The struct tracker contains callback functions for handling a I/O units
of a block device. When a write request is handled, the change block
tracking (CBT) map functions are called and initiates the process of
copying data from the original block device to the change store.
Attaching and detaching the tracker is provided by the functions
bdev_filter_*() of the kernel.
Signed-off-by: Sergei Shtepa <[email protected]>
---
drivers/block/blksnap/tracker.c | 683 ++++++++++++++++++++++++++++++++
drivers/block/blksnap/tracker.h | 74 ++++
2 files changed, 757 insertions(+)
create mode 100644 drivers/block/blksnap/tracker.c
create mode 100644 drivers/block/blksnap/tracker.h
diff --git a/drivers/block/blksnap/tracker.c b/drivers/block/blksnap/tracker.c
new file mode 100644
index 000000000000..03e828d4a22f
--- /dev/null
+++ b/drivers/block/blksnap/tracker.c
@@ -0,0 +1,683 @@
+// SPDX-License-Identifier: GPL-2.0
+#define pr_fmt(fmt) KBUILD_MODNAME "-tracker: " fmt
+
+#include <linux/slab.h>
+#include <linux/blk-mq.h>
+#include <linux/sched/mm.h>
+#include <uapi/linux/blksnap.h>
+#include "params.h"
+#include "tracker.h"
+#include "cbt_map.h"
+#include "diff_area.h"
+
+struct tracked_device {
+ struct list_head link;
+ dev_t dev_id;
+};
+
+DEFINE_PERCPU_RWSEM(tracker_submit_lock);
+LIST_HEAD(tracked_device_list);
+DEFINE_SPINLOCK(tracked_device_lock);
+static refcount_t trackers_counter = REFCOUNT_INIT(1);
+
+struct tracker_release_worker {
+ struct work_struct work;
+ struct list_head list;
+ spinlock_t lock;
+};
+static struct tracker_release_worker tracker_release_worker;
+
+void tracker_lock(void)
+{
+ pr_debug("Lock trackers\n");
+ percpu_down_write(&tracker_submit_lock);
+};
+void tracker_unlock(void)
+{
+ percpu_up_write(&tracker_submit_lock);
+ pr_debug("Trackers have been unlocked\n");
+};
+
+static void tracker_free(struct tracker *tracker)
+{
+ might_sleep();
+
+ pr_debug("Free tracker for device [%u:%u].\n", MAJOR(tracker->dev_id),
+ MINOR(tracker->dev_id));
+
+ diff_area_put(tracker->diff_area);
+ cbt_map_put(tracker->cbt_map);
+
+ kfree(tracker);
+
+ refcount_dec(&trackers_counter);
+}
+
+static inline struct tracker *tracker_get_by_dev(struct block_device *bdev)
+{
+ struct bdev_filter *flt = bdev->bd_filter;
+
+ if (!flt)
+ return NULL;
+
+ bdev_filter_get(flt);
+
+ return container_of(flt, struct tracker, flt);
+}
+
+static bool tracker_submit_bio_cb(struct bio *bio)
+{
+ struct bdev_filter *flt = bio->bi_bdev->bd_filter;
+ struct bio_list bio_list_on_stack[2] = { };
+ struct bio *new_bio;
+ bool ret = true;
+ struct tracker *tracker = container_of(flt, struct tracker, flt);
+ int err;
+ sector_t sector;
+ sector_t count;
+ unsigned int current_flag;
+
+ WARN_ON_ONCE(!flt);
+ if (unlikely(!flt))
+ return true;
+
+ if (bio->bi_opf & REQ_NOWAIT) {
+ if (!percpu_down_read_trylock(&tracker_submit_lock)) {
+ bio_wouldblock_error(bio);
+ return false;
+ }
+ } else
+ percpu_down_read(&tracker_submit_lock);
+
+ if (!op_is_write(bio_op(bio)))
+ goto out;
+
+ count = bio_sectors(bio);
+ if (!count)
