Skip to contentSkip to conversation

ethtool ntuple filters

PATCH Superseded Jacob Keller 2 participants 19 messages Feb 6, 2017 – Mar 20, 2017

Subsystems (4)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

cover ethtool ntuple filters

This series (labeled PART2) contains patches to implement flexible data filters for the ntuple ethtool API, making use of the user-def field. First, it resolves a longstanding issue with the API not properly honoring mask fields.

Second, it implements support for partial filtering, allowing the user to filter only partially on some fields, rather than requiring all fields to be used.

Finally, it implements the flexible byte ntuple filters on top of the cleaned up code.

It is dependent on the series labeled PART1 which has self-contained bug fixes.

Jacob Keller (9):
i40e: correctly honor the mask fields for ETHTOOL_SRXCLSRLINS
i40e: check current configured input set when adding ntuple filters
i40e: restore default input set for each flow type
i40e: allow changing input set for ntuple filters
i40e: partition the ring_cookie to get VF index
i40e: add parsing of flexible filter fields from userdef
i40e: implement support for flexible word payload
i40e: add support for SCTPv4 FDir filters
i40e: document drivers use of ntuple filters

 Documentation/networking/i40e.txt              |  72 ++
 drivers/net/ethernet/intel/i40e/i40e.h         | 130 ++++
 drivers/net/ethernet/intel/i40e/i40e_ethtool.c | 970 ++++++++++++++++++++++++-
 drivers/net/ethernet/intel/i40e/i40e_main.c    |  37 +
 drivers/net/ethernet/intel/i40e/i40e_txrx.c    | 107 +++
 5 files changed, 1283 insertions(+), 33 deletions(-)
Signature · 1 line
2.12.0.rc0.151.g8a5726c42288
#1 lore raw

PATCH 1/9 i40e: correctly honor the mask fields for ETHTOOL_SRXCLSRLINS

The current implementation of .set_rxnfc does not properly read the mask field for filter entries. This results in incorrect driver behavior, as we do not reject filters which have masks set to ignore some fields. The current implementation simply assumes that every part of the tuple or "input set" is specified. This results in filters not behaving as expected, and not working correctly.

As a first step in supporting some partial filters, add code which checks the mask fields and rejects any filters which do not have an acceptable mask. For now, we just assume that all fields must be set.

This will get the driver one step towards allowing some partial filters. At a minimum, the ethtool commands which previously installed filters that would not function will now return a non-zero exit code indicating failure instead.

We should now be meeting the minimum requirements of the .set_rxnfc API, by ensuring that all filters we program have a valid mask value for each field.

Finally, add code to report the mask correctly so that the ethtool command properly reports the mask to the user.

Change-Id Ia020149e07c87aa3fcec7b2283621b887ef0546f
---
 drivers/net/ethernet/intel/i40e/i40e_ethtool.c | 83 ++++++++++++++++++++++++++
 1 file changed, 83 insertions(+)
drivers/net/ethernet/intel/i40e/i40e_ethtool.c +83 -0
index f825181ec7c2..772dd234d581 100644
@@ -2371,6 +2371,12 @@ static int i40e_get_ethtool_fdir_entry(struct i40e_pf *pf,
 	fsp->h_u.tcp_ip4_spec.ip4src = rule->dst_ip;
 	fsp->h_u.tcp_ip4_spec.ip4dst = rule->src_ip;
 
+	/* Set the mask fields */
+	fsp->m_u.tcp_ip4_spec.psrc = htons(0xFFFF);
+	fsp->m_u.tcp_ip4_spec.pdst = htons(0xFFFF);
+	fsp->m_u.tcp_ip4_spec.ip4src = htonl(0xFFFFFFFF);
+	fsp->m_u.tcp_ip4_spec.ip4dst = htonl(0xFFFFFFFF);
+
 	if (rule->dest_ctl == I40E_FILTER_PROGRAM_DESC_DEST_DROP_PACKET)
 		fsp->ring_cookie = RX_CLS_FLOW_DISC;
 	else
@@ -2679,6 +2685,79 @@ static int i40e_del_fdir_entry(struct i40e_vsi *vsi,
 	return ret;
 }
 
+/**
+ * i40e_check_fdir_input_set - Check that a given rx_flow_spec mask is valid
+ * @fsp: pointer to Rx flow specification
+ *
+ * Ensures that a given ethtool_rx_flow_spec has a valid mask.
+ **/
+static int i40e_check_fdir_input_set(struct ethtool_rx_flow_spec *fsp)
+{
+	struct ethtool_tcpip4_spec *tcp_ip4_spec;
+	struct ethtool_usrip4_spec *usr_ip4_spec;
+
+	/* Verify the provided mask is valid. */
+	switch (fsp->flow_type & ~FLOW_EXT) {
+	case SCTP_V4_FLOW:
+	case TCP_V4_FLOW:
+	case UDP_V4_FLOW:
+		tcp_ip4_spec = &fsp->m_u.tcp_ip4_spec;
+
+		/* IPv4 source address */
+		if (!tcp_ip4_spec->ip4src || ~tcp_ip4_spec->ip4src)
+			return -EOPNOTSUPP;
+
+		/* IPv4 destination address */
+		if (!tcp_ip4_spec->ip4dst || ~tcp_ip4_spec->ip4dst)
+			return -EOPNOTSUPP;
+
+		/* L4 source port */
+		if (!tcp_ip4_spec->psrc || ~tcp_ip4_spec->psrc)
+			return -EOPNOTSUPP;
+
+		/* L4 destination port */
+		if (!tcp_ip4_spec->pdst || ~tcp_ip4_spec->pdst)
+			return -EOPNOTSUPP;
+
+		/* Filtering on Type of Service is not supported. */
+		if (tcp_ip4_spec->tos)
+			return -EOPNOTSUPP;
+
+		break;
+	case IP_USER_FLOW:
+		usr_ip4_spec = &fsp->m_u.usr_ip4_spec;
+
+		/* IPv4 source address */
+		if (!usr_ip4_spec->ip4src || ~usr_ip4_spec->ip4src)
+			return -EOPNOTSUPP;
+
+		/* IPv4 destination address */
+		if (!usr_ip4_spec->ip4dst || ~usr_ip4_spec->ip4dst)
+			return -EOPNOTSUPP;
+
+		/* First 4 bytes of L4 header */
+		if (!usr_ip4_spec->l4_4_bytes || ~usr_ip4_spec->l4_4_bytes)
+			return -EOPNOTSUPP;
+
+		/* Filtering on Type of Service is not supported. */
+		if (usr_ip4_spec->tos)
+			return -EOPNOTSUPP;
+
+		/* IP version does not have a mask field. */
+		if (usr_ip4_spec->ip_ver)
+			return -EINVAL;
+
+		/* L4 protocol doesn't have a mask field. */
+		if (usr_ip4_spec->proto)
+			return -EINVAL;
+		break;
+	default:
+		return -EOPNOTSUPP;
+	}
+
+	return 0;
+}
+
 /**
  * i40e_add_fdir_ethtool - Add/Remove Flow Director filters
  * @vsi: pointer to the targeted VSI
@@ -2719,6 +2798,10 @@ static int i40e_add_fdir_ethtool(struct i40e_vsi *vsi,
 	if (fsp->flow_type & FLOW_MAC_EXT)
 		return -EINVAL;
 
+	ret = i40e_check_fdir_input_set(fsp);
+	if (ret)
+		return ret;
+
 	if (fsp->location >= (pf->hw.func_caps.fd_filters_best_effort +
 			      pf->hw.func_caps.fd_filters_guaranteed)) {
 		return -EINVAL;
Signature · 1 line
2.12.0.rc0.151.g8a5726c42288
#2 lore raw

PATCH 1/9 i40e: correctly honor the mask fields for ETHTOOL_SRXCLSRLINS

Quoted 31 lines

-----Original Message-----
From: Intel-wired-lan [mailto:intel-wired-lan-bounces at lists.osuosl.org] On Behalf Of Jacob Keller
Sent: Monday, February 6, 2017 2:39 PM
To: Intel Wired LAN <[email protected]>
Subject: [Intel-wired-lan] [PART2 PATCH 1/9] i40e: correctly honor the mask fields for ETHTOOL_SRXCLSRLINS

The current implementation of .set_rxnfc does not properly read the mask field for filter entries. This results in incorrect driver behavior, as we do not reject filters which have masks set to ignore some fields. The current implementation simply assumes that every part of the tuple or "input set" is specified. This results in filters not behaving as expected, and not working correctly.

As a first step in supporting some partial filters, add code which checks the mask fields and rejects any filters which do not have an acceptable mask. For now, we just assume that all fields must be set.

This will get the driver one step towards allowing some partial filters. At a minimum, the ethtool commands which previously installed filters that would not function will now return a non-zero exit code indicating failure instead.

We should now be meeting the minimum requirements of the .set_rxnfc API, by ensuring that all filters we program have a valid mask value for each field.

Finally, add code to report the mask correctly so that the ethtool command properly reports the mask to the user.

Quoted 6 lines
Change-Id Ia020149e07c87aa3fcec7b2283621b887ef0546f
---
 drivers/net/ethernet/intel/i40e/i40e_ethtool.c | 83
++++++++++++++++++++++++++
 1 file changed, 83 insertions(+)
#3 lore raw

PATCH 2/9 i40e: check current configured input set when adding ntuple filters

In reply to Jacob Keller

Do not assume that hardware has been programmed with the default mask, but instead read the input set registers to determine what is currently programmed. This ensures that all programmed filters match exactly how the hardware will interpret them, avoiding confusion regarding filter behavior.

This sets the initial ground-work for allowing custom input sets where some fields are disabled. A future patch will fully implement this feature.

The use of ntohl and ntohs is not strictly necessary for correctness, but it helps avoid a sparse warning due to the construction of "!~<be16>" degrades the be16 value into an integer when doing negation. This could alternatively be fixed by the following code construction without the byteswaps,

if (!<value>)
<disable field>
else if (~<value>)
<return error>
else
<enable field>

However, this construction is confusing because the error return occurs in the middle, which is misleading to other readers. Additionally, it makes the condition for the <enable field> much more difficult to process due to it being in an else statement.

Change-Id 3d8db46cb28ea0afdaac8c5b31a2bfb90e3a4102
---
 drivers/net/ethernet/intel/i40e/i40e.h         |  19 ++++
 drivers/net/ethernet/intel/i40e/i40e_ethtool.c | 117 +++++++++++++++++++++----
 2 files changed, 121 insertions(+), 15 deletions(-)
drivers/net/ethernet/intel/i40e/i40e.h +19 -0
index 9957eca8f71d..c32658ca5c22 100644
@@ -742,6 +742,25 @@ static inline int i40e_get_fd_cnt_all(struct i40e_pf *pf)
 	return pf->hw.fdir_shared_filter_count + pf->fdir_pf_filter_count;
 }
 
+/**
+ * i40e_read_fd_input_set - reads value of flow director input set register
+ * @pf: pointer to the PF struct
+ * @addr: register addr
+ *
+ * This function reads value of flow director input set register
+ * specified by 'addr' (which is specific to flow-type)
+ **/
+static inline u64 i40e_read_fd_input_set(struct i40e_pf *pf, u16 addr)
+{
+	u64 val;
+
+	val = i40e_read_rx_ctl(&pf->hw, I40E_PRTQF_FD_INSET(addr, 1));
+	val <<= 32;
+	val += i40e_read_rx_ctl(&pf->hw, I40E_PRTQF_FD_INSET(addr, 0));
+
+	return val;
+}
+
 /* needed by i40e_ethtool.c */
 int i40e_up(struct i40e_vsi *vsi);
 void i40e_down(struct i40e_vsi *vsi);
drivers/net/ethernet/intel/i40e/i40e_ethtool.c +102 -15

188 lines not on this page · raw message

Signature · 1 line
2.12.0.rc0.151.g8a5726c42288
#4 lore raw

PATCH 2/9 i40e: check current configured input set when adding ntuple filters

Quoted 34 lines

-----Original Message-----
From: Intel-wired-lan [mailto:intel-wired-lan-bounces at lists.osuosl.org] On Behalf Of Jacob Keller
Sent: Monday, February 6, 2017 2:39 PM
To: Intel Wired LAN <[email protected]>
Subject: [Intel-wired-lan] [PART2 PATCH 2/9] i40e: check current configured input set when adding ntuple filters

Do not assume that hardware has been programmed with the default mask, but instead read the input set registers to determine what is currently programmed. This ensures that all programmed filters match exactly how the hardware will interpret them, avoiding confusion regarding filter behavior.

This sets the initial ground-work for allowing custom input sets where some fields are disabled. A future patch will fully implement this feature.

The use of ntohl and ntohs is not strictly necessary for correctness, but it helps avoid a sparse warning due to the construction of "!~<be16>" degrades the be16 value into an integer when doing negation.
This could alternatively be fixed by the following code construction without the byteswaps,

if (!<value>)
<disable field>
else if (~<value>)
<return error>
else
<enable field>

However, this construction is confusing because the error return occurs in the middle, which is misleading to other readers. Additionally, it makes the condition for the <enable field> much more difficult to process due to it being in an else statement.

Quoted 7 lines
Change-Id 3d8db46cb28ea0afdaac8c5b31a2bfb90e3a4102
---
 drivers/net/ethernet/intel/i40e/i40e.h         |  19 ++++
 drivers/net/ethernet/intel/i40e/i40e_ethtool.c | 117
+++++++++++++++++++++----
 2 files changed, 121 insertions(+), 15 deletions(-)
#5 lore raw

PATCH 3/9 i40e: restore default input set for each flow type

In reply to Jacob Keller

Ensure that the default input set is correctly reprogrammed when cleaning up after disabling flow director support. This ensures that the programmed value will be in a clean state.

Although we do not yet have support for SCTPv4 filters, a future patch will add support for this protocol, so we will correctly restore the SCTPv4 input set here as well. Note that strictly speaking the default hardware value for SCTP includes matching the verification tag. However, the ethtool API does not have support for specifying this value, so there is no reason to keep the verification field enabled.

This patch is the next step on the way to enabling partial tuple filters which will be implemented in a following patch.

Change-Id c22e1c267ae37518bb036aca4a5694681449f283
---
 drivers/net/ethernet/intel/i40e/i40e.h      | 18 ++++++++++++++++++
 drivers/net/ethernet/intel/i40e/i40e_main.c | 19 +++++++++++++++++++
 2 files changed, 37 insertions(+)
drivers/net/ethernet/intel/i40e/i40e.h +18 -0
index c32658ca5c22..4d54d5005ad3 100644
@@ -761,6 +761,24 @@ static inline u64 i40e_read_fd_input_set(struct i40e_pf *pf, u16 addr)
 	return val;
 }
 
+/**
+ * i40e_write_fd_input_set - writes value into flow director input set register
+ * @pf: pointer to the PF struct
+ * @addr: register addr
+ * @val: value to be written
+ *
+ * This function writes specified value to the register specified by 'addr'.
+ * This register is input set register based on flow-type.
+ **/
+static inline void i40e_write_fd_input_set(struct i40e_pf *pf,
+					   u16 addr, u64 val)
+{
+	i40e_write_rx_ctl(&pf->hw, I40E_PRTQF_FD_INSET(addr, 1),
+			  (u32)(val >> 32));
+	i40e_write_rx_ctl(&pf->hw, I40E_PRTQF_FD_INSET(addr, 0),
+			  (u32)(val & 0xFFFFFFFFULL));
+}
+
 /* needed by i40e_ethtool.c */
 int i40e_up(struct i40e_vsi *vsi);
 void i40e_down(struct i40e_vsi *vsi);
drivers/net/ethernet/intel/i40e/i40e_main.c +19 -0
index c188559fc4ed..2c3a427f5b75 100644
@@ -5900,6 +5900,25 @@ static void i40e_fdir_filter_exit(struct i40e_pf *pf)
 	pf->fd_tcp4_filter_cnt = 0;
 	pf->fd_udp4_filter_cnt = 0;
 	pf->fd_ip4_filter_cnt = 0;
+
+	/* Reprogram the default input set for TCP/IPv4 */
+	i40e_write_fd_input_set(pf, I40E_FILTER_PCTYPE_NONF_IPV4_TCP,
+				I40E_L3_SRC_MASK | I40E_L3_DST_MASK |
+				I40E_L4_SRC_MASK | I40E_L4_DST_MASK);
+
+	/* Reprogram the default input set for UDP/IPv4 */
+	i40e_write_fd_input_set(pf, I40E_FILTER_PCTYPE_NONF_IPV4_UDP,
+				I40E_L3_SRC_MASK | I40E_L3_DST_MASK |
+				I40E_L4_SRC_MASK | I40E_L4_DST_MASK);
+
+	/* Reprogram the default input set for SCTP/IPv4 */
+	i40e_write_fd_input_set(pf, I40E_FILTER_PCTYPE_NONF_IPV4_SCTP,
+				I40E_L3_SRC_MASK | I40E_L3_DST_MASK |
+				I40E_L4_SRC_MASK | I40E_L4_DST_MASK);
+
+	/* Reprogram the default input set for Other/IPv4 */
+	i40e_write_fd_input_set(pf, I40E_FILTER_PCTYPE_NONF_IPV4_OTHER,
+				I40E_L3_SRC_MASK | I40E_L3_DST_MASK);
 }
 
