[MIPS] kgdb: add arch support for the kernel's kgdb core

The new kgdb architecture specific handler registers and unregisters
dynamically for exceptions depending on when you configure a kgdb I/O
driver.

Aside from initializing the exceptions earlier in the boot process,
kgdb should have no impact on a device when it is compiled in so long
as an I/O module is not configured for use.

There have been quite a number of contributors during the existence of
this patch (see arch/mips/kernel/kgdb.c).  Most recently Jason
re-wrote the mips kgdb logic to use the die notification handlers.

Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

arch/mips/Kconfig

@@ -3,6 +3,7 @@ config MIPS
 	default y
 	select HAVE_IDE
 	select HAVE_OPROFILE
+	select HAVE_ARCH_KGDB
 	# Horrible source of confusion.  Die, die, die ...
 	select EMBEDDED
 	select RTC_LIB

arch/mips/kernel/Makefile

@@ -71,6 +71,7 @@ obj-$(CONFIG_MIPS32_COMPAT)	+= linux32.o ptrace32.o signal32.o
 obj-$(CONFIG_MIPS32_N32)	+= binfmt_elfn32.o scall64-n32.o signal_n32.o
 obj-$(CONFIG_MIPS32_O32)	+= binfmt_elfo32.o scall64-o32.o
 
+obj-$(CONFIG_KGDB)		+= kgdb.o
 obj-$(CONFIG_PROC_FS)		+= proc.o
 
 obj-$(CONFIG_64BIT)		+= cpu-bugs64.o

arch/mips/kernel/irq.c

@@ -21,11 +21,16 @@
 #include <linux/sched.h>
 #include <linux/seq_file.h>
 #include <linux/kallsyms.h>
+#include <linux/kgdb.h>
 
 #include <asm/atomic.h>
 #include <asm/system.h>
 #include <asm/uaccess.h>
 
+#ifdef CONFIG_KGDB
+int kgdb_early_setup;
+#endif
+
 static unsigned long irq_map[NR_IRQS / BITS_PER_LONG];
 
 int allocate_irqno(void)
@@ -130,8 +135,18 @@ void __init init_IRQ(void)
 {
 	int i;
 
+#ifdef CONFIG_KGDB
+	if (kgdb_early_setup)
+		return;
+#endif
+
 	for (i = 0; i < NR_IRQS; i++)
 		set_irq_noprobe(i);
 
 	arch_init_irq();
+
+#ifdef CONFIG_KGDB
+	if (!kgdb_early_setup)
+		kgdb_early_setup = 1;
+#endif
 }

