Skip to contentSkip to conversation

mpi3mr: Support PCIe Error Recovery

PATCH In review Ranjan Kumar 2 participants 10 messages Dec 14, 2023 – Jan 5, 2024

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

v2 cover mpi3mr: Support PCIe Error Recovery

Patches are related to "Advance Error Recovery"
and are tightly coupled and the remaining two patches are defect fix.

v1->v2:
- AER patch split as suggested by Bjorn Helgaas
- Updated driver version to a new value

Ranjan Kumar (6):
mpi3mr: Creation of helper function
mpi3mr: Support PCIe Error Recovery callback handlers
mpi3mr: Prevent PCI writes from driver during PCI error recovery
mpi3mr: Improve Shutdown times when firmware has faulted
mpi3mr: Reset stop_bsgs flag post controller reset failure
mpi3mr: Update driver version to 8.6.1.0.50

 drivers/scsi/mpi3mr/mpi3mr.h           |  36 ++-
 drivers/scsi/mpi3mr/mpi3mr_app.c       |  64 ++---
 drivers/scsi/mpi3mr/mpi3mr_fw.c        |  65 +++--
 drivers/scsi/mpi3mr/mpi3mr_os.c        | 322 +++++++++++++++++++++----
 drivers/scsi/mpi3mr/mpi3mr_transport.c |  39 ++-
 5 files changed, 435 insertions(+), 91 deletions(-)
Signature · 1 line
2.31.1
#1 lore raw

PATCH v2 1/6 mpi3mr: Creation of helper function

Use of helper function to get controller and shost details.

---
 drivers/scsi/mpi3mr/mpi3mr_os.c | 54 ++++++++++++++++++++++++++-------
 1 file changed, 43 insertions(+), 11 deletions(-)
drivers/scsi/mpi3mr/mpi3mr_os.c +43 -11
index 1bffd629c124..76ba31a9517d 100644
@@ -5230,6 +5230,35 @@ mpi3mr_probe(struct pci_dev *pdev, const struct pci_device_id *id)
 	return retval;
 }
 
+/**
+ * mpi3mr_get_shost_and_mrioc - get shost and ioc reference if
+ *                     they are valid
+ * @pdev: PCI device struct
+ * @shost: address to store scsi host reference
+ * @mrioc: address store HBA adapter reference
+ *
+ * Return: 0 if *shost and *ioc are not NULL otherwise -1.
+ */
+
+static int
+mpi3mr_get_shost_and_mrioc(struct pci_dev *pdev,
+	struct Scsi_Host **shost, struct mpi3mr_ioc **mrioc)
+{
+	*shost = pci_get_drvdata(pdev);
+	if (*shost == NULL) {
+		dev_err(&pdev->dev, "pdev's driver data is null\n");
+		return -ENXIO;
+	}
+
+	*mrioc = shost_priv(*shost);
+	if (*mrioc == NULL) {
+		dev_err(&pdev->dev, "shost's private data is null\n");
+		*shost = NULL;
+		return -ENXIO;
+		}
+	return 0;
+}
+
 /**
  * mpi3mr_remove - PCI remove callback
  * @pdev: PCI device instance
@@ -5242,7 +5271,7 @@ mpi3mr_probe(struct pci_dev *pdev, const struct pci_device_id *id)
  */
 static void mpi3mr_remove(struct pci_dev *pdev)
 {
-	struct Scsi_Host *shost = pci_get_drvdata(pdev);
+	struct Scsi_Host *shost;
 	struct mpi3mr_ioc *mrioc;
 	struct workqueue_struct	*wq;
 	unsigned long flags;
@@ -5250,7 +5279,7 @@ static void mpi3mr_remove(struct pci_dev *pdev)
 	struct mpi3mr_hba_port *port, *hba_port_next;
 	struct mpi3mr_sas_node *sas_expander, *sas_expander_next;
 
-	if (!shost)
+	if (mpi3mr_get_shost_and_mrioc(pdev, &shost, &mrioc))
 		return;
 
 	mrioc = shost_priv(shost);
@@ -5328,12 +5357,12 @@ static void mpi3mr_remove(struct pci_dev *pdev)
  */
 static void mpi3mr_shutdown(struct pci_dev *pdev)
 {
-	struct Scsi_Host *shost = pci_get_drvdata(pdev);
+	struct Scsi_Host *shost;
 	struct mpi3mr_ioc *mrioc;
 	struct workqueue_struct	*wq;
 	unsigned long flags;
 
-	if (!shost)
+	if (mpi3mr_get_shost_and_mrioc(pdev, &shost, &mrioc))
 		return;
 
 	mrioc = shost_priv(shost);
@@ -5361,17 +5390,19 @@ static void mpi3mr_shutdown(struct pci_dev *pdev)
  * Change the power state to the given value and cleanup the IOC
  * by issuing MUR and shutdown notification
  *
- * Return: 0 always.
+ * Return: 0 on success, non-zero on failure
  */
 static int __maybe_unused
 mpi3mr_suspend(struct device *dev)
 {
 	struct pci_dev *pdev = to_pci_dev(dev);
-	struct Scsi_Host *shost = pci_get_drvdata(pdev);
+	struct Scsi_Host *shost;
 	struct mpi3mr_ioc *mrioc;
+	int rc;
 
-	if (!shost)
-		return 0;
+	rc = mpi3mr_get_shost_and_mrioc(pdev, &shost, &mrioc);
+	if (rc)
+		return rc;
 
 	mrioc = shost_priv(shost);
 	while (mrioc->reset_in_progress || mrioc->is_driver_loading)
@@ -5402,13 +5433,14 @@ static int __maybe_unused
 mpi3mr_resume(struct device *dev)
 {
 	struct pci_dev *pdev = to_pci_dev(dev);
-	struct Scsi_Host *shost = pci_get_drvdata(pdev);
+	struct Scsi_Host *shost;
 	struct mpi3mr_ioc *mrioc;
 	pci_power_t device_state = pdev->current_state;
 	int r;
 
-	if (!shost)
-		return 0;
+	r = mpi3mr_get_shost_and_mrioc(pdev, &shost, &mrioc);
+	if (r)
+		return r;
 
 	mrioc = shost_priv(shost);
 
Signature · 1 line
2.31.1
#2 lore raw

Re: [PATCH v2 1/6] mpi3mr: Creation of helper function

I wasn't cc'd on this series, so apologies if it's already been applied and these comments are moot.

BTW, the cover doesn't say what this applies to, and it doesn't apply cleanly to v6.7-rc1.

Quoted 16 lines · Ranjan Kumar

Use of helper function to get controller and shost details.

---
 drivers/scsi/mpi3mr/mpi3mr_os.c | 54 ++++++++++++++++++++++++++-------
 1 file changed, 43 insertions(+), 11 deletions(-)
drivers/scsi/mpi3mr/mpi3mr_os.c +0 -0
index 1bffd629c124..76ba31a9517d 100644
@@ -5230,6 +5230,35 @@ mpi3mr_probe(struct pci_dev *pdev, const struct pci_device_id *id)
 	return retval;
 }
 

drivers/scsi/mpi3mr/mpi3mr_os.c mpi3mr_probe(struct pci_dev *pdev, const struct pci_device_id *id) · line 5242

Quoted 10 lines
drivers/scsi/mpi3mr/mpi3mr_os.c +10 -0
+/**
+ * mpi3mr_get_shost_and_mrioc - get shost and ioc reference if
+ *                     they are valid
+ * @pdev: PCI device struct
+ * @shost: address to store scsi host reference
+ * @mrioc: address store HBA adapter reference
+ *
+ * Return: 0 if *shost and *ioc are not NULL otherwise -1.
+ */
+

Spurious blank line (I see that mpi3mr_probe() has a blank line here, but most functions in this file do not).

drivers/scsi/mpi3mr/mpi3mr_os.c mpi3mr_probe(struct pci_dev *pdev, const struct pci_device_id *id) · line 5244

Quoted 2 lines
drivers/scsi/mpi3mr/mpi3mr_os.c +2 -0
+static int
+mpi3mr_get_shost_and_mrioc(struct pci_dev *pdev,

Most functions in this file use this style:

static int mpi3mr_get_shost_and_mrioc(struct pci_dev *pdev,

Quoted 30 lines
drivers/scsi/mpi3mr/mpi3mr_os.c +18 -1
+	struct Scsi_Host **shost, struct mpi3mr_ioc **mrioc)
+{
+	*shost = pci_get_drvdata(pdev);
+	if (*shost == NULL) {
+		dev_err(&pdev->dev, "pdev's driver data is null\n");
+		return -ENXIO;
+	}
+
+	*mrioc = shost_priv(*shost);
+	if (*mrioc == NULL) {
+		dev_err(&pdev->dev, "shost's private data is null\n");
+		*shost = NULL;
+		return -ENXIO;
+		}
+	return 0;
+}
+
 /**
  * mpi3mr_remove - PCI remove callback
  * @pdev: PCI device instance
@@ -5242,7 +5271,7 @@ mpi3mr_probe(struct pci_dev *pdev, const struct pci_device_id *id)
  */
 static void mpi3mr_remove(struct pci_dev *pdev)
 {
-	struct Scsi_Host *shost = pci_get_drvdata(pdev);
+	struct Scsi_Host *shost;
 	struct mpi3mr_ioc *mrioc;
 	struct workqueue_struct	*wq;
 	unsigned long flags;

drivers/scsi/mpi3mr/mpi3mr_os.c static void mpi3mr_remove(struct pci_dev *pdev) · line 5285

Quoted 8 lines
drivers/scsi/mpi3mr/mpi3mr_os.c +1 -1
@@ -5250,7 +5279,7 @@ static void mpi3mr_remove(struct pci_dev *pdev)
 	struct mpi3mr_hba_port *port, *hba_port_next;
 	struct mpi3mr_sas_node *sas_expander, *sas_expander_next;
 
-	if (!shost)
+	if (mpi3mr_get_shost_and_mrioc(pdev, &shost, &mrioc))
 		return;
 
 	mrioc = shost_priv(shost);

Maybe I'm missing something, but it looks like this patch should remove "mrioc = shost_priv(shost)" every time it adds a call to mpi3mr_get_shost_and_mrioc().

