1
linux/arch/x86/kernel/suspend_64.c

295 lines
7.0 KiB
C
Raw Normal View History

/*
* Suspend support specific for i386.
*
* Distribute under GPLv2
*
* Copyright (c) 2002 Pavel Machek <pavel@suse.cz>
* Copyright (c) 2001 Patrick Mochel <mochel@osdl.org>
*/
#include <linux/smp.h>
#include <linux/suspend.h>
#include <asm/proto.h>
[PATCH] x86_64: Set up safe page tables during resume The following patch makes swsusp avoid the possible temporary corruption of page translation tables during resume on x86-64. This is achieved by creating a copy of the relevant page tables that will not be modified by swsusp and can be safely used by it on resume. The problem is that during resume on x86-64 swsusp may temporarily corrupt the page tables used for the direct mapping of RAM. If that happens, a page fault occurs and cannot be handled properly, which leads to the solid hang of the affected system. This leads to the loss of the system's state from before suspend and may result in the loss of data or the corruption of filesystems, so it is a serious issue. Also, it appears to happen quite often (for me, as often as 50% of the time). The problem is related to the fact that (at least) one of the PMD entries used in the direct memory mapping (starting at PAGE_OFFSET) points to a page table the physical address of which is much greater than the physical address of the PMD entry itself. Moreover, unfortunately, the physical address of the page table before suspend (i.e. the one stored in the suspend image) happens to be different to the physical address of the corresponding page table used during resume (i.e. the one that is valid right before swsusp_arch_resume() in arch/x86_64/kernel/suspend_asm.S is executed). Thus while the image is restored, the "offending" PMD entry gets overwritten, so it does not point to the right physical address any more (i.e. there's no page table at the address pointed to by it, because it points to the address the page table has been at during suspend). Consequently, if the PMD entry is used later on, and it _is_ used in the process of copying the image pages, a page fault occurs, but it cannot be handled in the normal way and the system hangs. In principle we can call create_resume_mapping() from swsusp_arch_resume() (ie. from suspend_asm.S), but then the memory allocations in create_resume_mapping(), resume_pud_mapping(), and resume_pmd_mapping() must be made carefully so that we use _only_ NosaveFree pages in them (the other pages are overwritten by the loop in swsusp_arch_resume()). Additionally, we are in atomic context at that time, so we cannot use GFP_KERNEL. Moreover, if one of the allocations fails, we should free all of the allocated pages, so we need to trace them somehow. All of this is done in the appended patch, except that the functions populating the page tables are located in arch/x86_64/kernel/suspend.c rather than in init.c. It may be done in a more elegan way in the future, with the help of some swsusp patches that are in the works now. [AK: move some externs into headers, renamed a function] Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Andi Kleen <ak@suse.de> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-09 12:19:40 -07:00
#include <asm/page.h>
#include <asm/pgtable.h>
#include <asm/mtrr.h>
/* References to section boundaries */
extern const void __nosave_begin, __nosave_end;
struct saved_context saved_context;
void __save_processor_state(struct saved_context *ctxt)
{
kernel_fpu_begin();
/*
* descriptor tables
*/
store_gdt((struct desc_ptr *)&ctxt->gdt_limit);
store_idt((struct desc_ptr *)&ctxt->idt_limit);
store_tr(ctxt->tr);
/* XMM0..XMM15 should be handled by kernel_fpu_begin(). */
/*
* segment registers
*/
asm volatile ("movw %%ds, %0" : "=m" (ctxt->ds));
asm volatile ("movw %%es, %0" : "=m" (ctxt->es));
asm volatile ("movw %%fs, %0" : "=m" (ctxt->fs));
asm volatile ("movw %%gs, %0" : "=m" (ctxt->gs));
asm volatile ("movw %%ss, %0" : "=m" (ctxt->ss));
rdmsrl(MSR_FS_BASE, ctxt->fs_base);
rdmsrl(MSR_GS_BASE, ctxt->gs_base);
rdmsrl(MSR_KERNEL_GS_BASE, ctxt->gs_kernel_base);
mtrr_save_fixed_ranges(NULL);
/*
* control registers
*/
rdmsrl(MSR_EFER, ctxt->efer);
ctxt->cr0 = read_cr0();
ctxt->cr2 = read_cr2();
ctxt->cr3 = read_cr3();
ctxt->cr4 = read_cr4();
ctxt->cr8 = read_cr8();
}
void save_processor_state(void)
{
__save_processor_state(&saved_context);
}
static void do_fpu_end(void)
{
/*
* Restore FPU regs if necessary
*/
kernel_fpu_end();
}
void __restore_processor_state(struct saved_context *ctxt)
{
/*
* control registers
*/
wrmsrl(MSR_EFER, ctxt->efer);
write_cr8(ctxt->cr8);
write_cr4(ctxt->cr4);
write_cr3(ctxt->cr3);
write_cr2(ctxt->cr2);
write_cr0(ctxt->cr0);
/*
* now restore the descriptor tables to their proper values
* ltr is done i fix_processor_context().
