2008-01-30 12:33:41 +00:00
|
|
|
/*
|
|
|
|
* Copyright 2002 Andi Kleen, SuSE Labs.
|
2005-04-16 22:20:36 +00:00
|
|
|
* Thanks to Ben LaHaise for precious feedback.
|
2008-01-30 12:33:41 +00:00
|
|
|
*/
|
2005-04-16 22:20:36 +00:00
|
|
|
#include <linux/highmem.h>
|
2008-01-30 12:34:04 +00:00
|
|
|
#include <linux/bootmem.h>
|
2005-04-16 22:20:36 +00:00
|
|
|
#include <linux/module.h>
|
2008-01-30 12:33:41 +00:00
|
|
|
#include <linux/sched.h>
|
2005-04-16 22:20:36 +00:00
|
|
|
#include <linux/slab.h>
|
2008-01-30 12:33:41 +00:00
|
|
|
#include <linux/mm.h>
|
2008-02-09 22:24:09 +00:00
|
|
|
#include <linux/interrupt.h>
|
2008-01-30 12:33:41 +00:00
|
|
|
|
2008-01-30 12:34:06 +00:00
|
|
|
#include <asm/e820.h>
|
2005-04-16 22:20:36 +00:00
|
|
|
#include <asm/processor.h>
|
|
|
|
#include <asm/tlbflush.h>
|
2006-01-06 08:12:10 +00:00
|
|
|
#include <asm/sections.h>
|
2008-01-30 12:33:41 +00:00
|
|
|
#include <asm/uaccess.h>
|
|
|
|
#include <asm/pgalloc.h>
|
2005-04-16 22:20:36 +00:00
|
|
|
|
2008-02-04 15:48:09 +00:00
|
|
|
/*
|
|
|
|
* The current flushing context - we pass it instead of 5 arguments:
|
|
|
|
*/
|
2008-02-04 15:48:07 +00:00
|
|
|
struct cpa_data {
|
|
|
|
unsigned long vaddr;
|
|
|
|
pgprot_t mask_set;
|
|
|
|
pgprot_t mask_clr;
|
2008-02-04 15:48:07 +00:00
|
|
|
int numpages;
|
2008-02-04 15:48:07 +00:00
|
|
|
int flushtlb;
|
2008-02-04 15:48:07 +00:00
|
|
|
};
|
|
|
|
|
2008-01-30 12:34:04 +00:00
|
|
|
static inline int
|
|
|
|
within(unsigned long addr, unsigned long start, unsigned long end)
|
2008-01-30 12:34:04 +00:00
|
|
|
{
|
2008-01-30 12:34:04 +00:00
|
|
|
return addr >= start && addr < end;
|
|
|
|
}
|
|
|
|
|
2008-01-30 12:34:07 +00:00
|
|
|
/*
|
|
|
|
* Flushing functions
|
|
|
|
*/
|
2008-01-30 12:34:08 +00:00
|
|
|
|
|
|
|
/**
|
|
|
|
* clflush_cache_range - flush a cache range with clflush
|
|
|
|
* @addr: virtual start address
|
|
|
|
* @size: number of bytes to flush
|
|
|
|
*
|
|
|
|
* clflush is an unordered instruction which needs fencing with mfence
|
|
|
|
* to avoid ordering issues.
|
|
|
|
*/
|
2008-01-30 12:34:09 +00:00
|
|
|
void clflush_cache_range(void *vaddr, unsigned int size)
|
2008-01-30 12:34:07 +00:00
|
|
|
{
|
2008-01-30 12:34:09 +00:00
|
|
|
void *vend = vaddr + size - 1;
|
2008-01-30 12:34:07 +00:00
|
|
|
|
2008-01-30 12:34:08 +00:00
|
|
|
mb();
|
2008-01-30 12:34:09 +00:00
|
|
|
|
|
|
|
for (; vaddr < vend; vaddr += boot_cpu_data.x86_clflush_size)
|
|
|
|
clflush(vaddr);
|
|
|
|
/*
|
|
|
|
* Flush any possible final partial cacheline:
|
|
|
|
*/
|
|
|
|
clflush(vend);
|
|
|
|
|
2008-01-30 12:34:08 +00:00
|
|
|
mb();
|
2008-01-30 12:34:07 +00:00
|
|
|
}
|
|
|
|
|
2008-01-30 12:34:08 +00:00
|
|
|
static void __cpa_flush_all(void *arg)
|
2008-01-30 12:34:07 +00:00
|
|
|
{
|
2008-02-04 15:48:06 +00:00
|
|
|
unsigned long cache = (unsigned long)arg;
|
|
|
|
|
2008-01-30 12:34:07 +00:00
|
|
|
/*
|
|
|
|
* Flush all to work around Errata in early athlons regarding
|
|
|
|
* large page flushing.
|
|
|
|
*/
|
|
|
|
__flush_tlb_all();
|
|
|
|
|
2008-02-04 15:48:06 +00:00
|
|
|
if (cache && boot_cpu_data.x86_model >= 4)
|
2008-01-30 12:34:07 +00:00
|
|
|
wbinvd();
|
|
|
|
}
|
|
|
|
|
2008-02-04 15:48:06 +00:00
|
|
|
static void cpa_flush_all(unsigned long cache)
|
2008-01-30 12:34:07 +00:00
|
|
|
{
|
|
|
|
BUG_ON(irqs_disabled());
|
|
|
|
|
2008-02-04 15:48:06 +00:00
|
|
|
on_each_cpu(__cpa_flush_all, (void *) cache, 1, 1);
|
2008-01-30 12:34:07 +00:00
|
|
|
}
|
|
|
|
|
2008-01-30 12:34:08 +00:00
|
|
|
static void __cpa_flush_range(void *arg)
|
|
|
|
{
|
|
|
|
/*
|
|
|
|
* We could optimize that further and do individual per page
|
|
|
|
* tlb invalidates for a low number of pages. Caveat: we must
|
|
|
|
* flush the high aliases on 64bit as well.
|
|
|
|
*/
|
|
|
|
__flush_tlb_all();
|
|
|
|
}
|
|
|
|
|
2008-02-04 15:48:06 +00:00
|
|
|
static void cpa_flush_range(unsigned long start, int numpages, int cache)
|
2008-01-30 12:34:08 +00:00
|
|
|
{
|
2008-01-30 12:34:09 +00:00
|
|
|
unsigned int i, level;
|
|
|
|
unsigned long addr;
|
|
|
|
|
2008-01-30 12:34:08 +00:00
|
|
|
BUG_ON(irqs_disabled());
|
2008-01-30 12:34:09 +00:00
|
|
|
WARN_ON(PAGE_ALIGN(start) != start);
|
2008-01-30 12:34:08 +00:00
|
|
|
|
2008-01-30 12:34:08 +00:00
|
|
|
on_each_cpu(__cpa_flush_range, NULL, 1, 1);
|
2008-01-30 12:34:08 +00:00
|
|
|
|
2008-02-04 15:48:06 +00:00
|
|
|
if (!cache)
|
|
|
|
return;
|
|
|
|
|
2008-01-30 12:34:08 +00:00
|
|
|
/*
|
|
|
|
* We only need to flush on one CPU,
|
|
|
|
* clflush is a MESI-coherent instruction that
|
|
|
|
* will cause all other CPUs to flush the same
|
|
|
|
* cachelines:
|
|
|
|
*/
|
2008-01-30 12:34:09 +00:00
|
|
|
for (i = 0, addr = start; i < numpages; i++, addr += PAGE_SIZE) {
|
|
|
|
pte_t *pte = lookup_address(addr, &level);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Only flush present addresses:
|
|
|
|
*/
|
2008-02-04 15:48:08 +00:00
|
|
|
if (pte && (pte_val(*pte) & _PAGE_PRESENT))
|
2008-01-30 12:34:09 +00:00
|
|
|
clflush_cache_range((void *) addr, PAGE_SIZE);
|
|
|
|
}
|
2008-01-30 12:34:08 +00:00
|
|
|
}
|
|
|
|
|
2008-02-04 15:48:05 +00:00
|
|
|
#define HIGH_MAP_START __START_KERNEL_map
|
|
|
|
#define HIGH_MAP_END (__START_KERNEL_map + KERNEL_TEXT_SIZE)
|
|
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Converts a virtual address to a X86-64 highmap address
|
|
|
|
*/
|
|
|
|
static unsigned long virt_to_highmap(void *address)
|
|
|
|
{
|
|
|
|
#ifdef CONFIG_X86_64
|
|
|
|
return __pa((unsigned long)address) + HIGH_MAP_START - phys_base;
|
|
|
|
#else
|
|
|
|
return (unsigned long)address;
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
2008-01-30 12:34:04 +00:00
|
|
|
/*
|
|
|
|
* Certain areas of memory on x86 require very specific protection flags,
|
|
|
|
* for example the BIOS area or kernel text. Callers don't always get this
|
|
|
|
* right (again, ioremap() on BIOS memory is not uncommon) so this function
|
|
|
|
* checks and fixes these known static required protection bits.
