[Cbe-oss-dev] [RFC, PATCH 3/4] Add support to OProfile for profiling Cell BE SPUs -- update 2
Carl Love
cel at us.ibm.com
Thu Feb 22 10:25:01 EST 2007
Subject: Enable SPU switch notification to detect currently active SPU tasks.
From: Maynard Johnson <maynardj at us.ibm.com>, Carl Love <carll at us.ibm.com>
This repost contains the change to move the notify_active flag from ctx->spu to ctx as
discussed and agreed to by Arnd on 1/30/07. This patch is dependent on the series of SPU
patches from Christopher Hellwig.
This patch adds to the capability of spu_switch_event_register so that the
caller is also notified of currently active SPU tasks. It also exports
spu_switch_event_register and spu_switch_event_unregister.
Signed-off-by: Maynard Johnson <mpjohn at us.ibm.com>
Signed-off-by: Carl Love <carll at us.ibm.com>
Index: linux-2.6.20-be0706.1.20070213/arch/powerpc/platforms/cell/spufs/sched.c
===================================================================
--- linux-2.6.20-be0706.1.20070213.orig/arch/powerpc/platforms/cell/spufs/sched.c 2007-02-19 17:24:40.379148840 -0600
+++ linux-2.6.20-be0706.1.20070213/arch/powerpc/platforms/cell/spufs/sched.c 2007-02-20 08:25:08.077145400 -0600
@@ -137,21 +137,46 @@
static BLOCKING_NOTIFIER_HEAD(spu_switch_notifier);
-static void spu_switch_notify(struct spu *spu, struct spu_context *ctx)
+void spu_switch_notify(struct spu *spu, struct spu_context *ctx)
{
blocking_notifier_call_chain(&spu_switch_notifier,
ctx ? ctx->object_id : 0, spu);
}
+static void notify_spus_active(void)
+{
+ int node;
+ /* Wake up the active spu_contexts. When the awakened processes
+ * see their "notify_active" flag is set, they will call
+ * spu_switch_notify();
+ */
+ for (node = 0; node < MAX_NUMNODES; node++) {
+ struct spu *spu;
+ mutex_lock(&spu_prio->active_mutex[node]);
+ list_for_each_entry(spu, &spu_prio->active_list[node], list) {
+ struct spu_context *ctx = spu->ctx;
+ set_bit(SPU_SCHED_NOTIFY_ACTIVE, &ctx->sched_flags);
+ wake_up_all(&ctx->stop_wq);
+ }
+ mutex_unlock(&spu_prio->active_mutex[node]);
+ }
+}
+
int spu_switch_event_register(struct notifier_block * n)
{
- return blocking_notifier_chain_register(&spu_switch_notifier, n);
+ int ret;
+ ret = blocking_notifier_chain_register(&spu_switch_notifier, n);
+ if (!ret)
+ notify_spus_active();
+ return ret;
}
+EXPORT_SYMBOL_GPL(spu_switch_event_register);
int spu_switch_event_unregister(struct notifier_block * n)
{
return blocking_notifier_chain_unregister(&spu_switch_notifier, n);
}
+EXPORT_SYMBOL_GPL(spu_switch_event_unregister);
/**
* spu_bind_context - bind spu context to physical spu
Index: linux-2.6.20-be0706.1.20070213/arch/powerpc/platforms/cell/spufs/spufs.h
===================================================================
--- linux-2.6.20-be0706.1.20070213.orig/arch/powerpc/platforms/cell/spufs/spufs.h 2007-02-15 21:17:21.000000000 -0600
+++ linux-2.6.20-be0706.1.20070213/arch/powerpc/platforms/cell/spufs/spufs.h 2007-02-20 08:21:31.985246936 -0600
@@ -42,6 +42,7 @@
/* ctx->sched_flags */
enum {
SPU_SCHED_WAKE = 0,
+ SPU_SCHED_NOTIFY_ACTIVE,
};
struct spu_context {
@@ -220,6 +221,7 @@
int spu_activate(struct spu_context *ctx, unsigned long flags);
void spu_deactivate(struct spu_context *ctx);
void spu_yield(struct spu_context *ctx);
+void spu_switch_notify(struct spu *spu, struct spu_context *ctx);
void spu_start_tick(struct spu_context *ctx);
void spu_stop_tick(struct spu_context *ctx);
void spu_sched_tick(struct work_struct *work);
Index: linux-2.6.20-be0706.1.20070213/arch/powerpc/platforms/cell/spufs/run.c
===================================================================
--- linux-2.6.20-be0706.1.20070213.orig/arch/powerpc/platforms/cell/spufs/run.c 2007-02-15 21:17:20.000000000 -0600
+++ linux-2.6.20-be0706.1.20070213/arch/powerpc/platforms/cell/spufs/run.c 2007-02-20 08:41:08.230217536 -0600
@@ -45,9 +45,11 @@
u64 pte_fault;
*stat = ctx->ops->status_read(ctx);
- if (ctx->state != SPU_STATE_RUNNABLE)
- return 1;
+
spu = ctx->spu;
+ if (ctx->state != SPU_STATE_RUNNABLE ||
+ test_bit(SPU_SCHED_NOTIFY_ACTIVE, &ctx->sched_flags))
+ return 1;
pte_fault = spu->dsisr &
(MFC_DSISR_PTE_NOT_FOUND | MFC_DSISR_ACCESS_DENIED);
return (!(*stat & 0x1) || pte_fault || spu->class_0_pending) ? 1 : 0;
@@ -308,6 +310,7 @@
u32 *npc, u32 *event)
{
int ret;
+ struct spu * spu;
u32 status;
if (down_interruptible(&ctx->run_sema))
@@ -321,8 +324,17 @@
do {
ret = spufs_wait(ctx->stop_wq, spu_stopped(ctx, &status));
+ spu = ctx->spu;
if (unlikely(ret))
break;
+ if (unlikely(test_bit(SPU_SCHED_NOTIFY_ACTIVE,
+ &ctx->sched_flags))) {
+ clear_bit(SPU_SCHED_NOTIFY_ACTIVE, &ctx->sched_flags);
+ if (!(status & SPU_STATUS_STOPPED_BY_STOP)) {
+ spu_switch_notify(spu, ctx);
+ continue;
+ }
+ }
if ((status & SPU_STATUS_STOPPED_BY_STOP) &&
(status >> SPU_STOP_STATUS_SHIFT == 0x2104)) {
ret = spu_process_callback(ctx);
More information about the cbe-oss-dev
mailing list