blob: fcd887703f953eebbfb7cc2e1aee9cd82113e5c9 [file] [log] [blame]
/*
* Copyright (C) 2011-2012 Red Hat UK.
*
* This file is released under the GPL.
*/
#include "dm-thin-metadata.h"
#include "dm-bio-prison-v1.h"
#include "dm.h"
#include <linux/device-mapper.h>
#include <linux/dm-io.h>
#include <linux/dm-kcopyd.h>
#include <linux/jiffies.h>
#include <linux/log2.h>
#include <linux/list.h>
#include <linux/rculist.h>
#include <linux/init.h>
#include <linux/module.h>
#include <linux/slab.h>
#include <linux/vmalloc.h>
#include <linux/sort.h>
#include <linux/rbtree.h>
#define DM_MSG_PREFIX "thin"
/*
* Tunable constants
*/
#define ENDIO_HOOK_POOL_SIZE 1024
#define MAPPING_POOL_SIZE 1024
#define COMMIT_PERIOD HZ
#define NO_SPACE_TIMEOUT_SECS 60
static unsigned no_space_timeout_secs = NO_SPACE_TIMEOUT_SECS;
DECLARE_DM_KCOPYD_THROTTLE_WITH_MODULE_PARM(snapshot_copy_throttle,
"A percentage of time allocated for copy on write");
/*
* The block size of the device holding pool data must be
* between 64KB and 1GB.
*/
#define DATA_DEV_BLOCK_SIZE_MIN_SECTORS (64 * 1024 >> SECTOR_SHIFT)
#define DATA_DEV_BLOCK_SIZE_MAX_SECTORS (1024 * 1024 * 1024 >> SECTOR_SHIFT)
/*
* Device id is restricted to 24 bits.
*/
#define MAX_DEV_ID ((1 << 24) - 1)
/*
* How do we handle breaking sharing of data blocks?
* =================================================
*
* We use a standard copy-on-write btree to store the mappings for the
* devices (note I'm talking about copy-on-write of the metadata here, not
* the data). When you take an internal snapshot you clone the root node
* of the origin btree. After this there is no concept of an origin or a
* snapshot. They are just two device trees that happen to point to the
* same data blocks.
*
* When we get a write in we decide if it's to a shared data block using
* some timestamp magic. If it is, we have to break sharing.
*
* Let's say we write to a shared block in what was the origin. The
* steps are:
*
* i) plug io further to this physical block. (see bio_prison code).
*
* ii) quiesce any read io to that shared data block. Obviously
* including all devices that share this block. (see dm_deferred_set code)
*
* iii) copy the data block to a newly allocate block. This step can be
* missed out if the io covers the block. (schedule_copy).
*
* iv) insert the new mapping into the origin's btree
* (process_prepared_mapping). This act of inserting breaks some
* sharing of btree nodes between the two devices. Breaking sharing only
* effects the btree of that specific device. Btrees for the other
* devices that share the block never change. The btree for the origin
* device as it was after the last commit is untouched, ie. we're using
* persistent data structures in the functional programming sense.
*
* v) unplug io to this physical block, including the io that triggered
* the breaking of sharing.
*
* Steps (ii) and (iii) occur in parallel.
*
* The metadata _doesn't_ need to be committed before the io continues. We
* get away with this because the io is always written to a _new_ block.
* If there's a crash, then:
*
* - The origin mapping will point to the old origin block (the shared
* one). This will contain the data as it was before the io that triggered
* the breaking of sharing came in.
*
* - The snap mapping still points to the old block. As it would after
* the commit.
*
* The downside of this scheme is the timestamp magic isn't perfect, and
* will continue to think that data block in the snapshot device is shared
* even after the write to the origin has broken sharing. I suspect data
* blocks will typically be shared by many different devices, so we're
* breaking sharing n + 1 times, rather than n, where n is the number of
* devices that reference this data block. At the moment I think the
* benefits far, far outweigh the disadvantages.
*/
/*----------------------------------------------------------------*/
/*
* Key building.
*/
enum lock_space {
VIRTUAL,
PHYSICAL
};
static void build_key(struct dm_thin_device *td, enum lock_space ls,
dm_block_t b, dm_block_t e, struct dm_cell_key *key)
{
key->virtual = (ls == VIRTUAL);
key->dev = dm_thin_dev_id(td);
key->block_begin = b;
key->block_end = e;
}
static void build_data_key(struct dm_thin_device *td, dm_block_t b,
struct dm_cell_key *key)
{
build_key(td, PHYSICAL, b, b + 1llu, key);
}
static void build_virtual_key(struct dm_thin_device *td, dm_block_t b,
struct dm_cell_key *key)
{
build_key(td, VIRTUAL, b, b + 1llu, key);
}
/*----------------------------------------------------------------*/
#define THROTTLE_THRESHOLD (1 * HZ)
struct throttle {
struct rw_semaphore lock;
unsigned long threshold;
bool throttle_applied;
};
static void throttle_init(struct throttle *t)
{
init_rwsem(&t->lock);
t->throttle_applied = false;
}
static void throttle_work_start(struct throttle *t)
{
t->threshold = jiffies + THROTTLE_THRESHOLD;
}
static void throttle_work_update(struct throttle *t)
{
if (!t->throttle_applied && jiffies > t->threshold) {
down_write(&t->lock);
t->throttle_applied = true;
}
}
static void throttle_work_complete(struct throttle *t)
{
if (t->throttle_applied) {
t->throttle_applied = false;
up_write(&t->lock);
}
}
static void throttle_lock(struct throttle *t)
{
down_read(&t->lock);
}
static void throttle_unlock(struct throttle *t)
{
up_read(&t->lock);
}
/*----------------------------------------------------------------*/
/*
* A pool device ties together a metadata device and a data device. It
* also provides the interface for creating and destroying internal
* devices.
*/
struct dm_thin_new_mapping;
/*
* The pool runs in various modes. Ordered in degraded order for comparisons.
*/
enum pool_mode {
PM_WRITE, /* metadata may be changed */
PM_OUT_OF_DATA_SPACE, /* metadata may be changed, though data may not be allocated */
/*
* Like READ_ONLY, except may switch back to WRITE on metadata resize. Reported as READ_ONLY.
*/
PM_OUT_OF_METADATA_SPACE,
PM_READ_ONLY, /* metadata may not be changed */
PM_FAIL, /* all I/O fails */
};
struct pool_features {
enum pool_mode mode;
bool zero_new_blocks:1;
bool discard_enabled:1;
bool discard_passdown:1;
bool error_if_no_space:1;
};
struct thin_c;
typedef void (*process_bio_fn)(struct thin_c *tc, struct bio *bio);
typedef void (*process_cell_fn)(struct thin_c *tc, struct dm_bio_prison_cell *cell);
typedef void (*process_mapping_fn)(struct dm_thin_new_mapping *m);
#define CELL_SORT_ARRAY_SIZE 8192
struct pool {
struct list_head list;
struct dm_target *ti; /* Only set if a pool target is bound */
struct mapped_device *pool_md;
struct block_device *md_dev;
struct dm_pool_metadata *pmd;
dm_block_t low_water_blocks;
uint32_t sectors_per_block;
int sectors_per_block_shift;
struct pool_features pf;
bool low_water_triggered:1; /* A dm event has been sent */
bool suspended:1;
bool out_of_data_space:1;
struct dm_bio_prison *prison;
struct dm_kcopyd_client *copier;
struct work_struct worker;
struct workqueue_struct *wq;
struct throttle throttle;
struct delayed_work waker;
struct delayed_work no_space_timeout;
unsigned long last_commit_jiffies;
unsigned ref_count;
spinlock_t lock;
struct bio_list deferred_flush_bios;
struct bio_list deferred_flush_completions;
struct list_head prepared_mappings;
struct list_head prepared_discards;
struct list_head prepared_discards_pt2;
struct list_head active_thins;
struct dm_deferred_set *shared_read_ds;
struct dm_deferred_set *all_io_ds;
struct dm_thin_new_mapping *next_mapping;
process_bio_fn process_bio;
process_bio_fn process_discard;
process_cell_fn process_cell;
process_cell_fn process_discard_cell;
process_mapping_fn process_prepared_mapping;
process_mapping_fn process_prepared_discard;
process_mapping_fn process_prepared_discard_pt2;
struct dm_bio_prison_cell **cell_sort_array;
mempool_t mapping_pool;
};
static void metadata_operation_failed(struct pool *pool, const char *op, int r);
static enum pool_mode get_pool_mode(struct pool *pool)
{
return pool->pf.mode;
}
static void notify_of_pool_mode_change(struct pool *pool)
{
const char *descs[] = {
"write",
"out-of-data-space",
"read-only",
"read-only",
"fail"
};
const char *extra_desc = NULL;
enum pool_mode mode = get_pool_mode(pool);
if (mode == PM_OUT_OF_DATA_SPACE) {
if (!pool->pf.error_if_no_space)
extra_desc = " (queue IO)";
else
extra_desc = " (error IO)";
}
dm_table_event(pool->ti->table);
DMINFO("%s: switching pool to %s%s mode",
dm_device_name(pool->pool_md),
descs[(int)mode], extra_desc ? : "");
}
/*
* Target context for a pool.
*/
struct pool_c {
struct dm_target *ti;
struct pool *pool;
struct dm_dev *data_dev;
struct dm_dev *metadata_dev;
struct dm_target_callbacks callbacks;
dm_block_t low_water_blocks;
struct pool_features requested_pf; /* Features requested during table load */
struct pool_features adjusted_pf; /* Features used after adjusting for constituent devices */
};
/*
* Target context for a thin.
*/
struct thin_c {
struct list_head list;
struct dm_dev *pool_dev;
struct dm_dev *origin_dev;
sector_t origin_size;
dm_thin_id dev_id;
struct pool *pool;
struct dm_thin_device *td;
struct mapped_device *thin_md;
bool requeue_mode:1;
spinlock_t lock;
struct list_head deferred_cells;
struct bio_list deferred_bio_list;
struct bio_list retry_on_resume_list;
struct rb_root sort_bio_list; /* sorted list of deferred bios */
/*
* Ensures the thin is not destroyed until the worker has finished
* iterating the active_thins list.
