[Skiboot] [PATCH v2 2/3] opal: Check Core FIRs to find the reason for Malfunction Alert.

Mahesh Jagannath Salgaonkar mahesh at linux.vnet.ibm.com
Thu Feb 12 20:08:31 AEDT 2015

On 02/12/2015 11:24 AM, Stewart Smith wrote:
> Mahesh J Salgaonkar <mahesh at linux.vnet.ibm.com> writes:
>> @@ -337,7 +454,7 @@ static int64_t opal_handle_hmi(void)
>>  	struct OpalHMIEvent hmi_evt;
>>  	memset(&hmi_evt, 0, sizeof(struct OpalHMIEvent));
>> -	hmi_evt.version = OpalHMIEvt_V1;
>> +	hmi_evt.version = OpalHMIEvt_V2;
>>  	lock(&hmi_lock);
>>  	hmer = mfspr(SPR_HMER);		/* Get HMER register value */
> It would be better if you only sent v2 in the event of something not
> supported by V1.

Wouldn't that make code bit messy ? Also, would it be correct for OPAL
layer to send two different versions of HMI event depending on what it
contains ? V2 is a super set of v1, and can contain data pertaining to
v1 and v2 both. V2 is backward compatible. Older Linux kernel receiving
V2 event (containing only v1 data) can still print all the info that
pertains to V1.


More information about the Skiboot mailing list