static int eeh_reset_device (struct pci_dn *pe_dn, struct pci_bus *bus)
{
int cnt, rc;
/* pcibios will clear the counter; save the value */
cnt = pe_dn->eeh_freeze_count;
if (bus)
pcibios_remove_pci_devices(bus);
/* Reset the pci controller. (Asserts RST#; resets config space).
* Reconfigure bridges and devices. Don't try to bring the system
* up if the reset failed for some reason. */
rc = rtas_set_slot_reset(pe_dn);
if (rc)
return rc;
/* New-style config addrs might be shared across multiple devices,
* Walk over all functions on this device */
if (pe_dn->eeh_pe_config_addr) {
struct device_node *pe = pe_dn->node;
pe = pe->parent->child;
while (pe) {
struct pci_dn *ppe = PCI_DN(pe);
if (pe_dn->eeh_pe_config_addr == ppe->eeh_pe_config_addr) {
rtas_configure_bridge(ppe);
eeh_restore_bars(ppe);
}
pe = pe->sibling;
}
} else {
rtas_configure_bridge(pe_dn);
eeh_restore_bars(pe_dn);
}
/* Give the system 5 seconds to finish running the user-space
* hotplug shutdown scripts, e.g. ifdown for ethernet. Yes,
* this is a hack, but if we don't do this, and try to bring
* the device up before the scripts have taken it down,
* potentially weird things happen.
*/
if (bus) {
ssleep (5);
pcibios_add_pci_devices(bus);
}
pe_dn->eeh_freeze_count = cnt;
return 0;
}
/**
* eeh_reset_device - Perform actual reset of a pci slot
* @edev: PE associated EEH device
* @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)
{
struct device_node *dn;
int cnt, rc;
/* pcibios will clear the counter; save the value */
cnt = edev->freeze_count;
if (bus)
pcibios_remove_pci_devices(bus);
/* Reset the pci controller. (Asserts RST#; resets config space).
* Reconfigure bridges and devices. Don't try to bring the system
* up if the reset failed for some reason.
*/
rc = eeh_reset_pe(edev);
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;
}
/* Give the system 5 seconds to finish running the user-space
* hotplug shutdown scripts, e.g. ifdown for ethernet. Yes,
* this is a hack, but if we don't do this, and try to bring
* the device up before the scripts have taken it down,
* potentially weird things happen.
*/
if (bus) {
ssleep(5);
pcibios_add_pci_devices(bus);
}
edev->freeze_count = cnt;
return 0;
}
static int eeh_reset_device(struct eeh_dev *edev, struct pci_bus *bus)
{
struct device_node *dn;
int cnt, rc;
cnt = edev->freeze_count;
if (bus)
pcibios_remove_pci_devices(bus);
rc = eeh_reset_pe(edev);
if (rc)
return rc;
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);
if (edev->pe_config_addr == pedev->pe_config_addr) {
eeh_ops->configure_bridge(dn);
eeh_restore_bars(pedev);
}
dn = dn->sibling;
}
if (bus) {
ssleep(5);
pcibios_add_pci_devices(bus);
}
edev->freeze_count = cnt;
return 0;
}
struct pci_dn * handle_eeh_events (struct eeh_event *event)
{
struct device_node *frozen_dn;
struct pci_dn *frozen_pdn;
struct pci_bus *frozen_bus;
int rc = 0;
enum pci_ers_result result = PCI_ERS_RESULT_NONE;
const char *location, *pci_str, *drv_str;
frozen_dn = find_device_pe(event->dn);
if (!frozen_dn) {
location = of_get_property(event->dn, "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, pci_name(event->dev));
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);
if (!frozen_bus) {
printk(KERN_ERR "EEH: Cannot find PCI bus "
"for location=%s dn=%s\n",
location, frozen_dn->full_name);
return NULL;
}
frozen_pdn = PCI_DN(frozen_dn);
frozen_pdn->eeh_freeze_count++;
if (frozen_pdn->pcidev) {
pci_str = pci_name (frozen_pdn->pcidev);
drv_str = pcid_name (frozen_pdn->pcidev);
} else {
pci_str = pci_name (event->dev);
drv_str = pcid_name (event->dev);
}
if (frozen_pdn->eeh_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_pdn->eeh_freeze_count);
printk(KERN_WARNING
"EEH: location=%s driver=%s pci addr=%s\n",
location, drv_str, pci_str);
/* Walk the various device drivers attached to this slot through
* a reset sequence, giving each an opportunity to do what it needs
* to accomplish the reset. Each child gets a report of the
* 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);
/* Get the current PCI slot state. This can take a long time,
* sometimes over 3 seconds for certain systems. */
rc = eeh_wait_for_slot_status (frozen_pdn, MAX_WAIT_FOR_RECOVERY*1000);
if (rc < 0) {
printk(KERN_WARNING "EEH: Permanent failure\n");
goto hard_fail;
}
/* Since rtas may enable MMIO when posting the error log,
* don't post the error log until after all dev drivers
* have been informed.
*/
eeh_slot_error_detail(frozen_pdn, EEH_LOG_TEMP_FAILURE);
/* 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_pdn, frozen_bus);
if (rc) {
printk(KERN_WARNING "EEH: Unable to reset, rc=%d\n", rc);
goto hard_fail;
}
}
/* If all devices reported they can proceed, then re-enable MMIO */
if (result == PCI_ERS_RESULT_CAN_RECOVER) {
rc = rtas_pci_enable(frozen_pdn, EEH_THAW_MMIO);
if (rc < 0)
goto hard_fail;
if (rc) {
result = PCI_ERS_RESULT_NEED_RESET;
} else {
result = PCI_ERS_RESULT_NONE;
pci_walk_bus(frozen_bus, eeh_report_mmio_enabled, &result);
}
}
/* If all devices reported they can proceed, then re-enable DMA */
if (result == PCI_ERS_RESULT_CAN_RECOVER) {
rc = rtas_pci_enable(frozen_pdn, EEH_THAW_DMA);
if (rc < 0)
goto hard_fail;
if (rc)
result = PCI_ERS_RESULT_NEED_RESET;
else
result = PCI_ERS_RESULT_RECOVERED;
}
/* If any device has a hard failure, then shut off everything. */
if (result == PCI_ERS_RESULT_DISCONNECT) {
printk(KERN_WARNING "EEH: Device driver gave up\n");
goto hard_fail;
}
/* If any device called out for a reset, then reset the slot */
if (result == PCI_ERS_RESULT_NEED_RESET) {
rc = eeh_reset_device(frozen_pdn, 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);
}
/* All devices should claim they have recovered by now. */
if ((result != PCI_ERS_RESULT_RECOVERED) &&
(result != PCI_ERS_RESULT_NONE)) {
printk(KERN_WARNING "EEH: Not recovered\n");
goto hard_fail;
}
/* Tell all device drivers that they can resume operations */
pci_walk_bus(frozen_bus, eeh_report_resume, NULL);
return frozen_pdn;
excess_failures:
/*
* About 90% of all real-life EEH failures in the field
* 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_pdn->eeh_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);
perm_error:
eeh_slot_error_detail(frozen_pdn, EEH_LOG_PERM_FAILURE);
/* Notify all devices that they're about to go down. */
pci_walk_bus(frozen_bus, eeh_report_failure, NULL);
/* Shut down the device drivers for good. */
pcibios_remove_pci_devices(frozen_bus);
return NULL;
}
/**
* eeh_reset_device - Perform actual reset of a pci slot
* @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_pe *pe, struct pci_bus *bus)
{
struct pci_bus *frozen_bus = eeh_pe_bus_get(pe);
struct timeval tstamp;
int cnt, rc, removed = 0;
/* pcibios will clear the counter; save the value */
cnt = pe->freeze_count;
tstamp = pe->tstamp;
/*
* We don't remove the corresponding PE instances because
* we need the information afterwords. The attached EEH
* devices are expected to be attached soon when calling
* into pcibios_add_pci_devices().
*/
eeh_pe_state_mark(pe, EEH_PE_KEEP);
if (bus) {
pci_lock_rescan_remove();
pcibios_remove_pci_devices(bus);
pci_unlock_rescan_remove();
} else if (frozen_bus) {
eeh_pe_dev_traverse(pe, eeh_rmv_device, &removed);
}
/*
* Reset the pci controller. (Asserts RST#; resets config space).
* Reconfigure bridges and devices. Don't try to bring the system
* up if the reset failed for some reason.
