[PATCH] swsusp: low level interface

Introduce the low level interface that can be used for handling the
snapshot of the system memory by the in-kernel swap-writing/reading code of
swsusp and the userland interface code (to be introduced shortly).

Also change the way in which swsusp records the allocated swap pages and,
consequently, simplifies the in-kernel swap-writing/reading code (this is
necessary for the userland interface too).  To this end, it introduces two
helper functions in mm/swapfile.c, so that the swsusp code does not refer
directly to the swap internals.
Signed-off-by: NRafael J. Wysocki <rjw@sisk.pl>
Acked-by: NPavel Machek <pavel@ucw.cz>
Signed-off-by: NAdrian Bunk <bunk@stusta.de>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

include/linux/swap.h

@@ -234,14 +234,15 @@ extern struct page * read_swap_cache_async(swp_entry_t, struct vm_area_struct *v
 /* linux/mm/swapfile.c */
 extern long total_swap_pages;
 extern unsigned int nr_swapfiles;
-extern struct swap_info_struct swap_info[];
 extern void si_swapinfo(struct sysinfo *);
 extern swp_entry_t get_swap_page(void);
-extern swp_entry_t get_swap_page_of_type(int type);
+extern swp_entry_t get_swap_page_of_type(int);
 extern int swap_duplicate(swp_entry_t);
 extern int valid_swaphandles(swp_entry_t, unsigned long *);
 extern void swap_free(swp_entry_t);
 extern void free_swap_and_cache(swp_entry_t);
+extern int swap_type_of(dev_t);
+extern unsigned int count_swap_pages(int, int);
 extern sector_t map_swap_page(struct swap_info_struct *, pgoff_t);
 extern struct swap_info_struct *get_swap_info_struct(unsigned);
 extern int can_share_swap_page(struct page *);

kernel/power/disk.c

@@ -26,9 +26,9 @@ extern suspend_disk_method_t pm_disk_mode;
 
 extern int swsusp_shrink_memory(void);
 extern int swsusp_suspend(void);
-extern int swsusp_write(struct pbe *pblist, unsigned int nr_pages);
+extern int swsusp_write(void);
 extern int swsusp_check(void);
-extern int swsusp_read(struct pbe **pblist_ptr);
+extern int swsusp_read(void);
 extern void swsusp_close(void);
 extern int swsusp_resume(void);
 
@@ -70,10 +70,6 @@ static void power_down(suspend_disk_method_t mode)
 	while(1);
 }
 
-
-static int in_suspend __nosavedata = 0;
-
-
 static inline void platform_finish(void)
 {
 	if (pm_disk_mode == PM_DISK_PLATFORM) {
@@ -145,7 +141,7 @@ int pm_suspend_disk(void)
 	if (in_suspend) {
 		device_resume();
 		pr_debug("PM: writing image.\n");
-		error = swsusp_write(pagedir_nosave, nr_copy_pages);
+		error = swsusp_write();
 		if (!error)
 			power_down(pm_disk_mode);
 		else {
@@ -216,7 +212,7 @@ static int software_resume(void)
 
 	pr_debug("PM: Reading swsusp image.\n");
 
-	if ((error = swsusp_read(&pagedir_nosave))) {
+	if ((error = swsusp_read())) {
 		swsusp_free();
 		goto Thaw;
 	}

kernel/power/power.h

@@ -37,21 +37,31 @@ extern struct subsystem power_subsys;
 /* References to section boundaries */
 extern const void __nosave_begin, __nosave_end;
 
-extern unsigned int nr_copy_pages;
 extern struct pbe *pagedir_nosave;
 
 /* Preferred image size in bytes (default 500 MB) */
 extern unsigned long image_size;
 