+ goto out;
+
+ sector = bio->bi_iter.bi_sector;
+ if (bio_flagged(bio, BIO_REMAPPED))
+ sector -= bio->bi_bdev->bd_start_sect;
+
+ current_flag = memalloc_noio_save();
+ err = cbt_map_set(tracker->cbt_map, sector, count);
+ memalloc_noio_restore(current_flag);
+ if (unlikely(err))
+ goto out;
+
+ if (!atomic_read(&tracker->snapshot_is_taken))
+ goto out;
+
+ if (diff_area_is_corrupted(tracker->diff_area))
+ goto out;
+
+ current_flag = memalloc_noio_save();
+ bio_list_init(&bio_list_on_stack[0]);
+ current->bio_list = bio_list_on_stack;
+ barrier();
+
+ err = diff_area_copy(tracker->diff_area, sector, count,
+ !!(bio->bi_opf & REQ_NOWAIT));
+
+ current->bio_list = NULL;
+ barrier();
+ memalloc_noio_restore(current_flag);
+
+ if (unlikely(err))
+ goto fail;
+
+ while ((new_bio = bio_list_pop(&bio_list_on_stack[0]))) {
+ /*
+ * The result from submitting a bio from the
+ * filter itself does not need to be processed,
+ * even if this function has a return code.
+ */
+
+ bio_set_flag(new_bio, BIO_FILTERED);
+ submit_bio_noacct(new_bio);
+ }
+ /*
+ * If a new bio was created during the handling, then new bios must
+ * be sent and returned to complete the processing of the original bio.
+ * Unfortunately, this has to be done for any bio, regardless of their
+ * flags and options.
+ * Otherwise, write I/O units may overtake read I/O units.
+ */
+ err = diff_area_wait(tracker->diff_area, sector, count,
+ !!(bio->bi_opf & REQ_NOWAIT));
+ if (likely(err == 0))
+ goto out;
+fail:
+ if (err == -EAGAIN) {
+ bio_wouldblock_error(bio);
+ ret = false;
+ } else
+ pr_err("Failed to copy data to diff storage with error %d\n", abs(err));
+out:
+ percpu_up_read(&tracker_submit_lock);
+ return ret;
+}
+
+
+static void tracker_release_work(struct work_struct *work)
+{
+ struct tracker *tracker = NULL;
+ struct tracker_release_worker *tracker_release =
+ container_of(work, struct tracker_release_worker, work);
+
+ do {
+ spin_lock(&tracker_release->lock);
+ tracker = list_first_entry_or_null(&tracker_release->list,
+ struct tracker, link);
+ if (tracker)
+ list_del(&tracker->link);
+ spin_unlock(&tracker_release->lock);
+
+ if (tracker)
+ tracker_free(tracker);
+ } while (tracker);
+}
+
+static void tracker_release_cb(struct kref *kref)
+{
+ struct bdev_filter *flt = container_of(kref, struct bdev_filter, kref);
+ struct tracker *tracker = container_of(flt, struct tracker, flt);
+
+ spin_lock(&tracker_release_worker.lock);
+ list_add_tail(&tracker->link, &tracker_release_worker.list);
+ spin_unlock(&tracker_release_worker.lock);
+
+ queue_work(system_wq, &tracker_release_worker.work);
+}
+
+static const struct bdev_filter_operations tracker_fops = {
+ .submit_bio_cb = tracker_submit_bio_cb,
+ .release_cb = tracker_release_cb
+};
+
+static int tracker_filter_attach(struct block_device *bdev,
+ struct tracker *tracker)
+{
+ int ret;
+ bool is_frozen = false;
+
+ pr_debug("Tracker attach filter\n");
+
+ if (freeze_bdev(bdev))
+ pr_err("Failed to freeze device [%u:%u]\n", MAJOR(bdev->bd_dev),
+ MINOR(bdev->bd_dev));
+ else {
