i40e: refactor VF reset flow

Fix the VF reset flow so that it works on real hardware. After
discussions with the HW team, the reset flow has been changed
somewhat.

- Change the i40e_reset_vf function to a void type, and fix
  up the callers to reflect this.
- Move the MSI-X disable code to i40e_free_vf_res since it must
  be done every time the VF is freed, regardless of whether or
  not it is reset.
- Ensure that the PCIe bus is quiet before polling the reset bit.
- Don't clear the VFGEN_RSTAT1 register at the beginning as it is
  cleared by the reset.
- Poll longer for the reset to be done.
- Disable the queues using an existing function rather than
  rolling our own.
- Free and reallocate the VSI after reset to avoid rx hang.

Change-Id: I11e2590431cb73e8663714d1cc5b23d59b809033
Signed-off-by: Mitch Williams <mitch.a.williams@intel.com>
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Tested-by: Kavindya Deegala <kavindya.s.deegala@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

drivers/net/ethernet/intel/i40e/i40e.h

@@ -543,6 +543,7 @@ struct i40e_vsi *i40e_vsi_setup(struct i40e_pf *pf, u8 type,
 int i40e_vsi_release(struct i40e_vsi *vsi);
 struct i40e_vsi *i40e_vsi_lookup(struct i40e_pf *pf, enum i40e_vsi_type type,
 				 struct i40e_vsi *start_vsi);
+int i40e_vsi_control_rings(struct i40e_vsi *vsi, bool enable);
 int i40e_reconfig_rss_queues(struct i40e_pf *pf, int queue_count);
 struct i40e_veb *i40e_veb_setup(struct i40e_pf *pf, u16 flags, u16 uplink_seid,
 				u16 downlink_seid, u8 enabled_tc);

drivers/net/ethernet/intel/i40e/i40e_main.c

@@ -3110,7 +3110,7 @@ static int i40e_vsi_control_rx(struct i40e_vsi *vsi, bool enable)
  * @vsi: the VSI being configured
  * @enable: start or stop the rings
  **/
-static int i40e_vsi_control_rings(struct i40e_vsi *vsi, bool request)
+int i40e_vsi_control_rings(struct i40e_vsi *vsi, bool request)
 {
 	int ret;
 

drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c

@@ -621,6 +621,9 @@ static void i40e_disable_vf_mappings(struct i40e_vf *vf)
 static void i40e_free_vf_res(struct i40e_vf *vf)
 {
 	struct i40e_pf *pf = vf->pf;
+	struct i40e_hw *hw = &pf->hw;
+	u32 reg_idx, reg;
+	int i, msix_vf;
 
 	/* free vsi & disconnect it from the parent uplink */
 	if (vf->lan_vsi_index) {
@@ -628,7 +631,34 @@ static void i40e_free_vf_res(struct i40e_vf *vf)
 		vf->lan_vsi_index = 0;
 		vf->lan_vsi_id = 0;
 	}
+	msix_vf = pf->hw.func_caps.num_msix_vectors_vf + 1;
+	/* disable interrupts so the VF starts in a known state */
+	for (i = 0; i < msix_vf; i++) {
+		/* format is same for both registers */
+		if (0 == i)
+			reg_idx = I40E_VFINT_DYN_CTL0(vf->vf_id);
+		else
+			reg_idx = I40E_VFINT_DYN_CTLN(((msix_vf - 1) *
+						      (vf->vf_id))
+						     + (i - 1));
+		wr32(hw, reg_idx, I40E_VFINT_DYN_CTLN_CLEARPBA_MASK);
+		i40e_flush(hw);
+	}
 
+	/* clear the irq settings */
+	for (i = 0; i < msix_vf; i++) {
+		/* format is same for both registers */
+		if (0 == i)
+			reg_idx = I40E_VPINT_LNKLST0(vf->vf_id);
+		else
+			reg_idx = I40E_VPINT_LNKLSTN(((msix_vf - 1) *
+						      (vf->vf_id))
+						     + (i - 1));
+		reg = (I40E_VPINT_LNKLSTN_FIRSTQ_TYPE_MASK |
+		       I40E_VPINT_LNKLSTN_FIRSTQ_INDX_MASK);
+		wr32(hw, reg_idx, reg);
+		i40e_flush(hw);
+	}
 	/* reset some of the state varibles keeping
 	 * track of the resources
 	 */
@@ -670,6 +700,36 @@ static int i40e_alloc_vf_res(struct i40e_vf *vf)
 	return ret;
 }
 
