aboutsummaryrefslogtreecommitdiffstats
path: root/drivers/md/dm-writecache.c
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/md/dm-writecache.c')
-rw-r--r--drivers/md/dm-writecache.c120
1 files changed, 80 insertions, 40 deletions
diff --git a/drivers/md/dm-writecache.c b/drivers/md/dm-writecache.c
index 1cb137f0ef9d..2727a525b5bb 100644
--- a/drivers/md/dm-writecache.c
+++ b/drivers/md/dm-writecache.c
@@ -280,6 +280,8 @@ static int persistent_memory_claim(struct dm_writecache *wc)
while (daa-- && i < p) {
pages[i++] = pfn_t_to_page(pfn);
pfn.val++;
+ if (!(i & 15))
+ cond_resched();
}
} while (i < p);
wc->memory_map = vmap(pages, p, VM_MAP, PAGE_KERNEL);
@@ -443,7 +445,13 @@ static void writecache_notify_io(unsigned long error, void *context)
complete(&endio->c);
}
-static void ssd_commit_flushed(struct dm_writecache *wc)
+static void writecache_wait_for_ios(struct dm_writecache *wc, int direction)
+{
+ wait_event(wc->bio_in_progress_wait[direction],
+ !atomic_read(&wc->bio_in_progress[direction]));
+}
+
+static void ssd_commit_flushed(struct dm_writecache *wc, bool wait_for_ios)
{
struct dm_io_region region;
struct dm_io_request req;
@@ -489,17 +497,20 @@ static void ssd_commit_flushed(struct dm_writecache *wc)
writecache_notify_io(0, &endio);
wait_for_completion_io(&endio.c);
+ if (wait_for_ios)
+ writecache_wait_for_ios(wc, WRITE);
+
writecache_disk_flush(wc, wc->ssd_dev);
memset(wc->dirty_bitmap, 0, wc->dirty_bitmap_size);
}
-static void writecache_commit_flushed(struct dm_writecache *wc)
+static void writecache_commit_flushed(struct dm_writecache *wc, bool wait_for_ios)
{
if (WC_MODE_PMEM(wc))
wmb();
else
- ssd_commit_flushed(wc);
+ ssd_commit_flushed(wc, wait_for_ios);
}
static void writecache_disk_flush(struct dm_writecache *wc, struct dm_dev *dev)
@@ -523,12 +534,6 @@ static void writecache_disk_flush(struct dm_writecache *wc, struct dm_dev *dev)
writecache_error(wc, r, "error flushing metadata: %d", r);
}
-static void writecache_wait_for_ios(struct dm_writecache *wc, int direction)
-{
- wait_event(wc->bio_in_progress_wait[direction],
- !atomic_read(&wc->bio_in_progress[direction]));
-}
-
#define WFE_RETURN_FOLLOWING 1
#define WFE_LOWEST_SEQ 2
@@ -623,6 +628,12 @@ static void writecache_add_to_freelist(struct dm_writecache *wc, struct wc_entry
wc->freelist_size++;
}
+static inline void writecache_verify_watermark(struct dm_writecache *wc)
+{
+ if (unlikely(wc->freelist_size + wc->writeback_size <= wc->freelist_high_watermark))
+ queue_work(wc->writeback_wq, &wc->writeback_work);
+}
+
static struct wc_entry *writecache_pop_from_freelist(struct dm_writecache *wc)
{
struct wc_entry *e;
@@ -644,8 +655,8 @@ static struct wc_entry *writecache_pop_from_freelist(struct dm_writecache *wc)
list_del(&e->lru);
}
wc->freelist_size--;
- if (unlikely(wc->freelist_size + wc->writeback_size <= wc->freelist_high_watermark))
- queue_work(wc->writeback_wq, &wc->writeback_work);
+
+ writecache_verify_watermark(wc);
