[v2 PATCH 1/2] ACPI, APEI, EINJ, new parameter to control trigger action

[v2 PATCH 1/2] ACPI, APEI, EINJ, new parameter to control trigger action

[Chen Gong]

Some APEI firmware implementation will access injected address
specified in param1 to trigger the error when injecting memory
error, which means if one SRAR error is injected, the crash
always happens because it is executed in kernel context. This
new parameter can disable trigger action and control is taken
over by the user. In this way, an SRAR error can happen in user
context instead of crashing the system. This function is highly
depended on BIOS implementation so please ensure you know the
BIOS trigger procedure before you enable this switch.

v2:
  notrigger should be created together with param1/param2

Tested-by: Tony Luck <tony.luck@lintel.com>
Signed-off-by: Chen Gong <gong.chen@linux.intel.com>
---
 drivers/acpi/apei/einj.c |   15 ++++++++++++---
 1 files changed, 12 insertions(+), 3 deletions(-)

diff --git a/drivers/acpi/apei/einj.c b/drivers/acpi/apei/einj.c
index 4ca087d..afc380c 100644
--- a/drivers/acpi/apei/einj.c
+++ b/drivers/acpi/apei/einj.c
@@ -74,6 +74,8 @@ struct vendor_error_type_extension {
 	u8	reserved[3];
 };
 
+static u32 notrigger;
+
 static u32 vendor_flags;
 static struct debugfs_blob_wrapper vendor_blob;
 static char vendor_dev[64];
@@ -496,9 +498,11 @@ static int __einj_error_inject(u32 type, u64 param1, u64 param2)
 	if (rc)
 		return rc;
 	trigger_paddr = apei_exec_ctx_get_output(&ctx);
-	rc = __einj_error_trigger(trigger_paddr, type, param1, param2);
-	if (rc)
-		return rc;
+	if (notrigger == 0) {
+		rc = __einj_error_trigger(trigger_paddr, type, param1, param2);
+		if (rc)
+			return rc;
+	}
 	rc = apei_exec_run_optional(&ctx, ACPI_EINJ_END_OPERATION);
 
 	return rc;
@@ -700,6 +704,11 @@ static int __init einj_init(void)
 					    einj_debug_dir, &error_param2);
 		if (!fentry)
 			goto err_unmap;
+
+		fentry = debugfs_create_x32("notrigger", S_IRUSR | S_IWUSR,
+					    einj_debug_dir, &notrigger);
+		if (!fentry)
+			goto err_unmap;
 	}
 
 	if (vendor_dev[0]) {
-- 
1.7.8.2.302.g17b4e

[PATCH 2/2] Update documentation for parameter *notrigger* in einj.txt

[Chen Gong]

Add description of parameter notrigger in the einj.txt.
One can utilize this new parameter to do some SRAR injection
test. Pay attention, the operation is highly depended on the
BIOS implementation. If no proper BIOS supports it, even if
enabling this parameter, expected result will not happen.

Signed-off-by: Chen Gong <gong.chen@linux.intel.com>
---
 Documentation/acpi/apei/einj.txt |    6 ++++++
 1 files changed, 6 insertions(+), 0 deletions(-)

diff --git a/Documentation/acpi/apei/einj.txt b/Documentation/acpi/apei/einj.txt
index e7cc363..45ff681 100644
--- a/Documentation/acpi/apei/einj.txt
+++ b/Documentation/acpi/apei/einj.txt
@@ -53,6 +53,12 @@ directory apei/einj. The following files are provided.
   This file is used to set the second error parameter value. Effect of
   parameter depends on error_type specified.
 
+- notrigger
+  This file is used to set whether or not exectuing trigger action by
+  the user. When set to 1, it means trigger action is executed by
+  the user, otherwise the trigger action is executed by the kernel.
+  It is useful when injecting one SRAR in the user context situation.
+
 BIOS versions based in the ACPI 4.0 specification have limited options
 to control where the errors are injected.  Your BIOS may support an
 extension (enabled with the param_extension=1 module parameter, or
-- 
1.7.8.2.302.g17b4e

RE: [PATCH 2/2] Update documentation for parameter *notrigger* in einj.txt

[Luck, Tony]

+- notrigger
+  This file is used to set whether or not exectuing trigger action by
+  the user. When set to 1, it means trigger action is executed by
+  the user, otherwise the trigger action is executed by the kernel.
+  It is useful when injecting one SRAR in the user context situation.
+

You hid the "this might not work in your BIOS" part up in the commit
message - which is harder for someone to track down and read than this
Documentation file.
 
How about:

notrigger
The EINJ mechanism is a two step process. First inject the error, then
perform some action to trigger it. Setting "notrigger" to 1 skips the
trigger phase, which *may* allow the user to cause the error in some other
context by a simple access to the cpu, memory location, or device that is
the target of the error injection. Whether this actually works depends
on what operations the BIOS actually includes in the trigger phase.


-Tony

Re: [v2 PATCH 1/2] ACPI, APEI, EINJ, new parameter to control trigger action

[Jiang Liu]

