2005-04-17 06:20:36 +08:00
|
|
|
/*
|
|
|
|
* linux/arch/x86_64/mm/init.c
|
|
|
|
*
|
|
|
|
* Copyright (C) 1995 Linus Torvalds
|
2010-07-18 20:27:13 +08:00
|
|
|
* Copyright (C) 2000 Pavel Machek <pavel@ucw.cz>
|
2005-04-17 06:20:36 +08:00
|
|
|
* Copyright (C) 2002,2003 Andi Kleen <ak@suse.de>
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include <linux/signal.h>
|
|
|
|
#include <linux/sched.h>
|
|
|
|
#include <linux/kernel.h>
|
|
|
|
#include <linux/errno.h>
|
|
|
|
#include <linux/string.h>
|
|
|
|
#include <linux/types.h>
|
|
|
|
#include <linux/ptrace.h>
|
|
|
|
#include <linux/mman.h>
|
|
|
|
#include <linux/mm.h>
|
|
|
|
#include <linux/swap.h>
|
|
|
|
#include <linux/smp.h>
|
|
|
|
#include <linux/init.h>
|
2008-05-12 21:43:36 +08:00
|
|
|
#include <linux/initrd.h>
|
2005-04-17 06:20:36 +08:00
|
|
|
#include <linux/pagemap.h>
|
|
|
|
#include <linux/bootmem.h>
|
2010-08-26 04:39:17 +08:00
|
|
|
#include <linux/memblock.h>
|
2005-04-17 06:20:36 +08:00
|
|
|
#include <linux/proc_fs.h>
|
2005-11-06 00:25:53 +08:00
|
|
|
#include <linux/pci.h>
|
[PATCH] x86: tighten kernel image page access rights
On x86-64, kernel memory freed after init can be entirely unmapped instead
of just getting 'poisoned' by overwriting with a debug pattern.
On i386 and x86-64 (under CONFIG_DEBUG_RODATA), kernel text and bug table
can also be write-protected.
Compared to the first version, this one prevents re-creating deleted
mappings in the kernel image range on x86-64, if those got removed
previously. This, together with the original changes, prevents temporarily
having inconsistent mappings when cacheability attributes are being
changed on such pages (e.g. from AGP code). While on i386 such duplicate
mappings don't exist, the same change is done there, too, both for
consistency and because checking pte_present() before using various other
pte_XXX functions is a requirement anyway. At once, i386 code gets
adjusted to use pte_huge() instead of open coding this.
AK: split out cpa() changes
Signed-off-by: Jan Beulich <jbeulich@novell.com>
Signed-off-by: Andi Kleen <ak@suse.de>
2007-05-03 01:27:10 +08:00
|
|
|
#include <linux/pfn.h>
|
2006-06-27 17:53:52 +08:00
|
|
|
#include <linux/poison.h>
|
2006-01-12 05:44:42 +08:00
|
|
|
#include <linux/dma-mapping.h>
|
2011-06-14 08:57:50 +08:00
|
|
|
#include <linux/memory.h>
|
2006-01-17 14:03:41 +08:00
|
|
|
#include <linux/memory_hotplug.h>
|
2016-01-16 08:56:22 +08:00
|
|
|
#include <linux/memremap.h>
|
2007-05-03 01:27:11 +08:00
|
|
|
#include <linux/nmi.h>
|
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h
percpu.h is included by sched.h and module.h and thus ends up being
included when building most .c files. percpu.h includes slab.h which
in turn includes gfp.h making everything defined by the two files
universally available and complicating inclusion dependencies.
percpu.h -> slab.h dependency is about to be removed. Prepare for
this change by updating users of gfp and slab facilities include those
headers directly instead of assuming availability. As this conversion
needs to touch large number of source files, the following script is
used as the basis of conversion.
http://userweb.kernel.org/~tj/misc/slabh-sweep.py
The script does the followings.
* Scan files for gfp and slab usages and update includes such that
only the necessary includes are there. ie. if only gfp is used,
gfp.h, if slab is used, slab.h.
* When the script inserts a new include, it looks at the include
blocks and try to put the new include such that its order conforms
to its surrounding. It's put in the include block which contains
core kernel includes, in the same order that the rest are ordered -
alphabetical, Christmas tree, rev-Xmas-tree or at the end if there
doesn't seem to be any matching order.
* If the script can't find a place to put a new include (mostly
because the file doesn't have fitting include block), it prints out
an error message indicating which .h file needs to be added to the
file.
The conversion was done in the following steps.
1. The initial automatic conversion of all .c files updated slightly
over 4000 files, deleting around 700 includes and adding ~480 gfp.h
and ~3000 slab.h inclusions. The script emitted errors for ~400
files.
2. Each error was manually checked. Some didn't need the inclusion,
some needed manual addition while adding it to implementation .h or
embedding .c file was more appropriate for others. This step added
inclusions to around 150 files.
3. The script was run again and the output was compared to the edits
from #2 to make sure no file was left behind.
4. Several build tests were done and a couple of problems were fixed.
e.g. lib/decompress_*.c used malloc/free() wrappers around slab
APIs requiring slab.h to be added manually.
5. The script was run on all .h files but without automatically
editing them as sprinkling gfp.h and slab.h inclusions around .h
files could easily lead to inclusion dependency hell. Most gfp.h
inclusion directives were ignored as stuff from gfp.h was usually
wildly available and often used in preprocessor macros. Each
slab.h inclusion directive was examined and added manually as
necessary.
6. percpu.h was updated not to include slab.h.
7. Build test were done on the following configurations and failures
were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my
distributed build env didn't work with gcov compiles) and a few
more options had to be turned off depending on archs to make things
build (like ipr on powerpc/64 which failed due to missing writeq).
* x86 and x86_64 UP and SMP allmodconfig and a custom test config.
* powerpc and powerpc64 SMP allmodconfig
* sparc and sparc64 SMP allmodconfig
* ia64 SMP allmodconfig
* s390 SMP allmodconfig
* alpha SMP allmodconfig
* um on x86_64 SMP allmodconfig
8. percpu.h modifications were reverted so that it could be applied as
a separate patch and serve as bisection point.
Given the fact that I had only a couple of failures from tests on step
6, I'm fairly confident about the coverage of this conversion patch.
If there is a breakage, it's likely to be something in one of the arch
headers which should be easily discoverable easily on most builds of
the specific arch.
Signed-off-by: Tejun Heo <tj@kernel.org>
Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 16:04:11 +08:00
|
|
|
#include <linux/gfp.h>
|
2013-04-12 07:10:25 +08:00
|
|
|
#include <linux/kcore.h>
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
#include <asm/processor.h>
|
2008-10-12 21:06:29 +08:00
|
|
|
#include <asm/bios_ebda.h>
|
2016-12-25 03:46:01 +08:00
|
|
|
#include <linux/uaccess.h>
|
2005-04-17 06:20:36 +08:00
|
|
|
#include <asm/pgtable.h>
|
|
|
|
#include <asm/pgalloc.h>
|
|
|
|
#include <asm/dma.h>
|
|
|
|
#include <asm/fixmap.h>
|
2017-01-27 17:27:10 +08:00
|
|
|
#include <asm/e820/api.h>
|
2005-04-17 06:20:36 +08:00
|
|
|
#include <asm/apic.h>
|
|
|
|
#include <asm/tlb.h>
|
|
|
|
#include <asm/mmu_context.h>
|
|
|
|
#include <asm/proto.h>
|
|
|
|
#include <asm/smp.h>
|
2005-11-06 00:25:53 +08:00
|
|
|
#include <asm/sections.h>
|
2008-01-30 20:30:17 +08:00
|
|
|
#include <asm/kdebug.h>
|
2008-01-30 20:30:17 +08:00
|
|
|
#include <asm/numa.h>
|
2017-05-09 06:58:11 +08:00
|
|
|
#include <asm/set_memory.h>
|
2009-03-05 20:55:08 +08:00
|
|
|
#include <asm/init.h>
|
2015-12-01 00:47:43 +08:00
|
|
|
#include <asm/uv/uv.h>
|
2011-02-18 19:30:30 +08:00
|
|
|
#include <asm/setup.h>
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2012-11-17 11:39:01 +08:00
|
|
|
#include "mm_internal.h"
|
|
|
|
|
2016-05-07 06:01:34 +08:00
|
|
|
#include "ident_map.c"
|
2013-01-25 04:19:48 +08:00
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
/*
|
|
|
|
* NOTE: pagetable_init alloc all the fixmap pagetables contiguous on the
|
|
|
|
* physical space so we can cache the place of the first one and move
|
|
|
|
* around without checking the pgd every time.
|
|
|
|
*/
|
|
|
|
|
2014-01-08 01:03:06 +08:00
|
|
|
pteval_t __supported_pte_mask __read_mostly = ~0;
|
2008-09-05 15:58:28 +08:00
|
|
|
EXPORT_SYMBOL_GPL(__supported_pte_mask);
|
|
|
|
|
|
|
|
int force_personality32;
|
|
|
|
|
2008-09-05 16:23:26 +08:00
|
|
|
/*
|
|
|
|
* noexec32=on|off
|
|
|
|
* Control non executable heap for 32bit processes.
|
|
|
|
* To control the stack too use noexec=off
|
|
|
|
*
|
|
|
|
* on PROT_READ does not imply PROT_EXEC for 32-bit processes (default)
|
|
|
|
* off PROT_READ implies PROT_EXEC
|
|
|
|
*/
|
2008-09-05 15:58:28 +08:00
|
|
|
static int __init nonx32_setup(char *str)
|
|
|
|
{
|
|
|
|
if (!strcmp(str, "on"))
|
|
|
|
force_personality32 &= ~READ_IMPLIES_EXEC;
|
|
|
|
else if (!strcmp(str, "off"))
|
|
|
|
force_personality32 |= READ_IMPLIES_EXEC;
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
__setup("noexec32=", nonx32_setup);
|
|
|
|
|
2010-05-19 17:42:14 +08:00
|
|
|
/*
|
2016-12-15 07:44:03 +08:00
|
|
|
* When memory was added make sure all the processes MM have
|
2010-05-19 17:42:14 +08:00
|
|
|
* suitable PGD entries in the local PGD level page.
|
|
|
|
*/
|
2017-06-06 19:31:29 +08:00
|
|
|
#ifdef CONFIG_X86_5LEVEL
|
|
|
|
void sync_global_pgds(unsigned long start, unsigned long end)
|
|
|
|
{
|
|
|
|
unsigned long addr;
|
|
|
|
|
|
|
|
for (addr = start; addr <= end; addr = ALIGN(addr + 1, PGDIR_SIZE)) {
|
|
|
|
const pgd_t *pgd_ref = pgd_offset_k(addr);
|
|
|
|
struct page *page;
|
|
|
|
|
|
|
|
/* Check for overflow */
|
|
|
|
if (addr < start)
|
|
|
|
break;
|
|
|
|
|
|
|
|
if (pgd_none(*pgd_ref))
|
|
|
|
continue;
|
|
|
|
|
|
|
|
spin_lock(&pgd_lock);
|
|
|
|
list_for_each_entry(page, &pgd_list, lru) {
|
|
|
|
pgd_t *pgd;
|
|
|
|
spinlock_t *pgt_lock;
|
|
|
|
|
|
|
|
pgd = (pgd_t *)page_address(page) + pgd_index(addr);
|
|
|
|
/* the pgt_lock only for Xen */
|
|
|
|
pgt_lock = &pgd_page_get_mm(page)->page_table_lock;
|
|
|
|
spin_lock(pgt_lock);
|
|
|
|
|
|
|
|
if (!pgd_none(*pgd_ref) && !pgd_none(*pgd))
|
|
|
|
BUG_ON(pgd_page_vaddr(*pgd) != pgd_page_vaddr(*pgd_ref));
|
|
|
|
|
|
|
|
if (pgd_none(*pgd))
|
|
|
|
set_pgd(pgd, *pgd_ref);
|
|
|
|
|
|
|
|
spin_unlock(pgt_lock);
|
|
|
|
}
|
|
|
|
spin_unlock(&pgd_lock);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
#else
|
2016-12-15 07:44:03 +08:00
|
|
|
void sync_global_pgds(unsigned long start, unsigned long end)
|
2010-05-19 17:42:14 +08:00
|
|
|
{
|
x86/mm: Fix boot crash caused by incorrect loop count calculation in sync_global_pgds()
Jeff Moyer reported that on his system with two memory regions 0~64G and
1T~1T+192G, and kernel option "memmap=192G!1024G" added, enabling KASLR
will make the system hang intermittently during boot. While adding 'nokaslr'
won't.
The back trace is:
Oops: 0000 [#1] SMP
RIP: memcpy_erms()
[ .... ]
Call Trace:
pmem_rw_page()
bdev_read_page()
do_mpage_readpage()
mpage_readpages()
blkdev_readpages()
__do_page_cache_readahead()
force_page_cache_readahead()
page_cache_sync_readahead()
generic_file_read_iter()
blkdev_read_iter()
__vfs_read()
vfs_read()
SyS_read()
entry_SYSCALL_64_fastpath()
This crash happens because the for loop count calculation in sync_global_pgds()
is not correct. When a mapping area crosses PGD entries, we should
calculate the starting address of region which next PGD covers and assign
it to next for loop count, but not add PGDIR_SIZE directly. The old
code works right only if the mapping area is an exact multiple of PGDIR_SIZE,
otherwize the end region could be skipped so that it can't be synchronized
to all other processes from kernel PGD init_mm.pgd.
In Jeff's system, emulated pmem area [1024G, 1216G) is smaller than
PGDIR_SIZE. While 'nokaslr' works because PAGE_OFFSET is 1T aligned, it
makes this area be mapped inside one PGD entry. With KASLR enabled,
this area could cross two PGD entries, then the next PGD entry won't
be synced to all other processes. That is why we saw empty PGD.
Fix it.
Reported-by: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Baoquan He <bhe@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Dave Young <dyoung@redhat.com>
Cc: Denys Vlasenko <dvlasenk@redhat.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Jinbum Park <jinb.park7@gmail.com>
Cc: Josh Poimboeuf <jpoimboe@redhat.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Garnier <thgarnie@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Yasuaki Ishimatsu <yasu.isimatu@gmail.com>
Cc: Yinghai Lu <yinghai@kernel.org>
Link: http://lkml.kernel.org/r/1493864747-8506-1-git-send-email-bhe@redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2017-05-04 10:25:47 +08:00
|
|
|
unsigned long addr;
|
2010-10-15 08:04:59 +08:00
|
|
|
|
x86/mm: Fix boot crash caused by incorrect loop count calculation in sync_global_pgds()
Jeff Moyer reported that on his system with two memory regions 0~64G and
1T~1T+192G, and kernel option "memmap=192G!1024G" added, enabling KASLR
will make the system hang intermittently during boot. While adding 'nokaslr'
won't.
The back trace is:
Oops: 0000 [#1] SMP
RIP: memcpy_erms()
[ .... ]
Call Trace:
pmem_rw_page()
bdev_read_page()
do_mpage_readpage()
mpage_readpages()
blkdev_readpages()
__do_page_cache_readahead()
force_page_cache_readahead()
page_cache_sync_readahead()
generic_file_read_iter()
blkdev_read_iter()
__vfs_read()
vfs_read()
SyS_read()
entry_SYSCALL_64_fastpath()
This crash happens because the for loop count calculation in sync_global_pgds()
is not correct. When a mapping area crosses PGD entries, we should
calculate the starting address of region which next PGD covers and assign
it to next for loop count, but not add PGDIR_SIZE directly. The old
code works right only if the mapping area is an exact multiple of PGDIR_SIZE,
otherwize the end region could be skipped so that it can't be synchronized
to all other processes from kernel PGD init_mm.pgd.
In Jeff's system, emulated pmem area [1024G, 1216G) is smaller than
PGDIR_SIZE. While 'nokaslr' works because PAGE_OFFSET is 1T aligned, it
makes this area be mapped inside one PGD entry. With KASLR enabled,
this area could cross two PGD entries, then the next PGD entry won't
be synced to all other processes. That is why we saw empty PGD.
Fix it.
Reported-by: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Baoquan He <bhe@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Dave Young <dyoung@redhat.com>
Cc: Denys Vlasenko <dvlasenk@redhat.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Jinbum Park <jinb.park7@gmail.com>
Cc: Josh Poimboeuf <jpoimboe@redhat.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Garnier <thgarnie@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Yasuaki Ishimatsu <yasu.isimatu@gmail.com>
Cc: Yinghai Lu <yinghai@kernel.org>
Link: http://lkml.kernel.org/r/1493864747-8506-1-git-send-email-bhe@redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2017-05-04 10:25:47 +08:00
|
|
|
for (addr = start; addr <= end; addr = ALIGN(addr + 1, PGDIR_SIZE)) {
|
|
|
|
pgd_t *pgd_ref = pgd_offset_k(addr);
|
2017-03-18 02:55:15 +08:00
|
|
|
const p4d_t *p4d_ref;
|
2010-10-15 08:04:59 +08:00
|
|
|
struct page *page;
|
|
|
|
|
2017-03-18 02:55:15 +08:00
|
|
|
/*
|
|
|
|
* With folded p4d, pgd_none() is always false, we need to
|
|
|
|
* handle synchonization on p4d level.
|
|
|
|
*/
|
|
|
|
BUILD_BUG_ON(pgd_none(*pgd_ref));
|
x86/mm: Fix boot crash caused by incorrect loop count calculation in sync_global_pgds()
Jeff Moyer reported that on his system with two memory regions 0~64G and
1T~1T+192G, and kernel option "memmap=192G!1024G" added, enabling KASLR
will make the system hang intermittently during boot. While adding 'nokaslr'
won't.
The back trace is:
Oops: 0000 [#1] SMP
RIP: memcpy_erms()
[ .... ]
Call Trace:
pmem_rw_page()
bdev_read_page()
do_mpage_readpage()
mpage_readpages()
blkdev_readpages()
__do_page_cache_readahead()
force_page_cache_readahead()
page_cache_sync_readahead()
generic_file_read_iter()
blkdev_read_iter()
__vfs_read()
vfs_read()
SyS_read()
entry_SYSCALL_64_fastpath()
This crash happens because the for loop count calculation in sync_global_pgds()
is not correct. When a mapping area crosses PGD entries, we should
calculate the starting address of region which next PGD covers and assign
it to next for loop count, but not add PGDIR_SIZE directly. The old
code works right only if the mapping area is an exact multiple of PGDIR_SIZE,
otherwize the end region could be skipped so that it can't be synchronized
to all other processes from kernel PGD init_mm.pgd.
