linux-kernel - Re: [PATCH v11 1/4] drm/msm/dp: do not initialize phy until plugin interrupt received

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]

Date:   Thu, 13 Jan 2022 13:52:25 -0800
From:   Stephen Boyd <swboyd@...omium.org>
To:     Kuogee Hsieh <quic_khsieh@...cinc.com>, agross@...nel.org,
        airlied@...ux.ie, bjorn.andersson@...aro.org, daniel@...ll.ch,
        dmitry.baryshkov@...aro.org, dri-devel@...ts.freedesktop.org,
        robdclark@...il.com, sean@...rly.run, vkoul@...nel.org
Cc:     quic_abhinavk@...cinc.com, aravindh@...eaurora.org,
        quic_sbillaka@...cinc.com, freedreno@...ts.freedesktop.org,
        linux-arm-msm@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v11 1/4] drm/msm/dp: do not initialize phy until plugin
 interrupt received

Quoting Kuogee Hsieh (2022-01-13 12:44:16)
>
> On 1/13/2022 11:47 AM, Stephen Boyd wrote:
> > Quoting Kuogee Hsieh (2022-01-13 09:51:42)
> >> On 1/12/2022 8:13 PM, Stephen Boyd wrote:
> >>>>>> -       if (dp->usbpd->orientation == ORIENTATION_CC2)
> >>>>>> -               flip = true;
> >>>>>> +       dp_power_init(dp->power, false);
> >>>>>> +       dp_ctrl_reset_irq_ctrl(dp->ctrl, true);
> >>>>>> +
> >>>>>> +       /*
> >>>>>> +        * eDP is the embedded primary display and has its own phy
> >>>>>> +        * initialize phy immediately
> >>>>> Question still stands why we can't wait for hpd high from the eDP panel.
> >>>>> Also, I think "has its own phy" means that it's not part of a combo
> >>>>> USB+DP phy? Can you please clarify?
> Correct, eDP has its dedicated phy which is not part of combo phy.

Why does that mean we can't wait for hpd high from the eDP panel?

> >>>>>> +        */
> >>>>>> +       if (dp->dp_display.connector_type == DRM_MODE_CONNECTOR_eDP)
> >>>>>> +               dp_display_host_phy_init(dp);
> >>>>>>
> >>>>>> -       dp_power_init(dp->power, flip);
> >>>>>> -       dp_ctrl_host_init(dp->ctrl, flip, reset);
> >>>>>>            dp_aux_init(dp->aux);
> >>>>>>            dp->core_initialized = true;
> >>>>>>     }
> >>>>>> @@ -1306,20 +1330,23 @@ static int dp_pm_resume(struct device *dev)
> >>>>>>            dp->hpd_state = ST_DISCONNECTED;
> >>>>>>
> >>>>>>            /* turn on dp ctrl/phy */
> >>>>>> -       dp_display_host_init(dp, true);
> >>>>>> +       dp_display_host_init(dp);
> >>>>>>
> >>>>>>            dp_catalog_ctrl_hpd_config(dp->catalog);
> >>>>>>
> >>>>>> -       /*
> >>>>>> -        * set sink to normal operation mode -- D0
> >>>>>> -        * before dpcd read
> >>>>>> -        */
> >>>>>> -       dp_link_psm_config(dp->link, &dp->panel->link_info, false);
> >>>>>>
> >>>>>>            if (dp_catalog_link_is_connected(dp->catalog)) {
> >>>>>> +               /*
> >>>>>> +                * set sink to normal operation mode -- D0
> >>>>>> +                * before dpcd read
> >>>>>> +                */
> >>>>>> +               dp_display_host_phy_init(dp);
> >>>>>> +               dp_link_psm_config(dp->link, &dp->panel->link_info, false);
> >>>>>>                    sink_count = drm_dp_read_sink_count(dp->aux);
> >>>>>>                    if (sink_count < 0)
> >>>>>>                            sink_count = 0;
> >>>>>> +
> >>>>>> +               dp_display_host_phy_exit(dp);
> >>>>> Why is the phy exited on resume when the link is still connected? Is
> >>>>> this supposed to be done only when the sink_count is 0? And how does
> >>>>> this interact with eDP where the phy is initialized by the call to
> >>>>> dp_display_host_init() earlier in this function.
>
> At beginning of dp_pm_resume bot core_initialized and phy_initialized
> should be off.
>
> However at the case of dongle still connected to DUT, we have to read
> dongle dpcd to decided any hdmi connect to dongle (sink_count != 0). in
> this case, we have to turn on phy to perform dpcd read and  turn off phy
> after read so the following plugged-in interrupt can be handled correctly.
>

That looks like a lot of wasted work. Why can't we turn on the core,
turn on the phy, check if it's connected, and then turn off the phy if
it isn't? At the least, please put a comment above this phy_exit() call
indicating that we'll turn the phy back on while processing a plugged in
interrupt.