[PATCH] ACPI/APEI: revert "accept validly sized ERST on Intel systems only for now"

[PATCH] ACPI/APEI: revert "accept validly sized ERST on Intel systems only for now"

[Jan Beulich]

[-- Attachment #1: Type: text/plain, Size: 991 bytes --]

With the recent two fixes to ERST handling, this should no longer be
necessary.

Signed-off-by: Jan Beulich <jbeulich@suse.com>

--- a/xen/drivers/acpi/apei/erst.c
+++ b/xen/drivers/acpi/apei/erst.c
@@ -738,19 +738,6 @@ static int __init erst_check_table(struc
 
 	switch (erst_tab->header_length) {
 	case sizeof(*erst_tab) - sizeof(erst_tab->header):
-#ifdef CONFIG_X86
-		/* XXX
-		 * While the rest of the ERST code appears to work on Intel
-		 * systems with properly sized tables, various AMD systems
-		 * appear to get hung (at boot time) by allowing this. Until
-		 * someone with access to suitable hardware can debug this,
-		 * disable the rest of the code by considering this case
-		 * invalid.
-		 */
-		if (boot_cpu_data.x86_vendor != X86_VENDOR_INTEL)
-			return -EINVAL;
-		/* fall through */
-#endif
 	/*
 	 * While invalid per specification, there are (early?) systems
 	 * indicating the full header size here, so accept that value too.




[-- Attachment #2: ACPI-APEI-revert-26074.patch --]
[-- Type: text/plain, Size: 1064 bytes --]

ACPI/APEI: revert "accept validly sized ERST on Intel systems only for now"

With the recent two fixes to ERST handling, this should no longer be
necessary.

Signed-off-by: Jan Beulich <jbeulich@suse.com>

--- a/xen/drivers/acpi/apei/erst.c
+++ b/xen/drivers/acpi/apei/erst.c
@@ -738,19 +738,6 @@ static int __init erst_check_table(struc
 
 	switch (erst_tab->header_length) {
 	case sizeof(*erst_tab) - sizeof(erst_tab->header):
-#ifdef CONFIG_X86
-		/* XXX
-		 * While the rest of the ERST code appears to work on Intel
-		 * systems with properly sized tables, various AMD systems
-		 * appear to get hung (at boot time) by allowing this. Until
-		 * someone with access to suitable hardware can debug this,
-		 * disable the rest of the code by considering this case
-		 * invalid.
-		 */
-		if (boot_cpu_data.x86_vendor != X86_VENDOR_INTEL)
-			return -EINVAL;
-		/* fall through */
-#endif
 	/*
 	 * While invalid per specification, there are (early?) systems
 	 * indicating the full header size here, so accept that value too.

[-- Attachment #3: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [PATCH] ACPI/APEI: revert "accept validly sized ERST on Intel systems only for now"

[Andrew Cooper]

On 26/03/2013 09:26, Jan Beulich wrote:
> With the recent two fixes to ERST handling, this should no longer be
> necessary.
>
> Signed-off-by: Jan Beulich <jbeulich@suse.com>

There are still no failures from our regression testing, so a tentative

Tested-by: Andrew Cooper <andrew.cooper3@citrix.com>

>
> --- a/xen/drivers/acpi/apei/erst.c
> +++ b/xen/drivers/acpi/apei/erst.c
> @@ -738,19 +738,6 @@ static int __init erst_check_table(struc
>  
>  	switch (erst_tab->header_length) {
>  	case sizeof(*erst_tab) - sizeof(erst_tab->header):
> -#ifdef CONFIG_X86
> -		/* XXX
> -		 * While the rest of the ERST code appears to work on Intel
> -		 * systems with properly sized tables, various AMD systems
> -		 * appear to get hung (at boot time) by allowing this. Until
> -		 * someone with access to suitable hardware can debug this,
> -		 * disable the rest of the code by considering this case
> -		 * invalid.
> -		 */
> -		if (boot_cpu_data.x86_vendor != X86_VENDOR_INTEL)
> -			return -EINVAL;
> -		/* fall through */
> -#endif
>  	/*
>  	 * While invalid per specification, there are (early?) systems
>  	 * indicating the full header size here, so accept that value too.
>
>
>

Re: [PATCH] ACPI/APEI: revert "accept validly sized ERST on Intel systems only for now"

[Jan Beulich]

>>> On 26.03.13 at 15:24, Andrew Cooper <andrew.cooper3@citrix.com> wrote:
> On 26/03/2013 09:26, Jan Beulich wrote:
>> With the recent two fixes to ERST handling, this should no longer be
>> necessary.
>>
>> Signed-off-by: Jan Beulich <jbeulich@suse.com>
> 
> There are still no failures from our regression testing, so a tentative
> 
> Tested-by: Andrew Cooper <andrew.cooper3@citrix.com>

Thanks for confirming!

Jan

Re: [PATCH] ACPI/APEI: revert "accept validly sized ERST on Intel systems only for now"

[Ian Jackson]

Jan Beulich writes ("[PATCH] ACPI/APEI: revert "accept validly sized ERST on Intel systems only for now""):
> With the recent two fixes to ERST handling, this should no longer be
> necessary.
...

Which recent two fixes, JOOI ?  I did a recent test of reverting this
and found I still had some problems but it
that was based on 37d930430c2f2ef4dedf27deb404d2fe2602c039.

Thanks,
Ian.

Re: [PATCH] ACPI/APEI: revert "accept validly sized ERST on Intel systems only for now"

[Jan Beulich]

>>> On 26.03.13 at 17:37, Ian Jackson <Ian.Jackson@eu.citrix.com> wrote:
> Jan Beulich writes ("[PATCH] ACPI/APEI: revert "accept validly sized ERST on 
> Intel systems only for now""):
>> With the recent two fixes to ERST handling, this should no longer be
>> necessary.
> ...
> 
> Which recent two fixes, JOOI ?  I did a recent test of reverting this
> and found I still had some problems but it
> that was based on 37d930430c2f2ef4dedf27deb404d2fe2602c039.

0611689 and 72af01b, both of which came after said commit.

Jan

Re: [PATCH] ACPI/APEI: revert "accept validly sized ERST on Intel systems only for now"

[Jan Beulich]

>>> On 26.03.13 at 17:37, Ian Jackson <Ian.Jackson@eu.citrix.com> wrote:
> Jan Beulich writes ("[PATCH] ACPI/APEI: revert "accept validly sized ERST on 
> Intel systems only for now""):
>> With the recent two fixes to ERST handling, this should no longer be
>> necessary.
> ...
> 
> Which recent two fixes, JOOI ?  I did a recent test of reverting this
> and found I still had some problems but it
> that was based on 37d930430c2f2ef4dedf27deb404d2fe2602c039.

Btw., do you prefer to retest the revert on top of these before
I commit it, or would you agree that Andrew's testing warrants
committing this and then wait for the results of a regular testing
round?

Jan

Re: [PATCH] ACPI/APEI: revert "accept validly sized ERST on Intel systems only for now"

[Ian Jackson]

Jan Beulich writes ("Re: [PATCH] ACPI/APEI: revert "accept validly sized ERST on Intel systems only for now""):
> Btw., do you prefer to retest the revert on top of these before
> I commit it, or would you agree that Andrew's testing warrants
> committing this and then wait for the results of a regular testing
> round?

Please go ahead and commit the revert.  If it breaks we can revert the
reversions of the revert :-).

Ian.