In Jeff's system, emulated pmem area [1024G, 1216G) is smaller than
PGDIR_SIZE. While 'nokaslr' works because PAGE_OFFSET is 1T aligned, it
makes this area be mapped inside one PGD entry. With KASLR enabled,
this area could cross two PGD entries, then the next PGD entry won't
be synced to all other processes. That is why we saw empty PGD.
Fix it.
Reported-by: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Baoquan He <bhe@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Dave Young <dyoung@redhat.com>
Cc: Denys Vlasenko <dvlasenk@redhat.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Jinbum Park <jinb.park7@gmail.com>
Cc: Josh Poimboeuf <jpoimboe@redhat.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Garnier <thgarnie@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Yasuaki Ishimatsu <yasu.isimatu@gmail.com>
Cc: Yinghai Lu <yinghai@kernel.org>
Link: http://lkml.kernel.org/r/1493864747-8506-1-git-send-email-bhe@redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2017-05-04 10:25:47 +08:00
|
|
|
p4d_ref = p4d_offset(pgd_ref, addr);
|
2017-03-18 02:55:15 +08:00
|
|
|
|
|
|
|
if (p4d_none(*p4d_ref))
|
2010-10-15 08:04:59 +08:00
|
|
|
continue;
|
|
|
|
|
2011-02-17 07:45:22 +08:00
|
|
|
spin_lock(&pgd_lock);
|
2010-10-15 08:04:59 +08:00
|
|
|
list_for_each_entry(page, &pgd_list, lru) {
|
2012-12-16 04:29:54 +08:00
|
|
|
pgd_t *pgd;
|
2017-03-18 02:55:15 +08:00
|
|
|
p4d_t *p4d;
|
2010-09-22 03:01:51 +08:00
|
|
|
spinlock_t *pgt_lock;
|
|
|
|
|
x86/mm: Fix boot crash caused by incorrect loop count calculation in sync_global_pgds()
Jeff Moyer reported that on his system with two memory regions 0~64G and
1T~1T+192G, and kernel option "memmap=192G!1024G" added, enabling KASLR
will make the system hang intermittently during boot. While adding 'nokaslr'
won't.
The back trace is:
Oops: 0000 [#1] SMP
RIP: memcpy_erms()
[ .... ]
Call Trace:
pmem_rw_page()
bdev_read_page()
do_mpage_readpage()
mpage_readpages()
blkdev_readpages()
__do_page_cache_readahead()
force_page_cache_readahead()
page_cache_sync_readahead()
generic_file_read_iter()
blkdev_read_iter()
__vfs_read()
vfs_read()
SyS_read()
entry_SYSCALL_64_fastpath()
This crash happens because the for loop count calculation in sync_global_pgds()
is not correct. When a mapping area crosses PGD entries, we should
calculate the starting address of region which next PGD covers and assign
it to next for loop count, but not add PGDIR_SIZE directly. The old
code works right only if the mapping area is an exact multiple of PGDIR_SIZE,
otherwize the end region could be skipped so that it can't be synchronized
to all other processes from kernel PGD init_mm.pgd.
In Jeff's system, emulated pmem area [1024G, 1216G) is smaller than
PGDIR_SIZE. While 'nokaslr' works because PAGE_OFFSET is 1T aligned, it
makes this area be mapped inside one PGD entry. With KASLR enabled,
this area could cross two PGD entries, then the next PGD entry won't
be synced to all other processes. That is why we saw empty PGD.
Fix it.
Reported-by: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Baoquan He <bhe@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Dave Young <dyoung@redhat.com>
Cc: Denys Vlasenko <dvlasenk@redhat.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Jinbum Park <jinb.park7@gmail.com>
Cc: Josh Poimboeuf <jpoimboe@redhat.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Garnier <thgarnie@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Yasuaki Ishimatsu <yasu.isimatu@gmail.com>
Cc: Yinghai Lu <yinghai@kernel.org>
Link: http://lkml.kernel.org/r/1493864747-8506-1-git-send-email-bhe@redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2017-05-04 10:25:47 +08:00
|
|
|
pgd = (pgd_t *)page_address(page) + pgd_index(addr);
|
|
|
|
p4d = p4d_offset(pgd, addr);
|
2011-02-17 07:45:22 +08:00
|
|
|
/* the pgt_lock only for Xen */
|
2010-09-22 03:01:51 +08:00
|
|
|
pgt_lock = &pgd_page_get_mm(page)->page_table_lock;
|
|
|
|
spin_lock(pgt_lock);
|
|
|
|
|
2017-03-18 02:55:15 +08:00
|
|
|
if (!p4d_none(*p4d_ref) && !p4d_none(*p4d))
|
|
|
|
BUG_ON(p4d_page_vaddr(*p4d)
|
|
|
|
!= p4d_page_vaddr(*p4d_ref));
|
2010-09-22 03:01:51 +08:00
|
|
|
|
2017-03-18 02:55:15 +08:00
|
|
|
if (p4d_none(*p4d))
|
|
|
|
set_p4d(p4d, *p4d_ref);
|
2014-08-23 04:27:34 +08:00
|
|
|
|
2010-09-22 03:01:51 +08:00
|
|
|
spin_unlock(pgt_lock);
|
2010-10-15 08:04:59 +08:00
|
|
|
}
|
2011-02-17 07:45:22 +08:00
|
|
|
spin_unlock(&pgd_lock);
|
2010-10-15 08:04:59 +08:00
|
|
|
}
|
2010-05-19 17:42:14 +08:00
|
|
|
}
|
2017-06-06 19:31:29 +08:00
|
|
|
#endif
|
2010-05-19 17:42:14 +08:00
|
|
|
|
2008-08-16 00:32:24 +08:00
|
|
|
/*
|
|
|
|
* NOTE: This function is marked __ref because it calls __init function
|
|
|
|
* (alloc_bootmem_pages). It's safe to do it ONLY when after_bootmem == 0.
|
|
|
|
*/
|
|
|
|
static __ref void *spp_getpage(void)
|
2008-01-30 20:34:10 +08:00
|
|
|
{
|
2005-04-17 06:20:36 +08:00
|
|
|
void *ptr;
|
2008-01-30 20:34:10 +08:00
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
if (after_bootmem)
|
2009-02-22 18:28:25 +08:00
|
|
|
ptr = (void *) get_zeroed_page(GFP_ATOMIC | __GFP_NOTRACK);
|
2005-04-17 06:20:36 +08:00
|
|
|
else
|
|
|
|
ptr = alloc_bootmem_pages(PAGE_SIZE);
|
2008-01-30 20:34:10 +08:00
|
|
|
|
|
|
|
if (!ptr || ((unsigned long)ptr & ~PAGE_MASK)) {
|
|
|
|
panic("set_pte_phys: cannot allocate page data %s\n",
|
|
|
|
after_bootmem ? "after bootmem" : "");
|
|
|
|
}
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2008-01-30 20:34:10 +08:00
|
|
|
pr_debug("spp_getpage %p\n", ptr);
|
2008-01-30 20:34:10 +08:00
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
return ptr;
|
2008-01-30 20:34:10 +08:00
|
|
|
}
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2017-03-18 02:55:15 +08:00
|
|
|
static p4d_t *fill_p4d(pgd_t *pgd, unsigned long vaddr)
|
2005-04-17 06:20:36 +08:00
|
|
|
{
|
2009-02-24 10:57:21 +08:00
|
|
|
if (pgd_none(*pgd)) {
|
2017-03-18 02:55:15 +08:00
|
|
|
p4d_t *p4d = (p4d_t *)spp_getpage();
|
|
|
|
pgd_populate(&init_mm, pgd, p4d);
|
|
|
|
if (p4d != p4d_offset(pgd, 0))
|
2009-02-24 10:57:21 +08:00
|
|
|
printk(KERN_ERR "PAGETABLE BUG #00! %p <-> %p\n",
|
2017-03-18 02:55:15 +08:00
|
|
|
p4d, p4d_offset(pgd, 0));
|
|
|
|
}
|
|
|
|
return p4d_offset(pgd, vaddr);
|
|
|
|
}
|
|
|
|
|
|
|
|
static pud_t *fill_pud(p4d_t *p4d, unsigned long vaddr)
|
|
|
|
{
|
|
|
|
if (p4d_none(*p4d)) {
|
|
|
|
pud_t *pud = (pud_t *)spp_getpage();
|
|
|
|
p4d_populate(&init_mm, p4d, pud);
|
|
|
|
if (pud != pud_offset(p4d, 0))
|
|
|
|
printk(KERN_ERR "PAGETABLE BUG #01! %p <-> %p\n",
|
|
|
|
pud, pud_offset(p4d, 0));
|
2009-02-24 10:57:21 +08:00
|
|
|
}
|
2017-03-18 02:55:15 +08:00
|
|
|
return pud_offset(p4d, vaddr);
|
2009-02-24 10:57:21 +08:00
|
|
|
}
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2009-03-04 04:02:57 +08:00
|
|
|
static pmd_t *fill_pmd(pud_t *pud, unsigned long vaddr)
|
2009-02-24 10:57:21 +08:00
|
|
|
{
|
2005-04-17 06:20:36 +08:00
|
|
|
if (pud_none(*pud)) {
|
2009-02-24 10:57:21 +08:00
|
|
|
pmd_t *pmd = (pmd_t *) spp_getpage();
|
2008-06-25 12:19:02 +08:00
|
|
|
pud_populate(&init_mm, pud, pmd);
|
2009-02-24 10:57:21 +08:00
|
|
|
if (pmd != pmd_offset(pud, 0))
|
2017-03-18 02:55:15 +08:00
|
|
|
printk(KERN_ERR "PAGETABLE BUG #02! %p <-> %p\n",
|
2009-02-24 10:57:21 +08:00
|
|
|
pmd, pmd_offset(pud, 0));
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
2009-02-24 10:57:21 +08:00
|
|
|
return pmd_offset(pud, vaddr);
|
|
|
|
}
|
|
|
|
|
2009-03-04 04:02:57 +08:00
|
|
|
static pte_t *fill_pte(pmd_t *pmd, unsigned long vaddr)
|
2009-02-24 10:57:21 +08:00
|
|
|
{
|
2005-04-17 06:20:36 +08:00
|
|
|
if (pmd_none(*pmd)) {
|
2009-02-24 10:57:21 +08:00
|
|
|
pte_t *pte = (pte_t *) spp_getpage();
|
2008-06-25 12:19:02 +08:00
|
|
|
pmd_populate_kernel(&init_mm, pmd, pte);
|
2009-02-24 10:57:21 +08:00
|
|
|
if (pte != pte_offset_kernel(pmd, 0))
|
2017-03-18 02:55:15 +08:00
|
|
|
printk(KERN_ERR "PAGETABLE BUG #03!\n");
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
2009-02-24 10:57:21 +08:00
|
|
|
return pte_offset_kernel(pmd, vaddr);
|
|
|
|
}
|
|
|
|
|
2017-03-18 02:55:15 +08:00
|
|
|
static void __set_pte_vaddr(pud_t *pud, unsigned long vaddr, pte_t new_pte)
|
2009-02-24 10:57:21 +08:00
|
|
|
{
|
2017-03-18 02:55:15 +08:00
|
|
|
pmd_t *pmd = fill_pmd(pud, vaddr);
|
|
|
|
pte_t *pte = fill_pte(pmd, vaddr);
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
set_pte(pte, new_pte);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* It's enough to flush this one mapping.
|
|
|
|
* (PGE mappings get flushed as well)
|
|
|
|
*/
|
|
|
|
__flush_tlb_one(vaddr);
|
|
|
|
}
|
|
|
|
|
2017-03-18 02:55:15 +08:00
|
|
|
void set_pte_vaddr_p4d(p4d_t *p4d_page, unsigned long vaddr, pte_t new_pte)
|
|
|
|
{
|
|
|
|
p4d_t *p4d = p4d_page + p4d_index(vaddr);
|
|
|
|
pud_t *pud = fill_pud(p4d, vaddr);
|
|
|
|
|
|
|
|
__set_pte_vaddr(pud, vaddr, new_pte);
|
|
|
|
}
|
|
|
|
|
|
|
|
void set_pte_vaddr_pud(pud_t *pud_page, unsigned long vaddr, pte_t new_pte)
|
|
|
|
{
|
|
|
|
pud_t *pud = pud_page + pud_index(vaddr);
|
|
|
|
|
|
|
|
__set_pte_vaddr(pud, vaddr, new_pte);
|
|
|
|
}
|
|
|
|
|
2009-02-24 10:57:21 +08:00
|
|
|
void set_pte_vaddr(unsigned long vaddr, pte_t pteval)
|
2008-06-25 12:19:22 +08:00
|
|
|
{
|
|
|
|
pgd_t *pgd;
|
2017-03-18 02:55:15 +08:00
|
|
|
p4d_t *p4d_page;
|
2008-06-25 12:19:22 +08:00
|
|
|
|
|
|
|
pr_debug("set_pte_vaddr %lx to %lx\n", vaddr, native_pte_val(pteval));
|
|
|
|
|
|
|
|
pgd = pgd_offset_k(vaddr);
|
|
|
|
if (pgd_none(*pgd)) {
|
|
|
|
printk(KERN_ERR
|
|
|
|
"PGD FIXMAP MISSING, it should be setup in head.S!\n");
|
|
|
|
return;
|
|
|
|
}
|
2017-03-18 02:55:15 +08:00
|
|
|
|
|
|
|
p4d_page = p4d_offset(pgd, 0);
|
|
|
|
set_pte_vaddr_p4d(p4d_page, vaddr, pteval);
|
2008-06-25 12:19:22 +08:00
|
|
|
}
|
|
|
|
|
2009-02-24 10:57:21 +08:00
|
|
|
pmd_t * __init populate_extra_pmd(unsigned long vaddr)
|
2009-02-20 15:29:09 +08:00
|
|
|
{
|
|
|
|
pgd_t *pgd;
|
2017-03-18 02:55:15 +08:00
|
|
|
p4d_t *p4d;
|
2009-02-20 15:29:09 +08:00
|
|
|
pud_t *pud;
|
|
|
|
|
|
|
|
pgd = pgd_offset_k(vaddr);
|
2017-03-18 02:55:15 +08:00
|
|
|
p4d = fill_p4d(pgd, vaddr);
|
|
|
|
pud = fill_pud(p4d, vaddr);
|
2009-02-24 10:57:21 +08:00
|
|
|
return fill_pmd(pud, vaddr);
|
|
|
|
}
|
|
|
|
|
|
|
|
pte_t * __init populate_extra_pte(unsigned long vaddr)
|
|
|
|
{
|
|
|
|
pmd_t *pmd;
|
2009-02-20 15:29:09 +08:00
|
|
|
|
2009-02-24 10:57:21 +08:00
|
|
|
pmd = populate_extra_pmd(vaddr);
|
|
|
|
return fill_pte(pmd, vaddr);
|
2009-02-20 15:29:09 +08:00
|
|
|
}
|
|
|
|
|
2008-07-02 03:45:32 +08:00
|
|
|
/*
|
|
|
|
* Create large page table mappings for a range of physical addresses.
|
|
|
|
*/
|
|
|
|
static void __init __init_extra_mapping(unsigned long phys, unsigned long size,
|
2014-11-03 21:01:52 +08:00
|
|
|
enum page_cache_mode cache)
|
2008-07-02 03:45:32 +08:00
|
|
|
{
|
|
|
|
pgd_t *pgd;
|
2017-03-18 02:55:15 +08:00
|
|
|
p4d_t *p4d;
|
2008-07-02 03:45:32 +08:00
|
|
|
pud_t *pud;
|
|
|
|
pmd_t *pmd;
|
2014-11-03 21:01:52 +08:00
|
|
|
pgprot_t prot;
|
2008-07-02 03:45:32 +08:00
|
|
|
|
2014-11-03 21:01:52 +08:00
|
|
|
pgprot_val(prot) = pgprot_val(PAGE_KERNEL_LARGE) |
|
|
|
|
pgprot_val(pgprot_4k_2_large(cachemode2pgprot(cache)));
|
2008-07-02 03:45:32 +08:00
|
|
|
BUG_ON((phys & ~PMD_MASK) || (size & ~PMD_MASK));
|
|
|
|
for (; size; phys += PMD_SIZE, size -= PMD_SIZE) {
|
|
|
|
pgd = pgd_offset_k((unsigned long)__va(phys));
|
|
|
|
if (pgd_none(*pgd)) {
|
2017-03-18 02:55:15 +08:00
|
|
|
p4d = (p4d_t *) spp_getpage();
|
|
|
|
set_pgd(pgd, __pgd(__pa(p4d) | _KERNPG_TABLE |
|
|
|
|
_PAGE_USER));
|
|
|
|
}
|
|
|
|
p4d = p4d_offset(pgd, (unsigned long)__va(phys));
|
|
|
|
if (p4d_none(*p4d)) {
|
2008-07-02 03:45:32 +08:00
|
|
|
pud = (pud_t *) spp_getpage();
|
2017-03-18 02:55:15 +08:00
|
|
|
set_p4d(p4d, __p4d(__pa(pud) | _KERNPG_TABLE |
|
2008-07-02 03:45:32 +08:00
|
|
|
_PAGE_USER));
|
|
|
|
}
|
2017-03-18 02:55:15 +08:00
|
|
|
pud = pud_offset(p4d, (unsigned long)__va(phys));
|
2008-07-02 03:45:32 +08:00
|
|
|
if (pud_none(*pud)) {
|
|
|
|
pmd = (pmd_t *) spp_getpage();
|
|
|
|
set_pud(pud, __pud(__pa(pmd) | _KERNPG_TABLE |
|
|
|
|
_PAGE_USER));
|
|
|
|
}
|
|
|
|
pmd = pmd_offset(pud, phys);
|
|
|
|
BUG_ON(!pmd_none(*pmd));
|
|
|
|
set_pmd(pmd, __pmd(phys | pgprot_val(prot)));
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
void __init init_extra_mapping_wb(unsigned long phys, unsigned long size)
|
|
|
|
{
|
2014-11-03 21:01:52 +08:00
|
|
|
__init_extra_mapping(phys, size, _PAGE_CACHE_MODE_WB);
|
2008-07-02 03:45:32 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
void __init init_extra_mapping_uc(unsigned long phys, unsigned long size)
|
|
|
|
{
|
2014-11-03 21:01:52 +08:00
|
|
|
__init_extra_mapping(phys, size, _PAGE_CACHE_MODE_UC);
|
2008-07-02 03:45:32 +08:00
|
|
|
}
|
|
|
|
|
2008-02-16 00:29:12 +08:00
|
|
|
/*
|
2008-02-21 18:04:11 +08:00
|
|
|
* The head.S code sets up the kernel high mapping:
|
|
|
|
*
|
|
|
|
* from __START_KERNEL_map to __START_KERNEL_map + size (== _end-_text)
|
2008-02-16 00:29:12 +08:00
|
|
|
*
|
2013-05-15 10:58:07 +08:00
|
|
|
* phys_base holds the negative offset to the kernel, which is added
|
2008-02-16 00:29:12 +08:00
|
|
|
* to the compile time generated pmds. This results in invalid pmds up
|
|
|
|
* to the point where we hit the physaddr 0 mapping.