+ is_frozen = true;
+ pr_debug("Device [%u:%u] was frozen\n", MAJOR(bdev->bd_dev),
+ MINOR(bdev->bd_dev));
+ }
+
+ ret = bdev_filter_attach(bdev, &tracker->flt);
+
+ if (is_frozen) {
+ if (thaw_bdev(bdev))
+ pr_err("Failed to thaw device [%u:%u]\n",
+ MAJOR(tracker->dev_id), MINOR(tracker->dev_id));
+ else
+ pr_debug("Device [%u:%u] was unfrozen\n",
+ MAJOR(bdev->bd_dev), MINOR(bdev->bd_dev));
+ }
+
+ if (ret)
+ pr_err("Failed to attach tracker to device [%u:%u]\n",
+ MAJOR(tracker->dev_id), MINOR(tracker->dev_id));
+
+ return ret;
+}
+
+static int tracker_filter_detach(struct block_device *bdev)
+{
+ int ret;
+ bool is_frozen = false;
+
+ pr_debug("Tracker delete filter\n");
+ if (freeze_bdev(bdev))
+ pr_err("Failed to freeze device [%u:%u]\n", MAJOR(bdev->bd_dev),
+ MINOR(bdev->bd_dev));
+ else {
+ is_frozen = true;
+ pr_debug("Device [%u:%u] was frozen\n", MAJOR(bdev->bd_dev),
+ MINOR(bdev->bd_dev));
+ }
+
+
+ ret = bdev_filter_detach(bdev);
+
+ if (is_frozen) {
+ if (thaw_bdev(bdev))
+ pr_err("Failed to thaw device [%u:%u]\n",
+ MAJOR(bdev->bd_dev), MINOR(bdev->bd_dev));
+ else
+ pr_debug("Device [%u:%u] was unfrozen\n",
+ MAJOR(bdev->bd_dev), MINOR(bdev->bd_dev));
+ }
+
+ if (ret)
+ pr_err("Failed to detach filter from device [%u:%u]\n",
+ MAJOR(bdev->bd_dev), MINOR(bdev->bd_dev));
+ return ret;
+}
+
+static struct tracker *tracker_new(struct block_device *bdev)
+{
+ int ret;
+ struct tracker *tracker = NULL;
+ struct cbt_map *cbt_map;
+
+ pr_debug("Creating tracker for device [%u:%u].\n", MAJOR(bdev->bd_dev),
+ MINOR(bdev->bd_dev));
+
+ tracker = kzalloc(sizeof(struct tracker), GFP_KERNEL);
+ if (tracker == NULL)
+ return ERR_PTR(-ENOMEM);
+
+ refcount_inc(&trackers_counter);
+ bdev_filter_init(&tracker->flt, &tracker_fops);
+ INIT_LIST_HEAD(&tracker->link);
+ atomic_set(&tracker->snapshot_is_taken, false);
+ tracker->dev_id = bdev->bd_dev;
+
+ pr_info("Create tracker for device [%u:%u]. Capacity 0x%llx sectors\n",
+ MAJOR(tracker->dev_id), MINOR(tracker->dev_id),
+ (unsigned long long)bdev_nr_sectors(bdev));
+
+ cbt_map = cbt_map_create(bdev);
+ if (!cbt_map) {
+ pr_err("Failed to create tracker for device [%u:%u]\n",
+ MAJOR(tracker->dev_id), MINOR(tracker->dev_id));
+ ret = -ENOMEM;
+ goto fail;
+ }
+ tracker->cbt_map = cbt_map;
+
+ ret = tracker_filter_attach(bdev, tracker);
+ if (ret) {
+ pr_err("Failed to attach tracker. errno=%d\n", abs(ret));
+ goto fail;
+ }
+
+ pr_debug("New tracker for device [%u:%u] was created.\n",
+ MAJOR(tracker->dev_id), MINOR(tracker->dev_id));
+
+ return tracker;
+fail:
+ tracker_put(tracker);
+ return ERR_PTR(ret);
+}
+
+int tracker_take_snapshot(struct tracker *tracker)
+{
+ int ret = 0;
+ bool cbt_reset_needed = false;
+ sector_t capacity;
+
+ if (tracker->cbt_map->is_corrupted) {
+ cbt_reset_needed = true;
+ pr_warn("Corrupted CBT table detected. CBT fault\n");
+ }
+
+ capacity = bdev_nr_sectors(tracker->diff_area->orig_bdev);
+ if (tracker->cbt_map->device_capacity != capacity) {
+ cbt_reset_needed = true;
+ pr_warn("Device resize detected. CBT fault\n");
