On 7/25/19 5:44 AM, Bharath Vedartham wrote:
> For pages that were retained via get_user_pages*(), release those pages
> via the new put_user_page*() routines, instead of via put_page().
>
> This is part a tree-wide conversion, as described in commit fc1d8e7cca2d
> ("mm: introduce put_user_page*(), placeholder versions").
>
Hi Bharath,
If you like, I could re-post your patch here, modified slightly, as part of
the next version of the miscellaneous call site conversion series [1].
As part of that, we should change this to use put_user_pages_dirty_lock()
(see below).
> Cc: Ira Weiny <ira.weiny@xxxxxxxxx>
> Cc: John Hubbard <jhubbard@xxxxxxxxxx>
> Cc: Jérôme Glisse <jglisse@xxxxxxxxxx>
> Cc: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
> Cc: Matt Sickler <Matt.Sickler@xxxxxxxxxxxxxx>
> Cc: devel@xxxxxxxxxxxxxxxxxxxx
> Cc: linux-kernel@xxxxxxxxxxxxxxx
> Cc: linux-mm@xxxxxxxxx
> Reviewed-by: John Hubbard <jhubbard@xxxxxxxxxx>
> Signed-off-by: Bharath Vedartham <linux.bhar@xxxxxxxxx>
> ---
> Changes since v1
> - Improved changelog by John's suggestion.
> - Moved logic to dirty pages below sg_dma_unmap
> and removed PageReserved check.
> Changes since v2
> - Added back PageResevered check as
> suggested by John Hubbard.
> Changes since v3
> - Changed the changelog as suggested by John.
> - Added John's Reviewed-By tag.
> Changes since v4
> - Rebased the patch on the staging tree.
> - Improved commit log by fixing a line wrap.
> ---
> drivers/staging/kpc2000/kpc_dma/fileops.c | 17 ++++++-----------
> 1 file changed, 6 insertions(+), 11 deletions(-)
>
> diff --git a/drivers/staging/kpc2000/kpc_dma/fileops.c b/drivers/staging/kpc2000/kpc_dma/fileops.c
> index 48ca88b..f15e292 100644
> --- a/drivers/staging/kpc2000/kpc_dma/fileops.c
> +++ b/drivers/staging/kpc2000/kpc_dma/fileops.c
> @@ -190,9 +190,7 @@ static int kpc_dma_transfer(struct dev_private_data *priv,
> sg_free_table(&acd->sgt);
> err_dma_map_sg:
> err_alloc_sg_table:
> - for (i = 0 ; i < acd->page_count ; i++) {
> - put_page(acd->user_pages[i]);
> - }
> + put_user_pages(acd->user_pages, acd->page_count);
> err_get_user_pages:
> kfree(acd->user_pages);
> err_alloc_userpages:
> @@ -211,16 +209,13 @@ void transfer_complete_cb(struct aio_cb_data *acd, size_t xfr_count, u32 flags)
> BUG_ON(acd->ldev == NULL);
> BUG_ON(acd->ldev->pldev == NULL);
>
> - for (i = 0 ; i < acd->page_count ; i++) {
> - if (!PageReserved(acd->user_pages[i])) {
> - set_page_dirty(acd->user_pages[i]);
> - }
> - }
> -
> dma_unmap_sg(&acd->ldev->pldev->dev, acd->sgt.sgl, acd->sgt.nents, acd->ldev->dir);
>
> - for (i = 0 ; i < acd->page_count ; i++) {
> - put_page(acd->user_pages[i]);
> + for (i = 0; i < acd->page_count; i++) {
> + if (!PageReserved(acd->user_pages[i]))
> + put_user_pages_dirty(&acd->user_pages[i], 1);
This would change to:
put_user_pages_dirty_lock(&acd->user_pages[i], 1, true);
...and we'd add this blurb (this time with CH's name spelled properly) to
the commit description:
Note that this effectively changes the code's behavior in
qp_release_pages(): it now ultimately calls set_page_dirty_lock(),
instead of set_page_dirty(). This is probably more accurate.
As Christoph Hellwig put it, "set_page_dirty() is only safe if we are
dealing with a file backed page where we have reference on the inode it
hangs off." [1]
[1] https://lore.kernel.org/r/20190723153640.GB720@xxxxxx
Also, future: I don't know the driver well enough to say, but maybe "true"
could be replaced by "acd->ldev->dir == DMA_FROM_DEVICE", there, but that
would be a separate patch.
thanks,
--
John Hubbard
NVIDIA
> + else
> + put_user_page(acd->user_pages[i]);
> }
>
> sg_free_table(&acd->sgt);
>