|
|
|
|
*/
|
|
|
|
static inline pgprot_t static_protections(pgprot_t prot, unsigned long address)
|
|
|
|
{
|
|
|
|
pgprot_t forbidden = __pgprot(0);
|
|
|
|
|
2008-01-30 12:34:04 +00:00
|
|
|
/*
|
2008-01-30 12:34:04 +00:00
|
|
|
* The BIOS area between 640k and 1Mb needs to be executable for
|
|
|
|
* PCI BIOS based config access (CONFIG_PCI_GOBIOS) support.
|
2008-01-30 12:34:04 +00:00
|
|
|
*/
|
2008-01-30 12:34:04 +00:00
|
|
|
if (within(__pa(address), BIOS_BEGIN, BIOS_END))
|
|
|
|
pgprot_val(forbidden) |= _PAGE_NX;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* The kernel text needs to be executable for obvious reasons
|
|
|
|
* Does not cover __inittext since that is gone later on
|
|
|
|
*/
|
|
|
|
if (within(address, (unsigned long)_text, (unsigned long)_etext))
|
|
|
|
pgprot_val(forbidden) |= _PAGE_NX;
|
2008-02-04 15:48:05 +00:00
|
|
|
/*
|
|
|
|
* Do the same for the x86-64 high kernel mapping
|
|
|
|
*/
|
|
|
|
if (within(address, virt_to_highmap(_text), virt_to_highmap(_etext)))
|
|
|
|
pgprot_val(forbidden) |= _PAGE_NX;
|
|
|
|
|
2008-01-30 12:34:04 +00:00
|
|
|
/* The .rodata section needs to be read-only */
|
|
|
|
if (within(address, (unsigned long)__start_rodata,
|
|
|
|
(unsigned long)__end_rodata))
|
|
|
|
pgprot_val(forbidden) |= _PAGE_RW;
|
2008-02-04 15:48:05 +00:00
|
|
|
/*
|
|
|
|
* Do the same for the x86-64 high kernel mapping
|
|
|
|
*/
|
|
|
|
if (within(address, virt_to_highmap(__start_rodata),
|
|
|
|
virt_to_highmap(__end_rodata)))
|
|
|
|
pgprot_val(forbidden) |= _PAGE_RW;
|
2008-01-30 12:34:04 +00:00
|
|
|
|
|
|
|
prot = __pgprot(pgprot_val(prot) & ~pgprot_val(forbidden));
|
2008-01-30 12:34:04 +00:00
|
|
|
|
|
|
|
return prot;
|
|
|
|
}
|
|
|
|
|
2008-02-04 15:48:07 +00:00
|
|
|
/*
|
|
|
|
* Lookup the page table entry for a virtual address. Return a pointer
|
|
|
|
* to the entry and the level of the mapping.
|
|
|
|
*
|
|
|
|
* Note: We return pud and pmd either when the entry is marked large
|
|
|
|
* or when the present bit is not set. Otherwise we would return a
|
|
|
|
* pointer to a nonexisting mapping.
|
|
|
|
*/
|
2008-02-09 22:24:08 +00:00
|
|
|
pte_t *lookup_address(unsigned long address, unsigned int *level)
|
2008-01-30 12:33:41 +00:00
|
|
|
{
|
2005-04-16 22:20:36 +00:00
|
|
|
pgd_t *pgd = pgd_offset_k(address);
|
|
|
|
pud_t *pud;
|
|
|
|
pmd_t *pmd;
|
2008-01-30 12:33:41 +00:00
|
|
|
|
2008-01-30 12:34:04 +00:00
|
|
|
*level = PG_LEVEL_NONE;
|
|
|
|
|
2005-04-16 22:20:36 +00:00
|
|
|
if (pgd_none(*pgd))
|
|
|
|
return NULL;
|
2008-02-04 15:48:09 +00:00
|
|
|
|
2005-04-16 22:20:36 +00:00
|
|
|
pud = pud_offset(pgd, address);
|
|
|
|
if (pud_none(*pud))
|
|
|
|
return NULL;
|
2008-02-04 15:48:09 +00:00
|
|
|
|
|
|
|
*level = PG_LEVEL_1G;
|
|
|
|
if (pud_large(*pud) || !pud_present(*pud))
|
|
|
|
return (pte_t *)pud;
|
|
|
|
|
2005-04-16 22:20:36 +00:00
|
|
|
pmd = pmd_offset(pud, address);
|
|
|
|
if (pmd_none(*pmd))
|
|
|
|
return NULL;
|
2008-01-30 12:34:04 +00:00
|
|
|
|
|
|
|
*level = PG_LEVEL_2M;
|
2008-02-04 15:48:07 +00:00
|
|
|
if (pmd_large(*pmd) || !pmd_present(*pmd))
|
2005-04-16 22:20:36 +00:00
|
|
|
return (pte_t *)pmd;
|
|
|
|
|
2008-01-30 12:34:04 +00:00
|
|
|
*level = PG_LEVEL_4K;
|
2008-02-04 15:48:09 +00:00
|
|
|
|
2008-01-30 12:33:41 +00:00
|
|
|
return pte_offset_kernel(pmd, address);
|
|
|
|
}
|
|
|
|
|
2008-02-04 15:48:09 +00:00
|
|
|
/*
|
|
|
|
* Set the new pmd in all the pgds we know about:
|
|
|
|
*/
|
2008-01-30 12:33:57 +00:00
|
|
|
static void __set_pmd_pte(pte_t *kpte, unsigned long address, pte_t pte)
|
2008-01-30 12:33:41 +00:00
|
|
|
{
|
|
|
|
/* change init_mm */
|
|
|
|
set_pte_atomic(kpte, pte);
|
2008-01-30 12:34:03 +00:00
|
|
|
#ifdef CONFIG_X86_32
|
2008-01-30 12:34:04 +00:00
|
|
|
if (!SHARED_KERNEL_PMD) {
|
2008-01-30 12:34:03 +00:00
|
|
|
struct page *page;
|
|
|
|
|
2008-01-30 12:34:11 +00:00
|
|
|
list_for_each_entry(page, &pgd_list, lru) {
|
2008-01-30 12:34:03 +00:00
|
|
|
pgd_t *pgd;
|
|
|
|
pud_t *pud;
|
|
|
|
pmd_t *pmd;
|
|
|
|
|
|
|
|
pgd = (pgd_t *)page_address(page) + pgd_index(address);
|
|
|
|
pud = pud_offset(pgd, address);
|
|
|
|
pmd = pmd_offset(pud, address);
|
|
|
|
set_pte_atomic((pte_t *)pmd, pte);
|
|
|
|
}
|
2005-04-16 22:20:36 +00:00
|
|
|
}
|
2008-01-30 12:34:03 +00:00
|
|
|
#endif
|
2005-04-16 22:20:36 +00:00
|
|
|
}
|
|
|
|
|
2008-02-04 15:48:09 +00:00
|
|
|
static int
|
|
|
|
try_preserve_large_page(pte_t *kpte, unsigned long address,
|
|
|
|
struct cpa_data *cpa)
|
2008-02-04 15:48:07 +00:00
|
|
|
{
|
|
|
|
unsigned long nextpage_addr, numpages, pmask, psize, flags;
|
|
|
|
pte_t new_pte, old_pte, *tmp;
|
|
|
|
pgprot_t old_prot, new_prot;
|
2008-02-09 22:24:08 +00:00
|
|
|
int do_split = 1;
|
|
|
|
unsigned int level;
|
2008-02-04 15:48:07 +00:00
|
|
|
|
|
|
|
spin_lock_irqsave(&pgd_lock, flags);
|
|
|
|
/*
|
|
|
|
* Check for races, another CPU might have split this page
|
|
|
|
* up already:
|
|
|
|
*/
|
|
|
|
tmp = lookup_address(address, &level);
|
|
|
|
