[PATCH v06 0/9] powerpc/hotplug: Update affinity for migrated CPUs

[PATCH v06 0/9] powerpc/hotplug: Update affinity for migrated CPUs

[Michael Bringmann]

The migration of LPARs across Power systems affects many attributes
including that of the associativity of CPUs.  The patches in this
set execute when a system is coming up fresh upon a migration target.
They are intended to,

* Recognize changes to the associativity of CPUs recorded in internal
  data structures when compared to the latest copies in the device tree.
* Generate calls to other code layers to reset the data structures
  related to associativity of the CPUs.
* Re-register the 'changed' entities into the target system.
  Re-registration of CPUs mostly entails acting as if they have been
  newly hot-added into the target system.

Signed-off-by: Michael Bringmann <mwb@linux.vnet.ibm.com>

Michael Bringmann (9):
  hotplug/cpu: Conditionally acquire/release DRC index
  hotplug/cpu: Add operation queuing function
  hotplug/cpu: Provide CPU readd operation
  mobility/numa: Ensure numa update does not overlap
  numa: Disable/enable arch_update_cpu_topology
  pmt/numa: Disable arch_update_cpu_topology during CPU readd
  powerpc/rtas: Allow disabling rtas_event_scan
  hotplug/rtas: No rtas_event_scan during PMT update
  hotplug/pmt: Update topology after PMT
---
Changes in patch:
  -- Restructure and rearrange content of patches to co-locate
     similar or related modifications
  -- Rename pseries_update_drconf_cpu to pseries_update_processor
  -- Simplify code to update CPU nodes during mobility checks.
     Remove functions to generate extra HP_ELOG messages in favor
     of direct function calls to dlpar_cpu_readd_by_index.
  -- Revise code order in dlpar_cpu_readd_by_index() to present
     more appropriate error codes from underlying layers of the
     implementation.
  -- Add hotplug device lock around all property updates
  -- Add call to rebuild_sched_domains in case of changes
  -- Various code cleanups and compaction
  -- Rebase to 4.18-rc1 kernel
  -- Change operation to run CPU readd after end of migration store.
  -- Improve descriptive text
  -- Cleanup patch reference to outdated function

[PATCH v06 1/9] hotplug/cpu: Conditionally acquire/release DRC index

[Michael Bringmann]

powerpc/cpu: Modify dlpar_cpu_add and dlpar_cpu_remove to allow the
skipping of DRC index acquire or release operations during the CPU
add or remove operations.  This is intended to support subsequent
changes to provide a 'CPU readd' operation.

Signed-off-by: Michael Bringmann <mwb@linux.vnet.ibm.com>
---
Changes in patch:
  -- Move new validity check added to pseries_smp_notifier
     to another patch
---
 arch/powerpc/platforms/pseries/hotplug-cpu.c |   68 +++++++++++++++-----------
 1 file changed, 39 insertions(+), 29 deletions(-)

diff --git a/arch/powerpc/platforms/pseries/hotplug-cpu.c b/arch/powerpc/platforms/pseries/hotplug-cpu.c
index 6ef77ca..3632db2 100644
--- a/arch/powerpc/platforms/pseries/hotplug-cpu.c
+++ b/arch/powerpc/platforms/pseries/hotplug-cpu.c
@@ -432,7 +432,7 @@ static bool valid_cpu_drc_index(struct device_node *parent, u32 drc_index)
 	return found;
 }
 
-static ssize_t dlpar_cpu_add(u32 drc_index)
+static ssize_t dlpar_cpu_add(u32 drc_index, bool acquire_drc)
 {
 	struct device_node *dn, *parent;
 	int rc, saved_rc;
@@ -457,19 +457,22 @@ static ssize_t dlpar_cpu_add(u32 drc_index)
 		return -EINVAL;
 	}
 
-	rc = dlpar_acquire_drc(drc_index);
-	if (rc) {
-		pr_warn("Failed to acquire DRC, rc: %d, drc index: %x\n",
-			rc, drc_index);
-		of_node_put(parent);
-		return -EINVAL;
+	if (acquire_drc) {
+		rc = dlpar_acquire_drc(drc_index);
+		if (rc) {
+			pr_warn("Failed to acquire DRC, rc: %d, drc index: %x\n",
+				rc, drc_index);
+			of_node_put(parent);
+			return -EINVAL;
+		}
 	}
 
 	dn = dlpar_configure_connector(cpu_to_be32(drc_index), parent);
 	if (!dn) {
 		pr_warn("Failed call to configure-connector, drc index: %x\n",
 			drc_index);
-		dlpar_release_drc(drc_index);
+		if (acquire_drc)
+			dlpar_release_drc(drc_index);
 		of_node_put(parent);
 		return -EINVAL;
 	}
@@ -484,8 +487,9 @@ static ssize_t dlpar_cpu_add(u32 drc_index)
 		pr_warn("Failed to attach node %s, rc: %d, drc index: %x\n",
 			dn->name, rc, drc_index);
 
-		rc = dlpar_release_drc(drc_index);
-		if (!rc)
+		if (acquire_drc)
+			rc = dlpar_release_drc(drc_index);
+		if (!rc || acquire_drc)
 			dlpar_free_cc_nodes(dn);
 
 		return saved_rc;
@@ -498,7 +502,7 @@ static ssize_t dlpar_cpu_add(u32 drc_index)
 			dn->name, rc, drc_index);
 
 		rc = dlpar_detach_node(dn);
-		if (!rc)
+		if (!rc && acquire_drc)
 			dlpar_release_drc(drc_index);
 
 		return saved_rc;
@@ -566,7 +570,8 @@ static int dlpar_offline_cpu(struct device_node *dn)
 
 }
 
-static ssize_t dlpar_cpu_remove(struct device_node *dn, u32 drc_index)
+static ssize_t dlpar_cpu_remove(struct device_node *dn, u32 drc_index,
+				bool release_drc)
 {
 	int rc;
 
@@ -579,12 +584,14 @@ static ssize_t dlpar_cpu_remove(struct device_node *dn, u32 drc_index)
 		return -EINVAL;
 	}
 
-	rc = dlpar_release_drc(drc_index);
-	if (rc) {
-		pr_warn("Failed to release drc (%x) for CPU %s, rc: %d\n",
-			drc_index, dn->name, rc);
-		dlpar_online_cpu(dn);
-		return rc;
+	if (release_drc) {
+		rc = dlpar_release_drc(drc_index);
+		if (rc) {
+			pr_warn("Failed to release drc (%x) for CPU %s, rc: %d\n",
+				drc_index, dn->name, rc);
+			dlpar_online_cpu(dn);
+			return rc;
+		}
 	}
 
 	rc = dlpar_detach_node(dn);
@@ -593,7 +600,10 @@ static ssize_t dlpar_cpu_remove(struct device_node *dn, u32 drc_index)
 
 		pr_warn("Failed to detach CPU %s, rc: %d", dn->name, rc);
 
-		rc = dlpar_acquire_drc(drc_index);
+		if (release_drc)
+			rc = dlpar_acquire_drc(drc_index);
+		else
+			rc = 0;
 		if (!rc)
 			dlpar_online_cpu(dn);
 
