Skip to contentSkip to conversation

net/i40e: fix flex payload rule conflict issue

PATCH Superseded [email protected] 3 participants 10 messages Dec 28, 2020 – Jan 5, 2021 Network drivers

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 net/i40e: fix flex payload rule conflict issue

From: Beilei Xing <[email protected]>

With the following commands, the second flow can't be created successfully.

1. flow create 0 ingress pattern eth / ipv4 / udp / raw relative is 1 pattern is 0102030405 / end actions drop / end
2. flow destroy 0 rule 0
3. flow create 0 ingress pattern eth / ipv4 / udp / raw relative is 1 pattern is 010203040506 / end actions drop / end

The root cause is that a flag for flex pit isn't reset.

---
 drivers/net/i40e/i40e_flow.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
drivers/net/i40e/i40e_flow.c +6 -1
index b09ff6590d..3a68274a23 100644
@@ -5284,6 +5284,7 @@ i40e_flow_destroy(struct rte_eth_dev *dev,
 	enum rte_filter_type filter_type = flow->filter_type;
 	struct i40e_fdir_info *fdir_info = &pf->fdir;
 	int ret = 0;
+	int i;
 
 	switch (filter_type) {
 	case RTE_ETH_FILTER_ETHERTYPE:
@@ -5299,9 +5300,10 @@ i40e_flow_destroy(struct rte_eth_dev *dev,
 				&((struct i40e_fdir_filter *)flow->rule)->fdir,
 				0);
 
-		/* If the last flow is destroyed, disable fdir. */
 		if (!ret && TAILQ_EMPTY(&pf->fdir.fdir_list)) {
 			i40e_fdir_rx_proc_enable(dev, 0);
+			for (i = 0; i < I40E_MAX_FLXPLD_LAYER; i++)
+				pf->fdir.flex_pit_flag[i] = 0;
 		}
 		break;
 	case RTE_ETH_FILTER_HASH:
@@ -5515,6 +5517,9 @@ i40e_flow_flush_fdir_filter(struct i40e_pf *pf)
 			pf->fdir.flex_mask_flag[pctype] = 0;
 		}
 
+		for (i = 0; i < I40E_MAX_FLXPLD_LAYER; i++)
+			pf->fdir.flex_pit_flag[i] = 0;
+
 		/* Disable FDIR processing as all FDIR rules are now flushed */
 		i40e_fdir_rx_proc_enable(dev, 0);
 	}
Signature · 1 line
2.26.2
#1 lore raw

PATCH v2 1/1 net/i40e: fix flex payload rule conflict issue

From: Beilei Xing <[email protected]>

With the following commands, the second flow can't be created successfully.

1. flow create 0 ingress pattern eth / ipv4 / udp / raw relative is 1 pattern is 0102030405 / end actions drop / end
2. flow destroy 0 rule 0
3. flow create 0 ingress pattern eth / ipv4 / udp / raw relative is 1 pattern is 010203040506 / end actions drop / end

The root cause is that a flag for flex pit isn't reset.

Fixes 6ced3dd72f5f ("net/i40e: support flexible payload parsing for FDIR")

---

v2 changes:
- Add fix line.
- Refine comments.

 drivers/net/i40e/i40e_flow.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)
drivers/net/i40e/i40e_flow.c +9 -1
index b09ff6590d..65e0b69356 100644
@@ -5284,6 +5284,7 @@ i40e_flow_destroy(struct rte_eth_dev *dev,
 	enum rte_filter_type filter_type = flow->filter_type;
 	struct i40e_fdir_info *fdir_info = &pf->fdir;
 	int ret = 0;
+	int i;
 
 	switch (filter_type) {
 	case RTE_ETH_FILTER_ETHERTYPE:
@@ -5299,9 +5300,13 @@ i40e_flow_destroy(struct rte_eth_dev *dev,
 				&((struct i40e_fdir_filter *)flow->rule)->fdir,
 				0);
 
-		/* If the last flow is destroyed, disable fdir. */
+		/* When the last flow is destroyed. */
 		if (!ret && TAILQ_EMPTY(&pf->fdir.fdir_list)) {
+			/* Disable FDIR processing. */
 			i40e_fdir_rx_proc_enable(dev, 0);
+			/* Reset the flex_pit_flag. */
+			for (i = 0; i < I40E_MAX_FLXPLD_LAYER; i++)
+				pf->fdir.flex_pit_flag[i] = 0;
 		}
 		break;
 	case RTE_ETH_FILTER_HASH:
@@ -5515,6 +5520,9 @@ i40e_flow_flush_fdir_filter(struct i40e_pf *pf)
 			pf->fdir.flex_mask_flag[pctype] = 0;
 		}
 
+		for (i = 0; i < I40E_MAX_FLXPLD_LAYER; i++)
+			pf->fdir.flex_pit_flag[i] = 0;
+
 		/* Disable FDIR processing as all FDIR rules are now flushed */
 		i40e_fdir_rx_proc_enable(dev, 0);
 	}
Signature · 1 line
2.26.2
#2 lore raw

Hi, beilei

Quoted 50 lines

-----Original Message-----
From: Xing, Beilei <[email protected]>
Sent: Tuesday, December 29, 2020 2:18 PM
To: Guo, Jia <[email protected]>; [email protected]
Cc: Xing, Beilei <[email protected]>; [email protected]; Sun, Chenmin <[email protected]>
Subject: [PATCH v2] net/i40e: fix flex payload rule conflict issue

From: Beilei Xing <[email protected]>

With the following commands, the second flow can't be created successfully.

1. flow create 0 ingress pattern eth / ipv4 / udp / raw relative is 1 pattern is 0102030405 / end actions drop / end
2. flow destroy 0 rule 0
3. flow create 0 ingress pattern eth / ipv4 / udp / raw relative is 1 pattern is 010203040506 / end actions drop / end

The root cause is that a flag for flex pit isn't reset.

Fixes 6ced3dd72f5f ("net/i40e: support flexible payload parsing for FDIR")

---

v2 changes:
- Add fix line.
- Refine comments.

 drivers/net/i40e/i40e_flow.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)
drivers/net/i40e/i40e_flow.c +1 -0
index b09ff6590d..65e0b69356 100644
@@ -5284,6 +5284,7 @@ i40e_flow_destroy(struct rte_eth_dev *dev,
 	enum rte_filter_type filter_type = flow->filter_type;
 	struct i40e_fdir_info *fdir_info = &pf->fdir;
 	int ret = 0;
+	int i;
 
 	switch (filter_type) {
 	case RTE_ETH_FILTER_ETHERTYPE:
@@ -5299,9 +5300,13 @@ i40e_flow_destroy(struct rte_eth_dev *dev,
 				&((struct i40e_fdir_filter *)flow->rule)->fdir,

drivers/net/i40e/i40e_flow.c i40e_flow_destroy(struct rte_eth_dev *dev, · line 5309

Quoted 10 lines
drivers/net/i40e/i40e_flow.c +5 -1
 				0);
 
-		/* If the last flow is destroyed, disable fdir. */
+		/* When the last flow is destroyed. */
 		if (!ret && TAILQ_EMPTY(&pf->fdir.fdir_list)) {
+			/* Disable FDIR processing. */
 			i40e_fdir_rx_proc_enable(dev, 0);
+			/* Reset the flex_pit_flag. */
+			for (i = 0; i < I40E_MAX_FLXPLD_LAYER; i++)
+				pf->fdir.flex_pit_flag[i] = 0;

Why reset all flex_pit_flag when destroy the last flow, if destroy other flow, is it no need to reset corresponding flex_pit_flag which set before when the flow added? And reset flag should be ahead of the FDIR disabling I think.

Quoted 16 lines
drivers/net/i40e/i40e_flow.c +3 -0
 		}
 		break;
 	case RTE_ETH_FILTER_HASH:
@@ -5515,6 +5520,9 @@ i40e_flow_flush_fdir_filter(struct i40e_pf *pf)
 			pf->fdir.flex_mask_flag[pctype] = 0;
 		}
 
+		for (i = 0; i < I40E_MAX_FLXPLD_LAYER; i++)
+			pf->fdir.flex_pit_flag[i] = 0;
+
 		/* Disable FDIR processing as all FDIR rules are now flushed
 */
 		i40e_fdir_rx_proc_enable(dev, 0);

}
--
2.26.2

#3 lore raw

PATCH v3 1/1 net/i40e: fix flex payload rule conflict issue

In reply to [email protected]

From: Beilei Xing <[email protected]>

With the following commands, the second flow can't be created successfully.

1. flow create 0 ingress pattern eth / ipv4 / udp / raw relative is 1 pattern is 0102030405 / end actions drop / end
2. flow destroy 0 rule 0
3. flow create 0 ingress pattern eth / ipv4 / udp / raw relative is 1 pattern is 010203040506 / end actions drop / end

The root cause is that a flag for flex pit isn't reset.

Fixes 6ced3dd72f5f ("net/i40e: support flexible payload parsing for FDIR")

---

v3 changes:
- Add flow count for flexible payload flow.
v2 changeds:
- Add fix line.

 drivers/net/i40e/i40e_ethdev.h |  3 +++
 drivers/net/i40e/i40e_fdir.c   | 16 ++++++++++++++--
 drivers/net/i40e/i40e_flow.c   |  4 ++++
 3 files changed, 21 insertions(+), 2 deletions(-)
drivers/net/i40e/i40e_ethdev.h +3 -0
index 696c5aaf7e..aac226999c 100644
@@ -636,6 +636,7 @@ struct i40e_fdir_flow_ext {
 	bool is_udp; /* ipv4|ipv6 udp flow */
 	enum i40e_flxpld_layer_idx layer_idx;
 	struct i40e_fdir_flex_pit flex_pit[I40E_MAX_FLXPLD_LAYER * I40E_MAX_FLXPLD_FIED];
+	bool is_flex_flow;
 };
 
 /* A structure used to define the input for a flow director filter entry */
@@ -784,6 +785,8 @@ struct i40e_fdir_info {
 	bool flex_mask_flag[I40E_FILTER_PCTYPE_MAX];
 
 	bool inset_flag[I40E_FILTER_PCTYPE_MAX]; /* Mark if input set is set */
+
+	uint32_t flex_flow_count[I40E_MAX_FLXPLD_LAYER];
 };
 
 /* Ethertype filter number HW supports */
drivers/net/i40e/i40e_fdir.c +14 -2
index 50c0eee9f2..4400b607c8 100644
@@ -355,6 +355,7 @@ i40e_init_flx_pld(struct i40e_pf *pf)
 			I40E_PRTQF_FLX_PIT(index + 1), 0x0000FC29);/*non-used*/
 		I40E_WRITE_REG(hw,
 			I40E_PRTQF_FLX_PIT(index + 2), 0x0000FC2A);/*non-used*/
+		pf->fdir.flex_pit_flag[i] = 0;
 	}
 
 	/* initialize the masks */
@@ -1513,8 +1514,6 @@ i40e_flow_set_fdir_flex_pit(struct i40e_pf *pf,
 		I40E_WRITE_REG(hw, I40E_PRTQF_FLX_PIT(field_idx), flx_pit);
 		min_next_off++;
 	}
-
-	pf->fdir.flex_pit_flag[layer_idx] = 1;
 }
 
 static int
@@ -1738,6 +1737,8 @@ i40e_flow_add_del_fdir_filter(struct rte_eth_dev *dev,
 				fdir_info->fdir_guarantee_free_space > 0)
 			wait_status = false;
 	} else {
+		layer_idx = filter->input.flow_ext.layer_idx;
+
 		node = i40e_sw_fdir_filter_lookup(fdir_info,
 				&check_filter.fdir.input);
 		if (!node) {
@@ -1785,6 +1786,17 @@ i40e_flow_add_del_fdir_filter(struct rte_eth_dev *dev,
 		goto error_op;
 	}
 
