Re: [RFC PATCH] mm/rmap: do not call mmu_notifier_invalidate_page() v3

2017-08-29 Thread Jerome Glisse
On Tue, Aug 29, 2017 at 12:58:45PM -0700, Linus Torvalds wrote: > On Tue, Aug 29, 2017 at 12:16 PM, Linus Torvalds > wrote: > > > > And then you can check if something actually happened by catching the > > *ATOMIC* call to mmu_notifier_invalidate_page(), setting a flag, and > > then doing somethin

Re: [RFC PATCH] mm/rmap: do not call mmu_notifier_invalidate_page() v3

2017-08-29 Thread Linus Torvalds
On Tue, Aug 29, 2017 at 12:16 PM, Linus Torvalds wrote: > > And then you can check if something actually happened by catching the > *ATOMIC* call to mmu_notifier_invalidate_page(), setting a flag, and > then doing something blocking at mmu_notifier_invalidate_range_end() > time. > > Maybe. Note t

Re: [RFC PATCH] mm/rmap: do not call mmu_notifier_invalidate_page() v3

2017-08-29 Thread Linus Torvalds
On Tue, Aug 29, 2017 at 12:09 PM, Linus Torvalds wrote: > > So any approach like this is fundamentally garbage. Really. Stop > sending crap. This is exactly tehe same thing that we already reverted > because it was broken shit. Why do you re-send it without actually > fixing the fundamental proble

Re: [RFC PATCH] mm/rmap: do not call mmu_notifier_invalidate_page() v3

2017-08-29 Thread Jerome Glisse
On Tue, Aug 29, 2017 at 12:09:45PM -0700, Linus Torvalds wrote: > On Tue, Aug 29, 2017 at 12:05 PM, Jérôme Glisse wrote: > > Some MMU notifier need to be able to sleep during callback. This was > > broken by c7ab0d2fdc84 ("mm: convert try_to_unmap_one() to use > > page_vma_mapped_walk()"). > > No

Re: [RFC PATCH] mm/rmap: do not call mmu_notifier_invalidate_page() v3

2017-08-29 Thread Linus Torvalds
On Tue, Aug 29, 2017 at 12:05 PM, Jérôme Glisse wrote: > Some MMU notifier need to be able to sleep during callback. This was > broken by c7ab0d2fdc84 ("mm: convert try_to_unmap_one() to use > page_vma_mapped_walk()"). No. No no no. Didn't you learn *anything* from the bug? You cannot replace "

Re: [RFC PATCH] mm/rmap: do not call mmu_notifier_invalidate_page() v3

2017-08-29 Thread Jerome Glisse
On Tue, Aug 29, 2017 at 03:05:26PM -0400, Jérôme Glisse wrote: > Some MMU notifier need to be able to sleep during callback. This was > broken by c7ab0d2fdc84 ("mm: convert try_to_unmap_one() to use > page_vma_mapped_walk()"). > > This patch restore the sleep ability and properly capture the range

[RFC PATCH] mm/rmap: do not call mmu_notifier_invalidate_page() v3

2017-08-29 Thread Jérôme Glisse
Some MMU notifier need to be able to sleep during callback. This was broken by c7ab0d2fdc84 ("mm: convert try_to_unmap_one() to use page_vma_mapped_walk()"). This patch restore the sleep ability and properly capture the range of address that needs to be invalidated. Relevent threads: https://lkml