[PATCH v3] ARM: L2 : Errata 588369: Clean & Invalidate do not invalidate clean lines

[Shilimkar, Santosh]

> -----Original Message-----
> From: Russell King - ARM Linux [mailto:linux at arm.linux.org.uk]
> Sent: Monday, January 11, 2010 8:15 PM
> To: Catalin Marinas
> Cc: Shilimkar, Santosh; linux-arm-kernel at lists.infradead.org; linux-omap at vger.kernel.org;
> tony at atomide.com; Woodruff, Richard
> Subject: Re: [PATCH v3] ARM: L2 : Errata 588369: Clean & Invalidate do not invalidate clean lines
> 
> On Mon, Jan 11, 2010 at 02:37:08PM +0000, Catalin Marinas wrote:
> > On Mon, 2009-12-21 at 10:09 +0000, Santosh Shilimkar wrote:
> > > @@ -139,7 +184,12 @@ static void l2x0_flush_range(unsigned long start, unsigned long end)
> > >                         spin_lock_irqsave(&l2x0_lock, flags);
> > >                 }
> > >         }
> > > +#ifdef CONFIG_PL310_ERRATA_588369
> > > +       cache_wait(base + L2X0_CLEAN_LINE_PA, 1);
> > > +       cache_wait(base + L2X0_INV_LINE_PA, 1);
> > > +#else
> > >         cache_wait(base + L2X0_CLEAN_INV_LINE_PA, 1);
> > > +#endif
> >
> > I don't think we need to way on two separate registers here. AFAICT, bit
> > 1 of those registers is shared for all the operations.
> >
> > As a general comment, maybe an inline function called something like
> > wait_writel(before/after) would be better than a lot of ifdefs in the
> > code, especially if one has a different workaround other than using TI's
> > secure monitor.
> 
> Since the bit is shared between all 'R7' registers, the ifdef above makes
> no sense.  We can just wait on any R7 register - I suggest that bit of
> code is left as-is.
Ok. will update the patch.