*/
refcount_t refcount;
struct completion can_destroy;
};
/*----------------------------------------------------------------*/
static bool block_size_is_power_of_two(struct pool *pool)
{
return pool->sectors_per_block_shift >= 0;
}
static sector_t block_to_sectors(struct pool *pool, dm_block_t b)
{
return block_size_is_power_of_two(pool) ?
(b << pool->sectors_per_block_shift) :
(b * pool->sectors_per_block);
}
/*----------------------------------------------------------------*/
struct discard_op {
struct thin_c *tc;
struct blk_plug plug;
struct bio *parent_bio;
struct bio *bio;
};
static void begin_discard(struct discard_op *op, struct thin_c *tc, struct bio *parent)
{
BUG_ON(!parent);
op->tc = tc;
blk_start_plug(&op->plug);
op->parent_bio = parent;
op->bio = NULL;
}
static int issue_discard(struct discard_op *op, dm_block_t data_b, dm_block_t data_e)
{
struct thin_c *tc = op->tc;
sector_t s = block_to_sectors(tc->pool, data_b);
sector_t len = block_to_sectors(tc->pool, data_e - data_b);
return __blkdev_issue_discard(tc->pool_dev->bdev, s, len,
GFP_NOWAIT, 0, &op->bio);
}
static void end_discard(struct discard_op *op, int r)
{
if (op->bio) {
/*
* Even if one of the calls to issue_discard failed, we
* need to wait for the chain to complete.
*/
bio_chain(op->bio, op->parent_bio);
bio_set_op_attrs(op->bio, REQ_OP_DISCARD, 0);
submit_bio(op->bio);
}
blk_finish_plug(&op->plug);
/*
* Even if r is set, there could be sub discards in flight that we
* need to wait for.
*/
if (r && !op->parent_bio->bi_status)
op->parent_bio->bi_status = errno_to_blk_status(r);
bio_endio(op->parent_bio);
}
/*----------------------------------------------------------------*/
/*
* wake_worker() is used when new work is queued and when pool_resume is
* ready to continue deferred IO processing.
*/
static void wake_worker(struct pool *pool)
{
queue_work(pool->wq, &pool->worker);
}
/*----------------------------------------------------------------*/
static int bio_detain(struct pool *pool, struct dm_cell_key *key, struct bio *bio,
struct dm_bio_prison_cell **cell_result)
{
int r;
struct dm_bio_prison_cell *cell_prealloc;
/*
* Allocate a cell from the prison's mempool.
* This might block but it can't fail.
*/
cell_prealloc = dm_bio_prison_alloc_cell(pool->prison, GFP_NOIO);
r = dm_bio_detain(pool->prison, key, bio, cell_prealloc, cell_result);
if (r)
/*
* We reused an old cell; we can get rid of
* the new one.
*/
dm_bio_prison_free_cell(pool->prison, cell_prealloc);
return r;
}
static void cell_release(struct pool *pool,
struct dm_bio_prison_cell *cell,
struct bio_list *bios)
{
dm_cell_release(pool->prison, cell, bios);
dm_bio_prison_free_cell(pool->prison, cell);
}
static void cell_visit_release(struct pool *pool,
void (*fn)(void *, struct dm_bio_prison_cell *),
void *context,
struct dm_bio_prison_cell *cell)
{
dm_cell_visit_release(pool->prison, fn, context, cell);
dm_bio_prison_free_cell(pool->prison, cell);
}
static void cell_release_no_holder(struct pool *pool,
struct dm_bio_prison_cell *cell,
struct bio_list *bios)
{
dm_cell_release_no_holder(pool->prison, cell, bios);
dm_bio_prison_free_cell(pool->prison, cell);
}
static void cell_error_with_code(struct pool *pool,
struct dm_bio_prison_cell *cell, blk_status_t error_code)
{
dm_cell_error(pool->prison, cell, error_code);
dm_bio_prison_free_cell(pool->prison, cell);
}
static blk_status_t get_pool_io_error_code(struct pool *pool)
{
return pool->out_of_data_space ? BLK_STS_NOSPC : BLK_STS_IOERR;
}
static void cell_error(struct pool *pool, struct dm_bio_prison_cell *cell)
{
cell_error_with_code(pool, cell, get_pool_io_error_code(pool));
}
static void cell_success(struct pool *pool, struct dm_bio_prison_cell *cell)
{
cell_error_with_code(pool, cell, 0);
}
static void cell_requeue(struct pool *pool, struct dm_bio_prison_cell *cell)
{
cell_error_with_code(pool, cell, BLK_STS_DM_REQUEUE);
}
/*----------------------------------------------------------------*/
/*
* A global list of pools that uses a struct mapped_device as a key.
*/
static struct dm_thin_pool_table {
struct mutex mutex;
struct list_head pools;
} dm_thin_pool_table;
static void pool_table_init(void)
{
mutex_init(&dm_thin_pool_table.mutex);
INIT_LIST_HEAD(&dm_thin_pool_table.pools);
}
static void pool_table_exit(void)
{
mutex_destroy(&dm_thin_pool_table.mutex);
}
static void __pool_table_insert(struct pool *pool)
{
BUG_ON(!mutex_is_locked(&dm_thin_pool_table.mutex));
list_add(&pool->list, &dm_thin_pool_table.pools);
}
static void __pool_table_remove(struct pool *pool)
{
BUG_ON(!mutex_is_locked(&dm_thin_pool_table.mutex));
list_del(&pool->list);
}
static struct pool *__pool_table_lookup(struct mapped_device *md)
{
struct pool *pool = NULL, *tmp;
BUG_ON(!mutex_is_locked(&dm_thin_pool_table.mutex));
list_for_each_entry(tmp, &dm_thin_pool_table.pools, list) {
if (tmp->pool_md == md) {
pool = tmp;
break;
}
}
return pool;
}
static struct pool *__pool_table_lookup_metadata_dev(struct block_device *md_dev)
{
struct pool *pool = NULL, *tmp;
BUG_ON(!mutex_is_locked(&dm_thin_pool_table.mutex));
list_for_each_entry(tmp, &dm_thin_pool_table.pools, list) {
if (tmp->md_dev == md_dev) {
pool = tmp;
break;
}
}
return pool;
}
/*----------------------------------------------------------------*/
struct dm_thin_endio_hook {
struct thin_c *tc;
struct dm_deferred_entry *shared_read_entry;
struct dm_deferred_entry *all_io_entry;
struct dm_thin_new_mapping *overwrite_mapping;
struct rb_node rb_node;
struct dm_bio_prison_cell *cell;
};
static void __merge_bio_list(struct bio_list *bios, struct bio_list *master)
{
bio_list_merge(bios, master);
bio_list_init(master);
}
static void error_bio_list(struct bio_list *bios, blk_status_t error)
{
struct bio *bio;
while ((bio = bio_list_pop(bios))) {
bio->bi_status = error;
bio_endio(bio);
}
}
static void error_thin_bio_list(struct thin_c *tc, struct bio_list *master,
blk_status_t error)
{
struct bio_list bios;
unsigned long flags;
bio_list_init(&bios);
spin_lock_irqsave(&tc->lock, flags);
__merge_bio_list(&bios, master);
spin_unlock_irqrestore(&tc->lock, flags);
error_bio_list(&bios, error);
}
static void requeue_deferred_cells(struct thin_c *tc)
{
struct pool *pool = tc->pool;
unsigned long flags;
struct list_head cells;
struct dm_bio_prison_cell *cell, *tmp;
INIT_LIST_HEAD(&cells);
spin_lock_irqsave(&tc->lock, flags);
list_splice_init(&tc->deferred_cells, &cells);
spin_unlock_irqrestore(&tc->lock, flags);
list_for_each_entry_safe(cell, tmp, &cells, user_list)
cell_requeue(pool, cell);
}
static void requeue_io(struct thin_c *tc)
{
struct bio_list bios;
unsigned long flags;
bio_list_init(&bios);
spin_lock_irqsave(&tc->lock, flags);
__merge_bio_list(&bios, &tc->deferred_bio_list);
__merge_bio_list(&bios, &tc->retry_on_resume_list);
spin_unlock_irqrestore(&tc->lock, flags);
error_bio_list(&bios, BLK_STS_DM_REQUEUE);
requeue_deferred_cells(tc);
}
static void error_retry_list_with_code(struct pool *pool, blk_status_t error)
{
struct thin_c *tc;
rcu_read_lock();
list_for_each_entry_rcu(tc, &pool->active_thins, list)
error_thin_bio_list(tc, &tc->retry_on_resume_list, error);
rcu_read_unlock();
}
static void error_retry_list(struct pool *pool)
{
error_retry_list_with_code(pool, get_pool_io_error_code(pool));
}
/*
* This section of code contains the logic for processing a thin device's IO.
* Much of the code depends on pool object resources (lists, workqueues, etc)
* but most is exclusively called from the thin target rather than the thin-pool
* target.
*/
static dm_block_t get_bio_block(struct thin_c *tc, struct bio *bio)
{
struct pool *pool = tc->pool;
sector_t block_nr = bio->bi_iter.bi_sector;
if (block_size_is_power_of_two(pool))
block_nr >>= pool->sectors_per_block_shift;
else
(void) sector_div(block_nr, pool->sectors_per_block);
return block_nr;
}
/*
* Returns the _complete_ blocks that this bio covers.
