Merge tag 'trace-v7.3-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-tracegrafted

Pull tracing fixes from Steven Rostedt:

 - Don't destroy user event fields when removal fails

   User event fields are destroyed before the event is removed from
   visibility. But that can fail leaving the still visible event with no
   fields. Move the destroying of the fields to after the event is
   successfully removed from visibility.

 - Initialize function graph state is fork before calling
   copy_exec_state()

   For non-CLONE_VM forks, copy_exec_state() allocates a new
   task_exec_state. If that allocation fails, ftrace_graph_exit_task()
   will free the tasks ret_stack pointer. Since that pointer is still
   using the parent's ret_stack, it mistakenly frees the parent's
   pointer too.

   Call ftrace_graph_init() on the task first which will NULL out the
   new tasks's ret_stack and if the copy fails, it will not free
   anything.

 - Remove FGRAPH_MAX_INDEX

   The macro FGRAPH_MAX_INDEX was added but never used. Remove it.

 - Save ent_size in function graph printing of nested functions

   The function graph tracer needs to look at the next event to see if
   the next event is the return of the current function entry. If it is,
   it prints a single line:

	ktime_get();

   Otherwise it prints it like a nested function:

	tick_nohz_irq_exit() {
	    ktime_get();
	    kcpustat_irq_exit();
	}

   In order to look at the next event, it must save the current event so
   that it has the information to print from it. It saves the event in
   the iterator descriptor called "ent". What it doesn't save is the
   ent_size of the event which is now used to know if the function graph
   arguments are to be printed. The peek doesn't save the size so the
   size used happens to be that of the size of the last event that was
   seen.

   Save the entry event size in the iterator descriptor so that the
   correct size is used.

 - Fix several errors with freeing data in the histogram code

   The histogram code had a lot of leaked or or incorrect accounting
   when failures happen. Correct them.

 - Fix histogram regression of .percent and .graph modifiers

   Up until 6.3 histogram values could have "percent" or "graph"
   modifiers that changed how they were printed. But a change that added
   restricting histograms values from being strings, stack traces and
   other modifiers inadvertently prevented them from using the percent
   and graph modifiers, which were legal use cases for values.

   Put back the percent and graph modifiers.

 - Fix various typos in the comments

 - Set the trace_clock before initializing a histogram with clock
   argument

   The histogram API allows the user to specific which trace clock to
   use via a "clock=" string. The histogram is set up first before the
   clock is checked. If the passed in clock is not valid, it exits
   without fully fixing up the histogram leaving it on the list and a
   use-after-free can trigger.

   Update the clock argument first and if it fails then exit gracefully
   before the histogram trigger is placed on any lists.

 - Restore :mod: trailer after parsing in ftrace_set_clr_event

   The function ftrace_set_clr_event() modifies the parse string and
   needs to put it back to what was passed in. It searches for ":mod:"
   via a strsep() but fails to put back the first ':' in the string.

   Add back the ':' in the passed in string.

 - Take trace_array reference when opening a tracer options file

   The options files are dynamically created and some tracers add their
   own options. When a tracer adds their own list of options, the
   trace_array holding them has an array to hold the list of options for
   each tracer. This array increases in size via a krealloc(), and the
   new entry gets a newly allocated array to hold the options of the new
   tracer being added.

   The element in each entry of the tracer's option array holds a
   pointer back to the trace_array, a pointer to the tracer it is
   associated to, a pointer to the flags of the option.

   The issue is that these arrays are freed when the trace_array is
   freed when its instance it represents is removed from the instances
   directory. There's a race that an open of one of these options files
   can happen when the instance is being removed.

   Add a new helper function to be called by the open function of the
   options file to iterate all existing trace_arrays under a lock and
   find the one that has the given option element in one of it's tracer
   arrays. If found, then update the associated trace_array's reference
   counter to keep it from being freed. If not found, have the open call
   return -ENODEV.

 - Disable interrupts when acquiring the lock in rb_wake_up_waiters()

   The function rb_wake_up_waiters() assumes it will be called in
   interrupt context and does not disable irqs when taking
   cpu_buffer->reader_lock, which can be called in hard interrupt
   context. The issue is in PREEMPT_RT, this function is called in
   thread context leaving this lock open to a deadlock.

   Take the lock with interrupts disabled.

 - Use rcu_assign_pointer() for tmp_ops filter hash

   The tmp_ops used in update_ftrace_direct_mod() assigns its
   filter_hash field directly, but that field is annotated as __rcu and
   sparse complains. Assign it with rcu_assign_pointer()

 - Fix use-after-free in enable_trigger_private_data_free()

   The trace_event_call is accessed through the event_trigger_data's
   trace_event_file pointer to put the trace_event_call on freeing. The
   issue is that the trace_event_file data may have been freed already
   causing a use-after-free. Add a field to the event_trigger_data that
   points directly to the trace_event_call so that it can decrement its
   reference directly without needing to go through the
   trace_event_file.