*
* During the reset, it's very dangerous to have uncontrolled PCI
* config accesses. So we prefer to block them. However, controlled
* PCI config accesses initiated from EEH itself are allowed.
*/
eeh_pe_state_mark(pe, EEH_PE_RESET);
rc = eeh_reset_pe(pe);
if (rc) {
eeh_pe_state_clear(pe, EEH_PE_RESET);
return rc;
}
pci_lock_rescan_remove();
/* Restore PE */
eeh_ops->configure_bridge(pe);
eeh_pe_restore_bars(pe);
eeh_pe_state_clear(pe, EEH_PE_RESET);
/* Clear frozen state */
rc = eeh_clear_pe_frozen_state(pe);
if (rc)
return rc;
/* Give the system 5 seconds to finish running the user-space
* hotplug shutdown scripts, e.g. ifdown for ethernet. Yes,
* this is a hack, but if we don't do this, and try to bring
* the device up before the scripts have taken it down,
* potentially weird things happen.
*/
if (bus) {
pr_info("EEH: Sleep 5s ahead of complete hotplug\n");
ssleep(5);
/*
* The EEH device is still connected with its parent
* PE. We should disconnect it so the binding can be
* rebuilt when adding PCI devices.
*/
eeh_pe_traverse(pe, eeh_pe_detach_dev, NULL);
pcibios_add_pci_devices(bus);
} else if (frozen_bus && removed) {
pr_info("EEH: Sleep 5s ahead of partial hotplug\n");
ssleep(5);
eeh_pe_traverse(pe, eeh_pe_detach_dev, NULL);
pcibios_add_pci_devices(frozen_bus);
}
eeh_pe_state_clear(pe, EEH_PE_KEEP);
pe->tstamp = tstamp;
pe->freeze_count = cnt;
pci_unlock_rescan_remove();
return 0;
}
struct pci_dn * handle_eeh_events (struct eeh_event *event)
{
struct device_node *frozen_dn;
struct pci_dn *frozen_pdn;
struct pci_bus *frozen_bus;
int rc = 0;
enum pci_ers_result result = PCI_ERS_RESULT_NONE;
const char *location, *pci_str, *drv_str;
frozen_dn = find_device_pe(event->dn);
frozen_bus = pcibios_find_pci_bus(frozen_dn);
if (!frozen_dn) {
location = (char *) get_property(event->dn, "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, pci_name(event->dev));
return NULL;
}
location = (char *) 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);
if (!frozen_bus) {
printk(KERN_ERR "EEH: Cannot find PCI bus "
"for location=%s dn=%s\n",
location, frozen_dn->full_name);
return NULL;
}
#if 0
/* We may get "permanent failure" messages on empty slots.
* These are false alarms. Empty slots have no child dn. */
if ((event->state == pci_channel_io_perm_failure) && (frozen_device == NULL))
return;
#endif
frozen_pdn = PCI_DN(frozen_dn);
frozen_pdn->eeh_freeze_count++;
if (frozen_pdn->pcidev) {
pci_str = pci_name (frozen_pdn->pcidev);
drv_str = pcid_name (frozen_pdn->pcidev);
} else {
pci_str = pci_name (event->dev);
drv_str = pcid_name (event->dev);
}
if (frozen_pdn->eeh_freeze_count > EEH_MAX_ALLOWED_FREEZES)
goto excess_failures;
/* If the reset state is a '5' and the time to reset is 0 (infinity)
* or is more then 15 seconds, then mark this as a permanent failure.
