2005-04-16 15:20:36 -07:00
|
|
|
/*
|
|
|
|
* linux/kernel/power/swsusp.c
|
|
|
|
*
|
2005-10-30 15:59:58 -07:00
|
|
|
* This file provides code to write suspend image to swap and read it back.
|
2005-04-16 15:20:36 -07:00
|
|
|
*
|
|
|
|
* Copyright (C) 1998-2001 Gabor Kuti <seasons@fornax.hu>
|
2005-10-30 15:59:56 -07:00
|
|
|
* Copyright (C) 1998,2001-2005 Pavel Machek <pavel@suse.cz>
|
2005-04-16 15:20:36 -07:00
|
|
|
*
|
|
|
|
* This file is released under the GPLv2.
|
|
|
|
*
|
|
|
|
* I'd like to thank the following people for their work:
|
2005-06-25 14:55:12 -07:00
|
|
|
*
|
2005-04-16 15:20:36 -07:00
|
|
|
* Pavel Machek <pavel@ucw.cz>:
|
|
|
|
* Modifications, defectiveness pointing, being with me at the very beginning,
|
|
|
|
* suspend to swap space, stop all tasks. Port to 2.4.18-ac and 2.5.17.
|
|
|
|
*
|
2005-06-25 14:55:12 -07:00
|
|
|
* Steve Doddi <dirk@loth.demon.co.uk>:
|
2005-04-16 15:20:36 -07:00
|
|
|
* Support the possibility of hardware state restoring.
|
|
|
|
*
|
|
|
|
* Raph <grey.havens@earthling.net>:
|
|
|
|
* Support for preserving states of network devices and virtual console
|
|
|
|
* (including X and svgatextmode)
|
|
|
|
*
|
|
|
|
* Kurt Garloff <garloff@suse.de>:
|
|
|
|
* Straightened the critical function in order to prevent compilers from
|
|
|
|
* playing tricks with local variables.
|
|
|
|
*
|
|
|
|
* Andreas Mohr <a.mohr@mailto.de>
|
|
|
|
*
|
|
|
|
* Alex Badea <vampire@go.ro>:
|
|
|
|
* Fixed runaway init
|
|
|
|
*
|
2006-01-06 01:13:05 -07:00
|
|
|
* Rafael J. Wysocki <rjw@sisk.pl>
|
2006-03-23 04:00:00 -07:00
|
|
|
* Reworked the freeing of memory and the handling of swap
|
2006-01-06 01:13:05 -07:00
|
|
|
*
|
2005-04-16 15:20:36 -07:00
|
|
|
* More state savers are welcome. Especially for the scsi layer...
|
|
|
|
*
|
|
|
|
* For TODOs,FIXMEs also look in Documentation/power/swsusp.txt
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include <linux/mm.h>
|
|
|
|
#include <linux/suspend.h>
|
|
|
|
#include <linux/spinlock.h>
|
|
|
|
#include <linux/kernel.h>
|
|
|
|
#include <linux/major.h>
|
|
|
|
#include <linux/swap.h>
|
|
|
|
#include <linux/pm.h>
|
|
|
|
#include <linux/swapops.h>
|
|
|
|
#include <linux/bootmem.h>
|
|
|
|
#include <linux/syscalls.h>
|
|
|
|
#include <linux/highmem.h>
|
|
|
|
|
|
|
|
#include "power.h"
|
|
|
|
|
2006-01-06 01:15:56 -07:00
|
|
|
/*
|
2006-02-01 04:05:07 -07:00
|
|
|
* Preferred image size in bytes (tunable via /sys/power/image_size).
|
2006-01-06 01:15:56 -07:00
|
|
|
* When it is set to N, swsusp will do its best to ensure the image
|
2006-02-01 04:05:07 -07:00
|
|
|
* size will not exceed N bytes, but if that is impossible, it will
|
2006-01-06 01:15:56 -07:00
|
|
|
* try to create the smallest image possible.
|
|
|
|
*/
|
2006-02-01 04:05:07 -07:00
|
|
|
unsigned long image_size = 500 * 1024 * 1024;
|
2006-01-06 01:15:56 -07:00
|
|
|
|
2006-03-23 03:59:59 -07:00
|
|
|
int in_suspend __nosavedata = 0;
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
/**
|
2006-03-23 03:59:59 -07:00
|
|
|
* The following functions are used for tracing the allocated
|
|
|
|
* swap pages, so that they can be freed in case of an error.
