Preparation the headers for the hardware offload
misses the outer ipv4 checksum offload.
It results in bad checksum computed by hardware NIC.
This patch fixes the issue by setting the outer ipv4
checksum field to 0.
Fixes: 4fb7e803eb1a ("ethdev: add Tx preparation")
Cc: stable@dpdk.org
Signed-off-by: Mohsin Kazmi <redacted>
Acked-by: Qi Zhang <redacted>
---
v2:
* Update the commit message with Fixes.
---
lib/net/rte_net.h | 12 +++++++++++-
1 file changed, 11 insertions(+), 1 deletion(-)
Hi Mohsin,
Hope you are fine!
Please see my comments below.
On Wed, Jun 30, 2021 at 01:04:04PM +0200, Mohsin Kazmi wrote:
Re: [PATCH v2] net: prepare the outer ipv4 hdr for checksum
I suggest to highlight that it this is the Intel-specific tx-prepare
function in the commit title. What about:
net: fix Intel-specific Tx preparation for outer checksums
quoted hunk
Preparation the headers for the hardware offload
misses the outer ipv4 checksum offload.
It results in bad checksum computed by hardware NIC.
This patch fixes the issue by setting the outer ipv4
checksum field to 0.
Fixes: 4fb7e803eb1a ("ethdev: add Tx preparation")
Cc: stable@dpdk.org
Signed-off-by: Mohsin Kazmi <redacted>
Acked-by: Qi Zhang <redacted>
---
v2:
* Update the commit message with Fixes.
---
lib/net/rte_net.h | 12 +++++++++++-
1 file changed, 11 insertions(+), 1 deletion(-)
Preparation the headers for the hardware offload
misses the outer ipv4 checksum offload.
It results in bad checksum computed by hardware NIC.
This patch fixes the issue by setting the outer ipv4
checksum field to 0.
Fixes: 4fb7e803eb1a ("ethdev: add Tx preparation")
Cc: stable@dpdk.org
Signed-off-by: Mohsin Kazmi <redacted>
Acked-by: Qi Zhang <redacted>
---
v3:
* Update the conditional test with PKT_TX_OUTER_IP_CKSUM.
* Update the commit title with "Intel-specific".
v2:
* Update the commit message with Fixes.
lib/net/rte_net.h | 15 +++++++++++++--
1 file changed, 13 insertions(+), 2 deletions(-)
From: Thomas Monjalon <hidden> Date: 2021-07-22 19:53:19
07/07/2021 11:14, Mohsin Kazmi:
On Wed, Jun 30, 2021 at 3:09 PM Olivier Matz [off-list ref] wrote:
quoted
quoted
+ if (ol_flags & (PKT_TX_OUTER_IPV4 | PKT_TX_OUTER_IPV6)) {
inner_l3_offset += m->outer_l2_len + m->outer_l3_len;
+ /*
+ * prepare outer ipv4 header checksum by setting it to 0,
+ * in order to be computed by hardware NICs.
+ */
+ if (ol_flags & PKT_TX_OUTER_IP_CKSUM) {
+ ipv4_hdr = rte_pktmbuf_mtod_offset(m,
+ struct rte_ipv4_hdr *,
m->outer_l2_len);
quoted
+ ipv4_hdr->hdr_checksum = 0;
+ }
+ }
What about outer L4 checksum? Does it requires the same than inner?
I am using XL710 for my testing with i40e dpdk driver. AFAIK, It doesn't
support outer l4 checksum. I am not sure if other Intel NICs support it.
This function is used by a lot of drivers.
Try git grep rte_net_intel_cksum_prepare
I think we need more reviews on the v3.
Given it is far from being a new bug, I suggest to wait the next release
in order to have more feedbacks.
From: Thomas Monjalon <hidden> Date: 2021-07-22 19:56:13
+Cc more people for reviews.
07/07/2021 11:40, Mohsin Kazmi:
Preparation the headers for the hardware offload
misses the outer ipv4 checksum offload.
It results in bad checksum computed by hardware NIC.
This patch fixes the issue by setting the outer ipv4
checksum field to 0.
On Thu, Jul 22, 2021 at 09:56:25PM +0200, Thomas Monjalon wrote:
+Cc more people for reviews.
07/07/2021 11:40, Mohsin Kazmi:
quoted
Preparation the headers for the hardware offload
"of" is missing after Preparation?
quoted
misses the outer ipv4 checksum offload.
It results in bad checksum computed by hardware NIC.
This patch fixes the issue by setting the outer ipv4
checksum field to 0.
