Re: [PATCH RFCv3 08/14] arm64: introduce aarch64_insn_gen_movewide()

[Will Deacon <will.deacon@arm.com>]

On Wed, Jul 16, 2014 at 11:04:22PM +0100, Zi Shen Lim wrote:
> On Wed, Jul 16, 2014 at 05:17:15PM +0100, Will Deacon wrote:
> > On Tue, Jul 15, 2014 at 07:25:06AM +0100, Zi Shen Lim wrote:
> > > Introduce function to generate move wide (immediate) instructions.
> > 
> > [...]
> > 
> > > +u32 aarch64_insn_gen_movewide(enum aarch64_insn_register dst,
> > > +			      int imm, int shift,
> > > +			      enum aarch64_insn_variant variant,
> > > +			      enum aarch64_insn_movewide_type type)
> > > +{
> > > +	u32 insn;
> > > +
> > > +	switch (type) {
> > > +	case AARCH64_INSN_MOVEWIDE_ZERO:
> > > +		insn = aarch64_insn_get_movz_value();
> > > +		break;
> > > +	case AARCH64_INSN_MOVEWIDE_KEEP:
> > > +		insn = aarch64_insn_get_movk_value();
> > > +		break;
> > > +	case AARCH64_INSN_MOVEWIDE_INVERSE:
> > > +		insn = aarch64_insn_get_movn_value();
> > > +		break;
> > > +	default:
> > > +		BUG_ON(1);
> > > +	}
> > > +
> > > +	BUG_ON(imm < 0 || imm > 65535);
> > 
> > Do this check with masking instead?
> 
> Ok, if you prefer, I can change it to:
> 
> 	BUG_ON(imm & ~GENMASK(15, 0));

Sure, that or use a named constant for the upper-bound (SZ_64K - 1).

> > > +	switch (variant) {
> > > +	case AARCH64_INSN_VARIANT_32BIT:
> > > +		BUG_ON(shift != 0 && shift != 16);
> > > +		break;
> > > +	case AARCH64_INSN_VARIANT_64BIT:
> > > +		insn |= BIT(31);
> > > +		BUG_ON(shift != 0 && shift != 16 && shift != 32 &&
> > > +		       shift != 48);
> > 
> > Would be neater as a nested switch, perhaps? If you reorder the
> > outer-switch, you could probably fall-through too and combine the shift
> > checks.
> 
> Not sure I picture what you had in mind... I couldn't come up with a
> neater version with the properties you described.
> 
> The alternative I had was using masks instead of integer values, but
> one could argue that while neater, it could also be harder to read:
> 
> 	switch (variant) {
> 	case AARCH64_INSN_VARIANT_32BIT:
> 		BUG_ON(shift & ~BIT(4));
> 		break;
> 	case AARCH64_INSN_VARIANT_64BIT:
> 		insn |= BIT(31);
> 		BUG_ON(shift & ~GENMASK(5, 4));
> 	...

I was thinking of using nested switches, but that doesn't fall out like I
hoped. How about:

	switch (variant) {
	case AARCH64_INSN_VARIANT_64BIT:
		BUG_ON(shift != 32 && shift != 48);
	case AARCH64_INSN_VARIANT_32BIT:
		BUG_ON(shift != 0 && shift != 16);
	};

?

Will