aboutsummaryrefslogtreecommitdiff
path: root/drivers/md
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/md')
-rw-r--r--drivers/md/Makefile2
-rw-r--r--drivers/md/dm-crypt.c56
-rw-r--r--drivers/md/dm-delay.c3
-rw-r--r--drivers/md/dm-exception-store.c108
-rw-r--r--drivers/md/dm-io.c2
-rw-r--r--drivers/md/dm-ioctl.c4
-rw-r--r--drivers/md/dm-kcopyd.c14
-rw-r--r--drivers/md/dm-linear.c15
-rw-r--r--drivers/md/dm-log.c2
-rw-r--r--drivers/md/dm-mpath.c19
-rw-r--r--drivers/md/dm-path-selector.c3
-rw-r--r--drivers/md/dm-raid1.c791
-rw-r--r--drivers/md/dm-region-hash.c704
-rw-r--r--drivers/md/dm-round-robin.c3
-rw-r--r--drivers/md/dm-snap.c11
-rw-r--r--drivers/md/dm-snap.h5
-rw-r--r--drivers/md/dm-stripe.c6
-rw-r--r--drivers/md/dm-table.c37
-rw-r--r--drivers/md/dm-zero.c2
-rw-r--r--drivers/md/dm.c76
-rw-r--r--drivers/md/dm.h9
-rw-r--r--drivers/md/md.c20
-rw-r--r--drivers/md/multipath.c4
23 files changed, 1020 insertions, 876 deletions
diff --git a/drivers/md/Makefile b/drivers/md/Makefile
index f1ef33dfd8c..1c615804ea7 100644
--- a/drivers/md/Makefile
+++ b/drivers/md/Makefile
@@ -34,7 +34,7 @@ obj-$(CONFIG_DM_CRYPT) += dm-crypt.o
obj-$(CONFIG_DM_DELAY) += dm-delay.o
obj-$(CONFIG_DM_MULTIPATH) += dm-multipath.o dm-round-robin.o
obj-$(CONFIG_DM_SNAPSHOT) += dm-snapshot.o
-obj-$(CONFIG_DM_MIRROR) += dm-mirror.o dm-log.o
+obj-$(CONFIG_DM_MIRROR) += dm-mirror.o dm-log.o dm-region-hash.o
obj-$(CONFIG_DM_ZERO) += dm-zero.o
quiet_cmd_unroll = UNROLL $@
diff --git a/drivers/md/dm-crypt.c b/drivers/md/dm-crypt.c
index 682ef9e6acd..ce26c84af06 100644
--- a/drivers/md/dm-crypt.c
+++ b/drivers/md/dm-crypt.c
@@ -23,7 +23,7 @@
#include <asm/page.h>
#include <asm/unaligned.h>
-#include "dm.h"
+#include <linux/device-mapper.h>
#define DM_MSG_PREFIX "crypt"
#define MESG_STR(x) x, sizeof(x)
@@ -56,6 +56,7 @@ struct dm_crypt_io {
atomic_t pending;
int error;
sector_t sector;
+ struct dm_crypt_io *base_io;
};
struct dm_crypt_request {
@@ -93,7 +94,6 @@ struct crypt_config {
struct workqueue_struct *io_queue;
struct workqueue_struct *crypt_queue;
- wait_queue_head_t writeq;
/*
* crypto related data
@@ -534,6 +534,7 @@ static struct dm_crypt_io *crypt_io_alloc(struct dm_target *ti,
io->base_bio = bio;
io->sector = sector;
io->error = 0;
+ io->base_io = NULL;
atomic_set(&io->pending, 0);
return io;
@@ -547,6 +548,7 @@ static void crypt_inc_pending(struct dm_crypt_io *io)
/*
* One of the bios was finished. Check for completion of
* the whole request and correctly clean up the buffer.
+ * If base_io is set, wait for the last fragment to complete.
