android_kernel_motorola_sm6225/drivers/pci/hotplug/fakephp.c

377 lines
8.8 KiB
C
Raw Normal View History

/*
* Fake PCI Hot Plug Controller Driver
*
* Copyright (C) 2003 Greg Kroah-Hartman <greg@kroah.com>
* Copyright (C) 2003 IBM Corp.
* Copyright (C) 2003 Rolf Eike Beer <eike-kernel@sf-tec.de>
*
* Based on ideas and code from:
* Vladimir Kondratiev <vladimir.kondratiev@intel.com>
* Rolf Eike Beer <eike-kernel@sf-tec.de>
*
* All rights reserved.
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation, version 2 of the License.
*
* Send feedback to <greg@kroah.com>
*/
/*
*
* This driver will "emulate" removing PCI devices from the system. If
* the "power" file is written to with "0" then the specified PCI device
* will be completely removed from the kernel.
*
* WARNING, this does NOT turn off the power to the PCI device. This is
* a "logical" removal, not a physical or electrical removal.
*
* Use this module at your own risk, you have been warned!
*
* Enabling PCI devices is left as an exercise for the reader...
*
*/
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/pci.h>
#include <linux/pci_hotplug.h>
#include <linux/init.h>
#include <linux/string.h>
#include <linux/slab.h>
PCI: Fix fakephp deadlock If the fakephp driver is used to emulate removal of a PCI device by writing text string "0" to the "power" sysfs attribute file, this causes its parent directory and its contents (including the "power" file) to be deleted before the write operation returns. Unfortunately, it ends up in a deadlock waiting for itself to complete. The deadlock is as follows: sysfs_write_file calls flush_write_buffer which calls sysfs_get_active_two before calling power_write_file in pci_hotplug_core.c via the sysfs store operation. The power_write_file function calls disable_slot in fakephp.c via the slot operation. The disable_slot function calls remove_slot which calls pci_hp_deregister (back in pci_hotplug_core.c) which calls fs_remove_slot which calls sysfs_remove_file to remove the "power" file. The sysfs_remove_file function calls sysfs_hash_and_remove which calls sysfs_addrm_finish which calls sysfs_deactivate. The sysfs_deactivate function sees that something has an active reference on the sysfs_dirent (from the previous call to sysfs_get_active_two back up the call stack somewhere) so waits for the active reference to go away, which is of course impossible. The problem has been present since 2.6.21. This patch breaks the deadlock by queuing work queue items on a single- threaded work queue to remove a slot from sysfs, and to rescan the PCI buses. There is also some protection against disabling a slot that is already being removed. Signed-off-by: Ian Abbott <abbotti@mev.co.uk> Cc: Kristen Accardi <kristen.c.accardi@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-01-25 17:23:56 +01:00
#include <linux/workqueue.h>
#include "../pci.h"
#if !defined(MODULE)
#define MY_NAME "fakephp"
#else
#define MY_NAME THIS_MODULE->name
#endif
#define dbg(format, arg...) \
do { \
if (debug) \
printk(KERN_DEBUG "%s: " format, \
MY_NAME , ## arg); \
} while (0)
#define err(format, arg...) printk(KERN_ERR "%s: " format, MY_NAME , ## arg)
#define info(format, arg...) printk(KERN_INFO "%s: " format, MY_NAME , ## arg)
#define DRIVER_AUTHOR "Greg Kroah-Hartman <greg@kroah.com>"
#define DRIVER_DESC "Fake PCI Hot Plug Controller Driver"
struct dummy_slot {
struct list_head node;
struct hotplug_slot *slot;
struct pci_dev *dev;
