PM: Remove legacy PM
authorPavel Machek <pavel@suse.cz>
Thu, 13 Mar 2008 22:52:49 +0000 (23:52 +0100)
committerLen Brown <len.brown@intel.com>
Tue, 15 Apr 2008 07:19:07 +0000 (03:19 -0400)
AFAICT pm_send_all is a nop when noone uses pm_register...

Hmm.. can we just force CONFIG_PM_LEGACY=n, and see what happens?

Or maybe this is better idea? It may break build somewhere, but it
should be easy to fix... (it builds here, i386 and x86-64).

Signed-off-by: Pavel Machek <pavel@suse.cz>
Acked-by: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Len Brown <len.brown@intel.com>
arch/frv/kernel/pm.c
arch/mips/au1000/common/power.c
arch/x86/kernel/apm_32.c
kernel/power/Kconfig
kernel/power/Makefile
kernel/power/pm.c [deleted file]

index c57ce3f..73f3aee 100644 (file)
@@ -163,14 +163,11 @@ static int sysctl_pm_do_suspend(ctl_table *ctl, int write, struct file *filp,
        if ((mode != 1) && (mode != 5))
                return -EINVAL;
 
-       retval = pm_send_all(PM_SUSPEND, (void *)3);
-
        if (retval == 0) {
                if (mode == 5)
                    retval = pm_do_bus_sleep();
                else
                    retval = pm_do_suspend();
-               pm_send_all(PM_RESUME, (void *)0);
        }
 
        return retval;
@@ -183,9 +180,6 @@ static int try_set_cmode(int new_cmode)
        if (!(clock_cmodes_permitted & (1<<new_cmode)))
                return -EINVAL;
 
-       /* tell all the drivers we're suspending */
-       pm_send_all(PM_SUSPEND, (void *)3);
-
        /* now change cmode */
        local_irq_disable();
        frv_dma_pause_all();
@@ -201,8 +195,6 @@ static int try_set_cmode(int new_cmode)
        frv_dma_resume_all();
        local_irq_enable();
 
-       /* tell all the drivers we're resuming */
-       pm_send_all(PM_RESUME, (void *)0);
        return 0;
 }
 
index 54047d6..a9f7f63 100644 (file)
@@ -258,7 +258,6 @@ int au_sleep(void)
 static int pm_do_sleep(ctl_table * ctl, int write, struct file *file,
                       void __user *buffer, size_t * len, loff_t *ppos)
 {
-       int retval = 0;
 #ifdef SLEEP_TEST_TIMEOUT
 #define TMPBUFLEN2 16
        char buf[TMPBUFLEN2], *p;
@@ -278,33 +277,21 @@ static int pm_do_sleep(ctl_table * ctl, int write, struct file *file,
                p = buf;
                sleep_ticks = simple_strtoul(p, &p, 0);
 #endif
-               retval = pm_send_all(PM_SUSPEND, (void *) 2);
-
-               if (retval)
-                       return retval;
 
                au_sleep();
-               retval = pm_send_all(PM_RESUME, (void *) 0);
        }
-       return retval;
+       return 0;
 }
 
 static int pm_do_suspend(ctl_table * ctl, int write, struct file *file,
                         void __user *buffer, size_t * len, loff_t *ppos)
 {
-       int retval = 0;
-
        if (!write) {
                *len = 0;
        } else {
-               retval = pm_send_all(PM_SUSPEND, (void *) 2);
-               if (retval)
-                       return retval;
                suspend_mode = 1;
-
-               retval = pm_send_all(PM_RESUME, (void *) 0);
        }
-       return retval;
+       return 0;
 }
 
 
index d4438ef..d7e92bf 100644 (file)
@@ -1189,19 +1189,6 @@ static int suspend(int vetoable)
        int err;
        struct apm_user *as;
 
