intel-xe.lists.freedesktop.org archive mirror
 help / color / mirror / Atom feed
From: "Ville Syrjälä" <ville.syrjala@linux.intel.com>
To: Jani Nikula <jani.nikula@linux.intel.com>
Cc: intel-gfx@lists.freedesktop.org, intel-xe@lists.freedesktop.org
Subject: Re: [PATCH] drm/i915: Walk crtcs in pipe order
Date: Thu, 20 Nov 2025 20:42:16 +0200	[thread overview]
Message-ID: <aR9hCIHDIJKtt_Su@intel.com> (raw)
In-Reply-To: <add11487eaa3a418d199d8f9b851e4dfbad0cf25@intel.com>

On Thu, Nov 20, 2025 at 07:47:54PM +0200, Jani Nikula wrote:
> On Thu, 20 Nov 2025, Ville Syrjala <ville.syrjala@linux.intel.com> wrote:
> > From: Ville Syrjälä <ville.syrjala@linux.intel.com>
> >
> > Currently our crtcs are registered in pipe order, and thus
> > all the for_intel_crtc*() iterators walk the crtcs in pipe
> > order. There are a bunch of places that more or less depend
> > on that. Eg. during plane updates and such we want joined
> > pipes to be processed back-to-back to give a better chance
> > of an atomic update across the whole set.
> >
> > When we start to register crtcs in a different order we don't
> > want to change the order in which the pipes get handled.
> > Decouple the for_each_intel_crtc*() iterators from the crtc
> > registration order by using a separate list which will be
> > sorted by the pipe rather than the crtc index.
> >
> > We could priobably use a simple array or something, but that
> > would require some kind of extra iterator variable for the
> > macros, and thus would require a lot more changes. Using
> > a linked list keeps the fallout minimal. We can look at
> > using a more optimal data structure later.
> 
> I think the list works fine.

I supose. Though we do walk these a lot, and we might be
hitting a few more cachelines now, but hopefully they
remain clean so there's no ping-pong between CPUs...

> 
> > I also added this extra junk to the atomic state iterators:
> > "(__i) = drm_crtc_index(&(crtc)->base), (void)(__i)"
> > even though the macro itself no longer needs the "__i" iterator.
> > This in case the "__i" is used by the caller, and to
> > avoid compiler warnings if it's completely unused now.
> 
> At a glance, I can't find any. Perhaps you could cook up some cocci to
> remove the parameter as follow-up?

IIRC I already took some kind of stab at hiding the __i inside
the macros (using the UNIQUE_ID stuff) but never posted it.
But I guess we can avoid that now. Might have to think about
doing the same to the drm core macros as well.

> 
> > Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
> > ---
> >  drivers/gpu/drm/i915/display/intel_crtc.c     | 20 +++++
> >  drivers/gpu/drm/i915/display/intel_display.h  | 90 ++++++++-----------
> >  .../gpu/drm/i915/display/intel_display_core.h |  3 +
> >  .../drm/i915/display/intel_display_driver.c   |  1 +
> >  .../drm/i915/display/intel_display_types.h    |  1 +
> >  drivers/gpu/drm/xe/display/xe_display.c       |  1 +
> >  6 files changed, 64 insertions(+), 52 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/i915/display/intel_crtc.c b/drivers/gpu/drm/i915/display/intel_crtc.c
> > index 153ff4b4b52c..7ebbde716238 100644
> > --- a/drivers/gpu/drm/i915/display/intel_crtc.c
> > +++ b/drivers/gpu/drm/i915/display/intel_crtc.c
> > @@ -209,6 +209,8 @@ static struct intel_crtc *intel_crtc_alloc(void)
> >  	crtc->base.state = &crtc_state->uapi;
> >  	crtc->config = crtc_state;
> >  
> > +	INIT_LIST_HEAD(&crtc->pipe_head);
> > +
> >  	return crtc;
> >  }
> >  
> > @@ -222,6 +224,8 @@ static void intel_crtc_destroy(struct drm_crtc *_crtc)
> >  {
> >  	struct intel_crtc *crtc = to_intel_crtc(_crtc);
> >  
> > +	list_del(&crtc->pipe_head);
> > +
> >  	cpu_latency_qos_remove_request(&crtc->vblank_pm_qos);
> >  
> >  	drm_crtc_cleanup(&crtc->base);
> > @@ -308,6 +312,20 @@ static const struct drm_crtc_funcs i8xx_crtc_funcs = {
> >  	.get_vblank_timestamp = intel_crtc_get_vblank_timestamp,
> >  };
> >  
> > +static void add_crtc_to_pipe_list(struct intel_display *display, struct intel_crtc *crtc)
> > +{
> > +	struct intel_crtc *iter;
> > +
> > +	list_for_each_entry(iter, &display->pipe_list, pipe_head) {
> > +		if (iter->pipe > crtc->pipe) {
> 
> Nitpick, somehow I found it easier to think of this in reverse order,
> "crtc->pipe < iter->pipe", but could be just me. *shrug*.