+ }
+
+ if (cbt_reset_needed) {
+ ret = cbt_map_reset(tracker->cbt_map, capacity);
+ if (ret) {
+ pr_err("Failed to create tracker. errno=%d\n",
+ abs(ret));
+ return ret;
+ }
+ }
+
+ cbt_map_switch(tracker->cbt_map);
+ atomic_set(&tracker->snapshot_is_taken, true);
+
+ return 0;
+}
+
+void tracker_release_snapshot(struct tracker *tracker)
+{
+ if (!tracker)
+ return;
+
+ pr_debug("Tracker for device [%u:%u] release snapshot\n",
+ MAJOR(tracker->dev_id), MINOR(tracker->dev_id));
+
+ atomic_set(&tracker->snapshot_is_taken, false);
+}
+
+int tracker_init(void)
+{
+ INIT_WORK(&tracker_release_worker.work, tracker_release_work);
+ INIT_LIST_HEAD(&tracker_release_worker.list);
+ spin_lock_init(&tracker_release_worker.lock);
+
+ return 0;
+}
+
+/**
+ * tracker_wait_for_release - Waiting for all trackers are released.
+ *
+ * Trackers are released in the worker thread. So, this function allows to wait
+ * for the end of the process of releasing trackers.
+ */
+static void tracker_wait_for_release(void)
+{
+ long inx = 0;
+ u64 start_waiting = jiffies_64;
+
+ while (refcount_read(&trackers_counter) > 1) {
+ schedule_timeout_interruptible(HZ);
+ if (jiffies_64 > (start_waiting + 10*HZ)) {
+ start_waiting = jiffies_64;
+ inx++;
+
+ if (inx <= 12)
+ pr_warn("Waiting for trackers release\n");
+
+ WARN_ONCE(inx > 12, "Failed to release trackers\n");
+ }
+ }
+}
+
+void tracker_done(void)
+{
+ struct tracked_device *tr_dev;
+
+ pr_debug("Cleanup trackers\n");
+ while (true) {
+ spin_lock(&tracked_device_lock);
+ tr_dev = list_first_entry_or_null(&tracked_device_list,
+ struct tracked_device, link);
+ if (tr_dev)
+ list_del(&tr_dev->link);
+ spin_unlock(&tracked_device_lock);
+
+ if (!tr_dev)
+ break;
+
+ tracker_remove(tr_dev->dev_id);
+ kfree(tr_dev);
+ }
+
+ tracker_wait_for_release();
+}
+
+struct tracker *tracker_create_or_get(dev_t dev_id)
+{
+ struct tracker *tracker;
+ struct block_device *bdev;
+ struct tracked_device *tr_dev;
+
+ bdev = blkdev_get_by_dev(dev_id, 0, NULL);
+ if (IS_ERR(bdev)) {
+ int err = PTR_ERR(bdev);
+
+ pr_info("Cannot open device [%u:%u]\n", MAJOR(dev_id),
+ MINOR(dev_id));
+ return ERR_PTR(err);
+ }
+
+ tracker = tracker_get_by_dev(bdev);
+ if (IS_ERR(tracker)) {
+ int err = PTR_ERR(tracker);
+
+ pr_err("Cannot get tracker for device [%u:%u]. errno=%d\n",
+ MAJOR(dev_id), MINOR(dev_id), abs(err));
+ goto put_bdev;
+ }
+ if (tracker) {
+ pr_debug("Device [%u:%u] is already under tracking\n",
+ MAJOR(dev_id), MINOR(dev_id));
+ goto put_bdev;
+ }
+
+ tr_dev = kzalloc(sizeof(struct tracked_device), GFP_KERNEL);
+ if (!tr_dev) {
+ tracker = ERR_PTR(-ENOMEM);
+ goto put_bdev;
+ }
+
+ INIT_LIST_HEAD(&tr_dev->link);
+ tr_dev->dev_id = dev_id;
+
+ tracker = tracker_new(bdev);
+ if (IS_ERR(tracker)) {
+ int err = PTR_ERR(tracker);
+
+ pr_err("Failed to create tracker. errno=%d\n", abs(err));
+ kfree(tr_dev);
+ } else {
+ /*
+ * It is normal that the new trackers filter will have
+ * a ref counter value of 2. This allows not to detach
+ * the filter when the snapshot is released.
