path: root/0020-xen-sched-carve-out-memory-allocation-and-freeing-fr.patch

From c377ceab0a007690a1e71c81a5232613c99e944d Mon Sep 17 00:00:00 2001
From: Juergen Gross <jgross@suse.com>
Date: Tue, 11 Oct 2022 15:00:05 +0200
Subject: [PATCH 20/26] xen/sched: carve out memory allocation and freeing from
 schedule_cpu_rm()

In order to prepare not allocating or freeing memory from
schedule_cpu_rm(), move this functionality to dedicated functions.

For now call those functions from schedule_cpu_rm().

No change of behavior expected.

Signed-off-by: Juergen Gross <jgross@suse.com>
Acked-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
master commit: d42be6f83480b3ada286dc18444331a816be88a3
master date: 2022-09-05 11:42:30 +0100
---
 xen/common/sched/core.c    | 143 ++++++++++++++++++++++---------------
 xen/common/sched/private.h |  11 +++
 2 files changed, 98 insertions(+), 56 deletions(-)

diff --git a/xen/common/sched/core.c b/xen/common/sched/core.c
index 065a83eca912..2decb1161a63 100644
--- a/xen/common/sched/core.c
+++ b/xen/common/sched/core.c
@@ -3221,6 +3221,75 @@ out:
     return ret;
 }
 
+/*
+ * Allocate all memory needed for free_cpu_rm_data(), as allocations cannot
+ * be made in stop_machine() context.
+ *
+ * Between alloc_cpu_rm_data() and the real cpu removal action the relevant
+ * contents of struct sched_resource can't change, as the cpu in question is
+ * locked against any other movement to or from cpupools, and the data copied
+ * by alloc_cpu_rm_data() is modified only in case the cpu in question is
+ * being moved from or to a cpupool.
+ */
+struct cpu_rm_data *alloc_cpu_rm_data(unsigned int cpu)
+{
+    struct cpu_rm_data *data;
+    const struct sched_resource *sr;
+    unsigned int idx;
+
+    rcu_read_lock(&sched_res_rculock);
+
+    sr = get_sched_res(cpu);
+    data = xmalloc_flex_struct(struct cpu_rm_data, sr, sr->granularity - 1);
+    if ( !data )
+        goto out;
+
+    data->old_ops = sr->scheduler;
+    data->vpriv_old = idle_vcpu[cpu]->sched_unit->priv;
+    data->ppriv_old = sr->sched_priv;
+
+    for ( idx = 0; idx < sr->granularity - 1; idx++ )
+    {
+        data->sr[idx] = sched_alloc_res();
+        if ( data->sr[idx] )
+        {
+            data->sr[idx]->sched_unit_idle = sched_alloc_unit_mem();
+            if ( !data->sr[idx]->sched_unit_idle )
+            {
+                sched_res_free(&data->sr[idx]->rcu);
+                data->sr[idx] = NULL;
+            }
+        }
+        if ( !data->sr[idx] )
+        {
+            while ( idx > 0 )
+                sched_res_free(&data->sr[--idx]->rcu);
+            XFREE(data);
+            goto out;
+        }
+
+        data->sr[idx]->curr = data->sr[idx]->sched_unit_idle;
+        data->sr[idx]->scheduler = &sched_idle_ops;
+        data->sr[idx]->granularity = 1;
+
+        /* We want the lock not to change when replacing the resource. */
+        data->sr[idx]->schedule_lock = sr->schedule_lock;
+    }
+
+ out:
+    rcu_read_unlock(&sched_res_rculock);
+
+    return data;
+}
+
+void free_cpu_rm_data(struct cpu_rm_data *mem, unsigned int cpu)
+{
+    sched_free_udata(mem->old_ops, mem->vpriv_old);
+    sched_free_pdata(mem->old_ops, mem->ppriv_old, cpu);
+
+    xfree(mem);
+}
+
 /*
  * Remove a pCPU from its cpupool. Its scheduler becomes &sched_idle_ops
  * (the idle scheduler).
@@ -3229,53 +3298,23 @@ out:
  */
 int schedule_cpu_rm(unsigned int cpu)
 {
-    void *ppriv_old, *vpriv_old;
-    struct sched_resource *sr, **sr_new = NULL;
+    struct sched_resource *sr;
+    struct cpu_rm_data *data;
     struct sched_unit *unit;
-    struct scheduler *old_ops;
     spinlock_t *old_lock;
     unsigned long flags;
-    int idx, ret = -ENOMEM;
+    int idx = 0;
     unsigned int cpu_iter;
 
+    data = alloc_cpu_rm_data(cpu);
+    if ( !data )
+        return -ENOMEM;
+
     rcu_read_lock(&sched_res_rculock);
 
     sr = get_sched_res(cpu);
-    old_ops = sr->scheduler;
 
-    if ( sr->granularity > 1 )
-    {
-        sr_new = xmalloc_array(struct sched_resource *, sr->granularity - 1);
-        if ( !sr_new )
-            goto out;
-        for ( idx = 0; idx < sr->granularity - 1; idx++ )
-        {
-            sr_new[idx] = sched_alloc_res();
-            if ( sr_new[idx] )
-            {
-                sr_new[idx]->sched_unit_idle = sched_alloc_unit_mem();
-                if ( !sr_new[idx]->sched_unit_idle )
-                {
-                    sched_res_free(&sr_new[idx]->rcu);
-                    sr_new[idx] = NULL;
-                }
-            }
-            if ( !sr_new[idx] )
-            {
-                for ( idx--; idx >= 0; idx-- )
-                    sched_res_free(&sr_new[idx]->rcu);
-                goto out;
-            }
-            sr_new[idx]->curr = sr_new[idx]->sched_unit_idle;
-            sr_new[idx]->scheduler = &sched_idle_ops;
-            sr_new[idx]->granularity = 1;
-
-            /* We want the lock not to change when replacing the resource. */
-            sr_new[idx]->schedule_lock = sr->schedule_lock;
-        }
-    }
-
-    ret = 0;
+    ASSERT(sr->granularity);
     ASSERT(sr->cpupool != NULL);
     ASSERT(cpumask_test_cpu(cpu, &cpupool_free_cpus));
     ASSERT(!cpumask_test_cpu(cpu, sr->cpupool->cpu_valid));
@@ -3283,10 +3322,6 @@ int schedule_cpu_rm(unsigned int cpu)
     /* See comment in schedule_cpu_add() regarding lock switching. */
     old_lock = pcpu_schedule_lock_irqsave(cpu, &flags);
 
-    vpriv_old = idle_vcpu[cpu]->sched_unit->priv;
-    ppriv_old = sr->sched_priv;
-
-    idx = 0;
     for_each_cpu ( cpu_iter, sr->cpus )
     {
         per_cpu(sched_res_idx, cpu_iter) = 0;
@@ -3300,27 +3335,27 @@ int schedule_cpu_rm(unsigned int cpu)
         else
         {
             /* Initialize unit. */
-            unit = sr_new[idx]->sched_unit_idle;
-            unit->res = sr_new[idx];
+            unit = data->sr[idx]->sched_unit_idle;
+            unit->res = data->sr[idx];
             unit->is_running = true;
             sched_unit_add_vcpu(unit, idle_vcpu[cpu_iter]);
             sched_domain_insert_unit(unit, idle_vcpu[cpu_iter]->domain);
 
             /* Adjust cpu masks of resources (old and new). */
             cpumask_clear_cpu(cpu_iter, sr->cpus);
-            cpumask_set_cpu(cpu_iter, sr_new[idx]->cpus);
+            cpumask_set_cpu(cpu_iter, data->sr[idx]->cpus);
             cpumask_set_cpu(cpu_iter, &sched_res_mask);
 
             /* Init timer. */
-            init_timer(&sr_new[idx]->s_timer, s_timer_fn, NULL, cpu_iter);
+            init_timer(&data->sr[idx]->s_timer, s_timer_fn, NULL, cpu_iter);
 
             /* Last resource initializations and insert resource pointer. */
-            sr_new[idx]->master_cpu = cpu_iter;
-            set_sched_res(cpu_iter, sr_new[idx]);
+            data->sr[idx]->master_cpu = cpu_iter;
+            set_sched_res(cpu_iter, data->sr[idx]);
 
             /* Last action: set the new lock pointer. */
             smp_mb();
-            sr_new[idx]->schedule_lock = &sched_free_cpu_lock;
+            data->sr[idx]->schedule_lock = &sched_free_cpu_lock;
 
             idx++;
         }
@@ -3336,16 +3371,12 @@ int schedule_cpu_rm(unsigned int cpu)
     /* _Not_ pcpu_schedule_unlock(): schedule_lock may have changed! */
     spin_unlock_irqrestore(old_lock, flags);
 
-    sched_deinit_pdata(old_ops, ppriv_old, cpu);
+    sched_deinit_pdata(data->old_ops, data->ppriv_old, cpu);
 
-    sched_free_udata(old_ops, vpriv_old);
-    sched_free_pdata(old_ops, ppriv_old, cpu);
-
-out:
     rcu_read_unlock(&sched_res_rculock);
-    xfree(sr_new);
+    free_cpu_rm_data(data, cpu);
 
-    return ret;
+    return 0;
 }
 
 struct scheduler *scheduler_get_default(void)
diff --git a/xen/common/sched/private.h b/xen/common/sched/private.h
index 2b04b01a0c0a..e286849a1312 100644
--- a/xen/common/sched/private.h
+++ b/xen/common/sched/private.h
@@ -600,6 +600,15 @@ struct affinity_masks {
 
 bool alloc_affinity_masks(struct affinity_masks *affinity);
 void free_affinity_masks(struct affinity_masks *affinity);
+
+/* Memory allocation related data for schedule_cpu_rm(). */
+struct cpu_rm_data {
+    const struct scheduler *old_ops;
+    void *ppriv_old;
+    void *vpriv_old;
+    struct sched_resource *sr[];
+};
+
 void sched_rm_cpu(unsigned int cpu);
 const cpumask_t *sched_get_opt_cpumask(enum sched_gran opt, unsigned int cpu);
 void schedule_dump(struct cpupool *c);
@@ -608,6 +617,8 @@ struct scheduler *scheduler_alloc(unsigned int sched_id);
 void scheduler_free(struct scheduler *sched);
 int cpu_disable_scheduler(unsigned int cpu);
 int schedule_cpu_add(unsigned int cpu, struct cpupool *c);
+struct cpu_rm_data *alloc_cpu_rm_data(unsigned int cpu);
+void free_cpu_rm_data(struct cpu_rm_data *mem, unsigned int cpu);
 int schedule_cpu_rm(unsigned int cpu);
 int sched_move_domain(struct domain *d, struct cpupool *c);
 struct cpupool *cpupool_get_by_id(unsigned int poolid);
-- 
2.37.3