+	if (filter->input.flow_ext.is_flex_flow) {
+		if (add) {
+			fdir_info->flex_flow_count[layer_idx]++;
+			pf->fdir.flex_pit_flag[layer_idx] = 1;
+		} else {
+			fdir_info->flex_flow_count[layer_idx]--;
+			if (!fdir_info->flex_flow_count[layer_idx])
+				pf->fdir.flex_pit_flag[layer_idx] = 0;
+		}
+	}
+
 	if (add) {
 		fdir_info->fdir_actual_cnt++;
 		if (fdir_info->fdir_invalprio == 1 &&
drivers/net/i40e/i40e_flow.c +4 -0
index b09ff6590d..bbd666b7a0 100644
@@ -3069,6 +3069,7 @@ i40e_flow_parse_fdir_pattern(struct rte_eth_dev *dev,
 			       &flex_pit, sizeof(struct i40e_fdir_flex_pit));
 			filter->input.flow_ext.layer_idx = layer_idx;
 			filter->input.flow_ext.raw_id = raw_id;
+			filter->input.flow_ext.is_flex_flow = true;
 			break;
 		case RTE_FLOW_ITEM_TYPE_VF:
 			vf_spec = item->spec;
@@ -5515,6 +5516,9 @@ i40e_flow_flush_fdir_filter(struct i40e_pf *pf)
 			pf->fdir.flex_mask_flag[pctype] = 0;
 		}
 
+		for (i = 0; i < I40E_MAX_FLXPLD_LAYER; i++)
+			pf->fdir.flex_pit_flag[i] = 0;
+
 		/* Disable FDIR processing as all FDIR rules are now flushed */
 		i40e_fdir_rx_proc_enable(dev, 0);
 	}
Signature · 1 line
2.26.2
#4 lore raw
Quoted 88 lines

-----Original Message-----
From: Xing, Beilei <[email protected]>
Sent: Friday, January 1, 2021 12:04 PM
To: Guo, Jia <[email protected]>; [email protected]
Cc: Xing, Beilei <[email protected]>; [email protected]; Sun, Chenmin <[email protected]>
Subject: [PATCH v3] net/i40e: fix flex payload rule conflict issue

From: Beilei Xing <[email protected]>

With the following commands, the second flow can't be created successfully.

1. flow create 0 ingress pattern eth / ipv4 / udp / raw relative is 1 pattern is 0102030405 / end actions drop / end
2. flow destroy 0 rule 0
3. flow create 0 ingress pattern eth / ipv4 / udp / raw relative is 1 pattern is 010203040506 / end actions drop / end

The root cause is that a flag for flex pit isn't reset.

Fixes 6ced3dd72f5f ("net/i40e: support flexible payload parsing for FDIR")

---

v3 changes:
- Add flow count for flexible payload flow.
v2 changeds:
- Add fix line.

 drivers/net/i40e/i40e_ethdev.h |  3 +++
 drivers/net/i40e/i40e_fdir.c   | 16 ++++++++++++++--
 drivers/net/i40e/i40e_flow.c   |  4 ++++
 3 files changed, 21 insertions(+), 2 deletions(-)
(quoted diff) +0 -0

b/drivers/net/i40e/i40e_ethdev.h index 696c5aaf7e..aac226999c 100644

drivers/net/i40e/i40e_ethdev.h +1 -0
@@ -636,6 +636,7 @@ struct i40e_fdir_flow_ext {
 	bool is_udp; /* ipv4|ipv6 udp flow */
 	enum i40e_flxpld_layer_idx layer_idx;
 	struct i40e_fdir_flex_pit flex_pit[I40E_MAX_FLXPLD_LAYER *
 I40E_MAX_FLXPLD_FIED];
+	bool is_flex_flow;
 };
 

/* A structure used to define the input for a flow director filter entry */ @@
-784,6 +785,8 @@ struct i40e_fdir_info {

drivers/net/i40e/i40e_ethdev.h +0 -0
 	bool flex_mask_flag[I40E_FILTER_PCTYPE_MAX];
 
 	bool inset_flag[I40E_FILTER_PCTYPE_MAX]; /* Mark if input set is set

*/

drivers/net/i40e/i40e_ethdev.h +2 -0
+
+	uint32_t flex_flow_count[I40E_MAX_FLXPLD_LAYER];
 };
 
 /* Ethertype filter number HW supports */ diff --git

a/drivers/net/i40e/i40e_fdir.c b/drivers/net/i40e/i40e_fdir.c index 50c0eee9f2..4400b607c8 100644

drivers/net/i40e/i40e_fdir.c +0 -0
@@ -355,6 +355,7 @@ i40e_init_flx_pld(struct i40e_pf *pf)
 			I40E_PRTQF_FLX_PIT(index + 1), 0x0000FC29);/*non-

used*/

drivers/net/i40e/i40e_fdir.c +0 -0
 		I40E_WRITE_REG(hw,
 			I40E_PRTQF_FLX_PIT(index + 2), 0x0000FC2A);/*non-

used*/

drivers/net/i40e/i40e_fdir.c +1 -2
+		pf->fdir.flex_pit_flag[i] = 0;
 	}
 
 	/* initialize the masks */
@@ -1513,8 +1514,6 @@ i40e_flow_set_fdir_flex_pit(struct i40e_pf *pf,
 		I40E_WRITE_REG(hw, I40E_PRTQF_FLX_PIT(field_idx),
 flx_pit);
 		min_next_off++;
 	}
-
-	pf->fdir.flex_pit_flag[layer_idx] = 1;
 }
 

static int

drivers/net/i40e/i40e_fdir.c i40e_flow_add_del_fdir_filter(struct rte_eth_dev · line 1742

Quoted 6 lines
drivers/net/i40e/i40e_fdir.c +2 -0
@@ -1738,6 +1737,8 @@ i40e_flow_add_del_fdir_filter(struct rte_eth_dev
 *dev,
 				fdir_info->fdir_guarantee_free_space > 0)
 			wait_status = false;
 	} else {
+		layer_idx = filter->input.flow_ext.layer_idx;
+

Assume that flow_ext.layer_idx should be assign to the layer_idx at the condition of " (!filter->input.flow_ext.customized_pctype) ", right?

