[PATCH 18/21] ppc/eeh: handle EEH error based on PE
Gavin Shan
shangw at linux.vnet.ibm.com
Thu Jun 28 02:01:48 EST 2012
The patch reworks the current implementation so that the eeh errors
will be handled basing on PE instead of eeh device.
Signed-off-by: Gavin Shan <shangw at linux.vnet.ibm.com>
---
arch/powerpc/include/asm/eeh.h | 1 +
arch/powerpc/include/asm/eeh_event.h | 2 +-
arch/powerpc/platforms/pseries/eeh_driver.c | 229 +++++++++++----------------
arch/powerpc/platforms/pseries/eeh_event.c | 2 +-
arch/powerpc/platforms/pseries/eeh_pe.c | 27 ++++
5 files changed, 124 insertions(+), 137 deletions(-)
diff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h
index 9a9fe28..e07ece1 100644
--- a/arch/powerpc/include/asm/eeh.h
+++ b/arch/powerpc/include/asm/eeh.h
@@ -174,6 +174,7 @@ int eeh_pe_remove(struct eeh_dev *edev);
void *eeh_pe_dev_traverse(struct eeh_pe *root,
eeh_traverse_func fn, void *flag);
void eeh_pe_restore_bars(struct eeh_pe *pe);
+struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe);
void * __devinit eeh_dev_init(struct device_node *dn, void *data);
void __devinit eeh_dev_phb_init_dynamic(struct pci_controller *phb);
diff --git a/arch/powerpc/include/asm/eeh_event.h b/arch/powerpc/include/asm/eeh_event.h
index dc722b5..de67d83 100644
--- a/arch/powerpc/include/asm/eeh_event.h
+++ b/arch/powerpc/include/asm/eeh_event.h
@@ -32,7 +32,7 @@ struct eeh_event {
};
int eeh_send_failure_event(struct eeh_pe *pe);
-struct eeh_dev *handle_eeh_events(struct eeh_event *);
+void eeh_handle_event(struct eeh_pe *pe);
#endif /* __KERNEL__ */
#endif /* ASM_POWERPC_EEH_EVENT_H */
diff --git a/arch/powerpc/platforms/pseries/eeh_driver.c b/arch/powerpc/platforms/pseries/eeh_driver.c
index baf92cd..343c807 100644
--- a/arch/powerpc/platforms/pseries/eeh_driver.c
+++ b/arch/powerpc/platforms/pseries/eeh_driver.c
@@ -116,28 +116,35 @@ static void eeh_enable_irq(struct pci_dev *dev)
/**
* eeh_report_error - Report pci error to each device driver
- * @dev: PCI device
+ * @data: eeh device
* @userdata: return value
*
* Report an EEH error to each device driver, collect up and
* merge the device driver responses. Cumulative response
* passed back in "userdata".
*/
-static int eeh_report_error(struct pci_dev *dev, void *userdata)
+static void *eeh_report_error(void *data, void *userdata)
{
+ struct eeh_dev *edev = (struct eeh_dev *)data;
+ struct pci_dev *dev = eeh_dev_to_pci_dev(edev);
enum pci_ers_result rc, *res = userdata;
struct pci_driver *driver = dev->driver;
+ /* We might not have the associated PCI device,
+ * then we should continue for next one.
+ */
+ if (!dev) return NULL;
+
dev->error_state = pci_channel_io_frozen;
if (!driver)
- return 0;
+ return NULL;
eeh_disable_irq(dev);
if (!driver->err_handler ||
!driver->err_handler->error_detected)
- return 0;
+ return NULL;
rc = driver->err_handler->error_detected(dev, pci_channel_io_frozen);
@@ -145,27 +152,31 @@ static int eeh_report_error(struct pci_dev *dev, void *userdata)
if (rc == PCI_ERS_RESULT_NEED_RESET) *res = rc;
if (*res == PCI_ERS_RESULT_NONE) *res = rc;
- return 0;
+ return NULL;
}
/**
* eeh_report_mmio_enabled - Tell drivers that MMIO has been enabled
- * @dev: PCI device
+ * @data: eeh device
* @userdata: return value
*
* Tells each device driver that IO ports, MMIO and config space I/O
* are now enabled. Collects up and merges the device driver responses.
