Re: [Qemu-devel] [PATCH v4] fpu: add mechanism to check for invalid long double formats

[Peter Maydell]

On 17 August 2016 at 01:06, Andrew Dutcher <address@hidden> wrote:
> Also- I'm having issues applying the new patch:
>
> @@ -5768,7 +5774,9 @@ int floatx80_lt(floatx80 a, floatx80 b,
> float_status *status)
>  
> *----------------------------------------------------------------------------*/
>  int floatx80_unordered(floatx80 a, floatx80 b, float_status *status)
>  {
> -    if (    (    ( extractFloatx80Exp( a ) == 0x7FFF )
> +    if (    floatx80_invalid_encoding( a )
> +         || floatx80_invalid_encoding( b )
> +         || (    ( extractFloatx80Exp( a ) == 0x7FFF )
>                && (uint64_t) ( extractFloatx80Frac( a )<<1 ) )
>           || (    ( extractFloatx80Exp( b ) == 0x7FFF )
>                && (uint64_t) ( extractFloatx80Frac( b )<<1 ) )
>
> When I do this, the style checker complains about the spaces after the
> open parens and before the close parens. I now have to change this
> entire stanza to be styled correctly, since I'm replacing the original
> first line...

Yes, that's the best approach. softfloat has some very weird
spacing patterns because it's third-party code. Generally
what we do is update it to match QEMU's coding style when
we have to touch a part of the code for some other reason,
which can mean needing to reindent some lines around the
ones being changed to keep checkpatch happy.

thanks
-- PMM