+extern int in_suspend;
+
 extern asmlinkage int swsusp_arch_suspend(void);
 extern asmlinkage int swsusp_arch_resume(void);
 
 extern unsigned int count_data_pages(void);
-extern void free_pagedir(struct pbe *pblist);
-extern void release_eaten_pages(void);
-extern struct pbe *alloc_pagedir(unsigned nr_pages, gfp_t gfp_mask, int safe_needed);
 extern void swsusp_free(void);
-extern int alloc_data_pages(struct pbe *pblist, gfp_t gfp_mask, int safe_needed);
-extern unsigned int snapshot_nr_pages(void);
-extern struct pbe *snapshot_pblist(void);
-extern void snapshot_pblist_set(struct pbe *pblist);
+
+struct snapshot_handle {
+	loff_t		offset;
+	unsigned int	page;
+	unsigned int	page_offset;
+	unsigned int	prev;
+	struct pbe	*pbe;
+	void		*buffer;
+	unsigned int	buf_offset;
+};
+
+#define data_of(handle)	((handle).buffer + (handle).buf_offset)
+
+extern int snapshot_read_next(struct snapshot_handle *handle, size_t count);
+extern int snapshot_write_next(struct snapshot_handle *handle, size_t count);
+int snapshot_image_loaded(struct snapshot_handle *handle);

kernel/power/snapshot.c

@@ -10,6 +10,7 @@
  */
 
 
+#include <linux/version.h>
 #include <linux/module.h>
 #include <linux/mm.h>
 #include <linux/suspend.h>
@@ -34,7 +35,8 @@
 #include "power.h"
 
 struct pbe *pagedir_nosave;
-unsigned int nr_copy_pages;
+static unsigned int nr_copy_pages;
+static unsigned int nr_meta_pages;
 
 #ifdef CONFIG_HIGHMEM
 unsigned int count_highmem_pages(void)
@@ -235,7 +237,7 @@ static void copy_data_pages(struct pbe *pblist)
  *	free_pagedir - free pages allocated with alloc_pagedir()
  */
 
-void free_pagedir(struct pbe *pblist)
+static void free_pagedir(struct pbe *pblist)
 {
 	struct pbe *pbe;
 
@@ -301,7 +303,7 @@ struct eaten_page {
 
 static struct eaten_page *eaten_pages = NULL;
 
-void release_eaten_pages(void)
+static void release_eaten_pages(void)
 {
 	struct eaten_page *p, *q;
 
@@ -376,7 +378,6 @@ struct pbe *alloc_pagedir(unsigned int nr_pages, gfp_t gfp_mask, int safe_needed
 	if (!nr_pages)
 		return NULL;
 
-	pr_debug("alloc_pagedir(): nr_pages = %d\n", nr_pages);
 	pblist = alloc_image_page(gfp_mask, safe_needed);
 	/* FIXME: rewrite this ugly loop */
 	for (pbe = pblist, num = PBES_PER_PAGE; pbe && num < nr_pages;
@@ -414,6 +415,9 @@ void swsusp_free(void)
 				}
 			}
 	}
+	nr_copy_pages = 0;
+	nr_meta_pages = 0;
+	pagedir_nosave = NULL;
 }
 
 
@@ -437,7 +441,7 @@ static int enough_free_mem(unsigned int nr_pages)
 		(nr_pages + PBES_PER_PAGE - 1) / PBES_PER_PAGE);
 }
 
-int alloc_data_pages(struct pbe *pblist, gfp_t gfp_mask, int safe_needed)
+static int alloc_data_pages(struct pbe *pblist, gfp_t gfp_mask, int safe_needed)
 {
 	struct pbe *p;
 
@@ -504,7 +508,319 @@ asmlinkage int swsusp_save(void)
 	 */
 
 	nr_copy_pages = nr_pages;
+	nr_meta_pages = (nr_pages * sizeof(long) + PAGE_SIZE - 1) >> PAGE_SHIFT;
 