 /**
Signature · 1 line
2.12.0.rc0.151.g8a5726c42288
#6 lore raw

PATCH 3/9 i40e: restore default input set for each flow type

Quoted 22 lines

-----Original Message-----
From: Intel-wired-lan [mailto:intel-wired-lan-bounces at lists.osuosl.org] On Behalf Of Jacob Keller
Sent: Monday, February 6, 2017 2:39 PM
To: Intel Wired LAN <[email protected]>
Subject: [Intel-wired-lan] [PART2 PATCH 3/9] i40e: restore default input set for each flow type

Ensure that the default input set is correctly reprogrammed when cleaning up after disabling flow director support. This ensures that the programmed value will be in a clean state.

Although we do not yet have support for SCTPv4 filters, a future patch will add support for this protocol, so we will correctly restore the SCTPv4 input set here as well. Note that strictly speaking the default hardware value for SCTP includes matching the verification tag. However, the ethtool API does not have support for specifying this value, so there is no reason to keep the verification field enabled.

This patch is the next step on the way to enabling partial tuple filters which will be implemented in a following patch.

Quoted 6 lines
Change-Id c22e1c267ae37518bb036aca4a5694681449f283
---
 drivers/net/ethernet/intel/i40e/i40e.h      | 18 ++++++++++++++++++
 drivers/net/ethernet/intel/i40e/i40e_main.c | 19 +++++++++++++++++++
 2 files changed, 37 insertions(+)
#7 lore raw

PATCH 4/9 i40e: allow changing input set for ntuple filters

In reply to Jacob Keller

Add support to detect when we can update the input set for each flow type.

Because the hardware only supports a single input set for all flows of that matching type, the driver shall only allow the input set to change if there are no other configured filters for that flow type.

Thus, the first filter added for each flow type is allowed to change the input set, and all future filters must match the same input set. Display a diagnostic message whenever the filter input set changes, and a warning whenever a filter cannot be accepted because it does not match the configured input set.

Change-Id c22e1c267ae37518bb036aca4a5694681449f283
---
 drivers/net/ethernet/intel/i40e/i40e_ethtool.c | 148 ++++++++++++++++++++++++-
 1 file changed, 145 insertions(+), 3 deletions(-)
drivers/net/ethernet/intel/i40e/i40e_ethtool.c +145 -3
index 5a7718e67c0a..595ede664c56 100644
@@ -2718,12 +2718,108 @@ static int i40e_del_fdir_entry(struct i40e_vsi *vsi,
 	return ret;
 }
 
+/**
+ * i40e_flow_str - Converts a flow_type into a human readable string
+ * @flow_type: the flow type from a flow specification
+ *
+ * Currently only flow types we support are included here, and the string
+ * value attempts to match what ethtool would use to configure this flow type.
+ **/
+static const char *i40e_flow_str(struct ethtool_rx_flow_spec *fsp)
+{
+	switch (fsp->flow_type & ~FLOW_EXT) {
+	case TCP_V4_FLOW:
+		return "tcp4";
+	case UDP_V4_FLOW:
+		return "udp4";
+	case SCTP_V4_FLOW:
+		return "sctp4";
+	case IP_USER_FLOW:
+		return "ip4";
+	default:
+		return "unknown";
+	}
+}
+
+/**
+ * i40e_print_input_set - Show changes between two input sets
+ * @vsi: the vsi being configured
+ * @old: the old input set
+ * @new: the new input set
+ *
+ * Print the difference between old and new input sets by showing which series
+ * of words are toggled on or off. Only displays the bits we actually support
+ * changing.
+ **/
+static void i40e_print_input_set(struct i40e_vsi *vsi, u64 old, u64 new)
+{
+	struct i40e_pf *pf = vsi->back;
+	bool old_value, new_value;
+
+	old_value = !!(old & I40E_L3_SRC_MASK);
+	new_value = !!(new & I40E_L3_SRC_MASK);
+	if (old_value != new_value)
+		netif_info(pf, drv, vsi->netdev, "L3 source address: %s -> %s\n",
+			   old_value ? "ON" : "OFF",
+			   new_value ? "ON" : "OFF");
+
+	old_value = !!(old & I40E_L3_DST_MASK);
+	new_value = !!(new & I40E_L3_DST_MASK);
+	if (old_value != new_value)
+		netif_info(pf, drv, vsi->netdev, "L3 destination address: %s -> %s\n",
+			   old_value ? "ON" : "OFF",
+			   new_value ? "ON" : "OFF");
+
+	old_value = !!(old & I40E_L4_SRC_MASK);
+	new_value = !!(new & I40E_L4_SRC_MASK);
+	if (old_value != new_value)
+		netif_info(pf, drv, vsi->netdev, "L4 source port: %s -> %s\n",
+			   old_value ? "ON" : "OFF",
+			   new_value ? "ON" : "OFF");
+
+	old_value = !!(old & I40E_L4_DST_MASK);
+	new_value = !!(new & I40E_L4_DST_MASK);
+	if (old_value != new_value)
+		netif_info(pf, drv, vsi->netdev, "L4 destination port: %s -> %s\n",
+			   old_value ? "ON" : "OFF",
+			   new_value ? "ON" : "OFF");
+
+	old_value = !!(old & I40E_VERIFY_TAG_MASK);
+	new_value = !!(new & I40E_VERIFY_TAG_MASK);
+	if (old_value != new_value)
+		netif_info(pf, drv, vsi->netdev, "SCTP verification tag: %s -> %s\n",
+			   old_value ? "ON" : "OFF",
+			   new_value ? "ON" : "OFF");
+
+	netif_info(pf, drv, vsi->netdev, "  Current input set: %0llx\n",
+		   old);
+	netif_info(pf, drv, vsi->netdev, "Requested input set: %0llx\n",
+		   new);
+}
+
 /**
  * i40e_check_fdir_input_set - Check that a given rx_flow_spec mask is valid
  * @vsi: pointer to the targeted VSI
  * @fsp: pointer to Rx flow specification
  *
- * Ensures that a given ethtool_rx_flow_spec has a valid mask.
+ * Ensures that a given ethtool_rx_flow_spec has a valid mask. Some support
+ * for partial matches exists with a few limitations. First, hardware only
+ * supports masking by word boundary (2 bytes) and not per individual bit.
+ * Second, hardware is limited to using one mask for a flow type and cannot
+ * use a separate mask for each filter.
+ *
+ * To support these limitations, if we already have a configured filter for
+ * the specified type, this function enforces that new filters of the type
+ * match the configured input set. Otherwise, if we do not have a filter of
+ * the specified type, we allow the input set to be updated to match the
+ * desired filter.
+ *
+ * To help ensure that administrators understand why filters weren't displayed
+ * as supported, we print a diagnostic message displaying how the input set
+ * would change and warning to delete the preexisting filters if required.
+ *
+ * Returns 0 on successful input set match, and a negative return code on
+ * failure.
  **/
 static int i40e_check_fdir_input_set(struct i40e_vsi *vsi,
 				     struct ethtool_rx_flow_spec *fsp)
@@ -2732,17 +2828,21 @@ static int i40e_check_fdir_input_set(struct i40e_vsi *vsi,
 	struct ethtool_tcpip4_spec *tcp_ip4_spec;
 	struct ethtool_usrip4_spec *usr_ip4_spec;
 	u64 current_mask, new_mask;
+	u16 *fdir_filter_count;
 	u16 index;
 
