Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
MIME-Version: 1.0
In-Reply-To: <20180617200432.krw36wrcwidb25cj@ziepe.ca>
References: <20180617012510.20139-1-jhubbard@nvidia.com> <20180617012510.20139-3-jhubbard@nvidia.com>
 <CAPcyv4i=eky-QrPcLUEqjsASuRUrFEWqf79hWe0mU8xtz6Jk-w@mail.gmail.com> <20180617200432.krw36wrcwidb25cj@ziepe.ca>
From:   Dan Williams <dan.j.williams@intel.com>
Date:   Sun, 17 Jun 2018 13:10:04 -0700
Message-ID: <CAPcyv4gayKk_zHDYAvntware12qMXWjnnL_FDJNUQsJS_zNfDw@mail.gmail.com>
Subject: Re: [PATCH 2/2] mm: set PG_dma_pinned on get_user_pages*()
To:     Jason Gunthorpe <jgg@ziepe.ca>
Cc:     john.hubbard@gmail.com, Matthew Wilcox <willy@infradead.org>,
        Michal Hocko <mhocko@kernel.org>,
        Christopher Lameter <cl@linux.com>, Jan Kara <jack@suse.cz>,
        Linux MM <linux-mm@kvack.org>,
        LKML <linux-kernel@vger.kernel.org>,
        linux-rdma <linux-rdma@vger.kernel.org>,
        John Hubbard <jhubbard@nvidia.com>,
        Christoph Hellwig <hch@lst.de>
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Sun, Jun 17, 2018 at 1:04 PM, Jason Gunthorpe <jgg@ziepe.ca> wrote:
> On Sun, Jun 17, 2018 at 12:53:04PM -0700, Dan Williams wrote:
>> > diff --git a/mm/rmap.c b/mm/rmap.c
>> > index 6db729dc4c50..37576f0a4645 100644
>> > +++ b/mm/rmap.c
>> > @@ -1360,6 +1360,8 @@ static bool try_to_unmap_one(struct page *page, struct vm_area_struct *vma,
>> >                                 flags & TTU_SPLIT_FREEZE, page);
>> >         }
>> >
>> > +       if (PageDmaPinned(page))
>> > +               return false;
>> >         /*
>> >          * We have to assume the worse case ie pmd for invalidation. Note that
>> >          * the page can not be free in this function as call of try_to_unmap()
>>
>> We have a similiar problem with DAX and the conclusion we came to is
>> that it is not acceptable for userspace to arbitrarily block kernel
>> actions. The conclusion there was: 'wait' if the DMA is transient, and
>> 'revoke' if the DMA is long lived, or otherwise 'block' long-lived DMA
>> if a revocation mechanism is not available.
>
> This might be the right answer for certain things, but it shouldn't be
> the immediate reaction to everthing. There are many user APIs that
> block kernel actions and hold kernel resources.
>
> IMHO, there should be an identifiable objection, eg is blocking going
> to create a DOS, dead-lock, insecurity, etc?

I believe kernel behavior regression is a primary concern as now
fallocate() and truncate() can randomly fail where they didn't before.