6 files changed, 57 insertions, 19 deletions

diff --git a/drivers/scsi/mpt2sas/mpt2sas_base.c b/drivers/scsi/mpt2sas/mpt2sas_base.c
index 1f22a764927a..57bcd5c9dcff 100644
--- a/drivers/scsi/mpt2sas/mpt2sas_base.c
+++ b/drivers/scsi/mpt2sas/mpt2sas_base.c
@@ -1311,6 +1311,9 @@ mpt2sas_base_map_resources(struct MPT2SAS_ADAPTER *ioc)
         printk(MPT2SAS_INFO_FMT "ioport(0x%016llx), size(%d)\n",
             ioc->name, (unsigned long long)pio_chip, pio_sz);
 
+        /* Save PCI configuration state for recovery from PCI AER/EEH errors */
+        pci_save_state(pdev);
+
         return 0;
 
  out_fail:
@@ -3407,6 +3410,9 @@ _base_make_ioc_ready(struct MPT2SAS_ADAPTER *ioc, int sleep_flag,
         dinitprintk(ioc, printk(MPT2SAS_INFO_FMT "%s\n", ioc->name,
             __func__));
 
+        if (ioc->pci_error_recovery)
+                return 0;
+
         ioc_state = mpt2sas_base_get_iocstate(ioc, 0);
         dhsprintk(ioc, printk(MPT2SAS_INFO_FMT "%s: ioc_state(0x%08x)\n",
             ioc->name, __func__, ioc_state));
@@ -3869,6 +3875,13 @@ mpt2sas_base_hard_reset_handler(struct MPT2SAS_ADAPTER *ioc, int sleep_flag,
         dtmprintk(ioc, printk(MPT2SAS_INFO_FMT "%s: enter\n", ioc->name,
             __func__));
 
+        if (ioc->pci_error_recovery) {
+                printk(MPT2SAS_ERR_FMT "%s: pci error recovery reset\n",
+                    ioc->name, __func__);
+                r = 0;
+                goto out;
+        }
+
         if (mpt2sas_fwfault_debug)
                 mpt2sas_halt_firmware(ioc);
 
diff --git a/drivers/scsi/mpt2sas/mpt2sas_base.h b/drivers/scsi/mpt2sas/mpt2sas_base.h
index 0b0ef5e7899a..0ebef0c0d949 100644
--- a/drivers/scsi/mpt2sas/mpt2sas_base.h
+++ b/drivers/scsi/mpt2sas/mpt2sas_base.h
@@ -477,6 +477,7 @@ typedef void (*MPT_ADD_SGE)(void *paddr, u32 flags_length, dma_addr_t dma_addr);
  * @ioc_link_reset_in_progress: phy/hard reset in progress
  * @ignore_loginfos: ignore loginfos during task managment
  * @remove_host: flag for when driver unloads, to avoid sending dev resets
+ * @pci_error_recovery: flag to prevent ioc access until slot reset completes
  * @wait_for_port_enable_to_complete:
  * @msix_enable: flag indicating msix is enabled
  * @msix_vector_count: number msix vectors
@@ -617,6 +618,7 @@ struct MPT2SAS_ADAPTER {
 
         u8              ignore_loginfos;
         u8              remove_host;
+        u8              pci_error_recovery;
         u8              wait_for_port_enable_to_complete;
 
