linux-kernel - Re: [PATCH v4 02/28] mm: Add an unlock function for PG_private_2/PG

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: <CAHk-=whWoJhGeMn85LOh9FX-5d2-Upzmv1m2ZmYxvD31TKpUTA@mail.gmail.com>
Date:   Tue, 16 Mar 2021 19:35:17 -0700
From:   Linus Torvalds <torvalds@...ux-foundation.org>
To:     Josef Bacik <josef@...icpanda.com>
Cc:     Matthew Wilcox <willy@...radead.org>, Chris Mason <clm@...com>,
        David Sterba <dsterba@...e.com>,
        David Howells <dhowells@...hat.com>,
        Trond Myklebust <trondmy@...merspace.com>,
        Anna Schumaker <anna.schumaker@...app.com>,
        Steve French <sfrench@...ba.org>,
        Dominique Martinet <asmadeus@...ewreck.org>,
        Christoph Hellwig <hch@....de>,
        Alexander Viro <viro@...iv.linux.org.uk>,
        Linux-MM <linux-mm@...ck.org>, linux-cachefs@...hat.com,
        linux-afs@...ts.infradead.org,
        "open list:NFS, SUNRPC, AND..." <linux-nfs@...r.kernel.org>,
        CIFS <linux-cifs@...r.kernel.org>, ceph-devel@...r.kernel.org,
        v9fs-developer@...ts.sourceforge.net,
        linux-fsdevel <linux-fsdevel@...r.kernel.org>,
        Jeff Layton <jlayton@...hat.com>,
        David Wysochanski <dwysocha@...hat.com>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v4 02/28] mm: Add an unlock function for PG_private_2/PG_fscache

On Tue, Mar 16, 2021 at 7:12 PM Josef Bacik <josef@...icpanda.com> wrote:
>
>
> Yeah it's just a flag, we use it to tell that the page is part of a range that
> has been allocated for IO.  The lifetime of the page is independent of the page,
> but is generally either dirty or under writeback, so either it goes through
> truncate and we clear PagePrivate2 there, or it actually goes through IO and is
> cleared before we drop the page in our endio.

Ok, that's what it looked like from my very limited "looking at a
couple of grep cases", but I didn't go any further than that.

> We _always_ have PG_private set on the page as long as we own it, and
> PG_private_2 is only set in this IO related context, so we're safe
> there because of the rules around PG_dirty/PG_writeback. We don't need
> it to have an extra ref for it being set.

Perfect. That means that at least as far as btrfs is concerned, we
could trivially remove PG_private_2 from that page_has_private() math
- you'd always see the same result anyway, exactly because you have
PG_private set.

And as far as I can tell, fscache doesn't want that PG_private_2 bit
to interact with the random VM lifetime or migration rules either, and
should rely entirely on the page count. David?

There's actually a fair number of page_has_private() users, so we'd
better make sure that's the case. But it's simplified by this but
really only being used by btrfs (which doesn't care) and fscache, so
this cleanup would basically be entirely up to the whole fscache
series.

Hmm? Objections?

            Linus