x86: call smp vmxoff in smp stop

RFC PATCH Superseded Xishi Qiu 4 participants 15 messages Jan 4, 2017 – Apr 19, 2017 x86

lore

Share & tools

lore

Xishi Qiu

author

2017-01-04 10:13 UTC #0 lore raw

RFC PATCH 1/1 x86: call smp vmxoff in smp stop

From: f00186668 <[email protected]>

We need to disable VMX on all CPUs before stop cpu when OS panic, otherwisewe risk hanging up the machine, because the CPU ignore INIT signals when VMX is enabled. In kernel mainline this issue existence.

Signed-off-by f00186668 <[email protected]>

---
 arch/x86/kernel/smp.c | 2 ++
 1 file changed, 2 insertions(+)

arch/x86/kernel/smp.c +2 -0

index 68f8cc2..6b64c6b 100644
@@ -162,6 +162,7 @@ static int smp_stop_nmi_callback(unsigned int val, struct pt_regs *regs)
 	if (raw_smp_processor_id() == atomic_read(&stopping_cpu))
 		return NMI_HANDLED;
 
+	cpu_emergency_vmxoff();
 	stop_this_cpu(NULL);
 
 	return NMI_HANDLED;

@@ -174,6 +175,7 @@ static int smp_stop_nmi_callback(unsigned int val, struct pt_regs *regs)
 asmlinkage __visible void smp_reboot_interrupt(void)
 {
 	ipi_entering_ack_irq();
+	cpu_emergency_vmxoff();
 	stop_this_cpu(NULL);
 	irq_exit();
 }

Signature · 1 line

1.8.3.1

Xishi Qiu

author

2017-01-05 01:46 UTC #1 lore raw

RFC PATCH v2 1/1 x86: call smp vmxoff in smp stop

From: f00186668 <[email protected]>

Signed-off-by f00186668 <[email protected]>

---
 arch/x86/kernel/smp.c | 3 +++
 1 file changed, 3 insertions(+)

arch/x86/kernel/smp.c +3 -0

index 68f8cc2..b574d55 100644
@@ -33,6 +33,7 @@
 #include <asm/mce.h>
 #include <asm/trace/irq_vectors.h>
 #include <asm/kexec.h>
+#include <asm/virtext.h>
 
 /*
  *	Some notes on x86 processor bugs affecting SMP operation:

@@ -162,6 +163,7 @@ static int smp_stop_nmi_callback(unsigned int val, struct pt_regs *regs)
 	if (raw_smp_processor_id() == atomic_read(&stopping_cpu))
 		return NMI_HANDLED;
 
+	cpu_emergency_vmxoff();
 	stop_this_cpu(NULL);
 
 	return NMI_HANDLED;

@@ -174,6 +176,7 @@ static int smp_stop_nmi_callback(unsigned int val, struct pt_regs *regs)
 asmlinkage __visible void smp_reboot_interrupt(void)
 {
 	ipi_entering_ack_irq();
+	cpu_emergency_vmxoff();
 	stop_this_cpu(NULL);
 	irq_exit();
 }

Signature · 1 line

1.8.3.1

Paolo Bonzini

2017-01-12 14:05 UTC #2 lore raw

Quoted 7 lines · Xishi Qiu

From: f00186668 <[email protected]>
We need to disable VMX on all CPUs before stop cpu when OS panic, otherwisewe risk hanging up the machine, because the CPU ignore INIT signals when VMX is enabled. In kernel mainline this issue existence.
Signed-off-by f00186668 <[email protected]>

Looks good, but you need to put your colleague's real name (Tiantian Feng?) in the Signed-off-by line, and you need another Signed-off-by line for yourself.