 - Fix accounting of buffer data remote headers

   trace_buffer_desc_size() and trace_remote_alloc_buffer() undercount
   the number of pages is needed for the asked for size as it doesn't
   take into account the meta data on each page. Add a helper function
   to do the calculation properly and use that in these functions.

 - Catch nr_page_va overflow in ring_buffer_desc sizing

   The number of pages per remote ring buffer is capped by
   ring_buffer_desc::nr_page_va (32 bits). A buffer_size large enough to
   overflow that field would silently allocate a descriptor smaller than
   what was asked for.

 - Do not resize the subbuf order if any per_cpu buffer is disabled

   The mmapping of ring buffers disables resizing the subbuffers, but it
   is done per-cpu whereas the subbuf size change is done for all the
   per_cpu buffers under the buffer->mutex. It could change the size of
   some while the mapping is happening on others. Have the resize of the
   subbuf order check all the per_cpu buffers under the lock to see if
   any of them is disabled before starting and causing an inconsistency
   between buffers that are being mapped.

* tag 'trace-v7.3-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace: (25 commits)
  ring-buffer: Check resize_disabled before publishing the new subbuf order
  tracing/remotes: Catch nr_page_va overflow in ring_buffer_desc sizing
  tracing/remotes: Account for ring buffer page header in size calculation
  tracing: Don't dereference trace_event_file in deferred trigger free
  ftrace: Use rcu_assign_pointer() for tmp_ops filter hash
  ring-buffer: Acquire the lock with irqsave in rb_wake_up_waiters()
  tracing: Take trace_array reference when opening a tracer options file
  tracing: Fix ring_buffer_read_page_size() kernel-doc
  tracing: Restore :mod: trailer after parsing in ftrace_set_clr_event()
  tracing: Fix memory corruption from a "STACKTRACE" histogram key
  tracing: Fix memory corruption from the histogram stacktrace modifier
  tracing: Undo the registration when enabling the histogram trigger fails
  tracing: Take the reference before publishing the named histogram trigger
  tracing: Set the trace clock before registering the histogram trigger
  tracing: Fix typo "preceeded" in comment
  tracing: Fix typo "availabe" in comment
  tracing: Let histogram values keep the percent and graph modifiers
  tracing: Keep the entry count when the histogram stats allocation fails
  tracing: Free histogram the field rejected for a bad modifier
  tracing: Free histogram the var ref when its initialization fails
  ...

