Skip to contentSkip to conversation

PCI / PCIe / AER: Disable native AER service if BIOS has precedence

PATCH In review Rafael J. Wysocki 4 participants 6 messages Sep 19, 2010 – Sep 25, 2010

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 PCI / PCIe / AER: Disable native AER service if BIOS has precedence

From: Rafael J. Wysocki <[email protected]>

There is a design issue related to PCIe AER and _OSC that the BIOS may be asked to grant control of the AER service even if some Hardware Error Source Table (HEST) entries contain information meaning that the BIOS really should control it. Namely, pcie_port_acpi_setup() calls pcie_aer_get_firmware_first() that determines whether or not the AER service should be controlled by the BIOS on the basis of the HEST information for the given PCIe port. The BIOS is asked to grant control of the AER service for a PCIe Root Complex if pcie_aer_get_firmware_first() returns 'false' for at least one root port in that complex, even if all of the other root ports' HEST entries have the FIRMWARE_FIRST flag set (and none of them has the GLOBAL flag set). However, if the AER service is controlled by the kernel, that may interfere with the BIOS' handling of the error sources having the FIRMWARE_FIRST flag. Moreover, there may be PCIe endpoints that have the FIRMWARE_FIRST flag set in HEST and are attached to the root ports in question, in which case it also may be unsafe to ask the BIOS for control of the AER service.

For this reason, introduce a function checking if there's at least one PCIe-related HEST entry with the FIRMWARE_FIRST flag set and disable the native AER service altogether if this function returns 'true'.

---
 drivers/pci/pcie/aer/aerdrv.c      |    2 +-
 drivers/pci/pcie/aer/aerdrv.h      |    3 +++
 drivers/pci/pcie/aer/aerdrv_acpi.c |   31 +++++++++++++++++++++++++++++++
 drivers/pci/pcie/portdrv_acpi.c    |    2 +-
 4 files changed, 36 insertions(+), 2 deletions(-)
drivers/pci/pcie/aer/aerdrv_acpi.c +31 -0
Index: linux-2.6/drivers/pci/pcie/aer/aerdrv_acpi.c
===================================================================
@@ -93,4 +93,35 @@ int pcie_aer_get_firmware_first(struct p
 		aer_set_firmware_first(dev);
 	return dev->__aer_firmware_first;
 }
+
+static bool aer_firmware_first;
+
+static int aer_hest_parse_aff(struct acpi_hest_header *hest_hdr, void *data)
+{
+	if (aer_firmware_first)
+		return 0;
+
+	switch (hest_hdr->type) {
+	case ACPI_HEST_TYPE_AER_ROOT_PORT:
+	case ACPI_HEST_TYPE_AER_ENDPOINT:
+	case ACPI_HEST_TYPE_AER_BRIDGE:
+		aer_firmware_first = !!(p->flags & ACPI_HEST_FIRMWARE_FIRST);
+	default:
+		return 0;
+	}
+}
+
+/**
+ * aer_acpi_firmware_first - Check if APEI should control AER.
+ */
+bool aer_acpi_firmware_first(void)
+{
+	static bool parsed = false;
+
+	if (!parsed) {
+		apei_hest_parse(aer_hest_parse_aff, NULL);
+		parsed = true;
+	}
+	return aer_firmware_first;
+}
 #endif
drivers/pci/pcie/aer/aerdrv.h +3 -0
Index: linux-2.6/drivers/pci/pcie/aer/aerdrv.h
===================================================================
@@ -132,6 +132,7 @@ static inline int aer_osc_setup(struct p
 
