[PATCH 2/2] x86/intel_rdt: Update task closid immediately on CPU in rmdir and unmount

[Fenghua Yu]

From: Fenghua Yu <fenghua.yu@xxxxxxxxx>

When removing a sub directory/rdtgroup by rmdir or umount, closid in a
task in the sub directory is set to default rdtgroup's closid which is 0.
If the task is running on a CPU, the PQR_ASSOC MSR is only updated
when the task runs through a context switch. Up to the context switch,
the task runs with the wrong closid.

Make the change immediately effective by invoking a smp function call
on all online CPUs which calls intel_rdt_sched_in() to update the
PQR_ASSOC MSR.

rdt_update_closid() (renamed from rdt_update_percpu_closid()) calls
intel_rdt_sched_in() to update closid in the PQR_ASSOC MSR on a CPU.
The task closid and percpu closid are set up before
rdt_update_closid() is called. Handling PQR_ASSOC MSR update for
both task closid and percpu closid in rdt_update_closid() reduces
redundant smp function calls.

Signed-off-by: Fenghua Yu <fenghua.yu@xxxxxxxxx>
---
 arch/x86/kernel/cpu/intel_rdt_rdtgroup.c | 88 +++++++++++++++++++++++---------
 1 file changed, 63 insertions(+), 25 deletions(-)

diff --git a/arch/x86/kernel/cpu/intel_rdt_rdtgroup.c b/arch/x86/kernel/cpu/intel_rdt_rdtgroup.c
index eccea8a..ff0ee57 100644
--- a/arch/x86/kernel/cpu/intel_rdt_rdtgroup.c
+++ b/arch/x86/kernel/cpu/intel_rdt_rdtgroup.c
@@ -194,12 +194,14 @@ static int rdtgroup_cpus_show(struct kernfs_open_file *of,
 /*
  * This is safe against intel_rdt_sched_in() called from __switch_to()
  * because __switch_to() is executed with interrupts disabled. A local call
- * from rdt_update_percpu_closid() is proteced against __switch_to() because
+ * from rdt_update_closid() is proteced against __switch_to() because
  * preemption is disabled.
+ *
+ * Task closid and percpu closid should be set up before calling
+ * this function.
  */
-static void rdt_update_cpu_closid(void *v)
+static void rdt_update_cpu_closid(void *unused)
 {
-	this_cpu_write(cpu_closid, *(int *)v);
 	/*
 	 * We cannot unconditionally write the MSR because the current
 	 * executing task might have its own closid selected. Just reuse
@@ -208,14 +210,14 @@ static void rdt_update_cpu_closid(void *v)
 	intel_rdt_sched_in();
 }
 
-/* Update the per cpu closid and eventually the PGR_ASSOC MSR */
-static void rdt_update_percpu_closid(const struct cpumask *cpu_mask, int closid)
+/* Update closid in the PGR_ASSOC MSR */
+static void rdt_update_closid(const struct cpumask *cpu_mask)
 {
 	int cpu = get_cpu();
 
 	if (cpumask_test_cpu(cpu, cpu_mask))
-		rdt_update_cpu_closid(&closid);
-	smp_call_function_many(cpu_mask, rdt_update_cpu_closid, &closid, 1);
+		rdt_update_cpu_closid(NULL);
+	smp_call_function_many(cpu_mask, rdt_update_cpu_closid, NULL, 1);
 	put_cpu();
 }
 
@@ -224,7 +226,7 @@ static ssize_t rdtgroup_cpus_write(struct kernfs_open_file *of,
 {
 	cpumask_var_t tmpmask, newmask;
 	struct rdtgroup *rdtgrp, *r;
-	int ret;
+	int cpu, ret;
 
 	if (!buf)
 		return -EINVAL;
@@ -264,7 +266,9 @@ static ssize_t rdtgroup_cpus_write(struct kernfs_open_file *of,
 		/* Give any dropped cpus to rdtgroup_default */
 		cpumask_or(&rdtgroup_default.cpu_mask,
 			   &rdtgroup_default.cpu_mask, tmpmask);
-		rdt_update_percpu_closid(tmpmask, rdtgroup_default.closid);
+		for_each_cpu(cpu, tmpmask)
+			per_cpu(cpu_closid, cpu) = rdtgroup_default.closid;
+		rdt_update_closid(tmpmask);
 	}
 
 	/*
@@ -278,7 +282,9 @@ static ssize_t rdtgroup_cpus_write(struct kernfs_open_file *of,
 				continue;
 			cpumask_andnot(&r->cpu_mask, &r->cpu_mask, tmpmask);
 		}
-		rdt_update_percpu_closid(tmpmask, rdtgrp->closid);
+		for_each_cpu(cpu, tmpmask)
+			per_cpu(cpu_closid, cpu) = rdtgrp->closid;
+		rdt_update_closid(tmpmask);
 	}
 