|
|
|
|
*
|
2011-02-18 19:30:30 +08:00
|
|
|
* We limit the mappings to the region from _text to _brk_end. _brk_end
|
|
|
|
* is rounded up to the 2MB boundary. This catches the invalid pmds as
|
2008-02-16 00:29:12 +08:00
|
|
|
* well, as they are located before _text:
|
|
|
|
*/
|
|
|
|
void __init cleanup_highmap(void)
|
|
|
|
{
|
|
|
|
unsigned long vaddr = __START_KERNEL_map;
|
2013-01-25 04:19:54 +08:00
|
|
|
unsigned long vaddr_end = __START_KERNEL_map + KERNEL_IMAGE_SIZE;
|
2011-02-18 19:30:30 +08:00
|
|
|
unsigned long end = roundup((unsigned long)_brk_end, PMD_SIZE) - 1;
|
2008-02-16 00:29:12 +08:00
|
|
|
pmd_t *pmd = level2_kernel_pgt;
|
|
|
|
|
2013-01-25 04:19:54 +08:00
|
|
|
/*
|
|
|
|
* Native path, max_pfn_mapped is not set yet.
|
|
|
|
* Xen has valid max_pfn_mapped set in
|
|
|
|
* arch/x86/xen/mmu.c:xen_setup_kernel_pagetable().
|
|
|
|
*/
|
|
|
|
if (max_pfn_mapped)
|
|
|
|
vaddr_end = __START_KERNEL_map + (max_pfn_mapped << PAGE_SHIFT);
|
|
|
|
|
2011-02-18 19:30:30 +08:00
|
|
|
for (; vaddr + PMD_SIZE - 1 < vaddr_end; pmd++, vaddr += PMD_SIZE) {
|
2008-05-29 02:36:07 +08:00
|
|
|
if (pmd_none(*pmd))
|
2008-02-16 00:29:12 +08:00
|
|
|
continue;
|
|
|
|
if (vaddr < (unsigned long) _text || vaddr > end)
|
|
|
|
set_pmd(pmd, __pmd(0));
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2016-06-22 08:46:59 +08:00
|
|
|
/*
|
|
|
|
* Create PTE level page table mapping for physical addresses.
|
|
|
|
* It returns the last physical address mapped.
|
|
|
|
*/
|
2008-07-10 11:15:02 +08:00
|
|
|
static unsigned long __meminit
|
2016-06-22 08:46:59 +08:00
|
|
|
phys_pte_init(pte_t *pte_page, unsigned long paddr, unsigned long paddr_end,
|
2008-10-08 04:58:46 +08:00
|
|
|
pgprot_t prot)
|
2008-06-25 12:19:19 +08:00
|
|
|
{
|
2016-06-22 08:46:59 +08:00
|
|
|
unsigned long pages = 0, paddr_next;
|
|
|
|
unsigned long paddr_last = paddr_end;
|
|
|
|
pte_t *pte;
|
2008-06-25 12:19:19 +08:00
|
|
|
int i;
|
2008-07-10 11:15:02 +08:00
|
|
|
|
2016-06-22 08:46:59 +08:00
|
|
|
pte = pte_page + pte_index(paddr);
|
|
|
|
i = pte_index(paddr);
|
2008-06-25 12:19:19 +08:00
|
|
|
|
2016-06-22 08:46:59 +08:00
|
|
|
for (; i < PTRS_PER_PTE; i++, paddr = paddr_next, pte++) {
|
|
|
|
paddr_next = (paddr & PAGE_MASK) + PAGE_SIZE;
|
|
|
|
if (paddr >= paddr_end) {
|
2012-11-17 11:38:56 +08:00
|
|
|
if (!after_bootmem &&
|
2017-01-28 21:14:25 +08:00
|
|
|
!e820__mapped_any(paddr & PAGE_MASK, paddr_next,
|
2017-01-29 00:09:33 +08:00
|
|
|
E820_TYPE_RAM) &&
|
2017-01-28 21:14:25 +08:00
|
|
|
!e820__mapped_any(paddr & PAGE_MASK, paddr_next,
|
2017-01-29 00:09:33 +08:00
|
|
|
E820_TYPE_RESERVED_KERN))
|
2012-11-17 11:38:56 +08:00
|
|
|
set_pte(pte, __pte(0));
|
|
|
|
continue;
|
2008-06-25 12:19:19 +08:00
|
|
|
}
|
|
|
|
|
2008-10-08 04:58:46 +08:00
|
|
|
/*
|
|
|
|
* We will re-use the existing mapping.
|
|
|
|
* Xen for example has some special requirements, like mapping
|
|
|
|
* pagetable pages as RO. So assume someone who pre-setup
|
|
|
|
* these mappings are more intelligent.
|
|
|
|
*/
|
2016-07-08 08:19:15 +08:00
|
|
|
if (!pte_none(*pte)) {
|
2012-10-04 21:48:10 +08:00
|
|
|
if (!after_bootmem)
|
|
|
|
pages++;
|
2008-06-25 12:19:19 +08:00
|
|
|
continue;
|
2008-10-26 13:58:21 +08:00
|
|
|
}
|
2008-06-25 12:19:19 +08:00
|
|
|
|
|
|
|
if (0)
|
2016-06-22 08:46:59 +08:00
|
|
|
pr_info(" pte=%p addr=%lx pte=%016lx\n", pte, paddr,
|
|
|
|
pfn_pte(paddr >> PAGE_SHIFT, PAGE_KERNEL).pte);
|
2008-06-25 12:19:19 +08:00
|
|
|
pages++;
|
2016-06-22 08:46:59 +08:00
|
|
|
set_pte(pte, pfn_pte(paddr >> PAGE_SHIFT, prot));
|
|
|
|
paddr_last = (paddr & PAGE_MASK) + PAGE_SIZE;
|
2008-06-25 12:19:19 +08:00
|
|
|
}
|
2008-09-24 05:00:38 +08:00
|
|
|
|
2008-06-25 12:19:19 +08:00
|
|
|
update_page_count(PG_LEVEL_4K, pages);
|
2008-07-10 11:15:02 +08:00
|
|
|
|
2016-06-22 08:46:59 +08:00
|
|
|
return paddr_last;
|
2008-06-25 12:19:19 +08:00
|
|
|
}
|
|
|
|
|
2016-06-22 08:46:59 +08:00
|
|
|
/*
|
|
|
|
* Create PMD level page table mapping for physical addresses. The virtual
|
|
|
|
* and physical address have to be aligned at this level.
|
|
|
|
* It returns the last physical address mapped.
|
|
|
|
*/
|
2008-03-12 10:53:28 +08:00
|
|
|
static unsigned long __meminit
|
2016-06-22 08:46:59 +08:00
|
|
|
phys_pmd_init(pmd_t *pmd_page, unsigned long paddr, unsigned long paddr_end,
|
2008-10-08 04:58:46 +08:00
|
|
|
unsigned long page_size_mask, pgprot_t prot)
|
2006-01-17 14:03:41 +08:00
|
|
|
{
|
2016-06-22 08:46:59 +08:00
|
|
|
unsigned long pages = 0, paddr_next;
|
|
|
|
unsigned long paddr_last = paddr_end;
|
2008-05-02 17:46:49 +08:00
|
|
|
|
2016-06-22 08:46:59 +08:00
|
|
|
int i = pmd_index(paddr);
|
2006-01-17 14:03:41 +08:00
|
|
|
|
2016-06-22 08:46:59 +08:00
|
|
|
for (; i < PTRS_PER_PMD; i++, paddr = paddr_next) {
|
|
|
|
pmd_t *pmd = pmd_page + pmd_index(paddr);
|
2008-06-25 12:19:19 +08:00
|
|
|
pte_t *pte;
|
2008-10-08 04:58:46 +08:00
|
|
|
pgprot_t new_prot = prot;
|
2006-01-17 14:03:41 +08:00
|
|
|
|
2016-06-22 08:46:59 +08:00
|
|
|
paddr_next = (paddr & PMD_MASK) + PMD_SIZE;
|
|
|
|
if (paddr >= paddr_end) {
|
2012-11-17 11:38:56 +08:00
|
|
|
if (!after_bootmem &&
|
2017-01-28 21:14:25 +08:00
|
|
|
!e820__mapped_any(paddr & PMD_MASK, paddr_next,
|
2017-01-29 00:09:33 +08:00
|
|
|
E820_TYPE_RAM) &&
|
2017-01-28 21:14:25 +08:00
|
|
|
!e820__mapped_any(paddr & PMD_MASK, paddr_next,
|
2017-01-29 00:09:33 +08:00
|
|
|
E820_TYPE_RESERVED_KERN))
|
2012-11-17 11:38:56 +08:00
|
|
|
set_pmd(pmd, __pmd(0));
|
|
|
|
continue;
|
2006-01-17 14:03:41 +08:00
|
|
|
}
|
2006-09-26 16:52:36 +08:00
|
|
|
|
2016-07-08 08:19:15 +08:00
|
|
|
if (!pmd_none(*pmd)) {
|
2008-08-21 21:27:22 +08:00
|
|
|
if (!pmd_large(*pmd)) {
|
|
|
|
spin_lock(&init_mm.page_table_lock);
|
2012-11-17 11:38:59 +08:00
|
|
|
pte = (pte_t *)pmd_page_vaddr(*pmd);
|
2016-06-22 08:46:59 +08:00
|
|
|
paddr_last = phys_pte_init(pte, paddr,
|
|
|
|
paddr_end, prot);
|
2008-08-21 21:27:22 +08:00
|
|
|
spin_unlock(&init_mm.page_table_lock);
|
2008-09-24 05:00:38 +08:00
|
|
|
continue;
|
2008-08-21 21:27:22 +08:00
|
|
|
}
|
2008-10-08 04:58:46 +08:00
|
|
|
/*
|
|
|
|
* If we are ok with PG_LEVEL_2M mapping, then we will
|
|
|
|
* use the existing mapping,
|
|
|
|
*
|
|
|
|
* Otherwise, we will split the large page mapping but
|
|
|
|
* use the same existing protection bits except for
|
|
|
|
* large page, so that we don't violate Intel's TLB
|
|
|
|
* Application note (317080) which says, while changing
|
|
|
|
* the page sizes, new and old translations should
|
|
|
|
* not differ with respect to page frame and
|
|
|
|
* attributes.
|
|
|
|
*/
|
2008-10-26 13:58:21 +08:00
|
|
|
if (page_size_mask & (1 << PG_LEVEL_2M)) {
|
2012-10-04 21:48:10 +08:00
|
|
|
if (!after_bootmem)
|
|
|
|
pages++;
|
2016-06-22 08:46:59 +08:00
|
|
|
paddr_last = paddr_next;
|
2008-10-08 04:58:46 +08:00
|
|
|
continue;
|
2008-10-26 13:58:21 +08:00
|
|
|
}
|
2008-10-08 04:58:46 +08:00
|
|
|
new_prot = pte_pgprot(pte_clrhuge(*(pte_t *)pmd));
|
2008-06-25 12:19:19 +08:00
|
|
|
}
|
|
|
|
|
2008-07-08 16:41:05 +08:00
|
|
|
if (page_size_mask & (1<<PG_LEVEL_2M)) {
|
2008-06-25 12:19:19 +08:00
|
|
|
pages++;
|
2008-08-21 21:27:22 +08:00
|
|
|
spin_lock(&init_mm.page_table_lock);
|
2008-06-25 12:19:19 +08:00
|
|
|
set_pte((pte_t *)pmd,
|
2016-06-22 08:46:59 +08:00
|
|
|
pfn_pte((paddr & PMD_MASK) >> PAGE_SHIFT,
|
2008-10-08 04:58:46 +08:00
|
|
|
__pgprot(pgprot_val(prot) | _PAGE_PSE)));
|
2008-08-21 21:27:22 +08:00
|
|
|
spin_unlock(&init_mm.page_table_lock);
|
2016-06-22 08:46:59 +08:00
|
|
|
paddr_last = paddr_next;
|
2006-09-26 16:52:36 +08:00
|
|
|
continue;
|
2008-06-25 12:19:19 +08:00
|
|
|
}
|
2006-09-26 16:52:36 +08:00
|
|
|
|
2012-11-17 11:39:00 +08:00
|
|
|
pte = alloc_low_page();
|
2016-06-22 08:46:59 +08:00
|
|
|
paddr_last = phys_pte_init(pte, paddr, paddr_end, new_prot);
|
2008-06-25 12:19:19 +08:00
|
|
|
|
2008-08-21 21:27:22 +08:00
|
|
|
spin_lock(&init_mm.page_table_lock);
|
2012-11-17 11:39:00 +08:00
|
|
|
pmd_populate_kernel(&init_mm, pmd, pte);
|
2008-08-21 21:27:22 +08:00
|
|
|
spin_unlock(&init_mm.page_table_lock);
|
2006-01-17 14:03:41 +08:00
|
|
|
}
|
2008-05-02 17:46:49 +08:00
|
|
|
update_page_count(PG_LEVEL_2M, pages);
|
2016-06-22 08:46:59 +08:00
|
|
|
return paddr_last;
|
2006-01-17 14:03:41 +08:00
|
|
|
}
|
|
|
|
|
2016-06-22 08:46:59 +08:00
|
|
|
/*
|
|
|
|
* Create PUD level page table mapping for physical addresses. The virtual
|
2016-06-22 08:47:00 +08:00
|
|
|
* and physical address do not have to be aligned at this level. KASLR can
|
|
|
|
* randomize virtual addresses up to this level.
|
2016-06-22 08:46:59 +08:00
|
|
|
* It returns the last physical address mapped.
|
|
|
|
*/
|
2008-03-12 10:53:28 +08:00
|
|
|
static unsigned long __meminit
|
2016-06-22 08:46:59 +08:00
|
|
|
phys_pud_init(pud_t *pud_page, unsigned long paddr, unsigned long paddr_end,
|
|
|
|
unsigned long page_size_mask)
|
2008-01-30 20:34:10 +08:00
|
|
|
{
|
2016-06-22 08:46:59 +08:00
|
|
|
unsigned long pages = 0, paddr_next;
|
|
|
|
unsigned long paddr_last = paddr_end;
|
2016-06-22 08:47:00 +08:00
|
|
|
unsigned long vaddr = (unsigned long)__va(paddr);
|
|
|
|
int i = pud_index(vaddr);
|
2006-01-17 14:03:41 +08:00
|
|
|
|
2016-06-22 08:46:59 +08:00
|
|
|
for (; i < PTRS_PER_PUD; i++, paddr = paddr_next) {
|
2016-06-22 08:47:00 +08:00
|
|
|
pud_t *pud;
|
2005-04-17 06:20:36 +08:00
|
|
|
pmd_t *pmd;
|
2008-10-08 04:58:46 +08:00
|
|
|
pgprot_t prot = PAGE_KERNEL;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2016-06-22 08:47:00 +08:00
|
|
|
vaddr = (unsigned long)__va(paddr);
|
|
|
|
pud = pud_page + pud_index(vaddr);
|
2016-06-22 08:46:59 +08:00
|
|
|
paddr_next = (paddr & PUD_MASK) + PUD_SIZE;
|
2016-06-22 08:47:00 +08:00
|
|
|
|
2016-06-22 08:46:59 +08:00
|
|
|
if (paddr >= paddr_end) {
|
2012-11-17 11:38:56 +08:00
|
|
|
if (!after_bootmem &&
|
2017-01-28 21:14:25 +08:00
|
|
|
!e820__mapped_any(paddr & PUD_MASK, paddr_next,
|
2017-01-29 00:09:33 +08:00
|
|
|
E820_TYPE_RAM) &&
|
2017-01-28 21:14:25 +08:00
|
|
|
!e820__mapped_any(paddr & PUD_MASK, paddr_next,
|
2017-01-29 00:09:33 +08:00
|
|
|
E820_TYPE_RESERVED_KERN))
|
2012-11-17 11:38:56 +08:00
|
|
|
set_pud(pud, __pud(0));
|
2005-04-17 06:20:36 +08:00
|
|
|
continue;
|
2008-01-30 20:34:10 +08:00
|
|
|
}
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2016-07-08 08:19:15 +08:00
|
|
|
if (!pud_none(*pud)) {
|
2008-09-24 05:00:38 +08:00
|
|
|
if (!pud_large(*pud)) {
|
2012-11-17 11:38:59 +08:00
|
|
|
pmd = pmd_offset(pud, 0);
|
2016-06-22 08:46:59 +08:00
|
|
|
paddr_last = phys_pmd_init(pmd, paddr,
|
|
|
|
paddr_end,
|
|
|
|
page_size_mask,
|
|
|
|
prot);
|
x86-64, mm: Put early page table high
While dubug kdump, found current kernel will have problem with crashkernel=512M.
It turns out that initial mapping is to 512M, and later initial mapping to 4G
(acutally is 2040M in my platform), will put page table near 512M.
then initial mapping to 128g will be near 2g.
before this patch:
[ 0.000000] initial memory mapped : 0 - 20000000
[ 0.000000] init_memory_mapping: [0x00000000000000-0x0000007f74ffff]
[ 0.000000] 0000000000 - 007f600000 page 2M
[ 0.000000] 007f600000 - 007f750000 page 4k
[ 0.000000] kernel direct mapping tables up to 7f750000 @ [0x1fffc000-0x1fffffff]
[ 0.000000] memblock_x86_reserve_range: [0x1fffc000-0x1fffdfff] PGTABLE
[ 0.000000] init_memory_mapping: [0x00000100000000-0x0000207fffffff]
[ 0.000000] 0100000000 - 2080000000 page 2M
[ 0.000000] kernel direct mapping tables up to 2080000000 @ [0x7bc01000-0x7bc83fff]
[ 0.000000] memblock_x86_reserve_range: [0x7bc01000-0x7bc7efff] PGTABLE
[ 0.000000] RAMDISK: 7bc84000 - 7f745000
[ 0.000000] crashkernel reservation failed - No suitable area found.
after patch:
[ 0.000000] initial memory mapped : 0 - 20000000
[ 0.000000] init_memory_mapping: [0x00000000000000-0x0000007f74ffff]
[ 0.000000] 0000000000 - 007f600000 page 2M
[ 0.000000] 007f600000 - 007f750000 page 4k
[ 0.000000] kernel direct mapping tables up to 7f750000 @ [0x7f74c000-0x7f74ffff]
[ 0.000000] memblock_x86_reserve_range: [0x7f74c000-0x7f74dfff] PGTABLE
[ 0.000000] init_memory_mapping: [0x00000100000000-0x0000207fffffff]
[ 0.000000] 0100000000 - 2080000000 page 2M
[ 0.000000] kernel direct mapping tables up to 2080000000 @ [0x207ff7d000-0x207fffffff]
[ 0.000000] memblock_x86_reserve_range: [0x207ff7d000-0x207fffafff] PGTABLE
[ 0.000000] RAMDISK: 7bc84000 - 7f745000
[ 0.000000] memblock_x86_reserve_range: [0x17000000-0x36ffffff] CRASH KERNEL
[ 0.000000] Reserving 512MB of memory at 368MB for crashkernel (System RAM: 133120MB)
It means with the patch, page table for [0, 2g) will need 2g, instead of under 512M,
page table for [4g, 128g) will be near 128g, instead of under 2g.
