[patch 01/15] KGDB: Core files

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



CC: Amit Kale <[email protected]>
Here is a different version of KGDB than found currently in Andrew Morton's
tree that Amit Kale and I (and many others) have been working on for some time
now (and submitted a while back).  I appologize for not resubmitting this
sooner, but solving some of the issues George Anzinger and others dug up
turned out to take more time than I would have liked.

Having said that, this version of KGDB is designed so that as much code as
possible is done in a core file shared by all architectures, and with I/O (ie
8250 serial, custom uart, ethernet) being common when possible and modular.
The rough splitup of this is that 95% of the interaction with KGDB is done in
files common to all implementations with a small set of architecture specific
things (setjmp/longjmp, actually formatting registers for GDB, single
stepping).

Much of how to use this version (it is slightly different from George
Anzinger's version) is written up in DocBook and viewable that way.  I've
tried to do this in as clean a way as possible (notifier when possible for
example) so that it's as minimally instrusive as possible.  But, in order to
allow for KGDB to be used very early on (some folks argue this is critical,
some argue it's not) there's a few hooks so we can know if pidhash_init has
been run, or to try and have KGDB break in as soon as possible, if it can't
right then (such as on x86_64 where stacks aren't ready for us to use just
yet).

Comments?  Thanks!

---

 linux-2.6.13-rc3-trini/Documentation/DocBook/Makefile  |    2 
 linux-2.6.13-rc3-trini/Documentation/DocBook/kgdb.tmpl |  231 ++
 linux-2.6.13-rc3-trini/MAINTAINERS                     |    9 
 linux-2.6.13-rc3-trini/include/linux/kgdb.h            |  260 ++
 linux-2.6.13-rc3-trini/kernel/Makefile                 |    1 
 linux-2.6.13-rc3-trini/kernel/kgdb.c                   | 1835 +++++++++++++++++
 linux-2.6.13-rc3-trini/kernel/pid.c                    |   11 
 linux-2.6.13-rc3-trini/kernel/sched.c                  |    4 
 linux-2.6.13-rc3-trini/lib/Kconfig.debug               |   47 
 9 files changed, 2398 insertions(+), 2 deletions(-)

diff -puN /dev/null Documentation/DocBook/kgdb.tmpl
--- /dev/null	2005-07-25 10:57:32.312383000 -0700
+++ linux-2.6.13-rc3-trini/Documentation/DocBook/kgdb.tmpl	2005-07-29 13:18:19.000000000 -0700
@@ -0,0 +1,231 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE book PUBLIC "-//OASIS//DTD DocBook XML V4.1.2//EN"
+	"http://www.oasis-open.org/docbook/xml/4.1.2/docbookx.dtd"; []>
+
+<book id="kgdbInternals">
+ <bookinfo>
+  <title>KGDB Internals</title>
+
+  <authorgroup>
+   <author>
+    <firstname>Tom</firstname>
+    <surname>Rini</surname>
+    <affiliation>
+     <address>
+      <email>[email protected]</email>
+     </address>
+    </affiliation>
+   </author>
+  </authorgroup>
+
+  <authorgroup>
+   <author>
+    <firstname>Amit S.</firstname>
+    <surname>Kale</surname>
+    <affiliation>
+     <address>
+      <email>[email protected]</email>
+     </address>
+    </affiliation>
+   </author>
+  </authorgroup>
+
+  <copyright>
+   <year>2004-2005</year>
+   <holder>MontaVista Software, Inc.</holder>
+  </copyright>
+  <copyright>
+   <year>2004</year>
+   <holder>Amit S. Kale</holder>
+  </copyright>
+
+  <legalnotice>
+   <para>
+   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.
+   </para>
+
+  </legalnotice>
+ </bookinfo>
+
+<toc></toc>
+  <chapter id="Introduction">
+    <title>Introduction</title>
+    <para>
+    kgdb is a source level debugger for linux kernel. It is used along
+    with gdb to debug a linux kernel. Kernel developers can debug a kernel
+    similar to application programs with the use of kgdb. It makes it
+    possible to place breakpoints in kernel code, step through the code
+    and observe variables.
+    </para>
+    <para>
+    Two machines are required for using kgdb. One of these machines is a
+    development machine and the other is a test machine. The machines are
+    typically connected through a serial line, a null-modem cable which
+    connects their serial ports.  It is also possible however, to use an
+    ethernet connection between the machines.  The kernel to be debugged
+    runs on the test machine. gdb runs on the development machine. The
+    serial line or ethernet connection is used by gdb to communicate to
+    the kernel being debugged.
+    </para>
+  </chapter>
+  <chapter id="CompilingAKernel">
+    <title>Compiling a kernel</title>
+    <para>
+    To enable <symbol>CONFIG_KGDB</symbol>, look under the "Kernel debugging"
+    and then select "KGDB: kernel debugging with remote gdb".
+    </para>
+    <para>
+    The first choice for I/O is <symbol>CONFIG_KGDB_ONLY_MODULES</symbol>.
+    This means that you will only be able to use KGDB after loading a
+    kernel module that defines how you want to be able to talk with
+    KGDB.  There are two other choices (more on some architectures) that
+    can be enabled as modules later, if not picked here.
+    </para>
+    <para>The first of these is <symbol>CONFIG_KGDB_8250_NOMODULE</symbol>.
+    This has sub-options such as <symbol>CONFIG_KGDB_SIMPLE_SERIAL</symbol>
+    which toggles choosing the serial port by ttyS number or by specifying
+    a port and IRQ number.
+    </para>
+    <para>
+    The second of these choices on most systems for I/O is
+    <symbol>CONFIG_KGDB_ETH</symbol>. This requires that the machine to be
+    debugged has an ethernet card which supports the netpoll API, such as
+    the cards supported by <symbol>CONFIG_E100</symbol>.  There are no
+    sub-options for this, but a kernel command line option is required.
+    </para>
+  </chapter>
+  <chapter id="BootingTheKernel">
+    <title>Booting the kernel</title>
+    <para>
+    The Kernel command line option <constant>kgdbwait</constant> makes kgdb
+    wait for gdb connection during booting of a kernel.  If the
+    <symbol>CONFIG_KGDB_8250</symbol> driver is used (or if applicable,
+    another serial driver) this breakpoint will happen very early on, before
+    console output.  If you wish to change serial port information and you
+    have enabled both <symbol>CONFIG_KGDB_8250</symbol> and
+    <symbol>CONFIG_KGDB_SIMPLE_SERIAL</symbol> then you must pass the option
+    <constant>kgdb8250=portnumber,speed</constant> or on IA64
+    <constant>kgdb8250=portnumber,speed,irq,iomembase</constant> before
+    <constant>kgdbwait</constant>.  The values for portnumber are 0-3 and
+    correspond to ttyS0 to ttyS3 respectively.  The supported speeds are
+    <constant>9600</constant>, <constant>19200</constant>,
+    <constant>38400</constant>, <constant>57600</constant>, and
+    <constant>115200</constant>.
+    </para>
+    <para>
+    To have KGDB stop the kernel and wait, with the compiled values for the
+    serial driver, pass in: <constant>kgdbwait</constant>.
+    </para>
+    <para>
+    To specify the values of the serial port at boot:
+    <constant>kgdb8250=0,115200</constant>.
+    On IA64 this could also be:
+    <constant>kgdb8250=1,115200,74,0xc0000000ff5e0000</constant>
+    And to have KGDB also stop the kernel and wait for GDB to connect, pass in
+    <constant>kgdbwait</constant> after this arguement.
+    </para>
+    <para>
+    To configure the <symbol>CONFIG_KGDB_ETH</symbol> driver, pass in
+    <constant>kgdboe=[src-port]@&lt;src-ip&gt;/[dev],[tgt-port]@&lt;tgt-ip&gt;/[tgt-macaddr]</constant>
+    where:
+    <itemizedlist>
+      <listitem><para>src-port (optional): source for UDP packets (defaults to <constant>6443</constant>)</para></listitem>
+      <listitem><para>src-ip: source IP to use (interface address)</para></listitem>
+      <listitem><para>dev (optional): network interface (<constant>eth0</constant>)</para></listitem>
+      <listitem><para>tgt-port (optional): port GDB will use (defaults to <constant>6442</constant>)</para></listitem>
+      <listitem><para>tgt-ip: IP address GDB will be connecting from</para></listitem>
+      <listitem><para>tgt-macaddr (optional): ethernet MAC address for logging agent (default is broadcast)</para></listitem>
+    </itemizedlist>
+    </para>
+  </chapter>
+  <chapter id="ConnectingGDB">
+  <title>Connecting gdb</title>
+    <para>
+    If you have used any of the methods to have KGDB stop and create
+    an initial breakpoint described in the previous chapter, kgdb prints
+    the message "Waiting for connection from remote gdb..." on the console
+    and waits for connection from gdb. At this point you connect gdb to kgdb.
+    </para>
+    <para>
+    Example (serial):
+    </para>
+    <programlisting>
+    % gdb ./vmlinux
+    (gdb) set remotebaud 115200
+    (gdb) target remote /dev/ttyS0
+    </programlisting>
+    <para>
+    Example (ethernet):
+    </para>
+    <programlisting>
+    % gdb ./vmlinux
+    (gdb) target remote udp:192.168.2.2:6443
+    </programlisting>
+    <para>
+    Once connected, you can debug a kernel the way you would debug an
+    application program.
+    </para>
+  </chapter>
+  <chapter id="CommonBackEndReq">
+    <title>The common backend (required)</title>
+      <para>
+      There are a few flags which must be set on every architecture in
+      their &lt;asm/kgdb.h&gt; file.  These are:
+      <itemizedlist>
+        <listitem>
+	  <para>
+	  NUMREGBYTES: The size in bytes of all of the registers, so
+	  that we can ensure they will all fit into a packet.
+	  </para>
+	  <para>
+	  BUFMAX: The size in bytes of the buffer GDB will read into.
+	  This must be larger than NUMREGBYTES.
+	  </para>
+	  <para>
+	  CACHE_FLUSH_IS_SAFE: Set to one if it always safe to call
+	  flush_cache_range or flush_icache_range.  On some architectures,
+	  these functions may not be safe to call on SMP since we keep other
+	  CPUs in a holding pattern.
+	  </para>
+	</listitem>
+	<listitem>
+	  <para>
+	  CHECK_EXCEPTION_STACK(): It is possible for the kernel to be
+	  trying to use KGDB before our exception stacks are setup.  In
+	  this case, we must not let the KGDB happen yet.  If we do not
+	  have to worry about this, it should be set to one.  If we do
+	  have to worry about this, it should be set to a suitable
+	  check.
+	  </para>
+	</listitem>
+      </itemizedlist>
+      </para>
+      <para>
+      There are also the following functions for the common backend,
+      found in kernel/kgdb.c that must be supplied by the
+      architecture-specific backend.  No weak version of these is provided.
+      </para>
+!Iinclude/linux/kgdb.h
+  </chapter>
+  <chapter id="CommonBackEndOpt">
+    <title>The common backend (optional)</title>
+      <para>
+      These functions are part of the common backend, found in kernel/kgdb.c
+      and are optionally implemented.  Some functions (with _hw_ in the name)
+      end up being required on arches which use hardware breakpoints.
+      </para>
+!Ikernel/kgdb.c
+  </chapter>
+  <chapter id="DriverSpecificFunctions">
+    <title>Driver-Specific Functions</title>
+      <para>
+      Some of the I/O drivers have additional functions that can be
+      called, that are specific to the driver.  Calls from other places
+      to these functions must be wrapped in #ifdefs for the driver in
+      question.