Quoted 48 lines
drivers/net/i40e/i40e_fdir.c +0 -0
 		node = i40e_sw_fdir_filter_lookup(fdir_info,
 				&check_filter.fdir.input);
drivers/net/i40e/i40e_fdir.c +11 -0
 		if (!node) {
@@ -1785,6 +1786,17 @@ i40e_flow_add_del_fdir_filter(struct rte_eth_dev
 *dev,
 		goto error_op;
 	}
 
+	if (filter->input.flow_ext.is_flex_flow) {
+		if (add) {
+			fdir_info->flex_flow_count[layer_idx]++;
+			pf->fdir.flex_pit_flag[layer_idx] = 1;
+		} else {
+			fdir_info->flex_flow_count[layer_idx]--;
+			if (!fdir_info->flex_flow_count[layer_idx])
+				pf->fdir.flex_pit_flag[layer_idx] = 0;
+		}
+	}
+
 	if (add) {
 		fdir_info->fdir_actual_cnt++;

if (fdir_info->fdir_invalprio == 1 && diff --git
a/drivers/net/i40e/i40e_flow.c b/drivers/net/i40e/i40e_flow.c index b09ff6590d..bbd666b7a0 100644

drivers/net/i40e/i40e_flow.c +1 -0
@@ -3069,6 +3069,7 @@ i40e_flow_parse_fdir_pattern(struct rte_eth_dev
 *dev,
 			       &flex_pit, sizeof(struct i40e_fdir_flex_pit));
 			filter->input.flow_ext.layer_idx = layer_idx;
 			filter->input.flow_ext.raw_id = raw_id;
+			filter->input.flow_ext.is_flex_flow = true;
 			break;
 		case RTE_FLOW_ITEM_TYPE_VF:
drivers/net/i40e/i40e_flow.c +3 -0
 			vf_spec = item->spec;
@@ -5515,6 +5516,9 @@ i40e_flow_flush_fdir_filter(struct i40e_pf *pf)
 			pf->fdir.flex_mask_flag[pctype] = 0;
 		}
 
+		for (i = 0; i < I40E_MAX_FLXPLD_LAYER; i++)
+			pf->fdir.flex_pit_flag[i] = 0;
+
 		/* Disable FDIR processing as all FDIR rules are now flushed
 */
 		i40e_fdir_rx_proc_enable(dev, 0);

}
--
2.26.2

#5 lore raw
Quoted 92 lines

-----Original Message-----
From: Guo, Jia <[email protected]>
Sent: Thursday, December 31, 2020 2:07 PM
To: Xing, Beilei <[email protected]>; [email protected]
Cc: [email protected]; Sun, Chenmin <[email protected]>
Subject: RE: [PATCH v3] net/i40e: fix flex payload rule conflict issue

Quoted 82 lines

-----Original Message-----
From: Xing, Beilei <[email protected]>
Sent: Friday, January 1, 2021 12:04 PM
To: Guo, Jia <[email protected]>; [email protected]
Cc: Xing, Beilei <[email protected]>; [email protected]; Sun, Chenmin <[email protected]>
Subject: [PATCH v3] net/i40e: fix flex payload rule conflict issue

From: Beilei Xing <[email protected]>

With the following commands, the second flow can't be created successfully.

1. flow create 0 ingress pattern eth / ipv4 / udp / raw relative is 1 pattern is 0102030405 / end actions drop / end
2. flow destroy 0 rule 0
3. flow create 0 ingress pattern eth / ipv4 / udp / raw relative is 1 pattern is 010203040506 / end actions drop / end

The root cause is that a flag for flex pit isn't reset.

Fixes 6ced3dd72f5f ("net/i40e: support flexible payload parsing for

FDIR")
Cc: [email protected]

---

v3 changes:
- Add flow count for flexible payload flow.
v2 changeds:
- Add fix line.

 drivers/net/i40e/i40e_ethdev.h |  3 +++
 drivers/net/i40e/i40e_fdir.c   | 16 ++++++++++++++--
 drivers/net/i40e/i40e_flow.c   |  4 ++++
 3 files changed, 21 insertions(+), 2 deletions(-)
(quoted diff) +0 -0

b/drivers/net/i40e/i40e_ethdev.h index 696c5aaf7e..aac226999c 100644

drivers/net/i40e/i40e_ethdev.h +0 -0
@@ -636,6 +636,7 @@ struct i40e_fdir_flow_ext {  bool is_udp; /*

ipv4|ipv6 udp flow */ enum i40e_flxpld_layer_idx layer_idx; struct i40e_fdir_flex_pit flex_pit[I40E_MAX_FLXPLD_LAYER * I40E_MAX_FLXPLD_FIED];
+bool is_flex_flow;
};

/* A structure used to define the input for a flow director filter
entry */ @@
-784,6 +785,8 @@ struct i40e_fdir_info { bool flex_mask_flag[I40E_FILTER_PCTYPE_MAX];

bool inset_flag[I40E_FILTER_PCTYPE_MAX]; /* Mark if input set is set
*/

drivers/net/i40e/i40e_ethdev.h +2 -0
+
+uint32_t flex_flow_count[I40E_MAX_FLXPLD_LAYER];
 };
 
 /* Ethertype filter number HW supports */ diff --git

a/drivers/net/i40e/i40e_fdir.c b/drivers/net/i40e/i40e_fdir.c index 50c0eee9f2..4400b607c8 100644

drivers/net/i40e/i40e_fdir.c +0 -0
@@ -355,6 +355,7 @@ i40e_init_flx_pld(struct i40e_pf *pf)

