ubd_kern.c

来自「linux 内核源代码」· C语言 代码 · 共 1,485 行 · 第 1/3 页

/*
 * Copyright (C) 2000 Jeff Dike ([e-mail])
 * Licensed under the GPL
 */

/* 2001-09-28...2002-04-17
 * Partition stuff by [e-mail]
 * old style ubd by setting UBD_SHIFT to 0
 * 2002-09-27...2002-10-18 massive tinkering for 2.5
 * partitions have changed in 2.5
 * 2003-01-29 more tinkering for 2.5.59-1
 * This should now address the sysfs problems and has
 * the symlink for devfs to allow for booting with
 * the common /dev/ubd/discX/... names rather than
 * only /dev/ubdN/discN this version also has lots of
 * clean ups preparing for ubd-many.
 * James McMechan
 */

#define MAJOR_NR UBD_MAJOR
#define UBD_SHIFT 4

#include "linux/kernel.h"
#include "linux/module.h"
#include "linux/blkdev.h"
#include "linux/hdreg.h"
#include "linux/init.h"
#include "linux/cdrom.h"
#include "linux/proc_fs.h"
#include "linux/ctype.h"
#include "linux/capability.h"
#include "linux/mm.h"
#include "linux/vmalloc.h"
#include "linux/blkpg.h"
#include "linux/genhd.h"
#include "linux/spinlock.h"
#include "linux/platform_device.h"
#include "linux/scatterlist.h"
#include "asm/segment.h"
#include "asm/uaccess.h"
#include "asm/irq.h"
#include "asm/types.h"
#include "asm/tlbflush.h"
#include "mem_user.h"
#include "kern_util.h"
#include "kern.h"
#include "mconsole_kern.h"
#include "init.h"
#include "irq_user.h"
#include "irq_kern.h"
#include "ubd_user.h"
#include "os.h"
#include "mem.h"
#include "mem_kern.h"
#include "cow.h"

enum ubd_req { UBD_READ, UBD_WRITE };

struct io_thread_req {
	struct request *req;
	enum ubd_req op;
	int fds[2];
	unsigned long offsets[2];
	unsigned long long offset;
	unsigned long length;
	char *buffer;
	int sectorsize;
	unsigned long sector_mask;
	unsigned long long cow_offset;
	unsigned long bitmap_words[2];
	int error;
};

extern int open_ubd_file(char *file, struct openflags *openflags, int shared,
			 char **backing_file_out, int *bitmap_offset_out,
			 unsigned long *bitmap_len_out, int *data_offset_out,
			 int *create_cow_out);
extern int create_cow_file(char *cow_file, char *backing_file,
			   struct openflags flags, int sectorsize,
			   int alignment, int *bitmap_offset_out,
			   unsigned long *bitmap_len_out,
			   int *data_offset_out);
extern int read_cow_bitmap(int fd, void *buf, int offset, int len);
extern void do_io(struct io_thread_req *req);

static inline int ubd_test_bit(__u64 bit, unsigned char *data)
{
	__u64 n;
	int bits, off;

	bits = sizeof(data[0]) * 8;
	n = bit / bits;
	off = bit % bits;
	return (data[n] & (1 << off)) != 0;
}

static inline void ubd_set_bit(__u64 bit, unsigned char *data)
{
	__u64 n;
	int bits, off;

	bits = sizeof(data[0]) * 8;
	n = bit / bits;
	off = bit % bits;
	data[n] |= (1 << off);
}
/*End stuff from ubd_user.h*/

#define DRIVER_NAME "uml-blkdev"

static DEFINE_MUTEX(ubd_lock);

static int ubd_open(struct inode * inode, struct file * filp);
static int ubd_release(struct inode * inode, struct file * file);
static int ubd_ioctl(struct inode * inode, struct file * file,
		     unsigned int cmd, unsigned long arg);
static int ubd_getgeo(struct block_device *bdev, struct hd_geometry *geo);

#define MAX_DEV (16)

static struct block_device_operations ubd_blops = {
        .owner		= THIS_MODULE,
        .open		= ubd_open,
        .release	= ubd_release,
        .ioctl		= ubd_ioctl,
	.getgeo		= ubd_getgeo,
};

/* Protected by ubd_lock */
static int fake_major = MAJOR_NR;
static struct gendisk *ubd_gendisk[MAX_DEV];
static struct gendisk *fake_gendisk[MAX_DEV];

#ifdef CONFIG_BLK_DEV_UBD_SYNC
#define OPEN_FLAGS ((struct openflags) { .r = 1, .w = 1, .s = 1, .c = 0, \
					 .cl = 1 })
#else
#define OPEN_FLAGS ((struct openflags) { .r = 1, .w = 1, .s = 0, .c = 0, \
					 .cl = 1 })
#endif
static struct openflags global_openflags = OPEN_FLAGS;

struct cow {
	/* backing file name */
	char *file;
	/* backing file fd */
	int fd;
	unsigned long *bitmap;
	unsigned long bitmap_len;
	int bitmap_offset;
	int data_offset;
};

