Skip to contentSkip to conversation

drm/i915: Wait until after wm optimization to drop runtime PM reference

PATCH Applied Matt Roper 7 participants 16 messages Mar 5, 2016 – Mar 23, 2016

Subsystems (2)
lore
Share & tools
lore

Matt Roper

author
#0 lore raw

PATCH 1/1 drm/i915: Wait until after wm optimization to drop runtime PM reference

At the end of an atomic commit, we currently wait for vblanks to complete, call put() on the various runtime PM references, and then try to optimize our watermarks (on platforms that need two-step watermark programming). This can lead to watermark registers being programmed while the power well is powered down. We need to wait until after watermark optimization is complete before dropping our runtime power references.

Note that in the future the watermark optimization is probably going to move to an asynchronous workqueue task that happens at some arbitrary point after vblank. When we make that change, we'll no longer necessarily be operating under the power reference held here, so we'll need to wrap the watermark register programmin in a call to intel_runtime_pm_get_if_in_use() or similar.

Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=94349

Fixes ed4a6a7ca853 ("drm/i915: Add two-stage ILK-style watermark programming (v11)")
---
 drivers/gpu/drm/i915/intel_display.c | 20 ++++++++++----------
 1 file changed, 10 insertions(+), 10 deletions(-)
drivers/gpu/drm/i915/intel_display.c +10 -10
index 62d36a7..0af08d7 100644
@@ -13789,16 +13789,6 @@ static int intel_atomic_commit(struct drm_device *dev,
 	if (!state->legacy_cursor_update)
 		intel_atomic_wait_for_vblanks(dev, dev_priv, crtc_vblank_mask);
 
-	for_each_crtc_in_state(state, crtc, crtc_state, i) {
-		intel_post_plane_update(to_intel_crtc(crtc));
-
-		if (put_domains[i])
-			modeset_put_power_domains(dev_priv, put_domains[i]);
-	}
-
-	if (intel_state->modeset)
-		intel_display_power_put(dev_priv, POWER_DOMAIN_MODESET);
-
 	/*
 	 * Now that the vblank has passed, we can go ahead and program the
 	 * optimal watermarks on platforms that need two-step watermark
@@ -13813,6 +13803,16 @@ static int intel_atomic_commit(struct drm_device *dev,
 			dev_priv->display.optimize_watermarks(intel_cstate);
 	}
 
+	for_each_crtc_in_state(state, crtc, crtc_state, i) {
+		intel_post_plane_update(to_intel_crtc(crtc));
+
+		if (put_domains[i])
+			modeset_put_power_domains(dev_priv, put_domains[i]);
+	}
+
+	if (intel_state->modeset)
+		intel_display_power_put(dev_priv, POWER_DOMAIN_MODESET);
+
 	mutex_lock(&dev->struct_mutex);
 	drm_atomic_helper_cleanup_planes(dev, state);
 	mutex_unlock(&dev->struct_mutex);
Signature · 6 lines
2.1.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#1 lore raw

Hi Matt,

Quoted 69 lines · Matt Roper

At the end of an atomic commit, we currently wait for vblanks to complete, call put() on the various runtime PM references, and then try
to optimize our watermarks (on platforms that need two-step watermark programming).  This can lead to watermark registers being programmed while the power well is powered down.  We need to wait until after watermark optimization is complete before dropping our runtime power references.

Note that in the future the watermark optimization is probably going to
move to an asynchronous workqueue task that happens at some arbitrary point after vblank.  When we make that change, we'll no longer necessarily be operating under the power reference held here, so we'll
need to wrap the watermark register programmin in a call to intel_runtime_pm_get_if_in_use() or similar.

Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=94349

Fixes ed4a6a7ca853 ("drm/i915: Add two-stage ILK-style watermark

programming (v11)")

---
 drivers/gpu/drm/i915/intel_display.c | 20 ++++++++++----------

 1 file changed, 10 insertions(+), 10 deletions(-)
drivers/gpu/drm/i915/intel_display.c +0 -0
index 62d36a7..0af08d7 100644
@@ -13789,16 +13789,6 @@ static int intel_atomic_commit(struct
 drm_device *dev,
  	if (!state->legacy_cursor_update)
  		intel_atomic_wait_for_vblanks(dev, dev_priv,
 crtc_vblank_mask);

drivers/gpu/drm/i915/intel_display.c +0 -5
-	for_each_crtc_in_state(state, crtc, crtc_state, i) {
-		intel_post_plane_update(to_intel_crtc(crtc));
-
-		if (put_domains[i])
-			modeset_put_power_domains(dev_priv,

put_domains[i]);

drivers/gpu/drm/i915/intel_display.c +0 -4
-	}
-
-	if (intel_state->modeset)
-		intel_display_power_put(dev_priv,

POWER_DOMAIN_MODESET);
-
  /*
   * Now that the vblank has passed, we can go ahead and program the
   * optimal watermarks on platforms that need two-step watermark

drivers/gpu/drm/i915/intel_display.c +0 -0
@@ -13813,6 +13803,16 @@ static int intel_atomic_commit(struct
 drm_device *dev,

  dev_priv-

Quoted 1 line

display.optimize_watermarks(intel_cstate);

  }

drivers/gpu/drm/i915/intel_display.c +5 -0
+	for_each_crtc_in_state(state, crtc, crtc_state, i) {
+		intel_post_plane_update(to_intel_crtc(crtc));
+
+		if (put_domains[i])
+			modeset_put_power_domains(dev_priv,

put_domains[i]);

drivers/gpu/drm/i915/intel_display.c static int intel_atomic_commit(struct drm_device *dev, · line 13815

Quoted 5 lines
drivers/gpu/drm/i915/intel_display.c +5 -0
+	}
+
+	if (intel_state->modeset)
+> 		intel_display_power_put(dev_priv, POWER_DOMAIN_MODESET);
+

Since the error was caused by writing some WM register while the HW is suspended, I don't see what would be the point of programming it if the register loses its value anyway after dropping the power reference. What would make sense to me is to avoid programming the WM registers if all the outputs are disabled (like they were when the bug triggered) and program them next time around when any output gets enabled.

--Imre

Quoted 3 lines

  mutex_lock(&dev->struct_mutex); drm_atomic_helper_cleanup_planes(dev, state); mutex_unlock(&dev->struct_mutex);

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

author
#2 lore raw
Quoted 76 lines · Imre Deak

Hi Matt,

Quoted 73 lines · Matt Roper

At the end of an atomic commit, we currently wait for vblanks to complete, call put() on the various runtime PM references, and then try
to optimize our watermarks (on platforms that need two-step watermark programming).  This can lead to watermark registers being programmed while the power well is powered down.  We need to wait until after watermark optimization is complete before dropping our runtime power references.

Note that in the future the watermark optimization is probably going to
move to an asynchronous workqueue task that happens at some arbitrary point after vblank.  When we make that change, we'll no longer necessarily be operating under the power reference held here, so we'll
need to wrap the watermark register programmin in a call to intel_runtime_pm_get_if_in_use() or similar.

Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=94349

Fixes ed4a6a7ca853 ("drm/i915: Add two-stage ILK-style watermark

programming (v11)")

---
 drivers/gpu/drm/i915/intel_display.c | 20 ++++++++++----------

 1 file changed, 10 insertions(+), 10 deletions(-)
drivers/gpu/drm/i915/intel_display.c +0 -0
index 62d36a7..0af08d7 100644
@@ -13789,16 +13789,6 @@ static int intel_atomic_commit(struct
 drm_device *dev,
  	if (!state->legacy_cursor_update)
  		intel_atomic_wait_for_vblanks(dev, dev_priv,
 crtc_vblank_mask);

drivers/gpu/drm/i915/intel_display.c +0 -5
-	for_each_crtc_in_state(state, crtc, crtc_state, i) {
-		intel_post_plane_update(to_intel_crtc(crtc));
-
-		if (put_domains[i])
-			modeset_put_power_domains(dev_priv,

put_domains[i]);

drivers/gpu/drm/i915/intel_display.c +0 -4
-	}
-
-	if (intel_state->modeset)
-		intel_display_power_put(dev_priv,

POWER_DOMAIN_MODESET);
-
  /*
   * Now that the vblank has passed, we can go ahead and program the
   * optimal watermarks on platforms that need two-step watermark

drivers/gpu/drm/i915/intel_display.c +0 -0
@@ -13813,6 +13803,16 @@ static int intel_atomic_commit(struct
 drm_device *dev,

  dev_priv-

Quoted 1 line

display.optimize_watermarks(intel_cstate);

  }

drivers/gpu/drm/i915/intel_display.c +5 -0
+	for_each_crtc_in_state(state, crtc, crtc_state, i) {
+		intel_post_plane_update(to_intel_crtc(crtc));
+
+		if (put_domains[i])
+			modeset_put_power_domains(dev_priv,

put_domains[i]);

drivers/gpu/drm/i915/intel_display.c +4 -0
+	}
+
+	if (intel_state->modeset)
+> 		intel_display_power_put(dev_priv, POWER_DOMAIN_MODESET);
Quoted 8 lines
Quoted 1 line

+

Since the error was caused by writing some WM register while the HW is suspended, I don't see what would be the point of programming it if the register loses its value anyway after dropping the power reference. What would make sense to me is to avoid programming the WM registers if all the outputs are disabled (like they were when the bug triggered) and program them next time around when any output gets enabled.

Yeah, that was actually the first approach I went with:

        https://patchwork.freedesktop.org/patch/75772/

Ville felt like it was more of a workaround than a fix though, so I posted this alternative instead.

Matt

Quoted 7 lines

--Imre

Quoted 3 lines

  mutex_lock(&dev->struct_mutex); drm_atomic_helper_cleanup_planes(dev, state); mutex_unlock(&dev->struct_mutex);

Signature · 9 lines
Matt Roper
Graphics Software Engineer
IoTG Platform Enabling & Development
Intel Corporation
(916) 356-2795
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#3 lore raw
Quoted 86 lines · Matt Roper
Quoted 76 lines · Imre Deak

Hi Matt,

Quoted 73 lines · Matt Roper

At the end of an atomic commit, we currently wait for vblanks to complete, call put() on the various runtime PM references, and then try
to optimize our watermarks (on platforms that need two-step watermark programming).  This can lead to watermark registers being programmed while the power well is powered down.  We need to wait until after watermark optimization is complete before dropping our runtime power references.

Note that in the future the watermark optimization is probably going to
move to an asynchronous workqueue task that happens at some arbitrary point after vblank.  When we make that change, we'll no longer necessarily be operating under the power reference held here, so we'll
need to wrap the watermark register programmin in a call to intel_runtime_pm_get_if_in_use() or similar.

Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=94349

Fixes ed4a6a7ca853 ("drm/i915: Add two-stage ILK-style watermark

programming (v11)")

---
 drivers/gpu/drm/i915/intel_display.c | 20 ++++++++++----------

 1 file changed, 10 insertions(+), 10 deletions(-)
drivers/gpu/drm/i915/intel_display.c +0 -0
index 62d36a7..0af08d7 100644
@@ -13789,16 +13789,6 @@ static int intel_atomic_commit(struct
 drm_device *dev,
  	if (!state->legacy_cursor_update)
  		intel_atomic_wait_for_vblanks(dev, dev_priv,
 crtc_vblank_mask);

drivers/gpu/drm/i915/intel_display.c +0 -5
-	for_each_crtc_in_state(state, crtc, crtc_state, i) {
-		intel_post_plane_update(to_intel_crtc(crtc));
-
-		if (put_domains[i])
-			modeset_put_power_domains(dev_priv,

put_domains[i]);

drivers/gpu/drm/i915/intel_display.c +0 -4
-	}
-
-	if (intel_state->modeset)
-		intel_display_power_put(dev_priv,

