Re: [PATCH] mm/autonuma: don't use set_pte_at when updating protnone ptes

2017-02-14 Thread Andrew Morton
On Tue, 14 Feb 2017 19:41:17 +0530 "Aneesh Kumar K.V" wrote: > "Aneesh Kumar K.V" writes: > > > Architectures like ppc64, use privilege access bit to mark pte non > > accessible. > > This implies that kernel can do a copy_to_user to an address marked for > > numa fault. > > This also implies

Re: [PATCH] mm/autonuma: don't use set_pte_at when updating protnone ptes

2017-02-14 Thread Aneesh Kumar K.V
"Aneesh Kumar K.V" writes: > Architectures like ppc64, use privilege access bit to mark pte non accessible. > This implies that kernel can do a copy_to_user to an address marked for numa > fault. > This also implies that there can be a parallel hardware update for the pte. > set_pte_at cannot be

Re: [PATCH] mm/autonuma: don't use set_pte_at when updating protnone ptes

2017-02-06 Thread Mel Gorman
On Mon, Feb 06, 2017 at 10:36:16PM +0530, Aneesh Kumar K.V wrote: > Architectures like ppc64, use privilege access bit to mark pte non accessible. > This implies that kernel can do a copy_to_user to an address marked for numa > fault. > This also implies that there can be a parallel hardware updat

Re: [PATCH] mm/autonuma: don't use set_pte_at when updating protnone ptes

2017-02-06 Thread Rik van Riel
On Mon, 2017-02-06 at 22:36 +0530, Aneesh Kumar K.V wrote: > Architectures like ppc64, use privilege access bit to mark pte non > accessible. > This implies that kernel can do a copy_to_user to an address marked > for numa fault. > This also implies that there can be a parallel hardware update for

[PATCH] mm/autonuma: don't use set_pte_at when updating protnone ptes

2017-02-06 Thread Aneesh Kumar K.V
Architectures like ppc64, use privilege access bit to mark pte non accessible. This implies that kernel can do a copy_to_user to an address marked for numa fault. This also implies that there can be a parallel hardware update for the pte. set_pte_at cannot be used in such scenarios. Hence switch t