         u8              msix_enable;
diff --git a/drivers/scsi/mpt2sas/mpt2sas_config.c b/drivers/scsi/mpt2sas/mpt2sas_config.c
index e26f9206a521..6afd67b324fe 100644
--- a/drivers/scsi/mpt2sas/mpt2sas_config.c
+++ b/drivers/scsi/mpt2sas/mpt2sas_config.c
@@ -401,7 +401,7 @@ _config_request(struct MPT2SAS_ADAPTER *ioc, Mpi2ConfigRequest_t
                 if (ioc->config_cmds.smid == smid)
                         mpt2sas_base_free_smid(ioc, smid);
                 if ((ioc->shost_recovery) || (ioc->config_cmds.status &
-                    MPT2_CMD_RESET))
+                    MPT2_CMD_RESET) || ioc->pci_error_recovery)
                         goto retry_config;
                 issue_host_reset = 1;
                 r = -EFAULT;
diff --git a/drivers/scsi/mpt2sas/mpt2sas_ctl.c b/drivers/scsi/mpt2sas/mpt2sas_ctl.c
index 55ac1cb34778..b774973f0765 100644
--- a/drivers/scsi/mpt2sas/mpt2sas_ctl.c
+++ b/drivers/scsi/mpt2sas/mpt2sas_ctl.c
@@ -2156,7 +2156,7 @@ _ctl_ioctl_main(struct file *file, unsigned int cmd, void __user *arg)
                     !ioc)
                         return -ENODEV;
 
-                if (ioc->shost_recovery)
+                if (ioc->shost_recovery || ioc->pci_error_recovery)
                         return -EAGAIN;
 
                 if (_IOC_SIZE(cmd) == sizeof(struct mpt2_ioctl_command)) {
@@ -2275,7 +2275,7 @@ _ctl_compat_mpt_command(struct file *file, unsigned cmd, unsigned long arg)
         if (_ctl_verify_adapter(karg32.hdr.ioc_number, &ioc) == -1 || !ioc)
                 return -ENODEV;
 
-        if (ioc->shost_recovery)
+        if (ioc->shost_recovery || ioc->pci_error_recovery)
                 return -EAGAIN;
 
         memset(&karg, 0, sizeof(struct mpt2_ioctl_command));
diff --git a/drivers/scsi/mpt2sas/mpt2sas_scsih.c b/drivers/scsi/mpt2sas/mpt2sas_scsih.c
index 854cc91e7aac..6273abd0535e 100644
--- a/drivers/scsi/mpt2sas/mpt2sas_scsih.c
+++ b/drivers/scsi/mpt2sas/mpt2sas_scsih.c
@@ -1997,7 +1997,8 @@ mpt2sas_scsih_issue_tm(struct MPT2SAS_ADAPTER *ioc, u16 handle, uint channel,
                 goto err_out;
         }
 
-        if (ioc->shost_recovery || ioc->remove_host) {
+        if (ioc->shost_recovery || ioc->remove_host ||
+            ioc->pci_error_recovery) {
                 printk(MPT2SAS_INFO_FMT "%s: host reset in progress!\n",
                     __func__, ioc->name);
                 rc = FAILED;
@@ -2644,7 +2645,8 @@ _scsih_tm_tr_send(struct MPT2SAS_ADAPTER *ioc, u16 handle)
         unsigned long flags;
         struct _tr_list *delayed_tr;
 
-        if (ioc->shost_recovery || ioc->remove_host) {
+        if (ioc->shost_recovery || ioc->remove_host ||
+            ioc->pci_error_recovery) {
                 dewtprintk(ioc, printk(MPT2SAS_INFO_FMT "%s: host reset in "
                    "progress!\n", __func__, ioc->name));
                 return;
@@ -2742,7 +2744,8 @@ _scsih_tm_tr_volume_send(struct MPT2SAS_ADAPTER *ioc, u16 handle)
         u16 smid;
         struct _tr_list *delayed_tr;
 
-        if (ioc->shost_recovery || ioc->remove_host) {
+        if (ioc->shost_recovery || ioc->remove_host ||
+            ioc->pci_error_recovery) {
                 dewtprintk(ioc, printk(MPT2SAS_INFO_FMT "%s: host reset in "
                    "progress!\n", __func__, ioc->name));
                 return;
@@ -2793,7 +2796,8 @@ _scsih_tm_volume_tr_complete(struct MPT2SAS_ADAPTER *ioc, u16 smid,
         Mpi2SCSITaskManagementReply_t *mpi_reply =
             mpt2sas_base_get_reply_virt_addr(ioc, reply);
 