*/
if ((event->state == pci_channel_io_perm_failure) &&
((event->time_unavail <= 0) ||
(event->time_unavail > MAX_WAIT_FOR_RECOVERY*1000)))
goto hard_fail;
eeh_slot_error_detail(frozen_pdn, 1 /* Temporary Error */);
printk(KERN_WARNING
"EEH: This PCI device has failed %d times since last reboot: "
"location=%s driver=%s pci addr=%s\n",
frozen_pdn->eeh_freeze_count, location, drv_str, pci_str);
/* Walk the various device drivers attached to this slot through
* a reset sequence, giving each an opportunity to do what it needs
* to accomplish the reset. Each child gets a report of the
* 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);
/* 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_pdn, frozen_bus);
if (rc)
goto hard_fail;
}
/* If any device called out for a reset, then reset the slot */
if (result == PCI_ERS_RESULT_NEED_RESET) {
rc = eeh_reset_device(frozen_pdn, NULL);
if (rc)
goto hard_fail;
pci_walk_bus(frozen_bus, eeh_report_reset, NULL);
}
/* If all devices reported they can proceed, the re-enable PIO */
if (result == PCI_ERS_RESULT_CAN_RECOVER) {
/* XXX Not supported; we brute-force reset the device */
rc = eeh_reset_device(frozen_pdn, NULL);
if (rc)
goto hard_fail;
pci_walk_bus(frozen_bus, eeh_report_reset, NULL);
}
/* Tell all device drivers that they can resume operations */
pci_walk_bus(frozen_bus, eeh_report_resume, NULL);
return frozen_pdn;
excess_failures:
/*
* About 90% of all real-life EEH failures in the field
* 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 and has been permanently disabled. \n"
"Please try reseating this device or replacing it.\n",
location, drv_str, pci_str, frozen_pdn->eeh_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);
perm_error:
eeh_slot_error_detail(frozen_pdn, 2 /* Permanent Error */);
/* Notify all devices that they're about to go down. */
pci_walk_bus(frozen_bus, eeh_report_failure, NULL);
/* Shut down the device drivers for good. */
pcibios_remove_pci_devices(frozen_bus);
return NULL;
}
struct eeh_dev *handle_eeh_events(struct eeh_event *event)
{
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";
if (!frozen_bus)
frozen_bus = pcibios_find_pci_bus(frozen_dn->parent);
if (!frozen_bus) {
printk(KERN_ERR "EEH: Cannot find PCI bus "
"for location=%s dn=%s\n",
location, frozen_dn->full_name);
return NULL;
}
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)
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);
pci_walk_bus(frozen_bus, eeh_report_error, &result);
rc = eeh_ops->wait_state(eeh_dev_to_of_node(frozen_edev), MAX_WAIT_FOR_RECOVERY*1000);
if (rc < 0 || rc == EEH_STATE_NOT_SUPPORT) {
printk(KERN_WARNING "EEH: Permanent failure\n");
goto hard_fail;
}
eeh_slot_error_detail(frozen_edev, EEH_LOG_TEMP);
if (result == PCI_ERS_RESULT_NONE) {
rc = eeh_reset_device(frozen_edev, frozen_bus);
if (rc) {
printk(KERN_WARNING "EEH: Unable to reset, rc=%d\n", rc);
goto hard_fail;
}
}
if (result == PCI_ERS_RESULT_CAN_RECOVER) {
rc = eeh_pci_enable(frozen_edev, EEH_OPT_THAW_MMIO);
if (rc < 0)
goto hard_fail;
if (rc) {
result = PCI_ERS_RESULT_NEED_RESET;
} else {
result = PCI_ERS_RESULT_NONE;
pci_walk_bus(frozen_bus, eeh_report_mmio_enabled, &result);
}
}
if (result == PCI_ERS_RESULT_CAN_RECOVER) {
rc = eeh_pci_enable(frozen_edev, EEH_OPT_THAW_DMA);
if (rc < 0)
goto hard_fail;
if (rc)
result = PCI_ERS_RESULT_NEED_RESET;
else
result = PCI_ERS_RESULT_RECOVERED;
}
if (result == PCI_ERS_RESULT_DISCONNECT) {
printk(KERN_WARNING "EEH: Device driver gave up\n");
goto hard_fail;
}
if (result == PCI_ERS_RESULT_NEED_RESET) {
rc = eeh_reset_device(frozen_edev, 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);
}
if ((result != PCI_ERS_RESULT_RECOVERED) &&
(result != PCI_ERS_RESULT_NONE)) {
printk(KERN_WARNING "EEH: Not recovered\n");
goto hard_fail;
}
pci_walk_bus(frozen_bus, eeh_report_resume, NULL);
return frozen_edev;
excess_failures:
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);
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);
perm_error:
eeh_slot_error_detail(frozen_edev, EEH_LOG_PERM);
pci_walk_bus(frozen_bus, eeh_report_failure, NULL);
pcibios_remove_pci_devices(frozen_bus);
return NULL;
}