arch/powerpc/platforms/pseries/eeh_event.c

   1 /*
   2  * eeh_event.c
   3  *
   4  * This program is free software; you can redistribute it and/or modify
   5  * it under the terms of the GNU General Public License as published by
   6  * the Free Software Foundation; either version 2 of the License, or
   7  * (at your option) any later version.
   8  *
   9  * This program is distributed in the hope that it will be useful,
  10  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  11  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  12  * GNU General Public License for more details.
  13  *
  14  * You should have received a copy of the GNU General Public License
  15  * along with this program; if not, write to the Free Software
  16  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
  17  *
  18  * Copyright (c) 2005 Linas Vepstas <linas@linas.org>
  19  */
  20
  21 #include <linux/list.h>
  22 #include <linux/mutex.h>
  23 #include <linux/pci.h>
  24 #include <linux/workqueue.h>
  25 #include <asm/eeh_event.h>
  26 #include <asm/ppc-pci.h>
  27
  28 /** Overview:
  29  *  EEH error states may be detected within exception handlers;
  30  *  however, the recovery processing needs to occur asynchronously
  31  *  in a normal kernel context and not an interrupt context.
  32  *  This pair of routines creates an event and queues it onto a
  33  *  work-queue, where a worker thread can drive recovery.
  34  */
  35
  36 /* EEH event workqueue setup. */
  37 static spinlock_t eeh_eventlist_lock = SPIN_LOCK_UNLOCKED;
  38 LIST_HEAD(eeh_eventlist);
  39 static void eeh_thread_launcher(void *);
  40 DECLARE_WORK(eeh_event_wq, eeh_thread_launcher, NULL);
  41
  42 /* Serialize reset sequences for a given pci device */
  43 DEFINE_MUTEX(eeh_event_mutex);
  44
  45 /**
  46  * eeh_event_handler - dispatch EEH events.
  47  * @dummy - unused
  48  *
  49  * The detection of a frozen slot can occur inside an interrupt,
  50  * where it can be hard to do anything about it.  The goal of this
  51  * routine is to pull these detection events out of the context
  52  * of the interrupt handler, and re-dispatch them for processing
  53  * at a later time in a normal context.
  54  */
  55 static int eeh_event_handler(void * dummy)
  56 {
  57         unsigned long flags;
  58         struct eeh_event        *event;
  59
  60         daemonize ("eehd");
  61
  62         while (1) {
  63                 set_current_state(TASK_INTERRUPTIBLE);
  64
  65                 spin_lock_irqsave(&eeh_eventlist_lock, flags);
  66                 event = NULL;
  67
  68                 /* Unqueue the event, get ready to process. */
  69                 if (!list_empty(&eeh_eventlist)) {
  70                         event = list_entry(eeh_eventlist.next, struct eeh_event, list);
  71                         list_del(&event->list);
  72                 }
  73                 spin_unlock_irqrestore(&eeh_eventlist_lock, flags);
  74
  75                 if (event == NULL)
  76                         break;
  77
  78                 /* Serialize processing of EEH events */
  79                 mutex_lock(&eeh_event_mutex);
  80                 eeh_mark_slot(event->dn, EEH_MODE_RECOVERING);
  81
  82                 printk(KERN_INFO "EEH: Detected PCI bus error on device %s\n",
  83                        pci_name(event->dev));
  84
  85                 handle_eeh_events(event);
  86
  87                 eeh_clear_slot(event->dn, EEH_MODE_RECOVERING);
  88                 pci_dev_put(event->dev);
  89                 kfree(event);
  90                 mutex_unlock(&eeh_event_mutex);
  91         }
  92
  93         return 0;
  94 }
  95
  96 /**
  97  * eeh_thread_launcher
  98  * @dummy - unused
  99  */
 100 static void eeh_thread_launcher(void *dummy)
 101 {
 102         if (kernel_thread(eeh_event_handler, NULL, CLONE_KERNEL) < 0)
 103                 printk(KERN_ERR "Failed to start EEH daemon\n");
 104 }
 105
 106 /**
 107  * eeh_send_failure_event - generate a PCI error event
 108  * @dev pci device
 109  *
 110  * This routine can be called within an interrupt context;
 111  * the actual event will be delivered in a normal context
 112  * (from a workqueue).
 113  */
 114 int eeh_send_failure_event (struct device_node *dn,
 115                             struct pci_dev *dev,
 116                             enum pci_channel_state state,
 117                             int time_unavail)
 118 {
 119         unsigned long flags;
 120         struct eeh_event *event;
 121         char *location;
 122
 123         if (!mem_init_done) {
 124                 printk(KERN_ERR "EEH: event during early boot not handled\n");
 125                 location = (char *) get_property(dn, "ibm,loc-code", NULL);
 126                 printk(KERN_ERR "EEH: device node = %s\n", dn->full_name);
 127                 printk(KERN_ERR "EEH: PCI location = %s\n", location);
 128                 return 1;
 129         }
 130         event = kmalloc(sizeof(*event), GFP_ATOMIC);
 131         if (event == NULL) {
 132                 printk (KERN_ERR "EEH: out of memory, event not handled\n");
 133                 return 1;
 134         }
 135
 136         if (dev)
 137                 pci_dev_get(dev);
 138
 139         event->dn = dn;
 140         event->dev = dev;
 141         event->state = state;
 142         event->time_unavail = time_unavail;
 143
 144         /* We may or may not be called in an interrupt context */
 145         spin_lock_irqsave(&eeh_eventlist_lock, flags);
 146         list_add(&event->list, &eeh_eventlist);
 147         spin_unlock_irqrestore(&eeh_eventlist_lock, flags);
 148
 149         schedule_work(&eeh_event_wq);
 150
 151         return 0;
 152 }
 153
 154 /********************** END OF FILE ******************************/