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/delay.h>
  22 #include <linux/list.h>
  23 #include <linux/mutex.h>
  24 #include <linux/pci.h>
  25 #include <linux/workqueue.h>
  26 #include <asm/eeh_event.h>
  27 #include <asm/ppc-pci.h>
  28
  29 /** Overview:
  30  *  EEH error states may be detected within exception handlers;
  31  *  however, the recovery processing needs to occur asynchronously
  32  *  in a normal kernel context and not an interrupt context.
  33  *  This pair of routines creates an event and queues it onto a
  34  *  work-queue, where a worker thread can drive recovery.
  35  */
  36
  37 /* EEH event workqueue setup. */
  38 static DEFINE_SPINLOCK(eeh_eventlist_lock);
  39 LIST_HEAD(eeh_eventlist);
  40 static void eeh_thread_launcher(void *);
  41 DECLARE_WORK(eeh_event_wq, eeh_thread_launcher, NULL);
  42
  43 /* Serialize reset sequences for a given pci device */
  44 DEFINE_MUTEX(eeh_event_mutex);
  45
  46 /**
  47  * eeh_event_handler - dispatch EEH events.
  48  * @dummy - unused
  49  *
  50  * The detection of a frozen slot can occur inside an interrupt,
  51  * where it can be hard to do anything about it.  The goal of this
  52  * routine is to pull these detection events out of the context
  53  * of the interrupt handler, and re-dispatch them for processing
  54  * at a later time in a normal context.
  55  */
  56 static int eeh_event_handler(void * dummy)
  57 {
  58         unsigned long flags;
  59         struct eeh_event        *event;
  60         struct pci_dn *pdn;
  61
  62         daemonize ("eehd");
  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                 return 0;
  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         pdn = 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         /* If there are no new errors after an hour, clear the counter. */
  93         if (pdn && pdn->eeh_freeze_count>0) {
  94                 msleep_interruptible (3600*1000);
  95                 if (pdn->eeh_freeze_count>0)
  96                         pdn->eeh_freeze_count--;
  97         }
  98
  99         return 0;
 100 }
 101
 102 /**
 103  * eeh_thread_launcher
 104  * @dummy - unused
 105  */
 106 static void eeh_thread_launcher(void *dummy)
 107 {
 108         if (kernel_thread(eeh_event_handler, NULL, CLONE_KERNEL) < 0)
 109                 printk(KERN_ERR "Failed to start EEH daemon\n");
 110 }
 111
 112 /**
 113  * eeh_send_failure_event - generate a PCI error event
 114  * @dev pci device
 115  *
 116  * This routine can be called within an interrupt context;
 117  * the actual event will be delivered in a normal context
 118  * (from a workqueue).
 119  */
 120 int eeh_send_failure_event (struct device_node *dn,
 121                             struct pci_dev *dev,
 122                             enum pci_channel_state state,
 123                             int time_unavail)
 124 {
 125         unsigned long flags;
 126         struct eeh_event *event;
 127         const char *location;
 128
 129         if (!mem_init_done) {
 130                 printk(KERN_ERR "EEH: event during early boot not handled\n");
 131                 location = get_property(dn, "ibm,loc-code", NULL);
 132                 printk(KERN_ERR "EEH: device node = %s\n", dn->full_name);
 133                 printk(KERN_ERR "EEH: PCI location = %s\n", location);
 134                 return 1;
 135         }
 136         event = kmalloc(sizeof(*event), GFP_ATOMIC);
 137         if (event == NULL) {
 138                 printk (KERN_ERR "EEH: out of memory, event not handled\n");
 139                 return 1;
 140         }
 141
 142         if (dev)
 143                 pci_dev_get(dev);
 144
 145         event->dn = dn;
 146         event->dev = dev;
 147         event->state = state;
 148         event->time_unavail = time_unavail;
 149
 150         /* We may or may not be called in an interrupt context */
 151         spin_lock_irqsave(&eeh_eventlist_lock, flags);
 152         list_add(&event->list, &eeh_eventlist);
 153         spin_unlock_irqrestore(&eeh_eventlist_lock, flags);
 154
 155         schedule_work(&eeh_event_wq);
 156
 157         return 0;
 158 }
 159
 160 /********************** END OF FILE ******************************/