@@ -622,7 +632,7 @@ static struct device_node *cpu_drc_index_to_dn(u32 drc_index)
 	return dn;
 }
 
-static int dlpar_cpu_remove_by_index(u32 drc_index)
+static int dlpar_cpu_remove_by_index(u32 drc_index, bool release_drc)
 {
 	struct device_node *dn;
 	int rc;
@@ -634,7 +644,7 @@ static int dlpar_cpu_remove_by_index(u32 drc_index)
 		return -ENODEV;
 	}
 
-	rc = dlpar_cpu_remove(dn, drc_index);
+	rc = dlpar_cpu_remove(dn, drc_index, release_drc);
 	of_node_put(dn);
 	return rc;
 }
@@ -699,7 +709,7 @@ static int dlpar_cpu_remove_by_count(u32 cpus_to_remove)
 	}
 
 	for (i = 0; i < cpus_to_remove; i++) {
-		rc = dlpar_cpu_remove_by_index(cpu_drcs[i]);
+		rc = dlpar_cpu_remove_by_index(cpu_drcs[i], true);
 		if (rc)
 			break;
 
@@ -710,7 +720,7 @@ static int dlpar_cpu_remove_by_count(u32 cpus_to_remove)
 		pr_warn("CPU hot-remove failed, adding back removed CPUs\n");
 
 		for (i = 0; i < cpus_removed; i++)
-			dlpar_cpu_add(cpu_drcs[i]);
+			dlpar_cpu_add(cpu_drcs[i], true);
 
 		rc = -EINVAL;
 	} else {
@@ -780,7 +790,7 @@ static int dlpar_cpu_add_by_count(u32 cpus_to_add)
 	}
 
 	for (i = 0; i < cpus_to_add; i++) {
-		rc = dlpar_cpu_add(cpu_drcs[i]);
+		rc = dlpar_cpu_add(cpu_drcs[i], true);
 		if (rc)
 			break;
 
@@ -791,7 +801,7 @@ static int dlpar_cpu_add_by_count(u32 cpus_to_add)
 		pr_warn("CPU hot-add failed, removing any added CPUs\n");
 
 		for (i = 0; i < cpus_added; i++)
-			dlpar_cpu_remove_by_index(cpu_drcs[i]);
+			dlpar_cpu_remove_by_index(cpu_drcs[i], true);
 
 		rc = -EINVAL;
 	} else {
@@ -817,7 +827,7 @@ int dlpar_cpu(struct pseries_hp_errorlog *hp_elog)
 		if (hp_elog->id_type == PSERIES_HP_ELOG_ID_DRC_COUNT)
 			rc = dlpar_cpu_remove_by_count(count);
 		else if (hp_elog->id_type == PSERIES_HP_ELOG_ID_DRC_INDEX)
-			rc = dlpar_cpu_remove_by_index(drc_index);
+			rc = dlpar_cpu_remove_by_index(drc_index, true);
 		else
 			rc = -EINVAL;
 		break;
@@ -825,7 +835,7 @@ int dlpar_cpu(struct pseries_hp_errorlog *hp_elog)
 		if (hp_elog->id_type == PSERIES_HP_ELOG_ID_DRC_COUNT)
 			rc = dlpar_cpu_add_by_count(count);
 		else if (hp_elog->id_type == PSERIES_HP_ELOG_ID_DRC_INDEX)
-			rc = dlpar_cpu_add(drc_index);
+			rc = dlpar_cpu_add(drc_index, true);
 		else
 			rc = -EINVAL;
 		break;
@@ -850,7 +860,7 @@ static ssize_t dlpar_cpu_probe(const char *buf, size_t count)
 	if (rc)
 		return -EINVAL;
 
-	rc = dlpar_cpu_add(drc_index);
+	rc = dlpar_cpu_add(drc_index, true);
 
 	return rc ? rc : count;
 }
@@ -871,7 +881,7 @@ static ssize_t dlpar_cpu_release(const char *buf, size_t count)
 		return -EINVAL;
 	}
 
-	rc = dlpar_cpu_remove(dn, drc_index);
+	rc = dlpar_cpu_remove(dn, drc_index, true);
 	of_node_put(dn);
 
 	return rc ? rc : count;

[PATCH v06 2/9] hotplug/cpu: Add operation queuing function

[Michael Bringmann]

migration/dlpar: This patch adds function dlpar_queue_action()
which will queued up information about a CPU/Memory 'readd'
operation according to resource type, action code, and DRC index.
At a subsequent point, the list of operations can be run/played
in series.  Examples of such oprations include 'readd' of CPU
and Memory blocks identified as having changed their associativity
during an LPAR migration event.

Signed-off-by: Michael Bringmann <mwb@linux.vnet.ibm.com>
---
Changes in patch:
  -- Correct drc_index before adding to pseries_hp_errorlog struct
  -- Correct text of notice
  -- Revise queuing model to save up all of the DLPAR actions for
     later execution.
  -- Restore list init statement missing from patch
  -- Move call to apply queued operations into 'mobility.c'
  -- Compress some code
  -- Rename some of queueing function APIs
  -- Revise implementation to push execution of queued operations
     to a workqueue task.
  -- Cleanup reference to outdated queuing operation.
---
 arch/powerpc/include/asm/rtas.h           |    2 +
 arch/powerpc/platforms/pseries/dlpar.c    |   61 +++++++++++++++++++++++++++++
 arch/powerpc/platforms/pseries/mobility.c |    4 ++
 arch/powerpc/platforms/pseries/pseries.h  |    2 +
 4 files changed, 69 insertions(+)

diff --git a/arch/powerpc/include/asm/rtas.h b/arch/powerpc/include/asm/rtas.h
index 71e393c..4f601c7 100644
--- a/arch/powerpc/include/asm/rtas.h
+++ b/arch/powerpc/include/asm/rtas.h
@@ -310,12 +310,14 @@ struct pseries_hp_errorlog {
 		struct { __be32 count, index; } ic;
 		char	drc_name[1];
 	} _drc_u;
+	struct list_head list;
 };
 
 #define PSERIES_HP_ELOG_RESOURCE_CPU	1
 #define PSERIES_HP_ELOG_RESOURCE_MEM	2
 #define PSERIES_HP_ELOG_RESOURCE_SLOT	3
 #define PSERIES_HP_ELOG_RESOURCE_PHB	4
+#define PSERIES_HP_ELOG_RESOURCE_PMT	5
 
 #define PSERIES_HP_ELOG_ACTION_ADD	1
 #define PSERIES_HP_ELOG_ACTION_REMOVE	2
diff --git a/arch/powerpc/platforms/pseries/dlpar.c b/arch/powerpc/platforms/pseries/dlpar.c
index a0b20c0..7264b8e 100644
--- a/arch/powerpc/platforms/pseries/dlpar.c
+++ b/arch/powerpc/platforms/pseries/dlpar.c
@@ -25,6 +25,7 @@
 #include <asm/prom.h>
 #include <asm/machdep.h>
 #include <linux/uaccess.h>
+#include <linux/delay.h>
 #include <asm/rtas.h>
 
 static struct workqueue_struct *pseries_hp_wq;
@@ -329,6 +330,8 @@ int dlpar_release_drc(u32 drc_index)
 	return 0;
 }
 