PCI: Fix fakephp deadlock If the fakephp driver is used to emulate removal of a PCI device by writing text string "0" to the "power" sysfs attribute file, this causes its parent directory and its contents (including the "power" file) to be deleted before the write operation returns. Unfortunately, it ends up in a deadlock waiting for itself to complete. The deadlock is as follows: sysfs_write_file calls flush_write_buffer which calls sysfs_get_active_two before calling power_write_file in pci_hotplug_core.c via the sysfs store operation. The power_write_file function calls disable_slot in fakephp.c via the slot operation. The disable_slot function calls remove_slot which calls pci_hp_deregister (back in pci_hotplug_core.c) which calls fs_remove_slot which calls sysfs_remove_file to remove the "power" file. The sysfs_remove_file function calls sysfs_hash_and_remove which calls sysfs_addrm_finish which calls sysfs_deactivate. The sysfs_deactivate function sees that something has an active reference on the sysfs_dirent (from the previous call to sysfs_get_active_two back up the call stack somewhere) so waits for the active reference to go away, which is of course impossible. The problem has been present since 2.6.21. This patch breaks the deadlock by queuing work queue items on a single- threaded work queue to remove a slot from sysfs, and to rescan the PCI buses. There is also some protection against disabling a slot that is already being removed. Signed-off-by: Ian Abbott <abbotti@mev.co.uk> Cc: Kristen Accardi <kristen.c.accardi@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-01-25 17:23:56 +01:00
struct work_struct remove_work;
unsigned long removed;
char name[8];
};
static int debug;
static LIST_HEAD(slot_list);
PCI: Fix fakephp deadlock If the fakephp driver is used to emulate removal of a PCI device by writing text string "0" to the "power" sysfs attribute file, this causes its parent directory and its contents (including the "power" file) to be deleted before the write operation returns. Unfortunately, it ends up in a deadlock waiting for itself to complete. The deadlock is as follows: sysfs_write_file calls flush_write_buffer which calls sysfs_get_active_two before calling power_write_file in pci_hotplug_core.c via the sysfs store operation. The power_write_file function calls disable_slot in fakephp.c via the slot operation. The disable_slot function calls remove_slot which calls pci_hp_deregister (back in pci_hotplug_core.c) which calls fs_remove_slot which calls sysfs_remove_file to remove the "power" file. The sysfs_remove_file function calls sysfs_hash_and_remove which calls sysfs_addrm_finish which calls sysfs_deactivate. The sysfs_deactivate function sees that something has an active reference on the sysfs_dirent (from the previous call to sysfs_get_active_two back up the call stack somewhere) so waits for the active reference to go away, which is of course impossible. The problem has been present since 2.6.21. This patch breaks the deadlock by queuing work queue items on a single- threaded work queue to remove a slot from sysfs, and to rescan the PCI buses. There is also some protection against disabling a slot that is already being removed. Signed-off-by: Ian Abbott <abbotti@mev.co.uk> Cc: Kristen Accardi <kristen.c.accardi@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-01-25 17:23:56 +01:00
static struct workqueue_struct *dummyphp_wq;
static void pci_rescan_worker(struct work_struct *work);
static DECLARE_WORK(pci_rescan_work, pci_rescan_worker);
static int enable_slot (struct hotplug_slot *slot);
static int disable_slot (struct hotplug_slot *slot);
static struct hotplug_slot_ops dummy_hotplug_slot_ops = {
.owner = THIS_MODULE,
.enable_slot = enable_slot,
.disable_slot = disable_slot,
};
static void dummy_release(struct hotplug_slot *slot)
{
struct dummy_slot *dslot = slot->private;
list_del(&dslot->node);
