Merge branch 'next' into for-linus
authorDmitry Torokhov <dmitry.torokhov@gmail.com>
Sat, 19 Mar 2011 06:38:50 +0000 (23:38 -0700)
committerDmitry Torokhov <dmitry.torokhov@gmail.com>
Sat, 19 Mar 2011 06:38:50 +0000 (23:38 -0700)
1  2 
drivers/input/gameport/gameport.c
drivers/input/input.c
drivers/input/mouse/bcm5974.c
drivers/input/tablet/wacom_sys.c
drivers/input/touchscreen/ads7846.c
drivers/tty/sysrq.c
include/linux/input/matrix_keypad.h

Simple merge
Simple merge
Simple merge
Simple merge
Simple merge
index 0000000000000000000000000000000000000000,8e0dd254eb11cf3d9f503c7bda32bcb3e75ba09d..81f13958e751095d869a2935460163d8111f4098
mode 000000,100644..100644
--- /dev/null
@@@ -1,0 -1,890 +1,905 @@@
 -                               * to * re-inject Alt-SysRq keysroke.
+ /*
+  *    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;
++      bool reinjecting;
+ };
+ 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) {
++              /* we do not want the assignment to be reordered */
++              sysrq->reinjecting = true;
++              mb();
++
+               /* 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);
++
++              mb();
++              sysrq->reinjecting = false;
+       }
+ }
+ 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;
++      /*
++       * Do not filter anything if we are in the process of re-injecting
++       * Alt+SysRq combination.
++       */
++      if (sysrq->reinjecting)
++              return false;
++
+       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);
Simple merge