 	switch (fsp->flow_type & ~FLOW_EXT) {
 	case TCP_V4_FLOW:
 		index = I40E_FILTER_PCTYPE_NONF_IPV4_TCP;
+		fdir_filter_count = &pf->fd_tcp4_filter_cnt;
 		break;
 	case UDP_V4_FLOW:
 		index = I40E_FILTER_PCTYPE_NONF_IPV4_UDP;
+		fdir_filter_count = &pf->fd_udp4_filter_cnt;
 		break;
 	case IP_USER_FLOW:
 		index = I40E_FILTER_PCTYPE_NONF_IPV4_OTHER;
+		fdir_filter_count = &pf->fd_ip4_filter_cnt;
 		break;
 	default:
 		return -EOPNOTSUPP;
@@ -2752,7 +2852,15 @@ static int i40e_check_fdir_input_set(struct i40e_vsi *vsi,
 	current_mask = i40e_read_fd_input_set(pf, index);
 	new_mask = current_mask;
 
-	/* Verify the provided mask is valid. */
+	/* Determine, if any, the required changes to the input set in order
+	 * to support the provided mask.
+	 *
+	 * Hardware only supports masking at word (2 byte) granularity and does
+	 * not support full bitwise masking. This implementation simplifies
+	 * even further and only supports fully enabled or fully disabled
+	 * masks for each field, even though we could split the ip4src and
+	 * ip4dst fields.
+	 */
 	switch (fsp->flow_type & ~FLOW_EXT) {
 	case TCP_V4_FLOW:
 	case UDP_V4_FLOW:
@@ -2839,8 +2947,42 @@ static int i40e_check_fdir_input_set(struct i40e_vsi *vsi,
 		return -EOPNOTSUPP;
 	}
 
-	if (new_mask != current_mask)
+	/* If the input set doesn't need any changes then this filter is safe
+	 * to apply.
+	 */
+	if (new_mask == current_mask)
+		return 0;
+
+	netif_info(pf, drv, vsi->netdev, "Input set change requested for %s flows:\n",
+		   i40e_flow_str(fsp));
+	i40e_print_input_set(vsi, current_mask, new_mask);
+
+	/* Hardware input sets are global across multiple ports, so even the
+	 * main port cannot change them when in MFP mode as this would impact
+	 * any filters on the other ports.
+	 */
+	if (pf->flags & I40E_FLAG_MFP_ENABLED) {
+		netif_err(pf, drv, vsi->netdev, "Cannot change Flow Director input sets while MFP is enabled\n");
 		return -EOPNOTSUPP;
+	}
+
+	/* This filter requires us to update the input set. However, hardware
+	 * only supports one input set per flow type, and does not support
+	 * separate masks for each filter. This means that we can only support
+	 * a single mask for all filters of a specific type.
+	 *
+	 * If we have preexisting filters, they obviously depend on the
+	 * current programmed input set. Display a diagnostic message in this
+	 * case explaining why the filter could not be accepted.
+	 */
+	if (*fdir_filter_count) {
+		netif_err(pf, drv, vsi->netdev, "Cannot change input set for %s flows until %d preexisting filters are removed\n",
+			  i40e_flow_str(fsp),
+			  *fdir_filter_count);
+		return -EOPNOTSUPP;
+	}
+
+	i40e_write_fd_input_set(pf, index, new_mask);
 
 	return 0;
 }
Signature · 1 line
2.12.0.rc0.151.g8a5726c42288
#8 lore raw

PATCH 4/9 i40e: allow changing input set for ntuple filters

Quoted 19 lines

-----Original Message-----
From: Intel-wired-lan [mailto:intel-wired-lan-bounces at lists.osuosl.org] On Behalf Of Jacob Keller
Sent: Monday, February 6, 2017 2:39 PM
To: Intel Wired LAN <[email protected]>
Subject: [Intel-wired-lan] [PART2 PATCH 4/9] i40e: allow changing input set for ntuple filters

Add support to detect when we can update the input set for each flow type.

Because the hardware only supports a single input set for all flows of that matching type, the driver shall only allow the input set to change if there are no other configured filters for that flow type.

Thus, the first filter added for each flow type is allowed to change the input set, and all future filters must match the same input set. Display a diagnostic message whenever the filter input set changes, and a warning whenever a filter cannot be accepted because it does not match the configured input set.

Quoted 6 lines
Change-Id c22e1c267ae37518bb036aca4a5694681449f283
---
 drivers/net/ethernet/intel/i40e/i40e_ethtool.c | 148
++++++++++++++++++++++++-
 1 file changed, 145 insertions(+), 3 deletions(-)
#9 lore raw

PATCH 5/9 i40e: partition the ring_cookie to get VF index

In reply to Jacob Keller

Do not use the user-def field for determining the VF target. Instead, similar to ixgbe, partition the ring_cookie value into 8bits of VF index, along with 32bits of queue number. This is better than using the user-def field, because it leaves the field open for extension in a future patch which will enable flexible data. Also, this matches with convention used by ixgbe and other drivers.

Change-Id e36745186d817216b12f0313b99ec95cb8a9130c
---
 drivers/net/ethernet/intel/i40e/i40e_ethtool.c | 74 ++++++++++++++------------
 1 file changed, 40 insertions(+), 34 deletions(-)
drivers/net/ethernet/intel/i40e/i40e_ethtool.c +40 -34
index 595ede664c56..f6cca3254fd4 100644
@@ -2420,8 +2420,13 @@ static int i40e_get_ethtool_fdir_entry(struct i40e_pf *pf,
 
 		vsi = i40e_find_vsi_from_id(pf, rule->dest_vsi);
 		if (vsi && vsi->type == I40E_VSI_SRIOV) {
-			fsp->h_ext.data[1] = htonl(vsi->vf_id);
-			fsp->m_ext.data[1] = htonl(0x1);
+			/* VFs are zero-indexed by the driver, but ethtool
+			 * expects them to be one-indexed, so add one here
+			 */
+			u64 ring_vf = vsi->vf_id + 1;
+
+			ring_vf <<= ETHTOOL_RX_FLOW_SPEC_RING_VF_OFF;
+			fsp->ring_cookie |= ring_vf;
 		}
 	}
 
@@ -3000,9 +3005,10 @@ static int i40e_add_fdir_ethtool(struct i40e_vsi *vsi,
 {
 	struct ethtool_rx_flow_spec *fsp;
 	struct i40e_fdir_filter *input;
+	u16 dest_vsi = 0, q_index = 0;
 	struct i40e_pf *pf;
 	int ret = -EINVAL;
-	u16 vf_id;
+	u8 dest_ctl;
 
 	if (!vsi)
 		return -EINVAL;
@@ -3036,9 +3042,32 @@ static int i40e_add_fdir_ethtool(struct i40e_vsi *vsi,
 		return -EINVAL;
 	}
 
-	if ((fsp->ring_cookie != RX_CLS_FLOW_DISC) &&
-	    (fsp->ring_cookie >= vsi->num_queue_pairs))
-		return -EINVAL;
+	/* ring_cookie is either the drop index, or is a mask of the queue
+	 * index and VF id we wish to target.
+	 */
+	if (fsp->ring_cookie == RX_CLS_FLOW_DISC) {
+		dest_ctl = I40E_FILTER_PROGRAM_DESC_DEST_DROP_PACKET;
+	} else {
+		u32 ring = ethtool_get_flow_spec_ring(fsp->ring_cookie);
+		u8 vf = ethtool_get_flow_spec_ring_vf(fsp->ring_cookie);
+
+		if (!vf) {
+			if (ring >= vsi->num_queue_pairs)
+				return -EINVAL;
+			dest_vsi = vsi->id;
+		} else {
+			/* VFs are zero-indexed, so we subtract one here */
+			vf--;
+
+			if (vf >= pf->num_alloc_vfs)
+				return -EINVAL;
+			if (ring >= pf->vf[vf].num_queue_pairs)
+				return -EINVAL;
+			dest_vsi = pf->vf[vf].lan_vsi_id;
+		}
+		dest_ctl = I40E_FILTER_PROGRAM_DESC_DEST_DIRECT_PACKET_QINDEX;
+		q_index = ring;
+	}
 
 	input = kzalloc(sizeof(*input), GFP_KERNEL);
 