Thanks,

Paolo

Quoted 33 lines

---
 arch/x86/kernel/smp.c | 3 +++
 1 file changed, 3 insertions(+)

arch/x86/kernel/smp.c +3 -0

index 68f8cc2..b574d55 100644
@@ -33,6 +33,7 @@
 #include <asm/mce.h>
 #include <asm/trace/irq_vectors.h>
 #include <asm/kexec.h>
+#include <asm/virtext.h>
 
 /*
  *	Some notes on x86 processor bugs affecting SMP operation:

@@ -162,6 +163,7 @@ static int smp_stop_nmi_callback(unsigned int val, struct pt_regs *regs)
 	if (raw_smp_processor_id() == atomic_read(&stopping_cpu))
 		return NMI_HANDLED;
 
+	cpu_emergency_vmxoff();
 	stop_this_cpu(NULL);
 
 	return NMI_HANDLED;

@@ -174,6 +176,7 @@ static int smp_stop_nmi_callback(unsigned int val, struct pt_regs *regs)
 asmlinkage __visible void smp_reboot_interrupt(void)
 {
 	ipi_entering_ack_irq();
+	cpu_emergency_vmxoff();
 	stop_this_cpu(NULL);
 	irq_exit();
 }

Xishi Qiu

author

2017-01-14 01:41 UTC #12 lore raw

PATCH 1/1 x86: call smp vmxoff in smp stop

In reply to Xishi Qiu

From: Tiantian Feng <[email protected]>

Signed-off-by Tiantian Feng <[email protected]>

---
 arch/x86/kernel/smp.c | 2 ++
 1 file changed, 2 insertions(+)

arch/x86/kernel/smp.c +2 -0

index 68f8cc2..6b64c6b 100644
@@ -162,6 +162,7 @@ static int smp_stop_nmi_callback(unsigned int val, struct pt_regs *regs)
 	if (raw_smp_processor_id() == atomic_read(&stopping_cpu))
 		return NMI_HANDLED;
 
+	cpu_emergency_vmxoff();
 	stop_this_cpu(NULL);
 
 	return NMI_HANDLED;

@@ -174,6 +175,7 @@ static int smp_stop_nmi_callback(unsigned int val, struct pt_regs *regs)
 asmlinkage __visible void smp_reboot_interrupt(void)
 {
 	ipi_entering_ack_irq();
+	cpu_emergency_vmxoff();
 	stop_this_cpu(NULL);
 	irq_exit();
 }

Signature · 1 line

1.8.3.1 .

Xishi Qiu

author

2017-01-14 01:42 UTC #13 lore raw

Quoted 8 lines · Xishi Qiu

From: Tiantian Feng <[email protected]>
We need to disable VMX on all CPUs before stop cpu when OS panic, otherwisewe risk hanging up the machine, because the CPU ignore INIT signals when VMX is enabled. In kernel mainline this issue existence.
Signed-off-by Tiantian Feng <[email protected]>
---

Sorry, I missed something, please ignore this one, thanks.

Quoted 23 lines

 arch/x86/kernel/smp.c | 2 ++
 1 file changed, 2 insertions(+)

arch/x86/kernel/smp.c +2 -0

index 68f8cc2..6b64c6b 100644
@@ -162,6 +162,7 @@ static int smp_stop_nmi_callback(unsigned int val, struct pt_regs *regs)
 	if (raw_smp_processor_id() == atomic_read(&stopping_cpu))
 		return NMI_HANDLED;
 
+	cpu_emergency_vmxoff();
 	stop_this_cpu(NULL);
 
 	return NMI_HANDLED;

@@ -174,6 +175,7 @@ static int smp_stop_nmi_callback(unsigned int val, struct pt_regs *regs)
 asmlinkage __visible void smp_reboot_interrupt(void)
 {
 	ipi_entering_ack_irq();
+	cpu_emergency_vmxoff();
 	stop_this_cpu(NULL);
 	irq_exit();
 }

