[Qemu-devel] [PATCH 1/4] AER-PCI: Add infrastructure for notification of errors to other subsystems

[Pandarathil, Vijaymohan R]

- Adds infrastructure support for error notifications from AER subsystem
to other subsystems. The generic notifier_chain functionality is used.

- When the AER rootport driver detects an uncorrected error, invoke the
callbacks registered for notifications as well as mark the device as
tainted.

- After the recovery is successful, clear the tainted flag on the device.

Signed-off-by: Vijay Mohan Pandarathil <address@hidden>

---
 drivers/pci/pcie/aer/aerdrv.c      | 20 ++++++++++++++++++++
 drivers/pci/pcie/aer/aerdrv_core.c |  9 ++++++++-
 include/linux/aer.h                |  4 ++++
 include/linux/pci.h                |  2 ++
 4 files changed, 34 insertions(+), 1 deletion(-)

diff --git a/drivers/pci/pcie/aer/aerdrv.c b/drivers/pci/pcie/aer/aerdrv.c
index 030cf12..92dc54c 100644
--- a/drivers/pci/pcie/aer/aerdrv.c
+++ b/drivers/pci/pcie/aer/aerdrv.c
@@ -68,6 +68,26 @@ static struct pcie_port_service_driver aerdriver = {
 
 static int pcie_aer_disable;
 
+ATOMIC_NOTIFIER_HEAD(aer_notifier_list);
+
+void aer_notifier_register(struct notifier_block *nb)
+{
+       atomic_notifier_chain_register(&aer_notifier_list, nb);
+}
+EXPORT_SYMBOL_GPL(aer_notifier_register);
+
+void aer_notifier_unregister(struct notifier_block *nb)
+{
+       atomic_notifier_chain_unregister(&aer_notifier_list, nb);
+}
+EXPORT_SYMBOL_GPL(aer_notifier_unregister);
+
+void aer_notify(unsigned long val, void *v)
+{
+       atomic_notifier_call_chain(&aer_notifier_list, val, v);
+}
+EXPORT_SYMBOL_GPL(aer_notify);
+
 void pci_no_aer(void)
 {
        pcie_aer_disable = 1;   /* has priority over 'forceload' */
diff --git a/drivers/pci/pcie/aer/aerdrv_core.c 
b/drivers/pci/pcie/aer/aerdrv_core.c
index af4e31c..be6c3ee 100644
--- a/drivers/pci/pcie/aer/aerdrv_core.c
+++ b/drivers/pci/pcie/aer/aerdrv_core.c
@@ -215,6 +215,8 @@ static int report_error_detected(struct pci_dev *dev, void 
*data)
 
        device_lock(&dev->dev);
        dev->error_state = result_data->state;
+       dev->dev_flags |= PCI_DEV_FLAGS_ERR_DETECTED;
+       aer_notify(0, NULL);
 
        if (!dev->driver ||
                !dev->driver->err_handler ||
@@ -291,6 +293,7 @@ static int report_resume(struct pci_dev *dev, void *data)
 
        device_lock(&dev->dev);
        dev->error_state = pci_channel_io_normal;
+       dev->dev_flags &= ~PCI_DEV_FLAGS_ERR_DETECTED;
 
        if (!dev->driver ||
                !dev->driver->err_handler ||
@@ -521,6 +524,7 @@ static void do_recovery(struct pci_dev *dev, int severity)
                                "resume",
                                report_resume);
 
+       dev->dev_flags &= ~PCI_DEV_FLAGS_ERR_DETECTED;
        dev_info(&dev->dev, "AER: Device recovery successful\n");
        return;
 
@@ -552,8 +556,11 @@ static void handle_error_source(struct pcie_device *aerdev,
                if (pos)
                        pci_write_config_dword(dev, pos + PCI_ERR_COR_STATUS,
                                        info->status);
-       } else
+       } else {
+               dev->dev_flags |= PCI_DEV_FLAGS_ERR_DETECTED;
+               aer_notify(0, NULL);
                do_recovery(dev, info->severity);
+       }
 }
 
 #ifdef CONFIG_ACPI_APEI_PCIEAER
diff --git a/include/linux/aer.h b/include/linux/aer.h
index 544abdb..f8df468 100644
--- a/include/linux/aer.h
+++ b/include/linux/aer.h
@@ -54,5 +54,9 @@ extern void cper_print_aer(const char *prefix, int 
cper_severity,
 extern int cper_severity_to_aer(int cper_severity);
 extern void aer_recover_queue(int domain, unsigned int bus, unsigned int devfn,
                              int severity);
+extern void aer_notifier_register(struct notifier_block *nb);
+extern void aer_notifier_unregister(struct notifier_block *nb);
+extern void aer_notify(unsigned long val, void *v);
+
 #endif //_AER_H_
 
diff --git a/include/linux/pci.h b/include/linux/pci.h
index ee21795..ab17a08 100644
--- a/include/linux/pci.h
+++ b/include/linux/pci.h
@@ -155,6 +155,8 @@ enum pci_dev_flags {
        PCI_DEV_FLAGS_NO_D3 = (__force pci_dev_flags_t) 2,
        /* Provide indication device is assigned by a Virtual Machine Manager */
        PCI_DEV_FLAGS_ASSIGNED = (__force pci_dev_flags_t) 4,
+       /* Indicates that hw has reported an uncorrected error for the device */
+       PCI_DEV_FLAGS_ERR_DETECTED = (__force pci_dev_flags_t) 8,
 };
 
 enum pci_irq_reroute_variant {
-- 
1.7.11.3