DPDK patches and discussions
 help / color / mirror / Atom feed
From: Qi Zhang <qi.z.zhang@intel.com>
To: ferruh.yigit@intel.com
Cc: dev@dpdk.org, beilei.xing@intel.com,
	Qi Zhang <qi.z.zhang@intel.com>, Ting Xu <ting.xu@intel.com>,
	Brett Creeley <brett.creeley@intel.com>
Subject: [dpdk-dev] [PATCH v2 2/2] common/iavf: add support for VFs with >16 queue pairs
Date: Fri, 16 Oct 2020 18:41:28 +0800
Message-ID: <20201016104128.1049430-3-qi.z.zhang@intel.com> (raw)
In-Reply-To: <20201016104128.1049430-1-qi.z.zhang@intel.com>

Currently there are limitations in the virthcnl.h interface that only
allow a maximum of 16 queues to be used by a VF driver. Add support in
virtchnl.h to allow a VF driver to request >16 queues. Also, the RSS
qregion size is currently assumed to be the max number of queues a VF
can request and/or is given on initialization. With larger VFs this
assumption can no longer be made, so add a new op to support querying
the max RSS qregion size.

In order to request more queues than the initially given queues the VF
driver needs to use the VIRTCHNL_OP_REQUEST_QUEUES opcode.

The VF is given more >16 queues it should use the new
VIRTCHNL_OP_GET_MAX_RSS_QREGION to determine its max qregion size. This
is needed to correctly configure the RSS LUT and/or configure filters
based on queue base/offset and queue region size.

If the VF is configuring >16 queues it should use the following opcodes
to configure the qeueus and interrupts after successfully requesting
them.

VIRTCHNL_OP_MAP_QUEUE_VECTOR
VIRTCHNL_OP_ENABLE_QUEUES_V2
VIRTCHNL_OP_DISABLE_QUEUES_V2

Also, add support in virtchnl_vc_validate_vf_msg() to validate the above
messages. As a part of this move the virtchnl_vector_limits enumeration
directly above the function it's used.

The patch also update base code release version in readme.

Signed-off-by: Ting Xu <ting.xu@intel.com>
Signed-off-by: Brett Creeley <brett.creeley@intel.com>
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Acked-by: Beilei Xing <beilei.xing@intel.com>
---
 drivers/common/iavf/README     |   2 +-
 drivers/common/iavf/virtchnl.h | 231 +++++++++++++++++++++++++++++----
 2 files changed, 205 insertions(+), 28 deletions(-)

diff --git a/drivers/common/iavf/README b/drivers/common/iavf/README
index f301ff665a..5a42750465 100644
--- a/drivers/common/iavf/README
+++ b/drivers/common/iavf/README
@@ -6,7 +6,7 @@ Intel® IAVF driver
 =================
 
 This directory contains source code of FreeBSD IAVF driver of version
-cid-avf.2020.04.09.tar.gz released by the team which develops
+cid-avf.2020.10.14.tar.gz released by the team which develops
 basic drivers for any IAVF NIC. The directory of base/ contains the
 original source package.
 
diff --git a/drivers/common/iavf/virtchnl.h b/drivers/common/iavf/virtchnl.h
index a5a300afe3..b931da61e5 100644
--- a/drivers/common/iavf/virtchnl.h
+++ b/drivers/common/iavf/virtchnl.h
@@ -140,6 +140,11 @@ enum virtchnl_ops {
 	VIRTCHNL_OP_ADD_FDIR_FILTER = 47,
 	VIRTCHNL_OP_DEL_FDIR_FILTER = 48,
 	VIRTCHNL_OP_QUERY_FDIR_FILTER = 49,
+	VIRTCHNL_OP_GET_MAX_RSS_QREGION = 50,
+	VIRTCHNL_OP_ENABLE_QUEUES_V2 = 107,
+	VIRTCHNL_OP_DISABLE_QUEUES_V2 = 108,
+	VIRTCHNL_OP_MAP_QUEUE_VECTOR = 111,
+	VIRTCHNL_OP_MAX,
 };
 
 /* These macros are used to generate compilation errors if a structure/union
@@ -244,6 +249,8 @@ VIRTCHNL_CHECK_STRUCT_LEN(16, virtchnl_vsi_resource);
 #define VIRTCHNL_VF_OFFLOAD_WB_ON_ITR		0x00000020
 #define VIRTCHNL_VF_OFFLOAD_REQ_QUEUES		0x00000040
 #define VIRTCHNL_VF_OFFLOAD_CRC			0x00000080
+	/* 0X00000100 is reserved */
+#define VIRTCHNL_VF_LARGE_NUM_QPAIRS		0x00000200
 #define VIRTCHNL_VF_OFFLOAD_VLAN		0x00010000
 #define VIRTCHNL_VF_OFFLOAD_RX_POLLING		0x00020000
 #define VIRTCHNL_VF_OFFLOAD_RSS_PCTYPE_V2	0x00040000
@@ -424,6 +431,35 @@ struct virtchnl_queue_select {
 