Bjorn

#3 lore raw

PATCH v2 2/6 mpi3mr: Support PCIe Error Recovery callback handlers

In reply to Ranjan Kumar

The driver has been upgraded to include support for the PCIe error recovery callback handler which is crucial for the recovery of the controllers. This feature is necessary for addressing the errors reported by the PCIe AER (Advanced Error Reporting) mechanism.

---
 drivers/scsi/mpi3mr/mpi3mr.h    |   5 +
 drivers/scsi/mpi3mr/mpi3mr_os.c | 197 ++++++++++++++++++++++++++++++++
 2 files changed, 202 insertions(+)
drivers/scsi/mpi3mr/mpi3mr.h +5 -0
index 3de1ee05c44e..25e6e3a09468 100644
@@ -23,6 +23,7 @@
 #include <linux/miscdevice.h>
 #include <linux/module.h>
 #include <linux/pci.h>
+#include <linux/aer.h>
 #include <linux/poll.h>
 #include <linux/sched.h>
 #include <linux/slab.h>
@@ -1055,6 +1056,8 @@ struct scmd_priv {
  * @ioctl_chain_sge: DMA buffer descriptor for IOCTL chain
  * @ioctl_resp_sge: DMA buffer descriptor for Mgmt cmd response
  * @ioctl_sges_allocated: Flag for IOCTL SGEs allocated or not
+ * @pcie_err_recovery: PCIe error recovery in progress
+ * @block_on_pcie_err: Block IO during PCI error recovery
  */
 struct mpi3mr_ioc {
 	struct list_head list;
@@ -1246,6 +1249,8 @@ struct mpi3mr_ioc {
 	struct dma_memory_desc ioctl_chain_sge;
 	struct dma_memory_desc ioctl_resp_sge;
 	bool ioctl_sges_allocated;
+	bool pcie_err_recovery;
+	bool block_on_pcie_err;
 };
 
 /**
drivers/scsi/mpi3mr/mpi3mr_os.c +197 -0

215 lines not on this page · raw message

Signature · 1 line
2.31.1
#4 lore raw

Re: [PATCH v2 2/6] mpi3mr: Support PCIe Error Recovery callback handlers

Quoted 5 lines · Ranjan Kumar

The driver has been upgraded to include support for the PCIe error recovery callback handler which is crucial for the recovery of the controllers. This feature is necessary for addressing the errors reported by the PCIe AER (Advanced Error Reporting) mechanism.

I think PCI error handling is not strictly limited to PCIe AER because the powerpc EEH framework also uses the pci_error_handlers callbacks, so I would refer to this as just "PCI error recovery" and maybe omit the AER part.

Commit log and comments appear to be wrapped to 60 or 70 columns, which is unnecessarily short. I'd recommend 75 columns for commit log, maybe 78 for comments.

Quoted 20 lines
---
 drivers/scsi/mpi3mr/mpi3mr.h    |   5 +
 drivers/scsi/mpi3mr/mpi3mr_os.c | 197 ++++++++++++++++++++++++++++++++
 2 files changed, 202 insertions(+)
drivers/scsi/mpi3mr/mpi3mr.h +1 -0
index 3de1ee05c44e..25e6e3a09468 100644
@@ -23,6 +23,7 @@
 #include <linux/miscdevice.h>
 #include <linux/module.h>
 #include <linux/pci.h>
+#include <linux/aer.h>
 #include <linux/poll.h>
 #include <linux/sched.h>
 #include <linux/slab.h>

drivers/scsi/mpi3mr/mpi3mr.h struct scmd_priv { · line 1060

Quoted 5 lines
drivers/scsi/mpi3mr/mpi3mr.h +2 -0
@@ -1055,6 +1056,8 @@ struct scmd_priv {
  * @ioctl_chain_sge: DMA buffer descriptor for IOCTL chain
  * @ioctl_resp_sge: DMA buffer descriptor for Mgmt cmd response
  * @ioctl_sges_allocated: Flag for IOCTL SGEs allocated or not
+ * @pcie_err_recovery: PCIe error recovery in progress
+ * @block_on_pcie_err: Block IO during PCI error recovery

s/PCIe error recovery/PCI error recovery/ so they match.

Quoted 21 lines
drivers/scsi/mpi3mr/mpi3mr.h +2 -0
 struct mpi3mr_ioc {
 	struct list_head list;
@@ -1246,6 +1249,8 @@ struct mpi3mr_ioc {
 	struct dma_memory_desc ioctl_chain_sge;
 	struct dma_memory_desc ioctl_resp_sge;
 	bool ioctl_sges_allocated;
+	bool pcie_err_recovery;
+	bool block_on_pcie_err;
 };
 
 /**
drivers/scsi/mpi3mr/mpi3mr_os.c +2 -0
index 76ba31a9517d..dea47ef53abb 100644
@@ -5472,6 +5472,195 @@ mpi3mr_resume(struct device *dev)
 	return 0;
 }
 
+/**
+ * mpi3mr_pcierr_detected - PCI error detected callback

drivers/scsi/mpi3mr/mpi3mr_os.c mpi3mr_resume(struct device *dev) · line 5486

Quoted 10 lines
drivers/scsi/mpi3mr/mpi3mr_os.c +10 -0
+ * @pdev: PCI device instance
+ * @state: channel state
+ *
+ * This function is called by the PCI error recovery driver and
+ * based on the state passed the driver decides what actions to
+ * be recommended back to PCI driver.
+ *
+ * For all of the states if there is no valid mrioc or scsi host
+ * references in the pci device then this function will retyrn
+ * the resul as disconnect.

s/pci device/PCI device/ to match other uses
s/retyrn/return/
s/resul/result/

drivers/scsi/mpi3mr/mpi3mr_os.c mpi3mr_resume(struct device *dev) · line 5491

Quoted 4 lines
drivers/scsi/mpi3mr/mpi3mr_os.c +4 -0
+ * For normal state, this function will return the result as can
+ * recover.
+ *
+ * For frozen state, this function will block for any pennding

s/pennding/pending/

Quoted 13 lines
drivers/scsi/mpi3mr/mpi3mr_os.c +13 -0
+ * controller initialization or re-initialization to complete,
+ * stop any new interactions with the controller and return
+ * status as reset required.
+ *
+ * For permanent failure state, this function will mark the
+ * controller as unrecoverable and return status as disconnect.
+ *
+ * Returns: PCI_ERS_RESULT_NEED_RESET or CAN_RECOVER or
+ * DISCONNECT based on the controller state.
+ */
+static pci_ers_result_t
+mpi3mr_pcierr_detected(struct pci_dev *pdev, pci_channel_state_t state)
+{

drivers/scsi/mpi3mr/mpi3mr_os.c mpi3mr_resume(struct device *dev) · line 5514

Quoted 10 lines
drivers/scsi/mpi3mr/mpi3mr_os.c +10 -0
+	struct Scsi_Host *shost;
+	struct mpi3mr_ioc *mrioc;
+	pci_ers_result_t ret_val = PCI_ERS_RESULT_DISCONNECT;
+
+	dev_info(&pdev->dev, "%s: callback invoked state(%d)\n", __func__,
+	    state);
+
+	if (mpi3mr_get_shost_and_mrioc(pdev, &shost, &mrioc)) {
+		dev_err(&pdev->dev, "device not available\n");
+		return ret_val;

Since you already know the return value here, use it explicitly:

return PCI_ERS_RESULT_DISCONNECT;

drivers/scsi/mpi3mr/mpi3mr_os.c mpi3mr_resume(struct device *dev) · line 5519

Quoted 5 lines
drivers/scsi/mpi3mr/mpi3mr_os.c +5 -0
+	}
+
+	switch (state) {
+	case pci_channel_io_normal:
+		ret_val = PCI_ERS_RESULT_CAN_RECOVER;

Personally, I would just return directly from each case instead of assigning "ret_val".

drivers/scsi/mpi3mr/mpi3mr_os.c mpi3mr_resume(struct device *dev) · line 5525

Quoted 6 lines
drivers/scsi/mpi3mr/mpi3mr_os.c +6 -0
+		break;
+	case pci_channel_io_frozen:
+		mrioc->pcie_err_recovery = true;
+		mrioc->block_on_pcie_err = true;
+		while (mrioc->reset_in_progress || mrioc->is_driver_loading)
+			ssleep(1);

This looks like a potential hang. What guarantees that one of these is eventually set?

drivers/scsi/mpi3mr/mpi3mr_os.c mpi3mr_resume(struct device *dev) · line 5528

Quoted 3 lines
drivers/scsi/mpi3mr/mpi3mr_os.c +3 -0
+		scsi_block_requests(mrioc->shost);
+		mpi3mr_stop_watchdog(mrioc);
+		mpi3mr_cleanup_resources(mrioc);

Why are mpi3mr_cleanup_resources() and mpi3mr_setup_resources() needed?

drivers/scsi/mpi3mr/mpi3mr_os.c mpi3mr_resume(struct device *dev) · line 5529

Quoted 1 line
drivers/scsi/mpi3mr/mpi3mr_os.c +1 -0
+		mrioc->pdev = NULL;

Seems weird to clear mrioc->pdev and then restore it later. What does this accomplish?

Quoted 4 lines
drivers/scsi/mpi3mr/mpi3mr_os.c +4 -0
+		ret_val = PCI_ERS_RESULT_NEED_RESET;
+		break;
+	case pci_channel_io_perm_failure:
+		mrioc->pcie_err_recovery = true;

drivers/scsi/mpi3mr/mpi3mr_os.c mpi3mr_resume(struct device *dev) · line 5543

Quoted 10 lines
drivers/scsi/mpi3mr/mpi3mr_os.c +10 -0
+		mrioc->block_on_pcie_err = true;
+		mrioc->unrecoverable = 1;
+		mpi3mr_stop_watchdog(mrioc);
+		mpi3mr_flush_cmds_for_unrecovered_controller(mrioc);
+		ret_val = PCI_ERS_RESULT_DISCONNECT;
+		break;
+	default:
+		break;
+	}
+	return ret_val;

Then this could become "return PCI_ERS_RESULT_DISCONNECT;" and you don't need "ret_val" at all.

