linux_old1/arch/s390/mm/vmem.c

401 lines
8.7 KiB
C
Raw Normal View History

/*
* arch/s390/mm/vmem.c
*
* Copyright IBM Corp. 2006
* Author(s): Heiko Carstens <heiko.carstens@de.ibm.com>
*/
#include <linux/bootmem.h>
#include <linux/pfn.h>
#include <linux/mm.h>
#include <linux/module.h>
#include <linux/list.h>
#include <asm/pgalloc.h>
#include <asm/pgtable.h>
#include <asm/setup.h>
#include <asm/tlbflush.h>
static DEFINE_MUTEX(vmem_mutex);
struct memory_segment {
struct list_head list;
unsigned long start;
unsigned long size;
};
static LIST_HEAD(mem_segs);
void __meminit memmap_init(unsigned long size, int nid, unsigned long zone,
unsigned long start_pfn)
{
struct page *start, *end;
struct page *map_start, *map_end;
int i;
start = pfn_to_page(start_pfn);
end = start + size;
for (i = 0; i < MEMORY_CHUNKS && memory_chunk[i].size > 0; i++) {
unsigned long cstart, cend;
cstart = PFN_DOWN(memory_chunk[i].addr);
cend = cstart + PFN_DOWN(memory_chunk[i].size);
map_start = mem_map + cstart;
map_end = mem_map + cend;
if (map_start < start)
map_start = start;
if (map_end > end)
map_end = end;
map_start -= ((unsigned long) map_start & (PAGE_SIZE - 1))
/ sizeof(struct page);
map_end += ((PFN_ALIGN((unsigned long) map_end)
- (unsigned long) map_end)
/ sizeof(struct page));
if (map_start < map_end)
memmap_init_zone((unsigned long)(map_end - map_start),
nid, zone, page_to_pfn(map_start),
MEMMAP_EARLY);
}
}
static void __ref *vmem_alloc_pages(unsigned int order)
{
if (slab_is_available())
return (void *)__get_free_pages(GFP_KERNEL, order);
return alloc_bootmem_pages((1 << order) * PAGE_SIZE);
}
#define vmem_pud_alloc() ({ BUG(); ((pud_t *) NULL); })
static inline pmd_t *vmem_pmd_alloc(void)
{
pmd_t *pmd = NULL;
#ifdef CONFIG_64BIT
pmd = vmem_alloc_pages(2);
if (!pmd)
return NULL;
clear_table((unsigned long *) pmd, _SEGMENT_ENTRY_EMPTY, PAGE_SIZE*4);
#endif
return pmd;
}
static pte_t __init_refok *vmem_pte_alloc(void)
{
pte_t *pte;
if (slab_is_available())
pte = (pte_t *) page_table_alloc(&init_mm);
else
pte = alloc_bootmem(PTRS_PER_PTE * sizeof(pte_t));
if (!pte)
return NULL;
clear_table((unsigned long *) pte, _PAGE_TYPE_EMPTY,
PTRS_PER_PTE * sizeof(pte_t));
return pte;
}
/*
* Add a physical memory range to the 1:1 mapping.