 VIRTCHNL_CHECK_STRUCT_LEN(12, virtchnl_queue_select);
 
+/* VIRTCHNL_OP_GET_MAX_RSS_QREGION
+ *
+ * if VIRTCHNL_VF_LARGE_NUM_QPAIRS was negotiated in VIRTCHNL_OP_GET_VF_RESOURCES
+ * then this op must be supported.
+ *
+ * VF sends this message in order to query the max RSS queue region
+ * size supported by PF, when VIRTCHNL_VF_LARGE_NUM_QPAIRS is enabled.
+ * This information should be used when configuring the RSS LUT and/or
+ * configuring queue region based filters.
+ *
+ * The maximum RSS queue region is 2^qregion_width. So, a qregion_width
+ * of 6 would inform the VF that the PF supports a maximum RSS queue region
+ * of 64.
+ *
+ * A queue region represents a range of queues that can be used to configure
+ * a RSS LUT. For example, if a VF is given 64 queues, but only a max queue
+ * region size of 16 (i.e. 2^qregion_width = 16) then it will only be able
+ * to configure the RSS LUT with queue indices from 0 to 15. However, other
+ * filters can be used to direct packets to queues >15 via specifying a queue
+ * base/offset and queue region width.
+ */
+struct virtchnl_max_rss_qregion {
+	u16 vport_id;
+	u16 qregion_width;
+	u8 pad[4];
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(8, virtchnl_max_rss_qregion);
+
 /* VIRTCHNL_OP_ADD_ETH_ADDR
  * VF sends this message in order to add one or more unicast or multicast
  * address filters for the specified VSI.
@@ -774,33 +810,6 @@ struct virtchnl_pf_event {
 VIRTCHNL_CHECK_STRUCT_LEN(16, virtchnl_pf_event);
 
 
-/* Since VF messages are limited by u16 size, precalculate the maximum possible
- * values of nested elements in virtchnl structures that virtual channel can
- * possibly handle in a single message.
- */
-enum virtchnl_vector_limits {
-	VIRTCHNL_OP_CONFIG_VSI_QUEUES_MAX	=
-		((u16)(~0) - sizeof(struct virtchnl_vsi_queue_config_info)) /
-		sizeof(struct virtchnl_queue_pair_info),
-
-	VIRTCHNL_OP_CONFIG_IRQ_MAP_MAX		=
-		((u16)(~0) - sizeof(struct virtchnl_irq_map_info)) /
-		sizeof(struct virtchnl_vector_map),
-
-	VIRTCHNL_OP_ADD_DEL_ETH_ADDR_MAX	=
-		((u16)(~0) - sizeof(struct virtchnl_ether_addr_list)) /
-		sizeof(struct virtchnl_ether_addr),
-
-	VIRTCHNL_OP_ADD_DEL_VLAN_MAX		=
-		((u16)(~0) - sizeof(struct virtchnl_vlan_filter_list)) /
-		sizeof(u16),
-
-
-	VIRTCHNL_OP_ENABLE_CHANNELS_MAX		=
-		((u16)(~0) - sizeof(struct virtchnl_tc_info)) /
-		sizeof(struct virtchnl_channel_info),
-};
-
 /* VF reset states - these are written into the RSTAT register:
  * VFGEN_RSTAT on the VF
  * When the PF initiates a reset, it writes 0
@@ -1142,6 +1151,143 @@ struct virtchnl_fdir_query {
 };
 