-        if (ioc->shost_recovery || ioc->remove_host) {
+        if (ioc->shost_recovery || ioc->remove_host ||
+            ioc->pci_error_recovery) {
                 dewtprintk(ioc, printk(MPT2SAS_INFO_FMT "%s: host reset in "
                    "progress!\n", __func__, ioc->name));
                 return 1;
@@ -2845,7 +2849,8 @@ _scsih_tm_tr_complete(struct MPT2SAS_ADAPTER *ioc, u16 smid, u8 msix_index,
         Mpi2SasIoUnitControlRequest_t *mpi_request;
         u16 smid_sas_ctrl;
 
-        if (ioc->shost_recovery || ioc->remove_host) {
+        if (ioc->shost_recovery || ioc->remove_host ||
+            ioc->pci_error_recovery) {
                 dewtprintk(ioc, printk(MPT2SAS_INFO_FMT "%s: host reset in "
                    "progress!\n", __func__, ioc->name));
                 return 1;
@@ -3187,7 +3192,10 @@ _scsih_flush_running_cmds(struct MPT2SAS_ADAPTER *ioc)
                 count++;
                 mpt2sas_base_free_smid(ioc, smid);
                 scsi_dma_unmap(scmd);
-                scmd->result = DID_RESET << 16;
+                if (ioc->pci_error_recovery)
+                        scmd->result = DID_NO_CONNECT << 16;
+                else
+                        scmd->result = DID_RESET << 16;
                 scmd->scsi_done(scmd);
         }
         dtmprintk(ioc, printk(MPT2SAS_INFO_FMT "completing %d cmds\n",
@@ -3324,6 +3332,12 @@ _scsih_qcmd(struct scsi_cmnd *scmd, void (*done)(struct scsi_cmnd *))
                 return 0;
         }
 
+        if (ioc->pci_error_recovery) {
+                scmd->result = DID_NO_CONNECT << 16;
+                scmd->scsi_done(scmd);
+                return 0;
+        }
+
         sas_target_priv_data = sas_device_priv_data->sas_target;
         /* invalid device handle */
         if (sas_target_priv_data->handle == MPT2SAS_INVALID_DEVICE_HANDLE) {
@@ -4156,7 +4170,7 @@ _scsih_expander_add(struct MPT2SAS_ADAPTER *ioc, u16 handle)
         if (!handle)
                 return -1;
 
-        if (ioc->shost_recovery)
+        if (ioc->shost_recovery || ioc->pci_error_recovery)
                 return -1;
 
         if ((mpt2sas_config_get_expander_pg0(ioc, &mpi_reply, &expander_pg0,
@@ -4734,7 +4748,7 @@ _scsih_sas_topology_change_event(struct MPT2SAS_ADAPTER *ioc,
                 _scsih_sas_topology_change_event_debug(ioc, event_data);
 #endif
 
-        if (ioc->shost_recovery || ioc->remove_host)
+        if (ioc->shost_recovery || ioc->remove_host || ioc->pci_error_recovery)
                 return;
 
         if (!ioc->sas_hba.num_phys)
@@ -4773,7 +4787,8 @@ _scsih_sas_topology_change_event(struct MPT2SAS_ADAPTER *ioc,
                             "expander event\n", ioc->name));
                         return;
                 }
-                if (ioc->shost_recovery || ioc->remove_host)
+                if (ioc->shost_recovery || ioc->remove_host ||
+                    ioc->pci_error_recovery)
                         return;
                 phy_number = event_data->StartPhyNum + i;
                 reason_code = event_data->PHY[i].PhyStatus &
@@ -6273,7 +6288,8 @@ _firmware_event_work(struct work_struct *work)
         struct MPT2SAS_ADAPTER *ioc = fw_event->ioc;
 
         /* the queue is being flushed so ignore this event */
-        if (ioc->remove_host || fw_event->cancel_pending_work) {
+        if (ioc->remove_host || fw_event->cancel_pending_work ||
+            ioc->pci_error_recovery) {
                 _scsih_fw_event_free(ioc, fw_event);
                 return;
         }
@@ -6355,7 +6371,7 @@ mpt2sas_scsih_event_callback(struct MPT2SAS_ADAPTER *ioc, u8 msix_index,
         u16 sz;
 
