From: Prarit Bhargava <prarit@redhat.com>
Date: Wed, 16 Dec 2009 18:42:15 -0500
Subject: : PCI AER HEST Firmware First Support
Message-id: <20091216183816.16974.53029.sendpatchset@prarit.bos.redhat.com>
Patchwork-id: 22004
O-Subject: [RHEL5 1/2]: PCI AER HEST Firmware First Support
Bugzilla: 547762

commit 527337e0fdb893fee671c3b534940d4c8f4ab42b
Author: Prarit Bhargava <prarit@redhat.com>
Date:   Wed Dec 16 11:49:39 2009 -0500

    PCI AER HEST Firmware First Support

    Backport of upstream commit 0584396157ad2d008e2cc76b4ed6254151183a25

    Some HP, Fujitsu, and Dell systems have firmware that will handle PCI AER
    errors.  Support detection and parsing of the HEST table in RHEL5.

    Successfully tested by me on HP and Dell systems.

    Resolves BZ 547762.

Signed-off-by: Jarod Wilson <jarod@redhat.com>

diff --git a/drivers/acpi/Makefile b/drivers/acpi/Makefile
index 137f59f..aeffa25 100644
--- a/drivers/acpi/Makefile
+++ b/drivers/acpi/Makefile
@@ -65,3 +65,4 @@ obj-$(CONFIG_ACPI_SBS)		+= i2c_ec.o sbs.o
 obj-$(CONFIG_ACPI_POWER_METER)        += power_meter.o
 obj-$(CONFIG_DELL_WMI)		+= dell-wmi.o
 obj-$(CONFIG_THINKPAD_ACPI)	+= thinkpad_acpi.o
+obj-$(CONFIG_PCIEAER)		+= hest.o
diff --git a/drivers/acpi/hest.c b/drivers/acpi/hest.c
new file mode 100644
index 0000000..e3dfee7
--- /dev/null
+++ b/drivers/acpi/hest.c
@@ -0,0 +1,136 @@
+#include <linux/acpi.h>
+#include <linux/pci.h>
+
+#define PREFIX "ACPI: "
+
+static inline unsigned long parse_acpi_hest_ia_machine_check(struct acpi_hest_ia_machine_check *p)
+{
+	return sizeof(*p) +
+		(sizeof(struct acpi_hest_ia_error_bank) * p->num_hardware_banks);
+}
+
+static inline unsigned long parse_acpi_hest_ia_corrected(struct acpi_hest_ia_corrected *p)
+{
+	return sizeof(*p) +
+		(sizeof(struct acpi_hest_ia_error_bank) * p->num_hardware_banks);
+}
+
+static inline unsigned long parse_acpi_hest_ia_nmi(struct acpi_hest_ia_nmi *p)
+{
+	return sizeof(*p);
+}
+
+static inline unsigned long parse_acpi_hest_generic(struct acpi_hest_generic *p)
+{
+	return sizeof(*p);
+}
+
+static inline unsigned int hest_match_pci(struct acpi_hest_aer_common *p, struct pci_dev *pci)
+{
+	return	(0           == pci_domain_nr(pci->bus) &&
+		 p->bus      == pci->bus->number &&
+		 p->device   == PCI_SLOT(pci->devfn) &&
+		 p->function == PCI_FUNC(pci->devfn));
+}
+
+static unsigned long parse_acpi_hest_aer(void *hdr, int type, struct pci_dev *pci, int *firmware_first)
+{
+	struct acpi_hest_aer_common *p = hdr + sizeof(struct acpi_hest_header);
+	unsigned long rc=0;
+	u8 pcie_type = 0;
+	u8 bridge = 0;
+	switch (type) {
+	case ACPI_HEST_TYPE_AER_ROOT_PORT:
+		rc = sizeof(struct acpi_hest_aer_root);
+		pcie_type = PCI_EXP_TYPE_ROOT_PORT;
+		break;
+	case ACPI_HEST_TYPE_AER_ENDPOINT:
+		rc = sizeof(struct acpi_hest_aer);
+		pcie_type = PCI_EXP_TYPE_ENDPOINT;
+		break;
+	case ACPI_HEST_TYPE_AER_BRIDGE:
