[PATCH] OLPC: Add XO-1 suspend/resume support

[PATCH] OLPC: Add XO-1 suspend/resume support

[Daniel Drake]

Add code needed for basic suspend/resume of the XO-1 laptop.

swsusp_pg_dir needs to be exposed as it is used by the assembly
code run in the wakeup path.

Signed-off-by: Daniel Drake <dsd@laptop.org>
---
 arch/x86/include/asm/olpc.h       |    5 +-
 arch/x86/kernel/Makefile          |    2 +-
 arch/x86/kernel/olpc-xo1-wakeup.S |  132 +++++++++++++++++++++++++++++++++++++
 arch/x86/kernel/olpc-xo1.c        |   79 ++++++++++++++++++++++
 arch/x86/mm/init_32.c             |    6 +-
 5 files changed, 219 insertions(+), 5 deletions(-)
 create mode 100644 arch/x86/kernel/olpc-xo1-wakeup.S

diff --git a/arch/x86/include/asm/olpc.h b/arch/x86/include/asm/olpc.h
index 101229b..54dfe92 100644
--- a/arch/x86/include/asm/olpc.h
+++ b/arch/x86/include/asm/olpc.h
@@ -88,7 +88,10 @@ extern int olpc_ec_mask_unset(uint8_t bits);
 
 /* EC commands */
 
-#define EC_FIRMWARE_REV		0x08
+#define EC_FIRMWARE_REV			0x08
+#define EC_WAKE_UP_WLAN			0x24
+#define EC_SET_SCI_INHIBIT		0x32
+#define EC_SET_SCI_INHIBIT_RELEASE	0x34
 
 /* SCI source values */
 
diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile
index 4983b61..1b23334 100644
--- a/arch/x86/kernel/Makefile
+++ b/arch/x86/kernel/Makefile
@@ -106,7 +106,7 @@ obj-$(CONFIG_SCx200)		+= scx200.o
 scx200-y			+= scx200_32.o
 
 obj-$(CONFIG_OLPC)		+= olpc.o
-obj-$(CONFIG_OLPC_XO1)		+= olpc-xo1.o
+obj-$(CONFIG_OLPC_XO1)		+= olpc-xo1.o olpc-xo1-wakeup.o
 obj-$(CONFIG_OLPC_OPENFIRMWARE)	+= olpc_ofw.o
 obj-$(CONFIG_X86_MRST)		+= mrst.o
 
diff --git a/arch/x86/kernel/olpc-xo1-wakeup.S b/arch/x86/kernel/olpc-xo1-wakeup.S
new file mode 100644
index 0000000..d335074
--- /dev/null
+++ b/arch/x86/kernel/olpc-xo1-wakeup.S
@@ -0,0 +1,132 @@
+.text
+#include <linux/linkage.h>
+#include <asm/segment.h>
+#include <asm/page.h>
+
+	.macro writepost,value
+		movb $0x34, %al
+		outb %al, $0x70
+		movb $\value, %al
+		outb %al, $0x71
+	.endm
+
+ALIGN
+	.align 4096
+
+wakeup_start:
+#	jmp wakeup_start
+
+	cli
+	cld
+
+	# Clear any dangerous flags
+
+	pushl $0
+	popfl
+
+	writepost 0x31
+
+	# Set up %cr3
+	movl $swsusp_pg_dir - __PAGE_OFFSET, %eax
+	movl %eax, %cr3
+
+	movl saved_cr4, %eax
+	movl %eax, %cr4
+
+	movl saved_cr0, %eax
+	movl %eax, %cr0
+
+	jmp 1f
+1:
+	ljmpl $__KERNEL_CS,$wakeup_return
+
+
+.org 0x1000
+
+wakeup_return:
+	movw    $__KERNEL_DS, %ax
+	movw    %ax, %ss
+	movw    %ax, %ds
+	movw    %ax, %es
+	movw    %ax, %fs
+	movw    %ax, %gs
+
+	lgdt    saved_gdt
+	lidt    saved_idt
+	lldt    saved_ldt
+	ljmp    $(__KERNEL_CS),$1f
+1:
+	movl    %cr3, %eax
+	movl    %eax, %cr3
+	wbinvd
+
+	# Go back to the return point
+	jmp ret_point
+
+save_registers:
+	sgdt  saved_gdt
+	sidt  saved_idt
+	sldt  saved_ldt
+
+	pushl %edx
+	movl %cr4, %edx
+	movl %edx, saved_cr4
+
+	movl %cr0, %edx
+	movl %edx, saved_cr0
+
+	popl %edx
+
+	movl %ebx, saved_context_ebx
+	movl %ebp, saved_context_ebp
+	movl %esi, saved_context_esi
+	movl %edi, saved_context_edi
+
+	pushfl
+	popl saved_context_eflags
+
+	ret
+
+
+restore_registers:
+	movl saved_context_ebp, %ebp
+	movl saved_context_ebx, %ebx
+	movl saved_context_esi, %esi
+	movl saved_context_edi, %edi
+
+	pushl saved_context_eflags
+	popfl
+
+	ret
+
+
+ENTRY(do_olpc_suspend_lowlevel)
+	call	save_processor_state
+	call	save_registers
+
+	# This is the stack context we want to remember
+	movl %esp, saved_context_esp
+
+	pushl	$3
+	call	olpc_xo1_do_sleep
+
+	jmp	wakeup_start
+	.p2align 4,,7
+ret_point:
+	movl    saved_context_esp, %esp
+
+	writepost 0x32
+
+	call	restore_registers
+	call	restore_processor_state
+	ret
+
+.data
+ALIGN
+
+saved_gdt:     .long   0,0
+saved_idt:     .long   0,0
+saved_ldt:     .long   0
+saved_cr4:     .long   0
+saved_cr0:     .long   0
+
diff --git a/arch/x86/kernel/olpc-xo1.c b/arch/x86/kernel/olpc-xo1.c
index f5442c0..9a06081 100644
--- a/arch/x86/kernel/olpc-xo1.c
+++ b/arch/x86/kernel/olpc-xo1.c
@@ -16,6 +16,7 @@
 #include <linux/pci_ids.h>
 #include <linux/platform_device.h>
 #include <linux/pm.h>
+#include <linux/suspend.h>
 
 #include <asm/io.h>
 #include <asm/olpc.h>
@@ -33,12 +34,83 @@
 #define PM_SSC		0x54
 
 /* PM registers (ACPI block) */
+#define PM1_STS		0x00
 #define PM1_CNT		0x08
 #define PM_GPE0_STS	0x18
 
+#define CS5536_PM_PWRBTN (1 << 8)
+
+extern void do_olpc_suspend_lowlevel(void);
+
 static unsigned long acpi_base;
 static unsigned long pms_base;
 
+static struct {
+	unsigned long address;
+	unsigned short segment;
+} ofw_bios_entry = { 0xF0000 + PAGE_OFFSET, __KERNEL_CS };
+
+static int xo1_power_state_enter(suspend_state_t pm_state)
+{
+	int r;
+
+	/* Only STR is supported */
+	if (pm_state != PM_SUSPEND_MEM)
+		return -EINVAL;
+
+	r = olpc_ec_cmd(EC_SET_SCI_INHIBIT, NULL, 0, NULL, 0);
+	if (r)
+		return r;
+
+	/* Save CPU state */
+	do_olpc_suspend_lowlevel();
+
+	/* Resume path starts here */
+
+	/* Tell the EC to stop inhibiting SCIs */
+	olpc_ec_cmd(EC_SET_SCI_INHIBIT_RELEASE, NULL, 0, NULL, 0);
+
+	/*
+	 * Tell the wireless module to restart USB communication.
+	 * Must be done twice.
+	 */
+	olpc_ec_cmd(EC_WAKE_UP_WLAN, NULL, 0, NULL, 0);
+	olpc_ec_cmd(EC_WAKE_UP_WLAN, NULL, 0, NULL, 0);
+
+	return 0;
+}
+
+asmlinkage int olpc_xo1_do_sleep(u8 sleep_state)
+{
+	void *pgd_addr = __va(read_cr3());
+	printk(KERN_ERR "xo1_do_sleep!\n"); /* this needs to remain here so
+					      * that gcc doesn't optimize
+					      * away our __va! */
+
+	/* Enable wakeup through power button */
+	outl((CS5536_PM_PWRBTN << 16) | 0xFFFF, acpi_base + PM1_STS);
+
+	__asm__ __volatile__("movl %0,%%eax" : : "r" (pgd_addr));
+	__asm__("call *(%%edi); cld"
+		: : "D" (&ofw_bios_entry));
+	__asm__ __volatile__("movb $0x34, %al\n\t"
+			     "outb %al, $0x70\n\t"
+			     "movb $0x30, %al\n\t"
+			     "outb %al, $0x71\n\t");
+	return 0;
+}
+
+static int xo1_power_state_valid(suspend_state_t pm_state)
+{
+	/* suspend-to-RAM only */
+	return pm_state == PM_SUSPEND_MEM;
+}
+
+static struct platform_suspend_ops xo1_suspend_ops = {
+	.valid = xo1_power_state_valid,
+	.enter = xo1_power_state_enter,
+};
+
 static void xo1_power_off(void)
 {
 	printk(KERN_INFO "OLPC XO-1 power off sequence...\n");
@@ -101,6 +173,13 @@ static int __devinit olpc_xo1_probe(struct platform_device *pdev)
 	if (r)
 		return r;
 