*/
load_gdt((const struct desc_ptr *)&ctxt->gdt_limit);
load_idt((const struct desc_ptr *)&ctxt->idt_limit);
/*
* segment registers
*/
asm volatile ("movw %0, %%ds" :: "r" (ctxt->ds));
asm volatile ("movw %0, %%es" :: "r" (ctxt->es));
asm volatile ("movw %0, %%fs" :: "r" (ctxt->fs));
load_gs_index(ctxt->gs);
asm volatile ("movw %0, %%ss" :: "r" (ctxt->ss));
wrmsrl(MSR_FS_BASE, ctxt->fs_base);
wrmsrl(MSR_GS_BASE, ctxt->gs_base);
wrmsrl(MSR_KERNEL_GS_BASE, ctxt->gs_kernel_base);
fix_processor_context();
do_fpu_end();
mtrr_ap_init();
}
void restore_processor_state(void)
{
__restore_processor_state(&saved_context);
}
void fix_processor_context(void)
{
int cpu = smp_processor_id();
struct tss_struct *t = &per_cpu(init_tss, cpu);
set_tss_desc(cpu,t); /* This just modifies memory; should not be necessary. But... This is necessary, because 386 hardware has concept of busy TSS or some similar stupidity. */
cpu_gdt(cpu)[GDT_ENTRY_TSS].type = 9;
syscall_init(); /* This sets MSR_*STAR and related */
load_TR_desc(); /* This does ltr */
load_LDT(&current->active_mm->context); /* This does lldt */
/*
* Now maybe reload the debug registers
*/
if (current->thread.debugreg7){
loaddebug(&current->thread, 0);
loaddebug(&current->thread, 1);
loaddebug(&current->thread, 2);
loaddebug(&current->thread, 3);
/* no 4 and 5 */
loaddebug(&current->thread, 6);
loaddebug(&current->thread, 7);
}
}
#ifdef CONFIG_HIBERNATION
[PATCH] x86_64: Set up safe page tables during resume The following patch makes swsusp avoid the possible temporary corruption of page translation tables during resume on x86-64. This is achieved by creating a copy of the relevant page tables that will not be modified by swsusp and can be safely used by it on resume. The problem is that during resume on x86-64 swsusp may temporarily corrupt the page tables used for the direct mapping of RAM. If that happens, a page fault occurs and cannot be handled properly, which leads to the solid hang of the affected system. This leads to the loss of the system's state from before suspend and may result in the loss of data or the corruption of filesystems, so it is a serious issue. Also, it appears to happen quite often (for me, as often as 50% of the time). The problem is related to the fact that (at least) one of the PMD entries used in the direct memory mapping (starting at PAGE_OFFSET) points to a page table the physical address of which is much greater than the physical address of the PMD entry itself. Moreover, unfortunately, the physical address of the page table before suspend (i.e. the one stored in the suspend image) happens to be different to the physical address of the corresponding page table used during resume (i.e. the one that is valid right before swsusp_arch_resume() in arch/x86_64/kernel/suspend_asm.S is executed). Thus while the image is restored, the "offending" PMD entry gets overwritten, so it does not point to the right physical address any more (i.e. there's no page table at the address pointed to by it, because it points to the address the page table has been at during suspend). Consequently, if the PMD entry is used later on, and it _is_ used in the process of copying the image pages, a page fault occurs, but it cannot be handled in the normal way and the system hangs. In principle we can call create_resume_mapping() from swsusp_arch_resume() (ie. from suspend_asm.S), but then the memory allocations in create_resume_mapping(), resume_pud_mapping(), and resume_pmd_mapping() must be made carefully so that we use _only_ NosaveFree pages in them (the other pages are overwritten by the loop in swsusp_arch_resume()). Additionally, we are in atomic context at that time, so we cannot use GFP_KERNEL. Moreover, if one of the allocations fails, we should free all of the allocated pages, so we need to trace them somehow. All of this is done in the appended patch, except that the functions populating the page tables are located in arch/x86_64/kernel/suspend.c rather than in init.c. It may be done in a more elegan way in the future, with the help of some swsusp patches that are in the works now. [AK: move some externs into headers, renamed a function] Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Andi Kleen <ak@suse.de> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-09 12:19:40 -07:00
/* Defined in arch/x86_64/kernel/suspend_asm.S */
extern int restore_image(void);
Hibernation: Arbitrary boot kernel support on x86_64 Make it possible to restore a hibernation image on x86_64 with the help of a kernel different from the one in the image. The idea is to split the core restoration code into two separate parts and to place each of them in a different page.  The first part belongs to the boot kernel and is executed as the last step of the image kernel's memory restoration procedure.  Before being executed, it is relocated to a safe page that won't be overwritten while copying the image kernel pages. The final operation performed by it is a jump to the second part of the core restoration code that belongs to the image kernel and has just been restored. This code makes the CPU switch to the image kernel's page tables and restores the state of general purpose registers (including the stack pointer) from before the hibernation. The main issue with this idea is that in order to jump to the second part of the core restoration code the boot kernel needs to know its address.  However, this address may be passed to it in the image header.  Namely, the part of the image header previously used for checking if the version of the image kernel is correct can be replaced with some architecture specific data that will allow the boot kernel to jump to the right address within the image kernel.  These data should also be used for checking if the image kernel is compatible with the boot kernel (as far as the memory restroration procedure is concerned). It can be done, for example, with the help of a "magic" value that has to be equal in both kernels, so that they can be regarded as compatible. Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Acked-by: Pavel Machek <pavel@ucw.cz> Cc: Andi Kleen <ak@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-18 03:04:53 -07:00
/*
* Address to jump to in the last phase of restore in order to get to the image
* kernel's text (this value is passed in the image header).
*/
unsigned long restore_jump_address;
/*
* Value of the cr3 register from before the hibernation (this value is passed
* in the image header).
*/
unsigned long restore_cr3;
[PATCH] x86_64: Set up safe page tables during resume The following patch makes swsusp avoid the possible temporary corruption of page translation tables during resume on x86-64. This is achieved by creating a copy of the relevant page tables that will not be modified by swsusp and can be safely used by it on resume. The problem is that during resume on x86-64 swsusp may temporarily corrupt the page tables used for the direct mapping of RAM. If that happens, a page fault occurs and cannot be handled properly, which leads to the solid hang of the affected system. This leads to the loss of the system's state from before suspend and may result in the loss of data or the corruption of filesystems, so it is a serious issue. Also, it appears to happen quite often (for me, as often as 50% of the time). The problem is related to the fact that (at least) one of the PMD entries used in the direct memory mapping (starting at PAGE_OFFSET) points to a page table the physical address of which is much greater than the physical address of the PMD entry itself. Moreover, unfortunately, the physical address of the page table before suspend (i.e. the one stored in the suspend image) happens to be different to the physical address of the corresponding page table used during resume (i.e. the one that is valid right before swsusp_arch_resume() in arch/x86_64/kernel/suspend_asm.S is executed). Thus while the image is restored, the "offending" PMD entry gets overwritten, so it does not point to the right physical address any more (i.e. there's no page table at the address pointed to by it, because it points to the address the page table has been at during suspend). Consequently, if the PMD entry is used later on, and it _is_ used in the process of copying the image pages, a page fault occurs, but it cannot be handled in the normal way and the system hangs. In principle we can call create_resume_mapping() from swsusp_arch_resume() (ie. from suspend_asm.S), but then the memory allocations in create_resume_mapping(), resume_pud_mapping(), and resume_pmd_mapping() must be made carefully so that we use _only_ NosaveFree pages in them (the other pages are overwritten by the loop in swsusp_arch_resume()). Additionally, we are in atomic context at that time, so we cannot use GFP_KERNEL. Moreover, if one of the allocations fails, we should free all of the allocated pages, so we need to trace them somehow. All of this is done in the appended patch, except that the functions populating the page tables are located in arch/x86_64/kernel/suspend.c rather than in init.c. It may be done in a more elegan way in the future, with the help of some swsusp patches that are in the works now. [AK: move some externs into headers, renamed a function] Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Andi Kleen <ak@suse.de> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-09 12:19:40 -07:00
