[dpdk-dev] [PATCH] net/mlx5: fix inlining segmented TSO packet

[dpdk-dev] [PATCH] net/mlx5: fix inlining segmented TSO packet

[Yongseok Koh]

When a multi-segmented packet is inlined, data can be further inlined even
after the first segment. In case of TSO packet, extra inline data after TSO
header should be carried by an inline DSEG which has 4B inline header
recording the length of the inline data. If more than one segment is
inlined, the length doesn't count from the second segment. This will cause
a fault in HW and CQE will have an error, which is ignored by PMD.

Fixes: f895536be4fa ("net/mlx5: enable inlining data from multiple segments")
Cc: stable@dpdk.org

Signed-off-by: Yongseok Koh <yskoh@mellanox.com>
---
 drivers/net/mlx5/mlx5_rxtx.c | 13 +++++++++++--
 1 file changed, 11 insertions(+), 2 deletions(-)

diff --git a/drivers/net/mlx5/mlx5_rxtx.c b/drivers/net/mlx5/mlx5_rxtx.c
index d960a73d5..2eed65642 100644
--- a/drivers/net/mlx5/mlx5_rxtx.c
+++ b/drivers/net/mlx5/mlx5_rxtx.c
@@ -680,8 +680,17 @@ mlx5_tx_burst(void *dpdk_txq, struct rte_mbuf **pkts, uint16_t pkts_n)
 			} else if (!segs_n) {
 				goto next_pkt;
 			} else {
-				raw += copy_b;
-				inline_room -= copy_b;
+				/*
+				 * Further inline the next segment only for
+				 * non-TSO packets.
+				 */
+				if (!tso) {
+					raw += copy_b;
+					inline_room -= copy_b;
+				} else {
+					inline_room = 0;
+				}
+				/* Move to the next segment. */
 				--segs_n;
 				buf = buf->next;
 				assert(buf);
-- 
2.11.0

Re: [dpdk-dev] [PATCH] net/mlx5: fix inlining segmented TSO packet

[Yongseok Koh]

> On May 10, 2018, at 5:50 PM, Yongseok Koh <yskoh@mellanox.com> wrote:
> 
> When a multi-segmented packet is inlined, data can be further inlined even
> after the first segment. In case of TSO packet, extra inline data after TSO
> header should be carried by an inline DSEG which has 4B inline header
> recording the length of the inline data. If more than one segment is
> inlined, the length doesn't count from the second segment. This will cause
> a fault in HW and CQE will have an error, which is ignored by PMD.
> 
> Fixes: f895536be4fa ("net/mlx5: enable inlining data from multiple segments")
> Cc: stable@dpdk.org
> 
> Signed-off-by: Yongseok Koh <yskoh@mellanox.com>

Shahaf,

If you apply this patch, please amend the signed-off-by tag.
I should've added Xueming as he found it for the first time.

> Signed-off-by: Xueming Li <xuemingl@mellanox.com>
> Signed-off-by: Yongseok Koh <yskoh@mellanox.com>


Thanks,
Yongseok

[dpdk-dev] [PATCH v2] net/mlx5: fix inlining segmented TSO packet

[Yongseok Koh]

When a multi-segmented packet is inlined, data can be further inlined even
after the first segment. In case of TSO packet, extra inline data after TSO
header should be carried by an inline DSEG which has 4B inline header
recording the length of the inline data. If more than one segment is
inlined, the length doesn't count from the second segment. This will cause
a fault in HW and CQE will have an error, which is ignored by PMD.

Fixes: f895536be4fa ("net/mlx5: enable inlining data from multiple segments")
Cc: stable@dpdk.org

Signed-off-by: Xueming Li <xuemingl@mellanox.com>
Signed-off-by: Yongseok Koh <yskoh@mellanox.com>
---

v2:
* replace unnecessary check of 'inl' with assert().
* add "Signed-off-by: Xueming Li".

 drivers/net/mlx5/mlx5_rxtx.c | 16 +++++++++++++---
 1 file changed, 13 insertions(+), 3 deletions(-)

diff --git a/drivers/net/mlx5/mlx5_rxtx.c b/drivers/net/mlx5/mlx5_rxtx.c
index d960a73d5..a2b373e21 100644
--- a/drivers/net/mlx5/mlx5_rxtx.c
+++ b/drivers/net/mlx5/mlx5_rxtx.c
@@ -645,7 +645,8 @@ mlx5_tx_burst(void *dpdk_txq, struct rte_mbuf **pkts, uint16_t pkts_n)
 				if (unlikely(max_wqe < n))
 					break;
 				max_wqe -= n;
-				if (tso && !inl) {
+				if (tso) {
+					assert(inl == 0);
 					inl = rte_cpu_to_be_32(copy_b |
 							       MLX5_INLINE_SEG);
 					rte_memcpy((void *)raw,
@@ -680,8 +681,17 @@ mlx5_tx_burst(void *dpdk_txq, struct rte_mbuf **pkts, uint16_t pkts_n)
 			} else if (!segs_n) {
 				goto next_pkt;
 			} else {
-				raw += copy_b;
-				inline_room -= copy_b;
+				/*
+				 * Further inline the next segment only for
+				 * non-TSO packets.
+				 */
+				if (!tso) {
+					raw += copy_b;
+					inline_room -= copy_b;
+				} else {
+					inline_room = 0;
+				}
+				/* Move to the next segment. */
 				--segs_n;
 				buf = buf->next;
 				assert(buf);
-- 
2.11.0

Re: [dpdk-dev] [PATCH v2] net/mlx5: fix inlining segmented TSO packet

[Shahaf Shuler]

Friday, May 11, 2018 8:39 PM, Yongseok Koh:
> Subject: [PATCH v2] net/mlx5: fix inlining segmented TSO packet
> 
> When a multi-segmented packet is inlined, data can be further inlined even
> after the first segment. In case of TSO packet, extra inline data after TSO
> header should be carried by an inline DSEG which has 4B inline header
> recording the length of the inline data. If more than one segment is inlined,
> the length doesn't count from the second segment. This will cause a fault in
> HW and CQE will have an error, which is ignored by PMD.
> 
> Fixes: f895536be4fa ("net/mlx5: enable inlining data from multiple
> segments")
> Cc: stable@dpdk.org
> 
> Signed-off-by: Xueming Li <xuemingl@mellanox.com>
> Signed-off-by: Yongseok Koh <yskoh@mellanox.com>

Applied to next-net-mlx, thanks.