EEH cleanups and reworks

EEH cleanups and reworks

[Oliver O'Halloran]

This is really two series joined together since they end up conflicting
with each other lighty slightly and I figured this is easier for all
involved.

Patches 1-5 streamline how the generic and platform specfic parts of EEH
are initialised at boot so more of the setup process happens in linear code
rather in initcalls.

Patches 6-9 re-work how the per-device EEH probing works on pseries to
make it line up better with the process outlined in PAPR. It also
removes the pe->config_addr field which has always confused me since
it's not really clear what or why it's needed (spoiler: it's not).

Oliver

[PATCH 1/9] powerpc/eeh: Rework EEH initialisation

[Oliver O'Halloran]

Drop the EEH register / unregister ops thing and have the platform pass the
ops structure into eeh_init() directly. This takes one initcall out of the
EEH setup path and it means we're only doing EEH setup on the platforms
which actually support it. It's also less code and generally easier to
follow.

No functional changes.

Signed-off-by: Oliver O'Halloran <oohall@gmail.com>
---
 arch/powerpc/include/asm/eeh.h               |  3 +-
 arch/powerpc/kernel/eeh.c                    | 87 ++++----------------
 arch/powerpc/platforms/powernv/eeh-powernv.c |  4 +-
 arch/powerpc/platforms/pseries/eeh_pseries.c |  5 +-
 4 files changed, 21 insertions(+), 78 deletions(-)

diff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h
index d5f369bcd130..765bcf63edea 100644
--- a/arch/powerpc/include/asm/eeh.h
+++ b/arch/powerpc/include/asm/eeh.h
@@ -295,8 +295,7 @@ const char *eeh_pe_loc_get(struct eeh_pe *pe);
 struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe);
 
 void eeh_show_enabled(void);
-int __init eeh_ops_register(struct eeh_ops *ops);
-int __exit eeh_ops_unregister(const char *name);
+int __init eeh_init(struct eeh_ops *ops);
 int eeh_check_failure(const volatile void __iomem *token);
 int eeh_dev_check_failure(struct eeh_dev *edev);
 void eeh_addr_cache_init(void);
diff --git a/arch/powerpc/kernel/eeh.c b/arch/powerpc/kernel/eeh.c
index 94682382fc8c..28a0ea5d9faa 100644
--- a/arch/powerpc/kernel/eeh.c
+++ b/arch/powerpc/kernel/eeh.c
@@ -929,56 +929,6 @@ void eeh_save_bars(struct eeh_dev *edev)
 		edev->config_space[1] |= PCI_COMMAND_MASTER;
 }
 
-/**
- * eeh_ops_register - Register platform dependent EEH operations
- * @ops: platform dependent EEH operations
- *
- * Register the platform dependent EEH operation callback
- * functions. The platform should call this function before
- * any other EEH operations.
- */
-int __init eeh_ops_register(struct eeh_ops *ops)
-{
-	if (!ops->name) {
-		pr_warn("%s: Invalid EEH ops name for %p\n",
-			__func__, ops);
-		return -EINVAL;
-	}
-
-	if (eeh_ops && eeh_ops != ops) {
-		pr_warn("%s: EEH ops of platform %s already existing (%s)\n",
-			__func__, eeh_ops->name, ops->name);
-		return -EEXIST;
-	}
-
-	eeh_ops = ops;
-
-	return 0;
-}
-
-/**
- * eeh_ops_unregister - Unreigster platform dependent EEH operations
- * @name: name of EEH platform operations
- *
- * Unregister the platform dependent EEH operation callback
- * functions.
- */
-int __exit eeh_ops_unregister(const char *name)
-{
-	if (!name || !strlen(name)) {
-		pr_warn("%s: Invalid EEH ops name\n",
-			__func__);
-		return -EINVAL;
-	}
-
-	if (eeh_ops && !strcmp(eeh_ops->name, name)) {
-		eeh_ops = NULL;
-		return 0;
-	}
-
-	return -EEXIST;
-}
-
 static int eeh_reboot_notifier(struct notifier_block *nb,
 			       unsigned long action, void *unused)
 {
@@ -991,25 +941,22 @@ static struct notifier_block eeh_reboot_nb = {
 };
 
 /**
- * eeh_init - EEH initialization
- *
- * Initialize EEH by trying to enable it for all of the adapters in the system.
- * As a side effect we can determine here if eeh is supported at all.
- * Note that we leave EEH on so failed config cycles won't cause a machine
- * check.  If a user turns off EEH for a particular adapter they are really
- * telling Linux to ignore errors.  Some hardware (e.g. POWER5) won't
- * grant access to a slot if EEH isn't enabled, and so we always enable
- * EEH for all slots/all devices.
+ * eeh_init - System wide EEH initialization
  *
- * The eeh-force-off option disables EEH checking globally, for all slots.
- * Even if force-off is set, the EEH hardware is still enabled, so that
- * newer systems can boot.
+ * It's the platform's job to call this from an arch_initcall().
  */
-static int eeh_init(void)
+int eeh_init(struct eeh_ops *ops)
 {
 	struct pci_controller *hose, *tmp;
 	int ret = 0;
 
+	/* the platform should only initialise EEH once */
+	if (WARN_ON(eeh_ops))
+		return -EEXIST;
+	if (WARN_ON(!ops))
+		return -ENOENT;
+	eeh_ops = ops;
+
 	/* Register reboot notifier */
 	ret = register_reboot_notifier(&eeh_reboot_nb);
 	if (ret) {
@@ -1018,13 +965,13 @@ static int eeh_init(void)
 		return ret;
 	}
 
-	/* call platform initialization function */
-	if (!eeh_ops) {
-		pr_warn("%s: Platform EEH operation not found\n",
-			__func__);
-		return -EEXIST;
-	} else if ((ret = eeh_ops->init()))
+	if (eeh_ops->init)
+		ret = eeh_ops->init();
+	if (ret) {
+		pr_warn("%s: platform EEH init failed (%d)\n",
+			__func__, ret);
 		return ret;
+	}
 
 	/* Initialize PHB PEs */
 	list_for_each_entry_safe(hose, tmp, &hose_list, list_node)
@@ -1036,8 +983,6 @@ static int eeh_init(void)
 	return eeh_event_init();
 }
 
-core_initcall_sync(eeh_init);
-
 static int eeh_device_notifier(struct notifier_block *nb,
 			       unsigned long action, void *data)
 {
diff --git a/arch/powerpc/platforms/powernv/eeh-powernv.c b/arch/powerpc/platforms/powernv/eeh-powernv.c
index 9af8c3b98853..a550f837ccb5 100644
--- a/arch/powerpc/platforms/powernv/eeh-powernv.c
+++ b/arch/powerpc/platforms/powernv/eeh-powernv.c
@@ -1717,7 +1717,7 @@ static int __init eeh_powernv_init(void)
 {
 	int ret = -EINVAL;
 
-	ret = eeh_ops_register(&pnv_eeh_ops);
+	ret = eeh_init(&pnv_eeh_ops);
 	if (!ret)
 		pr_info("EEH: PowerNV platform initialized\n");
 	else
@@ -1725,4 +1725,4 @@ static int __init eeh_powernv_init(void)
 
 	return ret;
 }
-machine_early_initcall(powernv, eeh_powernv_init);
+machine_core_initcall_sync(powernv, eeh_powernv_init);
diff --git a/arch/powerpc/platforms/pseries/eeh_pseries.c b/arch/powerpc/platforms/pseries/eeh_pseries.c
index cb2d9a970b7b..3cc569e8b6d4 100644
--- a/arch/powerpc/platforms/pseries/eeh_pseries.c
+++ b/arch/powerpc/platforms/pseries/eeh_pseries.c
@@ -994,13 +994,12 @@ static int __init eeh_pseries_init(void)
 {
 	int ret;
 
-	ret = eeh_ops_register(&pseries_eeh_ops);
+	ret = eeh_init(&pseries_eeh_ops);
 	if (!ret)
 		pr_info("EEH: pSeries platform initialized\n");
 	else
 		pr_info("EEH: pSeries platform initialization failure (%d)\n",
 			ret);
-
 	return ret;
 }
-machine_early_initcall(pseries, eeh_pseries_init);
+machine_core_initcall_sync(pseries, eeh_pseries_init);
-- 
2.26.2

[PATCH 2/9] powerpc/powernv: Stop using eeh_ops->init()

[Oliver O'Halloran]

Fold pnv_eeh_init() into eeh_powernv_init() rather than having eeh_init()
call it via eeh_ops->init(). It's simpler and it'll let us delete
eeh_ops.init.

