linux-kernel - Re: [RFC PATCH] drm/panel: Make backlight attachment lazy

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] [thread-next>] [day] [month] [year] [list]
Message-ID: <X8/yrXmEwuq6YZaD@builder.lan>
Date:   Tue, 8 Dec 2020 15:39:57 -0600
From:   Bjorn Andersson <bjorn.andersson@...aro.org>
To:     Sam Ravnborg <sam@...nborg.org>
Cc:     Thierry Reding <thierry.reding@...il.com>,
        Maarten Lankhorst <maarten.lankhorst@...ux.intel.com>,
        Maxime Ripard <mripard@...nel.org>,
        Thomas Zimmermann <tzimmermann@...e.de>,
        David Airlie <airlied@...ux.ie>,
        Daniel Vetter <daniel@...ll.ch>,
        dri-devel@...ts.freedesktop.org, linux-kernel@...r.kernel.org
Subject: Re: [RFC PATCH] drm/panel: Make backlight attachment lazy

On Mon 07 Dec 23:48 CST 2020, Sam Ravnborg wrote:

> Hi Bjorn,
> On Mon, Dec 07, 2020 at 10:44:46PM -0600, Bjorn Andersson wrote:
> > Some bridge chips, such as the TI SN65DSI86 DSI/eDP bridge, provides
> > means of generating a PWM signal for backlight control of the attached
> > panel. The provided PWM chip is typically controlled by the
> > pwm-backlight driver, which if tied to the panel will provide DPMS.
> > 
> > But with the current implementation the panel will refuse to probe
> > because the bridge driver has yet to probe and register the PWM chip,
> > and the bridge driver will refuse to probe because it's unable to find
> > the panel.
> > 
> > Mitigate this catch-22 situation by allowing the panel driver to probe
> > and retry the attachment of the backlight as the panel is turned on or
> > off.
> > 
> > Signed-off-by: Bjorn Andersson <bjorn.andersson@...aro.org>
> > ---
> >  drivers/gpu/drm/drm_panel.c | 47 +++++++++++++++++++++++++++----------
> >  include/drm/drm_panel.h     |  8 +++++++
> >  2 files changed, 43 insertions(+), 12 deletions(-)
> > 
> > diff --git a/drivers/gpu/drm/drm_panel.c b/drivers/gpu/drm/drm_panel.c
> > index f634371c717a..7487329bd22d 100644
> > --- a/drivers/gpu/drm/drm_panel.c
> > +++ b/drivers/gpu/drm/drm_panel.c
> > @@ -43,6 +43,34 @@ static LIST_HEAD(panel_list);
> >   * take look at drm_panel_bridge_add() and devm_drm_panel_bridge_add().
> >   */
> >  
> > +#if IS_REACHABLE(CONFIG_BACKLIGHT_CLASS_DEVICE)
> > +static int drm_panel_of_backlight_lazy(struct drm_panel *panel)
> > +{
> > +	struct backlight_device *backlight;
> > +
> > +	if (!panel || !panel->dev)
> > +		return -EINVAL;
> > +
> > +	backlight = devm_of_find_backlight(panel->dev);
> > +
> > +	if (IS_ERR(backlight)) {
> > +		if (PTR_ERR(backlight) == -EPROBE_DEFER) {
> > +			panel->backlight_init_pending = true;
> > +			return 0;
> > +		}
> > +
> > +		return PTR_ERR(backlight);
> Use dev_err_probe()
> 

I need special handling of EPROBE_DEFER, both in terms of remembering
that we should retry and to not pass the error back to the panel driver.

I also don't want to introduce an error print here.

> > +	}
> > +
> > +	panel->backlight = backlight;
> > +	panel->backlight_init_pending = false;
> > +
> > +	return 0;
> > +}
> > +#else
> > +static int drm_panel_of_backlight_lazy(struct drm_panel *panel) { return 0; }
> > +#endif
> > +
> >  /**
> >   * drm_panel_init - initialize a panel
> >   * @panel: DRM panel
> > @@ -161,6 +189,9 @@ int drm_panel_enable(struct drm_panel *panel)
> >  			return ret;
> >  	}
> >  
> > +	if (panel->backlight_init_pending)
> > +		drm_panel_of_backlight_lazy(panel);
> > +
> >  	ret = backlight_enable(panel->backlight);
> >  	if (ret < 0)
> >  		DRM_DEV_INFO(panel->dev, "failed to enable backlight: %d\n",
> > @@ -187,6 +218,9 @@ int drm_panel_disable(struct drm_panel *panel)
> >  	if (!panel)
> >  		return -EINVAL;
> >  
> > +	if (panel->backlight_init_pending)
> > +		drm_panel_of_backlight_lazy(panel);
> > +
> >  	ret = backlight_disable(panel->backlight);
> >  	if (ret < 0)
> >  		DRM_DEV_INFO(panel->dev, "failed to disable backlight: %d\n",
> > @@ -328,18 +362,7 @@ EXPORT_SYMBOL(of_drm_get_panel_orientation);
> >   */
> >  int drm_panel_of_backlight(struct drm_panel *panel)
> >  {
> > -	struct backlight_device *backlight;
> > -
> > -	if (!panel || !panel->dev)
> > -		return -EINVAL;
> > -
> > -	backlight = devm_of_find_backlight(panel->dev);
> > -
> > -	if (IS_ERR(backlight))
> > -		return PTR_ERR(backlight);
> > -
> > -	panel->backlight = backlight;
> > -	return 0;
> > +	return drm_panel_of_backlight_lazy(panel);
> Could you update the drm_panel_of_backlight() implementation (and
> do not forget the documentation) and avoid drm_panel_of_backlight_lazy()?
> 

That sounds reasonable, there's not really a reason for introducing a
new function for what I'm doing.

> 
> >  }
> >  EXPORT_SYMBOL(drm_panel_of_backlight);
> >  #endif
> > diff --git a/include/drm/drm_panel.h b/include/drm/drm_panel.h
> > index 33605c3f0eba..b126abebb2f3 100644
> > --- a/include/drm/drm_panel.h
> > +++ b/include/drm/drm_panel.h
> > @@ -149,6 +149,14 @@ struct drm_panel {
> >  	 */
> >  	struct backlight_device *backlight;
> >  
> > +	/**
> > +	 * @backlight_init_pending
> > +	 *
> > +	 * Backlight driver is not yet available so further attempts to
> > +	 * initialize @backlight is necessary.
> > +	 */
> > +	bool backlight_init_pending;
> > +
> 
> We have not done so today for other fields, but it would be good
> to document this is for drm_panel use only and drivers shall not touch.
> 

Of course.

Thanks,
Bjorn