Quoted 4 lines
drivers/scsi/mpi3mr/mpi3mr_os.c +4 -0
+}
+
+/**
+ * mpi3mr_pcierr_slot_reset_done - Post slot reset callback

drivers/scsi/mpi3mr/mpi3mr_os.c mpi3mr_resume(struct device *dev) · line 5557

Quoted 10 lines
drivers/scsi/mpi3mr/mpi3mr_os.c +10 -0
+ * @pdev: PCI device instance
+ *
+ * This function is called by the PCI error recovery driver
+ * after a slot or link reset issued by it for the recovery, the
+ * driver is expected to bring back the controller and
+ * initialize it.
+ *
+ * This function restores pci state and reinitializes controller
+ * resoruces and the controller, this blocks for any pending
+ * reset to complete.

s/pci state/PCI state/ to match other uses
s/resoruces/resources/

drivers/scsi/mpi3mr/mpi3mr_os.c mpi3mr_resume(struct device *dev) · line 5567

Quoted 9 lines
drivers/scsi/mpi3mr/mpi3mr_os.c +9 -0
+ * Returns: PCI_ERS_RESULT_DISCONNECT on failure or
+ * PCI_ERS_RESULT_RECOVERED
+ */
+static pci_ers_result_t mpi3mr_pcierr_slot_reset_done(struct pci_dev *pdev)
+{
+	struct Scsi_Host *shost;
+	struct mpi3mr_ioc *mrioc;
+
+

Spurious blank line.

drivers/scsi/mpi3mr/mpi3mr_os.c mpi3mr_resume(struct device *dev) · line 5576

Quoted 9 lines
drivers/scsi/mpi3mr/mpi3mr_os.c +9 -0
+	dev_info(&pdev->dev, "%s: callback invoked\n", __func__);
+
+	if (mpi3mr_get_shost_and_mrioc(pdev, &shost, &mrioc)) {
+		dev_err(&pdev->dev, "device not available\n");
+		return PCI_ERS_RESULT_DISCONNECT;
+	}
+
+	while (mrioc->reset_in_progress)
+		ssleep(1);

Another potential hang?

Quoted 10 lines
drivers/scsi/mpi3mr/mpi3mr_os.c +10 -0
+	mrioc->pdev = pdev;
+	pci_restore_state(pdev);
+
+	if (mpi3mr_setup_resources(mrioc)) {
+		ioc_err(mrioc, "setup resources failed\n");
+		goto out_failed;
+	}
+	mrioc->unrecoverable = 0;
+	mrioc->pcie_err_recovery = false;
+

drivers/scsi/mpi3mr/mpi3mr_os.c mpi3mr_resume(struct device *dev) · line 5597

Quoted 10 lines
drivers/scsi/mpi3mr/mpi3mr_os.c +10 -0
+	if (mpi3mr_soft_reset_handler(mrioc, MPI3MR_RESET_FROM_FIRMWARE, 0))
+		goto out_failed;
+
+	return PCI_ERS_RESULT_RECOVERED;
+
+out_failed:
+	mrioc->unrecoverable = 1;
+	mrioc->block_on_pcie_err = false;
+	scsi_unblock_requests(shost);
+	mpi3mr_start_watchdog(mrioc);

I'm not a SCSI person, but it seems odd to unblock requests and start the watchdog here. Isn't the device just dead at this point?

drivers/scsi/mpi3mr/mpi3mr_os.c mpi3mr_resume(struct device *dev) · line 5607

Quoted 10 lines
drivers/scsi/mpi3mr/mpi3mr_os.c +10 -0
+	return PCI_ERS_RESULT_DISCONNECT;
+}
+
+/**
+ * mpi3mr_pcierr_resume - PCI error recovery resume
+ * callback
+ * @pdev: PCI device instance
+ *
+ * This function enables all I/O and IOCTLs post reset issued as
+ * part of the PCI advacned error reporting and handling

s/advacned/advanced/ (actually, I'd just omit "advanced", since this covers more than just AER in the powerpc case).

Quoted 20 lines
drivers/scsi/mpi3mr/mpi3mr_os.c +20 -0
+ * Return: Nothing.
+ */
+static void mpi3mr_pcierr_resume(struct pci_dev *pdev)
+{
+	struct Scsi_Host *shost;
+	struct mpi3mr_ioc *mrioc;
+
+	dev_info(&pdev->dev, "%s: callback invoked\n", __func__);
+
+	if (mpi3mr_get_shost_and_mrioc(pdev, &shost, &mrioc)) {
+		dev_err(&pdev->dev, "device not available\n");
+		return;
+	}
+
+	pci_aer_clear_nonfatal_status(pdev);
+
+	if (mrioc->block_on_pcie_err) {
+		mrioc->block_on_pcie_err = false;
+		scsi_unblock_requests(shost);
+		mpi3mr_start_watchdog(mrioc);

drivers/scsi/mpi3mr/mpi3mr_os.c mpi3mr_resume(struct device *dev) · line 5638

Quoted 10 lines
drivers/scsi/mpi3mr/mpi3mr_os.c +10 -0
+	}
+
+}
+
+/**
+ * mpi3mr_pcierr_mmio_enabled - PCI error recovery callback
+ * @pdev: PCI device instance
+ *
+ * This is called only if _pcierr_error_detected returns
+ * PCI_ERS_RESULT_CAN_RECOVER.

s/_pcierr_error_detected/mpi3mr_pcierr_detected()/ ? (Or possible different name as suggested below)

drivers/scsi/mpi3mr/mpi3mr_os.c mpi3mr_resume(struct device *dev) · line 5642

Quoted 3 lines
drivers/scsi/mpi3mr/mpi3mr_os.c +3 -0
+ * Return: PCI_ERS_RESULT_DISCONNECT when the controller is
+ * unrecoverable or when the shost/mnrioc reference cannot be
+ * found, else return PCI_ERS_RESULT_RECOVERED

s/mnrioc/mrioc/

drivers/scsi/mpi3mr/mpi3mr_os.c mpi3mr_resume(struct device *dev) · line 5651

Quoted 8 lines
drivers/scsi/mpi3mr/mpi3mr_os.c +8 -0
+static pci_ers_result_t mpi3mr_pcierr_mmio_enabled(struct pci_dev *pdev)
+{
+
+	struct Scsi_Host *shost;
+	struct mpi3mr_ioc *mrioc;
+/*
+ *
+ */

Spurious comment?

Quoted 16 lines
drivers/scsi/mpi3mr/mpi3mr_os.c +12 -0
+	dev_info(&pdev->dev, "%s: callback invoked\n", __func__);
+
+	if (mpi3mr_get_shost_and_mrioc(pdev, &shost, &mrioc)) {
+		dev_err(&pdev->dev, "device not available\n");
+		return PCI_ERS_RESULT_DISCONNECT;
+	}
+	if (mrioc->unrecoverable)
+		return PCI_ERS_RESULT_DISCONNECT;
+
+	return PCI_ERS_RESULT_RECOVERED;
+}
+
 static const struct pci_device_id mpi3mr_pci_id_table[] = {
 	{
 		PCI_DEVICE_SUB(MPI3_MFGPAGE_VENDORID_BROADCOM,

drivers/scsi/mpi3mr/mpi3mr_os.c static const struct pci_device_id mpi3mr_pci_id_table[] = { · line 5684

Quoted 7 lines
drivers/scsi/mpi3mr/mpi3mr_os.c +4 -0
@@ -5489,6 +5678,13 @@ static const struct pci_device_id mpi3mr_pci_id_table[] = {
 };
 MODULE_DEVICE_TABLE(pci, mpi3mr_pci_id_table);
 
+static struct pci_error_handlers mpi3mr_err_handler = {
+	.error_detected = mpi3mr_pcierr_detected,
+	.mmio_enabled = mpi3mr_pcierr_mmio_enabled,
+	.slot_reset = mpi3mr_pcierr_slot_reset_done,

It would be helpful if you used "mpi3mr_pcierr_slot_reset()" to match the function pointer name. This would make it much easier to compare .slot_reset() implementations across drivers.

I agree that "slot_reset_done()" is more descriptive and would probably have been a better name for the function pointer, but that ship has sailed.

I'd suggest "mpi3mr_pcierr_error_detected()" for the same reason. Or maybe remove "pcierr_" from all of them if it feels too redundant.

Quoted 17 lines
drivers/scsi/mpi3mr/mpi3mr_os.c +4 -0
+	.resume = mpi3mr_pcierr_resume,
+};
+
 static SIMPLE_DEV_PM_OPS(mpi3mr_pm_ops, mpi3mr_suspend, mpi3mr_resume);
 
 static struct pci_driver mpi3mr_pci_driver = {
@@ -5497,6 +5693,7 @@ static struct pci_driver mpi3mr_pci_driver = {
 	.probe = mpi3mr_probe,
 	.remove = mpi3mr_remove,
 	.shutdown = mpi3mr_shutdown,
+	.err_handler = &mpi3mr_err_handler,
 	.driver.pm = &mpi3mr_pm_ops,
 };
 
Signature · 1 line
2.31.1
#5 lore raw

PATCH v2 3/6 mpi3mr: Prevent PCI writes from driver during PCI error recovery

In reply to Ranjan Kumar

Prevent interacting with the hardware while
the error recovery in progress.

