Received-SPF: pass (google.com: domain of linux-kernel+bounces-94132-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.80.249 as permitted sender) client-ip=147.75.80.249;
Message-ID: <f42ceee61ddb8b50c347589649d4131476ab5d81.camel@gmail.com>
Subject: Re: [PATCH v4 1/2] driver core: Introduce device_link_wait_removal()
From: Nuno =?ISO-8859-1?Q?S=E1?= <noname.nuno@gmail.com>
To: Herve Codina <herve.codina@bootlin.com>
Cc: "Rafael J. Wysocki" <rafael@kernel.org>, Greg Kroah-Hartman
 <gregkh@linuxfoundation.org>, Rob Herring <robh+dt@kernel.org>, Frank
 Rowand <frowand.list@gmail.com>, Saravana Kannan <saravanak@google.com>,
 Lizhi Hou <lizhi.hou@amd.com>, Max Zhen <max.zhen@amd.com>, Sonal Santan
 <sonal.santan@amd.com>, Stefano Stabellini <stefano.stabellini@xilinx.com>,
  Jonathan Cameron <Jonathan.Cameron@huawei.com>,
 linux-kernel@vger.kernel.org, devicetree@vger.kernel.org, Allan Nielsen
 <allan.nielsen@microchip.com>, Horatiu Vultur
 <horatiu.vultur@microchip.com>, Steen Hegelund
 <steen.hegelund@microchip.com>, Luca Ceresoli <luca.ceresoli@bootlin.com>,
 Nuno Sa <nuno.sa@analog.com>, Thomas Petazzoni
 <thomas.petazzoni@bootlin.com>,  stable@vger.kernel.org
Date: Wed, 06 Mar 2024 16:13:15 +0100
In-Reply-To: <20240306160101.25b45335@bootlin.com>
References: <20240306085007.169771-1-herve.codina@bootlin.com>
	 <20240306085007.169771-2-herve.codina@bootlin.com>
	 <1fff8742a13c28dd7e1dda47ad2d6fa8e21e421e.camel@gmail.com>
	 <CAJZ5v0gWCo9nDAHkzeD08tTKoE0DE0ocht-Qq4zA7P59y9KeuQ@mail.gmail.com>
	 <ed442b6916016b3a40782dc32538fc517715db6c.camel@gmail.com>
	 <CAJZ5v0iQNEj6e_L1=uBTPaWn7BqV4pnoWxUq7LRPe5iVWsaifw@mail.gmail.com>
	 <ec7705f410bc848e79b8ab878b5fbf7618d9456d.camel@gmail.com>
	 <CAJZ5v0iMUOJmm99H6SgfP9179hBsLdyC+1ixJwBxSP0b18V6XA@mail.gmail.com>
	 <86a0f91675197a00bbd921d6e57d2f3c57796e68.camel@gmail.com>
	 <20240306160101.25b45335@bootlin.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
User-Agent: Evolution 3.50.4 
Precedence: bulk
MIME-Version: 1.0

On Wed, 2024-03-06 at 16:01 +0100, Herve Codina wrote:
> Hi Nuno,
>=20
> On Wed, 06 Mar 2024 15:50:44 +0100
> Nuno S=C3=A1 <noname.nuno@gmail.com> wrote:
>=20
> ...
> > > > > >=20
> > > > > > That makes sense but then the only thing I still don't fully ge=
t is
> > > > > > why
> > > > > > we
> > > > > > have
> > > > > > a separate devlink_class_init() initcall for registering the de=
vlink
> > > > > > class
> > > > > > (which can also fail)...=C2=A0=20
> > > > >=20
> > > > > Well, I haven't added it. :-)
> > > > > =C2=A0=20
> > > > > > What I take from the above is that we should fail the
> > > > > > driver model if one of it's fundamental components fails so I w=
ould
> > > > > > say
> > > > > > we
> > > > > > should merge devlink_class_init() with device_init() otherwise =
it's
> > > > > > a
> > > > > > bit
> > > > > > confusing (at least to me) and gives the idea that it's ok for =
the
> > > > > > driver
> > > > > > model
> > > > > > to exist without the links (unless I'm missing some other reaso=
n for
> > > > > > the
> > > > > > devlink
> > > > > > init function).=C2=A0=20
> > > > >=20
> > > > > +1
> > > > >=20
> > > > > Feel free to send a patch along these lines, chances are that it =
will
> > > > > be popular. ;-)=C2=A0=20
> > > >=20
> > > > I was actually thinking about that but I think I encountered the re=
ason
> > > > why
> > > > we
> > > > have it like this... devices_init() is called from driver_init() an=
d
> > > > there
> > > > we
> > > > have:
> > > >=20
> > > > ...
> > > >=20
> > > > devices_init();
> > > > buses_init();
> > > > classes_init();
> > > >=20
> > > > ...
> > > >=20
> > > > So classes are initialized after devices which means we can't reall=
y do
> > > > class_register(&devlink_class) from devices_init(). Unless, of cour=
se,
> > > > we
> > > > re-
> > > > order things in driver_init() but that would be a questionable chan=
ge at
> > > > the
> > > > very least.
> > > >=20
> > > > So, while I agree with what you've said, I'm still not sure if mixi=
ng
> > > > devlink
> > > > stuff between devices_init() and devlink_class_init() is the best t=
hing
> > > > to
> > > > do
> > > > given that we already have the case where devlink_class_init() can =
fail
> > > > while
> > > > the driver model is up.=C2=A0=20
> > >=20
> > > So why don't you make devlink_class_init() do a BUG() on failure
> > > instead of returning an error?=C2=A0 IMO crashing early is better tha=
n
> > > crashing later or otherwise failing in a subtle way due to a missed
> > > dependency.=C2=A0=20
> >=20
> > Well, I do agree with that... Maybe that's something that Herve can sne=
ak in
> > this patch? Otherwise, I can later (after this one is applied) send a p=
atch
> > for
> > it.
>=20
> Well, I don't thing that this have to be part of this current series.
> It is an other topic and should be handled out of this current series.
>=20

Yeah, fair enough... IMHO, then I would say that we should still have the
workqueue moved to devlink_class_init() and I can then follow up with a pat=
ch to
do BUG_ON(ret) in it.

Alternatively I can also just move it in the follow up patch but I don't th=
ink
it makes much sense.

- Nuno S=C3=A1