1
0
Fork 0
remarkable-linux/drivers/pci/hotplug/cpci_hotplug_core.c

703 lines
16 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0+
/*
* CompactPCI Hot Plug Driver
*
* Copyright (C) 2002,2005 SOMA Networks, Inc.
* Copyright (C) 2001 Greg Kroah-Hartman (greg@kroah.com)
* Copyright (C) 2001 IBM Corp.
*
* All rights reserved.
*
* Send feedback to <scottm@somanetworks.com>
*/
#include <linux/module.h>
#include <linux/kernel.h>
#include <linux/sched/signal.h>
#include <linux/slab.h>
#include <linux/pci.h>
#include <linux/pci_hotplug.h>
#include <linux/init.h>
#include <linux/interrupt.h>
#include <linux/atomic.h>
#include <linux/delay.h>
#include <linux/kthread.h>
#include "cpci_hotplug.h"
#define DRIVER_AUTHOR "Scott Murray <scottm@somanetworks.com>"
#define DRIVER_DESC "CompactPCI Hot Plug Core"
#define MY_NAME "cpci_hotplug"
#define dbg(format, arg...) \
do { \
if (cpci_debug) \
printk(KERN_DEBUG "%s: " format "\n", \
MY_NAME, ## arg); \
} while (0)
#define err(format, arg...) printk(KERN_ERR "%s: " format "\n", MY_NAME, ## arg)
#define info(format, arg...) printk(KERN_INFO "%s: " format "\n", MY_NAME, ## arg)
#define warn(format, arg...) printk(KERN_WARNING "%s: " format "\n", MY_NAME, ## arg)
/* local variables */
static DECLARE_RWSEM(list_rwsem);
static LIST_HEAD(slot_list);
static int slots;
static atomic_t extracting;
int cpci_debug;
static struct cpci_hp_controller *controller;
static struct task_struct *cpci_thread;
static int thread_finished;
static int enable_slot(struct hotplug_slot *slot);
static int disable_slot(struct hotplug_slot *slot);
static int set_attention_status(struct hotplug_slot *slot, u8 value);
static int get_power_status(struct hotplug_slot *slot, u8 *value);
static int get_attention_status(struct hotplug_slot *slot, u8 *value);
static int get_adapter_status(struct hotplug_slot *slot, u8 *value);
static int get_latch_status(struct hotplug_slot *slot, u8 *value);
static struct hotplug_slot_ops cpci_hotplug_slot_ops = {
.enable_slot = enable_slot,
.disable_slot = disable_slot,
.set_attention_status = set_attention_status,
.get_power_status = get_power_status,
.get_attention_status = get_attention_status,
.get_adapter_status = get_adapter_status,
.get_latch_status = get_latch_status,
};
static int
update_latch_status(struct hotplug_slot *hotplug_slot, u8 value)
{
struct hotplug_slot_info info;
memcpy(&info, hotplug_slot->info, sizeof(struct hotplug_slot_info));
info.latch_status = value;
return pci_hp_change_slot_info(hotplug_slot, &info);
}
static int
update_adapter_status(struct hotplug_slot *hotplug_slot, u8 value)
{
struct hotplug_slot_info info;
memcpy(&info, hotplug_slot->info, sizeof(struct hotplug_slot_info));
info.adapter_status = value;
return pci_hp_change_slot_info(hotplug_slot, &info);
}
static int
enable_slot(struct hotplug_slot *hotplug_slot)
{
struct slot *slot = hotplug_slot->private;
int retval = 0;
dbg("%s - physical_slot = %s", __func__, slot_name(slot));
if (controller->ops->set_power)
retval = controller->ops->set_power(slot, 1);
return retval;
}
static int
disable_slot(struct hotplug_slot *hotplug_slot)
{
struct slot *slot = hotplug_slot->private;
int retval = 0;
dbg("%s - physical_slot = %s", __func__, slot_name(slot));
down_write(&list_rwsem);
/* Unconfigure device */
dbg("%s - unconfiguring slot %s", __func__, slot_name(slot));
retval = cpci_unconfigure_slot(slot);
if (retval) {
err("%s - could not unconfigure slot %s",
__func__, slot_name(slot));
goto disable_error;
}
dbg("%s - finished unconfiguring slot %s", __func__, slot_name(slot));
/* Clear EXT (by setting it) */
if (cpci_clear_ext(slot)) {
