Skip to contentSkip to conversation

x86: fix dummy apic read/write warning

PATCH In review Yinghai Lu 3 participants 4 messages Jun 6, 2009 – Jun 7, 2009 1 message not in the archive x86

lore
Share & tools
lore

A message that is not in the archive (sent off-list or not yet received): [email protected]

Yinghai Lu

author
#1 lore raw

PATCH 1/1 x86: fix dummy apic read/write warning

ingo got

[    0.000000] Using APIC driver default
[    0.000000] SMP: Allowing 1 CPUs, 0 hotplug CPUs
[    0.000000] Local APIC disabled by BIOS -- you can enable it with "lapic"
[    0.000000] APIC: disable apic facility
[    0.000000] ------------[ cut here ]------------
[    0.000000] WARNING: at arch/x86/kernel/apic/apic.c:254 native_apic_read_dummy+0x2d/0x3b()
[    0.000000] Hardware name: HP OmniBook PC

we have wrong check in dummy read and write.
[we only test on system that have APIC with "disableapic"...]

---
 arch/x86/kernel/apic/apic.c |    4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
arch/x86/kernel/apic/apic.c +2 -2
Index: linux-2.6/arch/x86/kernel/apic/apic.c
===================================================================
@@ -246,12 +246,12 @@ static int modern_apic(void)
  */
 static void native_apic_write_dummy(u32 reg, u32 v)
 {
-	WARN_ON_ONCE((cpu_has_apic || !disable_apic));
+	WARN_ON_ONCE((cpu_has_apic && !disable_apic));
 }
 
 static u32 native_apic_read_dummy(u32 reg)
 {
-	WARN_ON_ONCE((cpu_has_apic || !disable_apic));
+	WARN_ON_ONCE((cpu_has_apic && !disable_apic));
 	return 0;
 }
 
#2 lore raw

[Yinghai Lu - Sat, Jun 06, 2009 at 03:16:38PM -0700]

| ingo got
| [    0.000000] Using APIC driver default
| [    0.000000] SMP: Allowing 1 CPUs, 0 hotplug CPUs
| [    0.000000] Local APIC disabled by BIOS -- you can enable it with "lapic"
| [    0.000000] APIC: disable apic facility
| [    0.000000] ------------[ cut here ]------------
| [    0.000000] WARNING: at arch/x86/kernel/apic/apic.c:254 native_apic_read_dummy+0x2d/0x3b()
| [    0.000000] Hardware name: HP OmniBook PC
| 
| we have wrong check in dummy read and write.
| [we only test on system that have APIC with "disableapic"...]
| 
| Signed-off-by: yinghai Lu <[email protected]>
| 
| ---
|  arch/x86/kernel/apic/apic.c |    4 ++--
|  1 file changed, 2 insertions(+), 2 deletions(-)
| 
| Index: linux-2.6/arch/x86/kernel/apic/apic.c
| ===================================================================
| --- linux-2.6.orig/arch/x86/kernel/apic/apic.c
| +++ linux-2.6/arch/x86/kernel/apic/apic.c
| @@ -246,12 +246,12 @@ static int modern_apic(void)
|   */
|  static void native_apic_write_dummy(u32 reg, u32 v)
|  {
| -	WARN_ON_ONCE((cpu_has_apic || !disable_apic));
| +	WARN_ON_ONCE((cpu_has_apic && !disable_apic));
|  }
|  
|  static u32 native_apic_read_dummy(u32 reg)
|  {
| -	WARN_ON_ONCE((cpu_has_apic || !disable_apic));
| +	WARN_ON_ONCE((cpu_has_apic && !disable_apic));
|  	return 0;
|  }
|  
| 

No Yinghai, this is wrong. native_apic_write_dummy is not admissible to be called after apic->write has been NOP'ified.

If a site calls for apic->write after apic_disable() it's a plain *bug* in caller. So we should continute warning if cpu_has_apic _or_ !disable_apic.

Though since for SMP compiled kernel we still rely on apic->read the snippet is to be changed to (cpu_has_apic && !disable_apic) indeed.

-- Cyrill

#3 lore raw
Quoted 40 lines · Cyrill Gorcunov

[Yinghai Lu - Sat, Jun 06, 2009 at 03:16:38PM -0700]