Signed-off-by: Oliver O'Halloran <oohall@gmail.com>
---
 arch/powerpc/platforms/powernv/eeh-powernv.c | 94 ++++++++++----------
 1 file changed, 45 insertions(+), 49 deletions(-)

diff --git a/arch/powerpc/platforms/powernv/eeh-powernv.c b/arch/powerpc/platforms/powernv/eeh-powernv.c
index a550f837ccb5..03e566874595 100644
--- a/arch/powerpc/platforms/powernv/eeh-powernv.c
+++ b/arch/powerpc/platforms/powernv/eeh-powernv.c
@@ -44,54 +44,6 @@ void pnv_pcibios_bus_add_device(struct pci_dev *pdev)
 	eeh_probe_device(pdev);
 }
 
-static int pnv_eeh_init(void)
-{
-	struct pci_controller *hose;
-	struct pnv_phb *phb;
-	int max_diag_size = PNV_PCI_DIAG_BUF_SIZE;
-
-	if (!firmware_has_feature(FW_FEATURE_OPAL)) {
-		pr_warn("%s: OPAL is required !\n",
-			__func__);
-		return -EINVAL;
-	}
-
-	/* Set probe mode */
-	eeh_add_flag(EEH_PROBE_MODE_DEV);
-
-	/*
-	 * P7IOC blocks PCI config access to frozen PE, but PHB3
-	 * doesn't do that. So we have to selectively enable I/O
-	 * prior to collecting error log.
-	 */
-	list_for_each_entry(hose, &hose_list, list_node) {
-		phb = hose->private_data;
-
-		if (phb->model == PNV_PHB_MODEL_P7IOC)
-			eeh_add_flag(EEH_ENABLE_IO_FOR_LOG);
-
-		if (phb->diag_data_size > max_diag_size)
-			max_diag_size = phb->diag_data_size;
-
-		/*
-		 * PE#0 should be regarded as valid by EEH core
-		 * if it's not the reserved one. Currently, we
-		 * have the reserved PE#255 and PE#127 for PHB3
-		 * and P7IOC separately. So we should regard
-		 * PE#0 as valid for PHB3 and P7IOC.
-		 */
-		if (phb->ioda.reserved_pe_idx != 0)
-			eeh_add_flag(EEH_VALID_PE_ZERO);
-
-		break;
-	}
-
-	eeh_set_pe_aux_size(max_diag_size);
-	ppc_md.pcibios_bus_add_device = pnv_pcibios_bus_add_device;
-
-	return 0;
-}
-
 static irqreturn_t pnv_eeh_event(int irq, void *data)
 {
 	/*
@@ -1674,7 +1626,6 @@ static int pnv_eeh_restore_config(struct eeh_dev *edev)
 
 static struct eeh_ops pnv_eeh_ops = {
 	.name                   = "powernv",
-	.init                   = pnv_eeh_init,
 	.probe			= pnv_eeh_probe,
 	.set_option             = pnv_eeh_set_option,
 	.get_state              = pnv_eeh_get_state,
@@ -1715,8 +1666,53 @@ DECLARE_PCI_FIXUP_HEADER(PCI_ANY_ID, PCI_ANY_ID, pnv_pci_fixup_vf_mps);
  */
 static int __init eeh_powernv_init(void)
 {
+	int max_diag_size = PNV_PCI_DIAG_BUF_SIZE;
+	struct pci_controller *hose;
+	struct pnv_phb *phb;
 	int ret = -EINVAL;
 
+	if (!firmware_has_feature(FW_FEATURE_OPAL)) {
+		pr_warn("%s: OPAL is required !\n", __func__);
+		return -EINVAL;
+	}
+
+	/* Set probe mode */
+	eeh_add_flag(EEH_PROBE_MODE_DEV);
+
+	/*
+	 * P7IOC blocks PCI config access to frozen PE, but PHB3
+	 * doesn't do that. So we have to selectively enable I/O
+	 * prior to collecting error log.
+	 */
+	list_for_each_entry(hose, &hose_list, list_node) {
+		phb = hose->private_data;
+
+		if (phb->model == PNV_PHB_MODEL_P7IOC)
+			eeh_add_flag(EEH_ENABLE_IO_FOR_LOG);
+
+		if (phb->diag_data_size > max_diag_size)
+			max_diag_size = phb->diag_data_size;
+
+		/*
+		 * PE#0 should be regarded as valid by EEH core
+		 * if it's not the reserved one. Currently, we
+		 * have the reserved PE#255 and PE#127 for PHB3
+		 * and P7IOC separately. So we should regard
+		 * PE#0 as valid for PHB3 and P7IOC.
+		 */
+		if (phb->ioda.reserved_pe_idx != 0)
+			eeh_add_flag(EEH_VALID_PE_ZERO);
+
+		break;
+	}
+
+	/*
+	 * eeh_init() allocates the eeh_pe and its aux data buf so the
+	 * size needs to be set before calling eeh_init().
+	 */
+	eeh_set_pe_aux_size(max_diag_size);
+	ppc_md.pcibios_bus_add_device = pnv_pcibios_bus_add_device;
+
 	ret = eeh_init(&pnv_eeh_ops);
 	if (!ret)
 		pr_info("EEH: PowerNV platform initialized\n");
-- 
2.26.2

[PATCH 3/9] powerpc/pseries: Stop using eeh_ops->init()

[Oliver O'Halloran]

Fold pseries_eeh_init() into eeh_pseries_init() rather than having
eeh_init() call it via eeh_ops->init(). It's simpler and it'll let us
delete eeh_ops.init.

Signed-off-by: Oliver O'Halloran <oohall@gmail.com>
---
 arch/powerpc/platforms/pseries/eeh_pseries.c | 155 +++++++++----------
 1 file changed, 71 insertions(+), 84 deletions(-)

diff --git a/arch/powerpc/platforms/pseries/eeh_pseries.c b/arch/powerpc/platforms/pseries/eeh_pseries.c
index 3cc569e8b6d4..fd3444428632 100644
--- a/arch/powerpc/platforms/pseries/eeh_pseries.c
+++ b/arch/powerpc/platforms/pseries/eeh_pseries.c
@@ -239,88 +239,6 @@ static unsigned char slot_errbuf[RTAS_ERROR_LOG_MAX];
 static DEFINE_SPINLOCK(slot_errbuf_lock);
 static int eeh_error_buf_size;
 