err("%s - could not clear EXT for slot %s",
__func__, slot_name(slot));
retval = -ENODEV;
goto disable_error;
}
cpci_led_on(slot);
if (controller->ops->set_power) {
retval = controller->ops->set_power(slot, 0);
if (retval)
goto disable_error;
}
if (update_adapter_status(slot->hotplug_slot, 0))
warn("failure to update adapter file");
if (slot->extracting) {
slot->extracting = 0;
atomic_dec(&extracting);
}
disable_error:
up_write(&list_rwsem);
return retval;
}
static u8
cpci_get_power_status(struct slot *slot)
{
u8 power = 1;
if (controller->ops->get_power)
power = controller->ops->get_power(slot);
return power;
}
static int
get_power_status(struct hotplug_slot *hotplug_slot, u8 *value)
{
struct slot *slot = hotplug_slot->private;
*value = cpci_get_power_status(slot);
return 0;
}
static int
get_attention_status(struct hotplug_slot *hotplug_slot, u8 *value)
{
struct slot *slot = hotplug_slot->private;
*value = cpci_get_attention_status(slot);
return 0;
}
static int
set_attention_status(struct hotplug_slot *hotplug_slot, u8 status)
{
return cpci_set_attention_status(hotplug_slot->private, status);
}
static int
get_adapter_status(struct hotplug_slot *hotplug_slot, u8 *value)
{
*value = hotplug_slot->info->adapter_status;
return 0;
}
static int
get_latch_status(struct hotplug_slot *hotplug_slot, u8 *value)
{
*value = hotplug_slot->info->latch_status;
return 0;
}
PCI: hotplug: Demidlayer registration with the core When a hotplug driver calls pci_hp_register(), all steps necessary for registration are carried out in one go, including creation of a kobject and addition to sysfs. That's a problem for pciehp once it's converted to enable/disable the slot exclusively from the IRQ thread: The thread needs to be spawned after creation of the kobject (because it uses the kobject's name), but before addition to sysfs (because it will handle enable/disable requests submitted via sysfs). pci_hp_deregister() does offer a ->release callback that's invoked after deletion from sysfs and before destruction of the kobject. But because pci_hp_register() doesn't offer a counterpart, hotplug drivers' ->probe and ->remove code becomes asymmetric, which is error prone as recently discovered use-after-free bugs in pciehp's ->remove hook have shown. In a sense, this appears to be a case of the midlayer antipattern: "The core thesis of the "midlayer mistake" is that midlayers are bad and should not exist. That common functionality which it is so tempting to put in a midlayer should instead be provided as library routines which can [be] used, augmented, or ignored by each bottom level driver independently. Thus every subsystem that supports multiple implementations (or drivers) should provide a very thin top layer which calls directly into the bottom layer drivers, and a rich library of support code that eases the implementation of those drivers. This library is available to, but not forced upon, those drivers." -- Neil Brown (2009), https://lwn.net/Articles/336262/ The presence of midlayer traits in the PCI hotplug core might be ascribed to its age: When it was introduced in February 2002, the blessings of a library approach might not have been well known: https://git.kernel.org/tglx/history/c/a8a2069f432c For comparison, the driver core does offer split functions for creating a kobject (device_initialize()) and addition to sysfs (device_add()) as an alternative to carrying out everything at once (device_register()). This was introduced in October 2002: https://git.kernel.org/tglx/history/c/8b290eb19962 The odd ->release callback in the PCI hotplug core was added in 2003: https://git.kernel.org/tglx/history/c/69f8d663b595 Clearly, a library approach would not force every hotplug driver to implement a ->release callback, but rather allow the driver to remove the sysfs files, release its data structures and finally destroy the kobject. Alternatively, a driver may choose to remove everything with pci_hp_deregister(), then release its data structures. To this end, offer drivers pci_hp_initialize() and pci_hp_add() as a split-up version of pci_hp_register(). Likewise, offer pci_hp_del() and pci_hp_destroy() as a split-up version of pci_hp_deregister(). Eliminate the ->release callback and move its code into each driver's teardown routine. Declare pci_hp_deregister() void, in keeping with the usual kernel pattern that enablement can fail, but disablement cannot. It only returned an error if the caller passed in a NULL pointer or a slot which has never or is no longer registered or is sharing its name with another slot. Those would be bugs, so WARN about them. Few hotplug drivers actually checked the return value and those that did only printed a useless error message to dmesg. Remove that. For most drivers the conversion was straightforward since it doesn't matter whether the code in the ->release callback is executed before or after destruction of the kobject. But in the case of ibmphp, it was unclear to me whether setting slot_cur->ctrl and slot_cur->bus_on to NULL needs to happen before the kobject is destroyed, so I erred on the side of caution and ensured that the order stays the same. Another nontrivial case is pnv_php, I've found the list and kref logic difficult to understand, however my impression was that it is safe to delete the list element and drop the references until after the kobject is destroyed. Signed-off-by: Lukas Wunner <lukas@wunner.de> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Acked-by: Andy Shevchenko <andy.shevchenko@gmail.com> # drivers/platform/x86 Cc: Rafael J. Wysocki <rjw@rjwysocki.net> Cc: Len Brown <lenb@kernel.org> Cc: Scott Murray <scott@spiteful.org> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Michael Ellerman <mpe@ellerman.id.au> Cc: Gavin Shan <gwshan@linux.vnet.ibm.com> Cc: Sebastian Ott <sebott@linux.vnet.ibm.com> Cc: Gerald Schaefer <gerald.schaefer@de.ibm.com> Cc: Corentin Chary <corentin.chary@gmail.com> Cc: Darren Hart <dvhart@infradead.org> Cc: Andy Shevchenko <andy@infradead.org>
2018-07-19 16:27:43 -06:00
static void release_slot(struct slot *slot)
{
kfree(slot->hotplug_slot->info);
kfree(slot->hotplug_slot);
pci_dev_put(slot->dev);
kfree(slot);
}
#define SLOT_NAME_SIZE 6
int
cpci_hp_register_bus(struct pci_bus *bus, u8 first, u8 last)
{
struct slot *slot;
struct hotplug_slot *hotplug_slot;
struct hotplug_slot_info *info;
char name[SLOT_NAME_SIZE];
int status;
int i;
if (!(controller && bus))
return -ENODEV;
/*
* Create a structure for each slot, and register that slot
* with the pci_hotplug subsystem.
*/
for (i = first; i <= last; ++i) {
slot = kzalloc(sizeof(struct slot), GFP_KERNEL);
if (!slot) {
status = -ENOMEM;
goto error;
}
hotplug_slot =
kzalloc(sizeof(struct hotplug_slot), GFP_KERNEL);
if (!hotplug_slot) {
status = -ENOMEM;
goto error_slot;
}
slot->hotplug_slot = hotplug_slot;
info = kzalloc(sizeof(struct hotplug_slot_info), GFP_KERNEL);
if (!info) {
status = -ENOMEM;
goto error_hpslot;
}
hotplug_slot->info = info;
slot->bus = bus;
slot->number = i;
slot->devfn = PCI_DEVFN(i, 0);
snprintf(name, SLOT_NAME_SIZE, "%02x:%02x", bus->number, i);
hotplug_slot->private = slot;
hotplug_slot->ops = &cpci_hotplug_slot_ops;
/*
* Initialize the slot info structure with some known
* good values.