@@ -3046,19 +3075,13 @@ static int i40e_add_fdir_ethtool(struct i40e_vsi *vsi,
 		return -ENOMEM;
 
 	input->fd_id = fsp->location;
-
-	if (fsp->ring_cookie == RX_CLS_FLOW_DISC)
-		input->dest_ctl = I40E_FILTER_PROGRAM_DESC_DEST_DROP_PACKET;
-	else
-		input->dest_ctl =
-			     I40E_FILTER_PROGRAM_DESC_DEST_DIRECT_PACKET_QINDEX;
-
-	input->q_index = fsp->ring_cookie;
-	input->flex_off = 0;
-	input->pctype = 0;
-	input->dest_vsi = vsi->id;
+	input->q_index = q_index;
+	input->dest_vsi = dest_vsi;
+	input->dest_ctl = dest_ctl;
 	input->fd_status = I40E_FILTER_PROGRAM_DESC_FD_STATUS_FD_ID;
 	input->cnt_index  = I40E_FD_SB_STAT_IDX(pf->hw.pf_id);
+	input->dst_ip = fsp->h_u.tcp_ip4_spec.ip4src;
+	input->src_ip = fsp->h_u.tcp_ip4_spec.ip4dst;
 	input->flow_type = fsp->flow_type;
 	input->ip4_proto = fsp->h_u.usr_ip4_spec.proto;
 
@@ -3070,23 +3093,6 @@ static int i40e_add_fdir_ethtool(struct i40e_vsi *vsi,
 	input->dst_ip = fsp->h_u.tcp_ip4_spec.ip4src;
 	input->src_ip = fsp->h_u.tcp_ip4_spec.ip4dst;
 
-	if (ntohl(fsp->m_ext.data[1])) {
-		vf_id = ntohl(fsp->h_ext.data[1]);
-		if (vf_id >= pf->num_alloc_vfs) {
-			netif_info(pf, drv, vsi->netdev,
-				   "Invalid VF id %d\n", vf_id);
-			goto free_input;
-		}
-		/* Find vsi id from vf id and override dest vsi */
-		input->dest_vsi = pf->vf[vf_id].lan_vsi_id;
-		if (input->q_index >= pf->vf[vf_id].num_queue_pairs) {
-			netif_info(pf, drv, vsi->netdev,
-				   "Invalid queue id %d for VF %d\n",
-				   input->q_index, vf_id);
-			goto free_input;
-		}
-	}
-
 	ret = i40e_add_del_fdir(vsi, input, true);
 	if (ret)
 		goto free_input;
Signature · 1 line
2.12.0.rc0.151.g8a5726c42288
#10 lore raw

PATCH 5/9 i40e: partition the ring_cookie to get VF index

Quoted 15 lines

-----Original Message-----
From: Intel-wired-lan [mailto:intel-wired-lan-bounces at lists.osuosl.org] On Behalf Of Jacob Keller
Sent: Monday, February 6, 2017 2:39 PM
To: Intel Wired LAN <[email protected]>
Subject: [Intel-wired-lan] [PART2 PATCH 5/9] i40e: partition the ring_cookie to get VF index

Do not use the user-def field for determining the VF target. Instead, similar to ixgbe, partition the ring_cookie value into 8bits of VF index, along with 32bits of queue number. This is better than using the user-def field, because it leaves the field open for extension in a future patch which will enable flexible data. Also, this matches with convention used by ixgbe and other drivers.

Quoted 6 lines
Change-Id e36745186d817216b12f0313b99ec95cb8a9130c
---
 drivers/net/ethernet/intel/i40e/i40e_ethtool.c | 74 ++++++++++++++-------
-----
 1 file changed, 40 insertions(+), 34 deletions(-)
#11 lore raw

PATCH 6/9 i40e: add parsing of flexible filter fields from userdef

In reply to Jacob Keller

Add code to parse the user-def field into a data structure format. This code is intended to allow future extensions of the user-def field by keeping all code that actually reads and writes the field into a single location. This ensures that we do not litter the driver with references to the user-def field and minimizes the amount of bitwise operations we need to do on the data.

Add code which parses the lower 32bits into a flexible word and its offset. This will be used in a future patch to enable flexible filters which can match on some arbitrary data in the packet payload. For now, we just return -EOPNOTSUPP when this is used.

Add code to fill in the user-def field when reporting the filter back, even though we don't actually implement any user-def fields yet.

Additionally, ensure that we mask the extended FLOW_EXT bit from the flow_type now that we will be accepting filters which have the FLOW_EXT bit set (and thus make use of the user-def field).

Change-Id 238845035c179380a347baa8db8223304f5f6dd7
---
 drivers/net/ethernet/intel/i40e/i40e.h         |   9 +++
 drivers/net/ethernet/intel/i40e/i40e_ethtool.c | 108 ++++++++++++++++++++++++-
 2 files changed, 116 insertions(+), 1 deletion(-)
drivers/net/ethernet/intel/i40e/i40e.h +9 -0
index 4d54d5005ad3..bffcfc1041e7 100644
@@ -206,6 +206,15 @@ enum i40e_fd_stat_idx {
 #define I40E_FD_ATR_TUNNEL_STAT_IDX(pf_id) \
 			(I40E_FD_STAT_PF_IDX(pf_id) + I40E_FD_STAT_ATR_TUNNEL)
 
+/* The following structure contains the data parsed from the user-defined
+ * field of the ethtool_rx_flow_spec structure.
+ */
+struct i40e_rx_flow_userdef {
+	bool flex_filter;
+	u16 flex_word;
+	u16 flex_offset;
+};
+
 struct i40e_fdir_filter {
 	struct hlist_node fdir_node;
 	/* filter ipnut set */
drivers/net/ethernet/intel/i40e/i40e_ethtool.c +107 -1

144 lines not on this page · raw message

Signature · 1 line
2.12.0.rc0.151.g8a5726c42288
#12 lore raw

PATCH 6/9 i40e: add parsing of flexible filter fields from userdef

Quoted 26 lines

-----Original Message-----
From: Intel-wired-lan [mailto:intel-wired-lan-bounces at lists.osuosl.org] On Behalf Of Jacob Keller
Sent: Monday, February 6, 2017 2:39 PM
To: Intel Wired LAN <[email protected]>
Subject: [Intel-wired-lan] [PART2 PATCH 6/9] i40e: add parsing of flexible filter fields from userdef

Add code to parse the user-def field into a data structure format. This code is intended to allow future extensions of the user-def field by keeping all code that actually reads and writes the field into a single location. This ensures that we do not litter the driver with references to the user-def field and minimizes the amount of bitwise operations we need to do on the data.

Add code which parses the lower 32bits into a flexible word and its offset. This will be used in a future patch to enable flexible filters which can match on some arbitrary data in the packet payload. For now, we just return - EOPNOTSUPP when this is used.

Add code to fill in the user-def field when reporting the filter back, even though we don't actually implement any user-def fields yet.

Additionally, ensure that we mask the extended FLOW_EXT bit from the flow_type now that we will be accepting filters which have the FLOW_EXT bit set (and thus make use of the user-def field).

Quoted 7 lines
Change-Id 238845035c179380a347baa8db8223304f5f6dd7
---
 drivers/net/ethernet/intel/i40e/i40e.h         |   9 +++
 drivers/net/ethernet/intel/i40e/i40e_ethtool.c | 108
++++++++++++++++++++++++-
 2 files changed, 116 insertions(+), 1 deletion(-)
#13 lore raw

PATCH 7/9 i40e: implement support for flexible word payload

In reply to Jacob Keller

Add support for flexible payloads passed via ethtool user-def field. This support is somewhat limited due to hardware design. The input set can only be programmed once per filter type, and the flexible offset is part of this filter input set. This means that the user cannot program both a regular and a flexible filter at the same time for a given flow type. Additionally, the user may not program two flexible filters of the same flow type with different offsets, although they are allowed to configure different values at that offset location.

We support a single flexible word (2byte) value per protocol type, and we handle the FLX_PIT register using a list of flexible entries so that each flow type may be configured separately.

Due to hardware implementation, the flexible data is offset from the start of the packet payload, and thus may not be in part of the header data. For this reason, the offset provided by the user defined data is interpreted as a byte offset from the start of the matching payload. Previous implementations have tried to represent the offset as from the start of the frame, but this is not feasible because header sizes may change due to options.

Testing-hints:
Need to ensure that multiple filters with same flex offset behave correctly, and ensure that different L4 protocols inter-operate correctly. README changes for this will be forth-coming.

Change-Id 36ed27995e97de63f9aea5ade5778ff038d6f811
---
 drivers/net/ethernet/intel/i40e/i40e.h         |  83 +++++
 drivers/net/ethernet/intel/i40e/i40e_ethtool.c | 494 ++++++++++++++++++++++++-
 drivers/net/ethernet/intel/i40e/i40e_main.c    |  16 +
 drivers/net/ethernet/intel/i40e/i40e_txrx.c    |  27 ++
 4 files changed, 608 insertions(+), 12 deletions(-)
drivers/net/ethernet/intel/i40e/i40e.h +83 -0
index bffcfc1041e7..f218a5bb7a16 100644
@@ -226,6 +226,12 @@ struct i40e_fdir_filter {
 	__be16 src_port;
 	__be16 dst_port;
 	__be32 sctp_v_tag;
+
+	/* Flexible data to match within the packet payload */
+	__be16 flex_word;
+	u16 flex_offset;
+	bool flex_filter;
+
 	/* filter control */
 	u16 q_index;
 	u8  flex_off;
@@ -262,6 +268,75 @@ struct i40e_udp_port_config {
 	u8 type;
 };
 
+/* macros related to FLX_PIT */
+#define I40E_FLEX_SET_FSIZE(fsize) (((fsize) << \
+				    I40E_PRTQF_FLX_PIT_FSIZE_SHIFT) & \
+				    I40E_PRTQF_FLX_PIT_FSIZE_MASK)
+#define I40E_FLEX_SET_DST_WORD(dst) (((dst) << \
+				     I40E_PRTQF_FLX_PIT_DEST_OFF_SHIFT) & \
+				     I40E_PRTQF_FLX_PIT_DEST_OFF_MASK)
+#define I40E_FLEX_SET_SRC_WORD(src) (((src) << \
+				     I40E_PRTQF_FLX_PIT_SOURCE_OFF_SHIFT) & \
+				     I40E_PRTQF_FLX_PIT_SOURCE_OFF_MASK)
+#define I40E_FLEX_PREP_VAL(dst, fsize, src) (I40E_FLEX_SET_DST_WORD(dst) | \
+					     I40E_FLEX_SET_FSIZE(fsize) | \
+					     I40E_FLEX_SET_SRC_WORD(src))
+
+#define I40E_FLEX_PIT_GET_SRC(flex) (((flex) & \
+				     I40E_PRTQF_FLX_PIT_SOURCE_OFF_MASK) >> \
+				     I40E_PRTQF_FLX_PIT_SOURCE_OFF_SHIFT)
+#define I40E_FLEX_PIT_GET_DST(flex) (((flex) & \
+				     I40E_PRTQF_FLX_PIT_DEST_OFF_MASK) >> \
+				     I40E_PRTQF_FLX_PIT_DEST_OFF_SHIFT)
+#define I40E_FLEX_PIT_GET_FSIZE(flex) (((flex) & \
+				       I40E_PRTQF_FLX_PIT_FSIZE_MASK) >> \
+				       I40E_PRTQF_FLX_PIT_FSIZE_SHIFT)
+
+#define I40E_MAX_FLEX_SRC_OFFSET 0x1F
+
+/* macros related to GLQF_ORT */
+#define I40E_ORT_SET_IDX(idx)		(((idx) << \
+					  I40E_GLQF_ORT_PIT_INDX_SHIFT) & \
+					 I40E_GLQF_ORT_PIT_INDX_MASK)
+
+#define I40E_ORT_SET_COUNT(count)	(((count) << \
+					  I40E_GLQF_ORT_FIELD_CNT_SHIFT) & \
+					 I40E_GLQF_ORT_FIELD_CNT_MASK)
+
+#define I40E_ORT_SET_PAYLOAD(payload)	(((payload) << \
+					  I40E_GLQF_ORT_FLX_PAYLOAD_SHIFT) & \
+					 I40E_GLQF_ORT_FLX_PAYLOAD_MASK)
+
+#define I40E_ORT_PREP_VAL(idx, count, payload) (I40E_ORT_SET_IDX(idx) | \
+						I40E_ORT_SET_COUNT(count) | \
+						I40E_ORT_SET_PAYLOAD(payload))
+
+#define I40E_L3_GLQF_ORT_IDX		34
+#define I40E_L4_GLQF_ORT_IDX		35
+
+/* Flex PIT register index */
+#define I40E_FLEX_PIT_IDX_START_L2	0
+#define I40E_FLEX_PIT_IDX_START_L3	3
+#define I40E_FLEX_PIT_IDX_START_L4	6
+
+#define I40E_FLEX_PIT_TABLE_SIZE	3
+
+#define I40E_FLEX_DEST_UNUSED		63
+
+#define I40E_FLEX_INDEX_ENTRIES		8
+
+/* Flex MASK to disable all flexible entries */
+#define I40E_FLEX_INPUT_MASK	(I40E_FLEX_50_MASK | I40E_FLEX_51_MASK | \
+				 I40E_FLEX_52_MASK | I40E_FLEX_53_MASK | \
+				 I40E_FLEX_54_MASK | I40E_FLEX_55_MASK | \
+				 I40E_FLEX_56_MASK | I40E_FLEX_57_MASK)
+
+struct i40e_flex_pit {
+	struct list_head list;
+	u16 src_offset;
+	u8 pit_index;
+};
+
 /* struct that defines the Ethernet device */
 struct i40e_pf {
 	struct pci_dev *pdev;
@@ -308,6 +383,14 @@ struct i40e_pf {
 	u16 fd_udp4_filter_cnt;
 	u16 fd_ip4_filter_cnt;
 
+	/* Flexible filter table values that need to be programmed into
+	 * hardware, which expects L3 and L4 to be programmed separately. We
+	 * need to ensure that the values are in ascended order and don't have
+	 * duplicates, so we track each L3 and L4 values in separate lists.
+	 */
+	struct list_head l3_flex_pit_list;
+	struct list_head l4_flex_pit_list;
+
 	struct i40e_udp_port_config udp_ports[I40E_MAX_PF_UDP_OFFLOAD_PORTS];
 	u16 pending_udp_bitmap;
 
drivers/net/ethernet/intel/i40e/i40e_ethtool.c +482 -12

601 lines not on this page · raw message

drivers/net/ethernet/intel/i40e/i40e_main.c +16 -0
index 2c3a427f5b75..a2cd3979b81a 100644
@@ -5888,6 +5888,7 @@ int i40e_vsi_open(struct i40e_vsi *vsi)
 static void i40e_fdir_filter_exit(struct i40e_pf *pf)
 {
 	struct i40e_fdir_filter *filter;
+	struct i40e_flex_pit *pit_entry, *tmp;
 	struct hlist_node *node2;
 
 	hlist_for_each_entry_safe(filter, node2,
@@ -5896,6 +5897,18 @@ static void i40e_fdir_filter_exit(struct i40e_pf *pf)
 		kfree(filter);
 	}
 
+	list_for_each_entry_safe(pit_entry, tmp, &pf->l3_flex_pit_list, list) {
+		list_del(&pit_entry->list);
+		kfree(pit_entry);
+	}
+	INIT_LIST_HEAD(&pf->l3_flex_pit_list);
+
+	list_for_each_entry_safe(pit_entry, tmp, &pf->l4_flex_pit_list, list) {
+		list_del(&pit_entry->list);
+		kfree(pit_entry);
+	}
+	INIT_LIST_HEAD(&pf->l4_flex_pit_list);
+
 	pf->fdir_pf_active_filters = 0;
 	pf->fd_tcp4_filter_cnt = 0;
 	pf->fd_udp4_filter_cnt = 0;
@@ -11524,6 +11537,9 @@ static int i40e_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 	hw->bus.bus_id = pdev->bus->number;
 	pf->instance = pfs_found;
 
