Re: [REPOST PATCH] gbefb: cmap FIFO timeout

[Andrew Morton <akpm@linux-foundation.org>]

On Wed, 30 Jul 2008 00:12:04 +0200 (CEST)
Thomas Bogendoerfer <tsbogend@alpha.franken.de> wrote:

> +static void gbe_loadcmap(void)
> +{
> +     int i, j;
> +
> +     for (i = 0; i < 256; i++) {
> +             for (j = 0; j < 1000 && gbe->cm_fifo >= 63; j++)
> +                     udelay(10);
> +             if (j == 1000)
> +                     printk(KERN_ERR "gbefb: cmap FIFO timeout\n");
> +
> +             gbe->cmap[i] = gbe_cmap[i];
> +     }
>  }

This is polling the hardware, yes?

I'd have thought there's a terrible risk of the compiler "optimising"
the read of gbe->cm_fifo.  Shouldn't we be using readl() here?

<looks>

oh.  It's declared volatile.  ugh.