Received-SPF: pass (google.com: domain of linux-nfs-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
From:   Benjamin Coddington <bcodding@redhat.com>
To:     Jidong Xiao <jidong.xiao@gmail.com>
Cc:     linux-nfs@vger.kernel.org
Subject: Re: Is Linux NFS block size 512 bytes or is this a bug?
Date:   Mon, 05 Dec 2022 12:38:01 -0500
Message-ID: <60C55655-4A83-4B93-AEEB-89C69EBEE472@redhat.com>
In-Reply-To: <CAG4AFWYGYtYOoi8BuFos__GSNrLVXBwcaDpVR0D6fRv3ibPkmA@mail.gmail.com>
References: <CAG4AFWYGYtYOoi8BuFos__GSNrLVXBwcaDpVR0D6fRv3ibPkmA@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain
Precedence: bulk

On 5 Dec 2022, at 0:44, Jidong Xiao wrote:

> Hi,
>
> I saw this line in fs/nfsd/nfs3xdr.c:
>
> /* used */
> p = xdr_encode_hyper(p, ((u64)stat->blocks) << 9);
>
> It seems we are left-shifting the blocks 9 bits, to get the number of
> bytes used. Is this a bug or we know the block size is always 512
> bytes?

By calling vfs_getattr(), which needs to behave for stat(2) syscall as
returning the number of 512 byte blocks.  I think we're stuck with the
traditional idea that blocks are always 512 bytes in linux, at least from
the perspective of stat().

The history of such is beyond my time, but I don't think this is a bug.

Ben