+		rc = sizeof(struct acpi_hest_aer_bridge);
+		if ((pci->class >> 16) == PCI_BASE_CLASS_BRIDGE)
+			bridge = 1;
+		break;
+	}
+
+	if (p->flags & ACPI_HEST_GLOBAL) {
+		if ((pci->is_pcie && (pci->pcie_type == pcie_type)) || bridge)
+			*firmware_first = !!(p->flags & ACPI_HEST_FIRMWARE_FIRST);
+	}
+	else
+		if (hest_match_pci(p, pci))
+			*firmware_first = !!(p->flags & ACPI_HEST_FIRMWARE_FIRST);
+	return rc;
+}
+
+static int acpi_hest_firmware_first(struct acpi_table_header *stdheader, struct pci_dev *pci)
+{
+	struct acpi_table_hest *hest = (struct acpi_table_hest *)stdheader;
+	void *p = (void *)hest + sizeof(*hest); /* defined by the ACPI 4.0 spec */
+	struct acpi_hest_header *hdr = p;
+
+	int i;
+	int firmware_first = 0;
+	static unsigned char printed_unused = 0;
+	static unsigned char printed_reserved = 0;
+
+	for (i=0, hdr=p; p < (((void *)hest) + hest->header.length) && i < hest->error_source_count; i++) {
+		switch (hdr->type) {
+		case ACPI_HEST_TYPE_IA32_CHECK:
+			p += parse_acpi_hest_ia_machine_check(p);
+			break;
+		case ACPI_HEST_TYPE_IA32_CORRECTED_CHECK:
+			p += parse_acpi_hest_ia_corrected(p);
+			break;
+		case ACPI_HEST_TYPE_IA32_NMI:
+			p += parse_acpi_hest_ia_nmi(p);
+			break;
+		/* These three should never appear */
+		case ACPI_HEST_TYPE_NOT_USED3:
+		case ACPI_HEST_TYPE_NOT_USED4:
+		case ACPI_HEST_TYPE_NOT_USED5:
+			if (!printed_unused) {
+				printk(KERN_DEBUG PREFIX
+				       "HEST Error Source list contains an obsolete type (%d).\n", hdr->type);
+				printed_unused = 1;
+			}
+			break;
+		case ACPI_HEST_TYPE_AER_ROOT_PORT:
+		case ACPI_HEST_TYPE_AER_ENDPOINT:
+		case ACPI_HEST_TYPE_AER_BRIDGE:
+			p += parse_acpi_hest_aer(p, hdr->type, pci, &firmware_first);
+			break;
+		case ACPI_HEST_TYPE_GENERIC_ERROR:
+			p += parse_acpi_hest_generic(p);
+			break;
+		/* These should never appear either */
+		case ACPI_HEST_TYPE_RESERVED:
+		default:
+			if (!printed_reserved) {
+				printk(KERN_DEBUG PREFIX
+				       "HEST Error Source list contains a reserved type (%d).\n", hdr->type);
+				printed_reserved = 1;
+			}
+			break;
+		}
+	}
+	return firmware_first;
+}
+
+int acpi_hest_firmware_first_pci(struct pci_dev *pci)
+{
+	acpi_status status = AE_NOT_FOUND;
+	struct acpi_table_header *hest = NULL;
+	status = acpi_get_firmware_table("HEST", 1, ACPI_LOGICAL_ADDRESSING,
+					 (struct acpi_table_header **) &hest);
+
+	if (ACPI_SUCCESS(status)) {
+		if (acpi_hest_firmware_first(hest, pci)) {
+			return 1;
+		}
+	}
+	return 0;
+}
+EXPORT_SYMBOL_GPL(acpi_hest_firmware_first_pci);
diff --git a/drivers/pci/pcie/aer/aerdrv_core.c b/drivers/pci/pcie/aer/aerdrv_core.c
index 03a009f..50de6cf 100644
--- a/drivers/pci/pcie/aer/aerdrv_core.c
+++ b/drivers/pci/pcie/aer/aerdrv_core.c
@@ -35,6 +35,9 @@ int pci_enable_pcie_error_reporting(struct pci_dev *dev)
 	u16 reg16 = 0;
 	int pos;
 