POWER_DOMAIN_MODESET);
-
  /*
   * Now that the vblank has passed, we can go ahead and program the
   * optimal watermarks on platforms that need two-step watermark

drivers/gpu/drm/i915/intel_display.c +0 -0
@@ -13813,6 +13803,16 @@ static int intel_atomic_commit(struct
 drm_device *dev,

  dev_priv-

Quoted 1 line

display.optimize_watermarks(intel_cstate);

  }

drivers/gpu/drm/i915/intel_display.c +5 -0
+	for_each_crtc_in_state(state, crtc, crtc_state, i) {
+		intel_post_plane_update(to_intel_crtc(crtc));
+
+		if (put_domains[i])
+			modeset_put_power_domains(dev_priv,

put_domains[i]);

drivers/gpu/drm/i915/intel_display.c +4 -0
+	}
+
+	if (intel_state->modeset)
+> 		intel_display_power_put(dev_priv, POWER_DOMAIN_MODESET);
Quoted 8 lines
Quoted 1 line

+

Since the error was caused by writing some WM register while the HW is suspended, I don't see what would be the point of programming it if the register loses its value anyway after dropping the power reference. What would make sense to me is to avoid programming the WM registers if all the outputs are disabled (like they were when the bug triggered) and program them next time around when any output gets enabled.

Quoted 6 lines

Yeah, that was actually the first approach I went with:

        https://patchwork.freedesktop.org/patch/75772/

Ville felt like it was more of a workaround than a fix though, so I posted this alternative instead.

Ok, I haven't noticed that patch.

One other thing that I realized only after sending my email is that - on some old platforms at least - it would still make sense to program the WM state even with all outputs disabled: the description of FW_BLC_Self[15] on GEN3 for example could mean that memory self refresh is disabled whenever we set this bit to 0, independently of the display output state, although this description is somewhat vague.

--Imre
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#4 lore raw

✗ Fi.CI.BAT: failure for drm/i915: Wait until after wm optimization to drop runtime PM reference

In reply to Matt Roper
== Series Details ==

== Series Details ==

Series: drm/i915: Wait until after wm optimization to drop runtime PM reference
URL : https://patchwork.freedesktop.org/series/4135/
State : failure

== Summary ==

Series 4135v1 drm/i915: Wait until after wm optimization to drop runtime PM reference
http://patchwork.freedesktop.org/api/1.0/series/4135/revisions/1/mbox/

Test kms_flip:

        Subgroup basic-flip-vs-wf_vblank:
                fail       -> PASS       (bsw-nuc-2)
        Subgroup basic-plain-flip:
                dmesg-warn -> PASS       (hsw-gt2)
Test kms_force_connector_basic:
        Subgroup force-load-detect:
                pass       -> SKIP       (ivb-t430s)
Test kms_pipe_crc_basic:
        Subgroup read-crc-pipe-c:
                dmesg-warn -> PASS       (hsw-gt2)
        Subgroup suspend-read-crc-pipe-a:
                dmesg-warn -> PASS       (skl-i5k-2)
                skip       -> PASS       (hsw-brixbox)
        Subgroup suspend-read-crc-pipe-c:
                pass       -> DMESG-WARN (bsw-nuc-2)
                pass       -> INCOMPLETE (hsw-gt2)
Test pm_rpm:
        Subgroup basic-pci-d3-state:
                dmesg-warn -> PASS       (snb-dellxps)
        Subgroup basic-rte:
                pass       -> DMESG-WARN (bsw-nuc-2)
bdw-nuci7        total:183  pass:172  dwarn:0   dfail:0   fail:0   skip:11 
bdw-ultra        total:183  pass:165  dwarn:0   dfail:0   fail:0   skip:18 
bsw-nuc-2        total:183  pass:147  dwarn:2   dfail:0   fail:0   skip:34 
byt-nuc          total:183  pass:152  dwarn:0   dfail:0   fail:0   skip:31 
hsw-brixbox      total:183  pass:164  dwarn:0   dfail:0   fail:0   skip:19 
hsw-gt2          total:119  pass:109  dwarn:0   dfail:0   fail:0   skip:9  
ilk-hp8440p      total:183  pass:125  dwarn:0   dfail:0   fail:0   skip:58 
ivb-t430s        total:183  pass:161  dwarn:0   dfail:0   fail:0   skip:22 
skl-i5k-2        total:183  pass:163  dwarn:0   dfail:0   fail:0   skip:20 
skl-i7k-2        total:183  pass:163  dwarn:0   dfail:0   fail:0   skip:20 
snb-dellxps      total:183  pass:154  dwarn:0   dfail:0   fail:0   skip:29 

Results at /archive/results/CI_IGT_test/Patchwork_1532/

d369e0096716c6000139162b3b340f684f0a51da drm-intel-nightly: 2016y-03m-04d-17h-18m-08s UTC integration manifest b0744e82af66e41a8dca48bfa6ff8f38e1d9454f drm/i915: Wait until after wm optimization to drop runtime PM reference

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

author
#5 lore raw

Re: ✗ Fi.CI.BAT: failure for drm/i915: Wait until after wm optimization to drop runtime PM reference

Quoted 11 lines · Patchwork

== Series Details ==

Series: drm/i915: Wait until after wm optimization to drop runtime PM reference
URL : https://patchwork.freedesktop.org/series/4135/
State : failure

== Summary ==

Series 4135v1 drm/i915: Wait until after wm optimization to drop runtime PM reference
http://patchwork.freedesktop.org/api/1.0/series/4135/revisions/1/mbox/

Quoted 8 lines

Test kms_flip:

