Received-SPF: pass (google.com: domain of linux-kernel+bounces-132111-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:40f1:3f00::1 as permitted sender) client-ip=2604:1380:40f1:3f00::1;
Date: Thu, 4 Apr 2024 13:45:27 -0700
From: Roman Gushchin <roman.gushchin@linux.dev>
To: Frank van der Linden <fvdl@google.com>
Cc: linux-mm@kvack.org, muchun.song@linux.dev, akpm@linux-foundation.org,
	linux-kernel@vger.kernel.org
Subject: Re: [PATCH 2/2] mm/hugetlb: pass correct order_per_bit to
 cma_declare_contiguous_nid
Message-ID: <Zg8RZzYccpbqQ_ni@P9FQF9L96D.corp.robot.car>
References: <20240404162515.527802-1-fvdl@google.com>
 <20240404162515.527802-2-fvdl@google.com>
 <Zg7358ygxHBHUCy9@P9FQF9L96D.corp.robot.car>
 <CAPTztWYH5ANR2cYidf+frC2HBJiz6UUh5wC5khHJg8R-gYbcFw@mail.gmail.com>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <CAPTztWYH5ANR2cYidf+frC2HBJiz6UUh5wC5khHJg8R-gYbcFw@mail.gmail.com>

On Thu, Apr 04, 2024 at 12:40:58PM -0700, Frank van der Linden wrote:
> On Thu, Apr 4, 2024 at 11:56 AM Roman Gushchin <roman.gushchin@linux.dev> wrote:
> >
> > On Thu, Apr 04, 2024 at 04:25:15PM +0000, Frank van der Linden wrote:
> > > The hugetlb_cma code passes 0 in the order_per_bit argument to
> > > cma_declare_contiguous_nid (the alignment, computed using the
> > > page order, is correctly passed in).
> > >
> > > This causes a bit in the cma allocation bitmap to always represent
> > > a 4k page, making the bitmaps potentially very large, and slower.
> > >
> > > So, correctly pass in the order instead.
> > >
> > > Signed-off-by: Frank van der Linden <fvdl@google.com>
> > > Cc: Roman Gushchin <roman.gushchin@linux.dev>
> > > Fixes: cf11e85fc08c ("mm: hugetlb: optionally allocate gigantic hugepages using cma")
> >
> > Hi Frank,
> >
> > there is a comment just above your changes which explains why order_per_bit is 0.
> > Is this not true anymore? If so, please, fix the comment too. Please, clarify.
> >
> > Thanks!
> 
> Hi Roman,
> 
> I'm assuming you're referring to this comment:
> 
> /*
>  * Note that 'order per bit' is based on smallest size that
>  * may be returned to CMA allocator in the case of
>  * huge page demotion.
>  */
> 
> That comment was added in a01f43901cfb9 ("hugetlb: be sure to free
> demoted CMA pages to CMA").
> 
> It talks about HUGETLB_PAGE_ORDER being the minimum order being given
> back to the CMA allocator (after hugetlb demotion), therefore
> order_per_bit must be HUGETLB_PAGE_ORDER. See the commit message for
> that commit:
> 
> "Therefore, at region setup time we use HUGETLB_PAGE_ORDER as the
> smallest possible huge page size that can be given back to CMA."
> 
> But the commit, while correctly changing the alignment, left the
> order_per_bit argument at 0,  even though it clearly intended to set
> it at HUGETLB_PAGE_ORDER. The confusion may have been that
> cma_declare_contiguous_nid has 9 arguments, several of which can be
> left at 0 meaning 'use default', so it's easy to misread.
> 
> In other words, the comment was correct, but the code was not. After
> this patch, comment and code match.

Indeed the mentioned commit which added a comment which was not aligned
with the code was confusing. It all makes sense now, thank you for
the explanation!

Please, feel free to add
Acked-by: Roman Gushchin <roman.gushchin@linux.dev>
for your patch.

Thanks!