[LinuxBIOS] INIT detected.

[Steve Isaacs]

Is this message normal or does it indicate a problem that needs to be
corrected?

INIT detected from  ---- {APICID = 00 NODEID = 00 COREID = 00} ---

Issuing SOFT_RESET...

Looking at the code this is emitted in model_fxx/init_cpus.c when
cpu_init_detectedx is non-zero.

Thanks,

Steve

[Marc Jones]

Steve Isaacs wrote:
> Is this message normal or does it indicate a problem that needs to be
> corrected?
>
> INIT detected from  ---- {APICID = 00 NODEID = 00 COREID = 00} ---
>
> Issuing SOFT_RESET...
>
> Looking at the code this is emitted in model_fxx/init_cpus.c when
> cpu_init_detectedx is non-zero.
>
> Thanks,
>
> Steve

Steve,

This could indicate a real problem. It means that the core was 
unexpectedly reset at some point. This is checked in cache_as_ram.inc. 
Look for 	/* check if cpu_init_detected */.

If the reset is intentional, the reset code should set/clear the bit in 
the msr by calling set_bios_reset();

I have seen unintentional resets caused by AP init and/or CAR and memory 
init issues. You can get into some funny race conditions with AP and BSP 
using the same CAR location but that is just a guess.

I hope that gives you some ideas about what might be going wrong and 
where to start looking.

Marc

[Steve Isaacs]

On Fri, 2008-01-11 at 13:57 -0700, Marc Jones wrote:
>
> This could indicate a real problem. It means that the core was 
> unexpectedly reset at some point. This is checked in cache_as_ram.inc. 
> Look for 	/* check if cpu_init_detected */.

Yes, as it turns out this is a real problem. The southbridge for some
reason unknown to me is sending the INIT message when upstream messages
are enabled. I'm currently working with the chip supplier on that one.
> I have seen unintentional resets caused by AP init and/or CAR and memory 
> init issues. You can get into some funny race conditions with AP and BSP 
> using the same CAR location but that is just a guess.

This is what I was initially suspecting when I realized the INIT may be
a real problem. The INIT was causing other delays that eventually caused
a timeout on the AP in the fidvid code (previous post). I now realize
(just a short while ago) that with fidvid I've been chasing a symptom,
not another problem.

Thanks,

Steve