+static int dlpar_pmt(struct pseries_hp_errorlog *work);
+
 static int handle_dlpar_errorlog(struct pseries_hp_errorlog *hp_elog)
 {
 	int rc;
@@ -357,6 +360,9 @@ static int handle_dlpar_errorlog(struct pseries_hp_errorlog *hp_elog)
 	case PSERIES_HP_ELOG_RESOURCE_CPU:
 		rc = dlpar_cpu(hp_elog);
 		break;
+	case PSERIES_HP_ELOG_RESOURCE_PMT:
+		rc = dlpar_pmt(hp_elog);
+		break;
 	default:
 		pr_warn_ratelimited("Invalid resource (%d) specified\n",
 				    hp_elog->resource);
@@ -407,6 +413,61 @@ void queue_hotplug_event(struct pseries_hp_errorlog *hp_errlog,
 	}
 }
 
+LIST_HEAD(dlpar_delayed_list);
+
+int dlpar_queue_action(int resource, int action, u32 drc_index)
+{
+	struct pseries_hp_errorlog *hp_errlog;
+
+	hp_errlog = kmalloc(sizeof(struct pseries_hp_errorlog), GFP_KERNEL);
+	if (!hp_errlog)
+		return -ENOMEM;
+
+	hp_errlog->resource = resource;
+	hp_errlog->action = action;
+	hp_errlog->id_type = PSERIES_HP_ELOG_ID_DRC_INDEX;
+	hp_errlog->_drc_u.drc_index = cpu_to_be32(drc_index);
+
+	list_add_tail(&hp_errlog->list, &dlpar_delayed_list);
+
+	return 0;
+}
+
+static int dlpar_pmt(struct pseries_hp_errorlog *work)
+{
+	struct list_head *pos, *q;
+
+	ssleep(15);
+
+	list_for_each_safe(pos, q, &dlpar_delayed_list) {
+		struct pseries_hp_errorlog *tmp;
+
+		tmp = list_entry(pos, struct pseries_hp_errorlog, list);
+		handle_dlpar_errorlog(tmp);
+
+		list_del(pos);
+		kfree(tmp);
+
+		ssleep(10);
+	}
+
+	return 0;
+}
+
+int dlpar_queued_actions_run(void)
+{
+	if (!list_empty(&dlpar_delayed_list)) {
+		struct pseries_hp_errorlog hp_errlog;
+
+		hp_errlog.resource = PSERIES_HP_ELOG_RESOURCE_PMT;
+		hp_errlog.action = 0;
+		hp_errlog.id_type = 0;
+
+		queue_hotplug_event(&hp_errlog, 0, 0);
+	}
+	return 0;
+}
+
 static int dlpar_parse_resource(char **cmd, struct pseries_hp_errorlog *hp_elog)
 {
 	char *arg;
diff --git a/arch/powerpc/platforms/pseries/mobility.c b/arch/powerpc/platforms/pseries/mobility.c
index f6364d9..d0d1cae 100644
--- a/arch/powerpc/platforms/pseries/mobility.c
+++ b/arch/powerpc/platforms/pseries/mobility.c
@@ -378,6 +378,10 @@ static ssize_t migration_store(struct class *class,
 		return rc;
 
 	post_mobility_fixup();
+
+	/* Apply any necessary changes identified during fixup */
+	dlpar_queued_actions_run();
+
 	return count;
 }
 
diff --git a/arch/powerpc/platforms/pseries/pseries.h b/arch/powerpc/platforms/pseries/pseries.h
index 60db2ee..72ca996 100644
--- a/arch/powerpc/platforms/pseries/pseries.h
+++ b/arch/powerpc/platforms/pseries/pseries.h
@@ -61,6 +61,8 @@ extern struct device_node *dlpar_configure_connector(__be32,
 
 void queue_hotplug_event(struct pseries_hp_errorlog *hp_errlog,
 			 struct completion *hotplug_done, int *rc);
+int dlpar_queue_action(int resource, int action, u32 drc_index);
+int dlpar_queued_actions_run(void);
 #ifdef CONFIG_MEMORY_HOTPLUG
 int dlpar_memory(struct pseries_hp_errorlog *hp_elog);
 #else

[PATCH v06 3/9] hotplug/cpu: Provide CPU readd operation

[Michael Bringmann]

powerpc/dlpar: Provide hotplug CPU 'readd by index' operation to
support LPAR Post Migration state updates.  When such changes are
invoked by the PowerPC 'mobility' code, they will be queued up so
that modifications to CPU properties will take place after the new
property value is written to the device-tree.

Signed-off-by: Michael Bringmann <mwb@linux.vnet.ibm.com>
---
Changes in patch:
  -- Add CPU validity check to pseries_smp_notifier
  -- Improve check on 'ibm,associativity' property
  -- Add check for cpu type to new update property entry
  -- Cleanup reference to outdated queuing function.
---
 arch/powerpc/platforms/pseries/hotplug-cpu.c |   58 ++++++++++++++++++++++++++
 1 file changed, 58 insertions(+)

diff --git a/arch/powerpc/platforms/pseries/hotplug-cpu.c b/arch/powerpc/platforms/pseries/hotplug-cpu.c
index 3632db2..8f28160 100644
--- a/arch/powerpc/platforms/pseries/hotplug-cpu.c
+++ b/arch/powerpc/platforms/pseries/hotplug-cpu.c
@@ -305,6 +305,36 @@ static int pseries_add_processor(struct device_node *np)
 	return err;
 }
 
+static int pseries_update_processor(struct of_reconfig_data *pr)
+{
+	int old_entries, new_entries, rc = 0;
+	__be32 *old_assoc, *new_assoc;
+
+	/* We only handle changes due to 'ibm,associativity' property
+	 */
+	old_assoc = pr->old_prop->value;
+	old_entries = be32_to_cpu(*old_assoc++);
+
+	new_assoc = pr->prop->value;
+	new_entries = be32_to_cpu(*new_assoc++);
+
+	if (old_entries == new_entries) {
+		int sz = old_entries * sizeof(int);
+
+		if (memcmp(old_assoc, new_assoc, sz))
+			rc = dlpar_queue_action(
+					PSERIES_HP_ELOG_RESOURCE_CPU,
+					PSERIES_HP_ELOG_ACTION_READD,
+					pr->dn->phandle);
+	} else {
+		rc = dlpar_queue_action(PSERIES_HP_ELOG_RESOURCE_CPU,
+					PSERIES_HP_ELOG_ACTION_READD,
+					pr->dn->phandle);
+	}
+
+	return rc;
+}
+
 /*
  * Update the present map for a cpu node which is going away, and set
  * the hard id in the paca(s) to -1 to be consistent with boot time
@@ -649,6 +679,26 @@ static int dlpar_cpu_remove_by_index(u32 drc_index, bool release_drc)
 	return rc;
 }
 
