patches for DPDK stable branches
 help / color / mirror / Atom feed
From: luca.boccassi@gmail.com
To: Dariusz Sosnowski <dsosnowski@nvidia.com>
Cc: Bing Zhao <bingz@nvidia.com>,
	Raslan Darawsheh <rasland@nvidia.com>,
	dpdk stable <stable@dpdk.org>
Subject: patch 'net/mlx5: fix indirect flow age action handling' has been queued to stable release 22.11.11
Date: Mon, 27 Oct 2025 16:18:59 +0000	[thread overview]
Message-ID: <20251027162001.3710450-21-luca.boccassi@gmail.com> (raw)
In-Reply-To: <20251027162001.3710450-1-luca.boccassi@gmail.com>

Hi,

FYI, your patch has been queued to stable release 22.11.11

Note it hasn't been pushed to http://dpdk.org/browse/dpdk-stable yet.
It will be pushed if I get no objections before 10/29/25. So please
shout if anyone has objections.

Also note that after the patch there's a diff of the upstream commit vs the
patch applied to the branch. This will indicate if there was any rebasing
needed to apply to the stable branch. If there were code changes for rebasing
(ie: not only metadata diffs), please double check that the rebase was
correctly done.

Queued patches are on a temporary branch at:
https://github.com/bluca/dpdk-stable

This queued commit can be viewed at:
https://github.com/bluca/dpdk-stable/commit/4d16e6e800843b5c4e5134d15a2a856dca6d56df

Thanks.

Luca Boccassi

---
From 4d16e6e800843b5c4e5134d15a2a856dca6d56df Mon Sep 17 00:00:00 2001
From: Dariusz Sosnowski <dsosnowski@nvidia.com>
Date: Tue, 19 Aug 2025 13:27:42 +0200
Subject: [PATCH] net/mlx5: fix indirect flow age action handling

[ upstream commit 8bc72d9f277593f6d8b27278bad3a5bb92e7347f ]

Indirect AGE flow actions can be created either through synchronous
or asynchronous flow API.
mlx5 PMD stores the queue used to create that action to support
strict queueing. When action is created through synchronous API
invalid queue index is stored instead.

Whenever a flow rule is created with indirect AGE and
direct COUNT flow actions, PMD allocates a HW counter for ageing
that flow rule during rule creation.
During allocation of the counter a queue index is needed
to select a proper counter pool cache.

In case when indirect AGE action created through synchronous API
was used in that case, the associated queue index was used
to select pool cache. Since queue index was invalid, PMD crashed.

Counter can be allocated using the index of currently used queue and
it does not have to match the queue used to create AGE action.

This patch fixes the crash by using the index of currently used queue
for counter allocation.

This patch also adds missing validation for synchronous
and asynchronous AGE flow action creation:

- If strict queueing is disabled, only synchronous creation is allowed.
- If strict queueing is enabled, only asynchronous creation is allowed.

PMD documentation is updated accordingly.

It also updates validation of synchronous query
of aged flow rules in regards to strict queueing.
When strict queueing is enabled, synchronous query is rejected.
This aligns PMD behavior with API description.

Fixes: 04a4de756e14 ("net/mlx5: support flow age action with HWS")

Signed-off-by: Dariusz Sosnowski <dsosnowski@nvidia.com>
Acked-by: Bing Zhao <bingz@nvidia.com>
Acked-by: Raslan Darawsheh <rasland@nvidia.com>
---
 doc/guides/nics/mlx5.rst        |  4 ++++
 drivers/net/mlx5/mlx5_flow_hw.c | 24 +++++++++++++++++-------
 2 files changed, 21 insertions(+), 7 deletions(-)

diff --git a/doc/guides/nics/mlx5.rst b/doc/guides/nics/mlx5.rst
index d2f741a472..e74c1d8e8d 100644
--- a/doc/guides/nics/mlx5.rst
+++ b/doc/guides/nics/mlx5.rst
@@ -1700,6 +1700,10 @@ applications are allowed to:
 
    This option is supported only for Tx hairpin queues.
 
+#. With strict queueing enabled
+   (``RTE_FLOW_PORT_FLAG_STRICT_QUEUE`` passed to ``rte_flow_configure()``),
+   indirect age actions can be created only through asynchronous flow API.
+
 
 Notes for testpmd
 -----------------
diff --git a/drivers/net/mlx5/mlx5_flow_hw.c b/drivers/net/mlx5/mlx5_flow_hw.c
index 7a8455baf6..5fbbd487de 100644
--- a/drivers/net/mlx5/mlx5_flow_hw.c
+++ b/drivers/net/mlx5/mlx5_flow_hw.c
@@ -1886,6 +1886,7 @@ flow_hw_shared_action_construct(struct rte_eth_dev *dev, uint32_t queue,
 	uint32_t idx = act_idx &
 		       ((1u << MLX5_INDIRECT_ACTION_TYPE_OFFSET) - 1);
 	uint64_t item_flags;
+	uint32_t *cnt_queue;
 	cnt_id_t age_cnt;
 
 	memset(&act_data, 0, sizeof(act_data));
@@ -1931,9 +1932,8 @@ flow_hw_shared_action_construct(struct rte_eth_dev *dev, uint32_t queue,
 		if (param == NULL)
 			return -1;
 		if (action_flags & MLX5_FLOW_ACTION_COUNT) {
-			if (mlx5_hws_cnt_pool_get(priv->hws_cpool,
-						  &param->queue_id, &age_cnt,
-						  idx) < 0)
+			cnt_queue = mlx5_hws_cnt_get_queue(priv, &queue);
+			if (mlx5_hws_cnt_pool_get(priv->hws_cpool, cnt_queue, &age_cnt, idx) < 0)
 				return -1;
 			flow->cnt_id = age_cnt;
 			param->nb_cnts++;
@@ -8469,6 +8469,14 @@ flow_hw_action_create(struct rte_eth_dev *dev,
 		       const struct rte_flow_action *action,
 		       struct rte_flow_error *err)
 {
+	struct mlx5_priv *priv = dev->data->dev_private;
+
+	if (action->type == RTE_FLOW_ACTION_TYPE_AGE && priv->hws_strict_queue) {
+		rte_flow_error_set(err, EINVAL, RTE_FLOW_ERROR_TYPE_STATE, NULL,
+				   "Cannot create age action synchronously with strict queueing");
+		return NULL;
+	}
+
 	return flow_hw_action_handle_create(dev, MLX5_HW_INV_QUEUE,
 					    NULL, conf, action, NULL, err);
 }
@@ -8640,6 +8648,8 @@ flow_hw_get_q_aged_flows(struct rte_eth_dev *dev, uint32_t queue_id,
 					  RTE_FLOW_ERROR_TYPE_UNSPECIFIED,
 					  NULL, "No aging initialized");
 	if (priv->hws_strict_queue) {
+		/* Queue is invalid in sync query. Sync query and strict queueing is disallowed. */
+		MLX5_ASSERT(queue_id != MLX5_HW_INV_QUEUE);
 		if (queue_id >= age_info->hw_q_age->nb_rings)
 			return rte_flow_error_set(error, EINVAL,
 						RTE_FLOW_ERROR_TYPE_UNSPECIFIED,
@@ -8693,10 +8703,10 @@ flow_hw_get_aged_flows(struct rte_eth_dev *dev, void **contexts,
 	struct mlx5_priv *priv = dev->data->dev_private;
 
 	if (priv->hws_strict_queue)
-		DRV_LOG(WARNING,
-			"port %u get aged flows called in strict queue mode.",
-			dev->data->port_id);
-	return flow_hw_get_q_aged_flows(dev, 0, contexts, nb_contexts, error);
+		return rte_flow_error_set(error, EINVAL, RTE_FLOW_ERROR_TYPE_STATE, NULL,
+				"Cannot get aged flows synchronously with strict queueing");
+
+	return flow_hw_get_q_aged_flows(dev, MLX5_HW_INV_QUEUE, contexts, nb_contexts, error);
 }
 
