aboutsummaryrefslogtreecommitdiff
path: root/arch/x86/mm
diff options
context:
space:
mode:
Diffstat (limited to 'arch/x86/mm')
-rw-r--r--arch/x86/mm/fault.c2
-rw-r--r--arch/x86/mm/init_32.c1
-rw-r--r--arch/x86/mm/init_64.c28
-rw-r--r--arch/x86/mm/ioremap.c30
-rw-r--r--arch/x86/mm/numa_64.c5
-rw-r--r--arch/x86/mm/pageattr.c253
-rw-r--r--arch/x86/mm/srat_64.c3
7 files changed, 201 insertions, 121 deletions
diff --git a/arch/x86/mm/fault.c b/arch/x86/mm/fault.c
index 621afb6343d..fdc667422df 100644
--- a/arch/x86/mm/fault.c
+++ b/arch/x86/mm/fault.c
@@ -186,7 +186,7 @@ static int bad_address(void *p)
}
#endif
-void dump_pagetable(unsigned long address)
+static void dump_pagetable(unsigned long address)
{
#ifdef CONFIG_X86_32
__typeof__(pte_val(__pte(0))) page;
diff --git a/arch/x86/mm/init_32.c b/arch/x86/mm/init_32.c
index 8106bba41ec..ee1091a4696 100644
--- a/arch/x86/mm/init_32.c
+++ b/arch/x86/mm/init_32.c
@@ -47,6 +47,7 @@
#include <asm/sections.h>
#include <asm/paravirt.h>
#include <asm/setup.h>
+#include <asm/cacheflush.h>
unsigned int __VMALLOC_RESERVE = 128 << 20;
diff --git a/arch/x86/mm/init_64.c b/arch/x86/mm/init_64.c
index b59fc238151..bb652f5a93f 100644
--- a/arch/x86/mm/init_64.c
+++ b/arch/x86/mm/init_64.c
@@ -45,6 +45,7 @@
#include <asm/sections.h>
#include <asm/kdebug.h>
#include <asm/numa.h>
+#include <asm/cacheflush.h>
const struct dma_mapping_ops *dma_ops;
EXPORT_SYMBOL(dma_ops);
@@ -170,6 +171,33 @@ set_pte_phys(unsigned long vaddr, unsigned long phys, pgprot_t prot)
__flush_tlb_one(vaddr);
}
+/*
+ * The head.S code sets up the kernel high mapping from:
+ * __START_KERNEL_map to __START_KERNEL_map + KERNEL_TEXT_SIZE
+ *
+ * phys_addr holds the negative offset to the kernel, which is added
+ * to the compile time generated pmds. This results in invalid pmds up
+ * to the point where we hit the physaddr 0 mapping.
+ *
+ * We limit the mappings to the region from _text to _end. _end is
+ * rounded up to the 2MB boundary. This catches the invalid pmds as
+ * well, as they are located before _text:
+ */
+void __init cleanup_highmap(void)
+{
+ unsigned long vaddr = __START_KERNEL_map;
+ unsigned long end = round_up((unsigned long)_end, PMD_SIZE) - 1;
+ pmd_t *pmd = level2_kernel_pgt;
+ pmd_t *last_pmd = pmd + PTRS_PER_PMD;
+
+ for (; pmd < last_pmd; pmd++, vaddr += PMD_SIZE) {
+ if (!pmd_present(*pmd))
+ continue;
+ if (vaddr < (unsigned long) _text || vaddr > end)
+ set_pmd(pmd, __pmd(0));
+ }
+}
+
/* NOTE: this is meant to be run only at boot */
void __init
__set_fixmap(enum fixed_addresses idx, unsigned long phys, pgprot_t prot)
diff --git a/arch/x86/mm/ioremap.c b/arch/x86/mm/ioremap.c
index a4897a85268..882328efc3d 100644
--- a/arch/x86/mm/ioremap.c
+++ b/arch/x86/mm/ioremap.c
@@ -42,6 +42,22 @@ int page_is_ram(unsigned long pagenr)
unsigned long addr, end;
int i;
+ /*
+ * A special case is the first 4Kb of memory;
+ * This is a BIOS owned area, not kernel ram, but generally
+ * not listed as such in the E820 table.
+ */
+ if (pagenr == 0)
+ return 0;
+
+ /*
+ * Second special case: Some BIOSen report the PC BIOS
+ * area (640->1Mb) as ram even though it is not.