That would good, if we have lots of memory above 4g, like 1024g, or 2048g or 16T, will not put
related page table under 2g. that would be have chance to fill the under 2g if 1G or 2M page is
not used.
the code change will use add map_low_page() and update unmap_low_page() for 64bit, and use them
to get access the corresponding high memory for page table setting.
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
LKML-Reference: <4D0C0734.7060900@kernel.org>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
2010-12-18 08:58:28 +08:00
|
|
|
__flush_tlb_all();
|
2008-09-24 05:00:38 +08:00
|
|
|
continue;
|
|
|
|
}
|
2008-10-08 04:58:46 +08:00
|
|
|
/*
|
|
|
|
* If we are ok with PG_LEVEL_1G mapping, then we will
|
|
|
|
* use the existing mapping.
|
|
|
|
*
|
|
|
|
* Otherwise, we will split the gbpage mapping but use
|
|
|
|
* the same existing protection bits except for large
|
|
|
|
* page, so that we don't violate Intel's TLB
|
|
|
|
* Application note (317080) which says, while changing
|
|
|
|
* the page sizes, new and old translations should
|
|
|
|
* not differ with respect to page frame and
|
|
|
|
* attributes.
|
|
|
|
*/
|
2008-10-26 13:58:21 +08:00
|
|
|
if (page_size_mask & (1 << PG_LEVEL_1G)) {
|
2012-10-04 21:48:10 +08:00
|
|
|
if (!after_bootmem)
|
|
|
|
pages++;
|
2016-06-22 08:46:59 +08:00
|
|
|
paddr_last = paddr_next;
|
2008-10-08 04:58:46 +08:00
|
|
|
continue;
|
2008-10-26 13:58:21 +08:00
|
|
|
}
|
2008-10-08 04:58:46 +08:00
|
|
|
prot = pte_pgprot(pte_clrhuge(*(pte_t *)pud));
|
2008-04-17 23:40:45 +08:00
|
|
|
}
|
|
|
|
|
2008-07-08 16:41:05 +08:00
|
|
|
if (page_size_mask & (1<<PG_LEVEL_1G)) {
|
2008-05-02 17:46:49 +08:00
|
|
|
pages++;
|
2008-08-21 21:27:22 +08:00
|
|
|
spin_lock(&init_mm.page_table_lock);
|
2008-04-17 23:40:45 +08:00
|
|
|
set_pte((pte_t *)pud,
|
2016-06-22 08:46:59 +08:00
|
|
|
pfn_pte((paddr & PUD_MASK) >> PAGE_SHIFT,
|
2012-11-17 11:38:54 +08:00
|
|
|
PAGE_KERNEL_LARGE));
|
2008-08-21 21:27:22 +08:00
|
|
|
spin_unlock(&init_mm.page_table_lock);
|
2016-06-22 08:46:59 +08:00
|
|
|
paddr_last = paddr_next;
|
2006-09-26 16:52:36 +08:00
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
2012-11-17 11:39:00 +08:00
|
|
|
pmd = alloc_low_page();
|
2016-06-22 08:46:59 +08:00
|
|
|
paddr_last = phys_pmd_init(pmd, paddr, paddr_end,
|
|
|
|
page_size_mask, prot);
|
2008-08-21 21:27:22 +08:00
|
|
|
|
|
|
|
spin_lock(&init_mm.page_table_lock);
|
2012-11-17 11:39:00 +08:00
|
|
|
pud_populate(&init_mm, pud, pmd);
|
2006-01-17 14:03:41 +08:00
|
|
|
spin_unlock(&init_mm.page_table_lock);
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
2008-01-30 20:33:54 +08:00
|
|
|
__flush_tlb_all();
|
2008-09-24 05:00:38 +08:00
|
|
|
|
2008-05-02 17:46:49 +08:00
|
|
|
update_page_count(PG_LEVEL_1G, pages);
|
2008-03-12 10:53:28 +08:00
|
|
|
|
2016-06-22 08:46:59 +08:00
|
|
|
return paddr_last;
|
2008-01-30 20:34:10 +08:00
|
|
|
}
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2017-06-06 19:31:30 +08:00
|
|
|
static unsigned long __meminit
|
|
|
|
phys_p4d_init(p4d_t *p4d_page, unsigned long paddr, unsigned long paddr_end,
|
|
|
|
unsigned long page_size_mask)
|
|
|
|
{
|
|
|
|
unsigned long paddr_next, paddr_last = paddr_end;
|
|
|
|
unsigned long vaddr = (unsigned long)__va(paddr);
|
|
|
|
int i = p4d_index(vaddr);
|
|
|
|
|
|
|
|
if (!IS_ENABLED(CONFIG_X86_5LEVEL))
|
|
|
|
return phys_pud_init((pud_t *) p4d_page, paddr, paddr_end, page_size_mask);
|
|
|
|
|
|
|
|
for (; i < PTRS_PER_P4D; i++, paddr = paddr_next) {
|
|
|
|
p4d_t *p4d;
|
|
|
|
pud_t *pud;
|
|
|
|
|
|
|
|
vaddr = (unsigned long)__va(paddr);
|
|
|
|
p4d = p4d_page + p4d_index(vaddr);
|
|
|
|
paddr_next = (paddr & P4D_MASK) + P4D_SIZE;
|
|
|
|
|
|
|
|
if (paddr >= paddr_end) {
|
|
|
|
if (!after_bootmem &&
|
|
|
|
!e820__mapped_any(paddr & P4D_MASK, paddr_next,
|
|
|
|
E820_TYPE_RAM) &&
|
|
|
|
!e820__mapped_any(paddr & P4D_MASK, paddr_next,
|
|
|
|
E820_TYPE_RESERVED_KERN))
|
|
|
|
set_p4d(p4d, __p4d(0));
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!p4d_none(*p4d)) {
|
|
|
|
pud = pud_offset(p4d, 0);
|
|
|
|
paddr_last = phys_pud_init(pud, paddr,
|
|
|
|
paddr_end,
|
|
|
|
page_size_mask);
|
|
|
|
__flush_tlb_all();
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
|
|
|
pud = alloc_low_page();
|
|
|
|
paddr_last = phys_pud_init(pud, paddr, paddr_end,
|
|
|
|
page_size_mask);
|
|
|
|
|
|
|
|
spin_lock(&init_mm.page_table_lock);
|
|
|
|
p4d_populate(&init_mm, p4d, pud);
|
|
|
|
spin_unlock(&init_mm.page_table_lock);
|
|
|
|
}
|
|
|
|
__flush_tlb_all();
|
|
|
|
|
|
|
|
return paddr_last;
|
|
|
|
}
|
|
|
|
|
2016-06-22 08:46:59 +08:00
|
|
|
/*
|
|
|
|
* Create page table mapping for the physical memory for specific physical
|
2016-06-22 08:47:00 +08:00
|
|
|
* addresses. The virtual and physical addresses have to be aligned on PMD level
|
2016-06-22 08:46:59 +08:00
|
|
|
* down. It returns the last physical address mapped.
|
|
|
|
*/
|
2009-06-12 12:57:52 +08:00
|
|
|
unsigned long __meminit
|
2016-06-22 08:46:59 +08:00
|
|
|
kernel_physical_mapping_init(unsigned long paddr_start,
|
|
|
|
unsigned long paddr_end,
|
2009-03-05 20:55:05 +08:00
|
|
|
unsigned long page_size_mask)
|
2008-01-30 20:34:10 +08:00
|
|
|
{
|
2010-08-20 17:50:16 +08:00
|
|
|
bool pgd_changed = false;
|
2016-06-22 08:46:59 +08:00
|
|
|
unsigned long vaddr, vaddr_start, vaddr_end, vaddr_next, paddr_last;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2016-06-22 08:46:59 +08:00
|
|
|
paddr_last = paddr_end;
|
|
|
|
vaddr = (unsigned long)__va(paddr_start);
|
|
|
|
vaddr_end = (unsigned long)__va(paddr_end);
|
|
|
|
vaddr_start = vaddr;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2016-06-22 08:46:59 +08:00
|
|
|
for (; vaddr < vaddr_end; vaddr = vaddr_next) {
|
|
|
|
pgd_t *pgd = pgd_offset_k(vaddr);
|
2017-03-18 02:55:15 +08:00
|
|
|
p4d_t *p4d;
|
2006-01-17 14:03:41 +08:00
|
|
|
|
2016-06-22 08:46:59 +08:00
|
|
|
vaddr_next = (vaddr & PGDIR_MASK) + PGDIR_SIZE;
|
2008-06-25 12:19:19 +08:00
|
|
|
|
2017-06-06 19:31:30 +08:00
|
|
|
if (pgd_val(*pgd)) {
|
|
|
|
p4d = (p4d_t *)pgd_page_vaddr(*pgd);
|
|
|
|
paddr_last = phys_p4d_init(p4d, __pa(vaddr),
|
2016-06-22 08:46:59 +08:00
|
|
|
__pa(vaddr_end),
|
|
|
|
page_size_mask);
|
2008-06-25 12:19:19 +08:00
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
2017-06-06 19:31:30 +08:00
|
|
|
p4d = alloc_low_page();
|
|
|
|
paddr_last = phys_p4d_init(p4d, __pa(vaddr), __pa(vaddr_end),
|
2016-06-22 08:46:59 +08:00
|
|
|
page_size_mask);
|
2008-08-21 21:27:22 +08:00
|
|
|
|
|
|
|
spin_lock(&init_mm.page_table_lock);
|
2017-06-06 19:31:30 +08:00
|
|
|
if (IS_ENABLED(CONFIG_X86_5LEVEL))
|
|
|
|
pgd_populate(&init_mm, pgd, p4d);
|
|
|
|
else
|
|
|
|
p4d_populate(&init_mm, p4d_offset(pgd, vaddr), (pud_t *) p4d);
|
2008-08-21 21:27:22 +08:00
|
|
|
spin_unlock(&init_mm.page_table_lock);
|
2010-08-20 17:50:16 +08:00
|
|
|
pgd_changed = true;
|
2008-01-30 20:34:10 +08:00
|
|
|
}
|
2010-08-20 17:50:16 +08:00
|
|
|
|
|
|
|
if (pgd_changed)
|
2016-12-15 07:44:03 +08:00
|
|
|
sync_global_pgds(vaddr_start, vaddr_end - 1);
|
2010-08-20 17:50:16 +08:00
|
|
|
|
2008-09-24 05:00:38 +08:00
|
|
|
__flush_tlb_all();
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2016-06-22 08:46:59 +08:00
|
|
|
return paddr_last;
|
2008-07-08 16:41:05 +08:00
|
|
|
}
|
2008-07-10 11:15:02 +08:00
|
|
|
|
2005-06-23 15:08:06 +08:00
|
|
|
#ifndef CONFIG_NUMA
|
2011-02-16 19:13:06 +08:00
|
|
|
void __init initmem_init(void)
|
2008-06-22 17:44:49 +08:00
|
|
|
{
|
2014-01-22 07:49:26 +08:00
|
|
|
memblock_set_node(0, (phys_addr_t)ULLONG_MAX, &memblock.memory, 0);
|
2008-06-22 17:44:49 +08:00
|
|
|
}
|
2009-05-07 20:35:41 +08:00
|
|
|
#endif
|
2008-06-22 17:44:49 +08:00
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
void __init paging_init(void)
|
|
|
|
{
|
2009-05-07 20:35:41 +08:00
|
|
|
sparse_memory_present_with_active_regions(MAX_NUMNODES);
|
2006-01-17 14:03:41 +08:00
|
|
|
sparse_init();
|
2009-07-09 00:50:19 +08:00
|
|
|
|
|
|
|
/*
|
|
|
|
* clear the default setting with node 0
|
|
|
|
* note: don't use nodes_clear here, that is really clearing when
|
|
|
|
* numa support is not compiled in, and later node_set_state
|
|
|
|
* will not set it back.
|
|
|
|
*/
|
2012-12-13 05:51:46 +08:00
|
|
|
node_clear_state(0, N_MEMORY);
|
|
|
|
if (N_MEMORY != N_NORMAL_MEMORY)
|
|
|
|
node_clear_state(0, N_NORMAL_MEMORY);
|
2009-07-09 00:50:19 +08:00
|
|
|
|
2011-11-01 21:58:17 +08:00
|
|
|
zone_sizes_init();
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
|
|
|
|
2006-01-17 14:03:41 +08:00
|
|
|
/*
|
|
|
|
* Memory hotplug specific functions
|
|
|
|
*/
|
2006-06-27 17:53:30 +08:00
|
|
|
#ifdef CONFIG_MEMORY_HOTPLUG
|
2010-02-03 05:44:16 +08:00
|
|
|
/*
|
|
|
|
* After memory hotplug the variables max_pfn, max_low_pfn and high_memory need
|
|
|
|
* updating.
|
|
|
|
*/
|
|
|
|
static void update_end_of_memory_vars(u64 start, u64 size)
|
|
|
|
{
|
|
|
|
unsigned long end_pfn = PFN_UP(start + size);
|
|
|
|
|
|
|
|
if (end_pfn > max_pfn) {
|
|
|
|
max_pfn = end_pfn;
|
|
|
|
max_low_pfn = end_pfn;
|
|
|
|
high_memory = (void *)__va(max_pfn * PAGE_SIZE - 1) + 1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2017-07-07 06:38:21 +08:00
|
|
|
int arch_add_memory(int nid, u64 start, u64 size, bool want_memblock)
|
2006-01-17 14:03:41 +08:00
|
|
|
{
|
2012-11-17 11:38:52 +08:00
|
|
|
unsigned long start_pfn = start >> PAGE_SHIFT;
|
2006-01-17 14:03:41 +08:00
|
|
|
unsigned long nr_pages = size >> PAGE_SHIFT;
|
|
|
|
int ret;
|
|
|
|
|
2012-11-17 11:38:52 +08:00
|
|
|
init_memory_mapping(start, start + size);
|
2006-10-01 14:27:09 +08:00
|
|
|
|
2017-07-07 06:38:21 +08:00
|
|
|
ret = __add_pages(nid, start_pfn, nr_pages, want_memblock);
|
2008-10-29 07:43:14 +08:00
|
|
|
WARN_ON_ONCE(ret);
|
2006-01-17 14:03:41 +08:00
|
|
|
|
2010-02-03 05:44:16 +08:00
|
|
|
/* update max_pfn, max_low_pfn and high_memory */
|
|
|
|
update_end_of_memory_vars(start, size);
|
|
|
|
|
2006-01-17 14:03:41 +08:00
|
|
|
return ret;
|
|
|
|
}
|
2006-06-27 17:53:30 +08:00
|
|
|
EXPORT_SYMBOL_GPL(arch_add_memory);
|
2006-01-17 14:03:41 +08:00
|
|
|
|
memory-hotplug: common APIs to support page tables hot-remove
When memory is removed, the corresponding pagetables should alse be
removed. This patch introduces some common APIs to support vmemmap
pagetable and x86_64 architecture direct mapping pagetable removing.
All pages of virtual mapping in removed memory cannot be freed if some
pages used as PGD/PUD include not only removed memory but also other
memory. So this patch uses the following way to check whether a page
can be freed or not.
1) When removing memory, the page structs of the removed memory are
filled with 0FD.
2) All page structs are filled with 0xFD on PT/PMD, PT/PMD can be
cleared. In this case, the page used as PT/PMD can be freed.
For direct mapping pages, update direct_pages_count[level] when we freed
their pagetables. And do not free the pages again because they were
freed when offlining.
For vmemmap pages, free the pages and their pagetables.
For larger pages, do not split them into smaller ones because there is
no way to know if the larger page has been split. As a result, there is
no way to decide when to split. We deal the larger pages in the
following way:
1) For direct mapped pages, all the pages were freed when they were
offlined. And since menmory offline is done section by section, all
the memory ranges being removed are aligned to PAGE_SIZE. So only need
to deal with unaligned pages when freeing vmemmap pages.
2) For vmemmap pages being used to store page_struct, if part of the
larger page is still in use, just fill the unused part with 0xFD. And
when the whole page is fulfilled with 0xFD, then free the larger page.
[akpm@linux-foundation.org: fix typo in comment]
[tangchen@cn.fujitsu.com: do not calculate direct mapping pages when freeing vmemmap pagetables]
[tangchen@cn.fujitsu.com: do not free direct mapping pages twice]
[tangchen@cn.fujitsu.com: do not free page split from hugepage one by one]
[tangchen@cn.fujitsu.com: do not split pages when freeing pagetable pages]
[akpm@linux-foundation.org: use pmd_page_vaddr()]
[akpm@linux-foundation.org: fix used-uninitialised bug]
Signed-off-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Signed-off-by: Jianguo Wu <wujianguo@huawei.com>
Signed-off-by: Wen Congyang <wency@cn.fujitsu.com>
Signed-off-by: Tang Chen <tangchen@cn.fujitsu.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Jiang Liu <jiang.liu@huawei.com>
Cc: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Cc: Wu Jianguo <wujianguo@huawei.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-23 08:33:04 +08:00
|
|
|
#define PAGE_INUSE 0xFD
|
|
|
|
|
|
|
|
static void __meminit free_pagetable(struct page *page, int order)
|
|
|
|
{
|
|
|
|
unsigned long magic;
|
|
|
|
unsigned int nr_pages = 1 << order;
|
2016-01-16 08:56:22 +08:00
|
|
|
struct vmem_altmap *altmap = to_vmem_altmap((unsigned long) page);
|
|
|
|
|
|
|
|
if (altmap) {
|
|
|
|
vmem_altmap_free(altmap, nr_pages);
|
|
|
|
return;
|
|
|
|
}
|
memory-hotplug: common APIs to support page tables hot-remove
When memory is removed, the corresponding pagetables should alse be
removed. This patch introduces some common APIs to support vmemmap
pagetable and x86_64 architecture direct mapping pagetable removing.
All pages of virtual mapping in removed memory cannot be freed if some
pages used as PGD/PUD include not only removed memory but also other
memory. So this patch uses the following way to check whether a page
can be freed or not.