Yeah, I suppose that might be better. Would at least match the order
of arguments to list_add_tail(), which does match the order they
end up on the list.

> 
> > +			list_add_tail(&crtc->pipe_head, &iter->pipe_head);
> > +			return;
> > +		}
> > +	}
> > +
> > +	list_add_tail(&crtc->pipe_head, &display->pipe_list);
> > +}
> > +
> >  int intel_crtc_init(struct intel_display *display, enum pipe pipe)
> >  {
> >  	struct intel_plane *primary, *cursor;
> > @@ -398,6 +416,8 @@ int intel_crtc_init(struct intel_display *display, enum pipe pipe)
> >  	if (HAS_CASF(display))
> >  		drm_crtc_create_sharpness_strength_property(&crtc->base);
> >  
> > +	add_crtc_to_pipe_list(display, crtc);
> > +
> >  	return 0;
> >  
> >  fail:
> > diff --git a/drivers/gpu/drm/i915/display/intel_display.h b/drivers/gpu/drm/i915/display/intel_display.h
> > index bcc6ccb69d2b..ac83d4f09bb9 100644
> > --- a/drivers/gpu/drm/i915/display/intel_display.h
> > +++ b/drivers/gpu/drm/i915/display/intel_display.h
> > @@ -238,22 +238,23 @@ enum phy_fia {
> >  			    base.head)					\
> >  		for_each_if((intel_plane)->pipe == (intel_crtc)->pipe)
> >  
> > -#define for_each_intel_crtc(dev, intel_crtc)				\
> > -	list_for_each_entry(intel_crtc,					\
> > -			    &(dev)->mode_config.crtc_list,		\
> > -			    base.head)
> > +#define for_each_intel_crtc(dev, crtc) \
> > +	list_for_each_entry((crtc), \
> > +			    &to_intel_display(dev)->pipe_list, \
> > +			    pipe_head)
> >  
> > -#define for_each_intel_crtc_in_pipe_mask(dev, intel_crtc, pipe_mask)	\
> > -	list_for_each_entry(intel_crtc,					\
> > -			    &(dev)->mode_config.crtc_list,		\
> > -			    base.head)					\
> > -		for_each_if((pipe_mask) & BIT(intel_crtc->pipe))
> > +#define for_each_intel_crtc_reverse(dev, crtc) \
> > +	list_for_each_entry_reverse((crtc), \
> > +				    &to_intel_display(dev)->pipe_list, \
> > +				    pipe_head)
> >  
> > -#define for_each_intel_crtc_in_pipe_mask_reverse(dev, intel_crtc, pipe_mask)	\
> > -	list_for_each_entry_reverse((intel_crtc),				\
> > -				    &(dev)->mode_config.crtc_list,		\
> > -				    base.head)					\
> > -		for_each_if((pipe_mask) & BIT((intel_crtc)->pipe))
> > +#define for_each_intel_crtc_in_pipe_mask(dev, crtc, pipe_mask) \
> > +	for_each_intel_crtc((dev), (crtc)) \
> > +		for_each_if((pipe_mask) & BIT((crtc)->pipe))
> > +
> > +#define for_each_intel_crtc_in_pipe_mask_reverse(dev, crtc, pipe_mask) \
> > +	for_each_intel_crtc_reverse((dev), (crtc)) \
> > +		for_each_if((pipe_mask) & BIT((crtc)->pipe))
> >  
> >  #define for_each_intel_encoder(dev, intel_encoder)		\
> >  	list_for_each_entry(intel_encoder,			\
> > @@ -295,14 +296,6 @@ enum phy_fia {
> >  	     (__i)++) \
> >  		for_each_if(plane)
> >  
> > -#define for_each_old_intel_crtc_in_state(__state, crtc, old_crtc_state, __i) \
> > -	for ((__i) = 0; \
> > -	     (__i) < (__state)->base.dev->mode_config.num_crtc && \
> > -		     ((crtc) = to_intel_crtc((__state)->base.crtcs[__i].ptr), \
> > -		      (old_crtc_state) = to_intel_crtc_state((__state)->base.crtcs[__i].old_state), 1); \
> > -	     (__i)++) \
> > -		for_each_if(crtc)
> > -
> >  #define for_each_new_intel_plane_in_state(__state, plane, new_plane_state, __i) \