if (tmp != kpte)
|
|
|
|
goto out_unlock;
|
|
|
|
|
|
|
|
switch (level) {
|
|
|
|
case PG_LEVEL_2M:
|
2008-02-04 15:48:08 +00:00
|
|
|
psize = PMD_PAGE_SIZE;
|
|
|
|
pmask = PMD_PAGE_MASK;
|
2008-02-04 15:48:07 +00:00
|
|
|
break;
|
2008-02-04 15:48:09 +00:00
|
|
|
#ifdef CONFIG_X86_64
|
2008-02-04 15:48:07 +00:00
|
|
|
case PG_LEVEL_1G:
|
2008-02-04 15:48:09 +00:00
|
|
|
psize = PMD_PAGE_SIZE;
|
|
|
|
pmask = PMD_PAGE_MASK;
|
|
|
|
break;
|
|
|
|
#endif
|
2008-02-04 15:48:07 +00:00
|
|
|
default:
|
2008-02-04 15:48:09 +00:00
|
|
|
do_split = -EINVAL;
|
2008-02-04 15:48:07 +00:00
|
|
|
goto out_unlock;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Calculate the number of pages, which fit into this large
|
|
|
|
* page starting at address:
|
|
|
|
*/
|
|
|
|
nextpage_addr = (address + psize) & pmask;
|
|
|
|
numpages = (nextpage_addr - address) >> PAGE_SHIFT;
|
|
|
|
if (numpages < cpa->numpages)
|
|
|
|
cpa->numpages = numpages;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* We are safe now. Check whether the new pgprot is the same:
|
|
|
|
*/
|
|
|
|
old_pte = *kpte;
|
|
|
|
old_prot = new_prot = pte_pgprot(old_pte);
|
|
|
|
|
|
|
|
pgprot_val(new_prot) &= ~pgprot_val(cpa->mask_clr);
|
|
|
|
pgprot_val(new_prot) |= pgprot_val(cpa->mask_set);
|
|
|
|
new_prot = static_protections(new_prot, address);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* If there are no changes, return. maxpages has been updated
|
|
|
|
* above:
|
|
|
|
*/
|
|
|
|
if (pgprot_val(new_prot) == pgprot_val(old_prot)) {
|
2008-02-04 15:48:09 +00:00
|
|
|
do_split = 0;
|
2008-02-04 15:48:07 +00:00
|
|
|
goto out_unlock;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* We need to change the attributes. Check, whether we can
|
|
|
|
* change the large page in one go. We request a split, when
|
|
|
|
* the address is not aligned and the number of pages is
|
|
|
|
* smaller than the number of pages in the large page. Note
|
|
|
|
* that we limited the number of possible pages already to
|
|
|
|
* the number of pages in the large page.
|
|
|
|
*/
|
|
|
|
if (address == (nextpage_addr - psize) && cpa->numpages == numpages) {
|
|
|
|
/*
|
|
|
|
* The address is aligned and the number of pages
|
|
|
|
* covers the full page.
|
|
|
|
*/
|
|
|
|
new_pte = pfn_pte(pte_pfn(old_pte), canon_pgprot(new_prot));
|
|
|
|
__set_pmd_pte(kpte, address, new_pte);
|
|
|
|
cpa->flushtlb = 1;
|
2008-02-04 15:48:09 +00:00
|
|
|
do_split = 0;
|
2008-02-04 15:48:07 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
out_unlock:
|
|
|
|
spin_unlock_irqrestore(&pgd_lock, flags);
|
2008-02-04 15:48:09 +00:00
|
|
|
|
2008-02-04 15:48:09 +00:00
|
|
|
return do_split;
|
2008-02-04 15:48:07 +00:00
|
|
|
}
|
|
|
|
|
2008-02-09 22:24:09 +00:00
|
|
|
static LIST_HEAD(page_pool);
|
|
|
|
static unsigned long pool_size, pool_pages, pool_low;
|
|
|
|
static unsigned long pool_used, pool_failed, pool_refill;
|
|
|
|
|
|
|
|
static void cpa_fill_pool(void)
|
|
|
|
{
|
|
|
|
struct page *p;
|
|
|
|
gfp_t gfp = GFP_KERNEL;
|
|
|
|
|
|
|
|
/* Do not allocate from interrupt context */
|
|
|
|
if (in_irq() || irqs_disabled())
|
|
|
|
return;
|
|
|
|
/*
|
|
|
|
* Check unlocked. I does not matter when we have one more
|
|
|
|
* page in the pool. The bit lock avoids recursive pool
|
|
|
|
* allocations:
|
|
|
|
*/
|
|
|
|
if (pool_pages >= pool_size || test_and_set_bit_lock(0, &pool_refill))
|
|
|
|
return;
|
|
|
|
|
|
|
|
#ifdef CONFIG_DEBUG_PAGEALLOC
|
|
|
|
/*
|
|
|
|
* We could do:
|
|
|
|
* gfp = in_atomic() ? GFP_ATOMIC : GFP_KERNEL;
|
|
|
|
* but this fails on !PREEMPT kernels
|
|
|
|
*/
|
|
|
|
gfp = GFP_ATOMIC | __GFP_NORETRY | __GFP_NOWARN;
|
|
|
|
#endif
|
|
|
|
|
|
|
|
while (pool_pages < pool_size) {
|
|
|
|
p = alloc_pages(gfp, 0);
|
|
|
|
if (!p) {
|
|
|
|
pool_failed++;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
spin_lock_irq(&pgd_lock);
|
|
|
|
list_add(&p->lru, &page_pool);
|
|
|
|
pool_pages++;
|
|
|
|
spin_unlock_irq(&pgd_lock);
|
|
|
|
}
|
|
|
|
clear_bit_unlock(0, &pool_refill);
|
|
|
|
}
|
|
|
|
|
|
|
|
#define SHIFT_MB (20 - PAGE_SHIFT)
|
|
|
|
#define ROUND_MB_GB ((1 << 10) - 1)
|
|
|
|
#define SHIFT_MB_GB 10
|
|
|
|
#define POOL_PAGES_PER_GB 16
|
|
|
|
|
|
|
|
void __init cpa_init(void)
|
|
|
|
{
|
|
|
|
struct sysinfo si;
|
|
|
|
unsigned long gb;
|
|
|
|
|
|
|
|
si_meminfo(&si);
|
|
|
|
/*
|
|
|
|
* Calculate the number of pool pages:
|
|
|
|
*
|
|
|
|
* Convert totalram (nr of pages) to MiB and round to the next
|
|
|
|
* GiB. Shift MiB to Gib and multiply the result by
|
|
|
|
* POOL_PAGES_PER_GB:
|
|
|
|
*/
|
|
|
|
gb = ((si.totalram >> SHIFT_MB) + ROUND_MB_GB) >> SHIFT_MB_GB;
|
|
|
|
pool_size = POOL_PAGES_PER_GB * gb;
|
|
|
|
pool_low = pool_size;
|
|
|
|
|
|
|
|
cpa_fill_pool();
|
|
|
|
printk(KERN_DEBUG
|
|
|
|
"CPA: page pool initialized %lu of %lu pages preallocated\n",
|
|
|
|
pool_pages, pool_size);
|
|
|
|
}
|
|
|
|
|
2008-01-30 12:33:57 +00:00
|
|
|
static int split_large_page(pte_t *kpte, unsigned long address)