-/**
- * pseries_eeh_init - EEH platform dependent initialization
- *
- * EEH platform dependent initialization on pseries.
- */
-static int pseries_eeh_init(void)
-{
-	struct pci_controller *phb;
-	struct pci_dn *pdn;
-	int addr, config_addr;
-
-	/* figure out EEH RTAS function call tokens */
-	ibm_set_eeh_option		= rtas_token("ibm,set-eeh-option");
-	ibm_set_slot_reset		= rtas_token("ibm,set-slot-reset");
-	ibm_read_slot_reset_state2	= rtas_token("ibm,read-slot-reset-state2");
-	ibm_read_slot_reset_state	= rtas_token("ibm,read-slot-reset-state");
-	ibm_slot_error_detail		= rtas_token("ibm,slot-error-detail");
-	ibm_get_config_addr_info2	= rtas_token("ibm,get-config-addr-info2");
-	ibm_get_config_addr_info	= rtas_token("ibm,get-config-addr-info");
-	ibm_configure_pe		= rtas_token("ibm,configure-pe");
-
-	/*
-	 * ibm,configure-pe and ibm,configure-bridge have the same semantics,
-	 * however ibm,configure-pe can be faster.  If we can't find
-	 * ibm,configure-pe then fall back to using ibm,configure-bridge.
-	 */
-	if (ibm_configure_pe == RTAS_UNKNOWN_SERVICE)
-		ibm_configure_pe 	= rtas_token("ibm,configure-bridge");
-
-	/*
-	 * Necessary sanity check. We needn't check "get-config-addr-info"
-	 * and its variant since the old firmware probably support address
-	 * of domain/bus/slot/function for EEH RTAS operations.
-	 */
-	if (ibm_set_eeh_option == RTAS_UNKNOWN_SERVICE		||
-	    ibm_set_slot_reset == RTAS_UNKNOWN_SERVICE		||
-	    (ibm_read_slot_reset_state2 == RTAS_UNKNOWN_SERVICE &&
-	     ibm_read_slot_reset_state == RTAS_UNKNOWN_SERVICE)	||
-	    ibm_slot_error_detail == RTAS_UNKNOWN_SERVICE	||
-	    ibm_configure_pe == RTAS_UNKNOWN_SERVICE) {
-		pr_info("EEH functionality not supported\n");
-		return -EINVAL;
-	}
-
-	/* Initialize error log lock and size */
-	spin_lock_init(&slot_errbuf_lock);
-	eeh_error_buf_size = rtas_token("rtas-error-log-max");
-	if (eeh_error_buf_size == RTAS_UNKNOWN_SERVICE) {
-		pr_info("%s: unknown EEH error log size\n",
-			__func__);
-		eeh_error_buf_size = 1024;
-	} else if (eeh_error_buf_size > RTAS_ERROR_LOG_MAX) {
-		pr_info("%s: EEH error log size %d exceeds the maximal %d\n",
-			__func__, eeh_error_buf_size, RTAS_ERROR_LOG_MAX);
-		eeh_error_buf_size = RTAS_ERROR_LOG_MAX;
-	}
-
-	/* Set EEH probe mode */
-	eeh_add_flag(EEH_PROBE_MODE_DEVTREE | EEH_ENABLE_IO_FOR_LOG);
-
-	/* Set EEH machine dependent code */
-	ppc_md.pcibios_bus_add_device = pseries_pcibios_bus_add_device;
-
-	if (is_kdump_kernel() || reset_devices) {
-		pr_info("Issue PHB reset ...\n");
-		list_for_each_entry(phb, &hose_list, list_node) {
-			pdn = list_first_entry(&PCI_DN(phb->dn)->child_list, struct pci_dn, list);
-			addr = (pdn->busno << 16) | (pdn->devfn << 8);
-			config_addr = pseries_eeh_get_config_addr(phb, addr);
-			/* invalid PE config addr */
-			if (config_addr == 0)
-				continue;
-
-			pseries_eeh_phb_reset(phb, config_addr, EEH_RESET_FUNDAMENTAL);
-			pseries_eeh_phb_reset(phb, config_addr, EEH_RESET_DEACTIVATE);
-			pseries_eeh_phb_configure_bridge(phb, config_addr);
-		}
-	}
-
-	return 0;
-}
-
 static int pseries_eeh_cap_start(struct pci_dn *pdn)
 {
 	u32 status;
@@ -967,7 +885,6 @@ static int pseries_notify_resume(struct eeh_dev *edev)
 
 static struct eeh_ops pseries_eeh_ops = {
 	.name			= "pseries",
-	.init			= pseries_eeh_init,
 	.probe			= pseries_eeh_probe,
 	.set_option		= pseries_eeh_set_option,
 	.get_state		= pseries_eeh_get_state,
@@ -992,7 +909,77 @@ static struct eeh_ops pseries_eeh_ops = {
  */
 static int __init eeh_pseries_init(void)
 {
-	int ret;
+	struct pci_controller *phb;
+	struct pci_dn *pdn;
+	int ret, addr, config_addr;
+
+	/* figure out EEH RTAS function call tokens */
+	ibm_set_eeh_option		= rtas_token("ibm,set-eeh-option");
+	ibm_set_slot_reset		= rtas_token("ibm,set-slot-reset");
+	ibm_read_slot_reset_state2	= rtas_token("ibm,read-slot-reset-state2");
+	ibm_read_slot_reset_state	= rtas_token("ibm,read-slot-reset-state");
+	ibm_slot_error_detail		= rtas_token("ibm,slot-error-detail");
+	ibm_get_config_addr_info2	= rtas_token("ibm,get-config-addr-info2");
+	ibm_get_config_addr_info	= rtas_token("ibm,get-config-addr-info");
+	ibm_configure_pe		= rtas_token("ibm,configure-pe");
+
+	/*
+	 * ibm,configure-pe and ibm,configure-bridge have the same semantics,
+	 * however ibm,configure-pe can be faster.  If we can't find
+	 * ibm,configure-pe then fall back to using ibm,configure-bridge.
+	 */
+	if (ibm_configure_pe == RTAS_UNKNOWN_SERVICE)
+		ibm_configure_pe	= rtas_token("ibm,configure-bridge");
+
+	/*
+	 * Necessary sanity check. We needn't check "get-config-addr-info"
+	 * and its variant since the old firmware probably support address
+	 * of domain/bus/slot/function for EEH RTAS operations.
+	 */
+	if (ibm_set_eeh_option == RTAS_UNKNOWN_SERVICE		||
+	    ibm_set_slot_reset == RTAS_UNKNOWN_SERVICE		||
+	    (ibm_read_slot_reset_state2 == RTAS_UNKNOWN_SERVICE &&
+	     ibm_read_slot_reset_state == RTAS_UNKNOWN_SERVICE)	||
+	    ibm_slot_error_detail == RTAS_UNKNOWN_SERVICE	||
+	    ibm_configure_pe == RTAS_UNKNOWN_SERVICE) {
+		pr_info("EEH functionality not supported\n");
+		return -EINVAL;
+	}
+
+	/* Initialize error log lock and size */
+	spin_lock_init(&slot_errbuf_lock);
+	eeh_error_buf_size = rtas_token("rtas-error-log-max");
+	if (eeh_error_buf_size == RTAS_UNKNOWN_SERVICE) {
+		pr_info("%s: unknown EEH error log size\n",
+			__func__);
+		eeh_error_buf_size = 1024;
+	} else if (eeh_error_buf_size > RTAS_ERROR_LOG_MAX) {
+		pr_info("%s: EEH error log size %d exceeds the maximal %d\n",
+			__func__, eeh_error_buf_size, RTAS_ERROR_LOG_MAX);
+		eeh_error_buf_size = RTAS_ERROR_LOG_MAX;
+	}
+
+	/* Set EEH probe mode */
+	eeh_add_flag(EEH_PROBE_MODE_DEVTREE | EEH_ENABLE_IO_FOR_LOG);
+
+	/* Set EEH machine dependent code */
+	ppc_md.pcibios_bus_add_device = pseries_pcibios_bus_add_device;
+
+	if (is_kdump_kernel() || reset_devices) {
+		pr_info("Issue PHB reset ...\n");
+		list_for_each_entry(phb, &hose_list, list_node) {
+			pdn = list_first_entry(&PCI_DN(phb->dn)->child_list, struct pci_dn, list);
+			addr = (pdn->busno << 16) | (pdn->devfn << 8);
+			config_addr = pseries_eeh_get_config_addr(phb, addr);
+			/* invalid PE config addr */
+			if (config_addr == 0)
+				continue;
+
+			pseries_eeh_phb_reset(phb, config_addr, EEH_RESET_FUNDAMENTAL);
+			pseries_eeh_phb_reset(phb, config_addr, EEH_RESET_DEACTIVATE);
+			pseries_eeh_phb_configure_bridge(phb, config_addr);
+		}
+	}
 
 	ret = eeh_init(&pseries_eeh_ops);
 	if (!ret)
-- 
2.26.2

[PATCH 4/9] powerpc/eeh: Delete eeh_ops->init

[Oliver O'Halloran]

No longer used since the platforms perform their EEH initialisation before
calling eeh_init().

