[U-Boot] [PATCH] ARM: cache-cp15: Make sure EAE is not enabled

[Albert ARIBAUD]

Hello Tomeu,

On Fri, 19 Jun 2015 17:16:21 +0200, Tomeu Vizoso
<tomeu.vizoso at collabora.com> wrote:
> On 18 June 2015 at 17:13, Simon Glass <sjg at chromium.org> wrote:
> > Hi,
> >
> > On 18 June 2015 at 01:19, Tomeu Vizoso <tomeu.vizoso at collabora.com> wrote:
> >>
> >> Hello,
> >>
> >> any news on this one?
> >>
> >> Thanks,
> >>
> >> Tomeu
> >>
> >> On 30 April 2015 at 09:23, Sjoerd Simons <sjoerd.simons at collabora.co.uk> wrote:
> >> > +Albert, +Marek,
> >> >
> >> > On Tue, 2015-04-28 at 14:21 +0200, Tomeu Vizoso wrote:
> >> >> This could happen if we are being chainloaded by Coreboot with LPAE
> >> >> enabled, as is the case on the Tegra-based Chromebooks.
> >> >>
> >> >> Signed-off-by: Tomeu Vizoso <tomeu.vizoso at collabora.com>
> >> >
> >> > From the documentation of recent ARM cores it looks like 0 is the reset
> >> > value for that register, so explicitly ensuring it has that value seems
> >> > sane. I'm wary of giving a reviewed-by though as i don't know that part
> >> > of u-boot well enough to say it's the right spot.
> >> >
> >> > However, as we're now using successfully using a Nyan big board with
> >> > this u-boot patch in kernelci[0] i can happily say:
> >> >
> >> > Tested-By: Sjoerd Simons <sjoerd.simons at collabora.co.uk>
> >> >
> >> > 0: http://kernelci.org/boot/tegra124-nyan-big/
> >> >
> >> >
> >> >> ---
> >> >>  arch/arm/lib/cache-cp15.c | 4 ++++
> >> >>  1 file changed, 4 insertions(+)
> >> >>
> >> >> diff --git a/arch/arm/lib/cache-cp15.c b/arch/arm/lib/cache-cp15.c
> >> >> index 0291afa..78fb429 100644
> >> >> --- a/arch/arm/lib/cache-cp15.c
> >> >> +++ b/arch/arm/lib/cache-cp15.c
> >> >> @@ -96,6 +96,10 @@ static inline void mmu_setup(void)
> >> >>               dram_bank_mmu_setup(i);
> >> >>       }
> >> >>
> >> >> +     /* Make sure EAE is not enabled */
> >> >> +     asm volatile("mcr p15, 0, %0, c2, c0, 2"
> >> >> +                  : : "r" (0) : "memory");
> >> >> +
> >> >>       /* Copy the page table address to cp15 */
> >> >>       asm volatile("mcr p15, 0, %0, c2, c0, 0"
> >> >>                    : : "r" (gd->arch.tlb_addr) : "memory");
> >
> > Does this apply to all ARM cores? Should it have #ifdef CONFIG_ARMV7 or similar?
> 
> Don't know, TBH. Resetting that register will select the TTBR0 and a
> page table boundary size of 16KB, which AFAICS is expected in U-Boot
> anyways.
> 
> On ARMv6, there isn't a EAE bit, so only the above will happen.
> 
> I cannot say if it's a good idea or not to reset the TTBCR just in case.

Then the best is to just change in that register what must be changed.
If EAE is the only bit in TTBCR that needs resetting for the Nyan big
board to boot properly, then I would prefer that all other bits be left
untouched.

> Thanks,
> 
> Tomeu

Amicalement,
-- 
Albert.