout" in GDB keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1004). Workaround: increase "set remotetimeout" in GDB RCLK - adaptive dcc downloads are enabled fast memory access is enabled NAND flash device 'NAND 256MiB 3,3V 8-bit' found (gdb) I added additional keep alive steps in areas that troubleshooting revealed were causing problems. I only did this however for non-fast write memory accesses. I don't think most people would be using fast memory accesses to write to memory when the jtag and system clocks are slow anyway. If you disagree with my feeling, think there is a more elegant way to handle the problem, or think the patch will cause other unforeseen problems with other targets, let me know. As you can see below, the patch does eliminate the problem on my development station and I suspect that it will benefit others. *** AFTER PATCH *** (gdb) monitor reset init fast memory access is disabled 2 kHz JTAG tap: at91sam9g20.cpu tap/device found: 0x0792603f (mfg: 0x01f, part: 0x7926, ver: 0x0) target state: halted target halted in ARM state due to breakpoint, current mode: Supervisor cpsr: 0x000000d3 pc: 0x00000000 MMU: disabled, D-Cache: disabled, I-Cache: disabled RCLK - adaptive dcc downloads are enabled fast memory access is enabled NAND flash device 'NAND 256MiB 3,3V 8-bit' found (gdb) Gary Carlson Gary Carlson, MSEE Principal Engineer Carlson-Minot Inc.
2010-05-19 03:59:00 +00:00
/*
* if memory writes are made when the clock is running slow
* (i.e. 32 kHz) which is necessary in some scripts to reconfigure
* processor operations after a "reset halt" or "reset init",
* need to immediately stroke the keep alive or will end up with
* gdb "keep alive not sent error message" problem.
*/
target: slow targets could cause GDB to time out This second half of the patch is proposed to clean up some GDB keep alive issues on arm7_9 targets that start up with very slow clocks. If an attempt is made to write to key registers on the processor with a slow jtag speed, GDB timeout warnings appear on the console (at least mine) when "reset halt" or "reset init" commands are issued from the gdb client: *** BEFORE PATCH *** (gdb) monitor reset init fast memory access is disabled 2 kHz keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1026). Workaround: increase "set remotetimeout" in GDB JTAG tap: at91sam9g20.cpu tap/device found: 0x0792603f (mfg: 0x01f, part: 0x7926, ver: 0x0) target state: halted target halted in ARM state due to breakpoint, current mode: Supervisor cpsr: 0x000000d3 pc: 0x00000000 MMU: disabled, D-Cache: disabled, I-Cache: disabled keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1027). Workaround: increase "set remotetimeout" in GDB keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1006). Workaround: increase "set remotetimeout" in GDB keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1006). Workaround: increase "set remotetimeout" in GDB keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1006). Workaround: increase "set remotetimeout" in GDB keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1004). Workaround: increase "set remotetimeout" in GDB RCLK - adaptive dcc downloads are enabled fast memory access is enabled NAND flash device 'NAND 256MiB 3,3V 8-bit' found (gdb) I added additional keep alive steps in areas that troubleshooting revealed were causing problems. I only did this however for non-fast write memory accesses. I don't think most people would be using fast memory accesses to write to memory when the jtag and system clocks are slow anyway. If you disagree with my feeling, think there is a more elegant way to handle the problem, or think the patch will cause other unforeseen problems with other targets, let me know. As you can see below, the patch does eliminate the problem on my development station and I suspect that it will benefit others. *** AFTER PATCH *** (gdb) monitor reset init fast memory access is disabled 2 kHz JTAG tap: at91sam9g20.cpu tap/device found: 0x0792603f (mfg: 0x01f, part: 0x7926, ver: 0x0) target state: halted target halted in ARM state due to breakpoint, current mode: Supervisor cpsr: 0x000000d3 pc: 0x00000000 MMU: disabled, D-Cache: disabled, I-Cache: disabled RCLK - adaptive dcc downloads are enabled fast memory access is enabled NAND flash device 'NAND 256MiB 3,3V 8-bit' found (gdb) Gary Carlson Gary Carlson, MSEE Principal Engineer Carlson-Minot Inc.