From: Andrew Rybchenko <hidden> Date: 2021-07-28 15:46:58
On 7/7/21 12:40 PM, Mohsin Kazmi wrote:
quoted hunk
Preparation the headers for the hardware offload
misses the outer ipv4 checksum offload.
It results in bad checksum computed by hardware NIC.
This patch fixes the issue by setting the outer ipv4
checksum field to 0.
Fixes: 4fb7e803eb1a ("ethdev: add Tx preparation")
Cc: stable@dpdk.org
Signed-off-by: Mohsin Kazmi <redacted>
Acked-by: Qi Zhang <redacted>
---
v3:
* Update the conditional test with PKT_TX_OUTER_IP_CKSUM.
* Update the commit title with "Intel-specific".
v2:
* Update the commit message with Fixes.
lib/net/rte_net.h | 15 +++++++++++++--
1 file changed, 13 insertions(+), 2 deletions(-)
On Wed, Jul 28, 2021 at 06:46:53PM +0300, Andrew Rybchenko wrote:
On 7/7/21 12:40 PM, Mohsin Kazmi wrote:
quoted
Preparation the headers for the hardware offload
misses the outer ipv4 checksum offload.
It results in bad checksum computed by hardware NIC.
This patch fixes the issue by setting the outer ipv4
checksum field to 0.
Fixes: 4fb7e803eb1a ("ethdev: add Tx preparation")
Cc: stable@dpdk.org
Signed-off-by: Mohsin Kazmi <redacted>
Acked-by: Qi Zhang <redacted>
---
v3:
* Update the conditional test with PKT_TX_OUTER_IP_CKSUM.
* Update the commit title with "Intel-specific".
v2:
* Update the commit message with Fixes.
lib/net/rte_net.h | 15 +++++++++++++--
1 file changed, 13 insertions(+), 2 deletions(-)
From: Andrew Rybchenko <hidden> Date: 2021-07-31 12:49:07
On 7/30/21 2:11 PM, Olivier Matz wrote:
On Wed, Jul 28, 2021 at 06:46:53PM +0300, Andrew Rybchenko wrote:
quoted
On 7/7/21 12:40 PM, Mohsin Kazmi wrote:
quoted
Preparation the headers for the hardware offload
misses the outer ipv4 checksum offload.
It results in bad checksum computed by hardware NIC.
This patch fixes the issue by setting the outer ipv4
checksum field to 0.
Fixes: 4fb7e803eb1a ("ethdev: add Tx preparation")
Cc: stable@dpdk.org
Signed-off-by: Mohsin Kazmi <redacted>
Acked-by: Qi Zhang <redacted>
---
v3:
* Update the conditional test with PKT_TX_OUTER_IP_CKSUM.
* Update the commit title with "Intel-specific".
v2:
* Update the commit message with Fixes.
lib/net/rte_net.h | 15 +++++++++++++--
1 file changed, 13 insertions(+), 2 deletions(-)
Hi Thomas,
Thanks for the review.
I did the git grep rte_net_intel_cksum_prepare and git grep
PKT_TX_OUTER_UDP_CKSUM. Following are the two drivers that use the function
to prepare headers for checksum which also uses the outer_udp_checksum
offload within drivers.
1) Hisilicon hns3
2) Wangxun txgbe
1) has implemented its own version of functions to prepare for outer header
checksum. It may benefit/impact from the change.
The function "rte_net_intel_cksum_prepare" is intel specific and intel
cards do not support outer l4 checksum offload. DPDK may provide a generic
version of the same function which can be used in different drivers.
-br
Mohsin
On Thu, Jul 22, 2021 at 8:53 PM Thomas Monjalon [off-list ref] wrote:
07/07/2021 11:14, Mohsin Kazmi:
quoted
On Wed, Jun 30, 2021 at 3:09 PM Olivier Matz [off-list ref]
wrote:
quoted
quoted
quoted
+ if (ol_flags & (PKT_TX_OUTER_IPV4 | PKT_TX_OUTER_IPV6)) {
inner_l3_offset += m->outer_l2_len + m->outer_l3_len;
+ /*
+ * prepare outer ipv4 header checksum by setting it to
0,
quoted
quoted
quoted
+ * in order to be computed by hardware NICs.
+ */
+ if (ol_flags & PKT_TX_OUTER_IP_CKSUM) {
+ ipv4_hdr = rte_pktmbuf_mtod_offset(m,
+ struct rte_ipv4_hdr *,
m->outer_l2_len);
quoted
+ ipv4_hdr->hdr_checksum = 0;
+ }
+ }
What about outer L4 checksum? Does it requires the same than inner?