*/
static int vmem_add_range(unsigned long start, unsigned long size)
{
unsigned long address;
pgd_t *pg_dir;
pud_t *pu_dir;
pmd_t *pm_dir;
pte_t *pt_dir;
pte_t pte;
int ret = -ENOMEM;
for (address = start; address < start + size; address += PAGE_SIZE) {
pg_dir = pgd_offset_k(address);
if (pgd_none(*pg_dir)) {
pu_dir = vmem_pud_alloc();
if (!pu_dir)
goto out;
pgd_populate_kernel(&init_mm, pg_dir, pu_dir);
}
pu_dir = pud_offset(pg_dir, address);
if (pud_none(*pu_dir)) {
pm_dir = vmem_pmd_alloc();
if (!pm_dir)
goto out;
pud_populate_kernel(&init_mm, pu_dir, pm_dir);
}
pm_dir = pmd_offset(pu_dir, address);
if (pmd_none(*pm_dir)) {
pt_dir = vmem_pte_alloc();
if (!pt_dir)
goto out;
pmd_populate_kernel(&init_mm, pm_dir, pt_dir);
}
pt_dir = pte_offset_kernel(pm_dir, address);
pte = pfn_pte(address >> PAGE_SHIFT, PAGE_KERNEL);
[S390] noexec protection This provides a noexec protection on s390 hardware. Our hardware does not have any bits left in the pte for a hw noexec bit, so this is a different approach using shadow page tables and a special addressing mode that allows separate address spaces for code and data. As a special feature of our "secondary-space" addressing mode, separate page tables can be specified for the translation of data addresses (storage operands) and instruction addresses. The shadow page table is used for the instruction addresses and the standard page table for the data addresses. The shadow page table is linked to the standard page table by a pointer in page->lru.next of the struct page corresponding to the page that contains the standard page table (since page->private is not really private with the pte_lock and the page table pages are not in the LRU list). Depending on the software bits of a pte, it is either inserted into both page tables or just into the standard (data) page table. Pages of a vma that does not have the VM_EXEC bit set get mapped only in the data address space. Any try to execute code on such a page will cause a page translation exception. The standard reaction to this is a SIGSEGV with two exceptions: the two system call opcodes 0x0a77 (sys_sigreturn) and 0x0aad (sys_rt_sigreturn) are allowed. They are stored by the kernel to the signal stack frame. Unfortunately, the signal return mechanism cannot be modified to use an SA_RESTORER because the exception unwinding code depends on the system call opcode stored behind the signal stack frame. This feature requires that user space is executed in secondary-space mode and the kernel in home-space mode, which means that the addressing modes need to be switched and that the noexec protection only works for user space. After switching the addressing modes, we cannot use the mvcp/mvcs instructions anymore to copy between kernel and user space. A new mvcos instruction has been added to the z9 EC/BC hardware which allows to copy between arbitrary address spaces, but on older hardware the page tables need to be walked manually. Signed-off-by: Gerald Schaefer <geraldsc@de.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
2007-02-06 04:18:17 +08:00
*pt_dir = pte;
}
ret = 0;
out:
flush_tlb_kernel_range(start, start + size);
return ret;
}
/*
* Remove a physical memory range from the 1:1 mapping.
* Currently only invalidates page table entries.
*/
static void vmem_remove_range(unsigned long start, unsigned long size)
{
unsigned long address;
pgd_t *pg_dir;
pud_t *pu_dir;
pmd_t *pm_dir;
pte_t *pt_dir;
pte_t pte;
pte_val(pte) = _PAGE_TYPE_EMPTY;
for (address = start; address < start + size; address += PAGE_SIZE) {
pg_dir = pgd_offset_k(address);
pu_dir = pud_offset(pg_dir, address);
if (pud_none(*pu_dir))
continue;
pm_dir = pmd_offset(pu_dir, address);
if (pmd_none(*pm_dir))
continue;
pt_dir = pte_offset_kernel(pm_dir, address);
[S390] noexec protection This provides a noexec protection on s390 hardware. Our hardware does not have any bits left in the pte for a hw noexec bit, so this is a different approach using shadow page tables and a special addressing mode that allows separate address spaces for code and data. As a special feature of our "secondary-space" addressing mode, separate page tables can be specified for the translation of data addresses (storage operands) and instruction addresses. The shadow page table is used for the instruction addresses and the standard page table for the data addresses. The shadow page table is linked to the standard page table by a pointer in page->lru.next of the struct page corresponding to the page that contains the standard page table (since page->private is not really private with the pte_lock and the page table pages are not in the LRU list). Depending on the software bits of a pte, it is either inserted into both page tables or just into the standard (data) page table. Pages of a vma that does not have the VM_EXEC bit set get mapped only in the data address space. Any try to execute code on such a page will cause a page translation exception. The standard reaction to this is a SIGSEGV with two exceptions: the two system call opcodes 0x0a77 (sys_sigreturn) and 0x0aad (sys_rt_sigreturn) are allowed. They are stored by the kernel to the signal stack frame. Unfortunately, the signal return mechanism cannot be modified to use an SA_RESTORER because the exception unwinding code depends on the system call opcode stored behind the signal stack frame. This feature requires that user space is executed in secondary-space mode and the kernel in home-space mode, which means that the addressing modes need to be switched and that the noexec protection only works for user space. After switching the addressing modes, we cannot use the mvcp/mvcs instructions anymore to copy between kernel and user space. A new mvcos instruction has been added to the z9 EC/BC hardware which allows to copy between arbitrary address spaces, but on older hardware the page tables need to be walked manually. Signed-off-by: Gerald Schaefer <geraldsc@de.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
2007-02-06 04:18:17 +08:00
*pt_dir = pte;
}
flush_tlb_kernel_range(start, start + size);
}
/*
* Add a backed mem_map array to the virtual mem_map array.