I40E_PRTQF_FLX_PIT(index + 1), 0x0000FC29);/*non- used*/ I40E_WRITE_REG(hw, I40E_PRTQF_FLX_PIT(index + 2), 0x0000FC2A);/*non- used*/

drivers/net/i40e/i40e_fdir.c +1 -2
+pf->fdir.flex_pit_flag[i] = 0;
 }
 
 /* initialize the masks */
@@ -1513,8 +1514,6 @@ i40e_flow_set_fdir_flex_pit(struct i40e_pf *pf,
 I40E_WRITE_REG(hw, I40E_PRTQF_FLX_PIT(field_idx), flx_pit);
 min_next_off++;  }
-
-pf->fdir.flex_pit_flag[layer_idx] = 1;
 }
 
Quoted 9 lines

drivers/net/i40e/i40e_fdir.c i40e_flow_add_del_fdir_filter(struct rte_eth_dev · line 1740

Quoted 6 lines
drivers/net/i40e/i40e_fdir.c +2 -0
 static int
@@ -1738,6 +1737,8 @@ i40e_flow_add_del_fdir_filter(struct rte_eth_dev
 *dev,  fdir_info->fdir_guarantee_free_space > 0)  wait_status = false;
 } else {
+layer_idx = filter->input.flow_ext.layer_idx;
+

Assume that flow_ext.layer_idx should be assign to the layer_idx at the condition of " (!filter->input.flow_ext.customized_pctype) ", right?

Yes, thanks for the comment, will fix in next version.

Quoted 44 lines
Quoted 43 lines
drivers/net/i40e/i40e_fdir.c +10 -0
 node = i40e_sw_fdir_filter_lookup(fdir_info,
 &check_filter.fdir.input);
 if (!node) {
@@ -1785,6 +1786,17 @@ i40e_flow_add_del_fdir_filter(struct
 rte_eth_dev *dev,  goto error_op;  }
 
+if (filter->input.flow_ext.is_flex_flow) { if (add) {
+fdir_info->flex_flow_count[layer_idx]++;
+pf->fdir.flex_pit_flag[layer_idx] = 1;
+} else {
+fdir_info->flex_flow_count[layer_idx]--;
+if (!fdir_info->flex_flow_count[layer_idx])
+pf->fdir.flex_pit_flag[layer_idx] = 0;
+}
+}
+
 if (add) {
 fdir_info->fdir_actual_cnt++;
 if (fdir_info->fdir_invalprio == 1 && diff --git

a/drivers/net/i40e/i40e_flow.c b/drivers/net/i40e/i40e_flow.c index b09ff6590d..bbd666b7a0 100644

drivers/net/i40e/i40e_flow.c +1 -0
@@ -3069,6 +3069,7 @@ i40e_flow_parse_fdir_pattern(struct rte_eth_dev
 *dev,
        &flex_pit, sizeof(struct i40e_fdir_flex_pit));
 filter->input.flow_ext.layer_idx = layer_idx;
 filter->input.flow_ext.raw_id = raw_id;
+filter->input.flow_ext.is_flex_flow = true;
 break;
 case RTE_FLOW_ITEM_TYPE_VF:

vf_spec = item->spec;

drivers/net/i40e/i40e_flow.c +3 -1
@@ -5515,6 +5516,9 @@ i40e_flow_flush_fdir_filter(struct i40e_pf *pf)
 pf->fdir.flex_mask_flag[pctype] = 0;  }
 
+for (i = 0; i < I40E_MAX_FLXPLD_LAYER; i++)
+pf->fdir.flex_pit_flag[i] = 0;
+
 /* Disable FDIR processing as all FDIR rules are now flushed */
 i40e_fdir_rx_proc_enable(dev, 0);  }
--

2.26.2

#6 lore raw

PATCH v4 1/1 net/i40e: fix flex payload rule conflict issue

In reply to [email protected]

From: Beilei Xing <[email protected]>

With the following commands, the second flow can't be created successfully.

1. flow create 0 ingress pattern eth / ipv4 / udp / raw relative is 1 pattern is 0102030405 / end actions drop / end
2. flow destroy 0 rule 0
3. flow create 0 ingress pattern eth / ipv4 / udp / raw relative is 1 pattern is 010203040506 / end actions drop / end

The root cause is that a flag for flex pit isn't reset.

Fixes 6ced3dd72f5f ("net/i40e: support flexible payload parsing for FDIR")

---
v4 changes:
- Code refine.
v3 changes:
- Add flow count for flexible payload flow.
v2 changeds:
- Add fix line.

 drivers/net/i40e/i40e_ethdev.h |  3 +++
 drivers/net/i40e/i40e_fdir.c   | 19 ++++++++++++++++---
 drivers/net/i40e/i40e_flow.c   |  4 ++++
 3 files changed, 23 insertions(+), 3 deletions(-)
drivers/net/i40e/i40e_ethdev.h +3 -0
index 696c5aaf7e..aac226999c 100644
@@ -636,6 +636,7 @@ struct i40e_fdir_flow_ext {
 	bool is_udp; /* ipv4|ipv6 udp flow */
 	enum i40e_flxpld_layer_idx layer_idx;
 	struct i40e_fdir_flex_pit flex_pit[I40E_MAX_FLXPLD_LAYER * I40E_MAX_FLXPLD_FIED];
+	bool is_flex_flow;
 };
 
 /* A structure used to define the input for a flow director filter entry */
@@ -784,6 +785,8 @@ struct i40e_fdir_info {
 	bool flex_mask_flag[I40E_FILTER_PCTYPE_MAX];
 
 	bool inset_flag[I40E_FILTER_PCTYPE_MAX]; /* Mark if input set is set */
+
+	uint32_t flex_flow_count[I40E_MAX_FLXPLD_LAYER];
 };
 
 /* Ethertype filter number HW supports */
drivers/net/i40e/i40e_fdir.c +16 -3
index 50c0eee9f2..0343e8b09a 100644
@@ -355,6 +355,7 @@ i40e_init_flx_pld(struct i40e_pf *pf)
 			I40E_PRTQF_FLX_PIT(index + 1), 0x0000FC29);/*non-used*/
 		I40E_WRITE_REG(hw,
 			I40E_PRTQF_FLX_PIT(index + 2), 0x0000FC2A);/*non-used*/
+		pf->fdir.flex_pit_flag[i] = 0;
 	}
 
