Skip to contentSkip to conversation

drm/i915/icl: Fix context RPCS programming

PATCH Superseded Tvrtko Ursulin 4 participants 11 messages Aug 31, 2018 – Sep 4, 2018

Subsystems (2)
lore
Share & tools
lore

Tvrtko Ursulin

author maintainer
#0 lore raw

PATCH 1/1 drm/i915/icl: Fix context RPCS programming

From: Tvrtko Ursulin <[email protected]>

There are two issues with the current RPCS programming for Icelake:

Expansion of the slice count bitfield has been missed, as well as the required programming workaround for the subslice count bitfield size limitation.

1)

Bitfield width for configuring the active slice count has grown so we need to program the GEN8_R_PWR_CLK_STATE accordingly.

Current code was always requesting eight times the number of slices (due writting to a bitfield starting three bits higher than it should). These requests were luckily a) capped by the hardware to the available number of slices, and b) we haven't yet exported the code to ask for reduced slice configurations.

Due both of the above there was no impact from this incorrect programming but we should still fix it.

2)

Due subslice count bitfield being only three bits wide and furthermore capped to a maximum documented value of four, special programming workaround is needed to enable more than four subslices.

With this programming driver has to consider the GT configuration as 2x4x8, while the hardware internally translates this to 1x8x8.

A limitation stemming from this is that either a subslice count between one and four can be selected, or a subslice count equaling the total number of subslices in all selected slices. In other words, odd subslice counts greater than four are impossible, as are odd subslice counts greater than a single slice subslice count.

This also had no impact in the current code base due breakage from 1) always reqesting more than one slice.

While fixing this we also add some asserts to flag up any future bitfield overflows.

Bspec: 12247

---
 drivers/gpu/drm/i915/i915_reg.h  |  2 +
 drivers/gpu/drm/i915/intel_lrc.c | 89 +++++++++++++++++++++++++++-----
 2 files changed, 78 insertions(+), 13 deletions(-)
drivers/gpu/drm/i915/i915_reg.h +2 -0
index f2321785cbd6..09bc8e730ee1 100644
@@ -344,6 +344,8 @@ static inline bool i915_mmio_reg_valid(i915_reg_t reg)
 #define   GEN8_RPCS_S_CNT_ENABLE	(1 << 18)
 #define   GEN8_RPCS_S_CNT_SHIFT		15
 #define   GEN8_RPCS_S_CNT_MASK		(0x7 << GEN8_RPCS_S_CNT_SHIFT)
+#define   GEN11_RPCS_S_CNT_SHIFT	12
+#define   GEN11_RPCS_S_CNT_MASK		(0x3f << GEN11_RPCS_S_CNT_SHIFT)
 #define   GEN8_RPCS_SS_CNT_ENABLE	(1 << 11)
 #define   GEN8_RPCS_SS_CNT_SHIFT	8
 #define   GEN8_RPCS_SS_CNT_MASK		(0x7 << GEN8_RPCS_SS_CNT_SHIFT)
drivers/gpu/drm/i915/intel_lrc.c +76 -13

112 lines not on this page · raw message

Signature · 6 lines
2.17.1

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

✗ Fi.CI.CHECKPATCH: warning for drm/i915/icl: Fix context RPCS programming

== Series Details ==

== Series Details ==

Series: drm/i915/icl: Fix context RPCS programming
URL : https://patchwork.freedesktop.org/series/49005/
State : warning

== Summary ==

$ dim checkpatch origin/drm-tip
8b03edf949e3 drm/i915/icl: Fix context RPCS programming
-:18: WARNING:TYPO_SPELLING: 'writting' may be misspelled - perhaps 'writing'?
#18: 
writting to a bitfield starting three bits higher than it should). These

total: 0 errors, 1 warnings, 0 checks, 120 lines checked

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

#2 lore raw

✓ Fi.CI.BAT: success for drm/i915/icl: Fix context RPCS programming

In reply to Tvrtko Ursulin
== Series Details ==

== Series Details ==

Series: drm/i915/icl: Fix context RPCS programming
URL : https://patchwork.freedesktop.org/series/49005/
State : success

== Summary ==

= CI Bug Log - changes from CI_DRM_4748 -> Patchwork_10059 =

== Summary - SUCCESS ==

No regressions found.