> >  	for ((__i) = 0; \
> >  	     (__i) < (__state)->base.dev->mode_config.num_total_plane && \
> > @@ -311,22 +304,6 @@ enum phy_fia {
> >  	     (__i)++) \
> >  		for_each_if(plane)
> >  
> > -#define for_each_new_intel_crtc_in_state(__state, crtc, new_crtc_state, __i) \
> > -	for ((__i) = 0; \
> > -	     (__i) < (__state)->base.dev->mode_config.num_crtc && \
> > -		     ((crtc) = to_intel_crtc((__state)->base.crtcs[__i].ptr), \
> > -		      (new_crtc_state) = to_intel_crtc_state((__state)->base.crtcs[__i].new_state), 1); \
> > -	     (__i)++) \
> > -		for_each_if(crtc)
> > -
> > -#define for_each_new_intel_crtc_in_state_reverse(__state, crtc, new_crtc_state, __i) \
> > -	for ((__i) = (__state)->base.dev->mode_config.num_crtc - 1; \
> > -	     (__i) >= 0  && \
> > -	     ((crtc) = to_intel_crtc((__state)->base.crtcs[__i].ptr), \
> > -	      (new_crtc_state) = to_intel_crtc_state((__state)->base.crtcs[__i].new_state), 1); \
> > -	     (__i)--) \
> > -		for_each_if(crtc)
> > -
> >  #define for_each_oldnew_intel_plane_in_state(__state, plane, old_plane_state, new_plane_state, __i) \
> >  	for ((__i) = 0; \
> >  	     (__i) < (__state)->base.dev->mode_config.num_total_plane && \
> > @@ -336,23 +313,32 @@ enum phy_fia {
> >  	     (__i)++) \
> >  		for_each_if(plane)
> >  
> > +#define for_each_old_intel_crtc_in_state(__state, crtc, old_crtc_state, __i) \
> > +	for_each_intel_crtc((__state)->base.dev, (crtc)) \
> > +		for_each_if(((__i) = drm_crtc_index(&(crtc)->base), (void)(__i), \
> > +			     (old_crtc_state) = intel_atomic_get_old_crtc_state((__state), (crtc))))
> > +
> > +#define for_each_new_intel_crtc_in_state(__state, crtc, new_crtc_state, __i) \
> > +	for_each_intel_crtc((__state)->base.dev, (crtc)) \
> > +		for_each_if(((__i) = drm_crtc_index(&(crtc)->base), (void)(__i), \
> > +			     (new_crtc_state) = intel_atomic_get_new_crtc_state((__state), (crtc))))
> > +
> > +#define for_each_new_intel_crtc_in_state_reverse(__state, crtc, new_crtc_state, __i) \
> > +	for_each_intel_crtc_reverse((__state)->base.dev, (crtc)) \
> > +		for_each_if(((__i) = drm_crtc_index(&(crtc)->base), (void)(__i), \
> > +			     (new_crtc_state) = intel_atomic_get_new_crtc_state((__state), (crtc))))
> > +
> >  #define for_each_oldnew_intel_crtc_in_state(__state, crtc, old_crtc_state, new_crtc_state, __i) \
> > -	for ((__i) = 0; \
> > -	     (__i) < (__state)->base.dev->mode_config.num_crtc && \
> > -		     ((crtc) = to_intel_crtc((__state)->base.crtcs[__i].ptr), \
> > -		      (old_crtc_state) = to_intel_crtc_state((__state)->base.crtcs[__i].old_state), \
> > -		      (new_crtc_state) = to_intel_crtc_state((__state)->base.crtcs[__i].new_state), 1); \
> > -	     (__i)++) \
> > -		for_each_if(crtc)
> > +	for_each_intel_crtc((__state)->base.dev, (crtc)) \
> > +		for_each_if(((__i) = drm_crtc_index(&(crtc)->base), (void)(__i), \
> > +			     (old_crtc_state) = intel_atomic_get_old_crtc_state((__state), (crtc)), \
> > +			     (new_crtc_state) = intel_atomic_get_new_crtc_state((__state), (crtc))))
> >  
> >  #define for_each_oldnew_intel_crtc_in_state_reverse(__state, crtc, old_crtc_state, new_crtc_state, __i) \
> > -	for ((__i) = (__state)->base.dev->mode_config.num_crtc - 1; \
