On Tue, Jul 08, 2014 at 05:59:57PM -0400, j.glisse@xxxxxxxxx wrote: > So here is updated patchset. I believe the first patch despite Joerg > opposition, is still a welcome change to mmput. Wether or not kfd and > hmm could register might still be debated but i strongly believe that > the fact that we are tie to mm_struct and the file lifespan is not tie > to a specific mm_struct is a testimony that we should interface with > mm_struct destruction. > > The third patch have updated comment and i hope address all Linus' > worries. The event description have been updated and i hope is clear > enough. > > The last patch also been updated and now only pass vma to various > mmu_notifier callback. This does means that instead of calling once > the mmu_notifier for zapping whole address range, it will be call > once per vma. I believe that given this change only impact process > that use vma and that by restricting ourself to existing vma the > impact will be small and might even turn to be a win as mmu listener > will not have to traverse whole secondary page table full of non > existant entry but only do job on thing that migth exist in the > secondary page table. > > Hope this address any previous concern about those patches. > > Cheers, > Jérôme > > I manage to forget mailing list. -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@xxxxxxxxx. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>