* Cumulative response passed back in "userdata".
*/
-static int eeh_report_mmio_enabled(struct pci_dev *dev, void *userdata)
+static void *eeh_report_mmio_enabled(void *data, void *userdata)
{
+ struct eeh_dev *edev = (struct eeh_dev *)data;
+ struct pci_dev *dev = eeh_dev_to_pci_dev(edev);
enum pci_ers_result rc, *res = userdata;
- struct pci_driver *driver = dev->driver;
+ struct pci_driver *driver;
- if (!driver ||
+ if (!dev) return NULL;
+
+ if (!(driver = dev->driver) ||
!driver->err_handler ||
!driver->err_handler->mmio_enabled)
- return 0;
+ return NULL;
rc = driver->err_handler->mmio_enabled(dev);
@@ -173,12 +184,12 @@ static int eeh_report_mmio_enabled(struct pci_dev *dev, void *userdata)
if (rc == PCI_ERS_RESULT_NEED_RESET) *res = rc;
if (*res == PCI_ERS_RESULT_NONE) *res = rc;
- return 0;
+ return NULL;
}
/**
* eeh_report_reset - Tell device that slot has been reset
- * @dev: PCI device
+ * @data: eeh device
* @userdata: return value
*
* This routine must be called while EEH tries to reset particular
@@ -186,13 +197,15 @@ static int eeh_report_mmio_enabled(struct pci_dev *dev, void *userdata)
* some actions, usually to save data the driver needs so that the
* driver can work again while the device is recovered.
*/
-static int eeh_report_reset(struct pci_dev *dev, void *userdata)
+static void *eeh_report_reset(void *data, void *userdata)
{
+ struct eeh_dev *edev = (struct eeh_dev *)data;
+ struct pci_dev *dev = eeh_dev_to_pci_dev(edev);
enum pci_ers_result rc, *res = userdata;
- struct pci_driver *driver = dev->driver;
+ struct pci_driver *driver;
- if (!driver)
- return 0;
+ if (!dev || !(driver = dev->driver))
+ return NULL;
dev->error_state = pci_channel_io_normal;
@@ -200,7 +213,7 @@ static int eeh_report_reset(struct pci_dev *dev, void *userdata)
if (!driver->err_handler ||
!driver->err_handler->slot_reset)
- return 0;
+ return NULL;
rc = driver->err_handler->slot_reset(dev);
if ((*res == PCI_ERS_RESULT_NONE) ||
@@ -208,82 +221,89 @@ static int eeh_report_reset(struct pci_dev *dev, void *userdata)
if (*res == PCI_ERS_RESULT_DISCONNECT &&
rc == PCI_ERS_RESULT_NEED_RESET) *res = rc;
- return 0;
+ return NULL;
}
/**
* eeh_report_resume - Tell device to resume normal operations
- * @dev: PCI device
+ * @data: eeh device
* @userdata: return value
*
* This routine must be called to notify the device driver that it
* could resume so that the device driver can do some initialization
* to make the recovered device work again.
*/
-static int eeh_report_resume(struct pci_dev *dev, void *userdata)
+static void *eeh_report_resume(void *data, void *userdata)
{
- struct pci_driver *driver = dev->driver;
+ struct eeh_dev *edev = (struct eeh_dev *)data;
+ struct pci_dev *dev = eeh_dev_to_pci_dev(edev);
+ struct pci_driver *driver;
+
+ if (!dev) return NULL;
dev->error_state = pci_channel_io_normal;
- if (!driver)
- return 0;
+ if (!(driver = dev->driver))
+ return NULL;
eeh_enable_irq(dev);
if (!driver->err_handler ||
!driver->err_handler->resume)
- return 0;
+ return NULL;
driver->err_handler->resume(dev);
- return 0;
+ return NULL;
}
/**
* eeh_report_failure - Tell device driver that device is dead.