+ */
+ if (pagenr >= (BIOS_BEGIN >> PAGE_SHIFT) &&
+ pagenr < (BIOS_END >> PAGE_SHIFT))
+ return 0;
+
for (i = 0; i < e820.nr_map; i++) {
/*
* Not usable memory:
@@ -51,14 +67,6 @@ int page_is_ram(unsigned long pagenr)
addr = (e820.map[i].addr + PAGE_SIZE-1) >> PAGE_SHIFT;
end = (e820.map[i].addr + e820.map[i].size) >> PAGE_SHIFT;
- /*
- * Sanity check: Some BIOSen report areas as RAM that
- * are not. Notably the 640->1Mb area, which is the
- * PCI BIOS area.
- */
- if (addr >= (BIOS_BEGIN >> PAGE_SHIFT) &&
- end < (BIOS_END >> PAGE_SHIFT))
- continue;
if ((pagenr >= addr) && (pagenr < end))
return 1;
@@ -126,6 +134,8 @@ static void __iomem *__ioremap(unsigned long phys_addr, unsigned long size,
return NULL;
}
+ WARN_ON_ONCE(page_is_ram(pfn));
+
switch (mode) {
case IOR_MODE_UNCACHED:
default:
@@ -265,7 +275,9 @@ static __initdata pte_t bm_pte[PAGE_SIZE/sizeof(pte_t)]
static inline pmd_t * __init early_ioremap_pmd(unsigned long addr)
{
- pgd_t *pgd = &swapper_pg_dir[pgd_index(addr)];
+ /* Don't assume we're using swapper_pg_dir at this point */
+ pgd_t *base = __va(read_cr3());
+ pgd_t *pgd = &base[pgd_index(addr)];
pud_t *pud = pud_offset(pgd, addr);
pmd_t *pmd = pmd_offset(pud, addr);
diff --git a/arch/x86/mm/numa_64.c b/arch/x86/mm/numa_64.c
index 1aecc658cd7..59898fb0a4a 100644
--- a/arch/x86/mm/numa_64.c
+++ b/arch/x86/mm/numa_64.c
@@ -494,11 +494,13 @@ void __init numa_initmem_init(unsigned long start_pfn, unsigned long end_pfn)
int i;
nodes_clear(node_possible_map);
+ nodes_clear(node_online_map);
#ifdef CONFIG_NUMA_EMU
if (cmdline && !numa_emulation(start_pfn, end_pfn))
return;
nodes_clear(node_possible_map);
+ nodes_clear(node_online_map);
#endif
#ifdef CONFIG_ACPI_NUMA
@@ -506,6 +508,7 @@ void __init numa_initmem_init(unsigned long start_pfn, unsigned long end_pfn)
end_pfn << PAGE_SHIFT))
return;
nodes_clear(node_possible_map);
+ nodes_clear(node_online_map);
#endif
#ifdef CONFIG_K8_NUMA
@@ -513,6 +516,7 @@ void __init numa_initmem_init(unsigned long start_pfn, unsigned long end_pfn)
end_pfn<<PAGE_SHIFT))
return;
nodes_clear(node_possible_map);
+ nodes_clear(node_online_map);
#endif
printk(KERN_INFO "%s\n",
numa_off ? "NUMA turned off" : "No NUMA configuration found");
@@ -524,7 +528,6 @@ void __init numa_initmem_init(unsigned long start_pfn, unsigned long end_pfn)
memnode_shift = 63;
memnodemap = memnode.embedded_map;
memnodemap[0] = 0;
- nodes_clear(node_online_map);
node_set_online(0);
node_set(0, node_possible_map);
for (i = 0; i < NR_CPUS; i++)
diff --git a/arch/x86/mm/pageattr.c b/arch/x86/mm/pageattr.c
index 440210a2277..464d8fc21ce 100644
--- a/arch/x86/mm/pageattr.c
+++ b/arch/x86/mm/pageattr.c
@@ -16,6 +16,7 @@
#include <asm/sections.h>
#include <asm/uaccess.h>
#include <asm/pgalloc.h>
+#include <asm/proto.h>
/*
* The current flushing context - we pass it instead of 5 arguments:
@@ -26,8 +27,23 @@ struct cpa_data {
pgprot_t mask_clr;
int numpages;
int flushtlb;
+ unsigned long pfn;
};
+#ifdef CONFIG_X86_64
+
+static inline unsigned long highmap_start_pfn(void)
+{
+ return __pa(_text) >> PAGE_SHIFT;
+}
+
+static inline unsigned long highmap_end_pfn(void)
+{
+ return __pa(round_up((unsigned long)_end, PMD_SIZE)) >> PAGE_SHIFT;
+}
+
+#endif
+
static inline int
within(unsigned long addr, unsigned long start, unsigned long end)
{
@@ -123,29 +139,14 @@ static void cpa_flush_range(unsigned long start, int numpages, int cache)
}
}
-#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
-}
-
/*
* 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)
+static inline pgprot_t static_protections(pgprot_t prot, unsigned long address,
+ unsigned long pfn)
{
pgprot_t forbidden = __pgprot(0);
@@ -153,30 +154,23 @@ static inline pgprot_t static_protections(pgprot_t prot, unsigned long address)
* The BIOS area between 640k and 1Mb needs to be executable for
* PCI BIOS based config access (CONFIG_PCI_GOBIOS) support.
