[PATCH 1/2] mt76: mt7915: retry pci probe logic up to 3 times.

[PATCH 1/2] mt76: mt7915: retry pci probe logic up to 3 times.

[greearb]

From: Ben Greear <greearb@candelatech.com>

On my x86-64 system with 6 mtk7915 radios, one or more radios
often fails to start upon reboot.  So, do up to 3 retries in
the pci probe logic.  Example of this mitigation working:

[    8.637908] mt7915e 0000:06:00.0: ASIC revision: 79150000
[    8.638026] mt7915e 0000:06:00.0: mt7915:  register_device  Driver-Version: 5.17.0-ct
[   28.863203] mt7915e 0000:06:00.0: MCU: Initial Failure: Message 00000010 (cid 10 ext_cid: 0 seq 1) timeout.  Last successful cmd: 0x0
[   28.873960] mt7915e 0000:06:00.0: Failed to get patch semaphore: -110
[   28.879141] mt7915e 0000:06:00.0: mcu-init: Failed to load firmware, err: -11
[   28.879143] mt7915e 0000:06:00.0: mt7915_register_device failed, ret: -11
[   28.884796] mt7915e 0000:06:00.0: mt7915_pci_probe had error on try 1/3, ret: -11
[   28.891161] mt7915e 0000:06:00.0: ASIC revision: 79150094
[   28.891261] mt7915e 0000:06:00.0: mt7915:  register_device  Driver-Version: 5.17.0-ct
[   29.022875] mt7915e 0000:06:00.0: HW/SW Version: 0x8a108a10, Build Time: 20211222184017a
[   29.033934] mt7915e 0000:06:00.0: WM Firmware Version: ____000000, Build Time: 20211222184052
[   29.051992] mt7915e 0000:06:00.0: WA Firmware Version: DEV_000000, Build Time: 20211222184111
[   29.446014] mt7915e 0000:06:00.0: mt7915_pci_probe succeeded on try 2/3

A quick test of traffic shows the radio works fine after recovery.

Signed-off-by: Ben Greear <greearb@candelatech.com>
---
 .../net/wireless/mediatek/mt76/mt7915/pci.c   | 61 +++++++++++++------
 1 file changed, 44 insertions(+), 17 deletions(-)

diff --git a/drivers/net/wireless/mediatek/mt76/mt7915/pci.c b/drivers/net/wireless/mediatek/mt76/mt7915/pci.c
index 6f819c41a4c4..07cb8a699b75 100644
--- a/drivers/net/wireless/mediatek/mt76/mt7915/pci.c
+++ b/drivers/net/wireless/mediatek/mt76/mt7915/pci.c
@@ -92,8 +92,8 @@ static int mt7915_pci_hif2_probe(struct pci_dev *pdev)
 	return 0;
 }
 
-static int mt7915_pci_probe(struct pci_dev *pdev,
-			    const struct pci_device_id *id)
+static int _mt7915_pci_probe(struct pci_dev *pdev,
+			     const struct pci_device_id *id)
 {
 	struct mt7915_dev *dev;
 	struct mt76_dev *mdev;
@@ -101,20 +101,6 @@ static int mt7915_pci_probe(struct pci_dev *pdev,
 	int irq;
 	int ret;
 
-	ret = pcim_enable_device(pdev);
-	if (ret)
-		return ret;
-
-	ret = pcim_iomap_regions(pdev, BIT(0), pci_name(pdev));
-	if (ret)
-		return ret;
-
-	pci_set_master(pdev);
-
-	ret = dma_set_mask(&pdev->dev, DMA_BIT_MASK(32));
-	if (ret)
-		return ret;
-
 	mt76_pci_disable_aspm(pdev);
 
 	if (id->device == 0x7916 || id->device == 0x790a)
@@ -161,8 +147,11 @@ static int mt7915_pci_probe(struct pci_dev *pdev,
 	}
 
 	ret = mt7915_register_device(dev);
-	if (ret)
+	if (ret) {
+		dev_err(dev->mt76.dev, "mt7915_register_device failed, ret: %d",
+			ret);
 		goto free_hif2_irq;
+	}
 
 	return 0;
 
@@ -181,6 +170,44 @@ static int mt7915_pci_probe(struct pci_dev *pdev,
 	return ret;
 }
 