Signed-off-by: Oliver O'Halloran <oohall@gmail.com>
---
 arch/powerpc/include/asm/eeh.h | 1 -
 arch/powerpc/kernel/eeh.c      | 8 --------
 2 files changed, 9 deletions(-)

diff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h
index 765bcf63edea..85030c05e67e 100644
--- a/arch/powerpc/include/asm/eeh.h
+++ b/arch/powerpc/include/asm/eeh.h
@@ -216,7 +216,6 @@ enum {
 
 struct eeh_ops {
 	char *name;
-	int (*init)(void);
 	struct eeh_dev *(*probe)(struct pci_dev *pdev);
 	int (*set_option)(struct eeh_pe *pe, int option);
 	int (*get_state)(struct eeh_pe *pe, int *delay);
diff --git a/arch/powerpc/kernel/eeh.c b/arch/powerpc/kernel/eeh.c
index 28a0ea5d9faa..98faf139e676 100644
--- a/arch/powerpc/kernel/eeh.c
+++ b/arch/powerpc/kernel/eeh.c
@@ -965,14 +965,6 @@ int eeh_init(struct eeh_ops *ops)
 		return ret;
 	}
 
-	if (eeh_ops->init)
-		ret = eeh_ops->init();
-	if (ret) {
-		pr_warn("%s: platform EEH init failed (%d)\n",
-			__func__, ret);
-		return ret;
-	}
-
 	/* Initialize PHB PEs */
 	list_for_each_entry_safe(hose, tmp, &hose_list, list_node)
 		eeh_phb_pe_create(hose);
-- 
2.26.2

[PATCH 5/9] powerpc/eeh: Move EEH initialisation to an arch initcall

[Oliver O'Halloran]

The initialisation of EEH mostly happens in a core_initcall_sync initcall,
followed by registering a bus notifier later on in an arch_initcall.
Anything involving initcall dependecies is mostly incomprehensible unless
you've spent a while staring at code so here's the full sequence:

ppc_md.setup_arch       <-- pci_controllers are created here

...time passes...

core_initcall           <-- pci_dns are created from DT nodes
core_initcall_sync      <-- platforms call eeh_init()
postcore_initcall       <-- PCI bus type is registered
postcore_initcall_sync
arch_initcall           <-- EEH pci_bus notifier registered
subsys_initcall         <-- PHBs are scanned here

There's no real requirement to do the EEH setup at the core_initcall_sync
level. It just needs to be done after pci_dn's are created and before we
start scanning PHBs. Simplify the flow a bit by moving the platform EEH
inititalisation to an arch_initcall so we can fold the bus notifier
registration into eeh_init().

Signed-off-by: Oliver O'Halloran <oohall@gmail.com>
---
 arch/powerpc/kernel/eeh.c                    | 64 ++++++++++----------
 arch/powerpc/platforms/powernv/eeh-powernv.c |  2 +-
 arch/powerpc/platforms/pseries/eeh_pseries.c |  2 +-
 3 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/arch/powerpc/kernel/eeh.c b/arch/powerpc/kernel/eeh.c
index 98faf139e676..c9e25cfce8f0 100644
--- a/arch/powerpc/kernel/eeh.c
+++ b/arch/powerpc/kernel/eeh.c
@@ -940,6 +940,30 @@ static struct notifier_block eeh_reboot_nb = {
 	.notifier_call = eeh_reboot_notifier,
 };
 
+static int eeh_device_notifier(struct notifier_block *nb,
+			       unsigned long action, void *data)
+{
+	struct device *dev = data;
+
+	switch (action) {
+	/*
+	 * Note: It's not possible to perform EEH device addition (i.e.
+	 * {pseries,pnv}_pcibios_bus_add_device()) here because it depends on
+	 * the device's resources, which have not yet been set up.
+	 */
+	case BUS_NOTIFY_DEL_DEVICE:
+		eeh_remove_device(to_pci_dev(dev));
+		break;
+	default:
+		break;
+	}
+	return NOTIFY_DONE;
+}
+
+static struct notifier_block eeh_device_nb = {
+	.notifier_call = eeh_device_notifier,
+};
+
 /**
  * eeh_init - System wide EEH initialization
  *
@@ -960,7 +984,14 @@ int eeh_init(struct eeh_ops *ops)
 	/* Register reboot notifier */
 	ret = register_reboot_notifier(&eeh_reboot_nb);
 	if (ret) {
-		pr_warn("%s: Failed to register notifier (%d)\n",
+		pr_warn("%s: Failed to register reboot notifier (%d)\n",
+			__func__, ret);
+		return ret;
+	}
+
+	ret = bus_register_notifier(&pci_bus_type, &eeh_device_nb);
+	if (ret) {
+		pr_warn("%s: Failed to register bus notifier (%d)\n",
 			__func__, ret);
 		return ret;
 	}
@@ -975,37 +1006,6 @@ int eeh_init(struct eeh_ops *ops)
 	return eeh_event_init();
 }
 
-static int eeh_device_notifier(struct notifier_block *nb,
-			       unsigned long action, void *data)
-{
-	struct device *dev = data;
-
-	switch (action) {
-	/*
-	 * Note: It's not possible to perform EEH device addition (i.e.
-	 * {pseries,pnv}_pcibios_bus_add_device()) here because it depends on
-	 * the device's resources, which have not yet been set up.
-	 */
-	case BUS_NOTIFY_DEL_DEVICE:
-		eeh_remove_device(to_pci_dev(dev));
-		break;
-	default:
-		break;
-	}
-	return NOTIFY_DONE;
-}
-
-static struct notifier_block eeh_device_nb = {
-	.notifier_call = eeh_device_notifier,
-};
-
-static __init int eeh_set_bus_notifier(void)
-{
-	bus_register_notifier(&pci_bus_type, &eeh_device_nb);
-	return 0;
-}
-arch_initcall(eeh_set_bus_notifier);
-
 /**
  * eeh_probe_device() - Perform EEH initialization for the indicated pci device
  * @dev: pci device for which to set up EEH
diff --git a/arch/powerpc/platforms/powernv/eeh-powernv.c b/arch/powerpc/platforms/powernv/eeh-powernv.c
index 03e566874595..d03c5873defc 100644
--- a/arch/powerpc/platforms/powernv/eeh-powernv.c
+++ b/arch/powerpc/platforms/powernv/eeh-powernv.c
@@ -1721,4 +1721,4 @@ static int __init eeh_powernv_init(void)
 
 	return ret;
 }
-machine_core_initcall_sync(powernv, eeh_powernv_init);
+machine_arch_initcall(powernv, eeh_powernv_init);
diff --git a/arch/powerpc/platforms/pseries/eeh_pseries.c b/arch/powerpc/platforms/pseries/eeh_pseries.c
index fd3444428632..b1561961c7ff 100644
--- a/arch/powerpc/platforms/pseries/eeh_pseries.c
+++ b/arch/powerpc/platforms/pseries/eeh_pseries.c
@@ -989,4 +989,4 @@ static int __init eeh_pseries_init(void)
 			ret);
 	return ret;
 }
-machine_core_initcall_sync(pseries, eeh_pseries_init);
+machine_arch_initcall(pseries, eeh_pseries_init);
-- 
2.26.2

[PATCH 6/9] powerpc/pseries/eeh: Clean up pe_config_addr lookups

[Oliver O'Halloran]

De-duplicate, and fix up the comments, and make the prototype just take a
pci_dn since the job of the function is to return the pe_config_addr of the
PE which contains a given device.

Signed-off-by: Oliver O'Halloran <oohall@gmail.com>
---
 arch/powerpc/platforms/pseries/eeh_pseries.c | 80 +++-----------------
 1 file changed, 11 insertions(+), 69 deletions(-)

diff --git a/arch/powerpc/platforms/pseries/eeh_pseries.c b/arch/powerpc/platforms/pseries/eeh_pseries.c
index b1561961c7ff..10303de3d8d5 100644
--- a/arch/powerpc/platforms/pseries/eeh_pseries.c
+++ b/arch/powerpc/platforms/pseries/eeh_pseries.c
@@ -33,8 +33,6 @@
 #include <asm/ppc-pci.h>
 #include <asm/rtas.h>
 
-static int pseries_eeh_get_pe_addr(struct pci_dn *pdn);
-
 /* RTAS tokens */
 static int ibm_set_eeh_option;
 static int ibm_set_slot_reset;
@@ -86,7 +84,8 @@ void pseries_pcibios_bus_add_device(struct pci_dev *pdev)
 
 
 /**
- * pseries_eeh_get_config_addr - Retrieve config address
+ * pseries_eeh_get_pe_config_addr - Find the pe_config_addr for a device
+ * @pdn: pci_dn of the input device
  *
  * Retrieve the assocated config address. Actually, there're 2 RTAS
  * function calls dedicated for the purpose. We need implement
@@ -97,16 +96,17 @@ void pseries_pcibios_bus_add_device(struct pci_dev *pdev)
  * It's notable that zero'ed return value means invalid PE config
  * address.
  */
-static int pseries_eeh_get_config_addr(struct pci_controller *phb, int config_addr)
+static int pseries_eeh_get_pe_config_addr(struct pci_dn *pdn)
 {
+	int config_addr = rtas_config_addr(pdn->busno, pdn->devfn, 0);
+	struct pci_controller *phb = pdn->phb;
 	int ret = 0;
 	int rets[3];
 
 	if (ibm_get_config_addr_info2 != RTAS_UNKNOWN_SERVICE) {
 		/*
-		 * First of all, we need to make sure there has one PE
-		 * associated with the device. Otherwise, PE address is
-		 * meaningless.
+		 * First of all, use function 1 to determine if this device is
+		 * part of a PE or not. ret[0] being zero indicates it's not.
 		 */
 		ret = rtas_call(ibm_get_config_addr_info2, 4, 2, rets,
 				config_addr, BUID_HI(phb->buid),
@@ -431,7 +431,7 @@ void pseries_eeh_init_edev(struct pci_dn *pdn)
 		struct eeh_pe *parent;
 
