Re: set_pte_at_notify regression

2014-01-15 Thread Haggai Eran
On 12/01/2014 19:50, Andrea Arcangeli wrote: > On Sun, Jan 12, 2014 at 01:56:12PM +0200, Haggai Eran wrote: >> Hi, >> >> On 10/01/2014 18:57, Andrea Arcangeli wrote: >>> Hi! >>> >>> On Fri, Jan 10, 2014 at 06:38:06PM +0200, Izik Eidus wrote: It look like commit 6bdb913f0a70a4dfb7f066fb15e2d6f9

Re: set_pte_at_notify regression

2014-01-12 Thread Andrea Arcangeli
On Sun, Jan 12, 2014 at 01:56:12PM +0200, Haggai Eran wrote: > Hi, > > On 10/01/2014 18:57, Andrea Arcangeli wrote: > > Hi! > > > > On Fri, Jan 10, 2014 at 06:38:06PM +0200, Izik Eidus wrote: > >> It look like commit 6bdb913f0a70a4dfb7f066fb15e2d6f960701d00 break the > >> semantic of set_pte_at_n

Re: set_pte_at_notify regression

2014-01-12 Thread Haggai Eran
Hi, On 10/01/2014 18:57, Andrea Arcangeli wrote: > Hi! > > On Fri, Jan 10, 2014 at 06:38:06PM +0200, Izik Eidus wrote: >> It look like commit 6bdb913f0a70a4dfb7f066fb15e2d6f960701d00 break the >> semantic of set_pte_at_notify. >> The change of calling first to mmu_notifier_invalidate_range_start,

Re: set_pte_at_notify regression

2014-01-10 Thread Andrea Arcangeli
Hi! On Fri, Jan 10, 2014 at 06:38:06PM +0200, Izik Eidus wrote: > It look like commit 6bdb913f0a70a4dfb7f066fb15e2d6f960701d00 break the > semantic of set_pte_at_notify. > The change of calling first to mmu_notifier_invalidate_range_start, then > to set_pte_at_notify, and then to mmu_notifier_in

set_pte_at_notify regression

2014-01-10 Thread Izik Eidus
Hi, It look like commit 6bdb913f0a70a4dfb7f066fb15e2d6f960701d00 break the semantic of set_pte_at_notify. The change of calling first to mmu_notifier_invalidate_range_start, then to set_pte_at_notify, and then to mmu_notifier_invalidate_range_end not only increase the amount of locks kvm have t