1 files changed, 980 insertions, 0 deletions

diff --git a/drivers/md/dm-snap-persistent.c b/drivers/md/dm-snap-persistent.c
new file mode 100644
index 000000000..e86d929b0
--- /dev/null
+++ b/drivers/md/dm-snap-persistent.c
@@ -0,0 +1,980 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (C) 2001-2002 Sistina Software (UK) Limited.
+ * Copyright (C) 2006-2008 Red Hat GmbH
+ *
+ * This file is released under the GPL.
+ */
+
+#include "dm-exception-store.h"
+
+#include <linux/ctype.h>
+#include <linux/mm.h>
+#include <linux/pagemap.h>
+#include <linux/vmalloc.h>
+#include <linux/export.h>
+#include <linux/slab.h>
+#include <linux/dm-io.h>
+#include <linux/dm-bufio.h>
+
+#define DM_MSG_PREFIX "persistent snapshot"
+#define DM_CHUNK_SIZE_DEFAULT_SECTORS 32U	/* 16KB */
+
+#define DM_PREFETCH_CHUNKS		12
+
+/*
+ *---------------------------------------------------------------
+ * Persistent snapshots, by persistent we mean that the snapshot
+ * will survive a reboot.
+ *---------------------------------------------------------------
+ */
+
+/*
+ * We need to store a record of which parts of the origin have
+ * been copied to the snapshot device.  The snapshot code
+ * requires that we copy exception chunks to chunk aligned areas
+ * of the COW store.  It makes sense therefore, to store the
+ * metadata in chunk size blocks.
+ *
+ * There is no backward or forward compatibility implemented,
+ * snapshots with different disk versions than the kernel will
+ * not be usable.  It is expected that "lvcreate" will blank out
+ * the start of a fresh COW device before calling the snapshot
+ * constructor.
+ *
+ * The first chunk of the COW device just contains the header.
+ * After this there is a chunk filled with exception metadata,
+ * followed by as many exception chunks as can fit in the
+ * metadata areas.
+ *
+ * All on disk structures are in little-endian format.  The end
+ * of the exceptions info is indicated by an exception with a
+ * new_chunk of 0, which is invalid since it would point to the
+ * header chunk.
+ */
+
+/*
+ * Magic for persistent snapshots: "SnAp" - Feeble isn't it.
+ */
+#define SNAP_MAGIC 0x70416e53
+
+/*
+ * The on-disk version of the metadata.
+ */
+#define SNAPSHOT_DISK_VERSION 1
+
+#define NUM_SNAPSHOT_HDR_CHUNKS 1
+
+struct disk_header {
+	__le32 magic;
+
+	/*
+	 * Is this snapshot valid.  There is no way of recovering
+	 * an invalid snapshot.
+	 */
+	__le32 valid;
+
+	/*
+	 * Simple, incrementing version. no backward
+	 * compatibility.
+	 */
+	__le32 version;
+
+	/* In sectors */
+	__le32 chunk_size;
+} __packed;
+
+struct disk_exception {
+	__le64 old_chunk;
+	__le64 new_chunk;
+} __packed;
+
+struct core_exception {
+	uint64_t old_chunk;
+	uint64_t new_chunk;
+};
+
+struct commit_callback {
+	void (*callback)(void *ref, int success);
+	void *context;
+};
+
+/*
+ * The top level structure for a persistent exception store.
+ */
+struct pstore {
+	struct dm_exception_store *store;
+	int version;
+	int valid;
+	uint32_t exceptions_per_area;
+
+	/*
+	 * Now that we have an asynchronous kcopyd there is no
+	 * need for large chunk sizes, so it wont hurt to have a
+	 * whole chunks worth of metadata in memory at once.
+	 */
+	void *area;
+
+	/*
+	 * An area of zeros used to clear the next area.
+	 */
+	void *zero_area;
+
+	/*
+	 * An area used for header. The header can be written
+	 * concurrently with metadata (when invalidating the snapshot),
+	 * so it needs a separate buffer.
+	 */
+	void *header_area;
+
+	/*
+	 * Used to keep track of which metadata area the data in
+	 * 'chunk' refers to.
+	 */
+	chunk_t current_area;
+
+	/*
+	 * The next free chunk for an exception.
+	 *
+	 * When creating exceptions, all the chunks here and above are
+	 * free.  It holds the next chunk to be allocated.  On rare
+	 * occasions (e.g. after a system crash) holes can be left in
+	 * the exception store because chunks can be committed out of
+	 * order.
+	 *
