[PATCH v1 8/8] cpukit/aarch64: Add libdebugger support
Kinsey Moore
kinsey.moore at oarcorp.com
Fri Oct 29 21:17:21 UTC 2021
This adds support for libdebugger under AArch64 using software
breakpoints and the single-step execution mode present in all AArch64
CPUs.
---
bsps/aarch64/include/bsp/aarch64-mmu.h | 11 +
cpukit/libdebugger/rtems-debugger-aarch64.c | 1884 +++++++++++++++++++
spec/build/cpukit/libdebugger.yml | 2 +
spec/build/cpukit/objdbgaarch64.yml | 15 +
spec/build/cpukit/optlibdebugger.yml | 1 +
5 files changed, 1913 insertions(+)
create mode 100644 cpukit/libdebugger/rtems-debugger-aarch64.c
create mode 100644 spec/build/cpukit/objdbgaarch64.yml
diff --git a/bsps/aarch64/include/bsp/aarch64-mmu.h b/bsps/aarch64/include/bsp/aarch64-mmu.h
index 9926eb97f6..b1a471d534 100644
--- a/bsps/aarch64/include/bsp/aarch64-mmu.h
+++ b/bsps/aarch64/include/bsp/aarch64-mmu.h
@@ -371,6 +371,17 @@ aarch64_mmu_enable( void )
_AArch64_Write_sctlr_el1( sctlr );
}
+BSP_START_TEXT_SECTION static inline void
+aarch64_mmu_disable( void )
+{
+ uint64_t sctlr;
+
+ /* Enable MMU and cache */
+ sctlr = _AArch64_Read_sctlr_el1();
+ sctlr &= ~(AARCH64_SCTLR_EL1_M);
+ _AArch64_Write_sctlr_el1( sctlr );
+}
+
BSP_START_TEXT_SECTION static inline void aarch64_mmu_setup( void )
{
/* Set TCR */
diff --git a/cpukit/libdebugger/rtems-debugger-aarch64.c b/cpukit/libdebugger/rtems-debugger-aarch64.c
new file mode 100644
index 0000000000..279c2d61ef
--- /dev/null
+++ b/cpukit/libdebugger/rtems-debugger-aarch64.c
@@ -0,0 +1,1884 @@
+/*
+ * Copyright (c) 2016-2019 Chris Johns <chrisj at rtems.org>.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ */
+
+#define TARGET_DEBUG 0
+
+#ifdef HAVE_CONFIG_H
+#include "config.h"
+#endif
+
+#include <errno.h>
+#include <inttypes.h>
+#include <stdlib.h>
+
+/* Defined by linkcmds.base */
+extern char bsp_section_text_begin[];
+extern char bsp_section_text_end[];
+extern char bsp_section_fast_text_begin[];
+extern char bsp_section_fast_text_end[];
+
+#include <libcpu/mmu-vmsav8-64.h>
+
+#include <rtems.h>
+#include <rtems/score/aarch64-system-registers.h>
+#include <rtems/score/cpu.h>
+#include <rtems/score/threadimpl.h>
+
+#include <rtems/debugger/rtems-debugger-bsp.h>
+
+#include "rtems-debugger-target.h"
+#include "rtems-debugger-threads.h"
+
+#if TARGET_DEBUG
+#include <rtems/bspIo.h>
+#endif
+
+/*
+ * Structure used to manage a task executing a function on available cores on
+ * a scheduler.
+ */
+typedef struct {
+ rtems_id allCPUsBarrier;
+ rtems_task_entry work_function;
+ rtems_task_argument arg;
+ rtems_status_code sc;
+} run_across_cpus_context;
+
+/*
+ * The function that runs as the body of the task which moves itself among the
+ * various cores registered to a scheduler.
+ */
+static rtems_task run_across_cpus_task( rtems_task_argument arg )
+{
+ uint32_t released = 0;
+ rtems_status_code sc;
+ run_across_cpus_context *ctx = (run_across_cpus_context *) arg;
+ cpu_set_t set;
+ cpu_set_t scheduler_set;
+ rtems_id scheduler_id;
+
+ sc = rtems_task_get_scheduler( RTEMS_SELF, &scheduler_id );
+
+ if ( sc != RTEMS_SUCCESSFUL ) {
+ ctx->sc = sc;
+ rtems_task_exit();
+ }
+
+ CPU_ZERO( &scheduler_set );
+ sc = rtems_scheduler_get_processor_set(
+ scheduler_id,
+ sizeof( scheduler_set ),
+ &scheduler_set
+ );
+
+ if ( sc != RTEMS_SUCCESSFUL ) {
+ ctx->sc = sc;
+ rtems_task_exit();
+ }
+
+ for (
+ int cpu_index = 0;
+ cpu_index < rtems_scheduler_get_processor_maximum();
+ cpu_index++
+ ) {
+ if ( !CPU_ISSET( cpu_index, &scheduler_set ) ) {
+ continue;
+ }
+
+ CPU_ZERO( &set );
+ CPU_SET( cpu_index, &set );
+ sc = rtems_task_set_affinity( RTEMS_SELF, sizeof( set ), &set );
+
+ if ( sc != RTEMS_SUCCESSFUL ) {
+ ctx->sc = sc;
+ rtems_task_exit();
+ }
+
+ /* execute task on selected CPU */
+ ctx->work_function( ctx->arg );
+ }
+
+ sc = rtems_barrier_release( ctx->allCPUsBarrier, &released );
+
+ if ( sc != RTEMS_SUCCESSFUL ) {
+ ctx->sc = sc;
+ }
+
+ rtems_task_exit();
+}
+
+/*
+ * The function used to run a provided function with arbitrary argument across
+ * all cores registered to the current scheduler. This is similar to the Linux
+ * kernel's on_each_cpu() call and always waits for the task to complete before
+ * returning.
+ */
+static rtems_status_code run_across_cpus(
+ rtems_task_entry task_entry,
+ rtems_task_argument arg
+)
+{
+ rtems_status_code sc;
+ rtems_id Task_id;
+ run_across_cpus_context ctx;
+
+ ctx.work_function = task_entry;
+ ctx.arg = arg;
+ ctx.sc = RTEMS_SUCCESSFUL;
+
+ memset( &ctx.allCPUsBarrier, 0, sizeof( ctx.allCPUsBarrier ) );
+ sc = rtems_barrier_create(
+ rtems_build_name( 'B', 'c', 'p', 'u' ),
+ RTEMS_BARRIER_MANUAL_RELEASE,
+ 2,
+ &ctx.allCPUsBarrier
+ );
+
+ if ( sc != RTEMS_SUCCESSFUL ) {
+ return sc;
+ }
+
+ sc = rtems_task_create(
+ rtems_build_name( 'T', 'c', 'p', 'u' ),
+ 1,
+ RTEMS_MINIMUM_STACK_SIZE * 2,
+ RTEMS_DEFAULT_MODES,
+ RTEMS_FLOATING_POINT | RTEMS_DEFAULT_ATTRIBUTES,
+ &Task_id
+ );
+
+ if ( sc != RTEMS_SUCCESSFUL ) {
+ rtems_barrier_delete( ctx.allCPUsBarrier );
+ return sc;
+ }
+
+ sc = rtems_task_start(
+ Task_id,
+ run_across_cpus_task,
+ ( rtems_task_argument ) & ctx
+ );
+
+ if ( sc != RTEMS_SUCCESSFUL ) {
+ rtems_task_delete( Task_id );
+ rtems_barrier_delete( ctx.allCPUsBarrier );
+ return sc;
+ }
+
+ /* wait on task */
+ sc = rtems_barrier_wait( ctx.allCPUsBarrier, RTEMS_NO_TIMEOUT );
+
+ if ( sc != RTEMS_SUCCESSFUL ) {
+ rtems_task_delete( Task_id );
+ rtems_barrier_delete( ctx.allCPUsBarrier );
+ return sc;
+ }
+
+ rtems_barrier_delete( ctx.allCPUsBarrier );
+
+ if ( ctx.sc != RTEMS_SUCCESSFUL ) {
+ return ctx.sc;
+ }
+
+ return sc;
+}
+
+/*
+ * Number of registers.
+ */
+#define RTEMS_DEBUGGER_NUMREGS 68
+
+/*
+ * Number of bytes per type of register.
+ */
+#define RTEMS_DEBUGGER_REG_BYTES 8
+
+/* Debugger registers layout. See aarch64-core.xml in GDB source. */
+#define REG_X0 0
+#define REG_X1 1
+#define REG_X2 2
+#define REG_X3 3
+#define REG_X4 4
+#define REG_X5 5
+#define REG_X6 6
+#define REG_X7 7
+#define REG_X8 8
+#define REG_X9 9
+#define REG_X10 10
+#define REG_X11 11
+#define REG_X12 12
+#define REG_X13 13
+#define REG_X14 14
+#define REG_X15 15
+#define REG_X16 16
+#define REG_X17 17
+#define REG_X18 18
+#define REG_X19 19
+#define REG_X20 20
+#define REG_X21 21
+#define REG_X22 22
+#define REG_X23 23
+#define REG_X24 24
+#define REG_X25 25
+#define REG_X26 26
+#define REG_X27 27
+#define REG_X28 28
+#define REG_FP 29
+#define REG_LR 30
+#define REG_SP 31
+/*
+ * PC isn't a real directly accessible register on AArch64, but is exposed via
+ * ELR_EL1 in exception context.