        Subgroup basic-flip-vs-wf_vblank:
                fail       -> PASS       (bsw-nuc-2)
        Subgroup basic-plain-flip:
                dmesg-warn -> PASS       (hsw-gt2)
Test kms_force_connector_basic:
        Subgroup force-load-detect:
                pass       -> SKIP       (ivb-t430s)

https://bugs.freedesktop.org/show_bug.cgi?id=93769

Quoted 8 lines

Test kms_pipe_crc_basic:

        Subgroup read-crc-pipe-c:
                dmesg-warn -> PASS       (hsw-gt2)
        Subgroup suspend-read-crc-pipe-a:
                dmesg-warn -> PASS       (skl-i5k-2)
                skip       -> PASS       (hsw-brixbox)
        Subgroup suspend-read-crc-pipe-c:
                pass       -> DMESG-WARN (bsw-nuc-2)

https://bugs.freedesktop.org/show_bug.cgi?id=93294

Quoted 1 line
                pass       -> INCOMPLETE (hsw-gt2)

Seems like the machine just died completely? No stdout/stderr/command/dmesg output available from CI and time is given as 0:00:00. Doesn't seem like it could be related to this patch.

Quoted 5 lines

Test pm_rpm:

        Subgroup basic-pci-d3-state:
                dmesg-warn -> PASS       (snb-dellxps)
        Subgroup basic-rte:
                pass       -> DMESG-WARN (bsw-nuc-2)

https://bugs.freedesktop.org/show_bug.cgi?id=94164

Matt

Quoted 18 lines
bdw-nuci7        total:183  pass:172  dwarn:0   dfail:0   fail:0   skip:11 
bdw-ultra        total:183  pass:165  dwarn:0   dfail:0   fail:0   skip:18 
bsw-nuc-2        total:183  pass:147  dwarn:2   dfail:0   fail:0   skip:34 
byt-nuc          total:183  pass:152  dwarn:0   dfail:0   fail:0   skip:31 
hsw-brixbox      total:183  pass:164  dwarn:0   dfail:0   fail:0   skip:19 
hsw-gt2          total:119  pass:109  dwarn:0   dfail:0   fail:0   skip:9  
ilk-hp8440p      total:183  pass:125  dwarn:0   dfail:0   fail:0   skip:58 
ivb-t430s        total:183  pass:161  dwarn:0   dfail:0   fail:0   skip:22 
skl-i5k-2        total:183  pass:163  dwarn:0   dfail:0   fail:0   skip:20 
skl-i7k-2        total:183  pass:163  dwarn:0   dfail:0   fail:0   skip:20 
snb-dellxps      total:183  pass:154  dwarn:0   dfail:0   fail:0   skip:29 

Results at /archive/results/CI_IGT_test/Patchwork_1532/

d369e0096716c6000139162b3b340f684f0a51da drm-intel-nightly: 2016y-03m-04d-17h-18m-08s UTC integration manifest b0744e82af66e41a8dca48bfa6ff8f38e1d9454f drm/i915: Wait until after wm optimization to drop runtime PM reference

Signature · 9 lines
Matt Roper
Graphics Software Engineer
IoTG Platform Enabling & Development
Intel Corporation
(916) 356-2795
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Applied
#6 lore raw

Re: ✗ Fi.CI.BAT: failure for drm/i915: Wait until after wm optimization to drop runtime PM reference

Quoted 51 lines · Matt Roper
Quoted 14 lines · Patchwork

== Series Details ==

Series: drm/i915: Wait until after wm optimization to drop runtime PM reference
URL   : https://patchwork.freedesktop.org/series/4135/
State : failure

== Summary ==

Series 4135v1 drm/i915: Wait until after wm optimization to drop runtime PM reference
http://patchwork.freedesktop.org/api/1.0/series/4135/revisions/1/mb
ox/

Quoted 8 lines

Test kms_flip:
        Subgroup basic-flip-vs-wf_vblank:
                fail       -> PASS       (bsw-nuc-2) Subgroup basic-plain-flip:
                dmesg-warn -> PASS       (hsw-gt2) Test kms_force_connector_basic:
        Subgroup force-load-detect:
                pass       -> SKIP       (ivb-t430s)

https://bugs.freedesktop.org/show_bug.cgi?id=93769

Quoted 8 lines

Test kms_pipe_crc_basic:
        Subgroup read-crc-pipe-c:
                dmesg-warn -> PASS       (hsw-gt2) Subgroup suspend-read-crc-pipe-a:
                dmesg-warn -> PASS       (skl-i5k-2) skip       -> PASS       (hsw-brixbox) Subgroup suspend-read-crc-pipe-c:
                pass       -> DMESG-WARN (bsw-nuc-2)

https://bugs.freedesktop.org/show_bug.cgi?id=93294

Quoted 1 line

                pass       -> INCOMPLETE (hsw-gt2)

Seems like the machine just died completely?  No stdout/stderr/command/dmesg output available from CI and time is given
as 0:00:00.  Doesn't seem like it could be related to this patch.

Quoted 5 lines

Test pm_rpm:
        Subgroup basic-pci-d3-state:
                dmesg-warn -> PASS       (snb-dellxps) Subgroup basic-rte:
                pass       -> DMESG-WARN (bsw-nuc-2)

Quoted 4 lines

https://bugs.freedesktop.org/show_bug.cgi?id=94164

Matt

Pushed to -dinq, thanks for the patch and the review.

I had to rebase it on top of Ville's recent s/crtc_state/old_crtc_state/ change, please double check it.

Jani, this is for -fixes.