*/
static int vmem_add_mem_map(unsigned long start, unsigned long size)
{
unsigned long address, start_addr, end_addr;
struct page *map_start, *map_end;
pgd_t *pg_dir;
pud_t *pu_dir;
pmd_t *pm_dir;
pte_t *pt_dir;
pte_t pte;
int ret = -ENOMEM;
map_start = VMEM_MAP + PFN_DOWN(start);
map_end = VMEM_MAP + PFN_DOWN(start + size);
start_addr = (unsigned long) map_start & PAGE_MASK;
end_addr = PFN_ALIGN((unsigned long) map_end);
for (address = start_addr; address < end_addr; address += PAGE_SIZE) {
pg_dir = pgd_offset_k(address);
if (pgd_none(*pg_dir)) {
pu_dir = vmem_pud_alloc();
if (!pu_dir)
goto out;
pgd_populate_kernel(&init_mm, pg_dir, pu_dir);
}
pu_dir = pud_offset(pg_dir, address);
if (pud_none(*pu_dir)) {
pm_dir = vmem_pmd_alloc();
if (!pm_dir)
goto out;
pud_populate_kernel(&init_mm, pu_dir, pm_dir);
}
pm_dir = pmd_offset(pu_dir, address);
if (pmd_none(*pm_dir)) {
pt_dir = vmem_pte_alloc();
if (!pt_dir)
goto out;
pmd_populate_kernel(&init_mm, pm_dir, pt_dir);
}
pt_dir = pte_offset_kernel(pm_dir, address);
if (pte_none(*pt_dir)) {
unsigned long new_page;
new_page =__pa(vmem_alloc_pages(0));
if (!new_page)
goto out;
pte = pfn_pte(new_page >> PAGE_SHIFT, PAGE_KERNEL);
[S390] noexec protection This provides a noexec protection on s390 hardware. Our hardware does not have any bits left in the pte for a hw noexec bit, so this is a different approach using shadow page tables and a special addressing mode that allows separate address spaces for code and data. As a special feature of our "secondary-space" addressing mode, separate page tables can be specified for the translation of data addresses (storage operands) and instruction addresses. The shadow page table is used for the instruction addresses and the standard page table for the data addresses. The shadow page table is linked to the standard page table by a pointer in page->lru.next of the struct page corresponding to the page that contains the standard page table (since page->private is not really private with the pte_lock and the page table pages are not in the LRU list). Depending on the software bits of a pte, it is either inserted into both page tables or just into the standard (data) page table. Pages of a vma that does not have the VM_EXEC bit set get mapped only in the data address space. Any try to execute code on such a page will cause a page translation exception. The standard reaction to this is a SIGSEGV with two exceptions: the two system call opcodes 0x0a77 (sys_sigreturn) and 0x0aad (sys_rt_sigreturn) are allowed. They are stored by the kernel to the signal stack frame. Unfortunately, the signal return mechanism cannot be modified to use an SA_RESTORER because the exception unwinding code depends on the system call opcode stored behind the signal stack frame. This feature requires that user space is executed in secondary-space mode and the kernel in home-space mode, which means that the addressing modes need to be switched and that the noexec protection only works for user space. After switching the addressing modes, we cannot use the mvcp/mvcs instructions anymore to copy between kernel and user space. A new mvcos instruction has been added to the z9 EC/BC hardware which allows to copy between arbitrary address spaces, but on older hardware the page tables need to be walked manually. Signed-off-by: Gerald Schaefer <geraldsc@de.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
2007-02-06 04:18:17 +08:00
*pt_dir = pte;
}
}
ret = 0;
out:
flush_tlb_kernel_range(start_addr, end_addr);
return ret;
}
static int vmem_add_mem(unsigned long start, unsigned long size)
{
int ret;
ret = vmem_add_mem_map(start, size);
if (ret)
return ret;
return vmem_add_range(start, size);
}
/*
* Add memory segment to the segment list if it doesn't overlap with
* an already present segment.
