[PATCH 0/8] powerpc/powernv: PCI code cleanup

[PATCH 0/8] powerpc/powernv: PCI code cleanup

[Gavin Shan]

The patchset cleans up PowerNV PCI code. Also, the compound PEs are
affected to each other with help of PELTV, which was missed from
commit 262af557 ("powerpc/powernv: Enable M64 aperatus for PHB3").

Gavin Shan (8):
  powerpc/powernv: Check PHB type in advance
  powerpc/powernv: Fix condition to remove M64
  powerpc/powernv: Rename alloc_m64_pe() to reserve_m64_pe()
  powerpc/powernv: Initialize M64 PE in time
  powerpc/powernv: Set PELTV for compound PEs
  powerpc/powernv: Simplify pnv_ioda_configure_pe()
  powerpc/powernv: Bail upon invalid master PE
  powerpc/powernv: Fix potential zero devisor

 arch/powerpc/platforms/powernv/pci-ioda.c   | 205 +++++++++++++++++++++-------
 arch/powerpc/platforms/powernv/pci-p5ioc2.c |  28 ++--
 arch/powerpc/platforms/powernv/pci.h        |   2 +-
 3 files changed, 172 insertions(+), 63 deletions(-)

-- 
1.8.3.2

[PATCH 1/8] powerpc/powernv: Check PHB type in advance

[Gavin Shan]

The patch checks PHB type a bit early to save a bit cycles
for P7 because we don't support M64 for P7IOC no matter what
OPAL firmware we have.

Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
---
 arch/powerpc/platforms/powernv/pci-ioda.c | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/arch/powerpc/platforms/powernv/pci-ioda.c b/arch/powerpc/platforms/powernv/pci-ioda.c
index 468a0f2..7ab1dd7 100644
--- a/arch/powerpc/platforms/powernv/pci-ioda.c
+++ b/arch/powerpc/platforms/powernv/pci-ioda.c
@@ -313,6 +313,12 @@ static void __init pnv_ioda_parse_m64_window(struct pnv_phb *phb)
 	const u32 *r;
 	u64 pci_addr;
 
+	/* FIXME: Support M64 for P7IOC */
+	if (phb->type != PNV_PHB_IODA2) {
+		pr_info("  Not support M64 window\n");
+		return;
+	}
+
 	if (!firmware_has_feature(FW_FEATURE_OPALv3)) {
 		pr_info("  Firmware too old to support M64 window\n");
 		return;
@@ -325,12 +331,6 @@ static void __init pnv_ioda_parse_m64_window(struct pnv_phb *phb)
 		return;
 	}
 
-	/* FIXME: Support M64 for P7IOC */
-	if (phb->type != PNV_PHB_IODA2) {
-		pr_info("  Not support M64 window\n");
-		return;
-	}
-
 	res = &hose->mem_resources[1];
 	res->start = of_translate_address(dn, r + 2);
 	res->end = res->start + of_read_number(r + 4, 2) - 1;
-- 
1.8.3.2

[PATCH 2/8] powerpc/powernv: Fix condition to remove M64

[Gavin Shan]

The M64 resource should be removed if we don't have hook to
initialize it, or (not and) fail to do that.

Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
---
 arch/powerpc/platforms/powernv/pci-ioda.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/arch/powerpc/platforms/powernv/pci-ioda.c b/arch/powerpc/platforms/powernv/pci-ioda.c
index 7ab1dd7..7aa0404 100644
--- a/arch/powerpc/platforms/powernv/pci-ioda.c
+++ b/arch/powerpc/platforms/powernv/pci-ioda.c
@@ -2000,8 +2000,8 @@ static void __init pnv_pci_init_ioda_phb(struct device_node *np,
 		ioda_eeh_phb_reset(hose, OPAL_DEASSERT_RESET);
 	}
 
-	/* Configure M64 window */
-	if (phb->init_m64 && phb->init_m64(phb))
+	/* Remove M64 resource if we can't configure it successfully */
+	if (!phb->init_m64 || phb->init_m64(phb))
 		hose->mem_resources[1].flags = 0;
 }
 
-- 
1.8.3.2

[PATCH 3/8] powerpc/powernv: Rename alloc_m64_pe() to reserve_m64_pe()

[Gavin Shan]

The patch renames alloc_m64_pe() to reserve_m64_pe() to reflect
its real usage: We reserve PE numbers for M64 segments in advance
and then pick up the reserved PE numbers when building the mapping
between PE numbers and M64 segments.

Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
---
 arch/powerpc/platforms/powernv/pci-ioda.c | 8 ++++----
 arch/powerpc/platforms/powernv/pci.h      | 2 +-
 2 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/arch/powerpc/platforms/powernv/pci-ioda.c b/arch/powerpc/platforms/powernv/pci-ioda.c
index 7aa0404..cf90cce 100644
--- a/arch/powerpc/platforms/powernv/pci-ioda.c
+++ b/arch/powerpc/platforms/powernv/pci-ioda.c
@@ -172,7 +172,7 @@ fail:
 	return -EIO;
 }
 
-static void pnv_ioda2_alloc_m64_pe(struct pnv_phb *phb)
+static void pnv_ioda2_reserve_m64_pe(struct pnv_phb *phb)
 {
 	resource_size_t sgsz = phb->ioda.m64_segsize;
 	struct pci_dev *pdev;
@@ -345,7 +345,7 @@ static void __init pnv_ioda_parse_m64_window(struct pnv_phb *phb)
 	/* Use last M64 BAR to cover M64 window */
 	phb->ioda.m64_bar_idx = 15;
 	phb->init_m64 = pnv_ioda2_init_m64;
-	phb->alloc_m64_pe = pnv_ioda2_alloc_m64_pe;
+	phb->reserve_m64_pe = pnv_ioda2_reserve_m64_pe;
 	phb->pick_m64_pe = pnv_ioda2_pick_m64_pe;
 }
 
@@ -837,8 +837,8 @@ static void pnv_pci_ioda_setup_PEs(void)
 		phb = hose->private_data;
 
 		/* M64 layout might affect PE allocation */
-		if (phb->alloc_m64_pe)
-			phb->alloc_m64_pe(phb);
+		if (phb->reserve_m64_pe)
+			phb->reserve_m64_pe(phb);
 
 		pnv_ioda_setup_PEs(hose->bus);
 	}
diff --git a/arch/powerpc/platforms/powernv/pci.h b/arch/powerpc/platforms/powernv/pci.h
index 34d29eb..6c02ff8 100644
--- a/arch/powerpc/platforms/powernv/pci.h
+++ b/arch/powerpc/platforms/powernv/pci.h
@@ -130,7 +130,7 @@ struct pnv_phb {
 	u32 (*bdfn_to_pe)(struct pnv_phb *phb, struct pci_bus *bus, u32 devfn);
 	void (*shutdown)(struct pnv_phb *phb);
 	int (*init_m64)(struct pnv_phb *phb);
-	void (*alloc_m64_pe)(struct pnv_phb *phb);
+	void (*reserve_m64_pe)(struct pnv_phb *phb);
 	int (*pick_m64_pe)(struct pnv_phb *phb, struct pci_bus *bus, int all);
 	int (*get_pe_state)(struct pnv_phb *phb, int pe_no);
 	void (*freeze_pe)(struct pnv_phb *phb, int pe_no);
-- 
1.8.3.2

[PATCH 4/8] powerpc/powernv: Initialize M64 PE in time

[Gavin Shan]

The patch initializes PE instance when reserving PE number to
keep consistent things as we did before. Also, it replaces the
iteration on bridge's windows with the prefered way.

Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
---
 arch/powerpc/platforms/powernv/pci-ioda.c | 27 +++++++++++++++++++++------
 1 file changed, 21 insertions(+), 6 deletions(-)

diff --git a/arch/powerpc/platforms/powernv/pci-ioda.c b/arch/powerpc/platforms/powernv/pci-ioda.c
index cf90cce..7a7a688 100644
--- a/arch/powerpc/platforms/powernv/pci-ioda.c
+++ b/arch/powerpc/platforms/powernv/pci-ioda.c
@@ -91,6 +91,24 @@ static inline bool pnv_pci_is_mem_pref_64(unsigned long flags)
 		(IORESOURCE_MEM_64 | IORESOURCE_PREFETCH));
 }
 
+static void pnv_ioda_reserve_pe(struct pnv_phb *phb, int pe_no)
+{
+	if (!(pe_no >= 0 && pe_no < phb->ioda.total_pe)) {
+		pr_warn("%s: Invalid PE %d on PHB#%x\n",
+			__func__, pe_no, phb->hose->global_number);
+		return;
+	}
+
+	if (test_and_set_bit(pe_no, phb->ioda.pe_alloc)) {
+		pr_warn("%s: PE %d was assigned on PHB#%x\n",
+			__func__, pe_no, phb->hose->global_number);
+		return;
+	}
+
+	phb->ioda.pe_array[pe_no].phb = phb;
+	phb->ioda.pe_array[pe_no].pe_number = pe_no;
+}
+
 static int pnv_ioda_alloc_pe(struct pnv_phb *phb)
 {
 	unsigned long pe;
@@ -185,16 +203,15 @@ static void pnv_ioda2_reserve_m64_pe(struct pnv_phb *phb)
 	 * instead of root bus.
 	 */
 	list_for_each_entry(pdev, &phb->hose->bus->devices, bus_list) {
-		for (i = PCI_BRIDGE_RESOURCES;
-		     i <= PCI_BRIDGE_RESOURCE_END; i++) {
-			r = &pdev->resource[i];
+		for (i = 0; i < PCI_BRIDGE_RESOURCE_NUM; i++) {
+			r = &pdev->resource[PCI_BRIDGE_RESOURCES + i];
 			if (!r->parent ||
 			    !pnv_pci_is_mem_pref_64(r->flags))
 				continue;
 
 			base = (r->start - phb->ioda.m64_base) / sgsz;
 			for (step = 0; step < resource_size(r) / sgsz; step++)
-				set_bit(base + step, phb->ioda.pe_alloc);
+				pnv_ioda_reserve_pe(phb, base + step);
 		}
 	}
 }
@@ -287,8 +304,6 @@ done:
 	while ((i = find_next_bit(pe_alloc, phb->ioda.total_pe, i + 1)) <
 		phb->ioda.total_pe) {
 		pe = &phb->ioda.pe_array[i];
-		pe->phb = phb;
-		pe->pe_number = i;
 
 		if (!master_pe) {
 			pe->flags |= PNV_IODA_PE_MASTER;
-- 
1.8.3.2

[PATCH 5/8] powerpc/powernv: Set PELTV for compound PEs

[Gavin Shan]

Commit 262af55 ("powerpc/powernv: Enable M64 aperatus for PHB3")
introduced compound PEs in order to support M64 aperatus on PHB3.
However, we never configured PELTV for compound PEs. The patch
fixes that by: parent PE can freeze all child compound PEs. Any
compound PE affects the group.

Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
---
 arch/powerpc/platforms/powernv/pci-ioda.c | 118 ++++++++++++++++++++++++++----
 1 file changed, 102 insertions(+), 16 deletions(-)

diff --git a/arch/powerpc/platforms/powernv/pci-ioda.c b/arch/powerpc/platforms/powernv/pci-ioda.c
index 7a7a688..b96ba48 100644
--- a/arch/powerpc/platforms/powernv/pci-ioda.c
+++ b/arch/powerpc/platforms/powernv/pci-ioda.c
@@ -522,6 +522,106 @@ static struct pnv_ioda_pe *pnv_ioda_get_pe(struct pci_dev *dev)
 }
 #endif /* CONFIG_PCI_MSI */
 
