2007-07-19 08:49:36 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* edac_device.c
|
|
|
|
* (C) 2007 www.douglaskthompson.com
|
|
|
|
*
|
|
|
|
* This file may be distributed under the terms of the
|
|
|
|
* GNU General Public License.
|
|
|
|
*
|
|
|
|
* Written by Doug Thompson <norsk5@xmission.com>
|
|
|
|
*
|
|
|
|
* edac_device API implementation
|
|
|
|
* 19 Jan 2007
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include <linux/module.h>
|
|
|
|
#include <linux/types.h>
|
|
|
|
#include <linux/smp.h>
|
|
|
|
#include <linux/init.h>
|
|
|
|
#include <linux/sysctl.h>
|
|
|
|
#include <linux/highmem.h>
|
|
|
|
#include <linux/timer.h>
|
|
|
|
#include <linux/slab.h>
|
2007-07-19 08:50:20 +00:00
|
|
|
#include <linux/jiffies.h>
|
2007-07-19 08:49:36 +00:00
|
|
|
#include <linux/spinlock.h>
|
|
|
|
#include <linux/list.h>
|
|
|
|
#include <linux/ctype.h>
|
|
|
|
#include <linux/workqueue.h>
|
|
|
|
#include <asm/uaccess.h>
|
|
|
|
#include <asm/page.h>
|
|
|
|
|
|
|
|
#include "edac_core.h"
|
|
|
|
#include "edac_module.h"
|
|
|
|
|
2007-07-19 08:50:30 +00:00
|
|
|
/* lock for the list: 'edac_device_list', manipulation of this list
|
|
|
|
* is protected by the 'device_ctls_mutex' lock
|
|
|
|
*/
|
2007-07-19 08:50:22 +00:00
|
|
|
static DEFINE_MUTEX(device_ctls_mutex);
|
2008-04-29 08:03:17 +00:00
|
|
|
static LIST_HEAD(edac_device_list);
|
2007-07-19 08:49:36 +00:00
|
|
|
|
|
|
|
#ifdef CONFIG_EDAC_DEBUG
|
|
|
|
static void edac_device_dump_device(struct edac_device_ctl_info *edac_dev)
|
|
|
|
{
|
2012-04-29 20:08:39 +00:00
|
|
|
edac_dbg(3, "\tedac_dev = %p dev_idx=%d\n",
|
|
|
|
edac_dev, edac_dev->dev_idx);
|
|
|
|
edac_dbg(4, "\tedac_dev->edac_check = %p\n", edac_dev->edac_check);
|
|
|
|
edac_dbg(3, "\tdev = %p\n", edac_dev->dev);
|
|
|
|
edac_dbg(3, "\tmod_name:ctl_name = %s:%s\n",
|
|
|
|
edac_dev->mod_name, edac_dev->ctl_name);
|
|
|
|
edac_dbg(3, "\tpvt_info = %p\n\n", edac_dev->pvt_info);
|
2007-07-19 08:49:36 +00:00
|
|
|
}
|
2007-07-19 08:49:58 +00:00
|
|
|
#endif /* CONFIG_EDAC_DEBUG */
|
2007-07-19 08:49:36 +00:00
|
|
|
|
2007-07-19 08:50:29 +00:00
|
|
|
|
2007-07-19 08:49:36 +00:00
|
|
|
/*
|
2007-07-19 08:50:20 +00:00
|
|
|
* edac_device_alloc_ctl_info()
|
|
|
|
* Allocate a new edac device control info structure
|
|
|
|
*
|
|
|
|
* The control structure is allocated in complete chunk
|
|
|
|
* from the OS. It is in turn sub allocated to the
|
2012-04-17 18:30:52 +00:00
|
|
|
* various objects that compose the structure
|
2007-07-19 08:50:20 +00:00
|
|
|
*
|
|
|
|
* The structure has a 'nr_instance' array within itself.
|
|
|
|
* Each instance represents a major component
|
|
|
|
* Example: L1 cache and L2 cache are 2 instance components
|
|
|
|
*
|
|
|
|
* Within each instance is an array of 'nr_blocks' blockoffsets
|
2007-07-19 08:49:36 +00:00
|
|
|
*/
|
|
|
|
struct edac_device_ctl_info *edac_device_alloc_ctl_info(
|
|
|
|
unsigned sz_private,
|
2007-07-19 08:50:20 +00:00
|
|
|
char *edac_device_name, unsigned nr_instances,
|
|
|
|
char *edac_block_name, unsigned nr_blocks,
|
|
|
|
unsigned offset_value, /* zero, 1, or other based offset */
|
2007-07-19 08:50:27 +00:00
|
|
|
struct edac_dev_sysfs_block_attribute *attrib_spec, unsigned nr_attrib,
|
|
|
|
int device_index)
|
2007-07-19 08:49:36 +00:00
|
|
|
{
|
|
|
|
struct edac_device_ctl_info *dev_ctl;
|
|
|
|
struct edac_device_instance *dev_inst, *inst;
|
|
|
|
struct edac_device_block *dev_blk, *blk_p, *blk;
|
2007-07-19 08:50:25 +00:00
|
|
|
struct edac_dev_sysfs_block_attribute *dev_attrib, *attrib_p, *attrib;
|
2007-07-19 08:49:36 +00:00
|
|
|
unsigned total_size;
|
|
|
|
unsigned count;
|
|
|
|
unsigned instance, block, attr;
|
2012-04-16 13:18:12 +00:00
|
|
|
void *pvt, *p;
|
2007-07-19 08:50:29 +00:00
|
|
|
int err;
|
2007-07-19 08:49:36 +00:00
|
|
|
|
2012-04-29 20:08:39 +00:00
|
|
|
edac_dbg(4, "instances=%d blocks=%d\n", nr_instances, nr_blocks);
|
2007-07-19 08:49:36 +00:00
|
|
|
|
2007-07-19 08:50:25 +00:00
|
|
|
/* Calculate the size of memory we need to allocate AND
|
|
|
|
* determine the offsets of the various item arrays
|
|
|
|
* (instance,block,attrib) from the start of an allocated structure.
|
|
|
|
* We want the alignment of each item (instance,block,attrib)
|
2007-07-19 08:49:36 +00:00
|
|
|
* to be at least as stringent as what the compiler would
|
|
|
|
* provide if we could simply hardcode everything into a single struct.