*/
- if (within(__pa(address), BIOS_BEGIN, BIOS_END))
+ if (within(pfn, BIOS_BEGIN >> PAGE_SHIFT, BIOS_END >> PAGE_SHIFT))
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
+ * Does not cover __inittext since that is gone later on. On
+ * 64bit we do not enforce !NX on the low mapping
*/
if (within(address, (unsigned long)_text, (unsigned long)_etext))
pgprot_val(forbidden) |= _PAGE_NX;
- /*
- * 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;
- /* The .rodata section needs to be read-only */
- if (within(address, (unsigned long)__start_rodata,
- (unsigned long)__end_rodata))
- pgprot_val(forbidden) |= _PAGE_RW;
/*
- * Do the same for the x86-64 high kernel mapping
+ * The .rodata section needs to be read-only. Using the pfn
+ * catches all aliases.
*/
- if (within(address, virt_to_highmap(__start_rodata),
- virt_to_highmap(__end_rodata)))
+ if (within(pfn, __pa((unsigned long)__start_rodata) >> PAGE_SHIFT,
+ __pa((unsigned long)__end_rodata) >> PAGE_SHIFT))
pgprot_val(forbidden) |= _PAGE_RW;
prot = __pgprot(pgprot_val(prot) & ~pgprot_val(forbidden));
@@ -253,7 +247,7 @@ static int
try_preserve_large_page(pte_t *kpte, unsigned long address,
struct cpa_data *cpa)
{
- unsigned long nextpage_addr, numpages, pmask, psize, flags, addr;
+ unsigned long nextpage_addr, numpages, pmask, psize, flags, addr, pfn;
pte_t new_pte, old_pte, *tmp;
pgprot_t old_prot, new_prot;
int i, do_split = 1;
@@ -275,8 +269,8 @@ try_preserve_large_page(pte_t *kpte, unsigned long address,
break;
#ifdef CONFIG_X86_64
case PG_LEVEL_1G:
- psize = PMD_PAGE_SIZE;
- pmask = PMD_PAGE_MASK;
+ psize = PUD_PAGE_SIZE;
+ pmask = PUD_PAGE_MASK;
break;
#endif
default:
@@ -301,7 +295,15 @@ try_preserve_large_page(pte_t *kpte, unsigned long address,
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);
+
+ /*
+ * old_pte points to the large page base address. So we need
+ * to add the offset of the virtual address:
+ */
+ pfn = pte_pfn(old_pte) + ((address & (psize - 1)) >> PAGE_SHIFT);
+ cpa->pfn = pfn;
+
+ new_prot = static_protections(new_prot, address, pfn);
/*
* We need to check the full range, whether
@@ -309,8 +311,9 @@ try_preserve_large_page(pte_t *kpte, unsigned long address,
* the pages in the range we try to preserve:
*/
addr = address + PAGE_SIZE;
- for (i = 1; i < cpa->numpages; i++, addr += PAGE_SIZE) {
- pgprot_t chk_prot = static_protections(new_prot, addr);
+ pfn++;
+ for (i = 1; i < cpa->numpages; i++, addr += PAGE_SIZE, pfn++) {
+ pgprot_t chk_prot = static_protections(new_prot, addr, pfn);
if (pgprot_val(chk_prot) != pgprot_val(new_prot))
goto out_unlock;
@@ -505,46 +508,46 @@ out_unlock:
return 0;
}
-static int __change_page_attr(unsigned long address, struct cpa_data *cpa)
+static int __change_page_attr(struct cpa_data *cpa, int primary)
{
+ unsigned long address = cpa->vaddr;
int do_split, err;
unsigned int level;
- struct page *kpte_page;
- pte_t *kpte;
+ pte_t *kpte, old_pte;
repeat:
kpte = lookup_address(address, &level);
if (!kpte)
- return -EINVAL;
+ return primary ? -EINVAL : 0;
- kpte_page = virt_to_page(kpte);
- BUG_ON(PageLRU(kpte_page));
- BUG_ON(PageCompound(kpte_page));
+ old_pte = *kpte;
+ if (!pte_val(old_pte)) {
+ if (!primary)
+ return 0;
+ printk(KERN_WARNING "CPA: called for zero pte. "
+ "vaddr = %lx cpa->vaddr = %lx\n", address,
+ cpa->vaddr);
+ WARN_ON(1);
+ return -EINVAL;
+ }
if (level == PG_LEVEL_4K) {
- pte_t new_pte, old_pte = *kpte;
+ pte_t new_pte;
pgprot_t new_prot = pte_pgprot(old_pte);
-
- if(!pte_val(old_pte)) {
- printk(KERN_WARNING "CPA: called for zero pte. "
- "vaddr = %lx cpa->vaddr = %lx\n", address,
- cpa->vaddr);
- WARN_ON(1);
- return -EINVAL;
- }
+ unsigned long pfn = pte_pfn(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);
+ new_prot = static_protections(new_prot, address, pfn);
/*
* 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));
-
+ new_pte = pfn_pte(pfn, canon_pgprot(new_prot));
+ cpa->pfn = pfn;
/*
* Do we really change anything ?
