OpenCloudOS-Kernel/arch/ia64/pci/pci.c

825 lines
20 KiB
C
Raw Normal View History

/*
* pci.c - Low-Level PCI Access in IA-64
*
* Derived from bios32.c of i386 tree.
*
* (c) Copyright 2002, 2005 Hewlett-Packard Development Company, L.P.
* David Mosberger-Tang <davidm@hpl.hp.com>
* Bjorn Helgaas <bjorn.helgaas@hp.com>
* Copyright (C) 2004 Silicon Graphics, Inc.
*
* Note: Above list of copyright holders is incomplete...
*/
#include <linux/acpi.h>
#include <linux/types.h>
#include <linux/kernel.h>
#include <linux/pci.h>
#include <linux/pci-acpi.h>
#include <linux/init.h>
#include <linux/ioport.h>
#include <linux/slab.h>
#include <linux/spinlock.h>
#include <linux/bootmem.h>
#include <linux/export.h>
#include <asm/machvec.h>
#include <asm/page.h>
#include <asm/io.h>
#include <asm/sal.h>
#include <asm/smp.h>
#include <asm/irq.h>
#include <asm/hw_irq.h>
/*
* Low-level SAL-based PCI configuration access functions. Note that SAL
* calls are already serialized (via sal_lock), so we don't need another
* synchronization mechanism here.
*/
#define PCI_SAL_ADDRESS(seg, bus, devfn, reg) \
(((u64) seg << 24) | (bus << 16) | (devfn << 8) | (reg))
/* SAL 3.2 adds support for extended config space. */
#define PCI_SAL_EXT_ADDRESS(seg, bus, devfn, reg) \
(((u64) seg << 28) | (bus << 20) | (devfn << 12) | (reg))
int raw_pci_read(unsigned int seg, unsigned int bus, unsigned int devfn,
int reg, int len, u32 *value)
{
u64 addr, data = 0;
int mode, result;
if (!value || (seg > 65535) || (bus > 255) || (devfn > 255) || (reg > 4095))
return -EINVAL;
if ((seg | reg) <= 255) {
addr = PCI_SAL_ADDRESS(seg, bus, devfn, reg);
mode = 0;
} else if (sal_revision >= SAL_VERSION_CODE(3,2)) {
addr = PCI_SAL_EXT_ADDRESS(seg, bus, devfn, reg);
mode = 1;
} else {
return -EINVAL;
}
result = ia64_sal_pci_config_read(addr, mode, len, &data);
if (result != 0)
return -EINVAL;
*value = (u32) data;
return 0;
}
int raw_pci_write(unsigned int seg, unsigned int bus, unsigned int devfn,
int reg, int len, u32 value)
{
u64 addr;
int mode, result;
if ((seg > 65535) || (bus > 255) || (devfn > 255) || (reg > 4095))
return -EINVAL;
if ((seg | reg) <= 255) {
addr = PCI_SAL_ADDRESS(seg, bus, devfn, reg);
mode = 0;
} else if (sal_revision >= SAL_VERSION_CODE(3,2)) {
addr = PCI_SAL_EXT_ADDRESS(seg, bus, devfn, reg);
mode = 1;
} else {
return -EINVAL;
}
result = ia64_sal_pci_config_write(addr, mode, len, value);
if (result != 0)
return -EINVAL;
return 0;
}
static int pci_read(struct pci_bus *bus, unsigned int devfn, int where,
int size, u32 *value)
{
return raw_pci_read(pci_domain_nr(bus), bus->number,
devfn, where, size, value);
}
static int pci_write(struct pci_bus *bus, unsigned int devfn, int where,
int size, u32 value)
{
return raw_pci_write(pci_domain_nr(bus), bus->number,
devfn, where, size, value);
}
struct pci_ops pci_root_ops = {
.read = pci_read,
.write = pci_write,
};
/* Called by ACPI when it finds a new root bus. */
static struct pci_controller *alloc_pci_controller(int seg)
{
struct pci_controller *controller;
controller = kzalloc(sizeof(*controller), GFP_KERNEL);
if (!controller)
return NULL;
controller->segment = seg;
return controller;
}
struct pci_root_info {
struct acpi_device *bridge;
struct pci_controller *controller;
struct list_head resources;
struct resource *res;
resource_size_t *res_offset;
unsigned int res_num;
struct list_head io_resources;
char *name;
};
static unsigned int
new_space (u64 phys_base, int sparse)
{
u64 mmio_base;
int i;
if (phys_base == 0)
return 0; /* legacy I/O port space */
mmio_base = (u64) ioremap(phys_base, 0);
for (i = 0; i < num_io_spaces; i++)
if (io_space[i].mmio_base == mmio_base &&
io_space[i].sparse == sparse)
return i;
if (num_io_spaces == MAX_IO_SPACES) {
pr_err("PCI: Too many IO port spaces "
"(MAX_IO_SPACES=%lu)\n", MAX_IO_SPACES);
return ~0;
}
i = num_io_spaces++;
io_space[i].mmio_base = mmio_base;
io_space[i].sparse = sparse;
return i;
}
static u64 add_io_space(struct pci_root_info *info,
struct acpi_resource_address64 *addr)
{
struct iospace_resource *iospace;
struct resource *resource;
char *name;
unsigned long base, min, max, base_port;
unsigned int sparse = 0, space_nr, len;
len = strlen(info->name) + 32;
iospace = kzalloc(sizeof(*iospace) + len, GFP_KERNEL);
if (!iospace) {
dev_err(&info->bridge->dev,
"PCI: No memory for %s I/O port space\n",
info->name);
goto out;