On 02/17/2012 03:11 PM, Chen Gong wrote:
> Some APEI firmware implementation will access injected address
> specified in param1 to trigger the error when injecting memory
> error, which means if one SRAR error is injected, the crash
> always happens because it is executed in kernel context. This
> new parameter can disable trigger action and control is taken
> over by the user. In this way, an SRAR error can happen in user
> context instead of crashing the system. This function is highly
> depended on BIOS implementation so please ensure you know the
> BIOS trigger procedure before you enable this switch.
> 
> v2:
>   notrigger should be created together with param1/param2
> 
> Tested-by: Tony Luck <tony.luck@lintel.com>
> Signed-off-by: Chen Gong <gong.chen@linux.intel.com>
> ---
>  drivers/acpi/apei/einj.c |   15 ++++++++++++---
>  1 files changed, 12 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/acpi/apei/einj.c b/drivers/acpi/apei/einj.c
> index 4ca087d..afc380c 100644
> --- a/drivers/acpi/apei/einj.c
> +++ b/drivers/acpi/apei/einj.c
> @@ -74,6 +74,8 @@ struct vendor_error_type_extension {
>  	u8	reserved[3];
>  };
>  
> +static u32 notrigger;
> +
>  static u32 vendor_flags;
>  static struct debugfs_blob_wrapper vendor_blob;
>  static char vendor_dev[64];
> @@ -496,9 +498,11 @@ static int __einj_error_inject(u32 type, u64 param1, u64 param2)
>  	if (rc)
>  		return rc;
>  	trigger_paddr = apei_exec_ctx_get_output(&ctx);
> -	rc = __einj_error_trigger(trigger_paddr, type, param1, param2);
> -	if (rc)
> -		return rc;
> +	if (notrigger == 0) {
> +		rc = __einj_error_trigger(trigger_paddr, type, param1, param2);
> +		if (rc)
> +			return rc;
> +	}
>  	rc = apei_exec_run_optional(&ctx, ACPI_EINJ_END_OPERATION);
Suggest to reset the notrigger flag to default value (zero) after the injection
operation, so the test script doesn't need to explicitly set the notrigger flag
for each injection operation and it's backward compatible with existing scripts.
It would be better to reset the param2 too for the same reason.

>  
>  	return rc;
> @@ -700,6 +704,11 @@ static int __init einj_init(void)
>  					    einj_debug_dir, &error_param2);
>  		if (!fentry)
>  			goto err_unmap;
> +
> +		fentry = debugfs_create_x32("notrigger", S_IRUSR | S_IWUSR,
> +					    einj_debug_dir, &notrigger);
> +		if (!fentry)
> +			goto err_unmap;
>  	}
>  
>  	if (vendor_dev[0]) {

Re: [v2 PATCH 1/2] ACPI, APEI, EINJ, new parameter to control trigger action

[Chen Gong]

于 2012/2/20 0:21, Jiang Liu 写道:
> On 02/17/2012 03:11 PM, Chen Gong wrote:
>> Some APEI firmware implementation will access injected address
>> specified in param1 to trigger the error when injecting memory
>> error, which means if one SRAR error is injected, the crash
>> always happens because it is executed in kernel context. This
>> new parameter can disable trigger action and control is taken
>> over by the user. In this way, an SRAR error can happen in user
>> context instead of crashing the system. This function is highly
>> depended on BIOS implementation so please ensure you know the
>> BIOS trigger procedure before you enable this switch.
>>
>> v2:
>>    notrigger should be created together with param1/param2
>>
>> Tested-by: Tony Luck<tony.luck@lintel.com>
>> Signed-off-by: Chen Gong<gong.chen@linux.intel.com>
>> ---
>>   drivers/acpi/apei/einj.c |   15 ++++++++++++---
>>   1 files changed, 12 insertions(+), 3 deletions(-)
>>
>> diff --git a/drivers/acpi/apei/einj.c b/drivers/acpi/apei/einj.c
>> index 4ca087d..afc380c 100644
>> --- a/drivers/acpi/apei/einj.c
>> +++ b/drivers/acpi/apei/einj.c
>> @@ -74,6 +74,8 @@ struct vendor_error_type_extension {
>>   	u8	reserved[3];
>>   };
>>
>> +static u32 notrigger;
>> +
>>   static u32 vendor_flags;
>>   static struct debugfs_blob_wrapper vendor_blob;
>>   static char vendor_dev[64];
>> @@ -496,9 +498,11 @@ static int __einj_error_inject(u32 type, u64 param1, u64 param2)
>>   	if (rc)
>>   		return rc;
>>   	trigger_paddr = apei_exec_ctx_get_output(&ctx);
>> -	rc = __einj_error_trigger(trigger_paddr, type, param1, param2);
>> -	if (rc)
>> -		return rc;
>> +	if (notrigger == 0) {
>> +		rc = __einj_error_trigger(trigger_paddr, type, param1, param2);
>> +		if (rc)
>> +			return rc;
>> +	}
>>   	rc = apei_exec_run_optional(&ctx, ACPI_EINJ_END_OPERATION);
> Suggest to reset the notrigger flag to default value (zero) after the injection
> operation, so the test script doesn't need to explicitly set the notrigger flag
> for each injection operation and it's backward compatible with existing scripts.
> It would be better to reset the param2 too for the same reason.
>

I ever thought about this question, but it looks current implementation is
established by usage so I don't think it needs to be fixed, otherwise it has
strong requirement. As for the scripts, I assume you point to mce-test, I will
update them after this patch is merged into the kernel.

Thx for your comments.
--
To unsubscribe from this list: send the line "unsubscribe linux-acpi" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

RE: [v2 PATCH 1/2] ACPI, APEI, EINJ, new parameter to control trigger action

[Luck, Tony]

> Suggest to reset the notrigger flag to default value (zero) after the injection
> operation, so the test script doesn't need to explicitly set the notrigger flag
> for each injection operation and it's backward compatible with existing scripts.
> It would be better to reset the param2 too for the same reason.

That might cause more confusion. We haven't documented the behavior of the
other files here after an error is injected ... but their actual behavior has
always been that values written are sticky. I.e. you can write a script to
inject three errors to the same page like this:

echo 0x8 > error_type
echo 0x12340000 > param1
echo 0xfffffffffffff000 > param2

echo 1 > error_inject
sleep 1
echo 1 > error_inject
sleep 1
echo 1 > error_inject

so I think that it might be surprising if "notrigger" was not sticky
like everything else.

-Tony