powerpc/eeh: Single kthread to handle events

We possiblly have multiple kthreads running for multiple EEH errors
(events) and use one spinlock to make the process of handling those
EEH events serialized. That's unnecessary and the patch creates only
one kthread, which is started during EEH core initialization time in
eeh_init(). A new semaphore introduced to count the number of existing
EEH events in the queue and the kthread waiting on the semaphore.

Signed-off-by: Gavin Shan <shangw@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

1 files changed, 49 insertions, 47 deletions

diff --git a/arch/powerpc/kernel/eeh_event.c b/arch/powerpc/kernel/eeh_event.c
index 185bedd926df..62e532ddfb68 100644
--- a/arch/powerpc/kernel/eeh_event.c
+++ b/arch/powerpc/kernel/eeh_event.c
@@ -18,11 +18,10 @@
 
 #include <linux/delay.h>
 #include <linux/list.h>
-#include <linux/mutex.h>
 #include <linux/sched.h>
+#include <linux/semaphore.h>
 #include <linux/pci.h>
 #include <linux/slab.h>
-#include <linux/workqueue.h>
 #include <linux/kthread.h>
 #include <asm/eeh_event.h>
 #include <asm/ppc-pci.h>
@@ -35,14 +34,9 @@
  *  work-queue, where a worker thread can drive recovery.
  */
 
-/* EEH event workqueue setup. */
 static DEFINE_SPINLOCK(eeh_eventlist_lock);
+static struct semaphore eeh_eventlist_sem;
 LIST_HEAD(eeh_eventlist);
-static void eeh_thread_launcher(struct work_struct *);
-DECLARE_WORK(eeh_event_wq, eeh_thread_launcher);
-
-/* Serialize reset sequences for a given pci device */
-DEFINE_MUTEX(eeh_event_mutex);
 
 /**
  * eeh_event_handler - Dispatch EEH events.
@@ -60,55 +54,62 @@ static int eeh_event_handler(void * dummy)
 	struct eeh_event *event;
 	struct eeh_pe *pe;
 
-	spin_lock_irqsave(&eeh_eventlist_lock, flags);
-	event = NULL;
-
-	/* Unqueue the event, get ready to process. */
-	if (!list_empty(&eeh_eventlist)) {
-		event = list_entry(eeh_eventlist.next, struct eeh_event, list);
-		list_del(&event->list);
-	}
-	spin_unlock_irqrestore(&eeh_eventlist_lock, flags);
-
-	if (event == NULL)
-		return 0;
-
-	/* Serialize processing of EEH events */
-	mutex_lock(&eeh_event_mutex);
-	pe = event->pe;
-	eeh_pe_state_mark(pe, EEH_PE_RECOVERING);
-	pr_info("EEH: Detected PCI bus error on PHB#%d-PE#%x\n",
-		pe->phb->global_number, pe->addr);
-
-	set_current_state(TASK_INTERRUPTIBLE);	/* Don't add to load average */
-	eeh_handle_event(pe);
-	eeh_pe_state_clear(pe, EEH_PE_RECOVERING);
-
-	kfree(event);
-	mutex_unlock(&eeh_event_mutex);
-
-	/* If there are no new errors after an hour, clear the counter. */
-	if (pe && pe->freeze_count > 0) {
-		msleep_interruptible(3600*1000);
-		if (pe->freeze_count > 0)
-			pe->freeze_count--;
-
+	while (!kthread_should_stop()) {
+		down(&eeh_eventlist_sem);
+
+		/* Fetch EEH event from the queue */
+		spin_lock_irqsave(&eeh_eventlist_lock, flags);
+		event = NULL;
+		if (!list_empty(&eeh_eventlist)) {
+			event = list_entry(eeh_eventlist.next,
+					   struct eeh_event, list);
+			list_del(&event->list);
+		}
+		spin_unlock_irqrestore(&eeh_eventlist_lock, flags);
+		if (!event)
+			continue;
+
+		/* We might have event without binding PE */
+		pe = event->pe;
+		if (pe) {
+			eeh_pe_state_mark(pe, EEH_PE_RECOVERING);
+			pr_info("EEH: Detected PCI bus error on PHB#%d-PE#%x\n",
+				 pe->phb->global_number, pe->addr);
+			eeh_handle_event(pe);
+			eeh_pe_state_clear(pe, EEH_PE_RECOVERING);
+		} else {
+			eeh_handle_event(NULL);
+		}
+
+		kfree(event);
 	}
 
 	return 0;
 }
 
 /**
- * eeh_thread_launcher - Start kernel thread to handle EEH events
- * @dummy - unused
+ * eeh_event_init - Start kernel thread to handle EEH events
  *
  * This routine is called to start the kernel thread for processing
  * EEH event.
  */
-static void eeh_thread_launcher(struct work_struct *dummy)
+int eeh_event_init(void)
 {
-	if (IS_ERR(kthread_run(eeh_event_handler, NULL, "eehd")))
-		printk(KERN_ERR "Failed to start EEH daemon\n");
+	struct task_struct *t;
+	int ret = 0;
+
+	/* Initialize semaphore */
+	sema_init(&eeh_eventlist_sem, 0);
+
+	t = kthread_run(eeh_event_handler, NULL, "eehd");
+	if (IS_ERR(t)) {
+		ret = PTR_ERR(t);
+		pr_err("%s: Failed to start EEH daemon (%d)\n",
+			__func__, ret);
+		return ret;
+	}
+
+	return 0;
 }
 
 /**
@@ -136,7 +137,8 @@ int eeh_send_failure_event(struct eeh_pe *pe)
 	list_add(&event->list, &eeh_eventlist);
 	spin_unlock_irqrestore(&eeh_eventlist_lock, flags);
 
-	schedule_work(&eeh_event_wq);
+	/* For EEH deamon to knick in */
+	up(&eeh_eventlist_sem);
 
 	return 0;
 }