- * @dev: PCI device
+ * @data: eeh device
* @userdata: return value
*
* This informs the device driver that the device is permanently
* dead, and that no further recovery attempts will be made on it.
*/
-static int eeh_report_failure(struct pci_dev *dev, void *userdata)
+static void *eeh_report_failure(void *data, void *userdata)
{
- struct pci_driver *driver = dev->driver;
+ struct eeh_dev *edev = (struct eeh_dev *)data;
+ struct pci_dev *dev = eeh_dev_to_pci_dev(edev);
+ struct pci_driver *driver;
+
+ if (!dev) return NULL;
dev->error_state = pci_channel_io_perm_failure;
- if (!driver)
- return 0;
+ if (!(driver = dev->driver))
+ return NULL;
eeh_disable_irq(dev);
if (!driver->err_handler ||
!driver->err_handler->error_detected)
- return 0;
+ return NULL;
driver->err_handler->error_detected(dev, pci_channel_io_perm_failure);
- return 0;
+ return NULL;
}
/**
* eeh_reset_device - Perform actual reset of a pci slot
- * @edev: PE associated EEH device
+ * @pe: EEH PE
* @bus: PCI bus corresponding to the isolcated slot
*
* This routine must be called to do reset on the indicated PE.
* During the reset, udev might be invoked because those affected
* PCI devices will be removed and then added.
*/
-static int eeh_reset_device(struct eeh_dev *edev, struct pci_bus *bus)
+static int eeh_reset_device(struct eeh_pe *pe, struct pci_bus *bus)
{
- struct device_node *dn;
int cnt, rc;
/* pcibios will clear the counter; save the value */
- cnt = edev->freeze_count;
+ cnt = pe->freeze_count;
if (bus)
pcibios_remove_pci_devices(bus);
@@ -292,25 +312,13 @@ static int eeh_reset_device(struct eeh_dev *edev, struct pci_bus *bus)
* Reconfigure bridges and devices. Don't try to bring the system
* up if the reset failed for some reason.
*/
- rc = eeh_reset_pe(edev);
+ rc = eeh_reset_pe(pe);
if (rc)
return rc;
- /* Walk over all functions on this device. */
- dn = eeh_dev_to_of_node(edev);
- if (!pcibios_find_pci_bus(dn) && of_node_to_eeh_dev(dn->parent))
- dn = dn->parent->child;
-
- while (dn) {
- struct eeh_dev *pedev = of_node_to_eeh_dev(dn);
-
- /* On Power4, always true because eeh_pe_config_addr=0 */
- if (edev->pe_config_addr == pedev->pe_config_addr) {
- eeh_ops->configure_bridge(dn);
- eeh_restore_bars(pedev);
- }
- dn = dn->sibling;
- }
+ /* Restore PE */
+ eeh_ops->configure_bridge(pe);
+ eeh_pe_restore_bars(pe);
/* Give the system 5 seconds to finish running the user-space
* hotplug shutdown scripts, e.g. ifdown for ethernet. Yes,
@@ -322,7 +330,7 @@ static int eeh_reset_device(struct eeh_dev *edev, struct pci_bus *bus)
ssleep(5);
pcibios_add_pci_devices(bus);
}
- edev->freeze_count = cnt;
+ pe->freeze_count = cnt;
return 0;
}
@@ -334,7 +342,7 @@ static int eeh_reset_device(struct eeh_dev *edev, struct pci_bus *bus)
/**
* eeh_handle_event - Reset a PCI device after hard lockup.
- * @event: EEH event
+ * @pe: EEH PE
*
* While PHB detects address or data parity errors on particular PCI
* slot, the associated PE will be frozen. Besides, DMA's occurring
@@ -349,69 +357,24 @@ static int eeh_reset_device(struct eeh_dev *edev, struct pci_bus *bus)
* drivers (which cause a second set of hotplug events to go out to
* userspace).