kfree(dslot->slot->info);
kfree(dslot->slot);
pci_dev_put(dslot->dev);
kfree(dslot);
}
static int add_slot(struct pci_dev *dev)
{
struct dummy_slot *dslot;
struct hotplug_slot *slot;
int retval = -ENOMEM;
static int count = 1;
slot = kzalloc(sizeof(struct hotplug_slot), GFP_KERNEL);
if (!slot)
goto error;
slot->info = kzalloc(sizeof(struct hotplug_slot_info), GFP_KERNEL);
if (!slot->info)
goto error_slot;
slot->info->power_status = 1;
slot->info->max_bus_speed = PCI_SPEED_UNKNOWN;
slot->info->cur_bus_speed = PCI_SPEED_UNKNOWN;
PCI: Fix fakephp deadlock If the fakephp driver is used to emulate removal of a PCI device by writing text string "0" to the "power" sysfs attribute file, this causes its parent directory and its contents (including the "power" file) to be deleted before the write operation returns. Unfortunately, it ends up in a deadlock waiting for itself to complete. The deadlock is as follows: sysfs_write_file calls flush_write_buffer which calls sysfs_get_active_two before calling power_write_file in pci_hotplug_core.c via the sysfs store operation. The power_write_file function calls disable_slot in fakephp.c via the slot operation. The disable_slot function calls remove_slot which calls pci_hp_deregister (back in pci_hotplug_core.c) which calls fs_remove_slot which calls sysfs_remove_file to remove the "power" file. The sysfs_remove_file function calls sysfs_hash_and_remove which calls sysfs_addrm_finish which calls sysfs_deactivate. The sysfs_deactivate function sees that something has an active reference on the sysfs_dirent (from the previous call to sysfs_get_active_two back up the call stack somewhere) so waits for the active reference to go away, which is of course impossible. The problem has been present since 2.6.21. This patch breaks the deadlock by queuing work queue items on a single- threaded work queue to remove a slot from sysfs, and to rescan the PCI buses. There is also some protection against disabling a slot that is already being removed. Signed-off-by: Ian Abbott <abbotti@mev.co.uk> Cc: Kristen Accardi <kristen.c.accardi@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-01-25 17:23:56 +01:00
dslot = kzalloc(sizeof(struct dummy_slot), GFP_KERNEL);
if (!dslot)
goto error_info;
slot->name = dslot->name;
snprintf(slot->name, sizeof(dslot->name), "fake%d", count++);
dbg("slot->name = %s\n", slot->name);
slot->ops = &dummy_hotplug_slot_ops;
slot->release = &dummy_release;
slot->private = dslot;
PCI: introduce pci_slot Currently, /sys/bus/pci/slots/ only exposes hotplug attributes when a hotplug driver is loaded, but PCI slots have attributes such as address, speed, width, etc. that are not related to hotplug at all. Introduce pci_slot as the primary data structure and kobject model. Hotplug attributes described in hotplug_slot become a secondary structure associated with the pci_slot. This patch only creates the infrastructure that allows the separation of PCI slot attributes and hotplug attributes. In this patch, the PCI hotplug core remains the only user of this infrastructure, and thus, /sys/bus/pci/slots/ will still only become populated when a hotplug driver is loaded. A later patch in this series will add a second user of this new infrastructure and demonstrate splitting the task of exposing pci_slot attributes from hotplug_slot attributes. - Make pci_slot the primary sysfs entity. hotplug_slot becomes a subsidiary structure. o pci_create_slot() creates and registers a slot with the PCI core o pci_slot_add_hotplug() gives it hotplug capability - Change the prototype of pci_hp_register() to take the bus and slot number (on parent bus) as parameters. - Remove all the ->get_address methods since this functionality is now handled by pci_slot directly. [achiang@hp.com: rpaphp-correctly-pci_hp_register-for-empty-pci-slots] Tested-by: Badari Pulavarty <pbadari@us.ibm.com> Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> [akpm@linux-foundation.org: build fix] [akpm@linux-foundation.org: make headers_check happy] [akpm@linux-foundation.org: nuther build fix] [akpm@linux-foundation.org: fix typo in #include] Signed-off-by: Alex Chiang <achiang@hp.com> Signed-off-by: Matthew Wilcox <matthew@wil.cx> Cc: Greg KH <greg@kroah.com> Cc: Kristen Carlson Accardi <kristen.c.accardi@intel.com> Cc: Len Brown <lenb@kernel.org> Acked-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
2008-06-10 23:28:50 +02:00
retval = pci_hp_register(slot, dev->bus, PCI_SLOT(dev->devfn));
if (retval) {
err("pci_hp_register failed with error %d\n", retval);
goto error_dslot;
}
dslot->slot = slot;
dslot->dev = pci_dev_get(dev);
list_add (&dslot->node, &slot_list);
return retval;
error_dslot:
kfree(dslot);
error_info:
kfree(slot->info);
error_slot:
kfree(slot);
error:
return retval;
}
static int __init pci_scan_buses(void)
{
struct pci_dev *dev = NULL;
int lastslot = 0;
while ((dev = pci_get_device(PCI_ANY_ID, PCI_ANY_ID, dev)) != NULL) {
if (PCI_FUNC(dev->devfn) > 0 &&
lastslot == PCI_SLOT(dev->devfn))
continue;
lastslot = PCI_SLOT(dev->devfn);
add_slot(dev);
}
return 0;
}
static void remove_slot(struct dummy_slot *dslot)
{
int retval;
dbg("removing slot %s\n", dslot->slot->name);
retval = pci_hp_deregister(dslot->slot);
if (retval)
err("Problem unregistering a slot %s\n", dslot->slot->name);
}
PCI: Fix fakephp deadlock If the fakephp driver is used to emulate removal of a PCI device by writing text string "0" to the "power" sysfs attribute file, this causes its parent directory and its contents (including the "power" file) to be deleted before the write operation returns. Unfortunately, it ends up in a deadlock waiting for itself to complete. The deadlock is as follows: sysfs_write_file calls flush_write_buffer which calls sysfs_get_active_two before calling power_write_file in pci_hotplug_core.c via the sysfs store operation. The power_write_file function calls disable_slot in fakephp.c via the slot operation. The disable_slot function calls remove_slot which calls pci_hp_deregister (back in pci_hotplug_core.c) which calls fs_remove_slot which calls sysfs_remove_file to remove the "power" file. The sysfs_remove_file function calls sysfs_hash_and_remove which calls sysfs_addrm_finish which calls sysfs_deactivate. The sysfs_deactivate function sees that something has an active reference on the sysfs_dirent (from the previous call to sysfs_get_active_two back up the call stack somewhere) so waits for the active reference to go away, which is of course impossible. The problem has been present since 2.6.21. This patch breaks the deadlock by queuing work queue items on a single- threaded work queue to remove a slot from sysfs, and to rescan the PCI buses. There is also some protection against disabling a slot that is already being removed. Signed-off-by: Ian Abbott <abbotti@mev.co.uk> Cc: Kristen Accardi <kristen.c.accardi@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-01-25 17:23:56 +01:00
/* called from the single-threaded workqueue handler to remove a slot */
static void remove_slot_worker(struct work_struct *work)
{
struct dummy_slot *dslot =
container_of(work, struct dummy_slot, remove_work);
remove_slot(dslot);
}
/**
* pci_rescan_slot - Rescan slot
* @temp: Device template. Should be set: bus and devfn.
*
* Tries hard not to re-enable already existing devices;
* also handles scanning of subfunctions.
