linux/drivers/char/agp/amd-k7-agp.c
Dave Airlie 07613ba2f4 agp: switch AGP to use page array instead of unsigned long array
This switches AGP to use an array of pages for tracking the
pages allocated to the GART. This should enable GEM on PAE to work
a lot better as we can pass highmem pages to the PAT code and it will
do the right thing with them.

Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-19 10:21:42 +10:00

575 lines
15 KiB
C

/*
* AMD K7 AGPGART routines.
*/
#include <linux/module.h>
#include <linux/pci.h>
#include <linux/init.h>
#include <linux/agp_backend.h>
#include <linux/gfp.h>
#include <linux/page-flags.h>
#include <linux/mm.h>
#include "agp.h"
#define AMD_MMBASE 0x14
#define AMD_APSIZE 0xac
#define AMD_MODECNTL 0xb0
#define AMD_MODECNTL2 0xb2
#define AMD_GARTENABLE 0x02 /* In mmio region (16-bit register) */
#define AMD_ATTBASE 0x04 /* In mmio region (32-bit register) */
#define AMD_TLBFLUSH 0x0c /* In mmio region (32-bit register) */
#define AMD_CACHEENTRY 0x10 /* In mmio region (32-bit register) */
static struct pci_device_id agp_amdk7_pci_table[];
struct amd_page_map {
unsigned long *real;
unsigned long __iomem *remapped;
};
static struct _amd_irongate_private {
volatile u8 __iomem *registers;
struct amd_page_map **gatt_pages;
int num_tables;
} amd_irongate_private;
static int amd_create_page_map(struct amd_page_map *page_map)
{
int i;
page_map->real = (unsigned long *) __get_free_page(GFP_KERNEL);
if (page_map->real == NULL)
return -ENOMEM;
#ifndef CONFIG_X86
SetPageReserved(virt_to_page(page_map->real));
global_cache_flush();
page_map->remapped = ioremap_nocache(virt_to_gart(page_map->real),
PAGE_SIZE);
if (page_map->remapped == NULL) {
ClearPageReserved(virt_to_page(page_map->real));
free_page((unsigned long) page_map->real);
page_map->real = NULL;
return -ENOMEM;
}
global_cache_flush();
#else
set_memory_uc((unsigned long)page_map->real, 1);
page_map->remapped = page_map->real;
#endif
for (i = 0; i < PAGE_SIZE / sizeof(unsigned long); i++) {
writel(agp_bridge->scratch_page, page_map->remapped+i);
readl(page_map->remapped+i); /* PCI Posting. */
}
return 0;
}
static void amd_free_page_map(struct amd_page_map *page_map)
{
#ifndef CONFIG_X86
iounmap(page_map->remapped);
ClearPageReserved(virt_to_page(page_map->real));
#else
set_memory_wb((unsigned long)page_map->real, 1);
#endif
free_page((unsigned long) page_map->real);
}
static void amd_free_gatt_pages(void)
{
int i;
struct amd_page_map **tables;
struct amd_page_map *entry;
tables = amd_irongate_private.gatt_pages;
for (i = 0; i < amd_irongate_private.num_tables; i++) {
entry = tables[i];
if (entry != NULL) {
if (entry->real != NULL)
amd_free_page_map(entry);
kfree(entry);
}
}
kfree(tables);
amd_irongate_private.gatt_pages = NULL;
}
static int amd_create_gatt_pages(int nr_tables)
{
struct amd_page_map **tables;
struct amd_page_map *entry;
int retval = 0;
int i;
tables = kzalloc((nr_tables + 1) * sizeof(struct amd_page_map *),GFP_KERNEL);
if (tables == NULL)
return -ENOMEM;
for (i = 0; i < nr_tables; i++) {
entry = kzalloc(sizeof(struct amd_page_map), GFP_KERNEL);
tables[i] = entry;
if (entry == NULL) {
retval = -ENOMEM;
break;
}
retval = amd_create_page_map(entry);
if (retval != 0)