+static int mt7915_pci_probe(struct pci_dev *pdev,
+			     const struct pci_device_id *id)
+{
+	int z;
+	int ret;
+
+	ret = pcim_enable_device(pdev);
+	if (ret)
+		return ret;
+
+	ret = pcim_iomap_regions(pdev, BIT(0), pci_name(pdev));
+	if (ret)
+		return ret;
+
+	pci_set_master(pdev);
+
+	ret = dma_set_mask(&pdev->dev, DMA_BIT_MASK(32));
+	if (ret)
+		return ret;
+
+	for (z = 0; z<3; z++) {
+		ret = _mt7915_pci_probe(pdev, id);
+		if (ret) {
+			dev_err(&pdev->dev, "mt7915_pci_probe had error on try %d/3, ret: %d",
+			       z + 1, ret);
+		} else {
+			/* It is worth a message to let user know we succeeded if
+			 * earlier attempts failed.
+			 */
+			if (z > 0)
+				dev_info(&pdev->dev, "mt7915_pci_probe succeeded on try %d/3",
+					 z + 1);
+			break;
+		}
+	}
+	return ret;
+}
+
 static void mt7915_hif_remove(struct pci_dev *pdev)
 {
 	struct mt7915_hif *hif = pci_get_drvdata(pdev);
-- 
2.20.1

[PATCH 2/2] mt76: mt7915: improve logging around unhandled events.

[greearb]

From: Ben Greear <greearb@candelatech.com>

It would be good to know of events that firmware is sending that
the driver does not currently handle.

Signed-off-by: Ben Greear <greearb@candelatech.com>
---
 .../net/wireless/mediatek/mt76/mt76_connac_mcu.h   |  1 +
 drivers/net/wireless/mediatek/mt76/mt7915/mcu.c    | 14 ++++++++++++--
 2 files changed, 13 insertions(+), 2 deletions(-)

diff --git a/drivers/net/wireless/mediatek/mt76/mt76_connac_mcu.h b/drivers/net/wireless/mediatek/mt76/mt76_connac_mcu.h
index c3c93338d56a..272b107b7e91 100644
--- a/drivers/net/wireless/mediatek/mt76/mt76_connac_mcu.h
+++ b/drivers/net/wireless/mediatek/mt76/mt76_connac_mcu.h
@@ -820,6 +820,7 @@ enum {
 	MCU_EXT_EVENT_CSA_NOTIFY = 0x4f,
 	MCU_EXT_EVENT_BCC_NOTIFY = 0x75,
 	MCU_EXT_EVENT_MURU_CTRL = 0x9f,
+	MCU_EXT_EVENT_IGMP_FLOODING = 0xBD, /* Seen on 7915 */
 };
 
 enum {
diff --git a/drivers/net/wireless/mediatek/mt76/mt7915/mcu.c b/drivers/net/wireless/mediatek/mt76/mt7915/mcu.c
index d09cf89c4505..83af8e2ca95f 100644
--- a/drivers/net/wireless/mediatek/mt76/mt7915/mcu.c
+++ b/drivers/net/wireless/mediatek/mt76/mt7915/mcu.c
@@ -487,7 +487,13 @@ mt7915_mcu_rx_ext_event(struct mt7915_dev *dev, struct sk_buff *skb)
 	case MCU_EXT_EVENT_BCC_NOTIFY:
 		mt7915_mcu_rx_bcc_notify(dev, skb);
 		break;
+	case MCU_EXT_EVENT_IGMP_FLOODING:
+	case MCU_EXT_EVENT_PS_SYNC:
+		/* ignore some we know we do not care about */
+		break;
 	default:
+		/* in SDK, grep for EventExtEventHandler */
+		dev_info(dev->mt76.dev, "mt7915, unhandled rx_ext_event: 0x%x", rxd->ext_eid);
 		break;
 	}
 }
@@ -502,6 +508,7 @@ mt7915_mcu_rx_unsolicited_event(struct mt7915_dev *dev, struct sk_buff *skb)
 		mt7915_mcu_rx_ext_event(dev, skb);
 		break;
 	default:
+		dev_info(dev->mt76.dev, "mt7915, unhandled unsolicited event: 0x%x", rxd->eid);
 		break;
 	}
 	dev_kfree_skb(skb);
@@ -2090,7 +2097,7 @@ static int mt7915_load_patch(struct mt7915_dev *dev)
 	case PATCH_NOT_DL_SEM_SUCCESS:
 		break;
 	default:
-		dev_err(dev->mt76.dev, "Failed to get patch semaphore\n");
+		dev_err(dev->mt76.dev, "Failed to get patch semaphore: %d", sem);
 		return -EAGAIN;
 	}
 
@@ -2530,8 +2537,11 @@ int mt7915_mcu_init(struct mt7915_dev *dev)
 	}
 
 	ret = mt7915_load_firmware(dev);
-	if (ret)
+	if (ret) {
+		dev_info(dev->mt76.dev, "mcu-init: Failed to load firmware, err: %d",
+			 ret);
 		return ret;
+	}
 
 	set_bit(MT76_STATE_MCU_RUNNING, &dev->mphy.state);
 
-- 
2.20.1