---
 drivers/scsi/mpi3mr/mpi3mr.h           | 26 ++++++++++
 drivers/scsi/mpi3mr/mpi3mr_app.c       | 64 +++++++++++++----------
 drivers/scsi/mpi3mr/mpi3mr_fw.c        | 28 +++++++---
 drivers/scsi/mpi3mr/mpi3mr_os.c        | 71 +++++++++++++++-----------
 drivers/scsi/mpi3mr/mpi3mr_transport.c | 39 ++++++++++++--
 5 files changed, 158 insertions(+), 70 deletions(-)
drivers/scsi/mpi3mr/mpi3mr.h +26 -0
index 25e6e3a09468..1ac7f88dc1cd 100644
@@ -481,6 +481,7 @@ struct mpi3mr_throttle_group_info {
 
 /* HBA port flags */
 #define MPI3MR_HBA_PORT_FLAG_DIRTY	0x01
+#define MPI3MR_HBA_PORT_FLAG_NEW       0x02
 
 /* IOCTL data transfer sge*/
 #define MPI3MR_NUM_IOCTL_SGE		256
@@ -900,6 +901,29 @@ struct scmd_priv {
 	u8 mpi3mr_scsiio_req[MPI3MR_ADMIN_REQ_FRAME_SZ];
 };
 
+/**
+ * struct mpi3mr_pdevinfo - PCI device information
+ *
+ * @dev_id: PCI device ID of the adapter
+ * @dev_hw_rev: PCI revision of the adapter
+ * @subsys_dev_id: PCI subsystem device ID of the adapter
+ * @subsys_ven_id: PCI subsystem vendor ID of the adapter
+ * @dev: PCI device
+ * @func: PCI function
+ * @bus: PCI bus
+ * @seg_id: PCI segment ID
+ */
+struct mpi3mr_pdevinfo {
+	u16 id;
+	u16 ssid;
+	u16 ssvid;
+	u16 segment;
+	u8 dev:5;
+	u8 func:3;
+	u8 bus;
+	u8 revision;
+};
+
 /**
  * struct mpi3mr_ioc - Adapter anchor structure stored in shost
  * private data
@@ -1058,6 +1082,7 @@ struct scmd_priv {
  * @ioctl_sges_allocated: Flag for IOCTL SGEs allocated or not
  * @pcie_err_recovery: PCIe error recovery in progress
  * @block_on_pcie_err: Block IO during PCI error recovery
+ * @pdevinfo: PCI device information
  */
 struct mpi3mr_ioc {
 	struct list_head list;
@@ -1251,6 +1276,7 @@ struct mpi3mr_ioc {
 	bool ioctl_sges_allocated;
 	bool pcie_err_recovery;
 	bool block_on_pcie_err;
+	struct mpi3mr_pdevinfo pdevinfo;
 };
 
 /**
drivers/scsi/mpi3mr/mpi3mr_app.c +37 -27

99 lines not on this page · raw message

drivers/scsi/mpi3mr/mpi3mr_fw.c +20 -8
index d8c57a0a518f..c50fc27231cd 100644
@@ -595,7 +595,7 @@ int mpi3mr_blk_mq_poll(struct Scsi_Host *shost, unsigned int queue_num)
 	mrioc = (struct mpi3mr_ioc *)shost->hostdata;
 
 	if ((mrioc->reset_in_progress || mrioc->prepare_for_reset ||
-	    mrioc->unrecoverable))
+	    mrioc->unrecoverable || mrioc->pcie_err_recovery))
 		return 0;
 
 	num_entries = mpi3mr_process_op_reply_q(mrioc,
@@ -1037,13 +1037,13 @@ enum mpi3mr_iocstate mpi3mr_get_iocstate(struct mpi3mr_ioc *mrioc)
 	u32 ioc_status, ioc_config;
 	u8 ready, enabled;
 
-	ioc_status = readl(&mrioc->sysif_regs->ioc_status);
-	ioc_config = readl(&mrioc->sysif_regs->ioc_configuration);
-
 	if (mrioc->unrecoverable)
 		return MRIOC_STATE_UNRECOVERABLE;
+	ioc_status = readl(&mrioc->sysif_regs->ioc_status);
+
 	if (ioc_status & MPI3_SYSIF_IOC_STATUS_FAULT)
 		return MRIOC_STATE_FAULT;
+	ioc_config = readl(&mrioc->sysif_regs->ioc_configuration);
 
 	ready = (ioc_status & MPI3_SYSIF_IOC_STATUS_READY);
 	enabled = (ioc_config & MPI3_SYSIF_IOC_CONFIG_ENABLE_IOC);
@@ -1667,6 +1667,12 @@ int mpi3mr_admin_request_post(struct mpi3mr_ioc *mrioc, void *admin_req,
 		retval = -EAGAIN;
 		goto out;
 	}
+	if (mrioc->pcie_err_recovery) {
+		ioc_err(mrioc, "admin request queue submission failed due to pcie error recovery in progress\n");
+		retval = -EAGAIN;
+		goto out;
+	}
+
 	areq_entry = (u8 *)mrioc->admin_req_base +
 	    (areq_pi * MPI3MR_ADMIN_REQ_FRAME_SZ);
 	memset(areq_entry, 0, MPI3MR_ADMIN_REQ_FRAME_SZ);
@@ -2337,6 +2343,11 @@ int mpi3mr_op_request_post(struct mpi3mr_ioc *mrioc,
 		retval = -EAGAIN;
 		goto out;
 	}
+	if (mrioc->pcie_err_recovery) {
+		ioc_err(mrioc, "operational request queue submission failed due to pcie error recovery in progress\n");
+		retval = -EAGAIN;
+		goto out;
+	}
 
 	segment_base_addr = segments[pi / op_req_q->segment_qd].segment;
 	req_entry = (u8 *)segment_base_addr +
@@ -2585,7 +2596,7 @@ static void mpi3mr_watchdog_work(struct work_struct *work)
 	u32 fault, host_diagnostic, ioc_status;
 	u32 reset_reason = MPI3MR_RESET_FROM_FAULT_WATCH;
 
-	if (mrioc->reset_in_progress)
+	if (mrioc->reset_in_progress || mrioc->pcie_err_recovery)
 		return;
 
 	if (!mrioc->unrecoverable && !pci_device_is_present(mrioc->pdev)) {
@@ -4111,7 +4122,7 @@ int mpi3mr_reinit_ioc(struct mpi3mr_ioc *mrioc, u8 is_resume)
 		goto out_failed_noretry;
 	}
 
-	if (is_resume) {
+	if (is_resume || mrioc->block_on_pcie_err) {
 		dprint_reset(mrioc, "setting up single ISR\n");
 		retval = mpi3mr_setup_isr(mrioc, 1);
 		if (retval) {
@@ -4151,7 +4162,7 @@ int mpi3mr_reinit_ioc(struct mpi3mr_ioc *mrioc, u8 is_resume)
 		goto out_failed;
 	}
 
-	if (is_resume) {
+	if (is_resume || mrioc->block_on_pcie_err) {
 		dprint_reset(mrioc, "setting up multiple ISR\n");
 		retval = mpi3mr_setup_isr(mrioc, 0);
 		if (retval) {
@@ -4625,7 +4636,8 @@ void mpi3mr_cleanup_ioc(struct mpi3mr_ioc *mrioc)
 
 	ioc_state = mpi3mr_get_iocstate(mrioc);
 
-	if ((!mrioc->unrecoverable) && (!mrioc->reset_in_progress) &&
+	if (!mrioc->unrecoverable && !mrioc->reset_in_progress &&
+	    !mrioc->pcie_err_recovery &&
 	    (ioc_state == MRIOC_STATE_READY)) {
 		if (mpi3mr_issue_and_process_mur(mrioc,
 		    MPI3MR_RESET_FROM_CTLR_CLEANUP))
drivers/scsi/mpi3mr/mpi3mr_os.c +40 -31

132 lines not on this page · raw message

drivers/scsi/mpi3mr/mpi3mr_transport.c +35 -4
index c0c8ab586957..8c8368104a27 100644
@@ -149,6 +149,11 @@ static int mpi3mr_report_manufacture(struct mpi3mr_ioc *mrioc,
 		return -EFAULT;
 	}
 
+	if (mrioc->pcie_err_recovery) {
+		ioc_err(mrioc, "%s: pcie error recovery in progress!\n", __func__);
+		return -EFAULT;
+	}
+
 	data_out_sz = sizeof(struct rep_manu_request);
 	data_in_sz = sizeof(struct rep_manu_reply);
 	data_out = dma_alloc_coherent(&mrioc->pdev->dev,
@@ -792,6 +797,12 @@ static int mpi3mr_set_identify(struct mpi3mr_ioc *mrioc, u16 handle,
 		return -EFAULT;
 	}
 
+	if (mrioc->pcie_err_recovery) {
+		ioc_err(mrioc, "%s: pcie error recovery in progress!\n",
+		    __func__);
+		return -EFAULT;
+	}
+
 	if ((mpi3mr_cfg_get_dev_pg0(mrioc, &ioc_status, &device_pg0,
 	    sizeof(device_pg0), MPI3_DEVICE_PGAD_FORM_HANDLE, handle))) {
 		ioc_err(mrioc, "%s: device page0 read failed\n", __func__);
@@ -1009,6 +1020,9 @@ mpi3mr_alloc_hba_port(struct mpi3mr_ioc *mrioc, u16 port_id)
 	hba_port->port_id = port_id;
 	ioc_info(mrioc, "hba_port entry: %p, port: %d is added to hba_port list\n",
 	    hba_port, hba_port->port_id);
+	if (mrioc->reset_in_progress ||
+		mrioc->pcie_err_recovery)
+		hba_port->flags = MPI3MR_HBA_PORT_FLAG_NEW;
 	list_add_tail(&hba_port->list, &mrioc->hba_port_table_list);
 	return hba_port;
 }
@@ -1057,7 +1071,7 @@ void mpi3mr_update_links(struct mpi3mr_ioc *mrioc,
 	struct mpi3mr_sas_node *mr_sas_node;
 	struct mpi3mr_sas_phy *mr_sas_phy;
 
-	if (mrioc->reset_in_progress)
+	if (mrioc->reset_in_progress || mrioc->pcie_err_recovery)
 		return;
 
 	spin_lock_irqsave(&mrioc->sas_node_lock, flags);
@@ -1965,7 +1979,7 @@ int mpi3mr_expander_add(struct mpi3mr_ioc *mrioc, u16 handle)
 	if (!handle)
 		return -1;
 
-	if (mrioc->reset_in_progress)
+	if (mrioc->reset_in_progress || mrioc->pcie_err_recovery)
 		return -1;
 
 	if ((mpi3mr_cfg_get_sas_exp_pg0(mrioc, &ioc_status, &expander_pg0,
@@ -2171,7 +2185,7 @@ void mpi3mr_expander_node_remove(struct mpi3mr_ioc *mrioc,
 	/* remove sibling ports attached to this expander */
 	list_for_each_entry_safe(mr_sas_port, next,
 	   &sas_expander->sas_port_list, port_list) {
-		if (mrioc->reset_in_progress)
+		if (mrioc->reset_in_progress || mrioc->pcie_err_recovery)
 			return;
 		if (mr_sas_port->remote_identify.device_type ==
 		    SAS_END_DEVICE)
@@ -2221,7 +2235,7 @@ void mpi3mr_expander_remove(struct mpi3mr_ioc *mrioc, u64 sas_address,
 	struct mpi3mr_sas_node *sas_expander;
 	unsigned long flags;
 