+static int pnv_ioda_set_one_peltv(struct pnv_phb *phb,
+				  struct pnv_ioda_pe *parent,
+				  struct pnv_ioda_pe *child,
+				  bool is_add)
+{
+	const char *desc = is_add ? "adding" : "removing";
+	uint8_t op = is_add ? OPAL_ADD_PE_TO_DOMAIN :
+			      OPAL_REMOVE_PE_FROM_DOMAIN;
+	struct pnv_ioda_pe *slave;
+	long rc;
+
+	/* Parent PE affects child PE */
+	rc = opal_pci_set_peltv(phb->opal_id, parent->pe_number,
+				child->pe_number, op);
+	if (rc != OPAL_SUCCESS) {
+		pe_warn(child, "OPAL error %ld %s to parent PELTV\n",
+			rc, desc);
+		return -ENXIO;
+	}
+
+	if (!(child->flags & PNV_IODA_PE_MASTER))
+		return 0;
+
+	/* Compound case: parent PE affects slave PEs */
+	list_for_each_entry(slave, &child->slaves, list) {
+		rc = opal_pci_set_peltv(phb->opal_id, parent->pe_number,
+					slave->pe_number, op);
+		if (rc != OPAL_SUCCESS) {
+			pe_warn(slave, "OPAL error %ld %s to parent PELTV\n",
+				rc, desc);
+			return -ENXIO;
+		}
+	}
+
+	return 0;
+}
+
+static int pnv_ioda_set_peltv(struct pnv_phb *phb,
+			      struct pnv_ioda_pe *pe,
+			      bool is_add)
+{
+	struct pnv_ioda_pe *slave;
+	struct pci_dev *pdev;
+	int ret;
+
+	/*
+	 * Clear PE frozen state. If it's master PE, we need
+	 * clear slave PE frozen state as well.
+	 */
+	if (is_add) {
+		opal_pci_eeh_freeze_clear(phb->opal_id, pe->pe_number,
+					  OPAL_EEH_ACTION_CLEAR_FREEZE_ALL);
+		if (pe->flags & PNV_IODA_PE_MASTER) {
+			list_for_each_entry(slave, &pe->slaves, list)
+				opal_pci_eeh_freeze_clear(phb->opal_id,
+							  slave->pe_number,
+							  OPAL_EEH_ACTION_CLEAR_FREEZE_ALL);
+		}
+	}
+
+	/*
+	 * Associate PE in PELT. We need add the PE into the
+	 * corresponding PELT-V as well. Otherwise, the error
+	 * originated from the PE might contribute to other
+	 * PEs.
+	 */
+	ret = pnv_ioda_set_one_peltv(phb, pe, pe, is_add);
+	if (ret)
+		return ret;
+
+	/* For compound PEs, any one affects all of them */
+	if (pe->flags & PNV_IODA_PE_MASTER) {
+		list_for_each_entry(slave, &pe->slaves, list) {
+			ret = pnv_ioda_set_one_peltv(phb, slave, pe, is_add);
+			if (ret)
+				return ret;
+		}
+	}
+
+	if (pe->flags & (PNV_IODA_PE_BUS_ALL | PNV_IODA_PE_BUS))
+		pdev = pe->pbus->self;
+	else
+		pdev = pe->pdev->bus->self;
+	while (pdev) {
+		struct pci_dn *pdn = pci_get_pdn(pdev);
+		struct pnv_ioda_pe *parent;
+
+		if (pdn && pdn->pe_number != IODA_INVALID_PE) {
+			parent = &phb->ioda.pe_array[pdn->pe_number];
+			ret = pnv_ioda_set_one_peltv(phb, parent, pe, is_add);
+			if (ret)
+				return ret;
+		}
+
+		pdev = pdev->bus->self;
+	}
+
+	return 0;
+}
+
 static int pnv_ioda_configure_pe(struct pnv_phb *phb, struct pnv_ioda_pe *pe)
 {
 	struct pci_dev *parent;
@@ -576,23 +676,9 @@ static int pnv_ioda_configure_pe(struct pnv_phb *phb, struct pnv_ioda_pe *pe)
 		return -ENXIO;
 	}
 
