On 10/10/2018 09:51 AM, Sascha Hauer wrote:
>> +static inline int hab_get_status(void)
>> +{
>> +	if (IS_ENABLED(CONFIG_MMU))
>> +		return -EIO;
> 
> This restriction makes this patch rather unhandy to use. Internally we
> use imx*_hab_get_status() indeed with MMU enabled and it works fine, but
> we call it before mmu_init is called. In this state the MMU is enabled
> already, but the zero page is not yet set to faulting. So I think the
> problem is not the MMU as such, but instead the faulting zero page.

BTW:

The faulting zero page problem doesn't occur on mx25 and mx28 as the HAB
ROM doesn't live at 0x0:

#define HABV4_RVT_IMX28 0xffff8af8

int imx25_hab_get_status(void)
{
        return imx_habv3_get_status(readl(IOMEM(0x780018d4)));
}

> It would be great if we could overcome this issue. One way to do this
> would be to add a function pair mmu_zero_page_map() and
> mmu_zero_page_set_faulting() to be called in hab_get_status().
> 
> I don't know if you are willing to add that, but I don't want to
> motivate people to disable the MMU, so the patch is not acceptable as
> is.
Marc

-- 
Pengutronix e.K.                  | Marc Kleine-Budde           |
Industrial Linux Solutions        | Phone: +49-231-2826-924     |
Vertretung West/Dortmund          | Fax:   +49-5121-206917-5555 |
Amtsgericht Hildesheim, HRA 2686  | http://www.pengutronix.de   |