ubd_kern.c

来自「h内核」· C语言 代码 · 共 1,656 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/config.h"
#include "linux/module.h"
#include "linux/blkdev.h"
#include "linux/hdreg.h"
#include "linux/init.h"
#include "linux/devfs_fs_kernel.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 "asm/segment.h"
#include "asm/uaccess.h"
#include "asm/irq.h"
#include "asm/types.h"
#include "asm/tlbflush.h"
#include "user_util.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 "2_5compat.h"
#include "os.h"
#include "mem.h"
#include "mem_kern.h"
#include "cow.h"

enum ubd_req { UBD_READ, UBD_WRITE, UBD_MMAP };

struct io_thread_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 map_fd;
	unsigned long long map_offset;
	int error;
};

extern int open_ubd_file(char *file, struct openflags *openflags,
			 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_SPINLOCK(ubd_io_lock);
static DEFINE_SPINLOCK(ubd_lock);

static void (*do_ubd)(void);

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);

#define MAX_DEV (8)

/* Changed in early boot */
static int ubd_do_mmap = 0;
#define UBD_MMAP_BLOCK_SIZE PAGE_SIZE

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

/* Protected by the queue_lock */
static request_queue_t *ubd_queue;

/* 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

/* Not protected - changed only in ubd_setup_common and then only to
 * to enable O_SYNC.
 */
static struct openflags global_openflags = OPEN_FLAGS;

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

struct ubd {
	char *file;
	int count;
	int fd;
	__u64 size;
	struct openflags boot_openflags;
	struct openflags openflags;
	int no_cow;
	struct cow cow;
	struct platform_device pdev;

	int map_writes;
	int map_reads;
	int nomap_writes;
	int nomap_reads;
	int write_maps;
};

#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, \
        .cow =			DEFAULT_COW, \
	.map_writes		= 0, \
	.map_reads		= 0, \
	.nomap_writes		= 0, \
	.nomap_reads		= 0, \
	.write_maps		= 0, \
}

struct ubd ubd_dev[MAX_DEV] = { [ 0 ... MAX_DEV - 1 ] = DEFAULT_UBD };

static int ubd0_init(void)
{
	struct ubd *dev = &ubd_dev[0];

	if(dev->file == NULL)
		dev->file = "root_fs";
	return(0);
}

__initcall(ubd0_init);

/* 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->nlink = 1;
	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 <= 'h')) {
		n = *str - 'a';
		str++;
		*ptr = str;
	}
	return(n);
}

static int ubd_setup_common(char *str, int *index_out)
{
	struct ubd *dev;
	struct openflags flags = global_openflags;
	char *backing_file;
	int n, err, i;

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

		str++;
		if(!strcmp(str, "mmap")){
			CHOOSE_MODE(printk("mmap not supported by the ubd "
					   "driver in tt mode\n"),
				    ubd_do_mmap = 1);
			return(0);
		}

		if(!strcmp(str, "sync")){
			global_openflags = of_sync(global_openflags);
			return(0);
		}
		major = simple_strtoul(str, &end, 0);
		if((*end != '\0') || (end == str)){
			printk(KERN_ERR 
			       "ubd_setup : didn't parse major number\n");
			return(1);
		}

		err = 1;
 		spin_lock(&ubd_lock);
 		if(fake_major != MAJOR_NR){
 			printk(KERN_ERR "Can't assign a fake major twice\n");
 			goto out1;
 		}
 
 		fake_major = major;

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

	n = parse_unit(&str);
	if(n < 0){
		printk(KERN_ERR "ubd_setup : couldn't parse unit number "
		       "'%s'\n", str);
		return(1);
	}
	if(n >= MAX_DEV){
		printk(KERN_ERR "ubd_setup : index %d out of range "
		       "(%d devices, from 0 to %d)\n", n, MAX_DEV, MAX_DEV - 1);
		return(1);
	}

	err = 1;
	spin_lock(&ubd_lock);

	dev = &ubd_dev[n];
	if(dev->file != NULL){
		printk(KERN_ERR "ubd_setup : device already configured\n");
		goto out;
	}

	if (index_out)
		*index_out = n;

	for (i = 0; i < 4; i++) {
		switch (*str) {
		case 'r':
			flags.w = 0;
			break;
		case 's':
			flags.s = 1;
			break;
		case 'd':
			dev->no_cow = 1;
			break;
		case '=':
			str++;
			goto break_loop;
		default:
			printk(KERN_ERR "ubd_setup : Expected '=' or flag letter (r,s or d)\n");
			goto out;
		}
		str++;
	}

        if (*str == '=')
		printk(KERN_ERR "ubd_setup : Too many flags specified\n");
        else
		printk(KERN_ERR "ubd_setup : Expected '='\n");
	goto out;

break_loop:
	err = 0;
	backing_file = strchr(str, ',');

	if (!backing_file) {
		backing_file = strchr(str, ':');
	}

	if(backing_file){
		if(dev->no_cow)
			printk(KERN_ERR "Can't specify both 'd' and a "
			       "cow file\n");
		else {
			*backing_file = '\0';
			backing_file++;
		}
	}
	dev->file = str;
	dev->cow.file = backing_file;
	dev->boot_openflags = flags;
out:
	spin_unlock(&ubd_lock);
	return(err);
}

static int ubd_setup(char *str)
{
	ubd_setup_common(str, NULL);
	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 thread 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 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(request_queue_t * q);

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

/* Changed by ubd_handler, which is serialized because interrupts only
 * happen on CPU 0.
 */
int intr_count = 0;

/* call ubd_finish if you need to serialize */
static void __ubd_finish(struct request *req, int error)
{
	int nsect;

	if(error){
		end_request(req, 0);
		return;
	}
	nsect = req->current_nr_sectors;
	req->sector += nsect;
	req->buffer += nsect << 9;
	req->errors = 0;
	req->nr_sectors -= nsect;
	req->current_nr_sectors = 0;
	end_request(req, 1);
}

static inline void ubd_finish(struct request *req, int error)
{
 	spin_lock(&ubd_io_lock);
	__ubd_finish(req, error);
	spin_unlock(&ubd_io_lock);
}

/* Called without ubd_io_lock held */
static void ubd_handler(void)
{
	struct io_thread_req req;
	struct request *rq = elv_next_request(ubd_queue);
	int n, err;

	do_ubd = NULL;
	intr_count++;
	n = os_read_file(thread_fd, &req, sizeof(req));
	if(n != sizeof(req)){
		printk(KERN_ERR "Pid %d - spurious interrupt in ubd_handler, "
		       "err = %d\n", os_getpid(), -n);
		spin_lock(&ubd_io_lock);
		end_request(rq, 0);
		spin_unlock(&ubd_io_lock);
		return;
	}
        
	if((req.op != UBD_MMAP) &&
	   ((req.offset != ((__u64) (rq->sector)) << 9) ||
	    (req.length != (rq->current_nr_sectors) << 9)))
		panic("I/O op mismatch");
	
	if(req.map_fd != -1){
		err = physmem_subst_mapping(req.buffer, req.map_fd,
					    req.map_offset, 1);
		if(err)
			printk("ubd_handler - physmem_subst_mapping failed, "
			       "err = %d\n", -err);
	}

	ubd_finish(rq, req.error);
	reactivate_fd(thread_fd, UBD_IRQ);	
	do_ubd_request(ubd_queue);
}

static irqreturn_t ubd_intr(int irq, void *dev, struct pt_regs *unused)
{
	ubd_handler();
	return(IRQ_HANDLED);
}

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

void kill_io_thread(void)
{