External URL: https://patchwork.freedesktop.org/api/1.0/series/49005/revisions/1/mbox/

== Known issues ==

Here are the changes found in Patchwork_10059 that come from known issues:

=== IGT changes ===

    ==== Issues hit ====

    igt@kms_frontbuffer_tracking@basic:
      fi-hsw-peppy:       PASS -> DMESG-WARN (fdo#102614)
      {fi-byt-clapper}:   PASS -> FAIL (fdo#103167)

    
    ==== Possible fixes ====

    igt@kms_pipe_crc_basic@suspend-read-crc-pipe-c:
      fi-bxt-dsi:         INCOMPLETE (fdo#103927) -> PASS

    {igt@kms_psr@primary_page_flip}:
      fi-cnl-psr:         FAIL (fdo#107336) -> PASS

{name}: This element is suppressed. This means it is ignored when computing
the status of the difference (SUCCESS, WARNING, or FAILURE).

fdo#102614 https://bugs.freedesktop.org/show_bug.cgi?id=102614
fdo#103167 https://bugs.freedesktop.org/show_bug.cgi?id=103167
fdo#103927 https://bugs.freedesktop.org/show_bug.cgi?id=103927
fdo#107336 https://bugs.freedesktop.org/show_bug.cgi?id=107336

== Participating hosts (53 -> 48) ==

Missing (5): fi-ctg-p8600 fi-ilk-m540 fi-byt-squawks fi-bsw-cyan fi-hsw-4200u

== Build changes ==

* Linux: CI_DRM_4748 -> Patchwork_10059

CI_DRM_4748: 6caeb081ceb9282503439565e7093c1032758289 @ git://anongit.freedesktop.org/gfx-ci/linux
IGT_4613: 3f89d7b02dcf662e994c7135b13d52bc8e09a4ea @ git://anongit.freedesktop.org/xorg/app/intel-gpu-tools
Patchwork_10059: 8b03edf949e3c9c6d97154e40c11f1f0d2a1aec7 @ git://anongit.freedesktop.org/gfx-ci/linux

== Linux commits ==

8b03edf949e3 drm/i915/icl: Fix context RPCS programming

== Logs ==

For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_10059/issues.html
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#3 lore raw

✓ Fi.CI.IGT: success for drm/i915/icl: Fix context RPCS programming

In reply to Tvrtko Ursulin
== Series Details ==

== Series Details ==

Series: drm/i915/icl: Fix context RPCS programming
URL : https://patchwork.freedesktop.org/series/49005/
State : success

== Summary ==

= CI Bug Log - changes from CI_DRM_4748_full -> Patchwork_10059_full =

== Summary - WARNING ==

Minor unknown changes coming with Patchwork_10059_full need to be verified manually.

If you think the reported changes have nothing to do with the changes introduced in Patchwork_10059_full, please notify your bug team to allow them to document this new failure mode, which will reduce false positives in CI.

== Possible new issues ==

Here are the unknown changes that may have been introduced in Patchwork_10059_full:

=== IGT changes ===

    ==== Warnings ====

    igt@pm_rc6_residency@rc6-accuracy:
      shard-kbl:          SKIP -> PASS

== Known issues ==

Here are the changes found in Patchwork_10059_full that come from known issues:

=== IGT changes ===

    ==== Issues hit ====

    igt@gem_ctx_isolation@vecs0-s3:
      shard-kbl:          PASS -> DMESG-WARN (fdo#103313)

    igt@gem_ppgtt@blt-vs-render-ctx0:
      shard-snb:          NOTRUN -> INCOMPLETE (fdo#106887, fdo#105411)
      shard-kbl:          PASS -> INCOMPLETE (fdo#106023, fdo#103665)

    igt@kms_cursor_legacy@cursor-vs-flip-atomic-transitions:
      shard-hsw:          PASS -> FAIL (fdo#103355)

    igt@kms_flip@flip-vs-dpms-off-vs-modeset:
      shard-apl:          PASS -> DMESG-WARN (fdo#103558, fdo#105602) +5