2010-05-19 03:59:00 +00:00
keep_alive();
}
target: slow targets could cause GDB to time out This second half of the patch is proposed to clean up some GDB keep alive issues on arm7_9 targets that start up with very slow clocks. If an attempt is made to write to key registers on the processor with a slow jtag speed, GDB timeout warnings appear on the console (at least mine) when "reset halt" or "reset init" commands are issued from the gdb client: *** BEFORE PATCH *** (gdb) monitor reset init fast memory access is disabled 2 kHz keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1026). Workaround: increase "set remotetimeout" in GDB JTAG tap: at91sam9g20.cpu tap/device found: 0x0792603f (mfg: 0x01f, part: 0x7926, ver: 0x0) target state: halted target halted in ARM state due to breakpoint, current mode: Supervisor cpsr: 0x000000d3 pc: 0x00000000 MMU: disabled, D-Cache: disabled, I-Cache: disabled keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1027). Workaround: increase "set remotetimeout" in GDB keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1006). Workaround: increase "set remotetimeout" in GDB keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1006). Workaround: increase "set remotetimeout" in GDB keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1006). Workaround: increase "set remotetimeout" in GDB keep_alive() was not invoked in the 1000ms timelimit. GDB alive packet not sent! (1004). Workaround: increase "set remotetimeout" in GDB RCLK - adaptive dcc downloads are enabled fast memory access is enabled NAND flash device 'NAND 256MiB 3,3V 8-bit' found (gdb) I added additional keep alive steps in areas that troubleshooting revealed were causing problems. I only did this however for non-fast write memory accesses. I don't think most people would be using fast memory accesses to write to memory when the jtag and system clocks are slow anyway. If you disagree with my feeling, think there is a more elegant way to handle the problem, or think the patch will cause other unforeseen problems with other targets, let me know. As you can see below, the patch does eliminate the problem on my development station and I suspect that it will benefit others. *** AFTER PATCH *** (gdb) monitor reset init fast memory access is disabled 2 kHz JTAG tap: at91sam9g20.cpu tap/device found: 0x0792603f (mfg: 0x01f, part: 0x7926, ver: 0x0) target state: halted target halted in ARM state due to breakpoint, current mode: Supervisor cpsr: 0x000000d3 pc: 0x00000000 MMU: disabled, D-Cache: disabled, I-Cache: disabled RCLK - adaptive dcc downloads are enabled fast memory access is enabled NAND flash device 'NAND 256MiB 3,3V 8-bit' found (gdb) Gary Carlson Gary Carlson, MSEE Principal Engineer Carlson-Minot Inc.
2010-05-19 03:59:00 +00:00
if (retval != ERROR_OK)
return retval;
}
num_accesses += thisrun_accesses;
}
break;
}
/* Re-Set DBGACK */
buf_set_u32(dbg_ctrl->value, EICE_DBG_CONTROL_DBGACK, 1, 1);
embeddedice_store_reg(dbg_ctrl);
if (!is_arm_mode(arm->core_mode))
return ERROR_FAIL;
for (i = 0; i <= last_reg; i++) {
struct reg *r = arm_reg_current(arm, i);
r->dirty = r->valid;
}
arm7_9->read_xpsr(target, &cpsr, 0);
retval = jtag_execute_queue();
if (retval != ERROR_OK) {
LOG_ERROR("JTAG error while reading cpsr");
return ERROR_TARGET_DATA_ABORT;
}
if (((cpsr & 0x1f) == ARM_MODE_ABT) && (arm->core_mode != ARM_MODE_ABT)) {
LOG_WARNING(
"memory write caused data abort "
"(address: 0x%8.8" TARGET_PRIxADDR ", size: 0x%" PRIx32 ", count: 0x%" PRIx32 ")",
address,
size,
count);
arm7_9->write_xpsr_im8(target,
buf_get_u32(arm->cpsr->value, 0, 8)
& ~0x20, 0, 0);
return ERROR_TARGET_DATA_ABORT;
}
return ERROR_OK;
}
int arm7_9_write_memory_opt(struct target *target,
target_addr_t address,
uint32_t size,
uint32_t count,
const uint8_t *buffer)
{
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
int retval;
if (size == 4 && count > 32 && arm7_9->bulk_write_memory) {
/* Attempt to do a bulk write */
retval = arm7_9->bulk_write_memory(target, address, count, buffer);
if (retval == ERROR_OK)
return ERROR_OK;
}
