[PATCH net-next 0/8][pull request] ice: cleanups and preparation for live migration

[PATCH net-next 0/8][pull request] ice: cleanups and preparation for live migration

[Tony Nguyen]

Jake Keller says:

Various cleanups and preparation to the ice driver code for supporting
SR-IOV live migration.

The logic for unpacking Rx queue context data is added. This is the inverse
of the existing packing logic. Thanks to <linux/packing.h> this is trivial
to add.

Code to enable both reading and writing the Tx queue context for a queue
over a shared hardware register interface is added. Thanks to ice_adapter,
this is locked across all PFs that need to use it, preventing concurrency
issues with multiple PFs.

The RSS hash configuration requested by a VF is cached within the VF
structure. This will be used to track and restore the same configuration
during migration load.

ice_sriov_set_msix_vec_count() is updated to use pci_iov_vf_id() instead of
open-coding a worse equivalent, and checks to avoid rebuilding MSI-X if the
current request is for the existing amount of vectors.

A new ice_get_vf_by_dev() helper function is added to simplify accessing a
VF from its PCI device structure. This will be used more heavily within the
live migration code itself.
---
This is the first eight patches of my full series to support live
migration. The full series (based on net-next) is available at [1] for
early preview if you want to see the changes in context.

Some of these changes are not "used" until the live migration patches
themselves. However, I felt they were sufficiently large and review-able on
their own. Additionally, if I keep them included within the live migration
series it is 15 patches which is at the limit of acceptable size for
netdev. I'd prefer to merge these cleanups first in order to reduce the
burden of review for the whole feature.

Link: [1] https://github.com/jacob-keller/linux/tree/e810-live-migration/jk-migration-tlv

The following are changes since commit e090f978054e1cfcd970234589168fcbcba33976:
  Merge branch 'net-dsa-rzn1_a5psw-add-compile_test'
and are available in the git repository at:
  git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/next-queue 100GbE

Jacob Keller (8):
  ice: add support for reading and unpacking Rx queue context
  ice: add functions to get and set Tx queue context
  ice: save RSS hash configuration for migration
  ice: move ice_vsi_update_l2tsel to ice_lib.c
  ice: expose VF functions used by live migration
  ice: use pci_iov_vf_id() to get VF ID
  ice: avoid rebuilding if MSI-X vector count is unchanged
  ice: introduce ice_get_vf_by_dev() wrapper

 drivers/net/ethernet/intel/ice/ice_adapter.c  |   1 +
 drivers/net/ethernet/intel/ice/ice_adapter.h  |   5 +-
 .../net/ethernet/intel/ice/ice_adminq_cmd.h   |  14 +-
 drivers/net/ethernet/intel/ice/ice_common.c   | 233 +++++++++++++++++-
 drivers/net/ethernet/intel/ice/ice_common.h   |   6 +
 .../net/ethernet/intel/ice/ice_hw_autogen.h   |  12 +
 drivers/net/ethernet/intel/ice/ice_lib.c      |  35 +++
 drivers/net/ethernet/intel/ice/ice_lib.h      |   8 +
 drivers/net/ethernet/intel/ice/ice_sriov.c    |  19 +-
 drivers/net/ethernet/intel/ice/ice_sriov.h    |   7 +
 drivers/net/ethernet/intel/ice/ice_vf_lib.c   |   3 +
 drivers/net/ethernet/intel/ice/ice_vf_lib.h   |  26 +-
 drivers/net/ethernet/intel/ice/ice_virtchnl.c |  59 +----
 drivers/net/ethernet/intel/ice/ice_virtchnl.h |  19 ++
 14 files changed, 378 insertions(+), 69 deletions(-)

-- 
2.47.1

[PATCH net-next 1/8] ice: add support for reading and unpacking Rx queue context

[Tony Nguyen]

From: Jacob Keller <jacob.e.keller@intel.com>

In order to support live migration, the ice driver will need to read
certain data from the Rx queue context. This is stored in the hardware in a
packed format.

Since we use <linux/packing.h> for the mapping between the packed hardware
format and the unpacked structure, it is trivial to enable unpacking
support via the unpack_fields() function.

Add the ice_unpack_rxq_ctx() function based on the unpack_fields() API.
Re-use the same field definitions from the packing implementation.

Add ice_copy_rxq_ctx_from_hw() to copy the Rx queue context data from the
hardware registers.

Use these to implement ice_read_rxq_ctx() which will return the Rx queue
context to the caller in its unpacked ice_rlan_ctx struct.

This will enable the migration logic access to the relevant data about the
Rx device queues. It can easily be copied to the target system as part of
the migration payload, where it will be used to configure the Rx queues.

Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Madhu Chittim <madhu.chittim@intel.com>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
---
 drivers/net/ethernet/intel/ice/ice_common.c | 60 +++++++++++++++++++++
 drivers/net/ethernet/intel/ice/ice_common.h |  2 +
 2 files changed, 62 insertions(+)

diff --git a/drivers/net/ethernet/intel/ice/ice_common.c b/drivers/net/ethernet/intel/ice/ice_common.c
index 84cd8c6dcf39..ddde3487aea9 100644
--- a/drivers/net/ethernet/intel/ice/ice_common.c
+++ b/drivers/net/ethernet/intel/ice/ice_common.c
@@ -1342,6 +1342,26 @@ static void ice_copy_rxq_ctx_to_hw(struct ice_hw *hw,
 	}
 }
 
+/**
+ * ice_copy_rxq_ctx_from_hw - Copy packed Rx Queue context from HW registers
+ * @hw: pointer to the hardware structure
+ * @rxq_ctx: pointer to the packed Rx queue context
+ * @rxq_index: the index of the Rx queue
+ */
+static void ice_copy_rxq_ctx_from_hw(struct ice_hw *hw,
+				     ice_rxq_ctx_buf_t *rxq_ctx,
+				     u32 rxq_index)
+{
+	u32 *ctx = (u32 *)rxq_ctx;
+
+	/* Copy each dword separately from HW */
+	for (int i = 0; i < ICE_RXQ_CTX_SIZE_DWORDS; i++, ctx++) {
+		*ctx = rd32(hw, QRX_CONTEXT(i, rxq_index));
+
+		ice_debug(hw, ICE_DBG_QCTX, "qrxdata[%d]: %08X\n", i, *ctx);
+	}
+}
+
 #define ICE_CTX_STORE(struct_name, struct_field, width, lsb) \
 	PACKED_FIELD((lsb) + (width) - 1, (lsb), struct struct_name, struct_field)
 
@@ -1385,6 +1405,21 @@ static void ice_pack_rxq_ctx(const struct ice_rlan_ctx *ctx,
 		    QUIRK_LITTLE_ENDIAN | QUIRK_LSW32_IS_FIRST);
 }
 
+/**
+ * ice_unpack_rxq_ctx - Unpack Rx queue context from a HW buffer
+ * @buf: the HW buffer to unpack from
+ * @ctx: the Rx queue context to unpack
+ *
+ * Unpack the Rx queue context from the HW buffer into the CPU-friendly
+ * structure.
+ */
+static void ice_unpack_rxq_ctx(const ice_rxq_ctx_buf_t *buf,
+			       struct ice_rlan_ctx *ctx)
+{
+	unpack_fields(buf, sizeof(*buf), ctx, ice_rlan_ctx_fields,
+		      QUIRK_LITTLE_ENDIAN | QUIRK_LSW32_IS_FIRST);
+}
+
 /**
  * ice_write_rxq_ctx - Write Rx Queue context to hardware
  * @hw: pointer to the hardware structure
@@ -1410,6 +1445,31 @@ int ice_write_rxq_ctx(struct ice_hw *hw, struct ice_rlan_ctx *rlan_ctx,
 	return 0;
 }
 
