+#include <asm/mach/arch.h>
+#include <asm/mach/map.h>
+
+#include "mm.h"
+
+DEFINE_PER_CPU(struct mmu_gather, mmu_gathers);
+
+extern void _stext, _etext, __data_start, _end;
+extern pgd_t swapper_pg_dir[PTRS_PER_PGD];
+
+/*
+ * empty_zero_page is a special page that is used for
+ * zero-initialized data and COW.
+ */
+struct page *empty_zero_page;
+
+/*
+ * The pmd table for the upper-most set of pages.
+ */
+pmd_t *top_pmd;
+
+#define CPOLICY_UNCACHED 0
+#define CPOLICY_BUFFERED 1
+#define CPOLICY_WRITETHROUGH 2
+#define CPOLICY_WRITEBACK 3
+#define CPOLICY_WRITEALLOC 4
+
+static unsigned int cachepolicy __initdata = CPOLICY_WRITEBACK;
+static unsigned int ecc_mask __initdata = 0;
+pgprot_t pgprot_kernel;
+
+EXPORT_SYMBOL(pgprot_kernel);
+
+struct cachepolicy {
+ const char policy[16];
+ unsigned int cr_mask;
+ unsigned int pmd;
+ unsigned int pte;
+};
+
+static struct cachepolicy cache_policies[] __initdata = {
+ {
+ .policy = "uncached",
+ .cr_mask = CR_W|CR_C,
+ .pmd = PMD_SECT_UNCACHED,
+ .pte = 0,
+ }, {
+ .policy = "buffered",
+ .cr_mask = CR_C,
+ .pmd = PMD_SECT_BUFFERED,
+ .pte = PTE_BUFFERABLE,
+ }, {
+ .policy = "writethrough",
+ .cr_mask = 0,
+ .pmd = PMD_SECT_WT,
+ .pte = PTE_CACHEABLE,
+ }, {
+ .policy = "writeback",
+ .cr_mask = 0,
+ .pmd = PMD_SECT_WB,
+ .pte = PTE_BUFFERABLE|PTE_CACHEABLE,
+ }, {
+ .policy = "writealloc",
+ .cr_mask = 0,
+ .pmd = PMD_SECT_WBWA,
+ .pte = PTE_BUFFERABLE|PTE_CACHEABLE,
+ }
+};
+
+/*
+ * These are useful for identifing cache coherency
+ * problems by allowing the cache or the cache and
+ * writebuffer to be turned off. (Note: the write
+ * buffer should not be on and the cache off).
+ */
+static void __init early_cachepolicy(char **p)
+{
+ int i;
+
+ for (i = 0; i < ARRAY_SIZE(cache_policies); i++) {
+ int len = strlen(cache_policies[i].policy);
+
+ if (memcmp(*p, cache_policies[i].policy, len) == 0) {
+ cachepolicy = i;
+ cr_alignment &= ~cache_policies[i].cr_mask;
+ cr_no_alignment &= ~cache_policies[i].cr_mask;
+ *p += len;
+ break;
+ }
+ }
+ if (i == ARRAY_SIZE(cache_policies))
+ printk(KERN_ERR "ERROR: unknown or unsupported cache policy\n");
+ flush_cache_all();
+ set_cr(cr_alignment);
+}
+__early_param("cachepolicy=", early_cachepolicy);
+
+static void __init early_nocache(char **__unused)
+{
+ char *p = "buffered";
+ printk(KERN_WARNING "nocache is deprecated; use cachepolicy=%s\n", p);
+ early_cachepolicy(&p);
+}
+__early_param("nocache", early_nocache);
+
+static void __init early_nowrite(char **__unused)
+{
+ char *p = "uncached";
+ printk(KERN_WARNING "nowb is deprecated; use cachepolicy=%s\n", p);
+ early_cachepolicy(&p);
+}
+__early_param("nowb", early_nowrite);
+
+static void __init early_ecc(char **p)
+{
+ if (memcmp(*p, "on", 2) == 0) {
+ ecc_mask = PMD_PROTECTION;
+ *p += 2;
+ } else if (memcmp(*p, "off", 3) == 0) {
+ ecc_mask = 0;
+ *p += 3;
+ }
+}
+__early_param("ecc=", early_ecc);
+
+static int __init noalign_setup(char *__unused)
+{
+ cr_alignment &= ~CR_A;
+ cr_no_alignment &= ~CR_A;
+ set_cr(cr_alignment);
+ return 1;
+}
+__setup("noalign", noalign_setup);
+
+#ifndef CONFIG_SMP
+void adjust_cr(unsigned long mask, unsigned long set)
+{
+ unsigned long flags;
+
+ mask &= ~CR_A;
+
+ set &= mask;
+
+ local_irq_save(flags);
+
+ cr_no_alignment = (cr_no_alignment & ~mask) | set;
+ cr_alignment = (cr_alignment & ~mask) | set;
+
+ set_cr((get_cr() & ~mask) | set);
+
+ local_irq_restore(flags);
+}
+#endif
+
+struct mem_types {
+ unsigned int prot_pte;
+ unsigned int prot_l1;
+ unsigned int prot_sect;
+ unsigned int domain;
+};
+
+static struct mem_types mem_types[] __initdata = {
+ [MT_DEVICE] = {
+ .prot_pte = L_PTE_PRESENT | L_PTE_YOUNG | L_PTE_DIRTY |
+ L_PTE_WRITE,
+ .prot_l1 = PMD_TYPE_TABLE,
+ .prot_sect = PMD_TYPE_SECT | PMD_BIT4 | PMD_SECT_UNCACHED |
+ PMD_SECT_AP_WRITE,
+ .domain = DOMAIN_IO,
+ },
+ [MT_CACHECLEAN] = {
+ .prot_sect = PMD_TYPE_SECT | PMD_BIT4,
+ .domain = DOMAIN_KERNEL,
+ },
+ [MT_MINICLEAN] = {
+ .prot_sect = PMD_TYPE_SECT | PMD_BIT4 | PMD_SECT_MINICACHE,
+ .domain = DOMAIN_KERNEL,
+ },
+ [MT_LOW_VECTORS] = {
+ .prot_pte = L_PTE_PRESENT | L_PTE_YOUNG | L_PTE_DIRTY |
+ L_PTE_EXEC,
+ .prot_l1 = PMD_TYPE_TABLE,
+ .domain = DOMAIN_USER,
+ },
+ [MT_HIGH_VECTORS] = {
+ .prot_pte = L_PTE_PRESENT | L_PTE_YOUNG | L_PTE_DIRTY |
+ L_PTE_USER | L_PTE_EXEC,
+ .prot_l1 = PMD_TYPE_TABLE,
+ .domain = DOMAIN_USER,
+ },
+ [MT_MEMORY] = {
+ .prot_sect = PMD_TYPE_SECT | PMD_BIT4 | PMD_SECT_AP_WRITE,
+ .domain = DOMAIN_KERNEL,
+ },
+ [MT_ROM] = {
+ .prot_sect = PMD_TYPE_SECT | PMD_BIT4,
+ .domain = DOMAIN_KERNEL,
+ },
+ [MT_IXP2000_DEVICE] = { /* IXP2400 requires XCB=101 for on-chip I/O */
+ .prot_pte = L_PTE_PRESENT | L_PTE_YOUNG | L_PTE_DIRTY |
+ L_PTE_WRITE,
+ .prot_l1 = PMD_TYPE_TABLE,
+ .prot_sect = PMD_TYPE_SECT | PMD_BIT4 | PMD_SECT_UNCACHED |
+ PMD_SECT_AP_WRITE | PMD_SECT_BUFFERABLE |
+ PMD_SECT_TEX(1),
+ .domain = DOMAIN_IO,
+ },
+ [MT_NONSHARED_DEVICE] = {
+ .prot_l1 = PMD_TYPE_TABLE,
+ .prot_sect = PMD_TYPE_SECT | PMD_BIT4 | PMD_SECT_NONSHARED_DEV |
+ PMD_SECT_AP_WRITE,
+ .domain = DOMAIN_IO,
+ }
+};
+
+/*
+ * Adjust the PMD section entries according to the CPU in use.
+ */
+static void __init build_mem_type_table(void)
+{
+ struct cachepolicy *cp;
+ unsigned int cr = get_cr();
+ unsigned int user_pgprot, kern_pgprot;
+ int cpu_arch = cpu_architecture();
+ int i;
+
+#if defined(CONFIG_CPU_DCACHE_DISABLE)
+ if (cachepolicy > CPOLICY_BUFFERED)
+ cachepolicy = CPOLICY_BUFFERED;
+#elif defined(CONFIG_CPU_DCACHE_WRITETHROUGH)
+ if (cachepolicy > CPOLICY_WRITETHROUGH)
+ cachepolicy = CPOLICY_WRITETHROUGH;
+#endif
+ if (cpu_arch < CPU_ARCH_ARMv5) {
+ if (cachepolicy >= CPOLICY_WRITEALLOC)
+ cachepolicy = CPOLICY_WRITEBACK;
+ ecc_mask = 0;
+ }
+
+ /*
+ * Xscale must not have PMD bit 4 set for section mappings.
+ */
+ if (cpu_is_xscale())
+ for (i = 0; i < ARRAY_SIZE(mem_types); i++)
+ mem_types[i].prot_sect &= ~PMD_BIT4;
+
+ /*
+ * ARMv5 and lower, excluding Xscale, bit 4 must be set for
+ * page tables.
+ */
+ if (cpu_arch < CPU_ARCH_ARMv6 && !cpu_is_xscale())
+ for (i = 0; i < ARRAY_SIZE(mem_types); i++)
+ if (mem_types[i].prot_l1)
+ mem_types[i].prot_l1 |= PMD_BIT4;
+
+ cp = &cache_policies[cachepolicy];
+ kern_pgprot = user_pgprot = cp->pte;
+
+ /*
+ * Enable CPU-specific coherency if supported.
+ * (Only available on XSC3 at the moment.)
+ */
+ if (arch_is_coherent()) {
+ if (cpu_is_xsc3()) {
+ mem_types[MT_MEMORY].prot_sect |= PMD_SECT_S;
+ mem_types[MT_MEMORY].prot_pte |= L_PTE_SHARED;
+ }
+ }
+
+ /*
+ * ARMv6 and above have extended page tables.
+ */
+ if (cpu_arch >= CPU_ARCH_ARMv6 && (cr & CR_XP)) {
+ /*
+ * bit 4 becomes XN which we must clear for the
+ * kernel memory mapping.
+ */
+ mem_types[MT_MEMORY].prot_sect &= ~PMD_SECT_XN;
+ mem_types[MT_ROM].prot_sect &= ~PMD_SECT_XN;
+
+ /*
+ * Mark cache clean areas and XIP ROM read only
+ * from SVC mode and no access from userspace.
+ */
+ mem_types[MT_ROM].prot_sect |= PMD_SECT_APX|PMD_SECT_AP_WRITE;
+ mem_types[MT_MINICLEAN].prot_sect |= PMD_SECT_APX|PMD_SECT_AP_WRITE;
+ mem_types[MT_CACHECLEAN].prot_sect |= PMD_SECT_APX|PMD_SECT_AP_WRITE;
+
+ /*
+ * Mark the device area as "shared device"
+ */
+ mem_types[MT_DEVICE].prot_pte |= L_PTE_BUFFERABLE;
+ mem_types[MT_DEVICE].prot_sect |= PMD_SECT_BUFFERED;
+
+#ifdef CONFIG_SMP
+ /*
+ * Mark memory with the "shared" attribute for SMP systems
+ */
+ user_pgprot |= L_PTE_SHARED;
+ kern_pgprot |= L_PTE_SHARED;
+ mem_types[MT_MEMORY].prot_sect |= PMD_SECT_S;
+#endif
+ }
+
+ for (i = 0; i < 16; i++) {
+ unsigned long v = pgprot_val(protection_map[i]);
+ v = (v & ~(L_PTE_BUFFERABLE|L_PTE_CACHEABLE)) | user_pgprot;
+ protection_map[i] = __pgprot(v);
+ }
+
+ mem_types[MT_LOW_VECTORS].prot_pte |= kern_pgprot;
+ mem_types[MT_HIGH_VECTORS].prot_pte |= kern_pgprot;
+
+ if (cpu_arch >= CPU_ARCH_ARMv5) {
+#ifndef CONFIG_SMP
+ /*
+ * Only use write-through for non-SMP systems
+ */
+ mem_types[MT_LOW_VECTORS].prot_pte &= ~L_PTE_BUFFERABLE;
+ mem_types[MT_HIGH_VECTORS].prot_pte &= ~L_PTE_BUFFERABLE;
+#endif
+ } else {
+ mem_types[MT_MINICLEAN].prot_sect &= ~PMD_SECT_TEX(1);
+ }
+
+ pgprot_kernel = __pgprot(L_PTE_PRESENT | L_PTE_YOUNG |
+ L_PTE_DIRTY | L_PTE_WRITE |
+ L_PTE_EXEC | kern_pgprot);
+
+ mem_types[MT_LOW_VECTORS].prot_l1 |= ecc_mask;
+ mem_types[MT_HIGH_VECTORS].prot_l1 |= ecc_mask;
+ mem_types[MT_MEMORY].prot_sect |= ecc_mask | cp->pmd;
+ mem_types[MT_ROM].prot_sect |= cp->pmd;
+
+ switch (cp->pmd) {
+ case PMD_SECT_WT:
+ mem_types[MT_CACHECLEAN].prot_sect |= PMD_SECT_WT;
+ break;
+ case PMD_SECT_WB:
+ case PMD_SECT_WBWA:
+ mem_types[MT_CACHECLEAN].prot_sect |= PMD_SECT_WB;
+ break;
+ }
+ printk("Memory policy: ECC %sabled, Data cache %s\n",
+ ecc_mask ? "en" : "dis", cp->policy);
+}
+
+#define vectors_base() (vectors_high() ? 0xffff0000 : 0)
+
+/*
+ * Create a SECTION PGD between VIRT and PHYS in domain
+ * DOMAIN with protection PROT. This operates on half-
+ * pgdir entry increments.
+ */
+static inline void
+alloc_init_section(unsigned long virt, unsigned long phys, int prot)
+{
+ pmd_t *pmdp = pmd_off_k(virt);
+
+ if (virt & (1 << 20))
+ pmdp++;
+
+ *pmdp = __pmd(phys | prot);
+ flush_pmd_entry(pmdp);
+}
+
+/*
+ * Create a SUPER SECTION PGD between VIRT and PHYS with protection PROT
+ */
+static inline void
+alloc_init_supersection(unsigned long virt, unsigned long phys, int prot)
+{
+ int i;
+
+ for (i = 0; i < 16; i += 1) {
+ alloc_init_section(virt, phys, prot | PMD_SECT_SUPER);
+
+ virt += (PGDIR_SIZE / 2);
+ }
+}