 VIRTCHNL_CHECK_STRUCT_LEN(48, virtchnl_fdir_query);
+
+/* TX and RX queue types are valid in legacy as well as split queue models.
+ * With Split Queue model, 2 additional types are introduced - TX_COMPLETION
+ * and RX_BUFFER. In split queue model, RX corresponds to the queue where HW
+ * posts completions.
+ */
+enum virtchnl_queue_type {
+	VIRTCHNL_QUEUE_TYPE_TX			= 0,
+	VIRTCHNL_QUEUE_TYPE_RX			= 1,
+	VIRTCHNL_QUEUE_TYPE_TX_COMPLETION	= 2,
+	VIRTCHNL_QUEUE_TYPE_RX_BUFFER		= 3,
+	VIRTCHNL_QUEUE_TYPE_CONFIG_TX		= 4,
+	VIRTCHNL_QUEUE_TYPE_CONFIG_RX		= 5
+};
+
+
+/* structure to specify a chunk of contiguous queues */
+struct virtchnl_queue_chunk {
+	enum virtchnl_queue_type type;
+	u16 start_queue_id;
+	u16 num_queues;
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(8, virtchnl_queue_chunk);
+
+/* structure to specify several chunks of contiguous queues */
+struct virtchnl_queue_chunks {
+	u16 num_chunks;
+	u16 rsvd;
+	struct virtchnl_queue_chunk chunks[1];
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(12, virtchnl_queue_chunks);
+
+
+/* VIRTCHNL_OP_ENABLE_QUEUES_V2
+ * VIRTCHNL_OP_DISABLE_QUEUES_V2
+ * VIRTCHNL_OP_DEL_QUEUES
+ *
+ * If VIRTCHNL_CAP_EXT_FEATURES was negotiated in VIRTCHNL_OP_GET_VF_RESOURCES
+ * then all of these ops are available.
+ *
+ * If VIRTCHNL_VF_LARGE_NUM_QPAIRS was negotiated in VIRTCHNL_OP_GET_VF_RESOURCES
+ * then VIRTCHNL_OP_ENABLE_QUEUES_V2 and VIRTCHNL_OP_DISABLE_QUEUES_V2 are
+ * available.
+ *
+ * PF sends these messages to enable, disable or delete queues specified in
+ * chunks. PF sends virtchnl_del_ena_dis_queues struct to specify the queues
+ * to be enabled/disabled/deleted. Also applicable to single queue RX or
+ * TX. CP performs requested action and returns status.
+ */
+struct virtchnl_del_ena_dis_queues {
+	u16 vport_id;
+	u16 pad;
+	struct virtchnl_queue_chunks chunks;
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(16, virtchnl_del_ena_dis_queues);
+
+/* Virtchannel interrupt throttling rate index */
+enum virtchnl_itr_idx {
+	VIRTCHNL_ITR_IDX_0	= 0,
+	VIRTCHNL_ITR_IDX_1	= 1,
+	VIRTCHNL_ITR_IDX_NO_ITR	= 3,
+};
+
+/* Queue to vector mapping */
+struct virtchnl_queue_vector {
+	u16 queue_id;
+	u16 vector_id;
+	u8 pad[4];
+	enum virtchnl_itr_idx itr_idx;
+	enum virtchnl_queue_type queue_type;
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(16, virtchnl_queue_vector);
+
+/* VIRTCHNL_OP_MAP_QUEUE_VECTOR
+ * VIRTCHNL_OP_UNMAP_QUEUE_VECTOR
+ *
+ * If VIRTCHNL_CAP_EXT_FEATURES was negotiated in VIRTCHNL_OP_GET_VF_RESOURCES
+ * then all of these ops are available.
+ *
+ * If VIRTCHNL_VF_LARGE_NUM_QPAIRS was negotiated in VIRTCHNL_OP_GET_VF_RESOURCES
+ * then only VIRTCHNL_OP_MAP_QUEUE_VECTOR is available.
+ *
+ * PF sends this message to map or unmap queues to vectors and ITR index
+ * registers. External data buffer contains virtchnl_queue_vector_maps structure
+ * that contains num_qv_maps of virtchnl_queue_vector structures.
+ * CP maps the requested queue vector maps after validating the queue and vector
+ * ids and returns a status code.
+ */
+struct virtchnl_queue_vector_maps {
+	u16 vport_id;
+	u16 num_qv_maps;
+	u8 pad[4];
+	struct virtchnl_queue_vector qv_maps[1];
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(24, virtchnl_queue_vector_maps);
+
+
+/* Since VF messages are limited by u16 size, precalculate the maximum possible
+ * values of nested elements in virtchnl structures that virtual channel can
+ * possibly handle in a single message.
+ */
+enum virtchnl_vector_limits {
+	VIRTCHNL_OP_CONFIG_VSI_QUEUES_MAX	=
+		((u16)(~0) - sizeof(struct virtchnl_vsi_queue_config_info)) /
+		sizeof(struct virtchnl_queue_pair_info),
+