*/
-struct eeh_dev *handle_eeh_events(struct eeh_event *event)
+void eeh_handle_event(struct eeh_pe *pe)
{
- struct device_node *frozen_dn;
- struct eeh_dev *frozen_edev;
struct pci_bus *frozen_bus;
int rc = 0;
enum pci_ers_result result = PCI_ERS_RESULT_NONE;
- const char *location, *pci_str, *drv_str, *bus_pci_str, *bus_drv_str;
-
- frozen_dn = eeh_find_device_pe(eeh_dev_to_of_node(event->edev));
- if (!frozen_dn) {
- location = of_get_property(eeh_dev_to_of_node(event->edev), "ibm,loc-code", NULL);
- location = location ? location : "unknown";
- printk(KERN_ERR "EEH: Error: Cannot find partition endpoint "
- "for location=%s pci addr=%s\n",
- location, eeh_pci_name(eeh_dev_to_pci_dev(event->edev)));
- return NULL;
- }
-
- frozen_bus = pcibios_find_pci_bus(frozen_dn);
- location = of_get_property(frozen_dn, "ibm,loc-code", NULL);
- location = location ? location : "unknown";
-
- /* There are two different styles for coming up with the PE.
- * In the old style, it was the highest EEH-capable device
- * which was always an EADS pci bridge. In the new style,
- * there might not be any EADS bridges, and even when there are,
- * the firmware marks them as "EEH incapable". So another
- * two-step is needed to find the pci bus..
- */
- if (!frozen_bus)
- frozen_bus = pcibios_find_pci_bus(frozen_dn->parent);
+ frozen_bus = eeh_pe_bus_get(pe);
if (!frozen_bus) {
- printk(KERN_ERR "EEH: Cannot find PCI bus "
- "for location=%s dn=%s\n",
- location, frozen_dn->full_name);
- return NULL;
+ pr_err("%s: Cannot find PCI bus for PHB#%d-PE#%x\n",
+ __func__, pe->phb->global_number, pe->addr);
+ return;
}
- frozen_edev = of_node_to_eeh_dev(frozen_dn);
- frozen_edev->freeze_count++;
- pci_str = eeh_pci_name(eeh_dev_to_pci_dev(event->edev));
- drv_str = eeh_pcid_name(eeh_dev_to_pci_dev(event->edev));
-
- if (frozen_edev->freeze_count > EEH_MAX_ALLOWED_FREEZES)
+ pe->freeze_count++;
+ if (pe->freeze_count > EEH_MAX_ALLOWED_FREEZES)
goto excess_failures;
-
- printk(KERN_WARNING
- "EEH: This PCI device has failed %d times in the last hour:\n",
- frozen_edev->freeze_count);
-
- if (frozen_edev->pdev) {
- bus_pci_str = pci_name(frozen_edev->pdev);
- bus_drv_str = eeh_pcid_name(frozen_edev->pdev);
- printk(KERN_WARNING
- "EEH: Bus location=%s driver=%s pci addr=%s\n",
- location, bus_drv_str, bus_pci_str);
- }
-
- printk(KERN_WARNING
- "EEH: Device location=%s driver=%s pci addr=%s\n",
- location, drv_str, pci_str);
+ pr_warning("EEH: This PCI device has failed %d times in the last hour\n",
+ pe->freeze_count);
/* Walk the various device drivers attached to this slot through
* a reset sequence, giving each an opportunity to do what it needs
@@ -419,12 +382,12 @@ struct eeh_dev *handle_eeh_events(struct eeh_event *event)
* status ... if any child can't handle the reset, then the entire
* slot is dlpar removed and added.
*/
- pci_walk_bus(frozen_bus, eeh_report_error, &result);
+ eeh_pe_dev_traverse(pe, eeh_report_error, &result);
/* Get the current PCI slot state. This can take a long time,
* sometimes over 3 seconds for certain systems.