+      </para>
+!Idrivers/serial/kgdb_8250.c
+   </chapter>
+</book>
diff -puN Documentation/DocBook/Makefile~core-lite Documentation/DocBook/Makefile
--- linux-2.6.13-rc3/Documentation/DocBook/Makefile~core-lite	2005-07-29 11:13:42.000000000 -0700
+++ linux-2.6.13-rc3-trini/Documentation/DocBook/Makefile	2005-07-29 11:13:42.000000000 -0700
@@ -10,7 +10,7 @@ DOCBOOKS := wanbook.xml z8530book.xml mc
 	    kernel-hacking.xml kernel-locking.xml deviceiobook.xml \
 	    procfs-guide.xml writing_usb_driver.xml \
 	    sis900.xml kernel-api.xml journal-api.xml lsm.xml usb.xml \
-	    gadget.xml libata.xml mtdnand.xml librs.xml
+	    gadget.xml libata.xml mtdnand.xml librs.xml kgdb.xml
 
 ###
 # The build process is as follows (targets):
diff -puN /dev/null include/linux/kgdb.h
--- /dev/null	2005-07-25 10:57:32.312383000 -0700
+++ linux-2.6.13-rc3-trini/include/linux/kgdb.h	2005-07-29 13:23:59.000000000 -0700
@@ -0,0 +1,260 @@
+/*
+ * include/linux/kgdb.h
+ *
+ * This provides the hooks and functions that KGDB needs to share between
+ * the core, I/O and arch-specific portions.
+ *
+ * Author: Amit Kale <[email protected]> and
+ *         Tom Rini <[email protected]>
+ *
+ * 2001-2004 (c) Amit S. Kale and 2003-2004 (c) MontaVista Software, Inc.
+ * 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.
+ */
+#ifdef __KERNEL__
+#ifndef _KGDB_H_
+#define _KGDB_H_
+
+#include <asm/atomic.h>
+
+#ifdef CONFIG_KGDB
+#include <asm/kgdb.h>
+#include <linux/serial_8250.h>
+#include <linux/linkage.h>
+
+struct tasklet_struct;
+struct pt_regs;
+struct task_struct;
+struct uart_port;
+
+
+/* To enter the debugger explicitly. */
+extern void breakpoint(void);
+extern int kgdb_connected;
+extern int kgdb_may_fault;
+extern struct tasklet_struct kgdb_tasklet_breakpoint;
+
+extern atomic_t kgdb_setting_breakpoint;
+extern atomic_t cpu_doing_single_step;
+
+extern struct task_struct *kgdb_usethread, *kgdb_contthread;
+
+enum kgdb_bptype {
+	bp_breakpoint = '0',
+	bp_hardware_breakpoint,
+	bp_write_watchpoint,
+	bp_read_watchpoint,
+	bp_access_watchpoint
+};
+
+enum kgdb_bpstate {
+	bp_disabled,
+	bp_enabled
+};
+
+struct kgdb_bkpt {
+	unsigned long bpt_addr;
+	unsigned char saved_instr[BREAK_INSTR_SIZE];
+	enum kgdb_bptype type;
+	enum kgdb_bpstate state;
+};
+
+/* The maximum number of KGDB I/O modules that can be loaded */
+#define MAX_KGDB_IO_HANDLERS 3
+
+#ifndef MAX_BREAKPOINTS
+#define MAX_BREAKPOINTS		16
+#endif
+
+#define KGDB_HW_BREAKPOINT	1
+
+/* Required functions. */
+/**
+ *	regs_to_gdb_regs - Convert ptrace regs to GDB regs
+ *	@gdb_regs: A pointer to hold the registers in the order GDB wants.
+ *	@regs: The &struct pt_regs of the current process.
+ *
+ *	Convert the pt_regs in @regs into the format for registers that
+ *	GDB expects, stored in @gdb_regs.
+ */
+extern void regs_to_gdb_regs(unsigned long *gdb_regs, struct pt_regs *regs);
+
+/**
+ *	sleeping_regs_to_gdb_regs - Convert ptrace regs to GDB regs
+ *	@gdb_regs: A pointer to hold the registers in the order GDB wants.
+ *	@p: The &struct task_struct of the desired process.
+ *
+ *	Convert the register values of the sleeping process in @p to
+ *	the format that GDB expects.
+ *	This function is called when kgdb does not have access to the
+ *	&struct pt_regs and therefore it should fill the gdb registers
+ *	@gdb_regs with what has	been saved in &struct thread_struct
+ *	thread field during switch_to.
+ */
+extern void sleeping_thread_to_gdb_regs(unsigned long *gdb_regs,
+					struct task_struct *p);
+
+/**
+ *	gdb_regs_to_regs - Convert GDB regs to ptrace regs.
+ *	@gdb_regs: A pointer to hold the registers we've recieved from GDB.
+ *	@regs: A pointer to a &struct pt_regs to hold these values in.
+ *
+ *	Convert the GDB regs in @gdb_regs into the pt_regs, and store them
+ *	in @regs.
+ */
+extern void gdb_regs_to_regs(unsigned long *gdb_regs, struct pt_regs *regs);
+
+/**
+ *	kgdb_arch_handle_exception - Handle architecture specific GDB packets.
+ *	@vector: The error vector of the exception that happened.
+ *	@signo: The signal number of the exception that happened.
+ *	@err_code: The error code of the exception that happened.
+ *	@remcom_in_buffer: The buffer of the packet we have read.
+ *	@remcom_out_buffer: The buffer, of %BUFMAX to write a packet into.
+ *	@regs: The &struct pt_regs of the current process.
+ *
+ *	This function MUST handle the 'c' and 's' command packets,
+ *	as well packets to set / remove a hardware breakpoint, if used.
+ *	If there are additional packets which the hardware needs to handle,
+ *	they are handled here.  The code should return -1 if it wants to
+ *	process more packets, and a %0 or %1 if it wants to exit from the
+ *	kgdb hook.
+ */
+extern int kgdb_arch_handle_exception(int vector, int signo, int err_code,
+				      char *remcom_in_buffer,
+				      char *remcom_out_buffer,
+				      struct pt_regs *regs);
+
+#ifndef JMP_REGS_ALIGNMENT
+#define JMP_REGS_ALIGNMENT
+#endif
+
+extern unsigned long kgdb_fault_jmp_regs[];
+
+/**
+ *	kgdb_fault_setjmp - Store state in case we fault.
+ *	@curr_context: An array to store state into.
+ *
+ *	Certain functions may try and access memory, and in doing so may
+ *	cause a fault.  When this happens, we trap it, restore state to
+ *	this call, and let ourself know that something bad has happened.
+ */
+extern asmlinkage int kgdb_fault_setjmp(unsigned long *curr_context);
+
+/**
+ *	kgdb_fault_longjmp - Restore state when we have faulted.
+ *	@curr_context: The previously stored state.
+ *
+ *	When something bad does happen, this function is called to
+ *	restore the known good state, and set the return value to 1, so
+ *	we know something bad happened.
+ */
+extern asmlinkage void kgdb_fault_longjmp(unsigned long *curr_context);
+
+/* Optional functions. */
+extern int kgdb_arch_init(void);
+extern void kgdb_disable_hw_debug(struct pt_regs *regs);
+extern void kgdb_post_master_code(struct pt_regs *regs, int e_vector,
+				  int err_code);
+extern void kgdb_roundup_cpus(unsigned long flags);
+extern int kgdb_set_hw_break(unsigned long addr);
+extern int kgdb_remove_hw_break(unsigned long addr);
+extern void kgdb_remove_all_hw_break(void);
+extern void kgdb_correct_hw_break(void);
+extern void kgdb_shadowinfo(struct pt_regs *regs, char *buffer,
+			    unsigned threadid);
+extern struct task_struct *kgdb_get_shadow_thread(struct pt_regs *regs,
+						  int threadid);
+extern struct pt_regs *kgdb_shadow_regs(struct pt_regs *regs, int threadid);
+
+/**
+ * struct kgdb_arch - Desribe architecture specific values.
+ * @gdb_bpt_instr: The instruction to trigger a breakpoint.
+ * @flags: Flags for the breakpoint, currently just %KGDB_HW_BREAKPOINT.
+ * @shadowth: A value of %1 indicates we shadow information on processes.
+ * @set_breakpoint: Allow an architecture to specify how to set a software
+ * breakpoint.
+ * @remove_breakpoint: Allow an architecture to specify how to remove a
+ * software breakpoint.
+ * @set_hw_breakpoint: Allow an architecture to specify how to set a hardware
+ * breakpoint.
+ * @remove_hw_breakpoint: Allow an architecture to specify how to remove a
+ * hardware breakpoint.
+ *
+ * The @shadowth flag is an option to shadow information not retrievable by
+ * gdb otherwise.  This is deprecated in favor of a binutils which supports
+ * CFI macros.
+ */
+struct kgdb_arch {
+	unsigned char gdb_bpt_instr[BREAK_INSTR_SIZE];
+	unsigned long flags;
+	unsigned shadowth;
+	int (*set_breakpoint) (unsigned long, char *);
+	int (*remove_breakpoint)(unsigned long, char *);
+	int (*set_hw_breakpoint)(unsigned long, int, enum kgdb_bptype);
+	int (*remove_hw_breakpoint)(unsigned long, int, enum kgdb_bptype);
+};
+
+/* Thread reference */
+typedef unsigned char threadref[8];
+
+/**
+ * struct kgdb_io - Desribe the interface for an I/O driver to talk with KGDB.
+ * @read_char: Pointer to a function that will return one char.
+ * @write_char: Pointer to a function that will write one char.
+ * @flush: Pointer to a function that will flush any pending writes.
+ * @init: Pointer to a function that will initialize the device.
+ * @late_init: Pointer to a function that will do any setup that has
+ * other dependencies.
+ * @pre_exception: Pointer to a function that will do any prep work for
+ * the I/O driver.
+ * @post_exception: Pointer to a function that will do any cleanup work
+ * for the I/O driver.
+ *
+ * The @init and @late_init function pointers allow for an I/O driver
+ * such as a serial driver to fully initialize the port with @init and
+ * be called very early, yet safely call request_irq() later in the boot
+ * sequence.
+ *
+ * @init is allowed to return a non-0 return value to indicate failure.
+ * If this is called early on, then KGDB will try again when it would call
+ * @late_init.  If it has failed later in boot as well, the user will be
+ * notified.
+ */
+struct kgdb_io {
+	int (*read_char) (void);
+	void (*write_char) (int);
+	void (*flush) (void);
+	int (*init) (void);
+	void (*late_init) (void);
+	void (*pre_exception) (void);
+	void (*post_exception) (void);
+};
+
+extern struct kgdb_io kgdb_io_ops;
+extern struct kgdb_arch arch_kgdb_ops;
+extern int kgdb_initialized;
+
+extern int kgdb_register_io_module(struct kgdb_io *local_kgdb_io_ops);
+extern void kgdb_unregister_io_module(struct kgdb_io *local_kgdb_io_ops);
+
+extern void kgdb8250_add_port(int i, struct uart_port *serial_req);
+extern void kgdb8250_add_platform_port(int i, struct plat_serial8250_port *serial_req);
+
+extern int kgdb_hex2long(char **ptr, long *long_val);
+extern char *kgdb_mem2hex(char *mem, char *buf, int count);
+extern char *kgdb_hex2mem(char *buf, char *mem, int count);
+extern int kgdb_get_mem(char *addr, unsigned char *buf, int count);
+extern int kgdb_set_mem(char *addr, unsigned char *buf, int count);
+extern int kgdb_handle_exception(int ex_vector, int signo, int err_code,
+				struct pt_regs *regs);
+extern void kgdb_nmihook(int cpu, void *regs);
+extern int debugger_step;
+extern atomic_t debugger_active;
+#else
+/* Stubs for when KGDB is not set. */
+static const atomic_t debugger_active = ATOMIC_INIT(0);
+#endif				/* CONFIG_KGDB */
+#endif				/* _KGDB_H_ */
+#endif				/* __KERNEL__ */
diff -puN /dev/null kernel/kgdb.c
--- /dev/null	2005-07-25 10:57:32.312383000 -0700
+++ linux-2.6.13-rc3-trini/kernel/kgdb.c	2005-07-29 11:13:42.000000000 -0700
@@ -0,0 +1,1835 @@
+/*
+ * 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, 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.
