diff options
Diffstat (limited to 'arch/parisc/mm')
-rw-r--r-- | arch/parisc/mm/Makefile | 5 | ||||
-rw-r--r-- | arch/parisc/mm/fault.c | 271 | ||||
-rw-r--r-- | arch/parisc/mm/init.c | 1019 | ||||
-rw-r--r-- | arch/parisc/mm/ioremap.c | 207 | ||||
-rw-r--r-- | arch/parisc/mm/kmap.c | 166 |
5 files changed, 1668 insertions, 0 deletions
diff --git a/arch/parisc/mm/Makefile b/arch/parisc/mm/Makefile new file mode 100644 index 000000000000..758ceefb373a --- /dev/null +++ b/arch/parisc/mm/Makefile @@ -0,0 +1,5 @@ +# +# Makefile for arch/parisc/mm +# + +obj-y := init.o fault.o ioremap.o diff --git a/arch/parisc/mm/fault.c b/arch/parisc/mm/fault.c new file mode 100644 index 000000000000..eaa701479f5f --- /dev/null +++ b/arch/parisc/mm/fault.c @@ -0,0 +1,271 @@ +/* $Id: fault.c,v 1.5 2000/01/26 16:20:29 jsm Exp $ + * + * This file is subject to the terms and conditions of the GNU General Public + * License. See the file "COPYING" in the main directory of this archive + * for more details. + * + * + * Copyright (C) 1995, 1996, 1997, 1998 by Ralf Baechle + * Copyright 1999 SuSE GmbH (Philipp Rumpf, prumpf@tux.org) + * Copyright 1999 Hewlett Packard Co. + * + */ + +#include <linux/mm.h> +#include <linux/ptrace.h> +#include <linux/sched.h> +#include <linux/interrupt.h> +#include <linux/module.h> + +#include <asm/uaccess.h> +#include <asm/traps.h> + +#define PRINT_USER_FAULTS /* (turn this on if you want user faults to be */ + /* dumped to the console via printk) */ + + +/* Defines for parisc_acctyp() */ +#define READ 0 +#define WRITE 1 + +/* Various important other fields */ +#define bit22set(x) (x & 0x00000200) +#define bits23_25set(x) (x & 0x000001c0) +#define isGraphicsFlushRead(x) ((x & 0xfc003fdf) == 0x04001a80) + /* extended opcode is 0x6a */ + +#define BITSSET 0x1c0 /* for identifying LDCW */ + + +DEFINE_PER_CPU(struct exception_data, exception_data); + +/* + * parisc_acctyp(unsigned int inst) -- + * Given a PA-RISC memory access instruction, determine if the + * the instruction would perform a memory read or memory write + * operation. + * + * This function assumes that the given instruction is a memory access + * instruction (i.e. you should really only call it if you know that + * the instruction has generated some sort of a memory access fault). + * + * Returns: + * VM_READ if read operation + * VM_WRITE if write operation + * VM_EXEC if execute operation + */ +static unsigned long +parisc_acctyp(unsigned long code, unsigned int inst) +{ + if (code == 6 || code == 16) + return VM_EXEC; + + switch (inst & 0xf0000000) { + case 0x40000000: /* load */ + case 0x50000000: /* new load */ + return VM_READ; + + case 0x60000000: /* store */ + case 0x70000000: /* new store */ + return VM_WRITE; + + case 0x20000000: /* coproc */ + case 0x30000000: /* coproc2 */ + if (bit22set(inst)) + return VM_WRITE; + + case 0x0: /* indexed/memory management */ + if (bit22set(inst)) { + /* + * Check for the 'Graphics Flush Read' instruction. + * It resembles an FDC instruction, except for bits + * 20 and 21. Any combination other than zero will + * utilize the block mover functionality on some + * older PA-RISC platforms. The case where a block + * move is performed from VM to graphics IO space + * should be treated as a READ. + * + * The significance of bits 20,21 in the FDC + * instruction is: + * + * 00 Flush data cache (normal instruction behavior) + * 01 Graphics flush write (IO space -> VM) + * 10 Graphics flush read (VM -> IO space) + * 11 Graphics flush read/write (VM <-> IO space) + */ + if (isGraphicsFlushRead(inst)) + return VM_READ; + return VM_WRITE; + } else { + /* + * Check for LDCWX and LDCWS (semaphore instructions). + * If bits 23 through 25 are all 1's it is one of + * the above two instructions and is a write. + * + * Note: With the limited bits we are looking at, + * this will also catch PROBEW and PROBEWI. However, + * these should never get in here because they don't + * generate exceptions of the type: + * Data TLB miss fault/data page fault + * Data memory protection trap + */ + if (bits23_25set(inst) == BITSSET) + return VM_WRITE; + } + return VM_READ; /* Default */ + } + return VM_READ; /* Default */ +} + +#undef bit22set +#undef bits23_25set +#undef isGraphicsFlushRead +#undef BITSSET + + +#if 0 +/* This is the treewalk to find a vma which is the highest that has + * a start < addr. We're using find_vma_prev instead right now, but + * we might want to use this at some point in the future. Probably + * not, but I want it committed to CVS so I don't lose it :-) + */ + while (tree != vm_avl_empty) { + if (tree->vm_start > addr) { + tree = tree->vm_avl_left; + } else { + prev = tree; + if (prev->vm_next == NULL) + break; + if (prev->vm_next->vm_start > addr) + break; + tree = tree->vm_avl_right; + } + } +#endif + +void do_page_fault(struct pt_regs *regs, unsigned long code, + unsigned long address) +{ + struct vm_area_struct *vma, *prev_vma; + struct task_struct *tsk = current; + struct mm_struct *mm = tsk->mm; + const struct exception_table_entry *fix; + unsigned long acc_type; + + if (in_interrupt() || !mm) + goto no_context; + + down_read(&mm->mmap_sem); + vma = find_vma_prev(mm, address, &prev_vma); + if (!vma || address < vma->vm_start) + goto check_expansion; +/* + * Ok, we have a good vm_area for this memory