Nouveau - Jun 2019 - [PATCH 04/25] mm: remove MEMORY_DEVICE_PUBLIC support

[ add Ira ]

On Wed, Jun 26, 2019 at 5:27 AM Christoph Hellwig <hch at lst.de>
wrote:
>
> The code hasn't been used since it was added to the tree, and
doesn't
> appear to actually be usable.
>
> Signed-off-by: Christoph Hellwig <hch at lst.de>
> Reviewed-by: Jason Gunthorpe <jgg at mellanox.com>
> Acked-by: Michal Hocko <mhocko at suse.com>
[..]
> diff --git a/mm/swap.c b/mm/swap.c
> index 7ede3eddc12a..83107410d29f 100644
> --- a/mm/swap.c
> +++ b/mm/swap.c
> @@ -740,17 +740,6 @@ void release_pages(struct page **pages, int nr)
>                 if (is_huge_zero_page(page))
>                         continue;
>
> -               /* Device public page can not be huge page */
> -               if (is_device_public_page(page)) {
> -                       if (locked_pgdat) {
> -                              
spin_unlock_irqrestore(&locked_pgdat->lru_lock,
> -                                                      flags);
> -                               locked_pgdat = NULL;
> -                       }
> -                       put_devmap_managed_page(page);
> -                       continue;
> -               }
> -
This collides with Ira's bug fix [1]. The MEMORY_DEVICE_FSDAX case
needs this to be converted to be independent of "public" pages.
Perhaps it should be pulled out of -mm and incorporated in this
series.

[1]: https://lore.kernel.org/lkml/20190605214922.17684-1-ira.weiny at intel.com/

On Wed, Jun 26, 2019 at 09:00:47AM -0700, Dan Williams
wrote:
> [ add Ira ]
> 
> On Wed, Jun 26, 2019 at 5:27 AM Christoph Hellwig <hch at lst.de>
wrote:
> >
> > The code hasn't been used since it was added to the tree, and
doesn't
> > appear to actually be usable.
> >
> > Signed-off-by: Christoph Hellwig <hch at lst.de>
> > Reviewed-by: Jason Gunthorpe <jgg at mellanox.com>
> > Acked-by: Michal Hocko <mhocko at suse.com>
> [..]
> > diff --git a/mm/swap.c b/mm/swap.c
> > index 7ede3eddc12a..83107410d29f 100644
> > --- a/mm/swap.c
> > +++ b/mm/swap.c
> > @@ -740,17 +740,6 @@ void release_pages(struct page **pages, int nr)
> >                 if (is_huge_zero_page(page))
> >                         continue;
> >
> > -               /* Device public page can not be huge page */
> > -               if (is_device_public_page(page)) {
> > -                       if (locked_pgdat) {
> > -                              
spin_unlock_irqrestore(&locked_pgdat->lru_lock,
> > -                                                      flags);
> > -                               locked_pgdat = NULL;
> > -                       }
> > -                       put_devmap_managed_page(page);
> > -                       continue;
> > -               }
> > -
> 
> This collides with Ira's bug fix [1]. The MEMORY_DEVICE_FSDAX case
> needs this to be converted to be independent of "public" pages.
> Perhaps it should be pulled out of -mm and incorporated in this
> series.
> 
> [1]: https://lore.kernel.org/lkml/20190605214922.17684-1-ira.weiny at
intel.com/
Agreed and Andrew picked the first 2 versions of it, mmotm commits:

3eed114b5b6b mm-swap-fix-release_pages-when-releasing-devmap-pages-v2
9b7d8d0f572f mm/swap.c: fix release_pages() when releasing devmap pages

I don't see v3 but there were no objections...

Ira

On Wed, Jun 26, 2019 at 10:14:45AM -0700, 'Ira Weiny'
wrote:
> On Wed, Jun 26, 2019 at 09:00:47AM -0700, Dan Williams wrote:
> > [ add Ira ]
> > 
> > On Wed, Jun 26, 2019 at 5:27 AM Christoph Hellwig <hch at
lst.de> wrote:
> > >
> > > The code hasn't been used since it was added to the tree, and
doesn't
> > > appear to actually be usable.
> > >
> > > Signed-off-by: Christoph Hellwig <hch at lst.de>
> > > Reviewed-by: Jason Gunthorpe <jgg at mellanox.com>
> > > Acked-by: Michal Hocko <mhocko at suse.com>
> > [..]
> > > diff --git a/mm/swap.c b/mm/swap.c
> > > index 7ede3eddc12a..83107410d29f 100644
> > > --- a/mm/swap.c
> > > +++ b/mm/swap.c
> > > @@ -740,17 +740,6 @@ void release_pages(struct page **pages, int
nr)
> > >                 if (is_huge_zero_page(page))
> > >                         continue;
> > >
> > > -               /* Device public page can not be huge page */
> > > -               if (is_device_public_page(page)) {
> > > -                       if (locked_pgdat) {
> > > -                              
spin_unlock_irqrestore(&locked_pgdat->lru_lock,
> > > -                                                      flags);
> > > -                               locked_pgdat = NULL;
> > > -                       }
> > > -                       put_devmap_managed_page(page);
> > > -                       continue;
> > > -               }
> > > -
> > 
> > This collides with Ira's bug fix [1]. The MEMORY_DEVICE_FSDAX case
> > needs this to be converted to be independent of "public"
pages.
> > Perhaps it should be pulled out of -mm and incorporated in this
> > series.
> > 
> > [1]: https://lore.kernel.org/lkml/20190605214922.17684-1-ira.weiny at
intel.com/
> 
> Agreed and Andrew picked the first 2 versions of it, mmotm commits:
> 
> 3eed114b5b6b mm-swap-fix-release_pages-when-releasing-devmap-pages-v2
> 9b7d8d0f572f mm/swap.c: fix release_pages() when releasing devmap pages
> 
> I don't see v3 but there were no objections...
Ok somehow I can't fetch mmotm right now...

Dan had and updated mmotm tree and it does have my v4 patch.

Does anyone else have issues with git://git.cmpxchg.org/linux-mmotm.git or is
it just me?  FWIW I have checked proxies etc... and can get to linus and other
sites just fine, so it looks like an issue there.  Although the web page is
fine...

Sorry,
Ira