+	/*
+	 * Take a reference on ourself to prevent module unloading. We can't
+	 * safely unload after changing the suspend handlers.
+	 */
+	__module_get(THIS_MODULE);
+
+	suspend_set_ops(&xo1_suspend_ops);
 	pm_power_off = xo1_power_off;
 
 	printk(KERN_INFO "OLPC XO-1 support registered\n");
diff --git a/arch/x86/mm/init_32.c b/arch/x86/mm/init_32.c
index 43bbc29..dbdab67 100644
--- a/arch/x86/mm/init_32.c
+++ b/arch/x86/mm/init_32.c
@@ -549,7 +549,7 @@ static void __init pagetable_init(void)
 	permanent_kmaps_init(pgd_base);
 }
 
-#ifdef CONFIG_ACPI_SLEEP
+#if defined(CONFIG_ACPI_SLEEP) || defined(CONFIG_OLPC_XO1)
 /*
  * ACPI suspend needs this for resume, because things like the intel-agp
  * driver might have split up a kernel 4MB mapping.
@@ -561,11 +561,11 @@ static inline void save_pg_dir(void)
 {
 	copy_page(swsusp_pg_dir, swapper_pg_dir);
 }
-#else /* !CONFIG_ACPI_SLEEP */
+#else /* !CONFIG_ACPI_SLEEP && !CONFIG_OLPC_XO1 */
 static inline void save_pg_dir(void)
 {
 }
-#endif /* !CONFIG_ACPI_SLEEP */
+#endif /* !CONFIG_ACPI_SLEEP && !CONFIG_OLPC_XO1 */
 
 void zap_low_mappings(bool early)
 {
-- 
1.7.2.3

Re: [PATCH] OLPC: Add XO-1 suspend/resume support

[Randy Dunlap]

On Tue, 19 Oct 2010 23:01:58 +0100 (BST) Daniel Drake wrote:

> Add code needed for basic suspend/resume of the XO-1 laptop.
> 
> swsusp_pg_dir needs to be exposed as it is used by the assembly
> code run in the wakeup path.
> 
> Signed-off-by: Daniel Drake <dsd@laptop.org>
> ---
>  arch/x86/include/asm/olpc.h       |    5 +-
>  arch/x86/kernel/Makefile          |    2 +-
>  arch/x86/kernel/olpc-xo1-wakeup.S |  132 +++++++++++++++++++++++++++++++++++++
>  arch/x86/kernel/olpc-xo1.c        |   79 ++++++++++++++++++++++
>  arch/x86/mm/init_32.c             |    6 +-
>  5 files changed, 219 insertions(+), 5 deletions(-)
>  create mode 100644 arch/x86/kernel/olpc-xo1-wakeup.S
> 
> diff --git a/arch/x86/include/asm/olpc.h b/arch/x86/include/asm/olpc.h
> index 101229b..54dfe92 100644
> --- a/arch/x86/include/asm/olpc.h
> +++ b/arch/x86/include/asm/olpc.h
> @@ -88,7 +88,10 @@ extern int olpc_ec_mask_unset(uint8_t bits);
>  
>  /* EC commands */
>  
> -#define EC_FIRMWARE_REV		0x08
> +#define EC_FIRMWARE_REV			0x08
> +#define EC_WAKE_UP_WLAN			0x24
> +#define EC_SET_SCI_INHIBIT		0x32
> +#define EC_SET_SCI_INHIBIT_RELEASE	0x34
>  
>  /* SCI source values */
>  
> diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile
> index 4983b61..1b23334 100644
> --- a/arch/x86/kernel/Makefile
> +++ b/arch/x86/kernel/Makefile
> @@ -106,7 +106,7 @@ obj-$(CONFIG_SCx200)		+= scx200.o
>  scx200-y			+= scx200_32.o
>  
>  obj-$(CONFIG_OLPC)		+= olpc.o
> -obj-$(CONFIG_OLPC_XO1)		+= olpc-xo1.o
> +obj-$(CONFIG_OLPC_XO1)		+= olpc-xo1.o olpc-xo1-wakeup.o
>  obj-$(CONFIG_OLPC_OPENFIRMWARE)	+= olpc_ofw.o
>  obj-$(CONFIG_X86_MRST)		+= mrst.o
>  

Hi,
Does this build OK when CONFIG_PM is not enabled?
or is that config not possible?


> diff --git a/arch/x86/kernel/olpc-xo1-wakeup.S b/arch/x86/kernel/olpc-xo1-wakeup.S
> new file mode 100644
> index 0000000..d335074
> --- /dev/null
> +++ b/arch/x86/kernel/olpc-xo1-wakeup.S
> @@ -0,0 +1,132 @@
> +.text
> +#include <linux/linkage.h>
> +#include <asm/segment.h>
> +#include <asm/page.h>
> +
> +	.macro writepost,value
> +		movb $0x34, %al
> +		outb %al, $0x70
> +		movb $\value, %al
> +		outb %al, $0x71
> +	.endm
> +
> +ALIGN
> +	.align 4096
> +
> +wakeup_start:
> +#	jmp wakeup_start
> +
> +	cli
> +	cld
> +
> +	# Clear any dangerous flags
> +
> +	pushl $0
> +	popfl
> +
> +	writepost 0x31
> +
> +	# Set up %cr3
> +	movl $swsusp_pg_dir - __PAGE_OFFSET, %eax
> +	movl %eax, %cr3
> +
> +	movl saved_cr4, %eax
> +	movl %eax, %cr4
> +
> +	movl saved_cr0, %eax
> +	movl %eax, %cr0
> +
> +	jmp 1f
> +1:
> +	ljmpl $__KERNEL_CS,$wakeup_return
> +
> +
> +.org 0x1000
> +
> +wakeup_return:
> +	movw    $__KERNEL_DS, %ax
> +	movw    %ax, %ss
> +	movw    %ax, %ds
> +	movw    %ax, %es
> +	movw    %ax, %fs
> +	movw    %ax, %gs
> +
> +	lgdt    saved_gdt
> +	lidt    saved_idt
> +	lldt    saved_ldt
> +	ljmp    $(__KERNEL_CS),$1f
> +1:
> +	movl    %cr3, %eax
> +	movl    %eax, %cr3
> +	wbinvd
> +
> +	# Go back to the return point
> +	jmp ret_point
> +
> +save_registers:
> +	sgdt  saved_gdt
> +	sidt  saved_idt
> +	sldt  saved_ldt
> +
> +	pushl %edx
> +	movl %cr4, %edx
> +	movl %edx, saved_cr4
> +
> +	movl %cr0, %edx
> +	movl %edx, saved_cr0
> +
> +	popl %edx
> +
> +	movl %ebx, saved_context_ebx
> +	movl %ebp, saved_context_ebp
> +	movl %esi, saved_context_esi
> +	movl %edi, saved_context_edi
> +
> +	pushfl
> +	popl saved_context_eflags
> +
> +	ret
> +
> +
> +restore_registers:
> +	movl saved_context_ebp, %ebp
> +	movl saved_context_ebx, %ebx
> +	movl saved_context_esi, %esi
> +	movl saved_context_edi, %edi
> +
> +	pushl saved_context_eflags
> +	popfl
> +
> +	ret
> +
> +
> +ENTRY(do_olpc_suspend_lowlevel)
> +	call	save_processor_state
> +	call	save_registers
> +
> +	# This is the stack context we want to remember
> +	movl %esp, saved_context_esp
> +
> +	pushl	$3
> +	call	olpc_xo1_do_sleep
> +
> +	jmp	wakeup_start
> +	.p2align 4,,7
> +ret_point:
> +	movl    saved_context_esp, %esp
> +
> +	writepost 0x32
> +
> +	call	restore_registers
> +	call	restore_processor_state
> +	ret
> +
> +.data
> +ALIGN
> +
> +saved_gdt:     .long   0,0
> +saved_idt:     .long   0,0
> +saved_ldt:     .long   0
> +saved_cr4:     .long   0
> +saved_cr0:     .long   0
> +
> diff --git a/arch/x86/kernel/olpc-xo1.c b/arch/x86/kernel/olpc-xo1.c
> index f5442c0..9a06081 100644
> --- a/arch/x86/kernel/olpc-xo1.c
> +++ b/arch/x86/kernel/olpc-xo1.c
> @@ -16,6 +16,7 @@
>  #include <linux/pci_ids.h>
>  #include <linux/platform_device.h>
>  #include <linux/pm.h>
> +#include <linux/suspend.h>
>  
>  #include <asm/io.h>
>  #include <asm/olpc.h>
> @@ -33,12 +34,83 @@
>  #define PM_SSC		0x54
>  
>  /* PM registers (ACPI block) */
> +#define PM1_STS		0x00
>  #define PM1_CNT		0x08
>  #define PM_GPE0_STS	0x18
>  
> +#define CS5536_PM_PWRBTN (1 << 8)
> +
> +extern void do_olpc_suspend_lowlevel(void);
> +
>  static unsigned long acpi_base;
>  static unsigned long pms_base;
>  
> +static struct {
> +	unsigned long address;
> +	unsigned short segment;
> +} ofw_bios_entry = { 0xF0000 + PAGE_OFFSET, __KERNEL_CS };
> +
> +static int xo1_power_state_enter(suspend_state_t pm_state)
> +{
> +	int r;
> +
> +	/* Only STR is supported */
> +	if (pm_state != PM_SUSPEND_MEM)
> +		return -EINVAL;
> +
> +	r = olpc_ec_cmd(EC_SET_SCI_INHIBIT, NULL, 0, NULL, 0);
> +	if (r)
> +		return r;
> +
> +	/* Save CPU state */
> +	do_olpc_suspend_lowlevel();
> +
> +	/* Resume path starts here */
> +
> +	/* Tell the EC to stop inhibiting SCIs */
> +	olpc_ec_cmd(EC_SET_SCI_INHIBIT_RELEASE, NULL, 0, NULL, 0);
> +
> +	/*
> +	 * Tell the wireless module to restart USB communication.
> +	 * Must be done twice.
> +	 */
> +	olpc_ec_cmd(EC_WAKE_UP_WLAN, NULL, 0, NULL, 0);
> +	olpc_ec_cmd(EC_WAKE_UP_WLAN, NULL, 0, NULL, 0);
> +
> +	return 0;
> +}
> +
> +asmlinkage int olpc_xo1_do_sleep(u8 sleep_state)
> +{
> +	void *pgd_addr = __va(read_cr3());
> +	printk(KERN_ERR "xo1_do_sleep!\n"); /* this needs to remain here so
> +					      * that gcc doesn't optimize
> +					      * away our __va! */
> +
> +	/* Enable wakeup through power button */
> +	outl((CS5536_PM_PWRBTN << 16) | 0xFFFF, acpi_base + PM1_STS);
> +
> +	__asm__ __volatile__("movl %0,%%eax" : : "r" (pgd_addr));
> +	__asm__("call *(%%edi); cld"
> +		: : "D" (&ofw_bios_entry));
> +	__asm__ __volatile__("movb $0x34, %al\n\t"
> +			     "outb %al, $0x70\n\t"
> +			     "movb $0x30, %al\n\t"
> +			     "outb %al, $0x71\n\t");
> +	return 0;
> +}
> +
> +static int xo1_power_state_valid(suspend_state_t pm_state)
> +{
> +	/* suspend-to-RAM only */
> +	return pm_state == PM_SUSPEND_MEM;
> +}
> +
> +static struct platform_suspend_ops xo1_suspend_ops = {
> +	.valid = xo1_power_state_valid,
> +	.enter = xo1_power_state_enter,
> +};
> +
>  static void xo1_power_off(void)
>  {
>  	printk(KERN_INFO "OLPC XO-1 power off sequence...\n");
> @@ -101,6 +173,13 @@ static int __devinit olpc_xo1_probe(struct platform_device *pdev)
>  	if (r)
>  		return r;
>  
> +	/*
> +	 * Take a reference on ourself to prevent module unloading. We can't
> +	 * safely unload after changing the suspend handlers.
> +	 */
> +	__module_get(THIS_MODULE);
> +
> +	suspend_set_ops(&xo1_suspend_ops);
>  	pm_power_off = xo1_power_off;
>  
>  	printk(KERN_INFO "OLPC XO-1 support registered\n");
> diff --git a/arch/x86/mm/init_32.c b/arch/x86/mm/init_32.c
> index 43bbc29..dbdab67 100644
> --- a/arch/x86/mm/init_32.c
> +++ b/arch/x86/mm/init_32.c
> @@ -549,7 +549,7 @@ static void __init pagetable_init(void)
>  	permanent_kmaps_init(pgd_base);
>  }
>  
> -#ifdef CONFIG_ACPI_SLEEP
> +#if defined(CONFIG_ACPI_SLEEP) || defined(CONFIG_OLPC_XO1)
>  /*
>   * ACPI suspend needs this for resume, because things like the intel-agp
>   * driver might have split up a kernel 4MB mapping.
> @@ -561,11 +561,11 @@ static inline void save_pg_dir(void)
>  {
>  	copy_page(swsusp_pg_dir, swapper_pg_dir);
>  }
> -#else /* !CONFIG_ACPI_SLEEP */
> +#else /* !CONFIG_ACPI_SLEEP && !CONFIG_OLPC_XO1 */
>  static inline void save_pg_dir(void)
>  {
>  }
> -#endif /* !CONFIG_ACPI_SLEEP */
> +#endif /* !CONFIG_ACPI_SLEEP && !CONFIG_OLPC_XO1 */
>  
>  void zap_low_mappings(bool early)
>  {
> -- 

