[PATCH] dax: remove silly single-page limitation in dax_zero_page_range

[PATCH] dax: remove silly single-page limitation in dax_zero_page_range

[Darrick J. Wong]

From: Darrick J. Wong <djwong@kernel.org>

It's totally silly that the dax zero_page_range implementations are
required to accept a page count, but one of the four implementations
silently ignores the page count and the wrapper itself errors out if you
try to do more than one page.

Fix the nvdimm implementation to loop over the page count and remove the
artificial limitation.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
---
 drivers/dax/super.c   |    7 -------
 drivers/nvdimm/pmem.c |   14 +++++++++++---
 2 files changed, 11 insertions(+), 10 deletions(-)

diff --git a/drivers/dax/super.c b/drivers/dax/super.c
index fc89e91beea7..ca61a01f9ccd 100644
--- a/drivers/dax/super.c
+++ b/drivers/dax/super.c
@@ -353,13 +353,6 @@ int dax_zero_page_range(struct dax_device *dax_dev, pgoff_t pgoff,
 {
 	if (!dax_alive(dax_dev))
 		return -ENXIO;
-	/*
-	 * There are no callers that want to zero more than one page as of now.
-	 * Once users are there, this check can be removed after the
-	 * device mapper code has been updated to split ranges across targets.
-	 */
-	if (nr_pages != 1)
-		return -EIO;
 
 	return dax_dev->ops->zero_page_range(dax_dev, pgoff, nr_pages);
 }
diff --git a/drivers/nvdimm/pmem.c b/drivers/nvdimm/pmem.c
index 72de88ff0d30..3ef40bf74168 100644
--- a/drivers/nvdimm/pmem.c
+++ b/drivers/nvdimm/pmem.c
@@ -288,10 +288,18 @@ static int pmem_dax_zero_page_range(struct dax_device *dax_dev, pgoff_t pgoff,
 				    size_t nr_pages)
 {
 	struct pmem_device *pmem = dax_get_private(dax_dev);
+	int ret = 0;
 
-	return blk_status_to_errno(pmem_do_write(pmem, ZERO_PAGE(0), 0,
-				   PFN_PHYS(pgoff) >> SECTOR_SHIFT,
-				   PAGE_SIZE));
+	for (; nr_pages > 0 && ret == 0; pgoff++, nr_pages--) {
+		blk_status_t status;
+
+		status = pmem_do_write(pmem, ZERO_PAGE(0), 0,
+				       PFN_PHYS(pgoff) >> SECTOR_SHIFT,
+				       PAGE_SIZE);
+		ret = blk_status_to_errno(status);
+	}
+
+	return ret;
 }
 
 static long pmem_dax_direct_access(struct dax_device *dax_dev,

Re: [PATCH] dax: remove silly single-page limitation in dax_zero_page_range

[Dan Williams]

On Wed, Sep 22, 2021 at 6:09 PM Darrick J. Wong <djwong@kernel.org> wrote:
>
> From: Darrick J. Wong <djwong@kernel.org>
>
> It's totally silly that the dax zero_page_range implementations are
> required to accept a page count, but one of the four implementations
> silently ignores the page count and the wrapper itself errors out if you
> try to do more than one page.
>
> Fix the nvdimm implementation to loop over the page count and remove the
> artificial limitation.
>
> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
> ---
>  drivers/dax/super.c   |    7 -------
>  drivers/nvdimm/pmem.c |   14 +++++++++++---
>  2 files changed, 11 insertions(+), 10 deletions(-)
>
> diff --git a/drivers/dax/super.c b/drivers/dax/super.c
> index fc89e91beea7..ca61a01f9ccd 100644
> --- a/drivers/dax/super.c
> +++ b/drivers/dax/super.c
> @@ -353,13 +353,6 @@ int dax_zero_page_range(struct dax_device *dax_dev, pgoff_t pgoff,
>  {
>         if (!dax_alive(dax_dev))
>                 return -ENXIO;
> -       /*
> -        * There are no callers that want to zero more than one page as of now.
> -        * Once users are there, this check can be removed after the
> -        * device mapper code has been updated to split ranges across targets.
> -        */

It's device-mapper that's the issue, you need to make sure that every
device-mapper zero_page_range implementation knows how to route a
multi-page operation. This is part of the motivation to drop that
support and move simple concatenation and striping into the PMEM
driver directly.