[ACPI] Support for PCI MMCONFIG for PCI Express (Matt Wilcox)

arch/i386/Kconfig

@@ -1030,12 +1030,16 @@ config PCI_GOBIOS
 	  PCI-based systems don't have any BIOS at all. Linux can also try to
 	  detect the PCI hardware directly without using the BIOS.
 
-	  With this option, you can specify how Linux should detect the PCI
-	  devices. If you choose "BIOS", the BIOS will be used, if you choose
-	  "Direct", the BIOS won't be used, and if you choose "Any", the
-	  kernel will try the direct access method and falls back to the BIOS
-	  if that doesn't work. If unsure, go with the default, which is
-	  "Any".
+	  With this option, you can specify how Linux should detect the
+	  PCI devices. If you choose "BIOS", the BIOS will be used,
+	  if you choose "Direct", the BIOS won't be used, and if you
+	  choose "MMConfig", then PCI Express MMCONFIG will be used.
+	  If you choose "Any", the kernel will try MMCONFIG, then the
+	  direct access method and falls back to the BIOS if that doesn't
+	  work. If unsure, go with the default, which is "Any".
+
+config PCI_GOMMCONFIG
+	bool "MMConfig"
 
 config PCI_GODIRECT
 	bool "Direct"
@@ -1055,6 +1059,12 @@ config PCI_DIRECT
  	depends on PCI && ((PCI_GODIRECT || PCI_GOANY) || X86_VISWS)
 	default y
 
+config PCI_MMCONFIG
+	bool
+	depends on PCI && (PCI_GOMMCONFIG || PCI_GOANY)
+	select ACPI_BOOT
+	default y
+
 config PCI_USE_VECTOR
 	bool "Vector-based interrupt indexing"
 	depends on X86_LOCAL_APIC && X86_IO_APIC

arch/i386/kernel/acpi/boot.c

@@ -104,6 +104,31 @@ char *__acpi_map_table(unsigned long phys, unsigned long size)
 }
 
 
+#ifdef CONFIG_PCI_MMCONFIG
+static int __init acpi_parse_mcfg(unsigned long phys_addr, unsigned long size)
+{
+	struct acpi_table_mcfg *mcfg;
+
+	if (!phys_addr || !size)
+		return -EINVAL;
+
+	mcfg = (struct acpi_table_mcfg *) __acpi_map_table(phys_addr, size);
+	if (!mcfg) {
+		printk(KERN_WARNING PREFIX "Unable to map MCFG\n");
+		return -ENODEV;
+	}
+
+	if (mcfg->base_reserved) {
+		printk(KERN_ERR PREFIX "MMCONFIG not in low 4GB of memory\n");
+		return -ENODEV;
+	}
+
+	pci_mmcfg_base_addr = mcfg->base_address;
+
+	return 0;
+}
+#endif /* CONFIG_PCI_MMCONFIG */
+
 #ifdef CONFIG_X86_LOCAL_APIC
 static int __init
 acpi_parse_madt (
@@ -598,6 +623,12 @@ acpi_boot_init (void)
 	(void) acpi_table_parse(ACPI_HPET, acpi_parse_hpet);
 #endif
 
+#ifdef CONFIG_PCI_MMCONFIG
+	error = acpi_table_parse(ACPI_MCFG, acpi_parse_mcfg);
+	if (error)
+		printk(KERN_ERR PREFIX "Error %d parsing MCFG\n", error);
+#endif
+
 	return 0;
 }
 

arch/i386/pci/Makefile

 obj-y				:= i386.o
 
 obj-$(CONFIG_PCI_BIOS)		+= pcbios.o
+obj-$(CONFIG_PCI_MMCONFIG)	+= mmconfig.o
 obj-$(CONFIG_PCI_DIRECT)	+= direct.o
 
 pci-y				:= fixup.o

arch/i386/pci/common.c

@@ -20,7 +20,8 @@
 extern  void pcibios_sort(void);
 #endif
 
-unsigned int pci_probe = PCI_PROBE_BIOS | PCI_PROBE_CONF1 | PCI_PROBE_CONF2;
+unsigned int pci_probe = PCI_PROBE_BIOS | PCI_PROBE_CONF1 | PCI_PROBE_CONF2 |
+				PCI_PROBE_MMCONF;
 
 int pcibios_last_bus = -1;
 struct pci_bus *pci_root_bus = NULL;
@@ -197,6 +198,12 @@ char * __devinit  pcibios_setup(char *str)
 		pci_probe = PCI_PROBE_CONF2 | PCI_NO_CHECKS;
 		return NULL;
 	}
+#endif
+#ifdef CONFIG_PCI_MMCONFIG
+	else if (!strcmp(str, "nommconf")) {
+		pci_probe &= ~PCI_PROBE_MMCONF;
+		return NULL;
+	}
 #endif
 	else if (!strcmp(str, "noacpi")) {
 		acpi_noirq_set();

arch/i386/pci/mmconfig.c

+/*
+ * mmconfig.c - Low-level direct PCI config space access via MMCONFIG
+ */
+
+#include <linux/pci.h>
+#include <linux/init.h>
+#include "pci.h"
+
+/* The physical address of the MMCONFIG aperture.  Set from ACPI tables. */
+u32 pci_mmcfg_base_addr;
+
+#define mmcfg_virt_addr (fix_to_virt(FIX_PCIE_MCFG))
+
+/* The base address of the last MMCONFIG device accessed */
+static u32 mmcfg_last_accessed_device;
+
+/*
+ * Functions for accessing PCI configuration space with MMCONFIG accesses
+ */
+
+static inline void pci_exp_set_dev_base(int bus, int devfn)
+{
+	u32 dev_base = pci_mmcfg_base_addr | (bus << 20) | (devfn << 12);
+	if (dev_base != mmcfg_last_accessed_device) {
+		mmcfg_last_accessed_device = dev_base;
+		set_fixmap(FIX_PCIE_MCFG, dev_base);
+	}
+}
+
+static int pci_mmcfg_read(int seg, int bus, int devfn, int reg, int len, u32 *value)
+{
+	unsigned long flags;
+
+	if (!value || (bus > 255) || (devfn > 255) || (reg > 4095))
+		return -EINVAL;
+
+	spin_lock_irqsave(&pci_config_lock, flags);
+
+	pci_exp_set_dev_base(bus, devfn);
+
+	switch (len) {
+	case 1:
+		*value = readb(mmcfg_virt_addr + reg);
+		break;
+	case 2:
+		*value = readw(mmcfg_virt_addr + reg);
+		break;
+	case 4:
+		*value = readl(mmcfg_virt_addr + reg);
+		break;
+	}
+
+	spin_unlock_irqrestore(&pci_config_lock, flags);
+
+	return 0;
+}
+
+static int pci_mmcfg_write(int seg, int bus, int devfn, int reg, int len, u32 value)
+{
+	unsigned long flags;
+
+	if ((bus > 255) || (devfn > 255) || (reg > 4095)) 
+		return -EINVAL;
+
+	spin_lock_irqsave(&pci_config_lock, flags);
+
+	pci_exp_set_dev_base(bus, devfn);
+
+	switch (len) {
+	case 1:
+		writeb(value, mmcfg_virt_addr + reg);
+		break;
+	case 2:
+		writew(value, mmcfg_virt_addr + reg);
+		break;
+	case 4:
+		writel(value, mmcfg_virt_addr + reg);
+		break;
+	}
+
+	/* Dummy read to flush PCI write */
+	readl(mmcfg_virt_addr);
+
+	spin_unlock_irqrestore(&pci_config_lock, flags);
+
+	return 0;
+}
+
+static struct pci_raw_ops pci_mmcfg = {
+	.read =		pci_mmcfg_read,
+	.write =	pci_mmcfg_write,
+};
+
+static int __init pci_mmcfg_init(void)
+{
+	if ((pci_probe & PCI_PROBE_MMCONF) == 0)
+		goto out;
+	if (!pci_mmcfg_base_addr)
+		goto out;
+
+	printk(KERN_INFO "PCI: Using MMCONFIG\n");
+	raw_pci_ops = &pci_mmcfg;
+	pci_probe = (pci_probe & ~PCI_PROBE_MASK) | PCI_PROBE_MMCONF;
+
+ out:
+	return 0;
+}
+
+arch_initcall(pci_mmcfg_init);

arch/i386/pci/pci.h

@@ -15,6 +15,9 @@
 #define PCI_PROBE_BIOS		0x0001
 #define PCI_PROBE_CONF1		0x0002
 #define PCI_PROBE_CONF2		0x0004
+#define PCI_PROBE_MMCONF	0x0008
+#define PCI_PROBE_MASK		0x000f
+
 #define PCI_NO_SORT		0x0100
 #define PCI_BIOS_SORT		0x0200
 #define PCI_NO_CHECKS		0x0400

drivers/acpi/tables.c

@@ -58,6 +58,7 @@ static char *acpi_table_signatures[ACPI_TABLE_COUNT] = {
 	[ACPI_SSDT]		= "SSDT",
 	[ACPI_SPMI]		= "SPMI",
 	[ACPI_HPET]		= "HPET",
+	[ACPI_MCFG]		= "MCFG",
 };
 
 static char *mps_inti_flags_polarity[] = { "dfl", "high", "res", "low" };

include/asm-i386/fixmap.h

@@ -70,6 +70,9 @@ enum fixed_addresses {
 #ifdef CONFIG_ACPI_BOOT
 	FIX_ACPI_BEGIN,
 	FIX_ACPI_END = FIX_ACPI_BEGIN + FIX_ACPI_PAGES - 1,
+#endif
+#ifdef CONFIG_PCI_MMCONFIG
+	FIX_PCIE_MCFG,
 #endif
 	__end_of_permanent_fixed_addresses,
 	/* temporary boot-time mappings, used before ioremap() is functional */

include/linux/acpi.h

@@ -317,6 +317,15 @@ struct acpi_table_ecdt {
 	char				ec_id[0];
 } __attribute__ ((packed));
 
+/* PCI MMCONFIG */
+
+struct acpi_table_mcfg {
+	struct acpi_table_header	header;
+	u8				reserved[8];
+	u32				base_address;
+	u32				base_reserved;
+} __attribute__ ((packed));
+
 /* Table Handlers */
 
 enum acpi_table_id {
@@ -338,6 +347,7 @@ enum acpi_table_id {
 	ACPI_SSDT,
 	ACPI_SPMI,
 	ACPI_HPET,
+	ACPI_MCFG,
 	ACPI_TABLE_COUNT
 };
 
@@ -369,6 +379,8 @@ void acpi_numa_arch_fixup(void);
 
 extern int acpi_mp_config;
 
+extern u32 pci_mmcfg_base_addr;
+
 #else	/*!CONFIG_ACPI_BOOT*/
 
 #define acpi_mp_config	0