|
2008-01-30 12:33:56 +00:00
|
|
|
{
|
2008-02-04 15:48:10 +00:00
|
|
|
unsigned long flags, pfn, pfninc = 1;
|
2008-02-04 15:48:09 +00:00
|
|
|
gfp_t gfp_flags = GFP_KERNEL;
|
|
|
|
unsigned int i, level;
|
2008-01-30 12:33:56 +00:00
|
|
|
pte_t *pbase, *tmp;
|
2008-02-04 15:48:09 +00:00
|
|
|
pgprot_t ref_prot;
|
2008-01-30 12:33:56 +00:00
|
|
|
struct page *base;
|
|
|
|
|
2008-01-30 12:33:58 +00:00
|
|
|
#ifdef CONFIG_DEBUG_PAGEALLOC
|
2008-01-30 12:34:09 +00:00
|
|
|
gfp_flags = GFP_ATOMIC | __GFP_NOWARN;
|
2008-01-30 12:33:58 +00:00
|
|
|
#endif
|
|
|
|
base = alloc_pages(gfp_flags, 0);
|
2008-01-30 12:33:56 +00:00
|
|
|
if (!base)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
2008-01-30 12:33:57 +00:00
|
|
|
spin_lock_irqsave(&pgd_lock, flags);
|
2008-01-30 12:33:56 +00:00
|
|
|
/*
|
|
|
|
* Check for races, another CPU might have split this page
|
|
|
|
* up for us already:
|
|
|
|
*/
|
|
|
|
tmp = lookup_address(address, &level);
|
2008-02-04 15:48:08 +00:00
|
|
|
if (tmp != kpte)
|
2008-01-30 12:33:56 +00:00
|
|
|
goto out_unlock;
|
|
|
|
|
|
|
|
pbase = (pte_t *)page_address(base);
|
2008-01-30 12:34:03 +00:00
|
|
|
#ifdef CONFIG_X86_32
|
2008-01-30 12:33:56 +00:00
|
|
|
paravirt_alloc_pt(&init_mm, page_to_pfn(base));
|
2008-01-30 12:34:03 +00:00
|
|
|
#endif
|
2008-02-04 15:48:08 +00:00
|
|
|
ref_prot = pte_pgprot(pte_clrhuge(*kpte));
|
2008-01-30 12:33:56 +00:00
|
|
|
|
2008-02-04 15:48:09 +00:00
|
|
|
#ifdef CONFIG_X86_64
|
|
|
|
if (level == PG_LEVEL_1G) {
|
|
|
|
pfninc = PMD_PAGE_SIZE >> PAGE_SHIFT;
|
|
|
|
pgprot_val(ref_prot) |= _PAGE_PSE;
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
2008-02-04 15:48:05 +00:00
|
|
|
/*
|
|
|
|
* Get the target pfn from the original entry:
|
|
|
|
*/
|
|
|
|
pfn = pte_pfn(*kpte);
|
2008-02-04 15:48:09 +00:00
|
|
|
for (i = 0; i < PTRS_PER_PTE; i++, pfn += pfninc)
|
2008-02-04 15:48:05 +00:00
|
|
|
set_pte(&pbase[i], pfn_pte(pfn, ref_prot));
|
2008-01-30 12:33:56 +00:00
|
|
|
|
|
|
|
/*
|
2008-02-04 15:48:08 +00:00
|
|
|
* Install the new, split up pagetable. Important details here:
|
2008-01-30 12:34:04 +00:00
|
|
|
*
|
|
|
|
* On Intel the NX bit of all levels must be cleared to make a
|
|
|
|
* page executable. See section 4.13.2 of Intel 64 and IA-32
|
|
|
|
* Architectures Software Developer's Manual).
|
2008-02-04 15:48:08 +00:00
|
|
|
*
|
|
|
|
* Mark the entry present. The current mapping might be
|
|
|
|
* set to not present, which we preserved above.
|
2008-01-30 12:33:56 +00:00
|
|
|
*/
|
2008-01-30 12:34:04 +00:00
|
|
|
ref_prot = pte_pgprot(pte_mkexec(pte_clrhuge(*kpte)));
|
2008-02-04 15:48:08 +00:00
|
|
|
pgprot_val(ref_prot) |= _PAGE_PRESENT;
|
2008-01-30 12:33:57 +00:00
|
|
|
__set_pmd_pte(kpte, address, mk_pte(base, ref_prot));
|
2008-01-30 12:33:56 +00:00
|
|
|
base = NULL;
|
|
|
|
|
|
|
|
out_unlock:
|
2008-01-30 12:33:57 +00:00
|
|
|
spin_unlock_irqrestore(&pgd_lock, flags);
|
2008-01-30 12:33:56 +00:00
|
|
|
|
|
|
|
if (base)
|
|
|
|
__free_pages(base, 0);
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2008-02-04 15:48:07 +00:00
|
|
|
static int __change_page_attr(unsigned long address, struct cpa_data *cpa)
|
2008-01-30 12:33:41 +00:00
|
|
|
{
|
2008-02-09 22:24:08 +00:00
|
|
|
int do_split, err;
|
|
|
|
unsigned int level;
|
2005-04-16 22:20:36 +00:00
|
|
|
struct page *kpte_page;
|
2008-01-30 12:33:41 +00:00
|
|
|
pte_t *kpte;
|
2005-04-16 22:20:36 +00:00
|
|
|
|
2008-01-30 12:33:55 +00:00
|
|
|
repeat:
|
2008-01-30 12:33:43 +00:00
|
|
|
kpte = lookup_address(address, &level);
|
2005-04-16 22:20:36 +00:00
|
|
|
if (!kpte)
|
|
|
|
return -EINVAL;
|
2008-01-30 12:33:41 +00:00
|
|
|
|
2005-04-16 22:20:36 +00:00
|
|
|
kpte_page = virt_to_page(kpte);
|
2007-07-21 15:09:51 +00:00
|
|
|
BUG_ON(PageLRU(kpte_page));
|
|
|
|
BUG_ON(PageCompound(kpte_page));
|
|
|
|
|
2008-01-30 12:34:04 +00:00
|
|
|
if (level == PG_LEVEL_4K) {
|
2008-01-30 12:34:09 +00:00
|
|
|
pte_t new_pte, old_pte = *kpte;
|
2008-02-04 15:48:05 +00:00
|
|
|
pgprot_t new_prot = pte_pgprot(old_pte);
|
|
|
|
|
|
|
|
if(!pte_val(old_pte)) {
|
2008-02-04 15:48:07 +00:00
|
|
|
printk(KERN_WARNING "CPA: called for zero pte. "
|
|
|
|
"vaddr = %lx cpa->vaddr = %lx\n", address,
|
|
|
|
cpa->vaddr);
|
|
|
|
WARN_ON(1);
|
2008-02-04 15:48:05 +00:00
|
|
|
return -EINVAL;
|
|
|
|
}
|
2008-01-30 12:34:09 +00:00
|
|
|
|
2008-02-04 15:48:07 +00:00
|
|
|
pgprot_val(new_prot) &= ~pgprot_val(cpa->mask_clr);
|
|
|
|
pgprot_val(new_prot) |= pgprot_val(cpa->mask_set);
|
2008-01-30 12:34:09 +00:00
|
|
|
|
|
|
|
new_prot = static_protections(new_prot, address);
|
|
|
|
|
2008-02-04 15:48:05 +00:00
|
|
|
/*
|
|
|
|
* We need to keep the pfn from the existing PTE,
|
|
|
|
* after all we're only going to change it's attributes
|
|
|
|
* not the memory it points to
|
|
|
|
*/
|
|
|
|
new_pte = pfn_pte(pte_pfn(old_pte), canon_pgprot(new_prot));
|
2008-02-04 15:48:07 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Do we really change anything ?