 		/* Retrieve PE address */
-		edev->pe_config_addr = pseries_eeh_get_pe_addr(pdn);
+		edev->pe_config_addr = pseries_eeh_get_pe_config_addr(pdn);
 		pe.addr = edev->pe_config_addr;
 
 		/* Some older systems (Power4) allow the ibm,set-eeh-option
@@ -551,64 +551,6 @@ static int pseries_eeh_set_option(struct eeh_pe *pe, int option)
 	return ret;
 }
 
-/**
- * pseries_eeh_get_pe_addr - Retrieve PE address
- * @pe: EEH PE
- *
- * Retrieve the assocated PE address. Actually, there're 2 RTAS
- * function calls dedicated for the purpose. We need implement
- * it through the new function and then the old one. Besides,
- * you should make sure the config address is figured out from
- * FDT node before calling the function.
- *
- * It's notable that zero'ed return value means invalid PE config
- * address.
- */
-static int pseries_eeh_get_pe_addr(struct pci_dn *pdn)
-{
-	int config_addr = rtas_config_addr(pdn->busno, pdn->devfn, 0);
-	unsigned long buid = pdn->phb->buid;
-	int ret = 0;
-	int rets[3];
-
-	if (ibm_get_config_addr_info2 != RTAS_UNKNOWN_SERVICE) {
-		/*
-		 * First of all, we need to make sure there has one PE
-		 * associated with the device. Otherwise, PE address is
-		 * meaningless.
-		 */
-		ret = rtas_call(ibm_get_config_addr_info2, 4, 2, rets,
-				config_addr, BUID_HI(buid), BUID_LO(buid), 1);
-		if (ret || (rets[0] == 0))
-			return 0;
-
-		/* Retrieve the associated PE config address */
-		ret = rtas_call(ibm_get_config_addr_info2, 4, 2, rets,
-				config_addr, BUID_HI(buid), BUID_LO(buid), 0);
-		if (ret) {
-			pr_warn("%s: Failed to get address for PHB#%x-PE#%x\n",
-				__func__, pdn->phb->global_number, config_addr);
-			return 0;
-		}
-
-		return rets[0];
-	}
-
-	if (ibm_get_config_addr_info != RTAS_UNKNOWN_SERVICE) {
-		ret = rtas_call(ibm_get_config_addr_info, 4, 2, rets,
-				config_addr, BUID_HI(buid), BUID_LO(buid), 0);
-		if (ret) {
-			pr_warn("%s: Failed to get address for PHB#%x-PE#%x\n",
-				__func__, pdn->phb->global_number, config_addr);
-			return 0;
-		}
-
-		return rets[0];
-	}
-
-	return ret;
-}
-
 /**
  * pseries_eeh_get_state - Retrieve PE state
  * @pe: EEH PE
@@ -911,7 +853,7 @@ static int __init eeh_pseries_init(void)
 {
 	struct pci_controller *phb;
 	struct pci_dn *pdn;
-	int ret, addr, config_addr;
+	int ret, config_addr;
 
 	/* figure out EEH RTAS function call tokens */
 	ibm_set_eeh_option		= rtas_token("ibm,set-eeh-option");
@@ -969,8 +911,8 @@ static int __init eeh_pseries_init(void)
 		pr_info("Issue PHB reset ...\n");
 		list_for_each_entry(phb, &hose_list, list_node) {
 			pdn = list_first_entry(&PCI_DN(phb->dn)->child_list, struct pci_dn, list);
-			addr = (pdn->busno << 16) | (pdn->devfn << 8);
-			config_addr = pseries_eeh_get_config_addr(phb, addr);
+			config_addr = pseries_eeh_get_pe_config_addr(pdn);
+
 			/* invalid PE config addr */
 			if (config_addr == 0)
 				continue;
-- 
2.26.2

[PATCH 7/9] powerpc/pseries/eeh: Rework device EEH PE determination

[Oliver O'Halloran]

The process Linux uses for determining if a device supports EEH or not
appears to be at odds with what PAPR+ says the OS should be doing. The
current flow is something like:

1. Assume pe_config_addr is equal the the device's config_addr.
2. Attempt to enable EEH on that PE
3. Verify EEH was enabled (POWER4 bug workaround)
4. Try find the pe_config_addr using the ibm,get-config-addr-info2 RTAS
   call.
5. If that fails walk the pci_dn tree upwards trying to find a parent
   device with EEH support. If we find one then add the device to that PE.

The first major flaw with this is that in order to enable EEH on a PE we
need to know the PE's configuration address since that's an input to the
ibm,set-eeh-option RTAS call which is used to enable EEH for the PE. We
hack around that by assuming that the PE address is equal to the device's
RTAS config address with the register fields set to zero (see
rtas_config_addr()). This assumption happens to be valid if:

a) The PCI device is the 0th function, and
b) The device is on the PE's root bus.

However, it this does also appear to work for devices where these
conditions are not true. At a guess PowerVM's RTAS has some workarounds to
accommodate Linux's quirks. However, it's a bit sketch and the code is
confusing since it's not implementing what PAPR claims is the correct way.

This patch re-works how we handle EEH init so that we find the PE config
address using the ibm,get-config-addr-info2 RTAS call, then use that to
finish the EEH init process. It also drops the Power4 workaround since as
of commit 471d7ff8b51b ("powerpc/64s: Remove POWER4 support") the kernel
does not support running on a Power4 CPU.

1. Find the pe_config_addr using the RTAS call.
2. Enable the PE (if needed)
3. Insert the edev into the tree and create an eeh_pe if needed.

The other change made here is ignoring unsupported devices entirely.
Currently the device's BARs are saved to the eeh_dev even if the device is
not part of an EEH PE. Not being part of a PE means that an EEH recovery
pass will never see that device so the saving the BARs is pointless.

Signed-off-by: Oliver O'Halloran <oohall@gmail.com>
---
 arch/powerpc/platforms/pseries/eeh_pseries.c | 57 ++++++++------------
 1 file changed, 22 insertions(+), 35 deletions(-)

diff --git a/arch/powerpc/platforms/pseries/eeh_pseries.c b/arch/powerpc/platforms/pseries/eeh_pseries.c
index 10303de3d8d5..c2ecc0db2f94 100644
--- a/arch/powerpc/platforms/pseries/eeh_pseries.c
+++ b/arch/powerpc/platforms/pseries/eeh_pseries.c
@@ -357,10 +357,10 @@ static struct eeh_pe *pseries_eeh_pe_get_parent(struct eeh_dev *edev)
  */
 void pseries_eeh_init_edev(struct pci_dn *pdn)
 {
+	struct eeh_pe pe, *parent;
 	struct eeh_dev *edev;
-	struct eeh_pe pe;
+	int addr;
 	u32 pcie_flags;
-	int enable = 0;
 	int ret;
 
 	if (WARN_ON_ONCE(!eeh_has_flag(EEH_PROBE_MODE_DEVTREE)))
@@ -417,51 +417,38 @@ void pseries_eeh_init_edev(struct pci_dn *pdn)
 		}
 	}
 
-	/* Initialize the fake PE */
+	/* first up, find the pe_config_addr for the PE containing the device */
+	addr = pseries_eeh_get_pe_config_addr(pdn);
+	if (addr == 0) {
+		eeh_edev_dbg(edev, "Unable to find pe_config_addr\n");
+		goto err;
+	}
+
+	/* Try enable EEH on the fake PE */
 	memset(&pe, 0, sizeof(struct eeh_pe));
 	pe.phb = pdn->phb;
-	pe.config_addr = (pdn->busno << 16) | (pdn->devfn << 8);
+	pe.addr = addr;
 
-	/* Enable EEH on the device */
 	eeh_edev_dbg(edev, "Enabling EEH on device\n");
 	ret = eeh_ops->set_option(&pe, EEH_OPT_ENABLE);
 	if (ret) {
 		eeh_edev_dbg(edev, "EEH failed to enable on device (code %d)\n", ret);
-	} else {
-		struct eeh_pe *parent;
+		goto err;
+	}
 