 	/* Done pushing/pulling - update this group with new mask */
@@ -806,19 +812,33 @@ static int reset_all_cbms(struct rdt_resource *r)
 	return 0;
 }
 
+static int rdt_move_task_closid(struct rdtgroup *from, struct rdtgroup *to)
+{
+	struct task_struct *p, *t;
+	bool moved_task = false;
+
+	read_lock(&tasklist_lock);
+	for_each_process_thread(p, t) {
+		if (!from || t->closid == from->closid) {
+			t->closid = to->closid;
+			moved_task = true;
+		}
+	}
+	read_unlock(&tasklist_lock);
+
+	return moved_task;
+}
+
 /*
  * Forcibly remove all of subdirectories under root.
  */
 static void rmdir_all_sub(void)
 {
 	struct rdtgroup *rdtgrp, *tmp;
-	struct task_struct *p, *t;
+	int cpu;
 
 	/* move all tasks to default resource group */
-	read_lock(&tasklist_lock);
-	for_each_process_thread(p, t)
-		t->closid = 0;
-	read_unlock(&tasklist_lock);
+	rdt_move_task_closid(NULL, &rdtgroup_default);
 
 	list_for_each_entry_safe(rdtgrp, tmp, &rdt_all_groups, rdtgroup_list) {
 		/* Remove each rdtgroup other than root */
@@ -833,13 +853,18 @@ static void rmdir_all_sub(void)
 		cpumask_or(&rdtgroup_default.cpu_mask,
 			   &rdtgroup_default.cpu_mask, &rdtgrp->cpu_mask);
 
-		rdt_update_percpu_closid(&rdtgrp->cpu_mask,
-					 rdtgroup_default.closid);
+		for_each_cpu(cpu, &rdtgrp->cpu_mask)
+			per_cpu(cpu_closid, cpu) = rdtgroup_default.closid;
 
 		kernfs_remove(rdtgrp->kn);
 		list_del(&rdtgrp->rdtgroup_list);
 		kfree(rdtgrp);
 	}
+	/* Simply notify every online CPU to update PQR_ASSOC MSR */
+	get_online_cpus();
+	rdt_update_closid(cpu_online_mask);
+	put_online_cpus();
+
 	kernfs_remove(kn_info);
 }
 
@@ -944,8 +969,9 @@ static int rdtgroup_mkdir(struct kernfs_node *parent_kn, const char *name,
 
 static int rdtgroup_rmdir(struct kernfs_node *kn)
 {
-	struct task_struct *p, *t;
 	struct rdtgroup *rdtgrp;
+	bool moved_task = false;
+	int cpu;
 
 	rdtgrp = rdtgroup_kn_lock_live(kn);
 	if (!rdtgrp) {
@@ -954,17 +980,29 @@ static int rdtgroup_rmdir(struct kernfs_node *kn)
 	}
 
 	/* Give any tasks back to the default group */
-	read_lock(&tasklist_lock);
-	for_each_process_thread(p, t) {
-		if (t->closid == rdtgrp->closid)
-			t->closid = 0;
-	}
-	read_unlock(&tasklist_lock);
+	moved_task = rdt_move_task_closid(rdtgrp, &rdtgroup_default);
 
 	/* Give any CPUs back to the default group */
 	cpumask_or(&rdtgroup_default.cpu_mask,
 		   &rdtgroup_default.cpu_mask, &rdtgrp->cpu_mask);
-	rdt_update_percpu_closid(&rdtgrp->cpu_mask, rdtgroup_default.closid);
+
+	/* Update per cpu closid storage in rdtgrp */
+	for_each_cpu(cpu, &rdtgrp->cpu_mask)
+		per_cpu(cpu_closid, cpu) = rdtgroup_default.closid;
+
+	/*
+	 * If we moved any tasks out of this group, then force an
+	 * update on *all* CPUs so this will take effect right away.
+	 * Otherwise if we reallocated some CPUs from this group
+	 * we can just update the affected CPUs.
+	 */
+	if (moved_task) {
+		get_online_cpus();
+		rdt_update_closid(cpu_online_mask);
+		put_online_cpus();
+	} else if (!cpumask_empty(&rdtgrp->cpu_mask)) {
+		rdt_update_closid(&rdtgrp->cpu_mask);
+	}
 
 	rdtgrp->flags = RDT_DELETED;
 	closid_free(rdtgrp->closid);
-- 
2.5.0