+ */
+#define REG_PC 32
+/* CPSR is defined as 32-bit by GDB */
+#define REG_CPS 33
+/* Debugger registers layout. See aarch64-fpu.xml in GDB source. */
+#define REG_V0 34
+#define REG_V1 35
+#define REG_V2 36
+#define REG_V3 37
+#define REG_V4 38
+#define REG_V5 39
+#define REG_V6 40
+#define REG_V7 41
+#define REG_V8 42
+#define REG_V9 43
+#define REG_V10 44
+#define REG_V11 45
+#define REG_V12 46
+#define REG_V13 47
+#define REG_V14 48
+#define REG_V15 49
+#define REG_V16 50
+#define REG_V17 51
+#define REG_V18 52
+#define REG_V19 53
+#define REG_V20 54
+#define REG_V21 55
+#define REG_V22 56
+#define REG_V23 57
+#define REG_V24 58
+#define REG_V25 59
+#define REG_V26 60
+#define REG_V27 61
+#define REG_V28 62
+#define REG_V29 63
+#define REG_V30 64
+#define REG_V31 65
+/* FPSR and FPCR are defined as 32-bit by GDB */
+#define REG_FPS 66
+#define REG_FPC 67
+
+/**
+ * Register offset table with the total as the last entry.
+ *
+ * Check this table in gdb with the command:
+ *
+ * maint print registers
+ */
+static const size_t aarch64_reg_offsets[ RTEMS_DEBUGGER_NUMREGS + 1 ] = {
+ REG_X0 * 8,
+ REG_X1 * 8,
+ REG_X2 * 8,
+ REG_X3 * 8,
+ REG_X4 * 8,
+ REG_X5 * 8,
+ REG_X6 * 8,
+ REG_X7 * 8,
+ REG_X8 * 8,
+ REG_X9 * 8,
+ REG_X10 * 8,
+ REG_X11 * 8,
+ REG_X12 * 8,
+ REG_X13 * 8,
+ REG_X14 * 8,
+ REG_X15 * 8,
+ REG_X16 * 8,
+ REG_X17 * 8,
+ REG_X18 * 8,
+ REG_X19 * 8,
+ REG_X20 * 8,
+ REG_X21 * 8,
+ REG_X22 * 8,
+ REG_X23 * 8,
+ REG_X24 * 8,
+ REG_X25 * 8,
+ REG_X26 * 8,
+ REG_X27 * 8,
+ REG_X28 * 8,
+ REG_FP * 8,
+ REG_LR * 8,
+ REG_SP * 8,
+ REG_PC * 8,
+ REG_CPS * 8,
+/* Floating point registers, CPS is 32-bit */
+#define V0_OFFSET ( REG_CPS * 8 + 4 )
+ V0_OFFSET,
+ V0_OFFSET + 16 * ( REG_V1 - REG_V0 ),
+ V0_OFFSET + 16 * ( REG_V2 - REG_V0 ),
+ V0_OFFSET + 16 * ( REG_V3 - REG_V0 ),
+ V0_OFFSET + 16 * ( REG_V4 - REG_V0 ),
+ V0_OFFSET + 16 * ( REG_V5 - REG_V0 ),
+ V0_OFFSET + 16 * ( REG_V6 - REG_V0 ),
+ V0_OFFSET + 16 * ( REG_V7 - REG_V0 ),
+ V0_OFFSET + 16 * ( REG_V8 - REG_V0 ),
+ V0_OFFSET + 16 * ( REG_V9 - REG_V0 ),
+ V0_OFFSET + 16 * ( REG_V10 - REG_V0 ),
+ V0_OFFSET + 16 * ( REG_V11 - REG_V0 ),
+ V0_OFFSET + 16 * ( REG_V12 - REG_V0 ),
+ V0_OFFSET + 16 * ( REG_V13 - REG_V0 ),
+ V0_OFFSET + 16 * ( REG_V14 - REG_V0 ),
+ V0_OFFSET + 16 * ( REG_V15 - REG_V0 ),
+ V0_OFFSET + 16 * ( REG_V16 - REG_V0 ),
+ V0_OFFSET + 16 * ( REG_V17 - REG_V0 ),
+ V0_OFFSET + 16 * ( REG_V18 - REG_V0 ),
+ V0_OFFSET + 16 * ( REG_V19 - REG_V0 ),
+ V0_OFFSET + 16 * ( REG_V20 - REG_V0 ),
+ V0_OFFSET + 16 * ( REG_V21 - REG_V0 ),
+ V0_OFFSET + 16 * ( REG_V22 - REG_V0 ),
+ V0_OFFSET + 16 * ( REG_V23 - REG_V0 ),
+ V0_OFFSET + 16 * ( REG_V24 - REG_V0 ),
+ V0_OFFSET + 16 * ( REG_V25 - REG_V0 ),
+ V0_OFFSET + 16 * ( REG_V26 - REG_V0 ),
+ V0_OFFSET + 16 * ( REG_V27 - REG_V0 ),
+ V0_OFFSET + 16 * ( REG_V28 - REG_V0 ),
+ V0_OFFSET + 16 * ( REG_V29 - REG_V0 ),
+ V0_OFFSET + 16 * ( REG_V30 - REG_V0 ),
+ V0_OFFSET + 16 * ( REG_V31 - REG_V0 ),
+/* FPSR and FPCR are defined as 32-bit by GDB */
+#define FPS_OFFSET ( V0_OFFSET + 16 * ( REG_V31 - REG_V0 ) + 16 )
+ FPS_OFFSET,
+/* FPC follows FPS */
+ FPS_OFFSET + 4,
+/* Total size */
+ FPS_OFFSET + 8,
+};
+
+/*
+ * Number of bytes of registers.
+ */
+#define RTEMS_DEBUGGER_NUMREGBYTES \
+ aarch64_reg_offsets[ RTEMS_DEBUGGER_NUMREGS ]
+
+/**
+ * Print the exception frame.
+ */
+#define EXC_FRAME_PRINT( _out, _prefix, _frame ) \
+ do { \
+ _out( \
+ _prefix " X0 = %016" PRIx64 " X1 = %016" PRIx64 \
+ " X2 = %016" PRIx64 " X3 = %016" PRIx64 "\n", \
+ _frame->register_x0, \
+ _frame->register_x1, \
+ _frame->register_x2, \
+ _frame->register_x3 \
+ ); \
+ _out( \
+ _prefix " X4 = %016" PRIx64 " X5 = %016" PRIx64 \
+ " X6 = %016" PRIx64 " X7 = %016" PRIx64 "\n", \
+ _frame->register_x4, \
+ _frame->register_x5, \
+ _frame->register_x6, \
+ _frame->register_x7 \
+ ); \
+ _out( \
+ _prefix " X8 = %016" PRIx64 " X9 = %016" PRIx64 \
+ " X10 = %016" PRIx64 " X11 = %016" PRIx64 "\n", \
+ _frame->register_x8, \
+ _frame->register_x9, \
+ _frame->register_x10, \
+ _frame->register_x11 \
+ ); \
+ _out( \
+ _prefix " X12 = %016" PRIx64 " X13 = %016" PRIx64 \
+ " X14 = %016" PRIx64 " X15 = %016" PRIx64 "\n", \
+ _frame->register_x12, \
+ _frame->register_x13, \
+ _frame->register_x14, \
+ _frame->register_x15 \
+ ); \
+ _out( \
+ _prefix " X16 = %016" PRIx64 " X17 = %016" PRIx64 \
+ " X18 = %016" PRIx64 " X19 = %016" PRIx64 "\n", \
+ _frame->register_x16, \
+ _frame->register_x17, \
+ _frame->register_x18, \
+ _frame->register_x19 \
+ ); \
+ _out( \
+ _prefix " X20 = %016" PRIx64 " X21 = %016" PRIx64 \
+ " X22 = %016" PRIx64 " X23 = %016" PRIx64 "\n", \
+ _frame->register_x20, \
+ _frame->register_x21, \
+ _frame->register_x22, \
+ _frame->register_x23 \
+ ); \
+ _out( \
+ _prefix " X24 = %016" PRIx64 " X25 = %016" PRIx64 \
+ " X26 = %016" PRIx64 " X27 = %016" PRIx64 "\n", \
+ _frame->register_x24, \
+ _frame->register_x25, \
+ _frame->register_x26, \
+ _frame->register_x27 \
+ ); \
+ _out( \
+ _prefix " X28 = %016" PRIx64 " FP = %016" PRIx64 \
+ " LR = %016" PRIxPTR " SP = %016" PRIxPTR "\n", \
+ _frame->register_x28, \
+ _frame->register_fp, \
+ (intptr_t) _frame->register_lr, \
+ (intptr_t) _frame->register_sp \
+ ); \
+ _out( \
+ _prefix " PC = %016" PRIxPTR "\n", \
+ (intptr_t) _frame->register_pc \
+ ); \
+ _out( \
+ _prefix " CPSR = %08" PRIx64 " %c%c%c%c%c%c%c%c%c" \
+ " M:%" PRIx64 " %s\n", \
+ _frame->register_cpsr, \
+ ( _frame->register_cpsr & ( 1 << 31 ) ) != 0 ? 'N' : '-', \
+ ( _frame->register_cpsr & ( 1 << 30 ) ) != 0 ? 'Z' : '-', \
+ ( _frame->register_cpsr & ( 1 << 29 ) ) != 0 ? 'C' : '-', \
+ ( _frame->register_cpsr & ( 1 << 28 ) ) != 0 ? 'V' : '-', \
+ ( _frame->register_cpsr & ( 1 << 21 ) ) != 0 ? 'S' : '-', \
+ ( _frame->register_cpsr & ( 1 << 9 ) ) != 0 ? 'D' : '-', \
+ ( _frame->register_cpsr & ( 1 << 8 ) ) != 0 ? 'A' : '-', \
+ ( _frame->register_cpsr & ( 1 << 7 ) ) != 0 ? 'I' : '-', \
+ ( _frame->register_cpsr & ( 1 << 6 ) ) != 0 ? 'F' : '-', \
+ _frame->register_cpsr & 0x1f, \
+ aarch64_mode_label( _frame->register_cpsr & 0x1f ) \
+ ); \
+ } while ( 0 )
+
+/**
+ * The breakpoint instruction.