*/
- rc = eeh_ops->wait_state(eeh_dev_to_of_node(frozen_edev), MAX_WAIT_FOR_RECOVERY*1000);
+ rc = eeh_ops->wait_state(pe, MAX_WAIT_FOR_RECOVERY*1000);
if (rc < 0 || rc == EEH_STATE_NOT_SUPPORT) {
printk(KERN_WARNING "EEH: Permanent failure\n");
goto hard_fail;
@@ -434,14 +397,14 @@ struct eeh_dev *handle_eeh_events(struct eeh_event *event)
* don't post the error log until after all dev drivers
* have been informed.
*/
- eeh_slot_error_detail(frozen_edev, EEH_LOG_TEMP);
+ eeh_slot_error_detail(pe, EEH_LOG_TEMP);
/* If all device drivers were EEH-unaware, then shut
* down all of the device drivers, and hope they
* go down willingly, without panicing the system.
*/
if (result == PCI_ERS_RESULT_NONE) {
- rc = eeh_reset_device(frozen_edev, frozen_bus);
+ rc = eeh_reset_device(pe, frozen_bus);
if (rc) {
printk(KERN_WARNING "EEH: Unable to reset, rc=%d\n", rc);
goto hard_fail;
@@ -450,7 +413,7 @@ struct eeh_dev *handle_eeh_events(struct eeh_event *event)
/* If all devices reported they can proceed, then re-enable MMIO */
if (result == PCI_ERS_RESULT_CAN_RECOVER) {
- rc = eeh_pci_enable(frozen_edev, EEH_OPT_THAW_MMIO);
+ rc = eeh_pci_enable(pe, EEH_OPT_THAW_MMIO);
if (rc < 0)
goto hard_fail;
@@ -458,13 +421,13 @@ struct eeh_dev *handle_eeh_events(struct eeh_event *event)
result = PCI_ERS_RESULT_NEED_RESET;
} else {
result = PCI_ERS_RESULT_NONE;
- pci_walk_bus(frozen_bus, eeh_report_mmio_enabled, &result);
+ eeh_pe_dev_traverse(pe, eeh_report_mmio_enabled, &result);
}
}
/* If all devices reported they can proceed, then re-enable DMA */
if (result == PCI_ERS_RESULT_CAN_RECOVER) {
- rc = eeh_pci_enable(frozen_edev, EEH_OPT_THAW_DMA);
+ rc = eeh_pci_enable(pe, EEH_OPT_THAW_DMA);
if (rc < 0)
goto hard_fail;
@@ -482,13 +445,13 @@ struct eeh_dev *handle_eeh_events(struct eeh_event *event)
/* If any device called out for a reset, then reset the slot */
if (result == PCI_ERS_RESULT_NEED_RESET) {
- rc = eeh_reset_device(frozen_edev, NULL);
+ rc = eeh_reset_device(pe, NULL);
if (rc) {
printk(KERN_WARNING "EEH: Cannot reset, rc=%d\n", rc);
goto hard_fail;
}
result = PCI_ERS_RESULT_NONE;
- pci_walk_bus(frozen_bus, eeh_report_reset, &result);
+ eeh_pe_dev_traverse(pe, eeh_report_reset, &result);
}
/* All devices should claim they have recovered by now. */
@@ -499,9 +462,9 @@ struct eeh_dev *handle_eeh_events(struct eeh_event *event)
}
/* Tell all device drivers that they can resume operations */
- pci_walk_bus(frozen_bus, eeh_report_resume, NULL);
+ eeh_pe_dev_traverse(pe, eeh_report_resume, NULL);
- return frozen_edev;
+ return;
excess_failures:
/*
@@ -509,30 +472,26 @@ excess_failures:
* are due to poorly seated PCI cards. Only 10% or so are
* due to actual, failed cards.
