Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::3:2 as permitted sender) client-ip=2620:137:e000::3:2;
MIME-Version: 1.0
References: <20230915183707.2707298-1-willy@infradead.org>
In-Reply-To: <20230915183707.2707298-1-willy@infradead.org>
From:   Linus Torvalds <torvalds@linux-foundation.org>
Date:   Fri, 15 Sep 2023 17:31:43 -0700
Message-ID: <CAHk-=wjU44TsEkoae6HuJi8JcTHMr01JSi_ZhiVTVSwpKvBtXA@mail.gmail.com>
Subject: Re: [PATCH 00/17] Add folio_end_read
To:     "Matthew Wilcox (Oracle)" <willy@infradead.org>
Cc:     linux-kernel@vger.kernel.org, linux-fsdevel@vger.kernel.org,
        linux-arch@vger.kernel.org, Nicholas Piggin <npiggin@gmail.com>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Fri, 15 Sept 2023 at 11:37, Matthew Wilcox (Oracle)
<willy@infradead.org> wrote:
>
> I don't have any performance numbers; I'm hoping Nick might provide some
> since PPC seems particularly unhappy with write-after-write hazards.

I suspect you can't see the extra atomic in the IO paths.

The existing trick with bit #7 is because we do a lot of
page_lock/unlock pairs even when there is no actual IO. So it's worth
it because page_unlock() really traditionally shows up quite a bit.
But once you actually do IO, I think the effect is not measurable.

That said, the series doesn't look *wrong*, although I did note a few
things that just made me go "that looks very strange to me" in there.

                      Linus