+ */
+static const uint8_t breakpoint[ 4 ] = { 0x00, 0x00, 0x20, 0xd4 };
+
+/**
+ * Target lock.
+ */
+RTEMS_INTERRUPT_LOCK_DEFINE( static, target_lock, "target_lock" )
+
+/**
+ * Is a session active?
+ */
+static bool debug_session_active;
+
+/*
+ * AArch64 debug hardware.
+ */
+static uint64_t hw_breakpoints;
+static uint64_t hw_watchpoints;
+
+#ifdef HARDWARE_BREAKPOINTS_NOT_USED
+/**
+ * Hardware break and watch points.
+ */
+typedef struct {
+ bool enabled;
+ bool loaded;
+ void *address;
+ size_t length;
+ CPU_Exception_frame *frame;
+ uint64_t control;
+ uint64_t value;
+} aarch64_debug_hwbreak;
+
+/*
+ * AArch64 guarantees that 2-16 breakpoints will be available in:
+ * DBGBCR<0-15>_EL1 (control)
+ * BT: BSP_FLD64(val, 20, 23) (breakpoint type, always 0x0 or 0x4, address match or mismatch)
+ * LBN: BSP_FLD64(val, 16, 19) (linked breakpoint number, always 0x0, not relevant given above)
+ * SSC: BSP_FLD64(val, 14, 15) (security state control, only 0x0 relevant)
+ * HMC: BSP_BIT64(13) (higher mode control, only 0x0 relevant)
+ * BAS: BSP_FLD64(val, 5, 8) (byte address select, always 0xF, other values denote debugging of AArch32 code)
+ * PMC: BSP_FLD64(val, 1, 2) (privelege mode control, only 0x1 relevant)
+ * E: BSP_BIT64(0) (enable, 0x1 when in use, 0x0 when disabled)
+ * DBGBVR<0-15>_EL1 (value, address)
+ * ID_AA64DFR0_EL1
+ * WRPs: BSP_FLD64(val, 20, 23) (watchpoints implemented - 1, 0x0 reserved so minimum 2)
+ * BRPs: BSP_FLD64(val, 12, 15) (breakpoints implemented - 1, 0x0 reserved so minimum 2)
+ * DebugVer: BSP_FLD64(val, 0, 3) (0x6 - 8, 0x7 - 8 w/ VHE, 0x8 - 8.2, 0x9 - 8.4)
+ */
+#define AARCH64_HW_BREAKPOINT_MAX ( 16 )
+
+/*
+ * Types of break points.
+ */
+#define AARCH64_HW_BP_TYPE_UNLINKED_INSTR_MATCH ( 0x0 << 20 )
+#define AARCH64_HW_BP_TYPE_UNLINKED_INSTR_MISMATCH ( 0x4 << 20 )
+
+/*
+ * Byte Address Select
+ */
+#define AARCH64_HW_BP_BAS_A64 ( 0xF << 5 )
+
+/*
+ * Privilege level, corresponds to PMC at 2:1
+ */
+#define AARCH64_HW_BP_PRIV_EL1 ( 0x1 << 1 )
+
+/*
+ * Breakpoint enable.
+ */
+#define AARCH64_HW_BP_ENABLE ( 0x1 )
+
+static aarch64_debug_hwbreak hw_breaks[ AARCH64_HW_BREAKPOINT_MAX ];
+#endif
+
+/*
+ * Target debugging support. Use this to debug the backend.
+ */
+#if TARGET_DEBUG
+
+void rtems_debugger_printk_lock( rtems_interrupt_lock_context *lock_context );
+
+void rtems_debugger_printk_unlock(
+ rtems_interrupt_lock_context *lock_context
+);
+
+static void target_printk( const char *format, ... ) RTEMS_PRINTFLIKE( 1, 2 );
+
+static void target_printk( const char *format, ... )
+{
+ rtems_interrupt_lock_context lock_context;
+ va_list ap;
+
+ va_start( ap, format );
+ rtems_debugger_printk_lock( &lock_context );
+ vprintk( format, ap );
+ rtems_debugger_printk_unlock( &lock_context );
+ va_end( ap );
+}
+
+#else
+#define target_printk( _fmt, ... )
+#endif
+
+static const char *aarch64_mode_label( int mode )
+{
+ switch ( mode ) {
+ case 0x0:
+ return "EL0t";
+ case 0x4:
+ return "EL1t";
+ case 0x5:
+ return "EL1h";
+ }
+
+ return "---";
+}
+
+static int aarch64_debug_probe( rtems_debugger_target *target )
+{
+ int debug_version;
+ uint64_t val;
+ const char *vl = "[Invalid version]";
+ const char * const labels[] = {
+ NULL,
+ NULL,
+ NULL,
+ NULL,
+ NULL,
+ NULL,
+ "ARMv8.0",
+ "ARMv8.0+VHE",
+ "ARMv8.2",
+ "ARMv8.4"
+ };
+
+ val = _AArch64_Read_midr_el1();
+ rtems_debugger_printf(
+ "rtems-db: aarch64 core: Architecture: %" PRIu64 " Variant: %" PRIu64 " " \
+ "Implementor: %" PRIu64 " Part Number: %" PRIu64 " Revision: %" PRIu64 "\n",
+ AARCH64_MIDR_EL1_ARCHITECTURE_GET( val ),
+ AARCH64_MIDR_EL1_VARIANT_GET( val ),
+ AARCH64_MIDR_EL1_IMPLEMENTER_GET( val ),
+ AARCH64_MIDR_EL1_PARTNUM_GET( val ),
+ AARCH64_MIDR_EL1_REVISION_GET( val )
+ );
+
+ val = _AArch64_Read_id_aa64dfr0_el1();
+
+ debug_version = AARCH64_ID_AA64DFR0_EL1_DEBUGVER_GET( val );
+
+ if ( debug_version < 6 || debug_version > 9 ) {
+ rtems_debugger_printf(
+ "rtems-db: aarch64 debug: %d not supported\n",
+ debug_version
+ );
+ errno = EIO;
+ return -1;
+ }
+
+ vl = labels[ debug_version ];
+ hw_breakpoints = AARCH64_ID_AA64DFR0_EL1_BRPS_GET( val );
+ hw_watchpoints = AARCH64_ID_AA64DFR0_EL1_WRPS_GET( val );
+
+ rtems_debugger_printf(
+ "rtems-db: aarch64 debug: %s (%d) " \
+ "breakpoints:%" PRIu64 " watchpoints:%" PRIu64 "\n",
+ vl,
+ debug_version,
+ hw_breakpoints,
+ hw_watchpoints
+ );
+
+ return 0;
+}
+
+#ifdef HARDWARE_BREAKPOINTS_NOT_USED
+static void aarch64_debug_break_write_control( int bp, uint64_t control )
+{
+ if ( bp < 15 ) {
+ switch ( bp ) {
+ case 0:
+ _AArch64_Write_dbgbcr0_el1( control );
+ break;
+ case 1:
+ _AArch64_Write_dbgbcr1_el1( control );
+ break;
+ case 2:
+ _AArch64_Write_dbgbcr2_el1( control );
+ break;
+ case 3:
+ _AArch64_Write_dbgbcr3_el1( control );
+ break;
+ case 4:
+ _AArch64_Write_dbgbcr4_el1( control );
+ break;
+ case 5:
+ _AArch64_Write_dbgbcr5_el1( control );
+ break;
+ case 6:
+ _AArch64_Write_dbgbcr6_el1( control );
+ break;
+ case 7:
+ _AArch64_Write_dbgbcr7_el1( control );
+ break;
+ case 8:
+ _AArch64_Write_dbgbcr8_el1( control );
+ break;
+ case 9:
+ _AArch64_Write_dbgbcr9_el1( control );
+ break;
+ case 10:
+ _AArch64_Write_dbgbcr10_el1( control );
+ break;
+ case 11:
+ _AArch64_Write_dbgbcr11_el1( control );
+ break;
+ case 12:
+ _AArch64_Write_dbgbcr12_el1( control );
+ break;
+ case 13:
+ _AArch64_Write_dbgbcr13_el1( control );
+ break;
+ case 14:
+ _AArch64_Write_dbgbcr14_el1( control );
+ break;
+ case 15:
+ _AArch64_Write_dbgbcr15_el1( control );
+ break;
+ }
+ }
+}
+
+static void aarch64_debug_break_write_value( int bp, uint64_t value )
+{
+ if ( bp < 15 ) {
+ switch ( bp ) {
+ case 0:
+ _AArch64_Write_dbgbvr0_el1( value );
+ break;
+ case 1:
+ _AArch64_Write_dbgbvr1_el1( value );
+ break;
+ case 2:
+ _AArch64_Write_dbgbvr2_el1( value );
+ break;
+ case 3:
+ _AArch64_Write_dbgbvr3_el1( value );
+ break;
+ case 4:
+ _AArch64_Write_dbgbvr4_el1( value );
+ break;
+ case 5:
+ _AArch64_Write_dbgbvr5_el1( value );
+ break;
+ case 6:
+ _AArch64_Write_dbgbvr6_el1( value );
+ break;
+ case 7:
+ _AArch64_Write_dbgbvr7_el1( value );
+ break;
+ case 8:
+ _AArch64_Write_dbgbvr8_el1( value );
+ break;
+ case 9:
+ _AArch64_Write_dbgbvr9_el1( value );
+ break;
+ case 10:
+ _AArch64_Write_dbgbvr10_el1( value );
+ break;
+ case 11:
+ _AArch64_Write_dbgbvr11_el1( value );
+ break;
+ case 12:
+ _AArch64_Write_dbgbvr12_el1( value );
+ break;
+ case 13:
+ _AArch64_Write_dbgbvr13_el1( value );
+ break;
+ case 14:
+ _AArch64_Write_dbgbvr14_el1( value );
+ break;
+ case 15:
+ _AArch64_Write_dbgbvr15_el1( value );
+ break;
+ }
+ }
+}
+
+static inline void aarch64_debug_break_setup(
+ uint8_t index,
+ uint64_t address
+)
+{
+ aarch64_debug_hwbreak *bp = &hw_breaks[ index ];
+
+ bp->control = AARCH64_HW_BP_TYPE_UNLINKED_INSTR_MISMATCH |
+ AARCH64_HW_BP_BAS_A64 |
+ AARCH64_HW_BP_PRIV_EL1 |
+ AARCH64_HW_BP_ENABLE;
+ uint64_t address_mask = 0x3;
+
+ bp->value = (intptr_t) ( address & ~address_mask );
+ aarch64_debug_break_write_value( index, bp->value );
+ aarch64_debug_break_write_control( index, bp->control );
+}
+
+static void aarch64_debug_break_clear( void )
+{
+ rtems_interrupt_lock_context lock_context;
+ aarch64_debug_hwbreak *bp = &hw_breaks[ 0 ];
+ int i;
+
+ rtems_interrupt_lock_acquire( &target_lock, &lock_context );
+
+ for ( i = 0; i < hw_breakpoints; ++i, ++bp ) {
+ bp->enabled = false;
+ bp->loaded = false;
+ }
+
+ rtems_interrupt_lock_release( &target_lock, &lock_context );
+}
+
+static void aarch64_debug_break_load( void )
+{
+ rtems_interrupt_lock_context lock_context;
+ aarch64_debug_hwbreak *bp = &hw_breaks[ 0 ];
+ int i;
+
+ rtems_interrupt_lock_acquire( &target_lock, &lock_context );
+
+ if ( bp->enabled && !bp->loaded ) {
+ aarch64_debug_set_context_id( 0xdead1111 );
+ aarch64_debug_break_write_value( 0, bp->value );
+ aarch64_debug_break_write_control( 0, bp->control );
+ }
+
+ ++bp;
+
+ for ( i = 1; i < hw_breakpoints; ++i, ++bp ) {
+ if ( bp->enabled && !bp->loaded ) {
+ bp->loaded = true;
+ aarch64_debug_break_write_value( i, bp->value );
+ aarch64_debug_break_write_control( i, bp->control );
+ }
+ }
+
+ rtems_interrupt_lock_release( &target_lock, &lock_context );
+}
+
+static void aarch64_debug_break_unload( void )
+{
+ rtems_interrupt_lock_context lock_context;
+ aarch64_debug_hwbreak *bp = &hw_breaks[ 0 ];
+ int i;
+
+ rtems_interrupt_lock_acquire( &target_lock, &lock_context );
+ aarch64_debug_set_context_id( 0 );
+
+ for ( i = 0; i < hw_breakpoints; ++i, ++bp ) {
+ bp->loaded = false;
+ aarch64_debug_break_write_control( i, 0 );
+ }
+
+ rtems_interrupt_lock_release( &target_lock, &lock_context );
+}
+
+static void aarch64_debug_break_dump( void )
+{
+#if TARGET_DEBUG
+ aarch64_debug_hwbreak *bp = &hw_breaks[ 0 ];
+ int i;
+
+ for ( i = 0; i < hw_breakpoints; ++i, ++bp ) {
+ if ( bp->enabled ) {
+ target_printk(
+ "[} bp: %d: control: %016" PRIx64 " addr: %016" PRIxPTR "\n",
+ i,
+ bp->control,
+ (uintptr_t) bp->value
+ );
+ }
+ }
+
+#endif
+}
+#endif
+
+static void aarch64_debug_disable_interrupts( void )
+{
+ __asm__ volatile ( "msr DAIFSet, #0x2" );
+}
+
+static void aarch64_debug_enable_interrupts( void )
+{
+ __asm__ volatile ( "msr DAIFClr, #2\n" );
+}
+
+static void aarch64_debug_disable_debug_exceptions( void )
+{
+ __asm__ volatile ( "msr DAIFSet, #0x8" );
+}
+
+static inline void aarch64_debug_set_context_id( const uint32_t id )
+{
+ _AArch64_Write_contextidr_el1( id );
+}
+
+int rtems_debugger_target_configure( rtems_debugger_target *target )
+{
+ target->capabilities = ( RTEMS_DEBUGGER_TARGET_CAP_SWBREAK );
+ target->reg_num = RTEMS_DEBUGGER_NUMREGS;
+ target->reg_offset = aarch64_reg_offsets;
+ target->breakpoint = &breakpoint[ 0 ];
+ target->breakpoint_size = sizeof( breakpoint );
+ return aarch64_debug_probe( target );
+}
+
+static void target_print_frame( CPU_Exception_frame *frame )
+{
+ EXC_FRAME_PRINT( target_printk, "[} ", frame );
+}
+
+/* returns true if cascade is required */
+static bool target_exception( CPU_Exception_frame *frame )
+{
+ target_printk(
+ "[} > frame = %016" PRIxPTR \
+ " sig=%d" \
+ " pra=%016" PRIxPTR "\n" \
+ "[} > esr=%016" PRIx64 \
+ " far=%016" PRIxPTR "\n",
+ (uintptr_t) frame,
+ rtems_debugger_target_exception_to_signal( frame ),
+ (uintptr_t) frame->register_pc,
+ (uint64_t) frame->register_syndrome,
+ (uintptr_t) frame->register_fault_address
+ );
+
+ target_print_frame( frame );
+
+ switch ( rtems_debugger_target_exception( frame ) ) {
+ case rtems_debugger_target_exc_consumed:
+ default:
+ break;
+ case rtems_debugger_target_exc_step:
+ break;
+ case rtems_debugger_target_exc_cascade:
+ target_printk( "rtems-db: unhandled exception: cascading\n" );
+ /* Continue in fatal error handler chain */
+ return true;
+ break;
+ }
+
+ target_printk(
+ "[} < resuming frame = %016" PRIxPTR "\n",
+ (uintptr_t) frame
+ );
+ target_print_frame( frame );
+
+#if TARGET_DEBUG
+ uint64_t mdscr = _AArch64_Read_mdscr_el1();
+#endif
+ target_printk(
+ "[} global stepping: %s\n",
+ mdscr & AARCH64_MDSCR_EL1_SS ? "yes" : "no"
+ );
+ target_printk(
+ "[} kernel self-debug: %s\n",
+ mdscr & AARCH64_MDSCR_EL1_KDE ? "yes" : "no"
+ );
+ target_printk(
+ "[} non-step/non-BRK debug events: %s\n",
+ mdscr & AARCH64_MDSCR_EL1_MDE ? "yes" : "no"
+ );
+ target_printk(
+ "[} OSLSR(should be 0x8): 0x%016" PRIx64 "\n",
+ _AArch64_Read_oslsr_el1()
+ );
+#ifdef HARDWARE_BREAKPOINTS_NOT_USED
+ aarch64_debug_break_dump();
+#endif
+ return false;
+}
+
+#define xstr( a ) str( a )
+#define str( a ) #a
+#define FRAME_SIZE_STR xstr( AARCH64_EXCEPTION_FRAME_SIZE )
+
+/*
+ * This block of assembly must have a target branch function because GCC
+ * requires that SP not accumulate changes across an ASM block. Instead of
+ * changing the SP, we branch to a new function and never return since it was
+ * never going to return anyway.
+ */
+#define SWITCH_STACKS_AND_ALLOC( new_mode, old_frame, jump_target ) \
+ __asm__ volatile ( \
+ "msr spsel, #" new_mode "\n" /* switch to thread stack */ \
+ "sub sp, sp, #" FRAME_SIZE_STR "\n" /* reserve space for CEF */ \
+ "mov x0, sp\n" /* Set x0 to the new exception frame */ \
+ "mov x1, %[old_frame]\n" /* Set x1 to the old exception frame */ \
+ "b " #jump_target "\n" /* Jump to the specified function */ \
+ : \
+ : [ old_frame ] "r" ( old_frame ) \
+ : "x0", "x1" )
+
+#define SWITCH_STACKS_AND_ALLOC_WITH_CASCADE( new_mode, app_frame, \
+ jump_target ) \
+ __asm__ volatile ( \
+ "msr spsel, #" new_mode "\n" /* switch to thread stack */ \
+ "sub sp, sp, #" FRAME_SIZE_STR "\n" /* reserve space for CEF */ \
+ "mov x0, sp\n" /* Set x0 to the new exception frame */ \
+ "mov x1, %[app_frame]\n" /* Set x1 to the old exception frame */ \
+ "mov x2, %[needs_cascade]\n" /* pass on whether cascade is needed */ \
+ "b " #jump_target "\n" /* Jump to the specified function */ \
+ : \
+ : [ app_frame ] "r" ( app_frame ), \
+ [ needs_cascade ] "r" ( needs_cascade ) \
+ : "x0", "x1" )
+
+/*
+ * This block does not have an overall effect on SP since the spsel mode change
+ * preserves the original SP
+ */
+#define DROP_OLD_FRAME( old_frame, old_mode, new_mode ) \
+ __asm__ volatile ( \
+ "msr spsel, #" old_mode "\n" /* switch to exception stack */ \
+ "mov sp, %0\n" /* Reset SP to the beginning of the CEF */ \
+ "add sp, sp, #" FRAME_SIZE_STR "\n" /* release space for CEF on exception stack */ \
+ "msr spsel, #" new_mode "\n" /* switch to thread stack */ \
+ : \
+ : "r" ( old_frame ) ) \
+
+#define THREAD_MODE "1"
+#define EXCEPTION_MODE "0"
+
+void target_exception_stack_stage_3(
+ CPU_Exception_frame *exc_frame,
+ CPU_Exception_frame *app_frame,
+ bool needs_cascade
+);
+
+void target_exception_stack_stage_3(
+ CPU_Exception_frame *exc_frame,
+ CPU_Exception_frame *app_frame,
+ bool needs_cascade
+)
+{
+ _AArch64_Exception_frame_copy( exc_frame, app_frame );
+ DROP_OLD_FRAME( app_frame, THREAD_MODE, EXCEPTION_MODE );
+
+ if ( needs_cascade ) {
+ /* does not return */
+ _AArch64_Exception_default( exc_frame );
+ }
+
+ /* does not return */
+ _CPU_Exception_resume( exc_frame );
+}
+
+void target_exception_stack_stage_2(
+ CPU_Exception_frame *app_frame,
+ CPU_Exception_frame *exc_frame
+);
+
+void target_exception_stack_stage_2(
+ CPU_Exception_frame *app_frame,
+ CPU_Exception_frame *exc_frame
+)
+{
+ _AArch64_Exception_frame_copy( app_frame, exc_frame );
+ DROP_OLD_FRAME( exc_frame, EXCEPTION_MODE, THREAD_MODE );
+ /* breakpoints must be disabled here since other tasks could run that don't have debug masked */
+#ifdef HARDWARE_BREAKPOINTS_NOT_USED
+ aarch64_debug_break_unload();
+#endif
+ /* enable interrupts here to allow this thread to be suspended as necessary */
+ aarch64_debug_enable_interrupts();
+ bool needs_cascade = target_exception( app_frame );
+
+ /* disable interrupts to return to normal operation */
+ aarch64_debug_disable_interrupts();
+ /* re-enable breakpoints disabled above */
+#ifdef HARDWARE_BREAKPOINTS_NOT_USED
+ aarch64_debug_break_load();
+#endif
+ SWITCH_STACKS_AND_ALLOC_WITH_CASCADE(
+ EXCEPTION_MODE,
+ app_frame,
+ target_exception_stack_stage_3
+ );
+}
+
+/* not allowed to return since it unwinds the stack */
+static void target_exception_thread_stack( CPU_Exception_frame *old_frame )
+{
+ SWITCH_STACKS_AND_ALLOC(
+ THREAD_MODE,
+ old_frame,
+ target_exception_stack_stage_2
+ );
+}
+
+static void target_exception_application( CPU_Exception_frame *ef )
+{
+ /* Continue in fatal error handler chain */
+ if ( !debug_session_active ) {
+ /* does not return */
+ _AArch64_Exception_default( ef );
+ }
+
+ /*
+ * Set CPSR.D to disable single-step operation, this will be cleared before
+ * the thread is resumed if necessary.
+ */
+ ef->register_cpsr |= AARCH64_DSPSR_EL0_D;
+
+ /*
+ * Switching to the user stack is not possible if the stack pointer is bad.
+ * This should be a relatively rare occurrance and signals a severe problem
+ * with the application code or system.
+ */
+ if ( AARCH64_ESR_EL1_EC_GET( ef->register_syndrome ) == 0x26 ) {
+ if ( target_exception( ef ) ) {
+ /* does not return */
+ _AArch64_Exception_default( ef );
+ }
+
+ /* does not return */
+ _CPU_Exception_resume( ef );
+ }
+
+ target_exception_thread_stack( ef );
+}
+
+static void target_exception_kernel( CPU_Exception_frame *ef )
+{
+ /*
+ * If there is a stack alignment problem in exception mode, it really
+ * shouldn't happen and execution won't even make it this far.
+ */
+ if ( !debug_session_active ) {
+ /* does not return */
+ _AArch64_Exception_default( ef );
+ }
+
+ /*
+ * Set CPSR.D to disable single-step operation, this will be cleared before
+ * the thread is resumed if necessary.
+ */
+ ef->register_cpsr |= AARCH64_DSPSR_EL0_D;
+
+ if ( target_exception( ef ) ) {
+ /* does not return */
+ _AArch64_Exception_default( ef );
+ }
+
+ /* does not return */
+ _CPU_Exception_resume( ef );
+}
+
+static void rtems_debugger_target_set_vectors( void )
+{
+ /* Set vectors for both application and kernel modes */
+ AArch64_set_exception_handler(
+ AARCH64_EXCEPTION_SPx_SYNCHRONOUS,
+ (void *) target_exception_application
+ );
+ AArch64_set_exception_handler(
+ AARCH64_EXCEPTION_SP0_SYNCHRONOUS,
+ (void *) target_exception_kernel
+ );
+}
+
+static bool rtems_debugger_is_int_reg( size_t reg )
+{
+ const size_t size = aarch64_reg_offsets[ reg + 1 ] -
+ aarch64_reg_offsets[ reg ];
+
+ return size == RTEMS_DEBUGGER_REG_BYTES;
+}
+
+static void rtems_debugger_set_int_reg(
+ rtems_debugger_thread *thread,
+ size_t reg,
+ const uint64_t value
+)
+{
+ const size_t offset = aarch64_reg_offsets[ reg ];
+
+ memcpy( &thread->registers[ offset ], &value, sizeof( uint64_t ) );
+}
+
+static const uint64_t rtems_debugger_get_int_reg(
+ rtems_debugger_thread *thread,
+ size_t reg
+)
+{
+ const size_t offset = aarch64_reg_offsets[ reg ];
+ uint64_t value;
+
+ memcpy( &value, &thread->registers[ offset ], sizeof( uint64_t ) );
+ return value;
+}
+
+static void rtems_debugger_set_halfint_reg(
+ rtems_debugger_thread *thread,
+ size_t reg,
+ const uint32_t value
+)
+{
+ const size_t offset = aarch64_reg_offsets[ reg ];
+
+ memcpy( &thread->registers[ offset ], &value, sizeof( uint32_t ) );
+}
+
+static const uint32_t rtems_debugger_get_halfint_reg(
+ rtems_debugger_thread *thread,
+ size_t reg
+)
+{
+ const size_t offset = aarch64_reg_offsets[ reg ];
+ uint32_t value;
+
+ memcpy( &value, &thread->registers[ offset ], sizeof( uint32_t ) );
+ return value;
+}
+
+static void rtems_debugger_set_fp_reg(
+ rtems_debugger_thread *thread,
+ size_t reg,
+ const uint128_t value
+)
+{
+ const size_t offset = aarch64_reg_offsets[ reg ];
+
+ memcpy( &thread->registers[ offset ], &value, sizeof( uint128_t ) );
+}
+
+static const uint128_t rtems_debugger_get_fp_reg(
+ rtems_debugger_thread *thread,
+ size_t reg
+)
+{
+ const size_t offset = aarch64_reg_offsets[ reg ];
+ uint128_t value;
+
+ memcpy( &value, &thread->registers[ offset ], sizeof( uint128_t ) );
+ return value;
+}
+
+static rtems_status_code rtems_debugger_target_set_text_writable(
+ bool writable
+)
+{
+ uintptr_t text_begin = (uintptr_t) bsp_section_text_begin;
+ uintptr_t text_end = (uintptr_t) bsp_section_text_end;
+ uintptr_t fast_text_begin = (uintptr_t) bsp_section_fast_text_begin;
+ uintptr_t fast_text_end = (uintptr_t) bsp_section_fast_text_end;
+ uint64_t mmu_flags = AARCH64_MMU_CODE_RW_CACHED;
+ rtems_status_code sc;
+
+ if ( !writable ) {
+ mmu_flags = AARCH64_MMU_CODE_CACHED;
+ }
+
+ target_printk(
+ "[} MMU edit: text_begin: 0x%016" PRIxPTR
+ " text_end: 0x%016" PRIxPTR "\n",
+ text_begin,
+ text_end
+ );
+ sc = aarch64_mmu_map(
+ text_begin,
+ text_end - text_begin,
+ mmu_flags
+ );
+
+ if ( sc != RTEMS_SUCCESSFUL ) {
+ target_printk( "[} MMU edit failed\n" );
+ return sc;
+ }
+
+ target_printk(
+ "[} MMU edit: fast_text_begin: 0x%016" PRIxPTR
+ " fast_text_end: 0x%016" PRIxPTR "\n",
+ fast_text_begin,
+ fast_text_end
+ );
+ sc = aarch64_mmu_map(
+ fast_text_begin,
+ fast_text_end - fast_text_begin,
+ mmu_flags
+ );
+
+ if ( sc != RTEMS_SUCCESSFUL ) {
+ target_printk( "[} MMU edit failed\n" );
+ }
+
+ return sc;
+}
+
+static rtems_task setup_debugger_on_cpu( rtems_task_argument arg )
+{
+ rtems_status_code sc;
+ rtems_status_code *init_error = (rtems_status_code *) arg;
+ rtems_interrupt_lock_context lock_context;
+
+ rtems_interrupt_lock_acquire( &target_lock, &lock_context );
+ sc = rtems_debugger_target_set_text_writable( true );
+
+ if ( sc != RTEMS_SUCCESSFUL ) {
+ *init_error = sc;
+ }
+
+ rtems_debugger_target_set_vectors();
+
+ /* enable single-step debugging */
+ uint64_t mdscr = _AArch64_Read_mdscr_el1();
+
+ mdscr |= AARCH64_MDSCR_EL1_SS;
+ mdscr |= AARCH64_MDSCR_EL1_KDE;
+ mdscr |= AARCH64_MDSCR_EL1_MDE;
+ _AArch64_Write_mdscr_el1( mdscr );
+
+ /* clear the OS lock */
+ _AArch64_Write_oslar_el1( 0 );
+ rtems_interrupt_lock_release( &target_lock, &lock_context );
+}
+
+int rtems_debugger_target_enable( void )
+{
+ rtems_status_code sc;
+ rtems_status_code init_error = RTEMS_SUCCESSFUL;
+
+ debug_session_active = true;
+#ifdef HARDWARE_BREAKPOINTS_NOT_USED
+ aarch64_debug_break_unload();
+ aarch64_debug_break_clear();
+#endif
+ aarch64_debug_disable_debug_exceptions();
+ sc = run_across_cpus(
+ setup_debugger_on_cpu,
+ ( rtems_task_argument ) & init_error
+ );
+
+ if ( init_error != RTEMS_SUCCESSFUL ) {
+ return init_error;
+ }
+
+ return sc;
+}
+
+static rtems_task teardown_debugger_on_cpu( rtems_task_argument arg )
+{
+ rtems_status_code sc;
+ rtems_status_code *deinit_error = (rtems_status_code *) arg;
+ rtems_interrupt_lock_context lock_context;
+
+ rtems_interrupt_lock_acquire( &target_lock, &lock_context );
+ sc = rtems_debugger_target_set_text_writable( false );
+
+ if ( sc != RTEMS_SUCCESSFUL ) {
+ *deinit_error = sc;
+ }
+
+ /* disable single-step debugging */
+ uint64_t mdscr = _AArch64_Read_mdscr_el1();
+
+ mdscr &= ~AARCH64_MDSCR_EL1_SS;
+ mdscr &= ~AARCH64_MDSCR_EL1_KDE;
+ mdscr &= ~AARCH64_MDSCR_EL1_MDE;
+ _AArch64_Write_mdscr_el1( mdscr );
+
+ rtems_interrupt_lock_release( &target_lock, &lock_context );
+}
+
+int rtems_debugger_target_disable( void )
+{
+ rtems_status_code sc;
+ rtems_status_code deinit_error = RTEMS_SUCCESSFUL;
+
+ debug_session_active = false;
+#ifdef HARDWARE_BREAKPOINTS_NOT_USED
+ aarch64_debug_break_unload();
+ aarch64_debug_break_clear();
+#endif
+ sc = run_across_cpus(
+ teardown_debugger_on_cpu,
+ ( rtems_task_argument ) & deinit_error
+ );
+
+ if ( deinit_error != RTEMS_SUCCESSFUL ) {
+ return deinit_error;
+ }
+
+ return sc;
+}
+
+int rtems_debugger_target_read_regs( rtems_debugger_thread *thread )
+{
+ if (
+ !rtems_debugger_thread_flag(
+ thread,
+ RTEMS_DEBUGGER_THREAD_FLAG_REG_VALID
+ )
+ ) {
+ static const uintptr_t good_address = (uintptr_t) &good_address;
+ int i;
+
+ memset( &thread->registers[ 0 ], 0, RTEMS_DEBUGGER_NUMREGBYTES );
+
+ /* set all integer register to a known valid address */
+ for ( i = 0; i < RTEMS_DEBUGGER_NUMREGS; ++i ) {
+ if ( rtems_debugger_is_int_reg( i ) ) {
+ rtems_debugger_set_int_reg( thread, i, (uintptr_t) &good_address );
+ }
+ }
+
+ if ( thread->frame ) {
+ CPU_Exception_frame *frame = thread->frame;
+
+ *( (CPU_Exception_frame *) thread->registers ) = *frame;
+ rtems_debugger_set_int_reg( thread, REG_X0, frame->register_x0 );
+ rtems_debugger_set_int_reg( thread, REG_X1, frame->register_x1 );
+ rtems_debugger_set_int_reg( thread, REG_X2, frame->register_x2 );
+ rtems_debugger_set_int_reg( thread, REG_X3, frame->register_x3 );
+ rtems_debugger_set_int_reg( thread, REG_X4, frame->register_x4 );
+ rtems_debugger_set_int_reg( thread, REG_X5, frame->register_x5 );
+ rtems_debugger_set_int_reg( thread, REG_X6, frame->register_x6 );
+ rtems_debugger_set_int_reg( thread, REG_X7, frame->register_x7 );
+ rtems_debugger_set_int_reg( thread, REG_X8, frame->register_x8 );
+ rtems_debugger_set_int_reg( thread, REG_X9, frame->register_x9 );
+ rtems_debugger_set_int_reg( thread, REG_X10, frame->register_x10 );
+ rtems_debugger_set_int_reg( thread, REG_X11, frame->register_x11 );
+ rtems_debugger_set_int_reg( thread, REG_X12, frame->register_x12 );
+ rtems_debugger_set_int_reg( thread, REG_X13, frame->register_x13 );
+ rtems_debugger_set_int_reg( thread, REG_X14, frame->register_x14 );
+ rtems_debugger_set_int_reg( thread, REG_X15, frame->register_x15 );
+ rtems_debugger_set_int_reg( thread, REG_X16, frame->register_x16 );
+ rtems_debugger_set_int_reg( thread, REG_X17, frame->register_x17 );
+ rtems_debugger_set_int_reg( thread, REG_X18, frame->register_x18 );
+ rtems_debugger_set_int_reg( thread, REG_X19, frame->register_x19 );
+ rtems_debugger_set_int_reg( thread, REG_X20, frame->register_x20 );
+ rtems_debugger_set_int_reg( thread, REG_X21, frame->register_x21 );
+ rtems_debugger_set_int_reg( thread, REG_X22, frame->register_x22 );
+ rtems_debugger_set_int_reg( thread, REG_X23, frame->register_x23 );
+ rtems_debugger_set_int_reg( thread, REG_X24, frame->register_x24 );
+ rtems_debugger_set_int_reg( thread, REG_X25, frame->register_x25 );
+ rtems_debugger_set_int_reg( thread, REG_X26, frame->register_x26 );
+ rtems_debugger_set_int_reg( thread, REG_X27, frame->register_x27 );
+ rtems_debugger_set_int_reg( thread, REG_X28, frame->register_x28 );
+ rtems_debugger_set_int_reg( thread, REG_FP, frame->register_fp );
+ rtems_debugger_set_int_reg(
+ thread,
+ REG_LR,
+ (intptr_t) frame->register_lr
+ );
+ rtems_debugger_set_int_reg(
+ thread,
+ REG_SP,
+ (intptr_t) frame->register_sp
+ );
+ rtems_debugger_set_int_reg(
+ thread,
+ REG_PC,
+ (intptr_t) frame->register_pc
+ );
+ /* GDB considers CPSR to be 32-bit because bits 63:32 are RES0 */
+ rtems_debugger_set_halfint_reg(
+ thread,
+ REG_CPS,
+ (uint32_t) frame->register_cpsr
+ );
+ rtems_debugger_set_fp_reg( thread, REG_V0, frame->register_q0 );
+ rtems_debugger_set_fp_reg( thread, REG_V1, frame->register_q1 );
+ rtems_debugger_set_fp_reg( thread, REG_V2, frame->register_q2 );
+ rtems_debugger_set_fp_reg( thread, REG_V3, frame->register_q3 );
+ rtems_debugger_set_fp_reg( thread, REG_V4, frame->register_q4 );
+ rtems_debugger_set_fp_reg( thread, REG_V5, frame->register_q5 );
+ rtems_debugger_set_fp_reg( thread, REG_V6, frame->register_q6 );
+ rtems_debugger_set_fp_reg( thread, REG_V7, frame->register_q7 );
+ rtems_debugger_set_fp_reg( thread, REG_V8, frame->register_q8 );
+ rtems_debugger_set_fp_reg( thread, REG_V9, frame->register_q9 );
+ rtems_debugger_set_fp_reg( thread, REG_V10, frame->register_q10 );
+ rtems_debugger_set_fp_reg( thread, REG_V11, frame->register_q11 );
+ rtems_debugger_set_fp_reg( thread, REG_V12, frame->register_q12 );
+ rtems_debugger_set_fp_reg( thread, REG_V13, frame->register_q13 );
+ rtems_debugger_set_fp_reg( thread, REG_V14, frame->register_q14 );
+ rtems_debugger_set_fp_reg( thread, REG_V15, frame->register_q15 );
+ rtems_debugger_set_fp_reg( thread, REG_V16, frame->register_q16 );
+ rtems_debugger_set_fp_reg( thread, REG_V17, frame->register_q17 );
+ rtems_debugger_set_fp_reg( thread, REG_V18, frame->register_q18 );
+ rtems_debugger_set_fp_reg( thread, REG_V19, frame->register_q19 );
+ rtems_debugger_set_fp_reg( thread, REG_V20, frame->register_q20 );
+ rtems_debugger_set_fp_reg( thread, REG_V21, frame->register_q21 );
+ rtems_debugger_set_fp_reg( thread, REG_V22, frame->register_q22 );
+ rtems_debugger_set_fp_reg( thread, REG_V23, frame->register_q23 );
+ rtems_debugger_set_fp_reg( thread, REG_V24, frame->register_q24 );
+ rtems_debugger_set_fp_reg( thread, REG_V25, frame->register_q25 );
+ rtems_debugger_set_fp_reg( thread, REG_V26, frame->register_q26 );
+ rtems_debugger_set_fp_reg( thread, REG_V27, frame->register_q27 );
+ rtems_debugger_set_fp_reg( thread, REG_V28, frame->register_q28 );
+ rtems_debugger_set_fp_reg( thread, REG_V29, frame->register_q29 );
+ rtems_debugger_set_fp_reg( thread, REG_V30, frame->register_q30 );
+ rtems_debugger_set_fp_reg( thread, REG_V31, frame->register_q31 );
+ /* GDB considers FPSR and FPCR to be 32-bit because bits 63:32 are RES0 */
+ rtems_debugger_set_halfint_reg( thread, REG_FPS, frame->register_fpsr );
+ rtems_debugger_set_halfint_reg( thread, REG_FPC, frame->register_fpcr );
+ /*
+ * Get the signal from the frame.
+ */
+ thread->signal = rtems_debugger_target_exception_to_signal( frame );
+ } else {
+ rtems_debugger_set_int_reg(
+ thread,
+ REG_X19,
+ thread->tcb->Registers.register_x19
+ );
+ rtems_debugger_set_int_reg(
+ thread,
+ REG_X20,
+ thread->tcb->Registers.register_x20
+ );
+ rtems_debugger_set_int_reg(
+ thread,
+ REG_X21,
+ thread->tcb->Registers.register_x21
+ );
+ rtems_debugger_set_int_reg(
+ thread,
+ REG_X22,
+ thread->tcb->Registers.register_x22
+ );
+ rtems_debugger_set_int_reg(
+ thread,
+ REG_X23,
+ thread->tcb->Registers.register_x23
+ );
+ rtems_debugger_set_int_reg(
+ thread,
+ REG_X24,
+ thread->tcb->Registers.register_x24
+ );
+ rtems_debugger_set_int_reg(
+ thread,
+ REG_X25,
+ thread->tcb->Registers.register_x25
+ );
+ rtems_debugger_set_int_reg(
+ thread,
+ REG_X26,
+ thread->tcb->Registers.register_x26
+ );
+ rtems_debugger_set_int_reg(
+ thread,
+ REG_X27,
+ thread->tcb->Registers.register_x27
+ );
+ rtems_debugger_set_int_reg(
+ thread,
+ REG_X28,
+ thread->tcb->Registers.register_x28
+ );
+ rtems_debugger_set_int_reg(
+ thread,
+ REG_FP,
+ thread->tcb->Registers.register_fp
+ );
+ rtems_debugger_set_int_reg(
+ thread,
+ REG_LR,
+ (intptr_t) thread->tcb->Registers.register_lr
+ );
+ rtems_debugger_set_int_reg(
+ thread,
+ REG_SP,
+ (intptr_t) thread->tcb->Registers.register_sp
+ );
+ rtems_debugger_set_int_reg(
+ thread,
+ REG_PC,
+ (intptr_t) thread->tcb->Registers.register_lr
+ );
+ rtems_debugger_set_int_reg(
+ thread,
+ REG_V8,
+ thread->tcb->Registers.register_d8
+ );
+ rtems_debugger_set_int_reg(
+ thread,
+ REG_V9,
+ thread->tcb->Registers.register_d9
+ );
+ rtems_debugger_set_int_reg(
+ thread,
+ REG_V10,
+ thread->tcb->Registers.register_d10
+ );
+ rtems_debugger_set_int_reg(
+ thread,
+ REG_V11,
+ thread->tcb->Registers.register_d11
+ );
+ rtems_debugger_set_int_reg(
+ thread,
+ REG_V12,
+ thread->tcb->Registers.register_d12
+ );
+ rtems_debugger_set_int_reg(
+ thread,
+ REG_V13,
+ thread->tcb->Registers.register_d13
+ );
+ rtems_debugger_set_int_reg(
+ thread,
+ REG_V14,
+ thread->tcb->Registers.register_d14
+ );
+ rtems_debugger_set_int_reg(
+ thread,
+ REG_V15,
+ thread->tcb->Registers.register_d15
+ );
+ /*
+ * Blocked threads have no signal.
+ */
+ thread->signal = 0;
+ }
+
+ thread->flags |= RTEMS_DEBUGGER_THREAD_FLAG_REG_VALID;
+ thread->flags &= ~RTEMS_DEBUGGER_THREAD_FLAG_REG_DIRTY;
+ }
+
+ return 0;
+}
+
+int rtems_debugger_target_write_regs( rtems_debugger_thread *thread )
+{
+ if (
+ rtems_debugger_thread_flag(
+ thread,
+ RTEMS_DEBUGGER_THREAD_FLAG_REG_DIRTY
+ )
+ ) {
+ /*
+ * Only write to debugger controlled exception threads. Do not touch the
+ * registers for threads blocked in the context switcher.
+ */
+ if (
+ rtems_debugger_thread_flag(
+ thread,
+ RTEMS_DEBUGGER_THREAD_FLAG_EXCEPTION
+ )
+ ) {
+ CPU_Exception_frame *frame = thread->frame;
+ frame->register_x0 = rtems_debugger_get_int_reg( thread, REG_X0 );
+ frame->register_x1 = rtems_debugger_get_int_reg( thread, REG_X1 );
+ frame->register_x2 = rtems_debugger_get_int_reg( thread, REG_X2 );
+ frame->register_x3 = rtems_debugger_get_int_reg( thread, REG_X3 );
+ frame->register_x4 = rtems_debugger_get_int_reg( thread, REG_X4 );
+ frame->register_x5 = rtems_debugger_get_int_reg( thread, REG_X5 );
+ frame->register_x6 = rtems_debugger_get_int_reg( thread, REG_X6 );
+ frame->register_x7 = rtems_debugger_get_int_reg( thread, REG_X7 );
+ frame->register_x8 = rtems_debugger_get_int_reg( thread, REG_X8 );
+ frame->register_x9 = rtems_debugger_get_int_reg( thread, REG_X9 );
+ frame->register_x10 = rtems_debugger_get_int_reg( thread, REG_X10 );
+ frame->register_x11 = rtems_debugger_get_int_reg( thread, REG_X11 );
+ frame->register_x12 = rtems_debugger_get_int_reg( thread, REG_X12 );
+ frame->register_x13 = rtems_debugger_get_int_reg( thread, REG_X13 );
+ frame->register_x14 = rtems_debugger_get_int_reg( thread, REG_X14 );
+ frame->register_x15 = rtems_debugger_get_int_reg( thread, REG_X15 );
+ frame->register_x16 = rtems_debugger_get_int_reg( thread, REG_X16 );
+ frame->register_x17 = rtems_debugger_get_int_reg( thread, REG_X17 );
+ frame->register_x18 = rtems_debugger_get_int_reg( thread, REG_X18 );
+ frame->register_x19 = rtems_debugger_get_int_reg( thread, REG_X19 );
+ frame->register_x20 = rtems_debugger_get_int_reg( thread, REG_X20 );
+ frame->register_x21 = rtems_debugger_get_int_reg( thread, REG_X21 );
+ frame->register_x22 = rtems_debugger_get_int_reg( thread, REG_X22 );
+ frame->register_x23 = rtems_debugger_get_int_reg( thread, REG_X23 );
+ frame->register_x24 = rtems_debugger_get_int_reg( thread, REG_X24 );
+ frame->register_x25 = rtems_debugger_get_int_reg( thread, REG_X25 );
+ frame->register_x26 = rtems_debugger_get_int_reg( thread, REG_X26 );
+ frame->register_x27 = rtems_debugger_get_int_reg( thread, REG_X27 );
+ frame->register_x28 = rtems_debugger_get_int_reg( thread, REG_X28 );
+ frame->register_fp = (uintptr_t) rtems_debugger_get_int_reg(
+ thread,
+ REG_FP
+ );
+ frame->register_lr = (void *) (uintptr_t) rtems_debugger_get_int_reg(
+ thread,
+ REG_LR
+ );
+ frame->register_sp = (uintptr_t) rtems_debugger_get_int_reg(
+ thread,
+ REG_SP
+ );
+ frame->register_pc = (void *) (uintptr_t) rtems_debugger_get_int_reg(
+ thread,
+ REG_PC
+ );
+ frame->register_cpsr = rtems_debugger_get_halfint_reg( thread, REG_CPS );
+ frame->register_q0 = rtems_debugger_get_fp_reg( thread, REG_V0 );
+ frame->register_q1 = rtems_debugger_get_fp_reg( thread, REG_V1 );
+ frame->register_q2 = rtems_debugger_get_fp_reg( thread, REG_V2 );
+ frame->register_q3 = rtems_debugger_get_fp_reg( thread, REG_V3 );
+ frame->register_q4 = rtems_debugger_get_fp_reg( thread, REG_V4 );
+ frame->register_q5 = rtems_debugger_get_fp_reg( thread, REG_V5 );
+ frame->register_q6 = rtems_debugger_get_fp_reg( thread, REG_V6 );
+ frame->register_q7 = rtems_debugger_get_fp_reg( thread, REG_V7 );
+ frame->register_q8 = rtems_debugger_get_fp_reg( thread, REG_V8 );
+ frame->register_q9 = rtems_debugger_get_fp_reg( thread, REG_V9 );
+ frame->register_q10 = rtems_debugger_get_fp_reg( thread, REG_V10 );
+ frame->register_q11 = rtems_debugger_get_fp_reg( thread, REG_V11 );
+ frame->register_q12 = rtems_debugger_get_fp_reg( thread, REG_V12 );
+ frame->register_q13 = rtems_debugger_get_fp_reg( thread, REG_V13 );
+ frame->register_q14 = rtems_debugger_get_fp_reg( thread, REG_V14 );
+ frame->register_q15 = rtems_debugger_get_fp_reg( thread, REG_V15 );
