[PATCH] spufs: Improved SPU preemptability [part 2].

This patch reduces lock complexity of SPU scheduler, particularly
for involuntary preemptive switches.  As a result the new code
does a better job of mapping the highest priority tasks to SPUs.

Lock complexity is reduced by using the system default workqueue
to perform involuntary saves.  In this way we avoid nasty lock
ordering problems that the previous code had.  A "minimum timeslice"
for SPU contexts is also introduced.  The intent here is to avoid
thrashing.

While the new scheduler does a better job at prioritization it
still does nothing for fairness.

From: Mark Nutter <mnutter@us.ibm.com>
Signed-off-by: Arnd Bergmann <arndb@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>

1 files changed, 55 insertions, 26 deletions

diff --git a/arch/powerpc/platforms/cell/spufs/sched.c b/arch/powerpc/platforms/cell/spufs/sched.c
index e2f10b5b8a6a..fccc7709adbe 100644
--- a/arch/powerpc/platforms/cell/spufs/sched.c
+++ b/arch/powerpc/platforms/cell/spufs/sched.c
@@ -45,6 +45,8 @@
 #include <asm/spu_csa.h>
 #include "spufs.h"
 
+#define SPU_MIN_TIMESLICE 	(100 * HZ / 1000))
+
 #define SPU_BITMAP_SIZE (((MAX_PRIO+BITS_PER_LONG)/BITS_PER_LONG)+1)
 struct spu_prio_array {
 	atomic_t nr_blocked;
@@ -168,6 +170,7 @@ static inline void bind_context(struct spu *spu, struct spu_context *ctx)
 		 spu->number);
 	spu->ctx = ctx;
 	spu->flags = 0;
+	ctx->flags = 0;
 	ctx->spu = spu;
 	ctx->ops = &spu_hw_ops;
 	spu->pid = current->pid;
@@ -180,6 +183,7 @@ static inline void bind_context(struct spu *spu, struct spu_context *ctx)
 	mb();
 	spu_unmap_mappings(ctx);
 	spu_restore(&ctx->csa, spu);
+	spu->timestamp = jiffies;
 }
 
 static inline void unbind_context(struct spu *spu, struct spu_context *ctx)
@@ -188,6 +192,7 @@ static inline void unbind_context(struct spu *spu, struct spu_context *ctx)
 		 spu->pid, spu->number);
 	spu_unmap_mappings(ctx);
 	spu_save(&ctx->csa, spu);
+	spu->timestamp = jiffies;
 	ctx->state = SPU_STATE_SAVED;
 	spu->ibox_callback = NULL;
 	spu->wbox_callback = NULL;
@@ -197,38 +202,62 @@ static inline void unbind_context(struct spu *spu, struct spu_context *ctx)
 	spu->prio = MAX_PRIO;
 	ctx->ops = &spu_backing_ops;
 	ctx->spu = NULL;
+	ctx->flags = 0;
+	spu->flags = 0;
 	spu->ctx = NULL;
 }
 
-static struct spu *preempt_active(struct spu_runqueue *rq)
+static void spu_reaper(void *data)
 {
-	struct list_head *p;
-	struct spu *worst, *spu;
+	struct spu_context *ctx = data;
+	struct spu *spu;
 
-	worst = list_entry(rq->active_list.next, struct spu, sched_list);
-	list_for_each(p, &rq->active_list) {
-		spu = list_entry(p, struct spu, sched_list);
-		if (spu->prio > worst->prio) {
-			worst = spu;
+	down_write(&ctx->state_sema);
+	spu = ctx->spu;
+	if (spu && (ctx->flags & SPU_CONTEXT_PREEMPT)) {
+		if (atomic_read(&spu->rq->prio.nr_blocked)) {
+			pr_debug("%s: spu=%d\n", __func__, spu->number);
+			ctx->ops->runcntl_stop(ctx);
+			spu_deactivate(ctx);
+			wake_up_all(&ctx->stop_wq);
+		} else {
+			clear_bit(SPU_CONTEXT_PREEMPT_nr, &ctx->flags);
 		}
 	}
-	if (current->prio < worst->prio) {
-		struct spu_context *ctx = worst->ctx;
+	up_write(&ctx->state_sema);
+	put_spu_context(ctx);
+}
 
-		spu = worst;
-		if (down_write_trylock(&ctx->state_sema)) {
-			pr_debug("%s: booting pid=%d from SPU %d\n",
-				 __FUNCTION__, spu->pid, spu->number);
-			del_active(rq, spu);
-			up(&rq->sem);
-			wake_up_all(&ctx->stop_wq);
-			ctx->ops->runcntl_stop(ctx);
-			unbind_context(spu, ctx);
-			up_write(&ctx->state_sema);
-			return spu;
+static void schedule_spu_reaper(struct spu_runqueue *rq, struct spu *spu)
+{
+	struct spu_context *ctx = get_spu_context(spu->ctx);
+	unsigned long now = jiffies;
+	unsigned long expire = spu->timestamp + SPU_MIN_TIMESLICE;
+
+	set_bit(SPU_CONTEXT_PREEMPT_nr, &ctx->flags);
+	INIT_WORK(&ctx->reap_work, spu_reaper, ctx);
+	if (time_after(now, expire))
+		schedule_work(&ctx->reap_work);
+	else
+		schedule_delayed_work(&ctx->reap_work, expire - now);
+}
+
+static void check_preempt_active(struct spu_runqueue *rq)
+{
+	struct list_head *p;
+	struct spu *worst = NULL;
+
+	list_for_each(p, &rq->active_list) {
+		struct spu *spu = list_entry(p, struct spu, sched_list);
+		struct spu_context *ctx = spu->ctx;
+		if (!(ctx->flags & SPU_CONTEXT_PREEMPT)) {
+			if (!worst || (spu->prio > worst->prio)) {
+				worst = spu;
+			}
 		}
 	}
-	return NULL;
+	if (worst && (current->prio < worst->prio))
+		schedule_spu_reaper(rq, worst);
 }
 
 static struct spu *get_idle_spu(struct spu_context *ctx, u64 flags)
@@ -256,10 +285,7 @@ static struct spu *get_idle_spu(struct spu_context *ctx, u64 flags)
 				continue;
 			}
 		} else {
-			if (is_best_prio(rq)) {
-				if ((spu = preempt_active(rq)) != NULL)
-					return spu;
-			}
+			check_preempt_active(rq);
 			prio_wait(rq, ctx, flags);
 			if (signal_pending(current)) {
 				prio_wakeup(rq);
@@ -361,6 +387,8 @@ void spu_yield(struct spu_context *ctx)
 		spu_deactivate(ctx);
 		ctx->state = SPU_STATE_SAVED;
 		need_yield = 1;
+	} else if (spu) {
+		spu->prio = MAX_PRIO;
 	}
 	up_write(&ctx->state_sema);
 	if (unlikely(need_yield))
@@ -399,6 +427,7 @@ int __init spu_sched_init(void)
 		pr_debug("%s: adding SPU[%d]\n", __FUNCTION__, spu->number);
 		add_idle(rq, spu);
 		spu->rq = rq;
+		spu->timestamp = jiffies;
 	}
 	if (!rq->nr_idle) {
 		printk(KERN_WARNING "%s: No available SPUs.\n", __FUNCTION__);