From: Dariusz Sosnowski <dsosnowski@nvidia.com>
To: Viacheslav Ovsiienko <viacheslavo@nvidia.com>,
Bing Zhao <bingz@nvidia.com>, Ori Kam <orika@nvidia.com>,
Suanming Mou <suanmingm@nvidia.com>,
Matan Azrad <matan@nvidia.com>
Cc: <dev@dpdk.org>, <stable@dpdk.org>
Subject: [PATCH] net/mlx5: fix modify field action on group 0
Date: Fri, 25 Apr 2025 21:32:57 +0200 [thread overview]
Message-ID: <20250425193258.38994-1-dsosnowski@nvidia.com> (raw)
HW modify header commands generated for multiple modify field
flow actions, which modify/access the same packet fields
do not have to be separated by NOPs when used on group 0.
This is because:
- On group > 0, HW uses Modify Header Pattern objects
which require NOP explicitly.
- On group 0, modify field action is implemented using
Modify Header Context object managed by FW.
FW inserts requires NOPs internally.
mlx5 PMD inserted NOP always, which caused flow/table creation
failures on group 0 flow rules.
This patch addresses that.
Fixes: 0f4aa72b99da ("net/mlx5: support flow modify field with HWS")
Cc: suanmingm@nvidia.com
Cc: stable@dpdk.org
Signed-off-by: Dariusz Sosnowski <dsosnowski@nvidia.com>
Acked-by: Bing Zhao <bingz@nvidia.com>
---
drivers/net/mlx5/mlx5_flow_hw.c | 21 ++++++++++++++++-----
1 file changed, 16 insertions(+), 5 deletions(-)
diff --git a/drivers/net/mlx5/mlx5_flow_hw.c b/drivers/net/mlx5/mlx5_flow_hw.c
index 20d38ce414..5ae6ac1cfa 100644
--- a/drivers/net/mlx5/mlx5_flow_hw.c
+++ b/drivers/net/mlx5/mlx5_flow_hw.c
@@ -1422,7 +1422,8 @@ flow_hw_action_modify_field_is_shared(const struct rte_flow_action *action,
static __rte_always_inline bool
flow_hw_should_insert_nop(const struct mlx5_hw_modify_header_action *mhdr,
- const struct mlx5_modification_cmd *cmd)
+ const struct mlx5_modification_cmd *cmd,
+ const struct rte_flow_attr *attr)
{
struct mlx5_modification_cmd last_cmd = { { 0 } };
struct mlx5_modification_cmd new_cmd = { { 0 } };
@@ -1430,6 +1431,15 @@ flow_hw_should_insert_nop(const struct mlx5_hw_modify_header_action *mhdr,
unsigned int last_type;
bool should_insert = false;
+ /*
+ * Modify header action list does not require NOPs in root table,
+ * because different type of underlying object is used:
+ * - in root table - MODIFY_HEADER_CONTEXT (does not require NOPs),
+ * - in non-root - either inline modify action or based on Modify Header Pattern
+ * (which requires NOPs).
+ */
+ if (attr->group == 0)
+ return false;
if (cmds_num == 0)
return false;
last_cmd = *(&mhdr->mhdr_cmds[cmds_num - 1]);
@@ -1508,7 +1518,8 @@ flow_hw_mhdr_cmd_append(struct mlx5_hw_modify_header_action *mhdr,
static __rte_always_inline int
flow_hw_converted_mhdr_cmds_append(struct mlx5_hw_modify_header_action *mhdr,
- struct mlx5_flow_dv_modify_hdr_resource *resource)
+ struct mlx5_flow_dv_modify_hdr_resource *resource,
+ const struct rte_flow_attr *attr)
{
uint32_t idx;
int ret;
@@ -1516,7 +1527,7 @@ flow_hw_converted_mhdr_cmds_append(struct mlx5_hw_modify_header_action *mhdr,
for (idx = 0; idx < resource->actions_num; ++idx) {
struct mlx5_modification_cmd *src = &resource->actions[idx];
- if (flow_hw_should_insert_nop(mhdr, src)) {
+ if (flow_hw_should_insert_nop(mhdr, src, attr)) {
ret = flow_hw_mhdr_cmd_nop_append(mhdr);
if (ret)
return ret;
@@ -1639,14 +1650,14 @@ flow_hw_modify_field_compile(struct rte_eth_dev *dev,
* This NOP command will not be a part of action's command range used to update commands
* on rule creation.
*/
- if (flow_hw_should_insert_nop(mhdr, &resource->actions[0])) {
+ if (flow_hw_should_insert_nop(mhdr, &resource->actions[0], attr)) {
ret = flow_hw_mhdr_cmd_nop_append(mhdr);
if (ret)
return rte_flow_error_set(error, ret, RTE_FLOW_ERROR_TYPE_UNSPECIFIED,
NULL, "too many modify field operations specified");
}
cmds_start = mhdr->mhdr_cmds_num;
- ret = flow_hw_converted_mhdr_cmds_append(mhdr, resource);
+ ret = flow_hw_converted_mhdr_cmds_append(mhdr, resource, attr);
if (ret)
return rte_flow_error_set(error, ret, RTE_FLOW_ERROR_TYPE_UNSPECIFIED,
NULL, "too many modify field operations specified");
--
2.39.5
reply other threads:[~2025-04-25 19:33 UTC|newest]
Thread overview: [no followups] expand[flat|nested] mbox.gz Atom feed
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20250425193258.38994-1-dsosnowski@nvidia.com \
--to=dsosnowski@nvidia.com \
--cc=bingz@nvidia.com \
--cc=dev@dpdk.org \
--cc=matan@nvidia.com \
--cc=orika@nvidia.com \
--cc=stable@dpdk.org \
--cc=suanmingm@nvidia.com \
--cc=viacheslavo@nvidia.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).