---
~Randy
*** Remember to use Documentation/SubmitChecklist when testing your code ***

Re: [PATCH] OLPC: Add XO-1 suspend/resume support

[H. Peter Anvin]

On 10/19/2010 03:01 PM, Daniel Drake wrote:
> Add code needed for basic suspend/resume of the XO-1 laptop.
> 
> swsusp_pg_dir needs to be exposed as it is used by the assembly
> code run in the wakeup path.
> 

Okay... this is Yet Another Reason why we need to unify all the bloody
trampoline page tables.

What it comes down to is that there are several users which need a 1:1
mapped pagetable and a chunk of memory < 1 MiB, and they should all be
combined and linked together, and share a single pagetable set that
sticks around.

Borislav has already been doing some of this work:

[PATCH] x86-32, mm: Add an initial page table for core bootstrapping

I thought that patch was already in -tip, but it looks like it's not...
Borislav/Joerg... was there a newer patch or did we just miss the final
version?  I remember we talked about this at some length, but it looks
like I dropped the ball (I was on vacation when the above message was
posted.)

	-hpa

Re: [PATCH] OLPC: Add XO-1 suspend/resume support

[Borislav Petkov]

From: "H. Peter Anvin" <hpa@zytor.com>
Date: Tue, Oct 19, 2010 at 06:11:13PM -0400

> On 10/19/2010 03:01 PM, Daniel Drake wrote:
> > Add code needed for basic suspend/resume of the XO-1 laptop.
> > 
> > swsusp_pg_dir needs to be exposed as it is used by the assembly
> > code run in the wakeup path.
> > 
> 
> Okay... this is Yet Another Reason why we need to unify all the bloody
> trampoline page tables.
> 
> What it comes down to is that there are several users which need a 1:1
> mapped pagetable and a chunk of memory < 1 MiB, and they should all be
> combined and linked together, and share a single pagetable set that
> sticks around.
> 
> Borislav has already been doing some of this work:
> 
> [PATCH] x86-32, mm: Add an initial page table for core bootstrapping
> 
> I thought that patch was already in -tip, but it looks like it's not...
> Borislav/Joerg... was there a newer patch or did we just miss the final
> version?  I remember we talked about this at some length, but it looks
> like I dropped the ball (I was on vacation when the above message was
> posted.)

The final version with fixed ACPI sleep is this one:
http://www.gossamer-threads.com/lists/linux/kernel/1270589

I ran it for a couple of weeks and it didn't show any regressions. If
you want to get it into the merge window, I can get you an updated
version against a -tip branch of your liking ASAP.

-- 
Regards/Gruss,
Boris.

Advanced Micro Devices GmbH
Einsteinring 24, 85609 Dornach
General Managers: Alberto Bozzo, Andrew Bowd
Registration: Dornach, Gemeinde Aschheim, Landkreis Muenchen
Registergericht Muenchen, HRB Nr. 43632

Re: [PATCH] OLPC: Add XO-1 suspend/resume support

[H. Peter Anvin]

On 10/19/2010 10:47 PM, Borislav Petkov wrote:
> 
> The final version with fixed ACPI sleep is this one:
> http://www.gossamer-threads.com/lists/linux/kernel/1270589
> 
> I ran it for a couple of weeks and it didn't show any regressions. If
> you want to get it into the merge window, I can get you an updated
> version against a -tip branch of your liking ASAP.
> 

If you could prep something against tip:master we *might* be able to get
it in... not sure yet (and it partly depends on when Linus decides to
push the button, of course.)

	-hpa

-- 
H. Peter Anvin, Intel Open Source Technology Center
I work for Intel.  I don't speak on their behalf.

[PATCH -tip/master] x86-32, mm: Add an initial page table for core bootstrapping

[Borislav Petkov]

From: "H. Peter Anvin" <hpa@zytor.com>
Date: Tue, Oct 19, 2010 at 10:58:30PM -0700

> If you could prep something against tip:master we *might* be able to get
> it in... not sure yet (and it partly depends on when Linus decides to
> push the button, of course.)

Here you go, suspend-to-{ram,disk} works, will run it in the next couple
of days just in case.

--
From: Borislav Petkov <bp@alien8.de>
Date: Sat, 28 Aug 2010 15:58:33 +0200
Subject: [PATCH -tip/master] x86-32, mm: Add an initial page table for core bootstrapping

This patch adds an initial page table with low mappings used exclusively
for booting APs/resuming after ACPI suspend/machine restart. After this,
there's no need to add low mappings to swapper_pg_dir and zap them later
or create own swsusp PGD page solely for ACPI sleep needs - we have
initial_page_table for that.

