[PATCHv3 03/11] arm64: Introduce helpers for page table levels

Christoffer Dall christoffer.dall at linaro.org
Thu Oct 15 06:30:01 PDT 2015


On Thu, Oct 15, 2015 at 02:14:22PM +0100, Suzuki K. Poulose wrote:
> On 15/10/15 13:44, Mark Rutland wrote:
> >On Thu, Oct 15, 2015 at 01:37:35PM +0200, Christoffer Dall wrote:
> >>On Wed, Oct 14, 2015 at 12:20:26PM +0100, Suzuki K. Poulose wrote:
> >>>Introduce helpers for finding the number of page table
> >>>levels required for a given VA width, shift for a particular
> >>>page table level.
> 
> >>>+/*
> >>>+ * Size mapped by an entry at level n
> >>>+ * We map PAGE_SHIFT - 3 at all levels, except the PAGE_SHIFT bits at the last level
> >>>+ */
> >>>+#define ARM64_HW_PGTABLE_LEVEL_SHIFT(n)	((PAGE_SHIFT - 3) * (4 - (n)) + 3)
> >>
> >>I feel like I'm partially failing the interview question again, in that
> >>I don't fully understand the '+ 3' in the end?
> >
> >The last level handles PAGE_SHIFT bits (the bits from the VA that are
> >the same in the PA). We only accounted for (PAGE_SHIFT - 3) bits at each
> >level when multiplying, so we add those 3 missing bits back at the end.
> >
> 
> Something like :
> 
> /*
>  * Size mapped by an entry at level n
>  * We map PAGE_SHIFT - 3 at all levels, except the last, where we map PAGE_SHIFT bits.
>  * The maximum number of levels supported by the architecture is 4. Hence at starting
>  * at level n, we hanve (4 - n) levels of translation. So, the total number of bits

nit: s/hanve/have/

>  * mapped by an entry at level n is :
>  *
>  *  ((4 - n) - 1) * (PAGE_SHIFT - 3) + PAGE_SHIFT
>  *
>  * Rearranging it a bit we get :
>  *   (4 - n) * (PAGE_SHIFT - 3) + 3
>  */
> 
> Or we could use the formula without rearranging.
> 
Either way, even I get it now.

Thanks for the explanation!!

Assuming some version of this goes in:

Acked-by: Christoffer Dall <christoffer.dall at linaro.org>



More information about the linux-arm-kernel mailing list