|
|
|
|
*/
|
|
|
|
if (pte_val(old_pte) != pte_val(new_pte)) {
|
|
|
|
set_pte_atomic(kpte, new_pte);
|
|
|
|
cpa->flushtlb = 1;
|
|
|
|
}
|
2008-02-04 15:48:07 +00:00
|
|
|
cpa->numpages = 1;
|
|
|
|
return 0;
|
2005-04-16 22:20:36 +00:00
|
|
|
}
|
2008-02-04 15:48:07 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Check, whether we can keep the large page intact
|
|
|
|
* and just change the pte:
|
|
|
|
*/
|
2008-02-04 15:48:09 +00:00
|
|
|
do_split = try_preserve_large_page(kpte, address, cpa);
|
2008-02-04 15:48:07 +00:00
|
|
|
/*
|
|
|
|
* When the range fits into the existing large page,
|
|
|
|
* return. cp->numpages and cpa->tlbflush have been updated in
|
|
|
|
* try_large_page:
|
|
|
|
*/
|
2008-02-04 15:48:10 +00:00
|
|
|
if (do_split <= 0)
|
|
|
|
return do_split;
|
2008-02-04 15:48:07 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* We have to split the large page:
|
|
|
|
*/
|
2008-02-04 15:48:10 +00:00
|
|
|
err = split_large_page(kpte, address);
|
|
|
|
if (!err) {
|
|
|
|
cpa->flushtlb = 1;
|
|
|
|
goto repeat;
|
|
|
|
}
|
2008-02-04 15:48:09 +00:00
|
|
|
|
2008-02-04 15:48:10 +00:00
|
|
|
return err;
|
2008-01-30 12:33:41 +00:00
|
|
|
}
|
2005-04-16 22:20:36 +00:00
|
|
|
|
2008-01-30 12:34:03 +00:00
|
|
|
/**
|
|
|
|
* change_page_attr_addr - Change page table attributes in linear mapping
|
|
|
|
* @address: Virtual address in linear mapping.
|
|
|
|
* @prot: New page table attribute (PAGE_*)
|
2005-04-16 22:20:36 +00:00
|
|
|
*
|
2008-01-30 12:34:03 +00:00
|
|
|
* Change page attributes of a page in the direct mapping. This is a variant
|
|
|
|
* of change_page_attr() that also works on memory holes that do not have
|
|
|
|
* mem_map entry (pfn_valid() is false).
|
2008-01-30 12:33:41 +00:00
|
|
|
*
|
2008-01-30 12:34:03 +00:00
|
|
|
* See change_page_attr() documentation for more details.
|
x86: a new API for drivers/etc to control cache and other page attributes
Right now, if drivers or other code want to change, say, a cache attribute of a
page, the only API they have is change_page_attr(). c-p-a is a really bad API
for this, because it forces the caller to know *ALL* the attributes he wants
for the page, not just the 1 thing he wants to change. So code that wants to
set a page uncachable, needs to be aware of the NX status as well etc etc etc.
This patch introduces a set of new APIs for this, set_pages_<attr> and
set_memory_<attr>, that offer a logical change to the user, and leave all
attributes not implied by the requested logical change alone.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-01-30 12:34:06 +00:00
|
|
|
*
|
|
|
|
* Modules and drivers should use the set_memory_* APIs instead.
|
2005-04-16 22:20:36 +00:00
|
|
|
*/
|
2008-02-04 15:48:07 +00:00
|
|
|
static int change_page_attr_addr(struct cpa_data *cpa)
|
2005-04-16 22:20:36 +00:00
|
|
|
{
|
2008-01-30 12:34:09 +00:00
|
|
|
int err;
|
2008-02-04 15:48:07 +00:00
|
|
|
unsigned long address = cpa->vaddr;
|
2008-01-30 12:34:03 +00:00
|
|
|
|
|
|
|
#ifdef CONFIG_X86_64
|
2008-02-04 15:48:05 +00:00
|
|
|
unsigned long phys_addr = __pa(address);
|
|
|
|
|
2008-01-30 12:34:09 +00:00
|
|
|
/*
|
|
|
|
* If we are inside the high mapped kernel range, then we
|
|
|
|
* fixup the low mapping first. __va() returns the virtual
|
|
|
|
* address in the linear mapping:
|
|
|
|
*/
|
|
|
|
if (within(address, HIGH_MAP_START, HIGH_MAP_END))
|
|
|
|
address = (unsigned long) __va(phys_addr);
|
2008-01-30 12:34:03 +00:00
|
|
|
#endif
|
|
|
|
|
2008-02-04 15:48:07 +00:00
|
|
|
err = __change_page_attr(address, cpa);
|
2008-01-30 12:34:09 +00:00
|
|
|
if (err)
|
|
|
|
return err;
|
2008-01-30 12:34:03 +00:00
|
|
|
|
|
|
|
#ifdef CONFIG_X86_64
|
2008-01-30 12:34:07 +00:00
|
|
|
/*
|
2008-01-30 12:34:09 +00:00
|
|
|
* If the physical address is inside the kernel map, we need
|
|
|
|
* to touch the high mapped kernel as well:
|
2008-01-30 12:34:07 +00:00
|
|
|
*/
|
2008-01-30 12:34:09 +00:00
|
|
|
if (within(phys_addr, 0, KERNEL_TEXT_SIZE)) {
|
|
|
|
/*
|
|
|
|
* Calc the high mapping address. See __phys_addr()
|
|
|
|
* for the non obvious details.
|
2008-02-04 15:48:05 +00:00
|
|
|
*
|
|
|
|
* Note that NX and other required permissions are
|
|
|
|
* checked in static_protections().