+/**
+ * ice_read_rxq_ctx - Read Rx queue context from HW
+ * @hw: pointer to the hardware structure
+ * @rlan_ctx: pointer to the Rx queue context
+ * @rxq_index: the index of the Rx queue
+ *
+ * Read the Rx queue context from the hardware registers, and unpack it into
+ * the sparse Rx queue context structure.
+ *
+ * Returns: 0 on success, or -EINVAL if the Rx queue index is invalid.
+ */
+int ice_read_rxq_ctx(struct ice_hw *hw, struct ice_rlan_ctx *rlan_ctx,
+		     u32 rxq_index)
+{
+	ice_rxq_ctx_buf_t buf = {};
+
+	if (rxq_index > QRX_CTRL_MAX_INDEX)
+		return -EINVAL;
+
+	ice_copy_rxq_ctx_from_hw(hw, &buf, rxq_index);
+	ice_unpack_rxq_ctx(&buf, rlan_ctx);
+
+	return 0;
+}
+
 /* LAN Tx Queue Context */
 static const struct packed_field_u8 ice_tlan_ctx_fields[] = {
 				    /* Field			Width	LSB */
diff --git a/drivers/net/ethernet/intel/ice/ice_common.h b/drivers/net/ethernet/intel/ice/ice_common.h
index e8979b80c2f0..01992440bb9f 100644
--- a/drivers/net/ethernet/intel/ice/ice_common.h
+++ b/drivers/net/ethernet/intel/ice/ice_common.h
@@ -118,6 +118,8 @@ void ice_set_safe_mode_caps(struct ice_hw *hw);
 
 int ice_write_rxq_ctx(struct ice_hw *hw, struct ice_rlan_ctx *rlan_ctx,
 		      u32 rxq_index);
+int ice_read_rxq_ctx(struct ice_hw *hw, struct ice_rlan_ctx *rlan_ctx,
+		     u32 rxq_index);
 
 int
 ice_aq_get_rss_lut(struct ice_hw *hw, struct ice_aq_get_set_rss_lut_params *get_params);
-- 
2.47.1

[PATCH net-next 2/8] ice: add functions to get and set Tx queue context

[Tony Nguyen]

From: Jacob Keller <jacob.e.keller@intel.com>

The live migration driver will need to save and restore the Tx queue
context state from the hardware registers. This state contains both static
fields which do not change during Tx traffic as well as dynamic fields
which may change during Tx traffic.

Unlike the Rx context, the Tx queue context is accessed indirectly from
GLCOMM_QTX_CNTX_CTL and GLCOMM_QTX_CNTX_DATA registers. These registers are
shared by multiple PFs on the same PCIe card. Multiple PFs cannot safely
access the registers simultaneously, and there is no hardware semaphore or
logic to control access. To handle this, introduce the txq_ctx_lock to the
ice_adapter structure. This is similar to the ptp_gltsyn_time_lock. All PFs
on the same adapter share this structure, and use it to serialize access to
the registers to prevent error.

Add a new functions to get and set the Tx queue context through the
GLCOMM_QTX_CNTX_CTL interface. The hardware context values are stored in
the registers using the same packed format as the Admin Queue buffer.

The hardware buffer is 40 bytes wide, as it contains an additional 18 bytes
of internal state not sent with the Admin Queue buffer. For this reason, a
separate typedef and packing function must be used. We can share the same
packed fields definitions because we never need to unpack the internal
state. This is preferred, as it ensures the internal state is zero'd when
writing into HW, and avoids issues with reading by u32 registers into a
buffer of 22 bytes in length. Thanks to the typedefs, misuse of the API
with the wrong size buffer can easily be caught at compile time.

Note reading this data from hardware is essential because the current Tx
queue context may be different from the context as initially programmed by
the driver during VF initialization. When migrating a VF we must ensure the
target VF has identical context as the source VF did.

Co-developed-by: Yahui Cao <yahui.cao@intel.com>
Signed-off-by: Yahui Cao <yahui.cao@intel.com>
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Madhu Chittim <madhu.chittim@intel.com>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
---
 drivers/net/ethernet/intel/ice/ice_adapter.c  |   1 +
 drivers/net/ethernet/intel/ice/ice_adapter.h  |   5 +-
 .../net/ethernet/intel/ice/ice_adminq_cmd.h   |  14 +-
 drivers/net/ethernet/intel/ice/ice_common.c   | 173 +++++++++++++++++-
 drivers/net/ethernet/intel/ice/ice_common.h   |   4 +
 .../net/ethernet/intel/ice/ice_hw_autogen.h   |  12 ++
 6 files changed, 204 insertions(+), 5 deletions(-)

diff --git a/drivers/net/ethernet/intel/ice/ice_adapter.c b/drivers/net/ethernet/intel/ice/ice_adapter.c
index 66e070095d1b..9e4adc43e474 100644
--- a/drivers/net/ethernet/intel/ice/ice_adapter.c
+++ b/drivers/net/ethernet/intel/ice/ice_adapter.c
@@ -32,6 +32,7 @@ static struct ice_adapter *ice_adapter_new(u64 dsn)
 
 	adapter->device_serial_number = dsn;
 	spin_lock_init(&adapter->ptp_gltsyn_time_lock);
+	spin_lock_init(&adapter->txq_ctx_lock);
 	refcount_set(&adapter->refcount, 1);
 
 	mutex_init(&adapter->ports.lock);
diff --git a/drivers/net/ethernet/intel/ice/ice_adapter.h b/drivers/net/ethernet/intel/ice/ice_adapter.h
index ac15c0d2bc1a..db66d03c9f96 100644
--- a/drivers/net/ethernet/intel/ice/ice_adapter.h
+++ b/drivers/net/ethernet/intel/ice/ice_adapter.h
@@ -27,9 +27,10 @@ struct ice_port_list {
 
 /**
  * struct ice_adapter - PCI adapter resources shared across PFs
+ * @refcount: Reference count. struct ice_pf objects hold the references.
  * @ptp_gltsyn_time_lock: Spinlock protecting access to the GLTSYN_TIME
  *                        register of the PTP clock.
- * @refcount: Reference count. struct ice_pf objects hold the references.
+ * @txq_ctx_lock: Spinlock protecting access to the GLCOMM_QTX_CNTX_CTL register
  * @ctrl_pf: Control PF of the adapter
  * @ports: Ports list
  * @device_serial_number: DSN cached for collision detection on 32bit systems
@@ -38,6 +39,8 @@ struct ice_adapter {
 	refcount_t refcount;
 	/* For access to the GLTSYN_TIME register */
 	spinlock_t ptp_gltsyn_time_lock;
+	/* For access to GLCOMM_QTX_CNTX_CTL register */
+	spinlock_t txq_ctx_lock;
 
 	struct ice_pf *ctrl_pf;
 	struct ice_port_list ports;
diff --git a/drivers/net/ethernet/intel/ice/ice_adminq_cmd.h b/drivers/net/ethernet/intel/ice/ice_adminq_cmd.h
index 712f7ef2a00a..97f9ebd62d93 100644
--- a/drivers/net/ethernet/intel/ice/ice_adminq_cmd.h
+++ b/drivers/net/ethernet/intel/ice/ice_adminq_cmd.h
@@ -14,11 +14,23 @@
 