-	rc = opal_pci_set_peltv(phb->opal_id, pe->pe_number,
-				pe->pe_number, OPAL_ADD_PE_TO_DOMAIN);
-	if (rc)
-		pe_warn(pe, "OPAL error %d adding self to PELTV\n", rc);
-	opal_pci_eeh_freeze_clear(phb->opal_id, pe->pe_number,
-				  OPAL_EEH_ACTION_CLEAR_FREEZE_ALL);
+	/* Configure PELTV */
+	pnv_ioda_set_peltv(phb, pe, true);
 
-	/* Add to all parents PELT-V */
-	while (parent) {
-		struct pci_dn *pdn = pci_get_pdn(parent);
-		if (pdn && pdn->pe_number != IODA_INVALID_PE) {
-			rc = opal_pci_set_peltv(phb->opal_id, pdn->pe_number,
-						pe->pe_number, OPAL_ADD_PE_TO_DOMAIN);
-			/* XXX What to do in case of error ? */
-		}
-		parent = parent->bus->self;
-	}
 	/* Setup reverse map */
 	for (rid = pe->rid; rid < rid_end; rid++)
 		phb->ioda.pe_rmap[rid] = pe->pe_number;
-- 
1.8.3.2

[PATCH 6/8] powerpc/powernv: Simplify pnv_ioda_configure_pe()

[Gavin Shan]

Nested if statements are always bad and the patch avoids one by
checking PHB type and bail in advance if necessary.

Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
---
 arch/powerpc/platforms/powernv/pci-ioda.c | 32 ++++++++++++++++---------------
 1 file changed, 17 insertions(+), 15 deletions(-)

diff --git a/arch/powerpc/platforms/powernv/pci-ioda.c b/arch/powerpc/platforms/powernv/pci-ioda.c
index b96ba48..762ca14 100644
--- a/arch/powerpc/platforms/powernv/pci-ioda.c
+++ b/arch/powerpc/platforms/powernv/pci-ioda.c
@@ -684,26 +684,28 @@ static int pnv_ioda_configure_pe(struct pnv_phb *phb, struct pnv_ioda_pe *pe)
 		phb->ioda.pe_rmap[rid] = pe->pe_number;
 
 	/* Setup one MVTs on IODA1 */
-	if (phb->type == PNV_PHB_IODA1) {
-		pe->mve_number = pe->pe_number;
-		rc = opal_pci_set_mve(phb->opal_id, pe->mve_number,
-				      pe->pe_number);
+	if (phb->type != PNV_PHB_IODA1) {
+		pe->mve_number = 0;
+		goto out;
+	}
+
+	pe->mve_number = pe->pe_number;
+	rc = opal_pci_set_mve(phb->opal_id, pe->mve_number, pe->pe_number);
+	if (rc != OPAL_SUCCESS) {
+		pe_err(pe, "OPAL error %ld setting up MVE %d\n",
+		       rc, pe->mve_number);
+		pe->mve_number = -1;
+	} else {
+		rc = opal_pci_set_mve_enable(phb->opal_id,
+					     pe->mve_number, OPAL_ENABLE_MVE);
 		if (rc) {
-			pe_err(pe, "OPAL error %ld setting up MVE %d\n",
+			pe_err(pe, "OPAL error %ld enabling MVE %d\n",
 			       rc, pe->mve_number);
 			pe->mve_number = -1;
-		} else {
-			rc = opal_pci_set_mve_enable(phb->opal_id,
-						     pe->mve_number, OPAL_ENABLE_MVE);
-			if (rc) {
-				pe_err(pe, "OPAL error %ld enabling MVE %d\n",
-				       rc, pe->mve_number);
-				pe->mve_number = -1;
-			}
 		}
-	} else if (phb->type == PNV_PHB_IODA2)
-		pe->mve_number = 0;
+	}
 
+out:
 	return 0;
 }
 
-- 
1.8.3.2

[PATCH 7/8] powerpc/powernv: Bail upon invalid master PE

[Gavin Shan]

When freezing compound PEs in pnv_ioda_freeze_pe(), we should bail
upon illegal master PE. We needn't freeze slave PE because it should
have been put into frozen state by hardware.