*/
static void get_bio_block_range(struct thin_c *tc, struct bio *bio,
dm_block_t *begin, dm_block_t *end)
{
struct pool *pool = tc->pool;
sector_t b = bio->bi_iter.bi_sector;
sector_t e = b + (bio->bi_iter.bi_size >> SECTOR_SHIFT);
b += pool->sectors_per_block - 1ull; /* so we round up */
if (block_size_is_power_of_two(pool)) {
b >>= pool->sectors_per_block_shift;
e >>= pool->sectors_per_block_shift;
} else {
(void) sector_div(b, pool->sectors_per_block);
(void) sector_div(e, pool->sectors_per_block);
}
if (e < b)
/* Can happen if the bio is within a single block. */
e = b;
*begin = b;
*end = e;
}
static void remap(struct thin_c *tc, struct bio *bio, dm_block_t block)
{
struct pool *pool = tc->pool;
sector_t bi_sector = bio->bi_iter.bi_sector;
bio_set_dev(bio, tc->pool_dev->bdev);
if (block_size_is_power_of_two(pool))
bio->bi_iter.bi_sector =
(block << pool->sectors_per_block_shift) |
(bi_sector & (pool->sectors_per_block - 1));
else
bio->bi_iter.bi_sector = (block * pool->sectors_per_block) +
sector_div(bi_sector, pool->sectors_per_block);
}
static void remap_to_origin(struct thin_c *tc, struct bio *bio)
{
bio_set_dev(bio, tc->origin_dev->bdev);
}
static int bio_triggers_commit(struct thin_c *tc, struct bio *bio)
{
return op_is_flush(bio->bi_opf) &&
dm_thin_changed_this_transaction(tc->td);
}
static void inc_all_io_entry(struct pool *pool, struct bio *bio)
{
struct dm_thin_endio_hook *h;
if (bio_op(bio) == REQ_OP_DISCARD)
return;
h = dm_per_bio_data(bio, sizeof(struct dm_thin_endio_hook));
h->all_io_entry = dm_deferred_entry_inc(pool->all_io_ds);
}
static void issue(struct thin_c *tc, struct bio *bio)
{
struct pool *pool = tc->pool;
unsigned long flags;
if (!bio_triggers_commit(tc, bio)) {
generic_make_request(bio);
return;
}
/*
* Complete bio with an error if earlier I/O caused changes to
* the metadata that can't be committed e.g, due to I/O errors
* on the metadata device.
*/
if (dm_thin_aborted_changes(tc->td)) {
bio_io_error(bio);
return;
}
/*
* Batch together any bios that trigger commits and then issue a
* single commit for them in process_deferred_bios().
*/
spin_lock_irqsave(&pool->lock, flags);
bio_list_add(&pool->deferred_flush_bios, bio);
spin_unlock_irqrestore(&pool->lock, flags);
}
static void remap_to_origin_and_issue(struct thin_c *tc, struct bio *bio)
{
remap_to_origin(tc, bio);
issue(tc, bio);
}
static void remap_and_issue(struct thin_c *tc, struct bio *bio,
dm_block_t block)
{
remap(tc, bio, block);
issue(tc, bio);
}
/*----------------------------------------------------------------*/
/*
* Bio endio functions.
*/
struct dm_thin_new_mapping {
struct list_head list;
bool pass_discard:1;
bool maybe_shared:1;
/*
* Track quiescing, copying and zeroing preparation actions. When this
* counter hits zero the block is prepared and can be inserted into the
* btree.
*/
atomic_t prepare_actions;
blk_status_t status;
struct thin_c *tc;
dm_block_t virt_begin, virt_end;
dm_block_t data_block;
struct dm_bio_prison_cell *cell;
/*
* If the bio covers the whole area of a block then we can avoid
* zeroing or copying. Instead this bio is hooked. The bio will
* still be in the cell, so care has to be taken to avoid issuing
* the bio twice.
*/
struct bio *bio;
bio_end_io_t *saved_bi_end_io;
};
static void __complete_mapping_preparation(struct dm_thin_new_mapping *m)
{
struct pool *pool = m->tc->pool;
if (atomic_dec_and_test(&m->prepare_actions)) {
list_add_tail(&m->list, &pool->prepared_mappings);
wake_worker(pool);
}
}
static void complete_mapping_preparation(struct dm_thin_new_mapping *m)
{
unsigned long flags;
struct pool *pool = m->tc->pool;
spin_lock_irqsave(&pool->lock, flags);
__complete_mapping_preparation(m);
spin_unlock_irqrestore(&pool->lock, flags);
}
static void copy_complete(int read_err, unsigned long write_err, void *context)
{
struct dm_thin_new_mapping *m = context;
m->status = read_err || write_err ? BLK_STS_IOERR : 0;
complete_mapping_preparation(m);
}
static void overwrite_endio(struct bio *bio)
{
struct dm_thin_endio_hook *h = dm_per_bio_data(bio, sizeof(struct dm_thin_endio_hook));
struct dm_thin_new_mapping *m = h->overwrite_mapping;
bio->bi_end_io = m->saved_bi_end_io;
m->status = bio->bi_status;
complete_mapping_preparation(m);
}
/*----------------------------------------------------------------*/
/*
* Workqueue.
*/
/*
* Prepared mapping jobs.
*/
/*
* This sends the bios in the cell, except the original holder, back
* to the deferred_bios list.
*/
static void cell_defer_no_holder(struct thin_c *tc, struct dm_bio_prison_cell *cell)
{
struct pool *pool = tc->pool;
unsigned long flags;
spin_lock_irqsave(&tc->lock, flags);
cell_release_no_holder(pool, cell, &tc->deferred_bio_list);
spin_unlock_irqrestore(&tc->lock, flags);
wake_worker(pool);
}
static void thin_defer_bio(struct thin_c *tc, struct bio *bio);
struct remap_info {
struct thin_c *tc;
struct bio_list defer_bios;
struct bio_list issue_bios;
};
static void __inc_remap_and_issue_cell(void *context,
struct dm_bio_prison_cell *cell)
{
struct remap_info *info = context;
struct bio *bio;
while ((bio = bio_list_pop(&cell->bios))) {
if (op_is_flush(bio->bi_opf) || bio_op(bio) == REQ_OP_DISCARD)
bio_list_add(&info->defer_bios, bio);
else {
inc_all_io_entry(info->tc->pool, bio);
/*
* We can't issue the bios with the bio prison lock
* held, so we add them to a list to issue on
* return from this function.
*/
bio_list_add(&info->issue_bios, bio);
}
}
}
static void inc_remap_and_issue_cell(struct thin_c *tc,
struct dm_bio_prison_cell *cell,
dm_block_t block)
{
struct bio *bio;
struct remap_info info;
info.tc = tc;
bio_list_init(&info.defer_bios);
bio_list_init(&info.issue_bios);
/*
* We have to be careful to inc any bios we're about to issue
* before the cell is released, and avoid a race with new bios
* being added to the cell.
*/
cell_visit_release(tc->pool, __inc_remap_and_issue_cell,
&info, cell);
while ((bio = bio_list_pop(&info.defer_bios)))
thin_defer_bio(tc, bio);
while ((bio = bio_list_pop(&info.issue_bios)))
remap_and_issue(info.tc, bio, block);
}
static void process_prepared_mapping_fail(struct dm_thin_new_mapping *m)
{
cell_error(m->tc->pool, m->cell);
list_del(&m->list);
mempool_free(m, &m->tc->pool->mapping_pool);
}
static void complete_overwrite_bio(struct thin_c *tc, struct bio *bio)
{
struct pool *pool = tc->pool;
unsigned long flags;
/*
* If the bio has the REQ_FUA flag set we must commit the metadata
* before signaling its completion.
*/
if (!bio_triggers_commit(tc, bio)) {
bio_endio(bio);
return;
}
/*
* Complete bio with an error if earlier I/O caused changes to the
* metadata that can't be committed, e.g, due to I/O errors on the
* metadata device.
*/
if (dm_thin_aborted_changes(tc->td)) {
bio_io_error(bio);
return;
}
/*
* Batch together any bios that trigger commits and then issue a
* single commit for them in process_deferred_bios().
*/
spin_lock_irqsave(&pool->lock, flags);
bio_list_add(&pool->deferred_flush_completions, bio);
spin_unlock_irqrestore(&pool->lock, flags);
}
static void process_prepared_mapping(struct dm_thin_new_mapping *m)
{
struct thin_c *tc = m->tc;
struct pool *pool = tc->pool;
struct bio *bio = m->bio;
int r;
if (m->status) {
cell_error(pool, m->cell);
goto out;
}
/*
* Commit the prepared block into the mapping btree.
* Any I/O for this block arriving after this point will get
* remapped to it directly.
*/
r = dm_thin_insert_block(tc->td, m->virt_begin, m->data_block);
if (r) {
metadata_operation_failed(pool, "dm_thin_insert_block", r);
cell_error(pool, m->cell);
goto out;
}
/*
* Release any bios held while the block was being provisioned.
* If we are processing a write bio that completely covers the block,
* we already processed it so can ignore it now when processing
* the bios in the cell.
*/
if (bio) {
inc_remap_and_issue_cell(tc, m->cell, m->data_block);
complete_overwrite_bio(tc, bio);
} else {
inc_all_io_entry(tc->pool, m->cell->holder);
remap_and_issue(tc, m->cell->holder, m->data_block);
inc_remap_and_issue_cell(tc, m->cell, m->data_block);
}
out:
list_del(&m->list);
mempool_free(m, &pool->mapping_pool);
}
/*----------------------------------------------------------------*/
static void free_discard_mapping(struct dm_thin_new_mapping *m)
{
struct thin_c *tc = m->tc;
if (m->cell)
cell_defer_no_holder(tc, m->cell);
mempool_free(m, &tc->pool->mapping_pool);
}
static void process_prepared_discard_fail(struct dm_thin_new_mapping *m)
{
bio_io_error(m->bio);
free_discard_mapping(m);
}
static void process_prepared_discard_success(struct dm_thin_new_mapping *m)
{
bio_endio(m->bio);
free_discard_mapping(m);
}
static void process_prepared_discard_no_passdown(struct dm_thin_new_mapping *m)
{
int r;
struct thin_c *tc = m->tc;
r = dm_thin_remove_range(tc->td, m->cell->key.block_begin, m->cell->key.block_end);
if (r) {
metadata_operation_failed(tc->pool, "dm_thin_remove_range", r);
bio_io_error(m->bio);
} else
bio_endio(m->bio);
cell_defer_no_holder(tc, m->cell);
mempool_free(m, &tc->pool->mapping_pool);
}
/*----------------------------------------------------------------*/
static void passdown_double_checking_shared_status(struct dm_thin_new_mapping *m,
struct bio *discard_parent)
{
/*
* We've already unmapped this range of blocks, but before we
* passdown we have to check that these blocks are now unused.
*/
int r = 0;
bool shared = true;
struct thin_c *tc = m->tc;
struct pool *pool = tc->pool;
dm_block_t b = m->data_block, e, end = m->data_block + m->virt_end - m->virt_begin;
struct discard_op op;
begin_discard(&op, tc, discard_parent);
while (b != end) {
/* find start of unmapped run */
for (; b < end; b++) {
r = dm_pool_block_is_shared(pool->pmd, b, &shared);
if (r)
goto out;
if (!shared)
break;
}
if (b == end)
break;
/* find end of run */
for (e = b + 1; e != end; e++) {
r = dm_pool_block_is_shared(pool->pmd, e, &shared);
if (r)
goto out;
if (shared)
break;
}
r = issue_discard(&op, b, e);
if (r)
goto out;
b = e;
}
out:
end_discard(&op, r);
}
static void queue_passdown_pt2(struct dm_thin_new_mapping *m)
{
unsigned long flags;
struct pool *pool = m->tc->pool;
spin_lock_irqsave(&pool->lock, flags);
list_add_tail(&m->list, &pool->prepared_discards_pt2);
spin_unlock_irqrestore(&pool->lock, flags);
wake_worker(pool);
}
static void passdown_endio(struct bio *bio)
{
/*
* It doesn't matter if the passdown discard failed, we still want
* to unmap (we ignore err).
*/
queue_passdown_pt2(bio->bi_private);
bio_put(bio);
}
static void process_prepared_discard_passdown_pt1(struct dm_thin_new_mapping *m)
{
int r;
struct thin_c *tc = m->tc;
struct pool *pool = tc->pool;
struct bio *discard_parent;
dm_block_t data_end = m->data_block + (m->virt_end - m->virt_begin);
/*
* Only this thread allocates blocks, so we can be sure that the
* newly unmapped blocks will not be allocated before the end of
* the function.
*/
r = dm_thin_remove_range(tc->td, m->virt_begin, m->virt_end);
if (r) {
metadata_operation_failed(pool, "dm_thin_remove_range", r);
bio_io_error(m->bio);
cell_defer_no_holder(tc, m->cell);
mempool_free(m, &pool->mapping_pool);
return;
}
/*
* Increment the unmapped blocks. This prevents a race between the
* passdown io and reallocation of freed blocks.
*/
r = dm_pool_inc_data_range(pool->pmd, m->data_block, data_end);
if (r) {
metadata_operation_failed(pool, "dm_pool_inc_data_range", r);
bio_io_error(m->bio);
cell_defer_no_holder(tc, m->cell);
mempool_free(m, &pool->mapping_pool);
return;
}
discard_parent = bio_alloc(GFP_NOIO, 1);
if (!discard_parent) {
DMWARN("%s: unable to allocate top level discard bio for passdown. Skipping passdown.",
dm_device_name(tc->pool->pool_md));
queue_passdown_pt2(m);
} else {
discard_parent->bi_end_io = passdown_endio;
discard_parent->bi_private = m;
if (m->maybe_shared)
passdown_double_checking_shared_status(m, discard_parent);
else {
struct discard_op op;
begin_discard(&op, tc, discard_parent);
r = issue_discard(&op, m->data_block, data_end);
end_discard(&op, r);
}
}
}
static void process_prepared_discard_passdown_pt2(struct dm_thin_new_mapping *m)
{
int r;
struct thin_c *tc = m->tc;
struct pool *pool = tc->pool;
/*
* The passdown has completed, so now we can decrement all those
* unmapped blocks.
*/
r = dm_pool_dec_data_range(pool->pmd, m->data_block,
m->data_block + (m->virt_end - m->virt_begin));
if (r) {
metadata_operation_failed(pool, "dm_pool_dec_data_range", r);
bio_io_error(m->bio);
} else
bio_endio(m->bio);
cell_defer_no_holder(tc, m->cell);
mempool_free(m, &pool->mapping_pool);
}
static void process_prepared(struct pool *pool, struct list_head *head,
process_mapping_fn *fn)
{
unsigned long flags;
struct list_head maps;
struct dm_thin_new_mapping *m, *tmp;
INIT_LIST_HEAD(&maps);
spin_lock_irqsave(&pool->lock, flags);
list_splice_init(head, &maps);
spin_unlock_irqrestore(&pool->lock, flags);
list_for_each_entry_safe(m, tmp, &maps, list)
(*fn)(m);
}
/*
* Deferred bio jobs.
*/
static int io_overlaps_block(struct pool *pool, struct bio *bio)
{
return bio->bi_iter.bi_size ==
(pool->sectors_per_block << SECTOR_SHIFT);
}
static int io_overwrites_block(struct pool *pool, struct bio *bio)
{
return (bio_data_dir(bio) == WRITE) &&
io_overlaps_block(pool, bio);
}
static void save_and_set_endio(struct bio *bio, bio_end_io_t **save,
bio_end_io_t *fn)
{
*save = bio->bi_end_io;
bio->bi_end_io = fn;
}
static int ensure_next_mapping(struct pool *pool)
{
if (pool->next_mapping)
return 0;
pool->next_mapping = mempool_alloc(&pool->mapping_pool, GFP_ATOMIC);
return pool->next_mapping ? 0 : -ENOMEM;
}
static struct dm_thin_new_mapping *get_next_mapping(struct pool *pool)
{
struct dm_thin_new_mapping *m = pool->next_mapping;
BUG_ON(!pool->next_mapping);
memset(m, 0, sizeof(struct dm_thin_new_mapping));
INIT_LIST_HEAD(&m->list);
m->bio = NULL;
pool->next_mapping = NULL;
return m;
}
static void ll_zero(struct thin_c *tc, struct dm_thin_new_mapping *m,
sector_t begin, sector_t end)
{
struct dm_io_region to;
to.bdev = tc->pool_dev->bdev;
to.sector = begin;
to.count = end - begin;
dm_kcopyd_zero(tc->pool->copier, 1, &to, 0, copy_complete, m);
}
static void remap_and_issue_overwrite(struct thin_c *tc, struct bio *bio,
dm_block_t data_begin,
struct dm_thin_new_mapping *m)
{
struct pool *pool = tc->pool;
struct dm_thin_endio_hook *h = dm_per_bio_data(bio, sizeof(struct dm_thin_endio_hook));
h->overwrite_mapping = m;
m->bio = bio;
save_and_set_endio(bio, &m->saved_bi_end_io, overwrite_endio);
inc_all_io_entry(pool, bio);
remap_and_issue(tc, bio, data_begin);
}
/*
* A partial copy also needs to zero the uncopied region.
*/
static void schedule_copy(struct thin_c *tc, dm_block_t virt_block,
struct dm_dev *origin, dm_block_t data_origin,
dm_block_t data_dest,
struct dm_bio_prison_cell *cell, struct bio *bio,
sector_t len)
{
struct pool *pool = tc->pool;
struct dm_thin_new_mapping *m = get_next_mapping(pool);
m->tc = tc;
m->virt_begin = virt_block;
m->virt_end = virt_block + 1u;
m->data_block = data_dest;
m->cell = cell;
/*
* quiesce action + copy action + an extra reference held for the
* duration of this function (we may need to inc later for a
* partial zero).
*/
atomic_set(&m->prepare_actions, 3);
if (!dm_deferred_set_add_work(pool->shared_read_ds, &m->list))
complete_mapping_preparation(m); /* already quiesced */
/*
* IO to pool_dev remaps to the pool target's data_dev.
*
* If the whole block of data is being overwritten, we can issue the
* bio immediately. Otherwise we use kcopyd to clone the data first.
*/
if (io_overwrites_block(pool, bio))
remap_and_issue_overwrite(tc, bio, data_dest, m);
else {
struct dm_io_region from, to;
from.bdev = origin->bdev;
from.sector = data_origin * pool->sectors_per_block;
from.count = len;
to.bdev = tc->pool_dev->bdev;
to.sector = data_dest * pool->sectors_per_block;
to.count = len;
dm_kcopyd_copy(pool->copier, &from, 1, &to,
0, copy_complete, m);
/*
* Do we need to zero a tail region?
*/
if (len < pool->sectors_per_block && pool->pf.zero_new_blocks) {
atomic_inc(&m->prepare_actions);
ll_zero(tc, m,
data_dest * pool->sectors_per_block + len,
(data_dest + 1) * pool->sectors_per_block);
}
}
complete_mapping_preparation(m); /* drop our ref */
}
static void schedule_internal_copy(struct thin_c *tc, dm_block_t virt_block,
dm_block_t data_origin, dm_block_t data_dest,
struct dm_bio_prison_cell *cell, struct bio *bio)
{
schedule_copy(tc, virt_block, tc->pool_dev,
data_origin, data_dest, cell, bio,
tc->pool->sectors_per_block);
}
static void schedule_zero(struct thin_c *tc, dm_block_t virt_block,
dm_block_t data_block, struct dm_bio_prison_cell *cell,
struct bio *bio)
{
struct pool *pool = tc->pool;
struct dm_thin_new_mapping *m = get_next_mapping(pool);
atomic_set(&m->prepare_actions, 1); /* no need to quiesce */
m->tc = tc;
m->virt_begin = virt_block;
m->virt_end = virt_block + 1u;
m->data_block = data_block;
m->cell = cell;
/*
* If the whole block of data is being overwritten or we are not
* zeroing pre-existing data, we can issue the bio immediately.
* Otherwise we use kcopyd to zero the data first.
*/
if (pool->pf.zero_new_blocks) {
if (io_overwrites_block(pool, bio))
remap_and_issue_overwrite(tc, bio, data_block, m);
else
ll_zero(tc, m, data_block * pool->sectors_per_block,
(data_block + 1) * pool->sectors_per_block);
} else
process_prepared_mapping(m);
}
static void schedule_external_copy(struct thin_c *tc, dm_block_t virt_block,
dm_block_t data_dest,
struct dm_bio_prison_cell *cell, struct bio *bio)
{
struct pool *pool = tc->pool;
sector_t virt_block_begin = virt_block * pool->sectors_per_block;
sector_t virt_block_end = (virt_block + 1) * pool->sectors_per_block;
if (virt_block_end <= tc->origin_size)
schedule_copy(tc, virt_block, tc->origin_dev,
virt_block, data_dest, cell, bio,
pool->sectors_per_block);
else if (virt_block_begin < tc->origin_size)
schedule_copy(tc, virt_block, tc->origin_dev,
virt_block, data_dest, cell, bio,
tc->origin_size - virt_block_begin);
else
schedule_zero(tc, virt_block, data_dest, cell, bio);
}
static void set_pool_mode(struct pool *pool, enum pool_mode new_mode);
static void requeue_bios(struct pool *pool);
static bool is_read_only_pool_mode(enum pool_mode mode)
{
return (mode == PM_OUT_OF_METADATA_SPACE || mode == PM_READ_ONLY);
}
static bool is_read_only(struct pool *pool)
{
return is_read_only_pool_mode(get_pool_mode(pool));
}
static void check_for_metadata_space(struct pool *pool)
{
int r;
const char *ooms_reason = NULL;
dm_block_t nr_free;
r = dm_pool_get_free_metadata_block_count(pool->pmd, &nr_free);
if (r)
ooms_reason = "Could not get free metadata blocks";
else if (!nr_free)
ooms_reason = "No free metadata blocks";
if (ooms_reason && !is_read_only(pool)) {
DMERR("%s", ooms_reason);
set_pool_mode(pool, PM_OUT_OF_METADATA_SPACE);
}
}
static void check_for_data_space(struct pool *pool)
{
int r;
dm_block_t nr_free;
if (get_pool_mode(pool) != PM_OUT_OF_DATA_SPACE)
return;
r = dm_pool_get_free_block_count(pool->pmd, &nr_free);
if (r)
return;
if (nr_free) {
set_pool_mode(pool, PM_WRITE);
requeue_bios(pool);
}
}
/*
* A non-zero return indicates read_only or fail_io mode.