 const struct mlx5_flow_driver_ops mlx5_flow_hw_drv_ops = {
-- 
2.47.3

---
  Diff of the applied patch vs upstream commit (please double-check if non-empty:
---
--- -	2025-10-27 15:54:35.632116018 +0000
+++ 0021-net-mlx5-fix-indirect-flow-age-action-handling.patch	2025-10-27 15:54:34.771948947 +0000
@@ -1 +1 @@
-From 8bc72d9f277593f6d8b27278bad3a5bb92e7347f Mon Sep 17 00:00:00 2001
+From 4d16e6e800843b5c4e5134d15a2a856dca6d56df Mon Sep 17 00:00:00 2001
@@ -5,0 +6,2 @@
+[ upstream commit 8bc72d9f277593f6d8b27278bad3a5bb92e7347f ]
+
@@ -42 +43,0 @@
-Cc: stable@dpdk.org
@@ -53 +54 @@
-index 761dcad542..20056f61d6 100644
+index d2f741a472..e74c1d8e8d 100644
@@ -56,3 +57,3 @@
-@@ -2740,6 +2740,10 @@ With :ref:`HW steering <mlx5_hws>`,
-      in addition to flow rules using only age (without count action).
-    - ``nb_aging_objects`` is the number of flow rules containing age action.
+@@ -1700,6 +1700,10 @@ applications are allowed to:
+ 
+    This option is supported only for Tx hairpin queues.
@@ -65,2 +66,2 @@
- .. _mlx5_quota:
- 
+ Notes for testpmd
+ -----------------
@@ -68 +69 @@
-index 84f39c467e..c84ae726a7 100644
+index 7a8455baf6..5fbbd487de 100644
@@ -71,2 +72 @@
-@@ -3180,6 +3180,7 @@ flow_hw_shared_action_construct(struct rte_eth_dev *dev, uint32_t queue,
- 	uint32_t type = act_idx >> MLX5_INDIRECT_ACTION_TYPE_OFFSET;
+@@ -1886,6 +1886,7 @@ flow_hw_shared_action_construct(struct rte_eth_dev *dev, uint32_t queue,
@@ -74,0 +75 @@
+ 	uint64_t item_flags;
@@ -79 +80 @@
-@@ -3230,9 +3231,8 @@ flow_hw_shared_action_construct(struct rte_eth_dev *dev, uint32_t queue,
+@@ -1931,9 +1932,8 @@ flow_hw_shared_action_construct(struct rte_eth_dev *dev, uint32_t queue,
@@ -89 +89,0 @@
- 			flow->flags |= MLX5_FLOW_HW_FLOW_FLAG_CNT_ID;
@@ -91 +91,2 @@
-@@ -13142,6 +13142,14 @@ flow_hw_action_create(struct rte_eth_dev *dev,
+ 			param->nb_cnts++;
+@@ -8469,6 +8469,14 @@ flow_hw_action_create(struct rte_eth_dev *dev,
@@ -106 +107 @@
-@@ -13361,6 +13369,8 @@ flow_hw_get_q_aged_flows(struct rte_eth_dev *dev, uint32_t queue_id,
+@@ -8640,6 +8648,8 @@ flow_hw_get_q_aged_flows(struct rte_eth_dev *dev, uint32_t queue_id,
@@ -115 +116 @@
-@@ -13414,10 +13424,10 @@ flow_hw_get_aged_flows(struct rte_eth_dev *dev, void **contexts,
+@@ -8693,10 +8703,10 @@ flow_hw_get_aged_flows(struct rte_eth_dev *dev, void **contexts,
@@ -128,2 +129,2 @@
- /**
-  * Initialization function for non template API which calls
+ 
+ const struct mlx5_flow_driver_ops mlx5_flow_hw_drv_ops = {

  parent reply	other threads:[~2025-10-27 16:21 UTC|newest]