 #ifdef CONFIG_ACPI_APEI
 extern int pcie_aer_get_firmware_first(struct pci_dev *pci_dev);
+extern bool aer_acpi_firmware_first(void);
 #else
 static inline int pcie_aer_get_firmware_first(struct pci_dev *pci_dev)
 {
@@ -139,6 +140,8 @@ static inline int pcie_aer_get_firmware_
 		return pci_dev->__aer_firmware_first;
 	return 0;
 }
+
+static inline bool aer_acpi_firmware_first(void) { return false; }
 #endif
 
 static inline void pcie_aer_force_firmware_first(struct pci_dev *pci_dev,
drivers/pci/pcie/aer/aerdrv.c +1 -1
Index: linux-2.6/drivers/pci/pcie/aer/aerdrv.c
===================================================================
@@ -416,7 +416,7 @@ static void aer_error_resume(struct pci_
  */
 static int __init aer_service_init(void)
 {
-	if (!pci_aer_available())
+	if (!pci_aer_available() || aer_acpi_firmware_first())
 		return -ENXIO;
 	return pcie_port_service_register(&aerdriver);
 }
drivers/pci/pcie/portdrv_acpi.c +1 -1
Index: linux-2.6/drivers/pci/pcie/portdrv_acpi.c
===================================================================
@@ -49,7 +49,7 @@ int pcie_port_acpi_setup(struct pci_dev 
 		| OSC_PCI_EXPRESS_PME_CONTROL;
 
 	if (pci_aer_available()) {
-		if (pcie_aer_get_firmware_first(port))
+		if (aer_acpi_firmware_first())
 			dev_dbg(&port->dev, "PCIe errors handled by BIOS.\n");
 		else
 			flags |= OSC_PCI_EXPRESS_AER_CONTROL;
#1 lore raw
Quoted 44 lines · Rafael J. Wysocki

From: Rafael J. Wysocki<[email protected]>

There is a design issue related to PCIe AER and _OSC that the BIOS may be asked to grant control of the AER service even if some Hardware Error Source Table (HEST) entries contain information meaning that the BIOS really should control it. Namely, pcie_port_acpi_setup() calls pcie_aer_get_firmware_first() that determines whether or not the AER service should be controlled by the BIOS on the basis of the HEST information for the given PCIe port. The BIOS is asked to grant control of the AER service for a PCIe Root Complex if pcie_aer_get_firmware_first() returns 'false' for at least one root port in that complex, even if all of the other root ports' HEST entries have the FIRMWARE_FIRST flag set (and none of them has the GLOBAL flag set). However, if the AER service is controlled by the kernel, that may interfere with the BIOS' handling of the error sources having the FIRMWARE_FIRST flag. Moreover, there may be PCIe endpoints that have the FIRMWARE_FIRST flag set in HEST and are attached to the root ports in question, in which case it also may be unsafe to ask the BIOS for control of the AER service.

For this reason, introduce a function checking if there's at least one PCIe-related HEST entry with the FIRMWARE_FIRST flag set and disable the native AER service altogether if this function returns 'true'.

---
  drivers/pci/pcie/aer/aerdrv.c      |    2 +-
  drivers/pci/pcie/aer/aerdrv.h      |    3 +++
  drivers/pci/pcie/aer/aerdrv_acpi.c |   31 +++++++++++++++++++++++++++++++
  drivers/pci/pcie/portdrv_acpi.c    |    2 +-
  4 files changed, 36 insertions(+), 2 deletions(-)

Index: linux-2.6/drivers/pci/pcie/aer/aerdrv_acpi.c
===================================================================

drivers/pci/pcie/aer/aerdrv_acpi.c +3 -0
@@ -93,4 +93,35 @@ int pcie_aer_get_firmware_first(struct p
  		aer_set_firmware_first(dev);
  	return dev->__aer_firmware_first;
  }
+
+static bool aer_firmware_first;
+

drivers/pci/pcie/aer/aerdrv_acpi.c int pcie_aer_get_firmware_first(struct p · line 108

Quoted 10 lines
drivers/pci/pcie/aer/aerdrv_acpi.c +10 -0
+static int aer_hest_parse_aff(struct acpi_hest_header *hest_hdr, void *data)
+{
+	if (aer_firmware_first)
+		return 0;
+
+	switch (hest_hdr->type) {
+	case ACPI_HEST_TYPE_AER_ROOT_PORT:
+	case ACPI_HEST_TYPE_AER_ENDPOINT:
+	case ACPI_HEST_TYPE_AER_BRIDGE:
+		aer_firmware_first = !!(p->flags&  ACPI_HEST_FIRMWARE_FIRST);