I am using XL710 for my testing with i40e dpdk driver. AFAIK, It doesn't
support outer l4 checksum. I am not sure if other Intel NICs support it.
This function is used by a lot of drivers.
Try git grep rte_net_intel_cksum_prepare
I think we need more reviews on the v3.
Given it is far from being a new bug, I suggest to wait the next release
in order to have more feedbacks.
On Sat, Jul 31, 2021 at 1:49 PM Andrew Rybchenko <
andrew.rybchenko@oktetlabs.ru> wrote:
On 7/30/21 2:11 PM, Olivier Matz wrote:
quoted
On Wed, Jul 28, 2021 at 06:46:53PM +0300, Andrew Rybchenko wrote:
quoted
On 7/7/21 12:40 PM, Mohsin Kazmi wrote:
quoted
Preparation the headers for the hardware offload
misses the outer ipv4 checksum offload.
It results in bad checksum computed by hardware NIC.
This patch fixes the issue by setting the outer ipv4
checksum field to 0.
Fixes: 4fb7e803eb1a ("ethdev: add Tx preparation")
Cc: stable@dpdk.org
Signed-off-by: Mohsin Kazmi <redacted>
Acked-by: Qi Zhang <redacted>
---
v3:
* Update the conditional test with PKT_TX_OUTER_IP_CKSUM.
* Update the commit title with "Intel-specific".
v2:
* Update the commit message with Fixes.
lib/net/rte_net.h | 15 +++++++++++++--
1 file changed, 13 insertions(+), 2 deletions(-)
* Mainly it is required to avoid fragmented headers check if
* no offloads are requested.
*/
- if (!(ol_flags & (PKT_TX_IP_CKSUM | PKT_TX_L4_MASK |
PKT_TX_TCP_SEG)))
quoted
quoted
quoted
+ if (!(ol_flags & (PKT_TX_IP_CKSUM | PKT_TX_L4_MASK |
PKT_TX_TCP_SEG |
quoted
quoted
quoted
+ PKT_TX_OUTER_IP_CKSUM)))
return 0;
- if (ol_flags & (PKT_TX_OUTER_IPV4 | PKT_TX_OUTER_IPV6))
+ if (ol_flags & (PKT_TX_OUTER_IPV4 | PKT_TX_OUTER_IPV6)) {
inner_l3_offset += m->outer_l2_len + m->outer_l3_len;
+ /*
+ * prepare outer ipv4 header checksum by setting it to 0,
+ * in order to be computed by hardware NICs.
+ */
+ if (ol_flags & PKT_TX_OUTER_IP_CKSUM) {
+ ipv4_hdr = rte_pktmbuf_mtod_offset(m,
+ struct rte_ipv4_hdr *,
m->outer_l2_len);
quoted
quoted
quoted
+ ipv4_hdr->hdr_checksum = 0;
Here we assume that the field is located in the first segment.
Unlikely but it still could be false. We must handle it properly.
This is specified in the API comment, so I think it has to be checked
by the caller.
If no, what's the point to spoil memory here if stricter check is
done few lines below.
We have two possibilities:
1) take the whole block of above code after the strict check: Then strict
check will use m->outer_l2_len + m->outer_l3_len directly without any
condition and we will be on the mercy of drivers to initialize these to 0
if outer headers are not use. Drivers usually don't set the fields which
they are not interested in because of performance reasons as
setting these values per packet will cost them additional cycles.
2) Taking just PKT_TX_OUTER_IP_CKSUM conditional check after the strict
fragmented check: In that case, each packet will hit an extra conditional
check without getting benefit from it, again with a performance penalty.
I am more inclined towards solution 1. But I also welcome other
suggestions/comments.
Are we good with this patch with the current state? @Olivier: Any comments
on the above suggestions?
On Tue, Aug 3, 2021 at 1:49 PM Mohsin Kazmi [off-list ref]
wrote:
On Sat, Jul 31, 2021 at 1:49 PM Andrew Rybchenko <
andrew.rybchenko@oktetlabs.ru> wrote:
quoted
On 7/30/21 2:11 PM, Olivier Matz wrote:
quoted
On Wed, Jul 28, 2021 at 06:46:53PM +0300, Andrew Rybchenko wrote:
quoted
On 7/7/21 12:40 PM, Mohsin Kazmi wrote:
quoted
Preparation the headers for the hardware offload
misses the outer ipv4 checksum offload.
It results in bad checksum computed by hardware NIC.
This patch fixes the issue by setting the outer ipv4
checksum field to 0.