pgd_t *temp_level4_pgt;
Hibernation: Arbitrary boot kernel support on x86_64 Make it possible to restore a hibernation image on x86_64 with the help of a kernel different from the one in the image. The idea is to split the core restoration code into two separate parts and to place each of them in a different page.  The first part belongs to the boot kernel and is executed as the last step of the image kernel's memory restoration procedure.  Before being executed, it is relocated to a safe page that won't be overwritten while copying the image kernel pages. The final operation performed by it is a jump to the second part of the core restoration code that belongs to the image kernel and has just been restored. This code makes the CPU switch to the image kernel's page tables and restores the state of general purpose registers (including the stack pointer) from before the hibernation. The main issue with this idea is that in order to jump to the second part of the core restoration code the boot kernel needs to know its address.  However, this address may be passed to it in the image header.  Namely, the part of the image header previously used for checking if the version of the image kernel is correct can be replaced with some architecture specific data that will allow the boot kernel to jump to the right address within the image kernel.  These data should also be used for checking if the image kernel is compatible with the boot kernel (as far as the memory restroration procedure is concerned). It can be done, for example, with the help of a "magic" value that has to be equal in both kernels, so that they can be regarded as compatible. Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Acked-by: Pavel Machek <pavel@ucw.cz> Cc: Andi Kleen <ak@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-18 03:04:53 -07:00
void *relocated_restore_code;
static int res_phys_pud_init(pud_t *pud, unsigned long address, unsigned long end)
[PATCH] x86_64: Set up safe page tables during resume The following patch makes swsusp avoid the possible temporary corruption of page translation tables during resume on x86-64. This is achieved by creating a copy of the relevant page tables that will not be modified by swsusp and can be safely used by it on resume. The problem is that during resume on x86-64 swsusp may temporarily corrupt the page tables used for the direct mapping of RAM. If that happens, a page fault occurs and cannot be handled properly, which leads to the solid hang of the affected system. This leads to the loss of the system's state from before suspend and may result in the loss of data or the corruption of filesystems, so it is a serious issue. Also, it appears to happen quite often (for me, as often as 50% of the time). The problem is related to the fact that (at least) one of the PMD entries used in the direct memory mapping (starting at PAGE_OFFSET) points to a page table the physical address of which is much greater than the physical address of the PMD entry itself. Moreover, unfortunately, the physical address of the page table before suspend (i.e. the one stored in the suspend image) happens to be different to the physical address of the corresponding page table used during resume (i.e. the one that is valid right before swsusp_arch_resume() in arch/x86_64/kernel/suspend_asm.S is executed). Thus while the image is restored, the "offending" PMD entry gets overwritten, so it does not point to the right physical address any more (i.e. there's no page table at the address pointed to by it, because it points to the address the page table has been at during suspend). Consequently, if the PMD entry is used later on, and it _is_ used in the process of copying the image pages, a page fault occurs, but it cannot be handled in the normal way and the system hangs. In principle we can call create_resume_mapping() from swsusp_arch_resume() (ie. from suspend_asm.S), but then the memory allocations in create_resume_mapping(), resume_pud_mapping(), and resume_pmd_mapping() must be made carefully so that we use _only_ NosaveFree pages in them (the other pages are overwritten by the loop in swsusp_arch_resume()). Additionally, we are in atomic context at that time, so we cannot use GFP_KERNEL. Moreover, if one of the allocations fails, we should free all of the allocated pages, so we need to trace them somehow. All of this is done in the appended patch, except that the functions populating the page tables are located in arch/x86_64/kernel/suspend.c rather than in init.c. It may be done in a more elegan way in the future, with the help of some swsusp patches that are in the works now. [AK: move some externs into headers, renamed a function] Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Andi Kleen <ak@suse.de> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-09 12:19:40 -07:00
{
long i, j;
i = pud_index(address);
pud = pud + i;
for (; i < PTRS_PER_PUD; pud++, i++) {
unsigned long paddr;
pmd_t *pmd;
paddr = address + i*PUD_SIZE;
if (paddr >= end)
break;
pmd = (pmd_t *)get_safe_page(GFP_ATOMIC);
if (!pmd)
return -ENOMEM;
[PATCH] x86_64: Set up safe page tables during resume The following patch makes swsusp avoid the possible temporary corruption of page translation tables during resume on x86-64. This is achieved by creating a copy of the relevant page tables that will not be modified by swsusp and can be safely used by it on resume. The problem is that during resume on x86-64 swsusp may temporarily corrupt the page tables used for the direct mapping of RAM. If that happens, a page fault occurs and cannot be handled properly, which leads to the solid hang of the affected system. This leads to the loss of the system's state from before suspend and may result in the loss of data or the corruption of filesystems, so it is a serious issue. Also, it appears to happen quite often (for me, as often as 50% of the time). The problem is related to the fact that (at least) one of the PMD entries used in the direct memory mapping (starting at PAGE_OFFSET) points to a page table the physical address of which is much greater than the physical address of the PMD entry itself. Moreover, unfortunately, the physical address of the page table before suspend (i.e. the one stored in the suspend image) happens to be different to the physical address of the corresponding page table used during resume (i.e. the one that is valid right before swsusp_arch_resume() in arch/x86_64/kernel/suspend_asm.S is executed). Thus while the image is restored, the "offending" PMD entry gets overwritten, so it does not point to the right physical address any more (i.e. there's no page table at the address pointed to by it, because it points to the address the page table has been at during suspend). Consequently, if the PMD entry is used later on, and it _is_ used in the process of copying the image pages, a page fault occurs, but it cannot be handled in the normal way and the system hangs. In principle we can call create_resume_mapping() from swsusp_arch_resume() (ie. from suspend_asm.S), but then the memory allocations in create_resume_mapping(), resume_pud_mapping(), and resume_pmd_mapping() must be made carefully so that we use _only_ NosaveFree pages in them (the other pages are overwritten by the loop in swsusp_arch_resume()). Additionally, we are in atomic context at that time, so we cannot use GFP_KERNEL. Moreover, if one of the allocations fails, we should free all of the allocated pages, so we need to trace them somehow. All of this is done in the appended patch, except that the functions populating the page tables are located in arch/x86_64/kernel/suspend.c rather than in init.c. It may be done in a more elegan way in the future, with the help of some swsusp patches that are in the works now. [AK: move some externs into headers, renamed a function] Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Andi Kleen <ak@suse.de> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-09 12:19:40 -07:00
set_pud(pud, __pud(__pa(pmd) | _KERNPG_TABLE));
for (j = 0; j < PTRS_PER_PMD; pmd++, j++, paddr += PMD_SIZE) {
unsigned long pe;
if (paddr >= end)
break;