-		/* Retrieve PE address */
-		edev->pe_config_addr = pseries_eeh_get_pe_config_addr(pdn);
-		pe.addr = edev->pe_config_addr;
+	edev->pe_config_addr = addr;
 
-		/* Some older systems (Power4) allow the ibm,set-eeh-option
-		 * call to succeed even on nodes where EEH is not supported.
-		 * Verify support explicitly.
-		 */
-		ret = eeh_ops->get_state(&pe, NULL);
-		if (ret > 0 && ret != EEH_STATE_NOT_SUPPORT)
-			enable = 1;
+	eeh_add_flag(EEH_ENABLED);
 
-		/*
-		 * This device doesn't support EEH, but it may have an
-		 * EEH parent. In this case any error on the device will
-		 * freeze the PE of it's upstream bridge, so added it to
-		 * the upstream PE.
-		 */
-		parent = pseries_eeh_pe_get_parent(edev);
-		if (parent && !enable)
-			edev->pe_config_addr = parent->addr;
+	parent = pseries_eeh_pe_get_parent(edev);
+	eeh_pe_tree_insert(edev, parent);
+	eeh_save_bars(edev);
+	eeh_edev_dbg(edev, "EEH enabled for device");
 
-		if (enable || parent) {
-			eeh_add_flag(EEH_ENABLED);
-			eeh_pe_tree_insert(edev, parent);
-		}
-		eeh_edev_dbg(edev, "EEH is %s on device (code %d)\n",
-			     (enable ? "enabled" : "unsupported"), ret);
-	}
+	return;
 
-	/* Save memory bars */
-	eeh_save_bars(edev);
+err:
+	eeh_edev_dbg(edev, "EEH is unsupported on device (code = %d)\n", ret);
 }
 
 static struct eeh_dev *pseries_eeh_probe(struct pci_dev *pdev)
-- 
2.26.2

[PATCH 8/9] powerpc/pseries/eeh: Allow zero to be a valid PE configuration address

[Oliver O'Halloran]

There's no real reason why zero can't be a valid PE configuration address.
Under qemu each sPAPR PHB (i.e. EEH supporting) has the passed-though
devices on bus zero, so the PE address of bus <dddd>:00 should be zero.

However, all previous versions of Linux will reject that, so Qemu at least
goes out of it's way to avoid it. The Qemu implementation of
ibm,get-config-addr-info2 RTAS has the following comment:

> /*
>  * We always have PE address of form "00BB0001". "BB"
>  * represents the bus number of PE's primary bus.
>  */

So qemu puts a one into the register portion of the PE's config_addr to
avoid it being zero. The whole is pretty silly considering that RTAS will
return a negative error code if it can't map the device's config_addr to a
PE's.

Fix Linux to treat zero as a valid PE address. This shouldn't have any real
effects due to the Qemu hack mentioned above, and the fact that this has
worked historically on PowerVM means they never pass through devices on bus
zero.

Signed-off-by: Oliver O'Halloran <oohall@gmail.com>
---
 arch/powerpc/platforms/pseries/eeh_pseries.c | 38 +++++++++++---------
 1 file changed, 22 insertions(+), 16 deletions(-)

diff --git a/arch/powerpc/platforms/pseries/eeh_pseries.c b/arch/powerpc/platforms/pseries/eeh_pseries.c
index c2ecc0db2f94..e42c026392aa 100644
--- a/arch/powerpc/platforms/pseries/eeh_pseries.c
+++ b/arch/powerpc/platforms/pseries/eeh_pseries.c
@@ -87,21 +87,20 @@ void pseries_pcibios_bus_add_device(struct pci_dev *pdev)
  * pseries_eeh_get_pe_config_addr - Find the pe_config_addr for a device
  * @pdn: pci_dn of the input device
  *
- * Retrieve the assocated config address. Actually, there're 2 RTAS
- * function calls dedicated for the purpose. We need implement
- * it through the new function and then the old one. Besides,
- * you should make sure the config address is figured out from
- * FDT node before calling the function.
+ * The EEH RTAS calls use a tuple consisting of: (buid_hi, buid_lo,
+ * pe_config_addr) as a handle to a given PE. This function finds the
+ * pe_config_addr based on the device's config addr.
  *
- * It's notable that zero'ed return value means invalid PE config
- * address.
+ * Keep in mind that the pe_config_addr *might* be numerically identical to the
+ * device's config addr, but the two are conceptually distinct.
+ *
+ * Returns the pe_config_addr, or a negative error code.
  */
 static int pseries_eeh_get_pe_config_addr(struct pci_dn *pdn)
 {
 	int config_addr = rtas_config_addr(pdn->busno, pdn->devfn, 0);
 	struct pci_controller *phb = pdn->phb;
-	int ret = 0;
-	int rets[3];
+	int ret, rets[3];
 
 	if (ibm_get_config_addr_info2 != RTAS_UNKNOWN_SERVICE) {
 		/*
@@ -112,16 +111,16 @@ static int pseries_eeh_get_pe_config_addr(struct pci_dn *pdn)
 				config_addr, BUID_HI(phb->buid),
 				BUID_LO(phb->buid), 1);
 		if (ret || (rets[0] == 0))
-			return 0;
+			return -ENOENT;
 
-		/* Retrieve the associated PE config address */
+		/* Retrieve the associated PE config address with function 0 */
 		ret = rtas_call(ibm_get_config_addr_info2, 4, 2, rets,
 				config_addr, BUID_HI(phb->buid),
 				BUID_LO(phb->buid), 0);
 		if (ret) {
 			pr_warn("%s: Failed to get address for PHB#%x-PE#%x\n",
 				__func__, phb->global_number, config_addr);
-			return 0;
+			return -ENXIO;
 		}
 
 		return rets[0];
@@ -134,13 +133,20 @@ static int pseries_eeh_get_pe_config_addr(struct pci_dn *pdn)
 		if (ret) {
 			pr_warn("%s: Failed to get address for PHB#%x-PE#%x\n",
 				__func__, phb->global_number, config_addr);
-			return 0;
+			return -ENXIO;
 		}
 
 		return rets[0];
 	}
 
-	return ret;
+	/*
+	 * PAPR does describe a process for finding the pe_config_addr that was
+	 * used before the ibm,get-config-addr-info calls were added. However,
+	 * I haven't found *any* systems that don't have that RTAS call
+	 * implemented. If you happen to find one that needs the old DT based
+	 * process, patches are welcome!
+	 */
+	return -ENOENT;
 }
 
 /**
@@ -419,7 +425,7 @@ void pseries_eeh_init_edev(struct pci_dn *pdn)
 
 	/* first up, find the pe_config_addr for the PE containing the device */
 	addr = pseries_eeh_get_pe_config_addr(pdn);
-	if (addr == 0) {
+	if (addr < 0) {
 		eeh_edev_dbg(edev, "Unable to find pe_config_addr\n");
 		goto err;
 	}
@@ -901,7 +907,7 @@ static int __init eeh_pseries_init(void)
 			config_addr = pseries_eeh_get_pe_config_addr(pdn);
 
 			/* invalid PE config addr */
-			if (config_addr == 0)
+			if (config_addr < 0)
 				continue;
 
 			pseries_eeh_phb_reset(phb, config_addr, EEH_RESET_FUNDAMENTAL);
-- 
2.26.2

[PATCH 9/9] powerpc/eeh: Clean up PE addressing

[Oliver O'Halloran]

At some point in the distant past we only supported EEH on pseries. The
various EEH RTAS call use the "PE config address" as a handle to the PE
being manipulated so we need to find that address a PE.

There's three ways to determine the address of a PE starting from a device
inside of that PE: The old way, which requires traversing the DT until you
find a built-in device and finding its PE using the
ibm,read-slot-reset-state2 RTAS call. The new way, which requires using the
ibm,get-config-addr-info RTAS call to map the device address to a PE
address. And then there's the Linux way, which is "blindly assume the
device and PE addresses are the same." Naturally, PAPR doesn't recommend
(or even mention) that last one, but it's a technique
Linux on Power has used since the dawn of time.