+ */
+ bdev_filter_get(&tracker->flt);
+
+ spin_lock(&tracked_device_lock);
+ list_add_tail(&tr_dev->link, &tracked_device_list);
+ spin_unlock(&tracked_device_lock);
+ }
+put_bdev:
+ blkdev_put(bdev, 0);
+ return tracker;
+}
+
+int tracker_remove(dev_t dev_id)
+{
+ int ret;
+ struct tracker *tracker;
+ struct block_device *bdev;
+
+ pr_info("Removing device [%u:%u] from tracking\n", MAJOR(dev_id),
+ MINOR(dev_id));
+
+ bdev = blkdev_get_by_dev(dev_id, 0, NULL);
+ if (IS_ERR(bdev)) {
+ pr_info("Cannot open device [%u:%u]\n", MAJOR(dev_id),
+ MINOR(dev_id));
+ return PTR_ERR(bdev);
+ }
+
+ tracker = tracker_get_by_dev(bdev);
+ if (!tracker) {
+ pr_info("Unable to remove device [%u:%u] from tracking: ",
+ MAJOR(dev_id), MINOR(dev_id));
+ pr_info("tracker not found\n");
+ ret = -ENODATA;
+ goto put_bdev;
+ }
+
+ if (atomic_read(&tracker->snapshot_is_taken)) {
+ pr_err("Tracker for device [%u:%u] is busy with a snapshot\n",
+ MAJOR(dev_id), MINOR(dev_id));
+ ret = -EBUSY;
+ goto put_tracker;
+ }
+
+ ret = tracker_filter_detach(bdev);
+ if (ret)
+ pr_err("Failed to remove tracker from device [%u:%u]\n",
+ MAJOR(dev_id), MINOR(dev_id));
+ else {
+ struct tracked_device *tr_dev = NULL;
+ struct tracked_device *iter_tr_dev;
+
+ spin_lock(&tracked_device_lock);
+ list_for_each_entry(iter_tr_dev, &tracked_device_list, link) {
+ if (iter_tr_dev->dev_id == dev_id) {
+ list_del(&iter_tr_dev->link);
+ tr_dev = iter_tr_dev;
+ break;
+ }
+ }
+ spin_unlock(&tracked_device_lock);
+
+ kfree(tr_dev);
+ }
+put_tracker:
+ tracker_put(tracker);
+put_bdev:
+ blkdev_put(bdev, 0);
+ return ret;
+}
+
+int tracker_read_cbt_bitmap(dev_t dev_id, unsigned int offset, size_t length,
+ char __user *user_buff)
+{
+ int ret;
+ struct tracker *tracker;
+ struct block_device *bdev;
+
+ bdev = blkdev_get_by_dev(dev_id, 0, NULL);
+ if (IS_ERR(bdev)) {
+ pr_info("Cannot open device [%u:%u]\n", MAJOR(dev_id),
+ MINOR(dev_id));
+ return PTR_ERR(bdev);
+ }
+
+ tracker = tracker_get_by_dev(bdev);
+ if (!tracker) {
+ pr_err("Cannot get tracker for device [%u:%u]\n",
+ MAJOR(dev_id), MINOR(dev_id));
+ ret = PTR_ERR(tracker);
+ goto put_bdev;
+ }
+
+ if (atomic_read(&tracker->snapshot_is_taken)) {
+ ret = cbt_map_read_to_user(tracker->cbt_map, user_buff,
+ offset, length);
+ } else {
+ pr_err("Unable to read CBT bitmap for device [%u:%u]: ",
+ MAJOR(dev_id), MINOR(dev_id));
+ pr_err("device is not captured by snapshot\n");
+ ret = -EPERM;
+ }
+ tracker_put(tracker);
+put_bdev:
+ blkdev_put(bdev, 0);
+ return ret;
+}
+
+static inline void collect_cbt_info(dev_t dev_id,
+ struct blk_snap_cbt_info *cbt_info)
+{
+ struct block_device *bdev;
+ struct tracker *tracker;