 	printk("swsusp: critical section/: done (%d pages copied)\n", nr_pages);
 	return 0;
 }
+
+static void init_header(struct swsusp_info *info)
+{
+	memset(info, 0, sizeof(struct swsusp_info));
+	info->version_code = LINUX_VERSION_CODE;
+	info->num_physpages = num_physpages;
+	memcpy(&info->uts, &system_utsname, sizeof(system_utsname));
+	info->cpus = num_online_cpus();
+	info->image_pages = nr_copy_pages;
+	info->pages = nr_copy_pages + nr_meta_pages + 1;
+}
+
+/**
+ *	pack_orig_addresses - the .orig_address fields of the PBEs from the
+ *	list starting at @pbe are stored in the array @buf[] (1 page)
+ */
+
+static inline struct pbe *pack_orig_addresses(unsigned long *buf, struct pbe *pbe)
+{
+	int j;
+
+	for (j = 0; j < PAGE_SIZE / sizeof(long) && pbe; j++) {
+		buf[j] = pbe->orig_address;
+		pbe = pbe->next;
+	}
+	if (!pbe)
+		for (; j < PAGE_SIZE / sizeof(long); j++)
+			buf[j] = 0;
+	return pbe;
+}
+
+/**
+ *	snapshot_read_next - used for reading the system memory snapshot.
+ *
+ *	On the first call to it @handle should point to a zeroed
+ *	snapshot_handle structure.  The structure gets updated and a pointer
+ *	to it should be passed to this function every next time.
+ *
+ *	The @count parameter should contain the number of bytes the caller
+ *	wants to read from the snapshot.  It must not be zero.
+ *
+ *	On success the function returns a positive number.  Then, the caller
+ *	is allowed to read up to the returned number of bytes from the memory
+ *	location computed by the data_of() macro.  The number returned
+ *	may be smaller than @count, but this only happens if the read would
+ *	cross a page boundary otherwise.
+ *
+ *	The function returns 0 to indicate the end of data stream condition,
+ *	and a negative number is returned on error.  In such cases the
+ *	structure pointed to by @handle is not updated and should not be used
+ *	any more.
+ */
+
+int snapshot_read_next(struct snapshot_handle *handle, size_t count)
+{
+	static unsigned long *buffer;
+
+	if (handle->page > nr_meta_pages + nr_copy_pages)
+		return 0;
+	if (!buffer) {
+		/* This makes the buffer be freed by swsusp_free() */
+		buffer = alloc_image_page(GFP_ATOMIC, 0);
+		if (!buffer)
+			return -ENOMEM;
+	}
+	if (!handle->offset) {
+		init_header((struct swsusp_info *)buffer);
+		handle->buffer = buffer;
+		handle->pbe = pagedir_nosave;
+	}
+	if (handle->prev < handle->page) {
+		if (handle->page <= nr_meta_pages) {
+			handle->pbe = pack_orig_addresses(buffer, handle->pbe);
+			if (!handle->pbe)
+				handle->pbe = pagedir_nosave;
+		} else {
+			handle->buffer = (void *)handle->pbe->address;
+			handle->pbe = handle->pbe->next;
+		}
+		handle->prev = handle->page;
+	}
+	handle->buf_offset = handle->page_offset;
+	if (handle->page_offset + count >= PAGE_SIZE) {
+		count = PAGE_SIZE - handle->page_offset;
+		handle->page_offset = 0;
+		handle->page++;
+	} else {
+		handle->page_offset += count;
+	}
+	handle->offset += count;
+	return count;
+}
+
+/**
+ *	mark_unsafe_pages - mark the pages that cannot be used for storing
+ *	the image during resume, because they conflict with the pages that
+ *	had been used before suspend
+ */
+
+static int mark_unsafe_pages(struct pbe *pblist)
+{
+	struct zone *zone;
+	unsigned long zone_pfn;
+	struct pbe *p;
+
+	if (!pblist) /* a sanity check */
+		return -EINVAL;
+
+	/* Clear page flags */
+	for_each_zone (zone) {
+		for (zone_pfn = 0; zone_pfn < zone->spanned_pages; ++zone_pfn)
+			if (pfn_valid(zone_pfn + zone->zone_start_pfn))
+				ClearPageNosaveFree(pfn_to_page(zone_pfn +
+					zone->zone_start_pfn));
+	}
+
+	/* Mark orig addresses */
+	for_each_pbe (p, pblist) {
+		if (virt_addr_valid(p->orig_address))
+			SetPageNosaveFree(virt_to_page(p->orig_address));
+		else
+			return -EFAULT;
+	}
+
+	return 0;
+}
+
+static void copy_page_backup_list(struct pbe *dst, struct pbe *src)
+{
+	/* We assume both lists contain the same number of elements */
+	while (src) {
+		dst->orig_address = src->orig_address;
+		dst = dst->next;
+		src = src->next;
+	}
+}
+
+static int check_header(struct swsusp_info *info)
+{
+	char *reason = NULL;
+
+	if (info->version_code != LINUX_VERSION_CODE)
+		reason = "kernel version";
+	if (info->num_physpages != num_physpages)
+		reason = "memory size";
+	if (strcmp(info->uts.sysname,system_utsname.sysname))
+		reason = "system type";
+	if (strcmp(info->uts.release,system_utsname.release))
+		reason = "kernel release";
+	if (strcmp(info->uts.version,system_utsname.version))