+	 * When merging exceptions, it does not necessarily mean all the
+	 * chunks here and above are free.  It holds the value it would
+	 * have held if all chunks had been committed in order of
+	 * allocation.  Consequently the value may occasionally be
+	 * slightly too low, but since it's only used for 'status' and
+	 * it can never reach its minimum value too early this doesn't
+	 * matter.
+	 */
+
+	chunk_t next_free;
+
+	/*
+	 * The index of next free exception in the current
+	 * metadata area.
+	 */
+	uint32_t current_committed;
+
+	atomic_t pending_count;
+	uint32_t callback_count;
+	struct commit_callback *callbacks;
+	struct dm_io_client *io_client;
+
+	struct workqueue_struct *metadata_wq;
+};
+
+static int alloc_area(struct pstore *ps)
+{
+	int r = -ENOMEM;
+	size_t len;
+
+	len = ps->store->chunk_size << SECTOR_SHIFT;
+
+	/*
+	 * Allocate the chunk_size block of memory that will hold
+	 * a single metadata area.
+	 */
+	ps->area = vmalloc(len);
+	if (!ps->area)
+		goto err_area;
+
+	ps->zero_area = vzalloc(len);
+	if (!ps->zero_area)
+		goto err_zero_area;
+
+	ps->header_area = vmalloc(len);
+	if (!ps->header_area)
+		goto err_header_area;
+
+	return 0;
+
+err_header_area:
+	vfree(ps->zero_area);
+
+err_zero_area:
+	vfree(ps->area);
+
+err_area:
+	return r;
+}
+
+static void free_area(struct pstore *ps)
+{
+	vfree(ps->area);
+	ps->area = NULL;
+	vfree(ps->zero_area);
+	ps->zero_area = NULL;
+	vfree(ps->header_area);
+	ps->header_area = NULL;
+}
+
+struct mdata_req {
+	struct dm_io_region *where;
+	struct dm_io_request *io_req;
+	struct work_struct work;
+	int result;
+};
+
+static void do_metadata(struct work_struct *work)
+{
+	struct mdata_req *req = container_of(work, struct mdata_req, work);
+
+	req->result = dm_io(req->io_req, 1, req->where, NULL, NULL, IOPRIO_DEFAULT);
+}
+
+/*
+ * Read or write a chunk aligned and sized block of data from a device.
+ */
+static int chunk_io(struct pstore *ps, void *area, chunk_t chunk, blk_opf_t opf,
+		    int metadata)
+{
+	struct dm_io_region where = {
+		.bdev = dm_snap_cow(ps->store->snap)->bdev,
+		.sector = ps->store->chunk_size * chunk,
+		.count = ps->store->chunk_size,
+	};
+	struct dm_io_request io_req = {
+		.bi_opf = opf,
+		.mem.type = DM_IO_VMA,
+		.mem.ptr.vma = area,
+		.client = ps->io_client,
+		.notify.fn = NULL,
+	};
+	struct mdata_req req;
+
+	if (!metadata)
+		return dm_io(&io_req, 1, &where, NULL, NULL, IOPRIO_DEFAULT);
+
+	req.where = &where;
+	req.io_req = &io_req;
+
+	/*
+	 * Issue the synchronous I/O from a different thread
+	 * to avoid submit_bio_noacct recursion.
+	 */
+	INIT_WORK_ONSTACK(&req.work, do_metadata);
+	queue_work(ps->metadata_wq, &req.work);
+	flush_workqueue(ps->metadata_wq);
+	destroy_work_on_stack(&req.work);
+
+	return req.result;
+}
+
+/*
+ * Convert a metadata area index to a chunk index.
+ */
+static chunk_t area_location(struct pstore *ps, chunk_t area)
+{
+	return NUM_SNAPSHOT_HDR_CHUNKS + ((ps->exceptions_per_area + 1) * area);
+}
+
+static void skip_metadata(struct pstore *ps)
+{
+	uint32_t stride = ps->exceptions_per_area + 1;
+	chunk_t next_free = ps->next_free;
+
+	if (sector_div(next_free, stride) == NUM_SNAPSHOT_HDR_CHUNKS)
+		ps->next_free++;
+}
+
+/*
+ * Read or write a metadata area.  Remembering to skip the first
+ * chunk which holds the header.
+ */
+static int area_io(struct pstore *ps, blk_opf_t opf)
+{
+	chunk_t chunk = area_location(ps, ps->current_area);
+
+	return chunk_io(ps, ps->area, chunk, opf, 0);
+}
+
+static void zero_memory_area(struct pstore *ps)
+{
+	memset(ps->area, 0, ps->store->chunk_size << SECTOR_SHIFT);
+}
+
+static int zero_disk_area(struct pstore *ps, chunk_t area)
+{
+	return chunk_io(ps, ps->zero_area, area_location(ps, area),
+			REQ_OP_WRITE, 0);
+}
+
+static int read_header(struct pstore *ps, int *new_snapshot)
+{
+	int r;
+	struct disk_header *dh;
+	unsigned int chunk_size;
+	int chunk_size_supplied = 1;
+	char *chunk_err;
+
+	/*