+static int dlpar_cpu_readd_by_index(u32 drc_index)
+{
+	int rc = 0;
+
+	pr_info("Attempting to re-add CPU, drc index %x\n", drc_index);
+
+	rc = dlpar_cpu_remove_by_index(drc_index, false);
+	if (!rc)
+		rc = dlpar_cpu_add(drc_index, false);
+
+	if (rc)
+		pr_info("Failed to update cpu at drc_index %lx\n",
+				(unsigned long int)drc_index);
+	else
+		pr_info("CPU at drc_index %lx was updated\n",
+				(unsigned long int)drc_index);
+
+	return rc;
+}
+
 static int find_dlpar_cpus_to_remove(u32 *cpu_drcs, int cpus_to_remove)
 {
 	struct device_node *dn;
@@ -839,6 +889,9 @@ int dlpar_cpu(struct pseries_hp_errorlog *hp_elog)
 		else
 			rc = -EINVAL;
 		break;
+	case PSERIES_HP_ELOG_ACTION_READD:
+		rc = dlpar_cpu_readd_by_index(drc_index);
+		break;
 	default:
 		pr_err("Invalid action (%d) specified\n", hp_elog->action);
 		rc = -EINVAL;
@@ -902,6 +955,11 @@ static int pseries_smp_notifier(struct notifier_block *nb,
 	case OF_RECONFIG_DETACH_NODE:
 		pseries_remove_processor(rd->dn);
 		break;
+	case OF_RECONFIG_UPDATE_PROPERTY:
+		if (!strcmp(rd->dn->type, "cpu") &&
+		    !strcmp(rd->prop->name, "ibm,associativity"))
+			pseries_update_processor(rd);
+		break;
 	}
 	return notifier_from_errno(err);
 }

[PATCH v06 4/9] mobility/numa: Ensure numa update does not overlap

[Michael Bringmann]

mobility/numa: Ensure that numa_update_cpu_topology() can not be
entered multiple times concurrently.  It may be accessed through
many different paths / concurrent work functions, and the lock
ordering may be difficult to ensure otherwise.

Signed-off-by: Michael Bringmann <mwb@linux.vnet.ibm.com>
---
 arch/powerpc/mm/numa.c |    9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/arch/powerpc/mm/numa.c b/arch/powerpc/mm/numa.c
index a789d57..b22e27a 100644
--- a/arch/powerpc/mm/numa.c
+++ b/arch/powerpc/mm/numa.c
@@ -1079,6 +1079,7 @@ struct topology_update_data {
 static int topology_timer_secs = 1;
 static int topology_inited;
 static int topology_update_needed;
+static struct mutex topology_update_lock;
 
 /*
  * Change polling interval for associativity changes.
@@ -1320,6 +1321,11 @@ int numa_update_cpu_topology(bool cpus_locked)
 	if (!updates)
 		return 0;
 
+	if (!mutex_trylock(&topology_update_lock)) {
+		kfree(updates);
+		return 0;
+	}
+
 	cpumask_clear(&updated_cpus);
 
 	for_each_cpu(cpu, &cpu_associativity_changes_mask) {
@@ -1424,6 +1430,7 @@ int numa_update_cpu_topology(bool cpus_locked)
 out:
 	kfree(updates);
 	topology_update_needed = 0;
+	mutex_unlock(&topology_update_lock);
 	return changed;
 }
 
@@ -1598,6 +1605,8 @@ static ssize_t topology_write(struct file *file, const char __user *buf,
 
 static int topology_update_init(void)
 {
+	mutex_init(&topology_update_lock);
+
 	/* Do not poll for changes if disabled at boot */
 	if (topology_updates_enabled)
 		start_topology_update();

[PATCH v06 5/9] numa: Disable/enable arch_update_cpu_topology

[Michael Bringmann]

numa: Provide mechanism to disable/enable operation of
arch_update_cpu_topology/numa_update_cpu_topology.  This is
a simple tool to eliminate some avenues for thread deadlock
observed during system execution.

Signed-off-by: Michael Bringmann <mwb@linux.vnet.ibm.com>
---
 arch/powerpc/include/asm/topology.h |   10 ++++++++++
 arch/powerpc/mm/numa.c              |   14 ++++++++++++++
 2 files changed, 24 insertions(+)

diff --git a/arch/powerpc/include/asm/topology.h b/arch/powerpc/include/asm/topology.h
index 16b0778..d9ceba6 100644
--- a/arch/powerpc/include/asm/topology.h
+++ b/arch/powerpc/include/asm/topology.h
@@ -43,6 +43,8 @@ static inline int pcibus_to_node(struct pci_bus *bus)
 extern int sysfs_add_device_to_node(struct device *dev, int nid);
 extern void sysfs_remove_device_from_node(struct device *dev, int nid);
 extern int numa_update_cpu_topology(bool cpus_locked);
+extern void arch_update_cpu_topology_suspend(void);
+extern void arch_update_cpu_topology_resume(void);
 
 static inline void update_numa_cpu_lookup_table(unsigned int cpu, int node)
 {
@@ -82,6 +84,14 @@ static inline int numa_update_cpu_topology(bool cpus_locked)
 	return 0;
 }
 
+static inline void arch_update_cpu_topology_suspend(void)
+{
+}
+
+static inline void arch_update_cpu_topology_resume(void)
+{
+}
+
 static inline void update_numa_cpu_lookup_table(unsigned int cpu, int node) {}
 
 #endif /* CONFIG_NUMA */
diff --git a/arch/powerpc/mm/numa.c b/arch/powerpc/mm/numa.c
index b22e27a..2352489 100644
--- a/arch/powerpc/mm/numa.c
+++ b/arch/powerpc/mm/numa.c
@@ -1079,6 +1079,7 @@ struct topology_update_data {
 static int topology_timer_secs = 1;
 static int topology_inited;
 static int topology_update_needed;
+static int topology_update_enabled = 1;
 static struct mutex topology_update_lock;
 
 /*
@@ -1313,6 +1314,9 @@ int numa_update_cpu_topology(bool cpus_locked)
 		return 0;
 	}
 
+	if (!topology_update_enabled)
+		return 0;
+
 	weight = cpumask_weight(&cpu_associativity_changes_mask);
 	if (!weight)
 		return 0;
@@ -1439,6 +1443,16 @@ int arch_update_cpu_topology(void)
 	return numa_update_cpu_topology(true);
 }
 
+void arch_update_cpu_topology_suspend(void)
+{
+	topology_update_enabled = 0;
+}
+
+void arch_update_cpu_topology_resume(void)
+{
+	topology_update_enabled = 1;
+}
+
 static void topology_work_fn(struct work_struct *work)
 {
 	rebuild_sched_domains();

[PATCH v06 6/9] pmt/numa: Disable arch_update_cpu_topology during CPU readd

[Michael Bringmann]

pmt/numa: Disable arch_update_cpu_topology during post migration
CPU readd updates when evaluating device-tree changes after LPM
to avoid thread deadlocks trying to update node assignments.
System timing between all of the threads and timers restarted in
a migrated system overlapped frequently allowing tasks to start
acquiring resources (get_online_cpus) needed by rebuild_sched_domains.
Defer the operation of that function until after the CPU readd has
completed.