}
name = (char *)(iospace + 1);
min = addr->address.minimum;
max = min + addr->address.address_length - 1;
if (addr->info.io.translation_type == ACPI_SPARSE_TRANSLATION)
sparse = 1;
space_nr = new_space(addr->address.translation_offset, sparse);
if (space_nr == ~0)
goto free_resource;
base = __pa(io_space[space_nr].mmio_base);
base_port = IO_SPACE_BASE(space_nr);
snprintf(name, len, "%s I/O Ports %08lx-%08lx", info->name,
base_port + min, base_port + max);
/*
* The SDM guarantees the legacy 0-64K space is sparse, but if the
* mapping is done by the processor (not the bridge), ACPI may not
* mark it as sparse.
*/
if (space_nr == 0)
sparse = 1;
resource = &iospace->res;
resource->name = name;
resource->flags = IORESOURCE_MEM;
resource->start = base + (sparse ? IO_SPACE_SPARSE_ENCODING(min) : min);
resource->end = base + (sparse ? IO_SPACE_SPARSE_ENCODING(max) : max);
if (insert_resource(&iomem_resource, resource)) {
dev_err(&info->bridge->dev,
"can't allocate host bridge io space resource %pR\n",
resource);
goto free_resource;
}
list_add_tail(&iospace->list, &info->io_resources);
return base_port;
free_resource:
kfree(iospace);
out:
return ~0;
}
static acpi_status resource_to_window(struct acpi_resource *resource,
struct acpi_resource_address64 *addr)
{
acpi_status status;
/*
* We're only interested in _CRS descriptors that are
* - address space descriptors for memory or I/O space
* - non-zero size
*/
status = acpi_resource_to_address64(resource, addr);
if (ACPI_SUCCESS(status) &&
(addr->resource_type == ACPI_MEMORY_RANGE ||
addr->resource_type == ACPI_IO_RANGE) &&
ia64/PCI: Treat all host bridge Address Space Descriptors (even consumers) as windows Prior to c770cb4cb505 ("PCI: Mark invalid BARs as unassigned"), if we tried to claim a PCI BAR but could not find an upstream bridge window that matched it, we complained but still allowed the device to be enabled. c770cb4cb505 broke devices that previously worked (mptsas and igb in the case Tony reported, but it could be any devices) because it marks those BARs as IORESOURCE_UNSET, which makes pci_enable_device() complain and return failure: igb 0000:81:00.0: can't enable device: BAR 0 [mem size 0x00020000] not assigned igb: probe of 0000:81:00.0 failed with error -22 The underlying cause is an ACPI Address Space Descriptor for a PCI host bridge window that is marked as "consumer". This is a firmware defect: resources that are produced on the downstream side of a bridge should be marked "producer". But rejecting these BARs that we previously allowed is a functionality regression, and firmware has not used the producer/consumer bit consistently, so we can't rely on it anyway. Stop checking the producer/consumer bit, and assume all bridge Address Space Descriptors are for bridge windows. Note that this change does not affect I/O Port or Fixed Location I/O Port Descriptors, which are commonly used for the [io 0x0cf8-0x0cff] config access range. That range is a "consumer" range and should not be treated as a window. Fixes: c770cb4cb505 ("PCI: Mark invalid BARs as unassigned") Link: https://bugzilla.kernel.org/show_bug.cgi?id=96961 Reported-and-tested-by: Tony Luck <tony.luck@intel.com> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Acked-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2015-04-22 03:57:26 +08:00
addr->address.address_length)
return AE_OK;
return AE_ERROR;
}
static acpi_status count_window(struct acpi_resource *resource, void *data)
{
unsigned int *windows = (unsigned int *) data;
struct acpi_resource_address64 addr;
acpi_status status;
status = resource_to_window(resource, &addr);
if (ACPI_SUCCESS(status))
(*windows)++;
return AE_OK;
}
static acpi_status add_window(struct acpi_resource *res, void *data)
{
struct pci_root_info *info = data;
struct resource *resource;
struct acpi_resource_address64 addr;
acpi_status status;
unsigned long flags, offset = 0;
struct resource *root;
/* Return AE_OK for non-window resources to keep scanning for more */
status = resource_to_window(res, &addr);
if (!ACPI_SUCCESS(status))
return AE_OK;
if (addr.resource_type == ACPI_MEMORY_RANGE) {
flags = IORESOURCE_MEM;
root = &iomem_resource;
offset = addr.address.translation_offset;
} else if (addr.resource_type == ACPI_IO_RANGE) {
flags = IORESOURCE_IO;
root = &ioport_resource;
offset = add_io_space(info, &addr);
if (offset == ~0)