+	 * Use default chunk size (or logical_block_size, if larger)
+	 * if none supplied
+	 */
+	if (!ps->store->chunk_size) {
+		ps->store->chunk_size = max(DM_CHUNK_SIZE_DEFAULT_SECTORS,
+		    bdev_logical_block_size(dm_snap_cow(ps->store->snap)->
+					    bdev) >> 9);
+		ps->store->chunk_mask = ps->store->chunk_size - 1;
+		ps->store->chunk_shift = __ffs(ps->store->chunk_size);
+		chunk_size_supplied = 0;
+	}
+
+	ps->io_client = dm_io_client_create();
+	if (IS_ERR(ps->io_client))
+		return PTR_ERR(ps->io_client);
+
+	r = alloc_area(ps);
+	if (r)
+		return r;
+
+	r = chunk_io(ps, ps->header_area, 0, REQ_OP_READ, 1);
+	if (r)
+		goto bad;
+
+	dh = ps->header_area;
+
+	if (le32_to_cpu(dh->magic) == 0) {
+		*new_snapshot = 1;
+		return 0;
+	}
+
+	if (le32_to_cpu(dh->magic) != SNAP_MAGIC) {
+		DMWARN("Invalid or corrupt snapshot");
+		r = -ENXIO;
+		goto bad;
+	}
+
+	*new_snapshot = 0;
+	ps->valid = le32_to_cpu(dh->valid);
+	ps->version = le32_to_cpu(dh->version);
+	chunk_size = le32_to_cpu(dh->chunk_size);
+
+	if (ps->store->chunk_size == chunk_size)
+		return 0;
+
+	if (chunk_size_supplied)
+		DMWARN("chunk size %u in device metadata overrides table chunk size of %u.",
+		       chunk_size, ps->store->chunk_size);
+
+	/* We had a bogus chunk_size. Fix stuff up. */
+	free_area(ps);
+
+	r = dm_exception_store_set_chunk_size(ps->store, chunk_size,
+					      &chunk_err);
+	if (r) {
+		DMERR("invalid on-disk chunk size %u: %s.",
+		      chunk_size, chunk_err);
+		return r;
+	}
+
+	r = alloc_area(ps);
+	return r;
+
+bad:
+	free_area(ps);
+	return r;
+}
+
+static int write_header(struct pstore *ps)
+{
+	struct disk_header *dh;
+
+	memset(ps->header_area, 0, ps->store->chunk_size << SECTOR_SHIFT);
+
+	dh = ps->header_area;
+	dh->magic = cpu_to_le32(SNAP_MAGIC);
+	dh->valid = cpu_to_le32(ps->valid);
+	dh->version = cpu_to_le32(ps->version);
+	dh->chunk_size = cpu_to_le32(ps->store->chunk_size);
+
+	return chunk_io(ps, ps->header_area, 0, REQ_OP_WRITE, 1);
+}
+
+/*
+ * Access functions for the disk exceptions, these do the endian conversions.
+ */
+static struct disk_exception *get_exception(struct pstore *ps, void *ps_area,
+					    uint32_t index)
+{
+	BUG_ON(index >= ps->exceptions_per_area);
+
+	return ((struct disk_exception *) ps_area) + index;
+}
+
+static void read_exception(struct pstore *ps, void *ps_area,
+			   uint32_t index, struct core_exception *result)
+{
+	struct disk_exception *de = get_exception(ps, ps_area, index);
+
+	/* copy it */
+	result->old_chunk = le64_to_cpu(de->old_chunk);
+	result->new_chunk = le64_to_cpu(de->new_chunk);
+}
+
+static void write_exception(struct pstore *ps,
+			    uint32_t index, struct core_exception *e)
+{
+	struct disk_exception *de = get_exception(ps, ps->area, index);
+
+	/* copy it */
+	de->old_chunk = cpu_to_le64(e->old_chunk);
+	de->new_chunk = cpu_to_le64(e->new_chunk);
+}
+
+static void clear_exception(struct pstore *ps, uint32_t index)
+{
+	struct disk_exception *de = get_exception(ps, ps->area, index);
+
+	/* clear it */
+	de->old_chunk = 0;
+	de->new_chunk = 0;
+}
+
+/*
+ * Registers the exceptions that are present in the current area.
+ * 'full' is filled in to indicate if the area has been
+ * filled.
+ */
+static int insert_exceptions(struct pstore *ps, void *ps_area,
+			     int (*callback)(void *callback_context,
+					     chunk_t old, chunk_t new),
+			     void *callback_context,
+			     int *full)
+{
+	int r;
+	unsigned int i;
+	struct core_exception e;
+
+	/* presume the area is full */
+	*full = 1;
+
+	for (i = 0; i < ps->exceptions_per_area; i++) {
+		read_exception(ps, ps_area, i, &e);
+
+		/*
+		 * If the new_chunk is pointing at the start of
+		 * the COW device, where the first metadata area
+		 * is we know that we've hit the end of the
+		 * exceptions.  Therefore the area is not full.
+		 */
+		if (e.new_chunk == 0LL) {
+			ps->current_committed = i;
+			*full = 0;
+			break;
+		}
+
+		/*
+		 * Keep track of the start of the free chunks.
