From: Maynard Johnson <maynardj@us.ibm.com>
To: michael@ellerman.id.au
Cc: Arnd Bergmann <arnd.bergmann@de.ibm.com>,
linuxppc-dev@ozlabs.org, cbe-oss-dev@ozlabs.org,
linux-kernel@vger.kernel.org
Subject: Re: [PATCH] Cell SPU task notification -- updated patch: #1
Date: Mon, 15 Jan 2007 16:39:32 -0600 [thread overview]
Message-ID: <45AC02A4.8080301@us.ibm.com> (raw)
In-Reply-To: <1168826871.4622.32.camel@concordia.ozlabs.ibm.com>
[-- Attachment #1: Type: text/plain, Size: 565 bytes --]
Attached is an updated patch that addresses Michael Ellerman's comments.
One comment made by Michael has not yet been addressed:
The comment was in regard to the for-loop in
spufs/sched.c:notify_spus_active(). He wondered if the scheduler can
swap a context from one node to another. If so, there's a small window
in this loop (where we switch the lock from one node's active list to
the next) where it may be possible we might miss waking up a context and
send a spurious wakeup to another.
Arnd . . . can you comment on this question?
Thanks.
-Maynard
[-- Attachment #2: spu-notifier.patch --]
[-- Type: text/x-diff, Size: 4285 bytes --]
Subject: Enable SPU switch notification to detect currently active SPU tasks.
From: Maynard Johnson <maynardj@us.ibm.com>
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@us.ibm.com>
Index: linux-2.6.19-rc6-arnd1+patches/arch/powerpc/platforms/cell/spufs/sched.c
===================================================================
--- linux-2.6.19-rc6-arnd1+patches.orig/arch/powerpc/platforms/cell/spufs/sched.c 2006-12-04 10:56:04.730698720 -0600
+++ linux-2.6.19-rc6-arnd1+patches/arch/powerpc/platforms/cell/spufs/sched.c 2007-01-15 16:22:31.808461448 -0600
@@ -84,15 +84,42 @@
ctx ? ctx->object_id : 0, spu);
}
+static void notify_spus_active(void)
+{
+ int node;
+ /* Wake up the active spu_contexts. When the awakened processes
+ * sees their notify_active flag is set, they will call
+ * spu_notify_already_active().
+ */
+ 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;
+ spu->notify_active = 1;
+ wake_up_all(&ctx->stop_wq);
+ smp_wmb();
+ }
+ mutex_unlock(&spu_prio->active_mutex[node]);
+ }
+ yield();
+}
+
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);
static inline void bind_context(struct spu *spu, struct spu_context *ctx)
@@ -250,6 +277,14 @@
return spu_get_idle(ctx, flags);
}
+void spu_notify_already_active(struct spu_context *ctx)
+{
+ struct spu *spu = ctx->spu;
+ if (!spu)
+ return;
+ spu_switch_notify(spu, ctx);
+}
+
/* The three externally callable interfaces
* for the scheduler begin here.
*
Index: linux-2.6.19-rc6-arnd1+patches/arch/powerpc/platforms/cell/spufs/spufs.h
===================================================================
--- linux-2.6.19-rc6-arnd1+patches.orig/arch/powerpc/platforms/cell/spufs/spufs.h 2007-01-08 18:18:40.093354608 -0600
+++ linux-2.6.19-rc6-arnd1+patches/arch/powerpc/platforms/cell/spufs/spufs.h 2007-01-08 18:31:03.610345792 -0600
@@ -183,6 +183,7 @@
void spu_yield(struct spu_context *ctx);
int __init spu_sched_init(void);
void __exit spu_sched_exit(void);
+void spu_notify_already_active(struct spu_context *ctx);
extern char *isolated_loader;
Index: linux-2.6.19-rc6-arnd1+patches/arch/powerpc/platforms/cell/spufs/run.c
===================================================================
--- linux-2.6.19-rc6-arnd1+patches.orig/arch/powerpc/platforms/cell/spufs/run.c 2007-01-08 18:33:51.979311680 -0600
+++ linux-2.6.19-rc6-arnd1+patches/arch/powerpc/platforms/cell/spufs/run.c 2007-01-15 16:31:30.104499992 -0600
@@ -45,9 +45,11 @@
u64 pte_fault;
*stat = ctx->ops->status_read(ctx);
- if (ctx->state != SPU_STATE_RUNNABLE)
- return 1;
+ smp_rmb();
+
spu = ctx->spu;
+ if (ctx->state != SPU_STATE_RUNNABLE || spu->notify_active)
+ 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;
@@ -304,6 +306,7 @@
u32 *npc, u32 *event)
{
int ret;
+ struct * spu;
u32 status;
if (down_interruptible(&ctx->run_sema))
@@ -317,8 +320,16 @@
do {
ret = spufs_wait(ctx->stop_wq, spu_stopped(ctx, &status));
+ spu = ctx->spu;
if (unlikely(ret))
break;
+ if (unlikely(spu->notify_active)) {
+ spu->notify_active = 0;
+ if (!(status & SPU_STATUS_STOPPED_BY_STOP)) {
+ spu_notify_already_active(ctx);
+ continue;
+ }
+ }
if ((status & SPU_STATUS_STOPPED_BY_STOP) &&
(status >> SPU_STOP_STATUS_SHIFT == 0x2104)) {
ret = spu_process_callback(ctx);
WARNING: multiple messages have this Message-ID (diff)
From: Maynard Johnson <maynardj@us.ibm.com>
To: michael@ellerman.id.au
Cc: cbe-oss-dev@ozlabs.org, linux-kernel@vger.kernel.org,
linuxppc-dev@ozlabs.org, Arnd Bergmann <arnd.bergmann@de.ibm.com>
Subject: Re: [PATCH] Cell SPU task notification -- updated patch: #1
Date: Mon, 15 Jan 2007 16:39:32 -0600 [thread overview]
Message-ID: <45AC02A4.8080301@us.ibm.com> (raw)
In-Reply-To: <1168826871.4622.32.camel@concordia.ozlabs.ibm.com>
[-- Attachment #1: Type: text/plain, Size: 565 bytes --]
Attached is an updated patch that addresses Michael Ellerman's comments.
One comment made by Michael has not yet been addressed:
The comment was in regard to the for-loop in
spufs/sched.c:notify_spus_active(). He wondered if the scheduler can
swap a context from one node to another. If so, there's a small window
in this loop (where we switch the lock from one node's active list to
the next) where it may be possible we might miss waking up a context and
send a spurious wakeup to another.
Arnd . . . can you comment on this question?
Thanks.
-Maynard
[-- Attachment #2: spu-notifier.patch --]
[-- Type: text/x-diff, Size: 4285 bytes --]
Subject: Enable SPU switch notification to detect currently active SPU tasks.
From: Maynard Johnson <maynardj@us.ibm.com>
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@us.ibm.com>
Index: linux-2.6.19-rc6-arnd1+patches/arch/powerpc/platforms/cell/spufs/sched.c
===================================================================
--- linux-2.6.19-rc6-arnd1+patches.orig/arch/powerpc/platforms/cell/spufs/sched.c 2006-12-04 10:56:04.730698720 -0600
+++ linux-2.6.19-rc6-arnd1+patches/arch/powerpc/platforms/cell/spufs/sched.c 2007-01-15 16:22:31.808461448 -0600
@@ -84,15 +84,42 @@
ctx ? ctx->object_id : 0, spu);
}
+static void notify_spus_active(void)
+{
+ int node;
+ /* Wake up the active spu_contexts. When the awakened processes
+ * sees their notify_active flag is set, they will call
+ * spu_notify_already_active().
+ */
+ 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;
+ spu->notify_active = 1;
+ wake_up_all(&ctx->stop_wq);
+ smp_wmb();
+ }
+ mutex_unlock(&spu_prio->active_mutex[node]);
+ }
+ yield();
+}
+
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);
static inline void bind_context(struct spu *spu, struct spu_context *ctx)
@@ -250,6 +277,14 @@
return spu_get_idle(ctx, flags);
}
+void spu_notify_already_active(struct spu_context *ctx)
+{
+ struct spu *spu = ctx->spu;
+ if (!spu)
+ return;
+ spu_switch_notify(spu, ctx);
+}
+
/* The three externally callable interfaces
* for the scheduler begin here.
*
Index: linux-2.6.19-rc6-arnd1+patches/arch/powerpc/platforms/cell/spufs/spufs.h
===================================================================
--- linux-2.6.19-rc6-arnd1+patches.orig/arch/powerpc/platforms/cell/spufs/spufs.h 2007-01-08 18:18:40.093354608 -0600
+++ linux-2.6.19-rc6-arnd1+patches/arch/powerpc/platforms/cell/spufs/spufs.h 2007-01-08 18:31:03.610345792 -0600
@@ -183,6 +183,7 @@
void spu_yield(struct spu_context *ctx);
int __init spu_sched_init(void);
void __exit spu_sched_exit(void);
+void spu_notify_already_active(struct spu_context *ctx);
extern char *isolated_loader;
Index: linux-2.6.19-rc6-arnd1+patches/arch/powerpc/platforms/cell/spufs/run.c
===================================================================
--- linux-2.6.19-rc6-arnd1+patches.orig/arch/powerpc/platforms/cell/spufs/run.c 2007-01-08 18:33:51.979311680 -0600
+++ linux-2.6.19-rc6-arnd1+patches/arch/powerpc/platforms/cell/spufs/run.c 2007-01-15 16:31:30.104499992 -0600
@@ -45,9 +45,11 @@
u64 pte_fault;
*stat = ctx->ops->status_read(ctx);
- if (ctx->state != SPU_STATE_RUNNABLE)
- return 1;
+ smp_rmb();
+
spu = ctx->spu;
+ if (ctx->state != SPU_STATE_RUNNABLE || spu->notify_active)
+ 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;
@@ -304,6 +306,7 @@
u32 *npc, u32 *event)
{
int ret;
+ struct * spu;
u32 status;
if (down_interruptible(&ctx->run_sema))
@@ -317,8 +320,16 @@
do {
ret = spufs_wait(ctx->stop_wq, spu_stopped(ctx, &status));
+ spu = ctx->spu;
if (unlikely(ret))
break;
+ if (unlikely(spu->notify_active)) {
+ spu->notify_active = 0;
+ if (!(status & SPU_STATUS_STOPPED_BY_STOP)) {
+ spu_notify_already_active(ctx);
+ continue;
+ }
+ }
if ((status & SPU_STATUS_STOPPED_BY_STOP) &&
(status >> SPU_STOP_STATUS_SHIFT == 0x2104)) {
ret = spu_process_callback(ctx);
next prev parent reply other threads:[~2007-01-15 22:39 UTC|newest]
Thread overview: 15+ messages / expand[flat|nested] mbox.gz Atom feed top
2007-01-12 22:03 [PATCH] Cell SPU task notification Maynard Johnson
2007-01-15 2:07 ` Michael Ellerman
2007-01-15 2:07 ` Michael Ellerman
2007-01-15 20:21 ` Maynard Johnson
2007-01-15 20:21 ` Maynard Johnson
2007-01-15 22:39 ` Maynard Johnson [this message]
2007-01-15 22:39 ` [PATCH] Cell SPU task notification -- updated patch: #1 Maynard Johnson
2007-01-17 0:30 ` [Cbe-oss-dev] [PATCH] Cell SPU task notification Christoph Hellwig
2007-01-17 0:30 ` Christoph Hellwig
2007-01-17 15:56 ` Maynard Johnson
2007-01-17 15:56 ` Maynard Johnson
2007-01-19 3:19 ` Christoph Hellwig
2007-01-19 3:19 ` Christoph Hellwig
2007-01-26 22:39 ` [Cbe-oss-dev] [PATCH] Cell SPU task notification - repost of patch with updates Maynard Johnson
2007-01-26 22:39 ` Maynard Johnson
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=45AC02A4.8080301@us.ibm.com \
--to=maynardj@us.ibm.com \
--cc=arnd.bergmann@de.ibm.com \
--cc=cbe-oss-dev@ozlabs.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linuxppc-dev@ozlabs.org \
--cc=michael@ellerman.id.au \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.