|
|
|
|
*/
|
2012-04-16 13:18:12 +00:00
|
|
|
p = NULL;
|
|
|
|
dev_ctl = edac_align_ptr(&p, sizeof(*dev_ctl), 1);
|
2007-07-19 08:49:36 +00:00
|
|
|
|
2007-07-19 08:50:25 +00:00
|
|
|
/* Calc the 'end' offset past end of ONE ctl_info structure
|
|
|
|
* which will become the start of the 'instance' array
|
|
|
|
*/
|
2012-04-16 13:18:12 +00:00
|
|
|
dev_inst = edac_align_ptr(&p, sizeof(*dev_inst), nr_instances);
|
2007-07-19 08:49:36 +00:00
|
|
|
|
2007-07-19 08:50:25 +00:00
|
|
|
/* Calc the 'end' offset past the instance array within the ctl_info
|
|
|
|
* which will become the start of the block array
|
|
|
|
*/
|
2012-04-16 13:18:12 +00:00
|
|
|
count = nr_instances * nr_blocks;
|
|
|
|
dev_blk = edac_align_ptr(&p, sizeof(*dev_blk), count);
|
2007-07-19 08:49:36 +00:00
|
|
|
|
2007-07-19 08:50:25 +00:00
|
|
|
/* Calc the 'end' offset past the dev_blk array
|
|
|
|
* which will become the start of the attrib array, if any.
|
|
|
|
*/
|
2012-04-16 13:18:12 +00:00
|
|
|
/* calc how many nr_attrib we need */
|
|
|
|
if (nr_attrib > 0)
|
2007-07-19 08:50:25 +00:00
|
|
|
count *= nr_attrib;
|
2012-04-16 13:18:12 +00:00
|
|
|
dev_attrib = edac_align_ptr(&p, sizeof(*dev_attrib), count);
|
2007-07-19 08:49:36 +00:00
|
|
|
|
2012-04-16 13:18:12 +00:00
|
|
|
/* Calc the 'end' offset past the attributes array */
|
|
|
|
pvt = edac_align_ptr(&p, sz_private, 1);
|
2007-07-19 08:50:25 +00:00
|
|
|
|
|
|
|
/* 'pvt' now points to where the private data area is.
|
|
|
|
* At this point 'pvt' (like dev_inst,dev_blk and dev_attrib)
|
|
|
|
* is baselined at ZERO
|
|
|
|
*/
|
2007-07-19 08:49:58 +00:00
|
|
|
total_size = ((unsigned long)pvt) + sz_private;
|
2007-07-19 08:49:36 +00:00
|
|
|
|
|
|
|
/* Allocate the amount of memory for the set of control structures */
|
2007-07-19 08:50:20 +00:00
|
|
|
dev_ctl = kzalloc(total_size, GFP_KERNEL);
|
|
|
|
if (dev_ctl == NULL)
|
2007-07-19 08:49:36 +00:00
|
|
|
return NULL;
|
|
|
|
|
2007-07-19 08:50:25 +00:00
|
|
|
/* Adjust pointers so they point within the actual memory we
|
|
|
|
* just allocated rather than an imaginary chunk of memory
|
|
|
|
* located at address 0.
|
|
|
|
* 'dev_ctl' points to REAL memory, while the others are
|
|
|
|
* ZERO based and thus need to be adjusted to point within
|
|
|
|
* the allocated memory.
|
2007-07-19 08:49:36 +00:00
|
|
|
*/
|
|
|
|
dev_inst = (struct edac_device_instance *)
|
2007-07-19 08:50:13 +00:00
|
|
|
(((char *)dev_ctl) + ((unsigned long)dev_inst));
|
2007-07-19 08:49:36 +00:00
|
|
|
dev_blk = (struct edac_device_block *)
|
2007-07-19 08:50:13 +00:00
|
|
|
(((char *)dev_ctl) + ((unsigned long)dev_blk));
|
2007-07-19 08:50:25 +00:00
|
|
|
dev_attrib = (struct edac_dev_sysfs_block_attribute *)
|
2007-07-19 08:50:13 +00:00
|
|
|
(((char *)dev_ctl) + ((unsigned long)dev_attrib));
|
2007-07-19 08:49:58 +00:00
|
|
|
pvt = sz_private ? (((char *)dev_ctl) + ((unsigned long)pvt)) : NULL;
|
2007-07-19 08:49:36 +00:00
|
|
|
|
2007-07-19 08:50:25 +00:00
|
|
|
/* Begin storing the information into the control info structure */
|
2007-07-19 08:50:27 +00:00
|
|
|
dev_ctl->dev_idx = device_index;
|
2007-07-19 08:49:36 +00:00
|
|
|
dev_ctl->nr_instances = nr_instances;
|
|
|
|
dev_ctl->instances = dev_inst;
|
|
|
|
dev_ctl->pvt_info = pvt;
|
|
|
|
|
2008-02-07 08:14:51 +00:00
|
|
|
/* Default logging of CEs and UEs */
|
|
|
|
dev_ctl->log_ce = 1;
|
|
|
|
dev_ctl->log_ue = 1;
|
|
|
|
|
2007-07-19 08:50:20 +00:00
|
|
|
/* Name of this edac device */
|
|
|
|
snprintf(dev_ctl->name,sizeof(dev_ctl->name),"%s",edac_device_name);
|
2007-07-19 08:49:36 +00:00
|
|
|
|
2012-04-29 20:08:39 +00:00
|
|
|
edac_dbg(4, "edac_dev=%p next after end=%p\n",
|
|
|
|
dev_ctl, pvt + sz_private);
|
2007-07-19 08:50:33 +00:00
|
|
|
|
2007-07-19 08:49:36 +00:00
|
|
|
/* Initialize every Instance */
|
|
|
|
for (instance = 0; instance < nr_instances; instance++) {
|
|
|
|
inst = &dev_inst[instance];
|
|
|
|
inst->ctl = dev_ctl;
|
|
|
|
inst->nr_blocks = nr_blocks;
|
|
|
|
blk_p = &dev_blk[instance * nr_blocks];
|
|
|
|
inst->blocks = blk_p;
|
|
|
|
|
|
|
|
/* name of this instance */
|
|
|
|
snprintf(inst->name, sizeof(inst->name),
|
2007-07-19 08:49:58 +00:00
|
|
|
"%s%u", edac_device_name, instance);
|
2007-07-19 08:49:36 +00:00
|
|
|
|
|
|
|
/* Initialize every block in each instance */
|
2007-07-19 08:49:58 +00:00
|
|
|
for (block = 0; block < nr_blocks; block++) {
|
2007-07-19 08:49:36 +00:00
|
|
|
blk = &blk_p[block];
|
|
|
|
blk->instance = inst;
|
|
|
|
snprintf(blk->name, sizeof(blk->name),
|
2007-07-19 08:50:11 +00:00