+		 */
+		if (ps->next_free <= e.new_chunk)
+			ps->next_free = e.new_chunk + 1;
+
+		/*
+		 * Otherwise we add the exception to the snapshot.
+		 */
+		r = callback(callback_context, e.old_chunk, e.new_chunk);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+static int read_exceptions(struct pstore *ps,
+			   int (*callback)(void *callback_context, chunk_t old,
+					   chunk_t new),
+			   void *callback_context)
+{
+	int r, full = 1;
+	struct dm_bufio_client *client;
+	chunk_t prefetch_area = 0;
+
+	client = dm_bufio_client_create(dm_snap_cow(ps->store->snap)->bdev,
+					ps->store->chunk_size << SECTOR_SHIFT,
+					1, 0, NULL, NULL, 0);
+
+	if (IS_ERR(client))
+		return PTR_ERR(client);
+
+	/*
+	 * Setup for one current buffer + desired readahead buffers.
+	 */
+	dm_bufio_set_minimum_buffers(client, 1 + DM_PREFETCH_CHUNKS);
+
+	/*
+	 * Keeping reading chunks and inserting exceptions until
+	 * we find a partially full area.
+	 */
+	for (ps->current_area = 0; full; ps->current_area++) {
+		struct dm_buffer *bp;
+		void *area;
+		chunk_t chunk;
+
+		if (unlikely(prefetch_area < ps->current_area))
+			prefetch_area = ps->current_area;
+
+		if (DM_PREFETCH_CHUNKS) {
+			do {
+				chunk_t pf_chunk = area_location(ps, prefetch_area);
+
+				if (unlikely(pf_chunk >= dm_bufio_get_device_size(client)))
+					break;
+				dm_bufio_prefetch(client, pf_chunk, 1);
+				prefetch_area++;
+				if (unlikely(!prefetch_area))
+					break;
+			} while (prefetch_area <= ps->current_area + DM_PREFETCH_CHUNKS);
+		}
+
+		chunk = area_location(ps, ps->current_area);
+
+		area = dm_bufio_read(client, chunk, &bp);
+		if (IS_ERR(area)) {
+			r = PTR_ERR(area);
+			goto ret_destroy_bufio;
+		}
+
+		r = insert_exceptions(ps, area, callback, callback_context,
+				      &full);
+
+		if (!full)
+			memcpy(ps->area, area, ps->store->chunk_size << SECTOR_SHIFT);
+
+		dm_bufio_release(bp);
+
+		dm_bufio_forget(client, chunk);
+
+		if (unlikely(r))
+			goto ret_destroy_bufio;
+	}
+
+	ps->current_area--;
+
+	skip_metadata(ps);
+
+	r = 0;
+
+ret_destroy_bufio:
+	dm_bufio_client_destroy(client);
+
+	return r;
+}
+
+static struct pstore *get_info(struct dm_exception_store *store)
+{
+	return store->context;
+}
+
+static void persistent_usage(struct dm_exception_store *store,
+			     sector_t *total_sectors,
+			     sector_t *sectors_allocated,
+			     sector_t *metadata_sectors)
+{
+	struct pstore *ps = get_info(store);
+
+	*sectors_allocated = ps->next_free * store->chunk_size;
+	*total_sectors = get_dev_size(dm_snap_cow(store->snap)->bdev);
+
+	/*
+	 * First chunk is the fixed header.
+	 * Then there are (ps->current_area + 1) metadata chunks, each one
+	 * separated from the next by ps->exceptions_per_area data chunks.
+	 */
+	*metadata_sectors = (ps->current_area + 1 + NUM_SNAPSHOT_HDR_CHUNKS) *
+			    store->chunk_size;
+}
+
+static void persistent_dtr(struct dm_exception_store *store)
+{
+	struct pstore *ps = get_info(store);
+
+	destroy_workqueue(ps->metadata_wq);
+
+	/* Created in read_header */
+	if (ps->io_client)
+		dm_io_client_destroy(ps->io_client);
+	free_area(ps);
+
+	/* Allocated in persistent_read_metadata */
+	kvfree(ps->callbacks);
+
+	kfree(ps);
+}
+
+static int persistent_read_metadata(struct dm_exception_store *store,
+				    int (*callback)(void *callback_context,
+						    chunk_t old, chunk_t new),
+				    void *callback_context)
+{
+	int r, new_snapshot;
+	struct pstore *ps = get_info(store);
+
+	/*
+	 * Read the snapshot header.
+	 */
+	r = read_header(ps, &new_snapshot);
+	if (r)
+		return r;
+
+	/*
+	 * Now we know correct chunk_size, complete the initialisation.
+	 */
+	ps->exceptions_per_area = (ps->store->chunk_size << SECTOR_SHIFT) /
+				  sizeof(struct disk_exception);
+	ps->callbacks = kvzalloc_objs(*ps->callbacks, ps->exceptions_per_area);
+	if (!ps->callbacks)
+		return -ENOMEM;
+
+	/*
+	 * Do we need to setup a new snapshot ?
+	 */
+	if (new_snapshot) {
