From 7e4e7867b1e551b7b8f326da3604c47332972bc6 Mon Sep 17 00:00:00 2001 From: Gavin Shan Date: Wed, 15 Jan 2014 13:16:11 +0800 Subject: powerpc/eeh: Handle multiple EEH errors For one PCI error relevant OPAL event, we possibly have multiple EEH errors for that. For example, multiple frozen PEs detected on different PHBs. Unfortunately, we didn't cover the case. The patch enumarates the return value from eeh_ops::next_error() and change eeh_handle_special_event() and eeh_ops::next_error() to handle all existing EEH errors. As Ben pointed out, we needn't list_for_each_entry_safe() since we are not deleting any PHB from the hose_list and the EEH serialized lock should be held while purging EEH events. The patch covers those suggestions as well. Signed-off-by: Gavin Shan Signed-off-by: Benjamin Herrenschmidt --- arch/powerpc/kernel/eeh_driver.c | 150 ++++++++++++++++++++------------------- 1 file changed, 78 insertions(+), 72 deletions(-) (limited to 'arch/powerpc/kernel') diff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c index 7db39203a073..baa3b06e6dab 100644 --- a/arch/powerpc/kernel/eeh_driver.c +++ b/arch/powerpc/kernel/eeh_driver.c @@ -630,84 +630,90 @@ static void eeh_handle_special_event(void) { struct eeh_pe *pe, *phb_pe; struct pci_bus *bus; - struct pci_controller *hose, *tmp; + struct pci_controller *hose; unsigned long flags; - int rc = 0; + int rc; - /* - * The return value from next_error() has been classified as follows. - * It might be good to enumerate them. However, next_error() is only - * supported by PowerNV platform for now. So it would be fine to use - * integer directly: - * - * 4 - Dead IOC 3 - Dead PHB - * 2 - Fenced PHB 1 - Frozen PE - * 0 - No error found - * - */ - rc = eeh_ops->next_error(&pe); - if (rc <= 0) - return; - switch (rc) { - case 4: - /* Mark all PHBs in dead state */ - eeh_serialize_lock(&flags); - list_for_each_entry_safe(hose, tmp, - &hose_list, list_node) { - phb_pe = eeh_phb_pe_get(hose); - if (!phb_pe) continue; - - eeh_pe_state_mark(phb_pe, - EEH_PE_ISOLATED | EEH_PE_PHB_DEAD); + do { + rc = eeh_ops->next_error(&pe); + + switch (rc) { + case EEH_NEXT_ERR_DEAD_IOC: + /* Mark all PHBs in dead state */ + eeh_serialize_lock(&flags); + + /* Purge all events */ + eeh_remove_event(NULL); + + list_for_each_entry(hose, &hose_list, list_node) { + phb_pe = eeh_phb_pe_get(hose); + if (!phb_pe) continue; + + eeh_pe_state_mark(phb_pe, + EEH_PE_ISOLATED | EEH_PE_PHB_DEAD); + } + + eeh_serialize_unlock(flags); + + break; + case EEH_NEXT_ERR_FROZEN_PE: + case EEH_NEXT_ERR_FENCED_PHB: + case EEH_NEXT_ERR_DEAD_PHB: + /* Mark the PE in fenced state */ + eeh_serialize_lock(&flags); + + /* Purge all events of the PHB */ + eeh_remove_event(pe); + + if (rc == EEH_NEXT_ERR_DEAD_PHB) + eeh_pe_state_mark(pe, + EEH_PE_ISOLATED | EEH_PE_PHB_DEAD); + else + eeh_pe_state_mark(pe, + EEH_PE_ISOLATED | EEH_PE_RECOVERING); + + eeh_serialize_unlock(flags); + + break; + case EEH_NEXT_ERR_NONE: + return; + default: + pr_warn("%s: Invalid value %d from next_error()\n", + __func__, rc); + return; } - eeh_serialize_unlock(flags); - - /* Purge all events */ - eeh_remove_event(NULL); - break; - case 3: - case 2: - case 1: - /* Mark the PE in fenced state */ - eeh_serialize_lock(&flags); - if (rc == 3) - eeh_pe_state_mark(pe, - EEH_PE_ISOLATED | EEH_PE_PHB_DEAD); - else - eeh_pe_state_mark(pe, - EEH_PE_ISOLATED | EEH_PE_RECOVERING); - eeh_serialize_unlock(flags); - - /* Purge all events of the PHB */ - eeh_remove_event(pe); - break; - default: - pr_err("%s: Invalid value %d from next_error()\n", - __func__, rc); - return; - } - /* - * For fenced PHB and frozen PE, it's handled as normal - * event. We have to remove the affected PHBs for dead - * PHB and IOC - */ - if (rc == 2 || rc == 1) - eeh_handle_normal_event(pe); - else { - list_for_each_entry_safe(hose, tmp, - &hose_list, list_node) { - phb_pe = eeh_phb_pe_get(hose); - if (!phb_pe || !(phb_pe->state & EEH_PE_PHB_DEAD)) - continue; - - bus = eeh_pe_bus_get(phb_pe); - /* Notify all devices that they're about to go down. */ - eeh_pe_dev_traverse(pe, eeh_report_failure, NULL); - pcibios_remove_pci_devices(bus); + /* + * For fenced PHB and frozen PE, it's handled as normal + * event. We have to remove the affected PHBs for dead + * PHB and IOC + */ + if (rc == EEH_NEXT_ERR_FROZEN_PE || + rc == EEH_NEXT_ERR_FENCED_PHB) { + eeh_handle_normal_event(pe); + } else { + list_for_each_entry(hose, &hose_list, list_node) { + phb_pe = eeh_phb_pe_get(hose); + if (!phb_pe || + !(phb_pe->state & EEH_PE_PHB_DEAD)) + continue; + + /* Notify all devices to be down */ + bus = eeh_pe_bus_get(phb_pe); + eeh_pe_dev_traverse(pe, + eeh_report_failure, NULL); + pcibios_remove_pci_devices(bus); + } } - } + + /* + * If we have detected dead IOC, we needn't proceed + * any more since all PHBs would have been removed + */ + if (rc == EEH_NEXT_ERR_DEAD_IOC) + break; + } while (rc != EEH_NEXT_ERR_NONE); } /** -- cgit v1.2.3