Thread overview: 79+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-10-27 16:18 patch 'net/gve: allocate Rx QPL pages using malloc' " luca.boccassi
2025-10-27 16:18 ` patch 'eal: fix plugin dir walk' " luca.boccassi
2025-10-27 16:18 ` patch 'cmdline: fix port list parsing' " luca.boccassi
2025-10-27 16:18 ` patch 'cmdline: fix highest bit " luca.boccassi
2025-10-27 16:18 ` patch 'tailq: fix lookup macro' " luca.boccassi
2025-10-27 16:18 ` patch 'hash: fix unaligned access in predictable RSS' " luca.boccassi
2025-10-27 16:18 ` patch 'graph: fix unaligned access in stats' " luca.boccassi
2025-10-27 16:18 ` patch 'eventdev: fix listing timer adapters with telemetry' " luca.boccassi
2025-10-27 16:18 ` patch 'cfgfile: fix section count with no name' " luca.boccassi
2025-10-27 16:18 ` patch 'net/vmxnet3: fix mapping of mempools to queues' " luca.boccassi
2025-10-27 16:18 ` patch 'app/testpmd: increase size of set cores list command' " luca.boccassi
2025-10-27 16:18 ` patch 'net/dpaa2: fix shaper rate' " luca.boccassi
2025-10-27 16:18 ` patch 'app/testpmd: monitor state of primary process' " luca.boccassi
2025-10-27 16:18 ` patch 'app/testpmd: fix conntrack action query' " luca.boccassi
2025-10-27 16:18 ` patch 'doc: add conntrack state inspect command to testpmd guide' " luca.boccassi
2025-10-27 16:18 ` patch 'app/testpmd: validate DSCP and VLAN for meter creation' " luca.boccassi
2025-10-27 16:18 ` patch 'net/mlx5: fix min and max MTU reporting' " luca.boccassi
2025-10-27 16:18 ` patch 'net/mlx5: fix unsupported flow rule port action' " luca.boccassi
2025-10-27 16:18 ` patch 'net/mlx5: fix non-template age rules flush' " luca.boccassi
2025-10-27 16:18 ` patch 'net/mlx5: fix connection tracking state item validation' " luca.boccassi
2025-10-27 16:18 ` luca.boccassi [this message]
2025-10-27 16:19 ` patch 'net/mlx5: fix Direct Verbs counter offset detection' " luca.boccassi
2025-10-27 16:19 ` patch 'net/mlx5: fix interface name parameter definition' " luca.boccassi
2025-10-27 16:19 ` patch 'net/intel: fix assumption about tag placement order' " luca.boccassi
2025-10-27 16:19 ` patch 'net/ice/base: fix adding special words' " luca.boccassi
2025-10-27 16:19 ` patch 'net/ice/base: fix memory leak in HW profile handling' " luca.boccassi
2025-10-27 16:19 ` patch 'net/ice/base: fix memory leak in recipe " luca.boccassi
2025-10-27 16:19 ` patch 'eal: fix DMA mask validation with IOVA mode option' " luca.boccassi
2025-10-27 16:19 ` patch 'eal: fix MP socket cleanup' " luca.boccassi
2025-10-27 16:19 ` patch 'crypto/ipsec_mb: fix QP release in secondary' " luca.boccassi
2025-10-27 16:19 ` patch 'efd: fix AVX2 support' " luca.boccassi
2025-10-27 16:19 ` patch 'common/cnxk: fix async event handling' " luca.boccassi
2025-10-27 16:19 ` patch 'doc: fix feature list of ice driver' " luca.boccassi
2025-10-27 16:19 ` patch 'doc: fix feature list of iavf " luca.boccassi
2025-10-27 16:19 ` patch 'baseband/acc: fix exported header' " luca.boccassi
2025-10-27 16:19 ` patch 'gpudev: fix driver header for Windows' " luca.boccassi
2025-10-27 16:19 ` patch 'drivers: fix some exported headers' " luca.boccassi
2025-10-27 16:19 ` patch 'test/debug: fix crash with mlx5 devices' " luca.boccassi
2025-10-27 16:19 ` patch 'bus/pci: fix build with MinGW 13' " luca.boccassi
2025-10-27 16:19 ` patch 'net/mlx5: " luca.boccassi
2025-10-27 16:19 ` patch 'dma/hisilicon: fix stop with pending transfers' " luca.boccassi
2025-10-27 16:19 ` patch 'test/dma: fix failure condition' " luca.boccassi