 #define ICE_RXQ_CTX_SIZE_DWORDS		8
 #define ICE_RXQ_CTX_SZ			(ICE_RXQ_CTX_SIZE_DWORDS * sizeof(u32))
-#define ICE_TXQ_CTX_SZ			22
 
 typedef struct __packed { u8 buf[ICE_RXQ_CTX_SZ]; } ice_rxq_ctx_buf_t;
+
+/* The Tx queue context is 40 bytes, and includes some internal state. The
+ * Admin Queue buffers don't include the internal state, so only include the
+ * first 22 bytes of the context.
+ */
+#define ICE_TXQ_CTX_SZ			22
+
 typedef struct __packed { u8 buf[ICE_TXQ_CTX_SZ]; } ice_txq_ctx_buf_t;
 
+#define ICE_TXQ_CTX_FULL_SIZE_DWORDS	10
+#define ICE_TXQ_CTX_FULL_SZ \
+	(ICE_TXQ_CTX_FULL_SIZE_DWORDS * sizeof(u32))
+
+typedef struct __packed { u8 buf[ICE_TXQ_CTX_FULL_SZ]; } ice_txq_ctx_buf_full_t;
+
 struct ice_aqc_generic {
 	__le32 param0;
 	__le32 param1;
diff --git a/drivers/net/ethernet/intel/ice/ice_common.c b/drivers/net/ethernet/intel/ice/ice_common.c
index ddde3487aea9..1b435e108d3c 100644
--- a/drivers/net/ethernet/intel/ice/ice_common.c
+++ b/drivers/net/ethernet/intel/ice/ice_common.c
@@ -1503,12 +1503,12 @@ static const struct packed_field_u8 ice_tlan_ctx_fields[] = {
 };
 
 /**
- * ice_pack_txq_ctx - Pack Tx queue context into a HW buffer
+ * ice_pack_txq_ctx - Pack Tx queue context into Admin Queue buffer
  * @ctx: the Tx queue context to pack
- * @buf: the HW buffer to pack into
+ * @buf: the Admin Queue HW buffer to pack into
  *
  * Pack the Tx queue context from the CPU-friendly unpacked buffer into its
- * bit-packed HW layout.
+ * bit-packed Admin Queue layout.
  */
 void ice_pack_txq_ctx(const struct ice_tlan_ctx *ctx, ice_txq_ctx_buf_t *buf)
 {
@@ -1516,6 +1516,173 @@ void ice_pack_txq_ctx(const struct ice_tlan_ctx *ctx, ice_txq_ctx_buf_t *buf)
 		    QUIRK_LITTLE_ENDIAN | QUIRK_LSW32_IS_FIRST);
 }
 
+/**
+ * ice_pack_txq_ctx_full - Pack Tx queue context into a HW buffer
+ * @ctx: the Tx queue context to pack
+ * @buf: the HW buffer to pack into
+ *
+ * Pack the Tx queue context from the CPU-friendly unpacked buffer into its
+ * bit-packed HW layout, including the internal data portion.
+ */
+static void ice_pack_txq_ctx_full(const struct ice_tlan_ctx *ctx,
+				  ice_txq_ctx_buf_full_t *buf)
+{
+	pack_fields(buf, sizeof(*buf), ctx, ice_tlan_ctx_fields,
+		    QUIRK_LITTLE_ENDIAN | QUIRK_LSW32_IS_FIRST);
+}
+
+/**
+ * ice_unpack_txq_ctx_full - Unpack Tx queue context from a HW buffer
+ * @buf: the HW buffer to unpack from
+ * @ctx: the Tx queue context to unpack
+ *
+ * Unpack the Tx queue context from the HW buffer (including the full internal
+ * state) into the CPU-friendly structure.
+ */
+static void ice_unpack_txq_ctx_full(const ice_txq_ctx_buf_full_t *buf,
+				    struct ice_tlan_ctx *ctx)
+{
+	unpack_fields(buf, sizeof(*buf), ctx, ice_tlan_ctx_fields,
+		      QUIRK_LITTLE_ENDIAN | QUIRK_LSW32_IS_FIRST);
+}
+
+/**
+ * ice_copy_txq_ctx_from_hw - Copy Tx Queue context from HW registers
+ * @hw: pointer to the hardware structure
+ * @txq_ctx: pointer to the packed Tx queue context, including internal state
+ * @txq_index: the index of the Tx queue
+ *
+ * Copy Tx Queue context from HW register space to dense structure
+ */
+static void ice_copy_txq_ctx_from_hw(struct ice_hw *hw,
+				     ice_txq_ctx_buf_full_t *txq_ctx,
+				     u32 txq_index)
+{
+	struct ice_pf *pf = container_of(hw, struct ice_pf, hw);
+	u32 *ctx = (u32 *)txq_ctx;
+	u32 txq_base, reg;
+
+	/* Get Tx queue base within card space */
+	txq_base = rd32(hw, PFLAN_TX_QALLOC(hw->pf_id));
+	txq_base = FIELD_GET(PFLAN_TX_QALLOC_FIRSTQ_M, txq_base);
+
+	reg = FIELD_PREP(GLCOMM_QTX_CNTX_CTL_CMD_M,
+			 GLCOMM_QTX_CNTX_CTL_CMD_READ) |
+	      FIELD_PREP(GLCOMM_QTX_CNTX_CTL_QUEUE_ID_M,
+			 txq_base + txq_index) |
+	      GLCOMM_QTX_CNTX_CTL_CMD_EXEC_M;
+
+	/* Prevent other PFs on the same adapter from accessing the Tx queue
+	 * context interface concurrently.
+	 */
+	spin_lock(&pf->adapter->txq_ctx_lock);
+
+	wr32(hw, GLCOMM_QTX_CNTX_CTL, reg);
+	ice_flush(hw);
+
+	/* Copy each dword separately from HW */
+	for (int i = 0; i < ICE_TXQ_CTX_FULL_SIZE_DWORDS; i++, ctx++) {
+		*ctx = rd32(hw, GLCOMM_QTX_CNTX_DATA(i));
+
+		ice_debug(hw, ICE_DBG_QCTX, "qtxdata[%d]: %08X\n", i, *ctx);
+	}
+
+	spin_unlock(&pf->adapter->txq_ctx_lock);
+}
+
+/**
+ * ice_copy_txq_ctx_to_hw - Copy Tx Queue context into HW registers
+ * @hw: pointer to the hardware structure
+ * @txq_ctx: pointer to the packed Tx queue context, including internal state
+ * @txq_index: the index of the Tx queue
+ */
+static void ice_copy_txq_ctx_to_hw(struct ice_hw *hw,
+				   const ice_txq_ctx_buf_full_t *txq_ctx,
+				   u32 txq_index)
+{
+	struct ice_pf *pf = container_of(hw, struct ice_pf, hw);
+	u32 txq_base, reg;
+
+	/* Get Tx queue base within card space */
+	txq_base = rd32(hw, PFLAN_TX_QALLOC(hw->pf_id));
+	txq_base = FIELD_GET(PFLAN_TX_QALLOC_FIRSTQ_M, txq_base);
+
+	reg = FIELD_PREP(GLCOMM_QTX_CNTX_CTL_CMD_M,
+			 GLCOMM_QTX_CNTX_CTL_CMD_WRITE_NO_DYN) |
+	      FIELD_PREP(GLCOMM_QTX_CNTX_CTL_QUEUE_ID_M,
+			 txq_base + txq_index) |
+	      GLCOMM_QTX_CNTX_CTL_CMD_EXEC_M;
+
+	/* Prevent other PFs on the same adapter from accessing the Tx queue
+	 * context interface concurrently.
+	 */
+	spin_lock(&pf->adapter->txq_ctx_lock);
+
+	/* Copy each dword separately to HW */
+	for (int i = 0; i < ICE_TXQ_CTX_FULL_SIZE_DWORDS; i++) {
+		u32 ctx = ((const u32 *)txq_ctx)[i];
+
+		wr32(hw, GLCOMM_QTX_CNTX_DATA(i), ctx);
+
+		ice_debug(hw, ICE_DBG_QCTX, "qtxdata[%d]: %08X\n", i, ctx);
+	}
+
+	wr32(hw, GLCOMM_QTX_CNTX_CTL, reg);
+	ice_flush(hw);
+
+	spin_unlock(&pf->adapter->txq_ctx_lock);
+}
+
+/**
+ * ice_read_txq_ctx - Read Tx queue context from HW
+ * @hw: pointer to the hardware structure
+ * @tlan_ctx: pointer to the Tx queue context
+ * @txq_index: the index of the Tx queue
+ *
+ * Read the Tx queue context from the HW registers, then unpack it into the
+ * ice_tlan_ctx structure for use.
+ *
+ * Returns: 0 on success, or -EINVAL on an invalid Tx queue index.
+ */
+int ice_read_txq_ctx(struct ice_hw *hw, struct ice_tlan_ctx *tlan_ctx,
+		     u32 txq_index)
+{
+	ice_txq_ctx_buf_full_t buf = {};
+
+	if (txq_index > QTX_COMM_HEAD_MAX_INDEX)
+		return -EINVAL;
+
+	ice_copy_txq_ctx_from_hw(hw, &buf, txq_index);
+	ice_unpack_txq_ctx_full(&buf, tlan_ctx);
+
+	return 0;
+}
+
+/**
+ * ice_write_txq_ctx - Write Tx queue context to HW
+ * @hw: pointer to the hardware structure
+ * @tlan_ctx: pointer to the Tx queue context
+ * @txq_index: the index of the Tx queue
+ *
+ * Pack the Tx queue context into the dense HW layout, then write it into the
+ * HW registers.
+ *
+ * Returns: 0 on success, or -EINVAL on an invalid Tx queue index.
+ */
+int ice_write_txq_ctx(struct ice_hw *hw, struct ice_tlan_ctx *tlan_ctx,
+		      u32 txq_index)
+{
+	ice_txq_ctx_buf_full_t buf = {};
+
+	if (txq_index > QTX_COMM_HEAD_MAX_INDEX)
+		return -EINVAL;
+
+	ice_pack_txq_ctx_full(tlan_ctx, &buf);
+	ice_copy_txq_ctx_to_hw(hw, &buf, txq_index);
+
+	return 0;
+}
+
 /* Sideband Queue command wrappers */
 