Hibernation: Arbitrary boot kernel support on x86_64 Make it possible to restore a hibernation image on x86_64 with the help of a kernel different from the one in the image. The idea is to split the core restoration code into two separate parts and to place each of them in a different page.  The first part belongs to the boot kernel and is executed as the last step of the image kernel's memory restoration procedure.  Before being executed, it is relocated to a safe page that won't be overwritten while copying the image kernel pages. The final operation performed by it is a jump to the second part of the core restoration code that belongs to the image kernel and has just been restored. This code makes the CPU switch to the image kernel's page tables and restores the state of general purpose registers (including the stack pointer) from before the hibernation. The main issue with this idea is that in order to jump to the second part of the core restoration code the boot kernel needs to know its address.  However, this address may be passed to it in the image header.  Namely, the part of the image header previously used for checking if the version of the image kernel is correct can be replaced with some architecture specific data that will allow the boot kernel to jump to the right address within the image kernel.  These data should also be used for checking if the image kernel is compatible with the boot kernel (as far as the memory restroration procedure is concerned). It can be done, for example, with the help of a "magic" value that has to be equal in both kernels, so that they can be regarded as compatible. Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Acked-by: Pavel Machek <pavel@ucw.cz> Cc: Andi Kleen <ak@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-18 03:04:53 -07:00
pe = __PAGE_KERNEL_LARGE_EXEC | paddr;
[PATCH] x86_64: Set up safe page tables during resume The following patch makes swsusp avoid the possible temporary corruption of page translation tables during resume on x86-64. This is achieved by creating a copy of the relevant page tables that will not be modified by swsusp and can be safely used by it on resume. The problem is that during resume on x86-64 swsusp may temporarily corrupt the page tables used for the direct mapping of RAM. If that happens, a page fault occurs and cannot be handled properly, which leads to the solid hang of the affected system. This leads to the loss of the system's state from before suspend and may result in the loss of data or the corruption of filesystems, so it is a serious issue. Also, it appears to happen quite often (for me, as often as 50% of the time). The problem is related to the fact that (at least) one of the PMD entries used in the direct memory mapping (starting at PAGE_OFFSET) points to a page table the physical address of which is much greater than the physical address of the PMD entry itself. Moreover, unfortunately, the physical address of the page table before suspend (i.e. the one stored in the suspend image) happens to be different to the physical address of the corresponding page table used during resume (i.e. the one that is valid right before swsusp_arch_resume() in arch/x86_64/kernel/suspend_asm.S is executed). Thus while the image is restored, the "offending" PMD entry gets overwritten, so it does not point to the right physical address any more (i.e. there's no page table at the address pointed to by it, because it points to the address the page table has been at during suspend). Consequently, if the PMD entry is used later on, and it _is_ used in the process of copying the image pages, a page fault occurs, but it cannot be handled in the normal way and the system hangs. In principle we can call create_resume_mapping() from swsusp_arch_resume() (ie. from suspend_asm.S), but then the memory allocations in create_resume_mapping(), resume_pud_mapping(), and resume_pmd_mapping() must be made carefully so that we use _only_ NosaveFree pages in them (the other pages are overwritten by the loop in swsusp_arch_resume()). Additionally, we are in atomic context at that time, so we cannot use GFP_KERNEL. Moreover, if one of the allocations fails, we should free all of the allocated pages, so we need to trace them somehow. All of this is done in the appended patch, except that the functions populating the page tables are located in arch/x86_64/kernel/suspend.c rather than in init.c. It may be done in a more elegan way in the future, with the help of some swsusp patches that are in the works now. [AK: move some externs into headers, renamed a function] Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Andi Kleen <ak@suse.de> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-09 12:19:40 -07:00
pe &= __supported_pte_mask;
set_pmd(pmd, __pmd(pe));
}
}
return 0;
[PATCH] x86_64: Set up safe page tables during resume The following patch makes swsusp avoid the possible temporary corruption of page translation tables during resume on x86-64. This is achieved by creating a copy of the relevant page tables that will not be modified by swsusp and can be safely used by it on resume. The problem is that during resume on x86-64 swsusp may temporarily corrupt the page tables used for the direct mapping of RAM. If that happens, a page fault occurs and cannot be handled properly, which leads to the solid hang of the affected system. This leads to the loss of the system's state from before suspend and may result in the loss of data or the corruption of filesystems, so it is a serious issue. Also, it appears to happen quite often (for me, as often as 50% of the time). The problem is related to the fact that (at least) one of the PMD entries used in the direct memory mapping (starting at PAGE_OFFSET) points to a page table the physical address of which is much greater than the physical address of the PMD entry itself. Moreover, unfortunately, the physical address of the page table before suspend (i.e. the one stored in the suspend image) happens to be different to the physical address of the corresponding page table used during resume (i.e. the one that is valid right before swsusp_arch_resume() in arch/x86_64/kernel/suspend_asm.S is executed). Thus while the image is restored, the "offending" PMD entry gets overwritten, so it does not point to the right physical address any more (i.e. there's no page table at the address pointed to by it, because it points to the address the page table has been at during suspend). Consequently, if the PMD entry is used later on, and it _is_ used in the process of copying the image pages, a page fault occurs, but it cannot be handled in the normal way and the system hangs. In principle we can call create_resume_mapping() from swsusp_arch_resume() (ie. from suspend_asm.S), but then the memory allocations in create_resume_mapping(), resume_pud_mapping(), and resume_pmd_mapping() must be made carefully so that we use _only_ NosaveFree pages in them (the other pages are overwritten by the loop in swsusp_arch_resume()). Additionally, we are in atomic context at that time, so we cannot use GFP_KERNEL. Moreover, if one of the allocations fails, we should free all of the allocated pages, so we need to trace them somehow. All of this is done in the appended patch, except that the functions populating the page tables are located in arch/x86_64/kernel/suspend.c rather than in init.c. It may be done in a more elegan way in the future, with the help of some swsusp patches that are in the works now. [AK: move some externs into headers, renamed a function] Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Andi Kleen <ak@suse.de> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-09 12:19:40 -07:00
}
static int set_up_temporary_mappings(void)
[PATCH] x86_64: Set up safe page tables during resume The following patch makes swsusp avoid the possible temporary corruption of page translation tables during resume on x86-64. This is achieved by creating a copy of the relevant page tables that will not be modified by swsusp and can be safely used by it on resume. The problem is that during resume on x86-64 swsusp may temporarily corrupt the page tables used for the direct mapping of RAM. If that happens, a page fault occurs and cannot be handled properly, which leads to the solid hang of the affected system. This leads to the loss of the system's state from before suspend and may result in the loss of data or the corruption of filesystems, so it is a serious issue. Also, it appears to happen quite often (for me, as often as 50% of the time). The problem is related to the fact that (at least) one of the PMD entries used in the direct memory mapping (starting at PAGE_OFFSET) points to a page table the physical address of which is much greater than the physical address of the PMD entry itself. Moreover, unfortunately, the physical address of the page table before suspend (i.e. the one stored in the suspend image) happens to be different to the physical address of the corresponding page table used during resume (i.e. the one that is valid right before swsusp_arch_resume() in arch/x86_64/kernel/suspend_asm.S is executed). Thus while the image is restored, the "offending" PMD entry gets overwritten, so it does not point to the right physical address any more (i.e. there's no page table at the address pointed to by it, because it points to the address the page table has been at during suspend). Consequently, if the PMD entry is used later on, and it _is_ used in the process of copying the image pages, a page fault occurs, but it cannot be handled in the normal way and the system hangs. In principle we can call create_resume_mapping() from swsusp_arch_resume() (ie. from suspend_asm.S), but then the memory allocations in create_resume_mapping(), resume_pud_mapping(), and resume_pmd_mapping() must be made carefully so that we use _only_ NosaveFree pages in them (the other pages are overwritten by the loop in swsusp_arch_resume()). Additionally, we are in atomic context at that time, so we cannot use GFP_KERNEL. Moreover, if one of the allocations fails, we should free all of the allocated pages, so we need to trace them somehow. All of this is done in the appended patch, except that the functions populating the page tables are located in arch/x86_64/kernel/suspend.c rather than in init.c. It may be done in a more elegan way in the future, with the help of some swsusp patches that are in the works now. [AK: move some externs into headers, renamed a function] Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Andi Kleen <ak@suse.de> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-09 12:19:40 -07:00
{
unsigned long start, end, next;
int error;
[PATCH] x86_64: Set up safe page tables during resume The following patch makes swsusp avoid the possible temporary corruption of page translation tables during resume on x86-64. This is achieved by creating a copy of the relevant page tables that will not be modified by swsusp and can be safely used by it on resume. The problem is that during resume on x86-64 swsusp may temporarily corrupt the page tables used for the direct mapping of RAM. If that happens, a page fault occurs and cannot be handled properly, which leads to the solid hang of the affected system. This leads to the loss of the system's state from before suspend and may result in the loss of data or the corruption of filesystems, so it is a serious issue. Also, it appears to happen quite often (for me, as often as 50% of the time). The problem is related to the fact that (at least) one of the PMD entries used in the direct memory mapping (starting at PAGE_OFFSET) points to a page table the physical address of which is much greater than the physical address of the PMD entry itself. Moreover, unfortunately, the physical address of the page table before suspend (i.e. the one stored in the suspend image) happens to be different to the physical address of the corresponding page table used during resume (i.e. the one that is valid right before swsusp_arch_resume() in arch/x86_64/kernel/suspend_asm.S is executed). Thus while the image is restored, the "offending" PMD entry gets overwritten, so it does not point to the right physical address any more (i.e. there's no page table at the address pointed to by it, because it points to the address the page table has been at during suspend). Consequently, if the PMD entry is used later on, and it _is_ used in the process of copying the image pages, a page fault occurs, but it cannot be handled in the normal way and the system hangs. In principle we can call create_resume_mapping() from swsusp_arch_resume() (ie. from suspend_asm.S), but then the memory allocations in create_resume_mapping(), resume_pud_mapping(), and resume_pmd_mapping() must be made carefully so that we use _only_ NosaveFree pages in them (the other pages are overwritten by the loop in swsusp_arch_resume()). Additionally, we are in atomic context at that time, so we cannot use GFP_KERNEL. Moreover, if one of the allocations fails, we should free all of the allocated pages, so we need to trace them somehow. All of this is done in the appended patch, except that the functions populating the page tables are located in arch/x86_64/kernel/suspend.c rather than in init.c. It may be done in a more elegan way in the future, with the help of some swsusp patches that are in the works now. [AK: move some externs into headers, renamed a function] Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Andi Kleen <ak@suse.de> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-09 12:19:40 -07:00
temp_level4_pgt = (pgd_t *)get_safe_page(GFP_ATOMIC);
if (!temp_level4_pgt)
return -ENOMEM;
[PATCH] x86_64: Set up safe page tables during resume The following patch makes swsusp avoid the possible temporary corruption of page translation tables during resume on x86-64. This is achieved by creating a copy of the relevant page tables that will not be modified by swsusp and can be safely used by it on resume. The problem is that during resume on x86-64 swsusp may temporarily corrupt the page tables used for the direct mapping of RAM. If that happens, a page fault occurs and cannot be handled properly, which leads to the solid hang of the affected system. This leads to the loss of the system's state from before suspend and may result in the loss of data or the corruption of filesystems, so it is a serious issue. Also, it appears to happen quite often (for me, as often as 50% of the time). The problem is related to the fact that (at least) one of the PMD entries used in the direct memory mapping (starting at PAGE_OFFSET) points to a page table the physical address of which is much greater than the physical address of the PMD entry itself. Moreover, unfortunately, the physical address of the page table before suspend (i.e. the one stored in the suspend image) happens to be different to the physical address of the corresponding page table used during resume (i.e. the one that is valid right before swsusp_arch_resume() in arch/x86_64/kernel/suspend_asm.S is executed). Thus while the image is restored, the "offending" PMD entry gets overwritten, so it does not point to the right physical address any more (i.e. there's no page table at the address pointed to by it, because it points to the address the page table has been at during suspend). Consequently, if the PMD entry is used later on, and it _is_ used in the process of copying the image pages, a page fault occurs, but it cannot be handled in the normal way and the system hangs. In principle we can call create_resume_mapping() from swsusp_arch_resume() (ie. from suspend_asm.S), but then the memory allocations in create_resume_mapping(), resume_pud_mapping(), and resume_pmd_mapping() must be made carefully so that we use _only_ NosaveFree pages in them (the other pages are overwritten by the loop in swsusp_arch_resume()). Additionally, we are in atomic context at that time, so we cannot use GFP_KERNEL. Moreover, if one of the allocations fails, we should free all of the allocated pages, so we need to trace them somehow. All of this is done in the appended patch, except that the functions populating the page tables are located in arch/x86_64/kernel/suspend.c rather than in init.c. It may be done in a more elegan way in the future, with the help of some swsusp patches that are in the works now. [AK: move some externs into headers, renamed a function] Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Andi Kleen <ak@suse.de> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-09 12:19:40 -07:00
/* It is safe to reuse the original kernel mapping */
set_pgd(temp_level4_pgt + pgd_index(__START_KERNEL_map),
init_level4_pgt[pgd_index(__START_KERNEL_map)]);
[PATCH] x86_64: Set up safe page tables during resume The following patch makes swsusp avoid the possible temporary corruption of page translation tables during resume on x86-64. This is achieved by creating a copy of the relevant page tables that will not be modified by swsusp and can be safely used by it on resume. The problem is that during resume on x86-64 swsusp may temporarily corrupt the page tables used for the direct mapping of RAM. If that happens, a page fault occurs and cannot be handled properly, which leads to the solid hang of the affected system. This leads to the loss of the system's state from before suspend and may result in the loss of data or the corruption of filesystems, so it is a serious issue. Also, it appears to happen quite often (for me, as often as 50% of the time). The problem is related to the fact that (at least) one of the PMD entries used in the direct memory mapping (starting at PAGE_OFFSET) points to a page table the physical address of which is much greater than the physical address of the PMD entry itself. Moreover, unfortunately, the physical address of the page table before suspend (i.e. the one stored in the suspend image) happens to be different to the physical address of the corresponding page table used during resume (i.e. the one that is valid right before swsusp_arch_resume() in arch/x86_64/kernel/suspend_asm.S is executed). Thus while the image is restored, the "offending" PMD entry gets overwritten, so it does not point to the right physical address any more (i.e. there's no page table at the address pointed to by it, because it points to the address the page table has been at during suspend). Consequently, if the PMD entry is used later on, and it _is_ used in the process of copying the image pages, a page fault occurs, but it cannot be handled in the normal way and the system hangs. In principle we can call create_resume_mapping() from swsusp_arch_resume() (ie. from suspend_asm.S), but then the memory allocations in create_resume_mapping(), resume_pud_mapping(), and resume_pmd_mapping() must be made carefully so that we use _only_ NosaveFree pages in them (the other pages are overwritten by the loop in swsusp_arch_resume()). Additionally, we are in atomic context at that time, so we cannot use GFP_KERNEL. Moreover, if one of the allocations fails, we should free all of the allocated pages, so we need to trace them somehow. All of this is done in the appended patch, except that the functions populating the page tables are located in arch/x86_64/kernel/suspend.c rather than in init.c. It may be done in a more elegan way in the future, with the help of some swsusp patches that are in the works now. [AK: move some externs into headers, renamed a function] Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Andi Kleen <ak@suse.de> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-09 12:19:40 -07:00
/* Set up the direct mapping from scratch */
start = (unsigned long)pfn_to_kaddr(0);
end = (unsigned long)pfn_to_kaddr(end_pfn);
for (; start < end; start = next) {
pud_t *pud = (pud_t *)get_safe_page(GFP_ATOMIC);
if (!pud)
return -ENOMEM;
[PATCH] x86_64: Set up safe page tables during resume The following patch makes swsusp avoid the possible temporary corruption of page translation tables during resume on x86-64. This is achieved by creating a copy of the relevant page tables that will not be modified by swsusp and can be safely used by it on resume. The problem is that during resume on x86-64 swsusp may temporarily corrupt the page tables used for the direct mapping of RAM. If that happens, a page fault occurs and cannot be handled properly, which leads to the solid hang of the affected system. This leads to the loss of the system's state from before suspend and may result in the loss of data or the corruption of filesystems, so it is a serious issue. Also, it appears to happen quite often (for me, as often as 50% of the time). The problem is related to the fact that (at least) one of the PMD entries used in the direct memory mapping (starting at PAGE_OFFSET) points to a page table the physical address of which is much greater than the physical address of the PMD entry itself. Moreover, unfortunately, the physical address of the page table before suspend (i.e. the one stored in the suspend image) happens to be different to the physical address of the corresponding page table used during resume (i.e. the one that is valid right before swsusp_arch_resume() in arch/x86_64/kernel/suspend_asm.S is executed). Thus while the image is restored, the "offending" PMD entry gets overwritten, so it does not point to the right physical address any more (i.e. there's no page table at the address pointed to by it, because it points to the address the page table has been at during suspend). Consequently, if the PMD entry is used later on, and it _is_ used in the process of copying the image pages, a page fault occurs, but it cannot be handled in the normal way and the system hangs. In principle we can call create_resume_mapping() from swsusp_arch_resume() (ie. from suspend_asm.S), but then the memory allocations in create_resume_mapping(), resume_pud_mapping(), and resume_pmd_mapping() must be made carefully so that we use _only_ NosaveFree pages in them (the other pages are overwritten by the loop in swsusp_arch_resume()). Additionally, we are in atomic context at that time, so we cannot use GFP_KERNEL. Moreover, if one of the allocations fails, we should free all of the allocated pages, so we need to trace them somehow. All of this is done in the appended patch, except that the functions populating the page tables are located in arch/x86_64/kernel/suspend.c rather than in init.c. It may be done in a more elegan way in the future, with the help of some swsusp patches that are in the works now. [AK: move some externs into headers, renamed a function] Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Andi Kleen <ak@suse.de> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-09 12:19:40 -07:00
next = start + PGDIR_SIZE;
if (next > end)
next = end;
if ((error = res_phys_pud_init(pud, __pa(start), __pa(next))))
return error;
[PATCH] x86_64: Set up safe page tables during resume The following patch makes swsusp avoid the possible temporary corruption of page translation tables during resume on x86-64. This is achieved by creating a copy of the relevant page tables that will not be modified by swsusp and can be safely used by it on resume. The problem is that during resume on x86-64 swsusp may temporarily corrupt the page tables used for the direct mapping of RAM. If that happens, a page fault occurs and cannot be handled properly, which leads to the solid hang of the affected system. This leads to the loss of the system's state from before suspend and may result in the loss of data or the corruption of filesystems, so it is a serious issue. Also, it appears to happen quite often (for me, as often as 50% of the time). The problem is related to the fact that (at least) one of the PMD entries used in the direct memory mapping (starting at PAGE_OFFSET) points to a page table the physical address of which is much greater than the physical address of the PMD entry itself. Moreover, unfortunately, the physical address of the page table before suspend (i.e. the one stored in the suspend image) happens to be different to the physical address of the corresponding page table used during resume (i.e. the one that is valid right before swsusp_arch_resume() in arch/x86_64/kernel/suspend_asm.S is executed). Thus while the image is restored, the "offending" PMD entry gets overwritten, so it does not point to the right physical address any more (i.e. there's no page table at the address pointed to by it, because it points to the address the page table has been at during suspend). Consequently, if the PMD entry is used later on, and it _is_ used in the process of copying the image pages, a page fault occurs, but it cannot be handled in the normal way and the system hangs. In principle we can call create_resume_mapping() from swsusp_arch_resume() (ie. from suspend_asm.S), but then the memory allocations in create_resume_mapping(), resume_pud_mapping(), and resume_pmd_mapping() must be made carefully so that we use _only_ NosaveFree pages in them (the other pages are overwritten by the loop in swsusp_arch_resume()). Additionally, we are in atomic context at that time, so we cannot use GFP_KERNEL. Moreover, if one of the allocations fails, we should free all of the allocated pages, so we need to trace them somehow. All of this is done in the appended patch, except that the functions populating the page tables are located in arch/x86_64/kernel/suspend.c rather than in init.c. It may be done in a more elegan way in the future, with the help of some swsusp patches that are in the works now. [AK: move some externs into headers, renamed a function] Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Andi Kleen <ak@suse.de> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-09 12:19:40 -07:00
set_pgd(temp_level4_pgt + pgd_index(start),
mk_kernel_pgd(__pa(pud)));
}
return 0;
[PATCH] x86_64: Set up safe page tables during resume The following patch makes swsusp avoid the possible temporary corruption of page translation tables during resume on x86-64. This is achieved by creating a copy of the relevant page tables that will not be modified by swsusp and can be safely used by it on resume. The problem is that during resume on x86-64 swsusp may temporarily corrupt the page tables used for the direct mapping of RAM. If that happens, a page fault occurs and cannot be handled properly, which leads to the solid hang of the affected system. This leads to the loss of the system's state from before suspend and may result in the loss of data or the corruption of filesystems, so it is a serious issue. Also, it appears to happen quite often (for me, as often as 50% of the time). The problem is related to the fact that (at least) one of the PMD entries used in the direct memory mapping (starting at PAGE_OFFSET) points to a page table the physical address of which is much greater than the physical address of the PMD entry itself. Moreover, unfortunately, the physical address of the page table before suspend (i.e. the one stored in the suspend image) happens to be different to the physical address of the corresponding page table used during resume (i.e. the one that is valid right before swsusp_arch_resume() in arch/x86_64/kernel/suspend_asm.S is executed). Thus while the image is restored, the "offending" PMD entry gets overwritten, so it does not point to the right physical address any more (i.e. there's no page table at the address pointed to by it, because it points to the address the page table has been at during suspend). Consequently, if the PMD entry is used later on, and it _is_ used in the process of copying the image pages, a page fault occurs, but it cannot be handled in the normal way and the system hangs. In principle we can call create_resume_mapping() from swsusp_arch_resume() (ie. from suspend_asm.S), but then the memory allocations in create_resume_mapping(), resume_pud_mapping(), and resume_pmd_mapping() must be made carefully so that we use _only_ NosaveFree pages in them (the other pages are overwritten by the loop in swsusp_arch_resume()). Additionally, we are in atomic context at that time, so we cannot use GFP_KERNEL. Moreover, if one of the allocations fails, we should free all of the allocated pages, so we need to trace them somehow. All of this is done in the appended patch, except that the functions populating the page tables are located in arch/x86_64/kernel/suspend.c rather than in init.c. It may be done in a more elegan way in the future, with the help of some swsusp patches that are in the works now. [AK: move some externs into headers, renamed a function] Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Andi Kleen <ak@suse.de> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-09 12:19:40 -07:00
}
int swsusp_arch_resume(void)
{
int error;
[PATCH] x86_64: Set up safe page tables during resume The following patch makes swsusp avoid the possible temporary corruption of page translation tables during resume on x86-64. This is achieved by creating a copy of the relevant page tables that will not be modified by swsusp and can be safely used by it on resume. The problem is that during resume on x86-64 swsusp may temporarily corrupt the page tables used for the direct mapping of RAM. If that happens, a page fault occurs and cannot be handled properly, which leads to the solid hang of the affected system. This leads to the loss of the system's state from before suspend and may result in the loss of data or the corruption of filesystems, so it is a serious issue. Also, it appears to happen quite often (for me, as often as 50% of the time). The problem is related to the fact that (at least) one of the PMD entries used in the direct memory mapping (starting at PAGE_OFFSET) points to a page table the physical address of which is much greater than the physical address of the PMD entry itself. Moreover, unfortunately, the physical address of the page table before suspend (i.e. the one stored in the suspend image) happens to be different to the physical address of the corresponding page table used during resume (i.e. the one that is valid right before swsusp_arch_resume() in arch/x86_64/kernel/suspend_asm.S is executed). Thus while the image is restored, the "offending" PMD entry gets overwritten, so it does not point to the right physical address any more (i.e. there's no page table at the address pointed to by it, because it points to the address the page table has been at during suspend). Consequently, if the PMD entry is used later on, and it _is_ used in the process of copying the image pages, a page fault occurs, but it cannot be handled in the normal way and the system hangs. In principle we can call create_resume_mapping() from swsusp_arch_resume() (ie. from suspend_asm.S), but then the memory allocations in create_resume_mapping(), resume_pud_mapping(), and resume_pmd_mapping() must be made carefully so that we use _only_ NosaveFree pages in them (the other pages are overwritten by the loop in swsusp_arch_resume()). Additionally, we are in atomic context at that time, so we cannot use GFP_KERNEL. Moreover, if one of the allocations fails, we should free all of the allocated pages, so we need to trace them somehow. All of this is done in the appended patch, except that the functions populating the page tables are located in arch/x86_64/kernel/suspend.c rather than in init.c. It may be done in a more elegan way in the future, with the help of some swsusp patches that are in the works now. [AK: move some externs into headers, renamed a function] Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Andi Kleen <ak@suse.de> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-09 12:19:40 -07:00
/* We have got enough memory and from now on we cannot recover */
if ((error = set_up_temporary_mappings()))
return error;
Hibernation: Arbitrary boot kernel support on x86_64 Make it possible to restore a hibernation image on x86_64 with the help of a kernel different from the one in the image. The idea is to split the core restoration code into two separate parts and to place each of them in a different page.  The first part belongs to the boot kernel and is executed as the last step of the image kernel's memory restoration procedure.  Before being executed, it is relocated to a safe page that won't be overwritten while copying the image kernel pages. The final operation performed by it is a jump to the second part of the core restoration code that belongs to the image kernel and has just been restored. This code makes the CPU switch to the image kernel's page tables and restores the state of general purpose registers (including the stack pointer) from before the hibernation. The main issue with this idea is that in order to jump to the second part of the core restoration code the boot kernel needs to know its address.  However, this address may be passed to it in the image header.  Namely, the part of the image header previously used for checking if the version of the image kernel is correct can be replaced with some architecture specific data that will allow the boot kernel to jump to the right address within the image kernel.  These data should also be used for checking if the image kernel is compatible with the boot kernel (as far as the memory restroration procedure is concerned). It can be done, for example, with the help of a "magic" value that has to be equal in both kernels, so that they can be regarded as compatible. Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Acked-by: Pavel Machek <pavel@ucw.cz> Cc: Andi Kleen <ak@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-18 03:04:53 -07:00
relocated_restore_code = (void *)get_safe_page(GFP_ATOMIC);
if (!relocated_restore_code)
return -ENOMEM;
memcpy(relocated_restore_code, &core_restore_code,
&restore_registers - &core_restore_code);
[PATCH] x86_64: Set up safe page tables during resume The following patch makes swsusp avoid the possible temporary corruption of page translation tables during resume on x86-64. This is achieved by creating a copy of the relevant page tables that will not be modified by swsusp and can be safely used by it on resume. The problem is that during resume on x86-64 swsusp may temporarily corrupt the page tables used for the direct mapping of RAM. If that happens, a page fault occurs and cannot be handled properly, which leads to the solid hang of the affected system. This leads to the loss of the system's state from before suspend and may result in the loss of data or the corruption of filesystems, so it is a serious issue. Also, it appears to happen quite often (for me, as often as 50% of the time). The problem is related to the fact that (at least) one of the PMD entries used in the direct memory mapping (starting at PAGE_OFFSET) points to a page table the physical address of which is much greater than the physical address of the PMD entry itself. Moreover, unfortunately, the physical address of the page table before suspend (i.e. the one stored in the suspend image) happens to be different to the physical address of the corresponding page table used during resume (i.e. the one that is valid right before swsusp_arch_resume() in arch/x86_64/kernel/suspend_asm.S is executed). Thus while the image is restored, the "offending" PMD entry gets overwritten, so it does not point to the right physical address any more (i.e. there's no page table at the address pointed to by it, because it points to the address the page table has been at during suspend). Consequently, if the PMD entry is used later on, and it _is_ used in the process of copying the image pages, a page fault occurs, but it cannot be handled in the normal way and the system hangs. In principle we can call create_resume_mapping() from swsusp_arch_resume() (ie. from suspend_asm.S), but then the memory allocations in create_resume_mapping(), resume_pud_mapping(), and resume_pmd_mapping() must be made carefully so that we use _only_ NosaveFree pages in them (the other pages are overwritten by the loop in swsusp_arch_resume()). Additionally, we are in atomic context at that time, so we cannot use GFP_KERNEL. Moreover, if one of the allocations fails, we should free all of the allocated pages, so we need to trace them somehow. All of this is done in the appended patch, except that the functions populating the page tables are located in arch/x86_64/kernel/suspend.c rather than in init.c. It may be done in a more elegan way in the future, with the help of some swsusp patches that are in the works now. [AK: move some externs into headers, renamed a function] Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Andi Kleen <ak@suse.de> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-09 12:19:40 -07:00