+#define VF_DEVICE_STATUS 0xAA
+#define VF_TRANS_PENDING_MASK 0x20
+/**
+ * i40e_quiesce_vf_pci
+ * @vf: pointer to the vf structure
+ *
+ * Wait for VF PCI transactions to be cleared after reset. Returns -EIO
+ * if the transactions never clear.
+ **/
+static int i40e_quiesce_vf_pci(struct i40e_vf *vf)
+{
+	struct i40e_pf *pf = vf->pf;
+	struct i40e_hw *hw = &pf->hw;
+	int vf_abs_id, i;
+	u32 reg;
+
+	reg = rd32(hw, I40E_PF_VT_PFALLOC);
+	vf_abs_id = vf->vf_id + (reg & I40E_PF_VT_PFALLOC_FIRSTVF_MASK);
+
+	wr32(hw, I40E_PF_PCI_CIAA,
+	     VF_DEVICE_STATUS | (vf_abs_id << I40E_PF_PCI_CIAA_VF_NUM_SHIFT));
+	for (i = 0; i < 100; i++) {
+		reg = rd32(hw, I40E_PF_PCI_CIAD);
+		if ((reg & VF_TRANS_PENDING_MASK) == 0)
+			return 0;
+		udelay(1);
+	}
+	return -EIO;
+}
+
 /**
  * i40e_reset_vf
  * @vf: pointer to the vf structure
@@ -677,35 +737,36 @@ static int i40e_alloc_vf_res(struct i40e_vf *vf)
  *
  * reset the vf
  **/
-int i40e_reset_vf(struct i40e_vf *vf, bool flr)
+void i40e_reset_vf(struct i40e_vf *vf, bool flr)
 {
-	int ret = -ENOENT;
 	struct i40e_pf *pf = vf->pf;
 	struct i40e_hw *hw = &pf->hw;
-	u32 reg, reg_idx, msix_vf;
 	bool rsd = false;
-	u16 pf_queue_id;
-	int i, j;
+	int i;
+	u32 reg;
 
 	/* warn the VF */
-	wr32(hw, I40E_VFGEN_RSTAT1(vf->vf_id), I40E_VFR_INPROGRESS);
-
 	clear_bit(I40E_VF_STAT_ACTIVE, &vf->vf_states);
 
-	/* PF triggers VFR only when VF requests, in case of
-	 * VFLR, HW triggers VFR
+	/* In the case of a VFLR, the HW has already reset the VF and we
+	 * just need to clean up, so don't hit the VFRTRIG register.
 	 */
 	if (!flr) {
 		/* reset vf using VPGEN_VFRTRIG reg */
-		reg = I40E_VPGEN_VFRTRIG_VFSWR_MASK;
+		reg = rd32(hw, I40E_VPGEN_VFRTRIG(vf->vf_id));
+		reg |= I40E_VPGEN_VFRTRIG_VFSWR_MASK;
 		wr32(hw, I40E_VPGEN_VFRTRIG(vf->vf_id), reg);
 		i40e_flush(hw);
 	}
 
+	if (i40e_quiesce_vf_pci(vf))
+		dev_err(&pf->pdev->dev, "VF %d PCI transactions stuck\n",
+			vf->vf_id);
+
 	/* poll VPGEN_VFRSTAT reg to make sure
 	 * that reset is complete
 	 */
-	for (i = 0; i < 4; i++) {
+	for (i = 0; i < 100; i++) {
 		/* vf reset requires driver to first reset the
 		 * vf & than poll the status register to make sure
 		 * that the requested op was completed
@@ -720,84 +781,29 @@ int i40e_reset_vf(struct i40e_vf *vf, bool flr)
 	}
 
