Received-SPF: pass (google.com: domain of linux-wireless-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <cover.1660739276.git.duoming@zju.edu.cn> <b861ce56ba555109a67f85a146a785a69f0a3c95.1660739276.git.duoming@zju.edu.cn>
 <YvzicURy8t2JdQke@kroah.com> <176e7de7.8a223.182ac1fbc47.Coremail.duoming@zju.edu.cn>
 <Yv5TefZcrUPY1Qjc@kroah.com> <5108e03b.8c156.182b1a2973f.Coremail.duoming@zju.edu.cn>
In-Reply-To: <5108e03b.8c156.182b1a2973f.Coremail.duoming@zju.edu.cn>
From:   Brian Norris <briannorris@chromium.org>
Date:   Mon, 22 Aug 2022 11:21:36 -0700
Message-ID: <CA+ASDXNN81VczMTUt8=AyeytbMjy2vAGi_aVW_MNha9D99Z5VA@mail.gmail.com>
Subject: Re: [PATCH v7 1/2] devcoredump: remove the useless gfp_t parameter in
 dev_coredumpv and dev_coredumpm
To:     Duoming Zhou <duoming@zju.edu.cn>
Cc:     Greg KH <gregkh@linuxfoundation.org>,
        Linux Kernel <linux-kernel@vger.kernel.org>,
        linux-wireless <linux-wireless@vger.kernel.org>,
        amit karwar <amitkarwar@gmail.com>,
        Ganapathi Bhat <ganapathi017@gmail.com>,
        Sharvari Harisangam <sharvari.harisangam@nxp.com>,
        Xinming Hu <huxinming820@gmail.com>, kvalo@kernel.org,
        "David S. Miller" <davem@davemloft.net>,
        Eric Dumazet <edumazet@google.com>,
        Jakub Kicinski <kuba@kernel.org>,
        Paolo Abeni <pabeni@redhat.com>,
        "<netdev@vger.kernel.org>" <netdev@vger.kernel.org>,
        Johannes Berg <johannes@sipsolutions.net>,
        "Rafael J. Wysocki" <rafael@kernel.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

Hi,

On Thu, Aug 18, 2022 at 8:47 AM <duoming@zju.edu.cn> wrote:
> On Thu, 18 Aug 2022 16:58:01 +0200 Greg KH wrote:
> > No, that is not necessary.  Do the work now so that there is no flag day
> > and you don't have to worry about new users, it will all "just work".
>
> Do you mean we should replace dev_set_name() in dev_coredumpm() to some other
> functions that could work both in interrupt context and process context?

No.

I believe the suggestion is that rather than change the signature for
dev_coredumpv() (which means everyone has to agree on the new
signature on day 1), you should introduce a new API, like
dev_coredumpv_noatomic() (I'm not good at naming [1]) with the
signature you want, and then migrate users over. Once we have a
release with no users of the old API, we drop it.

There are plenty of examples of the kernel community doing similar
transitions. You can search around for examples, but a quick search of
my own shows something like this:
https://lwn.net/Articles/735887/
(In particular, timer_setup() was introduced, and all setup_timer()
users were migrated to it within a release or two.)

Brian

[1] Seriously, dev_coredumpv_noatomic() is not a name I want to see
last very long. Maybe some other trivial modification? Examples:

dev_core_dumpv()
dev_coredump_v()
device_coredumpv()
...