return e;
}
@@ -725,14 +736,12 @@ static void writecache_flush(struct dm_writecache *wc)
e = e2;
cond_resched();
}
- writecache_commit_flushed(wc);
-
- writecache_wait_for_ios(wc, WRITE);
+ writecache_commit_flushed(wc, true);
wc->seq_count++;
pmem_assign(sb(wc)->seq_count, cpu_to_le64(wc->seq_count));
writecache_flush_region(wc, &sb(wc)->seq_count, sizeof sb(wc)->seq_count);
- writecache_commit_flushed(wc);
+ writecache_commit_flushed(wc, false);
wc->overwrote_committed = false;
@@ -756,7 +765,7 @@ static void writecache_flush(struct dm_writecache *wc)
}
if (need_flush_after_free)
- writecache_commit_flushed(wc);
+ writecache_commit_flushed(wc, false);
}
static void writecache_flush_work(struct work_struct *work)
@@ -799,6 +808,8 @@ static void writecache_discard(struct dm_writecache *wc, sector_t start, sector_
writecache_wait_for_ios(wc, WRITE);
discarded_something = true;
}
+ if (!writecache_entry_is_committed(wc, e))
+ wc->uncommitted_blocks--;
writecache_free_entry(wc, e);
}
@@ -809,7 +820,7 @@ static void writecache_discard(struct dm_writecache *wc, sector_t start, sector_
}
if (discarded_something)
- writecache_commit_flushed(wc);
+ writecache_commit_flushed(wc, false);
}
static bool writecache_wait_for_writeback(struct dm_writecache *wc)
@@ -838,7 +849,7 @@ static void writecache_suspend(struct dm_target *ti)
}
wc_unlock(wc);
- flush_workqueue(wc->writeback_wq);
+ drain_workqueue(wc->writeback_wq);
wc_lock(wc);
if (flush_on_suspend)
@@ -866,11 +877,30 @@ static int writecache_alloc_entries(struct dm_writecache *wc)
struct wc_entry *e = &wc->entries[b];
e->index = b;
e->write_in_progress = false;
+ cond_resched();
}
return 0;
}
+static int writecache_read_metadata(struct dm_writecache *wc, sector_t n_sectors)
+{
+ struct dm_io_region region;
+ struct dm_io_request req;
+
+ region.bdev = wc->ssd_dev->bdev;
+ region.sector = wc->start_sector;
+ region.count = n_sectors;
+ req.bi_op = REQ_OP_READ;
+ req.bi_op_flags = REQ_SYNC;
+ req.mem.type = DM_IO_VMA;
+ req.mem.ptr.vma = (char *)wc->memory_map;
+ req.client = wc->dm_io;
+ req.notify.fn = NULL;
+
+ return dm_io(&req, 1, &region, NULL);
+}
+
static void writecache_resume(struct dm_target *ti)
{
struct dm_writecache *wc = ti->private;
@@ -881,8 +911,18 @@ static void writecache_resume(struct dm_target *ti)
wc_lock(wc);
- if (WC_MODE_PMEM(wc))
+ if (WC_MODE_PMEM(wc)) {
persistent_memory_invalidate_cache(wc->memory_map, wc->memory_map_size);
+ } else {
+ r = writecache_read_metadata(wc, wc->metadata_sectors);
+ if (r) {
+ size_t sb_entries_offset;
+ writecache_error(wc, r, "unable to read metadata: %d", r);
+ sb_entries_offset = offsetof(struct wc_memory_superblock, entries);
+ memset((char *)wc->memory_map + sb_entries_offset, -1,
+ (wc->metadata_sectors << SECTOR_SHIFT) - sb_entries_offset);
+ }
+ }
wc->tree = RB_ROOT;
INIT_LIST_HEAD(&wc->lru);
@@ -920,6 +960,7 @@ static void writecache_resume(struct dm_target *ti)
e->original_sector = le64_to_cpu(wme.original_sector);