 	if (!rsd)
-		dev_err(&pf->pdev->dev, "VF reset check timeout %d\n",
+		dev_err(&pf->pdev->dev, "VF reset check timeout on VF %d\n",
 			vf->vf_id);
-
-	/* fast disable qps */
-	for (j = 0; j < pf->vsi[vf->lan_vsi_index]->num_queue_pairs; j++) {
-		ret = i40e_ctrl_vsi_tx_queue(vf, vf->lan_vsi_index, j,
-					     I40E_QUEUE_CTRL_FASTDISABLE);
-		ret = i40e_ctrl_vsi_rx_queue(vf, vf->lan_vsi_index, j,
-					     I40E_QUEUE_CTRL_FASTDISABLE);
-	}
-
-	/* Queue enable/disable requires driver to
-	 * first reset the vf & than poll the status register
-	 * to make sure that the requested op was completed
-	 * successfully
-	 */
-	udelay(10);
-	for (j = 0; j < pf->vsi[vf->lan_vsi_index]->num_queue_pairs; j++) {
-		ret = i40e_ctrl_vsi_tx_queue(vf, vf->lan_vsi_index, j,
-					     I40E_QUEUE_CTRL_FASTDISABLECHECK);
-		if (ret)
-			dev_info(&pf->pdev->dev,
-				 "Queue control check failed on Tx queue %d of VSI %d VF %d\n",
-				 j, vf->lan_vsi_index, vf->vf_id);
-		ret = i40e_ctrl_vsi_rx_queue(vf, vf->lan_vsi_index, j,
-					     I40E_QUEUE_CTRL_FASTDISABLECHECK);
-		if (ret)
-			dev_info(&pf->pdev->dev,
-				 "Queue control check failed on Rx queue %d of VSI %d VF %d\n",
-				 j, vf->lan_vsi_index, vf->vf_id);
-	}
-
-	/* clear the irq settings */
-	msix_vf = pf->hw.func_caps.num_msix_vectors_vf;
-	for (i = 0; i < msix_vf; i++) {
-		/* format is same for both registers */
-		if (0 == i)
-			reg_idx = I40E_VPINT_LNKLST0(vf->vf_id);
-		else
-			reg_idx = I40E_VPINT_LNKLSTN(((msix_vf - 1) *
-						      (vf->vf_id))
-						     + (i - 1));
-		reg = (I40E_VPINT_LNKLSTN_FIRSTQ_TYPE_MASK |
-		       I40E_VPINT_LNKLSTN_FIRSTQ_INDX_MASK);
-		wr32(hw, reg_idx, reg);
-		i40e_flush(hw);
-	}
-	/* disable interrupts so the VF starts in a known state */
-	for (i = 0; i < msix_vf; i++) {
-		/* format is same for both registers */
-		if (0 == i)
-			reg_idx = I40E_VFINT_DYN_CTL0(vf->vf_id);
-		else
-			reg_idx = I40E_VFINT_DYN_CTLN(((msix_vf - 1) *
-						      (vf->vf_id))
-						     + (i - 1));
-		wr32(hw, reg_idx, I40E_VFINT_DYN_CTLN_CLEARPBA_MASK);
-		i40e_flush(hw);
-	}
-
-	/* set the defaults for the rqctl & tqctl registers */
-	reg = (I40E_QINT_RQCTL_NEXTQ_INDX_MASK | I40E_QINT_RQCTL_ITR_INDX_MASK |
-	       I40E_QINT_RQCTL_NEXTQ_TYPE_MASK);
-	for (j = 0; j < pf->vsi[vf->lan_vsi_index]->num_queue_pairs; j++) {
-		pf_queue_id = i40e_vc_get_pf_queue_id(vf, vf->lan_vsi_index, j);
-		wr32(hw, I40E_QINT_RQCTL(pf_queue_id), reg);
-		wr32(hw, I40E_QINT_TQCTL(pf_queue_id), reg);
-	}
-
+	wr32(hw, I40E_VFGEN_RSTAT1(vf->vf_id), I40E_VFR_COMPLETED);
 	/* clear the reset bit in the VPGEN_VFRTRIG reg */
 	reg = rd32(hw, I40E_VPGEN_VFRTRIG(vf->vf_id));
 	reg &= ~I40E_VPGEN_VFRTRIG_VFSWR_MASK;
 	wr32(hw, I40E_VPGEN_VFRTRIG(vf->vf_id), reg);
+
+	/* On initial reset, we won't have any queues */
+	if (vf->lan_vsi_index == 0)
+		goto complete_reset;
+
+	i40e_vsi_control_rings(pf->vsi[vf->lan_vsi_index], false);
+complete_reset:
+	/* reallocate vf resources to reset the VSI state */
+	i40e_free_vf_res(vf);
+	mdelay(10);
+	i40e_alloc_vf_res(vf);
+	i40e_enable_vf_mappings(vf);
+
 	/* tell the VF the reset is done */
-	wr32(hw, I40E_VFGEN_RSTAT1(vf->vf_id), I40E_VFR_COMPLETED);
+	wr32(hw, I40E_VFGEN_RSTAT1(vf->vf_id), I40E_VFR_VFACTIVE);
 	i40e_flush(hw);
-
-	return ret;
 }
 
