From a9364470e9198353e6ed8e28b0b86e6b2b7ce5b3 Mon Sep 17 00:00:00 2001
From: Dengdui Huang <[email protected]>
Date: Fri, 8 Dec 2023 15:44:14 +0800
Subject: [PATCH 25/30] net/hns3: fix VF multiple count on one reset

[ upstream commit 072a07a9dcbd604b1983bf2cb266d3dd4dc89824 ]

There are two ways for the hns3 VF driver to know reset event, namely,
interrupt task and periodic detection task. For the latter, the real
reset process will delay several microseconds to execute. Both tasks
cause the count to increase by 1.

However, the periodic detection task also detects a reset event A
after interrupt task receive a reset event A. As a result, the reset
count will be double.

So this patch adds the comparison of reset level for VF in case of the
multiple reset count.

Fixes: a5475d61fa34 ("net/hns3: support VF")
Cc: [email protected]

Signed-off-by: Dengdui Huang <[email protected]>
Signed-off-by: Jie Hai <[email protected]>
---
 drivers/net/hns3/hns3_ethdev_vf.c | 44 ++++++++++++++++++++-----------
 1 file changed, 29 insertions(+), 15 deletions(-)

diff --git a/drivers/net/hns3/hns3_ethdev_vf.c b/drivers/net/hns3/hns3_ethdev_vf.c
index f5a7a2b..83d3d66 100644
--- a/drivers/net/hns3/hns3_ethdev_vf.c
+++ b/drivers/net/hns3/hns3_ethdev_vf.c
@@ -569,13 +569,8 @@ hns3vf_check_event_cause(struct hns3_adapter *hns, uint32_t *clearval)
 		val = hns3_read_dev(hw, HNS3_VF_RST_ING);
 		hns3_write_dev(hw, HNS3_VF_RST_ING, val | HNS3_VF_RST_ING_BIT);
 		val = cmdq_stat_reg & ~BIT(HNS3_VECTOR0_RST_INT_B);
-		if (clearval) {
-			hw->reset.stats.global_cnt++;
-			hns3_warn(hw, "Global reset detected, clear reset status");
-		} else {
-			hns3_schedule_delayed_reset(hns);
-			hns3_warn(hw, "Global reset detected, don't clear reset status");
-		}
+		hw->reset.stats.global_cnt++;
+		hns3_warn(hw, "Global reset detected, clear reset status");
 
 		ret = HNS3VF_VECTOR0_EVENT_RST;
 		goto out;
@@ -590,9 +585,9 @@ hns3vf_check_event_cause(struct hns3_adapter *hns, uint32_t *clearval)
 
 	val = 0;
 	ret = HNS3VF_VECTOR0_EVENT_OTHER;
+
 out:
-	if (clearval)
-		*clearval = val;
+	*clearval = val;
 	return ret;
 }
 
@@ -1731,11 +1726,25 @@ is_vf_reset_done(struct hns3_hw *hw)
 	return true;
 }
 
+static enum hns3_reset_level
+hns3vf_detect_reset_event(struct hns3_hw *hw)
+{
+	enum hns3_reset_level reset = HNS3_NONE_RESET;
+	uint32_t cmdq_stat_reg;
+
+	cmdq_stat_reg = hns3_read_dev(hw, HNS3_VECTOR0_CMDQ_STAT_REG);
+	if (BIT(HNS3_VECTOR0_RST_INT_B) & cmdq_stat_reg)
+		reset = HNS3_VF_RESET;
+
+	return reset;
+}
+
 bool
 hns3vf_is_reset_pending(struct hns3_adapter *hns)
 {
+	enum hns3_reset_level last_req;
 	struct hns3_hw *hw = &hns->hw;
-	enum hns3_reset_level reset;
+	enum hns3_reset_level new_req;
 
 	/*
 	 * According to the protocol of PCIe, FLR to a PF device resets the PF
@@ -1758,13 +1767,18 @@ hns3vf_is_reset_pending(struct hns3_adapter *hns)
 	if (rte_eal_process_type() != RTE_PROC_PRIMARY)
 		return false;
 
-	hns3vf_check_event_cause(hns, NULL);
-	reset = hns3vf_get_reset_level(hw, &hw->reset.pending);
-	if (hw->reset.level != HNS3_NONE_RESET && reset != HNS3_NONE_RESET &&
-	    hw->reset.level < reset) {
-		hns3_warn(hw, "High level reset %d is pending", reset);
+	new_req = hns3vf_detect_reset_event(hw);
+	if (new_req == HNS3_NONE_RESET)
+		return false;
+
+	last_req = hns3vf_get_reset_level(hw, &hw->reset.pending);
+	if (last_req == HNS3_NONE_RESET || last_req < new_req) {
+		__atomic_store_n(&hw->reset.disable_cmd, 1, __ATOMIC_RELAXED);
+		hns3_schedule_delayed_reset(hns);
+		hns3_warn(hw, "High level reset detected, delay do reset");
 		return true;
 	}
+
 	return false;
 }
 
-- 
2.33.0