Signed-off-by: Michael Bringmann <mwb@linux.vnet.ibm.com>
---
 arch/powerpc/platforms/pseries/hotplug-cpu.c |    9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/arch/powerpc/platforms/pseries/hotplug-cpu.c b/arch/powerpc/platforms/pseries/hotplug-cpu.c
index 8f28160..6267b53 100644
--- a/arch/powerpc/platforms/pseries/hotplug-cpu.c
+++ b/arch/powerpc/platforms/pseries/hotplug-cpu.c
@@ -26,6 +26,7 @@
 #include <linux/sched.h>	/* for idle_task_exit */
 #include <linux/sched/hotplug.h>
 #include <linux/cpu.h>
+#include <linux/cpuset.h>
 #include <linux/of.h>
 #include <linux/slab.h>
 #include <asm/prom.h>
@@ -685,9 +686,15 @@ static int dlpar_cpu_readd_by_index(u32 drc_index)
 
 	pr_info("Attempting to re-add CPU, drc index %x\n", drc_index);
 
+	arch_update_cpu_topology_suspend();
 	rc = dlpar_cpu_remove_by_index(drc_index, false);
-	if (!rc)
+	arch_update_cpu_topology_resume();
+
+	if (!rc) {
+		arch_update_cpu_topology_suspend();
 		rc = dlpar_cpu_add(drc_index, false);
+		arch_update_cpu_topology_resume();
+	}
 
 	if (rc)
 		pr_info("Failed to update cpu at drc_index %lx\n",

[PATCH v06 7/9] powerpc/rtas: Allow disabling rtas_event_scan

[Michael Bringmann]

powerpc/rtas: Provide mechanism by which the rtas_event_scan can
be disabled/re-enabled by other portions of the powerpc code.
Among other things, this simplifies the usage of locking mechanisms
for shared kernel resources.

Signed-off-by: Michael Bringmann <mwb@linux.vnet.ibm.com>
---
 arch/powerpc/include/asm/rtas.h |    4 ++++
 arch/powerpc/kernel/rtasd.c     |   14 ++++++++++++++
 2 files changed, 18 insertions(+)

diff --git a/arch/powerpc/include/asm/rtas.h b/arch/powerpc/include/asm/rtas.h
index 4f601c7..4ab605a 100644
--- a/arch/powerpc/include/asm/rtas.h
+++ b/arch/powerpc/include/asm/rtas.h
@@ -386,8 +386,12 @@ extern int early_init_dt_scan_rtas(unsigned long node,
 
 #ifdef CONFIG_PPC_RTAS_DAEMON
 extern void rtas_cancel_event_scan(void);
+extern void rtas_event_scan_disable(void);
+extern void rtas_event_scan_enable(void);
 #else
 static inline void rtas_cancel_event_scan(void) { }
+static inline void rtas_event_scan_disable(void) { }
+static inline void rtas_event_scan_enable(void) { }
 #endif
 
 /* Error types logged.  */
diff --git a/arch/powerpc/kernel/rtasd.c b/arch/powerpc/kernel/rtasd.c
index 44d66c33d..af69e44 100644
--- a/arch/powerpc/kernel/rtasd.c
+++ b/arch/powerpc/kernel/rtasd.c
@@ -455,11 +455,25 @@ static void do_event_scan(void)
  */
 static unsigned long event_scan_delay = 1*HZ;
 static int first_pass = 1;
+static int res_enable = 1;
+
+void rtas_event_scan_disable(void)
+{
+	res_enable = 0;
+}
+
+void rtas_event_scan_enable(void)
+{
+	res_enable = 1;
+}
 
 static void rtas_event_scan(struct work_struct *w)
 {
 	unsigned int cpu;
 
+	if (!res_enable)
+		return;
+
 	do_event_scan();
 
 	get_online_cpus();

[PATCH v06 8/9] hotplug/rtas: No rtas_event_scan during PMT update

[Michael Bringmann]

hotplug/rtas: Disable rtas_event_scan during device-tree property
updates after migration to reduce conflicts with changes propagated
to other parts of the kernel configuration, such as CPUs or memory.

Signed-off-by: Michael Bringmann <mwb@linux.vnet.ibm.com>
---
 arch/powerpc/platforms/pseries/hotplug-cpu.c |    4 ++++
 1 file changed, 4 insertions(+)

diff --git a/arch/powerpc/platforms/pseries/hotplug-cpu.c b/arch/powerpc/platforms/pseries/hotplug-cpu.c
index 6267b53..f5c9e8f 100644
--- a/arch/powerpc/platforms/pseries/hotplug-cpu.c
+++ b/arch/powerpc/platforms/pseries/hotplug-cpu.c
@@ -686,14 +686,18 @@ static int dlpar_cpu_readd_by_index(u32 drc_index)
 
 	pr_info("Attempting to re-add CPU, drc index %x\n", drc_index);
 
+	rtas_event_scan_disable();
 	arch_update_cpu_topology_suspend();
 	rc = dlpar_cpu_remove_by_index(drc_index, false);
 	arch_update_cpu_topology_resume();
+	rtas_event_scan_enable();
 
 	if (!rc) {
+		rtas_event_scan_disable();
 		arch_update_cpu_topology_suspend();
 		rc = dlpar_cpu_add(drc_index, false);
 		arch_update_cpu_topology_resume();
+		rtas_event_scan_enable();
 	}
 
 	if (rc)

[PATCH v06 9/9] hotplug/pmt: Update topology after PMT

[Michael Bringmann]

hotplug/pmt: Call rebuild_sched_domains after applying changes
to update CPU associativity i.e. 'readd' CPUs.  This is to
ensure that the deferred calls to arch_update_cpu_topology are
now reflected in the system data structures.

Signed-off-by: Michael Bringmann <mwb@linux.vnet.ibm.com>
---
 arch/powerpc/platforms/pseries/dlpar.c |    4 ++++
 1 file changed, 4 insertions(+)

diff --git a/arch/powerpc/platforms/pseries/dlpar.c b/arch/powerpc/platforms/pseries/dlpar.c
index 7264b8e..ea3c08a 100644
--- a/arch/powerpc/platforms/pseries/dlpar.c
+++ b/arch/powerpc/platforms/pseries/dlpar.c
@@ -16,6 +16,7 @@
 #include <linux/notifier.h>
 #include <linux/spinlock.h>
 #include <linux/cpu.h>
+#include <linux/cpuset.h>
 #include <linux/slab.h>
 #include <linux/of.h>
 
@@ -451,6 +452,9 @@ static int dlpar_pmt(struct pseries_hp_errorlog *work)
 		ssleep(10);
 	}
 
+	ssleep(5);
+	rebuild_sched_domains();
+
 	return 0;
 }
 

Re: [PATCH v06 1/9] hotplug/cpu: Conditionally acquire/release DRC index

[Naveen N. Rao]

