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: <ZnvFkwy1hBwxFxQL@phenom.ffwll.local>
Date: Wed, 26 Jun 2024 09:38:59 +0200
From: Daniel Vetter <daniel@...ll.ch>
To: Jonas Karlman <jonas@...boo.se>
Cc: neil.armstrong@...aro.org, Andrzej Hajda <andrzej.hajda@...el.com>,
	Robert Foss <rfoss@...nel.org>,
	Maarten Lankhorst <maarten.lankhorst@...ux.intel.com>,
	Maxime Ripard <mripard@...nel.org>,
	Thomas Zimmermann <tzimmermann@...e.de>,
	David Airlie <airlied@...il.com>, Daniel Vetter <daniel@...ll.ch>,
	Laurent Pinchart <Laurent.pinchart@...asonboard.com>,
	Jernej Skrabec <jernej.skrabec@...il.com>,
	dri-devel@...ts.freedesktop.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 1/8] drm: bridge: dw_hdmi: Call poweron/poweroff from
 atomic enable/disable

On Mon, Jun 24, 2024 at 02:19:36PM +0200, Jonas Karlman wrote:
> Hi Neil,
> 
> On 2024-06-24 11:56, neil.armstrong@...aro.org wrote:
> > On 24/06/2024 11:41, Jonas Karlman wrote:
> >> Hi Neil,
> >>
> >> On 2024-06-24 11:23, Neil Armstrong wrote:
> >>> On 11/06/2024 17:50, Jonas Karlman wrote:
> >>>> Change to only call poweron/poweroff from atomic_enable/atomic_disable
> >>>> ops instead of trying to keep a bridge_is_on state and poweron/off in
> >>>> the hotplug irq handler.
> >>>>
> >>>> A benefit of this is that drm mode_config mutex is always held at
> >>>> poweron/off, something that may reduce the need for our own mutex.
> >>>>
> >>>> Signed-off-by: Jonas Karlman <jonas@...boo.se>
> >>>> ---
> >>>>    drivers/gpu/drm/bridge/synopsys/dw-hdmi.c | 33 ++---------------------
> >>>>    1 file changed, 2 insertions(+), 31 deletions(-)
> >>>>
> >>>> diff --git a/drivers/gpu/drm/bridge/synopsys/dw-hdmi.c b/drivers/gpu/drm/bridge/synopsys/dw-hdmi.c
> >>>> index 9f2bc932c371..34bc6f4754b8 100644
> >>>> --- a/drivers/gpu/drm/bridge/synopsys/dw-hdmi.c
> >>>> +++ b/drivers/gpu/drm/bridge/synopsys/dw-hdmi.c
> >>>> @@ -172,7 +172,6 @@ struct dw_hdmi {
> >>>>    	enum drm_connector_force force;	/* mutex-protected force state */
> >>>>    	struct drm_connector *curr_conn;/* current connector (only valid when !disabled) */
> >>>>    	bool disabled;			/* DRM has disabled our bridge */
> >>>> -	bool bridge_is_on;		/* indicates the bridge is on */
> >>>>    	bool rxsense;			/* rxsense state */
> >>>>    	u8 phy_mask;			/* desired phy int mask settings */
> >>>>    	u8 mc_clkdis;			/* clock disable register */
> >>>> @@ -2383,8 +2382,6 @@ static void initialize_hdmi_ih_mutes(struct dw_hdmi *hdmi)
> >>>>    
> >>>>    static void dw_hdmi_poweron(struct dw_hdmi *hdmi)
> >>>>    {
> >>>> -	hdmi->bridge_is_on = true;
> >>>> -
> >>>>    	/*
> >>>>    	 * The curr_conn field is guaranteed to be valid here, as this function
> >>>>    	 * is only be called when !hdmi->disabled.
> >>>> @@ -2398,30 +2395,6 @@ static void dw_hdmi_poweroff(struct dw_hdmi *hdmi)
> >>>>    		hdmi->phy.ops->disable(hdmi, hdmi->phy.data);
> >>>>    		hdmi->phy.enabled = false;
> >>>>    	}
> >>>> -
> >>>> -	hdmi->bridge_is_on = false;
> >>>> -}
> >>>> -
> >>>> -static void dw_hdmi_update_power(struct dw_hdmi *hdmi)
> >>>> -{
> >>>> -	int force = hdmi->force;
> >>>> -
> >>>> -	if (hdmi->disabled) {
> >>>> -		force = DRM_FORCE_OFF;
> >>>> -	} else if (force == DRM_FORCE_UNSPECIFIED) {
> >>>> -		if (hdmi->rxsense)
> >>>> -			force = DRM_FORCE_ON;
> >>>> -		else
> >>>> -			force = DRM_FORCE_OFF;
> >>>> -	}
> >>>
> >>> This means we always poweron the bridge even if rxsense is false ?
> >>
> >> If I follow the logic there should not be any difference, beside that
> >> power on now only happen in atomic_enable instead of sometime in irq
> >> handler.
> >>
> >> hdmi->rxsense is set to true based on hpd in dw_hdmi_setup_rx_sense().
> >> For both meson and dw-hdmi this means HPD=1 and not rxsense=1.
> > 
> > Yeah I know, I was worried for other platforms using rxsense
> 
> From what I can find only dw-hdmi.c and dw_hdmi_meson.c are caller of
> dw_hdmi_setup_rx_sense(). For meson the same value is passed for both
> hpd and rx_sense, everyone else hpd = HPD and rx_sense = RX_SENSE status.
> 
> My understanding and testing based on Rockchip has shown that rx_sense
> has no significant impact before and after this change.
> 
> > 
> >>
> >> drm core will call the force/detect ops and enable/disable based on
> >> forced/HPD state, so I am not expecting any difference in how this
> >> currently works.
> >>
> >> This change to only poweron/setup in atomic_enable should also fix
> >> issues/situations where dw-hdmi was poweron too early during bootup in
> >> irq handler, before the drm driver was fully probed.
> 
> I may have been wrong here, there is a check for disabled here so it
> should not have setup() before atomic_enable().
> 
> Still we should ensure configure happen in atomic_enable(). The
> hpd_irq_event/hpd_notify calls will trigger a detect() and signal core
> if the bridge should be enabled/disabled when connector status changes.
> 
> > 
> > Hmm, but I thought the code wouldn't poweron the bridge is rxsense was 0
> > even if a mode was set, this won't work like this anymore right ?
> 
> This is what I thought was happening, and the comment in code seem to
> indicate that. However, the code only seem to care about HPD=1 status to
> poweron() and possible both HPD=0 + RXSENSE=0 to poweroff().

