Filesystem-DAX flushes caches whenever it writes to the address returned
through dax_map_atomic() and when writing back dirty radix entries. That
flushing is only required in the pmem case, so add a dax operation to
allow pmem to take this extra action, but skip it for other dax capable
block_devices like brd.
We still do all the dirty tracking since the radix entry will already be
there for locking purposes. However, the work to clean the entry will be
a nop for some dax drivers.
Cc: Jan Kara <j...@suse.cz>
Cc: Jeff Moyer <jmo...@redhat.com>
Cc: Christoph Hellwig <h...@lst.de>
Cc: Matthew Wilcox <mawil...@microsoft.com>
Cc: Ross Zwisler <ross.zwis...@linux.intel.com>
Signed-off-by: Dan Williams <dan.j.willi...@intel.com>
---
drivers/nvdimm/pmem.c | 1 +
fs/dax.c | 16 ++++++++++++----
include/linux/blkdev.h | 1 +
3 files changed, 14 insertions(+), 4 deletions(-)
diff --git a/drivers/nvdimm/pmem.c b/drivers/nvdimm/pmem.c
index 71e5e365d3fc..68fc7599a053 100644
--- a/drivers/nvdimm/pmem.c
+++ b/drivers/nvdimm/pmem.c
@@ -229,6 +229,7 @@ static size_t pmem_copy_from_iter(void *addr, size_t bytes,
static const struct dax_operations pmem_dax_ops = {
.direct_access = pmem_direct_access,
.copy_from_iter = pmem_copy_from_iter,
+ .flush = wb_cache_pmem,
};
static const struct block_device_operations pmem_fops = {
diff --git a/fs/dax.c b/fs/dax.c
index 22cd57424a55..160024e403f6 100644
--- a/fs/dax.c
+++ b/fs/dax.c
@@ -756,10 +756,19 @@ static void dax_mapping_entry_mkclean(struct
address_space *mapping,
i_mmap_unlock_read(mapping);
}
+static const struct dax_operations *to_dax_ops(struct block_device *bdev)
+{
+ const struct block_device_operations *ops = bdev->bd_disk->fops;
+ const struct dax_operations *dax_ops = ops->dax_ops;
+
+ return dax_ops;
+}
+
static int dax_writeback_one(struct block_device *bdev,
struct address_space *mapping, pgoff_t index, void *entry)
{
struct radix_tree_root *page_tree = &mapping->page_tree;
+ const struct dax_operations *dax_ops = to_dax_ops(bdev);
struct blk_dax_ctl dax;
void *entry2, **slot;
int ret = 0;
@@ -830,7 +839,8 @@ static int dax_writeback_one(struct block_device *bdev,
}
dax_mapping_entry_mkclean(mapping, index, pfn_t_to_pfn(dax.pfn));
- wb_cache_pmem(dax.addr, dax.size);
+ if (dax_ops->flush)
+ dax_ops->flush(dax.addr, dax.size);
/*
* After we have flushed the cache, we can clear the dirty tag. There
* cannot be new dirty data in the pfn after the flush has completed as
@@ -1006,10 +1016,8 @@ static loff_t
dax_iomap_actor(struct inode *inode, loff_t pos, loff_t length, void *data,
struct iomap *iomap)
{
- struct block_device *bdev = iomap->bdev;
size_t (*dax_copy_from_iter)(void *, size_t, struct iov_iter *);
- const struct block_device_operations *ops = bdev->bd_disk->fops;
- const struct dax_operations *dax_ops = ops->dax_ops;
+ const struct dax_operations *dax_ops = to_dax_ops(iomap->bdev);
struct iov_iter *iter = data;
loff_t end = pos + length, done = 0;
ssize_t ret = 0;
diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index 7ca559d124a3..692bdcd63db6 100644
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -1867,6 +1867,7 @@ struct dax_operations {
long (*direct_access)(struct block_device *, sector_t, void **, pfn_t *,
long);
size_t (*copy_from_iter)(void *, size_t, struct iov_iter *);
+ void (*flush)(void *, size_t);
};
struct block_device_operations {
