linux-kernel - Re: [PATCH v2 0/2] drm/nouveau: remove drm_encoder

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <bpvyq543f4o3uusbjrivlymxeiqjcrnzb3atae2bp2vjleiggc@eolxb6jdep57>
Date: Mon, 16 Dec 2024 16:15:06 +0200
From: Dmitry Baryshkov <dmitry.baryshkov@...aro.org>
To: Danilo Krummrich <dakr@...nel.org>
Cc: Laurent Pinchart <laurent.pinchart@...asonboard.com>, 
	Maarten Lankhorst <maarten.lankhorst@...ux.intel.com>, Maxime Ripard <mripard@...nel.org>, 
	Thomas Zimmermann <tzimmermann@...e.de>, David Airlie <airlied@...il.com>, 
	Simona Vetter <simona@...ll.ch>, Karol Herbst <kherbst@...hat.com>, Lyude Paul <lyude@...hat.com>, 
	dri-devel@...ts.freedesktop.org, linux-kernel@...r.kernel.org, nouveau@...ts.freedesktop.org
Subject: Re: [PATCH v2 0/2] drm/nouveau: remove drm_encoder_slave interface

On Mon, Dec 16, 2024 at 02:30:55PM +0100, Danilo Krummrich wrote:
> On Mon, Dec 16, 2024 at 02:58:59PM +0200, Dmitry Baryshkov wrote:
> > On Mon, Dec 16, 2024 at 01:41:56PM +0100, Danilo Krummrich wrote:
> > > On Mon, Dec 16, 2024 at 02:16:51PM +0200, Laurent Pinchart wrote:
> > > > On Mon, Dec 16, 2024 at 02:11:41PM +0200, Dmitry Baryshkov wrote:
> > > > > On Mon, Dec 16, 2024 at 12:45:15PM +0100, Danilo Krummrich wrote:
> > > > > > On Sun, Dec 15, 2024 at 12:19:22PM +0200, Dmitry Baryshkov wrote:
> > > > > > > The nouveau driver is the only user of the drm_encoder_slave interface.
> > > > > > > Demote it from KMS helpers module to the nouveau driver itself, moving
> > > > > > > corresponding I2C encoders to be handled by nouveau driver too.
> > > > > > 
> > > > > > I understand nouveau is the only driver using this interface (and the
> > > > > > corresponding i2c encoders).
> > > > > > 
> > > > > > However, I'm not quite seeing the advantage of folding the interface (including
> > > > > > the two i2c drivers) into nouveau. I don't think this legacy interface does harm
> > > > > > the subsystem in any way / does prevent the subsystem from moving forward.
> > > > > > 
> > > > > > Can't we just keep it as it is?
> > > > > 
> > > > > Well, drm_encoder_slave is a part of the DRM KMS helpers module, so it
> > > > > take (a little bit) of space on every system. The nouveau situation
> > > > > isn't unique, other drivers (i915, ast) also incorporate the code for
> > > > > I2C backends. For the further discussion see the thread starting from
> > > > > Laurent's email ([1]).
> > > > > 
> > > > > [1] https://lore.kernel.org/all/20241117205426.GE12409@pendragon.ideasonboard.com/
> > > 
> > > The drm_encoder_slave code it's rather small, but I guess this can be used as
> > > argument for both, keeping it where it is and moving it.
> > > 
> > > If you want to move it to nouveau, I'm not going to object. But please fold the
> > > helper code, such that we aren't left with unused functions and unnecessary
> > > function pointer indirections through struct drm_encoder_slave_funcs.
> > 
> > This is more or less what I've done. Or would you prefer to keep the
> > wrapping functions that just execute the callback? I can change the
> > patchset accordingly.
> 
> No, I think it's good indeed -- st a first glance it looked like there's more to
> get rid of.
> 
> There are just a few more nits, I'll go ahead and add comments in the
> corresponding patches.
> 
> > 
> > > 
> > > > 
> > > > It's also a question of whether maintenance of this code based used by
> > > > the nouveau driver only should be the responsibility of the drm-misc
> > > > community or the nouveau driver maintainers.
> > > 
> > > Good question. It's common infrastructure; do we expect / require the last user
> > > of such infrastructure to take ownership?
> > 
> > Unfortunately it's more like 'the only one' :-( In other words, if we
> 
> I can't see a major difference between "last one" and "only one" in this
> context.

I was thinkinkg between "the last one" and "the only one, there will be
more in future". Excuse me.

> 
> > were expecting other users, there would not be such a move. But
> > hopefully all new drivers will use bridges infrastructure.
> 
> Agreed, but I don't think it answers my question.

Anyway, yes (in my humble opinion). it's usually easier to push bits of
infrasrucure towards the last user. Frequently it allows us to simplify
both the framework and the last user.

-- 
With best wishes
Dmitry