2025-10-27 16:19 ` patch 'fib6: fix tbl8 allocation check logic' " luca.boccassi
2025-10-27 16:19 ` patch 'vhost: fix double fetch when dequeue offloading' " luca.boccassi
2025-10-27 16:19 ` patch 'net/ice/base: fix integer overflow on NVM init' " luca.boccassi
2025-10-27 16:19 ` patch 'net/ice: fix initialization with 8 ports' " luca.boccassi
2025-10-27 16:19 ` patch 'net/ice: remove indirection for FDIR filters' " luca.boccassi
2025-10-27 16:19 ` patch 'net/ice: fix memory leak in raw pattern parse' " luca.boccassi
2025-10-27 16:19 ` patch 'net/i40e: fix symmetric Toeplitz hashing for SCTP' " luca.boccassi
2025-10-27 16:19 ` patch 'net/mlx5: fix multicast' " luca.boccassi
2025-10-27 16:19 ` patch 'net/mlx5: fix MTU initialization' " luca.boccassi
2025-10-27 16:19 ` patch 'net/mlx5: fix leak of flow indexed pools' " luca.boccassi
2025-10-27 16:19 ` patch 'net/hns3: fix inconsistent lock' " luca.boccassi
2025-10-27 16:19 ` patch 'net/hns3: fix VLAN resources freeing' " luca.boccassi
2025-10-27 16:19 ` patch 'net/af_packet: fix crash in secondary process' " luca.boccassi
2025-10-27 16:19 ` patch 'net/ark: remove double mbuf free' " luca.boccassi
2025-10-27 16:19 ` patch 'net/hns3: fix VLAN tag loss for short tunnel frame' " luca.boccassi
2025-10-27 16:19 ` patch 'ethdev: fix VLAN filter parameter description' " luca.boccassi
2025-10-27 16:19 ` patch 'net/enetfec: fix file descriptor leak on read error' " luca.boccassi
2025-10-27 16:19 ` patch 'net/enetfec: fix out-of-bounds access in UIO mapping' " luca.boccassi
2025-10-27 16:19 ` patch 'net/enetfec: fix buffer descriptor size configuration' " luca.boccassi
2025-10-27 16:19 ` patch 'net/enetfec: fix Tx queue free' " luca.boccassi
2025-10-27 16:19 ` patch 'net/enetfec: fix checksum flag handling and error return' " luca.boccassi
2025-10-27 16:19 ` patch 'net/enetfec: reject multi-queue configuration' " luca.boccassi
2025-10-27 16:19 ` patch 'net/enetfec: fix memory leak in Rx buffer cleanup' " luca.boccassi
2025-10-27 16:19 ` patch 'net/enetfec: reject Tx deferred queue' " luca.boccassi
2025-10-27 16:19 ` patch 'net/tap: fix interrupt callback crash after failed start' " luca.boccassi
2025-10-27 16:19 ` patch 'net/ena: fix PCI BAR mapping on 64K page size' " luca.boccassi
2025-10-27 16:19 ` patch 'net/ena/base: fix unsafe memcpy on invalid memory' " luca.boccassi
2025-10-27 16:19 ` patch 'net/dpaa2: fix uninitialized variable' " luca.boccassi
2025-10-27 16:19 ` patch 'net/dpaa2: fix L3/L4 checksum results' " luca.boccassi
2025-10-27 16:19 ` patch 'net/dpaa2: receive packets with additional parse errors' " luca.boccassi
2025-10-27 16:19 ` patch 'crypto/qat: fix source buffer alignment' " luca.boccassi
2025-10-27 16:19 ` patch 'crypto/cnxk: refactor RSA verification' " luca.boccassi
2025-10-27 16:19 ` patch 'test/crypto: fix mbuf handling' " luca.boccassi
2025-10-27 16:19 ` patch 'app/crypto-perf: fix plaintext size exceeds buffer size' " luca.boccassi
2025-10-27 16:19 ` patch 'test/crypto: fix vector initialization' " luca.boccassi
2025-10-27 16:19 ` patch 'crypto/virtio: fix cookies leak' " luca.boccassi
2025-10-27 16:19 ` patch 'sched: fix WRR parameter data type' " luca.boccassi

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=20251027162001.3710450-21-luca.boccassi@gmail.com \
    --to=luca.boccassi@gmail.com \
    --cc=bingz@nvidia.com \
    --cc=dsosnowski@nvidia.com \
    --cc=rasland@nvidia.com \
    --cc=stable@dpdk.org \
    /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).