*/
static void crypt_dec_pending(struct dm_crypt_io *io)
{
@@ -555,7 +557,14 @@ static void crypt_dec_pending(struct dm_crypt_io *io)
if (!atomic_dec_and_test(&io->pending))
return;
- bio_endio(io->base_bio, io->error);
+ if (likely(!io->base_io))
+ bio_endio(io->base_bio, io->error);
+ else {
+ if (io->error && !io->base_io->error)
+ io->base_io->error = io->error;
+ crypt_dec_pending(io->base_io);
+ }
+
mempool_free(io, cc->io_pool);
}
@@ -646,10 +655,7 @@ static void kcryptd_io_read(struct dm_crypt_io *io)
static void kcryptd_io_write(struct dm_crypt_io *io)
{
struct bio *clone = io->ctx.bio_out;
- struct crypt_config *cc = io->target->private;
-
generic_make_request(clone);
- wake_up(&cc->writeq);
}
static void kcryptd_io(struct work_struct *work)
@@ -688,7 +694,6 @@ static void kcryptd_crypt_write_io_submit(struct dm_crypt_io *io,
BUG_ON(io->ctx.idx_out < clone->bi_vcnt);
clone->bi_sector = cc->start + io->sector;
- io->sector += bio_sectors(clone);
if (async)
kcryptd_queue_io(io);
@@ -700,16 +705,18 @@ static void kcryptd_crypt_write_convert(struct dm_crypt_io *io)
{
struct crypt_config *cc = io->target->private;
struct bio *clone;
+ struct dm_crypt_io *new_io;
int crypt_finished;
unsigned out_of_pages = 0;
unsigned remaining = io->base_bio->bi_size;
+ sector_t sector = io->sector;
int r;
/*
* Prevent io from disappearing until this function completes.
*/
crypt_inc_pending(io);
- crypt_convert_init(cc, &io->ctx, NULL, io->base_bio, io->sector);
+ crypt_convert_init(cc, &io->ctx, NULL, io->base_bio, sector);
/*
* The allocated buffers can be smaller than the whole bio,
@@ -726,6 +733,7 @@ static void kcryptd_crypt_write_convert(struct dm_crypt_io *io)
io->ctx.idx_out = 0;
remaining -= clone->bi_size;
+ sector += bio_sectors(clone);
crypt_inc_pending(io);
r = crypt_convert(cc, &io->ctx);
@@ -741,6 +749,8 @@ static void kcryptd_crypt_write_convert(struct dm_crypt_io *io)
*/
if (unlikely(r < 0))
break;
+
+ io->sector = sector;
}
/*
@@ -750,8 +760,33 @@ static void kcryptd_crypt_write_convert(struct dm_crypt_io *io)
if (unlikely(out_of_pages))
congestion_wait(WRITE, HZ/100);
- if (unlikely(remaining))
- wait_event(cc->writeq, !atomic_read(&io->ctx.pending));
+ /*
+ * With async crypto it is unsafe to share the crypto context
+ * between fragments, so switch to a new dm_crypt_io structure.
+ */
+ if (unlikely(!crypt_finished && remaining)) {
+ new_io = crypt_io_alloc(io->target, io->base_bio,
+ sector);
+ crypt_inc_pending(new_io);
+ crypt_convert_init(cc, &new_io->ctx, NULL,
+ io->base_bio, sector);
+ new_io->ctx.idx_in = io->ctx.idx_in;
+ new_io->ctx.offset_in = io->ctx.offset_in;
+
+ /*
+ * Fragments after the first use the base_io
+ * pending count.
+ */
+ if (!io->base_io)
+ new_io->base_io = io;
+ else {
+ new_io->base_io = io->base_io;
+ crypt_inc_pending(io->base_io);
+ crypt_dec_pending(io);
+ }
+
+ io = new_io;
+ }
}
crypt_dec_pending(io);
@@ -1078,7 +1113,6 @@ static int crypt_ctr(struct dm_target *ti, unsigned int argc, char **argv)
goto bad_crypt_queue;
}
- init_waitqueue_head(&cc->writeq);
ti->private = cc;
return 0;
diff --git a/drivers/md/dm-delay.c b/drivers/md/dm-delay.c
index bdd37f881c4..848b381f117 100644
--- a/drivers/md/dm-delay.c
+++ b/drivers/md/dm-delay.c
@@ -13,7 +13,8 @@
#include <linux/bio.h>
#include <linux/slab.h>
-#include "dm.h"
+#include <linux/device-mapper.h>
+
#include "dm-bio-list.h"
#define DM_MSG_PREFIX "delay"
diff --git a/drivers/md/dm-exception-store.c b/drivers/md/dm-exception-store.c
index 769ab677f8e..01590f3e000 100644
--- a/drivers/md/dm-exception-store.c
+++ b/drivers/md/dm-exception-store.c
@@ -7,7 +7,6 @@
* This file is released under the GPL.
