Re: [RESEND PATCH v1] MIPS: uasm: false warning on use of uasm_i_lui()

[Jim Quinlan]

On Thu, Sep 10, 2020 at 5:37 AM Thomas Bogendoerfer
<tsbogend@xxxxxxxxxxxxxxxx> wrote:
>
> On Tue, Sep 08, 2020 at 12:45:06PM -0400, Jim Quinlan wrote:
> > Currently, the example uasm code
> >
> >       uasm_i_lui(p, tmp, 0xa000);
> >
> > issues a warning at Linux boot when the code is "assembled".  This is
> > because the "lui" instruction is defined by the macro "Ip_u1s2(_lui)" -- I
> > believe it should be Ip_u1u2(_lui) -- and its definition is associated with
> > the SIMM macro -- I believe it should be the UIMM macro.  The current code
> > takes a 32bit number and checks that it can be converted to a 16bit signed
> > immediate.  This check fails of course for an immediate such as 0x0000a000.
>
> IMHO SIMM is correct as the upper 16bits will be sign extended on 64bit
> CPUs.
>
Hi Thomas,

Got it.

Thanks,
Jim

> Your example looks like to try to load a KSEG1 address, so just use
>
> uasm_i_lui(p, tmp, uasm_rel_hi(0xa0000000));
>
> which even makes it clearer, what this is about.
>
> Thomas.
>
> --
> Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
> good idea.                                                [ RFC1925, 2.3 ]