return AE_OK;
} else
return AE_OK;
resource = &info->res[info->res_num];
resource->name = info->name;
resource->flags = flags;
resource->start = addr.address.minimum + offset;
resource->end = resource->start + addr.address.address_length - 1;
info->res_offset[info->res_num] = offset;
if (insert_resource(root, resource)) {
dev_err(&info->bridge->dev,
"can't allocate host bridge window %pR\n",
resource);
} else {
if (offset)
dev_info(&info->bridge->dev, "host bridge window %pR "
"(PCI address [%#llx-%#llx])\n",
resource,
resource->start - offset,
resource->end - offset);
else
dev_info(&info->bridge->dev,
"host bridge window %pR\n", resource);
}
/* HP's firmware has a hack to work around a Windows bug.
* Ignore these tiny memory ranges */
if (!((resource->flags & IORESOURCE_MEM) &&
(resource->end - resource->start < 16)))
pci_add_resource_offset(&info->resources, resource,
info->res_offset[info->res_num]);
info->res_num++;
return AE_OK;
}
static void free_pci_root_info_res(struct pci_root_info *info)
{
struct iospace_resource *iospace, *tmp;
list_for_each_entry_safe(iospace, tmp, &info->io_resources, list)
kfree(iospace);
kfree(info->name);
kfree(info->res);
info->res = NULL;
kfree(info->res_offset);
info->res_offset = NULL;
info->res_num = 0;
kfree(info->controller);
info->controller = NULL;
}
static void __release_pci_root_info(struct pci_root_info *info)
{
int i;
struct resource *res;
struct iospace_resource *iospace;
list_for_each_entry(iospace, &info->io_resources, list)
release_resource(&iospace->res);
for (i = 0; i < info->res_num; i++) {
res = &info->res[i];
if (!res->parent)
continue;
if (!(res->flags & (IORESOURCE_MEM | IORESOURCE_IO)))
continue;
release_resource(res);
}
free_pci_root_info_res(info);
kfree(info);
}
static void release_pci_root_info(struct pci_host_bridge *bridge)
{
struct pci_root_info *info = bridge->release_data;
__release_pci_root_info(info);
}
static int
probe_pci_root_info(struct pci_root_info *info, struct acpi_device *device,
int busnum, int domain)
{
char *name;
name = kmalloc(16, GFP_KERNEL);
if (!name)
return -ENOMEM;
sprintf(name, "PCI Bus %04x:%02x", domain, busnum);
info->bridge = device;
info->name = name;
acpi_walk_resources(device->handle, METHOD_NAME__CRS, count_window,
&info->res_num);
if (info->res_num) {
info->res =
kzalloc_node(sizeof(*info->res) * info->res_num,
GFP_KERNEL, info->controller->node);
if (!info->res) {
kfree(name);
return -ENOMEM;
}
info->res_offset =
kzalloc_node(sizeof(*info->res_offset) * info->res_num,
GFP_KERNEL, info->controller->node);
if (!info->res_offset) {
kfree(name);
kfree(info->res);
info->res = NULL;
return -ENOMEM;
}
info->res_num = 0;
acpi_walk_resources(device->handle, METHOD_NAME__CRS,
add_window, info);
} else
kfree(name);
return 0;
}
struct pci_bus *pci_acpi_scan_root(struct acpi_pci_root *root)
{
struct acpi_device *device = root->device;
int domain = root->segment;
int bus = root->secondary.start;
struct pci_controller *controller;
struct pci_root_info *info = NULL;
int busnum = root->secondary.start;
struct pci_bus *pbus;
int ret;
controller = alloc_pci_controller(domain);
if (!controller)
return NULL;
ACPI / driver core: Store an ACPI device pointer in struct acpi_dev_node Modify struct acpi_dev_node to contain a pointer to struct acpi_device associated with the given device object (that is, its ACPI companion device) instead of an ACPI handle corresponding to it. Introduce two new macros for manipulating that pointer in a CONFIG_ACPI-safe way, ACPI_COMPANION() and ACPI_COMPANION_SET(), and rework the ACPI_HANDLE() macro to take the above changes into account. Drop the ACPI_HANDLE_SET() macro entirely and rework its users to use ACPI_COMPANION_SET() instead. For some of them who used to pass the result of acpi_get_child() directly to ACPI_HANDLE_SET() introduce a helper routine acpi_preset_companion() doing an equivalent thing. The main motivation for doing this is that there are things represented by struct acpi_device objects that don't have valid ACPI handles (so called fixed ACPI hardware features, such as power and sleep buttons) and we would like to create platform device objects for them and "glue" them to their ACPI companions in the usual way (which currently is impossible due to the lack of valid ACPI handles). However, there are more reasons why it may be useful. First, struct acpi_device pointers allow of much better type checking than void pointers which are ACPI handles, so it should be more difficult to write buggy code using modified struct acpi_dev_node and the new macros. Second, the change should help to reduce (over time) the number of places in which the result of ACPI_HANDLE() is passed to acpi_bus_get_device() in order to obtain a pointer to the struct acpi_device associated with the given "physical" device, because now that pointer is returned by ACPI_COMPANION() directly. Finally, the change should make it easier to write generic code that will build both for CONFIG_ACPI set and unset without adding explicit compiler directives to it. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Tested-by: Mika Westerberg <mika.westerberg@linux.intel.com> # on Haswell Reviewed-by: Mika Westerberg <mika.westerberg@linux.intel.com> Reviewed-by: Aaron Lu <aaron.lu@intel.com> # for ATA and SDIO part
2013-11-12 05:41:56 +08:00
controller->companion = device;
controller->node = acpi_get_node(device->handle);
info = kzalloc(sizeof(*info), GFP_KERNEL);
if (!info) {
dev_err(&device->dev,
"pci_bus %04x:%02x: ignored (out of memory)\n",
domain, busnum);
kfree(controller);
return NULL;
}
info->controller = controller;
INIT_LIST_HEAD(&info->io_resources);
INIT_LIST_HEAD(&info->resources);
ret = probe_pci_root_info(info, device, busnum, domain);
if (ret) {
kfree(info->controller);
kfree(info);
return NULL;
}
/* insert busn resource at first */
pci_add_resource(&info->resources, &root->secondary);
/*
* See arch/x86/pci/acpi.c.
* The desired pci bus might already be scanned in a quirk. We
* should handle the case here, but it appears that IA64 hasn't
* such quirk. So we just ignore the case now.
*/
pbus = pci_create_root_bus(NULL, bus, &pci_root_ops, controller,
&info->resources);
if (!pbus) {
pci_free_resource_list(&info->resources);
__release_pci_root_info(info);
return NULL;
}
pci_set_host_bridge_release(to_pci_host_bridge(pbus->bridge),
release_pci_root_info, info);
pci_scan_child_bus(pbus);
return pbus;
}
ACPI / PCI: Set root bridge ACPI handle in advance The ACPI handles of PCI root bridges need to be known to acpi_bind_one(), so that it can create the appropriate "firmware_node" and "physical_node" files for them, but currently the way it gets to know those handles is not exactly straightforward (to put it lightly). This is how it works, roughly: 1. acpi_bus_scan() finds the handle of a PCI root bridge, creates a struct acpi_device object for it and passes that object to acpi_pci_root_add(). 2. acpi_pci_root_add() creates a struct acpi_pci_root object, populates its "device" field with its argument's address (device->handle is the ACPI handle found in step 1). 3. The struct acpi_pci_root object created in step 2 is passed to pci_acpi_scan_root() and used to get resources that are passed to pci_create_root_bus(). 4. pci_create_root_bus() creates a struct pci_host_bridge object and passes its "dev" member to device_register(). 5. platform_notify(), which for systems with ACPI is set to acpi_platform_notify(), is called. So far, so good. Now it starts to be "interesting". 6. acpi_find_bridge_device() is used to find the ACPI handle of the given device (which is the PCI root bridge) and executes acpi_pci_find_root_bridge(), among other things, for the given device object. 7. acpi_pci_find_root_bridge() uses the name (sic!) of the given device object to extract the segment and bus numbers of the PCI root bridge and passes them to acpi_get_pci_rootbridge_handle(). 8. acpi_get_pci_rootbridge_handle() browses the list of ACPI PCI root bridges and finds the one that matches the given segment and bus numbers. Its handle is then used to initialize the ACPI handle of the PCI root bridge's device object by acpi_bind_one(). However, this is *exactly* the ACPI handle we started with in step 1. Needless to say, this is quite embarassing, but it may be avoided thanks to commit f3fd0c8 (ACPI: Allow ACPI handles of devices to be initialized in advance), which makes it possible to initialize the ACPI handle of a device before passing it to device_register(). Accordingly, add a new __weak routine, pcibios_root_bridge_prepare(), defaulting to an empty implementation that can be replaced by the interested architecutres (x86 and ia64 at the moment) with functions that will set the root bridge's ACPI handle before its dev member is passed to device_register(). Make both x86 and ia64 provide such implementations of pcibios_root_bridge_prepare() and remove acpi_pci_find_root_bridge() and acpi_get_pci_rootbridge_handle() that aren't necessary any more. Included is a fix for breakage on systems with non-ACPI PCI host bridges from Bjorn Helgaas. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
2013-01-10 05:33:37 +08:00
int pcibios_root_bridge_prepare(struct pci_host_bridge *bridge)
{
/*
* We pass NULL as parent to pci_create_root_bus(), so if it is not NULL
* here, pci_create_root_bus() has been called by someone else and
* sysdata is likely to be different from what we expect. Let it go in
* that case.
*/
if (!bridge->dev.parent) {
struct pci_controller *controller = bridge->bus->sysdata;
ACPI_COMPANION_SET(&bridge->dev, controller->companion);
}
ACPI / PCI: Set root bridge ACPI handle in advance The ACPI handles of PCI root bridges need to be known to acpi_bind_one(), so that it can create the appropriate "firmware_node" and "physical_node" files for them, but currently the way it gets to know those handles is not exactly straightforward (to put it lightly). This is how it works, roughly: 1. acpi_bus_scan() finds the handle of a PCI root bridge, creates a struct acpi_device object for it and passes that object to acpi_pci_root_add(). 2. acpi_pci_root_add() creates a struct acpi_pci_root object, populates its "device" field with its argument's address (device->handle is the ACPI handle found in step 1). 3. The struct acpi_pci_root object created in step 2 is passed to pci_acpi_scan_root() and used to get resources that are passed to pci_create_root_bus(). 4. pci_create_root_bus() creates a struct pci_host_bridge object and passes its "dev" member to device_register(). 5. platform_notify(), which for systems with ACPI is set to acpi_platform_notify(), is called. So far, so good. Now it starts to be "interesting". 6. acpi_find_bridge_device() is used to find the ACPI handle of the given device (which is the PCI root bridge) and executes acpi_pci_find_root_bridge(), among other things, for the given device object. 7. acpi_pci_find_root_bridge() uses the name (sic!) of the given device object to extract the segment and bus numbers of the PCI root bridge and passes them to acpi_get_pci_rootbridge_handle(). 8. acpi_get_pci_rootbridge_handle() browses the list of ACPI PCI root bridges and finds the one that matches the given segment and bus numbers. Its handle is then used to initialize the ACPI handle of the PCI root bridge's device object by acpi_bind_one(). However, this is *exactly* the ACPI handle we started with in step 1. Needless to say, this is quite embarassing, but it may be avoided thanks to commit f3fd0c8 (ACPI: Allow ACPI handles of devices to be initialized in advance), which makes it possible to initialize the ACPI handle of a device before passing it to device_register(). Accordingly, add a new __weak routine, pcibios_root_bridge_prepare(), defaulting to an empty implementation that can be replaced by the interested architecutres (x86 and ia64 at the moment) with functions that will set the root bridge's ACPI handle before its dev member is passed to device_register(). Make both x86 and ia64 provide such implementations of pcibios_root_bridge_prepare() and remove acpi_pci_find_root_bridge() and acpi_get_pci_rootbridge_handle() that aren't necessary any more. Included is a fix for breakage on systems with non-ACPI PCI host bridges from Bjorn Helgaas. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
2013-01-10 05:33:37 +08:00
return 0;
}
void pcibios_fixup_device_resources(struct pci_dev *dev)
{
int idx;
if (!dev->bus)
return;
for (idx = 0; idx < PCI_BRIDGE_RESOURCES; idx++) {
struct resource *r = &dev->resource[idx];
if (!r->flags || r->parent || !r->start)
continue;
pci_claim_resource(dev, idx);
}
}
EXPORT_SYMBOL_GPL(pcibios_fixup_device_resources);
static void pcibios_fixup_bridge_resources(struct pci_dev *dev)
{
int idx;
if (!dev->bus)
return;
for (idx = PCI_BRIDGE_RESOURCES; idx < PCI_NUM_RESOURCES; idx++) {
struct resource *r = &dev->resource[idx];
if (!r->flags || r->parent || !r->start)
continue;
pci_claim_bridge_resource(dev, idx);
}
}
/*
* Called after each bus is probed, but before its children are examined.
*/
void pcibios_fixup_bus(struct pci_bus *b)
{
struct pci_dev *dev;
PCI: Revert "PCI: Call pci_read_bridge_bases() from core instead of arch code" Revert dff22d2054b5 ("PCI: Call pci_read_bridge_bases() from core instead of arch code"). Reading PCI bridge windows is not arch-specific in itself, but there is PCI core code that doesn't work correctly if we read them too early. For example, Hannes found this case on an ARM Freescale i.mx6 board: pci_bus 0000:00: root bus resource [mem 0x01000000-0x01efffff] pci 0000:00:00.0: PCI bridge to [bus 01-ff] pci 0000:00:00.0: BAR 8: no space for [mem size 0x01000000] (mem window) pci 0000:01:00.0: BAR 2: failed to assign [mem size 0x00200000] pci 0000:01:00.0: BAR 1: failed to assign [mem size 0x00004000] pci 0000:01:00.0: BAR 0: failed to assign [mem size 0x00000100] The 00:00.0 mem window needs to be at least 3MB: the 01:00.0 device needs 0x204100 of space, and mem windows are megabyte-aligned. Bus sizing can increase a bridge window size, but never *decrease* it (see d65245c3297a ("PCI: don't shrink bridge resources")). Prior to dff22d2054b5, ARM didn't read bridge windows at all, so the "original size" was zero, and we assigned a 3MB window. After dff22d2054b5, we read the bridge windows before sizing the bus. The firmware programmed a 16MB window (size 0x01000000) in 00:00.0, and since we never decrease the size, we kept 16MB even though we only needed 3MB. But 16MB doesn't fit in the host bridge aperture, so we failed to assign space for the window and the downstream devices. I think this is a defect in the PCI core: we shouldn't rely on the firmware to assign sensible windows. Ray reported a similar problem, also on ARM, with Broadcom iProc. Issues like this are too hard to fix right now, so revert dff22d2054b5. Reported-by: Hannes <oe5hpm@gmail.com> Reported-by: Ray Jui <rjui@broadcom.com> Link: http://lkml.kernel.org/r/CAAa04yFQEUJm7Jj1qMT57-LG7ZGtnhNDBe=PpSRa70Mj+XhW-A@mail.gmail.com Link: http://lkml.kernel.org/r/55F75BB8.4070405@broadcom.com Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Acked-by: Yinghai Lu <yinghai@kernel.org> Acked-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
2015-09-16 02:18:04 +08:00
if (b->self) {
pci_read_bridge_bases(b);
pcibios_fixup_bridge_resources(b->self);
PCI: Revert "PCI: Call pci_read_bridge_bases() from core instead of arch code" Revert dff22d2054b5 ("PCI: Call pci_read_bridge_bases() from core instead of arch code"). Reading PCI bridge windows is not arch-specific in itself, but there is PCI core code that doesn't work correctly if we read them too early. For example, Hannes found this case on an ARM Freescale i.mx6 board: pci_bus 0000:00: root bus resource [mem 0x01000000-0x01efffff] pci 0000:00:00.0: PCI bridge to [bus 01-ff] pci 0000:00:00.0: BAR 8: no space for [mem size 0x01000000] (mem window) pci 0000:01:00.0: BAR 2: failed to assign [mem size 0x00200000] pci 0000:01:00.0: BAR 1: failed to assign [mem size 0x00004000] pci 0000:01:00.0: BAR 0: failed to assign [mem size 0x00000100] The 00:00.0 mem window needs to be at least 3MB: the 01:00.0 device needs 0x204100 of space, and mem windows are megabyte-aligned. Bus sizing can increase a bridge window size, but never *decrease* it (see d65245c3297a ("PCI: don't shrink bridge resources")). Prior to dff22d2054b5, ARM didn't read bridge windows at all, so the "original size" was zero, and we assigned a 3MB window. After dff22d2054b5, we read the bridge windows before sizing the bus. The firmware programmed a 16MB window (size 0x01000000) in 00:00.0, and since we never decrease the size, we kept 16MB even though we only needed 3MB. But 16MB doesn't fit in the host bridge aperture, so we failed to assign space for the window and the downstream devices. I think this is a defect in the PCI core: we shouldn't rely on the firmware to assign sensible windows. Ray reported a similar problem, also on ARM, with Broadcom iProc. Issues like this are too hard to fix right now, so revert dff22d2054b5. Reported-by: Hannes <oe5hpm@gmail.com> Reported-by: Ray Jui <rjui@broadcom.com> Link: http://lkml.kernel.org/r/CAAa04yFQEUJm7Jj1qMT57-LG7ZGtnhNDBe=PpSRa70Mj+XhW-A@mail.gmail.com Link: http://lkml.kernel.org/r/55F75BB8.4070405@broadcom.com Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Acked-by: Yinghai Lu <yinghai@kernel.org> Acked-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
2015-09-16 02:18:04 +08:00
}
list_for_each_entry(dev, &b->devices, bus_list)
pcibios_fixup_device_resources(dev);
platform_pci_fixup_bus(b);
}
void pcibios_add_bus(struct pci_bus *bus)
{
acpi_pci_add_bus(bus);
}
void pcibios_remove_bus(struct pci_bus *bus)
{
acpi_pci_remove_bus(bus);
}
void pcibios_set_master (struct pci_dev *dev)
{
/* No special bus mastering setup handling */
}
int
pcibios_enable_device (struct pci_dev *dev, int mask)
{
int ret;
ret = pci_enable_resources(dev, mask);
if (ret < 0)
return ret;
if (!dev->msi_enabled)
return acpi_pci_irq_enable(dev);
return 0;
}
void
pcibios_disable_device (struct pci_dev *dev)
{
BUG_ON(atomic_read(&dev->enable_cnt));
if (!dev->msi_enabled)
acpi_pci_irq_disable(dev);
}
resource_size_t
pcibios_align_resource (void *data, const struct resource *res,
resource_size_t size, resource_size_t align)
{
return res->start;
}
int
pci_mmap_page_range (struct pci_dev *dev, struct vm_area_struct *vma,
enum pci_mmap_state mmap_state, int write_combine)
{
unsigned long size = vma->vm_end - vma->vm_start;
pgprot_t prot;
/*
* I/O space cannot be accessed via normal processor loads and
* stores on this platform.
*/
if (mmap_state == pci_mmap_io)
/*
* XXX we could relax this for I/O spaces for which ACPI
* indicates that the space is 1-to-1 mapped. But at the
* moment, we don't support multiple PCI address spaces and
* the legacy I/O space is not 1-to-1 mapped, so this is moot.
*/
return -EINVAL;
if (!valid_mmap_phys_addr_range(vma->vm_pgoff, size))
return -EINVAL;
prot = phys_mem_access_prot(NULL, vma->vm_pgoff, size,
vma->vm_page_prot);
/*
* If the user requested WC, the kernel uses UC or WC for this region,
* and the chipset supports WC, we can use WC. Otherwise, we have to
* use the same attribute the kernel uses.
*/
if (write_combine &&
((pgprot_val(prot) & _PAGE_MA_MASK) == _PAGE_MA_UC ||
(pgprot_val(prot) & _PAGE_MA_MASK) == _PAGE_MA_WC) &&
efi_range_is_wc(vma->vm_start, vma->vm_end - vma->vm_start))
vma->vm_page_prot = pgprot_writecombine(vma->vm_page_prot);
else
vma->vm_page_prot = prot;
if (remap_pfn_range(vma, vma->vm_start, vma->vm_pgoff,
vma->vm_end - vma->vm_start, vma->vm_page_prot))
return -EAGAIN;
return 0;
}
/**
* ia64_pci_get_legacy_mem - generic legacy mem routine
* @bus: bus to get legacy memory base address for
*
* Find the base of legacy memory for @bus. This is typically the first
* megabyte of bus address space for @bus or is simply 0 on platforms whose
* chipsets support legacy I/O and memory routing. Returns the base address
* or an error pointer if an error occurred.
*
* This is the ia64 generic version of this routine. Other platforms
* are free to override it with a machine vector.
*/
char *ia64_pci_get_legacy_mem(struct pci_bus *bus)
{
return (char *)__IA64_UNCACHED_OFFSET;
}
/**
* pci_mmap_legacy_page_range - map legacy memory space to userland
* @bus: bus whose legacy space we're mapping
* @vma: vma passed in by mmap
*
* Map legacy memory space for this device back to userspace using a machine
* vector to get the base address.
*/
int
pci_mmap_legacy_page_range(struct pci_bus *bus, struct vm_area_struct *vma,
enum pci_mmap_state mmap_state)
{
unsigned long size = vma->vm_end - vma->vm_start;
pgprot_t prot;
char *addr;
/* We only support mmap'ing of legacy memory space */
if (mmap_state != pci_mmap_mem)
return -ENOSYS;
/*
* Avoid attribute aliasing. See Documentation/ia64/aliasing.txt
* for more details.
*/
if (!valid_mmap_phys_addr_range(vma->vm_pgoff, size))
return -EINVAL;
prot = phys_mem_access_prot(NULL, vma->vm_pgoff, size,
vma->vm_page_prot);
addr = pci_get_legacy_mem(bus);
if (IS_ERR(addr))
return PTR_ERR(addr);
vma->vm_pgoff += (unsigned long)addr >> PAGE_SHIFT;
vma->vm_page_prot = prot;
if (remap_pfn_range(vma, vma->vm_start, vma->vm_pgoff,
size, vma->vm_page_prot))
return -EAGAIN;
return 0;
}
/**
* ia64_pci_legacy_read - read from legacy I/O space
* @bus: bus to read
* @port: legacy port value
* @val: caller allocated storage for returned value
* @size: number of bytes to read
*
* Simply reads @size bytes from @port and puts the result in @val.
*
* Again, this (and the write routine) are generic versions that can be
* overridden by the platform. This is necessary on platforms that don't
* support legacy I/O routing or that hard fail on legacy I/O timeouts.
*/
int ia64_pci_legacy_read(struct pci_bus *bus, u16 port, u32 *val, u8 size)
{
int ret = size;
switch (size) {
case 1:
*val = inb(port);
break;
case 2:
*val = inw(port);
break;
case 4:
*val = inl(port);
break;
default:
ret = -EINVAL;
break;
}
return ret;
}
/**
* ia64_pci_legacy_write - perform a legacy I/O write
* @bus: bus pointer
* @port: port to write
* @val: value to write
* @size: number of bytes to write from @val
*
* Simply writes @size bytes of @val to @port.
*/
int ia64_pci_legacy_write(struct pci_bus *bus, u16 port, u32 val, u8 size)
{
int ret = size;
switch (size) {
case 1:
outb(val, port);
break;
case 2:
outw(val, port);
break;
case 4:
outl(val, port);
break;
default:
ret = -EINVAL;
break;
}
return ret;
}
/**
* set_pci_cacheline_size - determine cacheline size for PCI devices
*
* We want to use the line-size of the outer-most cache. We assume
* that this line-size is the same for all CPUs.
*
* Code mostly taken from arch/ia64/kernel/palinfo.c:cache_info().
*/
static void __init set_pci_dfl_cacheline_size(void)
{
unsigned long levels, unique_caches;
long status;
pal_cache_config_info_t cci;
status = ia64_pal_cache_summary(&levels, &unique_caches);
if (status != 0) {
pr_err("%s: ia64_pal_cache_summary() failed "
"(status=%ld)\n", __func__, status);
return;
}
status = ia64_pal_cache_config_info(levels - 1,
/* cache_type (data_or_unified)= */ 2, &cci);
if (status != 0) {
pr_err("%s: ia64_pal_cache_config_info() failed "
"(status=%ld)\n", __func__, status);
return;
}
pci_dfl_cache_line_size = (1 << cci.pcci_line_size) / 4;
}
u64 ia64_dma_get_required_mask(struct device *dev)
{
u32 low_totalram = ((max_pfn - 1) << PAGE_SHIFT);
u32 high_totalram = ((max_pfn - 1) >> (32 - PAGE_SHIFT));
u64 mask;
if (!high_totalram) {
/* convert to mask just covering totalram */
low_totalram = (1 << (fls(low_totalram) - 1));
low_totalram += low_totalram - 1;
mask = low_totalram;
} else {
high_totalram = (1 << (fls(high_totalram) - 1));
high_totalram += high_totalram - 1;
mask = (((u64)high_totalram) << 32) + 0xffffffff;
}
return mask;
}
EXPORT_SYMBOL_GPL(ia64_dma_get_required_mask);
u64 dma_get_required_mask(struct device *dev)
{
return platform_dma_get_required_mask(dev);
}
EXPORT_SYMBOL_GPL(dma_get_required_mask);
static int __init pcibios_init(void)
{
set_pci_dfl_cacheline_size();
return 0;
}
subsys_initcall(pcibios_init);