break;
}
amd_irongate_private.num_tables = i;
amd_irongate_private.gatt_pages = tables;
if (retval != 0)
amd_free_gatt_pages();
return retval;
}
/* Since we don't need contiguous memory we just try
* to get the gatt table once
*/
#define GET_PAGE_DIR_OFF(addr) (addr >> 22)
#define GET_PAGE_DIR_IDX(addr) (GET_PAGE_DIR_OFF(addr) - \
GET_PAGE_DIR_OFF(agp_bridge->gart_bus_addr))
#define GET_GATT_OFF(addr) ((addr & 0x003ff000) >> 12)
#define GET_GATT(addr) (amd_irongate_private.gatt_pages[\
GET_PAGE_DIR_IDX(addr)]->remapped)
static int amd_create_gatt_table(struct agp_bridge_data *bridge)
{
struct aper_size_info_lvl2 *value;
struct amd_page_map page_dir;
unsigned long addr;
int retval;
u32 temp;
int i;
value = A_SIZE_LVL2(agp_bridge->current_size);
retval = amd_create_page_map(&page_dir);
if (retval != 0)
return retval;
retval = amd_create_gatt_pages(value->num_entries / 1024);
if (retval != 0) {
amd_free_page_map(&page_dir);
return retval;
}
agp_bridge->gatt_table_real = (u32 *)page_dir.real;
agp_bridge->gatt_table = (u32 __iomem *)page_dir.remapped;
agp_bridge->gatt_bus_addr = virt_to_gart(page_dir.real);
/* Get the address for the gart region.
* This is a bus address even on the alpha, b/c its
* used to program the agp master not the cpu
*/
pci_read_config_dword(agp_bridge->dev, AGP_APBASE, &temp);
addr = (temp & PCI_BASE_ADDRESS_MEM_MASK);
agp_bridge->gart_bus_addr = addr;
/* Calculate the agp offset */
for (i = 0; i < value->num_entries / 1024; i++, addr += 0x00400000) {
writel(virt_to_gart(amd_irongate_private.gatt_pages[i]->real) | 1,
page_dir.remapped+GET_PAGE_DIR_OFF(addr));
readl(page_dir.remapped+GET_PAGE_DIR_OFF(addr)); /* PCI Posting. */
}
return 0;
}
static int amd_free_gatt_table(struct agp_bridge_data *bridge)
{
struct amd_page_map page_dir;
page_dir.real = (unsigned long *)agp_bridge->gatt_table_real;
page_dir.remapped = (unsigned long __iomem *)agp_bridge->gatt_table;
amd_free_gatt_pages();
amd_free_page_map(&page_dir);
return 0;
}
static int amd_irongate_fetch_size(void)
{
int i;
u32 temp;
struct aper_size_info_lvl2 *values;
pci_read_config_dword(agp_bridge->dev, AMD_APSIZE, &temp);
temp = (temp & 0x0000000e);
values = A_SIZE_LVL2(agp_bridge->driver->aperture_sizes);
for (i = 0; i < agp_bridge->driver->num_aperture_sizes; i++) {
if (temp == values[i].size_value) {
agp_bridge->previous_size =
agp_bridge->current_size = (void *) (values + i);
agp_bridge->aperture_size_idx = i;
return values[i].size;
}
}
return 0;
}
static int amd_irongate_configure(void)
{
struct aper_size_info_lvl2 *current_size;
u32 temp;
u16 enable_reg;
current_size = A_SIZE_LVL2(agp_bridge->current_size);
if (!amd_irongate_private.registers) {
/* Get the memory mapped registers */
pci_read_config_dword(agp_bridge->dev, AMD_MMBASE, &temp);
temp = (temp & PCI_BASE_ADDRESS_MEM_MASK);
amd_irongate_private.registers = (volatile u8 __iomem *) ioremap(temp, 4096);
if (!amd_irongate_private.registers)
return -ENOMEM;
}
/* Write out the address of the gatt table */
writel(agp_bridge->gatt_bus_addr, amd_irongate_private.registers+AMD_ATTBASE);
readl(amd_irongate_private.registers+AMD_ATTBASE); /* PCI Posting. */
/* Write the Sync register */
pci_write_config_byte(agp_bridge->dev, AMD_MODECNTL, 0x80);
