Skip to contentSkip to conversation

fix 2.6.32.23 suspend regression caused by commit 6f6198a

PATCH In review Mike Galbraith 2 participants 5 messages Oct 1, 2010

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 fix 2.6.32.23 suspend regression caused by commit 6f6198a

Greetings -stable/Greg,

Thoroughly tested scheduler fixes series was not so thoroughly tested with suspend.. like at all (my nvidia card doesn't do the resume part).

Sorry about that.

-Mike

From: Mike Galbraith <[email protected]>
Date: Fri, 1 Oct 2010 05:57:59 +0200
Subject: [PATCH] fix 2.6.32.23 suspend regression caused by commit 6f6198a

6f6198a sched: kill migration thread in CPU_POST_DEAD instead of CPU_DEAD leaves migration threads lying about. Mask out CPU_TASKS_FROZEN.

---
 kernel/sched.c |    7 +------
 1 files changed, 1 insertions(+), 6 deletions(-)
kernel/sched.c +1 -6
index 3480822..a675fd6 100644
@@ -7752,10 +7752,9 @@ migration_call(struct notifier_block *nfb, unsigned long action, void *hcpu)
 	unsigned long flags;
 	struct rq *rq;
 
-	switch (action) {
+	switch (action & ~CPU_TASKS_FROZEN) {
 
 	case CPU_UP_PREPARE:
-	case CPU_UP_PREPARE_FROZEN:
 		p = kthread_create(migration_thread, hcpu, "migration/%d", cpu);
 		if (IS_ERR(p))
 			return NOTIFY_BAD;
@@ -7770,7 +7769,6 @@ migration_call(struct notifier_block *nfb, unsigned long action, void *hcpu)
 		break;
 
 	case CPU_ONLINE:
-	case CPU_ONLINE_FROZEN:
 		/* Strictly unnecessary, as first user will wake it. */
 		wake_up_process(cpu_rq(cpu)->migration_thread);
 
@@ -7787,7 +7785,6 @@ migration_call(struct notifier_block *nfb, unsigned long action, void *hcpu)
 
 #ifdef CONFIG_HOTPLUG_CPU
 	case CPU_UP_CANCELED:
-	case CPU_UP_CANCELED_FROZEN:
 		if (!cpu_rq(cpu)->migration_thread)
 			break;
 		/* Unbind it from offline cpu so it can run. Fall thru. */
@@ -7812,7 +7809,6 @@ migration_call(struct notifier_block *nfb, unsigned long action, void *hcpu)
 		break;
 
 	case CPU_DEAD:
-	case CPU_DEAD_FROZEN:
 		migrate_live_tasks(cpu);
 		rq = cpu_rq(cpu);
 		/* Idle task back to normal (off runqueue, low prio) */
@@ -7846,7 +7842,6 @@ migration_call(struct notifier_block *nfb, unsigned long action, void *hcpu)
 		break;
 
 	case CPU_DYING:
-	case CPU_DYING_FROZEN:
 		/* Update our root-domain */
 		rq = cpu_rq(cpu);
 		spin_lock_irqsave(&rq->lock, flags);
Signature · 1 line
1.7.1
#1 lore raw

Again, with a less bouncy address.

From: Mike Galbraith <[email protected]>
Date: Fri, 1 Oct 2010 05:57:59 +0200
Subject: [PATCH] fix 2.6.32.23 suspend regression caused by commit 6f6198a

6f6198a sched: kill migration thread in CPU_POST_DEAD instead of CPU_DEAD
leaves migration threads lying about. Mask out CPU_TASKS_FROZEN.

---
 kernel/sched.c |    7 +------
 1 files changed, 1 insertions(+), 6 deletions(-)
kernel/sched.c +0 -0
index 3480822..a675fd6 100644
@@ -7752,10 +7752,9 @@ migration_call(struct notifier_block *nfb,

unsigned long action, void *hcpu)

        unsigned long flags;
        struct rq *rq;

- switch (action) {
+ switch (action & ~CPU_TASKS_FROZEN) {

case CPU_UP_PREPARE:
- case CPU_UP_PREPARE_FROZEN:
p = kthread_create(migration_thread, hcpu, "migration/%

d", cpu);
                if (IS_ERR(p))
                        return NOTIFY_BAD;
@@ -7770,7 +7769,6 @@ migration_call(struct notifier_block *nfb,
unsigned long action, void *hcpu)
                break;
 
        case CPU_ONLINE:
-       case CPU_ONLINE_FROZEN:
                /* Strictly unnecessary, as first user will wake it. */
                wake_up_process(cpu_rq(cpu)->migration_thread);

@@ -7787,7 +7785,6 @@ migration_call(struct notifier_block *nfb, unsigned long action, void *hcpu)

#ifdef CONFIG_HOTPLUG_CPU
case CPU_UP_CANCELED:
- case CPU_UP_CANCELED_FROZEN:

                if (!cpu_rq(cpu)->migration_thread)
                        break;
                /* Unbind it from offline cpu so it can run. Fall thru.
*/
@@ -7812,7 +7809,6 @@ migration_call(struct notifier_block *nfb,
unsigned long action, void *hcpu)
                break;
 
        case CPU_DEAD:
-       case CPU_DEAD_FROZEN:
                migrate_live_tasks(cpu);
                rq = cpu_rq(cpu);
                /* Idle task back to normal (off runqueue, low prio) */
@@ -7846,7 +7842,6 @@ migration_call(struct notifier_block *nfb,
unsigned long action, void *hcpu)
                break;
 
        case CPU_DYING:
-       case CPU_DYING_FROZEN:
                /* Update our root-domain */
                rq = cpu_rq(cpu);
                spin_lock_irqsave(&rq->lock, flags);
Signature · 1 line
1.7.1
#2 lore raw
Quoted 7 lines · Mike Galbraith

Again, with a less bouncy address.

From: Mike Galbraith <[email protected]>
Date: Fri, 1 Oct 2010 05:57:59 +0200
Subject: [PATCH] fix 2.6.32.23 suspend regression caused by commit 6f6198a

Quoted 5 lines

6f6198a sched: kill migration thread in CPU_POST_DEAD instead of CPU_DEAD
leaves migration threads lying about. Mask out CPU_TASKS_FROZEN.

Is this patch upstream already? If so, what's the git commit id of it? If not, why isn't it also needed there?

thanks,

greg k-h

#3 lore raw
Quoted 8 lines · Greg KH
Quoted 7 lines · Mike Galbraith

Again, with a less bouncy address.

From: Mike Galbraith <[email protected]>
Date: Fri, 1 Oct 2010 05:57:59 +0200
Subject: [PATCH] fix 2.6.32.23 suspend regression caused by commit 6f6198a

Quoted 8 lines
Quoted 5 lines

6f6198a sched: kill migration thread in CPU_POST_DEAD instead of CPU_DEAD
leaves migration threads lying about. Mask out CPU_TASKS_FROZEN.

Is this patch upstream already? If so, what's the git commit id of it? If not, why isn't it also needed there?

No, it's not upstream, and will never go upstream because they don't need it. It's only needed for -stable because the fix incorporated in -stable is incomplete. Upstream already does what this patch adds.

-Mike

#4 lore raw
Quoted 13 lines · Mike Galbraith
Quoted 12 lines · Greg KH
Quoted 11 lines · Mike Galbraith

Again, with a less bouncy address.

From: Mike Galbraith <[email protected]>
Date: Fri, 1 Oct 2010 05:57:59 +0200
Subject: [PATCH] fix 2.6.32.23 suspend regression caused by commit 6f6198a

6f6198a sched: kill migration thread in CPU_POST_DEAD instead of CPU_DEAD
leaves migration threads lying about. Mask out CPU_TASKS_FROZEN.

Quoted 8 lines
Quoted 4 lines
Quoted 1 line

Is this patch upstream already? If so, what's the git commit id of it? If not, why isn't it also needed there?

No, it's not upstream, and will never go upstream because they don't need it. It's only needed for -stable because the fix incorporated in -stable is incomplete. Upstream already does what this patch adds.

Ah, ok, that makes sense, thanks. I'll queue it up.

greg k-h