+		r = write_header(ps);
+		if (r) {
+			DMWARN("write_header failed");
+			return r;
+		}
+
+		ps->current_area = 0;
+		zero_memory_area(ps);
+		r = zero_disk_area(ps, 0);
+		if (r)
+			DMWARN("zero_disk_area(0) failed");
+		return r;
+	}
+	/*
+	 * Sanity checks.
+	 */
+	if (ps->version != SNAPSHOT_DISK_VERSION) {
+		DMWARN("unable to handle snapshot disk version %d",
+		       ps->version);
+		return -EINVAL;
+	}
+
+	/*
+	 * Metadata are valid, but snapshot is invalidated
+	 */
+	if (!ps->valid)
+		return 1;
+
+	/*
+	 * Read the metadata.
+	 */
+	r = read_exceptions(ps, callback, callback_context);
+
+	return r;
+}
+
+static int persistent_prepare_exception(struct dm_exception_store *store,
+					struct dm_exception *e)
+{
+	struct pstore *ps = get_info(store);
+	sector_t size = get_dev_size(dm_snap_cow(store->snap)->bdev);
+
+	/* Is there enough room ? */
+	if (size < ((ps->next_free + 1) * store->chunk_size))
+		return -ENOSPC;
+
+	e->new_chunk = ps->next_free;
+
+	/*
+	 * Move onto the next free pending, making sure to take
+	 * into account the location of the metadata chunks.
+	 */
+	ps->next_free++;
+	skip_metadata(ps);
+
+	atomic_inc(&ps->pending_count);
+	return 0;
+}
+
+static void persistent_commit_exception(struct dm_exception_store *store,
+					struct dm_exception *e, int valid,
+					void (*callback)(void *, int success),
+					void *callback_context)
+{
+	unsigned int i;
+	struct pstore *ps = get_info(store);
+	struct core_exception ce;
+	struct commit_callback *cb;
+
+	if (!valid)
+		ps->valid = 0;
+
+	ce.old_chunk = e->old_chunk;
+	ce.new_chunk = e->new_chunk;
+	write_exception(ps, ps->current_committed++, &ce);
+
+	/*
+	 * Add the callback to the back of the array.  This code
+	 * is the only place where the callback array is
+	 * manipulated, and we know that it will never be called
+	 * multiple times concurrently.
+	 */
+	cb = ps->callbacks + ps->callback_count++;
+	cb->callback = callback;
+	cb->context = callback_context;
+
+	/*
+	 * 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))
+		return;
+
+	/*
+	 * 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;
+
+	/*
+	 * Commit exceptions to disk.
+	 */
+	if (ps->valid && area_io(ps, REQ_OP_WRITE | REQ_PREFLUSH | REQ_FUA |
+				 REQ_SYNC))
+		ps->valid = 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 int persistent_prepare_merge(struct dm_exception_store *store,
+				    chunk_t *last_old_chunk,
+				    chunk_t *last_new_chunk)
+{
+	struct pstore *ps = get_info(store);
+	struct core_exception ce;
+	int nr_consecutive;
+	int r;
+
+	/*
+	 * When current area is empty, move back to preceding area.
+	 */
+	if (!ps->current_committed) {
+		/*
+		 * Have we finished?
+		 */
+		if (!ps->current_area)
+			return 0;
+
+		ps->current_area--;
+		r = area_io(ps, REQ_OP_READ);
+		if (r < 0)
+			return r;
+		ps->current_committed = ps->exceptions_per_area;
+	}
+
+	read_exception(ps, ps->area, ps->current_committed - 1, &ce);
+	*last_old_chunk = ce.old_chunk;
+	*last_new_chunk = ce.new_chunk;
+
+	/*
+	 * Find number of consecutive chunks within the current area,
+	 * working backwards.
+	 */
+	for (nr_consecutive = 1; nr_consecutive < ps->current_committed;
+	     nr_consecutive++) {
+		read_exception(ps, ps->area,
+			       ps->current_committed - 1 - nr_consecutive, &ce);
+		if (ce.old_chunk != *last_old_chunk - nr_consecutive ||
+		    ce.new_chunk != *last_new_chunk - nr_consecutive)
+			break;
+	}
+
+	return nr_consecutive;
+}
+
+static int persistent_commit_merge(struct dm_exception_store *store,
+				   int nr_merged)
+{
+	int r, i;
+	struct pstore *ps = get_info(store);
+
+	BUG_ON(nr_merged > ps->current_committed);
+
+	for (i = 0; i < nr_merged; i++)
+		clear_exception(ps, ps->current_committed - 1 - i);
+
+	r = area_io(ps, REQ_OP_WRITE | REQ_PREFLUSH | REQ_FUA);
+	if (r < 0)
+		return r;
+
+	ps->current_committed -= nr_merged;
+
+	/*
+	 * At this stage, only persistent_usage() uses ps->next_free, so