*/
-#include "dm.h"
#include "dm-snap.h"
#include <linux/mm.h>
@@ -105,6 +104,11 @@ struct pstore {
void *area;
/*
+ * An area of zeros used to clear the next area.
+ */
+ void *zero_area;
+
+ /*
* Used to keep track of which metadata area the data in
* 'chunk' refers to.
*/
@@ -149,6 +153,13 @@ static int alloc_area(struct pstore *ps)
if (!ps->area)
return r;
+ ps->zero_area = vmalloc(len);
+ if (!ps->zero_area) {
+ vfree(ps->area);
+ return r;
+ }
+ memset(ps->zero_area, 0, len);
+
return 0;
}
@@ -156,6 +167,8 @@ static void free_area(struct pstore *ps)
{
vfree(ps->area);
ps->area = NULL;
+ vfree(ps->zero_area);
+ ps->zero_area = NULL;
}
struct mdata_req {
@@ -220,25 +233,41 @@ static chunk_t area_location(struct pstore *ps, chunk_t area)
* Read or write a metadata area. Remembering to skip the first
* chunk which holds the header.
*/
-static int area_io(struct pstore *ps, chunk_t area, int rw)
+static int area_io(struct pstore *ps, int rw)
{
int r;
chunk_t chunk;
- chunk = area_location(ps, area);
+ chunk = area_location(ps, ps->current_area);
r = chunk_io(ps, chunk, rw, 0);
if (r)
return r;
- ps->current_area = area;
return 0;
}
-static int zero_area(struct pstore *ps, chunk_t area)
+static void zero_memory_area(struct pstore *ps)
{
memset(ps->area, 0, ps->snap->chunk_size << SECTOR_SHIFT);
- return area_io(ps, area, WRITE);
+}
+
+static int zero_disk_area(struct pstore *ps, chunk_t area)
+{
+ struct dm_io_region where = {
+ .bdev = ps->snap->cow->bdev,
+ .sector = ps->snap->chunk_size * area_location(ps, area),
+ .count = ps->snap->chunk_size,
+ };
+ struct dm_io_request io_req = {
+ .bi_rw = WRITE,
+ .mem.type = DM_IO_VMA,
+ .mem.ptr.vma = ps->zero_area,
+ .client = ps->io_client,
+ .notify.fn = NULL,
+ };
+
+ return dm_io(&io_req, 1, &where, NULL);
}
static int read_header(struct pstore *ps, int *new_snapshot)
@@ -411,15 +440,14 @@ static int insert_exceptions(struct pstore *ps, int *full)
static int read_exceptions(struct pstore *ps)
{
- chunk_t area;
int r, full = 1;
/*
* Keeping reading chunks and inserting exceptions until
* we find a partially full area.
*/
- for (area = 0; full; area++) {
- r = area_io(ps, area, READ);
+ for (ps->current_area = 0; full; ps->current_area++) {
+ r = area_io(ps, READ);
if (r)
return r;
@@ -428,6 +456,8 @@ static int read_exceptions(struct pstore *ps)
return r;
}
+ ps->current_area--;
+
return 0;
}
@@ -486,12 +516,13 @@ static int persistent_read_metadata(struct exception_store *store)
return r;
}
- r = zero_area(ps, 0);
+ ps->current_area = 0;
+ zero_memory_area(ps);
+ r = zero_disk_area(ps, 0);
if (r) {
- DMWARN("zero_area(0) failed");
+ DMWARN("zero_disk_area(0) failed");
return r;
}
-
} else {
/*
* Sanity checks.
@@ -551,7 +582,6 @@ static void persistent_commit(struct exception_store *store,
void (*callback) (void *, int success),
void *callback_context)
{
- int r;
unsigned int i;
struct pstore *ps = get_info(store);
struct disk_exception de;
@@ -572,33 +602,41 @@ static void persistent_commit(struct exception_store *store,
cb->context = callback_context;
/*
- * If there are no more exceptions in flight, or we have
- * filled this metadata area we commit the exceptions to
- * disk.
+ * If there are exceptions in flight and we have not yet
+ * filled this metadata area there's nothing more to do.