* Many callers don't care about the return value.
*/
static int commit(struct pool *pool)
{
int r;
if (get_pool_mode(pool) >= PM_OUT_OF_METADATA_SPACE)
return -EINVAL;
r = dm_pool_commit_metadata(pool->pmd);
if (r)
metadata_operation_failed(pool, "dm_pool_commit_metadata", r);
else {
check_for_metadata_space(pool);
check_for_data_space(pool);
}
return r;
}
static void check_low_water_mark(struct pool *pool, dm_block_t free_blocks)
{
unsigned long flags;
if (free_blocks <= pool->low_water_blocks && !pool->low_water_triggered) {
DMWARN("%s: reached low water mark for data device: sending event.",
dm_device_name(pool->pool_md));
spin_lock_irqsave(&pool->lock, flags);
pool->low_water_triggered = true;
spin_unlock_irqrestore(&pool->lock, flags);
dm_table_event(pool->ti->table);
}
}
static int alloc_data_block(struct thin_c *tc, dm_block_t *result)
{
int r;
dm_block_t free_blocks;
struct pool *pool = tc->pool;
if (WARN_ON(get_pool_mode(pool) != PM_WRITE))
return -EINVAL;
r = dm_pool_get_free_block_count(pool->pmd, &free_blocks);
if (r) {
metadata_operation_failed(pool, "dm_pool_get_free_block_count", r);
return r;
}
check_low_water_mark(pool, free_blocks);
if (!free_blocks) {
/*
* Try to commit to see if that will free up some
* more space.
*/
r = commit(pool);
if (r)
return r;
r = dm_pool_get_free_block_count(pool->pmd, &free_blocks);
if (r) {
metadata_operation_failed(pool, "dm_pool_get_free_block_count", r);
return r;
}
if (!free_blocks) {
set_pool_mode(pool, PM_OUT_OF_DATA_SPACE);
return -ENOSPC;
}
}
r = dm_pool_alloc_data_block(pool->pmd, result);
if (r) {
if (r == -ENOSPC)
set_pool_mode(pool, PM_OUT_OF_DATA_SPACE);
else
metadata_operation_failed(pool, "dm_pool_alloc_data_block", r);
return r;
}
r = dm_pool_get_free_metadata_block_count(pool->pmd, &free_blocks);
if (r) {
metadata_operation_failed(pool, "dm_pool_get_free_metadata_block_count", r);
return r;
}
if (!free_blocks) {
/* Let's commit before we use up the metadata reserve. */
r = commit(pool);
if (r)
return r;
}
return 0;
}
/*
* If we have run out of space, queue bios until the device is
* resumed, presumably after having been reloaded with more space.
*/
static void retry_on_resume(struct bio *bio)
{
struct dm_thin_endio_hook *h = dm_per_bio_data(bio, sizeof(struct dm_thin_endio_hook));
struct thin_c *tc = h->tc;
unsigned long flags;
spin_lock_irqsave(&tc->lock, flags);
bio_list_add(&tc->retry_on_resume_list, bio);
spin_unlock_irqrestore(&tc->lock, flags);
}
static blk_status_t should_error_unserviceable_bio(struct pool *pool)
{
enum pool_mode m = get_pool_mode(pool);
switch (m) {
case PM_WRITE:
/* Shouldn't get here */
DMERR_LIMIT("bio unserviceable, yet pool is in PM_WRITE mode");
return BLK_STS_IOERR;
case PM_OUT_OF_DATA_SPACE:
return pool->pf.error_if_no_space ? BLK_STS_NOSPC : 0;
case PM_OUT_OF_METADATA_SPACE:
case PM_READ_ONLY:
case PM_FAIL:
return BLK_STS_IOERR;
default:
/* Shouldn't get here */
DMERR_LIMIT("bio unserviceable, yet pool has an unknown mode");
return BLK_STS_IOERR;
}
}
static void handle_unserviceable_bio(struct pool *pool, struct bio *bio)
{
blk_status_t error = should_error_unserviceable_bio(pool);
if (error) {
bio->bi_status = error;
bio_endio(bio);
} else
retry_on_resume(bio);
}
static void retry_bios_on_resume(struct pool *pool, struct dm_bio_prison_cell *cell)
{
struct bio *bio;
struct bio_list bios;
blk_status_t error;
error = should_error_unserviceable_bio(pool);
if (error) {
cell_error_with_code(pool, cell, error);
return;
}
bio_list_init(&bios);
cell_release(pool, cell, &bios);
while ((bio = bio_list_pop(&bios)))
retry_on_resume(bio);
}
static void process_discard_cell_no_passdown(struct thin_c *tc,
struct dm_bio_prison_cell *virt_cell)
{
struct pool *pool = tc->pool;
struct dm_thin_new_mapping *m = get_next_mapping(pool);
/*
* We don't need to lock the data blocks, since there's no
* passdown. We only lock data blocks for allocation and breaking sharing.
*/
m->tc = tc;
m->virt_begin = virt_cell->key.block_begin;
m->virt_end = virt_cell->key.block_end;
m->cell = virt_cell;
m->bio = virt_cell->holder;
if (!dm_deferred_set_add_work(pool->all_io_ds, &m->list))
pool->process_prepared_discard(m);
}
static void break_up_discard_bio(struct thin_c *tc, dm_block_t begin, dm_block_t end,
struct bio *bio)
{
struct pool *pool = tc->pool;
int r;
bool maybe_shared;
struct dm_cell_key data_key;
struct dm_bio_prison_cell *data_cell;
struct dm_thin_new_mapping *m;
dm_block_t virt_begin, virt_end, data_begin;
while (begin != end) {
r = ensure_next_mapping(pool);
if (r)
/* we did our best */
return;
r = dm_thin_find_mapped_range(tc->td, begin, end, &virt_begin, &virt_end,
&data_begin, &maybe_shared);
if (r)
/*
* Silently fail, letting any mappings we've
* created complete.
*/
break;
build_key(tc->td, PHYSICAL, data_begin, data_begin + (virt_end - virt_begin), &data_key);
if (bio_detain(tc->pool, &data_key, NULL, &data_cell)) {
/* contention, we'll give up with this range */
begin = virt_end;
continue;
}
/*
* IO may still be going to the destination block. We must
* quiesce before we can do the removal.
*/
m = get_next_mapping(pool);
m->tc = tc;
m->maybe_shared = maybe_shared;
m->virt_begin = virt_begin;
m->virt_end = virt_end;
m->data_block = data_begin;
m->cell = data_cell;
m->bio = bio;
/*
* The parent bio must not complete before sub discard bios are
* chained to it (see end_discard's bio_chain)!
*
* This per-mapping bi_remaining increment is paired with
* the implicit decrement that occurs via bio_endio() in
* end_discard().
*/
bio_inc_remaining(bio);
if (!dm_deferred_set_add_work(pool->all_io_ds, &m->list))
pool->process_prepared_discard(m);
begin = virt_end;
}
}
static void process_discard_cell_passdown(struct thin_c *tc, struct dm_bio_prison_cell *virt_cell)
{
struct bio *bio = virt_cell->holder;
struct dm_thin_endio_hook *h = dm_per_bio_data(bio, sizeof(struct dm_thin_endio_hook));
/*
* The virt_cell will only get freed once the origin bio completes.
* This means it will remain locked while all the individual
* passdown bios are in flight.
*/
h->cell = virt_cell;
break_up_discard_bio(tc, virt_cell->key.block_begin, virt_cell->key.block_end, bio);
/*
* We complete the bio now, knowing that the bi_remaining field
* will prevent completion until the sub range discards have
* completed.
*/
bio_endio(bio);
}
static void process_discard_bio(struct thin_c *tc, struct bio *bio)
{
dm_block_t begin, end;
struct dm_cell_key virt_key;
struct dm_bio_prison_cell *virt_cell;
get_bio_block_range(tc, bio, &begin, &end);
if (begin == end) {
/*
* The discard covers less than a block.
*/
bio_endio(bio);
return;
}
build_key(tc->td, VIRTUAL, begin, end, &virt_key);
if (bio_detain(tc->pool, &virt_key, bio, &virt_cell))
/*
* Potential starvation issue: We're relying on the
* fs/application being well behaved, and not trying to
* send IO to a region at the same time as discarding it.
* If they do this persistently then it's possible this
* cell will never be granted.