|
2008-01-30 12:34:09 +00:00
|
|
|
*/
|
|
|
|
address = phys_addr + HIGH_MAP_START - phys_base;
|
|
|
|
|
2008-01-30 12:34:09 +00:00
|
|
|
/*
|
2008-01-30 12:34:09 +00:00
|
|
|
* Our high aliases are imprecise, because we check
|
|
|
|
* everything between 0 and KERNEL_TEXT_SIZE, so do
|
|
|
|
* not propagate lookup failures back to users:
|
2008-01-30 12:34:09 +00:00
|
|
|
*/
|
2008-02-04 15:48:07 +00:00
|
|
|
__change_page_attr(address, cpa);
|
2008-01-30 12:33:41 +00:00
|
|
|
}
|
2008-01-30 12:34:07 +00:00
|
|
|
#endif
|
2005-04-16 22:20:36 +00:00
|
|
|
return err;
|
|
|
|
}
|
|
|
|
|
2008-02-04 15:48:07 +00:00
|
|
|
static int __change_page_attr_set_clr(struct cpa_data *cpa)
|
2008-01-30 12:34:08 +00:00
|
|
|
{
|
2008-02-04 15:48:07 +00:00
|
|
|
int ret, numpages = cpa->numpages;
|
2008-01-30 12:34:08 +00:00
|
|
|
|
2008-02-04 15:48:07 +00:00
|
|
|
while (numpages) {
|
|
|
|
/*
|
|
|
|
* Store the remaining nr of pages for the large page
|
|
|
|
* preservation check.
|
|
|
|
*/
|
|
|
|
cpa->numpages = numpages;
|
2008-02-04 15:48:07 +00:00
|
|
|
ret = change_page_attr_addr(cpa);
|
2008-01-30 12:34:08 +00:00
|
|
|
if (ret)
|
|
|
|
return ret;
|
|
|
|
|
2008-02-04 15:48:07 +00:00
|
|
|
/*
|
|
|
|
* Adjust the number of pages with the result of the
|
|
|
|
* CPA operation. Either a large page has been
|
|
|
|
* preserved or a single page update happened.
|
|
|
|
*/
|
|
|
|
BUG_ON(cpa->numpages > numpages);
|
|
|
|
numpages -= cpa->numpages;
|
|
|
|
cpa->vaddr += cpa->numpages * PAGE_SIZE;
|
|
|
|
}
|
2008-01-30 12:34:08 +00:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2008-02-04 15:48:06 +00:00
|
|
|
static inline int cache_attr(pgprot_t attr)
|
|
|
|
{
|
|
|
|
return pgprot_val(attr) &
|
|
|
|
(_PAGE_PAT | _PAGE_PAT_LARGE | _PAGE_PWT | _PAGE_PCD);
|
|
|
|
}
|
|
|
|
|
2008-01-30 12:34:08 +00:00
|
|
|
static int change_page_attr_set_clr(unsigned long addr, int numpages,
|
|
|
|
pgprot_t mask_set, pgprot_t mask_clr)
|
|
|
|
{
|
2008-02-04 15:48:07 +00:00
|
|
|
struct cpa_data cpa;
|
2008-02-04 15:48:06 +00:00
|
|
|
int ret, cache;
|
2008-02-04 15:48:06 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Check, if we are requested to change a not supported
|
|
|
|
* feature:
|
|
|
|
*/
|
|
|
|
mask_set = canon_pgprot(mask_set);
|
|
|
|
mask_clr = canon_pgprot(mask_clr);
|
|
|
|
if (!pgprot_val(mask_set) && !pgprot_val(mask_clr))
|
|
|
|
return 0;
|
|
|
|
|
2008-02-04 15:48:07 +00:00
|
|
|
cpa.vaddr = addr;
|
|
|
|
cpa.numpages = numpages;
|
|
|
|
cpa.mask_set = mask_set;
|
|
|
|
cpa.mask_clr = mask_clr;
|
2008-02-04 15:48:07 +00:00
|
|
|
cpa.flushtlb = 0;
|
2008-02-04 15:48:07 +00:00
|
|
|
|
|
|
|
ret = __change_page_attr_set_clr(&cpa);
|
2008-01-30 12:34:08 +00:00
|
|
|
|
2008-02-04 15:48:07 +00:00
|
|
|
/*
|
|
|
|
* Check whether we really changed something:
|
|
|
|
*/
|
|
|
|
if (!cpa.flushtlb)
|
2008-02-09 22:24:09 +00:00
|
|
|
goto out;
|
2008-02-04 15:48:07 +00:00
|
|
|
|
2008-02-04 15:48:06 +00:00
|
|
|
/*
|
|
|
|
* No need to flush, when we did not set any of the caching
|
|
|
|
* attributes:
|
|
|
|
*/
|
|
|
|
cache = cache_attr(mask_set);
|
|
|
|
|
2008-01-30 12:34:08 +00:00
|
|
|
/*
|
|
|
|
* On success we use clflush, when the CPU supports it to
|
|
|
|
* avoid the wbindv. If the CPU does not support it and in the
|
2008-01-30 12:34:08 +00:00
|
|
|
* error case we fall back to cpa_flush_all (which uses
|
2008-01-30 12:34:08 +00:00
|
|
|
* wbindv):
|
|
|
|
*/
|
|
|
|
if (!ret && cpu_has_clflush)
|
2008-02-04 15:48:06 +00:00
|
|
|
cpa_flush_range(addr, numpages, cache);
|
2008-01-30 12:34:08 +00:00
|
|
|
else
|
2008-02-04 15:48:06 +00:00
|
|
|
cpa_flush_all(cache);
|
2008-01-30 12:34:08 +00:00
|
|
|
|
2008-02-09 22:24:09 +00:00
|
|
|
out:
|
|
|
|
cpa_fill_pool();
|
2008-01-30 12:34:08 +00:00
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2008-01-30 12:34:08 +00:00
|
|
|
static inline int change_page_attr_set(unsigned long addr, int numpages,
|
|
|
|
pgprot_t mask)
|
x86: a new API for drivers/etc to control cache and other page attributes
Right now, if drivers or other code want to change, say, a cache attribute of a
page, the only API they have is change_page_attr(). c-p-a is a really bad API
for this, because it forces the caller to know *ALL* the attributes he wants
for the page, not just the 1 thing he wants to change. So code that wants to
set a page uncachable, needs to be aware of the NX status as well etc etc etc.
This patch introduces a set of new APIs for this, set_pages_<attr> and
set_memory_<attr>, that offer a logical change to the user, and leave all
attributes not implied by the requested logical change alone.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-01-30 12:34:06 +00:00
|
|
|
{
|
2008-01-30 12:34:08 +00:00
|
|
|
return change_page_attr_set_clr(addr, numpages, mask, __pgprot(0));
|
x86: a new API for drivers/etc to control cache and other page attributes
Right now, if drivers or other code want to change, say, a cache attribute of a
page, the only API they have is change_page_attr(). c-p-a is a really bad API
for this, because it forces the caller to know *ALL* the attributes he wants
for the page, not just the 1 thing he wants to change. So code that wants to
set a page uncachable, needs to be aware of the NX status as well etc etc etc.
This patch introduces a set of new APIs for this, set_pages_<attr> and
set_memory_<attr>, that offer a logical change to the user, and leave all
attributes not implied by the requested logical change alone.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-01-30 12:34:06 +00:00
|
|
|
}
|
|
|
|
|
2008-01-30 12:34:08 +00:00
|
|
|
static inline int change_page_attr_clear(unsigned long addr, int numpages,
|
|
|
|
pgprot_t mask)
|
2008-01-30 12:34:08 +00:00
|
|
|
{
|
2008-01-31 21:05:43 +00:00
|
|
|
return change_page_attr_set_clr(addr, numpages, __pgprot(0), mask);
|
2008-01-30 12:34:08 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
int set_memory_uc(unsigned long addr, int numpages)
|
|
|
|
{
|
|
|
|
return change_page_attr_set(addr, numpages,
|
|
|
|
__pgprot(_PAGE_PCD | _PAGE_PWT));
|
x86: a new API for drivers/etc to control cache and other page attributes
Right now, if drivers or other code want to change, say, a cache attribute of a
page, the only API they have is change_page_attr(). c-p-a is a really bad API
for this, because it forces the caller to know *ALL* the attributes he wants
for the page, not just the 1 thing he wants to change. So code that wants to
set a page uncachable, needs to be aware of the NX status as well etc etc etc.