Xishi Qiu

author

2017-01-14 01:42 UTC #3 lore raw

PATCH v3 1/1 x86: call smp vmxoff in smp stop

In reply to Xishi Qiu

From: Tiantian Feng <[email protected]>

Signed-off-by Tiantian Feng <[email protected]>

---
 arch/x86/kernel/smp.c | 3 +++
 1 file changed, 3 insertions(+)

arch/x86/kernel/smp.c +3 -0

index 68f8cc2..b574d55 100644
@@ -33,6 +33,7 @@
 #include <asm/mce.h>
 #include <asm/trace/irq_vectors.h>
 #include <asm/kexec.h>
+#include <asm/virtext.h>
 
 /*
  *	Some notes on x86 processor bugs affecting SMP operation:

@@ -162,6 +163,7 @@ static int smp_stop_nmi_callback(unsigned int val, struct pt_regs *regs)
 	if (raw_smp_processor_id() == atomic_read(&stopping_cpu))
 		return NMI_HANDLED;
 
+	cpu_emergency_vmxoff();
 	stop_this_cpu(NULL);
 
 	return NMI_HANDLED;

@@ -174,6 +176,7 @@ static int smp_stop_nmi_callback(unsigned int val, struct pt_regs *regs)
 asmlinkage __visible void smp_reboot_interrupt(void)
 {
 	ipi_entering_ack_irq();
+	cpu_emergency_vmxoff();
 	stop_this_cpu(NULL);
 	irq_exit();
 }

Signature · 1 line

1.8.3.1

kbuild test robot

build bot

2017-01-15 00:45 UTC #14 lore raw

In reply to Xishi Qiu

url: https://github.com/0day-ci/linux/commits/Xishi-Qiu/x86-call-smp-vmxoff-in-smp-stop/20170115-075446 config: x86_64-randconfig-x017-201703 (attached as .config) compiler: gcc-6 (Debian 6.2.0-3)…

Hi Tiantian,

[auto build test ERROR on tip/auto-latest]
[also build test ERROR on v4.10-rc3 next-20170113]
[if your patch is applied to the wrong git tree, please drop us a note to help improve the system]

url: https://github.com/0day-ci/linux/commits/Xishi-Qiu/x86-call-smp-vmxoff-in-smp-stop/20170115-075446
config: x86_64-randconfig-x017-201703 (attached as .config)
compiler: gcc-6 (Debian 6.2.0-3) 6.2.0 20160901 reproduce:

        # save the attached .config to linux build tree
        make ARCH=x86_64

All errors (new ones prefixed by >>):

arch/x86/kernel/smp.c: In function 'smp_stop_nmi_callback':

>> arch/x86/kernel/smp.c:165:2: error: implicit declaration of function 'cpu_emergency_vmxoff' [-Werror=implicit-function-declaration]
     cpu_emergency_vmxoff();
     ^~~~~~~~~~~~~~~~~~~~

cc1: some warnings being treated as errors

vim +/cpu_emergency_vmxoff +165 arch/x86/kernel/smp.c

159 static int smp_stop_nmi_callback(unsigned int val, struct pt_regs *regs)
160 {
161 /* We are registered on stopping cpu too, avoid spurious NMI */
162 if (raw_smp_processor_id() == atomic_read(&stopping_cpu))
163 return NMI_HANDLED;
164

 > 165		cpu_emergency_vmxoff();

166 stop_this_cpu(NULL);
167
168 return NMI_HANDLED;

---
0-DAY kernel test infrastructure Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all Intel Corporation

Paolo Bonzini

2017-01-17 15:18 UTC #4 lore raw

In reply to Xishi Qiu

Quoted 7 lines · Xishi Qiu

From: Tiantian Feng <[email protected]>
We need to disable VMX on all CPUs before stop cpu when OS panic, otherwisewe risk hanging up the machine, because the CPU ignore INIT signals when VMX is enabled. In kernel mainline this issue existence.
Signed-off-by Tiantian Feng <[email protected]>

Xishi,

it's still missing your Signed-off-by.