*/
- printk(KERN_ERR
- "EEH: PCI device at location=%s driver=%s pci addr=%s\n"
- "has failed %d times in the last hour "
- "and has been permanently disabled.\n"
- "Please try reseating this device or replacing it.\n",
- location, drv_str, pci_str, frozen_edev->freeze_count);
+ pr_err("EEH: PHB#%d-PE#%x has failed %d times in the\n"
+ "last hour and has been permanently disabled.\n"
+ "Please try reseating or replacing it.\n",
+ pe->phb->global_number, pe->addr,
+ pe->freeze_count);
goto perm_error;
hard_fail:
- printk(KERN_ERR
- "EEH: Unable to recover from failure of PCI device "
- "at location=%s driver=%s pci addr=%s\n"
- "Please try reseating this device or replacing it.\n",
- location, drv_str, pci_str);
+ pr_err("EEH: Unable to recover from failure from PHB#%d-PE#%x.\n"
+ "Please try reseating or replacing it\n",
+ pe->phb->global_number, pe->addr);
perm_error:
- eeh_slot_error_detail(frozen_edev, EEH_LOG_PERM);
+ eeh_slot_error_detail(pe, EEH_LOG_PERM);
/* Notify all devices that they're about to go down. */
- pci_walk_bus(frozen_bus, eeh_report_failure, NULL);
+ eeh_pe_dev_traverse(pe, eeh_report_failure, NULL);
/* Shut down the device drivers for good. */
- pcibios_remove_pci_devices(frozen_bus);
-
- return NULL;
+ if (frozen_bus)
+ pcibios_remove_pci_devices(frozen_bus);
}
diff --git a/arch/powerpc/platforms/pseries/eeh_event.c b/arch/powerpc/platforms/pseries/eeh_event.c
index 67c82c2..7d48feb 100644
--- a/arch/powerpc/platforms/pseries/eeh_event.c
+++ b/arch/powerpc/platforms/pseries/eeh_event.c
@@ -81,7 +81,7 @@ static int eeh_event_handler(void * dummy)
pr_info("EEH: Detected PCI bus error on PHB#%d-PE#%x\n",
pe->phb->global_number, pe->addr);
- handle_eeh_events(event);
+ eeh_handle_event(pe);
eeh_pe_state_clear(pe, EEH_PE_RECOVERING);
diff --git a/arch/powerpc/platforms/pseries/eeh_pe.c b/arch/powerpc/platforms/pseries/eeh_pe.c
index 8bae0f6..d70a7e4 100644
--- a/arch/powerpc/platforms/pseries/eeh_pe.c
+++ b/arch/powerpc/platforms/pseries/eeh_pe.c
@@ -554,3 +554,30 @@ void eeh_pe_restore_bars(struct eeh_pe *pe)
eeh_pe_dev_traverse(pe, eeh_restore_one_device_bars, NULL);
}
+/**
+ * eeh_pe_bus_get - Retrieve PCI bus according to the given PE
+ * @pe: EEH PE
+ *
+ * Retrieve the PCI bus according to the given PE. Basically,
+ * there're 3 types of PEs: PHB/Bus/Device. For PHB PE, the
+ * primary PCI bus will be retrieved. The parent bus will be
+ * returned for BUS PE. However, we don't have associated PCI
+ * bus for DEVICE PE.
+ */
+struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe)
+{
+ struct pci_bus *bus = NULL;
+ struct eeh_dev *edev;
+ struct pci_dev *pdev;
+
+ if (pe->type == EEH_PE_PHB) {
+ bus = pe->phb->bus;
+ } else if (pe->type == EEH_PE_BUS) {
+ edev = list_first_entry(&pe->edevs, struct eeh_dev, list);
+ pdev = eeh_dev_to_pci_dev(edev);
+ if (pdev)
+ bus = pdev->bus;
+ }
+
+ return bus;
+}
--
1.7.9.5
More information about the Linuxppc-dev
mailing list