This patch introduces a set of new APIs for this, set_pages_<attr> and
set_memory_<attr>, that offer a logical change to the user, and leave all
attributes not implied by the requested logical change alone.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-01-30 12:34:06 +00:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(set_memory_uc);
|
|
|
|
|
|
|
|
int set_memory_wb(unsigned long addr, int numpages)
|
|
|
|
{
|
2008-01-30 12:34:08 +00:00
|
|
|
return change_page_attr_clear(addr, numpages,
|
|
|
|
__pgprot(_PAGE_PCD | _PAGE_PWT));
|
x86: a new API for drivers/etc to control cache and other page attributes
Right now, if drivers or other code want to change, say, a cache attribute of a
page, the only API they have is change_page_attr(). c-p-a is a really bad API
for this, because it forces the caller to know *ALL* the attributes he wants
for the page, not just the 1 thing he wants to change. So code that wants to
set a page uncachable, needs to be aware of the NX status as well etc etc etc.
This patch introduces a set of new APIs for this, set_pages_<attr> and
set_memory_<attr>, that offer a logical change to the user, and leave all
attributes not implied by the requested logical change alone.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-01-30 12:34:06 +00:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(set_memory_wb);
|
|
|
|
|
|
|
|
int set_memory_x(unsigned long addr, int numpages)
|
|
|
|
{
|
2008-01-30 12:34:08 +00:00
|
|
|
return change_page_attr_clear(addr, numpages, __pgprot(_PAGE_NX));
|
x86: a new API for drivers/etc to control cache and other page attributes
Right now, if drivers or other code want to change, say, a cache attribute of a
page, the only API they have is change_page_attr(). c-p-a is a really bad API
for this, because it forces the caller to know *ALL* the attributes he wants
for the page, not just the 1 thing he wants to change. So code that wants to
set a page uncachable, needs to be aware of the NX status as well etc etc etc.
This patch introduces a set of new APIs for this, set_pages_<attr> and
set_memory_<attr>, that offer a logical change to the user, and leave all
attributes not implied by the requested logical change alone.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-01-30 12:34:06 +00:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(set_memory_x);
|
|
|
|
|
|
|
|
int set_memory_nx(unsigned long addr, int numpages)
|
|
|
|
{
|
2008-01-30 12:34:08 +00:00
|
|
|
return change_page_attr_set(addr, numpages, __pgprot(_PAGE_NX));
|
x86: a new API for drivers/etc to control cache and other page attributes
Right now, if drivers or other code want to change, say, a cache attribute of a
page, the only API they have is change_page_attr(). c-p-a is a really bad API
for this, because it forces the caller to know *ALL* the attributes he wants
for the page, not just the 1 thing he wants to change. So code that wants to
set a page uncachable, needs to be aware of the NX status as well etc etc etc.
This patch introduces a set of new APIs for this, set_pages_<attr> and
set_memory_<attr>, that offer a logical change to the user, and leave all
attributes not implied by the requested logical change alone.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-01-30 12:34:06 +00:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(set_memory_nx);
|
|
|
|
|
|
|
|
int set_memory_ro(unsigned long addr, int numpages)
|
|
|
|
{
|
2008-01-30 12:34:08 +00:00
|
|
|
return change_page_attr_clear(addr, numpages, __pgprot(_PAGE_RW));
|
x86: a new API for drivers/etc to control cache and other page attributes
Right now, if drivers or other code want to change, say, a cache attribute of a
page, the only API they have is change_page_attr(). c-p-a is a really bad API
for this, because it forces the caller to know *ALL* the attributes he wants
for the page, not just the 1 thing he wants to change. So code that wants to
set a page uncachable, needs to be aware of the NX status as well etc etc etc.
This patch introduces a set of new APIs for this, set_pages_<attr> and
set_memory_<attr>, that offer a logical change to the user, and leave all
attributes not implied by the requested logical change alone.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-01-30 12:34:06 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
int set_memory_rw(unsigned long addr, int numpages)
|
|
|
|
{
|
2008-01-30 12:34:08 +00:00
|
|
|
return change_page_attr_set(addr, numpages, __pgprot(_PAGE_RW));
|
x86: a new API for drivers/etc to control cache and other page attributes
Right now, if drivers or other code want to change, say, a cache attribute of a
page, the only API they have is change_page_attr(). c-p-a is a really bad API
for this, because it forces the caller to know *ALL* the attributes he wants
for the page, not just the 1 thing he wants to change. So code that wants to
set a page uncachable, needs to be aware of the NX status as well etc etc etc.
This patch introduces a set of new APIs for this, set_pages_<attr> and
set_memory_<attr>, that offer a logical change to the user, and leave all
attributes not implied by the requested logical change alone.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-01-30 12:34:06 +00:00
|
|
|
}
|
2008-01-30 12:34:07 +00:00
|
|
|
|
|
|
|
int set_memory_np(unsigned long addr, int numpages)
|
|
|
|
{
|
2008-01-30 12:34:08 +00:00
|
|
|
return change_page_attr_clear(addr, numpages, __pgprot(_PAGE_PRESENT));
|
2008-01-30 12:34:07 +00:00
|
|
|
}
|
x86: a new API for drivers/etc to control cache and other page attributes
Right now, if drivers or other code want to change, say, a cache attribute of a
page, the only API they have is change_page_attr(). c-p-a is a really bad API
for this, because it forces the caller to know *ALL* the attributes he wants
for the page, not just the 1 thing he wants to change. So code that wants to
set a page uncachable, needs to be aware of the NX status as well etc etc etc.
This patch introduces a set of new APIs for this, set_pages_<attr> and
set_memory_<attr>, that offer a logical change to the user, and leave all
attributes not implied by the requested logical change alone.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-01-30 12:34:06 +00:00
|
|
|
|
|
|
|
int set_pages_uc(struct page *page, int numpages)
|
|
|
|
{
|
|
|
|
unsigned long addr = (unsigned long)page_address(page);
|
|
|
|
|
2008-01-30 12:34:07 +00:00
|
|
|
return set_memory_uc(addr, numpages);
|
x86: a new API for drivers/etc to control cache and other page attributes
Right now, if drivers or other code want to change, say, a cache attribute of a
page, the only API they have is change_page_attr(). c-p-a is a really bad API
for this, because it forces the caller to know *ALL* the attributes he wants
for the page, not just the 1 thing he wants to change. So code that wants to
set a page uncachable, needs to be aware of the NX status as well etc etc etc.
This patch introduces a set of new APIs for this, set_pages_<attr> and
set_memory_<attr>, that offer a logical change to the user, and leave all
attributes not implied by the requested logical change alone.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-01-30 12:34:06 +00:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(set_pages_uc);
|
|
|
|
|
|
|
|
int set_pages_wb(struct page *page, int numpages)
|
|
|
|
{
|
|
|
|
unsigned long addr = (unsigned long)page_address(page);
|
|
|
|
|
2008-01-30 12:34:07 +00:00
|
|
|
return set_memory_wb(addr, numpages);
|
x86: a new API for drivers/etc to control cache and other page attributes
Right now, if drivers or other code want to change, say, a cache attribute of a
page, the only API they have is change_page_attr(). c-p-a is a really bad API
for this, because it forces the caller to know *ALL* the attributes he wants
for the page, not just the 1 thing he wants to change. So code that wants to
set a page uncachable, needs to be aware of the NX status as well etc etc etc.