|
2006-01-06 01:13:05 -07:00
|
|
|
*
|
2006-03-23 03:59:59 -07:00
|
|
|
* The functions operate on a linked bitmap structure defined
|
2006-03-23 04:00:00 -07:00
|
|
|
* in power.h
|
2005-04-16 15:20:36 -07:00
|
|
|
*/
|
2006-01-06 01:13:05 -07:00
|
|
|
|
2006-03-23 04:00:00 -07:00
|
|
|
void free_bitmap(struct bitmap_page *bitmap)
|
2005-04-16 15:20:36 -07:00
|
|
|
{
|
2006-03-23 03:59:59 -07:00
|
|
|
struct bitmap_page *bp;
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2006-03-23 03:59:59 -07:00
|
|
|
while (bitmap) {
|
|
|
|
bp = bitmap->next;
|
|
|
|
free_page((unsigned long)bitmap);
|
|
|
|
bitmap = bp;
|
2006-01-06 01:13:05 -07:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2006-03-23 04:00:00 -07:00
|
|
|
struct bitmap_page *alloc_bitmap(unsigned int nr_bits)
|
2006-01-06 01:13:05 -07:00
|
|
|
{
|
2006-03-23 03:59:59 -07:00
|
|
|
struct bitmap_page *bitmap, *bp;
|
|
|
|
unsigned int n;
|
2006-01-06 01:13:05 -07:00
|
|
|
|
2006-03-23 03:59:59 -07:00
|
|
|
if (!nr_bits)
|
2006-01-06 01:13:05 -07:00
|
|
|
return NULL;
|
|
|
|
|
2006-03-23 03:59:59 -07:00
|
|
|
bitmap = (struct bitmap_page *)get_zeroed_page(GFP_KERNEL);
|
|
|
|
bp = bitmap;
|
|
|
|
for (n = BITMAP_PAGE_BITS; n < nr_bits; n += BITMAP_PAGE_BITS) {
|
|
|
|
bp->next = (struct bitmap_page *)get_zeroed_page(GFP_KERNEL);
|
|
|
|
bp = bp->next;
|
|
|
|
if (!bp) {
|
|
|
|
free_bitmap(bitmap);
|
2006-01-06 01:13:05 -07:00
|
|
|
return NULL;
|
|
|
|
}
|
2005-04-16 15:20:36 -07:00
|
|
|
}
|
2006-03-23 03:59:59 -07:00
|
|
|
return bitmap;
|
2005-04-16 15:20:36 -07:00
|
|
|
}
|
|
|
|
|
2006-03-23 03:59:59 -07:00
|
|
|
static int bitmap_set(struct bitmap_page *bitmap, unsigned long bit)
|
2005-04-16 15:20:36 -07:00
|
|
|
{
|
2006-03-23 03:59:59 -07:00
|
|
|
unsigned int n;
|
|
|
|
|
|
|
|
n = BITMAP_PAGE_BITS;
|
|
|
|
while (bitmap && n <= bit) {
|
|
|
|
n += BITMAP_PAGE_BITS;
|
|
|
|
bitmap = bitmap->next;
|
|
|
|
}
|
|
|
|
if (!bitmap)
|
|
|
|
return -EINVAL;
|
|
|
|
n -= BITMAP_PAGE_BITS;
|
|
|
|
bit -= n;
|
|
|
|
n = 0;
|
|
|
|
while (bit >= BITS_PER_CHUNK) {
|
|
|
|
bit -= BITS_PER_CHUNK;
|
|
|
|
n++;
|
2006-01-06 01:13:05 -07:00
|
|
|
}
|
2006-03-23 03:59:59 -07:00
|
|
|
bitmap->chunks[n] |= (1UL << bit);
|
|
|
|
return 0;
|
2006-01-06 01:13:05 -07:00
|
|
|
}
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2006-03-23 04:00:00 -07:00
|
|
|
unsigned long alloc_swap_page(int swap, struct bitmap_page *bitmap)
|
2006-01-06 01:13:05 -07:00
|
|
|
{
|
2006-03-23 03:59:59 -07:00
|
|
|
unsigned long offset;
|
|
|
|
|
|
|
|
offset = swp_offset(get_swap_page_of_type(swap));
|
|
|
|
if (offset) {
|
|
|
|
if (bitmap_set(bitmap, offset)) {
|
|
|
|
swap_free(swp_entry(swap, offset));
|
|
|
|
offset = 0;
|
|
|
|
}
|
2006-01-06 01:13:05 -07:00
|
|
|
}
|
2006-03-23 03:59:59 -07:00
|
|
|
return offset;
|
2006-01-06 01:13:05 -07:00
|
|
|
}
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2006-03-23 04:00:00 -07:00
|
|
|
void free_all_swap_pages(int swap, struct bitmap_page *bitmap)