-       if (pm_send_all(PM_SUSPEND, (void *)3)) {
-               /* Vetoed */
-               if (vetoable) {
-                       if (apm_info.connection_version > 0x100)
-                               set_system_power_state(APM_STATE_REJECT);
-                       err = -EBUSY;
-                       ignore_sys_suspend = 0;
-                       printk(KERN_WARNING "apm: suspend was vetoed.\n");
-                       goto out;
-               }
-               printk(KERN_CRIT "apm: suspend was vetoed, but suspending anyway.\n");
-       }
-
        device_suspend(PMSG_SUSPEND);
        local_irq_disable();
        device_power_down(PMSG_SUSPEND);
@@ -1224,7 +1211,6 @@ static int suspend(int vetoable)
        device_power_up();
        local_irq_enable();
        device_resume();
-       pm_send_all(PM_RESUME, (void *)0);
        queue_event(APM_NORMAL_RESUME, NULL);
  out:
        spin_lock(&user_list_lock);
@@ -1337,7 +1323,6 @@ static void check_events(void)
                        if ((event != APM_NORMAL_RESUME)
                            || (ignore_normal_resume == 0)) {
                                device_resume();
-                               pm_send_all(PM_RESUME, (void *)0);
                                queue_event(event, NULL);
                        }
                        ignore_normal_resume = 0;
index 6233f3b..b45da40 100644 (file)
@@ -19,16 +19,6 @@ config PM
          will issue the hlt instruction if nothing is to be done, thereby
          sending the processor to sleep and saving power.
 
-config PM_LEGACY
-       bool "Legacy Power Management API (DEPRECATED)"
-       depends on PM
-       default n
-       ---help---
-          Support for pm_register() and friends.  This old API is obsoleted
-          by the driver model.
-
-          If unsure, say N.
-
 config PM_DEBUG
        bool "Power Management Debug Support"
        depends on PM
index f7dfff2..597823b 100644 (file)
@@ -4,7 +4,6 @@ EXTRA_CFLAGS    +=      -DDEBUG
 endif
 
 obj-y                          := main.o
-obj-$(CONFIG_PM_LEGACY)                += pm.o
 obj-$(CONFIG_PM_SLEEP)         += process.o console.o
 obj-$(CONFIG_HIBERNATION)      += swsusp.o disk.o snapshot.o swap.o user.o
 