Paolo

Quoted 33 lines

---
 arch/x86/kernel/smp.c | 3 +++
 1 file changed, 3 insertions(+)

arch/x86/kernel/smp.c +3 -0

index 68f8cc2..b574d55 100644
@@ -33,6 +33,7 @@
 #include <asm/mce.h>
 #include <asm/trace/irq_vectors.h>
 #include <asm/kexec.h>
+#include <asm/virtext.h>
 
 /*
  *	Some notes on x86 processor bugs affecting SMP operation:

@@ -162,6 +163,7 @@ static int smp_stop_nmi_callback(unsigned int val, struct pt_regs *regs)
 	if (raw_smp_processor_id() == atomic_read(&stopping_cpu))
 		return NMI_HANDLED;
 
+	cpu_emergency_vmxoff();
 	stop_this_cpu(NULL);
 
 	return NMI_HANDLED;

@@ -174,6 +176,7 @@ static int smp_stop_nmi_callback(unsigned int val, struct pt_regs *regs)
 asmlinkage __visible void smp_reboot_interrupt(void)
 {
 	ipi_entering_ack_irq();
+	cpu_emergency_vmxoff();
 	stop_this_cpu(NULL);
 	irq_exit();
 }

Xishi Qiu

author

2017-01-18 02:20 UTC #5 lore raw

Quoted 9 lines · Paolo Bonzini

Quoted 6 lines · Xishi Qiu
From: Tiantian Feng <[email protected]>
We need to disable VMX on all CPUs before stop cpu when OS panic, otherwisewe risk hanging up the machine, because the CPU ignore INIT signals when VMX is enabled. In kernel mainline this issue existence.

Quoted 6 lines

Quoted 1 line
Signed-off-by Tiantian Feng <[email protected]>
Xishi,
it's still missing your Signed-off-by.

Hi Paolo,

This patch is from fengtiantian, and I just send it for him, so still should add my SOB?

Thanks,
Xishi Qiu

Quoted 38 lines

Paolo

Quoted 26 lines

---
 arch/x86/kernel/smp.c | 3 +++
 1 file changed, 3 insertions(+)

arch/x86/kernel/smp.c +2 -0

index 68f8cc2..b574d55 100644
@@ -33,6 +33,7 @@
 #include <asm/mce.h>
 #include <asm/trace/irq_vectors.h>
 #include <asm/kexec.h>
+#include <asm/virtext.h>
 
 /*
  *	Some notes on x86 processor bugs affecting SMP operation:

@@ -162,6 +163,7 @@ static int smp_stop_nmi_callback(unsigned int val, struct pt_regs *regs)
 	if (raw_smp_processor_id() == atomic_read(&stopping_cpu))
 		return NMI_HANDLED;
 
+	cpu_emergency_vmxoff();
 	stop_this_cpu(NULL);
 
 	return NMI_HANDLED;

arch/x86/kernel/smp.c static int smp_stop_nmi_callback(unsigned int val, struct pt_regs *regs) · line 182

Quoted 7 lines

arch/x86/kernel/smp.c +1 -0

@@ -174,6 +176,7 @@ static int smp_stop_nmi_callback(unsigned int val, struct pt_regs *regs)
 asmlinkage __visible void smp_reboot_interrupt(void)
 {
 	ipi_entering_ack_irq();
+	cpu_emergency_vmxoff();
 	stop_this_cpu(NULL);
 	irq_exit();
 }

Paolo Bonzini

2017-01-18 09:30 UTC #6 lore raw

Quoted 15 lines · Xishi Qiu

Quoted 13 lines · Paolo Bonzini
Quoted 7 lines · Xishi Qiu
From: Tiantian Feng <[email protected]>
We need to disable VMX on all CPUs before stop cpu when OS panic, otherwisewe risk hanging up the machine, because the CPU ignore INIT signals when VMX is enabled. In kernel mainline this issue existence.
Signed-off-by Tiantian Feng <[email protected]>
Xishi,

Quoted 7 lines

Quoted 2 lines
it's still missing your Signed-off-by.
Hi Paolo,
This patch is from fengtiantian, and I just send it for him, so still should add my SOB?