*/
dbg("initializing slot %s", name);
info->power_status = cpci_get_power_status(slot);
info->attention_status = cpci_get_attention_status(slot);
dbg("registering slot %s", name);
status = pci_hp_register(slot->hotplug_slot, bus, i, name);
if (status) {
err("pci_hp_register failed with error %d", status);
goto error_info;
}
dbg("slot registered with name: %s", slot_name(slot));
/* Add slot to our internal list */
down_write(&list_rwsem);
list_add(&slot->slot_list, &slot_list);
slots++;
up_write(&list_rwsem);
}
return 0;
error_info:
kfree(info);
error_hpslot:
kfree(hotplug_slot);
error_slot:
kfree(slot);
error:
return status;
}
EXPORT_SYMBOL_GPL(cpci_hp_register_bus);
int
cpci_hp_unregister_bus(struct pci_bus *bus)
{
struct slot *slot;
struct slot *tmp;
int status = 0;
down_write(&list_rwsem);
if (!slots) {
up_write(&list_rwsem);
return -1;
}
list_for_each_entry_safe(slot, tmp, &slot_list, slot_list) {
if (slot->bus == bus) {
list_del(&slot->slot_list);
slots--;
dbg("deregistering slot %s", slot_name(slot));
PCI: hotplug: Demidlayer registration with the core When a hotplug driver calls pci_hp_register(), all steps necessary for registration are carried out in one go, including creation of a kobject and addition to sysfs. That's a problem for pciehp once it's converted to enable/disable the slot exclusively from the IRQ thread: The thread needs to be spawned after creation of the kobject (because it uses the kobject's name), but before addition to sysfs (because it will handle enable/disable requests submitted via sysfs). pci_hp_deregister() does offer a ->release callback that's invoked after deletion from sysfs and before destruction of the kobject. But because pci_hp_register() doesn't offer a counterpart, hotplug drivers' ->probe and ->remove code becomes asymmetric, which is error prone as recently discovered use-after-free bugs in pciehp's ->remove hook have shown. In a sense, this appears to be a case of the midlayer antipattern: "The core thesis of the "midlayer mistake" is that midlayers are bad and should not exist. That common functionality which it is so tempting to put in a midlayer should instead be provided as library routines which can [be] used, augmented, or ignored by each bottom level driver independently. Thus every subsystem that supports multiple implementations (or drivers) should provide a very thin top layer which calls directly into the bottom layer drivers, and a rich library of support code that eases the implementation of those drivers. This library is available to, but not forced upon, those drivers." -- Neil Brown (2009), https://lwn.net/Articles/336262/ The presence of midlayer traits in the PCI hotplug core might be ascribed to its age: When it was introduced in February 2002, the blessings of a library approach might not have been well known: https://git.kernel.org/tglx/history/c/a8a2069f432c For comparison, the driver core does offer split functions for creating a kobject (device_initialize()) and addition to sysfs (device_add()) as an alternative to carrying out everything at once (device_register()). This was introduced in October 2002: https://git.kernel.org/tglx/history/c/8b290eb19962 The odd ->release callback in the PCI hotplug core was added in 2003: https://git.kernel.org/tglx/history/c/69f8d663b595 Clearly, a library approach would not force every hotplug driver to implement a ->release callback, but rather allow the driver to remove the sysfs files, release its data structures and finally destroy the kobject. Alternatively, a driver may choose to remove everything with pci_hp_deregister(), then release its data structures. To this end, offer drivers pci_hp_initialize() and pci_hp_add() as a split-up version of pci_hp_register(). Likewise, offer pci_hp_del() and pci_hp_destroy() as a split-up version of pci_hp_deregister(). Eliminate the ->release callback and move its code into each driver's teardown routine. Declare pci_hp_deregister() void, in keeping with the usual kernel pattern that enablement can fail, but disablement cannot. It only returned an error if the caller passed in a NULL pointer or a slot which has never or is no longer registered or is sharing its name with another slot. Those would be bugs, so WARN about them. Few hotplug drivers actually checked the return value and those that did only printed a useless error message to dmesg. Remove that. For most drivers the conversion was straightforward since it doesn't matter whether the code in the ->release callback is executed before or after destruction of the kobject. But in the case of ibmphp, it was