*/
return;
tc->pool->process_discard_cell(tc, virt_cell);
}
static void break_sharing(struct thin_c *tc, struct bio *bio, dm_block_t block,
struct dm_cell_key *key,
struct dm_thin_lookup_result *lookup_result,
struct dm_bio_prison_cell *cell)
{
int r;
dm_block_t data_block;
struct pool *pool = tc->pool;
r = alloc_data_block(tc, &data_block);
switch (r) {
case 0:
schedule_internal_copy(tc, block, lookup_result->block,
data_block, cell, bio);
break;
case -ENOSPC:
retry_bios_on_resume(pool, cell);
break;
default:
DMERR_LIMIT("%s: alloc_data_block() failed: error = %d",
__func__, r);
cell_error(pool, cell);
break;
}
}
static void __remap_and_issue_shared_cell(void *context,
struct dm_bio_prison_cell *cell)
{
struct remap_info *info = context;
struct bio *bio;
while ((bio = bio_list_pop(&cell->bios))) {
if (bio_data_dir(bio) == WRITE || op_is_flush(bio->bi_opf) ||
bio_op(bio) == REQ_OP_DISCARD)
bio_list_add(&info->defer_bios, bio);
else {
struct dm_thin_endio_hook *h = dm_per_bio_data(bio, sizeof(struct dm_thin_endio_hook));
h->shared_read_entry = dm_deferred_entry_inc(info->tc->pool->shared_read_ds);
inc_all_io_entry(info->tc->pool, bio);
bio_list_add(&info->issue_bios, bio);
}
}
}
static void remap_and_issue_shared_cell(struct thin_c *tc,
struct dm_bio_prison_cell *cell,
dm_block_t block)
{
struct bio *bio;
struct remap_info info;
info.tc = tc;
bio_list_init(&info.defer_bios);
bio_list_init(&info.issue_bios);
cell_visit_release(tc->pool, __remap_and_issue_shared_cell,
&info, cell);
while ((bio = bio_list_pop(&info.defer_bios)))
thin_defer_bio(tc, bio);
while ((bio = bio_list_pop(&info.issue_bios)))
remap_and_issue(tc, bio, block);
}
static void process_shared_bio(struct thin_c *tc, struct bio *bio,
dm_block_t block,
struct dm_thin_lookup_result *lookup_result,
struct dm_bio_prison_cell *virt_cell)
{
struct dm_bio_prison_cell *data_cell;
struct pool *pool = tc->pool;
struct dm_cell_key key;
/*
* If cell is already occupied, then sharing is already in the process
* of being broken so we have nothing further to do here.
*/
build_data_key(tc->td, lookup_result->block, &key);
if (bio_detain(pool, &key, bio, &data_cell)) {
cell_defer_no_holder(tc, virt_cell);
return;
}
if (bio_data_dir(bio) == WRITE && bio->bi_iter.bi_size) {
break_sharing(tc, bio, block, &key, lookup_result, data_cell);
cell_defer_no_holder(tc, virt_cell);
} else {
struct dm_thin_endio_hook *h = dm_per_bio_data(bio, sizeof(struct dm_thin_endio_hook));
h->shared_read_entry = dm_deferred_entry_inc(pool->shared_read_ds);
inc_all_io_entry(pool, bio);
remap_and_issue(tc, bio, lookup_result->block);
remap_and_issue_shared_cell(tc, data_cell, lookup_result->block);
remap_and_issue_shared_cell(tc, virt_cell, lookup_result->block);
}
}
static void provision_block(struct thin_c *tc, struct bio *bio, dm_block_t block,
struct dm_bio_prison_cell *cell)
{
int r;
dm_block_t data_block;
struct pool *pool = tc->pool;
/*
* Remap empty bios (flushes) immediately, without provisioning.
*/
if (!bio->bi_iter.bi_size) {
inc_all_io_entry(pool, bio);
cell_defer_no_holder(tc, cell);
remap_and_issue(tc, bio, 0);
return;
}
/*
* Fill read bios with zeroes and complete them immediately.
*/
if (bio_data_dir(bio) == READ) {
zero_fill_bio(bio);
cell_defer_no_holder(tc, cell);
bio_endio(bio);
return;
}
r = alloc_data_block(tc, &data_block);
switch (r) {
case 0:
if (tc->origin_dev)
schedule_external_copy(tc, block, data_block, cell, bio);
else
schedule_zero(tc, block, data_block, cell, bio);
break;
case -ENOSPC:
retry_bios_on_resume(pool, cell);
break;
default:
DMERR_LIMIT("%s: alloc_data_block() failed: error = %d",
__func__, r);
cell_error(pool, cell);
break;
}
}
static void process_cell(struct thin_c *tc, struct dm_bio_prison_cell *cell)
{
int r;
struct pool *pool = tc->pool;
struct bio *bio = cell->holder;
dm_block_t block = get_bio_block(tc, bio);
struct dm_thin_lookup_result lookup_result;
if (tc->requeue_mode) {
cell_requeue(pool, cell);
return;
}
r = dm_thin_find_block(tc->td, block, 1, &lookup_result);
switch (r) {
case 0:
if (lookup_result.shared)
process_shared_bio(tc, bio, block, &lookup_result, cell);
else {
inc_all_io_entry(pool, bio);
remap_and_issue(tc, bio, lookup_result.block);
inc_remap_and_issue_cell(tc, cell, lookup_result.block);
}
break;
case -ENODATA:
if (bio_data_dir(bio) == READ && tc->origin_dev) {
inc_all_io_entry(pool, bio);
cell_defer_no_holder(tc, cell);
if (bio_end_sector(bio) <= tc->origin_size)
remap_to_origin_and_issue(tc, bio);
else if (bio->bi_iter.bi_sector < tc->origin_size) {
zero_fill_bio(bio);
bio->bi_iter.bi_size = (tc->origin_size - bio->bi_iter.bi_sector) << SECTOR_SHIFT;
remap_to_origin_and_issue(tc, bio);
} else {
zero_fill_bio(bio);
bio_endio(bio);
}
} else
provision_block(tc, bio, block, cell);
break;
default:
DMERR_LIMIT("%s: dm_thin_find_block() failed: error = %d",
__func__, r);
cell_defer_no_holder(tc, cell);
bio_io_error(bio);
break;
}
}
static void process_bio(struct thin_c *tc, struct bio *bio)
{
struct pool *pool = tc->pool;
dm_block_t block = get_bio_block(tc, bio);
struct dm_bio_prison_cell *cell;
struct dm_cell_key key;
/*
* If cell is already occupied, then the block is already
* being provisioned so we have nothing further to do here.
*/
build_virtual_key(tc->td, block, &key);
if (bio_detain(pool, &key, bio, &cell))
return;
process_cell(tc, cell);
}
static void __process_bio_read_only(struct thin_c *tc, struct bio *bio,
struct dm_bio_prison_cell *cell)
{
int r;
int rw = bio_data_dir(bio);
dm_block_t block = get_bio_block(tc, bio);
struct dm_thin_lookup_result lookup_result;
r = dm_thin_find_block(tc->td, block, 1, &lookup_result);
switch (r) {
case 0:
if (lookup_result.shared && (rw == WRITE) && bio->bi_iter.bi_size) {
handle_unserviceable_bio(tc->pool, bio);
if (cell)
cell_defer_no_holder(tc, cell);
} else {
inc_all_io_entry(tc->pool, bio);
remap_and_issue(tc, bio, lookup_result.block);
if (cell)
inc_remap_and_issue_cell(tc, cell, lookup_result.block);
}
break;
case -ENODATA:
if (cell)
cell_defer_no_holder(tc, cell);
if (rw != READ) {
handle_unserviceable_bio(tc->pool, bio);
break;
}
if (tc->origin_dev) {
inc_all_io_entry(tc->pool, bio);
remap_to_origin_and_issue(tc, bio);
break;
}
zero_fill_bio(bio);
bio_endio(bio);
break;
default:
DMERR_LIMIT("%s: dm_thin_find_block() failed: error = %d",
__func__, r);
if (cell)
cell_defer_no_holder(tc, cell);
bio_io_error(bio);
break;
}
}
static void process_bio_read_only(struct thin_c *tc, struct bio *bio)
{
__process_bio_read_only(tc, bio, NULL);
}
static void process_cell_read_only(struct thin_c *tc, struct dm_bio_prison_cell *cell)
{
__process_bio_read_only(tc, cell->holder, cell);
}
static void process_bio_success(struct thin_c *tc, struct bio *bio)
{
bio_endio(bio);
}
static void process_bio_fail(struct thin_c *tc, struct bio *bio)
{
bio_io_error(bio);
}
static void process_cell_success(struct thin_c *tc, struct dm_bio_prison_cell *cell)
{
cell_success(tc->pool, cell);
}
static void process_cell_fail(struct thin_c *tc, struct dm_bio_prison_cell *cell)
{
cell_error(tc->pool, cell);
}
/*
* FIXME: should we also commit due to size of transaction, measured in
* metadata blocks?
*/
static int need_commit_due_to_time(struct pool *pool)
{
return !time_in_range(jiffies, pool->last_commit_jiffies,
pool->last_commit_jiffies + COMMIT_PERIOD);
}
#define thin_pbd(node) rb_entry((node), struct dm_thin_endio_hook, rb_node)
#define thin_bio(pbd) dm_bio_from_per_bio_data((pbd), sizeof(struct dm_thin_endio_hook))
static void __thin_bio_rb_add(struct thin_c *tc, struct bio *bio)
{
struct rb_node **rbp, *parent;
struct dm_thin_endio_hook *pbd;
sector_t bi_sector = bio->bi_iter.bi_sector;
rbp = &tc->sort_bio_list.rb_node;
parent = NULL;
while (*rbp) {
parent = *rbp;
pbd = thin_pbd(parent);
if (bi_sector < thin_bio(pbd)->bi_iter.bi_sector)
rbp = &(*rbp)->rb_left;
else
rbp = &(*rbp)->rb_right;
}
pbd = dm_per_bio_data(bio, sizeof(struct dm_thin_endio_hook));
rb_link_node(&pbd->rb_node, parent, rbp);
rb_insert_color(&pbd->rb_node, &tc->sort_bio_list);
}
static void __extract_sorted_bios(struct thin_c *tc)
{
struct rb_node *node;
struct dm_thin_endio_hook *pbd;
struct bio *bio;
for (node = rb_first(&tc->sort_bio_list); node; node = rb_next(node)) {
pbd = thin_pbd(node);
bio = thin_bio(pbd);
bio_list_add(&tc->deferred_bio_list, bio);
rb_erase(&pbd->rb_node, &tc->sort_bio_list);
}
WARN_ON(!RB_EMPTY_ROOT(&tc->sort_bio_list));
}
static void __sort_thin_deferred_bios(struct thin_c *tc)
{
struct bio *bio;
struct bio_list bios;
bio_list_init(&bios);
bio_list_merge(&bios, &tc->deferred_bio_list);
bio_list_init(&tc->deferred_bio_list);
/* Sort deferred_bio_list using rb-tree */
while ((bio = bio_list_pop(&bios)))
__thin_bio_rb_add(tc, bio);
/*
* Transfer the sorted bios in sort_bio_list back to
* deferred_bio_list to allow lockless submission of
* all bios.