Yes, both of them should be there. The "signed-off-by" is a sequence of all people that managed the patch---so that would be Tiantian first, then you, then an x86 maintainer.

Paolo

Quoted 44 lines

Thanks,
Xishi Qiu

Quoted 38 lines

Paolo

Quoted 26 lines

---
 arch/x86/kernel/smp.c | 3 +++
 1 file changed, 3 insertions(+)

arch/x86/kernel/smp.c +2 -0

index 68f8cc2..b574d55 100644
@@ -33,6 +33,7 @@
 #include <asm/mce.h>
 #include <asm/trace/irq_vectors.h>
 #include <asm/kexec.h>
+#include <asm/virtext.h>
 
 /*
  *	Some notes on x86 processor bugs affecting SMP operation:

@@ -162,6 +163,7 @@ static int smp_stop_nmi_callback(unsigned int val, struct pt_regs *regs)
 	if (raw_smp_processor_id() == atomic_read(&stopping_cpu))
 		return NMI_HANDLED;
 
+	cpu_emergency_vmxoff();
 	stop_this_cpu(NULL);
 
 	return NMI_HANDLED;

arch/x86/kernel/smp.c static int smp_stop_nmi_callback(unsigned int val, struct pt_regs *regs) · line 182

Quoted 7 lines

arch/x86/kernel/smp.c +1 -0

@@ -174,6 +176,7 @@ static int smp_stop_nmi_callback(unsigned int val, struct pt_regs *regs)
 asmlinkage __visible void smp_reboot_interrupt(void)
 {
 	ipi_entering_ack_irq();
+	cpu_emergency_vmxoff();
 	stop_this_cpu(NULL);
 	irq_exit();
 }

Xishi Qiu

author

2017-01-18 11:40 UTC #7 lore raw

PATCH v4 1/1 x86: call smp vmxoff in smp stop

In reply to Xishi Qiu

From: Tiantian Feng <[email protected]>

Signed-off-by Tiantian Feng <[email protected]>

Signed-off-by Xishi Qiu <[email protected]>

---
 arch/x86/kernel/smp.c | 3 +++
 1 file changed, 3 insertions(+)

arch/x86/kernel/smp.c +3 -0

index 68f8cc2..b574d55 100644
@@ -33,6 +33,7 @@
 #include <asm/mce.h>
 #include <asm/trace/irq_vectors.h>
 #include <asm/kexec.h>
+#include <asm/virtext.h>
 
 /*
  *	Some notes on x86 processor bugs affecting SMP operation:

@@ -162,6 +163,7 @@ static int smp_stop_nmi_callback(unsigned int val, struct pt_regs *regs)
 	if (raw_smp_processor_id() == atomic_read(&stopping_cpu))
 		return NMI_HANDLED;
 
+	cpu_emergency_vmxoff();
 	stop_this_cpu(NULL);
 
 	return NMI_HANDLED;

@@ -174,6 +176,7 @@ static int smp_stop_nmi_callback(unsigned int val, struct pt_regs *regs)
 asmlinkage __visible void smp_reboot_interrupt(void)
 {
 	ipi_entering_ack_irq();
+	cpu_emergency_vmxoff();
 	stop_this_cpu(NULL);
 	irq_exit();
 }

Signature · 3 lines

1.8.3.1 

.

Paolo Bonzini

2017-04-18 13:34 UTC #8 lore raw

Ingo, can you put this in tip?