Reported-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
---
 arch/powerpc/platforms/powernv/pci-ioda.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/arch/powerpc/platforms/powernv/pci-ioda.c b/arch/powerpc/platforms/powernv/pci-ioda.c
index 762ca14..d035035 100644
--- a/arch/powerpc/platforms/powernv/pci-ioda.c
+++ b/arch/powerpc/platforms/powernv/pci-ioda.c
@@ -373,7 +373,9 @@ static void pnv_ioda_freeze_pe(struct pnv_phb *phb, int pe_no)
 	/* Fetch master PE */
 	if (pe->flags & PNV_IODA_PE_SLAVE) {
 		pe = pe->master;
-		WARN_ON(!pe || !(pe->flags & PNV_IODA_PE_MASTER));
+		if (WARN_ON(!pe || !(pe->flags & PNV_IODA_PE_MASTER)))
+			return;
+
 		pe_no = pe->pe_number;
 	}
 
-- 
1.8.3.2

[PATCH 8/8] powerpc/powernv: Fix potential zero devisor

[Gavin Shan]

If there're no PHBs under P5IOC2 HUB device tree node, we should
bail early to avoid zero devisor and allocating TCE tables.

Reported-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
---
 arch/powerpc/platforms/powernv/pci-p5ioc2.c | 28 ++++++++++++++++------------
 1 file changed, 16 insertions(+), 12 deletions(-)

diff --git a/arch/powerpc/platforms/powernv/pci-p5ioc2.c b/arch/powerpc/platforms/powernv/pci-p5ioc2.c
index 94ce348..3336fcb 100644
--- a/arch/powerpc/platforms/powernv/pci-p5ioc2.c
+++ b/arch/powerpc/platforms/powernv/pci-p5ioc2.c
@@ -196,6 +196,22 @@ void __init pnv_pci_init_p5ioc2_hub(struct device_node *np)
 	hub_id = be64_to_cpup(prop64);
 	pr_info(" HUB-ID : 0x%016llx\n", hub_id);
 
+	/* Count child PHBs and calculate TCE space per PHB */
+	for_each_child_of_node(np, phbn) {
+		if (of_device_is_compatible(phbn, "ibm,p5ioc2-pcix") ||
+		    of_device_is_compatible(phbn, "ibm,p5ioc2-pciex"))
+			phb_count++;
+	}
+
+	if (phb_count <= 0) {
+		pr_info(" No PHBs for Hub %s\n", np->full_name);
+		return;
+	}
+
+	tce_per_phb = __rounddown_pow_of_two(P5IOC2_TCE_MEMORY / phb_count);
+	pr_info(" Allocating %lld MB of TCE memory per PHB\n",
+		tce_per_phb >> 20);
+
 	/* Currently allocate 16M of TCE memory for every Hub
 	 *
 	 * XXX TODO: Make it chip local if possible
@@ -215,18 +231,6 @@ void __init pnv_pci_init_p5ioc2_hub(struct device_node *np)
 		return;
 	}
 
-	/* Count child PHBs */
-	for_each_child_of_node(np, phbn) {
-		if (of_device_is_compatible(phbn, "ibm,p5ioc2-pcix") ||
-		    of_device_is_compatible(phbn, "ibm,p5ioc2-pciex"))
-			phb_count++;
-	}
-
-	/* Calculate how much TCE space we can give per PHB */
-	tce_per_phb = __rounddown_pow_of_two(P5IOC2_TCE_MEMORY / phb_count);
-	pr_info(" Allocating %lld MB of TCE memory per PHB\n",
-		tce_per_phb >> 20);
-
 	/* Initialize PHBs */
 	for_each_child_of_node(np, phbn) {
 		if (of_device_is_compatible(phbn, "ibm,p5ioc2-pcix") ||
-- 
1.8.3.2