+
+ bdev = blkdev_get_by_dev(dev_id, 0, NULL);
+ if (IS_ERR(bdev)) {
+ pr_err("Cannot open device [%u:%u]\n", MAJOR(dev_id),
+ MINOR(dev_id));
+ return;
+ }
+
+ tracker = tracker_get_by_dev(bdev);
+ if (!tracker)
+ goto put_bdev;
+ if (!tracker->cbt_map)
+ goto put_tracker;
+
+ cbt_info->device_capacity =
+ (__u64)(tracker->cbt_map->device_capacity << SECTOR_SHIFT);
+ cbt_info->blk_size = (__u32)cbt_map_blk_size(tracker->cbt_map);
+ cbt_info->blk_count = (__u32)tracker->cbt_map->blk_count;
+ cbt_info->snap_number = (__u8)tracker->cbt_map->snap_number_previous;
+
+ export_uuid(cbt_info->generation_id.b, &tracker->cbt_map->generation_id);
+put_tracker:
+ tracker_put(tracker);
+put_bdev:
+ blkdev_put(bdev, 0);
+}
+
+int tracker_collect(int max_count, struct blk_snap_cbt_info *cbt_info,
+ int *pcount)
+{
+ int ret = 0;
+ int count = 0;
+ int iter = 0;
+ struct tracked_device *tr_dev;
+
+ if (!cbt_info) {
+ /**
+ * Just calculate trackers list length.
+ */
+ spin_lock(&tracked_device_lock);
+ list_for_each_entry(tr_dev, &tracked_device_list, link)
+ ++count;
+ spin_unlock(&tracked_device_lock);
+ goto out;
+ }
+
+ spin_lock(&tracked_device_lock);
+ list_for_each_entry(tr_dev, &tracked_device_list, link) {
+ if (count >= max_count) {
+ ret = -ENOBUFS;
+ break;
+ }
+
+ cbt_info[count].dev_id.mj = MAJOR(tr_dev->dev_id);
+ cbt_info[count].dev_id.mn = MINOR(tr_dev->dev_id);
+ ++count;
+ }
+ spin_unlock(&tracked_device_lock);
+
+ if (ret)
+ return ret;
+
+ for (iter = 0; iter < count; iter++) {
+ dev_t dev_id = MKDEV(cbt_info[iter].dev_id.mj,
+ cbt_info[iter].dev_id.mn);
+
+ collect_cbt_info(dev_id, &cbt_info[iter]);
+ }
+out:
+ *pcount = count;
+ return 0;
+}
+
+int tracker_mark_dirty_blocks(dev_t dev_id,
+ struct blk_snap_block_range *block_ranges,
+ unsigned int count)
+{
+ int ret = 0;
+ struct tracker *tracker;
+ struct block_device *bdev;
+
+ bdev = blkdev_get_by_dev(dev_id, 0, NULL);
+ if (IS_ERR(bdev)) {
+ pr_err("Cannot open device [%u:%u]\n", MAJOR(dev_id),
+ MINOR(dev_id));
+ return PTR_ERR(bdev);
+ }
+
+ pr_debug("Marking [%d] dirty blocks for device [%u:%u]\n", count,
+ MAJOR(dev_id), MINOR(dev_id));
+
+ tracker = tracker_get_by_dev(bdev);
+ if (!tracker) {
+ pr_err("Cannot find tracker for device [%u:%u]\n",
+ MAJOR(dev_id), MINOR(dev_id));
+ ret = -ENODEV;
+ goto put_bdev;
+ }
+
+ ret = cbt_map_mark_dirty_blocks(tracker->cbt_map, block_ranges, count);
+ if (ret)
+ pr_err("Failed to set CBT table. errno=%d\n", abs(ret));
+
+ tracker_put(tracker);
+put_bdev:
+ blkdev_put(bdev, 0);
+ return ret;
+}
diff --git a/drivers/block/blksnap/tracker.h b/drivers/block/blksnap/tracker.h
new file mode 100644
