[Skiboot] [PATCH 1/3] init, occ: Initialise OCC earlier on POWER9

[Andrew Donnellan]

On 20/06/18 03:50, Shilpasri G Bhat wrote:
> Should this check be based on if we have FSP or not?
> 
> Even in P9, OCC is not started until OPAL acknowledges a dummy OCC load request
> sent by FSP. I tested your patch on a ZZ where OCC init took 11700 ms
> 
> [  165.184882884,6] OCC: Got OCC Load message, scope=0x2 dbob=0x0 seq=0x1c
> [  178.266409596,7] OCC: Chip 00 Data (0000203ffd8e2000) = 01900160361e0000
> [  178.266410866,7] OCC: Chip 08 Data (0000203ffdce2000) = 01900060361e0000
> [  178.266412093,5] OCC: All Chip Rdy after 11700 ms
> 
> On both P8 and P9,
> OpenPOWER : OCC is up before OPAL boots
> FSP : OCC is started after OPAL boots.
> 
> Not sure if this check works for you..

That's fine at least for witherspoon, which is currently the only 
platform where this is an issue. If we ever have another FSP platform 
which requires presence detection via the OCC this might be a problem, 
but I highly highly doubt that will ever happen.

I'll spin a v2 with the new check.

Thanks for the feedback!


Andrew


> 
> Thanks and Regards,
> Shilpa
> 
>>   
>>   	if (!is_reboot) {
>>   		/* We wait for the nvram read to complete here so we can
>> @@ -1079,6 +1080,16 @@ void __noreturn __nomcount main_cpu_entry(const void *fdt)
>>   
>>   	op_display(OP_LOG, OP_MOD_INIT, 0x0002);
>>   
>> +	/*
>> +	 * On POWER9, we need to initialise the OCC before the NPU to
>> +	 * facilitate NVLink/OpenCAPI presence detection, so we set it
>> +	 * up as early as possible. On POWER8, Hostboot starts booting
>> +	 * the OCC later, so we delay OCC initialisation as late as
>> +	 * possible to give it the maximum time to boot up.
>> +	 */
>> +	if (proc_gen >= proc_gen_p9)
>> +		occ_pstates_init();
>> +
>>   	pci_nvram_init();
>>   
>>   	preload_io_vpd();
>>

-- 
Andrew Donnellan              OzLabs, ADL Canberra
andrew.donnellan at au1.ibm.com  IBM Australia Limited