mirror of
https://github.com/AuxXxilium/linux_dsm_epyc7002.git
synced 2024-12-06 13:16:39 +07:00
879129d208
If there are several snapshots sharing an origin and one is removed while the origin is being written to, the snapshot's mempool may get deleted while elements are still referenced. Prior to dm-snapshot-use-per-device-mempools.patch the pending exceptions may still have been referenced after the snapshot was destroyed, but this was not a problem because the shared mempool was still there. This patch fixes the problem by tracking the number of mempool elements in use. The scenario: - You have an origin and two snapshots 1 and 2. - Someone writes to the origin. - It creates two exceptions in the snapshots, snapshot 1 will be primary exception, snapshot 2's pending_exception->primary_pe will point to the exception in snapshot 1. - The exceptions are being relocated, relocation of exception 1 finishes (but it's pending_exception is still allocated, because it is referenced by an exception from snapshot 2) - The user lvremoves snapshot 1 --- it calls just suspend (does nothing) and destructor. md->pending is zero (there is no I/O submitted to the snapshot by md layer), so it won't help us. - The destructor waits for kcopyd jobs to finish on snapshot 1 --- but there are none. - The destructor on snapshot 1 cleans up everything. - The relocation of exception on snapshot 2 finishes, it drops reference on primary_pe. This frees its primary_pe pointer. Primary_pe points to pending exception created for snapshot 1. So it frees memory into non-existing mempool. Signed-off-by: Mikulas Patocka <mpatocka@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
231 lines
5.1 KiB
C
231 lines
5.1 KiB
C
/*
|
|
* dm-snapshot.c
|
|
*
|
|
* Copyright (C) 2001-2002 Sistina Software (UK) Limited.
|
|
*
|
|
* This file is released under the GPL.
|
|
*/
|
|
|
|
#ifndef DM_SNAPSHOT_H
|
|
#define DM_SNAPSHOT_H
|
|
|
|
#include <linux/device-mapper.h>
|
|
#include "dm-bio-list.h"
|
|
#include <linux/blkdev.h>
|
|
#include <linux/workqueue.h>
|
|
|
|
struct exception_table {
|
|
uint32_t hash_mask;
|
|
unsigned hash_shift;
|
|
struct list_head *table;
|
|
};
|
|
|
|
/*
|
|
* The snapshot code deals with largish chunks of the disk at a
|
|
* time. Typically 32k - 512k.
|
|
*/
|
|
typedef sector_t chunk_t;
|
|
|
|
/*
|
|
* An exception is used where an old chunk of data has been
|
|
* replaced by a new one.
|
|
* If chunk_t is 64 bits in size, the top 8 bits of new_chunk hold the number
|
|
* of chunks that follow contiguously. Remaining bits hold the number of the
|
|
* chunk within the device.
|
|
*/
|
|
struct dm_snap_exception {
|
|
struct list_head hash_list;
|
|
|
|
chunk_t old_chunk;
|
|
chunk_t new_chunk;
|
|
};
|
|
|
|
/*
|
|
* Funtions to manipulate consecutive chunks
|
|
*/
|
|
# if defined(CONFIG_LBD) || (BITS_PER_LONG == 64)
|
|
# define DM_CHUNK_CONSECUTIVE_BITS 8
|
|
# define DM_CHUNK_NUMBER_BITS 56
|
|
|
|
static inline chunk_t dm_chunk_number(chunk_t chunk)
|
|
{
|
|
return chunk & (chunk_t)((1ULL << DM_CHUNK_NUMBER_BITS) - 1ULL);
|
|
}
|
|
|
|
static inline unsigned dm_consecutive_chunk_count(struct dm_snap_exception *e)
|
|
{
|
|
return e->new_chunk >> DM_CHUNK_NUMBER_BITS;
|
|
}
|
|
|
|
static inline void dm_consecutive_chunk_count_inc(struct dm_snap_exception *e)
|
|
{
|
|
e->new_chunk += (1ULL << DM_CHUNK_NUMBER_BITS);
|
|
|
|
BUG_ON(!dm_consecutive_chunk_count(e));
|
|
}
|
|
|
|
# else
|
|
# define DM_CHUNK_CONSECUTIVE_BITS 0
|
|
|
|
static inline chunk_t dm_chunk_number(chunk_t chunk)
|
|
{
|
|
return chunk;
|
|
}
|
|
|
|
static inline unsigned dm_consecutive_chunk_count(struct dm_snap_exception *e)
|
|
{
|
|
return 0;
|
|
}
|
|
|
|
static inline void dm_consecutive_chunk_count_inc(struct dm_snap_exception *e)
|
|
{
|
|
}
|
|
|
|
# endif
|
|
|
|
/*
|
|
* Abstraction to handle the meta/layout of exception stores (the
|
|
* COW device).
|
|
*/
|
|
struct exception_store {
|
|
|
|
/*
|
|
* Destroys this object when you've finished with it.
|
|
*/
|
|
void (*destroy) (struct exception_store *store);
|
|
|
|
/*
|
|
* The target shouldn't read the COW device until this is
|
|
* called.
|
|
*/
|
|
int (*read_metadata) (struct exception_store *store);
|
|
|
|
/*
|
|
* Find somewhere to store the next exception.
|
|
*/
|
|
int (*prepare_exception) (struct exception_store *store,
|
|
struct dm_snap_exception *e);
|
|
|
|
/*
|
|
* Update the metadata with this exception.
|
|
*/
|
|
void (*commit_exception) (struct exception_store *store,
|
|
struct dm_snap_exception *e,
|
|
void (*callback) (void *, int success),
|
|
void *callback_context);
|
|
|
|
/*
|
|
* The snapshot is invalid, note this in the metadata.
|
|
*/
|
|
void (*drop_snapshot) (struct exception_store *store);
|
|
|
|
/*
|
|
* Return how full the snapshot is.
|
|
*/
|
|
void (*fraction_full) (struct exception_store *store,
|
|
sector_t *numerator,
|
|
sector_t *denominator);
|
|
|
|
struct dm_snapshot *snap;
|
|
void *context;
|
|
};
|
|
|
|
#define DM_TRACKED_CHUNK_HASH_SIZE 16
|
|
#define DM_TRACKED_CHUNK_HASH(x) ((unsigned long)(x) & \
|
|
(DM_TRACKED_CHUNK_HASH_SIZE - 1))
|
|
|
|
struct dm_snapshot {
|
|
struct rw_semaphore lock;
|
|
struct dm_target *ti;
|
|
|
|
struct dm_dev *origin;
|
|
struct dm_dev *cow;
|
|
|
|
/* List of snapshots per Origin */
|
|
struct list_head list;
|
|
|
|
/* Size of data blocks saved - must be a power of 2 */
|
|
chunk_t chunk_size;
|
|
chunk_t chunk_mask;
|
|
chunk_t chunk_shift;
|
|
|
|
/* You can't use a snapshot if this is 0 (e.g. if full) */
|
|
int valid;
|
|
|
|
/* Origin writes don't trigger exceptions until this is set */
|
|
int active;
|
|
|
|
/* Used for display of table */
|
|
char type;
|
|
|
|
mempool_t *pending_pool;
|
|
|
|
atomic_t pending_exceptions_count;
|
|
|
|
struct exception_table pending;
|
|
struct exception_table complete;
|
|
|
|
/*
|
|
* pe_lock protects all pending_exception operations and access
|
|
* as well as the snapshot_bios list.
|
|
*/
|
|
spinlock_t pe_lock;
|
|
|
|
/* The on disk metadata handler */
|
|
struct exception_store store;
|
|
|
|
struct dm_kcopyd_client *kcopyd_client;
|
|
|
|
/* Queue of snapshot writes for ksnapd to flush */
|
|
struct bio_list queued_bios;
|
|
struct work_struct queued_bios_work;
|
|
|
|
/* Chunks with outstanding reads */
|
|
mempool_t *tracked_chunk_pool;
|
|
spinlock_t tracked_chunk_lock;
|
|
struct hlist_head tracked_chunk_hash[DM_TRACKED_CHUNK_HASH_SIZE];
|
|
};
|
|
|
|
/*
|
|
* Used by the exception stores to load exceptions hen
|
|
* initialising.
|
|
*/
|
|
int dm_add_exception(struct dm_snapshot *s, chunk_t old, chunk_t new);
|
|
|
|
/*
|
|
* Constructor and destructor for the default persistent
|
|
* store.
|
|
*/
|
|
int dm_create_persistent(struct exception_store *store);
|
|
|
|
int dm_create_transient(struct exception_store *store);
|
|
|
|
/*
|
|
* Return the number of sectors in the device.
|
|
*/
|
|
static inline sector_t get_dev_size(struct block_device *bdev)
|
|
{
|
|
return bdev->bd_inode->i_size >> SECTOR_SHIFT;
|
|
}
|
|
|
|
static inline chunk_t sector_to_chunk(struct dm_snapshot *s, sector_t sector)
|
|
{
|
|
return (sector & ~s->chunk_mask) >> s->chunk_shift;
|
|
}
|
|
|
|
static inline sector_t chunk_to_sector(struct dm_snapshot *s, chunk_t chunk)
|
|
{
|
|
return chunk << s->chunk_shift;
|
|
}
|
|
|
|
static inline int bdev_equal(struct block_device *lhs, struct block_device *rhs)
|
|
{
|
|
/*
|
|
* There is only ever one instance of a particular block
|
|
* device so we can compare pointers safely.
|
|
*/
|
|
return lhs == rhs;
|
|
}
|
|
|
|
#endif
|