[ v1.1 Fix suspend-to-ram ]

Signed-off-by: Borislav Petkov <bp@alien8.de>
---
 arch/x86/include/asm/pgtable_32.h |    2 +-
 arch/x86/include/asm/tlbflush.h   |    2 -
 arch/x86/include/asm/trampoline.h |    3 --
 arch/x86/kernel/acpi/sleep.c      |    7 ++++-
 arch/x86/kernel/head32.c          |    1 +
 arch/x86/kernel/head_32.S         |   55 +++++++++++++++++--------------------
 arch/x86/kernel/reboot.c          |   10 +-----
 arch/x86/kernel/setup.c           |   18 +++++++++++-
 arch/x86/kernel/smpboot.c         |   16 +++--------
 arch/x86/kernel/trampoline.c      |   16 -----------
 arch/x86/mm/init_32.c             |   45 ------------------------------
 11 files changed, 56 insertions(+), 119 deletions(-)

diff --git a/arch/x86/include/asm/pgtable_32.h b/arch/x86/include/asm/pgtable_32.h
index f686f49..8abde9e 100644
--- a/arch/x86/include/asm/pgtable_32.h
+++ b/arch/x86/include/asm/pgtable_32.h
@@ -26,7 +26,7 @@ struct mm_struct;
 struct vm_area_struct;
 
 extern pgd_t swapper_pg_dir[1024];
-extern pgd_t trampoline_pg_dir[1024];
+extern pgd_t initial_page_table[1024];
 
 static inline void pgtable_cache_init(void) { }
 static inline void check_pgt_cache(void) { }
diff --git a/arch/x86/include/asm/tlbflush.h b/arch/x86/include/asm/tlbflush.h
index 7f3eba0..169be89 100644
--- a/arch/x86/include/asm/tlbflush.h
+++ b/arch/x86/include/asm/tlbflush.h
@@ -172,6 +172,4 @@ static inline void flush_tlb_kernel_range(unsigned long start,
 	flush_tlb_all();
 }
 
-extern void zap_low_mappings(bool early);
-
 #endif /* _ASM_X86_TLBFLUSH_H */
diff --git a/arch/x86/include/asm/trampoline.h b/arch/x86/include/asm/trampoline.h
index 4dde797..f4500fb 100644
--- a/arch/x86/include/asm/trampoline.h
+++ b/arch/x86/include/asm/trampoline.h
@@ -13,16 +13,13 @@ extern unsigned char *trampoline_base;
 
 extern unsigned long init_rsp;
 extern unsigned long initial_code;
-extern unsigned long initial_page_table;
 extern unsigned long initial_gs;
 
 #define TRAMPOLINE_SIZE roundup(trampoline_end - trampoline_data, PAGE_SIZE)
 
 extern unsigned long setup_trampoline(void);
-extern void __init setup_trampoline_page_table(void);
 extern void __init reserve_trampoline_memory(void);
 #else
-static inline void setup_trampoline_page_table(void) {}
 static inline void reserve_trampoline_memory(void) {}
 #endif /* CONFIG_X86_TRAMPOLINE */
 
diff --git a/arch/x86/kernel/acpi/sleep.c b/arch/x86/kernel/acpi/sleep.c
index e125207..74a8478 100644
--- a/arch/x86/kernel/acpi/sleep.c
+++ b/arch/x86/kernel/acpi/sleep.c
@@ -13,6 +13,11 @@
 #include <asm/segment.h>
 #include <asm/desc.h>
 
+#ifdef CONFIG_X86_32
+#include <asm/pgtable.h>
+#include <asm/pgtable_32.h>
+#endif
+
 #include "realmode/wakeup.h"
 #include "sleep.h"
 
@@ -91,7 +96,7 @@ int acpi_save_state_mem(void)
 
 #ifndef CONFIG_64BIT
 	header->pmode_entry = (u32)&wakeup_pmode_return;
-	header->pmode_cr3 = (u32)(swsusp_pg_dir - __PAGE_OFFSET);
+	header->pmode_cr3 = (u32)__pa(&initial_page_table);
 	saved_magic = 0x12345678;
 #else /* CONFIG_64BIT */
 	header->trampoline_segment = setup_trampoline() >> 4;
diff --git a/arch/x86/kernel/head32.c b/arch/x86/kernel/head32.c
index 9a6ca23..7633101 100644
--- a/arch/x86/kernel/head32.c
+++ b/arch/x86/kernel/head32.c
@@ -18,6 +18,7 @@
 #include <asm/apic.h>
 #include <asm/io_apic.h>
 #include <asm/bios_ebda.h>
+#include <asm/tlbflush.h>
 
 static void __init i386_default_early_setup(void)
 {
diff --git a/arch/x86/kernel/head_32.S b/arch/x86/kernel/head_32.S
index fa8c1b8..bcece91 100644
--- a/arch/x86/kernel/head_32.S
+++ b/arch/x86/kernel/head_32.S
@@ -183,13 +183,12 @@ default_entry:
 #ifdef CONFIG_X86_PAE
 
 	/*
-	 * In PAE mode swapper_pg_dir is statically defined to contain enough
-	 * entries to cover the VMSPLIT option (that is the top 1, 2 or 3
-	 * entries). The identity mapping is handled by pointing two PGD
-	 * entries to the first kernel PMD.
+	 * In PAE mode initial_page_table is statically defined to contain
+	 * enough entries to cover the VMSPLIT option (that is the top 1, 2 or 3
+	 * entries). The identity mapping is handled by pointing two PGD entries
+	 * to the first kernel PMD.
 	 *
-	 * Note the upper half of each PMD or PTE are always zero at
-	 * this stage.
+	 * Note the upper half of each PMD or PTE are always zero at this stage.
 	 */
 
 #define KPMDS (((-__PAGE_OFFSET) >> 30) & 3) /* Number of kernel PMDs */
@@ -197,7 +196,7 @@ default_entry:
 	xorl %ebx,%ebx				/* %ebx is kept at zero */
 
 	movl $pa(__brk_base), %edi
-	movl $pa(swapper_pg_pmd), %edx
+	movl $pa(initial_pg_pmd), %edx
 	movl $PTE_IDENT_ATTR, %eax
 10:
 	leal PDE_IDENT_ATTR(%edi),%ecx		/* Create PMD entry */
@@ -226,14 +225,14 @@ default_entry:
 	movl %eax, pa(max_pfn_mapped)
 
 	/* Do early initialization of the fixmap area */
-	movl $pa(swapper_pg_fixmap)+PDE_IDENT_ATTR,%eax
-	movl %eax,pa(swapper_pg_pmd+0x1000*KPMDS-8)
+	movl $pa(initial_pg_fixmap)+PDE_IDENT_ATTR,%eax
+	movl %eax,pa(initial_pg_pmd+0x1000*KPMDS-8)
 #else	/* Not PAE */
 
 page_pde_offset = (__PAGE_OFFSET >> 20);
 
 	movl $pa(__brk_base), %edi
-	movl $pa(swapper_pg_dir), %edx
+	movl $pa(initial_page_table), %edx
 	movl $PTE_IDENT_ATTR, %eax
 10:
 	leal PDE_IDENT_ATTR(%edi),%ecx		/* Create PDE entry */
@@ -257,8 +256,8 @@ page_pde_offset = (__PAGE_OFFSET >> 20);
 	movl %eax, pa(max_pfn_mapped)
 
 	/* Do early initialization of the fixmap area */
-	movl $pa(swapper_pg_fixmap)+PDE_IDENT_ATTR,%eax
-	movl %eax,pa(swapper_pg_dir+0xffc)
+	movl $pa(initial_pg_fixmap)+PDE_IDENT_ATTR,%eax
+	movl %eax,pa(initial_page_table+0xffc)
 #endif
 	jmp 3f
 /*
@@ -334,7 +333,7 @@ ENTRY(startup_32_smp)
 /*
  * Enable paging
  */
-	movl pa(initial_page_table), %eax
+	movl $pa(initial_page_table), %eax
 	movl %eax,%cr3		/* set the page table pointer.. */
 	movl %cr0,%eax
 	orl  $X86_CR0_PG,%eax
@@ -614,8 +613,6 @@ ignore_int:
 .align 4
 ENTRY(initial_code)
 	.long i386_start_kernel
-ENTRY(initial_page_table)
-	.long pa(swapper_pg_dir)
 
 /*
  * BSS section
@@ -623,20 +620,18 @@ ENTRY(initial_page_table)
 __PAGE_ALIGNED_BSS
 	.align PAGE_SIZE_asm
 #ifdef CONFIG_X86_PAE
-swapper_pg_pmd:
+initial_pg_pmd:
 	.fill 1024*KPMDS,4,0
 #else
-ENTRY(swapper_pg_dir)
+ENTRY(initial_page_table)
 	.fill 1024,4,0
 #endif
-swapper_pg_fixmap:
+initial_pg_fixmap:
 	.fill 1024,4,0
-#ifdef CONFIG_X86_TRAMPOLINE
-ENTRY(trampoline_pg_dir)
-	.fill 1024,4,0
-#endif
 ENTRY(empty_zero_page)
 	.fill 4096,1,0
+ENTRY(swapper_pg_dir)
+	.fill 1024,4,0
 
 /*
  * This starts the data section.
@@ -645,20 +640,20 @@ ENTRY(empty_zero_page)
 __PAGE_ALIGNED_DATA
 	/* Page-aligned for the benefit of paravirt? */
 	.align PAGE_SIZE_asm
-ENTRY(swapper_pg_dir)
-	.long	pa(swapper_pg_pmd+PGD_IDENT_ATTR),0	/* low identity map */
+ENTRY(initial_page_table)
+	.long	pa(initial_pg_pmd+PGD_IDENT_ATTR),0	/* low identity map */
 # if KPMDS == 3
-	.long	pa(swapper_pg_pmd+PGD_IDENT_ATTR),0
-	.long	pa(swapper_pg_pmd+PGD_IDENT_ATTR+0x1000),0
-	.long	pa(swapper_pg_pmd+PGD_IDENT_ATTR+0x2000),0
+	.long	pa(initial_pg_pmd+PGD_IDENT_ATTR),0
+	.long	pa(initial_pg_pmd+PGD_IDENT_ATTR+0x1000),0
+	.long	pa(initial_pg_pmd+PGD_IDENT_ATTR+0x2000),0
 # elif KPMDS == 2
 	.long	0,0
-	.long	pa(swapper_pg_pmd+PGD_IDENT_ATTR),0
-	.long	pa(swapper_pg_pmd+PGD_IDENT_ATTR+0x1000),0
+	.long	pa(initial_pg_pmd+PGD_IDENT_ATTR),0
+	.long	pa(initial_pg_pmd+PGD_IDENT_ATTR+0x1000),0
 # elif KPMDS == 1
 	.long	0,0
 	.long	0,0
-	.long	pa(swapper_pg_pmd+PGD_IDENT_ATTR),0
+	.long	pa(initial_pg_pmd+PGD_IDENT_ATTR),0
 # else
 #  error "Kernel PMDs should be 1, 2 or 3"
 # endif
diff --git a/arch/x86/kernel/reboot.c b/arch/x86/kernel/reboot.c
index 7a4cf14..f7f53dc 100644
--- a/arch/x86/kernel/reboot.c
+++ b/arch/x86/kernel/reboot.c
@@ -371,16 +371,10 @@ void machine_real_restart(const unsigned char *code, int length)
 	CMOS_WRITE(0x00, 0x8f);
 	spin_unlock(&rtc_lock);
 