-	if (mrioc->reset_in_progress)
+	if (mrioc->reset_in_progress || mrioc->pcie_err_recovery)
 		return;
 
 	if (!hba_port)
@@ -2532,6 +2546,11 @@ static int mpi3mr_get_expander_phy_error_log(struct mpi3mr_ioc *mrioc,
 		return -EFAULT;
 	}
 
+	if (mrioc->pcie_err_recovery) {
+		ioc_err(mrioc, "%s: pcie error recovery in progress!\n", __func__);
+		return -EFAULT;
+	}
+
 	data_out_sz = sizeof(struct phy_error_log_request);
 	data_in_sz = sizeof(struct phy_error_log_reply);
 	sz = data_out_sz + data_in_sz;
@@ -2791,6 +2810,12 @@ mpi3mr_expander_phy_control(struct mpi3mr_ioc *mrioc,
 		return -EFAULT;
 	}
 
+	if (mrioc->pcie_err_recovery) {
+		ioc_err(mrioc, "%s: pcie error recovery in progress!\n",
+		    __func__);
+		return -EFAULT;
+	}
+
 	data_out_sz = sizeof(struct phy_control_request);
 	data_in_sz = sizeof(struct phy_control_reply);
 	sz = data_out_sz + data_in_sz;
@@ -3214,6 +3239,12 @@ mpi3mr_transport_smp_handler(struct bsg_job *job, struct Scsi_Host *shost,
 		goto out;
 	}
 
+	if (mrioc->pcie_err_recovery) {
+		ioc_err(mrioc, "%s: pcie error recovery in progress!\n", __func__);
+		rc = -EFAULT;
+		goto out;
+	}
+
 	rc = mpi3mr_map_smp_buffer(&mrioc->pdev->dev, &job->request_payload,
 	    &dma_addr_out, &dma_len_out, &addr_out);
 	if (rc)
Signature · 1 line
2.31.1
#6 lore raw

Re: [PATCH v2 3/6] mpi3mr: Prevent PCI writes from driver during PCI error recovery

Quoted 30 lines · Ranjan Kumar

Prevent interacting with the hardware while
the error recovery in progress.

---
 drivers/scsi/mpi3mr/mpi3mr.h           | 26 ++++++++++
 drivers/scsi/mpi3mr/mpi3mr_app.c       | 64 +++++++++++++----------
 drivers/scsi/mpi3mr/mpi3mr_fw.c        | 28 +++++++---
 drivers/scsi/mpi3mr/mpi3mr_os.c        | 71 +++++++++++++++-----------
 drivers/scsi/mpi3mr/mpi3mr_transport.c | 39 ++++++++++++--
 5 files changed, 158 insertions(+), 70 deletions(-)
drivers/scsi/mpi3mr/mpi3mr.h +2 -0
index 25e6e3a09468..1ac7f88dc1cd 100644
@@ -481,6 +481,7 @@ struct mpi3mr_throttle_group_info {
 
 /* HBA port flags */
 #define MPI3MR_HBA_PORT_FLAG_DIRTY	0x01
+#define MPI3MR_HBA_PORT_FLAG_NEW       0x02
 
 /* IOCTL data transfer sge*/
 #define MPI3MR_NUM_IOCTL_SGE		256
@@ -900,6 +901,29 @@ struct scmd_priv {
 	u8 mpi3mr_scsiio_req[MPI3MR_ADMIN_REQ_FRAME_SZ];
 };
 
+/**

drivers/scsi/mpi3mr/mpi3mr.h struct scmd_priv { · line 914

Quoted 10 lines
drivers/scsi/mpi3mr/mpi3mr.h +10 -0
+ * struct mpi3mr_pdevinfo - PCI device information
+ *
+ * @dev_id: PCI device ID of the adapter
+ * @dev_hw_rev: PCI revision of the adapter
+ * @subsys_dev_id: PCI subsystem device ID of the adapter
+ * @subsys_ven_id: PCI subsystem vendor ID of the adapter
+ * @dev: PCI device
+ * @func: PCI function
+ * @bus: PCI bus
+ * @seg_id: PCI segment ID

These names don't match the actual structure members, so I bet scripts/kernel-doc is going to complain about them.

Quoted 24 lines
drivers/scsi/mpi3mr/mpi3mr.h +13 -0
+ */
+struct mpi3mr_pdevinfo {
+	u16 id;
+	u16 ssid;
+	u16 ssvid;
+	u16 segment;
+	u8 dev:5;
+	u8 func:3;
+	u8 bus;
+	u8 revision;
+};
+
 /**
  * struct mpi3mr_ioc - Adapter anchor structure stored in shost
  * private data
@@ -1058,6 +1082,7 @@ struct scmd_priv {
  * @ioctl_sges_allocated: Flag for IOCTL SGEs allocated or not
  * @pcie_err_recovery: PCIe error recovery in progress
  * @block_on_pcie_err: Block IO during PCI error recovery
+ * @pdevinfo: PCI device information
  */
 struct mpi3mr_ioc {
 	struct list_head list;

drivers/scsi/mpi3mr/mpi3mr.h struct mpi3mr_ioc { · line 1279

Quoted 4 lines
drivers/scsi/mpi3mr/mpi3mr.h +1 -0
@@ -1251,6 +1276,7 @@ struct mpi3mr_ioc {
 	bool ioctl_sges_allocated;
 	bool pcie_err_recovery;
 	bool block_on_pcie_err;
+	struct mpi3mr_pdevinfo pdevinfo;

I don't understand the reason to add "pdevinfo". mpi3mr_probe() fills pdevinfo from mrioc->pdev. mpi3mr_bsg_populate_adpinfo() previously filled adpinfo from mrioc->pdev, and now fills it from pdevinfo instead.

It seems like the same data. Is this related to the fact that mpi3mr_pcierr_detected() set mrioc->pdev to NULL? If so, this patch would have to come first, otherwise if we bisect at the point between patch [2/6] and this [3/6], there's a potential NULL pointer dereference. Bisection should work correctly at every commit.

I don't think setting mrioc->pdev to NULL is a good idea, so hopefully pdevinfo can just go away.

Quoted 59 lines
drivers/scsi/mpi3mr/mpi3mr.h +0 -0
 };
 
 /**
drivers/scsi/mpi3mr/mpi3mr_app.c +14 -17
index 4b93b7440da6..a11d6f026f0e 100644
@@ -31,7 +31,7 @@ static int mpi3mr_bsg_pel_abort(struct mpi3mr_ioc *mrioc)
 		dprint_bsg_err(mrioc, "%s: reset in progress\n", __func__);
 		return -1;
 	}
-	if (mrioc->stop_bsgs) {
+	if (mrioc->stop_bsgs || mrioc->block_on_pcie_err) {
 		dprint_bsg_err(mrioc, "%s: bsgs are blocked\n", __func__);
 		return -1;
 	}
@@ -424,6 +424,9 @@ static long mpi3mr_bsg_adp_reset(struct mpi3mr_ioc *mrioc,
 		goto out;
 	}
 
+	if (mrioc->unrecoverable || mrioc->block_on_pcie_err)
+		return -EINVAL;
+
 	sg_copy_to_buffer(job->request_payload.sg_list,
 			  job->request_payload.sg_cnt,
 			  &adpreset, sizeof(adpreset));
@@ -470,25 +473,29 @@ static long mpi3mr_bsg_populate_adpinfo(struct mpi3mr_ioc *mrioc,
 
 	memset(&adpinfo, 0, sizeof(adpinfo));
 	adpinfo.adp_type = MPI3MR_BSG_ADPTYPE_AVGFAMILY;
-	adpinfo.pci_dev_id = mrioc->pdev->device;
-	adpinfo.pci_dev_hw_rev = mrioc->pdev->revision;
-	adpinfo.pci_subsys_dev_id = mrioc->pdev->subsystem_device;
-	adpinfo.pci_subsys_ven_id = mrioc->pdev->subsystem_vendor;
-	adpinfo.pci_bus = mrioc->pdev->bus->number;
-	adpinfo.pci_dev = PCI_SLOT(mrioc->pdev->devfn);
-	adpinfo.pci_func = PCI_FUNC(mrioc->pdev->devfn);
-	adpinfo.pci_seg_id = pci_domain_nr(mrioc->pdev->bus);
 	adpinfo.app_intfc_ver = MPI3MR_IOCTL_VERSION;
 
-	ioc_state = mpi3mr_get_iocstate(mrioc);
-	if (ioc_state == MRIOC_STATE_UNRECOVERABLE)
-		adpinfo.adp_state = MPI3MR_BSG_ADPSTATE_UNRECOVERABLE;
-	else if ((mrioc->reset_in_progress) || (mrioc->stop_bsgs))
+	if (mrioc->reset_in_progress || mrioc->stop_bsgs ||
+	    mrioc->block_on_pcie_err)
 		adpinfo.adp_state = MPI3MR_BSG_ADPSTATE_IN_RESET;
-	else if (ioc_state == MRIOC_STATE_FAULT)
-		adpinfo.adp_state = MPI3MR_BSG_ADPSTATE_FAULT;
-	else
-		adpinfo.adp_state = MPI3MR_BSG_ADPSTATE_OPERATIONAL;
+	else {
+		ioc_state = mpi3mr_get_iocstate(mrioc);
+		if (ioc_state == MRIOC_STATE_UNRECOVERABLE)
+			adpinfo.adp_state = MPI3MR_BSG_ADPSTATE_UNRECOVERABLE;
+		else if (ioc_state == MRIOC_STATE_FAULT)
+			adpinfo.adp_state = MPI3MR_BSG_ADPSTATE_FAULT;
+		else
+			adpinfo.adp_state = MPI3MR_BSG_ADPSTATE_OPERATIONAL;

drivers/scsi/mpi3mr/mpi3mr_app.c static long mpi3mr_bsg_populate_adpinfo(struct mpi3mr_ioc *mrioc, · line 498

Quoted 10 lines
drivers/scsi/mpi3mr/mpi3mr_app.c +10 -0
+	}
+
+	adpinfo.pci_dev_id = mrioc->pdevinfo.id;
+	adpinfo.pci_dev_hw_rev = mrioc->pdevinfo.revision;
+	adpinfo.pci_subsys_dev_id = mrioc->pdevinfo.ssid;
+	adpinfo.pci_subsys_ven_id = mrioc->pdevinfo.ssvid;
+	adpinfo.pci_bus = mrioc->pdevinfo.bus;
+	adpinfo.pci_dev = mrioc->pdevinfo.dev;
+	adpinfo.pci_func = mrioc->pdevinfo.func;
+	adpinfo.pci_seg_id = mrioc->pdevinfo.segment;

I don't know if it's possible, but if the adpinfo changes could be moved to their own patch (or even removed altogether), this patch would be more about what the commit log advertises: preventing interaction with the hardware during error recovery.