    
    ==== Possible fixes ====

    igt@gem_exec_store@pages-blt:
      shard-snb:          INCOMPLETE (fdo#105411) -> PASS

    igt@kms_cursor_crc@cursor-256x256-suspend:
      shard-apl:          INCOMPLETE (fdo#103927) -> PASS

    igt@kms_cursor_legacy@cursor-vs-flip-varying-size:
      shard-hsw:          FAIL (fdo#103355) -> PASS

    igt@kms_frontbuffer_tracking@fbc-1p-primscrn-cur-indfb-draw-render:
      shard-glk:          FAIL (fdo#103167) -> PASS

    igt@kms_setmode@basic:
      shard-kbl:          FAIL (fdo#99912) -> PASS

fdo#103167 https://bugs.freedesktop.org/show_bug.cgi?id=103167
fdo#103313 https://bugs.freedesktop.org/show_bug.cgi?id=103313
fdo#103355 https://bugs.freedesktop.org/show_bug.cgi?id=103355
fdo#103558 https://bugs.freedesktop.org/show_bug.cgi?id=103558
fdo#103665 https://bugs.freedesktop.org/show_bug.cgi?id=103665
fdo#103927 https://bugs.freedesktop.org/show_bug.cgi?id=103927
fdo#105411 https://bugs.freedesktop.org/show_bug.cgi?id=105411
fdo#105602 https://bugs.freedesktop.org/show_bug.cgi?id=105602
fdo#106023 https://bugs.freedesktop.org/show_bug.cgi?id=106023
fdo#106887 https://bugs.freedesktop.org/show_bug.cgi?id=106887
fdo#99912 https://bugs.freedesktop.org/show_bug.cgi?id=99912

== Participating hosts (5 -> 5) ==

No changes in participating hosts

== Build changes ==

* Linux: CI_DRM_4748 -> Patchwork_10059

CI_DRM_4748: 6caeb081ceb9282503439565e7093c1032758289 @ git://anongit.freedesktop.org/gfx-ci/linux
IGT_4613: 3f89d7b02dcf662e994c7135b13d52bc8e09a4ea @ git://anongit.freedesktop.org/xorg/app/intel-gpu-tools
Patchwork_10059: 8b03edf949e3c9c6d97154e40c11f1f0d2a1aec7 @ git://anongit.freedesktop.org/gfx-ci/linux
piglit_4509: fdc5a4ca11124ab8413c7988896eec4c97336694 @ git://anongit.freedesktop.org/piglit

== Logs ==

For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_10059/shards.html
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

Reviewed-by on v1 1/1
#4 lore raw
In reply to Tvrtko Ursulin
Quoted 105 lines · Tvrtko Ursulin

From: Tvrtko Ursulin <[email protected]>

There are two issues with the current RPCS programming for Icelake:

Expansion of the slice count bitfield has been missed, as well as the required programming workaround for the subslice count bitfield size limitation.

1)

Bitfield width for configuring the active slice count has grown so we need to program the GEN8_R_PWR_CLK_STATE accordingly.

Current code was always requesting eight times the number of slices (due writting to a bitfield starting three bits higher than it should). These requests were luckily a) capped by the hardware to the available number of slices, and b) we haven't yet exported the code to ask for reduced slice configurations.

Due both of the above there was no impact from this incorrect programming but we should still fix it.

2)

Due subslice count bitfield being only three bits wide and furthermore capped to a maximum documented value of four, special programming workaround is needed to enable more than four subslices.

With this programming driver has to consider the GT configuration as 2x4x8, while the hardware internally translates this to 1x8x8.

A limitation stemming from this is that either a subslice count between one and four can be selected, or a subslice count equaling the total number of subslices in all selected slices. In other words, odd subslice counts greater than four are impossible, as are odd subslice counts greater than a single slice subslice count.

This also had no impact in the current code base due breakage from 1) always reqesting more than one slice.

While fixing this we also add some asserts to flag up any future bitfield overflows.