 /**
diff --git a/drivers/net/ethernet/intel/ice/ice_common.h b/drivers/net/ethernet/intel/ice/ice_common.h
index 01992440bb9f..25d9785f32cc 100644
--- a/drivers/net/ethernet/intel/ice/ice_common.h
+++ b/drivers/net/ethernet/intel/ice/ice_common.h
@@ -120,6 +120,10 @@ int ice_write_rxq_ctx(struct ice_hw *hw, struct ice_rlan_ctx *rlan_ctx,
 		      u32 rxq_index);
 int ice_read_rxq_ctx(struct ice_hw *hw, struct ice_rlan_ctx *rlan_ctx,
 		     u32 rxq_index);
+int ice_read_txq_ctx(struct ice_hw *hw, struct ice_tlan_ctx *tlan_ctx,
+		     u32 txq_index);
+int ice_write_txq_ctx(struct ice_hw *hw, struct ice_tlan_ctx *tlan_ctx,
+		      u32 txq_index);
 
 int
 ice_aq_get_rss_lut(struct ice_hw *hw, struct ice_aq_get_set_rss_lut_params *get_params);
diff --git a/drivers/net/ethernet/intel/ice/ice_hw_autogen.h b/drivers/net/ethernet/intel/ice/ice_hw_autogen.h
index aa4bfbcf85d2..dd520aa4d1d6 100644
--- a/drivers/net/ethernet/intel/ice/ice_hw_autogen.h
+++ b/drivers/net/ethernet/intel/ice/ice_hw_autogen.h
@@ -16,6 +16,7 @@
 #define GLCOMM_QUANTA_PROF_MAX_DESC_M		ICE_M(0x3F, 24)
 #define QTX_COMM_DBELL(_DBQM)			(0x002C0000 + ((_DBQM) * 4))
 #define QTX_COMM_HEAD(_DBQM)			(0x000E0000 + ((_DBQM) * 4))
+#define QTX_COMM_HEAD_MAX_INDEX			16383
 #define QTX_COMM_HEAD_HEAD_S			0
 #define QTX_COMM_HEAD_HEAD_M			ICE_M(0x1FFF, 0)
 #define PF_FW_ARQBAH				0x00080180
@@ -272,6 +273,8 @@
 #define VPINT_ALLOC_PCI_VALID_M			BIT(31)
 #define VPINT_MBX_CTL(_VSI)			(0x0016A000 + ((_VSI) * 4))
 #define VPINT_MBX_CTL_CAUSE_ENA_M		BIT(30)
+#define PFLAN_TX_QALLOC(_PF)			(0x001D2580 + ((_PF) * 4))
+#define PFLAN_TX_QALLOC_FIRSTQ_M		GENMASK(13, 0)
 #define GLLAN_RCTL_0				0x002941F8
 #define QRX_CONTEXT(_i, _QRX)			(0x00280000 + ((_i) * 8192 + (_QRX) * 4))
 #define QRX_CTRL(_QRX)				(0x00120000 + ((_QRX) * 4))
@@ -376,6 +379,15 @@
 #define GLNVM_ULD_POR_DONE_1_M			BIT(8)
 #define GLNVM_ULD_PCIER_DONE_2_M		BIT(9)
 #define GLNVM_ULD_PE_DONE_M			BIT(10)
+#define GLCOMM_QTX_CNTX_CTL			0x002D2DC8
+#define GLCOMM_QTX_CNTX_CTL_QUEUE_ID_M		GENMASK(13, 0)
+#define GLCOMM_QTX_CNTX_CTL_CMD_M		GENMASK(18, 16)
+#define GLCOMM_QTX_CNTX_CTL_CMD_READ		0
+#define GLCOMM_QTX_CNTX_CTL_CMD_WRITE		1
+#define GLCOMM_QTX_CNTX_CTL_CMD_RESET		3
+#define GLCOMM_QTX_CNTX_CTL_CMD_WRITE_NO_DYN	4
+#define GLCOMM_QTX_CNTX_CTL_CMD_EXEC_M		BIT(19)
+#define GLCOMM_QTX_CNTX_DATA(_i)		(0x002D2D40 + ((_i) * 4))
 #define GLPCI_CNF2				0x000BE004
 #define GLPCI_CNF2_CACHELINE_SIZE_M		BIT(1)
 #define PF_FUNC_RID				0x0009E880
-- 
2.47.1

[PATCH net-next 3/8] ice: save RSS hash configuration for migration

[Tony Nguyen]

From: Jacob Keller <jacob.e.keller@intel.com>

The VF can program the RSS hash configuration over virtchnl. It does this
by sending a u64 bitmask which represents the current hash configuration.