+ */
+
+/*
+ * Copyright (C) 2000-2001 VERITAS Software Corporation.
+ * Copyright (C) 2002-2004 Timesys Corporation
+ * Copyright (C) 2003-2004 Amit S. Kale <[email protected]>
+ * Copyright (C) 2004 Pavel Machek <[email protected]>
+ * Copyright (C) 2004-2005 Tom Rini <[email protected]>
+ * Copyright (C) 2004 LinSysSoft Technologies Pvt. Ltd.
+ * Copyright (C) 2005 Wind River System, Inc.
+ *
+ * Added ability to setup kgdb I/O at runtime with a kernel module.
+ * Added ability to allow a single in-kernel I/O routine and override
+ * it with a kernel module.
+ * IE: rs232 for early kernel debug and kgdboe after the boot is completed
+ * - Jason Wessel ( [email protected] )
+ *
+ * Restructured KGDB for 2.6 kernels.
+ * thread support, support for multiple processors,support for ia-32(x86)
+ * hardware debugging, Console support, handling nmi watchdog
+ * - Amit S. Kale ( [email protected] )
+ *
+ * Several enhancements by George Anzinger <[email protected]>
+ * Generic KGDB Support
+ * Implemented by Anurekh Saxena ([email protected])
+ *
+ * Contributor:	Lake Stevens Instrument Division
+ * Written by:	Glenn Engel
+ *
+ * Modified for 386 by Jim Kingdon, Cygnus Support.
+ * Origianl kgdb, compatibility with 2.1.xx kernel by
+ * David Grothe <[email protected]>
+ * Integrated into 2.2.5 kernel by Tigran Aivazian <[email protected]>
+ */
+
+#include <linux/string.h>
+#include <linux/kernel.h>
+#include <linux/interrupt.h>
+#include <linux/sched.h>
+#include <linux/smp.h>
+#include <linux/spinlock.h>
+#include <linux/delay.h>
+#include <linux/mm.h>
+#include <linux/threads.h>
+#include <asm/system.h>
+#include <asm/ptrace.h>
+#include <asm/uaccess.h>
+#include <linux/kgdb.h>
+#include <asm/atomic.h>
+#include <linux/notifier.h>
+#include <linux/module.h>
+#include <asm/cacheflush.h>
+#include <linux/init.h>
+#include <linux/sysrq.h>
+#include <linux/console.h>
+#include <asm/byteorder.h>
+
+extern int pid_max;
+extern int pidhash_init_done;
+
+/* How many times to count all of the waiting CPUs */
+#define ROUNDUP_WAIT		64000
+#define BUF_THREAD_ID_SIZE 16
+
+/*
+ * kgdb_initialized with a value of 1 indicates that kgdb is setup and is
+ * all ready to serve breakpoints and other kernel exceptions.  A value of
+ * -1 indicates that we have tried to initialize early, and need to try
+ * again later.
+ */
+int kgdb_initialized = 0;
+/* Is a host GDB connected to us? */
+int kgdb_connected;
+/* Could we be about to try and access a bad memory location? If so we
+ * also need to flag this has happend. */
+int kgdb_may_fault;
+/* All the KGDB handlers are installed */
+int kgdb_from_module_registered = 0;
+
+/* We provide a kgdb_io_ops structure that may be overriden. */
+struct kgdb_io __attribute__ ((weak)) kgdb_io_ops;
+
+static struct kgdb_io kgdb_io_ops_prev[MAX_KGDB_IO_HANDLERS];
+static int kgdb_io_handler_cnt = 0;
+
+/* Export the following symbols for use with kernel modules */
+EXPORT_SYMBOL(kgdb_io_ops);
+EXPORT_SYMBOL(kgdb_tasklet_breakpoint);
+EXPORT_SYMBOL(kgdb_connected);
+EXPORT_SYMBOL(kgdb_register_io_module);
+EXPORT_SYMBOL(kgdb_unregister_io_module);
+EXPORT_SYMBOL(debugger_active);
+
+/*
+ * Holds information about breakpoints in a kernel. These breakpoints are
+ * added and removed by gdb.
+ */
+struct kgdb_bkpt kgdb_break[MAX_BREAKPOINTS];
+
+struct kgdb_arch *kgdb_ops = &arch_kgdb_ops;
+
+static const char hexchars[] = "0123456789abcdef";
+
+static spinlock_t slavecpulocks[NR_CPUS];
+static volatile int procindebug[NR_CPUS];
+atomic_t kgdb_setting_breakpoint;
+struct task_struct *kgdb_usethread, *kgdb_contthread;
+
+int debugger_step;
+atomic_t debugger_active;
+
+/* Our I/O buffers. */
+static char remcom_in_buffer[BUFMAX];
+static char remcom_out_buffer[BUFMAX];
+/* Storage for the registers, in GDB format. */
+static unsigned long gdb_regs[(NUMREGBYTES + sizeof(unsigned long) - 1) /
+			      sizeof(unsigned long)];
+/* Storage of registers for handling a fault. */
+unsigned long kgdb_fault_jmp_regs[NUMCRITREGBYTES / sizeof(unsigned long)]
+ JMP_REGS_ALIGNMENT;
+
+struct debuggerinfo_struct {
+	void *debuggerinfo;
+	struct task_struct *task;
+} kgdb_info[NR_CPUS];
+
+/* to keep track of the CPU which is doing the single stepping*/
+atomic_t cpu_doing_single_step = ATOMIC_INIT(-1);
+
+/**
+ *	kgdb_arch_init - Perform any architecture specific initalization.
+ *
+ *	RETURN:
+ *	The return value is ignored.
+ *
+ *	This function will handle the initalization of any architecture
+ *	specific hooks.
+ */
+int __attribute__ ((weak))
+    kgdb_arch_init(void)
+{
+	return 0;
+}
+
+/**
+ *	kgdb_disable_hw_debug - Disable hardware debugging while we in kgdb.
+ *	@regs: Current &struct pt_regs.
+ *
+ *	This function will be called if the particular architecture must
+ *	disable hardware debugging while it is processing gdb packets or
+ *	handling exception.
+ */
+void __attribute__ ((weak))
+    kgdb_disable_hw_debug(struct pt_regs *regs)
+{
+}
+
+/**
+ *	kgdb_set_hw_break - Set a hardware breakpoint at @addr.
+ *	@addr: The address to set a hardware breakpoint at.
+ */
+int __attribute__ ((weak))
+    kgdb_set_hw_break(unsigned long addr)
+{
+	return 0;
+}
+
+/**
+ *	kgdb_remove_hw_break - Remove a hardware breakpoint at @addr.
+ *	@addr: The address to remove a hardware breakpoint from.
+ */
+int __attribute__ ((weak))
+    kgdb_remove_hw_break(unsigned long addr)
+{
+	return 0;
+}
+
+/**
+ *	kgdb_remove_all_hw_break - Clear all hardware breakpoints.
+ */
+void __attribute__ ((weak))
+    kgdb_remove_all_hw_break(void)
+{
+}
+
+/**
+ *	kgdb_correct_hw_break - Correct hardware breakpoints.
+ *
+ *	A hook to allow for changes to the hardware breakpoint, called
+ *	after a single step (s) or continue (c) packet, and once we're about
+ *	to let the kernel continue running.
+ *
+ *	This is used to set the hardware breakpoint registers for all the
+ *	slave cpus on an SMP configuration. This must be called after any
+ *	changes are made to the hardware breakpoints (such as by a single
+ *	step (s) or continue (c) packet. This is only required on
+ *	architectures that support SMP and every processor has its own set
+ *	of breakpoint registers.
+ */
+void __attribute__ ((weak))
+    kgdb_correct_hw_break(void)
+{
+}
+
+/**
+ *	kgdb_post_master_code - Save error vector/code numbers.
+ *	@regs: Original pt_regs.
+ *	@e_vector: Original error vector.
+ *	@err_code: Original error code.
+ *
+ *	This is needed on architectures which support SMP and KGDB.
+ *	This function is called after all the slave cpus have been put
+ *	to a know spin state and the master CPU has control over KGDB.
+ */
+
+void __attribute__ ((weak))
+    kgdb_post_master_code(struct pt_regs *regs, int e_vector, int err_code)
+{
+}
+
+/**
+ * 	kgdb_roundup_cpus - Get other CPUs into a holding pattern
+ * 	@flags: Current IRQ state
+ *
+ * 	On SMP systems, we need to get the attention of the other CPUs
+ * 	and get them be in a known state.  This should do what is needed
+ * 	to get the other CPUs to call kgdb_wait(). Note that on some arches,
+ *	the NMI approach is not used for rounding up all the CPUs. For example,
+ *	in case of MIPS, smp_call_function() is used to roundup CPUs. In
+ *	this case, we have to make sure that interrupts are enabled before
+ *	calling smp_call_function(). The argument to this function is
+ *	the flags that will be used when restoring the interrupts. There is
+ *	local_irq_save() call before kgdb_roundup_cpus().
+ */
+void __attribute__ ((weak))
+    kgdb_roundup_cpus(unsigned long flags)
+{
+}
+
+/**
+ *	kgdb_shadowinfo - Get shadowed information on @threadid.
+ *	@regs: The &struct pt_regs of the current process.
+ *	@buffer: A buffer of %BUFMAX size.
+ *	@threadid: The thread id of the shadowed process to get information on.
+ */
+void __attribute__ ((weak))
+    kgdb_shadowinfo(struct pt_regs *regs, char *buffer, unsigned threadid)
+{
+}
+
+/**
+ *	kgdb_get_shadow_thread - Get the shadowed &task_struct of @threadid.
+ *	@regs: The &struct pt_regs of the current thread.
+ *	@threadid: The thread id of the shadowed process to get information on.
+ *
+ *	RETURN:
+ *	This returns a pointer to the &struct task_struct of the shadowed
+ *	thread, @threadid.
+ */
+struct task_struct __attribute__ ((weak))
+    * kgdb_get_shadow_thread(struct pt_regs *regs, int threadid)
+{
+	return NULL;
+}
+
+/**
+ *	kgdb_shadow_regs - Return the shadowed registers of @threadid.
+ *	@regs: The &struct pt_regs of the current thread.
+ *	@threadid: The thread id we want the &struct pt_regs for.
+ *
+ *	RETURN:
+ *	The a pointer to the &struct pt_regs of the shadowed thread @threadid.
+ */
+struct pt_regs __attribute__ ((weak))
+    * kgdb_shadow_regs(struct pt_regs *regs, int threadid)
+{
+	return NULL;
+}
+
+static int hex(char ch)
+{
+	if ((ch >= 'a') && (ch <= 'f'))
+		return (ch - 'a' + 10);
+	if ((ch >= '0') && (ch <= '9'))
+		return (ch - '0');
+	if ((ch >= 'A') && (ch <= 'F'))
+		return (ch - 'A' + 10);
+	return (-1);
+}
+
+/* scan for the sequence $<data>#<checksum>	*/
+static void get_packet(char *buffer)
+{
+	unsigned char checksum;
+	unsigned char xmitcsum;
+	int count;
+	char ch;
+	if (!kgdb_io_ops.read_char)
+		return;
+	do {
+		/* Spin and wait around for the start character, ignore all
+		 * other characters */
+		while ((ch = (kgdb_io_ops.read_char())) != '$') ;
+		kgdb_connected = 1;
+		checksum = 0;
+		xmitcsum = -1;
+
+		count = 0;
+
+		/* now, read until a # or end of buffer is found */
+		while (count < (BUFMAX - 1)) {
+			ch = kgdb_io_ops.read_char();
+			if (ch == '#')
+				break;
+			checksum = checksum + ch;
+			buffer[count] = ch;
+			count = count + 1;
+		}
+		buffer[count] = 0;
+
+		if (ch == '#') {
+			xmitcsum = hex(kgdb_io_ops.read_char()) << 4;
+			xmitcsum += hex(kgdb_io_ops.read_char());
+
+			if (checksum != xmitcsum)
+				/* failed checksum */
+				kgdb_io_ops.write_char('-');
+			else
+				/* successful transfer */
+				kgdb_io_ops.write_char('+');
+			if (kgdb_io_ops.flush)
+				kgdb_io_ops.flush();
+		}
+	} while (checksum != xmitcsum);
+}
+
+/*
+ * Send the packet in buffer.