-	/* Remap the kernel at virtual address zero, as well as offset zero
-	   from the kernel segment.  This assumes the kernel segment starts at
-	   virtual address PAGE_OFFSET. */
-	memcpy(swapper_pg_dir, swapper_pg_dir + KERNEL_PGD_BOUNDARY,
-		sizeof(swapper_pg_dir [0]) * KERNEL_PGD_PTRS);
-
 	/*
-	 * Use `swapper_pg_dir' as our page directory.
+	 * Switch back to the initial page table.
 	 */
-	load_cr3(swapper_pg_dir);
+	load_cr3(initial_page_table);
 
 	/* Write 0x1234 to absolute memory location 0x472.  The BIOS reads
 	   this on booting to tell it to "Bypass memory test (also warm
diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c
index 44e208e..3df995f 100644
--- a/arch/x86/kernel/setup.c
+++ b/arch/x86/kernel/setup.c
@@ -700,6 +700,17 @@ void __init setup_arch(char **cmdline_p)
 #ifdef CONFIG_X86_32
 	memcpy(&boot_cpu_data, &new_cpu_data, sizeof(new_cpu_data));
 	visws_early_detect();
+
+	/*
+	 * copy kernel address range established so far and switch
+	 * to the proper swapper page table
+	 */
+	clone_pgd_range(swapper_pg_dir     + KERNEL_PGD_BOUNDARY,
+			initial_page_table + KERNEL_PGD_BOUNDARY,
+			KERNEL_PGD_PTRS);
+
+	load_cr3(swapper_pg_dir);
+	__flush_tlb_all();
 #else
 	printk(KERN_INFO "Command line: %s\n", boot_command_line);
 #endif
@@ -985,7 +996,12 @@ void __init setup_arch(char **cmdline_p)
 	paging_init();
 	x86_init.paging.pagetable_setup_done(swapper_pg_dir);
 
-	setup_trampoline_page_table();
+#ifdef CONFIG_X86_32
+	/* sync back kernel address range */
+	clone_pgd_range(initial_page_table + KERNEL_PGD_BOUNDARY,
+			swapper_pg_dir     + KERNEL_PGD_BOUNDARY,
+			KERNEL_PGD_PTRS);
+#endif
 
 	tboot_probe();
 
diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c
index dfb5089..6af1185 100644
--- a/arch/x86/kernel/smpboot.c
+++ b/arch/x86/kernel/smpboot.c
@@ -299,22 +299,16 @@ notrace static void __cpuinit start_secondary(void *unused)
 	 * fragile that we want to limit the things done here to the
 	 * most necessary things.
 	 */
+	cpu_init();
+	preempt_disable();
+	smp_callin();
 
 #ifdef CONFIG_X86_32
-	/*
-	 * Switch away from the trampoline page-table
-	 *
-	 * Do this before cpu_init() because it needs to access per-cpu
-	 * data which may not be mapped in the trampoline page-table.
-	 */
+	/* switch away from the initial page table */
 	load_cr3(swapper_pg_dir);
 	__flush_tlb_all();
 #endif
 
-	cpu_init();
-	preempt_disable();
-	smp_callin();
-
 	/* otherwise gcc will move up smp_processor_id before the cpu_init */
 	barrier();
 	/*
@@ -785,7 +779,6 @@ do_rest:
 #ifdef CONFIG_X86_32
 	/* Stack for startup_32 can be just as for start_secondary onwards */
 	irq_ctx_init(cpu);
-	initial_page_table = __pa(&trampoline_pg_dir);
 #else
 	clear_tsk_thread_flag(c_idle.idle, TIF_FORK);
 	initial_gs = per_cpu_offset(cpu);
@@ -934,7 +927,6 @@ int __cpuinit native_cpu_up(unsigned int cpu)
 	per_cpu(cpu_state, cpu) = CPU_UP_PREPARE;
 
 	err = do_boot_cpu(apicid, cpu);
-
 	if (err) {
 		pr_debug("do_boot_cpu failed %d\n", err);
 		return -EIO;
diff --git a/arch/x86/kernel/trampoline.c b/arch/x86/kernel/trampoline.c
index 4c3da56..a375616 100644
--- a/arch/x86/kernel/trampoline.c
+++ b/arch/x86/kernel/trampoline.c
@@ -38,19 +38,3 @@ unsigned long __trampinit setup_trampoline(void)
 	memcpy(trampoline_base, trampoline_data, TRAMPOLINE_SIZE);
 	return virt_to_phys(trampoline_base);
 }
-
-void __init setup_trampoline_page_table(void)
-{
-#ifdef CONFIG_X86_32
-	/* Copy kernel address range */
-	clone_pgd_range(trampoline_pg_dir + KERNEL_PGD_BOUNDARY,
-			swapper_pg_dir + KERNEL_PGD_BOUNDARY,
-			KERNEL_PGD_PTRS);
-
-	/* Initialize low mappings */
-	clone_pgd_range(trampoline_pg_dir,
-			swapper_pg_dir + KERNEL_PGD_BOUNDARY,
-			min_t(unsigned long, KERNEL_PGD_PTRS,
-			      KERNEL_PGD_BOUNDARY));
-#endif
-}
diff --git a/arch/x86/mm/init_32.c b/arch/x86/mm/init_32.c
index 5d0a671..0e969f9 100644
--- a/arch/x86/mm/init_32.c
+++ b/arch/x86/mm/init_32.c
@@ -528,48 +528,6 @@ static void __init pagetable_init(void)
 	permanent_kmaps_init(pgd_base);
 }
 
-#ifdef CONFIG_ACPI_SLEEP
-/*
- * ACPI suspend needs this for resume, because things like the intel-agp
- * driver might have split up a kernel 4MB mapping.
- */
-char swsusp_pg_dir[PAGE_SIZE]
-	__attribute__ ((aligned(PAGE_SIZE)));
-
-static inline void save_pg_dir(void)
-{
-	copy_page(swsusp_pg_dir, swapper_pg_dir);
-}
-#else /* !CONFIG_ACPI_SLEEP */
-static inline void save_pg_dir(void)
-{
-}
-#endif /* !CONFIG_ACPI_SLEEP */
-
-void zap_low_mappings(bool early)
-{
-	int i;
-
-	/*
-	 * Zap initial low-memory mappings.
-	 *
-	 * Note that "pgd_clear()" doesn't do it for
-	 * us, because pgd_clear() is a no-op on i386.
-	 */
-	for (i = 0; i < KERNEL_PGD_BOUNDARY; i++) {
-#ifdef CONFIG_X86_PAE
-		set_pgd(swapper_pg_dir+i, __pgd(1 + __pa(empty_zero_page)));
-#else
-		set_pgd(swapper_pg_dir+i, __pgd(0));
-#endif
-	}
-
-	if (early)
-		__flush_tlb();
-	else
-		flush_tlb_all();
-}
-
 pteval_t __supported_pte_mask __read_mostly = ~(_PAGE_NX | _PAGE_GLOBAL | _PAGE_IOMAP);
 EXPORT_SYMBOL_GPL(__supported_pte_mask);
 
@@ -882,9 +840,6 @@ void __init mem_init(void)
 
 	if (boot_cpu_data.wp_works_ok < 0)
 		test_wp_bit();
-
-	save_pg_dir();
-	zap_low_mappings(true);
 }
 
 #ifdef CONFIG_MEMORY_HOTPLUG
-- 
1.7.2.3


-- 
Regards/Gruss,
    Boris.

[tip:x86/trampoline] x86-32, mm: Add an initial page table for core bootstrapping

[tip-bot for Borislav Petkov]

Commit-ID:  b40827fa7268fda8a62490728a61c2856f33830b
Gitweb:     http://git.kernel.org/tip/b40827fa7268fda8a62490728a61c2856f33830b
Author:     Borislav Petkov <bp@alien8.de>
AuthorDate: Sat, 28 Aug 2010 15:58:33 +0200
Committer:  H. Peter Anvin <hpa@linux.intel.com>
CommitDate: Wed, 20 Oct 2010 14:23:55 -0700

x86-32, mm: Add an initial page table for core bootstrapping

This patch adds an initial page table with low mappings used exclusively
for booting APs/resuming after ACPI suspend/machine restart. After this,
there's no need to add low mappings to swapper_pg_dir and zap them later
or create own swsusp PGD page solely for ACPI sleep needs - we have
initial_page_table for that.