--Imre
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#7 lore raw

Re: ✗ Fi.CI.BAT: failure for drm/i915: Wait until after wm optimization to drop runtime PM reference

Quoted 56 lines · Imre Deak

[ text/plain ]

Quoted 54 lines · Matt Roper
Quoted 14 lines · Patchwork

== Series Details ==

Series: drm/i915: Wait until after wm optimization to drop runtime PM reference
URL   : https://patchwork.freedesktop.org/series/4135/
State : failure

== Summary ==

Series 4135v1 drm/i915: Wait until after wm optimization to drop runtime PM reference
http://patchwork.freedesktop.org/api/1.0/series/4135/revisions/1/mb
ox/

Quoted 8 lines

Test kms_flip:
        Subgroup basic-flip-vs-wf_vblank:
                fail       -> PASS       (bsw-nuc-2) Subgroup basic-plain-flip:
                dmesg-warn -> PASS       (hsw-gt2) Test kms_force_connector_basic:
        Subgroup force-load-detect:
                pass       -> SKIP       (ivb-t430s)

https://bugs.freedesktop.org/show_bug.cgi?id=93769

Quoted 8 lines

Test kms_pipe_crc_basic:
        Subgroup read-crc-pipe-c:
                dmesg-warn -> PASS       (hsw-gt2) Subgroup suspend-read-crc-pipe-a:
                dmesg-warn -> PASS       (skl-i5k-2) skip       -> PASS       (hsw-brixbox) Subgroup suspend-read-crc-pipe-c:
                pass       -> DMESG-WARN (bsw-nuc-2)

https://bugs.freedesktop.org/show_bug.cgi?id=93294

Quoted 1 line

                pass       -> INCOMPLETE (hsw-gt2)

Seems like the machine just died completely?  No stdout/stderr/command/dmesg output available from CI and time is given
as 0:00:00.  Doesn't seem like it could be related to this patch.

Quoted 5 lines

Test pm_rpm:
        Subgroup basic-pci-d3-state:
                dmesg-warn -> PASS       (snb-dellxps) Subgroup basic-rte:
                pass       -> DMESG-WARN (bsw-nuc-2)

https://bugs.freedesktop.org/show_bug.cgi?id=94164

Quoted 8 lines
Quoted 1 line

Matt

Pushed to -dinq, thanks for the patch and the review.

I had to rebase it on top of Ville's recent s/crtc_state/old_crtc_state/ change, please double check it.

Jani, this is for -fixes.

Surely you added either

or

in the commit when you pushed, then?

BR,
Jani.

Signature · 5 lines
Jani Nikula, Intel Open Source Technology Center
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#8 lore raw

Re: ✗ Fi.CI.BAT: failure for drm/i915: Wait until after wm optimization to drop runtime PM reference

Quoted 71 lines · Jani Nikula
Quoted 59 lines · Imre Deak

[ text/plain ]

Quoted 57 lines · Matt Roper
Quoted 17 lines · Patchwork

== Series Details ==

Series: drm/i915: Wait until after wm optimization to drop runtime
PM reference
URL   : https://patchwork.freedesktop.org/series/4135/
State : failure

== Summary ==

Series 4135v1 drm/i915: Wait until after wm optimization to drop
runtime PM reference
http://patchwork.freedesktop.org/api/1.0/series/4135/revisions/
1/mb
ox/

Quoted 8 lines

Test kms_flip:
        Subgroup basic-flip-vs-wf_vblank:
                fail       -> PASS       (bsw-nuc-2) Subgroup basic-plain-flip:
                dmesg-warn -> PASS       (hsw-gt2) Test kms_force_connector_basic:
        Subgroup force-load-detect:
                pass       -> SKIP       (ivb-t430s)

https://bugs.freedesktop.org/show_bug.cgi?id=93769

Quoted 8 lines

Test kms_pipe_crc_basic:
        Subgroup read-crc-pipe-c:
                dmesg-warn -> PASS       (hsw-gt2) Subgroup suspend-read-crc-pipe-a:
                dmesg-warn -> PASS       (skl-i5k-2) skip       -> PASS       (hsw-brixbox) Subgroup suspend-read-crc-pipe-c:
                pass       -> DMESG-WARN (bsw-nuc-2)

https://bugs.freedesktop.org/show_bug.cgi?id=93294

Quoted 1 line

                pass       -> INCOMPLETE (hsw-gt2)

Seems like the machine just died completely?  No stdout/stderr/command/dmesg output available from CI and time is given
as 0:00:00.  Doesn't seem like it could be related to this patch.

Quoted 5 lines

Test pm_rpm:
        Subgroup basic-pci-d3-state:
                dmesg-warn -> PASS       (snb-dellxps) Subgroup basic-rte:
                pass       -> DMESG-WARN (bsw-nuc-2)

https://bugs.freedesktop.org/show_bug.cgi?id=94164

Quoted 8 lines
Quoted 1 line

Matt

Pushed to -dinq, thanks for the patch and the review.

I had to rebase it on top of Ville's recent s/crtc_state/old_crtc_state/ change, please double check it.

Jani, this is for -fixes.

Surely you added either

Quoted 7 lines

or

in the commit when you pushed, then?

No, I haven't will do that in the future. Btw, what's the rule for deciding that something is for -fixes or stable only after it's been pushed? Just ping you in case for -fixes and resend it in case of stable?

--Imre
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

author
#9 lore raw

Re: ✗ Fi.CI.BAT: failure for drm/i915: Wait until after wm optimization to drop runtime PM reference

Quoted 78 lines · Imre Deak
Quoted 77 lines · Jani Nikula
Quoted 59 lines · Imre Deak

[ text/plain ]

Quoted 57 lines · Matt Roper
Quoted 17 lines · Patchwork

== Series Details ==

Series: drm/i915: Wait until after wm optimization to drop runtime
PM reference
URL   : https://patchwork.freedesktop.org/series/4135/
State : failure

== Summary ==

Series 4135v1 drm/i915: Wait until after wm optimization to drop
runtime PM reference
http://patchwork.freedesktop.org/api/1.0/series/4135/revisions/
1/mb
ox/

Quoted 8 lines

Test kms_flip:
        Subgroup basic-flip-vs-wf_vblank:
                fail       -> PASS       (bsw-nuc-2) Subgroup basic-plain-flip:
                dmesg-warn -> PASS       (hsw-gt2) Test kms_force_connector_basic:
        Subgroup force-load-detect:
                pass       -> SKIP       (ivb-t430s)

https://bugs.freedesktop.org/show_bug.cgi?id=93769

Quoted 8 lines

Test kms_pipe_crc_basic:
        Subgroup read-crc-pipe-c:
                dmesg-warn -> PASS       (hsw-gt2) Subgroup suspend-read-crc-pipe-a:
                dmesg-warn -> PASS       (skl-i5k-2) skip       -> PASS       (hsw-brixbox) Subgroup suspend-read-crc-pipe-c:
                pass       -> DMESG-WARN (bsw-nuc-2)

https://bugs.freedesktop.org/show_bug.cgi?id=93294

Quoted 1 line

                pass       -> INCOMPLETE (hsw-gt2)

Seems like the machine just died completely?  No stdout/stderr/command/dmesg output available from CI and time is given
as 0:00:00.  Doesn't seem like it could be related to this patch.

Quoted 5 lines

Test pm_rpm:
        Subgroup basic-pci-d3-state:
                dmesg-warn -> PASS       (snb-dellxps) Subgroup basic-rte:
                pass       -> DMESG-WARN (bsw-nuc-2)

https://bugs.freedesktop.org/show_bug.cgi?id=94164

Quoted 8 lines
Quoted 1 line

Matt

Pushed to -dinq, thanks for the patch and the review.

I had to rebase it on top of Ville's recent s/crtc_state/old_crtc_state/ change, please double check it.

Jani, this is for -fixes.

Surely you added either

or

Quoted 8 lines
Quoted 1 line

in the commit when you pushed, then?

No, I haven't will do that in the future. Btw, what's the rule for deciding that something is for -fixes or stable only after it's been pushed? Just ping you in case for -fixes and resend it in case of stable?

--Imre

Are you sure we need this for -fixes? The patch that introduced the regression isn't in drm-next/4.6 as far as I know.

Matt

Signature · 9 lines
Matt Roper
Graphics Software Engineer
IoTG Platform Enabling & Development
Intel Corporation
(916) 356-2795
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#10 lore raw

Re: ✗ Fi.CI.BAT: failure for drm/i915: Wait until after wm optimization to drop runtime PM reference

Quoted 90 lines · Matt Roper
Quoted 89 lines · Imre Deak
Quoted 74 lines · Jani Nikula
Quoted 62 lines · Imre Deak

[ text/plain ]

Quoted 60 lines · Matt Roper
Quoted 18 lines · Patchwork

== Series Details ==

Series: drm/i915: Wait until after wm optimization to drop runtime
PM reference
URL   : https://patchwork.freedesktop.org/series/4135/
State : failure

== Summary ==

Series 4135v1 drm/i915: Wait until after wm optimization to drop
runtime PM reference
http://patchwork.freedesktop.org/api/1.0/series/4135/revisi
ons/
1/mb
ox/

Quoted 8 lines

Test kms_flip:
        Subgroup basic-flip-vs-wf_vblank:
                fail       -> PASS       (bsw-nuc-2) Subgroup basic-plain-flip:
                dmesg-warn -> PASS       (hsw-gt2) Test kms_force_connector_basic:
        Subgroup force-load-detect:
                pass       -> SKIP       (ivb-t430s)

https://bugs.freedesktop.org/show_bug.cgi?id=93769

Quoted 8 lines

Test kms_pipe_crc_basic:
        Subgroup read-crc-pipe-c:
                dmesg-warn -> PASS       (hsw-gt2) Subgroup suspend-read-crc-pipe-a:
                dmesg-warn -> PASS       (skl-i5k-2) skip       -> PASS       (hsw-brixbox) Subgroup suspend-read-crc-pipe-c:
                pass       -> DMESG-WARN (bsw-nuc-2)

https://bugs.freedesktop.org/show_bug.cgi?id=93294

Quoted 1 line

                pass       -> INCOMPLETE (hsw-gt2)

Seems like the machine just died completely?  No stdout/stderr/command/dmesg output available from CI and time is
given
as 0:00:00.  Doesn't seem like it could be related to this patch.

Quoted 5 lines

Test pm_rpm:
        Subgroup basic-pci-d3-state:
                dmesg-warn -> PASS       (snb-dellxps) Subgroup basic-rte:
                pass       -> DMESG-WARN (bsw-nuc-2)

https://bugs.freedesktop.org/show_bug.cgi?id=94164

Quoted 8 lines
Quoted 1 line

Matt

Pushed to -dinq, thanks for the patch and the review.

I had to rebase it on top of Ville's recent s/crtc_state/old_crtc_state/ change, please double check it.

Jani, this is for -fixes.

Surely you added either

Quoted 7 lines

or

in the commit when you pushed, then?

No, I haven't will do that in the future. Btw, what's the rule for deciding that something is for -fixes or stable only after it's been
pushed? Just ping you in case for -fixes and resend it in case of stable?

Quoted 4 lines
Quoted 1 line

--Imre

Are you sure we need this for -fixes?  The patch that introduced the regression isn't in drm-next/4.6 as far as I know.

The regressing commit ed4a6a7ca853 is in drm-intel-next, so I think we do need it for -fixes.