e->seq_count = le64_to_cpu(wme.seq_count);
}
+ cond_resched();
}
#endif
for (b = 0; b < wc->n_blocks; b++) {
@@ -958,9 +999,11 @@ erase_this:
if (need_flush) {
writecache_flush_all_metadata(wc);
- writecache_commit_flushed(wc);
+ writecache_commit_flushed(wc, false);
}
+ writecache_verify_watermark(wc);
+
wc_unlock(wc);
}
@@ -1218,7 +1261,8 @@ bio_copy:
}
} while (bio->bi_iter.bi_size);
- if (unlikely(wc->uncommitted_blocks >= wc->autocommit_blocks))
+ if (unlikely(bio->bi_opf & REQ_FUA ||
+ wc->uncommitted_blocks >= wc->autocommit_blocks))
writecache_flush(wc);
else
writecache_schedule_autocommit(wc);
@@ -1341,7 +1385,7 @@ static void __writecache_endio_pmem(struct dm_writecache *wc, struct list_head *
wc->writeback_size--;
n_walked++;
if (unlikely(n_walked >= ENDIO_LATENCY)) {
- writecache_commit_flushed(wc);
+ writecache_commit_flushed(wc, false);
wc_unlock(wc);
wc_lock(wc);
n_walked = 0;
@@ -1422,7 +1466,7 @@ pop_from_list:
writecache_wait_for_ios(wc, READ);
}
- writecache_commit_flushed(wc);
+ writecache_commit_flushed(wc, false);
wc_unlock(wc);
}
@@ -1748,14 +1792,16 @@ static int init_memory(struct dm_writecache *wc)
pmem_assign(sb(wc)->n_blocks, cpu_to_le64(wc->n_blocks));
pmem_assign(sb(wc)->seq_count, cpu_to_le64(0));
- for (b = 0; b < wc->n_blocks; b++)
+ for (b = 0; b < wc->n_blocks; b++) {
write_original_sector_seq_count(wc, &wc->entries[b], -1, -1);
+ cond_resched();
+ }
writecache_flush_all_metadata(wc);
- writecache_commit_flushed(wc);
+ writecache_commit_flushed(wc, false);
pmem_assign(sb(wc)->magic, cpu_to_le32(MEMORY_SUPERBLOCK_MAGIC));
writecache_flush_region(wc, &sb(wc)->magic, sizeof sb(wc)->magic);
- writecache_commit_flushed(wc);
+ writecache_commit_flushed(wc, false);
return 0;
}
@@ -1958,6 +2004,12 @@ static int writecache_ctr(struct dm_target *ti, unsigned argc, char **argv)
ti->error = "Invalid block size";
goto bad;
}
+ if (wc->block_size < bdev_logical_block_size(wc->dev->bdev) ||
+ wc->block_size < bdev_logical_block_size(wc->ssd_dev->bdev)) {
+ r = -EINVAL;
+ ti->error = "Block size is smaller than device logical block size";
+ goto bad;
+ }
wc->block_size_bits = __ffs(wc->block_size);
wc->max_writeback_jobs = MAX_WRITEBACK_JOBS;
@@ -2046,8 +2098,6 @@ invalid_optional:
goto bad;
}
} else {
- struct dm_io_region region;
- struct dm_io_request req;
size_t n_blocks, n_metadata_blocks;
uint64_t n_bitmap_bits;
@@ -2104,19 +2154,9 @@ invalid_optional:
goto bad;
}
- region.bdev = wc->ssd_dev->bdev;
- region.sector = wc->start_sector;
- region.count = wc->metadata_sectors;
- req.bi_op = REQ_OP_READ;
- req.bi_op_flags = REQ_SYNC;
- req.mem.type = DM_IO_VMA;
- req.mem.ptr.vma = (char *)wc->memory_map;
- req.client = wc->dm_io;
- req.notify.fn = NULL;
-
- r = dm_io(&req, 1, &region, NULL);
+ r = writecache_read_metadata(wc, wc->block_size >> SECTOR_SHIFT);
if (r) {
- ti->error = "Unable to read metadata";
+ ti->error = "Unable to read first block of metadata";
goto bad;
}
}