restore_image();
return 0;
}
/*
* pfn_is_nosave - check if given pfn is in the 'nosave' section
*/
int pfn_is_nosave(unsigned long pfn)
{
unsigned long nosave_begin_pfn = __pa_symbol(&__nosave_begin) >> PAGE_SHIFT;
unsigned long nosave_end_pfn = PAGE_ALIGN(__pa_symbol(&__nosave_end)) >> PAGE_SHIFT;
return (pfn >= nosave_begin_pfn) && (pfn < nosave_end_pfn);
}
Hibernation: Arbitrary boot kernel support on x86_64 Make it possible to restore a hibernation image on x86_64 with the help of a kernel different from the one in the image. The idea is to split the core restoration code into two separate parts and to place each of them in a different page.  The first part belongs to the boot kernel and is executed as the last step of the image kernel's memory restoration procedure.  Before being executed, it is relocated to a safe page that won't be overwritten while copying the image kernel pages. The final operation performed by it is a jump to the second part of the core restoration code that belongs to the image kernel and has just been restored. This code makes the CPU switch to the image kernel's page tables and restores the state of general purpose registers (including the stack pointer) from before the hibernation. The main issue with this idea is that in order to jump to the second part of the core restoration code the boot kernel needs to know its address.  However, this address may be passed to it in the image header.  Namely, the part of the image header previously used for checking if the version of the image kernel is correct can be replaced with some architecture specific data that will allow the boot kernel to jump to the right address within the image kernel.  These data should also be used for checking if the image kernel is compatible with the boot kernel (as far as the memory restroration procedure is concerned). It can be done, for example, with the help of a "magic" value that has to be equal in both kernels, so that they can be regarded as compatible. Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Acked-by: Pavel Machek <pavel@ucw.cz> Cc: Andi Kleen <ak@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-18 03:04:53 -07:00
struct restore_data_record {
unsigned long jump_address;
unsigned long cr3;
unsigned long magic;
Hibernation: Arbitrary boot kernel support on x86_64 Make it possible to restore a hibernation image on x86_64 with the help of a kernel different from the one in the image. The idea is to split the core restoration code into two separate parts and to place each of them in a different page.  The first part belongs to the boot kernel and is executed as the last step of the image kernel's memory restoration procedure.  Before being executed, it is relocated to a safe page that won't be overwritten while copying the image kernel pages. The final operation performed by it is a jump to the second part of the core restoration code that belongs to the image kernel and has just been restored. This code makes the CPU switch to the image kernel's page tables and restores the state of general purpose registers (including the stack pointer) from before the hibernation. The main issue with this idea is that in order to jump to the second part of the core restoration code the boot kernel needs to know its address.  However, this address may be passed to it in the image header.  Namely, the part of the image header previously used for checking if the version of the image kernel is correct can be replaced with some architecture specific data that will allow the boot kernel to jump to the right address within the image kernel.  These data should also be used for checking if the image kernel is compatible with the boot kernel (as far as the memory restroration procedure is concerned). It can be done, for example, with the help of a "magic" value that has to be equal in both kernels, so that they can be regarded as compatible. Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Acked-by: Pavel Machek <pavel@ucw.cz> Cc: Andi Kleen <ak@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-18 03:04:53 -07:00
};
#define RESTORE_MAGIC 0x0123456789ABCDEFUL
/**
* arch_hibernation_header_save - populate the architecture specific part
* of a hibernation image header
* @addr: address to save the data at
*/
int arch_hibernation_header_save(void *addr, unsigned int max_size)
{
struct restore_data_record *rdr = addr;
if (max_size < sizeof(struct restore_data_record))
return -EOVERFLOW;
rdr->jump_address = restore_jump_address;
rdr->cr3 = restore_cr3;
rdr->magic = RESTORE_MAGIC;
Hibernation: Arbitrary boot kernel support on x86_64 Make it possible to restore a hibernation image on x86_64 with the help of a kernel different from the one in the image. The idea is to split the core restoration code into two separate parts and to place each of them in a different page.  The first part belongs to the boot kernel and is executed as the last step of the image kernel's memory restoration procedure.  Before being executed, it is relocated to a safe page that won't be overwritten while copying the image kernel pages. The final operation performed by it is a jump to the second part of the core restoration code that belongs to the image kernel and has just been restored. This code makes the CPU switch to the image kernel's page tables and restores the state of general purpose registers (including the stack pointer) from before the hibernation. The main issue with this idea is that in order to jump to the second part of the core restoration code the boot kernel needs to know its address.  However, this address may be passed to it in the image header.  Namely, the part of the image header previously used for checking if the version of the image kernel is correct can be replaced with some architecture specific data that will allow the boot kernel to jump to the right address within the image kernel.  These data should also be used for checking if the image kernel is compatible with the boot kernel (as far as the memory restroration procedure is concerned). It can be done, for example, with the help of a "magic" value that has to be equal in both kernels, so that they can be regarded as compatible. Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Acked-by: Pavel Machek <pavel@ucw.cz> Cc: Andi Kleen <ak@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-18 03:04:53 -07:00
return 0;
}
/**
* arch_hibernation_header_restore - read the architecture specific data
* from the hibernation image header
* @addr: address to read the data from
*/
int arch_hibernation_header_restore(void *addr)
{
struct restore_data_record *rdr = addr;
restore_jump_address = rdr->jump_address;
restore_cr3 = rdr->cr3;
return (rdr->magic == RESTORE_MAGIC) ? 0 : -EINVAL;
Hibernation: Arbitrary boot kernel support on x86_64 Make it possible to restore a hibernation image on x86_64 with the help of a kernel different from the one in the image. The idea is to split the core restoration code into two separate parts and to place each of them in a different page.  The first part belongs to the boot kernel and is executed as the last step of the image kernel's memory restoration procedure.  Before being executed, it is relocated to a safe page that won't be overwritten while copying the image kernel pages. The final operation performed by it is a jump to the second part of the core restoration code that belongs to the image kernel and has just been restored. This code makes the CPU switch to the image kernel's page tables and restores the state of general purpose registers (including the stack pointer) from before the hibernation. The main issue with this idea is that in order to jump to the second part of the core restoration code the boot kernel needs to know its address.  However, this address may be passed to it in the image header.  Namely, the part of the image header previously used for checking if the version of the image kernel is correct can be replaced with some architecture specific data that will allow the boot kernel to jump to the right address within the image kernel.  These data should also be used for checking if the image kernel is compatible with the boot kernel (as far as the memory restroration procedure is concerned). It can be done, for example, with the help of a "magic" value that has to be equal in both kernels, so that they can be regarded as compatible. Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Acked-by: Pavel Machek <pavel@ucw.cz> Cc: Andi Kleen <ak@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-18 03:04:53 -07:00
}
#endif /* CONFIG_HIBERNATION */