1) When removing memory, the page structs of the removed memory are
filled with 0FD.
2) All page structs are filled with 0xFD on PT/PMD, PT/PMD can be
cleared. In this case, the page used as PT/PMD can be freed.
For direct mapping pages, update direct_pages_count[level] when we freed
their pagetables. And do not free the pages again because they were
freed when offlining.
For vmemmap pages, free the pages and their pagetables.
For larger pages, do not split them into smaller ones because there is
no way to know if the larger page has been split. As a result, there is
no way to decide when to split. We deal the larger pages in the
following way:
1) For direct mapped pages, all the pages were freed when they were
offlined. And since menmory offline is done section by section, all
the memory ranges being removed are aligned to PAGE_SIZE. So only need
to deal with unaligned pages when freeing vmemmap pages.
2) For vmemmap pages being used to store page_struct, if part of the
larger page is still in use, just fill the unused part with 0xFD. And
when the whole page is fulfilled with 0xFD, then free the larger page.
[akpm@linux-foundation.org: fix typo in comment]
[tangchen@cn.fujitsu.com: do not calculate direct mapping pages when freeing vmemmap pagetables]
[tangchen@cn.fujitsu.com: do not free direct mapping pages twice]
[tangchen@cn.fujitsu.com: do not free page split from hugepage one by one]
[tangchen@cn.fujitsu.com: do not split pages when freeing pagetable pages]
[akpm@linux-foundation.org: use pmd_page_vaddr()]
[akpm@linux-foundation.org: fix used-uninitialised bug]
Signed-off-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Signed-off-by: Jianguo Wu <wujianguo@huawei.com>
Signed-off-by: Wen Congyang <wency@cn.fujitsu.com>
Signed-off-by: Tang Chen <tangchen@cn.fujitsu.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Jiang Liu <jiang.liu@huawei.com>
Cc: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Cc: Wu Jianguo <wujianguo@huawei.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-23 08:33:04 +08:00
|
|
|
|
|
|
|
/* bootmem page has reserved flag */
|
|
|
|
if (PageReserved(page)) {
|
|
|
|
__ClearPageReserved(page);
|
|
|
|
|
2017-02-23 07:45:13 +08:00
|
|
|
magic = (unsigned long)page->freelist;
|
memory-hotplug: common APIs to support page tables hot-remove
When memory is removed, the corresponding pagetables should alse be
removed. This patch introduces some common APIs to support vmemmap
pagetable and x86_64 architecture direct mapping pagetable removing.
All pages of virtual mapping in removed memory cannot be freed if some
pages used as PGD/PUD include not only removed memory but also other
memory. So this patch uses the following way to check whether a page
can be freed or not.
1) When removing memory, the page structs of the removed memory are
filled with 0FD.
2) All page structs are filled with 0xFD on PT/PMD, PT/PMD can be
cleared. In this case, the page used as PT/PMD can be freed.
For direct mapping pages, update direct_pages_count[level] when we freed
their pagetables. And do not free the pages again because they were
freed when offlining.
For vmemmap pages, free the pages and their pagetables.
For larger pages, do not split them into smaller ones because there is
no way to know if the larger page has been split. As a result, there is
no way to decide when to split. We deal the larger pages in the
following way:
1) For direct mapped pages, all the pages were freed when they were
offlined. And since menmory offline is done section by section, all
the memory ranges being removed are aligned to PAGE_SIZE. So only need
to deal with unaligned pages when freeing vmemmap pages.
2) For vmemmap pages being used to store page_struct, if part of the
larger page is still in use, just fill the unused part with 0xFD. And
when the whole page is fulfilled with 0xFD, then free the larger page.
[akpm@linux-foundation.org: fix typo in comment]
[tangchen@cn.fujitsu.com: do not calculate direct mapping pages when freeing vmemmap pagetables]
[tangchen@cn.fujitsu.com: do not free direct mapping pages twice]
[tangchen@cn.fujitsu.com: do not free page split from hugepage one by one]
[tangchen@cn.fujitsu.com: do not split pages when freeing pagetable pages]
[akpm@linux-foundation.org: use pmd_page_vaddr()]
[akpm@linux-foundation.org: fix used-uninitialised bug]
Signed-off-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Signed-off-by: Jianguo Wu <wujianguo@huawei.com>
Signed-off-by: Wen Congyang <wency@cn.fujitsu.com>
Signed-off-by: Tang Chen <tangchen@cn.fujitsu.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Jiang Liu <jiang.liu@huawei.com>
Cc: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Cc: Wu Jianguo <wujianguo@huawei.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-23 08:33:04 +08:00
|
|
|
if (magic == SECTION_INFO || magic == MIX_SECTION_INFO) {
|
|
|
|
while (nr_pages--)
|
|
|
|
put_page_bootmem(page++);
|
|
|
|
} else
|
2013-07-04 06:03:17 +08:00
|
|
|
while (nr_pages--)
|
|
|
|
free_reserved_page(page++);
|
memory-hotplug: common APIs to support page tables hot-remove
When memory is removed, the corresponding pagetables should alse be
removed. This patch introduces some common APIs to support vmemmap
pagetable and x86_64 architecture direct mapping pagetable removing.
All pages of virtual mapping in removed memory cannot be freed if some
pages used as PGD/PUD include not only removed memory but also other
memory. So this patch uses the following way to check whether a page
can be freed or not.
1) When removing memory, the page structs of the removed memory are
filled with 0FD.
2) All page structs are filled with 0xFD on PT/PMD, PT/PMD can be
cleared. In this case, the page used as PT/PMD can be freed.
For direct mapping pages, update direct_pages_count[level] when we freed
their pagetables. And do not free the pages again because they were
freed when offlining.
For vmemmap pages, free the pages and their pagetables.
For larger pages, do not split them into smaller ones because there is
no way to know if the larger page has been split. As a result, there is
no way to decide when to split. We deal the larger pages in the
following way:
1) For direct mapped pages, all the pages were freed when they were
offlined. And since menmory offline is done section by section, all
the memory ranges being removed are aligned to PAGE_SIZE. So only need
to deal with unaligned pages when freeing vmemmap pages.
2) For vmemmap pages being used to store page_struct, if part of the
larger page is still in use, just fill the unused part with 0xFD. And
when the whole page is fulfilled with 0xFD, then free the larger page.
[akpm@linux-foundation.org: fix typo in comment]
[tangchen@cn.fujitsu.com: do not calculate direct mapping pages when freeing vmemmap pagetables]
[tangchen@cn.fujitsu.com: do not free direct mapping pages twice]
[tangchen@cn.fujitsu.com: do not free page split from hugepage one by one]
[tangchen@cn.fujitsu.com: do not split pages when freeing pagetable pages]
[akpm@linux-foundation.org: use pmd_page_vaddr()]
[akpm@linux-foundation.org: fix used-uninitialised bug]
Signed-off-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Signed-off-by: Jianguo Wu <wujianguo@huawei.com>
Signed-off-by: Wen Congyang <wency@cn.fujitsu.com>
Signed-off-by: Tang Chen <tangchen@cn.fujitsu.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Jiang Liu <jiang.liu@huawei.com>
Cc: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Cc: Wu Jianguo <wujianguo@huawei.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-23 08:33:04 +08:00
|
|
|
} else
|
|
|
|
free_pages((unsigned long)page_address(page), order);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void __meminit free_pte_table(pte_t *pte_start, pmd_t *pmd)
|
|
|
|
{
|
|
|
|
pte_t *pte;
|
|
|
|
int i;
|
|
|
|
|
|
|
|
for (i = 0; i < PTRS_PER_PTE; i++) {
|
|
|
|
pte = pte_start + i;
|
2016-07-08 08:19:15 +08:00
|
|
|
if (!pte_none(*pte))
|
memory-hotplug: common APIs to support page tables hot-remove
When memory is removed, the corresponding pagetables should alse be
removed. This patch introduces some common APIs to support vmemmap
pagetable and x86_64 architecture direct mapping pagetable removing.
All pages of virtual mapping in removed memory cannot be freed if some
pages used as PGD/PUD include not only removed memory but also other
memory. So this patch uses the following way to check whether a page
can be freed or not.
1) When removing memory, the page structs of the removed memory are
filled with 0FD.
2) All page structs are filled with 0xFD on PT/PMD, PT/PMD can be
cleared. In this case, the page used as PT/PMD can be freed.
For direct mapping pages, update direct_pages_count[level] when we freed
their pagetables. And do not free the pages again because they were
freed when offlining.
For vmemmap pages, free the pages and their pagetables.
For larger pages, do not split them into smaller ones because there is
no way to know if the larger page has been split. As a result, there is
no way to decide when to split. We deal the larger pages in the
following way:
1) For direct mapped pages, all the pages were freed when they were
offlined. And since menmory offline is done section by section, all
the memory ranges being removed are aligned to PAGE_SIZE. So only need
to deal with unaligned pages when freeing vmemmap pages.
2) For vmemmap pages being used to store page_struct, if part of the
larger page is still in use, just fill the unused part with 0xFD. And
when the whole page is fulfilled with 0xFD, then free the larger page.
[akpm@linux-foundation.org: fix typo in comment]
[tangchen@cn.fujitsu.com: do not calculate direct mapping pages when freeing vmemmap pagetables]
[tangchen@cn.fujitsu.com: do not free direct mapping pages twice]
[tangchen@cn.fujitsu.com: do not free page split from hugepage one by one]
[tangchen@cn.fujitsu.com: do not split pages when freeing pagetable pages]
[akpm@linux-foundation.org: use pmd_page_vaddr()]
[akpm@linux-foundation.org: fix used-uninitialised bug]
Signed-off-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Signed-off-by: Jianguo Wu <wujianguo@huawei.com>
Signed-off-by: Wen Congyang <wency@cn.fujitsu.com>
Signed-off-by: Tang Chen <tangchen@cn.fujitsu.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Jiang Liu <jiang.liu@huawei.com>
Cc: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Cc: Wu Jianguo <wujianguo@huawei.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-23 08:33:04 +08:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* free a pte talbe */
|
|
|
|
free_pagetable(pmd_page(*pmd), 0);
|
|
|
|
spin_lock(&init_mm.page_table_lock);
|
|
|
|
pmd_clear(pmd);
|
|
|
|
spin_unlock(&init_mm.page_table_lock);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void __meminit free_pmd_table(pmd_t *pmd_start, pud_t *pud)
|
|
|
|
{
|
|
|
|
pmd_t *pmd;
|
|
|
|
int i;
|
|
|
|
|
|
|
|
for (i = 0; i < PTRS_PER_PMD; i++) {
|
|
|
|
pmd = pmd_start + i;
|
2016-07-08 08:19:15 +08:00
|
|
|
if (!pmd_none(*pmd))
|
memory-hotplug: common APIs to support page tables hot-remove
When memory is removed, the corresponding pagetables should alse be
removed. This patch introduces some common APIs to support vmemmap
pagetable and x86_64 architecture direct mapping pagetable removing.
All pages of virtual mapping in removed memory cannot be freed if some
pages used as PGD/PUD include not only removed memory but also other
memory. So this patch uses the following way to check whether a page
can be freed or not.
1) When removing memory, the page structs of the removed memory are
filled with 0FD.
2) All page structs are filled with 0xFD on PT/PMD, PT/PMD can be
cleared. In this case, the page used as PT/PMD can be freed.
For direct mapping pages, update direct_pages_count[level] when we freed
their pagetables. And do not free the pages again because they were
freed when offlining.
For vmemmap pages, free the pages and their pagetables.
For larger pages, do not split them into smaller ones because there is
no way to know if the larger page has been split. As a result, there is
no way to decide when to split. We deal the larger pages in the
following way:
1) For direct mapped pages, all the pages were freed when they were
offlined. And since menmory offline is done section by section, all
the memory ranges being removed are aligned to PAGE_SIZE. So only need
to deal with unaligned pages when freeing vmemmap pages.
2) For vmemmap pages being used to store page_struct, if part of the
larger page is still in use, just fill the unused part with 0xFD. And
when the whole page is fulfilled with 0xFD, then free the larger page.
[akpm@linux-foundation.org: fix typo in comment]
[tangchen@cn.fujitsu.com: do not calculate direct mapping pages when freeing vmemmap pagetables]
[tangchen@cn.fujitsu.com: do not free direct mapping pages twice]
[tangchen@cn.fujitsu.com: do not free page split from hugepage one by one]
[tangchen@cn.fujitsu.com: do not split pages when freeing pagetable pages]
[akpm@linux-foundation.org: use pmd_page_vaddr()]
[akpm@linux-foundation.org: fix used-uninitialised bug]
Signed-off-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Signed-off-by: Jianguo Wu <wujianguo@huawei.com>
Signed-off-by: Wen Congyang <wency@cn.fujitsu.com>
Signed-off-by: Tang Chen <tangchen@cn.fujitsu.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Jiang Liu <jiang.liu@huawei.com>
Cc: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Cc: Wu Jianguo <wujianguo@huawei.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-23 08:33:04 +08:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* free a pmd talbe */
|
|
|
|
free_pagetable(pud_page(*pud), 0);
|
|
|
|
spin_lock(&init_mm.page_table_lock);
|
|
|
|
pud_clear(pud);
|
|
|
|
spin_unlock(&init_mm.page_table_lock);
|
|
|
|
}
|
|
|
|
|
2017-03-18 02:55:15 +08:00
|
|
|
static void __meminit free_pud_table(pud_t *pud_start, p4d_t *p4d)
|
|
|
|
{
|
|
|
|
pud_t *pud;
|
|
|
|
int i;
|
|
|
|
|
|
|
|
for (i = 0; i < PTRS_PER_PUD; i++) {
|
|
|
|
pud = pud_start + i;
|
|
|
|
if (!pud_none(*pud))
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* free a pud talbe */
|
|
|
|
free_pagetable(p4d_page(*p4d), 0);
|
|
|
|
spin_lock(&init_mm.page_table_lock);
|
|
|
|
p4d_clear(p4d);
|
|
|
|
spin_unlock(&init_mm.page_table_lock);
|
|
|
|
}
|
|
|
|
|
memory-hotplug: common APIs to support page tables hot-remove
When memory is removed, the corresponding pagetables should alse be
removed. This patch introduces some common APIs to support vmemmap
pagetable and x86_64 architecture direct mapping pagetable removing.
All pages of virtual mapping in removed memory cannot be freed if some
pages used as PGD/PUD include not only removed memory but also other
memory. So this patch uses the following way to check whether a page
can be freed or not.
1) When removing memory, the page structs of the removed memory are
filled with 0FD.
2) All page structs are filled with 0xFD on PT/PMD, PT/PMD can be
cleared. In this case, the page used as PT/PMD can be freed.
For direct mapping pages, update direct_pages_count[level] when we freed
their pagetables. And do not free the pages again because they were
freed when offlining.
For vmemmap pages, free the pages and their pagetables.
For larger pages, do not split them into smaller ones because there is
no way to know if the larger page has been split. As a result, there is
no way to decide when to split. We deal the larger pages in the
following way:
1) For direct mapped pages, all the pages were freed when they were
offlined. And since menmory offline is done section by section, all
the memory ranges being removed are aligned to PAGE_SIZE. So only need
to deal with unaligned pages when freeing vmemmap pages.
2) For vmemmap pages being used to store page_struct, if part of the
larger page is still in use, just fill the unused part with 0xFD. And
when the whole page is fulfilled with 0xFD, then free the larger page.
[akpm@linux-foundation.org: fix typo in comment]
[tangchen@cn.fujitsu.com: do not calculate direct mapping pages when freeing vmemmap pagetables]
[tangchen@cn.fujitsu.com: do not free direct mapping pages twice]
[tangchen@cn.fujitsu.com: do not free page split from hugepage one by one]
[tangchen@cn.fujitsu.com: do not split pages when freeing pagetable pages]
[akpm@linux-foundation.org: use pmd_page_vaddr()]
[akpm@linux-foundation.org: fix used-uninitialised bug]
Signed-off-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Signed-off-by: Jianguo Wu <wujianguo@huawei.com>
Signed-off-by: Wen Congyang <wency@cn.fujitsu.com>
Signed-off-by: Tang Chen <tangchen@cn.fujitsu.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Jiang Liu <jiang.liu@huawei.com>
Cc: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Cc: Wu Jianguo <wujianguo@huawei.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-23 08:33:04 +08:00
|
|
|
static void __meminit
|
|
|
|
remove_pte_table(pte_t *pte_start, unsigned long addr, unsigned long end,
|
|
|
|
bool direct)
|
|
|
|
{
|
|
|
|
unsigned long next, pages = 0;
|
|
|
|
pte_t *pte;
|
|
|
|
void *page_addr;
|
|
|
|
phys_addr_t phys_addr;
|
|
|
|
|
|
|
|
pte = pte_start + pte_index(addr);
|
|
|
|
for (; addr < end; addr = next, pte++) {
|
|
|
|
next = (addr + PAGE_SIZE) & PAGE_MASK;
|
|
|
|
if (next > end)
|
|
|
|
next = end;
|
|
|
|
|
|
|
|
if (!pte_present(*pte))
|
|
|
|
continue;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* We mapped [0,1G) memory as identity mapping when
|
|
|
|
* initializing, in arch/x86/kernel/head_64.S. These
|
|
|
|
* pagetables cannot be removed.
|
|
|
|
*/
|
|
|
|
phys_addr = pte_val(*pte) + (addr & PAGE_MASK);
|
|
|
|
if (phys_addr < (phys_addr_t)0x40000000)
|
|
|
|
return;
|
|
|
|
|
2016-01-12 10:19:30 +08:00
|
|
|
if (PAGE_ALIGNED(addr) && PAGE_ALIGNED(next)) {
|
memory-hotplug: common APIs to support page tables hot-remove
When memory is removed, the corresponding pagetables should alse be
removed. This patch introduces some common APIs to support vmemmap
pagetable and x86_64 architecture direct mapping pagetable removing.
All pages of virtual mapping in removed memory cannot be freed if some
pages used as PGD/PUD include not only removed memory but also other
memory. So this patch uses the following way to check whether a page
can be freed or not.
1) When removing memory, the page structs of the removed memory are
filled with 0FD.
2) All page structs are filled with 0xFD on PT/PMD, PT/PMD can be
cleared. In this case, the page used as PT/PMD can be freed.
For direct mapping pages, update direct_pages_count[level] when we freed
their pagetables. And do not free the pages again because they were
freed when offlining.
For vmemmap pages, free the pages and their pagetables.
For larger pages, do not split them into smaller ones because there is
no way to know if the larger page has been split. As a result, there is
no way to decide when to split. We deal the larger pages in the
following way:
1) For direct mapped pages, all the pages were freed when they were
offlined. And since menmory offline is done section by section, all
the memory ranges being removed are aligned to PAGE_SIZE. So only need
to deal with unaligned pages when freeing vmemmap pages.