It is not trivial to reverse the hardware configuration back to this hash
set for migration. Instead, save the value to the ice_vf structure when its
modified by the VF.

The rss_hashcfg value is an 8-byte field. Make room for it in ice_vf by
re-arranging some of the existing fields. There is a 4-byte gap after the
first_vector_idx, and a 4-byte gap between max_tx_rate and vf_states. Move
first_vector_idx into the later 4-byte gap, creating an 8 byte area where
rss_hashcfg can be placed. Also move the num_msix field near min_tx_rate,
filling 2 bytes of a 3 byte hole.

The end result of these changes enables placing the rss_hashcfg field into
the structure while also saving 8 bytes in size. It looks like there are a
handful of more possible cleanups to reduce the size even further, but
those have been left as a future cleanup.

Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Reviewed-by: Madhu Chittim <madhu.chittim@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
---
 drivers/net/ethernet/intel/ice/ice_vf_lib.c   | 3 +++
 drivers/net/ethernet/intel/ice/ice_vf_lib.h   | 8 +++++---
 drivers/net/ethernet/intel/ice/ice_virtchnl.c | 4 ++++
 3 files changed, 12 insertions(+), 3 deletions(-)

diff --git a/drivers/net/ethernet/intel/ice/ice_vf_lib.c b/drivers/net/ethernet/intel/ice/ice_vf_lib.c
index 48cd533e93b7..c639ce716d32 100644
--- a/drivers/net/ethernet/intel/ice/ice_vf_lib.c
+++ b/drivers/net/ethernet/intel/ice/ice_vf_lib.c
@@ -1022,6 +1022,9 @@ void ice_initialize_vf_entry(struct ice_vf *vf)
 	vf->num_msix = vfs->num_msix_per;
 	vf->num_vf_qs = vfs->num_qps_per;
 
+	/* set default RSS hash configuration */
+	vf->rss_hashcfg = ICE_DEFAULT_RSS_HASHCFG;
+
 	/* ctrl_vsi_idx will be set to a valid value only when iAVF
 	 * creates its first fdir rule.
 	 */
diff --git a/drivers/net/ethernet/intel/ice/ice_vf_lib.h b/drivers/net/ethernet/intel/ice/ice_vf_lib.h
index 482f4285fd35..a5ee380f8c9e 100644
--- a/drivers/net/ethernet/intel/ice/ice_vf_lib.h
+++ b/drivers/net/ethernet/intel/ice/ice_vf_lib.h
@@ -106,8 +106,7 @@ struct ice_vf {
 	u16 ctrl_vsi_idx;
 	struct ice_vf_fdir fdir;
 	struct ice_fdir_prof_info fdir_prof_info[ICE_MAX_PTGS];
-	/* first vector index of this VF in the PF space */
-	int first_vector_idx;
+	u64 rss_hashcfg;		/* RSS hash configuration */
 	struct ice_sw *vf_sw_id;	/* switch ID the VF VSIs connect to */
 	struct virtchnl_version_info vf_ver;
 	u32 driver_caps;		/* reported by VF driver */
@@ -126,10 +125,14 @@ struct ice_vf {
 	u8 link_up:1;			/* only valid if VF link is forced */
 	u8 lldp_tx_ena:1;
 
+	u16 num_msix;			/* num of MSI-X configured on this VF */
+
 	u32 ptp_caps;
 
 	unsigned int min_tx_rate;	/* Minimum Tx bandwidth limit in Mbps */
 	unsigned int max_tx_rate;	/* Maximum Tx bandwidth limit in Mbps */
+	/* first vector index of this VF in the PF space */
+	int first_vector_idx;
 	DECLARE_BITMAP(vf_states, ICE_VF_STATES_NBITS);	/* VF runtime states */
 
 	unsigned long vf_caps;		/* VF's adv. capabilities */
@@ -154,7 +157,6 @@ struct ice_vf {
 	u16 lldp_recipe_id;
 	u16 lldp_rule_id;
 
-	u16 num_msix;			/* num of MSI-X configured on this VF */
 	struct ice_vf_qs_bw qs_bw[ICE_MAX_RSS_QS_PER_VF];
 };
 
diff --git a/drivers/net/ethernet/intel/ice/ice_virtchnl.c b/drivers/net/ethernet/intel/ice/ice_virtchnl.c
index 24426dcd8aa2..0a8f15ecac1f 100644
--- a/drivers/net/ethernet/intel/ice/ice_virtchnl.c
+++ b/drivers/net/ethernet/intel/ice/ice_virtchnl.c
@@ -3094,6 +3094,10 @@ static int ice_vc_set_rss_hashcfg(struct ice_vf *vf, u8 *msg)
 		v_ret = ice_err_to_virt_err(status);
 	}
 
+	/* save the requested VF configuration */
+	if (!v_ret)
+		vf->rss_hashcfg = vrh->hashcfg;
+
 	/* send the response to the VF */
 err:
 	return ice_vc_send_msg_to_vf(vf, VIRTCHNL_OP_SET_RSS_HASHCFG, v_ret,
-- 
2.47.1

[PATCH net-next 4/8] ice: move ice_vsi_update_l2tsel to ice_lib.c

[Tony Nguyen]

From: Jacob Keller <jacob.e.keller@intel.com>

A future change is going to need to call ice_vsi_update_l2tsel from a new
context outside of ice_virtchnl.c

Since this function deals with a generic VSI, move it into ice_lib.c to
enable calling it from other places in the ice driver.

Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Madhu Chittim <madhu.chittim@intel.com>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
---
 drivers/net/ethernet/intel/ice/ice_lib.c      | 35 ++++++++++++++++
 drivers/net/ethernet/intel/ice/ice_lib.h      |  8 ++++
 drivers/net/ethernet/intel/ice/ice_virtchnl.c | 42 -------------------
 3 files changed, 43 insertions(+), 42 deletions(-)

diff --git a/drivers/net/ethernet/intel/ice/ice_lib.c b/drivers/net/ethernet/intel/ice/ice_lib.c
index 2f1782e9357f..1be1e429a7c8 100644
--- a/drivers/net/ethernet/intel/ice/ice_lib.c
+++ b/drivers/net/ethernet/intel/ice/ice_lib.c
@@ -4020,3 +4020,38 @@ ice_vsi_update_local_lb(struct ice_vsi *vsi, bool set)
 	vsi->info = ctx.info;
 	return 0;
 }
+
+/**
+ * ice_vsi_update_l2tsel - update l2tsel field for all Rx rings on this VSI
+ * @vsi: VSI used to update l2tsel on
+ * @l2tsel: l2tsel setting requested
+ *
+ * Use the l2tsel setting to update all of the Rx queue context bits for l2tsel.
+ * This will modify which descriptor field the first offloaded VLAN will be
+ * stripped into.
+ */
+void ice_vsi_update_l2tsel(struct ice_vsi *vsi, enum ice_l2tsel l2tsel)
+{
+	struct ice_hw *hw = &vsi->back->hw;
+	u32 l2tsel_bit;
+	int i;
+
+	if (l2tsel == ICE_L2TSEL_EXTRACT_FIRST_TAG_L2TAG2_2ND)
+		l2tsel_bit = 0;
+	else
+		l2tsel_bit = BIT(ICE_L2TSEL_BIT_OFFSET);
+
+	for (i = 0; i < vsi->alloc_rxq; i++) {
+		u16 pfq = vsi->rxq_map[i];
+		u32 qrx_context_offset;
+		u32 regval;
+
+		qrx_context_offset =
+			QRX_CONTEXT(ICE_L2TSEL_QRX_CONTEXT_REG_IDX, pfq);
+
+		regval = rd32(hw, qrx_context_offset);
+		regval &= ~BIT(ICE_L2TSEL_BIT_OFFSET);
+		regval |= l2tsel_bit;
+		wr32(hw, qrx_context_offset, regval);
+	}
+}
diff --git a/drivers/net/ethernet/intel/ice/ice_lib.h b/drivers/net/ethernet/intel/ice/ice_lib.h
index 654516c5fc3e..2cb1eb98b9da 100644
--- a/drivers/net/ethernet/intel/ice/ice_lib.h
+++ b/drivers/net/ethernet/intel/ice/ice_lib.h
@@ -11,6 +11,13 @@
 #define ICE_VSI_FLAG_INIT	BIT(0)
 #define ICE_VSI_FLAG_NO_INIT	0
 