arch/mips/kernel/kgdb.c

@@ -0,0 +1,281 @@
+/*
+ *  Originally written by Glenn Engel, Lake Stevens Instrument Division
+ *
+ *  Contributed by HP Systems
+ *
+ *  Modified for Linux/MIPS (and MIPS in general) by Andreas Busse
+ *  Send complaints, suggestions etc. to <andy@waldorf-gmbh.de>
+ *
+ *  Copyright (C) 1995 Andreas Busse
+ *
+ *  Copyright (C) 2003 MontaVista Software Inc.
+ *  Author: Jun Sun, jsun@mvista.com or jsun@junsun.net
+ *
+ *  Copyright (C) 2004-2005 MontaVista Software Inc.
+ *  Author: Manish Lachwani, mlachwani@mvista.com or manish@koffee-break.com
+ *
+ *  Copyright (C) 2007-2008 Wind River Systems, Inc.
+ *  Author/Maintainer: Jason Wessel, jason.wessel@windriver.com
+ *
+ *  This file is licensed under the terms of the GNU General Public License
+ *  version 2. This program is licensed "as is" without any warranty of any
+ *  kind, whether express or implied.
+ */
+
+#include <linux/ptrace.h>		/* for linux pt_regs struct */
+#include <linux/kgdb.h>
+#include <linux/kdebug.h>
+#include <linux/sched.h>
+#include <asm/inst.h>
+#include <asm/fpu.h>
+#include <asm/cacheflush.h>
+#include <asm/processor.h>
+#include <asm/sigcontext.h>
+
+static struct hard_trap_info {
+	unsigned char tt;	/* Trap type code for MIPS R3xxx and R4xxx */
+	unsigned char signo;	/* Signal that we map this trap into */
+} hard_trap_info[] = {
+	{ 6, SIGBUS },		/* instruction bus error */
+	{ 7, SIGBUS },		/* data bus error */
+	{ 9, SIGTRAP },		/* break */
+/*	{ 11, SIGILL },	*/	/* CPU unusable */
+	{ 12, SIGFPE },		/* overflow */
+	{ 13, SIGTRAP },	/* trap */
+	{ 14, SIGSEGV },	/* virtual instruction cache coherency */
+	{ 15, SIGFPE },		/* floating point exception */
+	{ 23, SIGSEGV },	/* watch */
+	{ 31, SIGSEGV },	/* virtual data cache coherency */
+	{ 0, 0}			/* Must be last */
+};
+
+void arch_kgdb_breakpoint(void)
+{
+	__asm__ __volatile__(
+		".globl breakinst\n\t"
+		".set\tnoreorder\n\t"
+		"nop\n"
+		"breakinst:\tbreak\n\t"
+		"nop\n\t"
+		".set\treorder");
+}
+
+static void kgdb_call_nmi_hook(void *ignored)
+{
+	kgdb_nmicallback(raw_smp_processor_id(), (void *)0);
+}
+
+void kgdb_roundup_cpus(unsigned long flags)
+{
+	local_irq_enable();
+	smp_call_function(kgdb_call_nmi_hook, NULL, NULL);
+	local_irq_disable();
+}
+
+static int compute_signal(int tt)
+{
+	struct hard_trap_info *ht;
+
+	for (ht = hard_trap_info; ht->tt && ht->signo; ht++)
+		if (ht->tt == tt)
+			return ht->signo;
+
+	return SIGHUP;		/* default for things we don't know about */
+}
+
+void pt_regs_to_gdb_regs(unsigned long *gdb_regs, struct pt_regs *regs)
+{
+	int reg;
+
+#if (KGDB_GDB_REG_SIZE == 32)
+	u32 *ptr = (u32 *)gdb_regs;
+#else
+	u64 *ptr = (u64 *)gdb_regs;
+#endif
+
+	for (reg = 0; reg < 32; reg++)
+		*(ptr++) = regs->regs[reg];
+
+	*(ptr++) = regs->cp0_status;
+	*(ptr++) = regs->lo;
+	*(ptr++) = regs->hi;
+	*(ptr++) = regs->cp0_badvaddr;
+	*(ptr++) = regs->cp0_cause;
+	*(ptr++) = regs->cp0_epc;
+
+	/* FP REGS */
+	if (!(current && (regs->cp0_status & ST0_CU1)))
+		return;
+
+	save_fp(current);
+	for (reg = 0; reg < 32; reg++)
+		*(ptr++) = current->thread.fpu.fpr[reg];
+}
+
+void gdb_regs_to_pt_regs(unsigned long *gdb_regs, struct pt_regs *regs)
+{
+	int reg;
+
+#if (KGDB_GDB_REG_SIZE == 32)
+	const u32 *ptr = (u32 *)gdb_regs;
+#else
+	const u64 *ptr = (u64 *)gdb_regs;
+#endif
+
+	for (reg = 0; reg < 32; reg++)
+		regs->regs[reg] = *(ptr++);
+
+	regs->cp0_status = *(ptr++);
+	regs->lo = *(ptr++);
+	regs->hi = *(ptr++);
+	regs->cp0_badvaddr = *(ptr++);
+	regs->cp0_cause = *(ptr++);
+	regs->cp0_epc = *(ptr++);
+