+		reason = "version";
+	if (strcmp(info->uts.machine,system_utsname.machine))
+		reason = "machine";
+	if (reason) {
+		printk(KERN_ERR "swsusp: Resume mismatch: %s\n", reason);
+		return -EPERM;
+	}
+	return 0;
+}
+
+/**
+ *	load header - check the image header and copy data from it
+ */
+
+static int load_header(struct snapshot_handle *handle,
+                              struct swsusp_info *info)
+{
+	int error;
+	struct pbe *pblist;
+
+	error = check_header(info);
+	if (!error) {
+		pblist = alloc_pagedir(info->image_pages, GFP_ATOMIC, 0);
+		if (!pblist)
+			return -ENOMEM;
+		pagedir_nosave = pblist;
+		handle->pbe = pblist;
+		nr_copy_pages = info->image_pages;
+		nr_meta_pages = info->pages - info->image_pages - 1;
+	}
+	return error;
+}
+
+/**
+ *	unpack_orig_addresses - copy the elements of @buf[] (1 page) to
+ *	the PBEs in the list starting at @pbe
+ */
+
+static inline struct pbe *unpack_orig_addresses(unsigned long *buf,
+                                                struct pbe *pbe)
+{
+	int j;
+
+	for (j = 0; j < PAGE_SIZE / sizeof(long) && pbe; j++) {
+		pbe->orig_address = buf[j];
+		pbe = pbe->next;
+	}
+	return pbe;
+}
+
+/**
+ *	create_image - use metadata contained in the PBE list
+ *	pointed to by pagedir_nosave to mark the pages that will
+ *	be overwritten in the process of restoring the system
+ *	memory state from the image and allocate memory for
+ *	the image avoiding these pages
+ */
+
+static int create_image(struct snapshot_handle *handle)
+{
+	int error = 0;
+	struct pbe *p, *pblist;
+
+	p = pagedir_nosave;
+	error = mark_unsafe_pages(p);
+	if (!error) {
+		pblist = alloc_pagedir(nr_copy_pages, GFP_ATOMIC, 1);
+		if (pblist)
+			copy_page_backup_list(pblist, p);
+		free_pagedir(p);
+		if (!pblist)
+			error = -ENOMEM;
+	}
+	if (!error)
+		error = alloc_data_pages(pblist, GFP_ATOMIC, 1);
+	if (!error) {
+		release_eaten_pages();
+		pagedir_nosave = pblist;
+	} else {
+		pagedir_nosave = NULL;
+		handle->pbe = NULL;
+		nr_copy_pages = 0;
+		nr_meta_pages = 0;
+	}
+	return error;
+}
+
+/**
+ *	snapshot_write_next - used for writing the system memory snapshot.
+ *
+ *	On the first call to it @handle should point to a zeroed
+ *	snapshot_handle structure.  The structure gets updated and a pointer
+ *	to it should be passed to this function every next time.
+ *
+ *	The @count parameter should contain the number of bytes the caller
+ *	wants to write to the image.  It must not be zero.
+ *
+ *	On success the function returns a positive number.  Then, the caller
+ *	is allowed to write up to the returned number of bytes to the memory
+ *	location computed by the data_of() macro.  The number returned
+ *	may be smaller than @count, but this only happens if the write would
+ *	cross a page boundary otherwise.
+ *
+ *	The function returns 0 to indicate the "end of file" condition,
+ *	and a negative number is returned on error.  In such cases the
+ *	structure pointed to by @handle is not updated and should not be used
+ *	any more.
+ */
+
+int snapshot_write_next(struct snapshot_handle *handle, size_t count)
+{
+	static unsigned long *buffer;
+	int error = 0;
+
+	if (handle->prev && handle->page > nr_meta_pages + nr_copy_pages)
+		return 0;
+	if (!buffer) {
+		/* This makes the buffer be freed by swsusp_free() */
+		buffer = alloc_image_page(GFP_ATOMIC, 0);
+		if (!buffer)
+			return -ENOMEM;
+	}
+	if (!handle->offset)
+		handle->buffer = buffer;
+	if (handle->prev < handle->page) {
+		if (!handle->prev) {
+			error = load_header(handle, (struct swsusp_info *)buffer);
+			if (error)
+				return error;
+		} else if (handle->prev <= nr_meta_pages) {
+			handle->pbe = unpack_orig_addresses(buffer, handle->pbe);
+			if (!handle->pbe) {
+				error = create_image(handle);
+				if (error)
+					return error;
+				handle->pbe = pagedir_nosave;
+				handle->buffer = (void *)handle->pbe->address;
+			}
+		} else {
+			handle->pbe = handle->pbe->next;
+			handle->buffer = (void *)handle->pbe->address;
+		}
+		handle->prev = handle->page;
+	}
+	handle->buf_offset = handle->page_offset;
+	if (handle->page_offset + count >= PAGE_SIZE) {
+		count = PAGE_SIZE - handle->page_offset;
+		handle->page_offset = 0;
+		handle->page++;
+	} else {
+		handle->page_offset += count;
+	}
+	handle->offset += count;
+	return count;
+}
+
+int snapshot_image_loaded(struct snapshot_handle *handle)
+{
+	return !(!handle->pbe || handle->pbe->next || !nr_copy_pages ||
+		handle->page <= nr_meta_pages + nr_copy_pages);
+}