Thanks,

Paolo

Quoted 41 lines · Xishi Qiu

From: Tiantian Feng <[email protected]>

Signed-off-by Tiantian Feng <[email protected]>

Signed-off-by Xishi Qiu <[email protected]>

---
 arch/x86/kernel/smp.c | 3 +++
 1 file changed, 3 insertions(+)

arch/x86/kernel/smp.c +3 -0

index 68f8cc2..b574d55 100644
@@ -33,6 +33,7 @@
 #include <asm/mce.h>
 #include <asm/trace/irq_vectors.h>
 #include <asm/kexec.h>
+#include <asm/virtext.h>
 
 /*
  *	Some notes on x86 processor bugs affecting SMP operation:

@@ -162,6 +163,7 @@ static int smp_stop_nmi_callback(unsigned int val, struct pt_regs *regs)
 	if (raw_smp_processor_id() == atomic_read(&stopping_cpu))
 		return NMI_HANDLED;
 
+	cpu_emergency_vmxoff();
 	stop_this_cpu(NULL);
 
 	return NMI_HANDLED;

@@ -174,6 +176,7 @@ static int smp_stop_nmi_callback(unsigned int val, struct pt_regs *regs)
 asmlinkage __visible void smp_reboot_interrupt(void)
 {
 	ipi_entering_ack_irq();
+	cpu_emergency_vmxoff();
 	stop_this_cpu(NULL);
 	irq_exit();
 }

Ingo Molnar

2017-04-19 08:03 UTC #9 lore raw

Quoted 12 lines · Paolo Bonzini

Ingo, can you put this in tip?
Thanks,
Paolo
Quoted 5 lines · Xishi Qiu
From: Tiantian Feng <[email protected]>
We need to disable VMX on all CPUs before stop cpu when OS panic, otherwisewe risk hanging up the machine, because the CPU ignore INIT signals when VMX is enabled. In kernel mainline this issue existence.

Yes, but the changelog is atrcious:

- title should describe the purpose, not the implementation

- CPU is spelled 'CPU' once, then 'cpu' _in the same sentence_!

- typos

- spelling

- the last sentence doesn't even parse ...

Still it's already at V4 and comes with two signoffs and what amounts to a maintainer Ack??

Thanks,

Ingo

Paolo Bonzini

2017-04-19 08:22 UTC #10 lore raw

Quoted 14 lines

Quoted 7 lines · Xishi Qiu
Quoted 5 lines
From: Tiantian Feng <[email protected]>
We need to disable VMX on all CPUs before stop cpu when OS panic, otherwisewe risk hanging up the machine, because the CPU ignore INIT signals when VMX is enabled. In kernel mainline this issue existence.
Yes, but the changelog is atrcious:
- title should describe the purpose, not the implementation
- CPU is spelled 'CPU' once, then 'cpu' _in the same sentence_!

Quoted 8 lines

- typos
- spelling
- the last sentence doesn't even parse ...
Still it's already at V4 and comes with two signoffs and what amounts to a maintainer Ack??

Well, the v2-v4 were really just about getting the signoffs right. At some point you just get desensitized about the changelog. :(

I'll post v5 with a rewritten commit message.

Paolo

Ingo Molnar

2017-04-19 09:50 UTC #11 lore raw

Quoted 20 lines · Paolo Bonzini

Quoted 20 lines
Quoted 7 lines · Xishi Qiu
Quoted 5 lines
From: Tiantian Feng <[email protected]>
We need to disable VMX on all CPUs before stop cpu when OS panic, otherwisewe risk hanging up the machine, because the CPU ignore INIT signals when VMX is enabled. In kernel mainline this issue existence.
Yes, but the changelog is atrcious:
- title should describe the purpose, not the implementation
- CPU is spelled 'CPU' once, then 'cpu' _in the same sentence_!
- typos
- spelling
- the last sentence doesn't even parse ...

Quoted 7 lines

Quoted 2 lines
Still it's already at V4 and comes with two signoffs and what amounts to a maintainer Ack??
Well, the v2-v4 were really just about getting the signoffs right. At some point you just get desensitized about the changelog. :(
I'll post v5 with a rewritten commit message.

Thanks!

Ingo