PCI: keystone: Use PCI_ERROR_RESPONSE to identify config read errors

Include PCI_ERROR_RESPONSE along with 0xffffffff in the comment about
identifying config read errors. This makes checks for config read errors
easier to find. Comment change only.

Link: https://lore.kernel.org/r/6ae6b071d92052dc511407513e2a7c0035aff9e7.1637243717.git.naveennaidu479@gmail.com
Signed-off-by: Naveen Naidu <naveennaidu479@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>

drivers/pci/controller/dwc/pci-keystone.c

@@ -747,9 +747,9 @@ static int ks_pcie_config_legacy_irq(struct keystone_pcie *ks_pcie)
 
 #ifdef CONFIG_ARM
 /*
- * When a PCI device does not exist during config cycles, keystone host gets a
- * bus error instead of returning 0xffffffff. This handler always returns 0
- * for this kind of faults.
+ * When a PCI device does not exist during config cycles, keystone host
+ * gets a bus error instead of returning 0xffffffff (PCI_ERROR_RESPONSE).
+ * This handler always returns 0 for this kind of fault.
  */
 static int ks_pcie_fault(unsigned long addr, unsigned int fsr,
 			 struct pt_regs *regs)