 	/* initialize the masks */
@@ -1513,8 +1514,6 @@ i40e_flow_set_fdir_flex_pit(struct i40e_pf *pf,
 		I40E_WRITE_REG(hw, I40E_PRTQF_FLX_PIT(field_idx), flx_pit);
 		min_next_off++;
 	}
-
-	pf->fdir.flex_pit_flag[layer_idx] = 1;
 }
 
 static int
@@ -1686,7 +1685,7 @@ i40e_flow_add_del_fdir_filter(struct rte_eth_dev *dev,
 	i40e_fdir_filter_convert(filter, &check_filter);
 
 	if (add) {
-		if (!filter->input.flow_ext.customized_pctype) {
+		if (filter->input.flow_ext.is_flex_flow) {
 			for (i = 0; i < filter->input.flow_ext.raw_id; i++) {
 				layer_idx = filter->input.flow_ext.layer_idx;
 				field_idx = layer_idx * I40E_MAX_FLXPLD_FIED + i;
@@ -1738,6 +1737,9 @@ i40e_flow_add_del_fdir_filter(struct rte_eth_dev *dev,
 				fdir_info->fdir_guarantee_free_space > 0)
 			wait_status = false;
 	} else {
+		if (filter->input.flow_ext.is_flex_flow)
+			layer_idx = filter->input.flow_ext.layer_idx;
+
 		node = i40e_sw_fdir_filter_lookup(fdir_info,
 				&check_filter.fdir.input);
 		if (!node) {
@@ -1785,6 +1787,17 @@ i40e_flow_add_del_fdir_filter(struct rte_eth_dev *dev,
 		goto error_op;
 	}
 
+	if (filter->input.flow_ext.is_flex_flow) {
+		if (add) {
+			fdir_info->flex_flow_count[layer_idx]++;
+			pf->fdir.flex_pit_flag[layer_idx] = 1;
+		} else {
+			fdir_info->flex_flow_count[layer_idx]--;
+			if (!fdir_info->flex_flow_count[layer_idx])
+				pf->fdir.flex_pit_flag[layer_idx] = 0;
+		}
+	}
+
 	if (add) {
 		fdir_info->fdir_actual_cnt++;
 		if (fdir_info->fdir_invalprio == 1 &&
drivers/net/i40e/i40e_flow.c +4 -0
index b09ff6590d..bbd666b7a0 100644
@@ -3069,6 +3069,7 @@ i40e_flow_parse_fdir_pattern(struct rte_eth_dev *dev,
 			       &flex_pit, sizeof(struct i40e_fdir_flex_pit));
 			filter->input.flow_ext.layer_idx = layer_idx;
 			filter->input.flow_ext.raw_id = raw_id;
+			filter->input.flow_ext.is_flex_flow = true;
 			break;
 		case RTE_FLOW_ITEM_TYPE_VF:
 			vf_spec = item->spec;
@@ -5515,6 +5516,9 @@ i40e_flow_flush_fdir_filter(struct i40e_pf *pf)
 			pf->fdir.flex_mask_flag[pctype] = 0;
 		}
 
+		for (i = 0; i < I40E_MAX_FLXPLD_LAYER; i++)
+			pf->fdir.flex_pit_flag[i] = 0;
+
 		/* Disable FDIR processing as all FDIR rules are now flushed */
 		i40e_fdir_rx_proc_enable(dev, 0);
 	}
Signature · 1 line
2.26.2
#7 lore raw

PATCH v5 1/1 net/i40e: fix flex payload rule conflict issue

In reply to [email protected]

From: Beilei Xing <[email protected]>

With the following commands, the second flow can't be created successfully.

1. flow create 0 ingress pattern eth / ipv4 / udp / raw relative is 1 pattern is 0102030405 / end actions drop / end
2. flow destroy 0 rule 0
3. flow create 0 ingress pattern eth / ipv4 / udp / raw relative is 1 pattern is 010203040506 / end actions drop / end

The root cause is that a flag for flex pit isn't reset.

Fixes 6ced3dd72f5f ("net/i40e: support flexible payload parsing for FDIR")

---

v4 changes:
- Code refine.
v3 changes:
- Add flow count for flexible payload flow.
v2 changeds:
- Add fix line.

 drivers/net/i40e/i40e_ethdev.h |  3 +++
 drivers/net/i40e/i40e_fdir.c   | 19 ++++++++++++++++---
 drivers/net/i40e/i40e_flow.c   |  4 ++++
 3 files changed, 23 insertions(+), 3 deletions(-)
drivers/net/i40e/i40e_ethdev.h +3 -0
index 696c5aaf7e..aac226999c 100644
@@ -636,6 +636,7 @@ struct i40e_fdir_flow_ext {
 	bool is_udp; /* ipv4|ipv6 udp flow */
 	enum i40e_flxpld_layer_idx layer_idx;
 	struct i40e_fdir_flex_pit flex_pit[I40E_MAX_FLXPLD_LAYER * I40E_MAX_FLXPLD_FIED];
+	bool is_flex_flow;
 };
 
 /* A structure used to define the input for a flow director filter entry */
@@ -784,6 +785,8 @@ struct i40e_fdir_info {
 	bool flex_mask_flag[I40E_FILTER_PCTYPE_MAX];
 
 	bool inset_flag[I40E_FILTER_PCTYPE_MAX]; /* Mark if input set is set */
+
+	uint32_t flex_flow_count[I40E_MAX_FLXPLD_LAYER];
 };
 
 /* Ethertype filter number HW supports */
drivers/net/i40e/i40e_fdir.c +16 -3
index 50c0eee9f2..0343e8b09a 100644
@@ -355,6 +355,7 @@ i40e_init_flx_pld(struct i40e_pf *pf)
 			I40E_PRTQF_FLX_PIT(index + 1), 0x0000FC29);/*non-used*/
 		I40E_WRITE_REG(hw,
 			I40E_PRTQF_FLX_PIT(index + 2), 0x0000FC2A);/*non-used*/
+		pf->fdir.flex_pit_flag[i] = 0;
 	}
 