+#define ICE_L2TSEL_QRX_CONTEXT_REG_IDX	3
+#define ICE_L2TSEL_BIT_OFFSET		23
+enum ice_l2tsel {
+	ICE_L2TSEL_EXTRACT_FIRST_TAG_L2TAG2_2ND,
+	ICE_L2TSEL_EXTRACT_FIRST_TAG_L2TAG1,
+};
+
 const char *ice_vsi_type_str(enum ice_vsi_type vsi_type);
 
 bool ice_pf_state_is_nominal(struct ice_pf *pf);
@@ -116,4 +123,5 @@ void ice_set_feature_support(struct ice_pf *pf, enum ice_feature f);
 void ice_clear_feature_support(struct ice_pf *pf, enum ice_feature f);
 void ice_init_feature_support(struct ice_pf *pf);
 bool ice_vsi_is_rx_queue_active(struct ice_vsi *vsi);
+void ice_vsi_update_l2tsel(struct ice_vsi *vsi, enum ice_l2tsel l2tsel);
 #endif /* !_ICE_LIB_H_ */
diff --git a/drivers/net/ethernet/intel/ice/ice_virtchnl.c b/drivers/net/ethernet/intel/ice/ice_virtchnl.c
index 0a8f15ecac1f..307252500f02 100644
--- a/drivers/net/ethernet/intel/ice/ice_virtchnl.c
+++ b/drivers/net/ethernet/intel/ice/ice_virtchnl.c
@@ -3860,48 +3860,6 @@ ice_vc_ena_vlan_offload(struct ice_vsi *vsi,
 	return 0;
 }
 
-#define ICE_L2TSEL_QRX_CONTEXT_REG_IDX	3
-#define ICE_L2TSEL_BIT_OFFSET		23
-enum ice_l2tsel {
-	ICE_L2TSEL_EXTRACT_FIRST_TAG_L2TAG2_2ND,
-	ICE_L2TSEL_EXTRACT_FIRST_TAG_L2TAG1,
-};
-
-/**
- * ice_vsi_update_l2tsel - update l2tsel field for all Rx rings on this VSI
- * @vsi: VSI used to update l2tsel on
- * @l2tsel: l2tsel setting requested
- *
- * Use the l2tsel setting to update all of the Rx queue context bits for l2tsel.
- * This will modify which descriptor field the first offloaded VLAN will be
- * stripped into.
- */
-static void ice_vsi_update_l2tsel(struct ice_vsi *vsi, enum ice_l2tsel l2tsel)
-{
-	struct ice_hw *hw = &vsi->back->hw;
-	u32 l2tsel_bit;
-	int i;
-
-	if (l2tsel == ICE_L2TSEL_EXTRACT_FIRST_TAG_L2TAG2_2ND)
-		l2tsel_bit = 0;
-	else
-		l2tsel_bit = BIT(ICE_L2TSEL_BIT_OFFSET);
-
-	for (i = 0; i < vsi->alloc_rxq; i++) {
-		u16 pfq = vsi->rxq_map[i];
-		u32 qrx_context_offset;
-		u32 regval;
-
-		qrx_context_offset =
-			QRX_CONTEXT(ICE_L2TSEL_QRX_CONTEXT_REG_IDX, pfq);
-
-		regval = rd32(hw, qrx_context_offset);
-		regval &= ~BIT(ICE_L2TSEL_BIT_OFFSET);
-		regval |= l2tsel_bit;
-		wr32(hw, qrx_context_offset, regval);
-	}
-}
-
 /**
  * ice_vc_ena_vlan_stripping_v2_msg
  * @vf: VF the message was received from
-- 
2.47.1

[PATCH net-next 5/8] ice: expose VF functions used by live migration

[Tony Nguyen]

From: Jacob Keller <jacob.e.keller@intel.com>

The live migration process will require configuring the target VF with the
data provided from the source host. A few helper functions in ice_sriov.c
and ice_virtchnl.c will be needed for this process, but are currently
static.

Expose these functions in their respective headers so that the live
migration module can use them during the migration process.

Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
---
 drivers/net/ethernet/intel/ice/ice_sriov.h    |  7 +++++++
 drivers/net/ethernet/intel/ice/ice_virtchnl.c | 13 ++++++-------
 drivers/net/ethernet/intel/ice/ice_virtchnl.h | 19 +++++++++++++++++++
 3 files changed, 32 insertions(+), 7 deletions(-)

