Re: [PATCH] x86/ioemul: Rewrite stub generation

[Roger Pau Monné <roger.pau@xxxxxxxxxx>]

On Mon, Apr 27, 2020 at 05:18:52PM +0100, Andrew Cooper wrote:
> On 27/04/2020 16:18, Roger Pau Monné wrote:
> > On Mon, Apr 27, 2020 at 01:20:41PM +0100, Andrew Cooper wrote:
> >> +    /* Helpers - Read outer scope but only modify p. */
> >> +#define APPEND_BUFF(b) ({ memcpy(p, b, sizeof(b)); p += sizeof(b); })
> >> +#define APPEND_CALL(f)                                                  \
> >> +    ({                                                                  \
> >> +        long disp = (long)(f) - (stub_va + p - ctxt->io_emul_stub + 5); \
> >> +        BUG_ON((int32_t)disp != disp);                                  \
> > I'm not sure I get the point of using signed integers instead of
> > unsigned ones, AFAICT you just want to check that the displacement is
> > < 4GB so that a relative call can be used?
> 
> Displacements are +/- 2G, not <4G.
> 
> Using unsigned here would be buggy.

Right, sorry for the noise:

Reviewed-by: Roger Pau Monné <roger.pau@xxxxxxxxxx>

With the minor nits pointed above in the ioemul_handle_quirk.

Thanks, Roger.