PCI/ERR: Clear PCIe Device Status errors only if OS owns AER

pcie_clear_device_status() resets the error bits in the PCIe Device Status
Register (PCI_EXP_DEVSTA).

Previously we did this unconditionally, but on ACPI systems, the _OSC AER
bit negotiates control of the AER capability.  Per sec 4.5.1 of the System
Firmware Intermediary _OSC and DPC Updates ECN [1], this bit also covers
other error enable/status bits including the following:

  Correctable Error Reporting Enable
  Non-Fatal Error Reporting Enable
  Fatal Error Reporting Enable
  Unsupported Request Reporting Enable

These bits are all in the PCIe Device Control register (the ECN omitted
"Reporting", but I think that's a typo), so by implication the _OSC AER bit
also applies to the error status bits in the PCIe Device Status register:

  Correctable Error Detected
  Non-Fatal Error Detected
  Fatal Error Detected
  Unsupported Request Detected

Clear the PCIe Device Status error bits only when the OS controls the AER
capability and related error enable/status bits.  If platform firmware
controls the AER capability, firmware is responsible for clearing these
bits.

One call path leading here is:

  ghes_do_proc
    ghes_handle_aer
      aer_recover_queue
        schedule_work(&aer_recover_work)
  ...
  aer_recover_work_func
    pcie_do_recovery
      pcie_clear_device_status

[1] System Firmware Intermediary (SFI) _OSC and DPC Updates ECN, Feb 24,
    2020, affecting PCI Firmware Specification, Rev. 3.2
    https://members.pcisig.com/wg/PCI-SIG/document/14076
[bhelgaas: commit log, move test from pcie_clear_device_status() to callers]
Link: https://lore.kernel.org/r/20200622113523.891666-1-Jonathan.Cameron@huawei.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>

diff --git a/drivers/pci/pcie/aer.c b/drivers/pci/pcie/aer.c
index f6d7783..cccd674 100644 (file)
--- a/drivers/pci/pcie/aer.c
+++ b/drivers/pci/pcie/aer.c
@@ -939,7 +939,8 @@ static void handle_error_source(struct pci_dev *dev, struct aer_err_info *info)
                if (aer)
                        pci_write_config_dword(dev, aer + PCI_ERR_COR_STATUS,
                                        info->status);
-               pcie_clear_device_status(dev);
+               if (pcie_aer_is_native(dev))
+                       pcie_clear_device_status(dev);
        } else if (info->severity == AER_NONFATAL)
                pcie_do_recovery(dev, pci_channel_io_normal, aer_root_reset);
        else if (info->severity == AER_FATAL)

diff --git a/drivers/pci/pcie/err.c b/drivers/pci/pcie/err.c
index 55755bc..c543f41 100644 (file)
--- a/drivers/pci/pcie/err.c
+++ b/drivers/pci/pcie/err.c
@@ -197,7 +197,8 @@ pci_ers_result_t pcie_do_recovery(struct pci_dev *dev,
        pci_dbg(dev, "broadcast resume message\n");
        pci_walk_bus(bus, report_resume, &status);
 
-       pcie_clear_device_status(dev);
+       if (pcie_aer_is_native(dev))
+               pcie_clear_device_status(dev);
        pci_aer_clear_nonfatal_status(dev);
        pci_info(dev, "device recovery successful\n");
        return status;