diff --git a/drivers/net/ethernet/intel/ice/ice_sriov.h b/drivers/net/ethernet/intel/ice/ice_sriov.h
index 96549ca5c52c..d1a998a4bef6 100644
--- a/drivers/net/ethernet/intel/ice/ice_sriov.h
+++ b/drivers/net/ethernet/intel/ice/ice_sriov.h
@@ -64,6 +64,7 @@ bool
 ice_vc_validate_pattern(struct ice_vf *vf, struct virtchnl_proto_hdrs *proto);
 u32 ice_sriov_get_vf_total_msix(struct pci_dev *pdev);
 int ice_sriov_set_msix_vec_count(struct pci_dev *vf_dev, int msix_vec_count);
+int ice_vf_vsi_dis_single_txq(struct ice_vf *vf, struct ice_vsi *vsi, u16 q_id);
 #else /* CONFIG_PCI_IOV */
 static inline void ice_process_vflr_event(struct ice_pf *pf) { }
 static inline void ice_free_vfs(struct ice_pf *pf) { }
@@ -164,5 +165,11 @@ ice_sriov_set_msix_vec_count(struct pci_dev *vf_dev, int msix_vec_count)
 {
 	return -EOPNOTSUPP;
 }
+
+static inline int ice_vf_vsi_dis_single_txq(struct ice_vf *vf,
+					    struct ice_vsi *vsi, u16 q_id)
+{
+	return -EOPNOTSUPP;
+}
 #endif /* CONFIG_PCI_IOV */
 #endif /* _ICE_SRIOV_H_ */
diff --git a/drivers/net/ethernet/intel/ice/ice_virtchnl.c b/drivers/net/ethernet/intel/ice/ice_virtchnl.c
index 307252500f02..9460b6561b69 100644
--- a/drivers/net/ethernet/intel/ice/ice_virtchnl.c
+++ b/drivers/net/ethernet/intel/ice/ice_virtchnl.c
@@ -1427,7 +1427,7 @@ static bool ice_vc_validate_vqs_bitmaps(struct virtchnl_queue_select *vqs)
  * @vsi: VSI of the VF to configure
  * @q_idx: VF queue index used to determine the queue in the PF's space
  */
-static void ice_vf_ena_txq_interrupt(struct ice_vsi *vsi, u32 q_idx)
+void ice_vf_ena_txq_interrupt(struct ice_vsi *vsi, u32 q_idx)
 {
 	struct ice_hw *hw = &vsi->back->hw;
 	u32 pfq = vsi->txq_map[q_idx];
@@ -1450,7 +1450,7 @@ static void ice_vf_ena_txq_interrupt(struct ice_vsi *vsi, u32 q_idx)
  * @vsi: VSI of the VF to configure
  * @q_idx: VF queue index used to determine the queue in the PF's space
  */
-static void ice_vf_ena_rxq_interrupt(struct ice_vsi *vsi, u32 q_idx)
+void ice_vf_ena_rxq_interrupt(struct ice_vsi *vsi, u32 q_idx)
 {
 	struct ice_hw *hw = &vsi->back->hw;
 	u32 pfq = vsi->rxq_map[q_idx];
@@ -1566,8 +1566,7 @@ static int ice_vc_ena_qs_msg(struct ice_vf *vf, u8 *msg)
  * disabled then clear q_id bit in the enabled queues bitmap and return
  * success. Otherwise return error.
  */
-static int
-ice_vf_vsi_dis_single_txq(struct ice_vf *vf, struct ice_vsi *vsi, u16 q_id)
+int ice_vf_vsi_dis_single_txq(struct ice_vf *vf, struct ice_vsi *vsi, u16 q_id)
 {
 	struct ice_txq_meta txq_meta = { 0 };
 	struct ice_tx_ring *ring;
@@ -2621,7 +2620,7 @@ static bool ice_vf_vlan_offload_ena(u32 caps)
  * ice_is_vlan_promisc_allowed - check if VLAN promiscuous config is allowed
  * @vf: VF used to determine if VLAN promiscuous config is allowed
  */
-static bool ice_is_vlan_promisc_allowed(struct ice_vf *vf)
+bool ice_is_vlan_promisc_allowed(struct ice_vf *vf)
 {
 	if ((test_bit(ICE_VF_STATE_UC_PROMISC, vf->vf_states) ||
 	     test_bit(ICE_VF_STATE_MC_PROMISC, vf->vf_states)) &&
@@ -2640,8 +2639,8 @@ static bool ice_is_vlan_promisc_allowed(struct ice_vf *vf)
  * This function should only be called if VLAN promiscuous mode is allowed,
  * which can be determined via ice_is_vlan_promisc_allowed().
  */
-static int ice_vf_ena_vlan_promisc(struct ice_vf *vf, struct ice_vsi *vsi,
-				   struct ice_vlan *vlan)
+int ice_vf_ena_vlan_promisc(struct ice_vf *vf, struct ice_vsi *vsi,
+			    struct ice_vlan *vlan)
 {
 	u8 promisc_m = 0;
 	int status;
diff --git a/drivers/net/ethernet/intel/ice/ice_virtchnl.h b/drivers/net/ethernet/intel/ice/ice_virtchnl.h
index b3eece8c6780..71bb456e2d71 100644
--- a/drivers/net/ethernet/intel/ice/ice_virtchnl.h
+++ b/drivers/net/ethernet/intel/ice/ice_virtchnl.h
@@ -92,12 +92,31 @@ ice_vc_send_msg_to_vf(struct ice_vf *vf, u32 v_opcode,
 bool ice_vc_isvalid_vsi_id(struct ice_vf *vf, u16 vsi_id);
 void ice_vc_process_vf_msg(struct ice_pf *pf, struct ice_rq_event_info *event,
 			   struct ice_mbx_data *mbxdata);
+void ice_vf_ena_rxq_interrupt(struct ice_vsi *vsi, u32 q_idx);
+void ice_vf_ena_txq_interrupt(struct ice_vsi *vsi, u32 q_idx);
+int ice_vf_ena_vlan_promisc(struct ice_vf *vf, struct ice_vsi *vsi,
+			    struct ice_vlan *vlan);
+bool ice_is_vlan_promisc_allowed(struct ice_vf *vf);
 #else /* CONFIG_PCI_IOV */
 static inline void ice_virtchnl_set_dflt_ops(struct ice_vf *vf) { }
 static inline void ice_virtchnl_set_repr_ops(struct ice_vf *vf) { }
 static inline void ice_vc_notify_vf_link_state(struct ice_vf *vf) { }
 static inline void ice_vc_notify_link_state(struct ice_pf *pf) { }
 static inline void ice_vc_notify_reset(struct ice_pf *pf) { }
+static inline void ice_vf_ena_rxq_interrupt(struct ice_vsi *vsi, u32 q_idx) { }
+static inline void ice_vf_ena_txq_interrupt(struct ice_vsi *vsi, u32 q_idx) { }
+
+static inline int ice_vf_ena_vlan_promisc(struct ice_vf *vf,
+					  struct ice_vsi *vsi,
+					  struct ice_vlan *vlan)
+{
+	return -EOPNOTSUPP;
+}
+
+static inline bool ice_is_vlan_promisc_allowed(struct ice_vf *vf)
+{
+	return false;
+}
 
 static inline int
 ice_vc_send_msg_to_vf(struct ice_vf *vf, u32 v_opcode,
-- 
2.47.1

[PATCH net-next 6/8] ice: use pci_iov_vf_id() to get VF ID

[Tony Nguyen]

From: Jacob Keller <jacob.e.keller@intel.com>

The ice_sriov_set_msix_vec_count() obtains the VF device ID in a strange
way by iterating over the possible VF IDs and calling
pci_iov_virtfn_devfn to calculate the device and function combos and
compare them to the pdev->devfn.

This is unnecessary. The pci_iov_vf_id() helper already exists which does
the reverse calculation of pci_iov_virtfn_devfn(), which is much simpler
and avoids the loop construction. Use this instead.

Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
---
 drivers/net/ethernet/intel/ice/ice_sriov.c | 12 +++---------
 1 file changed, 3 insertions(+), 9 deletions(-)

diff --git a/drivers/net/ethernet/intel/ice/ice_sriov.c b/drivers/net/ethernet/intel/ice/ice_sriov.c
index 0e4dc1a5cff0..f88bfd2f3f00 100644
--- a/drivers/net/ethernet/intel/ice/ice_sriov.c
+++ b/drivers/net/ethernet/intel/ice/ice_sriov.c
@@ -952,17 +952,11 @@ int ice_sriov_set_msix_vec_count(struct pci_dev *vf_dev, int msix_vec_count)
 	if (msix_vec_count < ICE_MIN_INTR_PER_VF)
 		return -EINVAL;
 
-	/* Transition of PCI VF function number to function_id */
-	for (id = 0; id < pci_num_vf(pdev); id++) {
-		if (vf_dev->devfn == pci_iov_virtfn_devfn(pdev, id))
-			break;
-	}
-
-	if (id == pci_num_vf(pdev))
-		return -ENOENT;
+	id = pci_iov_vf_id(vf_dev);
+	if (id < 0)
+		return id;
 
 	vf = ice_get_vf_by_id(pf, id);
-
 	if (!vf)
 		return -ENOENT;
 
-- 
2.47.1

[PATCH net-next 7/8] ice: avoid rebuilding if MSI-X vector count is unchanged

[Tony Nguyen]

From: Jacob Keller <jacob.e.keller@intel.com>

Commit 05c16687e0cc ("ice: set MSI-X vector count on VF") added support to
change the vector count for VFs as part of ice_sriov_set_msix_vec_count().
This function modifies and rebuilds the target VF with the requested number
of MSI-X vectors.