Most (all?) modern pseries systems will provide the get-addr-info RTAS call
so Linux uses that for everything except for the initial ibm,eeh-set-option
RTAS call to enable EEH on the PE. The Linux way is still broken in that
case, but it seems to work so who maybe firmware have hacks to support it,
who knows. For systems that don't support the RTAS call we'll use the Linux
way as a fallback. For some reason we don't just use the fallback address
to initialise eeh_dev->addr and instead eeh_pe has two addresses and we'll
choose which one to use at runtime. This results in code that looks
something like:

	config_addr = pe->config_addr;
	if (pe->addr)
	        config_addr = pe->addr;

	rtas_call(..., config_addr, ...);

In other words, if the result of the RTAS call is non-zero then Linux will
use that as the pe address. If not, it falls back to using the config_addr.
It's worth pointing out that both fields here used to be part of pci_dn and
ended up in eeh_pe after a while.

Storing both addresses in eeh_pe doesn't really make a whole lot of sense.
Why does the eeh_pe structure, which is platform independent, have two
addresses baked into it for the sake of a pseries platform quirk? Why
doesn't the pseries platform handle determining what the "correct" PE
address is during EEH initialisation for the device and just pass that
address to the EEH core? What does it even mean for a pe to have an
"address" if there's two of them?

There are no good answers to these questions; especially that last one. The
EEH core makes a token effort to support looking up a PE by either address
by having two arguments to eeh_pe_get(). However, a survey of all the
callers to eeh_pe_get() shows that all except one hard-code zero as the
config_addr argument. The only one that doesn't is in eeh_pe_tree_insert()a
which looks like this:

	if (!eeh_has_flag(EEH_VALID_PE_ZERO) && !edev->pe_config_addr)
		return -EINVAL;

	pe = eeh_pe_get(hose, edev->pe_config_addr, edev->bdfn);

The third argument (config_addr) is only used if the second (pe->addr)
argument is invalid. In this case that would require edev->pe_config_addr
to be zero and the EEH_VALID_PE_ZERO flag to be unset. The preceding check
ensure that can never be true so there is no situation where eeh_pe_get()
will search for a PE with the specified pe->config_addr.

Similarly, on pseries the EEH_VALID_PE_ZERO flag isn't set so the check
above also ensures that there will never be a PE with pe->addr == 0. As a
result all the logic to choose whether we pass pe->config_addr or pe->addr
to an RTAS call is also dead code. The pe->config_addr will never be used
since pe->addr must be non-zero. Otherwise it wouldn't be in the PE tree.

This patch tries to clean up this mess by:

1) Removing pe->config_addr
2) Removing the EEH_VALID_PE_ZERO flag
3) Removing the fallback address argument to eeh_pe_get().
4) Removing all the checks for pe->addr being zero in the pseries EEH code.

This leaves us with PE's only being identified by what's in their pe->addr
field and relying on the platform to ensure that eeh_dev's are only
inserted into the EEH tree if they're actually inside a PE.

Signed-off-by: Oliver O'Halloran <oohall@gmail.com>
---
 arch/powerpc/include/asm/eeh.h               |  4 +-
 arch/powerpc/kernel/eeh.c                    |  2 +-
 arch/powerpc/kernel/eeh_pe.c                 | 46 +++-----------------
 arch/powerpc/platforms/powernv/eeh-powernv.c | 16 ++-----
 arch/powerpc/platforms/pseries/eeh_pseries.c | 42 +++---------------
 5 files changed, 17 insertions(+), 93 deletions(-)

diff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h
index 85030c05e67e..dd6a4ac6c713 100644
--- a/arch/powerpc/include/asm/eeh.h
+++ b/arch/powerpc/include/asm/eeh.h
@@ -27,7 +27,6 @@ struct pci_dn;
 #define EEH_FORCE_DISABLED	0x02	/* EEH disabled			     */
 #define EEH_PROBE_MODE_DEV	0x04	/* From PCI device		     */
 #define EEH_PROBE_MODE_DEVTREE	0x08	/* From device tree		     */
-#define EEH_VALID_PE_ZERO	0x10	/* PE#0 is valid		     */
 #define EEH_ENABLE_IO_FOR_LOG	0x20	/* Enable IO for log		     */
 #define EEH_EARLY_DUMP_LOG	0x40	/* Dump log immediately		     */
 
@@ -280,8 +279,7 @@ int eeh_phb_pe_create(struct pci_controller *phb);
 int eeh_wait_state(struct eeh_pe *pe, int max_wait);
 struct eeh_pe *eeh_phb_pe_get(struct pci_controller *phb);
 struct eeh_pe *eeh_pe_next(struct eeh_pe *pe, struct eeh_pe *root);
-struct eeh_pe *eeh_pe_get(struct pci_controller *phb,
-			  int pe_no, int config_addr);
+struct eeh_pe *eeh_pe_get(struct pci_controller *phb, int pe_no);
 int eeh_pe_tree_insert(struct eeh_dev *edev, struct eeh_pe *new_pe_parent);
 int eeh_pe_tree_remove(struct eeh_dev *edev);
 void eeh_pe_update_time_stamp(struct eeh_pe *pe);
diff --git a/arch/powerpc/kernel/eeh.c b/arch/powerpc/kernel/eeh.c
index c9e25cfce8f0..87de8b798b2d 100644
--- a/arch/powerpc/kernel/eeh.c
+++ b/arch/powerpc/kernel/eeh.c
@@ -1657,7 +1657,7 @@ static ssize_t eeh_force_recover_write(struct file *filp,
 		return -ENODEV;
 
 	/* Retrieve PE */
-	pe = eeh_pe_get(hose, pe_no, 0);
+	pe = eeh_pe_get(hose, pe_no);
 	if (!pe)
 		return -ENODEV;
 
diff --git a/arch/powerpc/kernel/eeh_pe.c b/arch/powerpc/kernel/eeh_pe.c
index d2aaaa73fdd5..61b7d4019051 100644
--- a/arch/powerpc/kernel/eeh_pe.c
+++ b/arch/powerpc/kernel/eeh_pe.c
@@ -251,43 +251,21 @@ void eeh_pe_dev_traverse(struct eeh_pe *root,
 
 /**
  * __eeh_pe_get - Check the PE address
- * @data: EEH PE
- * @flag: EEH device
  *
  * For one particular PE, it can be identified by PE address
  * or tranditional BDF address. BDF address is composed of
  * Bus/Device/Function number. The extra data referred by flag
  * indicates which type of address should be used.
  */
-struct eeh_pe_get_flag {
-	int pe_no;
-	int config_addr;
-};
-
 static void *__eeh_pe_get(struct eeh_pe *pe, void *flag)
 {
-	struct eeh_pe_get_flag *tmp = (struct eeh_pe_get_flag *) flag;
+	int *target_pe = flag;
 
-	/* Unexpected PHB PE */
+	/* PHB PEs are special and should be ignored */
 	if (pe->type & EEH_PE_PHB)
 		return NULL;
 
-	/*
-	 * We prefer PE address. For most cases, we should
-	 * have non-zero PE address
-	 */
-	if (eeh_has_flag(EEH_VALID_PE_ZERO)) {
-		if (tmp->pe_no == pe->addr)
-			return pe;
-	} else {
-		if (tmp->pe_no &&
-		    (tmp->pe_no == pe->addr))
-			return pe;
-	}
-
-	/* Try BDF address */
-	if (tmp->config_addr &&
-	   (tmp->config_addr == pe->config_addr))
+	if (*target_pe == pe->addr)
 		return pe;
 
 	return NULL;
@@ -297,7 +275,6 @@ static void *__eeh_pe_get(struct eeh_pe *pe, void *flag)
  * eeh_pe_get - Search PE based on the given address
  * @phb: PCI controller
  * @pe_no: PE number
- * @config_addr: Config address
  *
  * Search the corresponding PE based on the specified address which
  * is included in the eeh device. The function is used to check if
@@ -306,16 +283,11 @@ static void *__eeh_pe_get(struct eeh_pe *pe, void *flag)
  * which is composed of PCI bus/device/function number, or unified
  * PE address.
  */
-struct eeh_pe *eeh_pe_get(struct pci_controller *phb,
-		int pe_no, int config_addr)
+struct eeh_pe *eeh_pe_get(struct pci_controller *phb, int pe_no)
 {
 	struct eeh_pe *root = eeh_phb_pe_get(phb);
-	struct eeh_pe_get_flag tmp = { pe_no, config_addr };
-	struct eeh_pe *pe;
-
-	pe = eeh_pe_traverse(root, __eeh_pe_get, &tmp);
 
