Re: [PATCH] kernel32: implement GetLargePageMinimum

[Alexandre Julliard]

Marcus Meissner  writes:
> +SIZE_T WINAPI GetLargePageMinimum(void)
> +{
> +#ifdef linux
> - FILE *f;
> - unsigned long hugepagesize = 0;
> -
> - f = fopen( "/proc/meminfo", "r" );
> - if (f)
> - {
> -
>    char buffer[256];
> -
> -
>    while (fgets( buffer, sizeof(buffer), f ))
> -
>    {
> -
>        if (sscanf( buffer, "Hugepagesize: %lu", &hugepagesize ))
> -
>            break;
> -
>    }
> -
>    fclose( f );
> - }
> - return (SIZE_T)hugepagesize*1024;
>
> +#else
> - FIXME("stub: not implemented on your platform.\n");
> - return 0;
>
> +#endif

This could probably be simply hard-coded based on the CPU architecture.

[PATCH] kernel32: implement GetLargePageMinimum

[Marcus Meissner]

On Mon, Jun 13, 2016 at 08:28:18PM +0900, Alexandre Julliard wrote:
> Marcus Meissner  writes:
> > +SIZE_T WINAPI GetLargePageMinimum(void)
> > +{
> > +#ifdef linux
> > - FILE *f;
> > - unsigned long hugepagesize = 0;
> > -
> > - f = fopen( "/proc/meminfo", "r" );
> > - if (f)
> > - {
> > -
> >    char buffer[256];
> > -
> > -
> >    while (fgets( buffer, sizeof(buffer), f ))
> > -
> >    {
> > -
> >        if (sscanf( buffer, "Hugepagesize: %lu", &hugepagesize ))
> > -
> >            break;
> > -
> >    }
> > -
> >    fclose( f );
> > - }
> > - return (SIZE_T)hugepagesize*1024;
> >
> > +#else
> > - FIXME("stub: not implemented on your platform.\n");
> > - return 0;
> >
> > +#endif
>
> This could probably be simply hard-coded based on the CPU architecture.

While it does depend on some processor flags and also kernel support, I think these
are default enabled everywhere, so we can just hardcode it. I sent a follow up patch.

Ciao, Marcus