Where "p" comes from ? Maybe it points to "struct acpi_hest_aer_common *p;"

Quoted 71 lines
drivers/pci/pcie/aer/aerdrv_acpi.c +18 -0
+	default:
+		return 0;
+	}
+}
+
+/**
+ * aer_acpi_firmware_first - Check if APEI should control AER.
+ */
+bool aer_acpi_firmware_first(void)
+{
+	static bool parsed = false;
+
+	if (!parsed) {
+		apei_hest_parse(aer_hest_parse_aff, NULL);
+		parsed = true;
+	}
+	return aer_firmware_first;
+}
  #endif

Index: linux-2.6/drivers/pci/pcie/aer/aerdrv.h
===================================================================

drivers/pci/pcie/aer/aerdrv.h +3 -0
@@ -132,6 +132,7 @@ static inline int aer_osc_setup(struct p
 
  #ifdef CONFIG_ACPI_APEI
  extern int pcie_aer_get_firmware_first(struct pci_dev *pci_dev);
+extern bool aer_acpi_firmware_first(void);
  #else
  static inline int pcie_aer_get_firmware_first(struct pci_dev *pci_dev)
  {
@@ -139,6 +140,8 @@ static inline int pcie_aer_get_firmware_
  		return pci_dev->__aer_firmware_first;
  	return 0;
  }
+
+static inline bool aer_acpi_firmware_first(void) { return false; }
  #endif
 
  static inline void pcie_aer_force_firmware_first(struct pci_dev *pci_dev,

Index: linux-2.6/drivers/pci/pcie/aer/aerdrv.c
===================================================================

drivers/pci/pcie/aer/aerdrv.c +1 -1
@@ -416,7 +416,7 @@ static void aer_error_resume(struct pci_
   */
  static int __init aer_service_init(void)
  {
-	if (!pci_aer_available())
+	if (!pci_aer_available() || aer_acpi_firmware_first())
  		return -ENXIO;
  	return pcie_port_service_register(&aerdriver);
  }

Index: linux-2.6/drivers/pci/pcie/portdrv_acpi.c
===================================================================

drivers/pci/pcie/portdrv_acpi.c +1 -2
@@ -49,7 +49,7 @@ int pcie_port_acpi_setup(struct pci_dev
  		| OSC_PCI_EXPRESS_PME_CONTROL;
 
  	if (pci_aer_available()) {
-		if (pcie_aer_get_firmware_first(port))
+		if (aer_acpi_firmware_first())
  			dev_dbg(&port->dev, "PCIe errors handled by BIOS.\n");
  		else
  			flags |= OSC_PCI_EXPRESS_AER_CONTROL;
--

To unsubscribe from this list: send the line "unsubscribe linux-acpi" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

#2 lore raw

PATCH 1/1 PCI / PCIe / AER: Disable native AER service if BIOS has precedence (v2) (was: Re: [PATCH] PCI / PCIe / AER ...)

Quoted 49 lines · Chen Gong
Quoted 48 lines · Rafael J. Wysocki

From: Rafael J. Wysocki<[email protected]>

There is a design issue related to PCIe AER and _OSC that the BIOS may be asked to grant control of the AER service even if some Hardware Error Source Table (HEST) entries contain information meaning that the BIOS really should control it. Namely, pcie_port_acpi_setup() calls pcie_aer_get_firmware_first() that determines whether or not the AER service should be controlled by the BIOS on the basis of the HEST information for the given PCIe port. The BIOS is asked to grant control of the AER service for a PCIe Root Complex if pcie_aer_get_firmware_first() returns 'false' for at least one root port in that complex, even if all of the other root ports' HEST entries have the FIRMWARE_FIRST flag set (and none of them has the GLOBAL flag set). However, if the AER service is controlled by the kernel, that may interfere with the BIOS' handling of the error sources having the FIRMWARE_FIRST flag. Moreover, there may be PCIe endpoints that have the FIRMWARE_FIRST flag set in HEST and are attached to the root ports in question, in which case it also may be unsafe to ask the BIOS for control of the AER service.

