Thank you for your contribution! Sashiko AI review found 1 potential issue(s) 
to consider:

Pre-existing issues:
- [High] Missing synchronization lock between DRM atomic disable and ALSA 
callbacks leaves a Time-Of-Check to Time-Of-Use (TOCTOU) race condition, 
failing to fix the kernel panic.
--

commit 41e47d6bcdba49646deba6ac74845c934529f2f4
Author: Frank Zhang <[email protected]>

drm/bridge: dw-hdmi-qp: Guard clear_audio_infoframe when PHY is down

This commit attempts to fix a kernel panic caused when an audio server
closes the HDMI audio device concurrently with an atomic_disable. It tries
to guard register accesses in the audio teardown path by checking the
tmds_char_rate state variable.

> diff --git a/drivers/gpu/drm/bridge/synopsys/dw-hdmi-qp.c 
> b/drivers/gpu/drm/bridge/synopsys/dw-hdmi-qp.c
> index cfd1b18c06e62..f0e9324249d66 100644
> --- a/drivers/gpu/drm/bridge/synopsys/dw-hdmi-qp.c
> +++ b/drivers/gpu/drm/bridge/synopsys/dw-hdmi-qp.c
[ ... ]
> @@ -893,6 +893,16 @@ static int 
> dw_hdmi_qp_bridge_clear_audio_infoframe(struct drm_bridge *bridge)
>       return 0;
>  }
>  
> +static int dw_hdmi_qp_bridge_clear_audio_infoframe(struct drm_bridge *bridge)
> +{
> +     struct dw_hdmi_qp *hdmi = bridge->driver_private;
> +
> +     if (hdmi->tmds_char_rate > 0)
> +             return __dw_hdmi_qp_bridge_clear_audio_infoframe(bridge);
> +
> +     return 0;
> +}
> +

[Severity: High]
This isn't a bug introduced by this patch, but does this lockless check
leave an open Time-Of-Check to Time-Of-Use (TOCTOU) race condition?

If a userspace audio server (like pipewire) closes the HDMI audio device at
the same time as a DRM modeset disables the display, the ALSA teardown path
can execute dw_hdmi_qp_bridge_clear_audio_infoframe() and successfully
evaluate this unguarded check.

Concurrently, dw_hdmi_qp_bridge_atomic_disable() can execute:

    hdmi->tmds_char_rate = 0;
    hdmi->phy.ops->disable(hdmi, hdmi->phy.data);

This zeroes the state variable and powers off the PHY without
synchronization. The ALSA thread can then resume and invoke
__dw_hdmi_qp_bridge_clear_audio_infoframe(), writing to the now-disabled
PHY registers and triggering the Asynchronous SError Interrupt panic this
patch is trying to fix.

Additionally, could the same race exist in dw_hdmi_qp_audio_prepare()?

    if (!hdmi->tmds_char_rate)
        return -ENODEV;
    
    // ... further down ...
    dw_hdmi_qp_set_audio_interface(hdmi, fmt, hparms);

This path configures multiple registers after a lockless check, making the
entire preparation sequence vulnerable to the same race condition.

-- 
Sashiko AI review ยท 
https://sashiko.dev/#/patchset/[email protected]?part=1

Reply via email to