*/
- if (atomic_dec_and_test(&ps->pending_count) ||
- (ps->current_committed == ps->exceptions_per_area)) {
- r = area_io(ps, ps->current_area, WRITE);
- if (r)
- ps->valid = 0;
+ if (!atomic_dec_and_test(&ps->pending_count) &&
+ (ps->current_committed != ps->exceptions_per_area))
+ return;
- /*
- * Have we completely filled the current area ?
- */
- if (ps->current_committed == ps->exceptions_per_area) {
- ps->current_committed = 0;
- r = zero_area(ps, ps->current_area + 1);
- if (r)
- ps->valid = 0;
- }
+ /*
+ * If we completely filled the current area, then wipe the next one.
+ */
+ if ((ps->current_committed == ps->exceptions_per_area) &&
+ zero_disk_area(ps, ps->current_area + 1))
+ ps->valid = 0;
- for (i = 0; i < ps->callback_count; i++) {
- cb = ps->callbacks + i;
- cb->callback(cb->context, r == 0 ? 1 : 0);
- }
+ /*
+ * Commit exceptions to disk.
+ */
+ if (ps->valid && area_io(ps, WRITE))
+ ps->valid = 0;
- ps->callback_count = 0;
+ /*
+ * Advance to the next area if this one is full.
+ */
+ if (ps->current_committed == ps->exceptions_per_area) {
+ ps->current_committed = 0;
+ ps->current_area++;
+ zero_memory_area(ps);
}
+
+ for (i = 0; i < ps->callback_count; i++) {
+ cb = ps->callbacks + i;
+ cb->callback(cb->context, ps->valid);
+ }
+
+ ps->callback_count = 0;
}
static void persistent_drop(struct exception_store *store)
diff --git a/drivers/md/dm-io.c b/drivers/md/dm-io.c
index 4789c42d9a3..2fd6d445063 100644
--- a/drivers/md/dm-io.c
+++ b/drivers/md/dm-io.c
@@ -5,7 +5,7 @@
* This file is released under the GPL.
*/
-#include "dm.h"
+#include <linux/device-mapper.h>
#include <linux/bio.h>
#include <linux/mempool.h>
diff --git a/drivers/md/dm-ioctl.c b/drivers/md/dm-ioctl.c
index dca401dc70a..777c948180f 100644
--- a/drivers/md/dm-ioctl.c
+++ b/drivers/md/dm-ioctl.c
@@ -988,9 +988,9 @@ static int dev_wait(struct dm_ioctl *param, size_t param_size)
return r;
}
-static inline int get_mode(struct dm_ioctl *param)
+static inline fmode_t get_mode(struct dm_ioctl *param)
{
- int mode = FMODE_READ | FMODE_WRITE;
+ fmode_t mode = FMODE_READ | FMODE_WRITE;
if (param->flags & DM_READONLY_FLAG)
mode = FMODE_READ;
diff --git a/drivers/md/dm-kcopyd.c b/drivers/md/dm-kcopyd.c
index 996802b8a45..3073618269e 100644
--- a/drivers/md/dm-kcopyd.c
+++ b/drivers/md/dm-kcopyd.c
@@ -22,6 +22,7 @@
#include <linux/vmalloc.h>
#include <linux/workqueue.h>
#include <linux/mutex.h>
+#include <linux/device-mapper.h>
#include <linux/dm-kcopyd.h>
#include "dm.h"
@@ -268,6 +269,17 @@ static void push(struct list_head *jobs, struct kcopyd_job *job)
spin_unlock_irqrestore(&kc->job_lock, flags);
}
+
+static void push_head(struct list_head *jobs, struct kcopyd_job *job)
+{
+ unsigned long flags;
+ struct dm_kcopyd_client *kc = job->kc;
+
+ spin_lock_irqsave(&kc->job_lock, flags);
+ list_add(&job->list, jobs);
+ spin_unlock_irqrestore(&kc->job_lock, flags);
+}
+
/*
* These three functions process 1 item from the corresponding
* job list.
@@ -398,7 +410,7 @@ static int process_jobs(struct list_head *jobs, struct dm_kcopyd_client *kc,
* We couldn't service this job ATM, so
* push this job back onto the list.