+	INIT_LIST_HEAD(&pf->l3_flex_pit_list);
+	INIT_LIST_HEAD(&pf->l4_flex_pit_list);
+
 	/* set up the locks for the AQ, do this only once in probe
 	 * and destroy them only once in remove
 	 */
drivers/net/ethernet/intel/i40e/i40e_txrx.c +27 -0
index 55ac87c1e40b..cb38a248c38f 100644
@@ -72,6 +72,9 @@ static void i40e_fdir(struct i40e_ring *tx_ring,
 	flex_ptype |= I40E_TXD_FLTR_QW0_PCTYPE_MASK &
 		      (fdata->pctype << I40E_TXD_FLTR_QW0_PCTYPE_SHIFT);
 
+	flex_ptype |= I40E_TXD_FLTR_QW0_PCTYPE_MASK &
+		      (fdata->flex_offset << I40E_TXD_FLTR_QW0_FLEXOFF_SHIFT);
+
 	/* Use LAN VSI Id if not programmed by user */
 	flex_ptype |= I40E_TXD_FLTR_QW0_DEST_VSI_MASK &
 		      ((u32)(fdata->dest_vsi ? : pf->vsi[pf->lan_vsi]->id) <<
@@ -224,6 +227,14 @@ static int i40e_add_del_fdir_udpv4(struct i40e_vsi *vsi,
 	ip->saddr = fd_data->src_ip;
 	udp->source = fd_data->src_port;
 
+	if (fd_data->flex_filter) {
+		u8 *payload = raw_packet + I40E_UDPIP_DUMMY_PACKET_LEN;
+		__be16 pattern = fd_data->flex_word;
+		u16 off = fd_data->flex_offset;
+
+		*((__force __be16 *)(payload + off)) = pattern;
+	}
+
 	fd_data->pctype = I40E_FILTER_PCTYPE_NONF_IPV4_UDP;
 	ret = i40e_program_fdir_filter(fd_data, raw_packet, pf, add);
 	if (ret) {
@@ -290,6 +301,14 @@ static int i40e_add_del_fdir_tcpv4(struct i40e_vsi *vsi,
 	ip->saddr = fd_data->src_ip;
 	tcp->source = fd_data->src_port;
 
+	if (fd_data->flex_filter) {
+		u8 *payload = raw_packet + I40E_TCPIP_DUMMY_PACKET_LEN;
+		__be16 pattern = fd_data->flex_word;
+		u16 off = fd_data->flex_offset;
+
+		*((__force __be16 *)(payload + off)) = pattern;
+	}
+
 	fd_data->pctype = I40E_FILTER_PCTYPE_NONF_IPV4_TCP;
 	ret = i40e_program_fdir_filter(fd_data, raw_packet, pf, add);
 	if (ret) {
@@ -363,6 +382,14 @@ static int i40e_add_del_fdir_ipv4(struct i40e_vsi *vsi,
 		ip->daddr = fd_data->dst_ip;
 		ip->protocol = 0;
 
+		if (fd_data->flex_filter) {
+			u8 *payload = raw_packet + I40E_IP_DUMMY_PACKET_LEN;
+			__be16 pattern = fd_data->flex_word;
+			u16 off = fd_data->flex_offset;
+
+			*((__force __be16 *)(payload + off)) = pattern;
+		}
+
 		fd_data->pctype = i;
 		ret = i40e_program_fdir_filter(fd_data, raw_packet, pf, add);
 		if (ret) {
Signature · 1 line
2.12.0.rc0.151.g8a5726c42288
#14 lore raw

PATCH 7/9 i40e: implement support for flexible word payload

Quoted 36 lines

-----Original Message-----
From: Intel-wired-lan [mailto:intel-wired-lan-bounces at lists.osuosl.org] On Behalf Of Jacob Keller
Sent: Monday, February 6, 2017 2:39 PM
To: Intel Wired LAN <[email protected]>
Subject: [Intel-wired-lan] [PART2 PATCH 7/9] i40e: implement support for flexible word payload

Add support for flexible payloads passed via ethtool user-def field. This support is somewhat limited due to hardware design. The input set can only be programmed once per filter type, and the flexible offset is part of this filter input set. This means that the user cannot program both a regular and a flexible filter at the same time for a given flow type. Additionally, the user may not program two flexible filters of the same flow type with different offsets, although they are allowed to configure different values at that offset location.

We support a single flexible word (2byte) value per protocol type, and we handle the FLX_PIT register using a list of flexible entries so that each flow type may be configured separately.

Due to hardware implementation, the flexible data is offset from the start of the packet payload, and thus may not be in part of the header data. For this reason, the offset provided by the user defined data is interpreted as a byte offset from the start of the matching payload.
Previous implementations have tried to represent the offset as from the start of the frame, but this is not feasible because header sizes may change due to options.

Testing-hints:
Need to ensure that multiple filters with same flex offset behave correctly, and ensure that different L4 protocols inter-operate correctly. README changes for this will be forth-coming.

Change-Id 36ed27995e97de63f9aea5ade5778ff038d6f811
Quoted 7 lines
---
 drivers/net/ethernet/intel/i40e/i40e.h         |  83 +++++
 drivers/net/ethernet/intel/i40e/i40e_ethtool.c | 494
++++++++++++++++++++++++-
 drivers/net/ethernet/intel/i40e/i40e_main.c    |  16 +
 drivers/net/ethernet/intel/i40e/i40e_txrx.c    |  27 ++
 4 files changed, 608 insertions(+), 12 deletions(-)
#15 lore raw

PATCH 8/9 i40e: add support for SCTPv4 FDir filters

In reply to Jacob Keller

Enable FDir filters for SCTPv4 packets using the ethtool ntuple interface to enable filters. The ethtool API does not allow masking on the verification tag.

Testing-hints:
Ethtool commands to enable sctp4 filters should not produce errors, and should properly direct traffic to the desired queue.

ethtool -N <device> flow-type sctp4 <input set> action 1

Change-Id I093e88a8143994c7e6f4b7b17a0bd5cf861d18e4
---
 drivers/net/ethernet/intel/i40e/i40e.h         |  1 +
 drivers/net/ethernet/intel/i40e/i40e_ethtool.c | 10 ++++
 drivers/net/ethernet/intel/i40e/i40e_main.c    |  2 +
 drivers/net/ethernet/intel/i40e/i40e_txrx.c    | 80 ++++++++++++++++++++++++++
 4 files changed, 93 insertions(+)
drivers/net/ethernet/intel/i40e/i40e.h +1 -0
index f218a5bb7a16..33d1730878b7 100644
@@ -381,6 +381,7 @@ struct i40e_pf {
 	 */
 	u16 fd_tcp4_filter_cnt;
 	u16 fd_udp4_filter_cnt;
+	u16 fd_sctp4_filter_cnt;
 	u16 fd_ip4_filter_cnt;
 