return arm7_9->write_memory(target, address, size, count, buffer);
}
int arm7_9_write_memory_no_opt(struct target *target,
uint32_t address,
uint32_t size,
uint32_t count,
const uint8_t *buffer)
{
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
return arm7_9->write_memory(target, address, size, count, buffer);
}
static int dcc_count;
static const uint8_t *dcc_buffer;
static int arm7_9_dcc_completion(struct target *target,
uint32_t exit_point,
int timeout_ms,
void *arch_info)
{
int retval = ERROR_OK;
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
retval = target_wait_state(target, TARGET_DEBUG_RUNNING, 500);
if (retval != ERROR_OK)
return retval;
int little = target->endianness == TARGET_LITTLE_ENDIAN;
int count = dcc_count;
const uint8_t *buffer = dcc_buffer;
if (count > 2) {
/* Handle first & last using standard embeddedice_write_reg and the middle ones w/the
* core function repeated. */
embeddedice_write_reg(&arm7_9->eice_cache->reg_list[EICE_COMMS_DATA],
fast_target_buffer_get_u32(buffer, little));
buffer += 4;
struct embeddedice_reg *ice_reg =
arm7_9->eice_cache->reg_list[EICE_COMMS_DATA].arch_info;
uint8_t reg_addr = ice_reg->addr & 0x1f;
struct jtag_tap *tap;
tap = ice_reg->jtag_info->tap;
embeddedice_write_dcc(tap, reg_addr, buffer, little, count-2);
buffer += (count-2)*4;
embeddedice_write_reg(&arm7_9->eice_cache->reg_list[EICE_COMMS_DATA],
fast_target_buffer_get_u32(buffer, little));
} else {
int i;
for (i = 0; i < count; i++) {
embeddedice_write_reg(&arm7_9->eice_cache->reg_list[EICE_COMMS_DATA],
fast_target_buffer_get_u32(buffer, little));
buffer += 4;
}
}
retval = target_halt(target);
if (retval != ERROR_OK)
return retval;
return target_wait_state(target, TARGET_HALTED, 500);
}
static const uint32_t dcc_code[] = {
/* r0 == input, points to memory buffer
* r1 == scratch
*/
/* spin until DCC control (c0) reports data arrived */
0xee101e10, /* w: mrc p14, #0, r1, c0, c0 */
0xe3110001, /* tst r1, #1 */
0x0afffffc, /* bne w */
/* read word from DCC (c1), write to memory */
0xee111e10, /* mrc p14, #0, r1, c1, c0 */
0xe4801004, /* str r1, [r0], #4 */
/* repeat */
0xeafffff9 /* b w */
};
int arm7_9_bulk_write_memory(struct target *target,
target_addr_t address,
uint32_t count,
const uint8_t *buffer)
{
int retval;
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
if (address % 4 != 0)
return ERROR_TARGET_UNALIGNED_ACCESS;
if (!arm7_9->dcc_downloads)
return ERROR_TARGET_RESOURCE_NOT_AVAILABLE;
/* regrab previously allocated working_area, or allocate a new one */
if (!arm7_9->dcc_working_area) {
uint8_t dcc_code_buf[6 * 4];
/* make sure we have a working area */
if (target_alloc_working_area(target, 24, &arm7_9->dcc_working_area) != ERROR_OK) {
LOG_INFO("no working area available, falling back to memory writes");
return ERROR_TARGET_RESOURCE_NOT_AVAILABLE;
}
/* copy target instructions to target endianness */
target_buffer_set_u32_array(target, dcc_code_buf, ARRAY_SIZE(dcc_code), dcc_code);
/* write DCC code to working area, using the non-optimized
* memory write to avoid ending up here again */
retval = arm7_9_write_memory_no_opt(target,
arm7_9->dcc_working_area->address, 4, 6, dcc_code_buf);
if (retval != ERROR_OK)
return retval;
}
struct arm_algorithm arm_algo;
struct reg_param reg_params[1];
arm_algo.common_magic = ARM_COMMON_MAGIC;
arm_algo.core_mode = ARM_MODE_SVC;
arm_algo.core_state = ARM_STATE_ARM;
init_reg_param(&reg_params[0], "r0", 32, PARAM_IN_OUT);
buf_set_u32(reg_params[0].value, 0, 32, address);
dcc_count = count;
dcc_buffer = buffer;
retval = armv4_5_run_algorithm_inner(target, 0, NULL, 1, reg_params,
arm7_9->dcc_working_area->address,
arm7_9->dcc_working_area->address + 6*4,
20*1000, &arm_algo, arm7_9_dcc_completion);
if (retval == ERROR_OK) {
uint32_t endaddress = buf_get_u32(reg_params[0].value, 0, 32);
if (endaddress != (address + count*4)) {
LOG_ERROR(
"DCC write failed, expected end address 0x%08" TARGET_PRIxADDR " got 0x%0" PRIx32 "",
(address + count*4),
endaddress);
retval = ERROR_FAIL;
}
}
destroy_reg_param(&reg_params[0]);
return retval;
}
/**
* Perform per-target setup that requires JTAG access.
*/
int arm7_9_examine(struct target *target)
{
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
int retval;
if (!target_was_examined(target)) {
struct reg_cache *t, **cache_p;
t = embeddedice_build_reg_cache(target, arm7_9);
if (t == NULL)
return ERROR_FAIL;
cache_p = register_get_last_cache_p(&target->reg_cache);
(*cache_p) = t;
arm7_9->eice_cache = (*cache_p);
if (arm7_9->arm.etm)
(*cache_p)->next = etm_build_reg_cache(target,
&arm7_9->jtag_info,
arm7_9->arm.etm);
target_set_examined(target);
}
retval = embeddedice_setup(target);
if (retval == ERROR_OK)
retval = arm7_9_setup(target);
if (retval == ERROR_OK && arm7_9->arm.etm)
retval = etm_setup(target);
return retval;
}
void arm7_9_deinit(struct target *target)
{
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
if (target_was_examined(target))
embeddedice_free_reg_cache(arm7_9->eice_cache);
arm_jtag_close_connection(&arm7_9->jtag_info);
}
int arm7_9_check_reset(struct target *target)
{
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
if (get_target_reset_nag() && !arm7_9->dcc_downloads)
LOG_WARNING(
"NOTE! DCC downloads have not been enabled, defaulting to slow memory writes. Type 'help dcc'.");
if (get_target_reset_nag() && (target->working_area_size == 0))
LOG_WARNING("NOTE! Severe performance degradation without working memory enabled.");
if (get_target_reset_nag() && !arm7_9->fast_memory_access)
LOG_WARNING(
"NOTE! Severe performance degradation without fast memory access enabled. Type 'help fast'.");
return ERROR_OK;
}
int arm7_9_endianness_callback(jtag_callback_data_t pu8_in,
jtag_callback_data_t i_size, jtag_callback_data_t i_be,
jtag_callback_data_t i_flip)
{
uint8_t *in = (uint8_t *)pu8_in;
int size = (int)i_size;
int be = (int)i_be;
int flip = (int)i_flip;
uint32_t readback;
switch (size) {
case 4:
readback = le_to_h_u32(in);
if (flip)
readback = flip_u32(readback, 32);
if (be)
h_u32_to_be(in, readback);
else
h_u32_to_le(in, readback);
break;
case 2:
readback = le_to_h_u16(in);
if (flip)
readback = flip_u32(readback, 16);
if (be)
h_u16_to_be(in, readback & 0xffff);
else
h_u16_to_le(in, readback & 0xffff);
break;
case 1:
readback = *in;
if (flip)
readback = flip_u32(readback, 8);
*in = readback & 0xff;
break;
}
return ERROR_OK;
}
COMMAND_HANDLER(handle_arm7_9_dbgrq_command)
{
struct target *target = get_current_target(CMD_CTX);
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
if (!is_arm7_9(arm7_9)) {
command_print(CMD, "current target isn't an ARM7/ARM9 target");
return ERROR_TARGET_INVALID;
}
if (CMD_ARGC > 0)
COMMAND_PARSE_ENABLE(CMD_ARGV[0], arm7_9->use_dbgrq);
command_print(CMD,
"use of EmbeddedICE dbgrq instead of breakpoint for target halt %s",
(arm7_9->use_dbgrq) ? "enabled" : "disabled");
return ERROR_OK;
}
COMMAND_HANDLER(handle_arm7_9_fast_memory_access_command)
{
struct target *target = get_current_target(CMD_CTX);
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
if (!is_arm7_9(arm7_9)) {
command_print(CMD, "current target isn't an ARM7/ARM9 target");
return ERROR_TARGET_INVALID;
}
if (CMD_ARGC > 0)
COMMAND_PARSE_ENABLE(CMD_ARGV[0], arm7_9->fast_memory_access);
command_print(CMD,
"fast memory access is %s",
(arm7_9->fast_memory_access) ? "enabled" : "disabled");
return ERROR_OK;
}
