[PATCH 04/13] misc/habana: Use FOLL_LONGTERM for userptr

John Hubbard jhubbard at nvidia.com
Wed Oct 7 20:46:04 UTC 2020


On 10/7/20 9:44 AM, Daniel Vetter wrote:
> These are persistent, not just for the duration of a dma operation.
> 
> Signed-off-by: Daniel Vetter <daniel.vetter at intel.com>
> Cc: Jason Gunthorpe <jgg at ziepe.ca>
> Cc: Andrew Morton <akpm at linux-foundation.org>
> Cc: John Hubbard <jhubbard at nvidia.com>
> Cc: Jérôme Glisse <jglisse at redhat.com>
> Cc: Jan Kara <jack at suse.cz>
> Cc: Dan Williams <dan.j.williams at intel.com>
> Cc: linux-mm at kvack.org
> Cc: linux-arm-kernel at lists.infradead.org
> Cc: linux-samsung-soc at vger.kernel.org
> Cc: linux-media at vger.kernel.org
> Cc: Oded Gabbay <oded.gabbay at gmail.com>
> Cc: Omer Shpigelman <oshpigelman at habana.ai>
> Cc: Ofir Bitton <obitton at habana.ai>
> Cc: Tomer Tayar <ttayar at habana.ai>
> Cc: Moti Haimovski <mhaimovski at habana.ai>
> Cc: Daniel Vetter <daniel.vetter at ffwll.ch>
> Cc: Greg Kroah-Hartman <gregkh at linuxfoundation.org>
> Cc: Pawel Piskorski <ppiskorski at habana.ai>
> ---
>   drivers/misc/habanalabs/common/memory.c | 3 ++-
>   1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/misc/habanalabs/common/memory.c b/drivers/misc/habanalabs/common/memory.c
> index ef89cfa2f95a..94bef8faa82a 100644
> --- a/drivers/misc/habanalabs/common/memory.c
> +++ b/drivers/misc/habanalabs/common/memory.c
> @@ -1288,7 +1288,8 @@ static int get_user_memory(struct hl_device *hdev, u64 addr, u64 size,
>   		return -ENOMEM;
>   	}
>   
> -	rc = pin_user_pages_fast(start, npages, FOLL_FORCE | FOLL_WRITE,
> +	rc = pin_user_pages_fast(start, npages,
> +				 FOLL_FORCE | FOLL_WRITE | FOLL_LONGTERM,
>   				 userptr->pages);
>   
>   	if (rc != npages) {
> 

Again, from a pin_user_pages_fast() point of view, and not being at all familiar
with the habana driver (but their use of this really does seem clearly _LONGTERM!):

Reviewed-by: John Hubbard <jhubbard at nvidia.com>

thanks,
-- 
John Hubbard
NVIDIA


More information about the dri-devel mailing list