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;
Date:   Sat, 19 Nov 2022 04:46:17 +0000
From:   Al Viro <viro@zeniv.linux.org.uk>
To:     Peilin Ye <yepeilin.cs@gmail.com>
Cc:     Peilin Ye <peilin.ye@bytedance.com>,
        Cong Wang <cong.wang@bytedance.com>,
        linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH v2] coredump: Use vmsplice_to_pipe() for pipes in
 dump_emit_page()
Message-ID: <Y3hfmYF6b5T35Xqi@ZenIV>
References: <20221029005147.2553-1-yepeilin.cs@gmail.com>
 <20221031210349.3346-1-yepeilin.cs@gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20221031210349.3346-1-yepeilin.cs@gmail.com>
Sender: Al Viro <viro@ftp.linux.org.uk>
Precedence: bulk

On Mon, Oct 31, 2022 at 02:03:49PM -0700, Peilin Ye wrote:

> +	n = vmsplice_to_pipe(file, &iter, 0);
> +	if (n == -EBADF)
> +		n = __kernel_write_iter(cprm->file, &iter, &pos);

Yuck.  If anything, I would rather put a flag into coredump_params
and check it instead; this check for -EBADF is both unidiomatic and
brittle.  Suppose someday somebody looks at vmsplice(2) and
decides that it would make sense to lift the "is it a pipe" check
into e.g. vmsplice_type().  There's no obvious reasons not to,
unless one happens to know that coredump relies upon that check done
in vmsplice_to_pipe().  It's asking for trouble several years down
the road.

Make it explicit and independent from details of error checking
in vmsplice(2).