LXR linux/arch/powerpc/platforms/pseries/eeh

   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/sched.h>
  25#include <linux/pci.h>
  26#include <linux/slab.h>
  27#include <linux/workqueue.h>
  28#include <asm/eeh_event.h>
  29#include <asm/ppc-pci.h>
  30
  31/** Overview:
  32 *  EEH error states may be detected within exception handlers;
  33 *  however, the recovery processing needs to occur asynchronously
  34 *  in a normal kernel context and not an interrupt context.
  35 *  This pair of routines creates an event and queues it onto a
  36 *  work-queue, where a worker thread can drive recovery.
  37 */
  38
  39/* EEH event workqueue setup. */
  40static DEFINE_SPINLOCK(eeh_eventlist_lock);
  41LIST_HEAD(eeh_eventlist);
  42static void eeh_thread_launcher(struct work_struct *);
  43DECLARE_WORK(eeh_event_wq, eeh_thread_launcher);
  44
  45/* Serialize reset sequences for a given pci device */
  46DEFINE_MUTEX(eeh_event_mutex);
  47
  48/**
  49 * eeh_event_handler - dispatch EEH events.
  50 * @dummy - unused
  51 *
  52 * The detection of a frozen slot can occur inside an interrupt,
  53 * where it can be hard to do anything about it.  The goal of this
  54 * routine is to pull these detection events out of the context
  55 * of the interrupt handler, and re-dispatch them for processing
  56 * at a later time in a normal context.
  57 */
  58static int eeh_event_handler(void * dummy)
  59{
  60        unsigned long flags;
  61        struct eeh_event        *event;
  62        struct pci_dn *pdn;
  63
  64        daemonize ("eehd");
  65        set_current_state(TASK_INTERRUPTIBLE);
  66
  67        spin_lock_irqsave(&eeh_eventlist_lock, flags);
  68        event = NULL;
  69
  70        /* Unqueue the event, get ready to process. */
  71        if (!list_empty(&eeh_eventlist)) {
  72                event = list_entry(eeh_eventlist.next, struct eeh_event, list);
  73                list_del(&event->list);
  74        }
  75        spin_unlock_irqrestore(&eeh_eventlist_lock, flags);
  76
  77        if (event == NULL)
  78                return 0;
  79
  80        /* Serialize processing of EEH events */
  81        mutex_lock(&eeh_event_mutex);
  82        eeh_mark_slot(event->dn, EEH_MODE_RECOVERING);
  83
  84        printk(KERN_INFO "EEH: Detected PCI bus error on device %s\n",
  85               eeh_pci_name(event->dev));
  86
  87        pdn = handle_eeh_events(event);
  88
  89        eeh_clear_slot(event->dn, EEH_MODE_RECOVERING);
  90        pci_dev_put(event->dev);
  91        kfree(event);
  92        mutex_unlock(&eeh_event_mutex);
  93
  94        /* If there are no new errors after an hour, clear the counter. */
  95        if (pdn && pdn->eeh_freeze_count>0) {
  96                msleep_interruptible (3600*1000);
  97                if (pdn->eeh_freeze_count>0)
  98                        pdn->eeh_freeze_count--;
  99        }
 100
 101        return 0;
 102}
 103
 104/**
 105 * eeh_thread_launcher
 106 * @dummy - unused
 107 */
 108static void eeh_thread_launcher(struct work_struct *dummy)
 109{
 110        if (kernel_thread(eeh_event_handler, NULL, CLONE_KERNEL) < 0)
 111                printk(KERN_ERR "Failed to start EEH daemon\n");
 112}
 113
 114/**
 115 * eeh_send_failure_event - generate a PCI error event
 116 * @dev pci device
 117 *
 118 * This routine can be called within an interrupt context;
 119 * the actual event will be delivered in a normal context
 120 * (from a workqueue).
 121 */
 122int eeh_send_failure_event (struct device_node *dn,
 123                            struct pci_dev *dev)
 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 = of_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
 148        /* We may or may not be called in an interrupt context */
 149        spin_lock_irqsave(&eeh_eventlist_lock, flags);
 150        list_add(&event->list, &eeh_eventlist);
 151        spin_unlock_irqrestore(&eeh_eventlist_lock, flags);
 152
 153        schedule_work(&eeh_event_wq);
 154
 155        return 0;
 156}
 157
 158/********************** END OF FILE ******************************/
 159