+	/* FP REGS from current */
+	if (!(current && (regs->cp0_status & ST0_CU1)))
+		return;
+
+	for (reg = 0; reg < 32; reg++)
+		current->thread.fpu.fpr[reg] = *(ptr++);
+	restore_fp(current);
+}
+
+/*
+ * Similar to regs_to_gdb_regs() except that process is sleeping and so
+ * we may not be able to get all the info.
+ */
+void sleeping_thread_to_gdb_regs(unsigned long *gdb_regs, struct task_struct *p)
+{
+	int reg;
+	struct thread_info *ti = task_thread_info(p);
+	unsigned long ksp = (unsigned long)ti + THREAD_SIZE - 32;
+	struct pt_regs *regs = (struct pt_regs *)ksp - 1;
+#if (KGDB_GDB_REG_SIZE == 32)
+	u32 *ptr = (u32 *)gdb_regs;
+#else
+	u64 *ptr = (u64 *)gdb_regs;
+#endif
+
+	for (reg = 0; reg < 16; reg++)
+		*(ptr++) = regs->regs[reg];
+
+	/* S0 - S7 */
+	for (reg = 16; reg < 24; reg++)
+		*(ptr++) = regs->regs[reg];
+
+	for (reg = 24; reg < 28; reg++)
+		*(ptr++) = 0;
+
+	/* GP, SP, FP, RA */
+	for (reg = 28; reg < 32; reg++)
+		*(ptr++) = regs->regs[reg];
+
+	*(ptr++) = regs->cp0_status;
+	*(ptr++) = regs->lo;
+	*(ptr++) = regs->hi;
+	*(ptr++) = regs->cp0_badvaddr;
+	*(ptr++) = regs->cp0_cause;
+	*(ptr++) = regs->cp0_epc;
+}
+
+/*
+ * Calls linux_debug_hook before the kernel dies. If KGDB is enabled,
+ * then try to fall into the debugger
+ */
+static int kgdb_mips_notify(struct notifier_block *self, unsigned long cmd,
+			    void *ptr)
+{
+	struct die_args *args = (struct die_args *)ptr;
+	struct pt_regs *regs = args->regs;
+	int trap = (regs->cp0_cause & 0x7c) >> 2;
+
+	if (fixup_exception(regs))
+		return NOTIFY_DONE;
+
+	/* Userpace events, ignore. */
+	if (user_mode(regs))
+		return NOTIFY_DONE;
+
+	if (atomic_read(&kgdb_active) != -1)
+		kgdb_nmicallback(smp_processor_id(), regs);
+
+	if (kgdb_handle_exception(trap, compute_signal(trap), 0, regs))
+		return NOTIFY_DONE;
+
+	if (atomic_read(&kgdb_setting_breakpoint))
+		if ((trap == 9) && (regs->cp0_epc == (unsigned long)breakinst))
+			regs->cp0_epc += 4;
+
+	/* In SMP mode, __flush_cache_all does IPI */
+	local_irq_enable();
+	__flush_cache_all();
+
+	return NOTIFY_STOP;
+}
+
+static struct notifier_block kgdb_notifier = {
+	.notifier_call = kgdb_mips_notify,
+};
+
+/*
+ * Handle the 's' and 'c' commands
+ */
+int kgdb_arch_handle_exception(int vector, int signo, int err_code,
+			       char *remcom_in_buffer, char *remcom_out_buffer,
+			       struct pt_regs *regs)
+{
+	char *ptr;
+	unsigned long address;
+	int cpu = smp_processor_id();
+
+	switch (remcom_in_buffer[0]) {
+	case 's':
+	case 'c':
+		/* handle the optional parameter */
+		ptr = &remcom_in_buffer[1];
+		if (kgdb_hex2long(&ptr, &address))
+			regs->cp0_epc = address;
+
+		atomic_set(&kgdb_cpu_doing_single_step, -1);
+		if (remcom_in_buffer[0] == 's')
+			if (kgdb_contthread)
+				atomic_set(&kgdb_cpu_doing_single_step, cpu);
+
+		return 0;
+	}
+
+	return -1;
+}
+
+struct kgdb_arch arch_kgdb_ops;
+
+/*
+ * We use kgdb_early_setup so that functions we need to call now don't
+ * cause trouble when called again later.
+ */
+int kgdb_arch_init(void)
+{
+	union mips_instruction insn = {
+		.r_format = {
+			.opcode = spec_op,
+			.func   = break_op,
+		}
+	};
+	memcpy(arch_kgdb_ops.gdb_bpt_instr, insn.byte, BREAK_INSTR_SIZE);
+
+	register_die_notifier(&kgdb_notifier);
+
+	return 0;
+}
+
+/*
+ *	kgdb_arch_exit - Perform any architecture specific uninitalization.
+ *
+ *	This function will handle the uninitalization of any architecture
+ *	specific callbacks, for dynamic registration and unregistration.
+ */
+void kgdb_arch_exit(void)
+{
+	unregister_die_notifier(&kgdb_notifier);
+}