+ * Check for gdb connection if asked for.
+ */
+static void put_packet(char *buffer)
+{
+	unsigned char checksum;
+	int count;
+	char ch;
+
+	if (!kgdb_io_ops.write_char)
+		return;
+	/* $<packet info>#<checksum>. */
+	while (1) {
+		kgdb_io_ops.write_char('$');
+		checksum = 0;
+		count = 0;
+
+		while ((ch = buffer[count])) {
+			kgdb_io_ops.write_char(ch);
+			checksum += ch;
+			count++;
+		}
+
+		kgdb_io_ops.write_char('#');
+		kgdb_io_ops.write_char(hexchars[checksum >> 4]);
+		kgdb_io_ops.write_char(hexchars[checksum % 16]);
+		if (kgdb_io_ops.flush)
+			kgdb_io_ops.flush();
+
+		/* Now see what we get in reply. */
+		ch = kgdb_io_ops.read_char();
+
+		if (ch == 3)
+			ch = kgdb_io_ops.read_char();
+
+		/* If we get an ACK, we are done. */
+		if (ch == '+')
+			return;
+
+		/* If we get the start of another packet, this means
+		 * that GDB is attempting to reconnect.  We will NAK
+		 * the packet being sent, and stop trying to send this
+		 * packet. */
+		if (ch == '$') {
+			kgdb_io_ops.write_char('-');
+			if (kgdb_io_ops.flush)
+				kgdb_io_ops.flush();
+			return;
+		}
+	}
+}
+
+/*
+ * convert the memory pointed to by mem into hex, placing result in buf
+ * return a pointer to the last char put in buf (null). May return an error.
+ */
+char *kgdb_mem2hex(char *mem, char *buf, int count)
+{
+	kgdb_may_fault = 1;
+	if ((kgdb_fault_setjmp(kgdb_fault_jmp_regs)) != 0) {
+		kgdb_may_fault = 0;
+		return ERR_PTR(-EINVAL);
+	}
+	/* Accessing some registers in a single load instruction is
+	 * required to avoid bad side effects for some I/O registers.
+	 */
+	if ((count == 2) && (((long)mem & 1) == 0)) {
+		unsigned short tmp_s = *(unsigned short *)mem;
+		mem += 2;
+#ifdef __BIG_ENDIAN
+		*buf++ = hexchars[(tmp_s >> 12) & 0xf];
+		*buf++ = hexchars[(tmp_s >> 8) & 0xf];
+		*buf++ = hexchars[(tmp_s >> 4) & 0xf];
+		*buf++ = hexchars[tmp_s & 0xf];
+#else
+		*buf++ = hexchars[(tmp_s >> 4) & 0xf];
+		*buf++ = hexchars[tmp_s & 0xf];
+		*buf++ = hexchars[(tmp_s >> 12) & 0xf];
+		*buf++ = hexchars[(tmp_s >> 8) & 0xf];
+#endif
+	} else if ((count == 4) && (((long)mem & 3) == 0)) {
+		unsigned long tmp_l = *(unsigned int *)mem;
+		mem += 4;
+#ifdef __BIG_ENDIAN
+		*buf++ = hexchars[(tmp_l >> 28) & 0xf];
+		*buf++ = hexchars[(tmp_l >> 24) & 0xf];
+		*buf++ = hexchars[(tmp_l >> 20) & 0xf];
+		*buf++ = hexchars[(tmp_l >> 16) & 0xf];
+		*buf++ = hexchars[(tmp_l >> 12) & 0xf];
+		*buf++ = hexchars[(tmp_l >> 8) & 0xf];
+		*buf++ = hexchars[(tmp_l >> 4) & 0xf];
+		*buf++ = hexchars[tmp_l & 0xf];
+#else
+		*buf++ = hexchars[(tmp_l >> 4) & 0xf];
+		*buf++ = hexchars[tmp_l & 0xf];
+		*buf++ = hexchars[(tmp_l >> 12) & 0xf];
+		*buf++ = hexchars[(tmp_l >> 8) & 0xf];
+		*buf++ = hexchars[(tmp_l >> 20) & 0xf];
+		*buf++ = hexchars[(tmp_l >> 16) & 0xf];
+		*buf++ = hexchars[(tmp_l >> 28) & 0xf];
+		*buf++ = hexchars[(tmp_l >> 24) & 0xf];
+#endif
+#ifdef CONFIG_64BIT
+	} else if ((count == 8) && (((long)mem & 7) == 0)) {
+		unsigned long long tmp_ll = *(unsigned long long *)mem;
+		mem += 8;
+#ifdef __BIG_ENDIAN
+		*buf++ = hexchars[(tmp_ll >> 60) & 0xf];
+		*buf++ = hexchars[(tmp_ll >> 56) & 0xf];
+		*buf++ = hexchars[(tmp_ll >> 52) & 0xf];
+		*buf++ = hexchars[(tmp_ll >> 48) & 0xf];
+		*buf++ = hexchars[(tmp_ll >> 44) & 0xf];
+		*buf++ = hexchars[(tmp_ll >> 40) & 0xf];
+		*buf++ = hexchars[(tmp_ll >> 36) & 0xf];
+		*buf++ = hexchars[(tmp_ll >> 32) & 0xf];
+		*buf++ = hexchars[(tmp_ll >> 28) & 0xf];
+		*buf++ = hexchars[(tmp_ll >> 24) & 0xf];
+		*buf++ = hexchars[(tmp_ll >> 20) & 0xf];
+		*buf++ = hexchars[(tmp_ll >> 16) & 0xf];
+		*buf++ = hexchars[(tmp_ll >> 12) & 0xf];
+		*buf++ = hexchars[(tmp_ll >> 8) & 0xf];
+		*buf++ = hexchars[(tmp_ll >> 4) & 0xf];
+		*buf++ = hexchars[tmp_ll & 0xf];
+#else
+		*buf++ = hexchars[(tmp_ll >> 4) & 0xf];
+		*buf++ = hexchars[tmp_ll & 0xf];
+		*buf++ = hexchars[(tmp_ll >> 12) & 0xf];
+		*buf++ = hexchars[(tmp_ll >> 8) & 0xf];
+		*buf++ = hexchars[(tmp_ll >> 20) & 0xf];
+		*buf++ = hexchars[(tmp_ll >> 16) & 0xf];
+		*buf++ = hexchars[(tmp_ll >> 28) & 0xf];
+		*buf++ = hexchars[(tmp_ll >> 24) & 0xf];
+		*buf++ = hexchars[(tmp_ll >> 36) & 0xf];
+		*buf++ = hexchars[(tmp_ll >> 32) & 0xf];
+		*buf++ = hexchars[(tmp_ll >> 44) & 0xf];
+		*buf++ = hexchars[(tmp_ll >> 40) & 0xf];
+		*buf++ = hexchars[(tmp_ll >> 52) & 0xf];
+		*buf++ = hexchars[(tmp_ll >> 48) & 0xf];
+		*buf++ = hexchars[(tmp_ll >> 60) & 0xf];
+		*buf++ = hexchars[(tmp_ll >> 56) & 0xf];
+#endif
+#endif
+	} else {
+		while (count-- > 0) {
+			unsigned char ch = *mem++;
+			*buf++ = hexchars[ch >> 4];
+			*buf++ = hexchars[ch & 0xf];
+		}
+	}
+	kgdb_may_fault = 0;
+	*buf = 0;
+	return (buf);
+}
+
+/*
+ * Copy the binary array pointed to by buf into mem.  Fix $, #, and
+ * 0x7d escaped with 0x7d.  Return a pointer to the character after
+ * the last byte written.
+ */
+static char *kgdb_ebin2mem(char *buf, char *mem, int count)
+{
+	kgdb_may_fault = 1;
+	if ((kgdb_fault_setjmp(kgdb_fault_jmp_regs)) != 0) {
+		kgdb_may_fault = 0;
+		return ERR_PTR(-EINVAL);
+	}
+	for (; count > 0; count--, buf++) {
+		if (*buf == 0x7d)
+			*mem++ = *(++buf) ^ 0x20;
+		else
+			*mem++ = *buf;
+	}
+	kgdb_may_fault = 0;
+	return mem;
+}
+
+/*
+ * convert the hex array pointed to by buf into binary to be placed in mem
+ * return a pointer to the character AFTER the last byte written
+ * May return an error.
+ */
+char *kgdb_hex2mem(char *buf, char *mem, int count)
+{
+	kgdb_may_fault = 1;
+	if ((kgdb_fault_setjmp(kgdb_fault_jmp_regs)) != 0) {
+		kgdb_may_fault = 0;
+		return ERR_PTR(-EINVAL);
+	}
+	if ((count == 2) && (((long)mem & 1) == 0)) {
+		unsigned short tmp_s = 0;
+#ifdef __BIG_ENDIAN
+		tmp_s |= hex(*buf++) << 12;
+		tmp_s |= hex(*buf++) << 8;
+		tmp_s |= hex(*buf++) << 4;
+		tmp_s |= hex(*buf++);
+#else
+		tmp_s |= hex(*buf++) << 4;
+		tmp_s |= hex(*buf++);
+		tmp_s |= hex(*buf++) << 12;
+		tmp_s |= hex(*buf++) << 8;
+#endif
+		*(unsigned short *)mem = tmp_s;
+		mem += 2;
+	} else if ((count == 4) && (((long)mem & 3) == 0)) {
+		unsigned long tmp_l = 0;
+#ifdef __BIG_ENDIAN
+		tmp_l |= hex(*buf++) << 28;
+		tmp_l |= hex(*buf++) << 24;
+		tmp_l |= hex(*buf++) << 20;
+		tmp_l |= hex(*buf++) << 16;
+		tmp_l |= hex(*buf++) << 12;
+		tmp_l |= hex(*buf++) << 8;
+		tmp_l |= hex(*buf++) << 4;
+		tmp_l |= hex(*buf++);
+#else
+		tmp_l |= hex(*buf++) << 4;
+		tmp_l |= hex(*buf++);
+		tmp_l |= hex(*buf++) << 12;
+		tmp_l |= hex(*buf++) << 8;
+		tmp_l |= hex(*buf++) << 20;
+		tmp_l |= hex(*buf++) << 16;
+		tmp_l |= hex(*buf++) << 28;
+		tmp_l |= hex(*buf++) << 24;
+#endif
+		*(unsigned long *)mem = tmp_l;
+		mem += 4;
+	} else {
+		int i;
+		for (i = 0; i < count; i++) {
+			unsigned char ch = hex(*buf++) << 4;
+			ch |= hex(*buf++);
+			*mem++ = ch;
+		}
+	}
+	kgdb_may_fault = 0;
+	return (mem);
+}
+
+/*
+ * While we find nice hex chars, build a long_val.
+ * Return number of chars processed.
+ */
+int kgdb_hex2long(char **ptr, long *long_val)
+{
+	int hex_val, num = 0;
+
+	*long_val = 0;
+
+	while (**ptr) {
+		hex_val = hex(**ptr);
+		if (hex_val >= 0) {
+			*long_val = (*long_val << 4) | hex_val;
+			num++;
+		} else
+			break;
+
+		(*ptr)++;
+	}
+
+	return (num);
+}
+
+/* Write memory due to an 'M' or 'X' packet. */
+static char *write_mem_msg(int binary)
+{
+	char *ptr = &remcom_in_buffer[1];
+	unsigned long addr, length;
+
+	if (kgdb_hex2long(&ptr, &addr) > 0 && *(ptr++) == ',' &&
+	    kgdb_hex2long(&ptr, &length) > 0 && *(ptr++) == ':') {
+		if (binary)
+			ptr = kgdb_ebin2mem(ptr, (char *)addr, length);
+		else
+			ptr = kgdb_hex2mem(ptr, (char *)addr, length);
+		if (CACHE_FLUSH_IS_SAFE)
+			flush_icache_range(addr, addr + length + 1);
+		if (IS_ERR(ptr))
+			return ptr;
+		return NULL;
+	}
+
+	return ERR_PTR(-EINVAL);
+}
+
+static inline char *pack_hex_byte(char *pkt, int byte)
+{
+	*pkt++ = hexchars[(byte >> 4) & 0xf];
+	*pkt++ = hexchars[(byte & 0xf)];
+	return pkt;
+}
+
+static inline void error_packet(char *pkt, int error)
+{
+	error = -error;
+	pkt[0] = 'E';
+	pkt[1] = hexchars[(error / 10)];
+	pkt[2] = hexchars[(error % 10)];
+	pkt[3] = '\0';
+}
+
+static char *pack_threadid(char *pkt, threadref * id)
+{
+	char *limit;
+	unsigned char *altid;
+
+	altid = (unsigned char *)id;
+	limit = pkt + BUF_THREAD_ID_SIZE;
+	while (pkt < limit)
+		pkt = pack_hex_byte(pkt, *altid++);
+
+	return pkt;
+}
+
+void int_to_threadref(threadref * id, int value)
+{
+	unsigned char *scan;
+	int i = 4;
+
+	scan = (unsigned char *)id;
+	while (i--)
+		*scan++ = 0;
+	*scan++ = (value >> 24) & 0xff;
+	*scan++ = (value >> 16) & 0xff;
+	*scan++ = (value >> 8) & 0xff;
+	*scan++ = (value & 0xff);
+}
+
+static struct task_struct *getthread(struct pt_regs *regs, int tid)
+{
+	if (!pidhash_init_done)
+		return current;
+
+	if (tid >= pid_max + num_online_cpus() + kgdb_ops->shadowth)
+		return NULL;
+
+	if (tid >= pid_max + num_online_cpus())
+		return kgdb_get_shadow_thread(regs, tid - pid_max -
+					      num_online_cpus());
+
+	if (tid >= pid_max)
+		return idle_task(tid - pid_max);
+
+	if (!tid)
+		return NULL;
+
+	return find_task_by_pid(tid);
+}
+
+#ifdef CONFIG_SMP
+static void kgdb_wait(struct pt_regs *regs)
+{
+	unsigned long flags;
+	int processor;
+
+	local_irq_save(flags);
+	processor = smp_processor_id();
+	procindebug[processor] = 1;
+	kgdb_info[processor].debuggerinfo = regs;
+	kgdb_info[processor].task = current;
+
+	/* Wait till master processor goes completely into the debugger.
+	 * FIXME: this looks racy */
+	while (!procindebug[atomic_read(&debugger_active) - 1]) {
+		int i = 10;	/* an arbitrary number */
+
+		while (--i)
+			cpu_relax();
+		barrier();
+	}
+
+	/* Wait till master processor is done with debugging */
+	spin_lock(&slavecpulocks[processor]);
+
+	/* This has been taken from x86 kgdb implementation and
+	 * will be needed by architectures that have SMP support
+	 */
+	kgdb_correct_hw_break();
+
+	kgdb_info[processor].debuggerinfo = NULL;
+	kgdb_info[processor].task = NULL;
+
+	/* Signal the master processor that we are done */
+	procindebug[processor] = 0;
+	spin_unlock(&slavecpulocks[processor]);
+	local_irq_restore(flags);
+}
+#endif
+
+int kgdb_get_mem(char *addr, unsigned char *buf, int count)
+{
+	while (count) {
+		if ((unsigned long)addr < TASK_SIZE)
+			return -EINVAL;
+		*buf++ = *addr++;
+		count--;
+	}
+	return 0;
+}
+
+int kgdb_set_mem(char *addr, unsigned char *buf, int count)
+{
+	while (count) {
+		if ((unsigned long)addr < TASK_SIZE)
+			return -EINVAL;
+		*addr++ = *buf++;
+		count--;
+	}
+	return 0;
+}
+
+static int kgdb_set_sw_break(unsigned long addr)
+{
+	int i, breakno = -1;
+	int error;
+
+	for (i = 0; i < MAX_BREAKPOINTS; i++) {
+		if ((kgdb_break[i].state == bp_enabled) &&
+		    (kgdb_break[i].bpt_addr == addr))
+			return -EEXIST;
+
+		if (kgdb_break[i].state == bp_disabled) {
+			if ((breakno == -1) || (kgdb_break[i].bpt_addr == addr))
+				breakno = i;
+		}
+	}
+	if (breakno == -1)
+		return -E2BIG;
+
+	if (kgdb_ops->set_breakpoint) {
+		if ((error = kgdb_ops->set_breakpoint(addr,
+					kgdb_break[breakno].saved_instr)) < 0)
+			return error;
+	} else {
+		if ((error = kgdb_get_mem((char *)addr,
+					  kgdb_break[breakno].saved_instr,
+					  BREAK_INSTR_SIZE)) < 0)
+			return error;
+
+		if ((error = kgdb_set_mem((char *)addr, kgdb_ops->gdb_bpt_instr,
+					  BREAK_INSTR_SIZE)) < 0)
+			return error;
+	}
+	if (CACHE_FLUSH_IS_SAFE) {
+		if (current->mm && addr < TASK_SIZE)
+			flush_cache_range(current->mm->mmap_cache, addr,
+					  addr + BREAK_INSTR_SIZE);
+		else
+			flush_icache_range(addr, addr + BREAK_INSTR_SIZE);
+	}
+	kgdb_break[breakno].state = bp_enabled;
+	kgdb_break[breakno].type = bp_breakpoint;
+	kgdb_break[breakno].bpt_addr = addr;
+
+	return 0;
+}
+
+static int kgdb_remove_sw_break(unsigned long addr)
+{
+	int i;
+	int error;
+
+	for (i = 0; i < MAX_BREAKPOINTS; i++) {
+		if ((kgdb_break[i].state == bp_enabled) &&
+		    (kgdb_break[i].bpt_addr == addr)) {
+			if (kgdb_ops->remove_breakpoint) {
+				if ((error = kgdb_ops->remove_breakpoint(addr,
+						kgdb_break[i].saved_instr)) < 0)
+					return error;
+			} else if ((error =
+				    kgdb_set_mem((char *)addr,
+						 kgdb_break[i].saved_instr,
+						 BREAK_INSTR_SIZE)) < 0)
+				return error;
+			if (CACHE_FLUSH_IS_SAFE && current->mm &&
+					addr < TASK_SIZE)
+				flush_cache_range(current->mm->mmap_cache,
+						addr, addr + BREAK_INSTR_SIZE);
+			else if (CACHE_FLUSH_IS_SAFE)
+				flush_icache_range(addr,
+						addr + BREAK_INSTR_SIZE);
+			kgdb_break[i].state = bp_disabled;
+			return 0;
+		}
+	}
+	return -ENOENT;
+}
+
+int remove_all_break(void)
+{
+	int i;
+	int error;
+
+	/* Clear memory breakpoints. */
+	for (i = 0; i < MAX_BREAKPOINTS; i++) {
+		if (kgdb_break[i].state == bp_enabled) {
+			unsigned long addr = kgdb_break[i].bpt_addr;
+			if ((error =
+			     kgdb_set_mem((char *)addr,
+					  kgdb_break[i].saved_instr,
+					  BREAK_INSTR_SIZE)) < 0)
+				return error;
+			if (CACHE_FLUSH_IS_SAFE && current->mm &&
+					addr < TASK_SIZE)
+				flush_cache_range(current->mm->mmap_cache,
+						addr, addr + BREAK_INSTR_SIZE);
+			else if (CACHE_FLUSH_IS_SAFE)
+				flush_icache_range(addr,
+						addr + BREAK_INSTR_SIZE);
+		}
+		kgdb_break[i].state = bp_disabled;
+	}
+
+	/* Clear hardware breakpoints. */
+	kgdb_remove_all_hw_break();
+
+	return 0;
+}
+
+static inline int shadow_pid(int realpid)
+{
+	if (realpid) {
+		return realpid;
+	}
+	return pid_max + smp_processor_id();
+}
+
+static char gdbmsgbuf[BUFMAX + 1];
+static void kgdb_msg_write(const char *s, int len)
+{
+	int i;
+	int wcount;
+	char *bufptr;
+
+	/* 'O'utput */
+	gdbmsgbuf[0] = 'O';
+
+	/* Fill and send buffers... */
+	while (len > 0) {
+		bufptr = gdbmsgbuf + 1;
+
+		/* Calculate how many this time */
+		if ((len << 1) > (BUFMAX - 2))
+			wcount = (BUFMAX - 2) >> 1;
+		else
+			wcount = len;
+
+		/* Pack in hex chars */
+		for (i = 0; i < wcount; i++)
+			bufptr = pack_hex_byte(bufptr, s[i]);
+		*bufptr = '\0';
+
+		/* Move up */
+		s += wcount;
+		len -= wcount;
+
+		/* Write packet */
+		put_packet(gdbmsgbuf);
+	}
+}
+
+/*
+ * This function does all command procesing for interfacing to gdb.
+ *
+ * Locking hierarchy:
+ *	interface locks, if any (begin_session)
+ *	kgdb lock (debugger_active)
+ *
+ * Note that since we can be in here prior to our cpumask being filled
+ * out, we err on the side of caution and loop over NR_CPUS instead
+ * of a for_each_online_cpu.
+ *
+ */
+int kgdb_handle_exception(int ex_vector, int signo, int err_code,
+			  struct pt_regs *linux_regs)
+{
+	unsigned long length, addr;
+	char *ptr;
+	unsigned long flags;
+	int i;
+	long threadid;
+	threadref thref;
+	struct task_struct *thread = NULL;
+	unsigned procid;
+	int numshadowth = num_online_cpus() + kgdb_ops->shadowth;
+	long kgdb_usethreadid = 0;
+	int error = 0, all_cpus_synced = 0;
+	struct pt_regs *shadowregs;
+	int processor = smp_processor_id();
+	void *local_debuggerinfo;
+
+	/* Panic on recursive debugger calls. */
+	if (atomic_read(&debugger_active) == smp_processor_id() + 1)
+		return 0;
+
+      acquirelock:
+
+	/* Call the I/O drivers pre_exception routine if the I/O
+	 * driver defined one
+	 */
+	if (kgdb_io_ops.pre_exception)
+		kgdb_io_ops.pre_exception();
+
+	/*
+	 * Interrupts will be restored by the 'trap return' code, except when
+	 * single stepping.
+	 */
+	local_irq_save(flags);
+
+	/* Hold debugger_active */
+	procid = smp_processor_id();
+	while (cmpxchg(&atomic_read(&debugger_active), 0, (procid + 1)) != 0) {
+		int i = 25;	/* an arbitrary number */
+
+		while (--i)
+			cpu_relax();
+	}
+	/*
+	 * Don't enter if the last instance of the exception handler wanted to
+	 * come into the debugger again.
+	 */
+	if (atomic_read(&cpu_doing_single_step) != -1 &&
+	    atomic_read(&cpu_doing_single_step) != procid) {
+		atomic_set(&debugger_active, 0);
+		local_irq_restore(flags);
+		goto acquirelock;
+	}
+
+	kgdb_info[processor].debuggerinfo = linux_regs;
+	kgdb_info[processor].task = current;
+
+	kgdb_disable_hw_debug(linux_regs);
+
+	if (!debugger_step || !kgdb_contthread)
+		for (i = 0; i < NR_CPUS; i++)
+			spin_lock(&slavecpulocks[i]);
+
+	/* Make sure we get the other CPUs */
+	if (!debugger_step || !kgdb_contthread)
+		kgdb_roundup_cpus(flags);
+
+	/* spin_lock code is good enough as a barrier so we don't
+	 * need one here */
+	procindebug[smp_processor_id()] = 1;
+
+	/* Wait a reasonable time for the other CPUs to be notified and
+	 * be waiting for us.  Very early on this could be imperfect
+	 * as num_online_cpus() could be 0.*/
+	for (i = 0; i < ROUNDUP_WAIT; i++) {
+		int cpu, num = 0;
+		for (cpu = 0; cpu < NR_CPUS; cpu++) {
+			if (procindebug[cpu])
+				num++;
+		}
+		if (num >= num_online_cpus()) {
+			all_cpus_synced = 1;
+			break;
+		}
+	}
+
+	/* Clear the out buffer. */
+	memset(remcom_out_buffer, 0, sizeof(remcom_out_buffer));
+
+	/* Master processor is completely in the debugger */
+	kgdb_post_master_code(linux_regs, ex_vector, err_code);
+
+	debugger_step = 0;
+	kgdb_contthread = NULL;
+
+	if (kgdb_connected) {
+		/* If we're still unable to roundup all of the CPUs,
+		 * send an 'O' packet informing the user again. */
+		if (!all_cpus_synced)
+			kgdb_msg_write("Not all CPUs have been synced for "
+				       "KGDB\n", 39);
+		/* Reply to host that an exception has occurred */
+		ptr = remcom_out_buffer;
+		*ptr++ = 'T';
+		*ptr++ = hexchars[(signo >> 4) % 16];
+		*ptr++ = hexchars[signo % 16];
+		ptr += strlen(strcpy(ptr, "thread:"));
+		int_to_threadref(&thref, shadow_pid(current->pid));
+		ptr = pack_threadid(ptr, &thref);
+		*ptr++ = ';';
+
+		put_packet(remcom_out_buffer);
+	}
+
+	kgdb_usethread = kgdb_info[processor].task;
+	kgdb_usethreadid = shadow_pid(kgdb_info[processor].task->pid);
+
+	while (kgdb_io_ops.init) {
+		char *bpt_type;
+		error = 0;
+
+		/* Clear the out buffer. */
+		memset(remcom_out_buffer, 0, sizeof(remcom_out_buffer));
+
+		get_packet(remcom_in_buffer);
+
+		switch (remcom_in_buffer[0]) {
+		case '?':
+			/* We know that this packet is only sent
+			 * during initial connect.  So to be safe,
+			 * we clear out our breakpoints now incase
+			 * GDB is reconnecting. */
+			remove_all_break();
+			/* Also, if we haven't been able to roundup all
+			 * CPUs, send an 'O' packet informing the user
+			 * as much.  Only need to do this once. */
+			if (!all_cpus_synced)
+				kgdb_msg_write("Not all CPUs have been "
+					       "synced for KGDB\n", 39);
+			remcom_out_buffer[0] = 'S';
+			remcom_out_buffer[1] = hexchars[signo >> 4];
+			remcom_out_buffer[2] = hexchars[signo % 16];
+			break;
+
+		case 'g':	/* return the value of the CPU registers */
+			thread = kgdb_usethread;
+
+			if (!thread) {
+				thread = kgdb_info[processor].task;
+				local_debuggerinfo =
+				    kgdb_info[processor].debuggerinfo;
+			} else {
+				local_debuggerinfo = NULL;
+				for (i = 0; i < NR_CPUS; i++) {
+					/* Try to find the task on some other
+					 * or possibly this node if we do not
+					 * find the matching task then we try
+					 * to approximate the results.
+					 */
+					if (thread == kgdb_info[i].task)
+						local_debuggerinfo =
+						    kgdb_info[i].debuggerinfo;
+				}
+			}
+
+			/* All threads that don't have debuggerinfo should be
+			 * in __schedule() sleeping, since all other CPUs
+			 * are in kgdb_wait, and thus have debuggerinfo. */
+			if (kgdb_usethreadid >= pid_max + num_online_cpus()) {
+				shadowregs = kgdb_shadow_regs(linux_regs,
+							      kgdb_usethreadid -
+							      pid_max -
+							      num_online_cpus
+							      ());
+				if (!shadowregs) {
+					error_packet(remcom_out_buffer,
+						     -EINVAL);
+					break;
+				}
+				regs_to_gdb_regs(gdb_regs, shadowregs);
+			} else if (local_debuggerinfo)
+				regs_to_gdb_regs(gdb_regs, local_debuggerinfo);
+			else {
+				/* Pull stuff saved during
+				 * switch_to; nothing else is
+				 * accessible (or even particularly relevant).
+				 * This should be enough for a stack trace. */
+				sleeping_thread_to_gdb_regs(gdb_regs, thread);
+			}
+			kgdb_mem2hex((char *)gdb_regs, remcom_out_buffer,
+				     NUMREGBYTES);
+			break;
+
+			/* set the value of the CPU registers - return OK */
+		case 'G':
+			kgdb_hex2mem(&remcom_in_buffer[1], (char *)gdb_regs,
+				     NUMREGBYTES);
+
+			if (kgdb_usethread && kgdb_usethread != current)
+				error_packet(remcom_out_buffer, -EINVAL);
+			else {
+				gdb_regs_to_regs(gdb_regs, linux_regs);
+				strcpy(remcom_out_buffer, "OK");
+			}
+			break;
+
+			/* mAA..AA,LLLL  Read LLLL bytes at address AA..AA */
+		case 'm':
+			ptr = &remcom_in_buffer[1];
+			if (kgdb_hex2long(&ptr, &addr) > 0 && *ptr++ == ',' &&
+			    kgdb_hex2long(&ptr, &length) > 0) {
+				if (IS_ERR(ptr = kgdb_mem2hex((char *)addr,
+							      remcom_out_buffer,
+							      length)))
+					error_packet(remcom_out_buffer,
+						     PTR_ERR(ptr));
+			} else
+				error_packet(remcom_out_buffer, -EINVAL);
+			break;
+
+			/* MAA..AA,LLLL: Write LLLL bytes at address AA..AA */
+		case 'M':
+			if (IS_ERR(ptr = write_mem_msg(0)))
+				error_packet(remcom_out_buffer, PTR_ERR(ptr));
+			else
+				strcpy(remcom_out_buffer, "OK");
+			break;
+			/* XAA..AA,LLLL: Write LLLL bytes at address AA..AA */
+		case 'X':
+			if (IS_ERR(ptr = write_mem_msg(1)))
+				error_packet(remcom_out_buffer, PTR_ERR(ptr));
+			else
+				strcpy(remcom_out_buffer, "OK");
+			break;
+
+			/* kill or detach. KGDB should treat this like a
+			 * continue.
+			 */
+		case 'D':
+			if ((error = remove_all_break()) < 0) {
+				error_packet(remcom_out_buffer, error);
+			} else {
+				strcpy(remcom_out_buffer, "OK");
+				kgdb_connected = 0;
+			}
+			put_packet(remcom_out_buffer);
+			goto default_handle;
+
+		case 'k':
+			/* Don't care about error from remove_all_break */
+			remove_all_break();
+			kgdb_connected = 0;
+			goto default_handle;
+
+			/* query */
+		case 'q':
+			switch (remcom_in_buffer[1]) {
+			case 's':
+			case 'f':
+				if (memcmp(remcom_in_buffer + 2, "ThreadInfo",
+					   10)) {
+					error_packet(remcom_out_buffer,
+						     -EINVAL);
+					break;
+				}
+
+				/*
+				 * If we have not yet completed in
+				 * pidhash_init() there isn't much we
+				 * can give back.
+				 */
+				if (!pidhash_init_done) {
+					if (remcom_in_buffer[1] == 'f')
+						strcpy(remcom_out_buffer,
+						       "m0000000000000001");
+					break;
+				}
+
+				if (remcom_in_buffer[1] == 'f') {
+					threadid = 1;
+				}
+				remcom_out_buffer[0] = 'm';
+				ptr = remcom_out_buffer + 1;
+				for (i = 0; i < 17 && threadid < pid_max +
+				     numshadowth; threadid++) {
+					thread = getthread(linux_regs,
+							   threadid);
+					if (thread) {
+						int_to_threadref(&thref,
+								 threadid);
+						pack_threadid(ptr, &thref);
+						ptr += 16;
+						*(ptr++) = ',';
+						i++;
+					}
+				}
+				*(--ptr) = '\0';
+				break;
+
+			case 'C':
+				/* Current thread id */
+				strcpy(remcom_out_buffer, "QC");
+
+				threadid = shadow_pid(current->pid);
+
+				int_to_threadref(&thref, threadid);
+				pack_threadid(remcom_out_buffer + 2, &thref);
+				break;
+			case 'T':
+				if (memcmp(remcom_in_buffer + 1,
+					   "ThreadExtraInfo,", 16)) {
+					error_packet(remcom_out_buffer,
+						     -EINVAL);
+					break;
+				}
+				threadid = 0;
+				ptr = remcom_in_buffer + 17;
+				kgdb_hex2long(&ptr, &threadid);
+				if (!getthread(linux_regs, threadid)) {
+					error_packet(remcom_out_buffer,
+						     -EINVAL);
+					break;
+				}
+				if (threadid < pid_max) {
+					kgdb_mem2hex(getthread(linux_regs,
+							       threadid)->comm,
+						     remcom_out_buffer, 16);
+				} else if (threadid >= pid_max +
+					   num_online_cpus()) {
+					kgdb_shadowinfo(linux_regs,
+							remcom_out_buffer,
+							threadid - pid_max -
+							num_online_cpus());
+				} else {
+					static char tmpstr[23 +
+							   BUF_THREAD_ID_SIZE];
+					sprintf(tmpstr, "Shadow task %d"
+						" for pid 0",
+						(int)(threadid - pid_max));
+					kgdb_mem2hex(tmpstr, remcom_out_buffer,
+						     strlen(tmpstr));
+				}
+				break;
+			}
+			break;
+
+			/* task related */
+		case 'H':
+			switch (remcom_in_buffer[1]) {
+			case 'g':
+				ptr = &remcom_in_buffer[2];
+				kgdb_hex2long(&ptr, &threadid);
+				thread = getthread(linux_regs, threadid);
+				if (!thread && threadid > 0) {
+					error_packet(remcom_out_buffer,
+						     -EINVAL);
+					break;
+				}
+				kgdb_usethread = thread;
+				kgdb_usethreadid = threadid;
+				strcpy(remcom_out_buffer, "OK");
+				break;
+
+			case 'c':
+				ptr = &remcom_in_buffer[2];
+				kgdb_hex2long(&ptr, &threadid);
+				if (!threadid) {
+					kgdb_contthread = NULL;
+				} else {
+					thread = getthread(linux_regs,
+							   threadid);
+					if (!thread && threadid > 0) {
+						error_packet(remcom_out_buffer,
+							     -EINVAL);
+						break;
+					}
+					kgdb_contthread = thread;
+				}
+				strcpy(remcom_out_buffer, "OK");
+				break;
+			}
+			break;
+
+			/* Query thread status */
+		case 'T':
+			ptr = &remcom_in_buffer[1];
+			kgdb_hex2long(&ptr, &threadid);
+			thread = getthread(linux_regs, threadid);
+			if (thread)
+				strcpy(remcom_out_buffer, "OK");
+			else
+				error_packet(remcom_out_buffer, -EINVAL);
+			break;
+		/* Since GDB-5.3, it's been drafted that '0' is a software
+		 * breakpoint, '1' is a hardware breakpoint, so let's do
+		 * that.
+		 */
+		case 'z':
+		case 'Z':
+			bpt_type = &remcom_in_buffer[1];
+			ptr = &remcom_in_buffer[2];
+
+			if (kgdb_ops->set_hw_breakpoint && *bpt_type >= '1') {
+				/* Unsupported */
+				if (*bpt_type > '4')
+					break;
+			} else if (*bpt_type != '0' && *bpt_type != '1')
+				/* Unsupported. */
+				break;
+			/* Test if this is a hardware breakpoint, and
+			 * if we support it. */
+			if (*bpt_type == '1' &&
+			    !kgdb_ops->flags & KGDB_HW_BREAKPOINT)
+				/* Unsupported. */
+				break;
+
+			if (*(ptr++) != ',') {
+				error_packet(remcom_out_buffer, -EINVAL);
+				break;
+			} else if (kgdb_hex2long(&ptr, &addr)) {
+				if (*(ptr++) != ',' ||
+				    !kgdb_hex2long(&ptr, &length)) {
+					error_packet(remcom_out_buffer,
+						     -EINVAL);
+					break;
+				}
+			} else {
+				error_packet(remcom_out_buffer, -EINVAL);
+				break;
+			}
+
+			if (remcom_in_buffer[0] == 'Z' && *bpt_type == '0')
+				error = kgdb_set_sw_break(addr);
+			else if (remcom_in_buffer[0] == 'Z' && *bpt_type == '1')
+				error = kgdb_set_hw_break(addr);
+			else if (remcom_in_buffer[0] == 'z' && *bpt_type == '0')
+				error = kgdb_remove_sw_break(addr);
+			else if (remcom_in_buffer[0] == 'z' && *bpt_type == '1')
+				error = kgdb_remove_hw_break(addr);
+			else if (remcom_in_buffer[0] == 'Z')
+				error = kgdb_ops->set_hw_breakpoint(addr,
+								    (int)length,
+								    *bpt_type);
+			else if (remcom_in_buffer[0] == 'z')
+				error = kgdb_ops->remove_hw_breakpoint(addr,
+								       (int)
+								       length,
+								       *bpt_type);
+
+			if (error == 0)
+				strcpy(remcom_out_buffer, "OK");
+			else
+				error_packet(remcom_out_buffer, error);
+
+			break;
+		case 'c':
+		case 's':
+			if (kgdb_contthread && kgdb_contthread != current) {
+				/* Can't switch threads in kgdb */
+				error_packet(remcom_out_buffer, -EINVAL);
+				break;
+			}
+
+			/* Followthrough to default processing */
+		default:
+		      default_handle:
+			error = kgdb_arch_handle_exception(ex_vector, signo,
+							   err_code,
+							   remcom_in_buffer,
+							   remcom_out_buffer,
+							   linux_regs);
+
+			if (error >= 0 || remcom_in_buffer[0] == 'D' ||
+			    remcom_in_buffer[0] == 'k')
+				goto kgdb_exit;
+
+		}		/* switch */
+
+		/* reply to the request */
+		put_packet(remcom_out_buffer);
+	}
+
+      kgdb_exit:
+	/* Call the I/O driver's post_exception routine if the I/O
+	 * driver defined one.
+	 */
+	if (kgdb_io_ops.post_exception)
+		kgdb_io_ops.post_exception();
+
+	kgdb_info[processor].debuggerinfo = NULL;
+	kgdb_info[processor].task = NULL;
+	procindebug[smp_processor_id()] = 0;
+
+	if (!debugger_step || !kgdb_contthread) {
+		for (i = 0; i < NR_CPUS; i++)
+			spin_unlock(&slavecpulocks[i]);
+		/* Wait till all the processors have quit
+		 * from the debugger. */
+		for (i = 0; i < NR_CPUS; i++) {
+			while (procindebug[i]) {
+				int j = 10;	/* an arbitrary number */
+
+				while (--j)
+					cpu_relax();
+				barrier();
+			}
+		}
+	}
+
+	/* Free debugger_active */
+	atomic_set(&debugger_active, 0);
+	local_irq_restore(flags);
+
+	return error;
+}
+
+/*
+ * GDB places a breakpoint at this function to know dynamically
+ * loaded objects. It's not defined static so that only one instance with this
+ * name exists in the kernel.
+ */
+
+int module_event(struct notifier_block *self, unsigned long val, void *data)
+{
+	return 0;
+}
+
+static struct notifier_block kgdb_module_load_nb = {
+	.notifier_call = module_event,
+};
+
+void kgdb_nmihook(int cpu, void *regs)
+{
+#ifdef CONFIG_SMP
+	if (!procindebug[cpu] && atomic_read(&debugger_active) != (cpu + 1))
+		kgdb_wait((struct pt_regs *)regs);
+#endif
+}
+
+/*
+ * This is called when a panic happens.  All we need to do is
+ * breakpoint().
+ */
+static int kgdb_panic_notify(struct notifier_block *self, unsigned long cmd,
+			     void *ptr)
+{
+	breakpoint();
+
+	return 0;
+}
+
+static struct notifier_block kgdb_panic_notifier = {
+	.notifier_call = kgdb_panic_notify,
+};
+
+/*
+ * Initialization that needs to be done in either of our entry points.
+ */
+static void __init kgdb_internal_init(void)
+{
+	int i;
+
+	/* Initialize our spinlocks. */
+	for (i = 0; i < NR_CPUS; i++)
+		spin_lock_init(&slavecpulocks[i]);
+
+	for (i = 0; i < MAX_BREAKPOINTS; i++)
+		kgdb_break[i].state = bp_disabled;
+
+	/* Initialize the I/O handles */
+	for (i = 0; i < MAX_KGDB_IO_HANDLERS; i++)
+		kgdb_io_ops_prev[i].init = NULL;
+
+	/* We can't do much if this fails */
+	register_module_notifier(&kgdb_module_load_nb);
+
+	kgdb_initialized = 1;
+}
+
+static void kgdb_register_for_panic(void)
+{
+	/* Register for panics(). */
+	/* The registration is done in the kgdb_register_for_panic
+	 * routine because KGDB should not try to handle a panic when
+	 * there are no kgdb_io_ops setup. It is assumed that the
+	 * kgdb_io_ops are setup at the time this method is called.
+	 */
+	if (!kgdb_from_module_registered) {
+		notifier_chain_register(&panic_notifier_list,
+					&kgdb_panic_notifier);
+		kgdb_from_module_registered = 1;
+	}
+}
+
+static void kgdb_unregister_for_panic(void)
+{
+	/* When this routine is called KGDB should unregister from the
+	 * panic handler and clean up, making sure it is not handling any
+	 * break exceptions at the time.
+	 */
+	if (kgdb_from_module_registered) {
+		kgdb_from_module_registered = 0;
+		notifier_chain_unregister(&panic_notifier_list,
+					  &kgdb_panic_notifier);
+	}
+
+	if (kgdb_connected) {
+		printk(KERN_CRIT "kgdb: ERROR the debug module unload was "
+		       "called while kgdb was conencted.  Recovery "
+		       "attempted, but further debugging may not be "
+		       "possible.\n");
+		if (remove_all_break() < 0) {
+			printk(KERN_CRIT "kgdb: recovery failed, rebooting "
+			       "is advised\n");
+		}
+		kgdb_connected = 0;
+	}
+}
+
+int kgdb_register_io_module(struct kgdb_io *local_kgdb_io_ops)
+{
+
+	if (kgdb_connected) {
+		printk(KERN_ERR "kgdb: Cannot load I/O module while KGDB "
+		       "connected.\n");
+		return -EINVAL;
+	}
+
+	/* Save the old values so they can be restored */
+	if (kgdb_io_handler_cnt >= MAX_KGDB_IO_HANDLERS) {
+		printk(KERN_ERR "kgdb: No more I/O handles available.\n");
+		return -EINVAL;
+	}
+
+	/* Check to see if there is an existing driver and if so save
+	 * its values.
+	 */
+	if (kgdb_io_ops.init != NULL) {
+		memcpy(&kgdb_io_ops_prev[kgdb_io_handler_cnt],
+		       &kgdb_io_ops, sizeof(struct kgdb_io));
+		kgdb_io_handler_cnt++;
+	}
+
+	/* Initialize the io values for this module */
+	memcpy(&kgdb_io_ops, local_kgdb_io_ops, sizeof(struct kgdb_io));
+
+	/* Make the call to register kgdb if is not initialized */
+	kgdb_register_for_panic();
+
+	return 0;
+}
+
+void kgdb_unregister_io_module(struct kgdb_io *local_kgdb_io_ops)
+{
+	int i;
+
+	/* Unregister KGDB if there were no other prior io hooks, else
+	 * restore the io hooks.
+	 */
+	if (kgdb_io_handler_cnt > 0 && kgdb_io_ops_prev[0].init != NULL) {
+		/* First check if the hook that is in use is the one being
+		 * removed */
+		if (kgdb_io_ops.init == local_kgdb_io_ops->init) {
+			/* Set 'i' to the value of where the list should be
+			 * shifed */
+			i = kgdb_io_handler_cnt - 1;
+			memcpy(&kgdb_io_ops, &kgdb_io_ops_prev[i],
+			       sizeof(struct kgdb_io));
+		} else {
+			/* Simple case to remove an entry for an I/O handler
+			 * that is not in use */
+			for (i = 0; i < kgdb_io_handler_cnt; i++) {
+				if (kgdb_io_ops_prev[i].init ==
+				    local_kgdb_io_ops->init)
+					break;
+			}
+		}
+
+		/* Shift all the entries in the handler array so it is
+		 * ordered from oldest to newest.
+		 */
+		kgdb_io_handler_cnt--;
+		for (; i < kgdb_io_handler_cnt; i++) {
+			memcpy(&kgdb_io_ops_prev[i], &kgdb_io_ops_prev[i + 1],
+			       sizeof(struct kgdb_io));
+		}
+		/* Handle the case if we are on the last element and set it
+		 * to NULL; */
+		kgdb_io_ops_prev[kgdb_io_handler_cnt].init = NULL;
+
+		if (kgdb_connected)
+			printk(KERN_ERR "kgdb: WARNING: I/O method changed "
+			       "while kgdb was connected state, "
+			       "further debugging may not be " "possible\n");
+	} else {
+		/* Unregister the KGDB hooks since this must be the
+		 * last I/O module for KGDB
+		 */
+		kgdb_unregister_for_panic();
+		memset(&kgdb_io_ops, 0, sizeof(struct kgdb_io));
+	}
+}
+
+/*
+ * There are times we need to call a tasklet to cause a breakpoint
+ * as calling breakpoint() at that point might be fatal.  We have to
+ * check that the exception stack is setup, as tasklets may be scheduled
+ * prior to this.  When that happens, it is up to the architecture to
+ * schedule this when it is safe to run.
+ */
+static void kgdb_tasklet_bpt(unsigned long ing)
+{
+	if (CHECK_EXCEPTION_STACK())
+		breakpoint();
+}
+
+DECLARE_TASKLET(kgdb_tasklet_breakpoint, kgdb_tasklet_bpt, 0);
+
+/*
+ * This function can be called very early, either via early_param() or
+ * an explicit breakpoint() early on.
+ */
+static void __init kgdb_early_entry(void)
+{
+	/*
+	 * Don't try and do anything until the architecture is able to
+	 * setup the exception stack.  In this case, it is up to the
+	 * architecture to hook in and look at us when they are ready.
+	 */
+	if (!CHECK_EXCEPTION_STACK()) {
+		kgdb_initialized = -1;
+		tasklet_schedule(&kgdb_tasklet_breakpoint);
+		return;
+	}
+
+	/* Let the architecture do any setup that it needs to. */
+	kgdb_arch_init();
+
+	/* Now try the I/O. */
+	/* For early entry kgdb_io_ops.init must be defined */
+	if (!kgdb_io_ops.init || kgdb_io_ops.init()) {
+		/* Try again later. */
+		kgdb_initialized = -1;
+		return;
+	}
+
+	/* Finish up. */
+	kgdb_internal_init();
+
+	/* KGDB can assume that if kgdb_io_ops.init was defined that the
+	 * panic registion should be performed at this time. This means
+	 * kgdb_io_ops.init did not come from a kernel module and was
+	 * initialized statically by a built in.
+	 */
+	if (kgdb_io_ops.init)
+		kgdb_register_for_panic();
+}
+
+/*
+ * This function will always be invoked to make sure that KGDB will grab
+ * what it needs to so that if something happens while the system is
+ * running, KGDB will get involved.  If kgdb_early_entry() has already
+ * been invoked, there is little we need to do.
+ */
+static int __init kgdb_late_entry(void)
+{
+	int need_break = 0;
+
+	/* If kgdb_initialized is -1 then we were passed kgdbwait. */
+	if (kgdb_initialized == -1)
+		need_break = 1;
+
+	/*
+	 * If we haven't tried to initialize KGDB yet, we need to call
+	 * kgdb_arch_init before moving onto the I/O.
+	 */
+	if (!kgdb_initialized)
+		kgdb_arch_init();
+
+	if (kgdb_initialized != 1) {
+		if (kgdb_io_ops.init && kgdb_io_ops.init()) {
+			/* When KGDB allows I/O via modules and the core
+			 * I/O init fails KGDB must default to defering the
+			 * I/O setup, and appropriately print an error about
+			 * it.
+			 */
+			printk(KERN_ERR "kgdb: Could not setup core I/O "
+			       "for KGDB.\n");
+			printk(KERN_INFO "kgdb: Defering I/O setup to kernel "
+			       "module.\n");
+			memset(&kgdb_io_ops, 0, sizeof(struct kgdb_io));
+		} else {
+			printk(KERN_INFO "kgdb: Defering I/O setup to kernel "
+			       "module.\n");
+		}
+
+		kgdb_internal_init();
+
+		/* KGDB can assume that if kgdb_io_ops.init was defined that
+		 * panic registion should be performed at this time. This means
+		 * kgdb_io_ops.init did not come from a kernel module and was
+		 * initialized statically by a built in.
+		 */
+		if (kgdb_io_ops.init)
+			kgdb_register_for_panic();
+	}
+
+	/* Now do any late init of the I/O. */
+	if (kgdb_io_ops.late_init)
+		kgdb_io_ops.late_init();
+
+	if (need_break) {
+		printk(KERN_CRIT "kgdb: Waiting for connection from remote"
+		       " gdb...\n");
+		breakpoint();
+	}
+
+	return 0;
+}
+
+late_initcall(kgdb_late_entry);
+
+/*
+ * This function will generate a breakpoint exception.  It is used at the
+ * beginning of a program to sync up with a debugger and can be used
+ * otherwise as a quick means to stop program execution and "break" into
+ * the debugger.
+ */
+void breakpoint(void)
+{
+	if (kgdb_initialized != 1) {
+		kgdb_early_entry();
+		if (kgdb_initialized == 1)
+			printk(KERN_CRIT "Waiting for connection from remote "
+			       "gdb...\n");
+		else {
+			printk(KERN_CRIT "KGDB cannot initialize I/O yet.\n");
+			return;
+		}
+	}
+
+	atomic_set(&kgdb_setting_breakpoint, 1);
+	wmb();
+	BREAKPOINT();
+	wmb();
+	atomic_set(&kgdb_setting_breakpoint, 0);
+}
+
+EXPORT_SYMBOL(breakpoint);
+
+#ifdef CONFIG_MAGIC_SYSRQ
+static void sysrq_handle_gdb(int key, struct pt_regs *pt_regs,
+			     struct tty_struct *tty)
+{
+	printk("Entering GDB stub\n");
+	breakpoint();
+}
+static struct sysrq_key_op sysrq_gdb_op = {
+	.handler = sysrq_handle_gdb,
+	.help_msg = "Gdb",
+	.action_msg = "GDB",
+};
+
+static int gdb_register_sysrq(void)
+{
+	printk("Registering GDB sysrq handler\n");
+	register_sysrq_key('g', &sysrq_gdb_op);
+	return 0;
+}
+
+module_init(gdb_register_sysrq);
+#endif
+
+#ifdef CONFIG_KGDB_CONSOLE
+void kgdb_console_write(struct console *co, const char *s, unsigned count)
+{
+	unsigned long flags;
+
+	/* If we're debugging, or KGDB has not connected, don't try
+	 * and print. */
+	if (!kgdb_connected || atomic_read(&debugger_active) != 0)
+		return;
+
+	local_irq_save(flags);
+	kgdb_msg_write(s, count);
+	local_irq_restore(flags);
+}
+
+static struct console kgdbcons = {
+	.name = "kgdb",
+	.write = kgdb_console_write,
+	.flags = CON_PRINTBUFFER | CON_ENABLED,
+};
+static int __init kgdb_console_init(void)
+{
+	register_console(&kgdbcons);
+	return 0;
+}
+
+console_initcall(kgdb_console_init);
+#endif
+
+static int __init opt_kgdb_enter(char *str)
+{
+	/* We've already done this by an explicit breakpoint() call. */
+	if (kgdb_initialized)
+		return 0;
+
+	/* Call breakpoint() which will take care of init. */
+	breakpoint();
+
+	return 0;
+}
+
+early_param("kgdbwait", opt_kgdb_enter);
diff -puN kernel/Makefile~core-lite kernel/Makefile
--- linux-2.6.13-rc3/kernel/Makefile~core-lite	2005-07-29 11:13:42.000000000 -0700
+++ linux-2.6.13-rc3-trini/kernel/Makefile	2005-07-29 11:13:42.000000000 -0700
@@ -26,6 +26,7 @@ obj-$(CONFIG_STOP_MACHINE) += stop_machi
 obj-$(CONFIG_AUDIT) += audit.o
 obj-$(CONFIG_AUDITSYSCALL) += auditsc.o
 obj-$(CONFIG_KPROBES) += kprobes.o
+obj-$(CONFIG_KGDB) += kgdb.o
 obj-$(CONFIG_SYSFS) += ksysfs.o
 obj-$(CONFIG_GENERIC_HARDIRQS) += irq/
 obj-$(CONFIG_CRASH_DUMP) += crash_dump.o
diff -puN kernel/pid.c~core-lite kernel/pid.c
--- linux-2.6.13-rc3/kernel/pid.c~core-lite	2005-07-29 11:13:42.000000000 -0700
+++ linux-2.6.13-rc3-trini/kernel/pid.c	2005-07-29 11:13:42.000000000 -0700
@@ -250,8 +250,13 @@ void switch_exec_pids(task_t *leader, ta
 /*
  * The pid hash table is scaled according to the amount of memory in the
  * machine.  From a minimum of 16 slots up to 4096 slots at one gigabyte or
- * more.
+ * more.  KGDB needs to know if this function has been called already,
+ * since we might have entered KGDB very early.
  */
+#ifdef CONFIG_KGDB
+int pidhash_init_done;
+#endif
+
 void __init pidhash_init(void)
 {
 	int i, j, pidhash_size;
@@ -273,6 +278,10 @@ void __init pidhash_init(void)
 		for (j = 0; j < pidhash_size; j++)
 			INIT_HLIST_HEAD(&pid_hash[i][j]);
 	}
+
+#ifdef CONFIG_KGDB
+	pidhash_init_done = 1;
+#endif
 }
 
 void __init pidmap_init(void)
diff -puN kernel/sched.c~core-lite kernel/sched.c
--- linux-2.6.13-rc3/kernel/sched.c~core-lite	2005-07-29 11:13:42.000000000 -0700
+++ linux-2.6.13-rc3-trini/kernel/sched.c	2005-07-29 11:13:42.000000000 -0700
@@ -47,6 +47,7 @@
 #include <linux/syscalls.h>
 #include <linux/times.h>
 #include <linux/acct.h>
+#include <linux/kgdb.h>
 #include <asm/tlb.h>
 
 #include <asm/unistd.h>
@@ -5215,6 +5216,9 @@ void __might_sleep(char *file, int line)
 #if defined(in_atomic)
 	static unsigned long prev_jiffy;	/* ratelimiting */
 
+	if (atomic_read(&debugger_active))
+		return;
+
 	if ((in_atomic() || irqs_disabled()) &&
 	    system_state == SYSTEM_RUNNING && !oops_in_progress) {
 		if (time_before(jiffies, prev_jiffy + HZ) && prev_jiffy)
diff -puN lib/Kconfig.debug~core-lite lib/Kconfig.debug
--- linux-2.6.13-rc3/lib/Kconfig.debug~core-lite	2005-07-29 11:13:42.000000000 -0700
+++ linux-2.6.13-rc3-trini/lib/Kconfig.debug	2005-07-29 13:24:02.000000000 -0700
@@ -119,6 +119,7 @@ config DEBUG_BUGVERBOSE
 config DEBUG_INFO
 	bool "Compile the kernel with debug info"
 	depends on DEBUG_KERNEL
+	default y if KGDB
 	help
           If you say Y here the resulting kernel image will include
 	  debugging info resulting in a larger kernel image.
@@ -153,9 +154,55 @@ config FRAME_POINTER
 	bool "Compile the kernel with frame pointers"
 	depends on DEBUG_KERNEL && ((X86 && !X86_64) || CRIS || M68K || M68KNOMMU || FRV || UML)
 	default y if DEBUG_INFO && UML
+	default y if KGDB
 	help
 	  If you say Y here the resulting kernel image will be slightly larger
 	  and slower, but it will give very useful debugging information.
 	  If you don't debug the kernel, you can say N, but we may not be able
 	  to solve problems without frame pointers.
 
+config KGDB
+	bool "KGDB: kernel debugging with remote gdb"
+	depends on DEBUG_KERNEL
+	help
+	  If you say Y here, it will be possible to remotely debug the
+	  kernel using gdb. It is strongly suggested that you enable
+	  DEBUG_INFO, and if available on your platform, FRAME_POINTER.
+	  Documentation of kernel debugger available at
+	  http://kgdb.sourceforge.net as well as in DocBook form
+	  in Documentation/DocBook/.  If unsure, say N.
+
+config KGDB_CONSOLE
+	bool "KGDB: Console messages through gdb"
+	depends on KGDB
+	  help
+	    If you say Y here, console messages will appear through gdb.
+	    Other consoles such as tty or ttyS will continue to work as usual.
+	    Note, that if you use this in conjunction with KGDB_ETH, if the
+	    ethernet driver runs into an error condition during use with KGDB
+	    it is possible to hit an infinite recusrion, causing the kernel
+	    to crash, and typically reboot.  For this reason, it is preferable
+	    to use NETCONSOLE in conjunction with KGDB_ETH instead of
+	    KGDB_CONSOLE.
+
+choice
+	prompt "Method for KGDB communication"
+	depends on KGDB
+	default KGDB_ONLY_MODULES
+	help
+	  There are a number of different ways in which you can communicate
+	  with KGDB.  The most common is via serial, with the 8250 driver
+	  (should your hardware have an 8250, or ns1655x style uart).
+	  Another option is to use the NETPOLL framework and UDP, should
+	  your ethernet card support this.  Other options may exist.
+	  You can elect to have one core I/O driver that is built into the
+	  kernel for debugging as the kernel is booting, or using only
+	  kernel modules.
+
+config KGDB_ONLY_MODULES
+	bool "KGDB: Use only kernel modules for I/O"
+	help
+	  Use only kernel modules to configure KGDB I/O after the
+	  kernel is booted.
+
+endchoice
diff -puN MAINTAINERS~core-lite MAINTAINERS
--- linux-2.6.13-rc3/MAINTAINERS~core-lite	2005-07-29 11:13:45.000000000 -0700
+++ linux-2.6.13-rc3-trini/MAINTAINERS	2005-07-29 11:15:17.000000000 -0700
@@ -1343,6 +1343,15 @@ L:	[email protected]
 L:	[email protected]
 S:	Maintained
 
+KGDB
+P:	Tom Rini
+P:	Amit S. Kale
+M:	[email protected]
+M:	[email protected]
+W:	http://sourceforge.net/projects/kgdb
+L:	[email protected]
+S:	Maintained
+
 LANMEDIA WAN CARD DRIVER
 P:	Andrew Stanley-Jones
 M:	[email protected]
_
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[Index of Archives]     [Kernel Newbies]     [Netfilter]     [Bugtraq]     [Photo]     [Gimp]     [Yosemite News]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux RAID]     [Video 4 Linux]     [Linux for the blind]
  Powered by Linux