Re: [PATCH v2 1/5] mm/hugetlb: fix races when looking up a CONT-PTE size hugetlb page

From: Baolin Wang <baolin.wang@linux.alibaba.com>
To: David Hildenbrand <david@redhat.com>,
	Mike Kravetz <mike.kravetz@oracle.com>
Cc: akpm@linux-foundation.org, songmuchun@bytedance.com,
	linux-mm@kvack.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH v2 1/5] mm/hugetlb: fix races when looking up a CONT-PTE size hugetlb page
Date: Thu, 25 Aug 2022 15:58:45 +0800	[thread overview]
Message-ID: <3c381602-26ea-883c-859a-4007cb7cb289@linux.alibaba.com> (raw)
In-Reply-To: <50703dda-bbee-7484-97ff-87d6ea75bf80@redhat.com>

On 8/25/2022 3:10 PM, David Hildenbrand wrote:
> 
>>> +		/*
>>> +		 * try_grab_page() should always succeed here, because we hold
>>> +		 * the ptl lock and have verified pte_present().
>>> +		 */
>>> +		if (WARN_ON_ONCE(!try_grab_page(page, flags))) {
>>> +			page = NULL;
>>> +			goto out;
>>> +		}
>>> +	} else {
>>
>> Should add FOLL_MIGRATION validation before waiting a migration entry.
> 
> We really only need FOLL_MIGRATION for KSM. As hugetlb pages cannot be
> KSM pages, we don't need this.
> 
> Actually, I do have patches in the works that rip out FOLL_MIGRATION
> completely by adjusting KSM code.
> 
> So let's try to not add dead code (although it would make sense for
> feature completeness as is -- but then, FOLL_MIGRATION really needs to go).

Make sense. Thanks for your explanation.