+	VIRTCHNL_OP_CONFIG_IRQ_MAP_MAX		=
+		((u16)(~0) - sizeof(struct virtchnl_irq_map_info)) /
+		sizeof(struct virtchnl_vector_map),
+
+	VIRTCHNL_OP_ADD_DEL_ETH_ADDR_MAX	=
+		((u16)(~0) - sizeof(struct virtchnl_ether_addr_list)) /
+		sizeof(struct virtchnl_ether_addr),
+
+	VIRTCHNL_OP_ADD_DEL_VLAN_MAX		=
+		((u16)(~0) - sizeof(struct virtchnl_vlan_filter_list)) /
+		sizeof(u16),
+
+
+	VIRTCHNL_OP_ENABLE_CHANNELS_MAX		=
+		((u16)(~0) - sizeof(struct virtchnl_tc_info)) /
+		sizeof(struct virtchnl_channel_info),
+
+	VIRTCHNL_OP_ENABLE_DISABLE_DEL_QUEUES_V2_MAX	=
+		((u16)(~0) - sizeof(struct virtchnl_del_ena_dis_queues)) /
+		sizeof(struct virtchnl_queue_chunk),
+
+	VIRTCHNL_OP_MAP_UNMAP_QUEUE_VECTOR_MAX	=
+		((u16)(~0) - sizeof(struct virtchnl_queue_vector_maps)) /
+		sizeof(struct virtchnl_queue_vector),
+};
+
 /**
  * virtchnl_vc_validate_vf_msg
  * @ver: Virtchnl version info
@@ -1212,6 +1358,8 @@ virtchnl_vc_validate_vf_msg(struct virtchnl_version_info *ver, u32 v_opcode,
 	case VIRTCHNL_OP_DISABLE_QUEUES:
 		valid_len = sizeof(struct virtchnl_queue_select);
 		break;
+	case VIRTCHNL_OP_GET_MAX_RSS_QREGION:
+		break;
 	case VIRTCHNL_OP_ADD_ETH_ADDR:
 	case VIRTCHNL_OP_DEL_ETH_ADDR:
 		valid_len = sizeof(struct virtchnl_ether_addr_list);
@@ -1338,6 +1486,35 @@ virtchnl_vc_validate_vf_msg(struct virtchnl_version_info *ver, u32 v_opcode,
 	case VIRTCHNL_OP_QUERY_FDIR_FILTER:
 		valid_len = sizeof(struct virtchnl_fdir_query);
 		break;
+	case VIRTCHNL_OP_ENABLE_QUEUES_V2:
+	case VIRTCHNL_OP_DISABLE_QUEUES_V2:
+		valid_len = sizeof(struct virtchnl_del_ena_dis_queues);
+		if (msglen >= valid_len) {
+			struct virtchnl_del_ena_dis_queues *qs =
+				(struct virtchnl_del_ena_dis_queues *)msg;
+			if (qs->chunks.num_chunks == 0 ||
+			    qs->chunks.num_chunks > VIRTCHNL_OP_ENABLE_DISABLE_DEL_QUEUES_V2_MAX) {
+				err_msg_format = true;
+				break;
+			}
+			valid_len += (qs->chunks.num_chunks - 1) *
+				      sizeof(struct virtchnl_queue_chunk);
+		}
+		break;
+	case VIRTCHNL_OP_MAP_QUEUE_VECTOR:
+		valid_len = sizeof(struct virtchnl_queue_vector_maps);
+		if (msglen >= valid_len) {
+			struct virtchnl_queue_vector_maps *v_qp =
+				(struct virtchnl_queue_vector_maps *)msg;
+			if (v_qp->num_qv_maps == 0 ||
+			    v_qp->num_qv_maps > VIRTCHNL_OP_MAP_UNMAP_QUEUE_VECTOR_MAX) {
+				err_msg_format = true;
+				break;
+			}
+			valid_len += (v_qp->num_qv_maps - 1) *
+				      sizeof(struct virtchnl_queue_vector);
+		}
+		break;
 	/* These are always errors coming from the VF. */
 	case VIRTCHNL_OP_EVENT:
 	case VIRTCHNL_OP_UNKNOWN:
-- 
2.25.4


      parent reply	other threads:[~2020-10-16 10:38 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-10-16 10:41 [dpdk-dev] [PATCH v2 0/2] iavf update base code Qi Zhang
2020-10-16 10:40 ` Zhang, Qi Z
2020-10-16 10:41 ` [dpdk-dev] [PATCH v2 1/2] common/iavf: replace macro for MAC address length Qi Zhang
2020-10-16 10:41 ` Qi Zhang [this message]

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=20201016104128.1049430-3-qi.z.zhang@intel.com \
    --to=qi.z.zhang@intel.com \
    --cc=beilei.xing@intel.com \
    --cc=brett.creeley@intel.com \
    --cc=dev@dpdk.org \
    --cc=ferruh.yigit@intel.com \
    --cc=ting.xu@intel.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

DPDK patches and discussions

This inbox may be cloned and mirrored by anyone:

	git clone --mirror https://inbox.dpdk.org/dev/0 dev/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 dev dev/ https://inbox.dpdk.org/dev \
		dev@dpdk.org
	public-inbox-index dev

Example config snippet for mirrors.
Newsgroup available over NNTP:
	nntp://inbox.dpdk.org/inbox.dpdk.dev


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git