[PATCH] ARCv2: intc: untangle SMP, MCIP and IDU

[Alexey Brodkin]

Hi Vineet,

On Thu, 2016-10-06 at 10:10 -0700, Vineet Gupta wrote:
> On 10/06/2016 02:10 AM, Alexey Brodkin wrote:
> > 
> > > 
> > > +struct mcip_bcr {
> > > +#ifdef CONFIG_CPU_BIG_ENDIAN
> > > +		unsigned int pad3:8,
> > > +			     idu:1, llm:1, num_cores:6,
> > > +			     iocoh:1,  gfrc:1, dbg:1, pad2:1,
> > > +			     msg:1, sem:1, ipi:1, pad:1,
> > > +			     ver:8;
> > > +#else
> > > +		unsigned int ver:8,
> > > +			     pad:1, ipi:1, sem:1, msg:1,
> > > +			     pad2:1, dbg:1, gfrc:1, iocoh:1,
> > > +			     num_cores:6, llm:1, idu:1,
> > > +			     pad3:8;
> > > +#endif
> > > +};
> > 
> > IMHO we should stop using this kind of constructions because they
> > are ugly and what's more important not portable.
> 
> They are ugly I agree - but not portable - really ? The whole point is to make
> this work on BE w/o changing the src code - this details remains hidden in an
> obscure header.

That's what I learned the hard way.
At least I was beaten a couple of times yet in both Linux kernel community and
U-Boot
one.

> > Even though we have it now working for both LE and BE configurations
> > it won't work for 64-bit cores. We'll need to add ifdeffed 32-bit paddings
> > then which will make that construction even more ugly.
> 
> When we get to 64-bit a lot things would have to change - and possibly the aux reg
> layout. There is no way to make this exact code 64-bit ready !

Probably but as of now I believe use of offsets for bit-fields is the safest
approach which makes code ugly as well but at least that way we reduce risk
of erroneous copy-paste in "mirrored" part.

-Alexey