[Skiboot] [PATCH v2 02/17] hdata: Fix MDST structure

[Vasant Hegde]

On 05/30/2018 12:07 PM, Stewart Smith wrote:
> Vasant Hegde <hegdevasant at linux.vnet.ibm.com> writes:
>> --- a/include/opal-dump.h
>> +++ b/include/opal-dump.h
>> @@ -21,10 +21,13 @@
>>   /*
>>    * Dump region ids
>>    *
>> - * 0x01 - 0x7F : OPAL
>> + * 0x00 - 0x00 : CPU data
>> + * 0x01 - 0x3F : OPAL
>> + * 0x40 - 0x79 : Reserved for future use
>>    * 0x80 - 0xFF : Kernel
>>    *
>>    */
>> +#define DUMP_REGION_CPU_DATA		0x00
>>   #define DUMP_REGION_OPAL_START		0x01
>>   #define DUMP_REGION_OPAL_END		0x7F
>>   #define DUMP_REGION_HOST_START		OPAL_DUMP_REGION_HOST_START
>> @@ -33,16 +36,23 @@
>>   #define DUMP_REGION_CONSOLE	0x01
>>   #define DUMP_REGION_HBRT_LOG	0x02
>>
>> +/* Mainstore memory to be captured by FSP SYSDUMP */
>> +#define DUMP_TYPE_SYSDUMP		0xF5
>> +/* Mainstore memory to preserve during IPL */
>> +#define DUMP_TYPE_FADUMP		0x00
>> +
>>   /*
>> - * Sapphire Memory Dump Source Table
>> + *  Memory Dump Source Table
>>    *
>>    * Format of this table is same as Memory Dump Source Table (MDST)
>>    * defined in HDAT spec.
>>    */
>>   struct mdst_table {
>>   	__be64	addr;
>> -	__be32	type; /* DUMP_REGION_* */
>> +	uint8_t	data_region;	/* DUMP_REGION_* */
>> +	uint8_t dump_type;	/* DUMP_TYPE_* */
>> +	__be16	reserved;
>>   	__be32	size;
>> -};
>> +} __packed;

Stewart,

> 
> Hrm... so this has changed over what we've had on FSP systems in the
> past? Or have we always been wrong? I guess we do get the log in a
> sysdump today, but will that break on existing systems with this patch?
> 


It won't break existing SYSDUMP.. as FSP ignores these fields. They just care 
about address and size field from MDST. Hence we are good here. May be I will 
update description with these details..

-Vasant