Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Organization: Red Hat UK Ltd. Registered Address: Red Hat UK Ltd, Amberley
        Place, 107-111 Peascod Street, Windsor, Berkshire, SI4 1TE, United
        Kingdom.
        Registered in England and Wales under Company Registration No. 3798903
From:   David Howells <dhowells@redhat.com>
In-Reply-To: <084d78a4-6052-f2ec-72f2-af9c4979f5dc@linux.alibaba.com>
References: <084d78a4-6052-f2ec-72f2-af9c4979f5dc@linux.alibaba.com> <166757987929.950645.12595273010425381286.stgit@warthog.procyon.org.uk> <166757988611.950645.7626959069846893164.stgit@warthog.procyon.org.uk>
To:     JeffleXu <jefflexu@linux.alibaba.com>
Cc:     dhowells@redhat.com, willy@infradead.org,
        Jeff Layton <jlayton@kernel.org>, linux-cachefs@redhat.com,
        linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH v2 2/2] netfs: Fix dodgy maths
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <2363339.1667923148.1@warthog.procyon.org.uk>
Date:   Tue, 08 Nov 2022 15:59:08 +0000
Message-ID: <2363340.1667923148@warthog.procyon.org.uk>
Precedence: bulk

JeffleXu <jefflexu@linux.alibaba.com> wrote:

> > Fix the dodgy maths in netfs_rreq_unlock_folios().  start_page could be
> > inside the folio, in which case the calculation of pgpos will be come up
> > with a negative number (though for the moment rreq->start is rounded down
> > earlier and folios would have to get merged whilst locked)
> 
> Hi, the patch itself seems fine. Just some questions about the scenario.
> 
> 1. "start_page could be inside the folio" Is that because
> .expand_readahead() called from netfs_readahead()? Since otherwise,
> req-start is always aligned to the folio boundary.

At the moment, rreq->start is always coincident with the start of the first
folio in the collection because we always read whole folios - however, it
might be best to assume that this might not always hold true if it's simple to
fix the maths to get rid of the assumption.

> 2. If start_page is indeed inside the folio, then only the trailing part
> of the first folio can be covered by the request, and this folio will be
> marked with uptodate, though the beginning part of the folio may have
> not been read from the cache. Is that expected? Or correct me if I'm wrong.

For the moment there's no scenario where this arises; I think we need to wait
until we have a scenario and then see how we'll need to juggle the flags.

David