*/
static int insert_memory_segment(struct memory_segment *seg)
{
struct memory_segment *tmp;
if (seg->start + seg->size >= VMEM_MAX_PHYS ||
seg->start + seg->size < seg->start)
return -ERANGE;
list_for_each_entry(tmp, &mem_segs, list) {
if (seg->start >= tmp->start + tmp->size)
continue;
if (seg->start + seg->size <= tmp->start)
continue;
return -ENOSPC;
}
list_add(&seg->list, &mem_segs);
return 0;
}
/*
* Remove memory segment from the segment list.
*/
static void remove_memory_segment(struct memory_segment *seg)
{
list_del(&seg->list);
}
static void __remove_shared_memory(struct memory_segment *seg)
{
remove_memory_segment(seg);
vmem_remove_range(seg->start, seg->size);
}
int remove_shared_memory(unsigned long start, unsigned long size)
{
struct memory_segment *seg;
int ret;
mutex_lock(&vmem_mutex);
ret = -ENOENT;
list_for_each_entry(seg, &mem_segs, list) {
if (seg->start == start && seg->size == size)
break;
}
if (seg->start != start || seg->size != size)
goto out;
ret = 0;
__remove_shared_memory(seg);
kfree(seg);
out:
mutex_unlock(&vmem_mutex);
return ret;
}
int add_shared_memory(unsigned long start, unsigned long size)
{
struct memory_segment *seg;
struct page *page;
unsigned long pfn, num_pfn, end_pfn;
int ret;
mutex_lock(&vmem_mutex);
ret = -ENOMEM;
seg = kzalloc(sizeof(*seg), GFP_KERNEL);
if (!seg)
goto out;
seg->start = start;
seg->size = size;
ret = insert_memory_segment(seg);
if (ret)
goto out_free;
ret = vmem_add_mem(start, size);
if (ret)
goto out_remove;
pfn = PFN_DOWN(start);
num_pfn = PFN_DOWN(size);
end_pfn = pfn + num_pfn;
page = pfn_to_page(pfn);
memset(page, 0, num_pfn * sizeof(struct page));
for (; pfn < end_pfn; pfn++) {
page = pfn_to_page(pfn);
init_page_count(page);
reset_page_mapcount(page);
SetPageReserved(page);
INIT_LIST_HEAD(&page->lru);
}
goto out;
out_remove:
__remove_shared_memory(seg);
out_free:
kfree(seg);
out:
mutex_unlock(&vmem_mutex);
return ret;
}
/*
* map whole physical memory to virtual memory (identity mapping)
* we reserve enough space in the vmalloc area for vmemmap to hotplug
* additional memory segments.
*/
void __init vmem_map_init(void)
{
int i;
INIT_LIST_HEAD(&init_mm.context.crst_list);
INIT_LIST_HEAD(&init_mm.context.pgtable_list);
init_mm.context.noexec = 0;
NODE_DATA(0)->node_mem_map = VMEM_MAP;
for (i = 0; i < MEMORY_CHUNKS && memory_chunk[i].size > 0; i++)
vmem_add_mem(memory_chunk[i].addr, memory_chunk[i].size);
}
/*
* Convert memory chunk array to a memory segment list so there is a single
* list that contains both r/w memory and shared memory segments.
*/
static int __init vmem_convert_memory_chunk(void)
{
struct memory_segment *seg;
int i;
mutex_lock(&vmem_mutex);
for (i = 0; i < MEMORY_CHUNKS; i++) {
if (!memory_chunk[i].size)
continue;
seg = kzalloc(sizeof(*seg), GFP_KERNEL);
if (!seg)
panic("Out of memory...\n");
seg->start = memory_chunk[i].addr;
seg->size = memory_chunk[i].size;
insert_memory_segment(seg);
}
mutex_unlock(&vmem_mutex);
return 0;
}
core_initcall(vmem_convert_memory_chunk);