[1/2] virtio_ring: Do not call dma_map_page if sg is already mapped.
diff mbox

Message ID 1481180353-11139-2-git-send-email-jliang@xilinx.com
State New
Headers show

Commit Message

Wendy Liang Dec. 8, 2016, 6:59 a.m. UTC
If sg is already dma mapped, do not call dma_map_page() in
vring_map_one_sg().

In case of rpmsg, rpmsg uses dma_alloc_coherent() to allocate
memory to share with the remote. There is no pages setup
in dma_alloc_coherent().

In this case, we cannot convert the virtual address back to the
physical address. In this case, we can setup the sg_dma_addr to
store the DMA address, and also mark the sg is already mapped.

In the vring, we can detect if the address is already mapped
by checking the sg_dma_addr. If yes, do not call dma_map_page().

Signed-off-by: Wendy Liang <jliang@xilinx.com>
---
 drivers/virtio/virtio_ring.c | 6 ++++++
 1 file changed, 6 insertions(+)

Comments

Michael S. Tsirkin Dec. 8, 2016, 4:46 p.m. UTC | #1
On Wed, Dec 07, 2016 at 10:59:12PM -0800, Wendy Liang wrote:
> If sg is already dma mapped, do not call dma_map_page() in
> vring_map_one_sg().
> 
> In case of rpmsg, rpmsg uses dma_alloc_coherent() to allocate
> memory to share with the remote. There is no pages setup
> in dma_alloc_coherent().
> 
> In this case, we cannot convert the virtual address back to the
> physical address. In this case, we can setup the sg_dma_addr to
> store the DMA address, and also mark the sg is already mapped.
> 
> In the vring, we can detect if the address is already mapped
> by checking the sg_dma_addr. If yes, do not call dma_map_page().
> 
> Signed-off-by: Wendy Liang <jliang@xilinx.com>
> ---
>  drivers/virtio/virtio_ring.c | 6 ++++++
>  1 file changed, 6 insertions(+)
> 
> diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c
> index 489bfc6..9793e1f 100644
> --- a/drivers/virtio/virtio_ring.c
> +++ b/drivers/virtio/virtio_ring.c
> @@ -180,6 +180,12 @@ static dma_addr_t vring_map_one_sg(const struct vring_virtqueue *vq,
>  	if (!vring_use_dma_api(vq->vq.vdev))
>  		return (dma_addr_t)sg_phys(sg);
>  
> +	/* If the sg is already mapped, return the DMA address */

How come we even reach this code for rpmsg?
Does vring_use_dma_api return true for rpmsg?

> +	if (sg_dma_address(sg)) {
> +		sg->length = sg_dma_len(sg);
> +		return sg_dma_address(sg);
> +	}
> +

Is there a rule that says 0 is not a valid address?

>  	/*
>  	 * We can't use dma_map_sg, because we don't use scatterlists in
>  	 * the way it expects (we don't guarantee that the scatterlist
> -- 
> 1.9.1
--
To unsubscribe from this list: send the line "unsubscribe linux-remoteproc" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Wendy Liang Dec. 9, 2016, 6:19 p.m. UTC | #2
HI Michael,

On Thu, Dec 8, 2016 at 8:46 AM, Michael S. Tsirkin <mst@redhat.com> wrote:
> On Wed, Dec 07, 2016 at 10:59:12PM -0800, Wendy Liang wrote:
>> If sg is already dma mapped, do not call dma_map_page() in
>> vring_map_one_sg().
>>
>> In case of rpmsg, rpmsg uses dma_alloc_coherent() to allocate
>> memory to share with the remote. There is no pages setup
>> in dma_alloc_coherent().
>>
>> In this case, we cannot convert the virtual address back to the
>> physical address. In this case, we can setup the sg_dma_addr to
>> store the DMA address, and also mark the sg is already mapped.
>>
>> In the vring, we can detect if the address is already mapped
>> by checking the sg_dma_addr. If yes, do not call dma_map_page().
>>
>> Signed-off-by: Wendy Liang <jliang@xilinx.com>
>> ---
>>  drivers/virtio/virtio_ring.c | 6 ++++++
>>  1 file changed, 6 insertions(+)
>>
>> diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c
>> index 489bfc6..9793e1f 100644
>> --- a/drivers/virtio/virtio_ring.c
>> +++ b/drivers/virtio/virtio_ring.c
>> @@ -180,6 +180,12 @@ static dma_addr_t vring_map_one_sg(const struct vring_virtqueue *vq,
>>       if (!vring_use_dma_api(vq->vq.vdev))
>>               return (dma_addr_t)sg_phys(sg);
>>
>> +     /* If the sg is already mapped, return the DMA address */
>
> How come we even reach this code for rpmsg?
> Does vring_use_dma_api return true for rpmsg?
I used vdev feature bit VIRTIO_F_IOMMU_PLATFORM.

>
>> +     if (sg_dma_address(sg)) {
>> +             sg->length = sg_dma_len(sg);
>> +             return sg_dma_address(sg);
>> +     }
>> +
>
> Is there a rule that says 0 is not a valid address?
OK, i see. Is it better to check the sg_dma_len() instead?

However, I just noticed yesterday, there is another patch to
virtio_rpmsg_bus posted to linux-remoteproc mailing list trying to
solve the same issue:
[PATCH v1 2/6] rpmsg: virtio_rpmsg_bus: fix sg_set_buf() when addr is
not a valid kernel address. he didn't use sg_dma_address. And doesn't
require to update the virtio_ring implementation.


>
>>       /*
>>        * We can't use dma_map_sg, because we don't use scatterlists in
>>        * the way it expects (we don't guarantee that the scatterlist
>> --
>> 1.9.1
> --
> To unsubscribe from this list: send the line "unsubscribe linux-remoteproc" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
--
To unsubscribe from this list: send the line "unsubscribe linux-remoteproc" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Patch
diff mbox

diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c
index 489bfc6..9793e1f 100644
--- a/drivers/virtio/virtio_ring.c
+++ b/drivers/virtio/virtio_ring.c
@@ -180,6 +180,12 @@  static dma_addr_t vring_map_one_sg(const struct vring_virtqueue *vq,
 	if (!vring_use_dma_api(vq->vq.vdev))
 		return (dma_addr_t)sg_phys(sg);
 
+	/* If the sg is already mapped, return the DMA address */
+	if (sg_dma_address(sg)) {
+		sg->length = sg_dma_len(sg);
+		return sg_dma_address(sg);
+	}
+
 	/*
 	 * We can't use dma_map_sg, because we don't use scatterlists in
 	 * the way it expects (we don't guarantee that the scatterlist