Quoted 58 lines
drivers/scsi/mpi3mr/mpi3mr_app.c +13 -10
 	memcpy((u8 *)&adpinfo.driver_info, (u8 *)&mrioc->driver_info,
 	    sizeof(adpinfo.driver_info));
@@ -1495,7 +1502,7 @@ static long mpi3mr_bsg_process_mpt_cmds(struct bsg_job *job)
 		mutex_unlock(&mrioc->bsg_cmds.mutex);
 		goto out;
 	}
-	if (mrioc->stop_bsgs) {
+	if (mrioc->stop_bsgs || mrioc->block_on_pcie_err) {
 		dprint_bsg_err(mrioc, "%s: bsgs are blocked\n", __func__);
 		rval = -EAGAIN;
 		mutex_unlock(&mrioc->bsg_cmds.mutex);
@@ -2020,17 +2027,20 @@ adp_state_show(struct device *dev, struct device_attribute *attr,
 	enum mpi3mr_iocstate ioc_state;
 	uint8_t adp_state;
 
-	ioc_state = mpi3mr_get_iocstate(mrioc);
-	if (ioc_state == MRIOC_STATE_UNRECOVERABLE)
-		adp_state = MPI3MR_BSG_ADPSTATE_UNRECOVERABLE;
-	else if ((mrioc->reset_in_progress) || (mrioc->stop_bsgs))
+	if (mrioc->reset_in_progress || mrioc->stop_bsgs ||
+		 mrioc->block_on_pcie_err)
 		adp_state = MPI3MR_BSG_ADPSTATE_IN_RESET;
-	else if (ioc_state == MRIOC_STATE_FAULT)
-		adp_state = MPI3MR_BSG_ADPSTATE_FAULT;
-	else
-		adp_state = MPI3MR_BSG_ADPSTATE_OPERATIONAL;
+	else {
+		ioc_state = mpi3mr_get_iocstate(mrioc);
+		if (ioc_state == MRIOC_STATE_UNRECOVERABLE)
+			adp_state = MPI3MR_BSG_ADPSTATE_UNRECOVERABLE;
+		else if (ioc_state == MRIOC_STATE_FAULT)
+			adp_state = MPI3MR_BSG_ADPSTATE_FAULT;
+		else
+			adp_state = MPI3MR_BSG_ADPSTATE_OPERATIONAL;
+	}
 
-	return sysfs_emit(buf, "%u\n", adp_state);
+	return snprintf(buf, PAGE_SIZE, "%u\n", adp_state);
 }
 
 static DEVICE_ATTR_RO(adp_state);
drivers/scsi/mpi3mr/mpi3mr_fw.c +1 -1
index d8c57a0a518f..c50fc27231cd 100644
@@ -595,7 +595,7 @@ int mpi3mr_blk_mq_poll(struct Scsi_Host *shost, unsigned int queue_num)
 	mrioc = (struct mpi3mr_ioc *)shost->hostdata;
 
 	if ((mrioc->reset_in_progress || mrioc->prepare_for_reset ||
-	    mrioc->unrecoverable))
+	    mrioc->unrecoverable || mrioc->pcie_err_recovery))
 		return 0;
 
 	num_entries = mpi3mr_process_op_reply_q(mrioc,
@@ -1037,13 +1037,13 @@ enum mpi3mr_iocstate mpi3mr_get_iocstate(struct mpi3mr_ioc *mrioc)
 	u32 ioc_status, ioc_config;
 	u8 ready, enabled;
 

drivers/scsi/mpi3mr/mpi3mr_fw.c enum mpi3mr_iocstate mpi3mr_get_iocstate(struct mpi3mr_ioc *mrioc) · line 1046

Quoted 10 lines
drivers/scsi/mpi3mr/mpi3mr_fw.c +3 -3
-	ioc_status = readl(&mrioc->sysif_regs->ioc_status);
-	ioc_config = readl(&mrioc->sysif_regs->ioc_configuration);
-
 	if (mrioc->unrecoverable)
 		return MRIOC_STATE_UNRECOVERABLE;
+	ioc_status = readl(&mrioc->sysif_regs->ioc_status);
+
 	if (ioc_status & MPI3_SYSIF_IOC_STATUS_FAULT)
 		return MRIOC_STATE_FAULT;
+	ioc_config = readl(&mrioc->sysif_regs->ioc_configuration);

These reads look suspect because you cannot rely on mrioc->unrecoverable to catch all cases where the device is inaccessible. For example, the device may be unplugged here:

  # device is operating normally

if (mrioc->unrecoverable)

    return MRIOC_STATE_UNRECOVERABLE;
  # device is hot-unplugged here
  readl(&mrioc->sysif_regs->ioc_status);
  # this readl() returns ~0 because the MMIO access failed on PCI
Quoted 78 lines
drivers/scsi/mpi3mr/mpi3mr_fw.c +16 -4
 	ready = (ioc_status & MPI3_SYSIF_IOC_STATUS_READY);
 	enabled = (ioc_config & MPI3_SYSIF_IOC_CONFIG_ENABLE_IOC);
@@ -1667,6 +1667,12 @@ int mpi3mr_admin_request_post(struct mpi3mr_ioc *mrioc, void *admin_req,
 		retval = -EAGAIN;
 		goto out;
 	}
+	if (mrioc->pcie_err_recovery) {
+		ioc_err(mrioc, "admin request queue submission failed due to pcie error recovery in progress\n");
+		retval = -EAGAIN;
+		goto out;
+	}
+
 	areq_entry = (u8 *)mrioc->admin_req_base +
 	    (areq_pi * MPI3MR_ADMIN_REQ_FRAME_SZ);
 	memset(areq_entry, 0, MPI3MR_ADMIN_REQ_FRAME_SZ);
@@ -2337,6 +2343,11 @@ int mpi3mr_op_request_post(struct mpi3mr_ioc *mrioc,
 		retval = -EAGAIN;
 		goto out;
 	}
+	if (mrioc->pcie_err_recovery) {
+		ioc_err(mrioc, "operational request queue submission failed due to pcie error recovery in progress\n");
+		retval = -EAGAIN;
+		goto out;
+	}
 
 	segment_base_addr = segments[pi / op_req_q->segment_qd].segment;
 	req_entry = (u8 *)segment_base_addr +
@@ -2585,7 +2596,7 @@ static void mpi3mr_watchdog_work(struct work_struct *work)
 	u32 fault, host_diagnostic, ioc_status;
 	u32 reset_reason = MPI3MR_RESET_FROM_FAULT_WATCH;
 
-	if (mrioc->reset_in_progress)
+	if (mrioc->reset_in_progress || mrioc->pcie_err_recovery)
 		return;
 
 	if (!mrioc->unrecoverable && !pci_device_is_present(mrioc->pdev)) {
@@ -4111,7 +4122,7 @@ int mpi3mr_reinit_ioc(struct mpi3mr_ioc *mrioc, u8 is_resume)
 		goto out_failed_noretry;
 	}
 
-	if (is_resume) {
+	if (is_resume || mrioc->block_on_pcie_err) {
 		dprint_reset(mrioc, "setting up single ISR\n");
 		retval = mpi3mr_setup_isr(mrioc, 1);
 		if (retval) {
@@ -4151,7 +4162,7 @@ int mpi3mr_reinit_ioc(struct mpi3mr_ioc *mrioc, u8 is_resume)
 		goto out_failed;
 	}
 
-	if (is_resume) {
+	if (is_resume || mrioc->block_on_pcie_err) {
 		dprint_reset(mrioc, "setting up multiple ISR\n");
 		retval = mpi3mr_setup_isr(mrioc, 0);
 		if (retval) {
@@ -4625,7 +4636,8 @@ void mpi3mr_cleanup_ioc(struct mpi3mr_ioc *mrioc)
 
 	ioc_state = mpi3mr_get_iocstate(mrioc);
 
-	if ((!mrioc->unrecoverable) && (!mrioc->reset_in_progress) &&
+	if (!mrioc->unrecoverable && !mrioc->reset_in_progress &&
+	    !mrioc->pcie_err_recovery &&
 	    (ioc_state == MRIOC_STATE_READY)) {
 		if (mpi3mr_issue_and_process_mur(mrioc,
 		    MPI3MR_RESET_FROM_CTLR_CLEANUP))
drivers/scsi/mpi3mr/mpi3mr_os.c +1 -1
index dea47ef53abb..d61fd105dfad 100644
@@ -919,7 +919,7 @@ static int mpi3mr_report_tgtdev_to_host(struct mpi3mr_ioc *mrioc,
 	int retval = 0;
 	struct mpi3mr_tgt_dev *tgtdev;
 
-	if (mrioc->reset_in_progress)
+	if (mrioc->reset_in_progress || mrioc->pcie_err_recovery)
 		return -1;
 
 	tgtdev = mpi3mr_get_tgtdev_by_perst_id(mrioc, perst_id);

drivers/scsi/mpi3mr/mpi3mr_os.c static void mpi3mr_fwevt_bh(struct mpi3mr_ioc *mrioc, · line 2006

Quoted 8 lines
drivers/scsi/mpi3mr/mpi3mr_os.c +3 -1
@@ -2000,8 +2000,13 @@ static void mpi3mr_fwevt_bh(struct mpi3mr_ioc *mrioc,
 	}
 	case MPI3_EVENT_WAIT_FOR_DEVICES_TO_REFRESH:
 	{
-		while (mrioc->device_refresh_on)
+		while ((mrioc->device_refresh_on || mrioc->block_on_pcie_err) &&
+		    !mrioc->unrecoverable && !mrioc->pcie_err_recovery) {
 			msleep(500);
+		}

Looks like a potential hang here. I guess it was always a potential hang; this just makes the conditional more complicated. Usually you want a timeout or other means of ensuring the loop exits.