Michael Bringmann wrote:
> powerpc/cpu: Modify dlpar_cpu_add and dlpar_cpu_remove to allow the
> skipping of DRC index acquire or release operations during the CPU
> add or remove operations.  This is intended to support subsequent
> changes to provide a 'CPU readd' operation.
>=20
> Signed-off-by: Michael Bringmann <mwb@linux.vnet.ibm.com>
> ---
> Changes in patch:
>   -- Move new validity check added to pseries_smp_notifier
>      to another patch
> ---
>  arch/powerpc/platforms/pseries/hotplug-cpu.c |   68 +++++++++++++++-----=
------
>  1 file changed, 39 insertions(+), 29 deletions(-)
>=20
> diff --git a/arch/powerpc/platforms/pseries/hotplug-cpu.c b/arch/powerpc/=
platforms/pseries/hotplug-cpu.c
> index 6ef77ca..3632db2 100644
> --- a/arch/powerpc/platforms/pseries/hotplug-cpu.c
> +++ b/arch/powerpc/platforms/pseries/hotplug-cpu.c
> @@ -432,7 +432,7 @@ static bool valid_cpu_drc_index(struct device_node *p=
arent, u32 drc_index)
>  	return found;
>  }
>=20
> -static ssize_t dlpar_cpu_add(u32 drc_index)
> +static ssize_t dlpar_cpu_add(u32 drc_index, bool acquire_drc)
>  {
>  	struct device_node *dn, *parent;
>  	int rc, saved_rc;
> @@ -457,19 +457,22 @@ static ssize_t dlpar_cpu_add(u32 drc_index)
>  		return -EINVAL;
>  	}
>=20
> -	rc =3D dlpar_acquire_drc(drc_index);
> -	if (rc) {
> -		pr_warn("Failed to acquire DRC, rc: %d, drc index: %x\n",
> -			rc, drc_index);
> -		of_node_put(parent);
> -		return -EINVAL;
> +	if (acquire_drc) {
> +		rc =3D dlpar_acquire_drc(drc_index);
> +		if (rc) {
> +			pr_warn("Failed to acquire DRC, rc: %d, drc index: %x\n",
> +				rc, drc_index);
> +			of_node_put(parent);
> +			return -EINVAL;
> +		}
>  	}
>=20
>  	dn =3D dlpar_configure_connector(cpu_to_be32(drc_index), parent);
>  	if (!dn) {
>  		pr_warn("Failed call to configure-connector, drc index: %x\n",
>  			drc_index);
> -		dlpar_release_drc(drc_index);
> +		if (acquire_drc)
> +			dlpar_release_drc(drc_index);
>  		of_node_put(parent);
>  		return -EINVAL;
>  	}
> @@ -484,8 +487,9 @@ static ssize_t dlpar_cpu_add(u32 drc_index)
>  		pr_warn("Failed to attach node %s, rc: %d, drc index: %x\n",
>  			dn->name, rc, drc_index);
>=20
> -		rc =3D dlpar_release_drc(drc_index);
> -		if (!rc)
> +		if (acquire_drc)
> +			rc =3D dlpar_release_drc(drc_index);
> +		if (!rc || acquire_drc)
>  			dlpar_free_cc_nodes(dn);

(!rc) can only be true if acquire_drc is true, so we seem to be only=20
invoking dlpar_free_cc_nodes() if acquire_drc is true. Would it be=20
better to frame that condition in this manner:

	if (acquire_drc) {
		rc =3D dlpar_release_drc(drc_index):
		if (!rc)
			dlpar_free_cc_nodes(dn);
	}

>=20
>  		return saved_rc;
> @@ -498,7 +502,7 @@ static ssize_t dlpar_cpu_add(u32 drc_index)
>  			dn->name, rc, drc_index);
>=20
>  		rc =3D dlpar_detach_node(dn);
> -		if (!rc)
> +		if (!rc && acquire_drc)
>  			dlpar_release_drc(drc_index);
>=20
>  		return saved_rc;
> @@ -566,7 +570,8 @@ static int dlpar_offline_cpu(struct device_node *dn)
>=20
>  }
>=20
> -static ssize_t dlpar_cpu_remove(struct device_node *dn, u32 drc_index)
> +static ssize_t dlpar_cpu_remove(struct device_node *dn, u32 drc_index,
> +				bool release_drc)
>  {
>  	int rc;
>=20
> @@ -579,12 +584,14 @@ static ssize_t dlpar_cpu_remove(struct device_node =
*dn, u32 drc_index)
>  		return -EINVAL;
>  	}
>=20
> -	rc =3D dlpar_release_drc(drc_index);
> -	if (rc) {
> -		pr_warn("Failed to release drc (%x) for CPU %s, rc: %d\n",
> -			drc_index, dn->name, rc);
> -		dlpar_online_cpu(dn);
> -		return rc;
> +	if (release_drc) {
> +		rc =3D dlpar_release_drc(drc_index);
> +		if (rc) {
> +			pr_warn("Failed to release drc (%x) for CPU %s, rc: %d\n",
> +				drc_index, dn->name, rc);
> +			dlpar_online_cpu(dn);
> +			return rc;
> +		}
>  	}
>=20
>  	rc =3D dlpar_detach_node(dn);
> @@ -593,7 +600,10 @@ static ssize_t dlpar_cpu_remove(struct device_node *=
dn, u32 drc_index)
>=20
>  		pr_warn("Failed to detach CPU %s, rc: %d", dn->name, rc);
>=20
> -		rc =3D dlpar_acquire_drc(drc_index);
> +		if (release_drc)
> +			rc =3D dlpar_acquire_drc(drc_index);
> +		else
> +			rc =3D 0;
>  		if (!rc)
>  			dlpar_online_cpu(dn);

Here, we seem to want to invoke dlpar_online_cpu() unconditionally if=20
release_drc is false. So, to make that explicit, would it be better to=20
frame that as:
	if (release_drc)
		rc =3D dlpar_acquire_drc(drc_index);
	if (!release_drc || !rc)
		dlpar_online_cpu(dn);