access. We still need to + * check the access permissions. + */ + +good_area: + + acc_type = parisc_acctyp(code,regs->iir); + + if ((vma->vm_flags & acc_type) != acc_type) + goto bad_area; + + /* + * If for any reason at all we couldn't handle the fault, make + * sure we exit gracefully rather than endlessly redo the + * fault. + */ + + switch (handle_mm_fault(mm, vma, address, (acc_type & VM_WRITE) != 0)) { + case 1: + ++current->min_flt; + break; + case 2: + ++current->maj_flt; + break; + case 0: + /* + * We ran out of memory, or some other thing happened + * to us that made us unable to handle the page fault + * gracefully. + */ + goto bad_area; + default: + goto out_of_memory; + } + up_read(&mm->mmap_sem); + return; + +check_expansion: + vma = prev_vma; + if (vma && (expand_stack(vma, address) == 0)) + goto good_area; + +/* + * Something tried to access memory that isn't in our memory map.. + */ +bad_area: + up_read(&mm->mmap_sem); + + if (user_mode(regs)) { + struct siginfo si; + +#ifdef PRINT_USER_FAULTS + printk(KERN_DEBUG "\n"); + printk(KERN_DEBUG "do_page_fault() pid=%d command='%s' type=%lu address=0x%08lx\n", + tsk->pid, tsk->comm, code, address); + if (vma) { + printk(KERN_DEBUG "vm_start = 0x%08lx, vm_end = 0x%08lx\n", + vma->vm_start, vma->vm_end); + } + show_regs(regs); +#endif + /* FIXME: actually we need to get the signo and code correct */ + si.si_signo = SIGSEGV; + si.si_errno = 0; + si.si_code = SEGV_MAPERR; + si.si_addr = (void __user *) address; + force_sig_info(SIGSEGV, &si, current); + return; + } + +no_context: + + if (!user_mode(regs)) { + fix = search_exception_tables(regs->iaoq[0]); + + if (fix) { + struct exception_data *d; + + d = &__get_cpu_var(exception_data); + d->fault_ip = regs->iaoq[0]; + d->fault_space = regs->isr; + d->fault_addr = regs->ior; + + regs->iaoq[0] = ((fix->fixup) & ~3); + + /* + * NOTE: In some cases the faulting instruction + * may be in the delay slot of a branch. We + * don't want to take the branch, so we don't + * increment iaoq[1], instead we set it to be + * iaoq[0]+4, and clear the B bit in the PSW + */ + + regs->iaoq[1] = regs->iaoq[0] + 4; + regs->gr[0] &= ~PSW_B; /* IPSW in gr[0] */ + + return; + } + } + + parisc_terminate("Bad Address (null pointer deref?)", regs, code, address); + + out_of_memory: + up_read(&mm->mmap_sem); + printk(KERN_CRIT "VM: killing process %s\n", current->comm); + if (user_mode(regs)) + do_exit(SIGKILL); + goto no_context; +} diff --git a/arch/parisc/mm/init.c b/arch/parisc/mm/init.c new file mode 100644 index 000000000000..cac37589e35c --- /dev/null +++ b/arch/parisc/mm/init.c @@ -0,0 +1,1019 @@ +/* + * linux/arch/parisc/mm/init.c + * + * Copyright (C) 1995 Linus Torvalds + * Copyright 1999 SuSE GmbH + * changed by Philipp Rumpf + * Copyright 1999 Philipp Rumpf (prumpf@tux.org) + * Copyright 2004 Randolph Chung (tausq@debian.org) + * + */ + +#include <linux/config.h> + +#include <linux/module.h> +#include <linux/mm.h> +#include <linux/bootmem.h> +#include <linux/delay.h> +#include <linux/init.h> +#include <linux/pci.h> /* for hppa_dma_ops and pcxl_dma_ops */ +#include <linux/initrd.h> +#include <linux/swap.h> +#include <linux/unistd.h> +#include <linux/nodemask.h> /* for node_online_map */ +#include <linux/pagemap.h> /* for release_pages and page_cache_release */ + +#include <asm/pgalloc.h> +#include <asm/tlb.h> +#include <asm/pdc_chassis.h> +#include <asm/mmzone.h> + +DEFINE_PER_CPU(struct mmu_gather, mmu_gathers); + +extern char _text; /* start of kernel code, defined by linker */ +extern int data_start; +extern char _end; /* end of BSS, defined by linker */ +extern char __init_begin, __init_end; + +#ifdef CONFIG_DISCONTIGMEM +struct node_map_data node_data[MAX_NUMNODES]; +bootmem_data_t bmem_data[MAX_NUMNODES]; +unsigned char pfnnid_map[PFNNID_MAP_MAX]; +#endif + +static struct resource data_resource = { + .name = "Kernel data", + .flags = IORESOURCE_BUSY | IORESOURCE_MEM, +}; + +static struct resource code_resource = { + .name = "Kernel code", + .flags = IORESOURCE_BUSY | IORESOURCE_MEM, +}; + +static struct resource pdcdata_resource = { + .name = "PDC data (Page Zero)", + .start = 0, + .end = 0x9ff, + .flags = IORESOURCE_BUSY | IORESOURCE_MEM, +}; + +static struct resource sysram_resources[MAX_PHYSMEM_RANGES]; + +/* The following array is initialized from the firmware specific + * information retrieved in kernel/inventory.c. + */ + +physmem_range_t pmem_ranges[MAX_PHYSMEM_RANGES]; +int npmem_ranges; + +#ifdef __LP64__ +#define MAX_MEM (~0UL) +#else /* !__LP64__ */ +#define MAX_MEM (3584U*1024U*1024U) +#endif /* !__LP64__ */ + +static unsigned long mem_limit = MAX_MEM; + +static void __init mem_limit_func(void) +{ + char *cp, *end; + unsigned long limit; + extern char saved_command_line[]; + + /* We need this before __setup() functions are called */ + + limit = MAX_MEM; + for (cp = saved_command_line; *cp; ) { + if (memcmp(cp, "mem=", 4) == 0) { + cp += 4; + limit = memparse(cp, &end); + if (end != cp) + break; + cp = end; + } else { + while (*cp != ' ' && *cp) + ++cp; + while (*cp == ' ') + ++cp; + } + } + + if (limit < mem_limit) + mem_limit = limit; +} + +#define MAX_GAP (0x40000000UL >> PAGE_SHIFT) + +static void __init setup_bootmem(void) +{ + unsigned long bootmap_size; + unsigned long mem_max; + unsigned long bootmap_pages; + unsigned long bootmap_start_pfn; + unsigned long bootmap_pfn; +#ifndef CONFIG_DISCONTIGMEM + physmem_range_t pmem_holes[MAX_PHYSMEM_RANGES - 1]; + int npmem_holes; +#endif + int i, sysram_resource_count; + + disable_sr_hashing(); /* Turn off space register hashing */ + + /* + * Sort the ranges. Since the number of ranges is typically + * small, and performance is not an issue here, just do + * a simple insertion sort. + */ + + for (i = 1; i < npmem_ranges; i++) { + int j; + + for (j = i; j > 0; j--) { + unsigned long tmp; + + if (pmem_ranges[j-1].start_pfn < + pmem_ranges[j].start_pfn) { + + break; + } + tmp = pmem_ranges[j-1].start_pfn; + pmem_ranges[j-1].start_pfn = pmem_ranges[j].start_pfn; + pmem_ranges[j].start_pfn = tmp; + tmp = pmem_ranges[j-1].pages; + pmem_ranges[j-1].pages = pmem_ranges[j].pages; + pmem_ranges[j].pages = tmp; + } + } + +#ifndef CONFIG_DISCONTIGMEM + /* + * Throw out ranges that are too far apart (controlled by + * MAX_GAP). + */ + + for (i = 1; i < npmem_ranges; i++) { + if (pmem_ranges[i].start_pfn - + (pmem_ranges[i-1].start_pfn + + pmem_ranges[i-1].pages) > MAX_GAP) { + npmem_ranges = i; + printk("Large gap in memory detected (%ld pages). " + "Consider turning on CONFIG_DISCONTIGMEM\n", + pmem_ranges[i].start_pfn - + (pmem_ranges[i-1].start_pfn + + pmem_ranges[i-1].pages)); + break; + } + } +#endif + + if (npmem_ranges > 1) { + + /* Print the memory ranges */ + + printk(KERN_INFO "Memory Ranges:\n"); + + for (i = 0; i < npmem_ranges; i++) { + unsigned long start; + unsigned long size; + + size = (pmem_ranges[i].pages << PAGE_SHIFT); + start = (pmem_ranges[i].start_pfn << PAGE_SHIFT); + printk(KERN_INFO "%2d) Start 0x%016lx End 0x%016lx Size %6ld MB\n", + i,start, start + (size - 1), size >> 20); + } + } + + sysram_resource_count = npmem_ranges; + for (i = 0; i < sysram_resource_count; i++) { + struct resource *res = &sysram_resources[i]; + res->name = "System RAM"; + res->start = pmem_ranges[i].start_pfn << PAGE_SHIFT; + res->end = res->start + (pmem_ranges[i].pages << PAGE_SHIFT)-1; + res->flags = IORESOURCE_MEM | IORESOURCE_BUSY; + request_resource(&iomem_resource, res); + } + + /* + * For 32 bit kernels we limit the amount of memory we can + * support, in order to preserve enough kernel address space + * for other purposes. For 64 bit kernels we don't normally + * limit the memory, but this mechanism can be used to + * artificially limit the amount of memory (and it is written + * to work with multiple memory ranges). + */ + + mem_limit_func(); /* check for "mem=" argument */ + + mem_max = 0; + num_physpages = 0; + for (i = 0; i < npmem_ranges; i++) { + unsigned long rsize; + + rsize = pmem_ranges[i].pages << PAGE_SHIFT; + if ((mem_max + rsize) > mem_limit) { + printk(KERN_WARNING "Memory truncated to %ld MB\n", mem_limit >> 20); + if (mem_max == mem_limit) + npmem_ranges = i; + else { + pmem_ranges[i].pages = (mem_limit >> PAGE_SHIFT) + - (mem_max >> PAGE_SHIFT); + npmem_ranges = i + 1; + mem_max = mem_limit; + } + num_physpages += pmem_ranges[i].pages; + break; + } + num_physpages += pmem_ranges[i].pages; + mem_max += rsize; + } + + printk(KERN_INFO "Total Memory: %ld MB\n",mem_max >> 20); + +#ifndef CONFIG_DISCONTIGMEM + /* Merge the ranges, keeping track of the holes */ + + { + unsigned long end_pfn; + unsigned long hole_pages; + + npmem_holes = 0; + end_pfn = pmem_ranges[0].start_pfn + pmem_ranges[0].pages; + for (i = 1; i < npmem_ranges; i++) { + + hole_pages = pmem_ranges[i].start_pfn - end_pfn; + if (hole_pages) { + pmem_holes[npmem_holes].start_pfn = end_pfn; + pmem_holes[npmem_holes++].pages = hole_pages; + end_pfn += hole_pages; + } + end_pfn += pmem_ranges[i].pages; + } + + pmem_ranges[0].pages = end_pfn - pmem_ranges[0].start_pfn; + npmem_ranges = 1; + } +#endif + + bootmap_pages = 0; + for (i = 0; i < npmem_ranges; i++) + bootmap_pages += bootmem_bootmap_pages(pmem_ranges[i].pages); + + bootmap_start_pfn = PAGE_ALIGN(__pa((unsigned long) &_end)) >> PAGE_SHIFT; + +#ifdef CONFIG_DISCONTIGMEM + for (i = 0; i < MAX_PHYSMEM_RANGES; i++) { + memset(NODE_DATA(i), 0, sizeof(pg_data_t)); + NODE_DATA(i)->bdata = &bmem_data[i]; + } + memset(pfnnid_map, 0xff, sizeof(pfnnid_map)); + + for (i = 0; i < npmem_ranges; i++) + node_set_online(i); +#endif + + /* + * Initialize and free the full range of memory in each range. + * Note that the only writing these routines do are to the bootmap, + * and we've made sure to locate the bootmap properly so that they + * won't be writing over anything important. + */ + + bootmap_pfn = bootmap_start_pfn; + max_pfn = 0; + for (i = 0; i < npmem_ranges; i++) { + unsigned long start_pfn; + unsigned long npages; + + start_pfn = pmem_ranges[i].start_pfn; + npages = pmem_ranges[i].pages; + + bootmap_size = init_bootmem_node(NODE_DATA(i), + bootmap_pfn, + start_pfn, + (start_pfn + npages) ); + free_bootmem_node(NODE_DATA(i), + (start_pfn << PAGE_SHIFT), + (npages << PAGE_SHIFT) ); + bootmap_pfn += (bootmap_size + PAGE_SIZE - 1) >> PAGE_SHIFT; + if ((start_pfn + npages) > max_pfn) + max_pfn = start_pfn + npages; + } + + if ((bootmap_pfn - bootmap_start_pfn) != bootmap_pages) { + printk(KERN_WARNING "WARNING! bootmap sizing is messed up!\n"); + BUG(); + } + + /* reserve PAGE0 pdc memory, kernel text/data/bss & bootmap */ + +#define PDC_CONSOLE_IO_IODC_SIZE 32768 + + reserve_bootmem_node(NODE_DATA(0), 0UL, + (unsigned long)(PAGE0->mem_free + PDC_CONSOLE_IO_IODC_SIZE)); + reserve_bootmem_node(NODE_DATA(0),__pa((unsigned long)&_text), + (unsigned long)(&_end - &_text)); + reserve_bootmem_node(NODE_DATA(0), (bootmap_start_pfn << PAGE_SHIFT), + ((bootmap_pfn - bootmap_start_pfn) << PAGE_SHIFT)); + +#ifndef CONFIG_DISCONTIGMEM + + /* reserve the holes */ + + for (i = 0; i < npmem_holes; i++) { + reserve_bootmem_node(NODE_DATA(0), + (pmem_holes[i].start_pfn << PAGE_SHIFT), + (pmem_holes[i].pages << PAGE_SHIFT)); + } +#endif + +#ifdef CONFIG_BLK_DEV_INITRD + if (initrd_start) { + printk(KERN_INFO "initrd: %08lx-%08lx\n", initrd_start, initrd_end); + if (__pa(initrd_start) < mem_max) { + unsigned long initrd_reserve; + + if (__pa(initrd_end) > mem_max) { + initrd_reserve = mem_max - __pa(initrd_start); + } else { + initrd_reserve = initrd_end - initrd_start; + } + initrd_below_start_ok = 1; + printk(KERN_INFO "initrd: reserving %08lx-%08lx (mem_max %08lx)\n", __pa(initrd_start), __pa(initrd_start) + initrd_reserve, mem_max); + + reserve_bootmem_node(NODE_DATA(0),__pa(initrd_start), initrd_reserve); + } + } +#endif + + data_resource.start = virt_to_phys(&data_start); + data_resource.end = virt_to_phys(&_end)-1; + code_resource.start = virt_to_phys(&_text); + code_resource.end = virt_to_phys(&data_start)-1; + + /* We don't know which region the kernel will be in, so try + * all of them. + */ + for (i = 0; i < sysram_resource_count; i++) { + struct resource *res = &sysram_resources[i]; + request_resource(res, &code_resource); + request_resource(res, &data_resource); + } + request_resource(&sysram_resources[0], &pdcdata_resource); +} + +void free_initmem(void) +{ + /* FIXME: */ +#if 0 + printk(KERN_INFO "NOT FREEING INITMEM (%dk)\n", + (&__init_end - &__init_begin) >> 10); + return; +#else + unsigned long addr; + + printk(KERN_INFO "Freeing unused kernel memory: "); + +#if 1 + /* Attempt to catch anyone trying to execute code here + * by filling the page with BRK insns. + * + * If we disable interrupts for all CPUs, then IPI stops working. + * Kinda breaks the global cache flushing. + */ + local_irq_disable(); + + memset(&__init_begin, 0x00, + (unsigned long)&__init_end - (unsigned long)&__init_begin); + + flush_data_cache(); + asm volatile("sync" : : ); + flush_icache_range((unsigned long)&__init_begin, (unsigned long)&__init_end); + asm volatile("sync" : : ); + + local_irq_enable(); +#endif + + addr = (unsigned long)(&__init_begin); + for (; addr < (unsigned long)(&__init_end); addr += PAGE_SIZE) { + ClearPageReserved(virt_to_page(addr)); + set_page_count(virt_to_page(addr), 1); + free_page(addr); + num_physpages++; + totalram_pages++; + } + + /* set up a new led state on systems shipped LED State panel */ + pdc_chassis_send_status(PDC_CHASSIS_DIRECT_BCOMPLETE); + + printk("%luk freed\n", (unsigned long)(&__init_end - &__init_begin) >> 10); +#endif +} + +/* + * Just an arbitrary offset to serve as a "hole" between mapping areas + * (between top of physical memory and a potential pcxl dma mapping + * area, and below the vmalloc mapping area). + * + * The current 32K value just means that there will be a 32K "hole" + * between mapping areas. That means that any out-of-bounds memory + * accesses will hopefully be caught. The vmalloc() routines leaves + * a hole of 4kB between each vmalloced area for the same reason. + */ + + /* Leave room for gateway page expansion */ +#if KERNEL_MAP_START < GATEWAY_PAGE_SIZE +#error KERNEL_MAP_START is in gateway reserved region +#endif +#define MAP_START (KERNEL_MAP_START) + +#define VM_MAP_OFFSET (32*1024) +#define SET_MAP_OFFSET(x) ((void *)(((unsigned long)(x) + VM_MAP_OFFSET) \ + & ~(VM_MAP_OFFSET-1))) + +void *vmalloc_start; +EXPORT_SYMBOL(vmalloc_start); + +#ifdef CONFIG_PA11 +unsigned long pcxl_dma_start; +#endif + +void __init mem_init(void) +{ + high_memory = __va((max_pfn << PAGE_SHIFT)); + +#ifndef CONFIG_DISCONTIGMEM + max_mapnr = page_to_pfn(virt_to_page(high_memory - 1)) + 1; + totalram_pages += free_all_bootmem(); +#else + { + int i; + + for (i = 0; i < npmem_ranges; i++) + totalram_pages += free_all_bootmem_node(NODE_DATA(i)); + } +#endif + + printk(KERN_INFO "Memory: %luk available\n", num_physpages << (PAGE_SHIFT-10)); + +#ifdef CONFIG_PA11 + if (hppa_dma_ops == &pcxl_dma_ops) { + pcxl_dma_start = (unsigned long)SET_MAP_OFFSET(MAP_START); + vmalloc_start = SET_MAP_OFFSET(pcxl_dma_start + PCXL_DMA_MAP_SIZE); + } else { + pcxl_dma_start = 0; + vmalloc_start = SET_MAP_OFFSET(MAP_START); + } +#else + vmalloc_start = SET_MAP_OFFSET(MAP_START); +#endif + +} + +int do_check_pgt_cache(int low, int high) +{ + return 0; +} + +unsigned long *empty_zero_page; + +void show_mem(void) +{ + int i,free = 0,total = 0,reserved = 0; + int shared = 0, cached = 0; + + printk(KERN_INFO "Mem-info:\n"); + show_free_areas(); + printk(KERN_INFO "Free swap: %6ldkB\n", + nr_swap_pages<<(PAGE_SHIFT-10)); +#ifndef CONFIG_DISCONTIGMEM + i = max_mapnr; + while (i-- > 0) { + total++; + if (PageReserved(mem_map+i)) + reserved++; + else if (PageSwapCache(mem_map+i)) + cached++; + else if (!page_count(&mem_map[i])) + free++; + else + shared += page_count(&mem_map[i]) - 1; + } +#else + for (i = 0; i < npmem_ranges; i++) { + int j; + + for (j = node_start_pfn(i); j < node_end_pfn(i); j++) { + struct page *p; + + p = node_mem_map(i) + j - node_start_pfn(i); + + total++; + if (PageReserved(p)) + reserved++; + else if (PageSwapCache(p)) + cached++; + else if (!page_count(p)) + free++; + else + shared += page_count(p) - 1; + } + } +#endif + printk(KERN_INFO "%d pages of RAM\n", total); + printk(KERN_INFO "%d reserved pages\n", reserved); + printk(KERN_INFO "%d pages shared\n", shared); + printk(KERN_INFO "%d pages swap cached\n", cached); + + +#ifdef CONFIG_DISCONTIGMEM + { + struct zonelist *zl; + int i, j, k; + + for (i = 0; i < npmem_ranges; i++) { + for (j = 0; j < MAX_NR_ZONES; j++) { + zl = NODE_DATA(i)->node_zonelists + j; + + printk("Zone list for zone %d on node %d: ", j, i); + for (k = 0; zl->zones[k] != NULL; k++) + printk("[%d/%s] ", zl->zones[k]->zone_pgdat->node_id, zl->zones[k]->name); + printk("\n"); + } + } + } +#endif +} + + +static void __init map_pages(unsigned long start_vaddr, unsigned long start_paddr, unsigned long size, pgprot_t pgprot) +{ + pgd_t *pg_dir; + pmd_t *pmd; + pte_t *pg_table; + unsigned long end_paddr; + unsigned long start_pmd; + unsigned long start_pte; + unsigned long tmp1; + unsigned long tmp2; + unsigned long address; + unsigned long ro_start; + unsigned long ro_end; + unsigned long fv_addr; + unsigned long gw_addr; + extern const unsigned long fault_vector_20; + extern void * const linux_gateway_page; + + ro_start = __pa((unsigned long)&_text); + ro_end = __pa((unsigned long)&data_start); + fv_addr = __pa((unsigned long)&fault_vector_20) & PAGE_MASK; + gw_addr = __pa((unsigned long)&linux_gateway_page) & PAGE_MASK; + + end_paddr = start_paddr + size; + + pg_dir = pgd_offset_k(start_vaddr); + +#if PTRS_PER_PMD == 1 + start_pmd = 0; +#else + start_pmd = ((start_vaddr >> PMD_SHIFT) & (PTRS_PER_PMD - 1)); +#endif + start_pte = ((start_vaddr >> PAGE_SHIFT) & (PTRS_PER_PTE - 1)); + + address = start_paddr; + while (address < end_paddr) { +#if PTRS_PER_PMD == 1 + pmd = (pmd_t *)__pa(pg_dir); +#else + pmd = (pmd_t *)pgd_address(*pg_dir); + + /* + * pmd is physical at this point + */ + + if (!pmd) { + pmd = (pmd_t *) alloc_bootmem_low_pages_node(NODE_DATA(0),PAGE_SIZE << PMD_ORDER); + pmd = (pmd_t *) __pa(pmd); + } + + pgd_populate(NULL, pg_dir, __va(pmd)); +#endif + pg_dir++; + + /* now change pmd to kernel virtual addresses */ + + pmd = (pmd_t *)__va(pmd) + start_pmd; + for (tmp1 = start_pmd; tmp1 < PTRS_PER_PMD; tmp1++,pmd++) { + + /* + * pg_table is physical at this point + */ + + pg_table = (pte_t *)pmd_address(*pmd); + if (!pg_table) { + pg_table = (pte_t *) + alloc_bootmem_low_pages_node(NODE_DATA(0),PAGE_SIZE); + pg_table = (pte_t *) __pa(pg_table); + } + + pmd_populate_kernel(NULL, pmd, __va(pg_table)); + + /* now change pg_table to kernel virtual addresses */ + + pg_table = (pte_t *) __va(pg_table) + start_pte; + for (tmp2 = start_pte; tmp2 < PTRS_PER_PTE; tmp2++,pg_table++) { + pte_t pte; + + /* + * Map the fault vector writable so we can + * write the HPMC checksum. + */ + if (address >= ro_start && address < ro_end + && address != fv_addr + && address != gw_addr) + pte = __mk_pte(address, PAGE_KERNEL_RO); + else + pte = __mk_pte(address, pgprot); + + if (address >= end_paddr) + pte_val(pte) = 0; + + set_pte(pg_table, pte); + + address += PAGE_SIZE; + } + start_pte = 0; + + if (address >= end_paddr) + break; + } + start_pmd = 0; + } +} + +/* + * pagetable_init() sets up the page tables + * + * Note that gateway_init() places the Linux gateway page at page 0. + * Since gateway pages cannot be dereferenced this has the desirable + * side effect of trapping those pesky NULL-reference errors in the + * kernel. + */ +static void __init pagetable_init(void) +{ + int range; + + /* Map each physical memory range to its kernel vaddr */ + + for (range = 0; range < npmem_ranges; range++) { + unsigned long start_paddr; + unsigned long end_paddr; + unsigned long size; + + start_paddr = pmem_ranges[range].start_pfn << PAGE_SHIFT; + end_paddr = start_paddr + (pmem_ranges[range].pages << PAGE_SHIFT); + size = pmem_ranges[range].pages << PAGE_SHIFT; + + map_pages((unsigned long)__va(start_paddr), start_paddr, + size, PAGE_KERNEL); + } + +#ifdef CONFIG_BLK_DEV_INITRD + if (initrd_end && initrd_end > mem_limit) { + printk("initrd: mapping %08lx-%08lx\n", initrd_start, initrd_end); + map_pages(initrd_start, __pa(initrd_start), + initrd_end - initrd_start, PAGE_KERNEL); + } +#endif + + empty_zero_page = alloc_bootmem_pages(PAGE_SIZE); + memset(empty_zero_page, 0, PAGE_SIZE); +} + +static void __init gateway_init(void) +{ + unsigned long linux_gateway_page_addr; + /* FIXME: This is 'const' in order to trick the compiler + into not treating it as DP-relative data. */ + extern void * const linux_gateway_page; + + linux_gateway_page_addr = LINUX_GATEWAY_ADDR & PAGE_MASK; + + /* + * Setup Linux Gateway page. + * + * The Linux gateway page will reside in kernel space (on virtual + * page 0), so it doesn't need to be aliased into user space. + */ + + map_pages(linux_gateway_page_addr, __pa(&linux_gateway_page), + PAGE_SIZE, PAGE_GATEWAY); +} + +#ifdef CONFIG_HPUX +void +map_hpux_gateway_page(struct task_struct *tsk, struct mm_struct *mm) +{ + pgd_t *pg_dir; + pmd_t *pmd; + pte_t *pg_table; + unsigned long start_pmd; + unsigned long start_pte; + unsigned long address; + unsigned long hpux_gw_page_addr; + /* FIXME: This is 'const' in order to trick the compiler + into not treating it as DP-relative data. */ + extern void * const hpux_gateway_page; + + hpux_gw_page_addr = HPUX_GATEWAY_ADDR & PAGE_MASK; + + /* + * Setup HP-UX Gateway page. + * + * The HP-UX gateway page resides in the user address space, + * so it needs to be aliased into each process. + */ + + pg_dir = pgd_offset(mm,hpux_gw_page_addr); + +#if PTRS_PER_PMD == 1 + start_pmd = 0; +#else + start_pmd = ((hpux_gw_page_addr >> PMD_SHIFT) & (PTRS_PER_PMD - 1)); +#endif + start_pte = ((hpux_gw_page_addr >> PAGE_SHIFT) & (PTRS_PER_PTE - 1)); + + address = __pa(&hpux_gateway_page); +#if PTRS_PER_PMD == 1 + pmd = (pmd_t *)__pa(pg_dir); +#else + pmd = (pmd_t *) pgd_address(*pg_dir); + + /* + * pmd is physical at this point + */ + + if (!pmd) { + pmd = (pmd_t *) get_zeroed_page(GFP_KERNEL); + pmd = (pmd_t *) __pa(pmd); + } + + __pgd_val_set(*pg_dir, PxD_FLAG_PRESENT | PxD_FLAG_VALID | (unsigned long) pmd); +#endif + /* now change pmd to kernel virtual addresses */ + + pmd = (pmd_t *)__va(pmd) + start_pmd; + + /* + * pg_table is physical at this point + */ + + pg_table = (pte_t *) pmd_address(*pmd); + if (!pg_table) + pg_table = (pte_t *) __pa(get_zeroed_page(GFP_KERNEL)); + + __pmd_val_set(*pmd, PxD_FLAG_PRESENT | PxD_FLAG_VALID | (unsigned long) pg_table); + + /* now change pg_table to kernel virtual addresses */ + + pg_table = (pte_t *) __va(pg_table) + start_pte; + set_pte(pg_table, __mk_pte(address, PAGE_GATEWAY)); +} +EXPORT_SYMBOL(map_hpux_gateway_page); +#endif + +extern void flush_tlb_all_local(void); + +void __init paging_init(void) +{ + int i; + + setup_bootmem(); + pagetable_init(); + gateway_init(); + flush_cache_all_local(); /* start with known state */ + flush_tlb_all_local(); + + for (i = 0; i < npmem_ranges; i++) { + unsigned long zones_size[MAX_NR_ZONES] = { 0, 0, 0 }; + + /* We have an IOMMU, so all memory can go into a single + ZONE_DMA zone. */ + zones_size[ZONE_DMA] = pmem_ranges[i].pages; + +#ifdef CONFIG_DISCONTIGMEM + /* Need to initialize the pfnnid_map before we can initialize + the zone */ + { + int j; + for (j = (pmem_ranges[i].start_pfn >> PFNNID_SHIFT); + j <= ((pmem_ranges[i].start_pfn + pmem_ranges[i].pages) >> PFNNID_SHIFT); + j++) { + pfnnid_map[j] = i; + } + } +#endif + + free_area_init_node(i, NODE_DATA(i), zones_size, + pmem_ranges[i].start_pfn, NULL); + } +} + +#ifdef CONFIG_PA20 + +/* + * Currently, all PA20 chips have 18 bit protection id's, which is the + * limiting factor (space ids are 32 bits). + */ + +#define NR_SPACE_IDS 262144 + +#else + +/* + * Currently we have a one-to-one relationship between space id's and + * protection id's. Older parisc chips (PCXS, PCXT, PCXL, PCXL2) only + * support 15 bit protection id's, so that is the limiting factor. + * PCXT' has 18 bit protection id's, but only 16 bit spaceids, so it's + * probably not worth the effort for a special case here. + */ + +#define NR_SPACE_IDS 32768 + +#endif /* !CONFIG_PA20 */ + +#define RECYCLE_THRESHOLD (NR_SPACE_IDS / 2) +#define SID_ARRAY_SIZE (NR_SPACE_IDS / (8 * sizeof(long))) + +static unsigned long space_id[SID_ARRAY_SIZE] = { 1 }; /* disallow space 0 */ +static unsigned long dirty_space_id[SID_ARRAY_SIZE]; +static unsigned long space_id_index; +static unsigned long free_space_ids = NR_SPACE_IDS - 1; +static unsigned long dirty_space_ids = 0; + +static DEFINE_SPINLOCK(sid_lock); + +unsigned long alloc_sid(void) +{ + unsigned long index; + + spin_lock(&sid_lock); + + if (free_space_ids == 0) { + if (dirty_space_ids != 0) { + spin_unlock(&sid_lock); + flush_tlb_all(); /* flush_tlb_all() calls recycle_sids() */ + spin_lock(&sid_lock); + } + if (free_space_ids == 0) + BUG(); + } + + free_space_ids--; + + index = find_next_zero_bit(space_id, NR_SPACE_IDS, space_id_index); + space_id[index >> SHIFT_PER_LONG] |= (1L << (index & (BITS_PER_LONG - 1))); + space_id_index = index; + + spin_unlock(&sid_lock); + + return index << SPACEID_SHIFT; +} + +void free_sid(unsigned long spaceid) +{ + unsigned long index = spaceid >> SPACEID_SHIFT; + unsigned long *dirty_space_offset; + + dirty_space_offset = dirty_space_id + (index >> SHIFT_PER_LONG); + index &= (BITS_PER_LONG - 1); + + spin_lock(&sid_lock); + + if (*dirty_space_offset & (1L << index)) + BUG(); /* attempt to free space id twice */ + + *dirty_space_offset |= (1L << index); + dirty_space_ids++; + + spin_unlock(&sid_lock); +} + + +#ifdef CONFIG_SMP +static void get_dirty_sids(unsigned long *ndirtyptr,unsigned long *dirty_array) +{ + int i; + + /* NOTE: sid_lock must be held upon entry */ + + *ndirtyptr = dirty_space_ids; + if (dirty_space_ids != 0) { + for (i = 0; i < SID_ARRAY_SIZE; i++) { + dirty_array[i] = dirty_space_id[i]; + dirty_space_id[i] = 0; + } + dirty_space_ids = 0; + } + + return; +} + +static void recycle_sids(unsigned long ndirty,unsigned long *dirty_array) +{ + int i; + + /* NOTE: sid_lock must be held upon entry */ + + if (ndirty != 0) { + for (i = 0; i < SID_ARRAY_SIZE; i++) { + space_id[i] ^= dirty_array[i]; + } + + free_space_ids += ndirty; + space_id_index = 0; + } +} + +#else /* CONFIG_SMP */ + +static void recycle_sids(void) +{ + int i; + + /* NOTE: sid_lock must be held upon entry */ + + if (dirty_space_ids != 0) { + for (i = 0; i < SID_ARRAY_SIZE; i++) { + space_id[i] ^= dirty_space_id[i]; + dirty_space_id[i] = 0; + } + + free_space_ids += dirty_space_ids; + dirty_space_ids = 0; + space_id_index = 0; + } +} +#endif + +/* + * flush_tlb_all() calls recycle_sids(), since whenever the entire tlb is + * purged, we can safely reuse the space ids that were released but + * not flushed from the tlb. + */ + +#ifdef CONFIG_SMP + +static unsigned long recycle_ndirty; +static unsigned long recycle_dirty_array[SID_ARRAY_SIZE]; +static unsigned int recycle_inuse = 0; + +void flush_tlb_all(void) +{ + int do_recycle; + + do_recycle = 0; + spin_lock(&sid_lock); + if (dirty_space_ids > RECYCLE_THRESHOLD) { + if (recycle_inuse) { + BUG(); /* FIXME: Use a semaphore/wait queue here */ + } + get_dirty_sids(&recycle_ndirty,recycle_dirty_array); + recycle_inuse++; + do_recycle++; + } + spin_unlock(&sid_lock); + on_each_cpu((void (*)(void *))flush_tlb_all_local, NULL, 1, 1); + if (do_recycle) { + spin_lock(&sid_lock); + recycle_sids(recycle_ndirty,recycle_dirty_array); + recycle_inuse = 0; + spin_unlock(&sid_lock); + } +} +#else +void flush_tlb_all(void) +{ + spin_lock(&sid_lock); + flush_tlb_all_local(); + recycle_sids(); + spin_unlock(&sid_lock); +} +#endif + +#ifdef CONFIG_BLK_DEV_INITRD +void free_initrd_mem(unsigned long start, unsigned long end) +{ +#if 0 + if (start < end) + printk(KERN_INFO "Freeing initrd memory: %ldk freed\n", (end - start) >> 10); + for (; start < end; start += PAGE_SIZE) { + ClearPageReserved(virt_to_page(start)); + set_page_count(virt_to_page(start), 1); + free_page(start); + num_physpages++; + totalram_pages++; + } +#endif +} +#endif diff --git a/arch/parisc/mm/ioremap.c b/arch/parisc/mm/ioremap.c new file mode 100644 index 000000000000..f2df502cdae3 --- /dev/null +++ b/arch/parisc/mm/ioremap.c @@ -0,0 +1,207 @@ +/* + * arch/parisc/mm/ioremap.c + * + * Re-map IO memory to kernel address space so that we can access it. + * This is needed for high PCI addresses that aren't mapped in the + * 640k-1MB IO memory area on PC's + * + * (C) Copyright 1995 1996 Linus Torvalds + * (C) Copyright 2001 Helge Deller <deller@gmx.de> + */ + +#include <linux/vmalloc.h> +#include <linux/errno.h> +#include <linux/module.h> +#include <asm/io.h> +#include <asm/pgalloc.h> + +static inline void remap_area_pte(pte_t * pte, unsigned long address, unsigned long size, + unsigned long phys_addr, unsigned long flags) +{ + unsigned long end; + + address &= ~PMD_MASK; + end = address + size; + if (end > PMD_SIZE) + end = PMD_SIZE; + if (address >= end) + BUG(); + do { + if (!pte_none(*pte)) { + printk(KERN_ERR "remap_area_pte: page already exists\n"); + BUG(); + } + set_pte(pte, mk_pte_phys(phys_addr, __pgprot(_PAGE_PRESENT | _PAGE_RW | + _PAGE_DIRTY | _PAGE_ACCESSED | flags))); + address += PAGE_SIZE; + phys_addr += PAGE_SIZE; + pte++; + } while (address && (address < end)); +} + +static inline int remap_area_pmd(pmd_t * pmd, unsigned long address, unsigned long size, + unsigned long phys_addr, unsigned long flags) +{ + unsigned long end; + + address &= ~PGDIR_MASK; + end = address + size; + if (end > PGDIR_SIZE) + end = PGDIR_SIZE; + phys_addr -= address; + if (address >= end) + BUG(); + do { + pte_t * pte = pte_alloc_kernel(NULL, pmd, address); + if (!pte) + return -ENOMEM; + remap_area_pte(pte, address, end - address, address + phys_addr, flags); + address = (address + PMD_SIZE) & PMD_MASK; + pmd++; + } while (address && (address < end)); + return 0; +} + +#if (USE_HPPA_IOREMAP) +static int remap_area_pages(unsigned long address, unsigned long phys_addr, + unsigned long size, unsigned long flags) +{ + int error; + pgd_t * dir; + unsigned long end = address + size; + + phys_addr -= address; + dir = pgd_offset(&init_mm, address); + flush_cache_all(); + if (address >= end) + BUG(); + spin_lock(&init_mm.page_table_lock); + do { + pmd_t *pmd; + pmd = pmd_alloc(dir, address); + error = -ENOMEM; + if (!pmd) + break; + if (remap_area_pmd(pmd, address, end - address, + phys_addr + address, flags)) + break; + error = 0; + address = (address + PGDIR_SIZE) & PGDIR_MASK; + dir++; + } while (address && (address < end)); + spin_unlock(&init_mm.page_table_lock); + flush_tlb_all(); + return error; +} +#endif /* USE_HPPA_IOREMAP */ + +#ifdef CONFIG_DEBUG_IOREMAP +static unsigned long last = 0; + +void gsc_bad_addr(unsigned long addr) +{ + if (time_after(jiffies, last + HZ*10)) { + printk("gsc_foo() called with bad address 0x%lx\n", addr); + dump_stack(); + last = jiffies; + } +} +EXPORT_SYMBOL(gsc_bad_addr); + +void __raw_bad_addr(const volatile void __iomem *addr) +{ + if (time_after(jiffies, last + HZ*10)) { + printk("__raw_foo() called with bad address 0x%p\n", addr); + dump_stack(); + last = jiffies; + } +} +EXPORT_SYMBOL(__raw_bad_addr); +#endif + +/* + * Generic mapping function (not visible outside): + */ + +/* + * Remap an arbitrary physical address space into the kernel virtual + * address space. Needed when the kernel wants to access high addresses + * directly. + * + * NOTE! We need to allow non-page-aligned mappings too: we will obviously + * have to convert them into an offset in a page-aligned mapping, but the + * caller shouldn't need to know that small detail. + */ +void __iomem * __ioremap(unsigned long phys_addr, unsigned long size, unsigned long flags) +{ +#if !(USE_HPPA_IOREMAP) + + unsigned long end = phys_addr + size - 1; + /* Support EISA addresses */ + if ((phys_addr >= 0x00080000 && end < 0x000fffff) + || (phys_addr >= 0x00500000 && end < 0x03bfffff)) { + phys_addr |= 0xfc000000; + } + +#ifdef CONFIG_DEBUG_IOREMAP + return (void __iomem *)(phys_addr - (0x1UL << NYBBLE_SHIFT)); +#else + return (void __iomem *)phys_addr; +#endif + +#else + void * addr; + struct vm_struct * area; + unsigned long offset, last_addr; + + /* Don't allow wraparound or zero size */ + last_addr = phys_addr + size - 1; + if (!size || last_addr < phys_addr) + return NULL; + + /* + * Don't allow anybody to remap normal RAM that we're using.. + */ + if (phys_addr < virt_to_phys(high_memory)) { + char *t_addr, *t_end; + struct page *page; + + t_addr = __va(phys_addr); + t_end = t_addr + (size - 1); + + for(page = virt_to_page(t_addr); page <= virt_to_page(t_end); page++) + if(!PageReserved(page)) + return NULL; + } + + /* + * Mappings have to be page-aligned + */ + offset = phys_addr & ~PAGE_MASK; + phys_addr &= PAGE_MASK; + size = PAGE_ALIGN(last_addr) - phys_addr; + + /* + * Ok, go for it.. + */ + area = get_vm_area(size, VM_IOREMAP); + if (!area) + return NULL; + addr = area->addr; + if (remap_area_pages((unsigned long) addr, phys_addr, size, flags)) { + vfree(addr); + return NULL; + } + return (void __iomem *) (offset + (char *)addr); +#endif +} + +void iounmap(void __iomem *addr) +{ +#if !(USE_HPPA_IOREMAP) + return; +#else + if (addr > high_memory) + return vfree((void *) (PAGE_MASK & (unsigned long __force) addr)); +#endif +} diff --git a/arch/parisc/mm/kmap.c b/arch/parisc/mm/kmap.c new file mode 100644 index 000000000000..1b1acd5e2f6e --- /dev/null +++ b/arch/parisc/mm/kmap.c @@ -0,0 +1,166 @@ +/* + * kmap/page table map and unmap support routines + * + * Copyright 1999,2000 Hewlett-Packard Company + * Copyright 2000 John Marvin <jsm at hp.com> + * Copyright 2000 Grant Grundler <grundler at parisc-linux.org> + * Copyright 2000 Philipp Rumpf <prumpf@tux.org> + * + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ +/* +** Stolen mostly from arch/parisc/kernel/pci-dma.c +*/ + +#include <linux/types.h> +#include <linux/mm.h> +#include <linux/string.h> +#include <linux/pci.h> + +#include <linux/slab.h> +#include <linux/vmalloc.h> + +#include <asm/uaccess.h> +#include <asm/pgalloc.h> + +#include <asm/io.h> +#include <asm/page.h> /* get_order */ + +#undef flush_cache_all +#define flush_cache_all flush_all_caches + +typedef void (*pte_iterator_t) (pte_t * pte, unsigned long arg); + +#if 0 +/* XXX This routine could be used with iterate_page() to replace + * unmap_uncached_page() and save a little code space but I didn't + * do that since I'm not certain whether this is the right path. -PB + */ +static void unmap_cached_pte(pte_t * pte, unsigned long addr, unsigned long arg) +{ + pte_t page = *pte; + pte_clear(&init_mm, addr, pte); + if (!pte_none(page)) { + if (pte_present(page)) { + unsigned long map_nr = pte_pagenr(page); + if (map_nr < max_mapnr) + __free_page(mem_map + map_nr); + } else { + printk(KERN_CRIT + "Whee.. Swapped out page in kernel page table\n"); + } + } +} +#endif + +/* These two routines should probably check a few things... */ +static void set_uncached(pte_t * pte, unsigned long arg) +{ + pte_val(*pte) |= _PAGE_NO_CACHE; +} + +static void set_cached(pte_t * pte, unsigned long arg) +{ + pte_val(*pte) &= ~_PAGE_NO_CACHE; +} + +static inline void iterate_pte(pmd_t * pmd, unsigned long address, + unsigned long size, pte_iterator_t op, + unsigned long arg) +{ + pte_t *pte; + unsigned long end; + + if (pmd_none(*pmd)) + return; + if (pmd_bad(*pmd)) { + pmd_ERROR(*pmd); + pmd_clear(pmd); + return; + } + pte = pte_offset(pmd, address); + address &= ~PMD_MASK; + end = address + size; + if (end > PMD_SIZE) + end = PMD_SIZE; + do { + op(pte, arg); + address += PAGE_SIZE; + pte++; + } while (address < end); +} + +static inline void iterate_pmd(pgd_t * dir, unsigned long address, + unsigned long size, pte_iterator_t op, + unsigned long arg) +{ + pmd_t *pmd; + unsigned long end; + + if (pgd_none(*dir)) + return; + if (pgd_bad(*dir)) { + pgd_ERROR(*dir); + pgd_clear(dir); + return; + } + pmd = pmd_offset(dir, address); + address &= ~PGDIR_MASK; + end = address + size; + if (end > PGDIR_SIZE) + end = PGDIR_SIZE; + do { + iterate_pte(pmd, address, end - address, op, arg); + address = (address + PMD_SIZE) & PMD_MASK; + pmd++; + } while (address < end); +} + +static void iterate_pages(unsigned long address, unsigned long size, + pte_iterator_t op, unsigned long arg) +{ + pgd_t *dir; + unsigned long end = address + size; + + dir = pgd_offset_k(address); + flush_cache_all(); + do { + iterate_pmd(dir, address, end - address, op, arg); + address = (address + PGDIR_SIZE) & PGDIR_MASK; + dir++; + } while (address && (address < end)); + flush_tlb_all(); +} + +void +kernel_set_cachemode(unsigned long vaddr, unsigned long size, int what) +{ + switch (what) { + case IOMAP_FULL_CACHING: + iterate_pages(vaddr, size, set_cached, 0); + flush_tlb_range(NULL, vaddr, size); + break; + case IOMAP_NOCACHE_SER: + iterate_pages(vaddr, size, set_uncached, 0); + flush_tlb_range(NULL, vaddr, size); + break; + default: + printk(KERN_CRIT + "kernel_set_cachemode mode %d not understood\n", + what); + break; + } +} |