*/
__extract_sorted_bios(tc);
}
static void process_thin_deferred_bios(struct thin_c *tc)
{
struct pool *pool = tc->pool;
unsigned long flags;
struct bio *bio;
struct bio_list bios;
struct blk_plug plug;
unsigned count = 0;
if (tc->requeue_mode) {
error_thin_bio_list(tc, &tc->deferred_bio_list,
BLK_STS_DM_REQUEUE);
return;
}
bio_list_init(&bios);
spin_lock_irqsave(&tc->lock, flags);
if (bio_list_empty(&tc->deferred_bio_list)) {
spin_unlock_irqrestore(&tc->lock, flags);
return;
}
__sort_thin_deferred_bios(tc);
bio_list_merge(&bios, &tc->deferred_bio_list);
bio_list_init(&tc->deferred_bio_list);
spin_unlock_irqrestore(&tc->lock, flags);
blk_start_plug(&plug);
while ((bio = bio_list_pop(&bios))) {
/*
* If we've got no free new_mapping structs, and processing
* this bio might require one, we pause until there are some
* prepared mappings to process.
*/
if (ensure_next_mapping(pool)) {
spin_lock_irqsave(&tc->lock, flags);
bio_list_add(&tc->deferred_bio_list, bio);
bio_list_merge(&tc->deferred_bio_list, &bios);
spin_unlock_irqrestore(&tc->lock, flags);
break;
}
if (bio_op(bio) == REQ_OP_DISCARD)
pool->process_discard(tc, bio);
else
pool->process_bio(tc, bio);
if ((count++ & 127) == 0) {
throttle_work_update(&pool->throttle);
dm_pool_issue_prefetches(pool->pmd);
}
}
blk_finish_plug(&plug);
}
static int cmp_cells(const void *lhs, const void *rhs)
{
struct dm_bio_prison_cell *lhs_cell = *((struct dm_bio_prison_cell **) lhs);
struct dm_bio_prison_cell *rhs_cell = *((struct dm_bio_prison_cell **) rhs);
BUG_ON(!lhs_cell->holder);
BUG_ON(!rhs_cell->holder);
if (lhs_cell->holder->bi_iter.bi_sector < rhs_cell->holder->bi_iter.bi_sector)
return -1;
if (lhs_cell->holder->bi_iter.bi_sector > rhs_cell->holder->bi_iter.bi_sector)
return 1;
return 0;
}
static unsigned sort_cells(struct pool *pool, struct list_head *cells)
{
unsigned count = 0;
struct dm_bio_prison_cell *cell, *tmp;
list_for_each_entry_safe(cell, tmp, cells, user_list) {
if (count >= CELL_SORT_ARRAY_SIZE)
break;
pool->cell_sort_array[count++] = cell;
list_del(&cell->user_list);
}
sort(pool->cell_sort_array, count, sizeof(cell), cmp_cells, NULL);
return count;
}
static void process_thin_deferred_cells(struct thin_c *tc)
{
struct pool *pool = tc->pool;
unsigned long flags;
struct list_head cells;
struct dm_bio_prison_cell *cell;
unsigned i, j, count;
INIT_LIST_HEAD(&cells);
spin_lock_irqsave(&tc->lock, flags);
list_splice_init(&tc->deferred_cells, &cells);
spin_unlock_irqrestore(&tc->lock, flags);
if (list_empty(&cells))
return;
do {
count = sort_cells(tc->pool, &cells);
for (i = 0; i < count; i++) {
cell = pool->cell_sort_array[i];
BUG_ON(!cell->holder);
/*
* If we've got no free new_mapping structs, and processing
* this bio might require one, we pause until there are some
* prepared mappings to process.
*/
if (ensure_next_mapping(pool)) {
for (j = i; j < count; j++)
list_add(&pool->cell_sort_array[j]->user_list, &cells);
spin_lock_irqsave(&tc->lock, flags);
list_splice(&cells, &tc->deferred_cells);
spin_unlock_irqrestore(&tc->lock, flags);
return;
}
if (bio_op(cell->holder) == REQ_OP_DISCARD)
pool->process_discard_cell(tc, cell);
else
pool->process_cell(tc, cell);
}
} while (!list_empty(&cells));
}
static void thin_get(struct thin_c *tc);
static void thin_put(struct thin_c *tc);
/*
* We can't hold rcu_read_lock() around code that can block. So we
* find a thin with the rcu lock held; bump a refcount; then drop
* the lock.
*/
static struct thin_c *get_first_thin(struct pool *pool)
{
struct thin_c *tc = NULL;
rcu_read_lock();
if (!list_empty(&pool->active_thins)) {
tc = list_entry_rcu(pool->active_thins.next, struct thin_c, list);
thin_get(tc);
}
rcu_read_unlock();
return tc;
}
static struct thin_c *get_next_thin(struct pool *pool, struct thin_c *tc)
{
struct thin_c *old_tc = tc;
rcu_read_lock();
list_for_each_entry_continue_rcu(tc, &pool->active_thins, list) {
thin_get(tc);
thin_put(old_tc);
rcu_read_unlock();
return tc;
}
thin_put(old_tc);
rcu_read_unlock();
return NULL;
}
static void process_deferred_bios(struct pool *pool)
{
unsigned long flags;
struct bio *bio;
struct bio_list bios, bio_completions;
struct thin_c *tc;
tc = get_first_thin(pool);
while (tc) {
process_thin_deferred_cells(tc);
process_thin_deferred_bios(tc);
tc = get_next_thin(pool, tc);
}
/*
* If there are any deferred flush bios, we must commit the metadata
* before issuing them or signaling their completion.
*/
bio_list_init(&bios);
bio_list_init(&bio_completions);
spin_lock_irqsave(&pool->lock, flags);
bio_list_merge(&bios, &pool->deferred_flush_bios);
bio_list_init(&pool->deferred_flush_bios);
bio_list_merge(&bio_completions, &pool->deferred_flush_completions);
bio_list_init(&pool->deferred_flush_completions);
spin_unlock_irqrestore(&pool->lock, flags);
if (bio_list_empty(&bios) && bio_list_empty(&bio_completions) &&
!(dm_pool_changed_this_transaction(pool->pmd) && need_commit_due_to_time(pool)))
return;
if (commit(pool)) {
bio_list_merge(&bios, &bio_completions);
while ((bio = bio_list_pop(&bios)))
bio_io_error(bio);
return;
}
pool->last_commit_jiffies = jiffies;
while ((bio = bio_list_pop(&bio_completions)))
bio_endio(bio);
while ((bio = bio_list_pop(&bios)))
generic_make_request(bio);
}
static void do_worker(struct work_struct *ws)
{
struct pool *pool = container_of(ws, struct pool, worker);
throttle_work_start(&pool->throttle);
dm_pool_issue_prefetches(pool->pmd);
throttle_work_update(&pool->throttle);
process_prepared(pool, &pool->prepared_mappings, &pool->process_prepared_mapping);
throttle_work_update(&pool->throttle);
process_prepared(pool, &pool->prepared_discards, &pool->process_prepared_discard);
throttle_work_update(&pool->throttle);
process_prepared(pool, &pool->prepared_discards_pt2, &pool->process_prepared_discard_pt2);
throttle_work_update(&pool->throttle);
process_deferred_bios(pool);
throttle_work_complete(&pool->throttle);
}
/*
* We want to commit periodically so that not too much
* unwritten data builds up.
*/
static void do_waker(struct work_struct *ws)
{
struct pool *pool = container_of(to_delayed_work(ws), struct pool, waker);
wake_worker(pool);
queue_delayed_work(pool->wq, &pool->waker, COMMIT_PERIOD);
}
/*
* We're holding onto IO to allow userland time to react. After the
* timeout either the pool will have been resized (and thus back in
* PM_WRITE mode), or we degrade to PM_OUT_OF_DATA_SPACE w/ error_if_no_space.
*/
static void do_no_space_timeout(struct work_struct *ws)
{
struct pool *pool = container_of(to_delayed_work(ws), struct pool,
no_space_timeout);
if (get_pool_mode(pool) == PM_OUT_OF_DATA_SPACE && !pool->pf.error_if_no_space) {
pool->pf.error_if_no_space = true;
notify_of_pool_mode_change(pool);
error_retry_list_with_code(pool, BLK_STS_NOSPC);
}
}
/*----------------------------------------------------------------*/
struct pool_work {
struct work_struct worker;
struct completion complete;
};
static struct pool_work *to_pool_work(struct work_struct *ws)
{
return container_of(ws, struct pool_work, worker);
}
static void pool_work_complete(struct pool_work *pw)
{
complete(&pw->complete);
}
static void pool_work_wait(struct pool_work *pw, struct pool *pool,
void (*fn)(struct work_struct *))
{
INIT_WORK_ONSTACK(&pw->worker, fn);
init_completion(&pw->complete);
queue_work(pool->wq, &pw->worker);
wait_for_completion(&pw->complete);
}
/*----------------------------------------------------------------*/
struct noflush_work {
struct pool_work pw;
struct thin_c *tc;
};
static struct noflush_work *to_noflush(struct work_struct *ws)
{
return container_of(to_pool_work(ws), struct noflush_work, pw);
}
static void do_noflush_start(struct work_struct *ws)
{
struct noflush_work *w = to_noflush(ws);
w->tc->requeue_mode = true;
requeue_io(w->tc);
pool_work_complete(&w->pw);
}
static void do_noflush_stop(struct work_struct *ws)
{
struct noflush_work *w = to_noflush(ws);
w->tc->requeue_mode = false;
pool_work_complete(&w->pw);
}
static void noflush_work(struct thin_c *tc, void (*fn)(struct work_struct *))
{
struct noflush_work w;
w.tc = tc;
pool_work_wait(&w.pw, tc->pool, fn);
}
/*----------------------------------------------------------------*/
static bool passdown_enabled(struct pool_c *pt)
{
return pt->adjusted_pf.discard_passdown;
}
static void set_discard_callbacks(struct pool *pool)
{
struct pool_c *pt = pool->ti->private;
if (passdown_enabled(pt)) {
pool->process_discard_cell = process_discard_cell_passdown;
pool->process_prepared_discard = process_prepared_discard_passdown_pt1;
pool->process_prepared_discard_pt2 = process_prepared_discard_passdown_pt2;
} else {
pool->process_discard_cell = process_discard_cell_no_passdown;
pool->process_prepared_discard = process_prepared_discard_no_passdown;
}
}
static void set_pool_mode(struct pool *pool, enum pool_mode new_mode)
{
struct pool_c *pt = pool->ti->private;
bool needs_check = dm_pool_metadata_needs_check(pool->pmd);
enum pool_mode old_mode = get_pool_mode(pool);
unsigned long no_space_timeout = READ_ONCE(no_space_timeout_secs) * HZ;
/*
* Never allow the pool to transition to PM_WRITE mode if user
* intervention is required to verify metadata and data consistency.