Quoted 21 lines
drivers/scsi/mpi3mr/mpi3mr_os.c +10 -0
+
+		if (mrioc->unrecoverable || mrioc->pcie_err_recovery)
+			break;
 
 		dprint_event_bh(mrioc,
 		    "scan for non responding and newly added devices after soft reset started\n");
@@ -3680,6 +3685,13 @@ int mpi3mr_issue_tm(struct mpi3mr_ioc *mrioc, u8 tm_type,
 		mutex_unlock(&drv_cmd->mutex);
 		goto out;
 	}
+	if (mrioc->block_on_pcie_err) {
+		retval = -1;
+		dprint_tm(mrioc, "sending task management failed due to\n"
+				"pcie error recovery in progress\n");
+		mutex_unlock(&drv_cmd->mutex);
+		goto out;
+	}
 
 	drv_cmd->state = MPI3MR_CMD_PENDING;
 	drv_cmd->is_waiting = 1;

drivers/scsi/mpi3mr/mpi3mr_os.c static int mpi3mr_eh_bus_reset(struct scsi_cmnd *scmd) · line 4092

Quoted 9 lines
drivers/scsi/mpi3mr/mpi3mr_os.c +5 -1
@@ -4073,12 +4085,19 @@ static int mpi3mr_eh_bus_reset(struct scsi_cmnd *scmd)
 	if (dev_type == MPI3_DEVICE_DEVFORM_VD) {
 		mpi3mr_wait_for_host_io(mrioc,
 			MPI3MR_RAID_ERRREC_RESET_TIMEOUT);
-		if (!mpi3mr_get_fw_pending_ios(mrioc))
+		if (!mpi3mr_get_fw_pending_ios(mrioc)) {
+			while (mrioc->reset_in_progress ||
+			       mrioc->prepare_for_reset ||
+			       mrioc->block_on_pcie_err)
+				ssleep(1);

Another possible hang?

Quoted 122 lines
drivers/scsi/mpi3mr/mpi3mr_os.c +21 -28
 			retval = SUCCESS;
+			goto out;
+		}
 	}
 	if (retval == FAILED)
 		mpi3mr_print_pending_host_io(mrioc);
 
+out:
 	sdev_printk(KERN_INFO, scmd->device,
 		"Bus reset is %s for scmd(%p)\n",
 		((retval == SUCCESS) ? "SUCCESS" : "FAILED"), scmd);
@@ -4779,7 +4798,8 @@ static int mpi3mr_qcmd(struct Scsi_Host *shost,
 		goto out;
 	}
 
-	if (mrioc->reset_in_progress) {
+	if (mrioc->reset_in_progress || mrioc->prepare_for_reset
+	    || mrioc->block_on_pcie_err) {
 		retval = SCSI_MLQUEUE_HOST_BUSY;
 		goto out;
 	}
@@ -5123,6 +5143,14 @@ mpi3mr_probe(struct pci_dev *pdev, const struct pci_device_id *id)
 	mrioc->logging_level = logging_level;
 	mrioc->shost = shost;
 	mrioc->pdev = pdev;
+	mrioc->pdevinfo.id = pdev->device;
+	mrioc->pdevinfo.revision = pdev->revision;
+	mrioc->pdevinfo.ssid = pdev->subsystem_device;
+	mrioc->pdevinfo.ssvid = pdev->subsystem_vendor;
+	mrioc->pdevinfo.bus = pdev->bus->number;
+	mrioc->pdevinfo.dev = PCI_SLOT(pdev->devfn);
+	mrioc->pdevinfo.func = PCI_FUNC(pdev->devfn);
+	mrioc->pdevinfo.segment = pci_domain_nr(pdev->bus);
 	mrioc->stop_bsgs = 1;
 
 	mrioc->max_sgl_entries = max_sgl_entries;
@@ -5276,17 +5304,21 @@ static void mpi3mr_remove(struct pci_dev *pdev)
 	struct workqueue_struct	*wq;
 	unsigned long flags;
 	struct mpi3mr_tgt_dev *tgtdev, *tgtdev_next;
-	struct mpi3mr_hba_port *port, *hba_port_next;
-	struct mpi3mr_sas_node *sas_expander, *sas_expander_next;
 
 	if (mpi3mr_get_shost_and_mrioc(pdev, &shost, &mrioc))
 		return;
 
-	mrioc = shost_priv(shost);
 	while (mrioc->reset_in_progress || mrioc->is_driver_loading)
 		ssleep(1);
 
-	if (!pci_device_is_present(mrioc->pdev)) {
+	if (mrioc->block_on_pcie_err) {
+		mrioc->block_on_pcie_err = false;
+		scsi_unblock_requests(shost);
+		mrioc->unrecoverable = 1;
+	}
+
+	if (!pci_device_is_present(mrioc->pdev) ||
+	    mrioc->pcie_err_recovery) {
 		mrioc->unrecoverable = 1;
 		mpi3mr_flush_cmds_for_unrecovered_controller(mrioc);
 	}
@@ -5316,29 +5348,6 @@ static void mpi3mr_remove(struct pci_dev *pdev)
 	mpi3mr_cleanup_ioc(mrioc);
 	mpi3mr_free_mem(mrioc);
 	mpi3mr_cleanup_resources(mrioc);
-
-	spin_lock_irqsave(&mrioc->sas_node_lock, flags);
-	list_for_each_entry_safe_reverse(sas_expander, sas_expander_next,
-	    &mrioc->sas_expander_list, list) {
-		spin_unlock_irqrestore(&mrioc->sas_node_lock, flags);
-		mpi3mr_expander_node_remove(mrioc, sas_expander);
-		spin_lock_irqsave(&mrioc->sas_node_lock, flags);
-	}
-	list_for_each_entry_safe(port, hba_port_next, &mrioc->hba_port_table_list, list) {
-		ioc_info(mrioc,
-		    "removing hba_port entry: %p port: %d from hba_port list\n",
-		    port, port->port_id);
-		list_del(&port->list);
-		kfree(port);
-	}
-	spin_unlock_irqrestore(&mrioc->sas_node_lock, flags);
-
-	if (mrioc->sas_hba.num_phys) {
-		kfree(mrioc->sas_hba.phy);
-		mrioc->sas_hba.phy = NULL;
-		mrioc->sas_hba.num_phys = 0;
-	}
-
 	spin_lock(&mrioc_list_lock);
 	list_del(&mrioc->list);
 	spin_unlock(&mrioc_list_lock);
drivers/scsi/mpi3mr/mpi3mr_transport.c +11 -0
index c0c8ab586957..8c8368104a27 100644
@@ -149,6 +149,11 @@ static int mpi3mr_report_manufacture(struct mpi3mr_ioc *mrioc,
 		return -EFAULT;
 	}
 
+	if (mrioc->pcie_err_recovery) {
+		ioc_err(mrioc, "%s: pcie error recovery in progress!\n", __func__);
+		return -EFAULT;
+	}
+
 	data_out_sz = sizeof(struct rep_manu_request);
 	data_in_sz = sizeof(struct rep_manu_reply);
 	data_out = dma_alloc_coherent(&mrioc->pdev->dev,
@@ -792,6 +797,12 @@ static int mpi3mr_set_identify(struct mpi3mr_ioc *mrioc, u16 handle,
 		return -EFAULT;
 	}
 
+	if (mrioc->pcie_err_recovery) {
+		ioc_err(mrioc, "%s: pcie error recovery in progress!\n",
+		    __func__);
+		return -EFAULT;
+	}
+
 	if ((mpi3mr_cfg_get_dev_pg0(mrioc, &ioc_status, &device_pg0,
 	    sizeof(device_pg0), MPI3_DEVICE_PGAD_FORM_HANDLE, handle))) {
 		ioc_err(mrioc, "%s: device page0 read failed\n", __func__);

drivers/scsi/mpi3mr/mpi3mr_transport.c mpi3mr_alloc_hba_port(struct mpi3mr_ioc *mrioc, u16 port_id) · line 1024

Quoted 5 lines
drivers/scsi/mpi3mr/mpi3mr_transport.c +2 -0
@@ -1009,6 +1020,9 @@ mpi3mr_alloc_hba_port(struct mpi3mr_ioc *mrioc, u16 port_id)
 	hba_port->port_id = port_id;
 	ioc_info(mrioc, "hba_port entry: %p, port: %d is added to hba_port list\n",
 	    hba_port, hba_port->port_id);
+	if (mrioc->reset_in_progress ||
+		mrioc->pcie_err_recovery)
  if (mrioc->reset_in_progress || mrioc->pcie_err_recovery)

would fit on one line and make this easier to read.