/* Set indexing mode */
pci_write_config_byte(agp_bridge->dev, AMD_MODECNTL2, 0x00);
/* Write the enable register */
enable_reg = readw(amd_irongate_private.registers+AMD_GARTENABLE);
enable_reg = (enable_reg | 0x0004);
writew(enable_reg, amd_irongate_private.registers+AMD_GARTENABLE);
readw(amd_irongate_private.registers+AMD_GARTENABLE); /* PCI Posting. */
/* Write out the size register */
pci_read_config_dword(agp_bridge->dev, AMD_APSIZE, &temp);
temp = (((temp & ~(0x0000000e)) | current_size->size_value) | 1);
pci_write_config_dword(agp_bridge->dev, AMD_APSIZE, temp);
/* Flush the tlb */
writel(1, amd_irongate_private.registers+AMD_TLBFLUSH);
readl(amd_irongate_private.registers+AMD_TLBFLUSH); /* PCI Posting.*/
return 0;
}
static void amd_irongate_cleanup(void)
{
struct aper_size_info_lvl2 *previous_size;
u32 temp;
u16 enable_reg;
previous_size = A_SIZE_LVL2(agp_bridge->previous_size);
enable_reg = readw(amd_irongate_private.registers+AMD_GARTENABLE);
enable_reg = (enable_reg & ~(0x0004));
writew(enable_reg, amd_irongate_private.registers+AMD_GARTENABLE);
readw(amd_irongate_private.registers+AMD_GARTENABLE); /* PCI Posting. */
/* Write back the previous size and disable gart translation */
pci_read_config_dword(agp_bridge->dev, AMD_APSIZE, &temp);
temp = ((temp & ~(0x0000000f)) | previous_size->size_value);
pci_write_config_dword(agp_bridge->dev, AMD_APSIZE, temp);
iounmap((void __iomem *) amd_irongate_private.registers);
}
/*
* This routine could be implemented by taking the addresses
* written to the GATT, and flushing them individually. However
* currently it just flushes the whole table. Which is probably
* more efficent, since agp_memory blocks can be a large number of
* entries.
*/
static void amd_irongate_tlbflush(struct agp_memory *temp)
{
writel(1, amd_irongate_private.registers+AMD_TLBFLUSH);
readl(amd_irongate_private.registers+AMD_TLBFLUSH); /* PCI Posting. */
}
static int amd_insert_memory(struct agp_memory *mem, off_t pg_start, int type)
{
int i, j, num_entries;
unsigned long __iomem *cur_gatt;
unsigned long addr;
num_entries = A_SIZE_LVL2(agp_bridge->current_size)->num_entries;
if (type != 0 || mem->type != 0)
return -EINVAL;
if ((pg_start + mem->page_count) > num_entries)
return -EINVAL;
j = pg_start;
while (j < (pg_start + mem->page_count)) {
addr = (j * PAGE_SIZE) + agp_bridge->gart_bus_addr;
cur_gatt = GET_GATT(addr);
if (!PGE_EMPTY(agp_bridge, readl(cur_gatt+GET_GATT_OFF(addr))))
return -EBUSY;
j++;
}
if (!mem->is_flushed) {
global_cache_flush();
mem->is_flushed = true;
}
for (i = 0, j = pg_start; i < mem->page_count; i++, j++) {
addr = (j * PAGE_SIZE) + agp_bridge->gart_bus_addr;
cur_gatt = GET_GATT(addr);
writel(agp_generic_mask_memory(agp_bridge,
mem->pages[i], mem->type), cur_gatt+GET_GATT_OFF(addr));
readl(cur_gatt+GET_GATT_OFF(addr)); /* PCI Posting. */
}
amd_irongate_tlbflush(mem);
return 0;
}
static int amd_remove_memory(struct agp_memory *mem, off_t pg_start, int type)
{
int i;
unsigned long __iomem *cur_gatt;
unsigned long addr;
if (type != 0 || mem->type != 0)
return -EINVAL;
for (i = pg_start; i < (mem->page_count + pg_start); i++) {
addr = (i * PAGE_SIZE) + agp_bridge->gart_bus_addr;
cur_gatt = GET_GATT(addr);
writel(agp_bridge->scratch_page, cur_gatt+GET_GATT_OFF(addr));
readl(cur_gatt+GET_GATT_OFF(addr)); /* PCI Posting. */
}
amd_irongate_tlbflush(mem);
return 0;
}
static const struct aper_size_info_lvl2 amd_irongate_sizes[7] =
{
{2048, 524288, 0x0000000c},
{1024, 262144, 0x0000000a},
{512, 131072, 0x00000008},
{256, 65536, 0x00000006},
{128, 32768, 0x00000004},
{64, 16384, 0x00000002},
{32, 8192, 0x00000000}
};
static const struct gatt_mask amd_irongate_masks[] =
{
{.mask = 1, .type = 0}
};
static const struct agp_bridge_driver amd_irongate_driver = {
.owner = THIS_MODULE,
.aperture_sizes = amd_irongate_sizes,
.size_type = LVL2_APER_SIZE,
.num_aperture_sizes = 7,
.configure = amd_irongate_configure,
.fetch_size = amd_irongate_fetch_size,
.cleanup = amd_irongate_cleanup,
.tlb_flush = amd_irongate_tlbflush,
.mask_memory = agp_generic_mask_memory,
.masks = amd_irongate_masks,
.agp_enable = agp_generic_enable,
.cache_flush = global_cache_flush,
.create_gatt_table = amd_create_gatt_table,
.free_gatt_table = amd_free_gatt_table,
.insert_memory = amd_insert_memory,
.remove_memory = amd_remove_memory,
.alloc_by_type = agp_generic_alloc_by_type,
.free_by_type = agp_generic_free_by_type,
.agp_alloc_page = agp_generic_alloc_page,
.agp_alloc_pages = agp_generic_alloc_pages,
.agp_destroy_page = agp_generic_destroy_page,
.agp_destroy_pages = agp_generic_destroy_pages,
.agp_type_to_mask_type = agp_generic_type_to_mask_type,
};
static struct agp_device_ids amd_agp_device_ids[] __devinitdata =
{
{
.device_id = PCI_DEVICE_ID_AMD_FE_GATE_7006,
.chipset_name = "Irongate",
},
{
.device_id = PCI_DEVICE_ID_AMD_FE_GATE_700E,
.chipset_name = "761",
},
{
.device_id = PCI_DEVICE_ID_AMD_FE_GATE_700C,
.chipset_name = "760MP",
},
{ }, /* dummy final entry, always present */
};
static int __devinit agp_amdk7_probe(struct pci_dev *pdev,
const struct pci_device_id *ent)
{
struct agp_bridge_data *bridge;
u8 cap_ptr;
int j;
cap_ptr = pci_find_capability(pdev, PCI_CAP_ID_AGP);
if (!cap_ptr)
return -ENODEV;
j = ent - agp_amdk7_pci_table;
dev_info(&pdev->dev, "AMD %s chipset\n",
amd_agp_device_ids[j].chipset_name);
bridge = agp_alloc_bridge();
if (!bridge)
return -ENOMEM;
bridge->driver = &amd_irongate_driver;
bridge->dev_private_data = &amd_irongate_private,
bridge->dev = pdev;
bridge->capndx = cap_ptr;
/* 751 Errata (22564_B-1.PDF)
erratum 20: strobe glitch with Nvidia NV10 GeForce cards.
system controller may experience noise due to strong drive strengths
*/
if (agp_bridge->dev->device == PCI_DEVICE_ID_AMD_FE_GATE_7006) {
struct pci_dev *gfxcard=NULL;
cap_ptr = 0;
while (!cap_ptr) {
gfxcard = pci_get_class(PCI_CLASS_DISPLAY_VGA<<8, gfxcard);
if (!gfxcard) {
dev_info(&pdev->dev, "no AGP VGA controller\n");
return -ENODEV;
}
cap_ptr = pci_find_capability(gfxcard, PCI_CAP_ID_AGP);
}
/* With so many variants of NVidia cards, it's simpler just
to blacklist them all, and then whitelist them as needed
(if necessary at all). */
if (gfxcard->vendor == PCI_VENDOR_ID_NVIDIA) {
agp_bridge->flags |= AGP_ERRATA_1X;
dev_info(&pdev->dev, "AMD 751 chipset with NVidia GeForce; forcing 1X due to errata\n");
}
pci_dev_put(gfxcard);
}
/* 761 Errata (23613_F.pdf)
* Revisions B0/B1 were a disaster.
* erratum 44: SYSCLK/AGPCLK skew causes 2X failures -- Force mode to 1X
* erratum 45: Timing problem prevents fast writes -- Disable fast write.
* erratum 46: Setup violation on AGP SBA pins - Disable side band addressing.
* With this lot disabled, we should prevent lockups. */
if (agp_bridge->dev->device == PCI_DEVICE_ID_AMD_FE_GATE_700E) {
if (pdev->revision == 0x10 || pdev->revision == 0x11) {
agp_bridge->flags = AGP_ERRATA_FASTWRITES;
agp_bridge->flags |= AGP_ERRATA_SBA;
agp_bridge->flags |= AGP_ERRATA_1X;
dev_info(&pdev->dev, "AMD 761 chipset with errata; disabling AGP fast writes & SBA and forcing to 1X\n");
}
}
/* Fill in the mode register */
pci_read_config_dword(pdev,
bridge->capndx+PCI_AGP_STATUS,
&bridge->mode);
pci_set_drvdata(pdev, bridge);
return agp_add_bridge(bridge);
}
static void __devexit agp_amdk7_remove(struct pci_dev *pdev)
{
struct agp_bridge_data *bridge = pci_get_drvdata(pdev);
agp_remove_bridge(bridge);
agp_put_bridge(bridge);
}
#ifdef CONFIG_PM
static int agp_amdk7_suspend(struct pci_dev *pdev, pm_message_t state)
{
pci_save_state(pdev);
pci_set_power_state(pdev, pci_choose_state(pdev, state));
return 0;
}
static int agp_amdk7_resume(struct pci_dev *pdev)
{
pci_set_power_state(pdev, PCI_D0);
pci_restore_state(pdev);
return amd_irongate_driver.configure();
}
#endif /* CONFIG_PM */
/* must be the same order as name table above */
static struct pci_device_id agp_amdk7_pci_table[] = {
{
.class = (PCI_CLASS_BRIDGE_HOST << 8),
.class_mask = ~0,
.vendor = PCI_VENDOR_ID_AMD,
.device = PCI_DEVICE_ID_AMD_FE_GATE_7006,
.subvendor = PCI_ANY_ID,
.subdevice = PCI_ANY_ID,
},
{
.class = (PCI_CLASS_BRIDGE_HOST << 8),
.class_mask = ~0,
.vendor = PCI_VENDOR_ID_AMD,
.device = PCI_DEVICE_ID_AMD_FE_GATE_700E,
.subvendor = PCI_ANY_ID,
.subdevice = PCI_ANY_ID,
},
{
.class = (PCI_CLASS_BRIDGE_HOST << 8),
.class_mask = ~0,
.vendor = PCI_VENDOR_ID_AMD,
.device = PCI_DEVICE_ID_AMD_FE_GATE_700C,
.subvendor = PCI_ANY_ID,
.subdevice = PCI_ANY_ID,
},
{ }
};
MODULE_DEVICE_TABLE(pci, agp_amdk7_pci_table);
static struct pci_driver agp_amdk7_pci_driver = {
.name = "agpgart-amdk7",
.id_table = agp_amdk7_pci_table,
.probe = agp_amdk7_probe,
.remove = agp_amdk7_remove,
#ifdef CONFIG_PM
.suspend = agp_amdk7_suspend,
.resume = agp_amdk7_resume,
#endif
};
static int __init agp_amdk7_init(void)
{
if (agp_off)
return -EINVAL;
return pci_register_driver(&agp_amdk7_pci_driver);
}
static void __exit agp_amdk7_cleanup(void)
{
pci_unregister_driver(&agp_amdk7_pci_driver);
}
module_init(agp_amdk7_init);
module_exit(agp_amdk7_cleanup);
MODULE_LICENSE("GPL and additional rights");