+ frame->register_q16 = rtems_debugger_get_fp_reg( thread, REG_V16 );
+ frame->register_q17 = rtems_debugger_get_fp_reg( thread, REG_V17 );
+ frame->register_q18 = rtems_debugger_get_fp_reg( thread, REG_V18 );
+ frame->register_q19 = rtems_debugger_get_fp_reg( thread, REG_V19 );
+ frame->register_q20 = rtems_debugger_get_fp_reg( thread, REG_V20 );
+ frame->register_q21 = rtems_debugger_get_fp_reg( thread, REG_V21 );
+ frame->register_q22 = rtems_debugger_get_fp_reg( thread, REG_V22 );
+ frame->register_q23 = rtems_debugger_get_fp_reg( thread, REG_V23 );
+ frame->register_q24 = rtems_debugger_get_fp_reg( thread, REG_V24 );
+ frame->register_q25 = rtems_debugger_get_fp_reg( thread, REG_V25 );
+ frame->register_q26 = rtems_debugger_get_fp_reg( thread, REG_V26 );
+ frame->register_q27 = rtems_debugger_get_fp_reg( thread, REG_V27 );
+ frame->register_q28 = rtems_debugger_get_fp_reg( thread, REG_V28 );
+ frame->register_q29 = rtems_debugger_get_fp_reg( thread, REG_V29 );
+ frame->register_q30 = rtems_debugger_get_fp_reg( thread, REG_V30 );
+ frame->register_q31 = rtems_debugger_get_fp_reg( thread, REG_V31 );
+ frame->register_fpsr = rtems_debugger_get_halfint_reg( thread, REG_FPS );
+ frame->register_fpcr = rtems_debugger_get_halfint_reg( thread, REG_FPC );
+ }
+
+ thread->flags &= ~RTEMS_DEBUGGER_THREAD_FLAG_REG_DIRTY;
+ }
+
+ return 0;
+}
+
+uintptr_t rtems_debugger_target_reg_pc( rtems_debugger_thread *thread )
+{
+ int r;
+
+ r = rtems_debugger_target_read_regs( thread );
+
+ if ( r >= 0 ) {
+ return rtems_debugger_get_int_reg( thread, REG_PC );
+ }
+
+ return 0;
+}
+
+uintptr_t rtems_debugger_target_frame_pc( CPU_Exception_frame *frame )
+{
+ return (uintptr_t) frame->register_pc;
+}
+
+uintptr_t rtems_debugger_target_reg_sp( rtems_debugger_thread *thread )
+{
+ int r;
+
+ r = rtems_debugger_target_read_regs( thread );
+
+ if ( r >= 0 ) {
+ return rtems_debugger_get_int_reg( thread, REG_SP );
+ }
+
+ return 0;
+}
+
+uintptr_t rtems_debugger_target_tcb_sp( rtems_debugger_thread *thread )
+{
+ return (uintptr_t) thread->tcb->Registers.register_sp;
+}
+
+int rtems_debugger_target_thread_stepping( rtems_debugger_thread *thread )
+{
+ CPU_Exception_frame *frame = thread->frame;
+
+ if ( rtems_debugger_thread_flag(
+ thread,
+ RTEMS_DEBUGGER_THREAD_FLAG_STEP_INSTR
+ ) ) {
+ /* Especially on first startup, frame isn't guaranteed to be non-NULL */
+ if ( frame == NULL ) {
+ return -1;
+ }
+
+ /*
+ * Single stepping uses AArch64-specific single-step mode and does not
+ * involve hardware breakpoints.
+ */
+
+ /* Breakpoint instruction exceptions occur even when D is not set. */
+ uint64_t stepping_enabled =
+ !( frame->register_cpsr & AARCH64_DSPSR_EL0_D );
+
+ target_printk( "[} stepping: %s\n", stepping_enabled ? "yes" : "no" );
+
+ /*
+ * This field is unset by the CPU during the software step process and must
+ * be set again each time the debugger needs to advance one instruction. If
+ * this is not set each time, the software step exception will trigger
+ * before executing an instruction.
+ */
+ frame->register_cpsr |= AARCH64_DSPSR_EL0_SS;
+
+ if ( !stepping_enabled ) {
+ /*
+ * Clear CPSR.D to enable single-step operation. The debug mask flag is
+ * set on taking an exception to prevent unwanted stepping. The way
+ * single-stepping works will need to change if hardware breakpoints and
+ * watchpoints are ever used.
+ */
+ frame->register_cpsr &= ~AARCH64_DSPSR_EL0_D;
+ }
+ }
+
+ return 0;
+}
+
+int rtems_debugger_target_exception_to_signal( CPU_Exception_frame *frame )
+{
+ uint64_t EC = AARCH64_ESR_EL1_EC_GET( frame->register_syndrome );
+
+ switch ( EC ) {
+ case 0x1: /* WFI */
+ case 0x7: /* SVE/SIMD/FP */
+ case 0xa: /* LD64B/ST64B* */
+ case 0x15:
+ case 0x18: /* MSR/MRS/system instruction */
+ case 0x19: /* SVE */
+ case 0x31:
+ case 0x33:
+ case 0x35:
+ case 0x3c:
+ return RTEMS_DEBUGGER_SIGNAL_TRAP;
+
+ case 0x2c:
+ return RTEMS_DEBUGGER_SIGNAL_FPE;
+
+ case 0x21:
+ case 0x25:
+ return RTEMS_DEBUGGER_SIGNAL_SEGV;
+
+ default:
+ /*
+ * Covers unknown, SP/PC alignment, illegal execution state, and any new
+ * exception classes that get added.
+ */
+ return RTEMS_DEBUGGER_SIGNAL_ILL;
+ }
+}
+
+void rtems_debugger_target_exception_print( CPU_Exception_frame *frame )
+{
+ EXC_FRAME_PRINT( rtems_debugger_printf, "", frame );
+}
+
+int rtems_debugger_target_hwbreak_insert( void )
+{
+ /*
+ * Do nothing, these are loaded elsewhere if needed.
+ */
+ return 0;
+}
+
+int rtems_debugger_target_hwbreak_remove( void )
+{
+#ifdef HARDWARE_BREAKPOINTS_NOT_USED
+ aarch64_debug_break_unload();
+#endif
+ return 0;
+}
+
+int rtems_debugger_target_hwbreak_control(
+ rtems_debugger_target_watchpoint wp,
+ bool insert,
+ uintptr_t addr,
+ DB_UINT kind
+)
+{
+ /* To do. */
+ return 0;
+}
+
+int rtems_debugger_target_cache_sync( rtems_debugger_target_swbreak *swbreak )
+{
+ /*
+ * Flush the data cache and invalidate the instruction cache.
+ */
+ rtems_cache_flush_multiple_data_lines(
+ swbreak->address,
+ sizeof( breakpoint )
+ );
+ rtems_cache_instruction_sync_after_code_change(
+ swbreak->address,
+ sizeof( breakpoint )
+ );
+ return 0;
+}
diff --git a/spec/build/cpukit/libdebugger.yml b/spec/build/cpukit/libdebugger.yml
index 5069571a76..9b50adbe77 100644
--- a/spec/build/cpukit/libdebugger.yml
+++ b/spec/build/cpukit/libdebugger.yml
@@ -10,6 +10,8 @@ includes: []
install: []
install-path: ${BSP_LIBDIR}
links:
+- role: build-dependency
+ uid: objdbgaarch64
- role: build-dependency
uid: objdbgarm
- role: build-dependency
diff --git a/spec/build/cpukit/objdbgaarch64.yml b/spec/build/cpukit/objdbgaarch64.yml
new file mode 100644
index 0000000000..5dded29c5f
--- /dev/null
+++ b/spec/build/cpukit/objdbgaarch64.yml
@@ -0,0 +1,15 @@
+SPDX-License-Identifier: CC-BY-SA-4.0 OR BSD-2-Clause
+build-type: objects
+cflags: []
+copyrights:
+- Copyright (C) 2021 On-Line Applications Research (OAR)
+cppflags: []
+cxxflags: []
+enabled-by:
+- aarch64
+includes: []
+install: []
+links: []
+source:
+- cpukit/libdebugger/rtems-debugger-aarch64.c
+type: build
diff --git a/spec/build/cpukit/optlibdebugger.yml b/spec/build/cpukit/optlibdebugger.yml
index 5c7df0e0c9..a89a495528 100644
--- a/spec/build/cpukit/optlibdebugger.yml
+++ b/spec/build/cpukit/optlibdebugger.yml
@@ -9,6 +9,7 @@ default: null
default-by-variant: []
description: ''
enabled-by:
+- aarch64
- arm
- i386
links: []
--
2.30.2
More information about the devel
mailing list