diff options
Diffstat (limited to 'kernel/arch/powerpc/platforms/powernv/eeh-powernv.c')
-rw-r--r-- | kernel/arch/powerpc/platforms/powernv/eeh-powernv.c | 165 |
1 files changed, 82 insertions, 83 deletions
diff --git a/kernel/arch/powerpc/platforms/powernv/eeh-powernv.c b/kernel/arch/powerpc/platforms/powernv/eeh-powernv.c index ce738ab3d..2ba602591 100644 --- a/kernel/arch/powerpc/platforms/powernv/eeh-powernv.c +++ b/kernel/arch/powerpc/platforms/powernv/eeh-powernv.c @@ -16,6 +16,7 @@ #include <linux/delay.h> #include <linux/export.h> #include <linux/init.h> +#include <linux/interrupt.h> #include <linux/list.h> #include <linux/msi.h> #include <linux/of.h> @@ -40,18 +41,13 @@ #include "pci.h" static bool pnv_eeh_nb_init = false; +static int eeh_event_irq = -EINVAL; -/** - * pnv_eeh_init - EEH platform dependent initialization - * - * EEH platform dependent initialization on powernv - */ static int pnv_eeh_init(void) { struct pci_controller *hose; struct pnv_phb *phb; - /* We require OPALv3 */ if (!firmware_has_feature(FW_FEATURE_OPALv3)) { pr_warn("%s: OPALv3 is required !\n", __func__); @@ -75,9 +71,9 @@ static int pnv_eeh_init(void) /* * PE#0 should be regarded as valid by EEH core * if it's not the reserved one. Currently, we - * have the reserved PE#0 and PE#127 for PHB3 + * have the reserved PE#255 and PE#127 for PHB3 * and P7IOC separately. So we should regard - * PE#0 as valid for P7IOC. + * PE#0 as valid for PHB3 and P7IOC. */ if (phb->ioda.reserved_pe != 0) eeh_add_flag(EEH_VALID_PE_ZERO); @@ -88,34 +84,22 @@ static int pnv_eeh_init(void) return 0; } -static int pnv_eeh_event(struct notifier_block *nb, - unsigned long events, void *change) +static irqreturn_t pnv_eeh_event(int irq, void *data) { - uint64_t changed_evts = (uint64_t)change; - /* - * We simply send special EEH event if EEH has - * been enabled, or clear pending events in - * case that we enable EEH soon + * We simply send a special EEH event if EEH has been + * enabled. We don't care about EEH events until we've + * finished processing the outstanding ones. Event processing + * gets unmasked in next_error() if EEH is enabled. */ - if (!(changed_evts & OPAL_EVENT_PCI_ERROR) || - !(events & OPAL_EVENT_PCI_ERROR)) - return 0; + disable_irq_nosync(irq); if (eeh_enabled()) eeh_send_failure_event(NULL); - else - opal_notifier_update_evt(OPAL_EVENT_PCI_ERROR, 0x0ul); - return 0; + return IRQ_HANDLED; } -static struct notifier_block pnv_eeh_nb = { - .notifier_call = pnv_eeh_event, - .next = NULL, - .priority = 0 -}; - #ifdef CONFIG_DEBUG_FS static ssize_t pnv_eeh_ei_write(struct file *filp, const char __user *user_buf, @@ -237,16 +221,28 @@ static int pnv_eeh_post_init(void) /* Register OPAL event notifier */ if (!pnv_eeh_nb_init) { - ret = opal_notifier_register(&pnv_eeh_nb); - if (ret) { - pr_warn("%s: Can't register OPAL event notifier (%d)\n", - __func__, ret); + eeh_event_irq = opal_event_request(ilog2(OPAL_EVENT_PCI_ERROR)); + if (eeh_event_irq < 0) { + pr_err("%s: Can't register OPAL event interrupt (%d)\n", + __func__, eeh_event_irq); + return eeh_event_irq; + } + + ret = request_irq(eeh_event_irq, pnv_eeh_event, + IRQ_TYPE_LEVEL_HIGH, "opal-eeh", NULL); + if (ret < 0) { + irq_dispose_mapping(eeh_event_irq); + pr_err("%s: Can't request OPAL event interrupt (%d)\n", + __func__, eeh_event_irq); return ret; } pnv_eeh_nb_init = true; } + if (!eeh_enabled()) + disable_irq(eeh_event_irq); + list_for_each_entry(hose, &hose_list, list_node) { phb = hose->private_data; @@ -282,33 +278,23 @@ static int pnv_eeh_post_init(void) #endif /* CONFIG_DEBUG_FS */ } - return ret; } -static int pnv_eeh_cap_start(struct pci_dn *pdn) +static int pnv_eeh_find_cap(struct pci_dn *pdn, int cap) { - u32 status; + int pos = PCI_CAPABILITY_LIST; + int cnt = 48; /* Maximal number of capabilities */ + u32 status, id; if (!pdn) return 0; + /* Check if the device supports capabilities */ pnv_pci_cfg_read(pdn, PCI_STATUS, 2, &status); if (!(status & PCI_STATUS_CAP_LIST)) return 0; - return PCI_CAPABILITY_LIST; -} - -static int pnv_eeh_find_cap(struct pci_dn *pdn, int cap) -{ - int pos = pnv_eeh_cap_start(pdn); - int cnt = 48; /* Maximal number of capabilities */ - u32 id; - - if (!pos) - return 0; - while (cnt--) { pnv_pci_cfg_read(pdn, pos, 1, &pos); if (pos < 0x40) @@ -441,11 +427,14 @@ static void *pnv_eeh_probe(struct pci_dn *pdn, void *data) * that PE to block its config space. * * Broadcom Austin 4-ports NICs (14e4:1657) + * Broadcom Shiner 4-ports 1G NICs (14e4:168a) * Broadcom Shiner 2-ports 10G NICs (14e4:168e) */ if ((pdn->vendor_id == PCI_VENDOR_ID_BROADCOM && pdn->device_id == 0x1657) || (pdn->vendor_id == PCI_VENDOR_ID_BROADCOM && + pdn->device_id == 0x168a) || + (pdn->vendor_id == PCI_VENDOR_ID_BROADCOM && pdn->device_id == 0x168e)) edev->pe->state |= EEH_PE_CFG_RESTRICTED; @@ -455,9 +444,12 @@ static void *pnv_eeh_probe(struct pci_dn *pdn, void *data) * PCI devices of the PE are expected to be removed prior * to PE reset. */ - if (!edev->pe->bus) + if (!(edev->pe->state & EEH_PE_PRI_BUS)) { edev->pe->bus = pci_find_bus(hose->global_number, pdn->busno); + if (edev->pe->bus) + edev->pe->state |= EEH_PE_PRI_BUS; + } /* * Enable EEH explicitly so that we will do EEH check @@ -485,10 +477,9 @@ static int pnv_eeh_set_option(struct eeh_pe *pe, int option) struct pci_controller *hose = pe->phb; struct pnv_phb *phb = hose->private_data; bool freeze_pe = false; - int opt, ret = 0; + int opt; s64 rc; - /* Sanity check on option */ switch (option) { case EEH_OPT_DISABLE: return -EPERM; @@ -509,38 +500,37 @@ static int pnv_eeh_set_option(struct eeh_pe *pe, int option) return -EINVAL; } - /* If PHB supports compound PE, to handle it */ + /* Freeze master and slave PEs if PHB supports compound PEs */ if (freeze_pe) { if (phb->freeze_pe) { phb->freeze_pe(phb, pe->addr); - } else { - rc = opal_pci_eeh_freeze_set(phb->opal_id, - pe->addr, opt); - if (rc != OPAL_SUCCESS) { - pr_warn("%s: Failure %lld freezing " - "PHB#%x-PE#%x\n", - __func__, rc, - phb->hose->global_number, pe->addr); - ret = -EIO; - } + return 0; } - } else { - if (phb->unfreeze_pe) { - ret = phb->unfreeze_pe(phb, pe->addr, opt); - } else { - rc = opal_pci_eeh_freeze_clear(phb->opal_id, - pe->addr, opt); - if (rc != OPAL_SUCCESS) { - pr_warn("%s: Failure %lld enable %d " - "for PHB#%x-PE#%x\n", - __func__, rc, option, - phb->hose->global_number, pe->addr); - ret = -EIO; - } + + rc = opal_pci_eeh_freeze_set(phb->opal_id, pe->addr, opt); + if (rc != OPAL_SUCCESS) { + pr_warn("%s: Failure %lld freezing PHB#%x-PE#%x\n", + __func__, rc, phb->hose->global_number, + pe->addr); + return -EIO; } + + return 0; } - return ret; + /* Unfreeze master and slave PEs if PHB supports */ + if (phb->unfreeze_pe) + return phb->unfreeze_pe(phb, pe->addr, opt); + + rc = opal_pci_eeh_freeze_clear(phb->opal_id, pe->addr, opt); + if (rc != OPAL_SUCCESS) { + pr_warn("%s: Failure %lld enable %d for PHB#%x-PE#%x\n", + __func__, rc, option, phb->hose->global_number, + pe->addr); + return -EIO; + } + + return 0; } /** @@ -979,7 +969,7 @@ static int pnv_eeh_reset(struct eeh_pe *pe, int option) /** * pnv_eeh_wait_state - Wait for PE state * @pe: EEH PE - * @max_wait: maximal period in microsecond + * @max_wait: maximal period in millisecond * * Wait for the state of associated PE. It might take some time * to retrieve the PE's state. @@ -1000,13 +990,13 @@ static int pnv_eeh_wait_state(struct eeh_pe *pe, int max_wait) if (ret != EEH_STATE_UNAVAILABLE) return ret; - max_wait -= mwait; if (max_wait <= 0) { pr_warn("%s: Timeout getting PE#%x's state (%d)\n", __func__, pe->addr, max_wait); return EEH_STATE_NOT_SUPPORT; } + max_wait -= mwait; msleep(mwait); } @@ -1063,7 +1053,6 @@ static int pnv_eeh_err_inject(struct eeh_pe *pe, int type, int func, struct pnv_phb *phb = hose->private_data; s64 rc; - /* Sanity check on error type */ if (type != OPAL_ERR_INJECT_TYPE_IOA_BUS_ERR && type != OPAL_ERR_INJECT_TYPE_IOA_BUS_ERR64) { pr_warn("%s: Invalid error type %d\n", @@ -1303,12 +1292,10 @@ static int pnv_eeh_next_error(struct eeh_pe **pe) int state, ret = EEH_NEXT_ERR_NONE; /* - * While running here, it's safe to purge the event queue. - * And we should keep the cached OPAL notifier event sychronized - * between the kernel and firmware. + * While running here, it's safe to purge the event queue. The + * event should still be masked. */ eeh_remove_event(NULL, false); - opal_notifier_update_evt(OPAL_EVENT_PCI_ERROR, 0x0ul); list_for_each_entry(hose, &hose_list, list_node) { /* @@ -1394,11 +1381,19 @@ static int pnv_eeh_next_error(struct eeh_pe **pe) */ if (pnv_eeh_get_pe(hose, be64_to_cpu(frozen_pe_no), pe)) { - /* Try best to clear it */ pr_info("EEH: Clear non-existing PHB#%x-PE#%llx\n", - hose->global_number, frozen_pe_no); + hose->global_number, be64_to_cpu(frozen_pe_no)); pr_info("EEH: PHB location: %s\n", eeh_pe_loc_get(phb_pe)); + + /* Dump PHB diag-data */ + rc = opal_pci_get_phb_diag_data2(phb->opal_id, + phb->diag.blob, PNV_PCI_DIAG_BUF_SIZE); + if (rc == OPAL_SUCCESS) + pnv_pci_dump_phb_diag_data(hose, + phb->diag.blob); + + /* Try best to clear it */ opal_pci_eeh_freeze_clear(phb->opal_id, frozen_pe_no, OPAL_EEH_ACTION_CLEAR_FREEZE_ALL); @@ -1477,6 +1472,10 @@ static int pnv_eeh_next_error(struct eeh_pe **pe) break; } + /* Unmask the event */ + if (ret == EEH_NEXT_ERR_NONE && eeh_enabled()) + enable_irq(eeh_event_irq); + return ret; } |