*/
- push(jobs, job);
+ push_head(jobs, job);
break;
}
diff --git a/drivers/md/dm-linear.c b/drivers/md/dm-linear.c
index 6449bcdf84c..44042becad8 100644
--- a/drivers/md/dm-linear.c
+++ b/drivers/md/dm-linear.c
@@ -5,12 +5,12 @@
*/
#include "dm.h"
-
#include <linux/module.h>
#include <linux/init.h>
#include <linux/blkdev.h>
#include <linux/bio.h>
#include <linux/slab.h>
+#include <linux/device-mapper.h>
#define DM_MSG_PREFIX "linear"
@@ -110,20 +110,11 @@ static int linear_status(struct dm_target *ti, status_type_t type,
return 0;
}
-static int linear_ioctl(struct dm_target *ti, struct inode *inode,
- struct file *filp, unsigned int cmd,
+static int linear_ioctl(struct dm_target *ti, unsigned int cmd,
unsigned long arg)
{
struct linear_c *lc = (struct linear_c *) ti->private;
- struct block_device *bdev = lc->dev->bdev;
- struct file fake_file = {};
- struct dentry fake_dentry = {};
-
- fake_file.f_mode = lc->dev->mode;
- fake_file.f_path.dentry = &fake_dentry;
- fake_dentry.d_inode = bdev->bd_inode;
-
- return blkdev_driver_ioctl(bdev->bd_inode, &fake_file, bdev->bd_disk, cmd, arg);
+ return __blkdev_driver_ioctl(lc->dev->bdev, lc->dev->mode, cmd, arg);
}
static int linear_merge(struct dm_target *ti, struct bvec_merge_data *bvm,
diff --git a/drivers/md/dm-log.c b/drivers/md/dm-log.c
index 5b48478c79f..a8c0fc79ca7 100644
--- a/drivers/md/dm-log.c
+++ b/drivers/md/dm-log.c
@@ -12,7 +12,7 @@
#include <linux/dm-io.h>
#include <linux/dm-dirty-log.h>
-#include "dm.h"
+#include <linux/device-mapper.h>
#define DM_MSG_PREFIX "dirty region log"
diff --git a/drivers/md/dm-mpath.c b/drivers/md/dm-mpath.c
index 103304c1e3b..4840733cd90 100644
--- a/drivers/md/dm-mpath.c
+++ b/drivers/md/dm-mpath.c
@@ -5,7 +5,8 @@
* This file is released under the GPL.
*/
-#include "dm.h"
+#include <linux/device-mapper.h>
+
#include "dm-path-selector.h"
#include "dm-bio-list.h"
#include "dm-bio-record.h"
@@ -849,7 +850,7 @@ static int multipath_map(struct dm_target *ti, struct bio *bio,
dm_bio_record(&mpio->details, bio);
map_context->ptr = mpio;
- bio->bi_rw |= (1 << BIO_RW_FAILFAST);
+ bio->bi_rw |= (1 << BIO_RW_FAILFAST_TRANSPORT);
r = map_io(m, bio, mpio, 0);
if (r < 0 || r == DM_MAPIO_REQUEUE)
mempool_free(mpio, m->mpio_pool);
@@ -1395,19 +1396,15 @@ error:
return -EINVAL;
}
-static int multipath_ioctl(struct dm_target *ti, struct inode *inode,
- struct file *filp, unsigned int cmd,
+static int multipath_ioctl(struct dm_target *ti, unsigned int cmd,
unsigned long arg)
{
struct multipath *m = (struct multipath *) ti->private;
struct block_device *bdev = NULL;
+ fmode_t mode = 0;
unsigned long flags;
- struct file fake_file = {};
- struct dentry fake_dentry = {};
int r = 0;
- fake_file.f_path.dentry = &fake_dentry;
-
spin_lock_irqsave(&m->lock, flags);
if (!m->current_pgpath)
@@ -1415,8 +1412,7 @@ static int multipath_ioctl(struct dm_target *ti, struct inode *inode,
if (m->current_pgpath) {
bdev = m->current_pgpath->path.dev->bdev;
- fake_dentry.d_inode = bdev->bd_inode;
- fake_file.f_mode = m->current_pgpath->path.dev->mode;
+ mode = m->current_pgpath->path.dev->mode;
}
if (m->queue_io)
@@ -1426,8 +1422,7 @@ static int multipath_ioctl(struct dm_target *ti, struct inode *inode,
spin_unlock_irqrestore(&m->lock, flags);
- return r ? : blkdev_driver_ioctl(bdev->bd_inode, &fake_file,
- bdev->bd_disk, cmd, arg);
+ return r ? : __blkdev_driver_ioctl(bdev, mode, cmd, arg);
}
/*-----------------------------------------------------------------
diff --git a/drivers/md/dm-path-selector.c b/drivers/md/dm-path-selector.c
index ca1bb636a3e..96ea226155b 100644
--- a/drivers/md/dm-path-selector.c
+++ b/drivers/md/dm-path-selector.c
@@ -9,7 +9,8 @@
* Path selector registration.