 /**
@@ -909,11 +915,8 @@ static int i40e_alloc_vfs(struct i40e_pf *pf, u16 num_alloc_vfs)
 
 		/* assign default capabilities */
 		set_bit(I40E_VIRTCHNL_VF_CAP_L2, &vfs[i].vf_caps);
-
-		ret = i40e_alloc_vf_res(&vfs[i]);
-		i40e_reset_vf(&vfs[i], true);
-		if (ret)
-			break;
+		/* vf resources get allocated during reset */
+		i40e_reset_vf(&vfs[i], false);
 
 		/* enable vf vplan_qtable mappings */
 		i40e_enable_vf_mappings(&vfs[i]);
@@ -1140,12 +1143,10 @@ static int i40e_vc_get_vf_resources_msg(struct i40e_vf *vf)
  * unlike other virtchnl messages, pf driver
  * doesn't send the response back to the vf
  **/
-static int i40e_vc_reset_vf_msg(struct i40e_vf *vf)
+static void i40e_vc_reset_vf_msg(struct i40e_vf *vf)
 {
-	if (!test_bit(I40E_VF_STAT_ACTIVE, &vf->vf_states))
-		return -ENOENT;
-
-	return i40e_reset_vf(vf, false);
+	if (test_bit(I40E_VF_STAT_ACTIVE, &vf->vf_states))
+		i40e_reset_vf(vf, false);
 }
 
 /**
@@ -1879,7 +1880,8 @@ int i40e_vc_process_vf_msg(struct i40e_pf *pf, u16 vf_id, u32 v_opcode,
 		ret = i40e_vc_get_vf_resources_msg(vf);
 		break;
 	case I40E_VIRTCHNL_OP_RESET_VF:
-		ret = i40e_vc_reset_vf_msg(vf);
+		i40e_vc_reset_vf_msg(vf);
+		ret = 0;
 		break;
 	case I40E_VIRTCHNL_OP_CONFIG_PROMISCUOUS_MODE:
 		ret = i40e_vc_config_promiscuous_mode_msg(vf, msg, msglen);
@@ -1950,19 +1952,7 @@ int i40e_vc_process_vflr_event(struct i40e_pf *pf)
 			/* clear the bit in GLGEN_VFLRSTAT */
 			wr32(hw, I40E_GLGEN_VFLRSTAT(reg_idx), (1 << bit_idx));
 
-			if (i40e_reset_vf(vf, true))
-				dev_err(&pf->pdev->dev,
-					"Unable to reset the VF %d\n", vf_id);
-			/* free up vf resources to destroy vsi state */
-			i40e_free_vf_res(vf);
-
-			/* allocate new vf resources with the default state */
-			if (i40e_alloc_vf_res(vf))
-				dev_err(&pf->pdev->dev,
-					"Unable to allocate VF resources %d\n",
-					vf_id);
-
-			i40e_enable_vf_mappings(vf);
+			i40e_reset_vf(vf, true);
 		}
 	}
 

drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.h

@@ -104,7 +104,7 @@ int i40e_pci_sriov_configure(struct pci_dev *dev, int num_vfs);
 int i40e_vc_process_vf_msg(struct i40e_pf *pf, u16 vf_id, u32 v_opcode,
 			   u32 v_retval, u8 *msg, u16 msglen);
 int i40e_vc_process_vflr_event(struct i40e_pf *pf);
-int i40e_reset_vf(struct i40e_vf *vf, bool flr);
+void i40e_reset_vf(struct i40e_vf *vf, bool flr);
 void i40e_vc_notify_vf_reset(struct i40e_vf *vf);
 
 /* vf configuration related iplink handlers */