-	return pe;
+	return eeh_pe_traverse(root, __eeh_pe_get, &pe_no);
 }
 
 /**
@@ -336,19 +308,13 @@ int eeh_pe_tree_insert(struct eeh_dev *edev, struct eeh_pe *new_pe_parent)
 	struct pci_controller *hose = edev->controller;
 	struct eeh_pe *pe, *parent;
 
-	/* Check if the PE number is valid */
-	if (!eeh_has_flag(EEH_VALID_PE_ZERO) && !edev->pe_config_addr) {
-		eeh_edev_err(edev, "PE#0 is invalid for this PHB!\n");
-		return -EINVAL;
-	}
-
 	/*
 	 * Search the PE has been existing or not according
 	 * to the PE address. If that has been existing, the
 	 * PE should be composed of PCI bus and its subordinate
 	 * components.
 	 */
-	pe = eeh_pe_get(hose, edev->pe_config_addr, edev->bdfn);
+	pe = eeh_pe_get(hose, edev->pe_config_addr);
 	if (pe) {
 		if (pe->type & EEH_PE_INVALID) {
 			list_add_tail(&edev->entry, &pe->edevs);
diff --git a/arch/powerpc/platforms/powernv/eeh-powernv.c b/arch/powerpc/platforms/powernv/eeh-powernv.c
index d03c5873defc..32e3ff28851b 100644
--- a/arch/powerpc/platforms/powernv/eeh-powernv.c
+++ b/arch/powerpc/platforms/powernv/eeh-powernv.c
@@ -87,7 +87,7 @@ static ssize_t pnv_eeh_ei_write(struct file *filp,
 		return -EINVAL;
 
 	/* Retrieve PE */
-	pe = eeh_pe_get(hose, pe_no, 0);
+	pe = eeh_pe_get(hose, pe_no);
 	if (!pe)
 		return -ENODEV;
 
@@ -306,7 +306,7 @@ static struct eeh_pe *pnv_eeh_get_upstream_pe(struct pci_dev *pdev)
 	if (parent) {
 		struct pnv_ioda_pe *ioda_pe = pnv_ioda_get_pe(parent);
 
-		return eeh_pe_get(phb->hose, ioda_pe->pe_number, 0);
+		return eeh_pe_get(phb->hose, ioda_pe->pe_number);
 	}
 
 	return NULL;
@@ -1358,7 +1358,7 @@ static int pnv_eeh_get_pe(struct pci_controller *hose,
 	}
 
 	/* Find the PE according to PE# */
-	dev_pe = eeh_pe_get(hose, pe_no, 0);
+	dev_pe = eeh_pe_get(hose, pe_no);
 	if (!dev_pe)
 		return -EEXIST;
 
@@ -1693,16 +1693,6 @@ static int __init eeh_powernv_init(void)
 		if (phb->diag_data_size > max_diag_size)
 			max_diag_size = phb->diag_data_size;
 
-		/*
-		 * PE#0 should be regarded as valid by EEH core
-		 * if it's not the reserved one. Currently, we
-		 * have the reserved PE#255 and PE#127 for PHB3
-		 * and P7IOC separately. So we should regard
-		 * PE#0 as valid for PHB3 and P7IOC.
-		 */
-		if (phb->ioda.reserved_pe_idx != 0)
-			eeh_add_flag(EEH_VALID_PE_ZERO);
-
 		break;
 	}
 
diff --git a/arch/powerpc/platforms/pseries/eeh_pseries.c b/arch/powerpc/platforms/pseries/eeh_pseries.c
index e42c026392aa..2ebfe9d8bf8c 100644
--- a/arch/powerpc/platforms/pseries/eeh_pseries.c
+++ b/arch/powerpc/platforms/pseries/eeh_pseries.c
@@ -511,7 +511,6 @@ EXPORT_SYMBOL_GPL(pseries_eeh_init_edev_recursive);
 static int pseries_eeh_set_option(struct eeh_pe *pe, int option)
 {
 	int ret = 0;
-	int config_addr;
 
 	/*
 	 * When we're enabling or disabling EEH functioality on
@@ -524,9 +523,6 @@ static int pseries_eeh_set_option(struct eeh_pe *pe, int option)
 	case EEH_OPT_ENABLE:
 	case EEH_OPT_THAW_MMIO:
 	case EEH_OPT_THAW_DMA:
-		config_addr = pe->config_addr;
-		if (pe->addr)
-			config_addr = pe->addr;
 		break;
 	case EEH_OPT_FREEZE_PE:
 		/* Not support */
@@ -538,7 +534,7 @@ static int pseries_eeh_set_option(struct eeh_pe *pe, int option)
 	}
 
 	ret = rtas_call(ibm_set_eeh_option, 4, 1, NULL,
-			config_addr, BUID_HI(pe->phb->buid),
+			pe->addr, BUID_HI(pe->phb->buid),
 			BUID_LO(pe->phb->buid), option);
 
 	return ret;
@@ -559,25 +555,19 @@ static int pseries_eeh_set_option(struct eeh_pe *pe, int option)
  */
 static int pseries_eeh_get_state(struct eeh_pe *pe, int *delay)
 {
-	int config_addr;
 	int ret;
 	int rets[4];
 	int result;
 
-	/* Figure out PE config address if possible */
-	config_addr = pe->config_addr;
-	if (pe->addr)
-		config_addr = pe->addr;
-
 	if (ibm_read_slot_reset_state2 != RTAS_UNKNOWN_SERVICE) {
 		ret = rtas_call(ibm_read_slot_reset_state2, 3, 4, rets,
-				config_addr, BUID_HI(pe->phb->buid),
+				pe->addr, BUID_HI(pe->phb->buid),
 				BUID_LO(pe->phb->buid));
 	} else if (ibm_read_slot_reset_state != RTAS_UNKNOWN_SERVICE) {
 		/* Fake PE unavailable info */
 		rets[2] = 0;
 		ret = rtas_call(ibm_read_slot_reset_state, 3, 3, rets,
-				config_addr, BUID_HI(pe->phb->buid),
+				pe->addr, BUID_HI(pe->phb->buid),
 				BUID_LO(pe->phb->buid));
 	} else {
 		return EEH_STATE_NOT_SUPPORT;
@@ -631,14 +621,7 @@ static int pseries_eeh_get_state(struct eeh_pe *pe, int *delay)
  */
 static int pseries_eeh_reset(struct eeh_pe *pe, int option)
 {
-	int config_addr;
-
-	/* Figure out PE address */
-	config_addr = pe->config_addr;
-	if (pe->addr)
-		config_addr = pe->addr;
-
-	return pseries_eeh_phb_reset(pe->phb, config_addr, option);
+	return pseries_eeh_phb_reset(pe->phb, pe->addr, option);
 }
 
 /**
@@ -654,19 +637,13 @@ static int pseries_eeh_reset(struct eeh_pe *pe, int option)
  */
 static int pseries_eeh_get_log(struct eeh_pe *pe, int severity, char *drv_log, unsigned long len)
 {
-	int config_addr;
 	unsigned long flags;
 	int ret;
 
 	spin_lock_irqsave(&slot_errbuf_lock, flags);
 	memset(slot_errbuf, 0, eeh_error_buf_size);
 
-	/* Figure out the PE address */
-	config_addr = pe->config_addr;
-	if (pe->addr)
-		config_addr = pe->addr;
-
-	ret = rtas_call(ibm_slot_error_detail, 8, 1, NULL, config_addr,
+	ret = rtas_call(ibm_slot_error_detail, 8, 1, NULL, pe->addr,
 			BUID_HI(pe->phb->buid), BUID_LO(pe->phb->buid),
 			virt_to_phys(drv_log), len,
 			virt_to_phys(slot_errbuf), eeh_error_buf_size,
@@ -685,14 +662,7 @@ static int pseries_eeh_get_log(struct eeh_pe *pe, int severity, char *drv_log, u
  */
 static int pseries_eeh_configure_bridge(struct eeh_pe *pe)
 {
-	int config_addr;
-
-	/* Figure out the PE address */
-	config_addr = pe->config_addr;
-	if (pe->addr)
-		config_addr = pe->addr;
-
-	return pseries_eeh_phb_configure_bridge(pe->phb, config_addr);
+	return pseries_eeh_phb_configure_bridge(pe->phb, pe->addr);
 }
 
 /**
-- 
2.26.2