+	 * we make no attempt to keep ps->next_free strictly accurate
+	 * as exceptions may have been committed out-of-order originally.
+	 * Once a snapshot has become merging, we set it to the value it
+	 * would have held had all the exceptions been committed in order.
+	 *
+	 * ps->current_area does not get reduced by prepare_merge() until
+	 * after commit_merge() has removed the nr_merged previous exceptions.
+	 */
+	ps->next_free = area_location(ps, ps->current_area) +
+			ps->current_committed + 1;
+
+	return 0;
+}
+
+static void persistent_drop_snapshot(struct dm_exception_store *store)
+{
+	struct pstore *ps = get_info(store);
+
+	ps->valid = 0;
+	if (write_header(ps))
+		DMWARN("write header failed");
+}
+
+static int persistent_ctr(struct dm_exception_store *store, char *options)
+{
+	struct pstore *ps;
+	int r;
+
+	/* allocate the pstore */
+	ps = kzalloc_obj(*ps);
+	if (!ps)
+		return -ENOMEM;
+
+	ps->store = store;
+	ps->valid = 1;
+	ps->version = SNAPSHOT_DISK_VERSION;
+	ps->area = NULL;
+	ps->zero_area = NULL;
+	ps->header_area = NULL;
+	ps->next_free = NUM_SNAPSHOT_HDR_CHUNKS + 1; /* header and 1st area */
+	ps->current_committed = 0;
+
+	ps->callback_count = 0;
+	atomic_set(&ps->pending_count, 0);
+	ps->callbacks = NULL;
+
+	ps->metadata_wq = alloc_workqueue("ksnaphd",
+					  WQ_MEM_RECLAIM | WQ_PERCPU, 0);
+	if (!ps->metadata_wq) {
+		DMERR("couldn't start header metadata update thread");
+		r = -ENOMEM;
+		goto err_workqueue;
+	}
+
+	if (options) {
+		char overflow = toupper(options[0]);
+
+		if (overflow == 'O')
+			store->userspace_supports_overflow = true;
+		else {
+			DMERR("Unsupported persistent store option: %s", options);
+			r = -EINVAL;
+			goto err_options;
+		}
+	}
+
+	store->context = ps;
+
+	return 0;
+
+err_options:
+	destroy_workqueue(ps->metadata_wq);
+err_workqueue:
+	kfree(ps);
+
+	return r;
+}
+
+static unsigned int persistent_status(struct dm_exception_store *store,
+				  status_type_t status, char *result,
+				  unsigned int maxlen)
+{
+	unsigned int sz = 0;
+
+	switch (status) {
+	case STATUSTYPE_INFO:
+		break;
+	case STATUSTYPE_TABLE:
+		DMEMIT(" %s %llu", store->userspace_supports_overflow ? "PO" : "P",
+		       (unsigned long long)store->chunk_size);
+		break;
+	case STATUSTYPE_IMA:
+		*result = '\0';
+		break;
+	}
+
+	return sz;
+}
+
+static struct dm_exception_store_type _persistent_type = {
+	.name = "persistent",
+	.module = THIS_MODULE,
+	.ctr = persistent_ctr,
+	.dtr = persistent_dtr,
+	.read_metadata = persistent_read_metadata,
+	.prepare_exception = persistent_prepare_exception,
+	.commit_exception = persistent_commit_exception,
+	.prepare_merge = persistent_prepare_merge,
+	.commit_merge = persistent_commit_merge,
+	.drop_snapshot = persistent_drop_snapshot,
+	.usage = persistent_usage,
+	.status = persistent_status,
+};
+
+static struct dm_exception_store_type _persistent_compat_type = {
+	.name = "P",
+	.module = THIS_MODULE,
+	.ctr = persistent_ctr,
+	.dtr = persistent_dtr,
+	.read_metadata = persistent_read_metadata,
+	.prepare_exception = persistent_prepare_exception,
+	.commit_exception = persistent_commit_exception,
+	.prepare_merge = persistent_prepare_merge,
+	.commit_merge = persistent_commit_merge,
+	.drop_snapshot = persistent_drop_snapshot,
+	.usage = persistent_usage,
+	.status = persistent_status,
+};
+
+int dm_persistent_snapshot_init(void)
+{
+	int r;
+
+	r = dm_exception_store_type_register(&_persistent_type);
+	if (r) {
+		DMERR("Unable to register persistent exception store type");
+		return r;
+	}
+
+	r = dm_exception_store_type_register(&_persistent_compat_type);
+	if (r) {
+		DMERR("Unable to register old-style persistent exception store type");
+		dm_exception_store_type_unregister(&_persistent_type);
+		return r;
+	}
+
+	return r;
+}
+
+void dm_persistent_snapshot_exit(void)
+{
+	dm_exception_store_type_unregister(&_persistent_type);
+	dm_exception_store_type_unregister(&_persistent_compat_type);
+}