 	/* initialize the masks */
@@ -1513,8 +1514,6 @@ i40e_flow_set_fdir_flex_pit(struct i40e_pf *pf,
 		I40E_WRITE_REG(hw, I40E_PRTQF_FLX_PIT(field_idx), flx_pit);
 		min_next_off++;
 	}
-
-	pf->fdir.flex_pit_flag[layer_idx] = 1;
 }
 
 static int
@@ -1686,7 +1685,7 @@ i40e_flow_add_del_fdir_filter(struct rte_eth_dev *dev,
 	i40e_fdir_filter_convert(filter, &check_filter);
 
 	if (add) {
-		if (!filter->input.flow_ext.customized_pctype) {
+		if (filter->input.flow_ext.is_flex_flow) {
 			for (i = 0; i < filter->input.flow_ext.raw_id; i++) {
 				layer_idx = filter->input.flow_ext.layer_idx;
 				field_idx = layer_idx * I40E_MAX_FLXPLD_FIED + i;
@@ -1738,6 +1737,9 @@ i40e_flow_add_del_fdir_filter(struct rte_eth_dev *dev,
 				fdir_info->fdir_guarantee_free_space > 0)
 			wait_status = false;
 	} else {
+		if (filter->input.flow_ext.is_flex_flow)
+			layer_idx = filter->input.flow_ext.layer_idx;
+
 		node = i40e_sw_fdir_filter_lookup(fdir_info,
 				&check_filter.fdir.input);
 		if (!node) {
@@ -1785,6 +1787,17 @@ i40e_flow_add_del_fdir_filter(struct rte_eth_dev *dev,
 		goto error_op;
 	}
 
+	if (filter->input.flow_ext.is_flex_flow) {
+		if (add) {
+			fdir_info->flex_flow_count[layer_idx]++;
+			pf->fdir.flex_pit_flag[layer_idx] = 1;
+		} else {
+			fdir_info->flex_flow_count[layer_idx]--;
+			if (!fdir_info->flex_flow_count[layer_idx])
+				pf->fdir.flex_pit_flag[layer_idx] = 0;
+		}
+	}
+
 	if (add) {
 		fdir_info->fdir_actual_cnt++;
 		if (fdir_info->fdir_invalprio == 1 &&
drivers/net/i40e/i40e_flow.c +4 -0
index b09ff6590d..bbd666b7a0 100644
@@ -3069,6 +3069,7 @@ i40e_flow_parse_fdir_pattern(struct rte_eth_dev *dev,
 			       &flex_pit, sizeof(struct i40e_fdir_flex_pit));
 			filter->input.flow_ext.layer_idx = layer_idx;
 			filter->input.flow_ext.raw_id = raw_id;
+			filter->input.flow_ext.is_flex_flow = true;
 			break;
 		case RTE_FLOW_ITEM_TYPE_VF:
 			vf_spec = item->spec;
@@ -5515,6 +5516,9 @@ i40e_flow_flush_fdir_filter(struct i40e_pf *pf)
 			pf->fdir.flex_mask_flag[pctype] = 0;
 		}
 
+		for (i = 0; i < I40E_MAX_FLXPLD_LAYER; i++)
+			pf->fdir.flex_pit_flag[i] = 0;
+
 		/* Disable FDIR processing as all FDIR rules are now flushed */
 		i40e_fdir_rx_proc_enable(dev, 0);
 	}
Signature · 1 line
2.26.2
Acked-by on v5 1/1
#8 lore raw
Quoted 154 lines

-----Original Message-----
From: Xing, Beilei <[email protected]>
Sent: Tuesday, January 5, 2021 11:13 AM
To: Guo, Jia <[email protected]>; [email protected]
Cc: Xing, Beilei <[email protected]>; [email protected]; Sun, Chenmin <[email protected]>
Subject: [PATCH v5] net/i40e: fix flex payload rule conflict issue

From: Beilei Xing <[email protected]>

With the following commands, the second flow can't be created successfully.

1. flow create 0 ingress pattern eth / ipv4 / udp / raw relative is 1 pattern is 0102030405 / end actions drop / end
2. flow destroy 0 rule 0
3. flow create 0 ingress pattern eth / ipv4 / udp / raw relative is 1 pattern is 010203040506 / end actions drop / end

The root cause is that a flag for flex pit isn't reset.

Fixes 6ced3dd72f5f ("net/i40e: support flexible payload parsing for FDIR")

---

v4 changes:
- Code refine.
v3 changes:
- Add flow count for flexible payload flow.
v2 changeds:
- Add fix line.

 drivers/net/i40e/i40e_ethdev.h |  3 +++
 drivers/net/i40e/i40e_fdir.c   | 19 ++++++++++++++++---
 drivers/net/i40e/i40e_flow.c   |  4 ++++
 3 files changed, 23 insertions(+), 3 deletions(-)
(quoted diff) +0 -0

b/drivers/net/i40e/i40e_ethdev.h index 696c5aaf7e..aac226999c 100644

drivers/net/i40e/i40e_ethdev.h +1 -0
@@ -636,6 +636,7 @@ struct i40e_fdir_flow_ext {
 	bool is_udp; /* ipv4|ipv6 udp flow */
 	enum i40e_flxpld_layer_idx layer_idx;
 	struct i40e_fdir_flex_pit flex_pit[I40E_MAX_FLXPLD_LAYER *
 I40E_MAX_FLXPLD_FIED];
+	bool is_flex_flow;
 };
 

/* A structure used to define the input for a flow director filter entry */ @@
-784,6 +785,8 @@ struct i40e_fdir_info {

drivers/net/i40e/i40e_ethdev.h +0 -0
 	bool flex_mask_flag[I40E_FILTER_PCTYPE_MAX];
 
 	bool inset_flag[I40E_FILTER_PCTYPE_MAX]; /* Mark if input set is set

*/

drivers/net/i40e/i40e_ethdev.h +2 -0
+
+	uint32_t flex_flow_count[I40E_MAX_FLXPLD_LAYER];
 };
 