2) For vmemmap pages being used to store page_struct, if part of the
larger page is still in use, just fill the unused part with 0xFD. And
when the whole page is fulfilled with 0xFD, then free the larger page.
[akpm@linux-foundation.org: fix typo in comment]
[tangchen@cn.fujitsu.com: do not calculate direct mapping pages when freeing vmemmap pagetables]
[tangchen@cn.fujitsu.com: do not free direct mapping pages twice]
[tangchen@cn.fujitsu.com: do not free page split from hugepage one by one]
[tangchen@cn.fujitsu.com: do not split pages when freeing pagetable pages]
[akpm@linux-foundation.org: use pmd_page_vaddr()]
[akpm@linux-foundation.org: fix used-uninitialised bug]
Signed-off-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Signed-off-by: Jianguo Wu <wujianguo@huawei.com>
Signed-off-by: Wen Congyang <wency@cn.fujitsu.com>
Signed-off-by: Tang Chen <tangchen@cn.fujitsu.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Jiang Liu <jiang.liu@huawei.com>
Cc: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Cc: Wu Jianguo <wujianguo@huawei.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-23 08:33:04 +08:00
|
|
|
/*
|
|
|
|
* Do not free direct mapping pages since they were
|
|
|
|
* freed when offlining, or simplely not in use.
|
|
|
|
*/
|
|
|
|
if (!direct)
|
|
|
|
free_pagetable(pte_page(*pte), 0);
|
|
|
|
|
|
|
|
spin_lock(&init_mm.page_table_lock);
|
|
|
|
pte_clear(&init_mm, addr, pte);
|
|
|
|
spin_unlock(&init_mm.page_table_lock);
|
|
|
|
|
|
|
|
/* For non-direct mapping, pages means nothing. */
|
|
|
|
pages++;
|
|
|
|
} else {
|
|
|
|
/*
|
|
|
|
* If we are here, we are freeing vmemmap pages since
|
|
|
|
* direct mapped memory ranges to be freed are aligned.
|
|
|
|
*
|
|
|
|
* If we are not removing the whole page, it means
|
|
|
|
* other page structs in this page are being used and
|
|
|
|
* we canot remove them. So fill the unused page_structs
|
|
|
|
* with 0xFD, and remove the page when it is wholly
|
|
|
|
* filled with 0xFD.
|
|
|
|
*/
|
|
|
|
memset((void *)addr, PAGE_INUSE, next - addr);
|
|
|
|
|
|
|
|
page_addr = page_address(pte_page(*pte));
|
|
|
|
if (!memchr_inv(page_addr, PAGE_INUSE, PAGE_SIZE)) {
|
|
|
|
free_pagetable(pte_page(*pte), 0);
|
|
|
|
|
|
|
|
spin_lock(&init_mm.page_table_lock);
|
|
|
|
pte_clear(&init_mm, addr, pte);
|
|
|
|
spin_unlock(&init_mm.page_table_lock);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Call free_pte_table() in remove_pmd_table(). */
|
|
|
|
flush_tlb_all();
|
|
|
|
if (direct)
|
|
|
|
update_page_count(PG_LEVEL_4K, -pages);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void __meminit
|
|
|
|
remove_pmd_table(pmd_t *pmd_start, unsigned long addr, unsigned long end,
|
|
|
|
bool direct)
|
|
|
|
{
|
|
|
|
unsigned long next, pages = 0;
|
|
|
|
pte_t *pte_base;
|
|
|
|
pmd_t *pmd;
|
|
|
|
void *page_addr;
|
|
|
|
|
|
|
|
pmd = pmd_start + pmd_index(addr);
|
|
|
|
for (; addr < end; addr = next, pmd++) {
|
|
|
|
next = pmd_addr_end(addr, end);
|
|
|
|
|
|
|
|
if (!pmd_present(*pmd))
|
|
|
|
continue;
|
|
|
|
|
|
|
|
if (pmd_large(*pmd)) {
|
|
|
|
if (IS_ALIGNED(addr, PMD_SIZE) &&
|
|
|
|
IS_ALIGNED(next, PMD_SIZE)) {
|
|
|
|
if (!direct)
|
|
|
|
free_pagetable(pmd_page(*pmd),
|
|
|
|
get_order(PMD_SIZE));
|
|
|
|
|
|
|
|
spin_lock(&init_mm.page_table_lock);
|
|
|
|
pmd_clear(pmd);
|
|
|
|
spin_unlock(&init_mm.page_table_lock);
|
|
|
|
pages++;
|
|
|
|
} else {
|
|
|
|
/* If here, we are freeing vmemmap pages. */
|
|
|
|
memset((void *)addr, PAGE_INUSE, next - addr);
|
|
|
|
|
|
|
|
page_addr = page_address(pmd_page(*pmd));
|
|
|
|
if (!memchr_inv(page_addr, PAGE_INUSE,
|
|
|
|
PMD_SIZE)) {
|
|
|
|
free_pagetable(pmd_page(*pmd),
|
|
|
|
get_order(PMD_SIZE));
|
|
|
|
|
|
|
|
spin_lock(&init_mm.page_table_lock);
|
|
|
|
pmd_clear(pmd);
|
|
|
|
spin_unlock(&init_mm.page_table_lock);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
|
|
|
pte_base = (pte_t *)pmd_page_vaddr(*pmd);
|
|
|
|
remove_pte_table(pte_base, addr, next, direct);
|
|
|
|
free_pte_table(pte_base, pmd);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Call free_pmd_table() in remove_pud_table(). */
|
|
|
|
if (direct)
|
|
|
|
update_page_count(PG_LEVEL_2M, -pages);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void __meminit
|
|
|
|
remove_pud_table(pud_t *pud_start, unsigned long addr, unsigned long end,
|
|
|
|
bool direct)
|
|
|
|
{
|
|
|
|
unsigned long next, pages = 0;
|
|
|
|
pmd_t *pmd_base;
|
|
|
|
pud_t *pud;
|
|
|
|
void *page_addr;
|
|
|
|
|
|
|
|
pud = pud_start + pud_index(addr);
|
|
|
|
for (; addr < end; addr = next, pud++) {
|
|
|
|
next = pud_addr_end(addr, end);
|
|
|
|
|
|
|
|
if (!pud_present(*pud))
|
|
|
|
continue;
|
|
|
|
|
|
|
|
if (pud_large(*pud)) {
|
|
|
|
if (IS_ALIGNED(addr, PUD_SIZE) &&
|
|
|
|
IS_ALIGNED(next, PUD_SIZE)) {
|
|
|
|
if (!direct)
|
|
|
|
free_pagetable(pud_page(*pud),
|
|
|
|
get_order(PUD_SIZE));
|
|
|
|
|
|
|
|
spin_lock(&init_mm.page_table_lock);
|
|
|
|
pud_clear(pud);
|
|
|
|
spin_unlock(&init_mm.page_table_lock);
|
|
|
|
pages++;
|
|
|
|
} else {
|
|
|
|
/* If here, we are freeing vmemmap pages. */
|
|
|
|
memset((void *)addr, PAGE_INUSE, next - addr);
|
|
|
|
|
|
|
|
page_addr = page_address(pud_page(*pud));
|
|
|
|
if (!memchr_inv(page_addr, PAGE_INUSE,
|
|
|
|
PUD_SIZE)) {
|
|
|
|
free_pagetable(pud_page(*pud),
|
|
|
|
get_order(PUD_SIZE));
|
|
|
|
|
|
|
|
spin_lock(&init_mm.page_table_lock);
|
|
|
|
pud_clear(pud);
|
|
|
|
spin_unlock(&init_mm.page_table_lock);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
2017-04-25 17:25:57 +08:00
|
|
|
pmd_base = pmd_offset(pud, 0);
|
memory-hotplug: common APIs to support page tables hot-remove
When memory is removed, the corresponding pagetables should alse be
removed. This patch introduces some common APIs to support vmemmap
pagetable and x86_64 architecture direct mapping pagetable removing.
All pages of virtual mapping in removed memory cannot be freed if some
pages used as PGD/PUD include not only removed memory but also other
memory. So this patch uses the following way to check whether a page
can be freed or not.
1) When removing memory, the page structs of the removed memory are
filled with 0FD.
2) All page structs are filled with 0xFD on PT/PMD, PT/PMD can be
cleared. In this case, the page used as PT/PMD can be freed.
For direct mapping pages, update direct_pages_count[level] when we freed
their pagetables. And do not free the pages again because they were
freed when offlining.
For vmemmap pages, free the pages and their pagetables.
For larger pages, do not split them into smaller ones because there is
no way to know if the larger page has been split. As a result, there is
no way to decide when to split. We deal the larger pages in the
following way:
1) For direct mapped pages, all the pages were freed when they were
offlined. And since menmory offline is done section by section, all
the memory ranges being removed are aligned to PAGE_SIZE. So only need
to deal with unaligned pages when freeing vmemmap pages.
2) For vmemmap pages being used to store page_struct, if part of the
larger page is still in use, just fill the unused part with 0xFD. And
when the whole page is fulfilled with 0xFD, then free the larger page.
[akpm@linux-foundation.org: fix typo in comment]
[tangchen@cn.fujitsu.com: do not calculate direct mapping pages when freeing vmemmap pagetables]
[tangchen@cn.fujitsu.com: do not free direct mapping pages twice]
[tangchen@cn.fujitsu.com: do not free page split from hugepage one by one]
[tangchen@cn.fujitsu.com: do not split pages when freeing pagetable pages]
[akpm@linux-foundation.org: use pmd_page_vaddr()]
[akpm@linux-foundation.org: fix used-uninitialised bug]
Signed-off-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Signed-off-by: Jianguo Wu <wujianguo@huawei.com>
Signed-off-by: Wen Congyang <wency@cn.fujitsu.com>
Signed-off-by: Tang Chen <tangchen@cn.fujitsu.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Jiang Liu <jiang.liu@huawei.com>
Cc: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Cc: Wu Jianguo <wujianguo@huawei.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-23 08:33:04 +08:00
|
|
|
remove_pmd_table(pmd_base, addr, next, direct);
|
|
|
|
free_pmd_table(pmd_base, pud);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (direct)
|
|
|
|
update_page_count(PG_LEVEL_1G, -pages);
|
|
|
|
}
|
|
|
|
|
2017-03-18 02:55:15 +08:00
|
|
|
static void __meminit
|
|
|
|
remove_p4d_table(p4d_t *p4d_start, unsigned long addr, unsigned long end,
|
|
|
|
bool direct)
|
|
|
|
{
|
|
|
|
unsigned long next, pages = 0;
|
|
|
|
pud_t *pud_base;
|
|
|
|
p4d_t *p4d;
|
|
|
|
|
|
|
|
p4d = p4d_start + p4d_index(addr);
|
|
|
|
for (; addr < end; addr = next, p4d++) {
|
|
|
|
next = p4d_addr_end(addr, end);
|
|
|
|
|
|
|
|
if (!p4d_present(*p4d))
|
|
|
|
continue;
|
|
|
|
|
|
|
|
BUILD_BUG_ON(p4d_large(*p4d));
|
|
|
|
|
2017-04-25 17:25:57 +08:00
|
|
|
pud_base = pud_offset(p4d, 0);
|
2017-03-18 02:55:15 +08:00
|
|
|
remove_pud_table(pud_base, addr, next, direct);
|
2017-06-25 02:05:14 +08:00
|
|
|
/*
|
|
|
|
* For 4-level page tables we do not want to free PUDs, but in the
|
|
|
|
* 5-level case we should free them. This code will have to change
|
|
|
|
* to adapt for boot-time switching between 4 and 5 level page tables.
|
|
|
|
*/
|
|
|
|
if (CONFIG_PGTABLE_LEVELS == 5)
|
|
|
|
free_pud_table(pud_base, p4d);
|
2017-03-18 02:55:15 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
if (direct)
|
|
|
|
update_page_count(PG_LEVEL_512G, -pages);
|
|
|
|
}
|
|
|
|
|
memory-hotplug: common APIs to support page tables hot-remove
When memory is removed, the corresponding pagetables should alse be
removed. This patch introduces some common APIs to support vmemmap
pagetable and x86_64 architecture direct mapping pagetable removing.
All pages of virtual mapping in removed memory cannot be freed if some
pages used as PGD/PUD include not only removed memory but also other
memory. So this patch uses the following way to check whether a page
can be freed or not.
1) When removing memory, the page structs of the removed memory are
filled with 0FD.
2) All page structs are filled with 0xFD on PT/PMD, PT/PMD can be
cleared. In this case, the page used as PT/PMD can be freed.
For direct mapping pages, update direct_pages_count[level] when we freed
their pagetables. And do not free the pages again because they were
freed when offlining.
For vmemmap pages, free the pages and their pagetables.
For larger pages, do not split them into smaller ones because there is
no way to know if the larger page has been split. As a result, there is
no way to decide when to split. We deal the larger pages in the
following way:
1) For direct mapped pages, all the pages were freed when they were
offlined. And since menmory offline is done section by section, all
the memory ranges being removed are aligned to PAGE_SIZE. So only need
to deal with unaligned pages when freeing vmemmap pages.
2) For vmemmap pages being used to store page_struct, if part of the
larger page is still in use, just fill the unused part with 0xFD. And
when the whole page is fulfilled with 0xFD, then free the larger page.
[akpm@linux-foundation.org: fix typo in comment]
[tangchen@cn.fujitsu.com: do not calculate direct mapping pages when freeing vmemmap pagetables]
[tangchen@cn.fujitsu.com: do not free direct mapping pages twice]
[tangchen@cn.fujitsu.com: do not free page split from hugepage one by one]
[tangchen@cn.fujitsu.com: do not split pages when freeing pagetable pages]
[akpm@linux-foundation.org: use pmd_page_vaddr()]
[akpm@linux-foundation.org: fix used-uninitialised bug]
Signed-off-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Signed-off-by: Jianguo Wu <wujianguo@huawei.com>
Signed-off-by: Wen Congyang <wency@cn.fujitsu.com>
Signed-off-by: Tang Chen <tangchen@cn.fujitsu.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Jiang Liu <jiang.liu@huawei.com>
Cc: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Cc: Wu Jianguo <wujianguo@huawei.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-23 08:33:04 +08:00
|
|
|
/* start and end are both virtual address. */
|
|
|
|
static void __meminit
|
|
|
|
remove_pagetable(unsigned long start, unsigned long end, bool direct)
|
|
|
|
{
|
|
|
|
unsigned long next;
|
2014-08-23 04:27:31 +08:00
|
|
|
unsigned long addr;
|
memory-hotplug: common APIs to support page tables hot-remove
When memory is removed, the corresponding pagetables should alse be
removed. This patch introduces some common APIs to support vmemmap
pagetable and x86_64 architecture direct mapping pagetable removing.
All pages of virtual mapping in removed memory cannot be freed if some
pages used as PGD/PUD include not only removed memory but also other
memory. So this patch uses the following way to check whether a page
can be freed or not.
1) When removing memory, the page structs of the removed memory are
filled with 0FD.
2) All page structs are filled with 0xFD on PT/PMD, PT/PMD can be
cleared. In this case, the page used as PT/PMD can be freed.
For direct mapping pages, update direct_pages_count[level] when we freed
their pagetables. And do not free the pages again because they were
freed when offlining.
For vmemmap pages, free the pages and their pagetables.
For larger pages, do not split them into smaller ones because there is
no way to know if the larger page has been split. As a result, there is
no way to decide when to split. We deal the larger pages in the
following way:
1) For direct mapped pages, all the pages were freed when they were
offlined. And since menmory offline is done section by section, all
the memory ranges being removed are aligned to PAGE_SIZE. So only need
to deal with unaligned pages when freeing vmemmap pages.
2) For vmemmap pages being used to store page_struct, if part of the
larger page is still in use, just fill the unused part with 0xFD. And
when the whole page is fulfilled with 0xFD, then free the larger page.
[akpm@linux-foundation.org: fix typo in comment]
[tangchen@cn.fujitsu.com: do not calculate direct mapping pages when freeing vmemmap pagetables]
[tangchen@cn.fujitsu.com: do not free direct mapping pages twice]
[tangchen@cn.fujitsu.com: do not free page split from hugepage one by one]
[tangchen@cn.fujitsu.com: do not split pages when freeing pagetable pages]
[akpm@linux-foundation.org: use pmd_page_vaddr()]
[akpm@linux-foundation.org: fix used-uninitialised bug]
Signed-off-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Signed-off-by: Jianguo Wu <wujianguo@huawei.com>
Signed-off-by: Wen Congyang <wency@cn.fujitsu.com>
Signed-off-by: Tang Chen <tangchen@cn.fujitsu.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Jiang Liu <jiang.liu@huawei.com>
Cc: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Cc: Wu Jianguo <wujianguo@huawei.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-23 08:33:04 +08:00
|
|
|
pgd_t *pgd;
|
2017-03-18 02:55:15 +08:00
|
|
|
p4d_t *p4d;
|
memory-hotplug: common APIs to support page tables hot-remove
When memory is removed, the corresponding pagetables should alse be
removed. This patch introduces some common APIs to support vmemmap
pagetable and x86_64 architecture direct mapping pagetable removing.
All pages of virtual mapping in removed memory cannot be freed if some
pages used as PGD/PUD include not only removed memory but also other
memory. So this patch uses the following way to check whether a page
can be freed or not.
1) When removing memory, the page structs of the removed memory are
filled with 0FD.
2) All page structs are filled with 0xFD on PT/PMD, PT/PMD can be
cleared. In this case, the page used as PT/PMD can be freed.
For direct mapping pages, update direct_pages_count[level] when we freed
their pagetables. And do not free the pages again because they were
freed when offlining.
For vmemmap pages, free the pages and their pagetables.
For larger pages, do not split them into smaller ones because there is
no way to know if the larger page has been split. As a result, there is
no way to decide when to split. We deal the larger pages in the
following way:
1) For direct mapped pages, all the pages were freed when they were
offlined. And since menmory offline is done section by section, all
the memory ranges being removed are aligned to PAGE_SIZE. So only need
to deal with unaligned pages when freeing vmemmap pages.
2) For vmemmap pages being used to store page_struct, if part of the
larger page is still in use, just fill the unused part with 0xFD. And
when the whole page is fulfilled with 0xFD, then free the larger page.