*/
-#include "dm.h"
+#include <linux/device-mapper.h>
+
#include "dm-path-selector.h"
#include <linux/slab.h>
diff --git a/drivers/md/dm-raid1.c b/drivers/md/dm-raid1.c
index 29913e42c4a..92dcc06832a 100644
--- a/drivers/md/dm-raid1.c
+++ b/drivers/md/dm-raid1.c
@@ -1,30 +1,30 @@
/*
* Copyright (C) 2003 Sistina Software Limited.
+ * Copyright (C) 2005-2008 Red Hat, Inc. All rights reserved.
*
* This file is released under the GPL.
*/
-#include "dm.h"
#include "dm-bio-list.h"
#include "dm-bio-record.h"
-#include <linux/ctype.h>
#include <linux/init.h>
#include <linux/mempool.h>
#include <linux/module.h>
#include <linux/pagemap.h>
#include <linux/slab.h>
-#include <linux/time.h>
-#include <linux/vmalloc.h>
#include <linux/workqueue.h>
-#include <linux/log2.h>
-#include <linux/hardirq.h>
+#include <linux/device-mapper.h>
#include <linux/dm-io.h>
#include <linux/dm-dirty-log.h>
#include <linux/dm-kcopyd.h>
+#include <linux/dm-region-hash.h>
#define DM_MSG_PREFIX "raid1"
+
+#define MAX_RECOVERY 1 /* Maximum number of regions recovered in parallel. */
#define DM_IO_PAGES 64
+#define DM_KCOPYD_PAGES 64
#define DM_RAID1_HANDLE_ERRORS 0x01
#define errors_handled(p) ((p)->features & DM_RAID1_HANDLE_ERRORS)
@@ -32,87 +32,6 @@
static DECLARE_WAIT_QUEUE_HEAD(_kmirrord_recovery_stopped);
/*-----------------------------------------------------------------
- * Region hash
- *
- * The mirror splits itself up into discrete regions. Each
- * region can be in one of three states: clean, dirty,
- * nosync. There is no need to put clean regions in the hash.
- *
- * In addition to being present in the hash table a region _may_
- * be present on one of three lists.
- *
- * clean_regions: Regions on this list have no io pending to
- * them, they are in sync, we are no longer interested in them,
- * they are dull. rh_update_states() will remove them from the
- * hash table.
- *
- * quiesced_regions: These regions have been spun down, ready
- * for recovery. rh_recovery_start() will remove regions from
- * this list and hand them to kmirrord, which will schedule the
- * recovery io with kcopyd.
- *
- * recovered_regions: Regions that kcopyd has successfully
- * recovered. rh_update_states() will now schedule any delayed
- * io, up the recovery_count, and remove the region from the
- * hash.
- *
- * There are 2 locks:
- * A rw spin lock 'hash_lock' protects just the hash table,
- * this is never held in write mode from interrupt context,
- * which I believe means that we only have to disable irqs when
- * doing a write lock.
- *
- * An ordinary spin lock 'region_lock' that protects the three
- * lists in the region_hash, with the 'state', 'list' and
- * 'bhs_delayed' fields of the regions. This is used from irq
- * context, so all other uses will have to suspend local irqs.
- *---------------------------------------------------------------*/
-struct mirror_set;
-struct region_hash {
- struct mirror_set *ms;
- uint32_t region_size;
- unsigned region_shift;
-
- /* holds persistent region state */
- struct dm_dirty_log *log;
-
- /* hash table */
- rwlock_t hash_lock;
- mempool_t *region_pool;
- unsigned int mask;
- unsigned int nr_buckets;
- struct list_head *buckets;
-
- spinlock_t region_lock;
- atomic_t recovery_in_flight;
- struct semaphore recovery_count;
- struct list_head clean_regions;
- struct list_head quiesced_regions;
- struct list_head recovered_regions;
- struct list_head failed_recovered_regions;
-};
-
-enum {
- RH_CLEAN,
- RH_DIRTY,
- RH_NOSYNC,
- RH_RECOVERING
-};
-
-struct region {
- struct region_hash *rh; /* FIXME: can we get rid of this ? */
- region_t key;
- int state;
-
- struct list_head hash_list;
- struct list_head list;
-
- atomic_t pending;
- struct bio_list delayed_bios;
-};
-
-
-/*-----------------------------------------------------------------
* Mirror set structures.