Fixes: 4fb7e803eb1a ("ethdev: add Tx preparation")
Cc: stable@dpdk.org
Signed-off-by: Mohsin Kazmi <redacted>
Acked-by: Qi Zhang <redacted>
---
v3:
* Update the conditional test with PKT_TX_OUTER_IP_CKSUM.
* Update the commit title with "Intel-specific".
v2:
* Update the commit message with Fixes.
lib/net/rte_net.h | 15 +++++++++++++--
1 file changed, 13 insertions(+), 2 deletions(-)
* Mainly it is required to avoid fragmented headers check if
* no offloads are requested.
*/
- if (!(ol_flags & (PKT_TX_IP_CKSUM | PKT_TX_L4_MASK |
PKT_TX_TCP_SEG)))
quoted
quoted
quoted
+ if (!(ol_flags & (PKT_TX_IP_CKSUM | PKT_TX_L4_MASK |
PKT_TX_TCP_SEG |
quoted
quoted
quoted
+ PKT_TX_OUTER_IP_CKSUM)))
return 0;
- if (ol_flags & (PKT_TX_OUTER_IPV4 | PKT_TX_OUTER_IPV6))
+ if (ol_flags & (PKT_TX_OUTER_IPV4 | PKT_TX_OUTER_IPV6)) {
inner_l3_offset += m->outer_l2_len + m->outer_l3_len;
+ /*
+ * prepare outer ipv4 header checksum by setting it to 0,
+ * in order to be computed by hardware NICs.
+ */
+ if (ol_flags & PKT_TX_OUTER_IP_CKSUM) {
+ ipv4_hdr = rte_pktmbuf_mtod_offset(m,
+ struct rte_ipv4_hdr *,
m->outer_l2_len);
quoted
quoted
quoted
+ ipv4_hdr->hdr_checksum = 0;
Here we assume that the field is located in the first segment.
Unlikely but it still could be false. We must handle it properly.
This is specified in the API comment, so I think it has to be checked
by the caller.
If no, what's the point to spoil memory here if stricter check is
done few lines below.
We have two possibilities:
1) take the whole block of above code after the strict check: Then strict
check will use m->outer_l2_len + m->outer_l3_len directly without any
condition and we will be on the mercy of drivers to initialize these to 0
if outer headers are not use. Drivers usually don't set the fields which
they are not interested in because of performance reasons as
setting these values per packet will cost them additional cycles.
2) Taking just PKT_TX_OUTER_IP_CKSUM conditional check after the strict
fragmented check: In that case, each packet will hit an extra conditional
check without getting benefit from it, again with a performance penalty.
I am more inclined towards solution 1. But I also welcome other
suggestions/comments.
Preparation of the headers for the hardware offload
misses the outer IPv4 checksum offload.
It results in bad checksum computed by hardware NIC.
This patch fixes the issue by setting the outer IPv4
checksum field to 0.
Fixes: 4fb7e803eb1a ("ethdev: add Tx preparation")
Cc: stable@dpdk.org
Signed-off-by: Mohsin Kazmi <redacted>
Acked-by: Qi Zhang <redacted>
Acked-by: Olivier Matz <redacted>
---
v4:
* Update the commit message
v3:
* Update the conditional test with PKT_TX_OUTER_IP_CKSUM.
* Update the commit title with "Intel-specific".
v2:
* Update the commit message with Fixes.
lib/net/rte_net.h | 15 +++++++++++++--
1 file changed, 13 insertions(+), 2 deletions(-)
Preparation of the headers for the hardware offload
misses the outer IPv4 checksum offload.
It results in bad checksum computed by hardware NIC.
This patch fixes the issue by setting the outer IPv4
checksum field to 0.
Fixes: 4fb7e803eb1a ("ethdev: add Tx preparation")
Cc: stable@dpdk.org
Signed-off-by: Mohsin Kazmi <redacted>
Acked-by: Qi Zhang <redacted>
Acked-by: Olivier Matz <redacted>
Not directly related with this patch, but is the function
'rte_net_intel_cksum_flags_prepare()' really Intel specific?
I can see sfc & ena are also using this function, should we rename it?
Preparation of the headers for the hardware offload
misses the outer IPv4 checksum offload.
It results in bad checksum computed by hardware NIC.
This patch fixes the issue by setting the outer IPv4
checksum field to 0.
Fixes: 4fb7e803eb1a ("ethdev: add Tx preparation")
Cc: stable@dpdk.org
Signed-off-by: Mohsin Kazmi <redacted>
Acked-by: Qi Zhang <redacted>
Acked-by: Olivier Matz <redacted>
Applied to dpdk-next-net/main, thanks.
Patch title updated as following while merging:
net: fix checksum offload for outer IPv4