Intel-XE Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: Imre Deak <imre.deak@intel.com>
To: Dibin Moolakadan Subrahmanian <dibin.moolakadan.subrahmanian@intel.com>
Cc: <intel-gfx@lists.freedesktop.org>,
	<intel-xe@lists.freedesktop.org>, <ankit.k.nautiyal@intel.com>,
	<uma.shankar@intel.com>
Subject: Re: [PATCH] drm/xe/display: Block hpd during suspend
Date: Tue, 22 Jul 2025 13:47:15 +0300	[thread overview]
Message-ID: <aH9sMxdRfHQB41PE@ideak-desk> (raw)
In-Reply-To: <20250717085425.1347043-1-dibin.moolakadan.subrahmanian@intel.com>

On Thu, Jul 17, 2025 at 02:24:25PM +0530, Dibin Moolakadan Subrahmanian wrote:
> It has been observed that during `xe_display_pm_suspend()` execution,
> an HPD interrupt can still be triggered, resulting in `dig_port_work`
> being scheduled. The issue arises when this work executes after
> `xe_display_pm_suspend_late()`, by which time the display is fully
> suspended.
> 
> This can lead to errors such as "DC state mismatch", as the dig_port
> work accesses display resources that are no longer available or
> powered.
> 
> To address this, introduce  'intel_hpd_block_all_encoders()' and
> 'intel_hpd_unblock_all_encoders()' functions, which iterate over all
> encoders and block/unblock HPD respectively.
> 
> These are used to:
> - Block HPD IRQs before calling 'intel_hpd_cancel_work' in suspend
> - Unblock HPD IRQs after 'intel_hpd_init' in resume
> 
> This will prevent 'dig_port_work' being scheduled during display
> suspend.
> 
> Continuation of previous patch discussion:
> https://patchwork.freedesktop.org/patch/663964/
> 
> Signed-off-by: Dibin Moolakadan Subrahmanian <dibin.moolakadan.subrahmanian@intel.com>
> ---
>  drivers/gpu/drm/i915/display/intel_hotplug.c | 24 ++++++++++++++++++--
>  drivers/gpu/drm/i915/display/intel_hotplug.h |  2 ++
>  drivers/gpu/drm/xe/display/xe_display.c      |  4 ++++
>  3 files changed, 28 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/gpu/drm/i915/display/intel_hotplug.c b/drivers/gpu/drm/i915/display/intel_hotplug.c
> index 265aa97fcc75..7ffd8ded1c26 100644
> --- a/drivers/gpu/drm/i915/display/intel_hotplug.c
> +++ b/drivers/gpu/drm/i915/display/intel_hotplug.c
> @@ -223,6 +223,28 @@ queue_detection_work(struct intel_display *display, struct work_struct *work)
>  	return queue_work(display->wq.unordered, work);
>  }
>  
> +void intel_hpd_unblock_all_encoders(struct intel_display *display)
> +{
> +	struct intel_encoder *encoder;
> +
> +	if (!HAS_DISPLAY(display))
> +		return;
> +
> +	for_each_intel_encoder(display->drm, encoder)
> +		intel_hpd_unblock(encoder);
> +}
> +
> +void intel_hpd_block_all_encoders(struct intel_display *display)
> +{
> +	struct intel_encoder *encoder;
> +
> +	if (!HAS_DISPLAY(display))
> +		return;
> +
> +	for_each_intel_encoder(display->drm, encoder)
> +		intel_hpd_block(encoder);
> +}

intel_encoder.c has the intel_encoder_{suspend,shutdown}_all() helpers
called in the same suspend/shutdown paths, so I'd also add these to there
calling them stg like intel_encoder_{block,unblock}_all_hpds().