*---------------------------------------------------------------*/
enum dm_raid1_error {
@@ -132,8 +51,7 @@ struct mirror {
struct mirror_set {
struct dm_target *ti;
struct list_head list;
- struct region_hash rh;
- struct dm_kcopyd_client *kcopyd_client;
+
uint64_t features;
spinlock_t lock; /* protects the lists */
@@ -141,6 +59,8 @@ struct mirror_set {
struct bio_list writes;
struct bio_list failures;
+ struct dm_region_hash *rh;
+ struct dm_kcopyd_client *kcopyd_client;
struct dm_io_client *io_client;
mempool_t *read_record_pool;
@@ -159,25 +79,14 @@ struct mirror_set {
struct work_struct trigger_event;
- unsigned int nr_mirrors;
+ unsigned nr_mirrors;
struct mirror mirror[0];
};
-/*
- * Conversion fns
- */
-static inline region_t bio_to_region(struct region_hash *rh, struct bio *bio)
-{
- return (bio->bi_sector - rh->ms->ti->begin) >> rh->region_shift;
-}
-
-static inline sector_t region_to_sector(struct region_hash *rh, region_t region)
+static void wakeup_mirrord(void *context)
{
- return region << rh->region_shift;
-}
+ struct mirror_set *ms = context;
-static void wake(struct mirror_set *ms)
-{
queue_work(ms->kmirrord_wq, &ms->kmirrord_work);
}
@@ -186,7 +95,7 @@ static void delayed_wake_fn(unsigned long data)
struct mirror_set *ms = (struct mirror_set *) data;
clear_bit(0, &ms->timer_pending);
- wake(ms);
+ wakeup_mirrord(ms);
}
static void delayed_wake(struct mirror_set *ms)
@@ -200,473 +109,34 @@ static void delayed_wake(struct mirror_set *ms)
add_timer(&ms->timer);
}
-/* FIXME move this */
-static void queue_bio(struct mirror_set *ms, struct bio *bio, int rw);
-
-#define MIN_REGIONS 64
-#define MAX_RECOVERY 1
-static int rh_init(struct region_hash *rh, struct mirror_set *ms,
- struct dm_dirty_log *log, uint32_t region_size,
- region_t nr_regions)
+static void wakeup_all_recovery_waiters(void *context)
{
- unsigned int nr_buckets, max_buckets;
- size_t i;
-
- /*
- * Calculate a suitable number of buckets for our hash
- * table.
- */
- max_buckets = nr_regions >> 6;
- for (nr_buckets = 128u; nr_buckets < max_buckets; nr_buckets <<= 1)
- ;
- nr_buckets >>= 1;
-
- rh->ms = ms;
- rh->log = log;
- rh->region_size = region_size;
- rh->region_shift = ffs(region_size) - 1;
- rwlock_init(&rh->hash_lock);
- rh->mask = nr_buckets - 1;
- rh->nr_buckets = nr_buckets;
-
- rh->buckets = vmalloc(nr_buckets * sizeof(*rh->buckets));
- if (!rh->buckets) {
- DMERR("unable to allocate region hash memory");
- return -ENOMEM;
- }
-
- for (i = 0; i < nr_buckets; i++)
- INIT_LIST_HEAD(rh->buckets + i);
-
- spin_lock_init(&rh->region_lock);
- sema_init(&rh->recovery_count, 0);
- atomic_set(&rh->recovery_in_flight, 0);
- INIT_LIST_HEAD(&rh->clean_regions);
- INIT_LIST_HEAD(&rh->quiesced_regions);
- INIT_LIST_HEAD(&rh->recovered_regions);
- INIT_LIST_HEAD(&rh->failed_recovered_regions);
-
- rh->region_pool = mempool_create_kmalloc_pool(MIN_REGIONS,
- sizeof(struct region));
- if (!rh->region_pool) {
- vfree(rh->buckets);
- rh->buckets = NULL;
- return -ENOMEM;
- }
-
- return 0;
+ wake_up_all(&_kmirrord_recovery_stopped);
}
-static void rh_exit(struct region_hash *rh)
-{