index 000000000000..a7ae5312d488
--- /dev/null
+++ b/drivers/block/blksnap/tracker.h
@@ -0,0 +1,74 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef __BLK_SNAP_TRACKER_H
+#define __BLK_SNAP_TRACKER_H
+
+#include <linux/kref.h>
+#include <linux/spinlock.h>
+#include <linux/list.h>
+#include <linux/rwsem.h>
+#include <linux/blkdev.h>
+#include <linux/fs.h>
+
+struct cbt_map;
+struct diff_area;
+
+/**
+ * struct tracker - Tracker for a block device.
+ *
+ * @flt:
+ * The block device filter structure.
+ * @link:
+ * List header. Tracker release cannot be performed in the release_cb()
+ * filters callback function. Therefore, the trackers are queued for
+ * release in the worker thread.
+ * @dev_id:
+ * Original block device ID.
+ * @snapshot_is_taken:
+ * Indicates that a snapshot was taken for the device whose I/O unit are
+ * handled by this tracker.
+ * @cbt_map:
+ * Pointer to a change block tracker map.
+ * @diff_area:
+ * Pointer to a difference area.
+ *
+ * The goal of the tracker is to handle I/O unit. The tracker detectes
+ * the range of sectors that will change and transmits them to the CBT map
+ * and to the difference area.
+ */
+struct tracker {
+ struct bdev_filter flt;
+ struct list_head link;
+ dev_t dev_id;
+
+ atomic_t snapshot_is_taken;
+
+ struct cbt_map *cbt_map;
+ struct diff_area *diff_area;
+};
+
+void tracker_lock(void);
+void tracker_unlock(void);
+
+static inline void tracker_put(struct tracker *tracker)
+{
+ if (likely(tracker))
+ bdev_filter_put(&tracker->flt);
+};
+
+int tracker_init(void);
+void tracker_done(void);
+
+struct tracker *tracker_create_or_get(dev_t dev_id);
+int tracker_remove(dev_t dev_id);
+int tracker_collect(int max_count, struct blk_snap_cbt_info *cbt_info,
+ int *pcount);
+int tracker_read_cbt_bitmap(dev_t dev_id, unsigned int offset, size_t length,
+ char __user *user_buff);
+int tracker_mark_dirty_blocks(dev_t dev_id,
+ struct blk_snap_block_range *block_ranges,
+ unsigned int count);
+
+int tracker_take_snapshot(struct tracker *tracker);
+void tracker_release_snapshot(struct tracker *tracker);
+
+#endif /* __BLK_SNAP_TRACKER_H */
--
2.20.1
> +static bool tracker_submit_bio_cb(struct bio *bio)
> +{
> + struct bdev_filter *flt = bio->bi_bdev->bd_filter;
> + struct bio_list bio_list_on_stack[2] = { };
> + struct bio *new_bio;
> + bool ret = true;
> + struct tracker *tracker = container_of(flt, struct tracker, flt);
> + int err;
> + sector_t sector;
> + sector_t count;
> + unsigned int current_flag;
> +
> + WARN_ON_ONCE(!flt);
> + if (unlikely(!flt))
> + return true;
We're called through the filter, so checking this again here (twice)
is a bit silly.
