Intel-GFX Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: Ville Syrjala <ville.syrjala@linux.intel.com>
To: intel-gfx@lists.freedesktop.org
Cc: intel-xe@lists.freedesktop.org, Uma Shankar <uma.shankar@intel.com>
Subject: [PATCH 6/9] drm/i915/dmc: Reload pipe DMC MMIO registers for pipe C/D on PTL+
Date: Wed, 11 Jun 2025 18:52:38 +0300	[thread overview]
Message-ID: <20250611155241.24191-7-ville.syrjala@linux.intel.com> (raw)
In-Reply-To: <20250611155241.24191-1-ville.syrjala@linux.intel.com>

From: Ville Syrjälä <ville.syrjala@linux.intel.com>

On PTL+ the pipe DMC on pipes C/D loses its MMIO state occasionally.
Not quite sure what the specific sequence is that makes this happen
(eg. simply disabling PG2 doesn't seem to be enough to trigger this
on its own).

Reload the MMIO registers for the affected pipes when enabling the
pipe DMC. So far I've not see this happen on PTL pipe A/B, nor on any
pipe on any other post-TGL platform.

The DMC program RAM doesn't appear to need manual restoring, though
Windows appears to be doing exactly that on most platforms (for some
of the pipes). None of this is properly documented anywhere it seems.

Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
---
 drivers/gpu/drm/i915/display/intel_dmc.c | 32 +++++++++++++++++++-----
 1 file changed, 26 insertions(+), 6 deletions(-)

diff --git a/drivers/gpu/drm/i915/display/intel_dmc.c b/drivers/gpu/drm/i915/display/intel_dmc.c
index fd99c4645260..76b88c9bea02 100644
--- a/drivers/gpu/drm/i915/display/intel_dmc.c
+++ b/drivers/gpu/drm/i915/display/intel_dmc.c
@@ -575,8 +575,18 @@ static u32 dmc_mmiodata(struct intel_display *display,
 		return dmc->dmc_info[dmc_id].mmiodata[i];
 }
 
-static void dmc_load_program(struct intel_display *display,
-			     enum intel_dmc_id dmc_id)
+static void dmc_load_mmio(struct intel_display *display, enum intel_dmc_id dmc_id)
+{
+	struct intel_dmc *dmc = display_to_dmc(display);
+	int i;
+
+	for (i = 0; i < dmc->dmc_info[dmc_id].mmio_count; i++) {
+		intel_de_write(display, dmc->dmc_info[dmc_id].mmioaddr[i],
+			       dmc_mmiodata(display, dmc, dmc_id, i));
+	}
+}
+
+static void dmc_load_program(struct intel_display *display, enum intel_dmc_id dmc_id)
 {
 	struct intel_dmc *dmc = display_to_dmc(display);
 	int i;
@@ -593,10 +603,7 @@ static void dmc_load_program(struct intel_display *display,
 
 	preempt_enable();
 
-	for (i = 0; i < dmc->dmc_info[dmc_id].mmio_count; i++) {
-		intel_de_write(display, dmc->dmc_info[dmc_id].mmioaddr[i],
-			       dmc_mmiodata(display, dmc, dmc_id, i));
-	}
+	dmc_load_mmio(display, dmc_id);
 }
 
 static bool need_pipedmc_load_program(struct intel_display *display)
@@ -605,6 +612,17 @@ static bool need_pipedmc_load_program(struct intel_display *display)
 	return DISPLAY_VER(display) == 12;
 }
 
+static bool need_pipedmc_load_mmio(struct intel_display *display, enum pipe pipe)
+{
+	/*
+	 * On PTL pipe C/D PIPEDMC MMIO state is lost sometimes
+	 *
+	 * TODO figure out when exactly this happens, so far it
+	 * didn't seem 100% deterministic...
+	 */
+	return DISPLAY_VER(display) >= 30 && pipe >= PIPE_C;
+}
+
 void intel_dmc_enable_pipe(struct intel_display *display, enum pipe pipe)
 {
 	enum intel_dmc_id dmc_id = PIPE_TO_DMC_ID(pipe);
@@ -614,6 +632,8 @@ void intel_dmc_enable_pipe(struct intel_display *display, enum pipe pipe)
 
 	if (need_pipedmc_load_program(display))
 		dmc_load_program(display, dmc_id);
+	else if (need_pipedmc_load_mmio(display, pipe))
+		dmc_load_mmio(display, dmc_id);
 
 	if (DISPLAY_VER(display) >= 20) {
 		intel_de_write(display, PIPEDMC_INTERRUPT(pipe), pipedmc_interrupt_mask(display));
-- 
2.49.0


  parent reply	other threads:[~2025-06-11 15:53 UTC|newest]

Thread overview: 24+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-06-11 15:52 [PATCH 0/9] drm/i915/dmc: Deal with loss of pipe DMC state Ville Syrjala
2025-06-11 15:52 ` [PATCH 1/9] drm/i915/dmc: Limit pipe DMC clock gating w/a to just ADL/DG2/MTL Ville Syrjala
2025-06-11 15:52 ` [PATCH 2/9] drm/i915/dmc: Parametrize MTL_PIPEDMC_GATING_DIS Ville Syrjala
2025-06-11 15:52 ` [PATCH 3/9] drm/i915/dmc: Shuffle code around Ville Syrjala
2025-06-11 15:52 ` [PATCH 4/9] drm/i915/dmc: Extract dmc_load_program() Ville Syrjala
2025-06-12 20:16   ` Shankar, Uma
2025-06-13 14:18     ` Ville Syrjälä
2025-06-17 18:58       ` Shankar, Uma
2025-06-11 15:52 ` [PATCH 5/9] drm/i915/dmc: Reload pipe DMC state on TGL when enabling pipe A Ville Syrjala
2025-06-12 20:32   ` Shankar, Uma
2025-06-13 14:09     ` Ville Syrjälä
2025-06-17 18:51       ` Shankar, Uma
2025-06-11 15:52 ` Ville Syrjala [this message]
2025-06-11 15:52 ` [PATCH 7/9] drm/i915/dmc: Assert DMC is loaded harder Ville Syrjala
2025-06-12 20:55   ` Shankar, Uma
2025-06-11 15:52 ` [PATCH 8/9] drm/i915/dmc: Pass crtc_state to intel_dmc_{enable, disable}_pipe() Ville Syrjala
2025-06-12 20:58   ` Shankar, Uma
2025-06-11 15:52 ` [PATCH 9/9] drm/i915/dmc: Do not enable the pipe DMC on TGL when PSR is possible Ville Syrjala
2025-06-12 21:02   ` Shankar, Uma
2025-06-11 17:37 ` ✗ Fi.CI.CHECKPATCH: warning for drm/i915/dmc: Deal with loss of pipe DMC state Patchwork
2025-06-11 17:59 ` ✓ i915.CI.BAT: success " Patchwork
2025-06-11 21:09 ` ✗ i915.CI.Full: failure " Patchwork
2025-06-14  9:13 ` ✓ i915.CI.BAT: success for drm/i915/dmc: Deal with loss of pipe DMC state (rev2) Patchwork
2025-06-14 11:01 ` ✗ i915.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=20250611155241.24191-7-ville.syrjala@linux.intel.com \
    --to=ville.syrjala@linux.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