Re: [PATCH] mm/hugetlb: Get rid of NODEMASK_ALLOC

From: Mike Kravetz <mike.kravetz@oracle.com>
To: Andrew Morton <akpm@linux-foundation.org>,
	Oscar Salvador <osalvador@suse.de>
Cc: n-horiguchi@ah.jp.nec.com, linux-mm@kvack.org,
	linux-kernel@vger.kernel.org
Subject: Re: [PATCH] mm/hugetlb: Get rid of NODEMASK_ALLOC
Date: Tue, 2 Apr 2019 13:09:33 -0700	[thread overview]
Message-ID: <360a3016-4b35-3ec6-0716-1c2a62836f0a@oracle.com> (raw)
In-Reply-To: <20190402130153.338e59c6cfda1ed3ec882517@linux-foundation.org>

On 4/2/19 1:01 PM, Andrew Morton wrote:
> On Tue,  2 Apr 2019 15:34:15 +0200 Oscar Salvador <osalvador@suse.de> wrote:
> 
>> NODEMASK_ALLOC is used to allocate a nodemask bitmap, ant it does it by
>> first determining whether it should be allocated in the stack or dinamically
>> depending on NODES_SHIFT.
>> Right now, it goes the dynamic path whenever the nodemask_t is above 32
>> bytes.
>>
>> Although we could bump it to a reasonable value, the largest a nodemask_t
>> can get is 128 bytes, so since __nr_hugepages_store_common is called from
>> a rather shore stack we can just get rid of the NODEMASK_ALLOC call here.
>>
>> This reduces some code churn and complexity.
> 
> It took a bit of sleuthing to figure out that this patch applies to
> Mike's "hugetlbfs: fix potential over/underflow setting node specific
> nr_hugepages".  Should they be folded together?  I'm thinking not.

No need to fold.  They are separate issues and that over/underflow patch
may already be doing too many things.

> (Also, should "hugetlbfs: fix potential over/underflow setting node
> specific nr_hugepages" have been -stableified?  I also think not, but I
> bet it happens anyway).

I don't see a great reason for sending to stable.  IIRC, nobody actually
hit this issue: it was found through code inspection.
-- 
Mike Kravetz