#define MAX_SG 64

struct ubd {
	struct list_head restart;
	/* name (and fd, below) of the file opened for writing, either the
	 * backing or the cow file. */
	char *file;
	int count;
	int fd;
	__u64 size;
	struct openflags boot_openflags;
	struct openflags openflags;
	unsigned shared:1;
	unsigned no_cow:1;
	struct cow cow;
	struct platform_device pdev;
	struct request_queue *queue;
	spinlock_t lock;
	struct scatterlist sg[MAX_SG];
	struct request *request;
	int start_sg, end_sg;
};

#define DEFAULT_COW { \
	.file =			NULL, \
	.fd =			-1,	\
	.bitmap =		NULL, \
	.bitmap_offset =	0, \
	.data_offset =		0, \
}

#define DEFAULT_UBD { \
	.file = 		NULL, \
	.count =		0, \
	.fd =			-1, \
	.size =			-1, \
	.boot_openflags =	OPEN_FLAGS, \
	.openflags =		OPEN_FLAGS, \
	.no_cow =               0, \
	.shared =		0, \
	.cow =			DEFAULT_COW, \
	.lock =			SPIN_LOCK_UNLOCKED,	\
	.request =		NULL, \
	.start_sg =		0, \
	.end_sg =		0, \
}

/* Protected by ubd_lock */
struct ubd ubd_devs[MAX_DEV] = { [ 0 ... MAX_DEV - 1 ] = DEFAULT_UBD };

/* Only changed by fake_ide_setup which is a setup */
static int fake_ide = 0;
static struct proc_dir_entry *proc_ide_root = NULL;
static struct proc_dir_entry *proc_ide = NULL;

static void make_proc_ide(void)
{
	proc_ide_root = proc_mkdir("ide", NULL);
	proc_ide = proc_mkdir("ide0", proc_ide_root);
}

static int proc_ide_read_media(char *page, char **start, off_t off, int count,
			       int *eof, void *data)
{
	int len;

	strcpy(page, "disk\n");
	len = strlen("disk\n");
	len -= off;
	if (len < count){
		*eof = 1;
		if (len <= 0) return 0;
	}
	else len = count;
	*start = page + off;
	return len;
}

static void make_ide_entries(char *dev_name)
{
	struct proc_dir_entry *dir, *ent;
	char name[64];

	if(proc_ide_root == NULL) make_proc_ide();

	dir = proc_mkdir(dev_name, proc_ide);
	if(!dir) return;

	ent = create_proc_entry("media", S_IFREG|S_IRUGO, dir);
	if(!ent) return;
	ent->data = NULL;
	ent->read_proc = proc_ide_read_media;
	ent->write_proc = NULL;
	sprintf(name,"ide0/%s", dev_name);
	proc_symlink(dev_name, proc_ide_root, name);
}

static int fake_ide_setup(char *str)
{
	fake_ide = 1;
	return 1;
}

__setup("fake_ide", fake_ide_setup);

__uml_help(fake_ide_setup,
"fake_ide\n"
"    Create ide0 entries that map onto ubd devices.\n\n"
);

static int parse_unit(char **ptr)
{
	char *str = *ptr, *end;
	int n = -1;

	if(isdigit(*str)) {
		n = simple_strtoul(str, &end, 0);
		if(end == str)
			return -1;
		*ptr = end;
	}
	else if (('a' <= *str) && (*str <= 'z')) {
		n = *str - 'a';
		str++;
		*ptr = str;
	}
	return n;
}

/* If *index_out == -1 at exit, the passed option was a general one;
 * otherwise, the str pointer is used (and owned) inside ubd_devs array, so it
 * should not be freed on exit.
 */
static int ubd_setup_common(char *str, int *index_out, char **error_out)
{
	struct ubd *ubd_dev;
	struct openflags flags = global_openflags;
	char *backing_file;
	int n, err = 0, i;

	if(index_out) *index_out = -1;
	n = *str;
	if(n == '='){
		char *end;
		int major;

		str++;
		if(!strcmp(str, "sync")){
			global_openflags = of_sync(global_openflags);
			goto out1;
		}

		err = -EINVAL;
		major = simple_strtoul(str, &end, 0);
		if((*end != '\0') || (end == str)){
			*error_out = "Didn't parse major number";
			goto out1;
		}

		mutex_lock(&ubd_lock);
		if(fake_major != MAJOR_NR){
			*error_out = "Can't assign a fake major twice";
			goto out1;
		}

		fake_major = major;

		printk(KERN_INFO "Setting extra ubd major number to %d\n",
		       major);
		err = 0;
	out1:
		mutex_unlock(&ubd_lock);
		return err;
	}