Signed-off-by: Borislav Petkov <bp@alien8.de>
LKML-Reference: <20101020070526.GA9588@liondog.tnic>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
---
 arch/x86/include/asm/pgtable_32.h |    2 +-
 arch/x86/include/asm/tlbflush.h   |    2 -
 arch/x86/include/asm/trampoline.h |    3 --
 arch/x86/kernel/acpi/sleep.c      |    7 ++++-
 arch/x86/kernel/head32.c          |    1 +
 arch/x86/kernel/head_32.S         |   55 +++++++++++++++++--------------------
 arch/x86/kernel/reboot.c          |   10 +-----
 arch/x86/kernel/setup.c           |   18 +++++++++++-
 arch/x86/kernel/smpboot.c         |   16 +++--------
 arch/x86/kernel/trampoline.c      |   16 -----------
 arch/x86/mm/init_32.c             |   45 ------------------------------
 11 files changed, 56 insertions(+), 119 deletions(-)

diff --git a/arch/x86/include/asm/pgtable_32.h b/arch/x86/include/asm/pgtable_32.h
index f686f49..8abde9e 100644
--- a/arch/x86/include/asm/pgtable_32.h
+++ b/arch/x86/include/asm/pgtable_32.h
@@ -26,7 +26,7 @@ struct mm_struct;
 struct vm_area_struct;
 
 extern pgd_t swapper_pg_dir[1024];
-extern pgd_t trampoline_pg_dir[1024];
+extern pgd_t initial_page_table[1024];
 
 static inline void pgtable_cache_init(void) { }
 static inline void check_pgt_cache(void) { }
diff --git a/arch/x86/include/asm/tlbflush.h b/arch/x86/include/asm/tlbflush.h
index 7f3eba0..169be89 100644
--- a/arch/x86/include/asm/tlbflush.h
+++ b/arch/x86/include/asm/tlbflush.h
@@ -172,6 +172,4 @@ static inline void flush_tlb_kernel_range(unsigned long start,
 	flush_tlb_all();
 }
 
-extern void zap_low_mappings(bool early);
-
 #endif /* _ASM_X86_TLBFLUSH_H */
diff --git a/arch/x86/include/asm/trampoline.h b/arch/x86/include/asm/trampoline.h
index 4dde797..f4500fb 100644
--- a/arch/x86/include/asm/trampoline.h
+++ b/arch/x86/include/asm/trampoline.h
@@ -13,16 +13,13 @@ extern unsigned char *trampoline_base;
 
 extern unsigned long init_rsp;
 extern unsigned long initial_code;
-extern unsigned long initial_page_table;
 extern unsigned long initial_gs;
 
 #define TRAMPOLINE_SIZE roundup(trampoline_end - trampoline_data, PAGE_SIZE)
 
 extern unsigned long setup_trampoline(void);
-extern void __init setup_trampoline_page_table(void);
 extern void __init reserve_trampoline_memory(void);
 #else
-static inline void setup_trampoline_page_table(void) {}
 static inline void reserve_trampoline_memory(void) {}
 #endif /* CONFIG_X86_TRAMPOLINE */
 
diff --git a/arch/x86/kernel/acpi/sleep.c b/arch/x86/kernel/acpi/sleep.c
index 33cec15..b35e1ab 100644
--- a/arch/x86/kernel/acpi/sleep.c
+++ b/arch/x86/kernel/acpi/sleep.c
@@ -12,6 +12,11 @@
 #include <asm/segment.h>
 #include <asm/desc.h>
 
+#ifdef CONFIG_X86_32
+#include <asm/pgtable.h>
+#include <asm/pgtable_32.h>
+#endif
+
 #include "realmode/wakeup.h"
 #include "sleep.h"
 
@@ -90,7 +95,7 @@ int acpi_save_state_mem(void)
 
 #ifndef CONFIG_64BIT
 	header->pmode_entry = (u32)&wakeup_pmode_return;
-	header->pmode_cr3 = (u32)(swsusp_pg_dir - __PAGE_OFFSET);
+	header->pmode_cr3 = (u32)__pa(&initial_page_table);
 	saved_magic = 0x12345678;
 #else /* CONFIG_64BIT */
 	header->trampoline_segment = setup_trampoline() >> 4;
diff --git a/arch/x86/kernel/head32.c b/arch/x86/kernel/head32.c
index 784360c..8b9c201 100644
--- a/arch/x86/kernel/head32.c
+++ b/arch/x86/kernel/head32.c
@@ -17,6 +17,7 @@
 #include <asm/apic.h>
 #include <asm/io_apic.h>
 #include <asm/bios_ebda.h>
+#include <asm/tlbflush.h>
 
 static void __init i386_default_early_setup(void)
 {
diff --git a/arch/x86/kernel/head_32.S b/arch/x86/kernel/head_32.S
index fa8c1b8..bcece91 100644
--- a/arch/x86/kernel/head_32.S
+++ b/arch/x86/kernel/head_32.S
@@ -183,13 +183,12 @@ default_entry:
 #ifdef CONFIG_X86_PAE
 
 	/*
-	 * In PAE mode swapper_pg_dir is statically defined to contain enough
-	 * entries to cover the VMSPLIT option (that is the top 1, 2 or 3
-	 * entries). The identity mapping is handled by pointing two PGD
-	 * entries to the first kernel PMD.
+	 * In PAE mode initial_page_table is statically defined to contain
+	 * enough entries to cover the VMSPLIT option (that is the top 1, 2 or 3
+	 * entries). The identity mapping is handled by pointing two PGD entries
+	 * to the first kernel PMD.
 	 *
-	 * Note the upper half of each PMD or PTE are always zero at
-	 * this stage.
+	 * Note the upper half of each PMD or PTE are always zero at this stage.
 	 */
 
 #define KPMDS (((-__PAGE_OFFSET) >> 30) & 3) /* Number of kernel PMDs */
@@ -197,7 +196,7 @@ default_entry:
 	xorl %ebx,%ebx				/* %ebx is kept at zero */
 
 	movl $pa(__brk_base), %edi
-	movl $pa(swapper_pg_pmd), %edx
+	movl $pa(initial_pg_pmd), %edx
 	movl $PTE_IDENT_ATTR, %eax
 10:
 	leal PDE_IDENT_ATTR(%edi),%ecx		/* Create PMD entry */
@@ -226,14 +225,14 @@ default_entry:
 	movl %eax, pa(max_pfn_mapped)
 
 	/* Do early initialization of the fixmap area */
-	movl $pa(swapper_pg_fixmap)+PDE_IDENT_ATTR,%eax
-	movl %eax,pa(swapper_pg_pmd+0x1000*KPMDS-8)
+	movl $pa(initial_pg_fixmap)+PDE_IDENT_ATTR,%eax
+	movl %eax,pa(initial_pg_pmd+0x1000*KPMDS-8)
 #else	/* Not PAE */
 
 page_pde_offset = (__PAGE_OFFSET >> 20);
 
 	movl $pa(__brk_base), %edi
-	movl $pa(swapper_pg_dir), %edx
+	movl $pa(initial_page_table), %edx
 	movl $PTE_IDENT_ATTR, %eax
 10:
 	leal PDE_IDENT_ATTR(%edi),%ecx		/* Create PDE entry */
@@ -257,8 +256,8 @@ page_pde_offset = (__PAGE_OFFSET >> 20);
 	movl %eax, pa(max_pfn_mapped)
 
 	/* Do early initialization of the fixmap area */
-	movl $pa(swapper_pg_fixmap)+PDE_IDENT_ATTR,%eax
-	movl %eax,pa(swapper_pg_dir+0xffc)
+	movl $pa(initial_pg_fixmap)+PDE_IDENT_ATTR,%eax
+	movl %eax,pa(initial_page_table+0xffc)
 #endif
 	jmp 3f
 /*
@@ -334,7 +333,7 @@ ENTRY(startup_32_smp)
 /*
  * Enable paging
  */
-	movl pa(initial_page_table), %eax
+	movl $pa(initial_page_table), %eax
 	movl %eax,%cr3		/* set the page table pointer.. */
 	movl %cr0,%eax
 	orl  $X86_CR0_PG,%eax
@@ -614,8 +613,6 @@ ignore_int:
 .align 4
 ENTRY(initial_code)
 	.long i386_start_kernel
-ENTRY(initial_page_table)
-	.long pa(swapper_pg_dir)
 
 /*
  * BSS section
@@ -623,20 +620,18 @@ ENTRY(initial_page_table)
 __PAGE_ALIGNED_BSS
 	.align PAGE_SIZE_asm
 #ifdef CONFIG_X86_PAE
-swapper_pg_pmd:
+initial_pg_pmd:
 	.fill 1024*KPMDS,4,0
 #else
-ENTRY(swapper_pg_dir)
+ENTRY(initial_page_table)
 	.fill 1024,4,0
 #endif
-swapper_pg_fixmap:
+initial_pg_fixmap:
 	.fill 1024,4,0
-#ifdef CONFIG_X86_TRAMPOLINE
-ENTRY(trampoline_pg_dir)
-	.fill 1024,4,0
-#endif
 ENTRY(empty_zero_page)
 	.fill 4096,1,0
+ENTRY(swapper_pg_dir)
+	.fill 1024,4,0
 
