[Skiboot] [PATCH v2 01/12] hdat: SPIRA-H/S changes for P9

Vasant Hegde hegdevasant at linux.vnet.ibm.com
Tue Jan 3 20:06:02 AEDT 2017


On 12/20/2016 09:02 AM, Stewart Smith wrote:
> Vasant Hegde <hegdevasant at linux.vnet.ibm.com> writes:
>> Add new ntuples to SPIRA-H/S structure.
>>
>> We have already switched to SPIRA-H/S from legacy SPIRA. Hence
>> I've not made any update to legacy SPIRA structure.
>>
>> Signed-off-by: Vasant Hegde <hegdevasant at linux.vnet.ibm.com>
>> ---
>>   hdata/spira.h | 14 +++++++++-----
>>   1 file changed, 9 insertions(+), 5 deletions(-)
>>
>> diff --git a/hdata/spira.h b/hdata/spira.h
>> index eabf7f9..b4facb5 100644
>> --- a/hdata/spira.h
>> +++ b/hdata/spira.h
>> @@ -103,6 +103,7 @@ struct spirah_ntuples {
>>   	struct spira_ntuple	mdump_src;	/* 0x0a0 */
>>   	struct spira_ntuple	mdump_dst;	/* 0x0c0 */
>>   	struct spira_ntuple	mdump_res;	/* 0x0e0 */
>> +	struct spira_ntuple	proc_dump_area;	/* 0x100 */
>>   };
>>
>>   struct spirah {
>> @@ -110,7 +111,7 @@ struct spirah {
>>   	struct HDIF_idata_ptr	ntuples_ptr;
>>   	__be64			pad;
>>   	struct spirah_ntuples	ntuples;
>> -	u8			reserved[0x100];
>> +	u8			reserved[0xE0];
>
> In v10.3g, this says still 0x0100 reserved, which has effectively bumped
> the size of the structure?

I think they forgot to update this field. I've made adjustment to make sure we 
don't bump up the size.  Anyway its reserved space. Hence we are fine here. I 
will talk to Venky on this.


>
>>   } __packed __align(0x100);
>>
>>   extern struct spirah spirah;
>> @@ -119,7 +120,8 @@ extern struct spirah spirah;
>>   #define SPIRAS_HDIF_SIG		"SPIRAS"
>>
>>   /* First version on 810 release */
>> -#define SPIRAS_VERSION		0x40
>> +#define SPIRAS_VERSION_P8	0x40
>> +#define SPIRAS_VERSION_P9	0x50
>>
>>   /* N-tuples in SPIRAS */
>>   #define SPIRAS_NTUPLES_COUNT	0x10
>> @@ -141,15 +143,17 @@ struct spiras_ntuples {
>>   	struct spira_ntuple	pcia;			/* 0x1c0 */
>>   	struct spira_ntuple	proc_chip;		/* 0x1e0 */
>>   	struct spira_ntuple	hs_data;		/* 0x200 */
>> -	struct spira_ntuple	ipmi_sensor;		/* 0x220 */
>> -} __packed __align(0x100);
>> +	struct spira_ntuple	hs_runtime;		/* 0x220 */
>
> The ipmi_sensor has moved?

Yes. Not sure when and why it got changed. Anyway we were not using this ntuple. 
Hence I just updated.

>
> should be hbrt rather than hs?

Makes sense. Updated as hbrt_data.

-Vasant



More information about the Skiboot mailing list