|
2006-01-06 01:13:05 -07:00
|
|
|
{
|
2006-03-23 03:59:59 -07:00
|
|
|
unsigned int bit, n;
|
|
|
|
unsigned long test;
|
2006-01-06 01:13:05 -07:00
|
|
|
|
2006-03-23 03:59:59 -07:00
|
|
|
bit = 0;
|
|
|
|
while (bitmap) {
|
|
|
|
for (n = 0; n < BITMAP_PAGE_CHUNKS; n++)
|
|
|
|
for (test = 1UL; test; test <<= 1) {
|
|
|
|
if (bitmap->chunks[n] & test)
|
|
|
|
swap_free(swp_entry(swap, bit));
|
|
|
|
bit++;
|
|
|
|
}
|
|
|
|
bitmap = bitmap->next;
|
2005-04-16 15:20:36 -07:00
|
|
|
}
|
2006-01-06 01:13:05 -07:00
|
|
|
}
|
|
|
|
|
2006-01-06 01:13:46 -07:00
|
|
|
/**
|
|
|
|
* swsusp_shrink_memory - Try to free as much memory as needed
|
|
|
|
*
|
|
|
|
* ... but do not OOM-kill anyone
|
|
|
|
*
|
|
|
|
* Notice: all userland should be stopped before it is called, or
|
|
|
|
* livelock is possible.
|
|
|
|
*/
|
|
|
|
|
|
|
|
#define SHRINK_BITE 10000
|
2006-06-23 02:03:18 -07:00
|
|
|
static inline unsigned long __shrink_memory(long tmp)
|
|
|
|
{
|
|
|
|
if (tmp > SHRINK_BITE)
|
|
|
|
tmp = SHRINK_BITE;
|
|
|
|
return shrink_all_memory(tmp);
|
|
|
|
}
|
2006-01-06 01:13:46 -07:00
|
|
|
|
|
|
|
int swsusp_shrink_memory(void)
|
|
|
|
{
|
2006-01-06 01:15:22 -07:00
|
|
|
long size, tmp;
|
2006-01-06 01:13:46 -07:00
|
|
|
struct zone *zone;
|
|
|
|
unsigned long pages = 0;
|
|
|
|
unsigned int i = 0;
|
|
|
|
char *p = "-\\|/";
|
|
|
|
|
|
|
|
printk("Shrinking memory... ");
|
|
|
|
do {
|
2006-06-23 02:04:44 -07:00
|
|
|
size = 2 * count_special_pages();
|
2006-01-06 01:15:22 -07:00
|
|
|
size += size / 50 + count_data_pages();
|
|
|
|
size += (size + PBES_PER_PAGE - 1) / PBES_PER_PAGE +
|
2006-01-06 01:13:46 -07:00
|
|
|
PAGES_FOR_IO;
|
2006-01-06 01:15:22 -07:00
|
|
|
tmp = size;
|
2006-01-06 01:13:46 -07:00
|
|
|
for_each_zone (zone)
|
2006-06-23 02:04:46 -07:00
|
|
|
if (!is_highmem(zone) && populated_zone(zone)) {
|
2006-01-06 01:13:46 -07:00
|
|
|
tmp -= zone->free_pages;
|
2006-06-23 02:04:46 -07:00
|
|
|
tmp += zone->lowmem_reserve[ZONE_NORMAL];
|
|
|
|
}
|
2006-01-06 01:13:46 -07:00
|
|
|
if (tmp > 0) {
|
2006-06-23 02:03:18 -07:00
|
|
|
tmp = __shrink_memory(tmp);
|
2006-01-06 01:13:46 -07:00
|
|
|
if (!tmp)
|
|
|
|
return -ENOMEM;
|
|
|
|
pages += tmp;
|
2006-02-01 04:05:07 -07:00
|
|
|
} else if (size > image_size / PAGE_SIZE) {
|
2006-06-23 02:03:18 -07:00
|
|
|
tmp = __shrink_memory(size - (image_size / PAGE_SIZE));
|
2006-01-06 01:15:22 -07:00
|
|
|
pages += tmp;
|
2006-01-06 01:13:46 -07:00
|
|
|
}
|
|
|
|
printk("\b%c", p[i++%4]);
|
|
|
|
} while (tmp > 0);
|
|
|
|
printk("\bdone (%lu pages freed)\n", pages);
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
int swsusp_suspend(void)
|
|
|
|
{
|
|
|
|
int error;
|
2005-11-08 22:34:41 -07:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
if ((error = arch_prepare_suspend()))
|
|
|
|
return error;
|
|
|
|
local_irq_disable();
|
|
|
|
/* At this point, device_suspend() has been called, but *not*
|
|
|
|
* device_power_down(). We *must* device_power_down() now.
