[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20190108224026.GL31793@dhcp22.suse.cz>
Date: Tue, 8 Jan 2019 23:40:26 +0100
From: Michal Hocko <mhocko@...nel.org>
To: Qian Cai <cai@....pw>
Cc: Andrew Morton <akpm@...ux-foundation.org>,
Pavel.Tatashin@...rosoft.com, mingo@...nel.org,
mgorman@...hsingularity.net, iamjoonsoo.kim@....com,
tglx@...utronix.de, linux-mm@...ck.org,
linux-kernel@...r.kernel.org
Subject: Re: [PATCH v3] mm/page_owner: fix for deferred struct page init
On Tue 08-01-19 17:13:41, Qian Cai wrote:
>
>
> On 1/8/19 5:02 PM, Andrew Morton wrote:
> >
> > It's unclear (to me) where we stand with this patch. Shold we proceed
> > with v3 for now, or is something else planned?
>
> I don't have anything else plan for this right now. Michal particular don't like
> that 4-line ifdef which supposes to avoid an immediately regression (arguably
> small) that existing page_owner users with DEFERRED_STRUCT_PAGE_INIT deselected
> that would start to miss tens of thousands early page allocation call sites. So,
> feel free to choose v2 of this which has no ifdef if you agree with Michal too.
> I am fine either way.
Yes I would prefer to revert the faulty commit (fe53ca54270) and work on
a more robust page_owner initialization to cover earlier allocations on
top of that.
Not that I would insist but this would be a more straightforward
approach and I hope it will result in a better long term maintainable
code in the end.
--
Michal Hocko
SUSE Labs
Powered by blists - more mailing lists