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;
Subject: Re: [PATCH v5 2/2] kvm: Use huge pages for DAX-backed files
To:     Liran Alon <liran.alon@oracle.com>
Cc:     Paolo Bonzini <pbonzini@redhat.com>,
        Dan Williams <dan.j.williams@intel.com>,
        David Hildenbrand <david@redhat.com>,
        Dave Jiang <dave.jiang@intel.com>,
        Alexander Duyck <alexander.h.duyck@linux.intel.com>,
        Sean Christopherson <sean.j.christopherson@intel.com>,
        linux-nvdimm@lists.01.org, x86@kernel.org, kvm@vger.kernel.org,
        linux-kernel@vger.kernel.org, jason.zeng@intel.com
References: <20191212182238.46535-1-brho@google.com>
 <20191212182238.46535-3-brho@google.com>
 <06108004-1720-41EB-BCAB-BFA8FEBF4772@oracle.com>
 <ED482280-CB47-4AB6-9E7E-EEE7848E0F8B@oracle.com>
 <f8e948ff-6a2a-a6d6-9d8e-92b93003354a@google.com>
 <65FB6CC1-3AD2-4D6F-9481-500BD7037203@oracle.com>
From:   Barret Rhoden <brho@google.com>
Message-ID: <90a9af31-304c-e8d5-b17c-0ddb4c98fddb@google.com>
Date:   Fri, 13 Dec 2019 09:13:05 -0500
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:60.0) Gecko/20100101
 Thunderbird/60.9.0
MIME-Version: 1.0
In-Reply-To: <65FB6CC1-3AD2-4D6F-9481-500BD7037203@oracle.com>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Language: en-GB
Content-Transfer-Encoding: 8bit
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On 12/12/19 8:07 PM, Liran Alon wrote:
>> I was a little hesitant to change the this to handle 1 GB pages with this patchset at first.  I didn't want to break the non-DAX case stuff by doing so.
> 
> Why would it affect non-DAX case?
> Your patch should just make hugepage_adjust() to parse page-tables only in case is_zone_device_page(). Otherwise, page tables shouldn’t be parsed.
> i.e. THP merged pages should still be detected by PageTransCompoundMap().

That's what I already do.  But if I wanted to make the hugepage_adjust() 
function also handle the change to 1 GB, then that code would apply to 
THP too.  I didn't want to do that without knowing the implications for THP.

>> Specifically, can a THP page be 1 GB, and if so, how can you tell?  If you can't tell easily, I could walk the page table for all cases, instead of just zone_device().
> 
> I prefer to walk page-tables only for is_zone_device_page().

Is there another way to tell if a THP page is 1 GB?  Anyway, this is the 
sort of stuff I didn't want to mess around with.

hugepage_adjust() seemed like a reasonable place to get a huge (2MB) 
page table entry out of a DAX mapping.  I didn't want to proliferate 
another special case for upgrading to a larger PTE size (i.e. how 
hugetlbfs and THP have separate mechanisms), so I hopped on to the "can 
we do a 2MB mapping even though host_mapping_level() didn't say so" case 
- which is my interpretation of what huge_adjust() is for.

Barret