mm/swapfile.c

@@ -45,7 +45,7 @@ static const char Unused_offset[] = "Unused swap offset entry ";
 
 struct swap_list_t swap_list = {-1, -1};
 
-struct swap_info_struct swap_info[MAX_SWAPFILES];
+static struct swap_info_struct swap_info[MAX_SWAPFILES];
 
 static DEFINE_MUTEX(swapon_mutex);
 
@@ -417,6 +417,59 @@ void free_swap_and_cache(swp_entry_t entry)
 	}
 }
 
+#ifdef CONFIG_SOFTWARE_SUSPEND
+/*
+ * Find the swap type that corresponds to given device (if any)
+ *
+ * This is needed for software suspend and is done in such a way that inode
+ * aliasing is allowed.
+ */
+int swap_type_of(dev_t device)
+{
+	int i;
+
+	if (!device)
+		return -EINVAL;
+	spin_lock(&swap_lock);
+	for (i = 0; i < nr_swapfiles; i++) {
+		struct inode *inode;
+
+		if (!(swap_info[i].flags & SWP_WRITEOK))
+			continue;
+		inode = swap_info->swap_file->f_dentry->d_inode;
+		if (S_ISBLK(inode->i_mode) &&
+		    device == MKDEV(imajor(inode), iminor(inode))) {
+			spin_unlock(&swap_lock);
+			return i;
+		}
+	}
+	spin_unlock(&swap_lock);
+	return -ENODEV;
+}
+
+/*
+ * Return either the total number of swap pages of given type, or the number
+ * of free pages of that type (depending on @free)
+ *
+ * This is needed for software suspend
+ */
+unsigned int count_swap_pages(int type, int free)
+{
+	unsigned int n = 0;
+
+	if (type < nr_swapfiles) {
+		spin_lock(&swap_lock);
+		if (swap_info[type].flags & SWP_WRITEOK) {
+			n = swap_info[type].pages;
+			if (free)
+				n -= swap_info[type].inuse_pages;
+		}
+		spin_unlock(&swap_lock);
+	}
+	return n;
+}
+#endif
+
 /*
  * No need to decide whether this PTE shares the swap entry with others,
  * just let do_wp_page work it out if a write is requested later - to