 	/* Flexible filter table values that need to be programmed into
drivers/net/ethernet/intel/i40e/i40e_ethtool.c +10 -0
index 2dce5f5a486e..d9b379ef203b 100644
@@ -2469,6 +2469,9 @@ static int i40e_get_ethtool_fdir_entry(struct i40e_pf *pf,
 	fsp->h_u.tcp_ip4_spec.ip4dst = rule->src_ip;
 
 	switch (rule->flow_type) {
+	case SCTP_V4_FLOW:
+		index = I40E_FILTER_PCTYPE_NONF_IPV4_SCTP;
+		break;
 	case TCP_V4_FLOW:
 		index = I40E_FILTER_PCTYPE_NONF_IPV4_TCP;
 		break;
@@ -3291,6 +3294,10 @@ static int i40e_check_fdir_input_set(struct i40e_vsi *vsi,
 	int err;
 
 	switch (fsp->flow_type & ~FLOW_EXT) {
+	case SCTP_V4_FLOW:
+		index = I40E_FILTER_PCTYPE_NONF_IPV4_SCTP;
+		fdir_filter_count = &pf->fd_sctp4_filter_cnt;
+		break;
 	case TCP_V4_FLOW:
 		index = I40E_FILTER_PCTYPE_NONF_IPV4_TCP;
 		fdir_filter_count = &pf->fd_tcp4_filter_cnt;
@@ -3322,6 +3329,9 @@ static int i40e_check_fdir_input_set(struct i40e_vsi *vsi,
 	 * ip4dst fields.
 	 */
 	switch (fsp->flow_type & ~FLOW_EXT) {
+	case SCTP_V4_FLOW:
+		new_mask &= ~I40E_VERIFY_TAG_MASK;
+		/* Fall through */
 	case TCP_V4_FLOW:
 	case UDP_V4_FLOW:
 		tcp_ip4_spec = &fsp->m_u.tcp_ip4_spec;
drivers/net/ethernet/intel/i40e/i40e_main.c +2 -0
index a2cd3979b81a..6223db6ac728 100644
@@ -3323,6 +3323,7 @@ static void i40e_fdir_filter_restore(struct i40e_vsi *vsi)
 	/* Reset FDir counters as we're replaying all existing filters */
 	pf->fd_tcp4_filter_cnt = 0;
 	pf->fd_udp4_filter_cnt = 0;
+	pf->fd_sctp4_filter_cnt = 0;
 	pf->fd_ip4_filter_cnt = 0;
 
 	hlist_for_each_entry_safe(filter, node,
@@ -5912,6 +5913,7 @@ static void i40e_fdir_filter_exit(struct i40e_pf *pf)
 	pf->fdir_pf_active_filters = 0;
 	pf->fd_tcp4_filter_cnt = 0;
 	pf->fd_udp4_filter_cnt = 0;
+	pf->fd_sctp4_filter_cnt = 0;
 	pf->fd_ip4_filter_cnt = 0;
 
 	/* Reprogram the default input set for TCP/IPv4 */
drivers/net/ethernet/intel/i40e/i40e_txrx.c +80 -0

98 lines not on this page · raw message

Signature · 1 line
2.12.0.rc0.151.g8a5726c42288
#16 lore raw

PATCH 8/9 i40e: add support for SCTPv4 FDir filters

Quoted 19 lines

-----Original Message-----
From: Intel-wired-lan [mailto:intel-wired-lan-bounces at lists.osuosl.org] On Behalf Of Jacob Keller
Sent: Monday, February 6, 2017 2:39 PM
To: Intel Wired LAN <[email protected]>
Subject: [Intel-wired-lan] [PART2 PATCH 8/9] i40e: add support for SCTPv4 FDir filters

Enable FDir filters for SCTPv4 packets using the ethtool ntuple interface to enable filters. The ethtool API does not allow masking on the verification tag.

Testing-hints:
Ethtool commands to enable sctp4 filters should not produce errors, and should properly direct traffic to the desired queue.

ethtool -N <device> flow-type sctp4 <input set> action 1

Change-Id I093e88a8143994c7e6f4b7b17a0bd5cf861d18e4
Quoted 7 lines
---
 drivers/net/ethernet/intel/i40e/i40e.h         |  1 +
 drivers/net/ethernet/intel/i40e/i40e_ethtool.c | 10 ++++
 drivers/net/ethernet/intel/i40e/i40e_main.c    |  2 +
 drivers/net/ethernet/intel/i40e/i40e_txrx.c    | 80
++++++++++++++++++++++++++
 4 files changed, 93 insertions(+)
#17 lore raw

PATCH 9/9 i40e: document drivers use of ntuple filters

In reply to Jacob Keller

Add documentation describing the drivers use of ethtool ntuple filters, including the limitations that it has due to hardware, as well as how it reads and parses the user-def data block.

---
 Documentation/networking/i40e.txt | 72 +++++++++++++++++++++++++++++++++++++++
 1 file changed, 72 insertions(+)
Documentation/networking/i40e.txt +72 -0
index a251bf4fe9c9..57e616ed10b0 100644
@@ -63,6 +63,78 @@ Additional Configurations
   The latest release of ethtool can be found from
   https://www.kernel.org/pub/software/network/ethtool
 
+
+  Flow Director n-ntuple traffic filters (FDir)
+  ---------------------------------------------
+  The driver utilizes the ethtool interface for configuring ntuple filters,
+  via "ethtool -N <device> <filter>".
+
+  The sctp4, ip4, udp4, and tcp4 flow types are supported with the standard
+  fields including src-ip, dst-ip, src-port and dst-port. The driver only
+  supports fully enabling or fully masking the fields, so use of the mask
+  fields for partial matches is not supported.
+
+  Additionally, the driver supports using the action to specify filters for a
+  Virtual Function. You can specify the action as a 64bit value, where the
+  lower 32 bits represents the queue number, while the next 8 bits represent
+  which VF. Note that 0 is the PF, so the VF identifier is offset by 1. For
+  example:
+
+    ... action 0x800000002 ...
+
+  Would indicate to direct traffic for Virtual Function 7 (8 minus 1) on queue
+  2 of that VF.
+
+  The driver also supports using the user-defined field to specify 2 bytes of
+  arbitrary data to match within the packet payload in addition to the regular
+  fields. The data is specified in the lower 32bits of the user-def field in
+  the following way:
+
+  +----------------------------+---------------------------+
+  | 31    28    24    20    16 | 15    12     8     4     0|
+  +----------------------------+---------------------------+
+  | offset into packet payload |  2 bytes of flexible data |
+  +----------------------------+---------------------------+
+
+  As an example,
+
+    ... user-def 0x4FFFF ....
+
+  means to match the value 0xFFFF 4 bytes into the packet payload. Note that
+  the offset is based on the beginning of the payload, and not the beginning
+  of the packet. Thus
+
+    flow-type tcp4 ... user-def 0x8BEAF ....
+
+  would match TCP/IPv4 packets which have the value 0xBEAF 8bytes into the
+  TCP/IPv4 payload.
+
+  For ICMP, the hardware parses the ICMP header as 4 bytes of header and 4
+  bytes of payload, so if you want to match an ICMP frames payload you may need
+  to add 4 to the offset in order to match the data.
+
+  Furthermore, the offset can only be up to a value of 64, as the hardware
+  will only read up to 64 bytes of data from the payload. It must also be even
+  as the flexible data is 2 bytes long and must be aligned to byte 0 of the
+  packet payload.
+
+  When programming filters, the hardware is limited to using a single input
+  set for each flow type. This means that it is an error to program two
+  different filters with the same type that don't match on the same fields.
+  Thus the second of the following two commands will fail:
+
+    ethtool -N <device> flow-type tcp4 src-ip 192.168.0.7 action 5
+    ethtool -N <device> flow-type tcp4 dst-ip 192.168.15.18 action 1
+
+  This is because the first filter will be accepted and reprogram the input
+  set for TCPv4 filters, but the second filter will be unable to reprogram the
+  input set until all the conflicting TCPv4 filters are first removed.
+
+  Note that the user-defined flexible offset is also considered part of the
+  input set and cannot be programmed separately for multiple filters of the
+  same type. However, the flexible data is not part of the input set and
+  multiple filters may use the same offset but match against different data.
+
   Data Center Bridging (DCB)
   --------------------------
   DCB configuration is not currently supported.
Signature · 1 line
2.12.0.rc0.151.g8a5726c42288
#18 lore raw

PATCH 9/9 i40e: document drivers use of ntuple filters

Quoted 12 lines

-----Original Message-----
From: Intel-wired-lan [mailto:intel-wired-lan-bounces at lists.osuosl.org] On Behalf Of Jacob Keller
Sent: Monday, February 6, 2017 2:39 PM
To: Intel Wired LAN <[email protected]>
Subject: [Intel-wired-lan] [PART2 PATCH 9/9] i40e: document drivers use of ntuple filters

Add documentation describing the drivers use of ethtool ntuple filters, including the limitations that it has due to hardware, as well as how it reads and parses the user-def data block.

Quoted 5 lines
---
 Documentation/networking/i40e.txt | 72
+++++++++++++++++++++++++++++++++++++++
 1 file changed, 72 insertions(+)