|
|
|
|
* Otherwise, drivers for some devices (e.g. interrupt controllers)
|
|
|
|
* become desynchronized with the actual state of the hardware
|
|
|
|
* at resume time, and evil weirdness ensues.
|
|
|
|
*/
|
|
|
|
if ((error = device_power_down(PMSG_FREEZE))) {
|
2005-09-03 15:57:05 -07:00
|
|
|
printk(KERN_ERR "Some devices failed to power down, aborting suspend\n");
|
2005-11-08 22:34:41 -07:00
|
|
|
goto Enable_irqs;
|
2005-04-16 15:20:36 -07:00
|
|
|
}
|
2005-07-07 17:56:44 -07:00
|
|
|
|
2006-06-23 02:04:44 -07:00
|
|
|
if ((error = save_special_mem())) {
|
2005-11-08 22:34:41 -07:00
|
|
|
printk(KERN_ERR "swsusp: Not enough free pages for highmem\n");
|
|
|
|
goto Restore_highmem;
|
2005-07-07 17:56:44 -07:00
|
|
|
}
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
save_processor_state();
|
|
|
|
if ((error = swsusp_arch_suspend()))
|
2005-09-03 15:57:05 -07:00
|
|
|
printk(KERN_ERR "Error %d suspending\n", error);
|
2005-04-16 15:20:36 -07:00
|
|
|
/* Restore control flow magically appears here */
|
|
|
|
restore_processor_state();
|
2005-11-08 22:34:41 -07:00
|
|
|
Restore_highmem:
|
2006-06-23 02:04:44 -07:00
|
|
|
restore_special_mem();
|
2005-04-16 15:20:36 -07:00
|
|
|
device_power_up();
|
2005-11-08 22:34:41 -07:00
|
|
|
Enable_irqs:
|
2005-04-16 15:20:36 -07:00
|
|
|
local_irq_enable();
|
|
|
|
return error;
|
|
|
|
}
|
|
|
|
|
|
|
|
int swsusp_resume(void)
|
|
|
|
{
|
|
|
|
int error;
|
|
|
|
local_irq_disable();
|
|
|
|
if (device_power_down(PMSG_FREEZE))
|
|
|
|
printk(KERN_ERR "Some devices failed to power down, very bad\n");
|
|
|
|
/* We'll ignore saved state, but this gets preempt count (etc) right */
|
|
|
|
save_processor_state();
|
|
|
|
error = swsusp_arch_resume();
|
|
|
|
/* Code below is only ever reached in case of failure. Otherwise
|
|
|
|
* execution continues at place where swsusp_arch_suspend was called
|
|
|
|
*/
|
|
|
|
BUG_ON(!error);
|
2005-10-30 15:59:58 -07:00
|
|
|
/* The only reason why swsusp_arch_resume() can fail is memory being
|
|
|
|
* very tight, so we have to free it as soon as we can to avoid
|
|
|
|
* subsequent failures
|
|
|
|
*/
|
|
|
|
swsusp_free();
|
2005-04-16 15:20:36 -07:00
|
|
|
restore_processor_state();
|
2006-06-23 02:04:44 -07:00
|
|
|
restore_special_mem();
|
2005-09-06 15:16:27 -07:00
|
|
|
touch_softlockup_watchdog();
|
2005-04-16 15:20:36 -07:00
|
|
|
device_power_up();
|
|
|
|
local_irq_enable();
|
|
|
|
return error;
|
|
|
|
}
|