There have been issues with coordination of various subsystems using
get_user_pages. These issues are better described in [1].
An implementation of tracking get_user_pages is currently underway
The implementation requires the use put_user_page*() variants to release
a reference rather than put_page(). The commit that introduced
put_user_pages, Commit fc1d8e7cca2daa18d2fe56b94874848adf89d7f5 ("mm: introduce
put_user_page*(), placeholder version").
The implementation currently simply calls put_page() within
put_user_page(). But in the future, it is to change to add a mechanism
to keep track of get_user_pages. Once a tracking mechanism is
implemented, we can make attempts to work on improving on coordination
between various subsystems using get_user_pages.
[1] https://lwn.net/Articles/753027/
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
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.
The PageReserved check needs a closer look and is not worth messing
around with for now.
Matt, Could you give any suggestions for testing this patch?
If in-case, you are willing to pick this up to test. Could you
apply this patch to this tree
https://github.com/johnhubbard/linux/tree/gup_dma_core
and test it with your devices?
---
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 6166587..75ad263 100644
--- a/drivers/staging/kpc2000/kpc_dma/fileops.c
+++ b/drivers/staging/kpc2000/kpc_dma/fileops.c
@@ -198,9 +198,7 @@ int kpc_dma_transfer(struct dev_private_data *priv, struct kiocb *kcb, unsigned
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:
@@ -221,16 +219,13 @@ void transfer_complete_cb(struct aio_cb_data *acd, size_t xfr_count, u32 flags)
dev_dbg(&acd->ldev->pldev->dev, "transfer_complete_cb(acd = [%p])\n", acd);
- 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);
+ else
+ put_user_page(acd->user_pages[i]);
}
sg_free_table(&acd->sgt);
--
2.7.4