 /*
  * This starts the data section.
@@ -645,20 +640,20 @@ ENTRY(empty_zero_page)
 __PAGE_ALIGNED_DATA
 	/* Page-aligned for the benefit of paravirt? */
 	.align PAGE_SIZE_asm
-ENTRY(swapper_pg_dir)
-	.long	pa(swapper_pg_pmd+PGD_IDENT_ATTR),0	/* low identity map */
+ENTRY(initial_page_table)
+	.long	pa(initial_pg_pmd+PGD_IDENT_ATTR),0	/* low identity map */
 # if KPMDS == 3
-	.long	pa(swapper_pg_pmd+PGD_IDENT_ATTR),0
-	.long	pa(swapper_pg_pmd+PGD_IDENT_ATTR+0x1000),0
-	.long	pa(swapper_pg_pmd+PGD_IDENT_ATTR+0x2000),0
+	.long	pa(initial_pg_pmd+PGD_IDENT_ATTR),0
+	.long	pa(initial_pg_pmd+PGD_IDENT_ATTR+0x1000),0
+	.long	pa(initial_pg_pmd+PGD_IDENT_ATTR+0x2000),0
 # elif KPMDS == 2
 	.long	0,0
-	.long	pa(swapper_pg_pmd+PGD_IDENT_ATTR),0
-	.long	pa(swapper_pg_pmd+PGD_IDENT_ATTR+0x1000),0
+	.long	pa(initial_pg_pmd+PGD_IDENT_ATTR),0
+	.long	pa(initial_pg_pmd+PGD_IDENT_ATTR+0x1000),0
 # elif KPMDS == 1
 	.long	0,0
 	.long	0,0
-	.long	pa(swapper_pg_pmd+PGD_IDENT_ATTR),0
+	.long	pa(initial_pg_pmd+PGD_IDENT_ATTR),0
 # else
 #  error "Kernel PMDs should be 1, 2 or 3"
 # endif
diff --git a/arch/x86/kernel/reboot.c b/arch/x86/kernel/reboot.c
index 7a4cf14..f7f53dc 100644
--- a/arch/x86/kernel/reboot.c
+++ b/arch/x86/kernel/reboot.c
@@ -371,16 +371,10 @@ void machine_real_restart(const unsigned char *code, int length)
 	CMOS_WRITE(0x00, 0x8f);
 	spin_unlock(&rtc_lock);
 
-	/* Remap the kernel at virtual address zero, as well as offset zero
-	   from the kernel segment.  This assumes the kernel segment starts at
-	   virtual address PAGE_OFFSET. */
-	memcpy(swapper_pg_dir, swapper_pg_dir + KERNEL_PGD_BOUNDARY,
-		sizeof(swapper_pg_dir [0]) * KERNEL_PGD_PTRS);
-
 	/*
-	 * Use `swapper_pg_dir' as our page directory.
+	 * Switch back to the initial page table.
 	 */
-	load_cr3(swapper_pg_dir);
+	load_cr3(initial_page_table);
 
 	/* Write 0x1234 to absolute memory location 0x472.  The BIOS reads
 	   this on booting to tell it to "Bypass memory test (also warm
diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c
index 322b24f..af6cf2b 100644
--- a/arch/x86/kernel/setup.c
+++ b/arch/x86/kernel/setup.c
@@ -728,6 +728,17 @@ void __init setup_arch(char **cmdline_p)
 #ifdef CONFIG_X86_32
 	memcpy(&boot_cpu_data, &new_cpu_data, sizeof(new_cpu_data));
 	visws_early_detect();
+
+	/*
+	 * copy kernel address range established so far and switch
+	 * to the proper swapper page table
+	 */
+	clone_pgd_range(swapper_pg_dir     + KERNEL_PGD_BOUNDARY,
+			initial_page_table + KERNEL_PGD_BOUNDARY,
+			KERNEL_PGD_PTRS);
+
+	load_cr3(swapper_pg_dir);
+	__flush_tlb_all();
 #else
 	printk(KERN_INFO "Command line: %s\n", boot_command_line);
 #endif
@@ -1009,7 +1020,12 @@ void __init setup_arch(char **cmdline_p)
 	paging_init();
 	x86_init.paging.pagetable_setup_done(swapper_pg_dir);
 
-	setup_trampoline_page_table();
+#ifdef CONFIG_X86_32
+	/* sync back kernel address range */
+	clone_pgd_range(initial_page_table + KERNEL_PGD_BOUNDARY,
+			swapper_pg_dir     + KERNEL_PGD_BOUNDARY,
+			KERNEL_PGD_PTRS);
+#endif
 
 	tboot_probe();
 
diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c
index 63a1a55..e63bb51 100644
--- a/arch/x86/kernel/smpboot.c
+++ b/arch/x86/kernel/smpboot.c
@@ -298,22 +298,16 @@ notrace static void __cpuinit start_secondary(void *unused)
 	 * fragile that we want to limit the things done here to the
 	 * most necessary things.
 	 */
+	cpu_init();
+	preempt_disable();
+	smp_callin();
 
 #ifdef CONFIG_X86_32
-	/*
-	 * Switch away from the trampoline page-table
-	 *
-	 * Do this before cpu_init() because it needs to access per-cpu
-	 * data which may not be mapped in the trampoline page-table.
-	 */
+	/* switch away from the initial page table */
 	load_cr3(swapper_pg_dir);
 	__flush_tlb_all();
 #endif
 
-	cpu_init();
-	preempt_disable();
-	smp_callin();
-
 	/* otherwise gcc will move up smp_processor_id before the cpu_init */
 	barrier();
 	/*
@@ -772,7 +766,6 @@ do_rest:
 #ifdef CONFIG_X86_32
 	/* Stack for startup_32 can be just as for start_secondary onwards */
 	irq_ctx_init(cpu);
-	initial_page_table = __pa(&trampoline_pg_dir);
 #else
 	clear_tsk_thread_flag(c_idle.idle, TIF_FORK);
 	initial_gs = per_cpu_offset(cpu);
@@ -921,7 +914,6 @@ int __cpuinit native_cpu_up(unsigned int cpu)
 	per_cpu(cpu_state, cpu) = CPU_UP_PREPARE;
 
 	err = do_boot_cpu(apicid, cpu);
-
 	if (err) {
 		pr_debug("do_boot_cpu failed %d\n", err);
 		return -EIO;
diff --git a/arch/x86/kernel/trampoline.c b/arch/x86/kernel/trampoline.c
index e2a5952..f1488a3 100644
--- a/arch/x86/kernel/trampoline.c
+++ b/arch/x86/kernel/trampoline.c
@@ -38,19 +38,3 @@ unsigned long __trampinit setup_trampoline(void)
 	memcpy(trampoline_base, trampoline_data, TRAMPOLINE_SIZE);
 	return virt_to_phys(trampoline_base);
 }
-
-void __init setup_trampoline_page_table(void)
-{
-#ifdef CONFIG_X86_32
-	/* Copy kernel address range */
-	clone_pgd_range(trampoline_pg_dir + KERNEL_PGD_BOUNDARY,
-			swapper_pg_dir + KERNEL_PGD_BOUNDARY,
-			KERNEL_PGD_PTRS);
-
-	/* Initialize low mappings */
-	clone_pgd_range(trampoline_pg_dir,
-			swapper_pg_dir + KERNEL_PGD_BOUNDARY,
-			min_t(unsigned long, KERNEL_PGD_PTRS,
-			      KERNEL_PGD_BOUNDARY));
-#endif
-}
diff --git a/arch/x86/mm/init_32.c b/arch/x86/mm/init_32.c
index 558f2d3..1aeac2d 100644
--- a/arch/x86/mm/init_32.c
+++ b/arch/x86/mm/init_32.c
@@ -548,48 +548,6 @@ static void __init pagetable_init(void)
 	permanent_kmaps_init(pgd_base);
 }
 
-#ifdef CONFIG_ACPI_SLEEP
-/*
- * ACPI suspend needs this for resume, because things like the intel-agp
- * driver might have split up a kernel 4MB mapping.
- */
-char swsusp_pg_dir[PAGE_SIZE]
-	__attribute__ ((aligned(PAGE_SIZE)));
-
-static inline void save_pg_dir(void)
-{
-	copy_page(swsusp_pg_dir, swapper_pg_dir);
-}
-#else /* !CONFIG_ACPI_SLEEP */
-static inline void save_pg_dir(void)
-{
-}
-#endif /* !CONFIG_ACPI_SLEEP */
-
-void zap_low_mappings(bool early)
-{
-	int i;
-
-	/*
-	 * Zap initial low-memory mappings.
-	 *
-	 * Note that "pgd_clear()" doesn't do it for
-	 * us, because pgd_clear() is a no-op on i386.
-	 */
-	for (i = 0; i < KERNEL_PGD_BOUNDARY; i++) {
-#ifdef CONFIG_X86_PAE
-		set_pgd(swapper_pg_dir+i, __pgd(1 + __pa(empty_zero_page)));
-#else
-		set_pgd(swapper_pg_dir+i, __pgd(0));
-#endif
-	}
-
-	if (early)
-		__flush_tlb();
-	else
-		flush_tlb_all();
-}
-
 pteval_t __supported_pte_mask __read_mostly = ~(_PAGE_NX | _PAGE_GLOBAL | _PAGE_IOMAP);
 EXPORT_SYMBOL_GPL(__supported_pte_mask);
 
@@ -958,9 +916,6 @@ void __init mem_init(void)
 
 	if (boot_cpu_data.wp_works_ok < 0)
 		test_wp_bit();
-
-	save_pg_dir();
-	zap_low_mappings(true);
 }
 
 #ifdef CONFIG_MEMORY_HOTPLUG

[PATCH] OLPC: Add XO-1 suspend/resume support

[Daniel Drake]

Add code needed for basic suspend/resume of the XO-1 laptop.