arch/mips/kernel/traps.c

@@ -23,6 +23,8 @@
 #include <linux/bootmem.h>
 #include <linux/interrupt.h>
 #include <linux/ptrace.h>
+#include <linux/kgdb.h>
+#include <linux/kdebug.h>
 
 #include <asm/bootinfo.h>
 #include <asm/branch.h>
@@ -425,6 +427,10 @@ asmlinkage void do_be(struct pt_regs *regs)
 	printk(KERN_ALERT "%s bus error, epc == %0*lx, ra == %0*lx\n",
 	       data ? "Data" : "Instruction",
 	       field, regs->cp0_epc, field, regs->regs[31]);
+	if (notify_die(DIE_OOPS, "bus error", regs, SIGBUS, 0, 0)
+	    == NOTIFY_STOP)
+		return;
+
 	die_if_kernel("Oops", regs);
 	force_sig(SIGBUS, current);
 }
@@ -623,6 +629,9 @@ asmlinkage void do_fpe(struct pt_regs *regs, unsigned long fcr31)
 {
 	siginfo_t info;
 
+	if (notify_die(DIE_FP, "FP exception", regs, SIGFPE, 0, 0)
+	    == NOTIFY_STOP)
+		return;
 	die_if_kernel("FP exception in kernel code", regs);
 
 	if (fcr31 & FPU_CSR_UNI_X) {
@@ -682,6 +691,9 @@ static void do_trap_or_bp(struct pt_regs *regs, unsigned int code,
 	siginfo_t info;
 	char b[40];
 
+	if (notify_die(DIE_TRAP, str, regs, code, 0, 0) == NOTIFY_STOP)
+		return;
+
 	/*
 	 * A short test says that IRIX 5.3 sends SIGTRAP for all trap
 	 * insns, even for trap and break codes that indicate arithmetic
@@ -762,6 +774,10 @@ asmlinkage void do_ri(struct pt_regs *regs)
 	unsigned int opcode = 0;
 	int status = -1;
 
+	if (notify_die(DIE_RI, "RI Fault", regs, SIGSEGV, 0, 0)
+	    == NOTIFY_STOP)
+		return;
+
 	die_if_kernel("Reserved instruction in kernel code", regs);
 
 	if (unlikely(compute_return_epc(regs) < 0))
@@ -1537,6 +1553,11 @@ void __init trap_init(void)
 	extern char except_vec4;
 	unsigned long i;
 
+#if defined(CONFIG_KGDB)
+	if (kgdb_early_setup)
+		return;	/* Already done */
+#endif
+
 	if (cpu_has_veic || cpu_has_vint)
 		ebase = (unsigned long) alloc_bootmem_low_pages(0x200 + VECTORSPACING*64);
 	else