--Imre
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#11 lore raw

Re: ✗ Fi.CI.BAT: failure for drm/i915: Wait until after wm optimization to drop runtime PM reference

Quoted 92 lines · Imre Deak

[ text/plain ]

Quoted 90 lines · Matt Roper
Quoted 89 lines · Imre Deak
Quoted 74 lines · Jani Nikula
Quoted 62 lines · Imre Deak

[ text/plain ]

Quoted 60 lines · Matt Roper
Quoted 18 lines · Patchwork

== Series Details ==

Series: drm/i915: Wait until after wm optimization to drop runtime
PM reference
URL   : https://patchwork.freedesktop.org/series/4135/
State : failure

== Summary ==

Series 4135v1 drm/i915: Wait until after wm optimization to drop
runtime PM reference
http://patchwork.freedesktop.org/api/1.0/series/4135/revisi
ons/
1/mb
ox/

Quoted 8 lines

Test kms_flip:
        Subgroup basic-flip-vs-wf_vblank:
                fail       -> PASS       (bsw-nuc-2) Subgroup basic-plain-flip:
                dmesg-warn -> PASS       (hsw-gt2) Test kms_force_connector_basic:
        Subgroup force-load-detect:
                pass       -> SKIP       (ivb-t430s)

https://bugs.freedesktop.org/show_bug.cgi?id=93769

Quoted 8 lines

Test kms_pipe_crc_basic:
        Subgroup read-crc-pipe-c:
                dmesg-warn -> PASS       (hsw-gt2) Subgroup suspend-read-crc-pipe-a:
                dmesg-warn -> PASS       (skl-i5k-2) skip       -> PASS       (hsw-brixbox) Subgroup suspend-read-crc-pipe-c:
                pass       -> DMESG-WARN (bsw-nuc-2)

https://bugs.freedesktop.org/show_bug.cgi?id=93294

Quoted 1 line

                pass       -> INCOMPLETE (hsw-gt2)

Seems like the machine just died completely?  No stdout/stderr/command/dmesg output available from CI and time is
given
as 0:00:00.  Doesn't seem like it could be related to this patch.

Quoted 5 lines

Test pm_rpm:
        Subgroup basic-pci-d3-state:
                dmesg-warn -> PASS       (snb-dellxps) Subgroup basic-rte:
                pass       -> DMESG-WARN (bsw-nuc-2)

https://bugs.freedesktop.org/show_bug.cgi?id=94164

Quoted 8 lines
Quoted 1 line

Matt

Pushed to -dinq, thanks for the patch and the review.

I had to rebase it on top of Ville's recent s/crtc_state/old_crtc_state/ change, please double check it.

Jani, this is for -fixes.

Surely you added either

Quoted 7 lines

or

in the commit when you pushed, then?

No, I haven't will do that in the future. Btw, what's the rule for deciding that something is for -fixes or stable only after it's been
pushed? Just ping you in case for -fixes and resend it in case of stable?

Quoted 7 lines
Quoted 4 lines
Quoted 1 line

--Imre

Are you sure we need this for -fixes?  The patch that introduced the regression isn't in drm-next/4.6 as far as I know.

The regressing commit ed4a6a7ca853 is in drm-intel-next, so I think we do need it for -fixes.

Ah. It's not in 4.5 and it's not on its way to 4.6. Not fixes material.

BR,
Jani.

Signature · 5 lines
Jani Nikula, Intel Open Source Technology Center
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#12 lore raw

Re: ✗ Fi.CI.BAT: failure for drm/i915: Wait until after wm optimization to drop runtime PM reference

Quoted 95 lines · Jani Nikula
Quoted 94 lines · Imre Deak

[ text/plain ]

Quoted 92 lines · Matt Roper
Quoted 90 lines · Imre Deak
Quoted 74 lines · Jani Nikula
Quoted 62 lines · Imre Deak

[ text/plain ]

Quoted 60 lines · Matt Roper
Quoted 18 lines · Patchwork

== Series Details ==

Series: drm/i915: Wait until after wm optimization to drop runtime
PM reference
URL   : https://patchwork.freedesktop.org/series/4135/
State : failure

== Summary ==

Series 4135v1 drm/i915: Wait until after wm optimization to drop
runtime PM reference
http://patchwork.freedesktop.org/api/1.0/series/4135/revisi
ons/
1/mb
ox/

Quoted 8 lines

Test kms_flip:
        Subgroup basic-flip-vs-wf_vblank:
                fail       -> PASS       (bsw-nuc-2) Subgroup basic-plain-flip:
                dmesg-warn -> PASS       (hsw-gt2) Test kms_force_connector_basic:
        Subgroup force-load-detect:
                pass       -> SKIP       (ivb-t430s)

https://bugs.freedesktop.org/show_bug.cgi?id=93769

Quoted 8 lines

Test kms_pipe_crc_basic:
        Subgroup read-crc-pipe-c:
                dmesg-warn -> PASS       (hsw-gt2) Subgroup suspend-read-crc-pipe-a:
                dmesg-warn -> PASS       (skl-i5k-2) skip       -> PASS       (hsw-brixbox) Subgroup suspend-read-crc-pipe-c:
                pass       -> DMESG-WARN (bsw-nuc-2)

https://bugs.freedesktop.org/show_bug.cgi?id=93294

Quoted 1 line

                pass       -> INCOMPLETE (hsw-gt2)

Seems like the machine just died completely?  No stdout/stderr/command/dmesg output available from CI and time is
given
as 0:00:00.  Doesn't seem like it could be related to this patch.

Quoted 5 lines

Test pm_rpm:
        Subgroup basic-pci-d3-state:
                dmesg-warn -> PASS       (snb-dellxps) Subgroup basic-rte:
                pass       -> DMESG-WARN (bsw-nuc-2)

https://bugs.freedesktop.org/show_bug.cgi?id=94164

Quoted 8 lines
Quoted 1 line

Matt

Pushed to -dinq, thanks for the patch and the review.

I had to rebase it on top of Ville's recent s/crtc_state/old_crtc_state/ change, please double check it.

Jani, this is for -fixes.

Surely you added either

Quoted 7 lines

or

in the commit when you pushed, then?

No, I haven't will do that in the future. Btw, what's the rule for deciding that something is for -fixes or stable only after it's been
pushed? Just ping you in case for -fixes and resend it in case of stable?

--Imre

Quoted 7 lines
Quoted 5 lines
Quoted 2 lines

Are you sure we need this for -fixes?  The patch that introduced the regression isn't in drm-next/4.6 as far as I know.

The regressing commit ed4a6a7ca853 is in drm-intel-next, so I think we do need it for -fixes.

Ah. It's not in 4.5 and it's not on its way to 4.6. Not fixes material.

For the future:

$ dim tag-contains $commit_that_youre_fixing

Will tell you either the kernel version of if it hasn't landed, the branches. And if it's in airlied/drm-next but dinq is already past the merge window then it needs Cc: stable. If it's only in dinq then not.

I guess we could improve the heuristics of the script a bit and directly print out the Cc: stable or Cc: drm-intel-fixes line you would need, if any ... Volunteers highly welcome ;-)

-Daniel

Signature · 7 lines
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#13 lore raw
In reply to Matt Roper

Op 05-03-16 om 00:59 schreef Matt Roper:

Quoted 15 lines

At the end of an atomic commit, we currently wait for vblanks to complete, call put() on the various runtime PM references, and then try to optimize our watermarks (on platforms that need two-step watermark programming). This can lead to watermark registers being programmed while the power well is powered down. We need to wait until after watermark optimization is complete before dropping our runtime power references.

Note that in the future the watermark optimization is probably going to move to an asynchronous workqueue task that happens at some arbitrary point after vblank. When we make that change, we'll no longer necessarily be operating under the power reference held here, so we'll need to wrap the watermark register programmin in a call to intel_runtime_pm_get_if_in_use() or similar.

Quoted 6 lines

Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=94349

Fixes ed4a6a7ca853 ("drm/i915: Add two-stage ILK-style watermark programming (v11)")

post_plane_update can call intel_update_watermarks, will this cause any unintended behavioral changes if intel_update_watermarks is called before optimize_watermarks?

~Maarten
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

author
#14 lore raw
Quoted 16 lines · Maarten Lankhorst

Op 05-03-16 om 00:59 schreef Matt Roper:

Quoted 15 lines

At the end of an atomic commit, we currently wait for vblanks to complete, call put() on the various runtime PM references, and then try to optimize our watermarks (on platforms that need two-step watermark programming). This can lead to watermark registers being programmed while the power well is powered down. We need to wait until after watermark optimization is complete before dropping our runtime power references.

Note that in the future the watermark optimization is probably going to move to an asynchronous workqueue task that happens at some arbitrary point after vblank. When we make that change, we'll no longer necessarily be operating under the power reference held here, so we'll need to wrap the watermark register programmin in a call to intel_runtime_pm_get_if_in_use() or similar.

Quoted 8 lines
Quoted 6 lines

Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=94349

Fixes ed4a6a7ca853 ("drm/i915: Add two-stage ILK-style watermark programming (v11)")

post_plane_update can call intel_update_watermarks, will this cause any unintended behavioral changes if intel_update_watermarks is called before optimize_watermarks?

It shouldn't; intel_update_watermarks is the legacy watermark programming function. Any platform that that's been converted to atomic shouldn't have a dev_priv->display.update_wm vfunc, so intel_update_watermarks will be a noop.

Matt

Quoted 2 lines

~Maarten

Signature · 9 lines
Matt Roper
Graphics Software Engineer
IoTG Platform Enabling & Development
Intel Corporation
(916) 356-2795
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Reviewed-by on v1 1/1
#15 lore raw
In reply to Matt Roper
Quoted 15 lines · Matt Roper

At the end of an atomic commit, we currently wait for vblanks to complete, call put() on the various runtime PM references, and then try to optimize our watermarks (on platforms that need two-step watermark programming). This can lead to watermark registers being programmed while the power well is powered down. We need to wait until after watermark optimization is complete before dropping our runtime power references.

Note that in the future the watermark optimization is probably going to move to an asynchronous workqueue task that happens at some arbitrary point after vblank. When we make that change, we'll no longer necessarily be operating under the power reference held here, so we'll need to wrap the watermark register programmin in a call to intel_runtime_pm_get_if_in_use() or similar.

Quoted 6 lines

Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=94349

Fixes ed4a6a7ca853 ("drm/i915: Add two-stage ILK-style watermark programming (v11)")
Quoted 44 lines
---
 drivers/gpu/drm/i915/intel_display.c | 20 ++++++++++----------
 1 file changed, 10 insertions(+), 10 deletions(-)
drivers/gpu/drm/i915/intel_display.c +10 -10
index 62d36a7..0af08d7 100644
@@ -13789,16 +13789,6 @@ static int intel_atomic_commit(struct drm_device *dev,
 	if (!state->legacy_cursor_update)
 		intel_atomic_wait_for_vblanks(dev, dev_priv, crtc_vblank_mask);
 
-	for_each_crtc_in_state(state, crtc, crtc_state, i) {
-		intel_post_plane_update(to_intel_crtc(crtc));
-
-		if (put_domains[i])
-			modeset_put_power_domains(dev_priv, put_domains[i]);
-	}
-
-	if (intel_state->modeset)
-		intel_display_power_put(dev_priv, POWER_DOMAIN_MODESET);
-
 	/*
 	 * Now that the vblank has passed, we can go ahead and program the
 	 * optimal watermarks on platforms that need two-step watermark
@@ -13813,6 +13803,16 @@ static int intel_atomic_commit(struct drm_device *dev,
 			dev_priv->display.optimize_watermarks(intel_cstate);
 	}
 
+	for_each_crtc_in_state(state, crtc, crtc_state, i) {
+		intel_post_plane_update(to_intel_crtc(crtc));
+
+		if (put_domains[i])
+			modeset_put_power_domains(dev_priv, put_domains[i]);
+	}
+
+	if (intel_state->modeset)
+		intel_display_power_put(dev_priv, POWER_DOMAIN_MODESET);
+
 	mutex_lock(&dev->struct_mutex);
 	drm_atomic_helper_cleanup_planes(dev, state);
 	mutex_unlock(&dev->struct_mutex);
Signature · 1 line
2.1.4
Signature · 6 lines
Ville Syrjälä
Intel OTC
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx