diff options
author | 2015-08-04 12:17:53 -0700 | |
---|---|---|
committer | 2015-08-04 15:44:42 -0700 | |
commit | 9ca8dbcc65cfc63d6f5ef3312a33184e1d726e00 (patch) | |
tree | 1c9cafbcd35f783a87880a10f85d1a060db1a563 /kernel/arch/parisc/mm | |
parent | 98260f3884f4a202f9ca5eabed40b1354c489b29 (diff) |
Add the rt linux 4.1.3-rt3 as base
Import the rt linux 4.1.3-rt3 as OPNFV kvm base.
It's from git://git.kernel.org/pub/scm/linux/kernel/git/rt/linux-rt-devel.git linux-4.1.y-rt and
the base is:
commit 0917f823c59692d751951bf5ea699a2d1e2f26a2
Author: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Date: Sat Jul 25 12:13:34 2015 +0200
Prepare v4.1.3-rt3
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
We lose all the git history this way and it's not good. We
should apply another opnfv project repo in future.
Change-Id: I87543d81c9df70d99c5001fbdf646b202c19f423
Signed-off-by: Yunhong Jiang <yunhong.jiang@intel.com>
Diffstat (limited to 'kernel/arch/parisc/mm')
-rw-r--r-- | kernel/arch/parisc/mm/Makefile | 5 | ||||
-rw-r--r-- | kernel/arch/parisc/mm/fault.c | 351 | ||||
-rw-r--r-- | kernel/arch/parisc/mm/init.c | 971 | ||||
-rw-r--r-- | kernel/arch/parisc/mm/ioremap.c | 99 |
4 files changed, 1426 insertions, 0 deletions
diff --git a/kernel/arch/parisc/mm/Makefile b/kernel/arch/parisc/mm/Makefile new file mode 100644 index 000000000..758ceefb3 --- /dev/null +++ b/kernel/arch/parisc/mm/Makefile @@ -0,0 +1,5 @@ +# +# Makefile for arch/parisc/mm +# + +obj-y := init.o fault.o ioremap.o diff --git a/kernel/arch/parisc/mm/fault.c b/kernel/arch/parisc/mm/fault.c new file mode 100644 index 000000000..15503addd --- /dev/null +++ b/kernel/arch/parisc/mm/fault.c @@ -0,0 +1,351 @@ +/* + * 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 <linux/uaccess.h> + +#include <asm/traps.h> + +/* 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); + +int show_unhandled_signals = 1; + +/* + * 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 + +int fixup_exception(struct pt_regs *regs) +{ + const struct exception_table_entry *fix; + + /* If we only stored 32bit addresses in the exception table we can drop + * out if we faulted on a 64bit address. */ + if ((sizeof(regs->iaoq[0]) > sizeof(fix->insn)) + && (regs->iaoq[0] >> 32)) + return 0; + + fix = search_exception_tables(regs->iaoq[0]); + if (fix) { + struct exception_data *d; + d = this_cpu_ptr(&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 1; + } + + return 0; +} + +/* + * Print out info about fatal segfaults, if the show_unhandled_signals + * sysctl is set: + */ +static inline void +show_signal_msg(struct pt_regs *regs, unsigned long code, + unsigned long address, struct task_struct *tsk, + struct vm_area_struct *vma) +{ + if (!unhandled_signal(tsk, SIGSEGV)) + return; + + if (!printk_ratelimit()) + return; + + pr_warn("\n"); + pr_warn("do_page_fault() command='%s' type=%lu address=0x%08lx", + tsk->comm, code, address); + print_vma_addr(KERN_CONT " in ", regs->iaoq[0]); + if (vma) + pr_warn(" vm_start = 0x%08lx, vm_end = 0x%08lx\n", + vma->vm_start, vma->vm_end); + + show_regs(regs); +} + +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; + struct mm_struct *mm; + unsigned long acc_type; + int fault; + unsigned int flags; + + if (pagefault_disabled()) + goto no_context; + + tsk = current; + mm = tsk->mm; + if (!mm) + goto no_context; + + flags = FAULT_FLAG_ALLOW_RETRY | FAULT_FLAG_KILLABLE; + if (user_mode(regs)) + flags |= FAULT_FLAG_USER; + + acc_type = parisc_acctyp(code, regs->iir); + if (acc_type & VM_WRITE) + flags |= FAULT_FLAG_WRITE; +retry: + 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: + + 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. + */ + + fault = handle_mm_fault(mm, vma, address, flags); + + if ((fault & VM_FAULT_RETRY) && fatal_signal_pending(current)) + return; + + if (unlikely(fault & VM_FAULT_ERROR)) { + /* + * We hit a shared mapping outside of the file, or some + * other thing happened to us that made us unable to + * handle the page fault gracefully. + */ + if (fault & VM_FAULT_OOM) + goto out_of_memory; + else if (fault & VM_FAULT_SIGSEGV) + goto bad_area; + else if (fault & VM_FAULT_SIGBUS) + goto bad_area; + BUG(); + } + if (flags & FAULT_FLAG_ALLOW_RETRY) { + if (fault & VM_FAULT_MAJOR) + current->maj_flt++; + else + current->min_flt++; + if (fault & VM_FAULT_RETRY) { + flags &= ~FAULT_FLAG_ALLOW_RETRY; + + /* + * No need to up_read(&mm->mmap_sem) as we would + * have already released it in __lock_page_or_retry + * in mm/filemap.c. + */ + + goto retry; + } + } + 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; + + show_signal_msg(regs, code, address, tsk, vma); + + switch (code) { + case 15: /* Data TLB miss fault/Data page fault */ + /* send SIGSEGV when outside of vma */ + if (!vma || + address < vma->vm_start || address > vma->vm_end) { + si.si_signo = SIGSEGV; + si.si_code = SEGV_MAPERR; + break; + } + + /* send SIGSEGV for wrong permissions */ + if ((vma->vm_flags & acc_type) != acc_type) { + si.si_signo = SIGSEGV; + si.si_code = SEGV_ACCERR; + break; + } + + /* probably address is outside of mapped file */ + /* fall through */ + case 17: /* NA data TLB miss / page fault */ + case 18: /* Unaligned access - PCXS only */ + si.si_signo = SIGBUS; + si.si_code = (code == 18) ? BUS_ADRALN : BUS_ADRERR; + break; + case 16: /* Non-access instruction TLB miss fault */ + case 26: /* PCXL: Data memory access rights trap */ + default: + si.si_signo = SIGSEGV; + si.si_code = (code == 26) ? SEGV_ACCERR : SEGV_MAPERR; + break; + } + si.si_errno = 0; + si.si_addr = (void __user *) address; + force_sig_info(si.si_signo, &si, current); + return; + } + +no_context: + + if (!user_mode(regs) && fixup_exception(regs)) { + return; + } + + parisc_terminate("Bad Address (null pointer deref?)", regs, code, address); + + out_of_memory: + up_read(&mm->mmap_sem); + if (!user_mode(regs)) + goto no_context; + pagefault_out_of_memory(); +} diff --git a/kernel/arch/parisc/mm/init.c b/kernel/arch/parisc/mm/init.c new file mode 100644 index 000000000..c229427fa --- /dev/null +++ b/kernel/arch/parisc/mm/init.c @@ -0,0 +1,971 @@ +/* + * 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) + * Copyright 2006-2007 Helge Deller (deller@gmx.de) + * + */ + + +#include <linux/module.h> +#include <linux/mm.h> +#include <linux/bootmem.h> +#include <linux/gfp.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/pgtable.h> +#include <asm/tlb.h> +#include <asm/pdc_chassis.h> +#include <asm/mmzone.h> +#include <asm/sections.h> + +extern int data_start; +extern void parisc_kernel_start(void); /* Kernel entry point in head.S */ + +#if CONFIG_PGTABLE_LEVELS == 3 +/* NOTE: This layout exactly conforms to the hybrid L2/L3 page table layout + * with the first pmd adjacent to the pgd and below it. gcc doesn't actually + * guarantee that global objects will be laid out in memory in the same order + * as the order of declaration, so put these in different sections and use + * the linker script to order them. */ +pmd_t pmd0[PTRS_PER_PMD] __attribute__ ((__section__ (".data..vm0.pmd"), aligned(PAGE_SIZE))); +#endif + +pgd_t swapper_pg_dir[PTRS_PER_PGD] __attribute__ ((__section__ (".data..vm0.pgd"), aligned(PAGE_SIZE))); +pte_t pg0[PT_INITIAL * PTRS_PER_PTE] __attribute__ ((__section__ (".data..vm0.pte"), aligned(PAGE_SIZE))); + +#ifdef CONFIG_DISCONTIGMEM +struct node_map_data node_data[MAX_NUMNODES] __read_mostly; +signed char pfnnid_map[PFNNID_MAP_MAX] __read_mostly; +#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] __read_mostly; + +/* The following array is initialized from the firmware specific + * information retrieved in kernel/inventory.c. + */ + +physmem_range_t pmem_ranges[MAX_PHYSMEM_RANGES] __read_mostly; +int npmem_ranges __read_mostly; + +#ifdef CONFIG_64BIT +#define MAX_MEM (~0UL) +#else /* !CONFIG_64BIT */ +#define MAX_MEM (3584U*1024U*1024U) +#endif /* !CONFIG_64BIT */ + +static unsigned long mem_limit __read_mostly = MAX_MEM; + +static void __init mem_limit_func(void) +{ + char *cp, *end; + unsigned long limit; + + /* We need this before __setup() functions are called */ + + limit = MAX_MEM; + for (cp = boot_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; + 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; + } + break; + } + 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 = &bootmem_node_data[i]; + } + memset(pfnnid_map, 0xff, sizeof(pfnnid_map)); + + for (i = 0; i < npmem_ranges; i++) { + node_set_state(i, N_NORMAL_MEMORY); + 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; + } + + /* IOMMU is always used to access "high mem" on those boxes + * that can support enough mem that a PCI device couldn't + * directly DMA to any physical addresses. + * ISA DMA support will need to revisit this. + */ + max_low_pfn = max_pfn; + + /* bootmap sizing messed up? */ + BUG_ON((bootmap_pfn - bootmap_start_pfn) != bootmap_pages); + + /* 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), BOOTMEM_DEFAULT); + reserve_bootmem_node(NODE_DATA(0), __pa(KERNEL_BINARY_TEXT_START), + (unsigned long)(_end - KERNEL_BINARY_TEXT_START), + BOOTMEM_DEFAULT); + reserve_bootmem_node(NODE_DATA(0), (bootmap_start_pfn << PAGE_SHIFT), + ((bootmap_pfn - bootmap_start_pfn) << PAGE_SHIFT), + BOOTMEM_DEFAULT); + +#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), + BOOTMEM_DEFAULT); + } +#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, BOOTMEM_DEFAULT); + } + } +#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); +} + +static int __init parisc_text_address(unsigned long vaddr) +{ + static unsigned long head_ptr __initdata; + + if (!head_ptr) + head_ptr = PAGE_MASK & (unsigned long) + dereference_function_descriptor(&parisc_kernel_start); + + return core_kernel_text(vaddr) || vaddr == head_ptr; +} + +static void __init map_pages(unsigned long start_vaddr, + unsigned long start_paddr, unsigned long size, + pgprot_t pgprot, int force) +{ + 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 vaddr; + 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; + vaddr = start_vaddr; + 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 (force) + pte = __mk_pte(address, pgprot); + else if (parisc_text_address(vaddr) && + address != fv_addr) + pte = __mk_pte(address, PAGE_KERNEL_EXEC); + else +#if defined(CONFIG_PARISC_PAGE_SIZE_4KB) + if (address >= ro_start && address < ro_end + && address != fv_addr + && address != gw_addr) + pte = __mk_pte(address, PAGE_KERNEL_RO); + else +#endif + pte = __mk_pte(address, pgprot); + + if (address >= end_paddr) { + if (force) + break; + else + pte_val(pte) = 0; + } + + set_pte(pg_table, pte); + + address += PAGE_SIZE; + vaddr += PAGE_SIZE; + } + start_pte = 0; + + if (address >= end_paddr) + break; + } + start_pmd = 0; + } +} + +void free_initmem(void) +{ + unsigned long init_begin = (unsigned long)__init_begin; + unsigned long init_end = (unsigned long)__init_end; + + /* The init text pages are marked R-X. We have to + * flush the icache and mark them RW- + * + * This is tricky, because map_pages is in the init section. + * Do a dummy remap of the data section first (the data + * section is already PAGE_KERNEL) to pull in the TLB entries + * for map_kernel */ + map_pages(init_begin, __pa(init_begin), init_end - init_begin, + PAGE_KERNEL_RWX, 1); + /* now remap at PAGE_KERNEL since the TLB is pre-primed to execute + * map_pages */ + map_pages(init_begin, __pa(init_begin), init_end - init_begin, + PAGE_KERNEL, 1); + + /* force the kernel to see the new TLB entries */ + __flush_tlb_range(0, init_begin, init_end); + /* Attempt to catch anyone trying to execute code here + * by filling the page with BRK insns. + */ + memset((void *)init_begin, 0x00, init_end - init_begin); + /* finally dump all the instructions which were cached, since the + * pages are no-longer executable */ + flush_icache_range(init_begin, init_end); + + free_initmem_default(-1); + + /* set up a new led state on systems shipped LED State panel */ + pdc_chassis_send_status(PDC_CHASSIS_DIRECT_BCOMPLETE); +} + + +#ifdef CONFIG_DEBUG_RODATA +void mark_rodata_ro(void) +{ + /* rodata memory was already mapped with KERNEL_RO access rights by + pagetable_init() and map_pages(). No need to do additional stuff here */ + printk (KERN_INFO "Write protecting the kernel read-only data: %luk\n", + (unsigned long)(__end_rodata - __start_rodata) >> 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 *parisc_vmalloc_start __read_mostly; +EXPORT_SYMBOL(parisc_vmalloc_start); + +#ifdef CONFIG_PA11 +unsigned long pcxl_dma_start __read_mostly; +#endif + +void __init mem_init(void) +{ + /* Do sanity checks on page table constants */ + BUILD_BUG_ON(PTE_ENTRY_SIZE != sizeof(pte_t)); + BUILD_BUG_ON(PMD_ENTRY_SIZE != sizeof(pmd_t)); + BUILD_BUG_ON(PGD_ENTRY_SIZE != sizeof(pgd_t)); + BUILD_BUG_ON(PAGE_SHIFT + BITS_PER_PTE + BITS_PER_PMD + BITS_PER_PGD + > BITS_PER_LONG); + + high_memory = __va((max_pfn << PAGE_SHIFT)); + set_max_mapnr(page_to_pfn(virt_to_page(high_memory - 1)) + 1); + free_all_bootmem(); + +#ifdef CONFIG_PA11 + if (hppa_dma_ops == &pcxl_dma_ops) { + pcxl_dma_start = (unsigned long)SET_MAP_OFFSET(MAP_START); + parisc_vmalloc_start = SET_MAP_OFFSET(pcxl_dma_start + + PCXL_DMA_MAP_SIZE); + } else { + pcxl_dma_start = 0; + parisc_vmalloc_start = SET_MAP_OFFSET(MAP_START); + } +#else + parisc_vmalloc_start = SET_MAP_OFFSET(MAP_START); +#endif + + mem_init_print_info(NULL); +#ifdef CONFIG_DEBUG_KERNEL /* double-sanity-check paranoia */ + printk("virtual kernel memory layout:\n" + " vmalloc : 0x%p - 0x%p (%4ld MB)\n" + " memory : 0x%p - 0x%p (%4ld MB)\n" + " .init : 0x%p - 0x%p (%4ld kB)\n" + " .data : 0x%p - 0x%p (%4ld kB)\n" + " .text : 0x%p - 0x%p (%4ld kB)\n", + + (void*)VMALLOC_START, (void*)VMALLOC_END, + (VMALLOC_END - VMALLOC_START) >> 20, + + __va(0), high_memory, + ((unsigned long)high_memory - (unsigned long)__va(0)) >> 20, + + __init_begin, __init_end, + ((unsigned long)__init_end - (unsigned long)__init_begin) >> 10, + + _etext, _edata, + ((unsigned long)_edata - (unsigned long)_etext) >> 10, + + _text, _etext, + ((unsigned long)_etext - (unsigned long)_text) >> 10); +#endif +} + +unsigned long *empty_zero_page __read_mostly; +EXPORT_SYMBOL(empty_zero_page); + +void show_mem(unsigned int filter) +{ + int total = 0,reserved = 0; + pg_data_t *pgdat; + + printk(KERN_INFO "Mem-info:\n"); + show_free_areas(filter); + + for_each_online_pgdat(pgdat) { + unsigned long flags; + int zoneid; + + pgdat_resize_lock(pgdat, &flags); + for (zoneid = 0; zoneid < MAX_NR_ZONES; zoneid++) { + struct zone *zone = &pgdat->node_zones[zoneid]; + if (!populated_zone(zone)) + continue; + + total += zone->present_pages; + reserved = zone->present_pages - zone->managed_pages; + } + pgdat_resize_unlock(pgdat, &flags); + } + + printk(KERN_INFO "%d pages of RAM\n", total); + printk(KERN_INFO "%d reserved pages\n", reserved); + +#ifdef CONFIG_DISCONTIGMEM + { + struct zonelist *zl; + int i, j; + + for (i = 0; i < npmem_ranges; i++) { + zl = node_zonelist(i, 0); + for (j = 0; j < MAX_NR_ZONES; j++) { + struct zoneref *z; + struct zone *zone; + + printk("Zone list for zone %d on node %d: ", j, i); + for_each_zone_zonelist(zone, z, zl, j) + printk("[%d/%s] ", zone_to_nid(zone), + zone->name); + printk("\n"); + } + } + } +#endif +} + +/* + * 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, 0); + } + +#ifdef CONFIG_BLK_DEV_INITRD + if (initrd_end && initrd_end > mem_limit) { + printk(KERN_INFO "initrd: mapping %08lx-%08lx\n", initrd_start, initrd_end); + map_pages(initrd_start, __pa(initrd_start), + initrd_end - initrd_start, PAGE_KERNEL, 0); + } +#endif + + empty_zero_page = alloc_bootmem_pages(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, 1); +} + +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(NULL); + + for (i = 0; i < npmem_ranges; i++) { + unsigned long zones_size[MAX_NR_ZONES] = { 0, }; + + zones_size[ZONE_NORMAL] = 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, zones_size, + pmem_ranges[i].start_pfn, NULL); + } +} + +#ifdef CONFIG_PA20 + +/* + * Currently, all PA20 chips have 18 bit protection IDs, 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 IDs and + * protection IDs. Older parisc chips (PCXS, PCXT, PCXL, PCXL2) only + * support 15 bit protection IDs, so that is the limiting factor. + * PCXT' has 18 bit protection IDs, 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); + } + BUG_ON(free_space_ids == 0); + } + + 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); + + BUG_ON(*dirty_space_offset & (1L << index)); /* 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; + +void flush_tlb_all(void) +{ + int do_recycle; + + __inc_irq_stat(irq_tlb_count); + do_recycle = 0; + spin_lock(&sid_lock); + if (dirty_space_ids > RECYCLE_THRESHOLD) { + BUG_ON(recycle_inuse); /* 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(flush_tlb_all_local, NULL, 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) +{ + __inc_irq_stat(irq_tlb_count); + spin_lock(&sid_lock); + flush_tlb_all_local(NULL); + recycle_sids(); + spin_unlock(&sid_lock); +} +#endif + +#ifdef CONFIG_BLK_DEV_INITRD +void free_initrd_mem(unsigned long start, unsigned long end) +{ + free_reserved_area((void *)start, (void *)end, -1, "initrd"); +} +#endif diff --git a/kernel/arch/parisc/mm/ioremap.c b/kernel/arch/parisc/mm/ioremap.c new file mode 100644 index 000000000..838d0259c --- /dev/null +++ b/kernel/arch/parisc/mm/ioremap.c @@ -0,0 +1,99 @@ +/* + * arch/parisc/mm/ioremap.c + * + * (C) Copyright 1995 1996 Linus Torvalds + * (C) Copyright 2001-2006 Helge Deller <deller@gmx.de> + * (C) Copyright 2005 Kyle McMartin <kyle@parisc-linux.org> + */ + +#include <linux/vmalloc.h> +#include <linux/errno.h> +#include <linux/module.h> +#include <linux/io.h> +#include <asm/pgalloc.h> + +/* + * Generic mapping function (not visible outside): + */ + +/* + * Remap an arbitrary physical address space into the kernel virtual + * address space. + * + * 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) +{ + void __iomem *addr; + struct vm_struct *area; + unsigned long offset, last_addr; + pgprot_t pgprot; + +#ifdef CONFIG_EISA + unsigned long end = phys_addr + size - 1; + /* Support EISA addresses */ + if ((phys_addr >= 0x00080000 && end < 0x000fffff) || + (phys_addr >= 0x00500000 && end < 0x03bfffff)) { + phys_addr |= F_EXTEND(0xfc000000); + flags |= _PAGE_NO_CACHE; + } +#endif + + /* 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; + } + } + + pgprot = __pgprot(_PAGE_PRESENT | _PAGE_RW | _PAGE_DIRTY | + _PAGE_ACCESSED | flags); + + /* + * Mappings have to be page-aligned + */ + offset = phys_addr & ~PAGE_MASK; + phys_addr &= PAGE_MASK; + size = PAGE_ALIGN(last_addr + 1) - phys_addr; + + /* + * Ok, go for it.. + */ + area = get_vm_area(size, VM_IOREMAP); + if (!area) + return NULL; + + addr = (void __iomem *) area->addr; + if (ioremap_page_range((unsigned long)addr, (unsigned long)addr + size, + phys_addr, pgprot)) { + vfree(addr); + return NULL; + } + + return (void __iomem *) (offset + (char __iomem *)addr); +} +EXPORT_SYMBOL(__ioremap); + +void iounmap(const volatile void __iomem *addr) +{ + if (addr > high_memory) + return vfree((void *) (PAGE_MASK & (unsigned long __force) addr)); +} +EXPORT_SYMBOL(iounmap); |