- Naveen

>=20
> @@ -622,7 +632,7 @@ static struct device_node *cpu_drc_index_to_dn(u32 dr=
c_index)
>  	return dn;
>  }
>=20
> -static int dlpar_cpu_remove_by_index(u32 drc_index)
> +static int dlpar_cpu_remove_by_index(u32 drc_index, bool release_drc)
>  {
>  	struct device_node *dn;
>  	int rc;
> @@ -634,7 +644,7 @@ static int dlpar_cpu_remove_by_index(u32 drc_index)
>  		return -ENODEV;
>  	}
>=20
> -	rc =3D dlpar_cpu_remove(dn, drc_index);
> +	rc =3D dlpar_cpu_remove(dn, drc_index, release_drc);
>  	of_node_put(dn);
>  	return rc;
>  }
> @@ -699,7 +709,7 @@ static int dlpar_cpu_remove_by_count(u32 cpus_to_remo=
ve)
>  	}
>=20
>  	for (i =3D 0; i < cpus_to_remove; i++) {
> -		rc =3D dlpar_cpu_remove_by_index(cpu_drcs[i]);
> +		rc =3D dlpar_cpu_remove_by_index(cpu_drcs[i], true);
>  		if (rc)
>  			break;
>=20
> @@ -710,7 +720,7 @@ static int dlpar_cpu_remove_by_count(u32 cpus_to_remo=
ve)
>  		pr_warn("CPU hot-remove failed, adding back removed CPUs\n");
>=20
>  		for (i =3D 0; i < cpus_removed; i++)
> -			dlpar_cpu_add(cpu_drcs[i]);
> +			dlpar_cpu_add(cpu_drcs[i], true);
>=20
>  		rc =3D -EINVAL;
>  	} else {
> @@ -780,7 +790,7 @@ static int dlpar_cpu_add_by_count(u32 cpus_to_add)
>  	}
>=20
>  	for (i =3D 0; i < cpus_to_add; i++) {
> -		rc =3D dlpar_cpu_add(cpu_drcs[i]);
> +		rc =3D dlpar_cpu_add(cpu_drcs[i], true);
>  		if (rc)
>  			break;
>=20
> @@ -791,7 +801,7 @@ static int dlpar_cpu_add_by_count(u32 cpus_to_add)
>  		pr_warn("CPU hot-add failed, removing any added CPUs\n");
>=20
>  		for (i =3D 0; i < cpus_added; i++)
> -			dlpar_cpu_remove_by_index(cpu_drcs[i]);
> +			dlpar_cpu_remove_by_index(cpu_drcs[i], true);
>=20
>  		rc =3D -EINVAL;
>  	} else {
> @@ -817,7 +827,7 @@ int dlpar_cpu(struct pseries_hp_errorlog *hp_elog)
>  		if (hp_elog->id_type =3D=3D PSERIES_HP_ELOG_ID_DRC_COUNT)
>  			rc =3D dlpar_cpu_remove_by_count(count);
>  		else if (hp_elog->id_type =3D=3D PSERIES_HP_ELOG_ID_DRC_INDEX)
> -			rc =3D dlpar_cpu_remove_by_index(drc_index);
> +			rc =3D dlpar_cpu_remove_by_index(drc_index, true);
>  		else
>  			rc =3D -EINVAL;
>  		break;
> @@ -825,7 +835,7 @@ int dlpar_cpu(struct pseries_hp_errorlog *hp_elog)
>  		if (hp_elog->id_type =3D=3D PSERIES_HP_ELOG_ID_DRC_COUNT)
>  			rc =3D dlpar_cpu_add_by_count(count);
>  		else if (hp_elog->id_type =3D=3D PSERIES_HP_ELOG_ID_DRC_INDEX)
> -			rc =3D dlpar_cpu_add(drc_index);
> +			rc =3D dlpar_cpu_add(drc_index, true);
>  		else
>  			rc =3D -EINVAL;
>  		break;
> @@ -850,7 +860,7 @@ static ssize_t dlpar_cpu_probe(const char *buf, size_=
t count)
>  	if (rc)
>  		return -EINVAL;
>=20
> -	rc =3D dlpar_cpu_add(drc_index);
> +	rc =3D dlpar_cpu_add(drc_index, true);
>=20
>  	return rc ? rc : count;
>  }
> @@ -871,7 +881,7 @@ static ssize_t dlpar_cpu_release(const char *buf, siz=
e_t count)
>  		return -EINVAL;
>  	}
>=20
> -	rc =3D dlpar_cpu_remove(dn, drc_index);
> +	rc =3D dlpar_cpu_remove(dn, drc_index, true);
>  	of_node_put(dn);
>=20
>  	return rc ? rc : count;
>=20
>=20
=

Re: [PATCH v06 1/9] hotplug/cpu: Conditionally acquire/release DRC index

[Michael Bringmann]

See below.  Code cleanups accepted for next revision.

On 07/11/2018 11:00 AM, Naveen N. Rao wrote:
> Michael Bringmann wrote:
>> powerpc/cpu: Modify dlpar_cpu_add and dlpar_cpu_remove to allow the
>> skipping of DRC index acquire or release operations during the CPU
>> add or remove operations.  This is intended to support subsequent
>> changes to provide a 'CPU readd' operation.
>>
>> Signed-off-by: Michael Bringmann <mwb@linux.vnet.ibm.com>
>> ---
>> Changes in patch:
>>   -- Move new validity check added to pseries_smp_notifier
>>      to another patch
>> ---
>>  arch/powerpc/platforms/pseries/hotplug-cpu.c |   68 +++++++++++++++-----------
>>  1 file changed, 39 insertions(+), 29 deletions(-)
>>
>> diff --git a/arch/powerpc/platforms/pseries/hotplug-cpu.c b/arch/powerpc/platforms/pseries/hotplug-cpu.c
>> index 6ef77ca..3632db2 100644
>> --- a/arch/powerpc/platforms/pseries/hotplug-cpu.c
>> +++ b/arch/powerpc/platforms/pseries/hotplug-cpu.c
>> @@ -432,7 +432,7 @@ static bool valid_cpu_drc_index(struct device_node *parent, u32 drc_index)
>>      return found;
>>  }
>>
>> -static ssize_t dlpar_cpu_add(u32 drc_index)
>> +static ssize_t dlpar_cpu_add(u32 drc_index, bool acquire_drc)
>>  {
>>      struct device_node *dn, *parent;
>>      int rc, saved_rc;
>> @@ -457,19 +457,22 @@ static ssize_t dlpar_cpu_add(u32 drc_index)
>>          return -EINVAL;
>>      }
>>
>> -    rc = dlpar_acquire_drc(drc_index);
>> -    if (rc) {
>> -        pr_warn("Failed to acquire DRC, rc: %d, drc index: %x\n",
>> -            rc, drc_index);
>> -        of_node_put(parent);
>> -        return -EINVAL;
>> +    if (acquire_drc) {
>> +        rc = dlpar_acquire_drc(drc_index);
>> +        if (rc) {
>> +            pr_warn("Failed to acquire DRC, rc: %d, drc index: %x\n",
>> +                rc, drc_index);
>> +            of_node_put(parent);
>> +            return -EINVAL;
>> +        }
>>      }
>>
>>      dn = dlpar_configure_connector(cpu_to_be32(drc_index), parent);
>>      if (!dn) {
>>          pr_warn("Failed call to configure-connector, drc index: %x\n",
>>              drc_index);
>> -        dlpar_release_drc(drc_index);
>> +        if (acquire_drc)
>> +            dlpar_release_drc(drc_index);
>>          of_node_put(parent);
>>          return -EINVAL;
>>      }
>> @@ -484,8 +487,9 @@ static ssize_t dlpar_cpu_add(u32 drc_index)
>>          pr_warn("Failed to attach node %s, rc: %d, drc index: %x\n",
>>              dn->name, rc, drc_index);
>>
>> -        rc = dlpar_release_drc(drc_index);
>> -        if (!rc)
>> +        if (acquire_drc)
>> +            rc = dlpar_release_drc(drc_index);
>> +        if (!rc || acquire_drc)
>>              dlpar_free_cc_nodes(dn);
> 
> (!rc) can only be true if acquire_drc is true, so we seem to be only invoking dlpar_free_cc_nodes() if acquire_drc is true. Would it be better to frame that condition in this manner:
> 
>     if (acquire_drc) {
>         rc = dlpar_release_drc(drc_index):
>         if (!rc)
>             dlpar_free_cc_nodes(dn);
>     }
> 

Okay.