 /* Ethertype filter number HW supports */ diff --git

a/drivers/net/i40e/i40e_fdir.c b/drivers/net/i40e/i40e_fdir.c index 50c0eee9f2..0343e8b09a 100644

drivers/net/i40e/i40e_fdir.c +0 -0
@@ -355,6 +355,7 @@ i40e_init_flx_pld(struct i40e_pf *pf)
 			I40E_PRTQF_FLX_PIT(index + 1), 0x0000FC29);/*non-

used*/

drivers/net/i40e/i40e_fdir.c +0 -0
 		I40E_WRITE_REG(hw,
 			I40E_PRTQF_FLX_PIT(index + 2), 0x0000FC2A);/*non-

used*/

drivers/net/i40e/i40e_fdir.c +1 -2
+		pf->fdir.flex_pit_flag[i] = 0;
 	}
 
 	/* initialize the masks */
@@ -1513,8 +1514,6 @@ i40e_flow_set_fdir_flex_pit(struct i40e_pf *pf,
 		I40E_WRITE_REG(hw, I40E_PRTQF_FLX_PIT(field_idx),
 flx_pit);
 		min_next_off++;
 	}
-
-	pf->fdir.flex_pit_flag[layer_idx] = 1;
 }
 

static int

drivers/net/i40e/i40e_fdir.c +1 -1
@@ -1686,7 +1685,7 @@ i40e_flow_add_del_fdir_filter(struct rte_eth_dev
 *dev,
 	i40e_fdir_filter_convert(filter, &check_filter);
 
 	if (add) {
-		if (!filter->input.flow_ext.customized_pctype) {
+		if (filter->input.flow_ext.is_flex_flow) {
 			for (i = 0; i < filter->input.flow_ext.raw_id; i++) {
 				layer_idx = filter->input.flow_ext.layer_idx;

field_idx = layer_idx *
I40E_MAX_FLXPLD_FIED + i; @@ -1738,6 +1737,9 @@ i40e_flow_add_del_fdir_filter(struct rte_eth_dev *dev,

drivers/net/i40e/i40e_fdir.c +14 -0
 				fdir_info->fdir_guarantee_free_space > 0)
 			wait_status = false;
 	} else {
+		if (filter->input.flow_ext.is_flex_flow)
+			layer_idx = filter->input.flow_ext.layer_idx;
+
 		node = i40e_sw_fdir_filter_lookup(fdir_info,
 				&check_filter.fdir.input);
 		if (!node) {
@@ -1785,6 +1787,17 @@ i40e_flow_add_del_fdir_filter(struct rte_eth_dev
 *dev,
 		goto error_op;
 	}
 
+	if (filter->input.flow_ext.is_flex_flow) {
+		if (add) {
+			fdir_info->flex_flow_count[layer_idx]++;
+			pf->fdir.flex_pit_flag[layer_idx] = 1;
+		} else {
+			fdir_info->flex_flow_count[layer_idx]--;
+			if (!fdir_info->flex_flow_count[layer_idx])
+				pf->fdir.flex_pit_flag[layer_idx] = 0;
+		}
+	}
+
 	if (add) {
 		fdir_info->fdir_actual_cnt++;

if (fdir_info->fdir_invalprio == 1 && diff --git
a/drivers/net/i40e/i40e_flow.c b/drivers/net/i40e/i40e_flow.c index b09ff6590d..bbd666b7a0 100644

drivers/net/i40e/i40e_flow.c +1 -0
@@ -3069,6 +3069,7 @@ i40e_flow_parse_fdir_pattern(struct rte_eth_dev
 *dev,
 			       &flex_pit, sizeof(struct i40e_fdir_flex_pit));
 			filter->input.flow_ext.layer_idx = layer_idx;
 			filter->input.flow_ext.raw_id = raw_id;
+			filter->input.flow_ext.is_flex_flow = true;
 			break;
 		case RTE_FLOW_ITEM_TYPE_VF:
drivers/net/i40e/i40e_flow.c +3 -0
 			vf_spec = item->spec;
@@ -5515,6 +5516,9 @@ i40e_flow_flush_fdir_filter(struct i40e_pf *pf)
 			pf->fdir.flex_mask_flag[pctype] = 0;
 		}
 
+		for (i = 0; i < I40E_MAX_FLXPLD_LAYER; i++)
+			pf->fdir.flex_pit_flag[i] = 0;
+
 		/* Disable FDIR processing as all FDIR rules are now flushed
 */
 		i40e_fdir_rx_proc_enable(dev, 0);

}
--
2.26.2

Applied
#9 lore raw
Quoted 39 lines

-----Original Message-----
From: dev <[email protected]> On Behalf Of Guo, Jia
Sent: Tuesday, January 5, 2021 11:40 AM
To: Xing, Beilei <[email protected]>; [email protected]
Cc: [email protected]; Sun, Chenmin <[email protected]>
Subject: Re: [dpdk-dev] [PATCH v5] net/i40e: fix flex payload rule conflict issue

Quoted 28 lines

-----Original Message-----
From: Xing, Beilei <[email protected]>
Sent: Tuesday, January 5, 2021 11:13 AM
To: Guo, Jia <[email protected]>; [email protected]
Cc: Xing, Beilei <[email protected]>; [email protected]; Sun, Chenmin <[email protected]>
Subject: [PATCH v5] net/i40e: fix flex payload rule conflict issue

From: Beilei Xing <[email protected]>

With the following commands, the second flow can't be created successfully.

1. flow create 0 ingress pattern eth / ipv4 / udp / raw relative is 1 pattern is 0102030405 / end actions drop / end
2. flow destroy 0 rule 0
3. flow create 0 ingress pattern eth / ipv4 / udp / raw relative is 1 pattern is 010203040506 / end actions drop / end

The root cause is that a flag for flex pit isn't reset.

Fixes 6ced3dd72f5f ("net/i40e: support flexible payload parsing for

FDIR")
Cc: [email protected]

Applied to dpdk-next-net-intel.

Thanks
Qi