- unsigned int h;
- struct region *reg, *nreg;
-
- BUG_ON(!list_empty(&rh->quiesced_regions));
- for (h = 0; h < rh->nr_buckets; h++) {
- list_for_each_entry_safe(reg, nreg, rh->buckets + h, hash_list) {
- BUG_ON(atomic_read(&reg->pending));
- mempool_free(reg, rh->region_pool);
- }
- }
-
- if (rh->log)
- dm_dirty_log_destroy(rh->log);
- if (rh->region_pool)
- mempool_destroy(rh->region_pool);
- vfree(rh->buckets);
-}
-
-#define RH_HASH_MULT 2654435387U
-
-static inline unsigned int rh_hash(struct region_hash *rh, region_t region)
-{
- return (unsigned int) ((region * RH_HASH_MULT) >> 12) & rh->mask;
-}
-
-static struct region *__rh_lookup(struct region_hash *rh, region_t region)
-{
- struct region *reg;
-
- list_for_each_entry (reg, rh->buckets + rh_hash(rh, region), hash_list)
- if (reg->key == region)
- return reg;
-
- return NULL;
-}
-
-static void __rh_insert(struct region_hash *rh, struct region *reg)
-{
- unsigned int h = rh_hash(rh, reg->key);
- list_add(&reg->hash_list, rh->buckets + h);
-}
-
-static struct region *__rh_alloc(struct region_hash *rh, region_t region)
-{
- struct region *reg, *nreg;
-
- read_unlock(&rh->hash_lock);
- nreg = mempool_alloc(rh->region_pool, GFP_ATOMIC);
- if (unlikely(!nreg))
- nreg = kmalloc(sizeof(struct region), GFP_NOIO);
- nreg->state = rh->log->type->in_sync(rh->log, region, 1) ?
- RH_CLEAN : RH_NOSYNC;
- nreg->rh = rh;
- nreg->key = region;
-
- INIT_LIST_HEAD(&nreg->list);
-
- atomic_set(&nreg->pending, 0);
- bio_list_init(&nreg->delayed_bios);
- write_lock_irq(&rh->hash_lock);
-
- reg = __rh_lookup(rh, region);
- if (reg)
- /* we lost the race */
- mempool_free(nreg, rh->region_pool);
-
- else {
- __rh_insert(rh, nreg);
- if (nreg->state == RH_CLEAN) {
- spin_lock(&rh->region_lock);
- list_add(&nreg->list, &rh->clean_regions);
- spin_unlock(&rh->region_lock);
- }
- reg = nreg;
- }
- write_unlock_irq(&rh->hash_lock);
- read_lock(&rh->hash_lock);
-
- return reg;
-}
-
-static inline struct region *__rh_find(struct region_hash *rh, region_t region)
-{
- struct region *reg;
-
- reg = __rh_lookup(rh, region);
- if (!reg)
- reg = __rh_alloc(rh, region);
-
- return reg;
-}
-
-static int rh_state(struct region_hash *rh, region_t region, int may_block)
-{
- int r;
- struct region *reg;
-
- read_lock(&rh->hash_lock);
- reg = __rh_lookup(rh, region);
- read_unlock(&rh->hash_lock);
-
- if (reg)
- return reg->state;
-
- /*
- * The region wasn't in the hash, so we fall back to the
- * dirty log.
- */
- r = rh->log->type->in_sync(rh->log, region, may_block);
-
- /*
- * Any error from the dirty log (eg. -EWOULDBLOCK) gets
- * taken as a RH_NOSYNC
- */
- return r == 1 ? RH_CLEAN : RH_NOSYNC;
-}
-
-static inline int rh_in_sync(struct region_hash *rh,
- region_t region, int may_block)
-{
- int state = rh_state(rh, region, may_block);
- return state == RH_CLEAN || state == RH_DIRTY;
-}
-
-static void dispatch_bios(struct mirror_set *ms, struct bio_list *bio_list)
-{
- struct bio *bio;
-
- while ((bio = bio_list_pop(bio_list))) {
- queue_bio(ms, bio, WRITE);
- }
-}
-
-static void complete_resync_work(struct region *reg, int success)
-{