	n = parse_unit(&str);
	if(n < 0){
		*error_out = "Couldn't parse device number";
		return -EINVAL;
	}
	if(n >= MAX_DEV){
		*error_out = "Device number out of range";
		return 1;
	}

	err = -EBUSY;
	mutex_lock(&ubd_lock);

	ubd_dev = &ubd_devs[n];
	if(ubd_dev->file != NULL){
		*error_out = "Device is already configured";
		goto out;
	}

	if (index_out)
		*index_out = n;

	err = -EINVAL;
	for (i = 0; i < sizeof("rscd="); i++) {
		switch (*str) {
		case 'r':
			flags.w = 0;
			break;
		case 's':
			flags.s = 1;
			break;
		case 'd':
			ubd_dev->no_cow = 1;
			break;
		case 'c':
			ubd_dev->shared = 1;
			break;
		case '=':
			str++;
			goto break_loop;
		default:
			*error_out = "Expected '=' or flag letter "
				"(r, s, c, or d)";
			goto out;
		}
		str++;
	}

	if (*str == '=')
		*error_out = "Too many flags specified";
	else
		*error_out = "Missing '='";
	goto out;

break_loop:
	backing_file = strchr(str, ',');

	if (backing_file == NULL)
		backing_file = strchr(str, ':');

	if(backing_file != NULL){
		if(ubd_dev->no_cow){
			*error_out = "Can't specify both 'd' and a cow file";
			goto out;
		}
		else {
			*backing_file = '\0';
			backing_file++;
		}
	}
	err = 0;
	ubd_dev->file = str;
	ubd_dev->cow.file = backing_file;
	ubd_dev->boot_openflags = flags;
out:
	mutex_unlock(&ubd_lock);
	return err;
}

static int ubd_setup(char *str)
{
	char *error;
	int err;

	err = ubd_setup_common(str, NULL, &error);
	if(err)
		printk(KERN_ERR "Failed to initialize device with \"%s\" : "
		       "%s\n", str, error);
	return 1;
}

__setup("ubd", ubd_setup);
__uml_help(ubd_setup,
"ubd<n><flags>=<filename>[(:|,)<filename2>]\n"
"    This is used to associate a device with a file in the underlying\n"
"    filesystem. When specifying two filenames, the first one is the\n"
"    COW name and the second is the backing file name. As separator you can\n"
"    use either a ':' or a ',': the first one allows writing things like;\n"
"	ubd0=~/Uml/root_cow:~/Uml/root_backing_file\n"
"    while with a ',' the shell would not expand the 2nd '~'.\n"
"    When using only one filename, UML will detect whether to treat it like\n"
"    a COW file or a backing file. To override this detection, add the 'd'\n"
"    flag:\n"
"	ubd0d=BackingFile\n"
"    Usually, there is a filesystem in the file, but \n"
"    that's not required. Swap devices containing swap files can be\n"
"    specified like this. Also, a file which doesn't contain a\n"
"    filesystem can have its contents read in the virtual \n"
"    machine by running 'dd' on the device. <n> must be in the range\n"
"    0 to 7. Appending an 'r' to the number will cause that device\n"
"    to be mounted read-only. For example ubd1r=./ext_fs. Appending\n"
"    an 's' will cause data to be written to disk on the host immediately.\n\n"
);

static int udb_setup(char *str)
{
	printk("udb%s specified on command line is almost certainly a ubd -> "
	       "udb TYPO\n", str);
	return 1;
}

__setup("udb", udb_setup);
__uml_help(udb_setup,
"udb\n"
"    This option is here solely to catch ubd -> udb typos, which can be\n"
"    to impossible to catch visually unless you specifically look for\n"
"    them.  The only result of any option starting with 'udb' is an error\n"
"    in the boot output.\n\n"
);

static int fakehd_set = 0;
static int fakehd(char *str)
{
	printk(KERN_INFO "fakehd : Changing ubd name to \"hd\".\n");
	fakehd_set = 1;
	return 1;
}

__setup("fakehd", fakehd);
__uml_help(fakehd,
"fakehd\n"
"    Change the ubd device name to \"hd\".\n\n"
);

static void do_ubd_request(struct request_queue * q);

/* Only changed by ubd_init, which is an initcall. */
int thread_fd = -1;

static void ubd_end_request(struct request *req, int bytes, int uptodate)
{
	if (!end_that_request_first(req, uptodate, bytes >> 9)) {
		struct ubd *dev = req->rq_disk->private_data;
		unsigned long flags;

		add_disk_randomness(req->rq_disk);
		spin_lock_irqsave(&dev->lock, flags);
		end_that_request_last(req, uptodate);
		spin_unlock_irqrestore(&dev->lock, flags);
	}
}

/* Callable only from interrupt context - otherwise you need to do
 * spin_lock_irq()/spin_lock_irqsave() */
static inline void ubd_finish(struct request *req, int bytes)
{
	if(bytes < 0){