diff --git a/kernel/power/pm.c b/kernel/power/pm.c
deleted file mode 100644 (file)
index 60c73fa..0000000
+++ /dev/null
@@ -1,205 +0,0 @@
-/*
- *  pm.c - Power management interface
- *
- *  Copyright (C) 2000 Andrew Henroid
- *
- *  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 of the License, 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.
- *
- *  You should have received a copy of the GNU General Public License
- *  along with this program; if not, write to the Free Software
- *  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
- */
-#include <linux/init.h>
-#include <linux/module.h>
-#include <linux/spinlock.h>
-#include <linux/mm.h>
-#include <linux/slab.h>
-#include <linux/pm.h>
-#include <linux/pm_legacy.h>
-#include <linux/interrupt.h>
-#include <linux/mutex.h>
-
-/*
- *     Locking notes:
- *             pm_devs_lock can be a semaphore providing pm ops are not called
- *     from an interrupt handler (already a bad idea so no change here). Each
- *     change must be protected so that an unlink of an entry doesn't clash
- *     with a pm send - which is permitted to sleep in the current architecture
- *
- *     Module unloads clashing with pm events now work out safely, the module 
- *     unload path will block until the event has been sent. It may well block
- *     until a resume but that will be fine.
- */
-static DEFINE_MUTEX(pm_devs_lock);
-static LIST_HEAD(pm_devs);
-
-/**
- *     pm_register - register a device with power management
- *     @type: device type 
- *     @id: device ID
- *     @callback: callback function
- *
- *     Add a device to the list of devices that wish to be notified about
- *     power management events. A &pm_dev structure is returned on success,
- *     on failure the return is %NULL.
- *
- *      The callback function will be called in process context and
- *      it may sleep.
- */
-struct pm_dev *pm_register(pm_dev_t type,
-                          unsigned long id,
-                          pm_callback callback)
-{
-       struct pm_dev *dev = kzalloc(sizeof(struct pm_dev), GFP_KERNEL);
-       if (dev) {
-               dev->type = type;
-               dev->id = id;
-               dev->callback = callback;
-
-               mutex_lock(&pm_devs_lock);
-               list_add(&dev->entry, &pm_devs);
-               mutex_unlock(&pm_devs_lock);
-       }
-       return dev;
-}
-
-/**
- *     pm_send - send request to a single device
- *     @dev: device to send to
- *     @rqst: power management request
- *     @data: data for the callback
- *
- *     Issue a power management request to a given device. The 
- *     %PM_SUSPEND and %PM_RESUME events are handled specially. The
- *     data field must hold the intended next state. No call is made
- *     if the state matches.
- *
- *     BUGS: what stops two power management requests occurring in parallel
- *     and conflicting.
- *
- *     WARNING: Calling pm_send directly is not generally recommended, in
- *     particular there is no locking against the pm_dev going away. The
- *     caller must maintain all needed locking or have 'inside knowledge'
- *     on the safety. Also remember that this function is not locked against
- *     pm_unregister. This means that you must handle SMP races on callback
- *     execution and unload yourself.
- */
-static int pm_send(struct pm_dev *dev, pm_request_t rqst, void *data)
-{
-       int status = 0;
-       unsigned long prev_state, next_state;
-
-       if (in_interrupt())
-               BUG();
-
-       switch (rqst) {
-       case PM_SUSPEND:
-       case PM_RESUME:
-               prev_state = dev->state;
-               next_state = (unsigned long) data;
-               if (prev_state != next_state) {
-                       if (dev->callback)
-                               status = (*dev->callback)(dev, rqst, data);
-                       if (!status) {
-                               dev->state = next_state;
-                               dev->prev_state = prev_state;
-                       }
-               }
-               else {
-                       dev->prev_state = prev_state;
-               }
-               break;
-       default:
-               if (dev->callback)
-                       status = (*dev->callback)(dev, rqst, data);
-               break;
-       }
-       return status;
-}
-
-/*
- * Undo incomplete request
- */
-static void pm_undo_all(struct pm_dev *last)
-{
-       struct list_head *entry = last->entry.prev;
-       while (entry != &pm_devs) {
-               struct pm_dev *dev = list_entry(entry, struct pm_dev, entry);
-               if (dev->state != dev->prev_state) {
-                       /* previous state was zero (running) resume or
-                        * previous state was non-zero (suspended) suspend
-                        */
-                       pm_request_t undo = (dev->prev_state
-                                            ? PM_SUSPEND:PM_RESUME);
-                       pm_send(dev, undo, (void*) dev->prev_state);
-               }
-               entry = entry->prev;
-       }
-}
-
-/**
- *     pm_send_all - send request to all managed devices
- *     @rqst: power management request
- *     @data: data for the callback
- *
- *     Issue a power management request to a all devices. The 
- *     %PM_SUSPEND events are handled specially. Any device is 
- *     permitted to fail a suspend by returning a non zero (error)
- *     value from its callback function. If any device vetoes a 
- *     suspend request then all other devices that have suspended 
- *     during the processing of this request are restored to their
- *     previous state.
- *
- *     WARNING:  This function takes the pm_devs_lock. The lock is not dropped until
- *     the callbacks have completed. This prevents races against pm locking
- *     functions, races against module unload pm_unregister code. It does
- *     mean however that you must not issue pm_ functions within the callback
- *     or you will deadlock and users will hate you.
- *
- *     Zero is returned on success. If a suspend fails then the status
- *     from the device that vetoes the suspend is returned.
- *
- *     BUGS: what stops two power management requests occurring in parallel
- *     and conflicting.
- */
-int pm_send_all(pm_request_t rqst, void *data)
-{
-       struct list_head *entry;
-       
-       mutex_lock(&pm_devs_lock);
-       entry = pm_devs.next;
-       while (entry != &pm_devs) {
-               struct pm_dev *dev = list_entry(entry, struct pm_dev, entry);
-               if (dev->callback) {
-                       int status = pm_send(dev, rqst, data);
-                       if (status) {
-                               /* return devices to previous state on
-                                * failed suspend request
-                                */
-                               if (rqst == PM_SUSPEND)
-                                       pm_undo_all(dev);
-                               mutex_unlock(&pm_devs_lock);
-                               return status;
-                       }
-               }
-               entry = entry->next;
-       }
-       mutex_unlock(&pm_devs_lock);
-       return 0;
-}
-
-EXPORT_SYMBOL(pm_register);
-EXPORT_SYMBOL(pm_send_all);
-