Signed-off-by: Daniel Drake <dsd@laptop.org>
---
 arch/x86/Kconfig                         |    2 +-
 arch/x86/include/asm/olpc.h              |    9 ++-
 arch/x86/platform/olpc/Makefile          |    2 +-
 arch/x86/platform/olpc/olpc-xo1-wakeup.S |  132 ++++++++++++++++++++++++++++++
 arch/x86/platform/olpc/olpc-xo1.c        |   79 ++++++++++++++++++
 5 files changed, 219 insertions(+), 5 deletions(-)
 create mode 100644 arch/x86/platform/olpc/olpc-xo1-wakeup.S

v2: add dependency on CONFIG_PM_SLEEP (thanks Randy), avoid requirement
on hacking swsusp_pg_dir by switching to initial_page_table

v3: rebase to fix conflict in olpc.h with the now-merged XO-1 rfkill driver

Resending after 2 weeks of no feedback.

diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
index e832768..a27b0dc 100644
--- a/arch/x86/Kconfig
+++ b/arch/x86/Kconfig
@@ -2041,7 +2041,7 @@ config OLPC
 
 config OLPC_XO1
 	tristate "OLPC XO-1 support"
-	depends on OLPC && PCI
+	depends on OLPC && PCI && PM_SLEEP
 	---help---
 	  Add support for non-essential features of the OLPC XO-1 laptop.
 
diff --git a/arch/x86/include/asm/olpc.h b/arch/x86/include/asm/olpc.h
index 42a978c..5bf0805 100644
--- a/arch/x86/include/asm/olpc.h
+++ b/arch/x86/include/asm/olpc.h
@@ -88,9 +88,12 @@ extern int olpc_ec_mask_unset(uint8_t bits);
 
 /* EC commands */
 
-#define EC_FIRMWARE_REV		0x08
-#define EC_WLAN_ENTER_RESET	0x35
-#define EC_WLAN_LEAVE_RESET	0x25
+#define EC_FIRMWARE_REV			0x08
+#define EC_WAKE_UP_WLAN			0x24
+#define EC_WLAN_LEAVE_RESET		0x25
+#define EC_SET_SCI_INHIBIT		0x32
+#define EC_SET_SCI_INHIBIT_RELEASE	0x34
+#define EC_WLAN_ENTER_RESET		0x35
 
 /* SCI source values */
 
diff --git a/arch/x86/platform/olpc/Makefile b/arch/x86/platform/olpc/Makefile
index c31b8fc..a1e7f02 100644
--- a/arch/x86/platform/olpc/Makefile
+++ b/arch/x86/platform/olpc/Makefile
@@ -1,3 +1,3 @@
 obj-$(CONFIG_OLPC)		+= olpc.o
-obj-$(CONFIG_OLPC_XO1)		+= olpc-xo1.o
+obj-$(CONFIG_OLPC_XO1)		+= olpc-xo1.o olpc-xo1-wakeup.o
 obj-$(CONFIG_OLPC_OPENFIRMWARE)	+= olpc_ofw.o
diff --git a/arch/x86/platform/olpc/olpc-xo1-wakeup.S b/arch/x86/platform/olpc/olpc-xo1-wakeup.S
new file mode 100644
index 0000000..d36e5f4
--- /dev/null
+++ b/arch/x86/platform/olpc/olpc-xo1-wakeup.S
@@ -0,0 +1,132 @@
+.text
+#include <linux/linkage.h>
+#include <asm/segment.h>
+#include <asm/page.h>
+#include <asm/pgtable_32.h>
+
+	.macro writepost,value
+		movb $0x34, %al
+		outb %al, $0x70
+		movb $\value, %al
+		outb %al, $0x71
+	.endm
+
+ALIGN
+	.align 4096
+
+wakeup_start:
+#	jmp wakeup_start
+
+	cli
+	cld
+
+	# Clear any dangerous flags
+
+	pushl $0
+	popfl
+
+	writepost 0x31
+
+	# Set up %cr3
+	movl $initial_page_table - __PAGE_OFFSET, %eax
+	movl %eax, %cr3
+
+	movl saved_cr4, %eax
+	movl %eax, %cr4
+
+	movl saved_cr0, %eax
+	movl %eax, %cr0
+
+	jmp 1f
+1:
+	ljmpl $__KERNEL_CS,$wakeup_return
+
+
+.org 0x1000
+
+wakeup_return:
+	movw    $__KERNEL_DS, %ax
+	movw    %ax, %ss
+	movw    %ax, %ds
+	movw    %ax, %es
+	movw    %ax, %fs
+	movw    %ax, %gs
+
+	lgdt    saved_gdt
+	lidt    saved_idt
+	lldt    saved_ldt
+	ljmp    $(__KERNEL_CS),$1f
+1:
+	movl    %cr3, %eax
+	movl    %eax, %cr3
+	wbinvd
+
+	# Go back to the return point
+	jmp ret_point
+
+save_registers:
+	sgdt  saved_gdt
+	sidt  saved_idt
+	sldt  saved_ldt
+
+	pushl %edx
+	movl %cr4, %edx
+	movl %edx, saved_cr4
+
+	movl %cr0, %edx
+	movl %edx, saved_cr0
+
+	popl %edx
+
+	movl %ebx, saved_context_ebx
+	movl %ebp, saved_context_ebp
+	movl %esi, saved_context_esi
+	movl %edi, saved_context_edi
+
+	pushfl
+	popl saved_context_eflags
+
+	ret
+
+
+restore_registers:
+	movl saved_context_ebp, %ebp
+	movl saved_context_ebx, %ebx
+	movl saved_context_esi, %esi
+	movl saved_context_edi, %edi
+
+	pushl saved_context_eflags
+	popfl
+
+	ret
+
+
+ENTRY(do_olpc_suspend_lowlevel)
+	call	save_processor_state
+	call	save_registers
+
+	# This is the stack context we want to remember
+	movl %esp, saved_context_esp
+
+	pushl	$3
+	call	olpc_xo1_do_sleep
+
+	jmp	wakeup_start
+	.p2align 4,,7
+ret_point:
+	movl    saved_context_esp, %esp
+
+	writepost 0x32
+
+	call	restore_registers
+	call	restore_processor_state
+	ret
+
+.data
+ALIGN
+
+saved_gdt:     .long   0,0
+saved_idt:     .long   0,0
+saved_ldt:     .long   0
+saved_cr4:     .long   0
+saved_cr0:     .long   0
diff --git a/arch/x86/platform/olpc/olpc-xo1.c b/arch/x86/platform/olpc/olpc-xo1.c
index f5442c0..9a06081 100644
--- a/arch/x86/platform/olpc/olpc-xo1.c
+++ b/arch/x86/platform/olpc/olpc-xo1.c
@@ -16,6 +16,7 @@
 #include <linux/pci_ids.h>
 #include <linux/platform_device.h>
 #include <linux/pm.h>
+#include <linux/suspend.h>
 
 #include <asm/io.h>
 #include <asm/olpc.h>
@@ -33,12 +34,83 @@
 #define PM_SSC		0x54
 
 /* PM registers (ACPI block) */
+#define PM1_STS		0x00
 #define PM1_CNT		0x08
 #define PM_GPE0_STS	0x18
 
+#define CS5536_PM_PWRBTN (1 << 8)
+
+extern void do_olpc_suspend_lowlevel(void);
+
 static unsigned long acpi_base;
 static unsigned long pms_base;
 
+static struct {
+	unsigned long address;
+	unsigned short segment;
+} ofw_bios_entry = { 0xF0000 + PAGE_OFFSET, __KERNEL_CS };
+
+static int xo1_power_state_enter(suspend_state_t pm_state)
+{
+	int r;
+
+	/* Only STR is supported */
+	if (pm_state != PM_SUSPEND_MEM)
+		return -EINVAL;
+
+	r = olpc_ec_cmd(EC_SET_SCI_INHIBIT, NULL, 0, NULL, 0);
+	if (r)
+		return r;
+
+	/* Save CPU state */
+	do_olpc_suspend_lowlevel();
+
+	/* Resume path starts here */
+
+	/* Tell the EC to stop inhibiting SCIs */
+	olpc_ec_cmd(EC_SET_SCI_INHIBIT_RELEASE, NULL, 0, NULL, 0);
+
+	/*
+	 * Tell the wireless module to restart USB communication.
+	 * Must be done twice.
+	 */
+	olpc_ec_cmd(EC_WAKE_UP_WLAN, NULL, 0, NULL, 0);
+	olpc_ec_cmd(EC_WAKE_UP_WLAN, NULL, 0, NULL, 0);
+
+	return 0;
+}
+
+asmlinkage int olpc_xo1_do_sleep(u8 sleep_state)
+{
+	void *pgd_addr = __va(read_cr3());
+	printk(KERN_ERR "xo1_do_sleep!\n"); /* this needs to remain here so
+					      * that gcc doesn't optimize
+					      * away our __va! */
+
+	/* Enable wakeup through power button */
+	outl((CS5536_PM_PWRBTN << 16) | 0xFFFF, acpi_base + PM1_STS);
+
+	__asm__ __volatile__("movl %0,%%eax" : : "r" (pgd_addr));
+	__asm__("call *(%%edi); cld"
+		: : "D" (&ofw_bios_entry));
+	__asm__ __volatile__("movb $0x34, %al\n\t"
+			     "outb %al, $0x70\n\t"
+			     "movb $0x30, %al\n\t"
+			     "outb %al, $0x71\n\t");
+	return 0;
+}
+
+static int xo1_power_state_valid(suspend_state_t pm_state)
+{
+	/* suspend-to-RAM only */
+	return pm_state == PM_SUSPEND_MEM;
+}
+
+static struct platform_suspend_ops xo1_suspend_ops = {
+	.valid = xo1_power_state_valid,
+	.enter = xo1_power_state_enter,
+};
+
 static void xo1_power_off(void)
 {
 	printk(KERN_INFO "OLPC XO-1 power off sequence...\n");
@@ -101,6 +173,13 @@ static int __devinit olpc_xo1_probe(struct platform_device *pdev)
 	if (r)
 		return r;
 
+	/*
+	 * Take a reference on ourself to prevent module unloading. We can't
+	 * safely unload after changing the suspend handlers.
+	 */
+	__module_get(THIS_MODULE);
+
+	suspend_set_ops(&xo1_suspend_ops);
 	pm_power_off = xo1_power_off;
 
 	printk(KERN_INFO "OLPC XO-1 support registered\n");
-- 
1.7.3.2