linux-kernel - [PATCH v3 4/6] bus: mhi: host: pci_generic: Remove MHI driver and ensure graceful device recovery

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20250821-vdev_next-20250821_sriov-v3-4-e1b017c48d4a@quicinc.com>
Date: Thu, 21 Aug 2025 18:25:36 +0530
From: Vivek.Pernamitta@...cinc.com
To: Manivannan Sadhasivam <mani@...nel.org>
CC: <mhi@...ts.linux.dev>, <linux-arm-msm@...r.kernel.org>,
        <linux-kernel@...r.kernel.org>,
        Vivek Pernamitta <quic_vpernami@...cinc.com>
Subject: [PATCH v3 4/6] bus: mhi: host: pci_generic: Remove MHI driver and
 ensure graceful device recovery

From: Vivek Pernamitta <quic_vpernami@...cinc.com>

So, When the MHI driver is removed from the host side, it is essential to
ensure a clean and stable recovery of the device. This commit introduces
the following steps to achieve that:

1. Disable SR-IOV for any SR-IOV-enabled devices on the Physical Function.
2. Perform a SOC_RESET on the PF to fully reset the device.

Disabling SR-IOV ensures all Virtual Functions (VFs) are properly shutdown,
preventing issues during the reset process. The SOC_RESET guarantees that
the PF is restored to a known good state.

Note:
- The QDU100 platform supports 1 PF and 16 VFs.
- QDU100 does not support Function Level Reset (FLR) due to a hardware
  limitation. As a result, SOC_RESET is used to reset the device.
- On QDU100, any VF failure can cause the entire endpoint (EP)
  to go down, making this recovery mechanism critical.

Signed-off-by: Vivek Pernamitta <quic_vpernami@...cinc.com>
---
 drivers/bus/mhi/host/pci_generic.c | 14 +++++++++++++-
 1 file changed, 13 insertions(+), 1 deletion(-)

diff --git a/drivers/bus/mhi/host/pci_generic.c b/drivers/bus/mhi/host/pci_generic.c
index 351b177cdf84057fb5a4e2f5b52279d7f1da41c2..f922cca0ab633aeae942587f0c40038342ce9c33 100644
--- a/drivers/bus/mhi/host/pci_generic.c
+++ b/drivers/bus/mhi/host/pci_generic.c
@@ -45,6 +45,8 @@
  * @sideband_wake: Devices using dedicated sideband GPIO for wakeup instead
  *		   of inband wake support (such as sdx24)
  * @no_m3: M3 not supported
+ * @reset_on_remove: Set true for devices support SOC reset and perform it
+ *		     while drivee remove
  */
 struct mhi_pci_dev_info {
 	const struct mhi_controller_config *config;
@@ -58,6 +60,7 @@ struct mhi_pci_dev_info {
 	unsigned int mru_default;
 	bool sideband_wake;
 	bool no_m3;
+	bool reset_on_remove;
 };
 
 #define MHI_CHANNEL_CONFIG_UL(ch_num, ch_name, el_count, ev_ring) \
@@ -300,6 +303,7 @@ static const struct mhi_pci_dev_info mhi_qcom_qdu100_info = {
 	.dma_data_width = 32,
 	.sideband_wake = false,
 	.no_m3 = true,
+	.reset_on_remove = true,
 };
 
 static const struct mhi_channel_config mhi_qcom_sa8775p_channels[] = {
@@ -1039,6 +1043,7 @@ struct mhi_pci_device {
 	struct work_struct recovery_work;
 	struct timer_list health_check_timer;
 	unsigned long status;
+	bool reset_on_remove;
 };
 
 static int mhi_pci_read_reg(struct mhi_controller *mhi_cntrl,
@@ -1323,7 +1328,6 @@ static int mhi_pci_probe(struct pci_dev *pdev, const struct pci_device_id *id)
 	else
 		mhi_cntrl_config = info->config;
 
-	/* Initialize health check monitor only for Physical functions */
 	if (pdev->is_physfn)
 		timer_setup(&mhi_pdev->health_check_timer, health_check, 0);
 
@@ -1343,6 +1347,9 @@ static int mhi_pci_probe(struct pci_dev *pdev, const struct pci_device_id *id)
 	mhi_cntrl->mru = info->mru_default;
 	mhi_cntrl->name = info->name;
 
+	if (pdev->is_physfn)
+		mhi_pdev->reset_on_remove = info->reset_on_remove;
+
 	if (info->edl_trigger)
 		mhi_cntrl->edl_trigger = mhi_pci_generic_edl_trigger;
 
@@ -1415,6 +1422,8 @@ static void mhi_pci_remove(struct pci_dev *pdev)
 	struct mhi_pci_device *mhi_pdev = pci_get_drvdata(pdev);
 	struct mhi_controller *mhi_cntrl = &mhi_pdev->mhi_cntrl;
 
+	pci_disable_sriov(pdev);
+
 	if (pdev->is_physfn)
 		timer_delete_sync(&mhi_pdev->health_check_timer);
 	cancel_work_sync(&mhi_pdev->recovery_work);
@@ -1428,6 +1437,9 @@ static void mhi_pci_remove(struct pci_dev *pdev)
 	if (pci_pme_capable(pdev, PCI_D3hot))
 		pm_runtime_get_noresume(&pdev->dev);
 
+	if (mhi_pdev->reset_on_remove)
+		mhi_soc_reset(mhi_cntrl);
+
 	mhi_unregister_controller(mhi_cntrl);
 }
 

-- 
2.34.1