Skip to contentSkip to conversation

x86, suspend: Save/restore extra MSR registers for suspend

PATCH Superseded Chen Yu 3 participants 6 messages Oct 11, 2015 – Nov 16, 2015

Subsystems (2)

v5 of 7 Superseded Latest is v7 →

lore
Share & tools
lore

Chen Yu

author
#0 lore raw

PATCH v5 1/1 x86, suspend: Save/restore extra MSR registers for suspend

A bug is reported(https://bugzilla.redhat.com/show_bug.cgi?id=1227208) that, after resumed from S3, CPU is running at a low speed. After investigation, it is found that, BIOS has modified the value of THERM_CONTROL register during S3, and changes it from 0 to 0x10 (thus changes the clock modulation from reserved to enabled), since value of 0x10 means CPU can only get 25% of the Duty Cycle, this triggers the problem.

Here is a simple scenario to reproduce the issue(only for above case):
1.Boot up the system
2.Get MSR with address 0x19a, it should be 0
3.Put the system into sleep, then wake it up
4.Get MSR with address 0x19a, it should be 0(actually it shows 0x10)

Although this is a BIOS issue, it would be more robust for linux to deal with this situation. This patch fixes this issue by introducing a framework to save/restore specified MSR registers(THERM_CONTROL in this case) for suspend/resume.

When user encounters a problematic platform and needs to protect the MSRs during suspending, he can simply add a quirk entry in msr_save_dmi_table, and customizes MSR registers inside the quirk callback, for example:

u32 msr_id_need_to_save[] = {MSR_ID0, MSR_ID1, MSR_ID2...};

and the quirk mechanism ensures that, once resumed from suspended, the MSRs indicated by these IDs will be restored to their original values before suspended.

Since both 64/32-bit kernels are affected, this patch covers 64/32-bit common code path. And because the MSRs specified by the user might not be available or readable in any situation, we use rdmsrl_safe to safely save these MSRs.

---
v5:
- Rename some structures and variables for better understanding. Put the defination of struct saved_msr and struct msr_save_data to
header: arch/x86/include/asm/msr.h.
Re-implement the msr_save_context and msr_restore_context for better maintaining.
Convert the msr_context_array to be dynamically allocated. Fix some typos in code comments.
v4:
- Revert v3 to v2, and fix some typos in changelog/comments. Use msr_info structure instead of msr_id + msr_value. Adjust some codes for better readability.
v3:
- Simplify the patch to only focus on THERM_CONTROL register. This will make things 'just work'.
v2:
- Cover both 64/32-bit common code path.
Use rdmsrl_safe to safely read MSR.
Introduce a quirk framework for save/restore specified MSR on different platforms.

---
 arch/x86/include/asm/msr.h        | 10 ++++
 arch/x86/include/asm/suspend_32.h |  1 +
 arch/x86/include/asm/suspend_64.h |  1 +
 arch/x86/power/cpu.c              | 98 +++++++++++++++++++++++++++++++++++++++
 4 files changed, 110 insertions(+)
arch/x86/include/asm/msr.h +10 -0
index 77d8b28..5ae24ed 100644
@@ -32,6 +32,16 @@ struct msr_regs_info {
 	int err;
 };
 
+struct msr_save_data {
+	bool msr_saved;
+	struct msr_info rv;
+};
+
+struct msr_context {
+	unsigned short num;
+	struct msr_save_data *msr_array;
+};
+
 static inline unsigned long long native_read_tscp(unsigned int *aux)
 {
 	unsigned long low, high;
arch/x86/include/asm/suspend_32.h +1 -0
index d1793f0..5057f65 100644
@@ -15,6 +15,7 @@ struct saved_context {
 	unsigned long cr0, cr2, cr3, cr4;
 	u64 misc_enable;
 	bool misc_enable_saved;
+	struct msr_context msr_to_save;
 	struct desc_ptr gdt_desc;
 	struct desc_ptr idt;
 	u16 ldt;
arch/x86/include/asm/suspend_64.h +1 -0
index 7ebf0eb..60941de 100644
@@ -24,6 +24,7 @@ struct saved_context {
 	unsigned long cr0, cr2, cr3, cr4, cr8;
 	u64 misc_enable;
 	bool misc_enable_saved;
+	struct msr_context msr_to_save;
 	unsigned long efer;
 	u16 gdt_pad; /* Unused */
 	struct desc_ptr gdt_desc;
arch/x86/power/cpu.c +98 -0