Future support for live migration will add a call to
ice_sriov_set_msix_vec_count() to ensure that a migrated VF has the proper
MSI-X vector count. In most cases, this request will be to set the MSI-X
vector count to its current value. In that case, no work is necessary.
Rather than requiring the caller to check this, update the function to
check and exit early if the vector count is already at the requested value.
This avoids an unnecessary VF rebuild.

Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
---
 drivers/net/ethernet/intel/ice/ice_sriov.c | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/drivers/net/ethernet/intel/ice/ice_sriov.c b/drivers/net/ethernet/intel/ice/ice_sriov.c
index f88bfd2f3f00..f78d5d8d516c 100644
--- a/drivers/net/ethernet/intel/ice/ice_sriov.c
+++ b/drivers/net/ethernet/intel/ice/ice_sriov.c
@@ -966,6 +966,12 @@ int ice_sriov_set_msix_vec_count(struct pci_dev *vf_dev, int msix_vec_count)
 		return -ENOENT;
 	}
 
+	/* No need to rebuild if we're setting to the same value */
+	if (msix_vec_count == vf->num_msix) {
+		ice_put_vf(vf);
+		return 0;
+	}
+
 	prev_msix = vf->num_msix;
 	prev_queues = vf->num_vf_qs;
 
-- 
2.47.1

[PATCH net-next 8/8] ice: introduce ice_get_vf_by_dev() wrapper

[Tony Nguyen]

From: Jacob Keller <jacob.e.keller@intel.com>

The ice_get_vf_by_id() function is used to obtain a reference to a VF
structure based on its ID. The ice_sriov_set_msix_vec_count() function
needs to get a VF reference starting from the VF PCI device, and uses
pci_iov_vf_id() to get the VF ID. This pattern is currently uncommon in the
ice driver. However, the live migration module will introduce many more
such locations.

Add a helper wrapper ice_get_vf_by_dev() which takes the VF PCI device and
calls ice_get_vf_by_id() using pci_iov_vf_id() to get the VF ID.

Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
---
 drivers/net/ethernet/intel/ice/ice_sriov.c  |  7 +------
 drivers/net/ethernet/intel/ice/ice_vf_lib.h | 18 ++++++++++++++++++
 2 files changed, 19 insertions(+), 6 deletions(-)

diff --git a/drivers/net/ethernet/intel/ice/ice_sriov.c b/drivers/net/ethernet/intel/ice/ice_sriov.c
index f78d5d8d516c..c434326a4694 100644
--- a/drivers/net/ethernet/intel/ice/ice_sriov.c
+++ b/drivers/net/ethernet/intel/ice/ice_sriov.c
@@ -933,7 +933,6 @@ int ice_sriov_set_msix_vec_count(struct pci_dev *vf_dev, int msix_vec_count)
 	bool needs_rebuild = false;
 	struct ice_vsi *vsi;
 	struct ice_vf *vf;
-	int id;
 
 	if (!ice_get_num_vfs(pf))
 		return -ENOENT;
@@ -952,11 +951,7 @@ int ice_sriov_set_msix_vec_count(struct pci_dev *vf_dev, int msix_vec_count)
 	if (msix_vec_count < ICE_MIN_INTR_PER_VF)
 		return -EINVAL;
 
-	id = pci_iov_vf_id(vf_dev);
-	if (id < 0)
-		return id;
-
-	vf = ice_get_vf_by_id(pf, id);
+	vf = ice_get_vf_by_dev(pf, vf_dev);
 	if (!vf)
 		return -ENOENT;
 
diff --git a/drivers/net/ethernet/intel/ice/ice_vf_lib.h b/drivers/net/ethernet/intel/ice/ice_vf_lib.h
index a5ee380f8c9e..ffe1f9f830ea 100644
--- a/drivers/net/ethernet/intel/ice/ice_vf_lib.h
+++ b/drivers/net/ethernet/intel/ice/ice_vf_lib.h
@@ -239,6 +239,18 @@ static inline bool ice_vf_is_lldp_ena(struct ice_vf *vf)
 
 #ifdef CONFIG_PCI_IOV
 struct ice_vf *ice_get_vf_by_id(struct ice_pf *pf, u16 vf_id);
+
+static inline struct ice_vf *ice_get_vf_by_dev(struct ice_pf *pf,
+					       struct pci_dev *vf_dev)
+{
+	int vf_id = pci_iov_vf_id(vf_dev);
+
+	if (vf_id < 0)
+		return NULL;
+
+	return ice_get_vf_by_id(pf, pci_iov_vf_id(vf_dev));
+}
+
 void ice_put_vf(struct ice_vf *vf);
 bool ice_has_vfs(struct ice_pf *pf);
 u16 ice_get_num_vfs(struct ice_pf *pf);
@@ -265,6 +277,12 @@ static inline struct ice_vf *ice_get_vf_by_id(struct ice_pf *pf, u16 vf_id)
 	return NULL;
 }
 
+static inline struct ice_vf *ice_get_vf_by_dev(struct ice_pf *pf,
+					       struct pci_dev *vf_dev)
+{
+	return NULL;
+}
+
 static inline void ice_put_vf(struct ice_vf *vf)
 {
 }
-- 
2.47.1

Re: [PATCH net-next 0/8][pull request] ice: cleanups and preparation for live migration

[patchwork-bot+netdevbpf]

Hello:

This series was applied to netdev/net-next.git (main)
by Tony Nguyen <anthony.l.nguyen@intel.com>:

On Thu, 10 Jul 2025 14:45:09 -0700 you wrote:
> Jake Keller says:
> 
> Various cleanups and preparation to the ice driver code for supporting
> SR-IOV live migration.
> 
> The logic for unpacking Rx queue context data is added. This is the inverse
> of the existing packing logic. Thanks to <linux/packing.h> this is trivial
> to add.
> 
> [...]

Here is the summary with links:
  - [net-next,1/8] ice: add support for reading and unpacking Rx queue context
    https://git.kernel.org/netdev/net-next/c/ef41603d09f1
  - [net-next,2/8] ice: add functions to get and set Tx queue context
    https://git.kernel.org/netdev/net-next/c/b6f82e9b79b1
  - [net-next,3/8] ice: save RSS hash configuration for migration
    https://git.kernel.org/netdev/net-next/c/5ff8d9562357
  - [net-next,4/8] ice: move ice_vsi_update_l2tsel to ice_lib.c
    https://git.kernel.org/netdev/net-next/c/4f98ac2d8e53
  - [net-next,5/8] ice: expose VF functions used by live migration
    https://git.kernel.org/netdev/net-next/c/066c2715ada8
  - [net-next,6/8] ice: use pci_iov_vf_id() to get VF ID
    https://git.kernel.org/netdev/net-next/c/4ef21c83ea4b
  - [net-next,7/8] ice: avoid rebuilding if MSI-X vector count is unchanged
    https://git.kernel.org/netdev/net-next/c/922683498e84
  - [net-next,8/8] ice: introduce ice_get_vf_by_dev() wrapper
    https://git.kernel.org/netdev/net-next/c/2d925db5b2c6

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html