*/
@@ -581,67 +584,59 @@ repeat:
return err;
}
-/**
- * change_page_attr_addr - Change page table attributes in linear mapping
- * @address: Virtual address in linear mapping.
- * @prot: New page table attribute (PAGE_*)
- *
- * 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).
- *
- * See change_page_attr() documentation for more details.
- *
- * Modules and drivers should use the set_memory_* APIs instead.
- */
-static int change_page_attr_addr(struct cpa_data *cpa)
+static int __change_page_attr_set_clr(struct cpa_data *cpa, int checkalias);
+
+static int cpa_process_alias(struct cpa_data *cpa)
{
- int err;
- unsigned long address = cpa->vaddr;
+ struct cpa_data alias_cpa;
+ int ret = 0;
-#ifdef CONFIG_X86_64
- unsigned long phys_addr = __pa(address);
+ if (cpa->pfn > max_pfn_mapped)
+ return 0;
/*
- * 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:
+ * No need to redo, when the primary call touched the direct
+ * mapping already:
*/
- if (within(address, HIGH_MAP_START, HIGH_MAP_END))
- address = (unsigned long) __va(phys_addr);
-#endif
+ if (!within(cpa->vaddr, PAGE_OFFSET,
+ PAGE_OFFSET + (max_pfn_mapped << PAGE_SHIFT))) {
- err = __change_page_attr(address, cpa);
- if (err)
- return err;
+ alias_cpa = *cpa;
+ alias_cpa.vaddr = (unsigned long) __va(cpa->pfn << PAGE_SHIFT);
+
+ ret = __change_page_attr_set_clr(&alias_cpa, 0);
+ }
#ifdef CONFIG_X86_64
+ if (ret)
+ return ret;
+ /*
+ * No need to redo, when the primary call touched the high
+ * mapping already:
+ */
+ if (within(cpa->vaddr, (unsigned long) _text, (unsigned long) _end))
+ return 0;
+
/*
* If the physical address is inside the kernel map, we need
* to touch the high mapped kernel as well:
*/
- if (within(phys_addr, 0, KERNEL_TEXT_SIZE)) {
- /*
- * Calc the high mapping address. See __phys_addr()
- * for the non obvious details.
- *
- * Note that NX and other required permissions are
- * checked in static_protections().
- */
- address = phys_addr + HIGH_MAP_START - phys_base;
+ if (!within(cpa->pfn, highmap_start_pfn(), highmap_end_pfn()))
+ return 0;
- /*
- * Our high aliases are imprecise, because we check
- * everything between 0 and KERNEL_TEXT_SIZE, so do
- * not propagate lookup failures back to users:
- */
- __change_page_attr(address, cpa);
- }
+ alias_cpa = *cpa;
+ alias_cpa.vaddr =
+ (cpa->pfn << PAGE_SHIFT) + __START_KERNEL_map - phys_base;
+
+ /*
+ * The high mapping range is imprecise, so ignore the return value.
+ */
+ __change_page_attr_set_clr(&alias_cpa, 0);
#endif
- return err;
+ return ret;
}
-static int __change_page_attr_set_clr(struct cpa_data *cpa)
+static int __change_page_attr_set_clr(struct cpa_data *cpa, int checkalias)
{
int ret, numpages = cpa->numpages;
@@ -651,10 +646,17 @@ static int __change_page_attr_set_clr(struct cpa_data *cpa)
* preservation check.