| ingo got
| [    0.000000] Using APIC driver default
| [    0.000000] SMP: Allowing 1 CPUs, 0 hotplug CPUs
| [    0.000000] Local APIC disabled by BIOS -- you can enable it with "lapic"
| [    0.000000] APIC: disable apic facility
| [    0.000000] ------------[ cut here ]------------
| [    0.000000] WARNING: at arch/x86/kernel/apic/apic.c:254 native_apic_read_dummy+0x2d/0x3b()
| [    0.000000] Hardware name: HP OmniBook PC
|
| we have wrong check in dummy read and write.
| [we only test on system that have APIC with "disableapic"...]
|
| Signed-off-by: yinghai Lu <[email protected]>
|
| ---
|  arch/x86/kernel/apic/apic.c |    4 ++--
|  1 file changed, 2 insertions(+), 2 deletions(-)
|
| Index: linux-2.6/arch/x86/kernel/apic/apic.c
| ===================================================================
| --- linux-2.6.orig/arch/x86/kernel/apic/apic.c
| +++ linux-2.6/arch/x86/kernel/apic/apic.c
| @@ -246,12 +246,12 @@ static int modern_apic(void)
|   */
|  static void native_apic_write_dummy(u32 reg, u32 v)
|  {
| -     WARN_ON_ONCE((cpu_has_apic || !disable_apic));
| +     WARN_ON_ONCE((cpu_has_apic && !disable_apic));
|  }
|
|  static u32 native_apic_read_dummy(u32 reg)
|  {
| -     WARN_ON_ONCE((cpu_has_apic || !disable_apic));
| +     WARN_ON_ONCE((cpu_has_apic && !disable_apic));
|       return 0;
|  }
|
|
Quoted 6 lines

No Yinghai, this is wrong. native_apic_write_dummy is not admissible to be called after apic->write has been NOP'ified.

If a site calls for apic->write after apic_disable() it's a plain *bug* in caller. So we should continute warning if cpu_has_apic _or_ !disable_apic.

ok, just change it to

     WARN_ON_ONCE(1);

there.

Quoted 4 lines

Though since for SMP compiled kernel we still rely on apic->read the snippet is to be changed to (cpu_has_apic && !disable_apic) indeed.

YH

#4 lore raw

[Yinghai Lu - Sun, Jun 07, 2009 at 12:28:07PM -0700]

| On Sun, Jun 7, 2009 at 12:01 AM, Cyrill Gorcunov<[email protected]> wrote:
| > [Yinghai Lu - Sat, Jun 06, 2009 at 03:16:38PM -0700]
| > | ingo got
| > | [    0.000000] Using APIC driver default
| > | [    0.000000] SMP: Allowing 1 CPUs, 0 hotplug CPUs
| > | [    0.000000] Local APIC disabled by BIOS -- you can enable it with "lapic"
| > | [    0.000000] APIC: disable apic facility
| > | [    0.000000] ------------[ cut here ]------------
| > | [    0.000000] WARNING: at arch/x86/kernel/apic/apic.c:254 native_apic_read_dummy+0x2d/0x3b()
| > | [    0.000000] Hardware name: HP OmniBook PC
| > |
| > | we have wrong check in dummy read and write.
| > | [we only test on system that have APIC with "disableapic"...]
| > |
| > | Signed-off-by: yinghai Lu <[email protected]>
| > |
| > | ---
| > |  arch/x86/kernel/apic/apic.c |    4 ++--
| > |  1 file changed, 2 insertions(+), 2 deletions(-)
| > |
| > | Index: linux-2.6/arch/x86/kernel/apic/apic.c
| > | ===================================================================
| > | --- linux-2.6.orig/arch/x86/kernel/apic/apic.c
| > | +++ linux-2.6/arch/x86/kernel/apic/apic.c
| > | @@ -246,12 +246,12 @@ static int modern_apic(void)
| > |   */
| > |  static void native_apic_write_dummy(u32 reg, u32 v)
| > |  {
| > | -     WARN_ON_ONCE((cpu_has_apic || !disable_apic));
| > | +     WARN_ON_ONCE((cpu_has_apic && !disable_apic));
| > |  }
| > |
| > |  static u32 native_apic_read_dummy(u32 reg)
| > |  {
| > | -     WARN_ON_ONCE((cpu_has_apic || !disable_apic));
| > | +     WARN_ON_ONCE((cpu_has_apic && !disable_apic));
| > |       return 0;
| > |  }
| > |
| > |
| >
| > No Yinghai, this is wrong. native_apic_write_dummy is not
| > admissible to be called after apic->write has been NOP'ified.
| >
| > If a site calls for apic->write after apic_disable() it's
| > a plain *bug* in caller. So we should continute warning if
| > cpu_has_apic _or_ !disable_apic.
| 
| ok, just change it to
| 
|      WARN_ON_ONCE(1);
| 
| there.

There was some scenario where we had legitime apic->write on SMP kernel when cpu goes to halt. Will recheck tomorrow. That is why we added conditional WARN to eliminate uglifying by additional checks (ie we know that apic->write happens but it's nothing serious). Thanks Yinghai!

| 
| >
| > Though since for SMP compiled kernel we still rely on
| > apic->read the snippet is to be changed to (cpu_has_apic && !disable_apic)
| > indeed.
| 
| YH
| 

-- Cyrill