*/
if (new_mode == PM_WRITE && needs_check) {
DMERR("%s: unable to switch pool to write mode until repaired.",
dm_device_name(pool->pool_md));
if (old_mode != new_mode)
new_mode = old_mode;
else
new_mode = PM_READ_ONLY;
}
/*
* If we were in PM_FAIL mode, rollback of metadata failed. We're
* not going to recover without a thin_repair. So we never let the
* pool move out of the old mode.
*/
if (old_mode == PM_FAIL)
new_mode = old_mode;
switch (new_mode) {
case PM_FAIL:
dm_pool_metadata_read_only(pool->pmd);
pool->process_bio = process_bio_fail;
pool->process_discard = process_bio_fail;
pool->process_cell = process_cell_fail;
pool->process_discard_cell = process_cell_fail;
pool->process_prepared_mapping = process_prepared_mapping_fail;
pool->process_prepared_discard = process_prepared_discard_fail;
error_retry_list(pool);
break;
case PM_OUT_OF_METADATA_SPACE:
case PM_READ_ONLY:
dm_pool_metadata_read_only(pool->pmd);
pool->process_bio = process_bio_read_only;
pool->process_discard = process_bio_success;
pool->process_cell = process_cell_read_only;
pool->process_discard_cell = process_cell_success;
pool->process_prepared_mapping = process_prepared_mapping_fail;
pool->process_prepared_discard = process_prepared_discard_success;
error_retry_list(pool);
break;
case PM_OUT_OF_DATA_SPACE:
/*
* Ideally we'd never hit this state; the low water mark
* would trigger userland to extend the pool before we
* completely run out of data space. However, many small
* IOs to unprovisioned space can consume data space at an
* alarming rate. Adjust your low water mark if you're
* frequently seeing this mode.
*/
pool->out_of_data_space = true;
pool->process_bio = process_bio_read_only;
pool->process_discard = process_discard_bio;
pool->process_cell = process_cell_read_only;
pool->process_prepared_mapping = process_prepared_mapping;
set_discard_callbacks(pool);
if (!pool->pf.error_if_no_space && no_space_timeout)
queue_delayed_work(pool->wq, &pool->no_space_timeout, no_space_timeout);
break;
case PM_WRITE:
if (old_mode == PM_OUT_OF_DATA_SPACE)
cancel_delayed_work_sync(&pool->no_space_timeout);
pool->out_of_data_space = false;
pool->pf.error_if_no_space = pt->requested_pf.error_if_no_space;
dm_pool_metadata_read_write(pool->pmd);
pool->process_bio = process_bio;
pool->process_discard = process_discard_bio;
pool->process_cell = process_cell;
pool->process_prepared_mapping = process_prepared_mapping;
set_discard_callbacks(pool);
break;
}
pool->pf.mode = new_mode;
/*
* The pool mode may have changed, sync it so bind_control_target()
* doesn't cause an unexpected mode transition on resume.
*/
pt->adjusted_pf.mode = new_mode;
if (old_mode != new_mode)
notify_of_pool_mode_change(pool);
}
static void abort_transaction(struct pool *pool)
{
const char *dev_name = dm_device_name(pool->pool_md);
DMERR_LIMIT("%s: aborting current metadata transaction", dev_name);
if (dm_pool_abort_metadata(pool->pmd)) {
DMERR("%s: failed to abort metadata transaction", dev_name);
set_pool_mode(pool, PM_FAIL);
}
if (dm_pool_metadata_set_needs_check(pool->pmd)) {
DMERR("%s: failed to set 'needs_check' flag in metadata", dev_name);
set_pool_mode(pool, PM_FAIL);
}
}
static void metadata_operation_failed(struct pool *pool, const char *op, int r)
{
DMERR_LIMIT("%s: metadata operation '%s' failed: error = %d",
dm_device_name(pool->pool_md), op, r);
abort_transaction(pool);
set_pool_mode(pool, PM_READ_ONLY);
}
/*----------------------------------------------------------------*/
/*
* Mapping functions.
*/
/*
* Called only while mapping a thin bio to hand it over to the workqueue.
*/
static void thin_defer_bio(struct thin_c *tc, struct bio *bio)
{
unsigned long flags;
struct pool *pool = tc->pool;
spin_lock_irqsave(&tc->lock, flags);
bio_list_add(&tc->deferred_bio_list, bio);
spin_unlock_irqrestore(&tc->lock, flags);
wake_worker(pool);
}
static void thin_defer_bio_with_throttle(struct thin_c *tc, struct bio *bio)
{
struct pool *pool = tc->pool;
throttle_lock(&pool->throttle);
thin_defer_bio(tc, bio);
throttle_unlock(&pool->throttle);
}
static void thin_defer_cell(struct thin_c *tc, struct dm_bio_prison_cell *cell)
{
unsigned long flags;
struct pool *pool = tc->pool;
throttle_lock(&pool->throttle);
spin_lock_irqsave(&tc->lock, flags);
list_add_tail(&cell->user_list, &tc->deferred_cells);
spin_unlock_irqrestore(&tc->lock, flags);
throttle_unlock(&pool->throttle);
wake_worker(pool);
}
static void thin_hook_bio(struct thin_c *tc, struct bio *bio)
{
struct dm_thin_endio_hook *h = dm_per_bio_data(bio, sizeof(struct dm_thin_endio_hook));
h->tc = tc;
h->shared_read_entry = NULL;
h->all_io_entry = NULL;
h->overwrite_mapping = NULL;
h->cell = NULL;
}
/*
* Non-blocking function called from the thin target's map function.
*/
static int thin_bio_map(struct dm_target *ti, struct bio *bio)
{
int r;
struct thin_c *tc = ti->private;
dm_block_t block = get_bio_block(tc, bio);
struct dm_thin_device *td = tc->td;
struct dm_thin_lookup_result result;
struct dm_bio_prison_cell *virt_cell, *data_cell;
struct dm_cell_key key;
thin_hook_bio(tc, bio);
if (tc->requeue_mode) {
bio->bi_status = BLK_STS_DM_REQUEUE;
bio_endio(bio);
return DM_MAPIO_SUBMITTED;
}
if (get_pool_mode(tc->pool) == PM_FAIL) {
bio_io_error(bio);
return DM_MAPIO_SUBMITTED;
}
if (op_is_flush(bio->bi_opf) || bio_op(bio) == REQ_OP_DISCARD) {
thin_defer_bio_with_throttle(tc, bio);
return DM_MAPIO_SUBMITTED;
}
/*
* We must hold the virtual cell before doing the lookup, otherwise
* there's a race with discard.
*/
build_virtual_key(tc->td, block, &key);
if (bio_detain(tc->pool, &key, bio, &virt_cell))
return DM_MAPIO_SUBMITTED;
r = dm_thin_find_block(td, block, 0, &result);
/*
* Note that we defer readahead too.
*/
switch (r) {
case 0:
if (unlikely(result.shared)) {
/*
* We have a race condition here between the
* result.shared value returned by the lookup and
* snapshot creation, which may cause new
* sharing.
*
* To avoid this always quiesce the origin before
* taking the snap. You want to do this anyway to
* ensure a consistent application view
* (i.e. lockfs).
*
* More distant ancestors are irrelevant. The
* shared flag will be set in their case.
*/
thin_defer_cell(tc, virt_cell);
return DM_MAPIO_SUBMITTED;
}
build_data_key(tc->td, result.block, &key);
if (bio_detain(tc->pool, &key, bio, &data_cell)) {
cell_defer_no_holder(tc, virt_cell);
return DM_MAPIO_SUBMITTED;
}
inc_all_io_entry(tc->pool, bio);
cell_defer_no_holder(tc, data_cell);
cell_defer_no_holder(tc, virt_cell);
remap(tc, bio, result.block);
return DM_MAPIO_REMAPPED;
case -ENODATA:
case -EWOULDBLOCK:
thin_defer_cell(tc, virt_cell);
return DM_MAPIO_SUBMITTED;
default:
/*
* Must always call bio_io_error on failure.
* dm_thin_find_block can fail with -EINVAL if the
* pool is switched to fail-io mode.
*/
bio_io_error(bio);
cell_defer_no_holder(tc, virt_cell);
return DM_MAPIO_SUBMITTED;
}
}
static int pool_is_congested(struct dm_target_callbacks *cb, int bdi_bits)
{
struct pool_c *pt = container_of(cb, struct pool_c, callbacks);
struct request_queue *q;
if (get_pool_mode(pt->pool) == PM_OUT_OF_DATA_SPACE)
return 1;
q = bdev_get_queue(pt->data_dev->bdev);
return bdi_congested(q->backing_dev_info, bdi_bits);
}
static void requeue_bios(struct pool *pool)
{
unsigned long flags;
struct thin_c *tc;
rcu_read_lock();
list_for_each_entry_rcu(tc, &pool->active_thins, list) {
spin_lock_irqsave(&tc->lock, flags);
bio_list_merge(&tc->deferred_bio_list, &tc->retry_on_resume_list);
bio_list_init(&tc->retry_on_resume_list);
spin_unlock_irqrestore(&tc->lock, flags);
}
rcu_read_unlock();
}
/*----------------------------------------------------------------
* Binding of control targets to a pool object
*--------------------------------------------------------------*/
static bool data_dev_supports_discard(struct pool_c *pt)
{
struct request_queue *q = bdev_get_queue(pt->data_dev->bdev);
return q && blk_queue_discard(q);
}
static bool is_factor<