*/
static void pci_rescan_slot(struct pci_dev *temp)
{
struct pci_bus *bus = temp->bus;
struct pci_dev *dev;
int func;
int retval;
u8 hdr_type;
if (!pci_read_config_byte(temp, PCI_HEADER_TYPE, &hdr_type)) {
temp->hdr_type = hdr_type & 0x7f;
if ((dev = pci_get_slot(bus, temp->devfn)) != NULL)
pci_dev_put(dev);
else {
dev = pci_scan_single_device(bus, temp->devfn);
if (dev) {
dbg("New device on %s function %x:%x\n",
bus->name, temp->devfn >> 3,
temp->devfn & 7);
retval = pci_bus_add_device(dev);
if (retval)
dev_err(&dev->dev, "error adding "
"device, continuing.\n");
else
add_slot(dev);
}
}
/* multifunction device? */
if (!(hdr_type & 0x80))
return;
/* continue scanning for other functions */
for (func = 1, temp->devfn++; func < 8; func++, temp->devfn++) {
if (pci_read_config_byte(temp, PCI_HEADER_TYPE, &hdr_type))
continue;
temp->hdr_type = hdr_type & 0x7f;
if ((dev = pci_get_slot(bus, temp->devfn)) != NULL)
pci_dev_put(dev);
else {
dev = pci_scan_single_device(bus, temp->devfn);
if (dev) {
dbg("New device on %s function %x:%x\n",
bus->name, temp->devfn >> 3,
temp->devfn & 7);
retval = pci_bus_add_device(dev);
if (retval)
dev_err(&dev->dev, "error adding "
"device, continuing.\n");
else
add_slot(dev);
}
}
}
}
}
/**
* pci_rescan_bus - Rescan PCI bus
* @bus: the PCI bus to rescan
*
* Call pci_rescan_slot for each possible function of the bus.
*/
static void pci_rescan_bus(const struct pci_bus *bus)
{
unsigned int devfn;
struct pci_dev *dev;
dev = alloc_pci_dev();
if (!dev)
return;
dev->bus = (struct pci_bus*)bus;
dev->sysdata = bus->sysdata;
for (devfn = 0; devfn < 0x100; devfn += 8) {
dev->devfn = devfn;
pci_rescan_slot(dev);
}
kfree(dev);
}
/* recursively scan all buses */
static void pci_rescan_buses(const struct list_head *list)
{
const struct list_head *l;
list_for_each(l,list) {
const struct pci_bus *b = pci_bus_b(l);
pci_rescan_bus(b);
pci_rescan_buses(&b->children);
}
}
/* initiate rescan of all pci buses */
static inline void pci_rescan(void) {
pci_rescan_buses(&pci_root_buses);
}
PCI: Fix fakephp deadlock If the fakephp driver is used to emulate removal of a PCI device by writing text string "0" to the "power" sysfs attribute file, this causes its parent directory and its contents (including the "power" file) to be deleted before the write operation returns. Unfortunately, it ends up in a deadlock waiting for itself to complete. The deadlock is as follows: sysfs_write_file calls flush_write_buffer which calls sysfs_get_active_two before calling power_write_file in pci_hotplug_core.c via the sysfs store operation. The power_write_file function calls disable_slot in fakephp.c via the slot operation. The disable_slot function calls remove_slot which calls pci_hp_deregister (back in pci_hotplug_core.c) which calls fs_remove_slot which calls sysfs_remove_file to remove the "power" file. The sysfs_remove_file function calls sysfs_hash_and_remove which calls sysfs_addrm_finish which calls sysfs_deactivate. The sysfs_deactivate function sees that something has an active reference on the sysfs_dirent (from the previous call to sysfs_get_active_two back up the call stack somewhere) so waits for the active reference to go away, which is of course impossible. The problem has been present since 2.6.21. This patch breaks the deadlock by queuing work queue items on a single- threaded work queue to remove a slot from sysfs, and to rescan the PCI buses. There is also some protection against disabling a slot that is already being removed. Signed-off-by: Ian Abbott <abbotti@mev.co.uk> Cc: Kristen Accardi <kristen.c.accardi@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-01-25 17:23:56 +01:00
/* called from the single-threaded workqueue handler to rescan all pci buses */
static void pci_rescan_worker(struct work_struct *work)
{
pci_rescan();
}
static int enable_slot(struct hotplug_slot *hotplug_slot)
{
/* mis-use enable_slot for rescanning of the pci bus */
PCI: Fix fakephp deadlock If the fakephp driver is used to emulate removal of a PCI device by writing text string "0" to the "power" sysfs attribute file, this causes its parent directory and its contents (including the "power" file) to be deleted before the write operation returns. Unfortunately, it ends up in a deadlock waiting for itself to complete. The deadlock is as follows: sysfs_write_file calls flush_write_buffer which calls sysfs_get_active_two before calling power_write_file in pci_hotplug_core.c via the sysfs store operation. The power_write_file function calls disable_slot in fakephp.c via the slot operation. The disable_slot function calls remove_slot which calls pci_hp_deregister (back in pci_hotplug_core.c) which calls fs_remove_slot which calls sysfs_remove_file to remove the "power" file. The sysfs_remove_file function calls sysfs_hash_and_remove which calls sysfs_addrm_finish which calls sysfs_deactivate. The sysfs_deactivate function sees that something has an active reference on the sysfs_dirent (from the previous call to sysfs_get_active_two back up the call stack somewhere) so waits for the active reference to go away, which is of course impossible. The problem has been present since 2.6.21. This patch breaks the deadlock by queuing work queue items on a single- threaded work queue to remove a slot from sysfs, and to rescan the PCI buses. There is also some protection against disabling a slot that is already being removed. Signed-off-by: Ian Abbott <abbotti@mev.co.uk> Cc: Kristen Accardi <kristen.c.accardi@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-01-25 17:23:56 +01:00
cancel_work_sync(&pci_rescan_work);
queue_work(dummyphp_wq, &pci_rescan_work);
PCI: Hotplug: fakephp: Return success, not ENODEV, when bus rescan is triggered The 'power' attribute of the fakephp driver originally only let one turn a slot off. If one tried to turn a slot on (echo 1 > .../power), it would return ENODEV, as fakephp did not support this function. An old (pre-git) patch changed this: 2004/11/11 16:33:31-08:00 jdittmer [PATCH] fakephp: add pci bus rescan ability http://article.gmane.org/gmane.linux.kernel/251183 Now writing "1" to the power attribute has the effect of triggering a bus rescan, but it still returns ENODEV, probably an oversight in the above patch. Using the BusyBox echo will not produce an error message, but will trigger *two* bus rescans (and return an exit code of 1): ~ # strace echo -n 1 > /sys/bus/pci/slots/0000:00:00.0/power ... write(1, "1", 1) = -1 ENODEV (No such device) write(1, "1", 1) = -1 ENODEV (No such device) exit(1) = ? Using cp gives a write error, even though the write did happen and a rescan was triggered: ~ # echo -n 1 > tmp ; cp tmp /sys/bus/pci/slots/0000:00:00.0/power cp: Write Error: No such device It seems much better to return success instead of failure. The actual status of the bus rescan is hard to return. It happens asynchronously in a work thread, so the sysfs store functions returns before any status is ready (the whole point of the work queue). And even if it didn't do this, the rescan doesn't have any clear status to return. Signed-off-by: Trent Piepho <tpiepho@freescale.com> CC: Jan Dittmer <jdittmer@ppp0.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-04-08 01:04:16 +02:00
return 0;
}
static int disable_slot(struct hotplug_slot *slot)
{
struct dummy_slot *dslot;
struct pci_dev *dev;
int func;
if (!slot)
return -ENODEV;
dslot = slot->private;
dbg("%s - physical_slot = %s\n", __func__, slot->name);
for (func = 7; func >= 0; func--) {
dev = pci_get_slot(dslot->dev->bus, dslot->dev->devfn + func);
if (!dev)
continue;
if (test_and_set_bit(0, &dslot->removed)) {
dbg("Slot already scheduled for removal\n");
return -ENODEV;
}
/* queue work item to blow away this sysfs entry and other
* parts.
*/
INIT_WORK(&dslot->remove_work, remove_slot_worker);
queue_work(dummyphp_wq, &dslot->remove_work);
/* blow away this sysfs entry and other parts. */
remove_slot(dslot);
pci_dev_put(dev);
}
return 0;
}
static void cleanup_slots (void)
{
struct list_head *tmp;
struct list_head *next;
struct dummy_slot *dslot;
PCI: Fix fakephp deadlock If the fakephp driver is used to emulate removal of a PCI device by writing text string "0" to the "power" sysfs attribute file, this causes its parent directory and its contents (including the "power" file) to be deleted before the write operation returns. Unfortunately, it ends up in a deadlock waiting for itself to complete. The deadlock is as follows: sysfs_write_file calls flush_write_buffer which calls sysfs_get_active_two before calling power_write_file in pci_hotplug_core.c via the sysfs store operation. The power_write_file function calls disable_slot in fakephp.c via the slot operation. The disable_slot function calls remove_slot which calls pci_hp_deregister (back in pci_hotplug_core.c) which calls fs_remove_slot which calls sysfs_remove_file to remove the "power" file. The sysfs_remove_file function calls sysfs_hash_and_remove which calls sysfs_addrm_finish which calls sysfs_deactivate. The sysfs_deactivate function sees that something has an active reference on the sysfs_dirent (from the previous call to sysfs_get_active_two back up the call stack somewhere) so waits for the active reference to go away, which is of course impossible. The problem has been present since 2.6.21. This patch breaks the deadlock by queuing work queue items on a single- threaded work queue to remove a slot from sysfs, and to rescan the PCI buses. There is also some protection against disabling a slot that is already being removed. Signed-off-by: Ian Abbott <abbotti@mev.co.uk> Cc: Kristen Accardi <kristen.c.accardi@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-01-25 17:23:56 +01:00
destroy_workqueue(dummyphp_wq);
list_for_each_safe (tmp, next, &slot_list) {
dslot = list_entry (tmp, struct dummy_slot, node);
remove_slot(dslot);
}
}
static int __init dummyphp_init(void)
{
info(DRIVER_DESC "\n");
PCI: Fix fakephp deadlock If the fakephp driver is used to emulate removal of a PCI device by writing text string "0" to the "power" sysfs attribute file, this causes its parent directory and its contents (including the "power" file) to be deleted before the write operation returns. Unfortunately, it ends up in a deadlock waiting for itself to complete. The deadlock is as follows: sysfs_write_file calls flush_write_buffer which calls sysfs_get_active_two before calling power_write_file in pci_hotplug_core.c via the sysfs store operation. The power_write_file function calls disable_slot in fakephp.c via the slot operation. The disable_slot function calls remove_slot which calls pci_hp_deregister (back in pci_hotplug_core.c) which calls fs_remove_slot which calls sysfs_remove_file to remove the "power" file. The sysfs_remove_file function calls sysfs_hash_and_remove which calls sysfs_addrm_finish which calls sysfs_deactivate. The sysfs_deactivate function sees that something has an active reference on the sysfs_dirent (from the previous call to sysfs_get_active_two back up the call stack somewhere) so waits for the active reference to go away, which is of course impossible. The problem has been present since 2.6.21. This patch breaks the deadlock by queuing work queue items on a single- threaded work queue to remove a slot from sysfs, and to rescan the PCI buses. There is also some protection against disabling a slot that is already being removed. Signed-off-by: Ian Abbott <abbotti@mev.co.uk> Cc: Kristen Accardi <kristen.c.accardi@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2008-01-25 17:23:56 +01:00
dummyphp_wq = create_singlethread_workqueue(MY_NAME);
if (!dummyphp_wq)
return -ENOMEM;
return pci_scan_buses();
}
static void __exit dummyphp_exit(void)
{
cleanup_slots();
}
module_init(dummyphp_init);
module_exit(dummyphp_exit);
MODULE_AUTHOR(DRIVER_AUTHOR);
MODULE_DESCRIPTION(DRIVER_DESC);
MODULE_LICENSE("GPL");
module_param(debug, bool, S_IRUGO | S_IWUSR);
MODULE_PARM_DESC(debug, "Debugging mode enabled or not");