> > -	     (__i) >= 0  && \
> > -	     ((crtc) = to_intel_crtc((__state)->base.crtcs[__i].ptr), \
> > -	      (old_crtc_state) = to_intel_crtc_state((__state)->base.crtcs[__i].old_state), \
> > -	      (new_crtc_state) = to_intel_crtc_state((__state)->base.crtcs[__i].new_state), 1); \
> > -	     (__i)--) \
> > -		for_each_if(crtc)
> > +	for_each_intel_crtc_reverse((__state)->base.dev, (crtc)) \
> > +		for_each_if(((__i) = drm_crtc_index(&(crtc)->base), (void)(__i), \
> > +			     (old_crtc_state) = intel_atomic_get_old_crtc_state((__state), (crtc)), \
> > +			     (new_crtc_state) = intel_atomic_get_new_crtc_state((__state), (crtc))))
> 
> These were a PITA to go through, but I think it's fine. Knocks wood.
> 
> Thanks for doing this.
> 
> Reviewed-by: Jani Nikula <jani.nikula@intel.com>
> 
> >  
> >  #define intel_atomic_crtc_state_for_each_plane_state( \
> >  		  plane, plane_state, \
> > diff --git a/drivers/gpu/drm/i915/display/intel_display_core.h b/drivers/gpu/drm/i915/display/intel_display_core.h
> > index 9b8414b77c15..4f4d5c314394 100644
> > --- a/drivers/gpu/drm/i915/display/intel_display_core.h
> > +++ b/drivers/gpu/drm/i915/display/intel_display_core.h
> > @@ -294,6 +294,9 @@ struct intel_display {
> >  	/* Parent, or core, driver functions exposed to display */
> >  	const struct intel_display_parent_interface *parent;
> >  
> > +	/* list of all intel_crtcs sorted by pipe */
> > +	struct list_head pipe_list;
> > +
> >  	/* Display functions */
> >  	struct {
> >  		/* Top level crtc-ish functions */
> > diff --git a/drivers/gpu/drm/i915/display/intel_display_driver.c b/drivers/gpu/drm/i915/display/intel_display_driver.c
> > index 7e000ba3e08b..32726906e550 100644
> > --- a/drivers/gpu/drm/i915/display/intel_display_driver.c
> > +++ b/drivers/gpu/drm/i915/display/intel_display_driver.c
> > @@ -119,6 +119,7 @@ static void intel_mode_config_init(struct intel_display *display)
> >  
> >  	drm_mode_config_init(display->drm);
> >  	INIT_LIST_HEAD(&display->global.obj_list);
> > +	INIT_LIST_HEAD(&display->pipe_list);
> >  
> >  	mode_config->min_width = 0;
> >  	mode_config->min_height = 0;
> > diff --git a/drivers/gpu/drm/i915/display/intel_display_types.h b/drivers/gpu/drm/i915/display/intel_display_types.h
> > index 38702a9e0f50..1c2bd9445795 100644
> > --- a/drivers/gpu/drm/i915/display/intel_display_types.h
> > +++ b/drivers/gpu/drm/i915/display/intel_display_types.h
> > @@ -1441,6 +1441,7 @@ struct intel_flipq {
> >  
> >  struct intel_crtc {
> >  	struct drm_crtc base;
> > +	struct list_head pipe_head;
> >  	enum pipe pipe;
> >  	/*
> >  	 * Whether the crtc and the connected output pipeline is active. Implies
> > diff --git a/drivers/gpu/drm/xe/display/xe_display.c b/drivers/gpu/drm/xe/display/xe_display.c
> > index e3320d9e6314..cfcbc7dd8638 100644
> > --- a/drivers/gpu/drm/xe/display/xe_display.c
> > +++ b/drivers/gpu/drm/xe/display/xe_display.c
> > @@ -22,6 +22,7 @@
> >  #include "intel_audio.h"
> >  #include "intel_bw.h"
> >  #include "intel_display.h"
> > +#include "intel_display_core.h"
> >  #include "intel_display_device.h"
> >  #include "intel_display_driver.h"
> >  #include "intel_display_irq.h"
> 
> -- 
> Jani Nikula, Intel

-- 
Ville Syrjälä
Intel

  reply	other threads:[~2025-11-20 18:42 UTC|newest]

Thread overview: 12+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-11-20 14:49 [PATCH] drm/i915: Walk crtcs in pipe order Ville Syrjala
2025-11-20 16:45 ` ✗ CI.checkpatch: warning for " Patchwork
2025-11-20 16:46 ` ✓ CI.KUnit: success " Patchwork
2025-11-20 17:43 ` ✓ Xe.CI.BAT: " Patchwork
2025-11-20 17:47 ` [PATCH] " Jani Nikula
2025-11-20 18:42   ` Ville Syrjälä [this message]
2025-11-20 19:21 ` [PATCH v2] " Ville Syrjala
2025-11-20 21:50 ` ✗ Xe.CI.Full: failure for " Patchwork
2025-11-20 22:49 ` ✗ CI.checkpatch: warning for drm/i915: Walk crtcs in pipe order (rev2) Patchwork
2025-11-20 22:50 ` ✓ CI.KUnit: success " Patchwork
2025-11-20 23:29 ` ✓ Xe.CI.BAT: " Patchwork
2025-11-21  4:19 ` ✗ Xe.CI.Full: failure " Patchwork

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=aR9hCIHDIJKtt_Su@intel.com \
    --to=ville.syrjala@linux.intel.com \
    --cc=intel-gfx@lists.freedesktop.org \
    --cc=intel-xe@lists.freedesktop.org \
    --cc=jani.nikula@linux.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;
as well as URLs for NNTP newsgroup(s).