+	if (dev->aer_firmware_first)
+		return -EIO;
+
 	pos = pci_find_ext_capability(dev, PCI_EXT_CAP_ID_ERR);
 	if (!pos)
 		return -EIO;
@@ -60,6 +63,9 @@ int pci_disable_pcie_error_reporting(struct pci_dev *dev)
 	u16 reg16 = 0;
 	int pos;
 
+	if (dev->aer_firmware_first)
+		return -EIO;
+
 	pos = pci_find_capability(dev, PCI_CAP_ID_EXP);
 	if (!pos)
 		return -EIO;
@@ -852,8 +858,22 @@ void aer_delete_rootport(struct aer_rpc *rpc)
  */
 int aer_init(struct pcie_device *dev)
 {
-	if (aer_osc_setup(dev) && !forceload)
-		return -ENXIO;
+	if (dev->port->aer_firmware_first) {
+		dev_printk(KERN_DEBUG, &dev->device,
+			   "PCIe errors handled by platform firmware.\n");
+		goto out;
+	}
+
+	if (aer_osc_setup(dev))
+		goto out;
 
 	return 0;
+out:
+	if (forceload) {
+		dev_printk(KERN_DEBUG, &dev->device,
+			   "aerdrv forceload requested.\n");
+		dev->port->aer_firmware_first = 0;
+		return 0;
+	}
+	return -ENXIO;
 }
diff --git a/drivers/pci/probe.c b/drivers/pci/probe.c
index 59ca63a..5588fb2 100644
--- a/drivers/pci/probe.c
+++ b/drivers/pci/probe.c
@@ -9,6 +9,7 @@
 #include <linux/module.h>
 #include <linux/pci.h>
 #include <linux/slab.h>
+#include <acpi/acpi_hest.h>
 #include "pci.h"
 
 #define CARDBUS_LATENCY_TIMER	176	/* secondary latency timer */
@@ -701,6 +702,12 @@ static void set_pcie_port_type(struct pci_dev *pdev)
 	pdev->pcie_type = (reg16 & PCI_EXP_FLAGS_TYPE) >> 4;
 }
 
+static void set_pci_aer_firmware_first(struct pci_dev *pdev)
+{
+	if (acpi_hest_firmware_first_pci(pdev))
+		pdev->aer_firmware_first = 1;
+}
+
 /**
  * pci_setup_device - fill in class and map information of a device
  * @dev: the device structure to fill
@@ -727,6 +734,7 @@ int pci_setup_device(struct pci_dev * dev)
 	dev->cfg_size = pci_cfg_space_size(dev);
 	dev->error_state = pci_channel_io_normal;
 	set_pcie_port_type(dev);
+	set_pci_aer_firmware_first(dev);
 
 	/* Assume 32-bit PCI; let 64-bit PCI cards (which are far rarer)
 	   set this higher, assuming the system even supports it.  */
@@ -1232,4 +1240,3 @@ void __init pci_sort_breadthfirst(void)
 	pci_sort_breadthfirst_devices();
 	pci_sort_breadthfirst_klist();
 }
-
diff --git a/include/acpi/acpi_hest.h b/include/acpi/acpi_hest.h
new file mode 100644
index 0000000..63194d0
--- /dev/null
+++ b/include/acpi/acpi_hest.h
@@ -0,0 +1,12 @@
+#ifndef __ACPI_HEST_H
+#define __ACPI_HEST_H
+
+#include <linux/pci.h>
+
+#ifdef CONFIG_ACPI
+extern int acpi_hest_firmware_first_pci(struct pci_dev *pci);
+#else
+static inline int acpi_hest_firmware_first_pci(struct pci_dev *pci) { return 0; }
+#endif
+
+#endif
diff --git a/include/acpi/actbl.h b/include/acpi/actbl.h
index b125cee..b5431ef 100644
--- a/include/acpi/actbl.h
+++ b/include/acpi/actbl.h
@@ -367,4 +367,233 @@ extern u8 acpi_fadt_is_v1;	/* is set to 1 if FADT is revision 1,
 #define ACPI_FADT_FLAG_OFFSET(f,o)      ACPI_FLAG_OFFSET (struct fadt_descriptor,f,o)
 #define ACPI_FACS_FLAG_OFFSET(f,o)      ACPI_FLAG_OFFSET (struct facs_descriptor,f,o)
 