This patch introduces a set of new APIs for this, set_pages_<attr> and
set_memory_<attr>, that offer a logical change to the user, and leave all
attributes not implied by the requested logical change alone.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-01-30 12:34:06 +00:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(set_pages_wb);
|
|
|
|
|
|
|
|
int set_pages_x(struct page *page, int numpages)
|
|
|
|
{
|
|
|
|
unsigned long addr = (unsigned long)page_address(page);
|
|
|
|
|
2008-01-30 12:34:07 +00:00
|
|
|
return set_memory_x(addr, numpages);
|
x86: a new API for drivers/etc to control cache and other page attributes
Right now, if drivers or other code want to change, say, a cache attribute of a
page, the only API they have is change_page_attr(). c-p-a is a really bad API
for this, because it forces the caller to know *ALL* the attributes he wants
for the page, not just the 1 thing he wants to change. So code that wants to
set a page uncachable, needs to be aware of the NX status as well etc etc etc.
This patch introduces a set of new APIs for this, set_pages_<attr> and
set_memory_<attr>, that offer a logical change to the user, and leave all
attributes not implied by the requested logical change alone.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-01-30 12:34:06 +00:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(set_pages_x);
|
|
|
|
|
|
|
|
int set_pages_nx(struct page *page, int numpages)
|
|
|
|
{
|
|
|
|
unsigned long addr = (unsigned long)page_address(page);
|
|
|
|
|
2008-01-30 12:34:07 +00:00
|
|
|
return set_memory_nx(addr, numpages);
|
x86: a new API for drivers/etc to control cache and other page attributes
Right now, if drivers or other code want to change, say, a cache attribute of a
page, the only API they have is change_page_attr(). c-p-a is a really bad API
for this, because it forces the caller to know *ALL* the attributes he wants
for the page, not just the 1 thing he wants to change. So code that wants to
set a page uncachable, needs to be aware of the NX status as well etc etc etc.
This patch introduces a set of new APIs for this, set_pages_<attr> and
set_memory_<attr>, that offer a logical change to the user, and leave all
attributes not implied by the requested logical change alone.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-01-30 12:34:06 +00:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(set_pages_nx);
|
|
|
|
|
|
|
|
int set_pages_ro(struct page *page, int numpages)
|
|
|
|
{
|
|
|
|
unsigned long addr = (unsigned long)page_address(page);
|
|
|
|
|
2008-01-30 12:34:07 +00:00
|
|
|
return set_memory_ro(addr, numpages);
|
x86: a new API for drivers/etc to control cache and other page attributes
Right now, if drivers or other code want to change, say, a cache attribute of a
page, the only API they have is change_page_attr(). c-p-a is a really bad API
for this, because it forces the caller to know *ALL* the attributes he wants
for the page, not just the 1 thing he wants to change. So code that wants to
set a page uncachable, needs to be aware of the NX status as well etc etc etc.
This patch introduces a set of new APIs for this, set_pages_<attr> and
set_memory_<attr>, that offer a logical change to the user, and leave all
attributes not implied by the requested logical change alone.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-01-30 12:34:06 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
int set_pages_rw(struct page *page, int numpages)
|
|
|
|
{
|
|
|
|
unsigned long addr = (unsigned long)page_address(page);
|
2008-01-30 12:34:06 +00:00
|
|
|
|
2008-01-30 12:34:07 +00:00
|
|
|
return set_memory_rw(addr, numpages);
|
2008-01-30 12:33:55 +00:00
|
|
|
}
|
|
|
|
|
2005-04-16 22:20:36 +00:00
|
|
|
#ifdef CONFIG_DEBUG_PAGEALLOC
|
2008-01-30 12:34:07 +00:00
|
|
|
|
|
|
|
static int __set_pages_p(struct page *page, int numpages)
|
|
|
|
{
|
2008-02-04 15:48:07 +00:00
|
|
|
struct cpa_data cpa = { .vaddr = (unsigned long) page_address(page),
|
|
|
|
.numpages = numpages,
|
|
|
|
.mask_set = __pgprot(_PAGE_PRESENT | _PAGE_RW),
|
|
|
|
.mask_clr = __pgprot(0)};
|
2008-01-30 12:34:08 +00:00
|
|
|
|
2008-02-04 15:48:07 +00:00
|
|
|
return __change_page_attr_set_clr(&cpa);
|
2008-01-30 12:34:07 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static int __set_pages_np(struct page *page, int numpages)
|
|
|
|
{
|
2008-02-04 15:48:07 +00:00
|
|
|
struct cpa_data cpa = { .vaddr = (unsigned long) page_address(page),
|
|
|
|
.numpages = numpages,
|
|
|
|
.mask_set = __pgprot(0),
|
|
|
|
.mask_clr = __pgprot(_PAGE_PRESENT | _PAGE_RW)};
|
2008-01-30 12:34:08 +00:00
|
|
|
|
2008-02-04 15:48:07 +00:00
|
|
|
return __change_page_attr_set_clr(&cpa);
|
2008-01-30 12:34:07 +00:00
|
|
|
}
|
|
|
|
|
2005-04-16 22:20:36 +00:00
|
|
|
void kernel_map_pages(struct page *page, int numpages, int enable)
|
|
|
|
{
|
|
|
|
if (PageHighMem(page))
|
|
|
|
return;
|
2008-01-30 12:33:41 +00:00
|
|
|
if (!enable) {
|
2006-06-27 09:54:49 +00:00
|
|
|
debug_check_no_locks_freed(page_address(page),
|
|
|
|
numpages * PAGE_SIZE);
|
2008-01-30 12:33:41 +00:00
|
|
|
}
|
2006-01-09 23:59:21 +00:00
|
|
|
|
2008-01-30 12:33:58 +00:00
|
|
|
/*
|
|
|
|
* If page allocator is not up yet then do not call c_p_a():
|
|
|
|
*/
|
|
|
|
if (!debug_pagealloc_enabled)
|
|
|
|
return;
|
|
|
|
|
2008-01-30 12:33:41 +00:00
|
|
|
/*
|
2008-01-30 12:34:04 +00:00
|
|
|
* The return value is ignored - the calls cannot fail,
|
|
|
|
* large pages are disabled at boot time:
|
2005-04-16 22:20:36 +00:00
|
|
|
*/
|
2008-01-30 12:34:07 +00:00
|
|
|
if (enable)
|
|
|
|
__set_pages_p(page, numpages);
|
|
|
|
else
|
|
|
|
__set_pages_np(page, numpages);
|
2008-01-30 12:33:41 +00:00
|
|
|
|
|
|
|
/*
|
2008-01-30 12:34:04 +00:00
|
|
|
* We should perform an IPI and flush all tlbs,
|
|
|
|
* but that can deadlock->flush only current cpu:
|
2005-04-16 22:20:36 +00:00
|
|
|
*/
|
|
|
|
__flush_tlb_all();
|
2008-02-09 22:24:09 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Try to refill the page pool here. We can do this only after
|
|
|
|
* the tlb flush.
|
|
|
|
*/
|
|
|
|
cpa_fill_pool();
|
2005-04-16 22:20:36 +00:00
|
|
|
}
|
|
|
|
#endif
|
2008-01-30 12:34:07 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* The testcases use internal knowledge of the implementation that shouldn't
|
|
|
|
* be exposed to the rest of the kernel. Include these directly here.
|
|
|
|
*/
|
|
|
|
#ifdef CONFIG_CPA_DEBUG
|
|
|
|
#include "pageattr-test.c"
|
|
|
|
#endif
|