> + if (bio->bi_opf & REQ_NOWAIT) {
> + if (!percpu_down_read_trylock(&tracker_submit_lock)) {
> + bio_wouldblock_error(bio);
> + return false;
> + }
> + } else
> + percpu_down_read(&tracker_submit_lock);
Does it make sense to make this a global lock vs per-struct tracker?
> + if (!op_is_write(bio_op(bio)))
> + goto out;
> +
> + count = bio_sectors(bio);
> + if (!count)
> + goto out;
Just nitpicking, but what about moving all the code below here
into a separate helper that is only called for op_is_write &&
bio_sectors? It's not going to change anything functionally, but
would make the code easier to follow.
> + current_flag = memalloc_noio_save();
> + bio_list_init(&bio_list_on_stack[0]);
> + current->bio_list = bio_list_on_stack;
> + barrier();
barrier is just a compiler barrier, so it is unlikely to do what
you want. But without a comment I can't even figure out what it is
trying to do.
> +static int tracker_filter_attach(struct block_device *bdev,
> + struct tracker *tracker)
> +{
> + int ret;
> + bool is_frozen = false;
> +
> + pr_debug("Tracker attach filter\n");
> +
> + if (freeze_bdev(bdev))
> + pr_err("Failed to freeze device [%u:%u]\n", MAJOR(bdev->bd_dev),
> + MINOR(bdev->bd_dev));
I think you need to fail the attachment if we can't freeze the device.
> +static int tracker_filter_detach(struct block_device *bdev)
> +{
> + int ret;
> + bool is_frozen = false;
> +
> + pr_debug("Tracker delete filter\n");
> + if (freeze_bdev(bdev))
> + pr_err("Failed to freeze device [%u:%u]\n", MAJOR(bdev->bd_dev),
> + MINOR(bdev->bd_dev));
Same here.
> +/**
> + * tracker_wait_for_release - Waiting for all trackers are released.
> + *
> + * Trackers are released in the worker thread. So, this function allows to wait
> + * for the end of the process of releasing trackers.
> + */
> +static void tracker_wait_for_release(void)
This defeats the reason to move it to the workqueue first, as you
can still deadlock on whatever problem that tried to solve, just
out of reach of lockdep.
> +struct tracker *tracker_create_or_get(dev_t dev_id)
> +{
> + struct tracker *tracker;
> + struct block_device *bdev;
> + struct tracked_device *tr_dev;
> +
> + bdev = blkdev_get_by_dev(dev_id, 0, NULL);
These blkdev_get_by_dev calls are a little problematic, as they
bypass any access restriction (LSMs, containers, etc). That's
why the kernel generally does a blkdev_get_by_name based on the
actual file name, which does all the proper checks. I think the
tracker creation also needs to happen based on names to fit into this
security model. Passing in names should also be much easier for
userspace to start with.
Now for remove, and the other operations on the tracked device:
Is there any reason to not simply add an ioctl method to
bdev_filter_operations, so that you can issue these ioctls against the
tracked device? That removes the need to find the tracked device
entirely and should simplify a lot of things.
In fact thinking wonder if attachment of a filter driver should
go through the block layer using an ioctl on the tracked device
as well, i.e.
- add a name field to bdev_filter_operations
- keep a list of all bdev_filter_operations in the block core
- new core block layer ioctl to associate a block device with a
bdev_filter_operations
- everything after that is done through bdev_filter_operations->ioctl.