Quoted 81 lines
drivers/scsi/mpi3mr/mpi3mr_transport.c +22 -4
+		hba_port->flags = MPI3MR_HBA_PORT_FLAG_NEW;
 	list_add_tail(&hba_port->list, &mrioc->hba_port_table_list);
 	return hba_port;
 }
@@ -1057,7 +1071,7 @@ void mpi3mr_update_links(struct mpi3mr_ioc *mrioc,
 	struct mpi3mr_sas_node *mr_sas_node;
 	struct mpi3mr_sas_phy *mr_sas_phy;
 
-	if (mrioc->reset_in_progress)
+	if (mrioc->reset_in_progress || mrioc->pcie_err_recovery)
 		return;
 
 	spin_lock_irqsave(&mrioc->sas_node_lock, flags);
@@ -1965,7 +1979,7 @@ int mpi3mr_expander_add(struct mpi3mr_ioc *mrioc, u16 handle)
 	if (!handle)
 		return -1;
 
-	if (mrioc->reset_in_progress)
+	if (mrioc->reset_in_progress || mrioc->pcie_err_recovery)
 		return -1;
 
 	if ((mpi3mr_cfg_get_sas_exp_pg0(mrioc, &ioc_status, &expander_pg0,
@@ -2171,7 +2185,7 @@ void mpi3mr_expander_node_remove(struct mpi3mr_ioc *mrioc,
 	/* remove sibling ports attached to this expander */
 	list_for_each_entry_safe(mr_sas_port, next,
 	   &sas_expander->sas_port_list, port_list) {
-		if (mrioc->reset_in_progress)
+		if (mrioc->reset_in_progress || mrioc->pcie_err_recovery)
 			return;
 		if (mr_sas_port->remote_identify.device_type ==
 		    SAS_END_DEVICE)
@@ -2221,7 +2235,7 @@ void mpi3mr_expander_remove(struct mpi3mr_ioc *mrioc, u64 sas_address,
 	struct mpi3mr_sas_node *sas_expander;
 	unsigned long flags;
 
-	if (mrioc->reset_in_progress)
+	if (mrioc->reset_in_progress || mrioc->pcie_err_recovery)
 		return;
 
 	if (!hba_port)
@@ -2532,6 +2546,11 @@ static int mpi3mr_get_expander_phy_error_log(struct mpi3mr_ioc *mrioc,
 		return -EFAULT;
 	}
 
+	if (mrioc->pcie_err_recovery) {
+		ioc_err(mrioc, "%s: pcie error recovery in progress!\n", __func__);
+		return -EFAULT;
+	}
+
 	data_out_sz = sizeof(struct phy_error_log_request);
 	data_in_sz = sizeof(struct phy_error_log_reply);
 	sz = data_out_sz + data_in_sz;
@@ -2791,6 +2810,12 @@ mpi3mr_expander_phy_control(struct mpi3mr_ioc *mrioc,
 		return -EFAULT;
 	}
 
+	if (mrioc->pcie_err_recovery) {
+		ioc_err(mrioc, "%s: pcie error recovery in progress!\n",
+		    __func__);
+		return -EFAULT;
+	}
+
 	data_out_sz = sizeof(struct phy_control_request);
 	data_in_sz = sizeof(struct phy_control_reply);
 	sz = data_out_sz + data_in_sz;
@@ -3214,6 +3239,12 @@ mpi3mr_transport_smp_handler(struct bsg_job *job, struct Scsi_Host *shost,
 		goto out;
 	}
 
+	if (mrioc->pcie_err_recovery) {
+		ioc_err(mrioc, "%s: pcie error recovery in progress!\n", __func__);
+		rc = -EFAULT;
+		goto out;
+	}
+
 	rc = mpi3mr_map_smp_buffer(&mrioc->pdev->dev, &job->request_payload,
 	    &dma_addr_out, &dma_len_out, &addr_out);
 	if (rc)
Signature · 1 line
2.31.1
#7 lore raw

PATCH v2 4/6 mpi3mr: Improve Shutdown times when firmware has faulted

In reply to Ranjan Kumar

The driver monitors the controller state periodically while waiting for the shutdown notification MPI request to complete. If the firmware is faulty, the driver resets the controller and re-issues the shutdown notification. The driver will make three attempts to complete the shutdown process and will not retry the notification request if the controller reset is unsuccessful.

---
 drivers/scsi/mpi3mr/mpi3mr.h    |  1 +
 drivers/scsi/mpi3mr/mpi3mr_fw.c | 36 +++++++++++++++++++++++++--------
 2 files changed, 29 insertions(+), 8 deletions(-)
drivers/scsi/mpi3mr/mpi3mr.h +1 -0
index 1ac7f88dc1cd..de953eb055d0 100644
@@ -159,6 +159,7 @@ extern atomic64_t event_counter;
 /* Controller Reset related definitions */
 #define MPI3MR_HOSTDIAG_UNLOCK_RETRY_COUNT	5
 #define MPI3MR_MAX_RESET_RETRY_COUNT		3
+#define MPI3MR_MAX_SHUTDOWN_RETRY_COUNT		2
 
 /* ResponseCode definitions */
 #define MPI3MR_RI_MASK_RESPCODE		(0x000000FF)
drivers/scsi/mpi3mr/mpi3mr_fw.c +28 -8
index c50fc27231cd..491ef854fdba 100644
@@ -4566,9 +4566,10 @@ void mpi3mr_free_mem(struct mpi3mr_ioc *mrioc)
  */
 static void mpi3mr_issue_ioc_shutdown(struct mpi3mr_ioc *mrioc)
 {
-	u32 ioc_config, ioc_status;
-	u8 retval = 1;
+	u32 ioc_config, ioc_status, shutdown_action;
+	u8 retval = 1, retry = 0;
 	u32 timeout = MPI3MR_DEFAULT_SHUTDOWN_TIME * 10;
+	u32 timeout_remaining = 0;
 
 	ioc_info(mrioc, "Issuing shutdown Notification\n");
 	if (mrioc->unrecoverable) {
@@ -4583,15 +4584,16 @@ static void mpi3mr_issue_ioc_shutdown(struct mpi3mr_ioc *mrioc)
 		return;
 	}
 
+	shutdown_action = MPI3_SYSIF_IOC_CONFIG_SHUTDOWN_NORMAL |
+	    MPI3_SYSIF_IOC_CONFIG_DEVICE_SHUTDOWN_SEND_REQ;
 	ioc_config = readl(&mrioc->sysif_regs->ioc_configuration);
-	ioc_config |= MPI3_SYSIF_IOC_CONFIG_SHUTDOWN_NORMAL;
-	ioc_config |= MPI3_SYSIF_IOC_CONFIG_DEVICE_SHUTDOWN_SEND_REQ;
+	ioc_config |= shutdown_action;
 
 	writel(ioc_config, &mrioc->sysif_regs->ioc_configuration);
 
 	if (mrioc->facts.shutdown_timeout)
 		timeout = mrioc->facts.shutdown_timeout * 10;
-
+	timeout_remaining = timeout;
 	do {
 		ioc_status = readl(&mrioc->sysif_regs->ioc_status);
 		if ((ioc_status & MPI3_SYSIF_IOC_STATUS_SHUTDOWN_MASK)
@@ -4599,8 +4601,26 @@ static void mpi3mr_issue_ioc_shutdown(struct mpi3mr_ioc *mrioc)
 			retval = 0;
 			break;
 		}
+		if (mrioc->unrecoverable)
+			break;
+		if (ioc_status & MPI3_SYSIF_IOC_STATUS_FAULT) {
+			mpi3mr_print_fault_info(mrioc);
+			if (retry >= MPI3MR_MAX_SHUTDOWN_RETRY_COUNT)
+				break;
+			if (mpi3mr_issue_reset(mrioc,
+			    MPI3_SYSIF_HOST_DIAG_RESET_ACTION_SOFT_RESET,
+			    MPI3MR_RESET_FROM_CTLR_CLEANUP))
+				break;
+			ioc_config =
+			    readl(&mrioc->sysif_regs->ioc_configuration);
+			ioc_config |= shutdown_action;
+			writel(ioc_config,
+			    &mrioc->sysif_regs->ioc_configuration);
+			timeout_remaining = timeout;
+			retry++;
+		}
 		msleep(100);
-	} while (--timeout);
+	} while (--timeout_remaining);
 
 	ioc_status = readl(&mrioc->sysif_regs->ioc_status);
 	ioc_config = readl(&mrioc->sysif_regs->ioc_configuration);
@@ -4609,11 +4629,11 @@ static void mpi3mr_issue_ioc_shutdown(struct mpi3mr_ioc *mrioc)
 		if ((ioc_status & MPI3_SYSIF_IOC_STATUS_SHUTDOWN_MASK)
 		    == MPI3_SYSIF_IOC_STATUS_SHUTDOWN_IN_PROGRESS)
 			ioc_warn(mrioc,
-			    "shutdown still in progress after timeout\n");
+			    "shutdown still in progress\n");
 	}
 
 	ioc_info(mrioc,
-	    "Base IOC Sts/Config after %s shutdown is (0x%x)/(0x%x)\n",
+	    "ioc_status/ioc_config after %s shutdown is (0x%x)/(0x%x)\n",
 	    (!retval) ? "successful" : "failed", ioc_status,
 	    ioc_config);
 }
Signature · 1 line
2.31.1
#8 lore raw

PATCH v2 5/6 mpi3mr: Reset stop_bsgs flag post controller reset failure

In reply to Ranjan Kumar

stop_bsgs flag blocks the ioctls during controller reset. currently, it remains set after a controller failure, so resetting it.

---
 drivers/scsi/mpi3mr/mpi3mr_fw.c | 1 +
 1 file changed, 1 insertion(+)
drivers/scsi/mpi3mr/mpi3mr_fw.c +1 -0
index 491ef854fdba..209920cd5ec2 100644
@@ -5136,6 +5136,7 @@ int mpi3mr_soft_reset_handler(struct mpi3mr_ioc *mrioc,
 		mrioc->device_refresh_on = 0;
 		mrioc->unrecoverable = 1;
 		mrioc->reset_in_progress = 0;
+		mrioc->stop_bsgs = 0;
 		retval = -1;
 		mpi3mr_flush_cmds_for_unrecovered_controller(mrioc);
 	}
Signature · 1 line
2.31.1
#9 lore raw

PATCH v2 6/6 mpi3mr: Update driver version to 8.6.1.0.50

In reply to Ranjan Kumar

Update driver version to 8.6.1.0.50

---
 drivers/scsi/mpi3mr/mpi3mr.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/scsi/mpi3mr/mpi3mr.h +2 -2
index de953eb055d0..d6e43862221f 100644
@@ -56,8 +56,8 @@ extern struct list_head mrioc_list;
 extern int prot_mask;
 extern atomic64_t event_counter;
 
-#define MPI3MR_DRIVER_VERSION	"8.5.1.0.0"
-#define MPI3MR_DRIVER_RELDATE	"5-December-2023"
+#define MPI3MR_DRIVER_VERSION	"8.6.1.0.50"
+#define MPI3MR_DRIVER_RELDATE	"6-December-2023"
 
 #define MPI3MR_DRIVER_NAME	"mpi3mr"
 #define MPI3MR_DRIVER_LICENSE	"GPL"
Signature · 1 line
2.31.1