125 lines not on this page · raw message

Signature · 1 line
1.8.4.2
#1 lore raw
Quoted 1 line · Chen Yu

A bug is reported(https://bugzilla.redhat.com/show_bug.cgi?id=1227208)

I get:

"You are not authorized to access bug #1227208.

Most likely the bug has been restricted for internal development processes and we cannot grant access."

Quoted 6 lines

that, after resumed from S3, CPU is running at a low speed. After investigation, it is found that, BIOS has modified the value of THERM_CONTROL register during S3, and changes it from 0 to 0x10 (thus changes the clock modulation from reserved to enabled), since value of 0x10 means CPU can only get 25% of the Duty Cycle, this triggers the problem.

Is this what the bug described above is? In any case, please remove the private bugzilla link and describe the bug in text here.

Also, from reading the other thread about the v4 patch, it sounds like intel_pstate can't handle the clock modulation properly, according to what Doug says.

So let's have this aspect sorted out properly first please before adding yet another ugly BIOS workaround.

Btw, why can't that BIOS be fixed instead?

Signature · 5 lines
Regards/Gruss,
    Boris.

SUSE Linux GmbH, GF: Felix Imendörffer, Jane Smithard, Graham Norton, HRB 21284 (AG Nürnberg)
-- 
#2 lore raw
Quoted 20 lines · Borislav Petkov
Quoted 1 line · Chen Yu

A bug is reported(https://bugzilla.redhat.com/show_bug.cgi?id=1227208)

I get:

"You are not authorized to access bug #1227208.

Most likely the bug has been restricted for internal development processes and we cannot grant access."

Quoted 6 lines

that, after resumed from S3, CPU is running at a low speed. After investigation, it is found that, BIOS has modified the value of THERM_CONTROL register during S3, and changes it from 0 to 0x10 (thus changes the clock modulation from reserved to enabled), since value of 0x10 means CPU can only get 25% of the Duty Cycle, this triggers the problem.

Is this what the bug described above is? In any case, please remove the private bugzilla link and describe the bug in text here.

Quoted 8 lines

Also, from reading the other thread about the v4 patch, it sounds like intel_pstate can't handle the clock modulation properly, according to what Doug says.

So let's have this aspect sorted out properly first please before adding yet another ugly BIOS workaround.

Btw, why can't that BIOS be fixed instead?

Because it's been shipped to users.

Thanks,
Rafael

author
#3 lore raw
In reply to Borislav Petkov

Hi, Boris, thanks for you reply

Quoted 28 lines

-----Original Message-----
From: Borislav Petkov [mailto:[email protected]]
Sent: Tuesday, October 13, 2015 12:38 AM
To: Chen, Yu C
Cc: [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; Zhang, Rui; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; Kaszewski, Marcin
Subject: Re: [PATCH][v5] x86, suspend: Save/restore extra MSR registers for suspend

Quoted 1 line · Chen Yu

A bug is reported(https://bugzilla.redhat.com/show_bug.cgi?id=1227208)

I get:

"You are not authorized to access bug #1227208.

Most likely the bug has been restricted for internal development processes and we cannot grant access."

Quoted 6 lines

that, after resumed from S3, CPU is running at a low speed. After investigation, it is found that, BIOS has modified the value of THERM_CONTROL register during S3, and changes it from 0 to 0x10 (thus changes the clock modulation from reserved to enabled), since value of 0x10 means CPU can only get 25% of the Duty Cycle, this triggers the problem.

Quoted 2 lines

Is this what the bug described above is? In any case, please remove the private bugzilla link and describe the bug in text here.

OK, will remove the link here and inside the code.

Quoted 8 lines

Also, from reading the other thread about the v4 patch, it sounds like intel_pstate can't handle the clock modulation properly, according to what Doug says.

So let's have this aspect sorted out properly first please before adding yet another ugly BIOS workaround.

Well, that might be another driver bug IMO. I'll dig into that one.

Quoted 1 line

Btw, why can't that BIOS be fixed instead?

This bug is found on a released redhat server, so I think it would be hard to update all users' machines.

Quoted 4 lines

--
Regards/Gruss,
Boris.

Best Regards,
Yu

author
#4 lore raw
In reply to Borislav Petkov

Hi, Boris,

Quoted 28 lines

-----Original Message-----
From: Borislav Petkov [mailto:[email protected]]
Sent: Tuesday, October 13, 2015 12:38 AM
To: Chen, Yu C
Cc: [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; Zhang, Rui; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; Kaszewski, Marcin
Subject: Re: [PATCH][v5] x86, suspend: Save/restore extra MSR registers for suspend

Quoted 1 line · Chen Yu

A bug is reported(https://bugzilla.redhat.com/show_bug.cgi?id=1227208)

I get:

"You are not authorized to access bug #1227208.

Most likely the bug has been restricted for internal development processes and we cannot grant access."

Quoted 6 lines

that, after resumed from S3, CPU is running at a low speed. After investigation, it is found that, BIOS has modified the value of THERM_CONTROL register during S3, and changes it from 0 to 0x10 (thus changes the clock modulation from reserved to enabled), since value of 0x10 means CPU can only get 25% of the Duty Cycle, this triggers the problem.

Quoted 2 lines

Is this what the bug described above is? In any case, please remove the private bugzilla link and describe the bug in text here.

Yes, this bug is described in the changelog, I'll remove the link.

Quoted 4 lines

Also, from reading the other thread about the v4 patch, it sounds like intel_pstate can't handle the clock modulation properly, according to what Doug says.

As discussed with Doug in the v4 patch, the problem brought by clock modulation is because intel_pstate use a theoretical pstate to calculate the cpu busy ratio, and Doug has proposed a patch to use compensatory algorithm in intel_pstate, to predict the pstate based on actual cpu freq(take clock modulation into account), which can fix his problem now.
So I'll resend a version 6 of current patch, because some server in production environment has suffered from this problem for some time.. thanks.

Best regards,
Yu

Acked-by on v5 1/1
#5 lore raw
In reply to Chen Yu
Quoted 30 lines · Chen Yu

A bug is reported(https://bugzilla.redhat.com/show_bug.cgi?id=1227208) that, after resumed from S3, CPU is running at a low speed. After investigation, it is found that, BIOS has modified the value of THERM_CONTROL register during S3, and changes it from 0 to 0x10 (thus changes the clock modulation from reserved to enabled), since value of 0x10 means CPU can only get 25% of the Duty Cycle, this triggers the problem.

Here is a simple scenario to reproduce the issue(only for above case):
1.Boot up the system
2.Get MSR with address 0x19a, it should be 0
3.Put the system into sleep, then wake it up
4.Get MSR with address 0x19a, it should be 0(actually it shows 0x10)

Although this is a BIOS issue, it would be more robust for linux to deal with this situation. This patch fixes this issue by introducing a framework to save/restore specified MSR registers(THERM_CONTROL in this case) for suspend/resume.

When user encounters a problematic platform and needs to protect the MSRs during suspending, he can simply add a quirk entry in msr_save_dmi_table, and customizes MSR registers inside the quirk callback, for example:

u32 msr_id_need_to_save[] = {MSR_ID0, MSR_ID1, MSR_ID2...};

and the quirk mechanism ensures that, once resumed from suspended, the MSRs indicated by these IDs will be restored to their original values before suspended.

Quoted 7 lines

Since both 64/32-bit kernels are affected, this patch covers 64/32-bit common code path. And because the MSRs specified by the user might not be available or readable in any situation, we use rdmsrl_safe to safely save these MSRs.

Quoted 208 lines

---
v5:
- Rename some structures and variables for better understanding. Put the defination of struct saved_msr and struct msr_save_data to
header: arch/x86/include/asm/msr.h.
Re-implement the msr_save_context and msr_restore_context for better maintaining.
Convert the msr_context_array to be dynamically allocated. Fix some typos in code comments.
v4:
- Revert v3 to v2, and fix some typos in changelog/comments. Use msr_info structure instead of msr_id + msr_value. Adjust some codes for better readability.
v3:
- Simplify the patch to only focus on THERM_CONTROL register. This will make things 'just work'.
v2:
- Cover both 64/32-bit common code path.
Use rdmsrl_safe to safely read MSR.
Introduce a quirk framework for save/restore specified MSR on different platforms.

---
 arch/x86/include/asm/msr.h        | 10 ++++
 arch/x86/include/asm/suspend_32.h |  1 +
 arch/x86/include/asm/suspend_64.h |  1 +
 arch/x86/power/cpu.c              | 98 +++++++++++++++++++++++++++++++++++++++
 4 files changed, 110 insertions(+)
arch/x86/include/asm/msr.h +10 -0
index 77d8b28..5ae24ed 100644
@@ -32,6 +32,16 @@ struct msr_regs_info {
 	int err;
 };
 
+struct msr_save_data {
+	bool msr_saved;
+	struct msr_info rv;
+};
+
+struct msr_context {
+	unsigned short num;
+	struct msr_save_data *msr_array;
+};
+
 static inline unsigned long long native_read_tscp(unsigned int *aux)
 {
 	unsigned long low, high;
arch/x86/include/asm/suspend_32.h +1 -0
index d1793f0..5057f65 100644
@@ -15,6 +15,7 @@ struct saved_context {
 	unsigned long cr0, cr2, cr3, cr4;
 	u64 misc_enable;
 	bool misc_enable_saved;
+	struct msr_context msr_to_save;
 	struct desc_ptr gdt_desc;
 	struct desc_ptr idt;
 	u16 ldt;
arch/x86/include/asm/suspend_64.h +1 -0
index 7ebf0eb..60941de 100644
@@ -24,6 +24,7 @@ struct saved_context {
 	unsigned long cr0, cr2, cr3, cr4, cr8;
 	u64 misc_enable;
 	bool misc_enable_saved;
+	struct msr_context msr_to_save;
 	unsigned long efer;
 	u16 gdt_pad; /* Unused */
 	struct desc_ptr gdt_desc;
arch/x86/power/cpu.c +98 -0
index 9ab5279..fd3243a 100644
@@ -23,6 +23,7 @@
 #include <asm/debugreg.h>
 #include <asm/cpu.h>
 #include <asm/mmu_context.h>
+#include <linux/dmi.h>
 
 #ifdef CONFIG_X86_32
 __visible unsigned long saved_context_ebx;
@@ -32,6 +33,29 @@ __visible unsigned long saved_context_eflags;
 #endif
 struct saved_context saved_context;
 
+static void msr_save_context(struct saved_context *ctxt)
+{
+	struct msr_save_data *msr = ctxt->msr_to_save.msr_array;
+	struct msr_save_data *end = msr + ctxt->msr_to_save.num;
+
+	while (msr < end) {
+		msr->msr_saved = !rdmsrl_safe(msr->rv.msr_no, &msr->rv.reg.q);
+		msr++;
+	}
+}
+
+static void msr_restore_context(struct saved_context *ctxt)
+{
+	struct msr_save_data *msr = ctxt->msr_to_save.msr_array;
+	struct msr_save_data *end = msr + ctxt->msr_to_save.num;
+
+	while (msr < end) {
+		if (msr->msr_saved)
+			wrmsrl(msr->rv.msr_no, msr->rv.reg.q);
+		msr++;
+	}
+}
+
 /**
  *	__save_processor_state - save CPU registers before creating a
  *		hibernation image and before restoring the memory state from it
@@ -111,6 +135,7 @@ static void __save_processor_state(struct saved_context *ctxt)
 #endif
 	ctxt->misc_enable_saved = !rdmsrl_safe(MSR_IA32_MISC_ENABLE,
 					       &ctxt->misc_enable);
+	msr_save_context(ctxt);
 }
 
 /* Needed by apm.c */
@@ -229,6 +254,7 @@ static void notrace __restore_processor_state(struct saved_context *ctxt)
 	x86_platform.restore_sched_clock_state();
 	mtrr_bp_restore();
 	perf_restore_debug_store();
+	msr_restore_context(ctxt);
 }
 
 /* Needed by apm.c */
@@ -320,3 +346,75 @@ static int __init bsp_pm_check_init(void)
 }
 
 core_initcall(bsp_pm_check_init);
+
+/*
+ * The following section is a quirk framework for problematic BIOS:
+ * Sometimes MSRs are modified by BIOS after suspended to
+ * RAM, this might cause unexpected behavior after wakeup.
+ * Thus we save/restore these specified MSRs during suspending
+ * in order to work around it.
+ * A typical bug is reported at:
+ * https://bugzilla.redhat.com/show_bug.cgi?id=1227208
+ */
+static int msr_init_context(const u32 *msr_id, const int total_num)
+{
+	int i = 0;
+	struct msr_save_data *msr_data = NULL;
+
+	if (saved_context.msr_to_save.msr_array ||
+			saved_context.msr_to_save.num > 0) {
+		pr_err("PM: quirk already applied, please check your dmi match table.\n");
+		return -EINVAL;
+	}
+
+	msr_data = kmalloc_array(total_num,
+			sizeof(struct msr_save_data), GFP_KERNEL);
+	if (!msr_data) {
+		pr_err("PM: can not allocate memory to save/restore MSRs during suspend.\n");
+		return -ENOMEM;
+	}
+
+	for (i = 0; i < total_num; i++) {
+		msr_data[i].rv.msr_no = msr_id[i];
+		msr_data[i].msr_saved = false;
+		msr_data[i].rv.reg.q = 0;
+	}
+	saved_context.msr_to_save.num = total_num;
+	saved_context.msr_to_save.msr_array = msr_data;
+	return 0;
+}
+
+/*
+ * For any further problematic BIOS/platforms,
+ * please add your own function similar to msr_initialize_bdw.
+ */
+static int msr_initialize_bdw(const struct dmi_system_id *d)
+{
+	/* Add any extra MSR ids into this array. */
+	u32 bdw_msr_id[] = {MSR_IA32_THERM_CONTROL};
+
+	pr_info("PM: %s detected, MSR saving is needed during suspending.\n",
+		d->ident);
+	return msr_init_context(bdw_msr_id, ARRAY_SIZE(bdw_msr_id));
+}
+
+static struct dmi_system_id msr_save_dmi_table[] = {
+	{
+	 .callback = msr_initialize_bdw,
+	 .ident = "BROADWELL BDX_EP",
+	 .matches = {
+		DMI_MATCH(DMI_SYS_VENDOR, "Intel Corporation"),
+		DMI_MATCH(DMI_PRODUCT_NAME, "GRANTLEY"),
+		DMI_MATCH(DMI_PRODUCT_VERSION, "E63448-400"),
+		},
+	},
+	{}
+};
+
+static int pm_check_save_msr(void)
+{
+	dmi_check_system(msr_save_dmi_table);
+	return 0;
+}
+
+late_initcall(pm_check_save_msr);
Signature · 2 lines
I speak only for myself.
Rafael J. Wysocki, Intel Open Source Technology Center.