> +
>  static void
>  intel_hpd_irq_storm_switch_to_polling(struct intel_display *display)
>  {
> @@ -971,8 +993,6 @@ void intel_hpd_cancel_work(struct intel_display *display)
>  
>  	spin_lock_irq(&display->irq.lock);
>  
> -	drm_WARN_ON(display->drm, get_blocked_hpd_pin_mask(display));
> -
>  	display->hotplug.long_hpd_pin_mask = 0;
>  	display->hotplug.short_hpd_pin_mask = 0;
>  	display->hotplug.event_bits = 0;
> diff --git a/drivers/gpu/drm/i915/display/intel_hotplug.h b/drivers/gpu/drm/i915/display/intel_hotplug.h
> index edc41c9d3d65..3938c93d2385 100644
> --- a/drivers/gpu/drm/i915/display/intel_hotplug.h
> +++ b/drivers/gpu/drm/i915/display/intel_hotplug.h
> @@ -34,5 +34,7 @@ void intel_hpd_debugfs_register(struct intel_display *display);
>  void intel_hpd_enable_detection_work(struct intel_display *display);
>  void intel_hpd_disable_detection_work(struct intel_display *display);
>  bool intel_hpd_schedule_detection(struct intel_display *display);
> +void intel_hpd_block_all_encoders(struct intel_display *display);
> +void intel_hpd_unblock_all_encoders(struct intel_display *display);
>  
>  #endif /* __INTEL_HOTPLUG_H__ */
> diff --git a/drivers/gpu/drm/xe/display/xe_display.c b/drivers/gpu/drm/xe/display/xe_display.c
> index e2e0771cf274..51584ea3ed09 100644
> --- a/drivers/gpu/drm/xe/display/xe_display.c
> +++ b/drivers/gpu/drm/xe/display/xe_display.c
> @@ -340,6 +340,8 @@ void xe_display_pm_suspend(struct xe_device *xe)
>  
>  	xe_display_flush_cleanup_work(xe);
>  
> +	intel_hpd_block_all_encoders(display);
> +

Yes, I still think blocking here only the HPD IRQs makes sense,
disabling all the display IRQs suggested by Maarten only later.

xe_display_pm_shutdown() also misses this.

Also xe_display_fini_early() - which is called after disabling all IRQs
- misses intel_hpd_cancel_work().

>  	intel_hpd_cancel_work(display);
>  
>  	if (has_display(xe)) {
> @@ -471,6 +473,8 @@ void xe_display_pm_resume(struct xe_device *xe)
>  
>  	intel_hpd_init(display);
>  
> +	intel_hpd_unblock_all_encoders(display);
> +
>  	if (has_display(xe)) {
>  		intel_display_driver_resume(display);
>  		drm_kms_helper_poll_enable(&xe->drm);
> -- 
> 2.43.0
> 

      parent reply	other threads:[~2025-07-22 10:47 UTC|newest]

Thread overview: 13+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-07-17  8:54 [PATCH] drm/xe/display: Block hpd during suspend Dibin Moolakadan Subrahmanian
2025-07-17  9:46 ` ✓ CI.KUnit: success for " Patchwork
2025-07-17 10:24 ` ✓ Xe.CI.BAT: " Patchwork
2025-07-17 11:54 ` [PATCH] " Maarten Lankhorst
2025-07-18  5:21   ` Dibin Moolakadan Subrahmanian
2025-07-18 11:31     ` Maarten Lankhorst
2025-07-18 11:21   ` Imre Deak
2025-07-18 13:54 ` ✗ Xe.CI.Full: failure for " Patchwork
2025-07-21 20:16 ` ✓ CI.KUnit: success for drm/xe/display: Block hpd during suspend (rev2) Patchwork
2025-07-21 21:37 ` ✓ Xe.CI.BAT: " Patchwork
2025-07-22  0:01 ` ✗ Xe.CI.Full: failure " Patchwork
2025-07-22  5:00   ` Dibin Moolakadan Subrahmanian
2025-07-22 10:47 ` Imre Deak [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=aH9sMxdRfHQB41PE@ideak-desk \
    --to=imre.deak@intel.com \
    --cc=ankit.k.nautiyal@intel.com \
    --cc=dibin.moolakadan.subrahmanian@intel.com \
    --cc=intel-gfx@lists.freedesktop.org \
    --cc=intel-xe@lists.freedesktop.org \
    --cc=uma.shankar@intel.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox