]> git.proxmox.com Git - mirror_ubuntu-eoan-kernel.git/commitdiff
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input
authorLinus Torvalds <torvalds@linux-foundation.org>
Wed, 26 Jan 2011 06:31:44 +0000 (16:31 +1000)
committerLinus Torvalds <torvalds@linux-foundation.org>
Wed, 26 Jan 2011 06:31:44 +0000 (16:31 +1000)
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input:
  Input: wacom - pass touch resolution to clients through input_absinfo
  Input: wacom - add 2 Bamboo Pen and touch models
  Input: sysrq - ensure sysrq_enabled and __sysrq_enabled are consistent
  Input: sparse-keymap - fix KEY_VSW handling in sparse_keymap_setup
  Input: tegra-kbc - add tegra keyboard driver
  Input: gpio_keys - switch to using request_any_context_irq
  Input: serio - allow registered drivers to get status flag
  Input: ct82710c - return proper error code for ct82c710_open
  Input: bu21013_ts - added regulator support
  Input: bu21013_ts - remove duplicate resolution parameters
  Input: tnetv107x-ts - don't treat NULL clk as an error
  Input: tnetv107x-keypad - don't treat NULL clk as an error

Fix up trivial conflicts in drivers/input/keyboard/Makefile due to
additions of tc3589x/Tegra drivers

1  2 
drivers/input/keyboard/Kconfig
drivers/input/keyboard/Makefile
drivers/tty/sysrq.c
kernel/sysctl.c

Simple merge
index 4e5571b72cda460ddda5363f1738dbeef2c103b7,3456b931e2547b5ed3cdd4ce977aa82c3b49a908..468c627a28447efb78b9fb32b9a35351719d4f06
@@@ -41,7 -41,7 +41,8 @@@ obj-$(CONFIG_KEYBOARD_SPEAR)          += spear-
  obj-$(CONFIG_KEYBOARD_STMPE)          += stmpe-keypad.o
  obj-$(CONFIG_KEYBOARD_STOWAWAY)               += stowaway.o
  obj-$(CONFIG_KEYBOARD_SUNKBD)         += sunkbd.o
 +obj-$(CONFIG_KEYBOARD_TC3589X)                += tc3589x-keypad.o
+ obj-$(CONFIG_KEYBOARD_TEGRA)          += tegra-kbc.o
  obj-$(CONFIG_KEYBOARD_TNETV107X)      += tnetv107x-keypad.o
  obj-$(CONFIG_KEYBOARD_TWL4030)                += twl4030_keypad.o
  obj-$(CONFIG_KEYBOARD_XTKBD)          += xtkbd.o
index c556ed9db13d82a4bd7dcff1bb599dce6abd8341,0000000000000000000000000000000000000000..8e0dd254eb11cf3d9f503c7bda32bcb3e75ba09d
mode 100644,000000..100644
--- /dev/null
@@@ -1,890 -1,0 +1,890 @@@
- static int __read_mostly sysrq_enabled = 1;
 +/*
 + *    Linux Magic System Request Key Hacks
 + *
 + *    (c) 1997 Martin Mares <mj@atrey.karlin.mff.cuni.cz>
 + *    based on ideas by Pavel Machek <pavel@atrey.karlin.mff.cuni.cz>
 + *
 + *    (c) 2000 Crutcher Dunnavant <crutcher+kernel@datastacks.com>
 + *    overhauled to use key registration
 + *    based upon discusions in irc://irc.openprojects.net/#kernelnewbies
 + *
 + *    Copyright (c) 2010 Dmitry Torokhov
 + *    Input handler conversion
 + */
 +
 +#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
 +
 +#include <linux/sched.h>
 +#include <linux/interrupt.h>
 +#include <linux/mm.h>
 +#include <linux/fs.h>
 +#include <linux/mount.h>
 +#include <linux/kdev_t.h>
 +#include <linux/major.h>
 +#include <linux/reboot.h>
 +#include <linux/sysrq.h>
 +#include <linux/kbd_kern.h>
 +#include <linux/proc_fs.h>
 +#include <linux/nmi.h>
 +#include <linux/quotaops.h>
 +#include <linux/perf_event.h>
 +#include <linux/kernel.h>
 +#include <linux/module.h>
 +#include <linux/suspend.h>
 +#include <linux/writeback.h>
 +#include <linux/buffer_head.h>                /* for fsync_bdev() */
 +#include <linux/swap.h>
 +#include <linux/spinlock.h>
 +#include <linux/vt_kern.h>
 +#include <linux/workqueue.h>
 +#include <linux/hrtimer.h>
 +#include <linux/oom.h>
 +#include <linux/slab.h>
 +#include <linux/input.h>
 +
 +#include <asm/ptrace.h>
 +#include <asm/irq_regs.h>
 +
 +/* Whether we react on sysrq keys or just ignore them */
++static int __read_mostly sysrq_enabled = SYSRQ_DEFAULT_ENABLE;
 +static bool __read_mostly sysrq_always_enabled;
 +
 +static bool sysrq_on(void)
 +{
 +      return sysrq_enabled || sysrq_always_enabled;
 +}
 +
 +/*
 + * A value of 1 means 'all', other nonzero values are an op mask:
 + */
 +static bool sysrq_on_mask(int mask)
 +{
 +      return sysrq_always_enabled ||
 +             sysrq_enabled == 1 ||
 +             (sysrq_enabled & mask);
 +}
 +
 +static int __init sysrq_always_enabled_setup(char *str)
 +{
 +      sysrq_always_enabled = true;
 +      pr_info("sysrq always enabled.\n");
 +
 +      return 1;
 +}
 +
 +__setup("sysrq_always_enabled", sysrq_always_enabled_setup);
 +
 +
 +static void sysrq_handle_loglevel(int key)
 +{
 +      int i;
 +
 +      i = key - '0';
 +      console_loglevel = 7;
 +      printk("Loglevel set to %d\n", i);
 +      console_loglevel = i;
 +}
 +static struct sysrq_key_op sysrq_loglevel_op = {
 +      .handler        = sysrq_handle_loglevel,
 +      .help_msg       = "loglevel(0-9)",
 +      .action_msg     = "Changing Loglevel",
 +      .enable_mask    = SYSRQ_ENABLE_LOG,
 +};
 +
 +#ifdef CONFIG_VT
 +static void sysrq_handle_SAK(int key)
 +{
 +      struct work_struct *SAK_work = &vc_cons[fg_console].SAK_work;
 +      schedule_work(SAK_work);
 +}
 +static struct sysrq_key_op sysrq_SAK_op = {
 +      .handler        = sysrq_handle_SAK,
 +      .help_msg       = "saK",
 +      .action_msg     = "SAK",
 +      .enable_mask    = SYSRQ_ENABLE_KEYBOARD,
 +};
 +#else
 +#define sysrq_SAK_op (*(struct sysrq_key_op *)NULL)
 +#endif
 +
 +#ifdef CONFIG_VT
 +static void sysrq_handle_unraw(int key)
 +{
 +      struct kbd_struct *kbd = &kbd_table[fg_console];
 +
 +      if (kbd)
 +              kbd->kbdmode = default_utf8 ? VC_UNICODE : VC_XLATE;
 +}
 +static struct sysrq_key_op sysrq_unraw_op = {
 +      .handler        = sysrq_handle_unraw,
 +      .help_msg       = "unRaw",
 +      .action_msg     = "Keyboard mode set to system default",
 +      .enable_mask    = SYSRQ_ENABLE_KEYBOARD,
 +};
 +#else
 +#define sysrq_unraw_op (*(struct sysrq_key_op *)NULL)
 +#endif /* CONFIG_VT */
 +
 +static void sysrq_handle_crash(int key)
 +{
 +      char *killer = NULL;
 +
 +      panic_on_oops = 1;      /* force panic */
 +      wmb();
 +      *killer = 1;
 +}
 +static struct sysrq_key_op sysrq_crash_op = {
 +      .handler        = sysrq_handle_crash,
 +      .help_msg       = "Crash",
 +      .action_msg     = "Trigger a crash",
 +      .enable_mask    = SYSRQ_ENABLE_DUMP,
 +};
 +
 +static void sysrq_handle_reboot(int key)
 +{
 +      lockdep_off();
 +      local_irq_enable();
 +      emergency_restart();
 +}
 +static struct sysrq_key_op sysrq_reboot_op = {
 +      .handler        = sysrq_handle_reboot,
 +      .help_msg       = "reBoot",
 +      .action_msg     = "Resetting",
 +      .enable_mask    = SYSRQ_ENABLE_BOOT,
 +};
 +
 +static void sysrq_handle_sync(int key)
 +{
 +      emergency_sync();
 +}
 +static struct sysrq_key_op sysrq_sync_op = {
 +      .handler        = sysrq_handle_sync,
 +      .help_msg       = "Sync",
 +      .action_msg     = "Emergency Sync",
 +      .enable_mask    = SYSRQ_ENABLE_SYNC,
 +};
 +
 +static void sysrq_handle_show_timers(int key)
 +{
 +      sysrq_timer_list_show();
 +}
 +
 +static struct sysrq_key_op sysrq_show_timers_op = {
 +      .handler        = sysrq_handle_show_timers,
 +      .help_msg       = "show-all-timers(Q)",
 +      .action_msg     = "Show clockevent devices & pending hrtimers (no others)",
 +};
 +
 +static void sysrq_handle_mountro(int key)
 +{
 +      emergency_remount();
 +}
 +static struct sysrq_key_op sysrq_mountro_op = {
 +      .handler        = sysrq_handle_mountro,
 +      .help_msg       = "Unmount",
 +      .action_msg     = "Emergency Remount R/O",
 +      .enable_mask    = SYSRQ_ENABLE_REMOUNT,
 +};
 +
 +#ifdef CONFIG_LOCKDEP
 +static void sysrq_handle_showlocks(int key)
 +{
 +      debug_show_all_locks();
 +}
 +
 +static struct sysrq_key_op sysrq_showlocks_op = {
 +      .handler        = sysrq_handle_showlocks,
 +      .help_msg       = "show-all-locks(D)",
 +      .action_msg     = "Show Locks Held",
 +};
 +#else
 +#define sysrq_showlocks_op (*(struct sysrq_key_op *)NULL)
 +#endif
 +
 +#ifdef CONFIG_SMP
 +static DEFINE_SPINLOCK(show_lock);
 +
 +static void showacpu(void *dummy)
 +{
 +      unsigned long flags;
 +
 +      /* Idle CPUs have no interesting backtrace. */
 +      if (idle_cpu(smp_processor_id()))
 +              return;
 +
 +      spin_lock_irqsave(&show_lock, flags);
 +      printk(KERN_INFO "CPU%d:\n", smp_processor_id());
 +      show_stack(NULL, NULL);
 +      spin_unlock_irqrestore(&show_lock, flags);
 +}
 +
 +static void sysrq_showregs_othercpus(struct work_struct *dummy)
 +{
 +      smp_call_function(showacpu, NULL, 0);
 +}
 +
 +static DECLARE_WORK(sysrq_showallcpus, sysrq_showregs_othercpus);
 +
 +static void sysrq_handle_showallcpus(int key)
 +{
 +      /*
 +       * Fall back to the workqueue based printing if the
 +       * backtrace printing did not succeed or the
 +       * architecture has no support for it:
 +       */
 +      if (!trigger_all_cpu_backtrace()) {
 +              struct pt_regs *regs = get_irq_regs();
 +
 +              if (regs) {
 +                      printk(KERN_INFO "CPU%d:\n", smp_processor_id());
 +                      show_regs(regs);
 +              }
 +              schedule_work(&sysrq_showallcpus);
 +      }
 +}
 +
 +static struct sysrq_key_op sysrq_showallcpus_op = {
 +      .handler        = sysrq_handle_showallcpus,
 +      .help_msg       = "show-backtrace-all-active-cpus(L)",
 +      .action_msg     = "Show backtrace of all active CPUs",
 +      .enable_mask    = SYSRQ_ENABLE_DUMP,
 +};
 +#endif
 +
 +static void sysrq_handle_showregs(int key)
 +{
 +      struct pt_regs *regs = get_irq_regs();
 +      if (regs)
 +              show_regs(regs);
 +      perf_event_print_debug();
 +}
 +static struct sysrq_key_op sysrq_showregs_op = {
 +      .handler        = sysrq_handle_showregs,
 +      .help_msg       = "show-registers(P)",
 +      .action_msg     = "Show Regs",
 +      .enable_mask    = SYSRQ_ENABLE_DUMP,
 +};
 +
 +static void sysrq_handle_showstate(int key)
 +{
 +      show_state();
 +}
 +static struct sysrq_key_op sysrq_showstate_op = {
 +      .handler        = sysrq_handle_showstate,
 +      .help_msg       = "show-task-states(T)",
 +      .action_msg     = "Show State",
 +      .enable_mask    = SYSRQ_ENABLE_DUMP,
 +};
 +
 +static void sysrq_handle_showstate_blocked(int key)
 +{
 +      show_state_filter(TASK_UNINTERRUPTIBLE);
 +}
 +static struct sysrq_key_op sysrq_showstate_blocked_op = {
 +      .handler        = sysrq_handle_showstate_blocked,
 +      .help_msg       = "show-blocked-tasks(W)",
 +      .action_msg     = "Show Blocked State",
 +      .enable_mask    = SYSRQ_ENABLE_DUMP,
 +};
 +
 +#ifdef CONFIG_TRACING
 +#include <linux/ftrace.h>
 +
 +static void sysrq_ftrace_dump(int key)
 +{
 +      ftrace_dump(DUMP_ALL);
 +}
 +static struct sysrq_key_op sysrq_ftrace_dump_op = {
 +      .handler        = sysrq_ftrace_dump,
 +      .help_msg       = "dump-ftrace-buffer(Z)",
 +      .action_msg     = "Dump ftrace buffer",
 +      .enable_mask    = SYSRQ_ENABLE_DUMP,
 +};
 +#else
 +#define sysrq_ftrace_dump_op (*(struct sysrq_key_op *)NULL)
 +#endif
 +
 +static void sysrq_handle_showmem(int key)
 +{
 +      show_mem();
 +}
 +static struct sysrq_key_op sysrq_showmem_op = {
 +      .handler        = sysrq_handle_showmem,
 +      .help_msg       = "show-memory-usage(M)",
 +      .action_msg     = "Show Memory",
 +      .enable_mask    = SYSRQ_ENABLE_DUMP,
 +};
 +
 +/*
 + * Signal sysrq helper function.  Sends a signal to all user processes.
 + */
 +static void send_sig_all(int sig)
 +{
 +      struct task_struct *p;
 +
 +      for_each_process(p) {
 +              if (p->mm && !is_global_init(p))
 +                      /* Not swapper, init nor kernel thread */
 +                      force_sig(sig, p);
 +      }
 +}
 +
 +static void sysrq_handle_term(int key)
 +{
 +      send_sig_all(SIGTERM);
 +      console_loglevel = 8;
 +}
 +static struct sysrq_key_op sysrq_term_op = {
 +      .handler        = sysrq_handle_term,
 +      .help_msg       = "terminate-all-tasks(E)",
 +      .action_msg     = "Terminate All Tasks",
 +      .enable_mask    = SYSRQ_ENABLE_SIGNAL,
 +};
 +
 +static void moom_callback(struct work_struct *ignored)
 +{
 +      out_of_memory(node_zonelist(0, GFP_KERNEL), GFP_KERNEL, 0, NULL);
 +}
 +
 +static DECLARE_WORK(moom_work, moom_callback);
 +
 +static void sysrq_handle_moom(int key)
 +{
 +      schedule_work(&moom_work);
 +}
 +static struct sysrq_key_op sysrq_moom_op = {
 +      .handler        = sysrq_handle_moom,
 +      .help_msg       = "memory-full-oom-kill(F)",
 +      .action_msg     = "Manual OOM execution",
 +      .enable_mask    = SYSRQ_ENABLE_SIGNAL,
 +};
 +
 +#ifdef CONFIG_BLOCK
 +static void sysrq_handle_thaw(int key)
 +{
 +      emergency_thaw_all();
 +}
 +static struct sysrq_key_op sysrq_thaw_op = {
 +      .handler        = sysrq_handle_thaw,
 +      .help_msg       = "thaw-filesystems(J)",
 +      .action_msg     = "Emergency Thaw of all frozen filesystems",
 +      .enable_mask    = SYSRQ_ENABLE_SIGNAL,
 +};
 +#endif
 +
 +static void sysrq_handle_kill(int key)
 +{
 +      send_sig_all(SIGKILL);
 +      console_loglevel = 8;
 +}
 +static struct sysrq_key_op sysrq_kill_op = {
 +      .handler        = sysrq_handle_kill,
 +      .help_msg       = "kill-all-tasks(I)",
 +      .action_msg     = "Kill All Tasks",
 +      .enable_mask    = SYSRQ_ENABLE_SIGNAL,
 +};
 +
 +static void sysrq_handle_unrt(int key)
 +{
 +      normalize_rt_tasks();
 +}
 +static struct sysrq_key_op sysrq_unrt_op = {
 +      .handler        = sysrq_handle_unrt,
 +      .help_msg       = "nice-all-RT-tasks(N)",
 +      .action_msg     = "Nice All RT Tasks",
 +      .enable_mask    = SYSRQ_ENABLE_RTNICE,
 +};
 +
 +/* Key Operations table and lock */
 +static DEFINE_SPINLOCK(sysrq_key_table_lock);
 +
 +static struct sysrq_key_op *sysrq_key_table[36] = {
 +      &sysrq_loglevel_op,             /* 0 */
 +      &sysrq_loglevel_op,             /* 1 */
 +      &sysrq_loglevel_op,             /* 2 */
 +      &sysrq_loglevel_op,             /* 3 */
 +      &sysrq_loglevel_op,             /* 4 */
 +      &sysrq_loglevel_op,             /* 5 */
 +      &sysrq_loglevel_op,             /* 6 */
 +      &sysrq_loglevel_op,             /* 7 */
 +      &sysrq_loglevel_op,             /* 8 */
 +      &sysrq_loglevel_op,             /* 9 */
 +
 +      /*
 +       * a: Don't use for system provided sysrqs, it is handled specially on
 +       * sparc and will never arrive.
 +       */
 +      NULL,                           /* a */
 +      &sysrq_reboot_op,               /* b */
 +      &sysrq_crash_op,                /* c & ibm_emac driver debug */
 +      &sysrq_showlocks_op,            /* d */
 +      &sysrq_term_op,                 /* e */
 +      &sysrq_moom_op,                 /* f */
 +      /* g: May be registered for the kernel debugger */
 +      NULL,                           /* g */
 +      NULL,                           /* h - reserved for help */
 +      &sysrq_kill_op,                 /* i */
 +#ifdef CONFIG_BLOCK
 +      &sysrq_thaw_op,                 /* j */
 +#else
 +      NULL,                           /* j */
 +#endif
 +      &sysrq_SAK_op,                  /* k */
 +#ifdef CONFIG_SMP
 +      &sysrq_showallcpus_op,          /* l */
 +#else
 +      NULL,                           /* l */
 +#endif
 +      &sysrq_showmem_op,              /* m */
 +      &sysrq_unrt_op,                 /* n */
 +      /* o: This will often be registered as 'Off' at init time */
 +      NULL,                           /* o */
 +      &sysrq_showregs_op,             /* p */
 +      &sysrq_show_timers_op,          /* q */
 +      &sysrq_unraw_op,                /* r */
 +      &sysrq_sync_op,                 /* s */
 +      &sysrq_showstate_op,            /* t */
 +      &sysrq_mountro_op,              /* u */
 +      /* v: May be registered for frame buffer console restore */
 +      NULL,                           /* v */
 +      &sysrq_showstate_blocked_op,    /* w */
 +      /* x: May be registered on ppc/powerpc for xmon */
 +      NULL,                           /* x */
 +      /* y: May be registered on sparc64 for global register dump */
 +      NULL,                           /* y */
 +      &sysrq_ftrace_dump_op,          /* z */
 +};
 +
 +/* key2index calculation, -1 on invalid index */
 +static int sysrq_key_table_key2index(int key)
 +{
 +      int retval;
 +
 +      if ((key >= '0') && (key <= '9'))
 +              retval = key - '0';
 +      else if ((key >= 'a') && (key <= 'z'))
 +              retval = key + 10 - 'a';
 +      else
 +              retval = -1;
 +      return retval;
 +}
 +
 +/*
 + * get and put functions for the table, exposed to modules.
 + */
 +struct sysrq_key_op *__sysrq_get_key_op(int key)
 +{
 +        struct sysrq_key_op *op_p = NULL;
 +        int i;
 +
 +      i = sysrq_key_table_key2index(key);
 +      if (i != -1)
 +              op_p = sysrq_key_table[i];
 +
 +        return op_p;
 +}
 +
 +static void __sysrq_put_key_op(int key, struct sysrq_key_op *op_p)
 +{
 +        int i = sysrq_key_table_key2index(key);
 +
 +        if (i != -1)
 +                sysrq_key_table[i] = op_p;
 +}
 +
 +void __handle_sysrq(int key, bool check_mask)
 +{
 +      struct sysrq_key_op *op_p;
 +      int orig_log_level;
 +      int i;
 +      unsigned long flags;
 +
 +      spin_lock_irqsave(&sysrq_key_table_lock, flags);
 +      /*
 +       * Raise the apparent loglevel to maximum so that the sysrq header
 +       * is shown to provide the user with positive feedback.  We do not
 +       * simply emit this at KERN_EMERG as that would change message
 +       * routing in the consumers of /proc/kmsg.
 +       */
 +      orig_log_level = console_loglevel;
 +      console_loglevel = 7;
 +      printk(KERN_INFO "SysRq : ");
 +
 +        op_p = __sysrq_get_key_op(key);
 +        if (op_p) {
 +              /*
 +               * Should we check for enabled operations (/proc/sysrq-trigger
 +               * should not) and is the invoked operation enabled?
 +               */
 +              if (!check_mask || sysrq_on_mask(op_p->enable_mask)) {
 +                      printk("%s\n", op_p->action_msg);
 +                      console_loglevel = orig_log_level;
 +                      op_p->handler(key);
 +              } else {
 +                      printk("This sysrq operation is disabled.\n");
 +              }
 +      } else {
 +              printk("HELP : ");
 +              /* Only print the help msg once per handler */
 +              for (i = 0; i < ARRAY_SIZE(sysrq_key_table); i++) {
 +                      if (sysrq_key_table[i]) {
 +                              int j;
 +
 +                              for (j = 0; sysrq_key_table[i] !=
 +                                              sysrq_key_table[j]; j++)
 +                                      ;
 +                              if (j != i)
 +                                      continue;
 +                              printk("%s ", sysrq_key_table[i]->help_msg);
 +                      }
 +              }
 +              printk("\n");
 +              console_loglevel = orig_log_level;
 +      }
 +      spin_unlock_irqrestore(&sysrq_key_table_lock, flags);
 +}
 +
 +void handle_sysrq(int key)
 +{
 +      if (sysrq_on())
 +              __handle_sysrq(key, true);
 +}
 +EXPORT_SYMBOL(handle_sysrq);
 +
 +#ifdef CONFIG_INPUT
 +
 +/* Simple translation table for the SysRq keys */
 +static const unsigned char sysrq_xlate[KEY_CNT] =
 +        "\000\0331234567890-=\177\t"                    /* 0x00 - 0x0f */
 +        "qwertyuiop[]\r\000as"                          /* 0x10 - 0x1f */
 +        "dfghjkl;'`\000\\zxcv"                          /* 0x20 - 0x2f */
 +        "bnm,./\000*\000 \000\201\202\203\204\205"      /* 0x30 - 0x3f */
 +        "\206\207\210\211\212\000\000789-456+1"         /* 0x40 - 0x4f */
 +        "230\177\000\000\213\214\000\000\000\000\000\000\000\000\000\000" /* 0x50 - 0x5f */
 +        "\r\000/";                                      /* 0x60 - 0x6f */
 +
 +struct sysrq_state {
 +      struct input_handle handle;
 +      struct work_struct reinject_work;
 +      unsigned long key_down[BITS_TO_LONGS(KEY_CNT)];
 +      unsigned int alt;
 +      unsigned int alt_use;
 +      bool active;
 +      bool need_reinject;
 +};
 +
 +static void sysrq_reinject_alt_sysrq(struct work_struct *work)
 +{
 +      struct sysrq_state *sysrq =
 +                      container_of(work, struct sysrq_state, reinject_work);
 +      struct input_handle *handle = &sysrq->handle;
 +      unsigned int alt_code = sysrq->alt_use;
 +
 +      if (sysrq->need_reinject) {
 +              /* Simulate press and release of Alt + SysRq */
 +              input_inject_event(handle, EV_KEY, alt_code, 1);
 +              input_inject_event(handle, EV_KEY, KEY_SYSRQ, 1);
 +              input_inject_event(handle, EV_SYN, SYN_REPORT, 1);
 +
 +              input_inject_event(handle, EV_KEY, KEY_SYSRQ, 0);
 +              input_inject_event(handle, EV_KEY, alt_code, 0);
 +              input_inject_event(handle, EV_SYN, SYN_REPORT, 1);
 +      }
 +}
 +
 +static bool sysrq_filter(struct input_handle *handle,
 +                       unsigned int type, unsigned int code, int value)
 +{
 +      struct sysrq_state *sysrq = handle->private;
 +      bool was_active = sysrq->active;
 +      bool suppress;
 +
 +      switch (type) {
 +
 +      case EV_SYN:
 +              suppress = false;
 +              break;
 +
 +      case EV_KEY:
 +              switch (code) {
 +
 +              case KEY_LEFTALT:
 +              case KEY_RIGHTALT:
 +                      if (!value) {
 +                              /* One of ALTs is being released */
 +                              if (sysrq->active && code == sysrq->alt_use)
 +                                      sysrq->active = false;
 +
 +                              sysrq->alt = KEY_RESERVED;
 +
 +                      } else if (value != 2) {
 +                              sysrq->alt = code;
 +                              sysrq->need_reinject = false;
 +                      }
 +                      break;
 +
 +              case KEY_SYSRQ:
 +                      if (value == 1 && sysrq->alt != KEY_RESERVED) {
 +                              sysrq->active = true;
 +                              sysrq->alt_use = sysrq->alt;
 +                              /*
 +                               * If nothing else will be pressed we'll need
 +                               * to * re-inject Alt-SysRq keysroke.
 +                               */
 +                              sysrq->need_reinject = true;
 +                      }
 +
 +                      /*
 +                       * Pretend that sysrq was never pressed at all. This
 +                       * is needed to properly handle KGDB which will try
 +                       * to release all keys after exiting debugger. If we
 +                       * do not clear key bit it KGDB will end up sending
 +                       * release events for Alt and SysRq, potentially
 +                       * triggering print screen function.
 +                       */
 +                      if (sysrq->active)
 +                              clear_bit(KEY_SYSRQ, handle->dev->key);
 +
 +                      break;
 +
 +              default:
 +                      if (sysrq->active && value && value != 2) {
 +                              sysrq->need_reinject = false;
 +                              __handle_sysrq(sysrq_xlate[code], true);
 +                      }
 +                      break;
 +              }
 +
 +              suppress = sysrq->active;
 +
 +              if (!sysrq->active) {
 +                      /*
 +                       * If we are not suppressing key presses keep track of
 +                       * keyboard state so we can release keys that have been
 +                       * pressed before entering SysRq mode.
 +                       */
 +                      if (value)
 +                              set_bit(code, sysrq->key_down);
 +                      else
 +                              clear_bit(code, sysrq->key_down);
 +
 +                      if (was_active)
 +                              schedule_work(&sysrq->reinject_work);
 +
 +              } else if (value == 0 &&
 +                         test_and_clear_bit(code, sysrq->key_down)) {
 +                      /*
 +                       * Pass on release events for keys that was pressed before
 +                       * entering SysRq mode.
 +                       */
 +                      suppress = false;
 +              }
 +              break;
 +
 +      default:
 +              suppress = sysrq->active;
 +              break;
 +      }
 +
 +      return suppress;
 +}
 +
 +static int sysrq_connect(struct input_handler *handler,
 +                       struct input_dev *dev,
 +                       const struct input_device_id *id)
 +{
 +      struct sysrq_state *sysrq;
 +      int error;
 +
 +      sysrq = kzalloc(sizeof(struct sysrq_state), GFP_KERNEL);
 +      if (!sysrq)
 +              return -ENOMEM;
 +
 +      INIT_WORK(&sysrq->reinject_work, sysrq_reinject_alt_sysrq);
 +
 +      sysrq->handle.dev = dev;
 +      sysrq->handle.handler = handler;
 +      sysrq->handle.name = "sysrq";
 +      sysrq->handle.private = sysrq;
 +
 +      error = input_register_handle(&sysrq->handle);
 +      if (error) {
 +              pr_err("Failed to register input sysrq handler, error %d\n",
 +                      error);
 +              goto err_free;
 +      }
 +
 +      error = input_open_device(&sysrq->handle);
 +      if (error) {
 +              pr_err("Failed to open input device, error %d\n", error);
 +              goto err_unregister;
 +      }
 +
 +      return 0;
 +
 + err_unregister:
 +      input_unregister_handle(&sysrq->handle);
 + err_free:
 +      kfree(sysrq);
 +      return error;
 +}
 +
 +static void sysrq_disconnect(struct input_handle *handle)
 +{
 +      struct sysrq_state *sysrq = handle->private;
 +
 +      input_close_device(handle);
 +      cancel_work_sync(&sysrq->reinject_work);
 +      input_unregister_handle(handle);
 +      kfree(sysrq);
 +}
 +
 +/*
 + * We are matching on KEY_LEFTALT instead of KEY_SYSRQ because not all
 + * keyboards have SysRq key predefined and so user may add it to keymap
 + * later, but we expect all such keyboards to have left alt.
 + */
 +static const struct input_device_id sysrq_ids[] = {
 +      {
 +              .flags = INPUT_DEVICE_ID_MATCH_EVBIT |
 +                              INPUT_DEVICE_ID_MATCH_KEYBIT,
 +              .evbit = { BIT_MASK(EV_KEY) },
 +              .keybit = { BIT_MASK(KEY_LEFTALT) },
 +      },
 +      { },
 +};
 +
 +static struct input_handler sysrq_handler = {
 +      .filter         = sysrq_filter,
 +      .connect        = sysrq_connect,
 +      .disconnect     = sysrq_disconnect,
 +      .name           = "sysrq",
 +      .id_table       = sysrq_ids,
 +};
 +
 +static bool sysrq_handler_registered;
 +
 +static inline void sysrq_register_handler(void)
 +{
 +      int error;
 +
 +      error = input_register_handler(&sysrq_handler);
 +      if (error)
 +              pr_err("Failed to register input handler, error %d", error);
 +      else
 +              sysrq_handler_registered = true;
 +}
 +
 +static inline void sysrq_unregister_handler(void)
 +{
 +      if (sysrq_handler_registered) {
 +              input_unregister_handler(&sysrq_handler);
 +              sysrq_handler_registered = false;
 +      }
 +}
 +
 +#else
 +
 +static inline void sysrq_register_handler(void)
 +{
 +}
 +
 +static inline void sysrq_unregister_handler(void)
 +{
 +}
 +
 +#endif /* CONFIG_INPUT */
 +
 +int sysrq_toggle_support(int enable_mask)
 +{
 +      bool was_enabled = sysrq_on();
 +
 +      sysrq_enabled = enable_mask;
 +
 +      if (was_enabled != sysrq_on()) {
 +              if (sysrq_on())
 +                      sysrq_register_handler();
 +              else
 +                      sysrq_unregister_handler();
 +      }
 +
 +      return 0;
 +}
 +
 +static int __sysrq_swap_key_ops(int key, struct sysrq_key_op *insert_op_p,
 +                                struct sysrq_key_op *remove_op_p)
 +{
 +      int retval;
 +      unsigned long flags;
 +
 +      spin_lock_irqsave(&sysrq_key_table_lock, flags);
 +      if (__sysrq_get_key_op(key) == remove_op_p) {
 +              __sysrq_put_key_op(key, insert_op_p);
 +              retval = 0;
 +      } else {
 +              retval = -1;
 +      }
 +      spin_unlock_irqrestore(&sysrq_key_table_lock, flags);
 +      return retval;
 +}
 +
 +int register_sysrq_key(int key, struct sysrq_key_op *op_p)
 +{
 +      return __sysrq_swap_key_ops(key, op_p, NULL);
 +}
 +EXPORT_SYMBOL(register_sysrq_key);
 +
 +int unregister_sysrq_key(int key, struct sysrq_key_op *op_p)
 +{
 +      return __sysrq_swap_key_ops(key, NULL, op_p);
 +}
 +EXPORT_SYMBOL(unregister_sysrq_key);
 +
 +#ifdef CONFIG_PROC_FS
 +/*
 + * writing 'C' to /proc/sysrq-trigger is like sysrq-C
 + */
 +static ssize_t write_sysrq_trigger(struct file *file, const char __user *buf,
 +                                 size_t count, loff_t *ppos)
 +{
 +      if (count) {
 +              char c;
 +
 +              if (get_user(c, buf))
 +                      return -EFAULT;
 +              __handle_sysrq(c, false);
 +      }
 +
 +      return count;
 +}
 +
 +static const struct file_operations proc_sysrq_trigger_operations = {
 +      .write          = write_sysrq_trigger,
 +      .llseek         = noop_llseek,
 +};
 +
 +static void sysrq_init_procfs(void)
 +{
 +      if (!proc_create("sysrq-trigger", S_IWUSR, NULL,
 +                       &proc_sysrq_trigger_operations))
 +              pr_err("Failed to register proc interface\n");
 +}
 +
 +#else
 +
 +static inline void sysrq_init_procfs(void)
 +{
 +}
 +
 +#endif /* CONFIG_PROC_FS */
 +
 +static int __init sysrq_init(void)
 +{
 +      sysrq_init_procfs();
 +
 +      if (sysrq_on())
 +              sysrq_register_handler();
 +
 +      return 0;
 +}
 +module_init(sysrq_init);
diff --cc kernel/sysctl.c
Simple merge