For this reason, introduce a function checking if there's at least one PCIe-related HEST entry with the FIRMWARE_FIRST flag set and disable the native AER service altogether if this function returns 'true'.

---
  drivers/pci/pcie/aer/aerdrv.c      |    2 +-
  drivers/pci/pcie/aer/aerdrv.h      |    3 +++
  drivers/pci/pcie/aer/aerdrv_acpi.c |   31 +++++++++++++++++++++++++++++++
  drivers/pci/pcie/portdrv_acpi.c    |    2 +-
  4 files changed, 36 insertions(+), 2 deletions(-)

Index: linux-2.6/drivers/pci/pcie/aer/aerdrv_acpi.c
===================================================================

drivers/pci/pcie/aer/aerdrv_acpi.c +7 -0
@@ -93,4 +93,35 @@ int pcie_aer_get_firmware_first(struct p
  		aer_set_firmware_first(dev);
  	return dev->__aer_firmware_first;
  }
+
+static bool aer_firmware_first;
+
+static int aer_hest_parse_aff(struct acpi_hest_header *hest_hdr, void *data)
+{
+	if (aer_firmware_first)
+		return 0;
Quoted 8 lines

drivers/pci/pcie/aer/aerdrv_acpi.c int pcie_aer_get_firmware_first(struct p · line 108

Quoted 6 lines
drivers/pci/pcie/aer/aerdrv_acpi.c +6 -0
+
+	switch (hest_hdr->type) {
+	case ACPI_HEST_TYPE_AER_ROOT_PORT:
+	case ACPI_HEST_TYPE_AER_ENDPOINT:
+	case ACPI_HEST_TYPE_AER_BRIDGE:
+		aer_firmware_first = !!(p->flags&  ACPI_HEST_FIRMWARE_FIRST);

Where "p" comes from ? Maybe it points to "struct acpi_hest_aer_common *p;"

My bad, terribly sorry.

I must have posted a wrong version of the patch. :-( Fixed patch is appended.

Thanks,
Rafael

---
From: Rafael J. Wysocki <[email protected]>
Subject: PCI / PCIe / AER: Disable native AER service if BIOS has precedence (v2)

There is a design issue related to PCIe AER and _OSC that the BIOS may be asked to grant control of the AER service even if some Hardware Error Source Table (HEST) entries contain information meaning that the BIOS really should control it. Namely, pcie_port_acpi_setup() calls pcie_aer_get_firmware_first() that determines whether or not the AER service should be controlled by the BIOS on the basis of the HEST information for the given PCIe port. The BIOS is asked to grant control of the AER service for a PCIe Root Complex if pcie_aer_get_firmware_first() returns 'false' for at least one root port in that complex, even if all of the other root ports' HEST entries have the FIRMWARE_FIRST flag set (and none of them has the GLOBAL flag set). However, if the AER service is controlled by the kernel, that may interfere with the BIOS' handling of the error sources having the FIRMWARE_FIRST flag. Moreover, there may be PCIe endpoints that have the FIRMWARE_FIRST flag set in HEST and are attached to the root ports in question, in which case it also may be unsafe to ask the BIOS for control of the AER service.

For this reason, introduce a function checking if there's at least one PCIe-related HEST entry with the FIRMWARE_FIRST flag set and disable the native AER service altogether if this function returns 'true'.

---

Changes since v1:

* Fix compilation problem with undefined p in aer_hest_parse_aff().