         /* events turned off due to host reset or driver unloading */
-        if (ioc->remove_host)
+        if (ioc->remove_host || ioc->pci_error_recovery)
                 return 1;
 
         mpi_reply = mpt2sas_base_get_reply_virt_addr(ioc, reply);
@@ -7058,12 +7074,17 @@ _scsih_pci_error_detected(struct pci_dev *pdev, pci_channel_state_t state)
         case pci_channel_io_normal:
                 return PCI_ERS_RESULT_CAN_RECOVER;
         case pci_channel_io_frozen:
+                /* Fatal error, prepare for slot reset */
+                ioc->pci_error_recovery = 1;
                 scsi_block_requests(ioc->shost);
                 mpt2sas_base_stop_watchdog(ioc);
                 mpt2sas_base_free_resources(ioc);
                 return PCI_ERS_RESULT_NEED_RESET;
         case pci_channel_io_perm_failure:
-                _scsih_remove(pdev);
+                /* Permanent error, prepare for device removal */
+                ioc->pci_error_recovery = 1;
+                mpt2sas_base_stop_watchdog(ioc);
+                _scsih_flush_running_cmds(ioc);
                 return PCI_ERS_RESULT_DISCONNECT;
         }
         return PCI_ERS_RESULT_NEED_RESET;
@@ -7087,7 +7108,9 @@ _scsih_pci_slot_reset(struct pci_dev *pdev)
         printk(MPT2SAS_INFO_FMT "PCI error: slot reset callback!!\n",
                 ioc->name);
 
+        ioc->pci_error_recovery = 0;
         ioc->pdev = pdev;
+        pci_restore_state(pdev);
         rc = mpt2sas_base_map_resources(ioc);
         if (rc)
                 return PCI_ERS_RESULT_DISCONNECT;
diff --git a/drivers/scsi/mpt2sas/mpt2sas_transport.c b/drivers/scsi/mpt2sas/mpt2sas_transport.c
index f29ea5e78bb3..b55c6dc07470 100644
--- a/drivers/scsi/mpt2sas/mpt2sas_transport.c
+++ b/drivers/scsi/mpt2sas/mpt2sas_transport.c
@@ -140,7 +140,7 @@ _transport_set_identify(struct MPT2SAS_ADAPTER *ioc, u16 handle,
         u32 device_info;
         u32 ioc_status;
 
-        if (ioc->shost_recovery) {
+        if (ioc->shost_recovery || ioc->pci_error_recovery) {
                 printk(MPT2SAS_INFO_FMT "%s: host reset in progress!\n",
                     __func__, ioc->name);
                 return -EFAULT;
@@ -302,7 +302,7 @@ _transport_expander_report_manufacture(struct MPT2SAS_ADAPTER *ioc,
         u64 *sas_address_le;
         u16 wait_state_count;
 
-        if (ioc->shost_recovery) {
+        if (ioc->shost_recovery || ioc->pci_error_recovery) {
                 printk(MPT2SAS_INFO_FMT "%s: host reset in progress!\n",
                     __func__, ioc->name);
                 return -EFAULT;
@@ -894,7 +894,7 @@ mpt2sas_transport_update_links(struct MPT2SAS_ADAPTER *ioc,
         struct _sas_node *sas_node;
         struct _sas_phy *mpt2sas_phy;
 
-        if (ioc->shost_recovery)
+        if (ioc->shost_recovery || ioc->pci_error_recovery)
                 return;
 
         spin_lock_irqsave(&ioc->sas_node_lock, flags);
@@ -997,7 +997,7 @@ _transport_get_expander_phy_error_log(struct MPT2SAS_ADAPTER *ioc,
         u64 *sas_address_le;
         u16 wait_state_count;
 
-        if (ioc->shost_recovery) {
+        if (ioc->shost_recovery || ioc->pci_error_recovery) {
                 printk(MPT2SAS_INFO_FMT "%s: host reset in progress!\n",
                     __func__, ioc->name);
                 return -EFAULT;