linux_dsm_epyc7002/drivers/pci/pcie/pme.c
Rafael J. Wysocki 379021d5c0 PCI / PM: Extend PME polling to all PCI devices
The land of PCI power management is a land of sorrow and ugliness,
especially in the area of signaling events by devices.  There are
devices that set their PME Status bits, but don't really bother
to send a PME message or assert PME#.  There are hardware vendors
who don't connect PME# lines to the system core logic (they know
who they are).  There are PCI Express Root Ports that don't bother
to trigger interrupts when they receive PME messages from the devices
below.  There are ACPI BIOSes that forget to provide _PRW methods for
devices capable of signaling wakeup.  Finally, there are BIOSes that
do provide _PRW methods for such devices, but then don't bother to
call Notify() for those devices from the corresponding _Lxx/_Exx
GPE-handling methods.  In all of these cases the kernel doesn't have
a chance to receive a proper notification that it should wake up a
device, so devices stay in low-power states forever.  Worse yet, in
some cases they continuously send PME Messages that are silently
ignored, because the kernel simply doesn't know that it should clear
the device's PME Status bit.

This problem was first observed for "parallel" (non-Express) PCI
devices on add-on cards and Matthew Garrett addressed it by adding
code that polls PME Status bits of such devices, if they are enabled
to signal PME, to the kernel.  Recently, however, it has turned out
that PCI Express devices are also affected by this issue and that it
is not limited to add-on devices, so it seems necessary to extend
the PME polling to all PCI devices, including PCI Express and planar
ones.  Still, it would be wasteful to poll the PME Status bits of
devices that are known to receive proper PME notifications, so make
the kernel (1) poll the PME Status bits of all PCI and PCIe devices
enabled to signal PME and (2) disable the PME Status polling for
devices for which correct PME notifications are received.

Tested-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
2011-10-14 09:05:31 -07:00

453 lines
11 KiB
C

/*
* PCIe Native PME support
*
* Copyright (C) 2007 - 2009 Intel Corp
* Copyright (C) 2007 - 2009 Shaohua Li <shaohua.li@intel.com>
* Copyright (C) 2009 Rafael J. Wysocki <rjw@sisk.pl>, Novell Inc.
*
* This file is subject to the terms and conditions of the GNU General Public
* License V2. See the file "COPYING" in the main directory of this archive
* for more details.
*/
#include <linux/module.h>
#include <linux/pci.h>
#include <linux/kernel.h>
#include <linux/errno.h>
#include <linux/slab.h>
#include <linux/init.h>
#include <linux/interrupt.h>
#include <linux/device.h>
#include <linux/pcieport_if.h>
#include <linux/acpi.h>
#include <linux/pci-acpi.h>
#include <linux/pm_runtime.h>
#include "../pci.h"
#include "portdrv.h"
/*
* If this switch is set, MSI will not be used for PCIe PME signaling. This
* causes the PCIe port driver to use INTx interrupts only, but it turns out
* that using MSI for PCIe PME signaling doesn't play well with PCIe PME-based
* wake-up from system sleep states.
*/
bool pcie_pme_msi_disabled;
static int __init pcie_pme_setup(char *str)
{
if (!strncmp(str, "nomsi", 5))
pcie_pme_msi_disabled = true;
return 1;
}
__setup("pcie_pme=", pcie_pme_setup);
struct pcie_pme_service_data {
spinlock_t lock;
struct pcie_device *srv;
struct work_struct work;
bool noirq; /* Don't enable the PME interrupt used by this service. */
};
/**
* pcie_pme_interrupt_enable - Enable/disable PCIe PME interrupt generation.
* @dev: PCIe root port or event collector.
* @enable: Enable or disable the interrupt.
*/
void pcie_pme_interrupt_enable(struct pci_dev *dev, bool enable)
{
int rtctl_pos;
u16 rtctl;
rtctl_pos = pci_pcie_cap(dev) + PCI_EXP_RTCTL;
pci_read_config_word(dev, rtctl_pos, &rtctl);
if (enable)
rtctl |= PCI_EXP_RTCTL_PMEIE;
else
rtctl &= ~PCI_EXP_RTCTL_PMEIE;
pci_write_config_word(dev, rtctl_pos, rtctl);
}
/**
* pcie_pme_walk_bus - Scan a PCI bus for devices asserting PME#.
* @bus: PCI bus to scan.
*
* Scan given PCI bus and all buses under it for devices asserting PME#.
*/
static bool pcie_pme_walk_bus(struct pci_bus *bus)
{
struct pci_dev *dev;
bool ret = false;
list_for_each_entry(dev, &bus->devices, bus_list) {
/* Skip PCIe devices in case we started from a root port. */
if (!pci_is_pcie(dev) && pci_check_pme_status(dev)) {
if (dev->pme_poll)
dev->pme_poll = false;
pci_wakeup_event(dev);
pm_request_resume(&dev->dev);
ret = true;
}
if (dev->subordinate && pcie_pme_walk_bus(dev->subordinate))
ret = true;
}
return ret;
}
/**
* pcie_pme_from_pci_bridge - Check if PCIe-PCI bridge generated a PME.