|
|
|
"%s%d", edac_block_name, block+offset_value);
|
2007-07-19 08:49:36 +00:00
|
|
|
|
2012-04-29 20:08:39 +00:00
|
|
|
edac_dbg(4, "instance=%d inst_p=%p block=#%d block_p=%p name='%s'\n",
|
|
|
|
instance, inst, block, blk, blk->name);
|
2007-07-19 08:49:36 +00:00
|
|
|
|
2007-07-19 08:50:25 +00:00
|
|
|
/* if there are NO attributes OR no attribute pointer
|
|
|
|
* then continue on to next block iteration
|
|
|
|
*/
|
|
|
|
if ((nr_attrib == 0) || (attrib_spec == NULL))
|
|
|
|
continue;
|
|
|
|
|
|
|
|
/* setup the attribute array for this block */
|
|
|
|
blk->nr_attribs = nr_attrib;
|
|
|
|
attrib_p = &dev_attrib[block*nr_instances*nr_attrib];
|
|
|
|
blk->block_attributes = attrib_p;
|
|
|
|
|
2012-04-29 20:08:39 +00:00
|
|
|
edac_dbg(4, "THIS BLOCK_ATTRIB=%p\n",
|
|
|
|
blk->block_attributes);
|
2007-07-19 08:50:33 +00:00
|
|
|
|
2007-07-19 08:50:25 +00:00
|
|
|
/* Initialize every user specified attribute in this
|
|
|
|
* block with the data the caller passed in
|
2007-07-19 08:50:33 +00:00
|
|
|
* Each block gets its own copy of pointers,
|
|
|
|
* and its unique 'value'
|
2007-07-19 08:50:25 +00:00
|
|
|
*/
|
|
|
|
for (attr = 0; attr < nr_attrib; attr++) {
|
|
|
|
attrib = &attrib_p[attr];
|
|
|
|
|
2007-07-19 08:50:33 +00:00
|
|
|
/* populate the unique per attrib
|
|
|
|
* with the code pointers and info
|
|
|
|
*/
|
|
|
|
attrib->attr = attrib_spec[attr].attr;
|
|
|
|
attrib->show = attrib_spec[attr].show;
|
|
|
|
attrib->store = attrib_spec[attr].store;
|
|
|
|
|
|
|
|
attrib->block = blk; /* up link */
|
|
|
|
|
2012-04-29 20:08:39 +00:00
|
|
|
edac_dbg(4, "alloc-attrib=%p attrib_name='%s' attrib-spec=%p spec-name=%s\n",
|
|
|
|
attrib, attrib->attr.name,
|
|
|
|
&attrib_spec[attr],
|
|
|
|
attrib_spec[attr].attr.name
|
2007-07-19 08:50:33 +00:00
|
|
|
);
|
2007-07-19 08:49:36 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Mark this instance as merely ALLOCATED */
|
|
|
|
dev_ctl->op_state = OP_ALLOC;
|
|
|
|
|
2007-07-19 08:50:29 +00:00
|
|
|
/*
|
|
|
|
* Initialize the 'root' kobj for the edac_device controller
|
|
|
|
*/
|
|
|
|
err = edac_device_register_sysfs_main_kobj(dev_ctl);
|
|
|
|
if (err) {
|
|
|
|
kfree(dev_ctl);
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* at this point, the root kobj is valid, and in order to
|
|
|
|
* 'free' the object, then the function:
|
|
|
|
* edac_device_unregister_sysfs_main_kobj() must be called
|
|
|
|
* which will perform kobj unregistration and the actual free
|
|
|
|
* will occur during the kobject callback operation
|
|
|
|
*/
|
|
|
|
|
2007-07-19 08:49:36 +00:00
|
|
|
return dev_ctl;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(edac_device_alloc_ctl_info);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* edac_device_free_ctl_info()
|
|
|
|
* frees the memory allocated by the edac_device_alloc_ctl_info()
|
|
|
|
* function
|
|
|
|
*/
|
2007-07-19 08:49:58 +00:00
|
|
|
void edac_device_free_ctl_info(struct edac_device_ctl_info *ctl_info)
|
|
|
|
{
|
2007-07-19 08:50:29 +00:00
|
|
|
edac_device_unregister_sysfs_main_kobj(ctl_info);
|
2007-07-19 08:49:36 +00:00
|
|
|
}
|
2007-07-19 08:49:58 +00:00
|
|
|
EXPORT_SYMBOL_GPL(edac_device_free_ctl_info);
|
2007-07-19 08:49:36 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* find_edac_device_by_dev
|
|
|
|
* scans the edac_device list for a specific 'struct device *'
|
2007-07-19 08:50:20 +00:00
|
|
|
*
|
|
|
|
* lock to be held prior to call: device_ctls_mutex
|
|
|
|
*
|
|
|
|
* Return:
|
|
|
|
* pointer to control structure managing 'dev'
|
|
|
|
* NULL if not found on list
|
2007-07-19 08:49:36 +00:00
|
|
|
*/
|
2007-07-19 08:49:58 +00:00
|
|
|
static struct edac_device_ctl_info *find_edac_device_by_dev(struct device *dev)
|
2007-07-19 08:49:36 +00:00
|
|
|
{
|
|
|
|
struct edac_device_ctl_info *edac_dev;
|
|
|
|
struct list_head *item;
|
|
|
|
|
2012-04-29 20:08:39 +00:00
|
|
|
edac_dbg(0, "\n");
|
2007-07-19 08:49:36 +00:00
|
|
|
|
|
|
|
list_for_each(item, &edac_device_list) {
|
|
|
|
edac_dev = list_entry(item, struct edac_device_ctl_info, link);
|
|
|
|
|
|
|
|
if (edac_dev->dev == dev)
|
|
|
|
return edac_dev;
|
|
|
|
}
|
|
|
|
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* add_edac_dev_to_global_list
|
|
|
|
* Before calling this function, caller must
|
|
|
|
* assign a unique value to edac_dev->dev_idx.
|
2007-07-19 08:50:20 +00:00
|
|
|
*
|
|
|
|
* lock to be held prior to call: device_ctls_mutex
|
|
|
|
*
|
2007-07-19 08:49:36 +00:00
|
|
|
* Return:
|
|
|
|
* 0 on success
|
|
|
|
* 1 on failure.