More fundamental question: Why do we even care what happens in this case?
Userspace lit up an output that's not connected either

- it's broken, it gets to keep to pieces

- it knows what it's doing, there is a display, it's just broken, and it
  _really_ wants the output to be lit up.

Drivers trying to be clever isn't a good idea here imo ...
-Sima

> To make matter more complex core also mainly care about detect() status
> and most/all related drivers seem to only use the the HPD status.
> 
> So I would say that there should not be any changes in impact of the
> rxsense signal.
> 
> Will also send a new patch in a v2 for dw_hdmi_irq() that may be related:
> 
> -		if (phy_stat & HDMI_PHY_HPD)
> +		if ((intr_stat & HDMI_IH_PHY_STAT0_HPD) &&
> +		    (phy_stat & HDMI_PHY_HPD))
> 			status = connector_status_connected;
> 
> or there may be a status_connected event triggered when rxsense goes
> from 1 to 0 and a second one with status_disconnected shortly after when
> hpd goes from 1 to 0 when the cable is removed.
> 
> Regards,
> Jonas
> 
> > 
> > Neil
> > 
> >>
> >> Regards,
> >> Jonas
> >>
> >>>
> >>> Neil
> >>>
> >>>> -
> >>>> -	if (force == DRM_FORCE_OFF) {
> >>>> -		if (hdmi->bridge_is_on)
> >>>> -			dw_hdmi_poweroff(hdmi);
> >>>> -	} else {
> >>>> -		if (!hdmi->bridge_is_on)
> >>>> -			dw_hdmi_poweron(hdmi);
> >>>> -	}
> >>>>    }
> >>>>    
> >>>>    /*
> >>>> @@ -2546,7 +2519,6 @@ static void dw_hdmi_connector_force(struct drm_connector *connector)
> >>>>    
> >>>>    	mutex_lock(&hdmi->mutex);
> >>>>    	hdmi->force = connector->force;
> >>>> -	dw_hdmi_update_power(hdmi);
> >>>>    	dw_hdmi_update_phy_mask(hdmi);
> >>>>    	mutex_unlock(&hdmi->mutex);
> >>>>    }
> >>>> @@ -2955,7 +2927,7 @@ static void dw_hdmi_bridge_atomic_disable(struct drm_bridge *bridge,
> >>>>    	mutex_lock(&hdmi->mutex);
> >>>>    	hdmi->disabled = true;
> >>>>    	hdmi->curr_conn = NULL;
> >>>> -	dw_hdmi_update_power(hdmi);
> >>>> +	dw_hdmi_poweroff(hdmi);
> >>>>    	dw_hdmi_update_phy_mask(hdmi);
> >>>>    	handle_plugged_change(hdmi, false);
> >>>>    	mutex_unlock(&hdmi->mutex);
> >>>> @@ -2974,7 +2946,7 @@ static void dw_hdmi_bridge_atomic_enable(struct drm_bridge *bridge,
> >>>>    	mutex_lock(&hdmi->mutex);
> >>>>    	hdmi->disabled = false;
> >>>>    	hdmi->curr_conn = connector;
> >>>> -	dw_hdmi_update_power(hdmi);
> >>>> +	dw_hdmi_poweron(hdmi);
> >>>>    	dw_hdmi_update_phy_mask(hdmi);
> >>>>    	handle_plugged_change(hdmi, true);
> >>>>    	mutex_unlock(&hdmi->mutex);
> >>>> @@ -3073,7 +3045,6 @@ void dw_hdmi_setup_rx_sense(struct dw_hdmi *hdmi, bool hpd, bool rx_sense)
> >>>>    		if (hpd)
> >>>>    			hdmi->rxsense = true;
> >>>>    
> >>>> -		dw_hdmi_update_power(hdmi);
> >>>>    		dw_hdmi_update_phy_mask(hdmi);
> >>>>    	}
> >>>>    	mutex_unlock(&hdmi->mutex);
> >>>
> >>
> > 
> 

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