[akpm@linux-foundation.org: fix typo in comment]
[tangchen@cn.fujitsu.com: do not calculate direct mapping pages when freeing vmemmap pagetables]
[tangchen@cn.fujitsu.com: do not free direct mapping pages twice]
[tangchen@cn.fujitsu.com: do not free page split from hugepage one by one]
[tangchen@cn.fujitsu.com: do not split pages when freeing pagetable pages]
[akpm@linux-foundation.org: use pmd_page_vaddr()]
[akpm@linux-foundation.org: fix used-uninitialised bug]
Signed-off-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Signed-off-by: Jianguo Wu <wujianguo@huawei.com>
Signed-off-by: Wen Congyang <wency@cn.fujitsu.com>
Signed-off-by: Tang Chen <tangchen@cn.fujitsu.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Jiang Liu <jiang.liu@huawei.com>
Cc: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Cc: Wu Jianguo <wujianguo@huawei.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-23 08:33:04 +08:00
|
|
|
|
2014-08-23 04:27:31 +08:00
|
|
|
for (addr = start; addr < end; addr = next) {
|
|
|
|
next = pgd_addr_end(addr, end);
|
memory-hotplug: common APIs to support page tables hot-remove
When memory is removed, the corresponding pagetables should alse be
removed. This patch introduces some common APIs to support vmemmap
pagetable and x86_64 architecture direct mapping pagetable removing.
All pages of virtual mapping in removed memory cannot be freed if some
pages used as PGD/PUD include not only removed memory but also other
memory. So this patch uses the following way to check whether a page
can be freed or not.
1) When removing memory, the page structs of the removed memory are
filled with 0FD.
2) All page structs are filled with 0xFD on PT/PMD, PT/PMD can be
cleared. In this case, the page used as PT/PMD can be freed.
For direct mapping pages, update direct_pages_count[level] when we freed
their pagetables. And do not free the pages again because they were
freed when offlining.
For vmemmap pages, free the pages and their pagetables.
For larger pages, do not split them into smaller ones because there is
no way to know if the larger page has been split. As a result, there is
no way to decide when to split. We deal the larger pages in the
following way:
1) For direct mapped pages, all the pages were freed when they were
offlined. And since menmory offline is done section by section, all
the memory ranges being removed are aligned to PAGE_SIZE. So only need
to deal with unaligned pages when freeing vmemmap pages.
2) For vmemmap pages being used to store page_struct, if part of the
larger page is still in use, just fill the unused part with 0xFD. And
when the whole page is fulfilled with 0xFD, then free the larger page.
[akpm@linux-foundation.org: fix typo in comment]
[tangchen@cn.fujitsu.com: do not calculate direct mapping pages when freeing vmemmap pagetables]
[tangchen@cn.fujitsu.com: do not free direct mapping pages twice]
[tangchen@cn.fujitsu.com: do not free page split from hugepage one by one]
[tangchen@cn.fujitsu.com: do not split pages when freeing pagetable pages]
[akpm@linux-foundation.org: use pmd_page_vaddr()]
[akpm@linux-foundation.org: fix used-uninitialised bug]
Signed-off-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Signed-off-by: Jianguo Wu <wujianguo@huawei.com>
Signed-off-by: Wen Congyang <wency@cn.fujitsu.com>
Signed-off-by: Tang Chen <tangchen@cn.fujitsu.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Jiang Liu <jiang.liu@huawei.com>
Cc: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Cc: Wu Jianguo <wujianguo@huawei.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-23 08:33:04 +08:00
|
|
|
|
2014-08-23 04:27:31 +08:00
|
|
|
pgd = pgd_offset_k(addr);
|
memory-hotplug: common APIs to support page tables hot-remove
When memory is removed, the corresponding pagetables should alse be
removed. This patch introduces some common APIs to support vmemmap
pagetable and x86_64 architecture direct mapping pagetable removing.
All pages of virtual mapping in removed memory cannot be freed if some
pages used as PGD/PUD include not only removed memory but also other
memory. So this patch uses the following way to check whether a page
can be freed or not.
1) When removing memory, the page structs of the removed memory are
filled with 0FD.
2) All page structs are filled with 0xFD on PT/PMD, PT/PMD can be
cleared. In this case, the page used as PT/PMD can be freed.
For direct mapping pages, update direct_pages_count[level] when we freed
their pagetables. And do not free the pages again because they were
freed when offlining.
For vmemmap pages, free the pages and their pagetables.
For larger pages, do not split them into smaller ones because there is
no way to know if the larger page has been split. As a result, there is
no way to decide when to split. We deal the larger pages in the
following way:
1) For direct mapped pages, all the pages were freed when they were
offlined. And since menmory offline is done section by section, all
the memory ranges being removed are aligned to PAGE_SIZE. So only need
to deal with unaligned pages when freeing vmemmap pages.
2) For vmemmap pages being used to store page_struct, if part of the
larger page is still in use, just fill the unused part with 0xFD. And
when the whole page is fulfilled with 0xFD, then free the larger page.
[akpm@linux-foundation.org: fix typo in comment]
[tangchen@cn.fujitsu.com: do not calculate direct mapping pages when freeing vmemmap pagetables]
[tangchen@cn.fujitsu.com: do not free direct mapping pages twice]
[tangchen@cn.fujitsu.com: do not free page split from hugepage one by one]
[tangchen@cn.fujitsu.com: do not split pages when freeing pagetable pages]
[akpm@linux-foundation.org: use pmd_page_vaddr()]
[akpm@linux-foundation.org: fix used-uninitialised bug]
Signed-off-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Signed-off-by: Jianguo Wu <wujianguo@huawei.com>
Signed-off-by: Wen Congyang <wency@cn.fujitsu.com>
Signed-off-by: Tang Chen <tangchen@cn.fujitsu.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Jiang Liu <jiang.liu@huawei.com>
Cc: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Cc: Wu Jianguo <wujianguo@huawei.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-23 08:33:04 +08:00
|
|
|
if (!pgd_present(*pgd))
|
|
|
|
continue;
|
|
|
|
|
2017-04-25 17:25:57 +08:00
|
|
|
p4d = p4d_offset(pgd, 0);
|
2017-03-18 02:55:15 +08:00
|
|
|
remove_p4d_table(p4d, addr, next, direct);
|
memory-hotplug: common APIs to support page tables hot-remove
When memory is removed, the corresponding pagetables should alse be
removed. This patch introduces some common APIs to support vmemmap
pagetable and x86_64 architecture direct mapping pagetable removing.
All pages of virtual mapping in removed memory cannot be freed if some
pages used as PGD/PUD include not only removed memory but also other
memory. So this patch uses the following way to check whether a page
can be freed or not.
1) When removing memory, the page structs of the removed memory are
filled with 0FD.
2) All page structs are filled with 0xFD on PT/PMD, PT/PMD can be
cleared. In this case, the page used as PT/PMD can be freed.
For direct mapping pages, update direct_pages_count[level] when we freed
their pagetables. And do not free the pages again because they were
freed when offlining.
For vmemmap pages, free the pages and their pagetables.
For larger pages, do not split them into smaller ones because there is
no way to know if the larger page has been split. As a result, there is
no way to decide when to split. We deal the larger pages in the
following way:
1) For direct mapped pages, all the pages were freed when they were
offlined. And since menmory offline is done section by section, all
the memory ranges being removed are aligned to PAGE_SIZE. So only need
to deal with unaligned pages when freeing vmemmap pages.
2) For vmemmap pages being used to store page_struct, if part of the
larger page is still in use, just fill the unused part with 0xFD. And
when the whole page is fulfilled with 0xFD, then free the larger page.
[akpm@linux-foundation.org: fix typo in comment]
[tangchen@cn.fujitsu.com: do not calculate direct mapping pages when freeing vmemmap pagetables]
[tangchen@cn.fujitsu.com: do not free direct mapping pages twice]
[tangchen@cn.fujitsu.com: do not free page split from hugepage one by one]
[tangchen@cn.fujitsu.com: do not split pages when freeing pagetable pages]
[akpm@linux-foundation.org: use pmd_page_vaddr()]
[akpm@linux-foundation.org: fix used-uninitialised bug]
Signed-off-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Signed-off-by: Jianguo Wu <wujianguo@huawei.com>
Signed-off-by: Wen Congyang <wency@cn.fujitsu.com>
Signed-off-by: Tang Chen <tangchen@cn.fujitsu.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Jiang Liu <jiang.liu@huawei.com>
Cc: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Cc: Wu Jianguo <wujianguo@huawei.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-23 08:33:04 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
flush_tlb_all();
|
|
|
|
}
|
|
|
|
|
2013-04-30 06:07:50 +08:00
|
|
|
void __ref vmemmap_free(unsigned long start, unsigned long end)
|
2013-02-23 08:33:08 +08:00
|
|
|
{
|
|
|
|
remove_pagetable(start, end, false);
|
|
|
|
}
|
|
|
|
|
2013-04-15 17:46:46 +08:00
|
|
|
#ifdef CONFIG_MEMORY_HOTREMOVE
|
2013-02-23 08:33:06 +08:00
|
|
|
static void __meminit
|
|
|
|
kernel_physical_mapping_remove(unsigned long start, unsigned long end)
|
|
|
|
{
|
|
|
|
start = (unsigned long)__va(start);
|
|
|
|
end = (unsigned long)__va(end);
|
|
|
|
|
|
|
|
remove_pagetable(start, end, true);
|
|
|
|
}
|
|
|
|
|
2013-02-23 08:32:58 +08:00
|
|
|
int __ref arch_remove_memory(u64 start, u64 size)
|
|
|
|
{
|
|
|
|
unsigned long start_pfn = start >> PAGE_SHIFT;
|
|
|
|
unsigned long nr_pages = size >> PAGE_SHIFT;
|
2016-01-16 08:56:22 +08:00
|
|
|
struct page *page = pfn_to_page(start_pfn);
|
|
|
|
struct vmem_altmap *altmap;
|
2013-02-23 08:32:58 +08:00
|
|
|
struct zone *zone;
|
|
|
|
int ret;
|
|
|
|
|
2016-01-16 08:56:22 +08:00
|
|
|
/* With altmap the first mapped page is offset from @start */
|
|
|
|
altmap = to_vmem_altmap((unsigned long) page);
|
|
|
|
if (altmap)
|
|
|
|
page += vmem_altmap_offset(altmap);
|
|
|
|
zone = page_zone(page);
|
2013-02-23 08:32:58 +08:00
|
|
|
ret = __remove_pages(zone, start_pfn, nr_pages);
|
|
|
|
WARN_ON_ONCE(ret);
|
2016-01-16 08:56:22 +08:00
|
|
|
kernel_physical_mapping_remove(start, start + size);
|
2013-02-23 08:32:58 +08:00
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
#endif
|
2006-10-01 14:27:09 +08:00
|
|
|
#endif /* CONFIG_MEMORY_HOTPLUG */
|
|
|
|
|
2009-09-23 07:45:49 +08:00
|
|
|
static struct kcore_list kcore_vsyscall;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2012-11-17 11:39:19 +08:00
|
|
|
static void __init register_page_bootmem_info(void)
|
|
|
|
{
|
|
|
|
#ifdef CONFIG_NUMA
|
|
|
|
int i;
|
|
|
|
|
|
|
|
for_each_online_node(i)
|
|
|
|
register_page_bootmem_info_node(NODE_DATA(i));
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
void __init mem_init(void)
|
|
|
|
{
|
2006-06-26 19:58:11 +08:00
|
|
|
pci_iommu_alloc();
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2008-01-30 20:32:36 +08:00
|
|
|
/* clear_bss() already clear the empty_zero_page */
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2012-11-17 11:39:19 +08:00
|
|
|
register_page_bootmem_info();
|
2013-04-30 06:06:53 +08:00
|
|
|
|
|
|
|
/* this will put all memory onto the freelists */
|
2013-07-04 06:03:24 +08:00
|
|
|
free_all_bootmem();
|
2005-04-17 06:20:36 +08:00
|
|
|
after_bootmem = 1;
|
|
|
|
|
|
|
|
/* Register memory areas for /proc/kcore */
|
2014-05-06 03:19:36 +08:00
|
|
|
kclist_add(&kcore_vsyscall, (void *)VSYSCALL_ADDR,
|
|
|
|
PAGE_SIZE, KCORE_OTHER);
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2013-07-04 06:04:19 +08:00
|
|
|
mem_init_print_info(NULL);
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
|
|
|
|
2009-10-29 10:46:56 +08:00
|
|
|
int kernel_set_to_readonly;
|
2009-02-18 06:57:30 +08:00
|
|
|
|
|
|
|
void set_kernel_text_rw(void)
|
|
|
|
{
|
2009-10-15 05:46:55 +08:00
|
|
|
unsigned long start = PFN_ALIGN(_text);
|
2009-10-29 10:46:58 +08:00
|
|
|
unsigned long end = PFN_ALIGN(__stop___ex_table);
|
2009-02-18 06:57:30 +08:00
|
|
|
|
|
|
|
if (!kernel_set_to_readonly)
|
|
|
|
return;
|
|
|
|
|
|
|
|
pr_debug("Set kernel text: %lx - %lx for read write\n",
|
|
|
|
start, end);
|
|
|
|
|
2009-10-29 10:46:58 +08:00
|
|
|
/*
|
|
|
|
* Make the kernel identity mapping for text RW. Kernel text
|
|
|
|
* mapping will always be RO. Refer to the comment in
|
|
|
|
* static_protections() in pageattr.c
|
|
|
|
*/
|
2009-02-18 06:57:30 +08:00
|
|
|
set_memory_rw(start, (end - start) >> PAGE_SHIFT);
|
|
|
|
}
|
|
|
|
|
|
|
|
void set_kernel_text_ro(void)
|
|
|
|
{
|
2009-10-15 05:46:55 +08:00
|
|
|
unsigned long start = PFN_ALIGN(_text);
|
2009-10-29 10:46:58 +08:00
|
|
|
unsigned long end = PFN_ALIGN(__stop___ex_table);
|
2009-02-18 06:57:30 +08:00
|
|
|
|
|
|
|
if (!kernel_set_to_readonly)
|
|
|
|
return;
|
|
|
|
|
|
|
|
pr_debug("Set kernel text: %lx - %lx for read only\n",
|
|
|
|
start, end);
|
|
|
|
|
2009-10-29 10:46:58 +08:00
|
|
|
/*
|
|
|
|
* Set the kernel identity mapping for text RO.
|
|
|
|
*/
|
2009-02-18 06:57:30 +08:00
|
|
|
set_memory_ro(start, (end - start) >> PAGE_SHIFT);
|
|
|
|
}
|
|
|
|
|
2006-01-06 16:12:04 +08:00
|
|
|
void mark_rodata_ro(void)
|
|
|
|
{
|
2009-10-15 05:46:56 +08:00
|
|
|
unsigned long start = PFN_ALIGN(_text);
|
2012-11-17 05:57:13 +08:00
|
|
|
unsigned long rodata_start = PFN_ALIGN(__start_rodata);
|
2009-10-15 05:46:56 +08:00
|
|
|
unsigned long end = (unsigned long) &__end_rodata_hpage_align;
|
2012-11-17 05:57:13 +08:00
|
|
|
unsigned long text_end = PFN_ALIGN(&__stop___ex_table);
|
|
|
|
unsigned long rodata_end = PFN_ALIGN(&__end_rodata);
|
2014-11-15 03:47:37 +08:00
|
|
|
unsigned long all_end;
|
2008-05-13 03:20:56 +08:00
|
|
|
|
[PATCH] x86: tighten kernel image page access rights
On x86-64, kernel memory freed after init can be entirely unmapped instead
of just getting 'poisoned' by overwriting with a debug pattern.
On i386 and x86-64 (under CONFIG_DEBUG_RODATA), kernel text and bug table
can also be write-protected.
Compared to the first version, this one prevents re-creating deleted
mappings in the kernel image range on x86-64, if those got removed
previously. This, together with the original changes, prevents temporarily
having inconsistent mappings when cacheability attributes are being
changed on such pages (e.g. from AGP code). While on i386 such duplicate
mappings don't exist, the same change is done there, too, both for
consistency and because checking pte_present() before using various other
pte_XXX functions is a requirement anyway. At once, i386 code gets
adjusted to use pte_huge() instead of open coding this.
AK: split out cpa() changes
Signed-off-by: Jan Beulich <jbeulich@novell.com>
Signed-off-by: Andi Kleen <ak@suse.de>
2007-05-03 01:27:10 +08:00
|
|
|
printk(KERN_INFO "Write protecting the kernel read-only data: %luk\n",
|
Revert "[PATCH] x86: __pa and __pa_symbol address space separation"
This was broken. It adds complexity, for no good reason. Rather than
separate __pa() and __pa_symbol(), we should deprecate __pa_symbol(),
and preferably __pa() too - and just use "virt_to_phys()" instead, which
is more readable and has nicer semantics.
However, right now, just undo the separation, and make __pa_symbol() be
the exact same as __pa(). That fixes the bugs this patch introduced,
and we can do the fairly obvious cleanups later.
Do the new __phys_addr() function (which is now the actual workhorse for
the unified __pa()/__pa_symbol()) as a real external function, that way
all the potential issues with compile/link-time optimizations of
constant symbol addresses go away, and we can also, if we choose to, add
more sanity-checking of the argument.
Cc: Eric W. Biederman <ebiederm@xmission.com>
Cc: Vivek Goyal <vgoyal@in.ibm.com>
Cc: Andi Kleen <ak@suse.de>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-05-07 23:44:24 +08:00
|
|
|
(end - start) >> 10);
|
2008-02-07 05:39:45 +08:00
|
|
|
set_memory_ro(start, (end - start) >> PAGE_SHIFT);
|
|
|
|
|
2009-02-18 06:57:30 +08:00
|
|
|
kernel_set_to_readonly = 1;
|
|
|
|
|
2008-02-07 05:39:45 +08:00
|
|
|
/*
|
2013-01-25 04:20:13 +08:00
|
|
|
* The rodata/data/bss/brk section (but not the kernel text!)
|
|
|
|
* should also be not-executable.
|
2014-11-15 03:47:37 +08:00
|
|
|
*
|
|
|
|
* We align all_end to PMD_SIZE because the existing mapping
|
|
|
|
* is a full PMD. If we would align _brk_end to PAGE_SIZE we
|
|
|
|
* split the PMD and the reminder between _brk_end and the end
|
|
|
|
* of the PMD will remain mapped executable.
|
|
|
|
*
|
|
|
|
* Any PMD which was setup after the one which covers _brk_end
|
|
|
|
* has been zapped already via cleanup_highmem().
|
2008-02-07 05:39:45 +08:00
|
|
|
*/
|
2014-11-15 03:47:37 +08:00
|
|
|
all_end = roundup((unsigned long)_brk_end, PMD_SIZE);
|
2015-10-01 21:04:22 +08:00
|
|
|
set_memory_nx(text_end, (all_end - text_end) >> PAGE_SHIFT);
|
2006-01-06 16:12:04 +08:00
|
|
|
|
2008-01-30 20:33:42 +08:00
|
|
|
#ifdef CONFIG_CPA_DEBUG
|
2008-01-30 20:34:10 +08:00
|
|
|
printk(KERN_INFO "Testing CPA: undo %lx-%lx\n", start, end);
|
2008-01-30 20:34:06 +08:00
|
|
|
set_memory_rw(start, (end-start) >> PAGE_SHIFT);
|
2008-01-30 20:33:42 +08:00
|
|
|
|
2008-01-30 20:34:10 +08:00
|
|
|
printk(KERN_INFO "Testing CPA: again\n");
|
2008-01-30 20:34:06 +08:00
|
|
|
set_memory_ro(start, (end-start) >> PAGE_SHIFT);
|
2008-01-30 20:33:42 +08:00
|
|
|
#endif
|
2009-10-15 05:46:56 +08:00
|
|
|
|
2013-07-04 06:02:58 +08:00
|
|
|
free_init_pages("unused kernel",
|
2012-11-17 05:57:13 +08:00
|
|
|
(unsigned long) __va(__pa_symbol(text_end)),
|
|
|
|
(unsigned long) __va(__pa_symbol(rodata_start)));
|
2013-07-04 06:02:58 +08:00
|
|
|
free_init_pages("unused kernel",
|
2012-11-17 05:57:13 +08:00
|
|
|
(unsigned long) __va(__pa_symbol(rodata_end)),
|
|
|
|
(unsigned long) __va(__pa_symbol(_sdata)));
|
2015-10-06 00:55:20 +08:00
|
|
|
|
|
|
|
debug_checkwx();
|
2006-01-06 16:12:04 +08:00
|
|
|
}
|
2008-02-03 04:42:20 +08:00
|
|
|
|
2008-01-30 20:34:10 +08:00
|
|
|
int kern_addr_valid(unsigned long addr)
|
|
|
|
{
|
2005-04-17 06:20:36 +08:00
|
|
|
unsigned long above = ((long)addr) >> __VIRTUAL_MASK_SHIFT;
|
2008-01-30 20:34:10 +08:00
|
|
|
pgd_t *pgd;
|
2017-03-18 02:55:15 +08:00
|
|
|
p4d_t *p4d;
|
2008-01-30 20:34:10 +08:00
|
|
|
pud_t *pud;
|
|
|
|
pmd_t *pmd;
|
|
|
|
pte_t *pte;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
if (above != 0 && above != -1UL)
|
2008-01-30 20:34:10 +08:00
|
|
|
return 0;
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
pgd = pgd_offset_k(addr);
|
|
|
|
if (pgd_none(*pgd))
|
|
|
|
return 0;
|
|
|
|
|
2017-03-18 02:55:15 +08:00
|
|
|
p4d = p4d_offset(pgd, addr);
|
|
|
|
if (p4d_none(*p4d))
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
pud = pud_offset(p4d, addr);
|
2005-04-17 06:20:36 +08:00
|
|
|
if (pud_none(*pud))
|
2008-01-30 20:34:10 +08:00
|
|
|
return 0;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2013-02-11 22:52:36 +08:00
|
|
|
if (pud_large(*pud))
|
|
|
|
return pfn_valid(pud_pfn(*pud));
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
pmd = pmd_offset(pud, addr);
|
|
|
|
if (pmd_none(*pmd))
|
|
|
|
return 0;
|
2008-01-30 20:34:10 +08:00
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
if (pmd_large(*pmd))
|
|
|
|
return pfn_valid(pmd_pfn(*pmd));
|
|
|
|
|
|
|
|
pte = pte_offset_kernel(pmd, addr);
|
|
|
|
if (pte_none(*pte))
|
|
|
|
return 0;
|
2008-01-30 20:34:10 +08:00
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
return pfn_valid(pte_pfn(*pte));
|
|
|
|
}
|
|
|
|
|
x86, mm: probe memory block size for generic x86 64bit
On system with 2TiB ram, current x86_64 have 128M as section size, and
one memory_block only include one section. So will have 16400 entries
under /sys/devices/system/memory/.
Current code try to use block id to find block pointer in /sys for any
section, and reuse that block pointer. that finding will take some time
even after commit 7c243c7168dc ("mm: speedup in __early_pfn_to_nid")
that will skip the search in that case during booting up.
So solution could be increase block size just like SGI UV system did.
(harded code to 2g).
This patch is trying to probe the block size to make it match mmio remap
size. for example, Intel Nehalem later system will have memory range [0,
TOML), [4g, TOMH]. If the memory hole is 2g and total is 128g, TOM will
be 2g, and TOM2 will be 130g.
We could use 2g as block size instead of default 128M. That will reduce
number of entries in /sys/devices/system/memory/
On system 6TiB system will reduce boot time by 35 seconds.
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-06-05 07:06:32 +08:00
|
|
|
static unsigned long probe_memory_block_size(void)
|
2011-01-21 00:46:15 +08:00
|
|
|
{
|
2015-12-01 00:47:43 +08:00
|
|
|
unsigned long bz = MIN_MEMORY_BLOCK_SIZE;
|
x86, mm: probe memory block size for generic x86 64bit
On system with 2TiB ram, current x86_64 have 128M as section size, and
one memory_block only include one section. So will have 16400 entries
under /sys/devices/system/memory/.
Current code try to use block id to find block pointer in /sys for any
section, and reuse that block pointer. that finding will take some time
even after commit 7c243c7168dc ("mm: speedup in __early_pfn_to_nid")
that will skip the search in that case during booting up.
So solution could be increase block size just like SGI UV system did.
(harded code to 2g).
This patch is trying to probe the block size to make it match mmio remap
size. for example, Intel Nehalem later system will have memory range [0,
TOML), [4g, TOMH]. If the memory hole is 2g and total is 128g, TOM will
be 2g, and TOM2 will be 130g.
We could use 2g as block size instead of default 128M. That will reduce
number of entries in /sys/devices/system/memory/
On system 6TiB system will reduce boot time by 35 seconds.
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-06-05 07:06:32 +08:00
|
|
|
|
2015-12-01 00:47:43 +08:00
|
|
|
/* if system is UV or has 64GB of RAM or more, use large blocks */
|
|
|
|
if (is_uv_system() || ((max_pfn << PAGE_SHIFT) >= (64UL << 30)))
|
|
|
|
bz = 2UL << 30; /* 2GB */
|
x86, mm: probe memory block size for generic x86 64bit
On system with 2TiB ram, current x86_64 have 128M as section size, and
one memory_block only include one section. So will have 16400 entries
under /sys/devices/system/memory/.
Current code try to use block id to find block pointer in /sys for any
section, and reuse that block pointer. that finding will take some time
even after commit 7c243c7168dc ("mm: speedup in __early_pfn_to_nid")
that will skip the search in that case during booting up.
So solution could be increase block size just like SGI UV system did.
(harded code to 2g).
This patch is trying to probe the block size to make it match mmio remap
size. for example, Intel Nehalem later system will have memory range [0,
TOML), [4g, TOMH]. If the memory hole is 2g and total is 128g, TOM will
be 2g, and TOM2 will be 130g.
We could use 2g as block size instead of default 128M. That will reduce
number of entries in /sys/devices/system/memory/
On system 6TiB system will reduce boot time by 35 seconds.
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-06-05 07:06:32 +08:00
|
|
|
|
2015-12-01 00:47:43 +08:00
|
|
|
pr_info("x86/mm: Memory block size: %ldMB\n", bz >> 20);
|
x86, mm: probe memory block size for generic x86 64bit
On system with 2TiB ram, current x86_64 have 128M as section size, and
one memory_block only include one section. So will have 16400 entries
under /sys/devices/system/memory/.
Current code try to use block id to find block pointer in /sys for any
section, and reuse that block pointer. that finding will take some time
even after commit 7c243c7168dc ("mm: speedup in __early_pfn_to_nid")
that will skip the search in that case during booting up.
So solution could be increase block size just like SGI UV system did.
(harded code to 2g).
This patch is trying to probe the block size to make it match mmio remap
size. for example, Intel Nehalem later system will have memory range [0,
TOML), [4g, TOMH]. If the memory hole is 2g and total is 128g, TOM will
be 2g, and TOM2 will be 130g.
We could use 2g as block size instead of default 128M. That will reduce
number of entries in /sys/devices/system/memory/
On system 6TiB system will reduce boot time by 35 seconds.
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-06-05 07:06:32 +08:00
|
|
|
|
|
|
|
return bz;
|
|
|
|
}
|
|
|
|
|
|
|
|
static unsigned long memory_block_size_probed;
|
|
|
|
unsigned long memory_block_size_bytes(void)
|
|
|
|
{
|
|
|
|
if (!memory_block_size_probed)
|
|
|
|
memory_block_size_probed = probe_memory_block_size();
|
|
|
|
|
|
|
|
return memory_block_size_probed;
|
|
|
|
}
|
|
|
|
|
2007-10-16 16:24:15 +08:00
|
|
|
#ifdef CONFIG_SPARSEMEM_VMEMMAP
|
|
|
|
/*
|
|
|
|
* Initialise the sparsemem vmemmap using huge-pages at the PMD level.
|
|
|
|
*/
|
2008-04-12 16:19:24 +08:00
|
|
|
static long __meminitdata addr_start, addr_end;
|
|
|
|
static void __meminitdata *p_start, *p_end;
|
|
|
|
static int __meminitdata node_start;
|
|
|
|
|
2013-04-30 06:07:54 +08:00
|
|
|
static int __meminit vmemmap_populate_hugepages(unsigned long start,
|
2016-01-16 08:56:22 +08:00
|
|
|
unsigned long end, int node, struct vmem_altmap *altmap)
|
2007-10-16 16:24:15 +08:00
|
|
|
{
|
2013-04-30 06:07:50 +08:00
|
|
|
unsigned long addr;
|
2007-10-16 16:24:15 +08:00
|
|
|
unsigned long next;
|
|
|
|
pgd_t *pgd;
|
2017-03-18 02:55:15 +08:00
|
|
|
p4d_t *p4d;
|
2007-10-16 16:24:15 +08:00
|
|
|
pud_t *pud;
|
|
|
|
pmd_t *pmd;
|
|
|
|
|
2013-04-30 06:07:50 +08:00
|
|
|
for (addr = start; addr < end; addr = next) {
|
2013-04-30 06:07:54 +08:00
|
|
|
next = pmd_addr_end(addr, end);
|
2007-10-16 16:24:15 +08:00
|
|
|
|
|
|
|
pgd = vmemmap_pgd_populate(addr, node);
|
|
|
|
if (!pgd)
|
|
|
|
return -ENOMEM;
|
2008-01-30 20:34:10 +08:00
|
|
|
|
2017-03-18 02:55:15 +08:00
|
|
|
p4d = vmemmap_p4d_populate(pgd, addr, node);
|
|
|
|
if (!p4d)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
pud = vmemmap_pud_populate(p4d, addr, node);
|
2007-10-16 16:24:15 +08:00
|
|
|
if (!pud)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
2013-04-30 06:07:54 +08:00
|
|
|
pmd = pmd_offset(pud, addr);
|
|
|
|
if (pmd_none(*pmd)) {
|
|
|
|
void *p;
|
2008-01-30 20:34:10 +08:00
|
|
|
|
2016-01-16 08:56:22 +08:00
|
|
|
p = __vmemmap_alloc_block_buf(PMD_SIZE, node, altmap);
|
2013-04-30 06:07:56 +08:00
|
|
|
if (p) {
|
|
|
|
pte_t entry;
|
|
|
|
|
|
|
|
entry = pfn_pte(__pa(p) >> PAGE_SHIFT,
|
|
|
|
PAGE_KERNEL_LARGE);
|
|
|
|
set_pmd(pmd, __pmd(pte_val(entry)));
|
|
|
|
|
|
|
|
/* check to see if we have contiguous blocks */
|
|
|
|
if (p_end != p || node_start != node) {
|
|
|
|
if (p_start)
|
2015-09-18 04:27:57 +08:00
|
|
|
pr_debug(" [%lx-%lx] PMD -> [%p-%p] on node %d\n",
|
2013-04-30 06:07:56 +08:00
|
|
|
addr_start, addr_end-1, p_start, p_end-1, node_start);
|
|
|
|
addr_start = addr;
|
|
|
|
node_start = node;
|
|
|
|
p_start = p;
|
|
|
|
}
|
2008-06-25 12:19:20 +08:00
|
|
|
|
2013-04-30 06:07:56 +08:00
|
|
|
addr_end = addr + PMD_SIZE;
|
|
|
|
p_end = p + PMD_SIZE;
|
|
|
|
continue;
|
2016-01-16 08:56:22 +08:00
|
|
|
} else if (altmap)
|
|
|
|
return -ENOMEM; /* no fallback */
|
2013-04-30 06:07:56 +08:00
|
|
|
} else if (pmd_large(*pmd)) {
|
2013-04-30 06:07:54 +08:00
|
|
|
vmemmap_verify((pte_t *)pmd, node, addr, next);
|
2013-04-30 06:07:56 +08:00
|
|
|
continue;
|
|
|
|
}
|
|
|
|
pr_warn_once("vmemmap: falling back to regular page backing\n");
|
|
|
|
if (vmemmap_populate_basepages(addr, next, node))
|
|
|
|
return -ENOMEM;
|
2007-10-16 16:24:15 +08:00
|
|
|
}
|
|
|
|
return 0;
|
|
|
|
}
|
2008-04-12 16:19:24 +08:00
|
|
|
|
2013-04-30 06:07:54 +08:00
|
|
|
int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node)
|
|
|
|
{
|
2016-01-16 08:56:22 +08:00
|
|
|
struct vmem_altmap *altmap = to_vmem_altmap(start);
|
2013-04-30 06:07:54 +08:00
|
|
|
int err;
|
|
|
|
|
2016-03-29 23:42:03 +08:00
|
|
|
if (boot_cpu_has(X86_FEATURE_PSE))
|
2016-01-16 08:56:22 +08:00
|
|
|
err = vmemmap_populate_hugepages(start, end, node, altmap);
|
|
|
|
else if (altmap) {
|
|
|
|
pr_err_once("%s: no cpu support for altmap allocations\n",
|
|
|
|
__func__);
|
|
|
|
err = -ENOMEM;
|
|
|
|
} else
|
2013-04-30 06:07:54 +08:00
|
|
|
err = vmemmap_populate_basepages(start, end, node);
|
|
|
|
if (!err)
|
2016-12-15 07:44:03 +08:00
|
|
|
sync_global_pgds(start, end - 1);
|
2013-04-30 06:07:54 +08:00
|
|
|
return err;
|
|
|
|
}
|
|
|
|
|
2013-02-23 08:33:00 +08:00
|
|
|
#if defined(CONFIG_MEMORY_HOTPLUG_SPARSE) && defined(CONFIG_HAVE_BOOTMEM_INFO_NODE)
|
|
|
|
void register_page_bootmem_memmap(unsigned long section_nr,
|
|
|
|
struct page *start_page, unsigned long size)
|
|
|
|
{
|
|
|
|
unsigned long addr = (unsigned long)start_page;
|
|
|
|
unsigned long end = (unsigned long)(start_page + size);
|
|
|
|
unsigned long next;
|
|
|
|
pgd_t *pgd;
|
2017-03-18 02:55:15 +08:00
|
|
|
p4d_t *p4d;
|
2013-02-23 08:33:00 +08:00
|
|
|
pud_t *pud;
|
|
|
|
pmd_t *pmd;
|
|
|
|
unsigned int nr_pages;
|
|
|
|
struct page *page;
|
|
|
|
|
|
|
|
for (; addr < end; addr = next) {
|
|
|
|
pte_t *pte = NULL;
|
|
|
|
|
|
|
|
pgd = pgd_offset_k(addr);
|
|
|
|
if (pgd_none(*pgd)) {
|
|
|
|
next = (addr + PAGE_SIZE) & PAGE_MASK;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
get_page_bootmem(section_nr, pgd_page(*pgd), MIX_SECTION_INFO);
|
|
|
|
|
2017-03-18 02:55:15 +08:00
|
|
|
p4d = p4d_offset(pgd, addr);
|
|
|
|
if (p4d_none(*p4d)) {
|
|
|
|
next = (addr + PAGE_SIZE) & PAGE_MASK;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
get_page_bootmem(section_nr, p4d_page(*p4d), MIX_SECTION_INFO);
|
|
|
|
|
|
|
|
pud = pud_offset(p4d, addr);
|
2013-02-23 08:33:00 +08:00
|
|
|
if (pud_none(*pud)) {
|
|
|
|
next = (addr + PAGE_SIZE) & PAGE_MASK;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
get_page_bootmem(section_nr, pud_page(*pud), MIX_SECTION_INFO);
|
|
|
|
|
2016-03-29 23:42:03 +08:00
|
|
|
if (!boot_cpu_has(X86_FEATURE_PSE)) {
|
2013-02-23 08:33:00 +08:00
|
|
|
next = (addr + PAGE_SIZE) & PAGE_MASK;
|
|
|
|
pmd = pmd_offset(pud, addr);
|
|
|
|
if (pmd_none(*pmd))
|
|
|
|
continue;
|
|
|
|
get_page_bootmem(section_nr, pmd_page(*pmd),
|
|
|
|
MIX_SECTION_INFO);
|
|
|
|
|
|
|
|
pte = pte_offset_kernel(pmd, addr);
|
|
|
|
if (pte_none(*pte))
|
|
|
|
continue;
|
|
|
|
get_page_bootmem(section_nr, pte_page(*pte),
|
|
|
|
SECTION_INFO);
|
|
|
|
} else {
|
|
|
|
next = pmd_addr_end(addr, end);
|
|
|
|
|
|
|
|
pmd = pmd_offset(pud, addr);
|
|
|
|
if (pmd_none(*pmd))
|
|
|
|
continue;
|
|
|
|
|
|
|
|
nr_pages = 1 << (get_order(PMD_SIZE));
|
|
|
|
page = pmd_page(*pmd);
|
|
|
|
while (nr_pages--)
|
|
|
|
get_page_bootmem(section_nr, page++,
|
|
|
|
SECTION_INFO);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
2008-04-12 16:19:24 +08:00
|
|
|
void __meminit vmemmap_populate_print_last(void)
|
|
|
|
{
|
|
|
|
if (p_start) {
|
2015-09-18 04:27:57 +08:00
|
|
|
pr_debug(" [%lx-%lx] PMD -> [%p-%p] on node %d\n",
|
2008-04-12 16:19:24 +08:00
|
|
|
addr_start, addr_end-1, p_start, p_end-1, node_start);
|
|
|
|
p_start = NULL;
|
|
|
|
p_end = NULL;
|
|
|
|
node_start = 0;
|
|
|
|
}
|
|
|
|
}
|
2007-10-16 16:24:15 +08:00
|
|
|
#endif
|