|
|
|
|
*/
|
2007-07-19 08:49:58 +00:00
|
|
|
static int add_edac_dev_to_global_list(struct edac_device_ctl_info *edac_dev)
|
2007-07-19 08:49:36 +00:00
|
|
|
{
|
|
|
|
struct list_head *item, *insert_before;
|
|
|
|
struct edac_device_ctl_info *rover;
|
|
|
|
|
|
|
|
insert_before = &edac_device_list;
|
|
|
|
|
|
|
|
/* Determine if already on the list */
|
2007-07-19 08:50:20 +00:00
|
|
|
rover = find_edac_device_by_dev(edac_dev->dev);
|
|
|
|
if (unlikely(rover != NULL))
|
2007-07-19 08:49:36 +00:00
|
|
|
goto fail0;
|
|
|
|
|
|
|
|
/* Insert in ascending order by 'dev_idx', so find position */
|
|
|
|
list_for_each(item, &edac_device_list) {
|
|
|
|
rover = list_entry(item, struct edac_device_ctl_info, link);
|
|
|
|
|
|
|
|
if (rover->dev_idx >= edac_dev->dev_idx) {
|
|
|
|
if (unlikely(rover->dev_idx == edac_dev->dev_idx))
|
|
|
|
goto fail1;
|
|
|
|
|
|
|
|
insert_before = item;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
list_add_tail_rcu(&edac_dev->link, insert_before);
|
|
|
|
return 0;
|
|
|
|
|
2007-07-19 08:50:13 +00:00
|
|
|
fail0:
|
2007-07-19 08:49:36 +00:00
|
|
|
edac_printk(KERN_WARNING, EDAC_MC,
|
2007-07-19 08:50:13 +00:00
|
|
|
"%s (%s) %s %s already assigned %d\n",
|
2009-01-06 22:42:57 +00:00
|
|
|
dev_name(rover->dev), edac_dev_name(rover),
|
2007-07-19 08:50:13 +00:00
|
|
|
rover->mod_name, rover->ctl_name, rover->dev_idx);
|
2007-07-19 08:49:36 +00:00
|
|
|
return 1;
|
|
|
|
|
2007-07-19 08:50:13 +00:00
|
|
|
fail1:
|
2007-07-19 08:49:36 +00:00
|
|
|
edac_printk(KERN_WARNING, EDAC_MC,
|
2007-07-19 08:50:13 +00:00
|
|
|
"bug in low-level driver: attempt to assign\n"
|
|
|
|
" duplicate dev_idx %d in %s()\n", rover->dev_idx,
|
|
|
|
__func__);
|
2007-07-19 08:49:36 +00:00
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* del_edac_device_from_global_list
|
|
|
|
*/
|
2007-07-19 08:49:58 +00:00
|
|
|
static void del_edac_device_from_global_list(struct edac_device_ctl_info
|
2007-07-19 08:50:13 +00:00
|
|
|
*edac_device)
|
2007-07-19 08:49:36 +00:00
|
|
|
{
|
|
|
|
list_del_rcu(&edac_device->link);
|
2011-05-26 23:25:58 +00:00
|
|
|
|
|
|
|
/* these are for safe removal of devices from global list while
|
|
|
|
* NMI handlers may be traversing list
|
|
|
|
*/
|
|
|
|
synchronize_rcu();
|
|
|
|
INIT_LIST_HEAD(&edac_device->link);
|
2007-07-19 08:49:36 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
2007-07-19 08:49:52 +00:00
|
|
|
* edac_device_workq_function
|
2007-07-19 08:49:36 +00:00
|
|
|
* performs the operation scheduled by a workq request
|
2007-07-19 08:50:30 +00:00
|
|
|
*
|
|
|
|
* this workq is embedded within an edac_device_ctl_info
|
|
|
|
* structure, that needs to be polled for possible error events.
|
|
|
|
*
|
|
|
|
* This operation is to acquire the list mutex lock
|
2012-05-22 08:56:21 +00:00
|
|
|
* (thus preventing insertation or deletion)
|
2007-07-19 08:50:30 +00:00
|
|
|
* and then call the device's poll function IFF this device is
|
|
|
|
* running polled and there is a poll function defined.
|
2007-07-19 08:49:36 +00:00
|
|
|
*/
|
2007-07-19 08:49:52 +00:00
|
|
|
static void edac_device_workq_function(struct work_struct *work_req)
|
2007-07-19 08:49:36 +00:00
|
|
|
{
|
2009-04-13 21:40:21 +00:00
|
|
|
struct delayed_work *d_work = to_delayed_work(work_req);
|
2007-07-19 08:49:58 +00:00
|
|
|
struct edac_device_ctl_info *edac_dev = to_edac_device_ctl_work(d_work);
|
2007-07-19 08:49:36 +00:00
|
|
|
|
2007-07-19 08:50:22 +00:00
|
|
|
mutex_lock(&device_ctls_mutex);
|
2007-07-19 08:49:36 +00:00
|
|
|
|
edac: fix edac core deadlock when removing a device
When deleting an edac device, we have to wait for its edac_dev.work to be
completed before deleting the whole edac_dev structure. Since we have no
idea which work in current edac_poller's workqueue is the work we are
conerned about, we wait for all work in the edac_poller's workqueue to be
proceseed. This is done via flush_cpu_workqueue() which inserts a
wq_barrier into the tail of the workqueue and then sleeping on the
completion of this wq_barrier. The edac_poller will wake up sleepers when
it is found.
EDAC core creates only one kernel worker thread, edac_poller, to run the
works of all current edac devices. They share the same callback function
of edac_device_workq_function(), which would grab the mutex of
device_ctls_mutex first before it checks the device. This is exactly
where edac_poller and rmmod would have a great chance to deadlock.
In below call trace of rmmod > ... >
edac_device_del_device >
edac_device_workq_teardown > flush_workqueue > flush_cpu_workqueue,
device_ctls_mutex would have already been grabbed by
edac_device_del_device(). So, on one hand rmmod would sleep on the
completion of a wq_barrier, holding device_ctls_mutex; on the other hand
edac_poller would be blocked on the same mutex when it's running any one
of works of existing edac evices(Note, this edac_dev.work is likely to be
totally irrelevant to the one that is being removed right now)and never
would have a chance to run the work of above wq_barrier to wake rmmod up.
edac_device_workq_teardown() should not be called within the critical
region of device_ctls_mutex. Just like is done in edac_pci_del_device()
and edac_mc_del_mc(), where edac_pci_workq_teardown() and
edac_mc_workq_teardown() are called after related mutex are released.
Moreover, an edac_dev.work should check first if it is being removed. If
this is the case, then it should bail out immediately. Since not all of
existing edac devices are to be removed, this "shutting flag" should be
contained to edac device being removed. The current edac_dev.op_state can
be used to serve this purpose.
The original deadlock problem and the solution have been witnessed and
tested on actual hardware. Without the solution, rmmod an edac driver
would result in below deadlock:
root@localhost:/root> rmmod mv64x60_edac
EDAC DEBUG: mv64x60_dma_err_remove()
EDAC DEBUG: edac_device_del_device()
EDAC DEBUG: find_edac_device_by_dev()
(hang for a moment)
INFO: task edac-poller:2030 blocked for more than 120 seconds.
"echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
edac-poller D 00000000 0 2030 2
Call Trace:
[df159dc0] [c0071e3c] free_hot_cold_page+0x17c/0x304 (unreliable)
[df159e80] [c000a024] __switch_to+0x6c/0xa0
[df159ea0] [c03587d8] schedule+0x2f4/0x4d8
[df159f00] [c03598a8] __mutex_lock_slowpath+0xa0/0x174
[df159f40] [e1030434] edac_device_workq_function+0x28/0xd8 [edac_core]
[df159f60] [c003beb4] run_workqueue+0x114/0x218
[df159f90] [c003c674] worker_thread+0x5c/0xc8
[df159fd0] [c004106c] kthread+0x5c/0xa0
[df159ff0] [c0013538] original_kernel_thread+0x44/0x60
INFO: task rmmod:2062 blocked for more than 120 seconds.
"echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
rmmod D 0ff2c9fc 0 2062 1839
Call Trace:
[df119c00] [c0437a74] 0xc0437a74 (unreliable)
[df119cc0] [c000a024] __switch_to+0x6c/0xa0
[df119ce0] [c03587d8] schedule+0x2f4/0x4d8
[df119d40] [c03591dc] schedule_timeout+0xb0/0xf4
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-12-23 21:57:16 +00:00
|
|
|
/* If we are being removed, bail out immediately */
|
|
|
|
if (edac_dev->op_state == OP_OFFLINE) {
|
|
|
|
mutex_unlock(&device_ctls_mutex);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
2007-07-19 08:49:36 +00:00
|
|
|
/* Only poll controllers that are running polled and have a check */
|
|
|
|
if ((edac_dev->op_state == OP_RUNNING_POLL) &&
|
2007-07-19 08:50:13 +00:00
|
|
|
(edac_dev->edac_check != NULL)) {
|
|
|
|
edac_dev->edac_check(edac_dev);
|
2007-07-19 08:49:36 +00:00
|
|
|
}
|
|
|
|
|
2007-07-19 08:50:22 +00:00
|
|
|
mutex_unlock(&device_ctls_mutex);
|
2007-07-19 08:49:36 +00:00
|
|
|
|
2007-07-19 08:50:30 +00:00
|
|
|
/* Reschedule the workq for the next time period to start again
|
|
|
|
* if the number of msec is for 1 sec, then adjust to the next
|
2012-04-17 18:30:52 +00:00
|
|
|
* whole one second to save timers firing all over the period
|
2007-07-19 08:50:30 +00:00
|
|
|
* between integral seconds
|
|
|
|
*/
|
|
|
|
if (edac_dev->poll_msec == 1000)
|
|
|
|
queue_delayed_work(edac_workqueue, &edac_dev->work,
|
2008-02-07 08:14:51 +00:00
|
|
|
round_jiffies_relative(edac_dev->delay));
|
2007-07-19 08:50:30 +00:00
|
|
|
else
|
|
|
|
queue_delayed_work(edac_workqueue, &edac_dev->work,
|
|
|
|
edac_dev->delay);
|
2007-07-19 08:49:36 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
2007-07-19 08:49:52 +00:00
|
|
|
* edac_device_workq_setup
|
2007-07-19 08:49:36 +00:00
|
|
|
* initialize a workq item for this edac_device instance
|
|
|
|
* passing in the new delay period in msec
|
|
|
|
*/
|
2007-07-19 08:49:52 +00:00
|
|
|
void edac_device_workq_setup(struct edac_device_ctl_info *edac_dev,
|
2007-07-19 08:50:13 +00:00
|
|
|
unsigned msec)
|
2007-07-19 08:49:36 +00:00
|
|
|
{
|
2012-04-29 20:08:39 +00:00
|
|
|
edac_dbg(0, "\n");
|
2007-07-19 08:49:36 +00:00
|
|
|
|
2007-07-19 08:50:30 +00:00
|
|
|
/* take the arg 'msec' and set it into the control structure
|
|
|
|
* to used in the time period calculation
|
|
|
|
* then calc the number of jiffies that represents
|
|
|
|
*/
|
2007-07-19 08:49:36 +00:00
|
|
|
edac_dev->poll_msec = msec;
|
2007-07-19 08:50:30 +00:00
|
|
|
edac_dev->delay = msecs_to_jiffies(msec);
|
2007-07-19 08:49:36 +00:00
|
|
|
|
2007-07-19 08:49:52 +00:00
|
|
|
INIT_DELAYED_WORK(&edac_dev->work, edac_device_workq_function);
|
2007-07-19 08:50:30 +00:00
|
|
|
|
|
|
|
/* optimize here for the 1 second case, which will be normal value, to
|
|
|
|
* fire ON the 1 second time event. This helps reduce all sorts of
|
|
|
|
* timers firing on sub-second basis, while they are happy
|
|
|
|
* to fire together on the 1 second exactly
|
|
|
|
*/
|
|
|
|
if (edac_dev->poll_msec == 1000)
|
|
|
|
queue_delayed_work(edac_workqueue, &edac_dev->work,
|
2008-02-07 08:14:51 +00:00
|
|
|
round_jiffies_relative(edac_dev->delay));
|
2007-07-19 08:50:30 +00:00
|
|
|
else
|
|
|
|
queue_delayed_work(edac_workqueue, &edac_dev->work,
|
|
|
|
edac_dev->delay);
|
2007-07-19 08:49:36 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
2007-07-19 08:49:52 +00:00
|
|
|
* edac_device_workq_teardown
|
2007-07-19 08:49:36 +00:00
|
|
|
* stop the workq processing on this edac_dev
|
|
|
|
*/
|
2007-07-19 08:49:52 +00:00
|
|
|
void edac_device_workq_teardown(struct edac_device_ctl_info *edac_dev)
|
2007-07-19 08:49:36 +00:00
|
|
|
{
|
|
|
|
int status;
|
|
|
|
|
2013-12-30 20:14:12 +00:00
|
|
|
if (!edac_dev->edac_check)
|
|
|
|
return;
|
|
|
|
|
2007-07-19 08:49:36 +00:00
|
|
|
status = cancel_delayed_work(&edac_dev->work);
|
|
|
|
if (status == 0) {
|
|
|
|
/* workq instance might be running, wait for it */
|
|
|
|
flush_workqueue(edac_workqueue);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* edac_device_reset_delay_period
|
2007-07-19 08:50:30 +00:00
|
|
|
*
|
|
|
|
* need to stop any outstanding workq queued up at this time
|
|
|
|
* because we will be resetting the sleep time.
|
|
|
|
* Then restart the workq on the new delay
|
2007-07-19 08:49:36 +00:00
|
|
|
*/
|
2007-07-19 08:49:58 +00:00
|
|
|
void edac_device_reset_delay_period(struct edac_device_ctl_info *edac_dev,
|
2007-07-19 08:50:13 +00:00
|
|
|
unsigned long value)
|
2007-07-19 08:49:36 +00:00
|
|
|
{
|
2007-07-19 08:50:30 +00:00
|
|
|
/* cancel the current workq request, without the mutex lock */
|
2007-07-19 08:49:52 +00:00
|
|
|
edac_device_workq_teardown(edac_dev);
|
2007-07-19 08:49:36 +00:00
|
|
|
|
2007-07-19 08:50:30 +00:00
|
|
|
/* acquire the mutex before doing the workq setup */
|
|
|
|
mutex_lock(&device_ctls_mutex);
|
|
|
|
|
2007-07-19 08:49:36 +00:00
|
|
|
/* restart the workq request, with new delay value */
|
2007-07-19 08:49:52 +00:00
|
|
|
edac_device_workq_setup(edac_dev, value);
|
2007-07-19 08:49:36 +00:00
|
|
|
|
2007-07-19 08:50:22 +00:00
|
|
|
mutex_unlock(&device_ctls_mutex);
|
2007-07-19 08:49:36 +00:00
|
|
|
}
|
|
|
|
|
2009-06-17 23:27:59 +00:00
|
|
|
/*
|
|
|
|
* edac_device_alloc_index: Allocate a unique device index number
|
|
|
|
*
|
|
|
|
* Return:
|
|
|
|
* allocated index number
|
|
|
|
*/
|
|
|
|
int edac_device_alloc_index(void)
|
|
|
|
{
|
|
|
|
static atomic_t device_indexes = ATOMIC_INIT(0);
|
|
|
|
|
|
|
|
return atomic_inc_return(&device_indexes) - 1;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(edac_device_alloc_index);
|
|
|
|
|
2007-07-19 08:49:36 +00:00
|
|
|
/**
|
|
|
|
* edac_device_add_device: Insert the 'edac_dev' structure into the
|
|
|
|
* edac_device global list and create sysfs entries associated with
|
|
|
|
* edac_device structure.
|
|
|
|
* @edac_device: pointer to the edac_device structure to be added to the list
|
|
|
|
* 'edac_device' structure.
|
|
|
|
*
|
|
|
|
* Return:
|
|
|
|
* 0 Success
|
|
|
|
* !0 Failure
|
|
|
|
*/
|
2007-07-19 08:50:27 +00:00
|
|
|
int edac_device_add_device(struct edac_device_ctl_info *edac_dev)
|
2007-07-19 08:49:36 +00:00
|
|
|
{
|
2012-04-29 20:08:39 +00:00
|
|
|
edac_dbg(0, "\n");
|
2007-07-19 08:49:36 +00:00
|
|
|
|
|
|
|
#ifdef CONFIG_EDAC_DEBUG
|
|
|
|
if (edac_debug_level >= 3)
|
|
|
|
edac_device_dump_device(edac_dev);
|
|
|
|
#endif
|
2007-07-19 08:50:22 +00:00
|
|
|
mutex_lock(&device_ctls_mutex);
|
2007-07-19 08:49:36 +00:00
|
|
|
|
|
|
|
if (add_edac_dev_to_global_list(edac_dev))
|
|
|
|
goto fail0;
|
|
|
|
|
|
|
|
/* set load time so that error rate can be tracked */
|
|
|
|
edac_dev->start_time = jiffies;
|
|
|
|
|
|
|
|
/* create this instance's sysfs entries */
|
|
|
|
if (edac_device_create_sysfs(edac_dev)) {
|
|
|
|
edac_device_printk(edac_dev, KERN_WARNING,
|
2007-07-19 08:50:13 +00:00
|
|
|
"failed to create sysfs device\n");
|
2007-07-19 08:49:36 +00:00
|
|
|
goto fail1;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* If there IS a check routine, then we are running POLLED */
|
|
|
|
if (edac_dev->edac_check != NULL) {
|
|
|
|
/* This instance is NOW RUNNING */
|
|
|
|
edac_dev->op_state = OP_RUNNING_POLL;
|
|
|
|
|
2007-07-19 08:49:52 +00:00
|
|
|
/*
|
|
|
|
* enable workq processing on this instance,
|
|
|
|
* default = 1000 msec
|
|
|
|
*/
|
|
|
|
edac_device_workq_setup(edac_dev, 1000);
|
2007-07-19 08:49:36 +00:00
|
|
|
} else {
|
|
|
|
edac_dev->op_state = OP_RUNNING_INTERRUPT;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Report action taken */
|
|
|
|
edac_device_printk(edac_dev, KERN_INFO,
|
2013-10-10 16:22:36 +00:00
|
|
|
"Giving out device to module %s controller %s: DEV %s (%s)\n",
|
|
|
|
edac_dev->mod_name, edac_dev->ctl_name, edac_dev->dev_name,
|
|
|
|
edac_op_state_to_string(edac_dev->op_state));
|
2007-07-19 08:49:36 +00:00
|
|
|
|
2007-07-19 08:50:22 +00:00
|
|
|
mutex_unlock(&device_ctls_mutex);
|
2007-07-19 08:49:36 +00:00
|
|
|
return 0;
|
|
|
|
|
2007-07-19 08:50:13 +00:00
|
|
|
fail1:
|
2007-07-19 08:49:36 +00:00
|
|
|
/* Some error, so remove the entry from the lsit */
|
|
|
|
del_edac_device_from_global_list(edac_dev);
|
|
|
|
|
2007-07-19 08:50:13 +00:00
|
|
|
fail0:
|
2007-07-19 08:50:22 +00:00
|
|
|
mutex_unlock(&device_ctls_mutex);
|
2007-07-19 08:49:36 +00:00
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(edac_device_add_device);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* edac_device_del_device:
|
|
|
|
* Remove sysfs entries for specified edac_device structure and
|
|
|
|
* then remove edac_device structure from global list
|
|
|
|
*
|
2012-04-17 18:30:52 +00:00
|
|
|
* @dev:
|
2007-07-19 08:49:36 +00:00
|
|
|
* Pointer to 'struct device' representing edac_device
|
|
|
|
* structure to remove.
|
|
|
|
*
|
|
|
|
* Return:
|
|
|
|
* Pointer to removed edac_device structure,
|
|
|
|
* OR NULL if device not found.
|
|
|
|
*/
|
2007-07-19 08:49:58 +00:00
|
|
|
struct edac_device_ctl_info *edac_device_del_device(struct device *dev)
|
2007-07-19 08:49:36 +00:00
|
|
|
{
|
|
|
|
struct edac_device_ctl_info *edac_dev;
|
|
|
|
|
2012-04-29 20:08:39 +00:00
|
|
|
edac_dbg(0, "\n");
|
2007-07-19 08:49:36 +00:00
|
|
|
|
2007-07-19 08:50:22 +00:00
|
|
|
mutex_lock(&device_ctls_mutex);
|
2007-07-19 08:49:36 +00:00
|
|
|
|
2007-07-19 08:50:20 +00:00
|
|
|
/* Find the structure on the list, if not there, then leave */
|
|
|
|
edac_dev = find_edac_device_by_dev(dev);
|
|
|
|
if (edac_dev == NULL) {
|
2007-07-19 08:50:22 +00:00
|
|
|
mutex_unlock(&device_ctls_mutex);
|
2007-07-19 08:49:36 +00:00
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* mark this instance as OFFLINE */
|
|
|
|
edac_dev->op_state = OP_OFFLINE;
|
|
|
|
|
|
|
|
/* deregister from global list */
|
|
|
|
del_edac_device_from_global_list(edac_dev);
|
|
|
|
|
2007-07-19 08:50:22 +00:00
|
|
|
mutex_unlock(&device_ctls_mutex);
|
2007-07-19 08:49:36 +00:00
|
|
|
|
edac: fix edac core deadlock when removing a device
When deleting an edac device, we have to wait for its edac_dev.work to be
completed before deleting the whole edac_dev structure. Since we have no
idea which work in current edac_poller's workqueue is the work we are
conerned about, we wait for all work in the edac_poller's workqueue to be
proceseed. This is done via flush_cpu_workqueue() which inserts a
wq_barrier into the tail of the workqueue and then sleeping on the
completion of this wq_barrier. The edac_poller will wake up sleepers when
it is found.
EDAC core creates only one kernel worker thread, edac_poller, to run the
works of all current edac devices. They share the same callback function
of edac_device_workq_function(), which would grab the mutex of
device_ctls_mutex first before it checks the device. This is exactly
where edac_poller and rmmod would have a great chance to deadlock.
In below call trace of rmmod > ... >
edac_device_del_device >
edac_device_workq_teardown > flush_workqueue > flush_cpu_workqueue,
device_ctls_mutex would have already been grabbed by
edac_device_del_device(). So, on one hand rmmod would sleep on the
completion of a wq_barrier, holding device_ctls_mutex; on the other hand
edac_poller would be blocked on the same mutex when it's running any one
of works of existing edac evices(Note, this edac_dev.work is likely to be
totally irrelevant to the one that is being removed right now)and never
would have a chance to run the work of above wq_barrier to wake rmmod up.
edac_device_workq_teardown() should not be called within the critical
region of device_ctls_mutex. Just like is done in edac_pci_del_device()
and edac_mc_del_mc(), where edac_pci_workq_teardown() and
edac_mc_workq_teardown() are called after related mutex are released.
Moreover, an edac_dev.work should check first if it is being removed. If
this is the case, then it should bail out immediately. Since not all of
existing edac devices are to be removed, this "shutting flag" should be
contained to edac device being removed. The current edac_dev.op_state can
be used to serve this purpose.
The original deadlock problem and the solution have been witnessed and
tested on actual hardware. Without the solution, rmmod an edac driver
would result in below deadlock:
root@localhost:/root> rmmod mv64x60_edac
EDAC DEBUG: mv64x60_dma_err_remove()
EDAC DEBUG: edac_device_del_device()
EDAC DEBUG: find_edac_device_by_dev()
(hang for a moment)
INFO: task edac-poller:2030 blocked for more than 120 seconds.
"echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
edac-poller D 00000000 0 2030 2
Call Trace:
[df159dc0] [c0071e3c] free_hot_cold_page+0x17c/0x304 (unreliable)
[df159e80] [c000a024] __switch_to+0x6c/0xa0
[df159ea0] [c03587d8] schedule+0x2f4/0x4d8
[df159f00] [c03598a8] __mutex_lock_slowpath+0xa0/0x174
[df159f40] [e1030434] edac_device_workq_function+0x28/0xd8 [edac_core]
[df159f60] [c003beb4] run_workqueue+0x114/0x218
[df159f90] [c003c674] worker_thread+0x5c/0xc8
[df159fd0] [c004106c] kthread+0x5c/0xa0
[df159ff0] [c0013538] original_kernel_thread+0x44/0x60
INFO: task rmmod:2062 blocked for more than 120 seconds.
"echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
rmmod D 0ff2c9fc 0 2062 1839
Call Trace:
[df119c00] [c0437a74] 0xc0437a74 (unreliable)
[df119cc0] [c000a024] __switch_to+0x6c/0xa0
[df119ce0] [c03587d8] schedule+0x2f4/0x4d8
[df119d40] [c03591dc] schedule_timeout+0xb0/0xf4
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-12-23 21:57:16 +00:00
|
|
|
/* clear workq processing on this instance */
|
|
|
|
edac_device_workq_teardown(edac_dev);
|
|
|
|
|
2007-07-19 08:50:29 +00:00
|
|
|
/* Tear down the sysfs entries for this instance */
|
|
|
|
edac_device_remove_sysfs(edac_dev);
|
|
|
|
|
2007-07-19 08:49:36 +00:00
|
|
|
edac_printk(KERN_INFO, EDAC_MC,
|
2007-07-19 08:50:13 +00:00
|
|
|
"Removed device %d for %s %s: DEV %s\n",
|
|
|
|
edac_dev->dev_idx,
|
2008-05-05 03:54:19 +00:00
|
|
|
edac_dev->mod_name, edac_dev->ctl_name, edac_dev_name(edac_dev));
|
2007-07-19 08:49:36 +00:00
|
|
|
|
|
|
|
return edac_dev;
|
|
|
|
}
|
2007-07-19 08:49:58 +00:00
|
|
|
EXPORT_SYMBOL_GPL(edac_device_del_device);
|
2007-07-19 08:49:36 +00:00
|
|
|
|
|
|
|
static inline int edac_device_get_log_ce(struct edac_device_ctl_info *edac_dev)
|
|
|
|
{
|
|
|
|
return edac_dev->log_ce;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline int edac_device_get_log_ue(struct edac_device_ctl_info *edac_dev)
|
|
|
|
{
|
|
|
|
return edac_dev->log_ue;
|
|
|
|
}
|
|
|
|
|
2007-07-19 08:49:58 +00:00
|
|
|
static inline int edac_device_get_panic_on_ue(struct edac_device_ctl_info
|
2007-07-19 08:50:13 +00:00
|
|
|
*edac_dev)
|
2007-07-19 08:49:36 +00:00
|
|
|
{
|
|
|
|
return edac_dev->panic_on_ue;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* edac_device_handle_ce
|
|
|
|
* perform a common output and handling of an 'edac_dev' CE event
|
|
|
|
*/
|
|
|
|
void edac_device_handle_ce(struct edac_device_ctl_info *edac_dev,
|
2007-07-19 08:50:13 +00:00
|
|
|
int inst_nr, int block_nr, const char *msg)
|
2007-07-19 08:49:36 +00:00
|
|
|
{
|
|
|
|
struct edac_device_instance *instance;
|
|
|
|
struct edac_device_block *block = NULL;
|
|
|
|
|
|
|
|
if ((inst_nr >= edac_dev->nr_instances) || (inst_nr < 0)) {
|
|
|
|
edac_device_printk(edac_dev, KERN_ERR,
|
2007-07-19 08:50:13 +00:00
|
|
|
"INTERNAL ERROR: 'instance' out of range "
|
|
|
|
"(%d >= %d)\n", inst_nr,
|
|
|
|
edac_dev->nr_instances);
|
2007-07-19 08:49:36 +00:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
instance = edac_dev->instances + inst_nr;
|
|
|
|
|
|
|
|
if ((block_nr >= instance->nr_blocks) || (block_nr < 0)) {
|
|
|
|
edac_device_printk(edac_dev, KERN_ERR,
|
2007-07-19 08:50:13 +00:00
|
|
|
"INTERNAL ERROR: instance %d 'block' "
|
|
|
|
"out of range (%d >= %d)\n",
|
|
|
|
inst_nr, block_nr,
|
|
|
|
instance->nr_blocks);
|
2007-07-19 08:49:36 +00:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (instance->nr_blocks > 0) {
|
|
|
|
block = instance->blocks + block_nr;
|
|
|
|
block->counters.ce_count++;
|
|
|
|
}
|
|
|
|
|
2011-03-31 01:57:33 +00:00
|
|
|
/* Propagate the count up the 'totals' tree */
|
2007-07-19 08:49:36 +00:00
|
|
|
instance->counters.ce_count++;
|
|
|
|
edac_dev->counters.ce_count++;
|
|
|
|
|
|
|
|
if (edac_device_get_log_ce(edac_dev))
|
|
|
|
edac_device_printk(edac_dev, KERN_WARNING,
|
2007-07-19 08:50:13 +00:00
|
|
|
"CE: %s instance: %s block: %s '%s'\n",
|
|
|
|
edac_dev->ctl_name, instance->name,
|
|
|
|
block ? block->name : "N/A", msg);
|
2007-07-19 08:49:36 +00:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(edac_device_handle_ce);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* edac_device_handle_ue
|
|
|
|
* perform a common output and handling of an 'edac_dev' UE event
|
|
|
|
*/
|
|
|
|
void edac_device_handle_ue(struct edac_device_ctl_info *edac_dev,
|
2007-07-19 08:50:13 +00:00
|
|
|
int inst_nr, int block_nr, const char *msg)
|
2007-07-19 08:49:36 +00:00
|
|
|
{
|
|
|
|
struct edac_device_instance *instance;
|
|
|
|
struct edac_device_block *block = NULL;
|
|
|
|
|
|
|
|
if ((inst_nr >= edac_dev->nr_instances) || (inst_nr < 0)) {
|
|
|
|
edac_device_printk(edac_dev, KERN_ERR,
|
2007-07-19 08:50:13 +00:00
|
|
|
"INTERNAL ERROR: 'instance' out of range "
|
|
|
|
"(%d >= %d)\n", inst_nr,
|
|
|
|
edac_dev->nr_instances);
|
2007-07-19 08:49:36 +00:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
instance = edac_dev->instances + inst_nr;
|
|
|
|
|
|
|
|
if ((block_nr >= instance->nr_blocks) || (block_nr < 0)) {
|
|
|
|
edac_device_printk(edac_dev, KERN_ERR,
|
2007-07-19 08:50:13 +00:00
|
|
|
"INTERNAL ERROR: instance %d 'block' "
|
|
|
|
"out of range (%d >= %d)\n",
|
|
|
|
inst_nr, block_nr,
|
|
|
|
instance->nr_blocks);
|
2007-07-19 08:49:36 +00:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (instance->nr_blocks > 0) {
|
|
|
|
block = instance->blocks + block_nr;
|
|
|
|
block->counters.ue_count++;
|
|
|
|
}
|
|
|
|
|
2011-03-31 01:57:33 +00:00
|
|
|
/* Propagate the count up the 'totals' tree */
|
2007-07-19 08:49:36 +00:00
|
|
|
instance->counters.ue_count++;
|
|
|
|
edac_dev->counters.ue_count++;
|
|
|
|
|
|
|
|
if (edac_device_get_log_ue(edac_dev))
|
|
|
|
edac_device_printk(edac_dev, KERN_EMERG,
|
2007-07-19 08:50:13 +00:00
|
|
|
"UE: %s instance: %s block: %s '%s'\n",
|
|
|
|
edac_dev->ctl_name, instance->name,
|
|
|
|
block ? block->name : "N/A", msg);
|
2007-07-19 08:49:36 +00:00
|
|
|
|
|
|
|
if (edac_device_get_panic_on_ue(edac_dev))
|
2007-07-19 08:50:11 +00:00
|
|
|
panic("EDAC %s: UE instance: %s block %s '%s'\n",
|
2007-07-19 08:50:13 +00:00
|
|
|
edac_dev->ctl_name, instance->name,
|
|
|
|
block ? block->name : "N/A", msg);
|
2007-07-19 08:49:36 +00:00
|
|
|
}
|
2007-07-19 08:49:58 +00:00
|
|
|
EXPORT_SYMBOL_GPL(edac_device_handle_ue);
|