+/*******************************************************************************
+ *
+ * HEST - Hardware Error Source Table (ACPI 4.0)
+ *        Version 1
+ *
+ ******************************************************************************/
+
+struct acpi_table_hest {
+	struct acpi_table_header header;	/* Common ACPI table header */
+	u32 error_source_count;
+};
+
+/* HEST subtable header */
+
+struct acpi_hest_header {
+	u16 type;
+	u16 source_id;
+};
+
+/* Values for Type field above for subtables */
+
+enum acpi_hest_types {
+	ACPI_HEST_TYPE_IA32_CHECK = 0,
+	ACPI_HEST_TYPE_IA32_CORRECTED_CHECK = 1,
+	ACPI_HEST_TYPE_IA32_NMI = 2,
+	ACPI_HEST_TYPE_NOT_USED3 = 3,
+	ACPI_HEST_TYPE_NOT_USED4 = 4,
+	ACPI_HEST_TYPE_NOT_USED5 = 5,
+	ACPI_HEST_TYPE_AER_ROOT_PORT = 6,
+	ACPI_HEST_TYPE_AER_ENDPOINT = 7,
+	ACPI_HEST_TYPE_AER_BRIDGE = 8,
+	ACPI_HEST_TYPE_GENERIC_ERROR = 9,
+	ACPI_HEST_TYPE_RESERVED = 10	/* 10 and greater are reserved */
+};
+
+/*
+ * HEST substructures contained in subtables
+ */
+
+/*
+ * IA32 Error Bank(s) - Follows the struct acpi_hest_ia_machine_check and
+ * struct acpi_hest_ia_corrected structures.
+ */
+struct acpi_hest_ia_error_bank {
+	u8 bank_number;
+	u8 clear_status_on_init;
+	u8 status_format;
+	u8 reserved;
+	u32 control_register;
+	u64 control_data;
+	u32 status_register;
+	u32 address_register;
+	u32 misc_register;
+};
+
+/* Common HEST sub-structure for PCI/AER structures below (6,7,8) */
+
+struct acpi_hest_aer_common {
+	u16 reserved1;
+	u8 flags;
+	u8 enabled;
+	u32 records_to_preallocate;
+	u32 max_sections_per_record;
+	u32 bus;
+	u16 device;
+	u16 function;
+	u16 device_control;
+	u16 reserved2;
+	u32 uncorrectable_mask;
+	u32 uncorrectable_severity;
+	u32 correctable_mask;
+	u32 advanced_capabilities;
+};
+
+/* Masks for HEST Flags fields */
+
+#define ACPI_HEST_FIRMWARE_FIRST        (1)
+#define ACPI_HEST_GLOBAL                (1<<1)
+
+/* Hardware Error Notification */
+
+struct acpi_hest_notify {
+	u8 type;
+	u8 length;
+	u16 config_write_enable;
+	u32 poll_interval;
+	u32 vector;
+	u32 polling_threshold_value;
+	u32 polling_threshold_window;
+	u32 error_threshold_value;
+	u32 error_threshold_window;
+};
+
+/* Values for Notify Type field above */
+
+enum acpi_hest_notify_types {
+	ACPI_HEST_NOTIFY_POLLED = 0,
+	ACPI_HEST_NOTIFY_EXTERNAL = 1,
+	ACPI_HEST_NOTIFY_LOCAL = 2,
+	ACPI_HEST_NOTIFY_SCI = 3,
+	ACPI_HEST_NOTIFY_NMI = 4,
+	ACPI_HEST_NOTIFY_RESERVED = 5	/* 5 and greater are reserved */
+};
+
+/* Values for config_write_enable bitfield above */
+
+#define ACPI_HEST_TYPE                  (1)
+#define ACPI_HEST_POLL_INTERVAL         (1<<1)
+#define ACPI_HEST_POLL_THRESHOLD_VALUE  (1<<2)
+#define ACPI_HEST_POLL_THRESHOLD_WINDOW (1<<3)
+#define ACPI_HEST_ERR_THRESHOLD_VALUE   (1<<4)
+#define ACPI_HEST_ERR_THRESHOLD_WINDOW  (1<<5)
+
+/*
+ * HEST subtables
+ */
+
+/* 0: IA32 Machine Check Exception */
+
+struct acpi_hest_ia_machine_check {
+	struct acpi_hest_header header;
+	u16 reserved1;
+	u8 flags;
+	u8 enabled;
+	u32 records_to_preallocate;
+	u32 max_sections_per_record;
+	u64 global_capability_data;
+	u64 global_control_data;
+	u8 num_hardware_banks;
+	u8 reserved3[7];
+};
+
+/* 1: IA32 Corrected Machine Check */
+
+struct acpi_hest_ia_corrected {
+	struct acpi_hest_header header;
+	u16 reserved1;
+	u8 flags;
+	u8 enabled;
+	u32 records_to_preallocate;
+	u32 max_sections_per_record;
+	struct acpi_hest_notify notify;
+	u8 num_hardware_banks;
+	u8 reserved2[3];
+};
+
+/* 2: IA32 Non-Maskable Interrupt */
+
+struct acpi_hest_ia_nmi {
+	struct acpi_hest_header header;
+	u32 reserved;
+	u32 records_to_preallocate;
+	u32 max_sections_per_record;
+	u32 max_raw_data_length;
+};
+
+/* 3,4,5: Not used */
+
+/* 6: PCI Express Root Port AER */
+
+struct acpi_hest_aer_root {
+	struct acpi_hest_header header;
+	struct acpi_hest_aer_common aer;
+	u32 root_error_command;
+};
+
+/* 7: PCI Express AER (AER Endpoint) */
+
+struct acpi_hest_aer {
+	struct acpi_hest_header header;
+	struct acpi_hest_aer_common aer;
+};
+
+/* 8: PCI Express/PCI-X Bridge AER */
+
+struct acpi_hest_aer_bridge {
+	struct acpi_hest_header header;
+	struct acpi_hest_aer_common aer;
+	u32 uncorrectable_mask2;
+	u32 uncorrectable_severity2;
+	u32 advanced_capabilities2;
+};
+
+/* 9: Generic Hardware Error Source */
+
+struct acpi_hest_generic {
+	struct acpi_hest_header header;
+	u16 related_source_id;
+	u8 reserved;
+	u8 enabled;
+	u32 records_to_preallocate;
+	u32 max_sections_per_record;
+	u32 max_raw_data_length;
+	struct acpi_generic_address error_status_address;
+	struct acpi_hest_notify notify;
+	u32 error_block_length;
+};
+
+/* Generic Error Status block */
+
+struct acpi_hest_generic_status {
+	u32 block_status;
+	u32 raw_data_offset;
+	u32 raw_data_length;
+	u32 data_length;
+	u32 error_severity;
+};
+
+/* Values for block_status flags above */
+
+#define ACPI_HEST_UNCORRECTABLE             (1)
+#define ACPI_HEST_CORRECTABLE               (1<<1)
+#define ACPI_HEST_MULTIPLE_UNCORRECTABLE    (1<<2)
+#define ACPI_HEST_MULTIPLE_CORRECTABLE      (1<<3)
+#define ACPI_HEST_ERROR_ENTRY_COUNT         (0xFF<<4)	/* 8 bits, error count */
+
+/* Generic Error Data entry */
+
+struct acpi_hest_generic_data {
+	u8 section_type[16];
+	u32 error_severity;
+	u16 revision;
+	u8 validation_bits;
+	u8 flags;
+	u32 error_data_length;
+	u8 fru_id[16];
+	u8 fru_text[20];
+};
+
 #endif				/* __ACTBL_H__ */
diff --git a/include/linux/pci.h b/include/linux/pci.h
index 452803a..6ee67f9 100644
--- a/include/linux/pci.h
+++ b/include/linux/pci.h
@@ -214,6 +214,7 @@ struct pci_dev {
 #ifdef CONFIG_PCI_IOV
 	struct pci_ats *ats; /* Address Translation Service */
 #endif
+	unsigned int aer_firmware_first:1;
 #endif /* !__GENKSYMS__ */
 };