---
 drivers/pci/pcie/aer/aerdrv.c      |    2 +-
 drivers/pci/pcie/aer/aerdrv.h      |    3 +++
 drivers/pci/pcie/aer/aerdrv_acpi.c |   34 ++++++++++++++++++++++++++++++++++
 drivers/pci/pcie/portdrv_acpi.c    |    2 +-
 4 files changed, 39 insertions(+), 2 deletions(-)
drivers/pci/pcie/aer/aerdrv_acpi.c +34 -0
Index: linux-2.6/drivers/pci/pcie/aer/aerdrv_acpi.c
===================================================================
@@ -93,4 +93,38 @@ int pcie_aer_get_firmware_first(struct p
 		aer_set_firmware_first(dev);
 	return dev->__aer_firmware_first;
 }
+
+static bool aer_firmware_first;
+
+static int aer_hest_parse_aff(struct acpi_hest_header *hest_hdr, void *data)
+{
+	struct acpi_hest_aer_common *p;
+
+	if (aer_firmware_first)
+		return 0;
+
+	switch (hest_hdr->type) {
+	case ACPI_HEST_TYPE_AER_ROOT_PORT:
+	case ACPI_HEST_TYPE_AER_ENDPOINT:
+	case ACPI_HEST_TYPE_AER_BRIDGE:
+		p = (struct acpi_hest_aer_common *)(hest_hdr + 1);
+		aer_firmware_first = !!(p->flags & ACPI_HEST_FIRMWARE_FIRST);
+	default:
+		return 0;
+	}
+}
+
+/**
+ * aer_acpi_firmware_first - Check if APEI should control AER.
+ */
+bool aer_acpi_firmware_first(void)
+{
+	static bool parsed = false;
+
+	if (!parsed) {
+		apei_hest_parse(aer_hest_parse_aff, NULL);
+		parsed = true;
+	}
+	return aer_firmware_first;
+}
 #endif
drivers/pci/pcie/aer/aerdrv.h +3 -0
Index: linux-2.6/drivers/pci/pcie/aer/aerdrv.h
===================================================================
@@ -132,6 +132,7 @@ static inline int aer_osc_setup(struct p
 
 #ifdef CONFIG_ACPI_APEI
 extern int pcie_aer_get_firmware_first(struct pci_dev *pci_dev);
+extern bool aer_acpi_firmware_first(void);
 #else
 static inline int pcie_aer_get_firmware_first(struct pci_dev *pci_dev)
 {
@@ -139,6 +140,8 @@ static inline int pcie_aer_get_firmware_
 		return pci_dev->__aer_firmware_first;
 	return 0;
 }
+
+static inline bool aer_acpi_firmware_first(void) { return false; }
 #endif
 
 static inline void pcie_aer_force_firmware_first(struct pci_dev *pci_dev,
drivers/pci/pcie/aer/aerdrv.c +1 -1
Index: linux-2.6/drivers/pci/pcie/aer/aerdrv.c
===================================================================
@@ -416,7 +416,7 @@ static void aer_error_resume(struct pci_
  */
 static int __init aer_service_init(void)
 {
-	if (!pci_aer_available())
+	if (!pci_aer_available() || aer_acpi_firmware_first())
 		return -ENXIO;
 	return pcie_port_service_register(&aerdriver);
 }
drivers/pci/pcie/portdrv_acpi.c +1 -1
Index: linux-2.6/drivers/pci/pcie/portdrv_acpi.c
===================================================================
@@ -49,7 +49,7 @@ int pcie_port_acpi_setup(struct pci_dev 
 		| OSC_PCI_EXPRESS_PME_CONTROL;
 
 	if (pci_aer_available()) {
-		if (pcie_aer_get_firmware_first(port))
+		if (aer_acpi_firmware_first())
 			dev_dbg(&port->dev, "PCIe errors handled by BIOS.\n");
 		else
 			flags |= OSC_PCI_EXPRESS_AER_CONTROL;

--
To unsubscribe from this list: send the line "unsubscribe linux-acpi" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

#3 lore raw

Re: [PATCH] PCI / PCIe / AER: Disable native AER service if BIOS has precedence (v2) (was: Re: [PATCH] PCI / PCIe / AER ...)

Quoted 22 lines · On Mon, 20 Sep 2010 18:50:00 +0200 "Rafael J. Wysocki" <[email protected]> wrote:
Quoted 21 lines · Chen Gong
Quoted 20 lines · Rafael J. Wysocki

From: Rafael J. Wysocki<[email protected]>

There is a design issue related to PCIe AER and _OSC that the BIOS may be asked to grant control of the AER service even if some Hardware Error Source Table (HEST) entries contain information meaning that the BIOS really should control it. Namely, pcie_port_acpi_setup() calls pcie_aer_get_firmware_first() that determines whether or not the AER service should be controlled by the BIOS on the basis of the HEST information for the given PCIe port. The BIOS is asked to grant control of the AER service for a PCIe Root Complex if pcie_aer_get_firmware_first() returns 'false' for at least one root port in that complex, even if all of the other root ports' HEST entries have the FIRMWARE_FIRST flag set (and none of them has the GLOBAL flag set). However, if the AER service is controlled by the kernel, that may interfere with the BIOS' handling of the error sources having the FIRMWARE_FIRST flag. Moreover, there may be PCIe endpoints that have the FIRMWARE_FIRST flag set in HEST and are attached to the root ports in question, in which case it also may be unsafe to ask the BIOS for control of the AER service.

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

For this reason, introduce a function checking if there's at least one PCIe-related HEST entry with the FIRMWARE_FIRST flag set and disable the native AER service altogether if this function returns 'true'.

Applied the second patch to my linux-next branch, thanks.

Signature · 5 lines
Jesse Barnes, Intel Open Source Technology Center
--
To unsubscribe from this list: send the line "unsubscribe linux-acpi" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html
#4 lore raw

Re: [PATCH] PCI / PCIe / AER: Disable native AER service if BIOS has precedence (v2) (was: Re: [PATCH] PCI / PCIe / AER ...)

In reply to Rafael J. Wysocki

Hi, Rafael,

drivers/pci/pcie/portdrv_acpi.c int pcie_port_acpi_setup(struct pci_dev · line 55

Quoted 11 lines · Rafael J. Wysocki
drivers/pci/pcie/portdrv_acpi.c +1 -1
@@ -49,7 +49,7 @@ int pcie_port_acpi_setup(struct pci_dev 
 		| OSC_PCI_EXPRESS_PME_CONTROL;
 
 	if (pci_aer_available()) {
-		if (pcie_aer_get_firmware_first(port))
+		if (aer_acpi_firmware_first())
 			dev_dbg(&port->dev, "PCIe errors handled by BIOS.\n");
 		else
 			flags |= OSC_PCI_EXPRESS_AER_CONTROL;

The original per-device firmware_first is overridden by the new global firmware_first? If it is, why keep the original per-device firmware_first?

Best Regards,
Huang Ying

#5 lore raw

Re: [PATCH] PCI / PCIe / AER: Disable native AER service if BIOS has precedence (v2) (was: Re: [PATCH] PCI / PCIe / AER ...)

Quoted 8 lines · Huang Ying

Hi, Rafael,

Quoted 5 lines · Rafael J. Wysocki
drivers/pci/pcie/portdrv_acpi.c +0 -0
@@ -49,7 +49,7 @@ int pcie_port_acpi_setup(struct pci_dev 
 		| OSC_PCI_EXPRESS_PME_CONTROL;
 
Quoted 10 lines

drivers/pci/pcie/portdrv_acpi.c int pcie_port_acpi_setup(struct pci_dev · line 55

Quoted 6 lines
drivers/pci/pcie/portdrv_acpi.c +1 -1
 	if (pci_aer_available()) {
-		if (pcie_aer_get_firmware_first(port))
+		if (aer_acpi_firmware_first())
 			dev_dbg(&port->dev, "PCIe errors handled by BIOS.\n");
 		else
 			flags |= OSC_PCI_EXPRESS_AER_CONTROL;

The original per-device firmware_first is overridden by the new global firmware_first? If it is, why keep the original per-device firmware_first?

I didn't want to remove it just yet, but I do think it should be removed.

Rafael