[PATCH v2] nvme-rdma: fix in-casule data send for chained sgls

[PATCH v2] nvme-rdma: fix in-casule data send for chained sgls

[Sagi Grimberg]

We have only 2 inline sg entries and we allow 4 sg entries for the send
wr sge. Larger sgls entries will be chained. However when we build
in-capsule send wr sge, we iterate without taking into account that the
sgl may be chained and still fit in-capsule (which can happen if the sgl
is bigger than 2, but lower-equal to 4).

Fix in-capsule data mapping to correctly iterate chained sgls.

Reported-by: Walker, Benjamin <benjamin.walker@intel.com>
Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
---
Changes from v1:
- fix silly compilation error

 drivers/nvme/host/rdma.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/drivers/nvme/host/rdma.c b/drivers/nvme/host/rdma.c
index 8d107b201f16..f61bf39c39a0 100644
--- a/drivers/nvme/host/rdma.c
+++ b/drivers/nvme/host/rdma.c
@@ -1320,16 +1320,17 @@ static int nvme_rdma_map_sg_inline(struct nvme_rdma_queue *queue,
 		int count)
 {
 	struct nvme_sgl_desc *sg = &c->common.dptr.sgl;
-	struct scatterlist *sgl = req->data_sgl.sg_table.sgl;
 	struct ib_sge *sge = &req->sge[1];
+	struct scatterlist *sgl;
 	u32 len = 0;
 	int i;
 
-	for (i = 0; i < count; i++, sgl++, sge++) {
+	for_each_sg(req->data_sgl.sg_table.sgl, sgl, count, i) {
 		sge->addr = sg_dma_address(sgl);
 		sge->length = sg_dma_len(sgl);
 		sge->lkey = queue->device->pd->local_dma_lkey;
 		len += sge->length;
+		sge++;
 	}
 
 	sg->addr = cpu_to_le64(queue->ctrl->ctrl.icdoff);
-- 
2.27.0


_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme

Re: [PATCH v2] nvme-rdma: fix in-casule data send for chained sgls

[Max Gurtovoy]

On 5/28/2021 4:16 AM, Sagi Grimberg wrote:
> We have only 2 inline sg entries and we allow 4 sg entries for the send
> wr sge. Larger sgls entries will be chained. However when we build
> in-capsule send wr sge, we iterate without taking into account that the
> sgl may be chained and still fit in-capsule (which can happen if the sgl
> is bigger than 2, but lower-equal to 4).
>
> Fix in-capsule data mapping to correctly iterate chained sgls.
>
> Reported-by: Walker, Benjamin <benjamin.walker@intel.com>
> Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
> ---
> Changes from v1:
> - fix silly compilation error
>
>   drivers/nvme/host/rdma.c | 5 +++--
>   1 file changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/nvme/host/rdma.c b/drivers/nvme/host/rdma.c
> index 8d107b201f16..f61bf39c39a0 100644
> --- a/drivers/nvme/host/rdma.c
> +++ b/drivers/nvme/host/rdma.c
> @@ -1320,16 +1320,17 @@ static int nvme_rdma_map_sg_inline(struct nvme_rdma_queue *queue,
>   		int count)
>   {
>   	struct nvme_sgl_desc *sg = &c->common.dptr.sgl;
> -	struct scatterlist *sgl = req->data_sgl.sg_table.sgl;
>   	struct ib_sge *sge = &req->sge[1];
> +	struct scatterlist *sgl;
>   	u32 len = 0;
>   	int i;
>   
> -	for (i = 0; i < count; i++, sgl++, sge++) {
> +	for_each_sg(req->data_sgl.sg_table.sgl, sgl, count, i) {
>   		sge->addr = sg_dma_address(sgl);
>   		sge->length = sg_dma_len(sgl);
>   		sge->lkey = queue->device->pd->local_dma_lkey;
>   		len += sge->length;
> +		sge++;
>   	}
>   
>   	sg->addr = cpu_to_le64(queue->ctrl->ctrl.icdoff);

Looks good,

Reviewed-by: Max Gurtovoy <mgurtovoy@nvidia.com>



_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme

Re: [PATCH v2] nvme-rdma: fix in-casule data send for chained sgls

[Christoph Hellwig]

Thanks,

applied to nvme-5.13.

_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme