Skip to contentSkip to conversation

ACPI, APEI: Make APEI core configurable built-in instead of module

PATCH In review Huang Ying 2 participants 7 messages Mar 2, 2010 – Mar 3, 2010

Subsystems (5)
lore
Share & tools
lore

Huang Ying

author
#0 lore raw

PATCH 1/2 ACPI, APEI: Make APEI core configurable built-in instead of module

PCIE AER optionally depends on APEI HEST tabling parsing. If AER is built-in, HEST should be built-in or not configured at all instead of module. It is hard to express this elegantly in Kconfig. It is better to make APEI code part configurable built-in instead of module.

On the other hand, APEI core code is used for hardware error processing. It may run in very bad condition. It is reasonable to keep it built-in if enabled.

---
 Documentation/kernel-parameters.txt |   10 ++++----
 drivers/acpi/apei/Kconfig           |    2 +-
 drivers/acpi/apei/apei-base.c       |   42 +++++------------------------------
 drivers/acpi/apei/apei-internal.h   |    4 +--
 drivers/acpi/apei/einj.c            |    2 +-
 drivers/acpi/apei/hest.c            |   34 +++++++++++++++++++++++-----
 6 files changed, 42 insertions(+), 52 deletions(-)
Documentation/kernel-parameters.txt +5 -5
index b41a486..212d63f 100644
@@ -350,11 +350,6 @@ and is between 256 and 4096 characters. It is defined in the file
 			not play well with APC CPU idle - disable it if you have
 			APC and your system crashes randomly.
 
-	apei.hest_disable= [ACPI]
-			Disable Hardware Error Source Table (HEST) support,
-			corresponding firmware-first mode error processing
-			logic will be disabled.
-
 	apic=		[APIC,X86-32] Advanced Programmable Interrupt Controller
 			Change the output verbosity whilst booting
 			Format: { quiet (default) | verbose | debug }
@@ -847,6 +842,11 @@ and is between 256 and 4096 characters. It is defined in the file
 	hd=		[EIDE] (E)IDE hard drive subsystem geometry
 			Format: <cyl>,<head>,<sect>
 
+	hest_disable	[ACPI]
+			Disable Hardware Error Source Table (HEST) support,
+			corresponding firmware-first mode error processing
+			logic will be disabled.
+
 	highmem=nn[KMG]	[KNL,BOOT] forces the highmem zone to have an exact
 			size of <nn>. This works even on boxes that have no
 			highmem otherwise. This also works to reduce highmem
drivers/acpi/apei/Kconfig +1 -1
index 22791e4..5f0a41c 100644
@@ -1,5 +1,5 @@
 config ACPI_APEI
-	tristate "ACPI Platform Error Interface (APEI)"
+	bool "ACPI Platform Error Interface (APEI)"
 	depends on X86
 	help
 	  APEI allows to report errors (for example from the chipset)
drivers/acpi/apei/apei-base.c +6 -36

55 lines not on this page · raw message

drivers/acpi/apei/apei-internal.h +1 -3
index 1ab6598..86e041a 100644
@@ -6,8 +6,6 @@
 #ifndef APEI_INTERNAL_H
 #define APEI_INTERNAL_H
 
-int hest_init(void);
-
 struct apei_exec_context;
 
 typedef int (*apei_exec_ins_func_t)(struct apei_exec_context *ctx,
@@ -93,5 +91,5 @@ int apei_exec_collect_resources(struct apei_exec_context *ctx,
 				struct apei_resources *resources);
 
 struct dentry;
-extern struct dentry *apei_debug_dir;
+struct dentry *apei_get_debugfs_dir(void);
 #endif
drivers/acpi/apei/einj.c +1 -1
index 5521609..788b990 100644
@@ -409,7 +409,7 @@ static int __init einj_init(void)
 	}
 
 	rc = -ENOMEM;
-	einj_debug_dir = debugfs_create_dir("einj", apei_debug_dir);
+	einj_debug_dir = debugfs_create_dir("einj", apei_get_debugfs_dir());
 	if (!einj_debug_dir)
 		goto err_cleanup;
 	fentry = debugfs_create_file("available_error_type", S_IRUSR,
drivers/acpi/apei/hest.c +28 -6
index f872e54..2bd23c1 100644
@@ -40,6 +40,9 @@
 
 #define HEST_PFX "HEST: "
 
+int hest_disable;
+EXPORT_SYMBOL_GPL(hest_disable);
+
 /* HEST table parsing */
 
 static struct acpi_table_hest *hest_tab;
@@ -118,30 +121,49 @@ int apei_hest_parse(apei_hest_func_t func, void *data)
 }
 EXPORT_SYMBOL_GPL(apei_hest_parse);
 
-int __init hest_init(void)
+static int __init setup_hest_disable(char *str)
+{
+	hest_disable = 1;
+	return 0;
+}
+
+__setup("hest_disable", setup_hest_disable);
+
+static int __init hest_init(void)
 {
 	acpi_status status;
-	int rc;
+	int rc = 0;
 
 	if (acpi_disabled)
-		return -ENODEV;
+		goto err;
+
+	if (hest_disable) {
+		pr_info(HEST_PFX "HEST tabling parsing is disabled.\n");
+		goto err;
+	}
 
 	status = acpi_get_table(ACPI_SIG_HEST, 0,
 				(struct acpi_table_header **)&hest_tab);
 	if (status == AE_NOT_FOUND) {
 		pr_info(HEST_PFX "Table is not found!\n");
-		return -ENODEV;
+		goto err;
 	} else if (ACPI_FAILURE(status)) {
 		const char *msg = acpi_format_exception(status);
 		pr_info(HEST_PFX "Failed to get table, %s\n", msg);
-		return -EINVAL;
+		rc = -EINVAL;
+		goto err;
 	}
 
 	rc = apei_hest_parse(hest_void_parse, NULL);
 	if (rc)
-		return rc;
+		goto err;
 
 	pr_info(HEST_PFX "HEST table parsing is initialized.\n");
 
 	return 0;
+err:
+	hest_disable = 1;
+	return rc;
 }
+
+subsys_initcall(hest_init);
Signature · 1 line
1.6.6.1
author
#1 lore raw

PATCH 2/2 ACPI, APEI, PCIE AER, use general HEST table parsing in AER firmware_first setup

Now, a dedicated HEST tabling parsing code is used for PCIE AER firmware_first setup. It is rebased on general HEST tabling parsing code of APEI. The firmware_first setup code is moved from PCI core to AER driver too, because it is only AER related.

---
 drivers/acpi/Makefile              |    1 -
 drivers/acpi/hest.c                |  135 ------------------------------------
 drivers/pci/pcie/aer/aerdrv.h      |    6 ++
 drivers/pci/pcie/aer/aerdrv_acpi.c |   70 +++++++++++++++++++
 drivers/pci/pcie/aer/aerdrv_core.c |    2 +
 drivers/pci/probe.c                |    8 --
 include/acpi/acpi_hest.h           |   12 ---
 7 files changed, 78 insertions(+), 156 deletions(-)
 delete mode 100644 drivers/acpi/hest.c
 delete mode 100644 include/acpi/acpi_hest.h
drivers/acpi/Makefile +0 -1
index c00d683..3ba7feb 100644
@@ -20,7 +20,6 @@ obj-y				+= acpi.o \
 # All the builtin files are in the "acpi." module_param namespace.
 acpi-y				+= osl.o utils.o reboot.o
 acpi-y				+= atomicio.o
-acpi-y				+= hest.o
 
 # sleep related files
 acpi-y				+= wakeup.o
drivers/acpi/hest.c deleted +0 -135

135 lines not on this page · raw message

drivers/pci/pcie/aer/aerdrv.h +6 -0
index bd833ea..84e9470 100644
@@ -134,4 +134,10 @@ static inline int aer_osc_setup(struct pcie_device *pciedev)
 }
 #endif
 
+#ifdef CONFIG_ACPI_APEI
+extern void aer_set_firmware_first(struct pcie_device *dev);
+#else
+static inline void aer_set_firmware_first(struct pcie_device *dev) { }
+#endif
+
 #endif /* _AERDRV_H_ */
drivers/pci/pcie/aer/aerdrv_acpi.c +70 -0

79 lines not on this page · raw message

drivers/pci/pcie/aer/aerdrv_core.c +2 -0
index c843a79..cc527c1 100644
@@ -858,6 +858,8 @@ void aer_delete_rootport(struct aer_rpc *rpc)
  */
 int aer_init(struct pcie_device *dev)
 {
+	aer_set_firmware_first(dev);
+
 	if (dev->port->aer_firmware_first) {
 		dev_printk(KERN_DEBUG, &dev->device,
 			   "PCIe errors handled by platform firmware.\n");
drivers/pci/probe.c +0 -8
index 2a94309..ccfaf19 100644
@@ -10,7 +10,6 @@
 #include <linux/module.h>
 #include <linux/cpumask.h>
 #include <linux/pci-aspm.h>
-#include <acpi/acpi_hest.h>
 #include "pci.h"
 
 #define CARDBUS_LATENCY_TIMER	176	/* secondary latency timer */
@@ -900,12 +899,6 @@ void set_pcie_hotplug_bridge(struct pci_dev *pdev)
 		pdev->is_hotplug_bridge = 1;
 }
 
-static void set_pci_aer_firmware_first(struct pci_dev *pdev)
-{
-	if (acpi_hest_firmware_first_pci(pdev))
-		pdev->aer_firmware_first = 1;
-}
-
 #define LEGACY_IO_RESOURCE	(IORESOURCE_IO | IORESOURCE_PCI_FIXED)
 
 /**
@@ -935,7 +928,6 @@ int pci_setup_device(struct pci_dev *dev)
 	dev->multifunction = !!(hdr_type & 0x80);
 	dev->error_state = pci_channel_io_normal;
 	set_pcie_port_type(dev);
-	set_pci_aer_firmware_first(dev);
 
 	list_for_each_entry(slot, &dev->bus->slots, list)
 		if (PCI_SLOT(dev->devfn) == slot->number)
include/acpi/acpi_hest.h deleted +0 -12
deleted file mode 100644
index 63194d0..0000000
@@ -1,12 +0,0 @@
-#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
Signature · 1 line
1.6.6.1
#2 lore raw

Re: [PATCH 2/2] ACPI, APEI, PCIE AER, use general HEST table parsing in AER firmware_first setup

Quoted 2 lines · Huang Ying

... The firmware_first setup code is moved from PCI core to AER driver too, because it is only AER related.

(snip)

Quoted 15 lines
drivers/pci/pcie/aer/aerdrv_core.c +2 -0
index c843a79..cc527c1 100644
@@ -858,6 +858,8 @@ void aer_delete_rootport(struct aer_rpc *rpc)
  */
 int aer_init(struct pcie_device *dev)
 {
+	aer_set_firmware_first(dev);
+
 	if (dev->port->aer_firmware_first) {
 		dev_printk(KERN_DEBUG, &dev->device,
 			   "PCIe errors handled by platform firmware.\n");
drivers/pci/probe.c +0 -0
index 2a94309..ccfaf19 100644

(snip)

drivers/pci/probe.c int pci_setup_device(struct pci_dev *dev) · line 933

Quoted 8 lines
drivers/pci/probe.c +0 -1
@@ -935,7 +928,6 @@ int pci_setup_device(struct pci_dev *dev)
 	dev->multifunction = !!(hdr_type & 0x80);
 	dev->error_state = pci_channel_io_normal;
 	set_pcie_port_type(dev);
-	set_pci_aer_firmware_first(dev);
 
 	list_for_each_entry(slot, &dev->bus->slots, list)
 		if (PCI_SLOT(dev->devfn) == slot->number)

The aer_init() will be called for root ports, but not for end point devices or so on. So please remain the firmware_first setup code in PCI core. Otherwise endpoint drivers will get success on call of pci_enable_pcie_error_reporting() regardless of the firmware first.

Thanks,
H.Seto

author
#3 lore raw

Re: [PATCH 2/2] ACPI, APEI, PCIE AER, use general HEST table parsing in AER firmware_first setup

Quoted 26 lines · Hidetoshi Seto
Quoted 2 lines · Huang Ying

... The firmware_first setup code is moved from PCI core to AER driver too, because it is only AER related.

(snip)

Quoted 15 lines
drivers/pci/pcie/aer/aerdrv_core.c +2 -0
index c843a79..cc527c1 100644
@@ -858,6 +858,8 @@ void aer_delete_rootport(struct aer_rpc *rpc)
  */
 int aer_init(struct pcie_device *dev)
 {
+	aer_set_firmware_first(dev);
+
 	if (dev->port->aer_firmware_first) {
 		dev_printk(KERN_DEBUG, &dev->device,
 			   "PCIe errors handled by platform firmware.\n");
drivers/pci/probe.c +0 -0
index 2a94309..ccfaf19 100644

(snip)

Quoted 6 lines
drivers/pci/probe.c +0 -1
@@ -935,7 +928,6 @@ int pci_setup_device(struct pci_dev *dev)
 	dev->multifunction = !!(hdr_type & 0x80);
 	dev->error_state = pci_channel_io_normal;
 	set_pcie_port_type(dev);
-	set_pci_aer_firmware_first(dev);
 
Quoted 7 lines

drivers/pci/probe.c int pci_setup_device(struct pci_dev *dev) · line 933

Quoted 2 lines
drivers/pci/probe.c +0 -0
 	list_for_each_entry(slot, &dev->bus->slots, list)
 		if (PCI_SLOT(dev->devfn) == slot->number)

The aer_init() will be called for root ports, but not for end point devices or so on. So please remain the firmware_first setup code in PCI core. Otherwise endpoint drivers will get success on call of pci_enable_pcie_error_reporting() regardless of the firmware first.

Or we can call firmware_first setup code in
pci_enable_pcie_error_reporting(), because

1. I think AER related code should be put in drivers/pci/pcie/aer instead of PCI core or drivers/acpi, if it is possible.

2. pci_setup_device is called so early, so that it is hard to do some HEST related initialization (such as checking bad format) before it.

Best Regards,
Huang Ying

#4 lore raw

Re: [PATCH 2/2] ACPI, APEI, PCIE AER, use general HEST table parsing in AER firmware_first setup

Quoted 6 lines · Huang Ying
Quoted 4 lines · Hidetoshi Seto

The aer_init() will be called for root ports, but not for end point devices or so on. So please remain the firmware_first setup code in PCI core. Otherwise endpoint drivers will get success on call of pci_enable_pcie_error_reporting() regardless of the firmware first.

Quoted 8 lines

Or we can call firmware_first setup code in
pci_enable_pcie_error_reporting(), because

1. I think AER related code should be put in drivers/pci/pcie/aer instead of PCI core or drivers/acpi, if it is possible.

2. pci_setup_device is called so early, so that it is hard to do some HEST related initialization (such as checking bad format) before it.

I understands the feeling, but before agreeing with your proposal, I'd like to have an answer of a question:

- Is it necessary to setup the firmware_first flag for an endpoint even if the endpoint's driver never call pci_enable_pcie_error_reporting()?

According to the current implementation, there are no driver referring the firmware_first flag other than that it owns. However I guess that the flag will be necessary for AER driver (i.e. aerdrv_core) in near future, because we can use the flag to determine whether the AER driver can check the device or not, when it is required to walk pci bus hierarchy to find an erroneous device.

For example, assume that there are 2 endpoints under a same root port. One is (likely on-board) "firmware first" endpoint, with driver which does not call pci_enable_pcie_error_reporting() (because of no interest in AER, or just not implemented yet, anyway). The other is (likely card seated on a slot) not firmware first, with better driver which can handle it's AER. If my understanding is correct and if everything goes well, errors on one should be reported via APEI while the other should be reported via AER driver.

We could call firmware_first setup for all endpoint from aer_init(), but it will need another care for hot-plugged endpoints.

Thanks,
H.Seto

author
#5 lore raw

Re: [PATCH 2/2] ACPI, APEI, PCIE AER, use general HEST table parsing in AER firmware_first setup

Quoted 31 lines · Hidetoshi Seto
Quoted 6 lines · Huang Ying
Quoted 4 lines · Hidetoshi Seto

The aer_init() will be called for root ports, but not for end point devices or so on. So please remain the firmware_first setup code in PCI core. Otherwise endpoint drivers will get success on call of pci_enable_pcie_error_reporting() regardless of the firmware first.

Quoted 8 lines

Or we can call firmware_first setup code in
pci_enable_pcie_error_reporting(), because

1. I think AER related code should be put in drivers/pci/pcie/aer instead of PCI core or drivers/acpi, if it is possible.

2. pci_setup_device is called so early, so that it is hard to do some HEST related initialization (such as checking bad format) before it.

I understands the feeling, but before agreeing with your proposal, I'd like to have an answer of a question:

- Is it necessary to setup the firmware_first flag for an endpoint even if the endpoint's driver never call pci_enable_pcie_error_reporting()?

According to the current implementation, there are no driver referring the firmware_first flag other than that it owns. However I guess that the flag will be necessary for AER driver (i.e. aerdrv_core) in near future, because we can use the flag to determine whether the AER driver can check the device or not, when it is required to walk pci bus hierarchy to find an erroneous device.

Quoted 9 lines

For example, assume that there are 2 endpoints under a same root port. One is (likely on-board) "firmware first" endpoint, with driver which does not call pci_enable_pcie_error_reporting() (because of no interest in AER, or just not implemented yet, anyway). The other is (likely card seated on a slot) not firmware first, with better driver which can handle it's AER. If my understanding is correct and if everything goes well, errors on one should be reported via APEI while the other should be reported via AER driver.

Yes. I think this should be supported. How about something as follow?

struct pci_dev {
	...
	unsigned int __firmware_first:2;
	...
};

int pcie_aer_get_firmware_first(struct pci_dev *dev)

{
	if (!dev->__firmware_first)
		aer_set_firmware_first(dev);
	return dev->__firmware_first & 0x1;
}

Then we use pcie_aer_get_firmware_first() instead of dev->firmware_first directly.

Best Regards,
Huang Ying

#6 lore raw

Re: [PATCH 2/2] ACPI, APEI, PCIE AER, use general HEST table parsing in AER firmware_first setup

Quoted 19 lines · Huang Ying
Quoted 9 lines

For example, assume that there are 2 endpoints under a same root port. One is (likely on-board) "firmware first" endpoint, with driver which does not call pci_enable_pcie_error_reporting() (because of no interest in AER, or just not implemented yet, anyway). The other is (likely card seated on a slot) not firmware first, with better driver which can handle it's AER. If my understanding is correct and if everything goes well, errors on one should be reported via APEI while the other should be reported via AER driver.

Yes. I think this should be supported. How about something as follow?

struct pci_dev {
	...
	unsigned int __firmware_first:2;
	...
};

int pcie_aer_get_firmware_first(struct pci_dev *dev)

Quoted 8 lines
{
	if (!dev->__firmware_first)
		aer_set_firmware_first(dev);
	return dev->__firmware_first & 0x1;
}

Then we use pcie_aer_get_firmware_first() instead of dev->firmware_first directly.

Looks reasonable. I think the following is more straightforward:

struct pci_dev {
	...
	unsigned int __firmware_first_valid:1;
	unsigned int __firmware_first:1;
	...
};

int pcie_aer_get_firmware_first(struct pci_dev *dev)

{
	if (!dev->__firmware_first_valid)
		aer_set_firmware_first(dev);
	return dev->__firmware_first;
}

Thanks,
H.Seto