lvm-snap.c

linux和2410结合开发 用他可以生成2410所需的zImage文件

/*
 * kernel/lvm-snap.c
 *
 * Copyright (C) 2000 Andrea Arcangeli <[e-mail]> SuSE
 *                    Heinz Mauelshagen, Sistina Software (persistent snapshots)
 *
 * LVM snapshot driver 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, or (at your option)
 * any later version.
 * 
 * LVM snapshot driver 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 GNU CC; see the file COPYING.  If not, write to
 * the Free Software Foundation, 59 Temple Place - Suite 330,
 * Boston, MA 02111-1307, USA. 
 *
 */

/*
 * Changelog
 *
 *    05/07/2000 - implemented persistent snapshot support
 *    23/11/2000 - used cpu_to_le64 rather than my own macro
 *    25/01/2001 - Put LockPage back in
 *    01/02/2001 - A dropped snapshot is now set as inactive
 *    12/03/2001 - lvm_pv_get_number changes:
 *                 o made it static
 *                 o renamed it to _pv_get_number
 *                 o pv number is returned in new uint * arg
 *                 o -1 returned on error
 *                 lvm_snapshot_fill_COW_table has a return value too.
 *
 */

#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/vmalloc.h>
#include <linux/blkdev.h>
#include <linux/smp_lock.h>
#include <linux/types.h>
#include <linux/iobuf.h>
#include <linux/lvm.h>


#include "lvm-internal.h"

static char *lvm_snap_version __attribute__ ((unused)) =
   "LVM "LVM_RELEASE_NAME" snapshot code ("LVM_RELEASE_DATE")\n";


extern const char *const lvm_name;
extern int lvm_blocksizes[];

void lvm_snapshot_release(lv_t *);
static int _write_COW_table_block(vg_t *vg, lv_t *lv, int idx,
                                 const char **reason);
static void _disable_snapshot(vg_t *vg, lv_t *lv);


static int _pv_get_number(vg_t * vg, kdev_t rdev, uint *pvn) {
	uint p;
	for(p = 0; p < vg->pv_max; p++) {
		if(vg->pv[p] == NULL)
			continue;

		if(vg->pv[p]->pv_dev == rdev)
			break;

	}

 	if(p >= vg->pv_max) {
		/* bad news, the snapshot COW table is probably corrupt */
		printk(KERN_ERR
		       "%s -- _pv_get_number failed for rdev = %u\n",
		       lvm_name, rdev);
		return -1;
	}

	*pvn = vg->pv[p]->pv_number;
	return 0;
}

#define hashfn(dev,block,mask,chunk_size) \
	((HASHDEV(dev)^((block)/(chunk_size))) & (mask))

static inline lv_block_exception_t *
lvm_find_exception_table(kdev_t org_dev, unsigned long org_start, lv_t * lv)
{
	struct list_head * hash_table = lv->lv_snapshot_hash_table, * next;
	unsigned long mask = lv->lv_snapshot_hash_mask;
	int chunk_size = lv->lv_chunk_size;
	lv_block_exception_t * ret;
	int i = 0;

	hash_table = &hash_table[hashfn(org_dev, org_start, mask, chunk_size)];
	ret = NULL;
	for (next = hash_table->next; next != hash_table; next = next->next)
	{
		lv_block_exception_t * exception;

		exception = list_entry(next, lv_block_exception_t, hash);
		if (exception->rsector_org == org_start &&
		    exception->rdev_org == org_dev)
		{
			if (i)
			{
				/* fun, isn't it? :) */
				list_del(next);
				list_add(next, hash_table);
			}
			ret = exception;
			break;
		}
		i++;
	}
	return ret;
}

inline void lvm_hash_link(lv_block_exception_t * exception,
			  kdev_t org_dev, unsigned long org_start,
			  lv_t * lv)
{
	struct list_head * hash_table = lv->lv_snapshot_hash_table;
	unsigned long mask = lv->lv_snapshot_hash_mask;
	int chunk_size = lv->lv_chunk_size;

	hash_table = &hash_table[hashfn(org_dev, org_start, mask, chunk_size)];
	list_add(&exception->hash, hash_table);
}

int lvm_snapshot_remap_block(kdev_t * org_dev, unsigned long * org_sector,
			     unsigned long pe_start, lv_t * lv)
{
	int ret;
	unsigned long pe_off, pe_adjustment, __org_start;
	kdev_t __org_dev;
	int chunk_size = lv->lv_chunk_size;
	lv_block_exception_t * exception;

	pe_off = pe_start % chunk_size;
	pe_adjustment = (*org_sector-pe_off) % chunk_size;
	__org_start = *org_sector - pe_adjustment;
	__org_dev = *org_dev;
	ret = 0;
	exception = lvm_find_exception_table(__org_dev, __org_start, lv);
	if (exception)
	{
		*org_dev = exception->rdev_new;
		*org_sector = exception->rsector_new + pe_adjustment;
		ret = 1;
	}
	return ret;
}

void lvm_drop_snapshot(vg_t *vg, lv_t *lv_snap, const char *reason)
{
	kdev_t last_dev;
	int i;

	/* no exception storage space available for this snapshot
	   or error on this snapshot --> release it */
	invalidate_buffers(lv_snap->lv_dev);