unclear to me whether setting slot_cur->ctrl and slot_cur->bus_on to NULL needs to happen before the kobject is destroyed, so I erred on the side of caution and ensured that the order stays the same. Another nontrivial case is pnv_php, I've found the list and kref logic difficult to understand, however my impression was that it is safe to delete the list element and drop the references until after the kobject is destroyed. Signed-off-by: Lukas Wunner <lukas@wunner.de> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Acked-by: Andy Shevchenko <andy.shevchenko@gmail.com> # drivers/platform/x86 Cc: Rafael J. Wysocki <rjw@rjwysocki.net> Cc: Len Brown <lenb@kernel.org> Cc: Scott Murray <scott@spiteful.org> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Michael Ellerman <mpe@ellerman.id.au> Cc: Gavin Shan <gwshan@linux.vnet.ibm.com> Cc: Sebastian Ott <sebott@linux.vnet.ibm.com> Cc: Gerald Schaefer <gerald.schaefer@de.ibm.com> Cc: Corentin Chary <corentin.chary@gmail.com> Cc: Darren Hart <dvhart@infradead.org> Cc: Andy Shevchenko <andy@infradead.org>
2018-07-19 16:27:43 -06:00
pci_hp_deregister(slot->hotplug_slot);
release_slot(slot);
}
}
up_write(&list_rwsem);
return status;
}
EXPORT_SYMBOL_GPL(cpci_hp_unregister_bus);
/* This is the interrupt mode interrupt handler */
static irqreturn_t
IRQ: Maintain regs pointer globally rather than passing to IRQ handlers Maintain a per-CPU global "struct pt_regs *" variable which can be used instead of passing regs around manually through all ~1800 interrupt handlers in the Linux kernel. The regs pointer is used in few places, but it potentially costs both stack space and code to pass it around. On the FRV arch, removing the regs parameter from all the genirq function results in a 20% speed up of the IRQ exit path (ie: from leaving timer_interrupt() to leaving do_IRQ()). Where appropriate, an arch may override the generic storage facility and do something different with the variable. On FRV, for instance, the address is maintained in GR28 at all times inside the kernel as part of general exception handling. Having looked over the code, it appears that the parameter may be handed down through up to twenty or so layers of functions. Consider a USB character device attached to a USB hub, attached to a USB controller that posts its interrupts through a cascaded auxiliary interrupt controller. A character device driver may want to pass regs to the sysrq handler through the input layer which adds another few layers of parameter passing. I've build this code with allyesconfig for x86_64 and i386. I've runtested the main part of the code on FRV and i386, though I can't test most of the drivers. I've also done partial conversion for powerpc and MIPS - these at least compile with minimal configurations. This will affect all archs. Mostly the changes should be relatively easy. Take do_IRQ(), store the regs pointer at the beginning, saving the old one: struct pt_regs *old_regs = set_irq_regs(regs); And put the old one back at the end: set_irq_regs(old_regs); Don't pass regs through to generic_handle_irq() or __do_IRQ(). In timer_interrupt(), this sort of change will be necessary: - update_process_times(user_mode(regs)); - profile_tick(CPU_PROFILING, regs); + update_process_times(user_mode(get_irq_regs())); + profile_tick(CPU_PROFILING); I'd like to move update_process_times()'s use of get_irq_regs() into itself, except that i386, alone of the archs, uses something other than user_mode(). Some notes on the interrupt handling in the drivers: (*) input_dev() is now gone entirely. The regs pointer is no longer stored in the input_dev struct. (*) finish_unlinks() in drivers/usb/host/ohci-q.c needs checking. It does something different depending on whether it's been supplied with a regs pointer or not. (*) Various IRQ handler function pointers have been moved to type irq_handler_t. Signed-Off-By: David Howells <dhowells@redhat.com> (cherry picked from 1b16e7ac850969f38b375e511e3fa2f474a33867 commit)
2006-10-05 07:55:46 -06:00
cpci_hp_intr(int irq, void *data)
{
dbg("entered cpci_hp_intr");
/* Check to see if it was our interrupt */
if ((controller->irq_flags & IRQF_SHARED) &&
!controller->ops->check_irq(controller->dev_id)) {
dbg("exited cpci_hp_intr, not our interrupt");
return IRQ_NONE;
}
/* Disable ENUM interrupt */
controller->ops->disable_irq();
/* Trigger processing by the event thread */
wake_up_process(cpci_thread);
return IRQ_HANDLED;
}
/*
* According to PICMG 2.1 R2.0, section 6.3.2, upon
* initialization, the system driver shall clear the
* INS bits of the cold-inserted devices.
*/
static int
init_slots(int clear_ins)
{
struct slot *slot;
struct pci_dev *dev;
dbg("%s - enter", __func__);
down_read(&list_rwsem);
if (!slots) {
up_read(&list_rwsem);
return -1;
}
list_for_each_entry(slot, &slot_list, slot_list) {
dbg("%s - looking at slot %s", __func__, slot_name(slot));
if (clear_ins && cpci_check_and_clear_ins(slot))
dbg("%s - cleared INS for slot %s",
__func__, slot_name(slot));
dev = pci_get_slot(slot->bus, PCI_DEVFN(slot->number, 0));
if (dev) {
if (update_adapter_status(slot->hotplug_slot, 1))
warn("failure to update adapter file");
if (update_latch_status(slot->hotplug_slot, 1))
warn("failure to update latch file");
slot->dev = dev;
}
}
up_read(&list_rwsem);
dbg("%s - exit", __func__);
return 0;
}
static int
check_slots(void)
{
struct slot *slot;
int extracted;
int inserted;
u16 hs_csr;
down_read(&list_rwsem);
if (!slots) {
up_read(&list_rwsem);
err("no slots registered, shutting down");
return -1;
}
extracted = inserted = 0;
list_for_each_entry(slot, &slot_list, slot_list) {
dbg("%s - looking at slot %s", __func__, slot_name(slot));
if (cpci_check_and_clear_ins(slot)) {
/*
* Some broken hardware (e.g. PLX 9054AB) asserts
* ENUM# twice...
*/
if (slot->dev) {
warn("slot %s already inserted",
slot_name(slot));
inserted++;
continue;
}
/* Process insertion */
dbg("%s - slot %s inserted", __func__, slot_name(slot));
/* GSM, debug */
hs_csr = cpci_get_hs_csr(slot);
dbg("%s - slot %s HS_CSR (1) = %04x",
__func__, slot_name(slot), hs_csr);
/* Configure device */
dbg("%s - configuring slot %s",
__func__, slot_name(slot));
if (cpci_configure_slot(slot)) {
err("%s - could not configure slot %s",
__func__, slot_name(slot));
continue;
}
dbg("%s - finished configuring slot %s",
__func__, slot_name(slot));
/* GSM, debug */
hs_csr = cpci_get_hs_csr(slot);
dbg("%s - slot %s HS_CSR (2) = %04x",
__func__, slot_name(slot), hs_csr);
if (update_latch_status(slot->hotplug_slot, 1))
warn("failure to update latch file");
if (update_adapter_status(slot->hotplug_slot, 1))
warn("failure to update adapter file");
cpci_led_off(slot);
/* GSM, debug */
hs_csr = cpci_get_hs_csr(slot);
dbg("%s - slot %s HS_CSR (3) = %04x",
__func__, slot_name(slot), hs_csr);
inserted++;
} else if (cpci_check_ext(slot)) {
/* Process extraction request */
dbg("%s - slot %s extracted",
__func__, slot_name(slot));
/* GSM, debug */
hs_csr = cpci_get_hs_csr(slot);
dbg("%s - slot %s HS_CSR = %04x",
__func__, slot_name(slot), hs_csr);
if (!slot->extracting) {
if (update_latch_status(slot->hotplug_slot, 0))
warn("failure to update latch file");
slot->extracting = 1;
atomic_inc(&extracting);
}
extracted++;
} else if (slot->extracting) {
hs_csr = cpci_get_hs_csr(slot);
if (hs_csr == 0xffff) {
/*
* Hmmm, we're likely hosed at this point, should we
* bother trying to tell the driver or not?
*/
err("card in slot %s was improperly removed",
slot_name(slot));
if (update_adapter_status(slot->hotplug_slot, 0))
warn("failure to update adapter file");
slot->extracting = 0;
atomic_dec(&extracting);
}
}
}
up_read(&list_rwsem);
dbg("inserted=%d, extracted=%d, extracting=%d",
inserted, extracted, atomic_read(&extracting));
if (inserted || extracted)
return extracted;
else if (!atomic_read(&extracting)) {
err("cannot find ENUM# source, shutting down");
return -1;
}
return 0;
}
/* This is the interrupt mode worker thread body */
static int
event_thread(void *data)
{
int rc;
dbg("%s - event thread started", __func__);
while (1) {
dbg("event thread sleeping");
set_current_state(TASK_INTERRUPTIBLE);
schedule();
if (kthread_should_stop())
break;
do {
rc = check_slots();
if (rc > 0) {
/* Give userspace a chance to handle extraction */
msleep(500);
} else if (rc < 0) {
dbg("%s - error checking slots", __func__);
thread_finished = 1;
goto out;
}
} while (atomic_read(&extracting) && !kthread_should_stop());
if (kthread_should_stop())
break;
/* Re-enable ENUM# interrupt */
dbg("%s - re-enabling irq", __func__);
controller->ops->enable_irq();
}
out:
return 0;
}
/* This is the polling mode worker thread body */
static int
poll_thread(void *data)
{
int rc;
while (1) {
if (kthread_should_stop() || signal_pending(current))
break;
if (controller->ops->query_enum()) {
do {
rc = check_slots();
if (rc > 0) {
/* Give userspace a chance to handle extraction */
msleep(500);
} else if (rc < 0) {
dbg("%s - error checking slots", __func__);
thread_finished = 1;
goto out;
}
} while (atomic_read(&extracting) && !kthread_should_stop());
}
msleep(100);
}
out:
return 0;
}
static int
cpci_start_thread(void)
{
if (controller->irq)
cpci_thread = kthread_run(event_thread, NULL, "cpci_hp_eventd");
else
cpci_thread = kthread_run(poll_thread, NULL, "cpci_hp_polld");
if (IS_ERR(cpci_thread)) {
err("Can't start up our thread");
return PTR_ERR(cpci_thread);
}
thread_finished = 0;
return 0;
}
static void
cpci_stop_thread(void)
{
kthread_stop(cpci_thread);
thread_finished = 1;
}
int
cpci_hp_register_controller(struct cpci_hp_controller *new_controller)
{
int status = 0;
if (controller)
return -1;
if (!(new_controller && new_controller->ops))
return -EINVAL;
if (new_controller->irq) {
if (!(new_controller->ops->enable_irq &&
new_controller->ops->disable_irq))
status = -EINVAL;
if (request_irq(new_controller->irq,
cpci_hp_intr,
new_controller->irq_flags,
MY_NAME,
new_controller->dev_id)) {
err("Can't get irq %d for the hotplug cPCI controller",
new_controller->irq);
status = -ENODEV;
}
dbg("%s - acquired controller irq %d",
__func__, new_controller->irq);
}
if (!status)
controller = new_controller;
return status;
}
EXPORT_SYMBOL_GPL(cpci_hp_register_controller);
static void
cleanup_slots(void)
{
struct slot *slot;
struct slot *tmp;
/*
* Unregister all of our slots with the pci_hotplug subsystem,
* and free up all memory that we had allocated.
*/
down_write(&list_rwsem);
if (!slots)
goto cleanup_null;
list_for_each_entry_safe(slot, tmp, &slot_list, slot_list) {
list_del(&slot->slot_list);
pci_hp_deregister(slot->hotplug_slot);
PCI: hotplug: Demidlayer registration with the core When a hotplug driver calls pci_hp_register(), all steps necessary for registration are carried out in one go, including creation of a kobject and addition to sysfs. That's a problem for pciehp once it's converted to enable/disable the slot exclusively from the IRQ thread: The thread needs to be spawned after creation of the kobject (because it uses the kobject's name), but before addition to sysfs (because it will handle enable/disable requests submitted via sysfs). pci_hp_deregister() does offer a ->release callback that's invoked after deletion from sysfs and before destruction of the kobject. But because pci_hp_register() doesn't offer a counterpart, hotplug drivers' ->probe and ->remove code becomes asymmetric, which is error prone as recently discovered use-after-free bugs in pciehp's ->remove hook have shown. In a sense, this appears to be a case of the midlayer antipattern: "The core thesis of the "midlayer mistake" is that midlayers are bad and should not exist. That common functionality which it is so tempting to put in a midlayer should instead be provided as library routines which can [be] used, augmented, or ignored by each bottom level driver independently. Thus every subsystem that supports multiple implementations (or drivers) should provide a very thin top layer which calls directly into the bottom layer drivers, and a rich library of support code that eases the implementation of those drivers. This library is available to, but not forced upon, those drivers." -- Neil Brown (2009), https://lwn.net/Articles/336262/ The presence of midlayer traits in the PCI hotplug core might be ascribed to its age: When it was introduced in February 2002, the blessings of a library approach might not have been well known: https://git.kernel.org/tglx/history/c/a8a2069f432c For comparison, the driver core does offer split functions for creating a kobject (device_initialize()) and addition to sysfs (device_add()) as an alternative to carrying out everything at once (device_register()). This was introduced in October 2002: https://git.kernel.org/tglx/history/c/8b290eb19962 The odd ->release callback in the PCI hotplug core was added in 2003: https://git.kernel.org/tglx/history/c/69f8d663b595 Clearly, a library approach would not force every hotplug driver to implement a ->release callback, but rather allow the driver to remove the sysfs files, release its data structures and finally destroy the kobject. Alternatively, a driver may choose to remove everything with pci_hp_deregister(), then release its data structures. To this end, offer drivers pci_hp_initialize() and pci_hp_add() as a split-up version of pci_hp_register(). Likewise, offer pci_hp_del() and pci_hp_destroy() as a split-up version of pci_hp_deregister(). Eliminate the ->release callback and move its code into each driver's teardown routine. Declare pci_hp_deregister() void, in keeping with the usual kernel pattern that enablement can fail, but disablement cannot. It only returned an error if the caller passed in a NULL pointer or a slot which has never or is no longer registered or is sharing its name with another slot. Those would be bugs, so WARN about them. Few hotplug drivers actually checked the return value and those that did only printed a useless error message to dmesg. Remove that. For most drivers the conversion was straightforward since it doesn't matter whether the code in the ->release callback is executed before or after destruction of the kobject. But in the case of ibmphp, it was unclear to me whether setting slot_cur->ctrl and slot_cur->bus_on to NULL needs to happen before the kobject is destroyed, so I erred on the side of caution and ensured that the order stays the same. Another nontrivial case is pnv_php, I've found the list and kref logic difficult to understand, however my impression was that it is safe to delete the list element and drop the references until after the kobject is destroyed. Signed-off-by: Lukas Wunner <lukas@wunner.de> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Acked-by: Andy Shevchenko <andy.shevchenko@gmail.com> # drivers/platform/x86 Cc: Rafael J. Wysocki <rjw@rjwysocki.net> Cc: Len Brown <lenb@kernel.org> Cc: Scott Murray <scott@spiteful.org> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Michael Ellerman <mpe@ellerman.id.au> Cc: Gavin Shan <gwshan@linux.vnet.ibm.com> Cc: Sebastian Ott <sebott@linux.vnet.ibm.com> Cc: Gerald Schaefer <gerald.schaefer@de.ibm.com> Cc: Corentin Chary <corentin.chary@gmail.com> Cc: Darren Hart <dvhart@infradead.org> Cc: Andy Shevchenko <andy@infradead.org>
2018-07-19 16:27:43 -06:00
release_slot(slot);
}
cleanup_null:
up_write(&list_rwsem);
return;
}
int
cpci_hp_unregister_controller(struct cpci_hp_controller *old_controller)
{
int status = 0;
if (controller) {
if (!thread_finished)
cpci_stop_thread();
if (controller->irq)
free_irq(controller->irq, controller->dev_id);
controller = NULL;
cleanup_slots();
} else
status = -ENODEV;
return status;
}
EXPORT_SYMBOL_GPL(cpci_hp_unregister_controller);
int
cpci_hp_start(void)
{
static int first = 1;
int status;
dbg("%s - enter", __func__);
if (!controller)
return -ENODEV;
down_read(&list_rwsem);
if (list_empty(&slot_list)) {
up_read(&list_rwsem);
return -ENODEV;
}
up_read(&list_rwsem);
status = init_slots(first);
if (first)
first = 0;
if (status)
return status;
status = cpci_start_thread();
if (status)
return status;
dbg("%s - thread started", __func__);
if (controller->irq) {
/* Start enum interrupt processing */
dbg("%s - enabling irq", __func__);
controller->ops->enable_irq();
}
dbg("%s - exit", __func__);
return 0;
}
EXPORT_SYMBOL_GPL(cpci_hp_start);
int
cpci_hp_stop(void)
{
if (!controller)
return -ENODEV;
if (controller->irq) {
/* Stop enum interrupt processing */
dbg("%s - disabling irq", __func__);
controller->ops->disable_irq();
}
cpci_stop_thread();
return 0;
}
EXPORT_SYMBOL_GPL(cpci_hp_stop);
int __init
cpci_hotplug_init(int debug)
{
cpci_debug = debug;
return 0;
}