linux-kernel - Re: [PATCH v4 0/4] Make alloc_contig

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <a6b47162-d975-afef-473f-dccb677fefe6@redhat.com>
Date:   Mon, 15 Mar 2021 11:28:56 +0100
From:   David Hildenbrand <david@...hat.com>
To:     Oscar Salvador <osalvador@...e.de>
Cc:     Andrew Morton <akpm@...ux-foundation.org>,
        Vlastimil Babka <vbabka@...e.cz>,
        Michal Hocko <mhocko@...nel.org>,
        Muchun Song <songmuchun@...edance.com>,
        Mike Kravetz <mike.kravetz@...cle.com>, linux-mm@...ck.org,
        linux-kernel@...r.kernel.org,
        Anshuman Khandual <anshuman.khandual@....com>
Subject: Re: [PATCH v4 0/4] Make alloc_contig_range handle Hugetlb pages

On 15.03.21 11:27, Oscar Salvador wrote:
> On Mon, Mar 15, 2021 at 10:06:40AM +0100, David Hildenbrand wrote:
>>
>> BTW, I stumbled yesterday over
>>
>> alloc_contig_pages()->pfn_range_valid_contig():
>>
>> 	if (page_count(page) > 0)
>> 		rerurn false;
>> 	if (PageHuge(page))
>> 		return false;
>>
>> As used by memtrace and for gigantic pages. We can now
>>
>> a) Drop these check completely, as it's best-effort only and racy.
>> alloc_contig_pages()/alloc_contig_range() will handle it properly.
>>
>> b) Similarly, check for gigantic pages and/or movability/migratability.
>>
>> Dropping both checks might be the right thing to do: might significantly
>> increase allocation chances -- as we actually end up migrating busy pages
>> ...
> 
> Oh, sorry David, my mail client tricked me and I did not see this till now.
> 
> I will have a look, but I would like to collect some more feedback from all
> pieces before going any further and write a new version.
> Vlastimil patch#1 and patch#2 and he was ok with them, but let see what others
> think as well.

Planning on having a detailed look at the patches. Fairly busy though 
... :(

-- 
Thanks,

David / dhildenb