       /* wipe the snapshot since it's inconsistent now */
       _disable_snapshot(vg, lv_snap);

	for (i = last_dev = 0; i < lv_snap->lv_remap_ptr; i++) {
		if ( lv_snap->lv_block_exception[i].rdev_new != last_dev) {
			last_dev = lv_snap->lv_block_exception[i].rdev_new;
			invalidate_buffers(last_dev);
		}
	}

	lvm_snapshot_release(lv_snap);
	lv_snap->lv_status &= ~LV_ACTIVE;

	printk(KERN_INFO
	       "%s -- giving up to snapshot %s on %s: %s\n",
	       lvm_name, lv_snap->lv_snapshot_org->lv_name, lv_snap->lv_name,
	       reason);
}

static inline int lvm_snapshot_prepare_blocks(unsigned long *blocks,
					      unsigned long start,
					      int nr_sectors,
					      int blocksize)
{
	int i, sectors_per_block, nr_blocks;

	sectors_per_block = blocksize / SECTOR_SIZE;

	if(start & (sectors_per_block - 1))
		return 0;

	nr_blocks = nr_sectors / sectors_per_block;
	start /= sectors_per_block;

	for (i = 0; i < nr_blocks; i++)
		blocks[i] = start++;

	return 1;
}

inline int lvm_get_blksize(kdev_t dev)
{
	int correct_size = BLOCK_SIZE, i, major;

	major = MAJOR(dev);
	if (blksize_size[major])
	{
		i = blksize_size[major][MINOR(dev)];
		if (i)
			correct_size = i;
	}
	return correct_size;
}

#ifdef DEBUG_SNAPSHOT
static inline void invalidate_snap_cache(unsigned long start, unsigned long nr,
					 kdev_t dev)
{
	struct buffer_head * bh;
	int sectors_per_block, i, blksize, minor;

	minor = MINOR(dev);
	blksize = lvm_blocksizes[minor];
	sectors_per_block = blksize >> 9;
	nr /= sectors_per_block;
	start /= sectors_per_block;

	for (i = 0; i < nr; i++)
	{
		bh = get_hash_table(dev, start++, blksize);
		if (bh)
			bforget(bh);
	}
}
#endif


int lvm_snapshot_fill_COW_page(vg_t * vg, lv_t * lv_snap)
{
       uint pvn;
       int id = 0, is = lv_snap->lv_remap_ptr;
       ulong blksize_snap;
       lv_COW_table_disk_t * lv_COW_table = (lv_COW_table_disk_t *)
               page_address(lv_snap->lv_COW_table_iobuf->maplist[0]);

       if (is == 0)
               return 0;

	is--;
        blksize_snap =
               lvm_get_blksize(lv_snap->lv_block_exception[is].rdev_new);
        is -= is % (blksize_snap / sizeof(lv_COW_table_disk_t));

	memset(lv_COW_table, 0, blksize_snap);
	for ( ; is < lv_snap->lv_remap_ptr; is++, id++) {
		/* store new COW_table entry */
               lv_block_exception_t *be = lv_snap->lv_block_exception + is;
               if(_pv_get_number(vg, be->rdev_org, &pvn))
                       goto bad;

               lv_COW_table[id].pv_org_number = cpu_to_le64(pvn);
               lv_COW_table[id].pv_org_rsector = cpu_to_le64(be->rsector_org);
               if(_pv_get_number(vg, be->rdev_new, &pvn))
                       goto bad;

               lv_COW_table[id].pv_snap_number = cpu_to_le64(pvn);
               lv_COW_table[id].pv_snap_rsector =
                       cpu_to_le64(be->rsector_new);
	}

       return 0;

 bad:
       printk(KERN_ERR "%s -- lvm_snapshot_fill_COW_page failed", lvm_name);
       return -1;
}


/*
 * writes a COW exception table sector to disk (HM)
 */

int lvm_write_COW_table_block(vg_t * vg, lv_t *lv_snap)
{
	int r;
	const char *err;
	if((r = _write_COW_table_block(vg, lv_snap,
				       lv_snap->lv_remap_ptr - 1, &err)))
		lvm_drop_snapshot(vg, lv_snap, err);
	return r;
}

/*
 * copy on write handler for one snapshot logical volume
 *
 * read the original blocks and store it/them on the new one(s).
 * if there is no exception storage space free any longer --> release snapshot.
 *
 * this routine gets called for each _first_ write to a physical chunk.
 */
int lvm_snapshot_COW(kdev_t org_phys_dev,
		     unsigned long org_phys_sector,
		     unsigned long org_pe_start,
		     unsigned long org_virt_sector,
		     vg_t *vg, lv_t* lv_snap)
{
	const char * reason;
	unsigned long org_start, snap_start, snap_phys_dev, virt_start, pe_off;
	int idx = lv_snap->lv_remap_ptr, chunk_size = lv_snap->lv_chunk_size;
	struct kiobuf * iobuf;
	int blksize_snap, blksize_org, min_blksize, max_blksize;
	int max_sectors, nr_sectors;

	/* check if we are out of snapshot space */
	if (idx >= lv_snap->lv_remap_end)
		goto fail_out_of_space;

	/* calculate physical boundaries of source chunk */
	pe_off = org_pe_start % chunk_size;
	org_start = org_phys_sector - ((org_phys_sector-pe_off) % chunk_size);
	virt_start = org_virt_sector - (org_phys_sector - org_start);

	/* calculate physical boundaries of destination chunk */
	snap_phys_dev = lv_snap->lv_block_exception[idx].rdev_new;
	snap_start = lv_snap->lv_block_exception[idx].rsector_new;

#ifdef DEBUG_SNAPSHOT
	printk(KERN_INFO
	       "%s -- COW: "
	       "org %s faulting %lu start %lu, snap %s start %lu, "
	       "size %d, pe_start %lu pe_off %lu, virt_sec %lu\n",
	       lvm_name,
	       kdevname(org_phys_dev), org_phys_sector, org_start,
	       kdevname(snap_phys_dev), snap_start,
	       chunk_size,