COMMAND_HANDLER(handle_arm7_9_dcc_downloads_command)
{
struct target *target = get_current_target(CMD_CTX);
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
if (!is_arm7_9(arm7_9)) {
command_print(CMD, "current target isn't an ARM7/ARM9 target");
return ERROR_TARGET_INVALID;
}
if (CMD_ARGC > 0)
COMMAND_PARSE_ENABLE(CMD_ARGV[0], arm7_9->dcc_downloads);
command_print(CMD,
"dcc downloads are %s",
(arm7_9->dcc_downloads) ? "enabled" : "disabled");
return ERROR_OK;
}
static int arm7_9_setup_semihosting(struct target *target, int enable)
{
struct arm7_9_common *arm7_9 = target_to_arm7_9(target);
if (!is_arm7_9(arm7_9)) {
LOG_USER("current target isn't an ARM7/ARM9 target");
return ERROR_TARGET_INVALID;
}
if (arm7_9->has_vector_catch) {
struct reg *vector_catch = &arm7_9->eice_cache
->reg_list[EICE_VEC_CATCH];
if (!vector_catch->valid)
embeddedice_read_reg(vector_catch);
buf_set_u32(vector_catch->value, 2, 1, enable);
embeddedice_store_reg(vector_catch);
} else {
/* TODO: allow optional high vectors and/or BKPT_HARD */
if (enable)
breakpoint_add(target, 8, 4, BKPT_SOFT);
else
breakpoint_remove(target, 8);
}
return ERROR_OK;
}
int arm7_9_init_arch_info(struct target *target, struct arm7_9_common *arm7_9)
{
int retval = ERROR_OK;
struct arm *arm = &arm7_9->arm;
arm7_9->common_magic = ARM7_9_COMMON_MAGIC;
retval = arm_jtag_setup_connection(&arm7_9->jtag_info);
if (retval != ERROR_OK)
return retval;
/* caller must have allocated via calloc(), so everything's zeroed */
arm7_9->wp_available_max = 2;
arm7_9->fast_memory_access = false;
arm7_9->dcc_downloads = false;
arm->arch_info = arm7_9;
arm->core_type = ARM_CORE_TYPE_STD;
arm->read_core_reg = arm7_9_read_core_reg;
arm->write_core_reg = arm7_9_write_core_reg;
arm->full_context = arm7_9_full_context;
arm->setup_semihosting = arm7_9_setup_semihosting;
retval = arm_init_arch_info(target, arm);
if (retval != ERROR_OK)
return retval;
return target_register_timer_callback(arm7_9_handle_target_request,
1, TARGET_TIMER_TYPE_PERIODIC, target);
}
2009-11-23 15:43:05 +00:00
static const struct command_registration arm7_9_any_command_handlers[] = {
{
.name = "dbgrq",
.handler = handle_arm7_9_dbgrq_command,
2009-11-23 15:43:05 +00:00
.mode = COMMAND_ANY,
.usage = "['enable'|'disable']",
2009-11-23 15:43:05 +00:00
.help = "use EmbeddedICE dbgrq instead of breakpoint "
"for target halt requests",
},
{
.name = "fast_memory_access",
.handler = handle_arm7_9_fast_memory_access_command,
2009-11-23 15:43:05 +00:00
.mode = COMMAND_ANY,
.usage = "['enable'|'disable']",
2009-11-23 15:43:05 +00:00
.help = "use fast memory accesses instead of slower "
"but potentially safer accesses",
},
{
.name = "dcc_downloads",
.handler = handle_arm7_9_dcc_downloads_command,
2009-11-23 15:43:05 +00:00
.mode = COMMAND_ANY,
.usage = "['enable'|'disable']",
2009-11-23 15:43:05 +00:00
.help = "use DCC downloads for larger memory writes",
},
COMMAND_REGISTRATION_DONE
};
const struct command_registration arm7_9_command_handlers[] = {
{
.chain = arm_command_handlers,
},
{
.chain = etm_command_handlers,
},
2009-11-23 15:43:05 +00:00
{
.name = "arm7_9",
.mode = COMMAND_ANY,
.help = "arm7/9 specific commands",
.usage = "",
2009-11-23 15:43:05 +00:00
.chain = arm7_9_any_command_handlers,
},
COMMAND_REGISTRATION_DONE
};