* @bus: Secondary bus of the bridge.
* @devfn: Device/function number to check.
*
* PME from PCI devices under a PCIe-PCI bridge may be converted to an in-band
* PCIe PME message. In such that case the bridge should use the Requester ID
* of device/function number 0 on its secondary bus.
*/
static bool pcie_pme_from_pci_bridge(struct pci_bus *bus, u8 devfn)
{
struct pci_dev *dev;
bool found = false;
if (devfn)
return false;
dev = pci_dev_get(bus->self);
if (!dev)
return false;
if (pci_is_pcie(dev) && dev->pcie_type == PCI_EXP_TYPE_PCI_BRIDGE) {
down_read(&pci_bus_sem);
if (pcie_pme_walk_bus(bus))
found = true;
up_read(&pci_bus_sem);
}
pci_dev_put(dev);
return found;
}
/**
* pcie_pme_handle_request - Find device that generated PME and handle it.
* @port: Root port or event collector that generated the PME interrupt.
* @req_id: PCIe Requester ID of the device that generated the PME.
*/
static void pcie_pme_handle_request(struct pci_dev *port, u16 req_id)
{
u8 busnr = req_id >> 8, devfn = req_id & 0xff;
struct pci_bus *bus;
struct pci_dev *dev;
bool found = false;
/* First, check if the PME is from the root port itself. */
if (port->devfn == devfn && port->bus->number == busnr) {
if (port->pme_poll)
port->pme_poll = false;
if (pci_check_pme_status(port)) {
pm_request_resume(&port->dev);
found = true;
} else {
/*
* Apparently, the root port generated the PME on behalf
* of a non-PCIe device downstream. If this is done by
* a root port, the Requester ID field in its status
* register may contain either the root port's, or the
* source device's information (PCI Express Base
* Specification, Rev. 2.0, Section 6.1.9).
*/
down_read(&pci_bus_sem);
found = pcie_pme_walk_bus(port->subordinate);
up_read(&pci_bus_sem);
}
goto out;
}
/* Second, find the bus the source device is on. */
bus = pci_find_bus(pci_domain_nr(port->bus), busnr);
if (!bus)
goto out;
/* Next, check if the PME is from a PCIe-PCI bridge. */
found = pcie_pme_from_pci_bridge(bus, devfn);
if (found)
goto out;
/* Finally, try to find the PME source on the bus. */
down_read(&pci_bus_sem);
list_for_each_entry(dev, &bus->devices, bus_list) {
pci_dev_get(dev);
if (dev->devfn == devfn) {
found = true;
break;
}
pci_dev_put(dev);
}
up_read(&pci_bus_sem);
if (found) {
/* The device is there, but we have to check its PME status. */
found = pci_check_pme_status(dev);
if (found) {
if (dev->pme_poll)
dev->pme_poll = false;
pci_wakeup_event(dev);
pm_request_resume(&dev->dev);
}
pci_dev_put(dev);
} else if (devfn) {
/*
* The device is not there, but we can still try to recover by
* assuming that the PME was reported by a PCIe-PCI bridge that
* used devfn different from zero.
*/
dev_dbg(&port->dev, "PME interrupt generated for "
"non-existent device %02x:%02x.%d\n",
busnr, PCI_SLOT(devfn), PCI_FUNC(devfn));
found = pcie_pme_from_pci_bridge(bus, 0);
}
out:
if (!found)
dev_dbg(&port->dev, "Spurious native PME interrupt!\n");
}
/**
* pcie_pme_work_fn - Work handler for PCIe PME interrupt.
* @work: Work structure giving access to service data.
*/
static void pcie_pme_work_fn(struct work_struct *work)
{
struct pcie_pme_service_data *data =
container_of(work, struct pcie_pme_service_data, work);
struct pci_dev *port = data->srv->port;
int rtsta_pos;
u32 rtsta;
rtsta_pos = pci_pcie_cap(port) + PCI_EXP_RTSTA;
spin_lock_irq(&data->lock);
for (;;) {
if (data->noirq)
break;
pci_read_config_dword(port, rtsta_pos, &rtsta);
if (rtsta & PCI_EXP_RTSTA_PME) {
/*
* Clear PME status of the port. If there are other
* pending PMEs, the status will be set again.
*/
pcie_clear_root_pme_status(port);
spin_unlock_irq(&data->lock);
pcie_pme_handle_request(port, rtsta & 0xffff);
spin_lock_irq(&data->lock);
continue;
}
/* No need to loop if there are no more PMEs pending. */
if (!(rtsta & PCI_EXP_RTSTA_PENDING))
break;
spin_unlock_irq(&data->lock);
cpu_relax();
spin_lock_irq(&data->lock);
}
if (!data->noirq)
pcie_pme_interrupt_enable(port, true);
spin_unlock_irq(&data->lock);
}
/**
* pcie_pme_irq - Interrupt handler for PCIe root port PME interrupt.
* @irq: Interrupt vector.
* @context: Interrupt context pointer.
*/
static irqreturn_t pcie_pme_irq(int irq, void *context)
{
struct pci_dev *port;
struct pcie_pme_service_data *data;
int rtsta_pos;
u32 rtsta;
unsigned long flags;
port = ((struct pcie_device *)context)->port;
data = get_service_data((struct pcie_device *)context);
rtsta_pos = pci_pcie_cap(port) + PCI_EXP_RTSTA;
spin_lock_irqsave(&data->lock, flags);
pci_read_config_dword(port, rtsta_pos, &rtsta);
if (!(rtsta & PCI_EXP_RTSTA_PME)) {
spin_unlock_irqrestore(&data->lock, flags);
return IRQ_NONE;
}
pcie_pme_interrupt_enable(port, false);
spin_unlock_irqrestore(&data->lock, flags);
/* We don't use pm_wq, because it's freezable. */
schedule_work(&data->work);
return IRQ_HANDLED;
}
/**
* pcie_pme_set_native - Set the PME interrupt flag for given device.
* @dev: PCI device to handle.
* @ign: Ignored.
*/
static int pcie_pme_set_native(struct pci_dev *dev, void *ign)
{
dev_info(&dev->dev, "Signaling PME through PCIe PME interrupt\n");
device_set_run_wake(&dev->dev, true);
dev->pme_interrupt = true;
return 0;
}
/**
* pcie_pme_mark_devices - Set the PME interrupt flag for devices below a port.
* @port: PCIe root port or event collector to handle.
*
* For each device below given root port, including the port itself (or for each
* root complex integrated endpoint if @port is a root complex event collector)
* set the flag indicating that it can signal run-time wake-up events via PCIe
* PME interrupts.
*/
static void pcie_pme_mark_devices(struct pci_dev *port)
{
pcie_pme_set_native(port, NULL);
if (port->subordinate) {
pci_walk_bus(port->subordinate, pcie_pme_set_native, NULL);
} else {
struct pci_bus *bus = port->bus;
struct pci_dev *dev;
/* Check if this is a root port event collector. */
if (port->pcie_type != PCI_EXP_TYPE_RC_EC || !bus)
return;
down_read(&pci_bus_sem);
list_for_each_entry(dev, &bus->devices, bus_list)
if (pci_is_pcie(dev)
&& dev->pcie_type == PCI_EXP_TYPE_RC_END)
pcie_pme_set_native(dev, NULL);
up_read(&pci_bus_sem);
}
}
/**
* pcie_pme_probe - Initialize PCIe PME service for given root port.
* @srv: PCIe service to initialize.
*/
static int pcie_pme_probe(struct pcie_device *srv)
{
struct pci_dev *port;
struct pcie_pme_service_data *data;
int ret;
data = kzalloc(sizeof(*data), GFP_KERNEL);
if (!data)
return -ENOMEM;
spin_lock_init(&data->lock);
INIT_WORK(&data->work, pcie_pme_work_fn);
data->srv = srv;
set_service_data(srv, data);
port = srv->port;
pcie_pme_interrupt_enable(port, false);
pcie_clear_root_pme_status(port);
ret = request_irq(srv->irq, pcie_pme_irq, IRQF_SHARED, "PCIe PME", srv);
if (ret) {
kfree(data);
} else {
pcie_pme_mark_devices(port);
pcie_pme_interrupt_enable(port, true);
}
return ret;
}
/**
* pcie_pme_suspend - Suspend PCIe PME service device.
* @srv: PCIe service device to suspend.
*/
static int pcie_pme_suspend(struct pcie_device *srv)
{
struct pcie_pme_service_data *data = get_service_data(srv);
struct pci_dev *port = srv->port;
spin_lock_irq(&data->lock);
pcie_pme_interrupt_enable(port, false);
pcie_clear_root_pme_status(port);
data->noirq = true;
spin_unlock_irq(&data->lock);
synchronize_irq(srv->irq);
return 0;
}
/**
* pcie_pme_resume - Resume PCIe PME service device.
* @srv - PCIe service device to resume.
*/
static int pcie_pme_resume(struct pcie_device *srv)
{
struct pcie_pme_service_data *data = get_service_data(srv);
struct pci_dev *port = srv->port;
spin_lock_irq(&data->lock);
data->noirq = false;
pcie_clear_root_pme_status(port);
pcie_pme_interrupt_enable(port, true);
spin_unlock_irq(&data->lock);
return 0;
}
/**
* pcie_pme_remove - Prepare PCIe PME service device for removal.
* @srv - PCIe service device to resume.
*/
static void pcie_pme_remove(struct pcie_device *srv)
{
pcie_pme_suspend(srv);
free_irq(srv->irq, srv);
kfree(get_service_data(srv));
}
static struct pcie_port_service_driver pcie_pme_driver = {
.name = "pcie_pme",
.port_type = PCI_EXP_TYPE_ROOT_PORT,
.service = PCIE_PORT_SERVICE_PME,
.probe = pcie_pme_probe,
.suspend = pcie_pme_suspend,
.resume = pcie_pme_resume,
.remove = pcie_pme_remove,
};
/**
* pcie_pme_service_init - Register the PCIe PME service driver.
*/
static int __init pcie_pme_service_init(void)
{
return pcie_port_service_register(&pcie_pme_driver);
}
module_init(pcie_pme_service_init);