>>
>>          return saved_rc;
>> @@ -498,7 +502,7 @@ static ssize_t dlpar_cpu_add(u32 drc_index)
>>              dn->name, rc, drc_index);
>>
>>          rc = dlpar_detach_node(dn);
>> -        if (!rc)
>> +        if (!rc && acquire_drc)
>>              dlpar_release_drc(drc_index);
>>
>>          return saved_rc;
>> @@ -566,7 +570,8 @@ static int dlpar_offline_cpu(struct device_node *dn)
>>
>>  }
>>
>> -static ssize_t dlpar_cpu_remove(struct device_node *dn, u32 drc_index)
>> +static ssize_t dlpar_cpu_remove(struct device_node *dn, u32 drc_index,
>> +                bool release_drc)
>>  {
>>      int rc;
>>
>> @@ -579,12 +584,14 @@ static ssize_t dlpar_cpu_remove(struct device_node *dn, u32 drc_index)
>>          return -EINVAL;
>>      }
>>
>> -    rc = dlpar_release_drc(drc_index);
>> -    if (rc) {
>> -        pr_warn("Failed to release drc (%x) for CPU %s, rc: %d\n",
>> -            drc_index, dn->name, rc);
>> -        dlpar_online_cpu(dn);
>> -        return rc;
>> +    if (release_drc) {
>> +        rc = dlpar_release_drc(drc_index);
>> +        if (rc) {
>> +            pr_warn("Failed to release drc (%x) for CPU %s, rc: %d\n",
>> +                drc_index, dn->name, rc);
>> +            dlpar_online_cpu(dn);
>> +            return rc;
>> +        }
>>      }
>>
>>      rc = dlpar_detach_node(dn);
>> @@ -593,7 +600,10 @@ static ssize_t dlpar_cpu_remove(struct device_node *dn, u32 drc_index)
>>
>>          pr_warn("Failed to detach CPU %s, rc: %d", dn->name, rc);
>>
>> -        rc = dlpar_acquire_drc(drc_index);
>> +        if (release_drc)
>> +            rc = dlpar_acquire_drc(drc_index);
>> +        else
>> +            rc = 0;
>>          if (!rc)
>>              dlpar_online_cpu(dn);
> 
> Here, we seem to want to invoke dlpar_online_cpu() unconditionally if release_drc is false. So, to make that explicit, would it be better to frame that as:
>     if (release_drc)
>         rc = dlpar_acquire_drc(drc_index);
>     if (!release_drc || !rc)
>         dlpar_online_cpu(dn);
> 
> - Naveen

Okay.

> 
>>
>> @@ -622,7 +632,7 @@ static struct device_node *cpu_drc_index_to_dn(u32 drc_index)
>>      return dn;
>>  }
>>
>> -static int dlpar_cpu_remove_by_index(u32 drc_index)
>> +static int dlpar_cpu_remove_by_index(u32 drc_index, bool release_drc)
>>  {
>>      struct device_node *dn;
>>      int rc;
>> @@ -634,7 +644,7 @@ static int dlpar_cpu_remove_by_index(u32 drc_index)
>>          return -ENODEV;
>>      }
>>
>> -    rc = dlpar_cpu_remove(dn, drc_index);
>> +    rc = dlpar_cpu_remove(dn, drc_index, release_drc);
>>      of_node_put(dn);
>>      return rc;
>>  }
>> @@ -699,7 +709,7 @@ static int dlpar_cpu_remove_by_count(u32 cpus_to_remove)
>>      }
>>
>>      for (i = 0; i < cpus_to_remove; i++) {
>> -        rc = dlpar_cpu_remove_by_index(cpu_drcs[i]);
>> +        rc = dlpar_cpu_remove_by_index(cpu_drcs[i], true);
>>          if (rc)
>>              break;
>>
>> @@ -710,7 +720,7 @@ static int dlpar_cpu_remove_by_count(u32 cpus_to_remove)
>>          pr_warn("CPU hot-remove failed, adding back removed CPUs\n");
>>
>>          for (i = 0; i < cpus_removed; i++)
>> -            dlpar_cpu_add(cpu_drcs[i]);
>> +            dlpar_cpu_add(cpu_drcs[i], true);
>>
>>          rc = -EINVAL;
>>      } else {
>> @@ -780,7 +790,7 @@ static int dlpar_cpu_add_by_count(u32 cpus_to_add)
>>      }
>>
>>      for (i = 0; i < cpus_to_add; i++) {
>> -        rc = dlpar_cpu_add(cpu_drcs[i]);
>> +        rc = dlpar_cpu_add(cpu_drcs[i], true);
>>          if (rc)
>>              break;
>>
>> @@ -791,7 +801,7 @@ static int dlpar_cpu_add_by_count(u32 cpus_to_add)
>>          pr_warn("CPU hot-add failed, removing any added CPUs\n");
>>
>>          for (i = 0; i < cpus_added; i++)
>> -            dlpar_cpu_remove_by_index(cpu_drcs[i]);
>> +            dlpar_cpu_remove_by_index(cpu_drcs[i], true);
>>
>>          rc = -EINVAL;
>>      } else {
>> @@ -817,7 +827,7 @@ int dlpar_cpu(struct pseries_hp_errorlog *hp_elog)
>>          if (hp_elog->id_type == PSERIES_HP_ELOG_ID_DRC_COUNT)
>>              rc = dlpar_cpu_remove_by_count(count);
>>          else if (hp_elog->id_type == PSERIES_HP_ELOG_ID_DRC_INDEX)
>> -            rc = dlpar_cpu_remove_by_index(drc_index);
>> +            rc = dlpar_cpu_remove_by_index(drc_index, true);
>>          else
>>              rc = -EINVAL;
>>          break;
>> @@ -825,7 +835,7 @@ int dlpar_cpu(struct pseries_hp_errorlog *hp_elog)
>>          if (hp_elog->id_type == PSERIES_HP_ELOG_ID_DRC_COUNT)
>>              rc = dlpar_cpu_add_by_count(count);
>>          else if (hp_elog->id_type == PSERIES_HP_ELOG_ID_DRC_INDEX)
>> -            rc = dlpar_cpu_add(drc_index);
>> +            rc = dlpar_cpu_add(drc_index, true);
>>          else
>>              rc = -EINVAL;
>>          break;
>> @@ -850,7 +860,7 @@ static ssize_t dlpar_cpu_probe(const char *buf, size_t count)
>>      if (rc)
>>          return -EINVAL;
>>
>> -    rc = dlpar_cpu_add(drc_index);
>> +    rc = dlpar_cpu_add(drc_index, true);
>>
>>      return rc ? rc : count;
>>  }
>> @@ -871,7 +881,7 @@ static ssize_t dlpar_cpu_release(const char *buf, size_t count)
>>          return -EINVAL;
>>      }
>>
>> -    rc = dlpar_cpu_remove(dn, drc_index);
>> +    rc = dlpar_cpu_remove(dn, drc_index, true);
>>      of_node_put(dn);
>>
>>      return rc ? rc : count;
>>
>>
> 
> 

-- 
Michael W. Bringmann
Linux Technology Center
IBM Corporation
Tie-Line  363-5196
External: (512) 286-5196
Cell:       (512) 466-0650
mwb@linux.vnet.ibm.com