*/
cpa->numpages = numpages;
- ret = change_page_attr_addr(cpa);
+
+ ret = __change_page_attr(cpa, checkalias);
if (ret)
return ret;
+ if (checkalias) {
+ ret = cpa_process_alias(cpa);
+ if (ret)
+ return ret;
+ }
+
/*
* Adjust the number of pages with the result of the
* CPA operation. Either a large page has been
@@ -677,7 +679,7 @@ static int change_page_attr_set_clr(unsigned long addr, int numpages,
pgprot_t mask_set, pgprot_t mask_clr)
{
struct cpa_data cpa;
- int ret, cache;
+ int ret, cache, checkalias;
/*
* Check, if we are requested to change a not supported
@@ -688,13 +690,25 @@ static int change_page_attr_set_clr(unsigned long addr, int numpages,
if (!pgprot_val(mask_set) && !pgprot_val(mask_clr))
return 0;
+ /* Ensure we are PAGE_SIZE aligned */
+ if (addr & ~PAGE_MASK) {
+ addr &= PAGE_MASK;
+ /*
+ * People should not be passing in unaligned addresses:
+ */
+ WARN_ON_ONCE(1);
+ }
+
cpa.vaddr = addr;
cpa.numpages = numpages;
cpa.mask_set = mask_set;
cpa.mask_clr = mask_clr;
cpa.flushtlb = 0;
- ret = __change_page_attr_set_clr(&cpa);
+ /* No alias checking for _NX bit modifications */
+ checkalias = (pgprot_val(mask_set) | pgprot_val(mask_clr)) != _PAGE_NX;
+
+ ret = __change_page_attr_set_clr(&cpa, checkalias);
/*
* Check whether we really changed something:
@@ -832,7 +846,7 @@ static int __set_pages_p(struct page *page, int numpages)
.mask_set = __pgprot(_PAGE_PRESENT | _PAGE_RW),
.mask_clr = __pgprot(0)};
- return __change_page_attr_set_clr(&cpa);
+ return __change_page_attr_set_clr(&cpa, 1);
}
static int __set_pages_np(struct page *page, int numpages)
@@ -842,7 +856,7 @@ static int __set_pages_np(struct page *page, int numpages)
.mask_set = __pgprot(0),
.mask_clr = __pgprot(_PAGE_PRESENT | _PAGE_RW)};
- return __change_page_attr_set_clr(&cpa);
+ return __change_page_attr_set_clr(&cpa, 1);
}
void kernel_map_pages(struct page *page, int numpages, int enable)
@@ -861,8 +875,12 @@ void kernel_map_pages(struct page *page, int numpages, int enable)
return;
/*
- * The return value is ignored - the calls cannot fail,
- * large pages are disabled at boot time:
+ * The return value is ignored as the calls cannot fail.
+ * Large pages are kept enabled at boot time, and are
+ * split up quickly with DEBUG_PAGEALLOC. If a splitup
+ * fails here (due to temporary memory shortage) no damage
+ * is done because we just keep the largepage intact up
+ * to the next attempt when it will likely be split up:
*/
if (enable)
__set_pages_p(page, numpages);
@@ -881,7 +899,24 @@ void kernel_map_pages(struct page *page, int numpages, int enable)
*/
cpa_fill_pool();
}
-#endif
+
+#ifdef CONFIG_HIBERNATION
+
+bool kernel_page_present(struct page *page)
+{
+ unsigned int level;
+ pte_t *pte;
+
+ if (PageHighMem(page))
+ return false;
+
+ pte = lookup_address((unsigned long)page_address(page), &level);
+ return (pte_val(*pte) & _PAGE_PRESENT);
+}
+
+#endif /* CONFIG_HIBERNATION */
+
+#endif /* CONFIG_DEBUG_PAGEALLOC */
/*
* The testcases use internal knowledge of the implementation that shouldn't
diff --git a/arch/x86/mm/srat_64.c b/arch/x86/mm/srat_64.c
index ecd91ea8a8a..845001c617c 100644
--- a/arch/x86/mm/srat_64.c
+++ b/arch/x86/mm/srat_64.c
@@ -166,7 +166,8 @@ static inline int save_add_info(void) {return 0;}
* Both SPARSE and RESERVE need nodes_add information.
* This code supports one contiguous hot add area per node.
*/
-static int reserve_hotadd(int node, unsigned long start, unsigned long end)
+static int __init
+reserve_hotadd(int node, unsigned long start, unsigned long end)
{
unsigned long s_pfn = start >> PAGE_SHIFT;
unsigned long e_pfn = end >> PAGE_SHIFT;
generated by cgit v1.2.3 (git 2.25.1) at 2024-12-28 17:25:00 +0000