Bspec: 12247

---
  drivers/gpu/drm/i915/i915_reg.h  |  2 +
  drivers/gpu/drm/i915/intel_lrc.c | 89 +++++++++++++++++++++++++++-----
  2 files changed, 78 insertions(+), 13 deletions(-)
drivers/gpu/drm/i915/i915_reg.h +2 -0
index f2321785cbd6..09bc8e730ee1 100644
@@ -344,6 +344,8 @@ static inline bool i915_mmio_reg_valid(i915_reg_t reg)
  #define   GEN8_RPCS_S_CNT_ENABLE	(1 << 18)
  #define   GEN8_RPCS_S_CNT_SHIFT		15
  #define   GEN8_RPCS_S_CNT_MASK		(0x7 << GEN8_RPCS_S_CNT_SHIFT)
+#define   GEN11_RPCS_S_CNT_SHIFT	12
+#define   GEN11_RPCS_S_CNT_MASK		(0x3f << GEN11_RPCS_S_CNT_SHIFT)
  #define   GEN8_RPCS_SS_CNT_ENABLE	(1 << 11)
  #define   GEN8_RPCS_SS_CNT_SHIFT	8
  #define   GEN8_RPCS_SS_CNT_MASK		(0x7 << GEN8_RPCS_SS_CNT_SHIFT)
drivers/gpu/drm/i915/intel_lrc.c +22 -0
index f8ceb9c99dd6..323c46319cb8 100644
@@ -2480,6 +2480,9 @@ int logical_xcs_ring_init(struct intel_engine_cs *engine)
  static u32
  make_rpcs(struct drm_i915_private *dev_priv)
  {
+	bool subslice_pg = INTEL_INFO(dev_priv)->sseu.has_subslice_pg;
+	u8 slices = hweight8(INTEL_INFO(dev_priv)->sseu.slice_mask);
+	u8 subslices = hweight8(INTEL_INFO(dev_priv)->sseu.subslice_mask[0]);
  	u32 rpcs = 0;
 
  	/*
@@ -2489,6 +2492,38 @@ make_rpcs(struct drm_i915_private *dev_priv)
  	if (INTEL_GEN(dev_priv) < 9)
  		return 0;
 
+	/*
+	 * Since the SScount bitfield in GEN8_R_PWR_CLK_STATE is only three bits
+	 * wide and Icelake has up to eight subslices, specfial programming is
+	 * needed in order to correctly enable all subslices.
+	 *
+	 * According to documentation software must consider the configuration
+	 * as 2x4x8 and hardware will translate this to 1x8x8.
+	 *
+	 * Furthemore, even though SScount is three bits, maximum documented
+	 * value for it is four. From this some rules/restrictions follow:
+	 *
+	 * 1.
+	 * If enabled subslice count is greater than four, two whole slices must
+	 * be enabled instead.
+	 *
+	 * 2.
+	 * When more than one slice is enabled, hardware ignores the subslice
+	 * count altogether.
+	 *

drivers/gpu/drm/i915/intel_lrc.c make_rpcs(struct drm_i915_private *dev_priv) · line 2523

Quoted 10 lines
drivers/gpu/drm/i915/intel_lrc.c +10 -0
+	 * From these restrictions it follows that it is not possible to enable
+	 * a count of subslices between the SScount maximum of four restriction,
+	 * and the maximum available number on a particular SKU. Either all
+	 * subslices are enabled, or a count between one and four on the first
+	 * slice.
+	 */
+	if (IS_GEN11(dev_priv) && slices == 1 && subslices >= 4) {
+		GEM_BUG_ON(subslices & 1);
+
+		subslice_pg = false;

Err... Now I'm looking at the documentation again and I see this for the subslice enable field :

Enable Subslice Count Request.

0 = Use Async subslice count

1 = Use SScount in this register

Searching for "Async subslice count" leads me to another register (BSpec 9497).

Really confused as to what this does :|

Is the hardware then reading for that second register if you set SSCountEn to 0?

We could put max(hweight8(sseu.subslice_mask[0]), 4) in SSCount, but it's unclear what's right...

If you feel like this needs to be investigated, go ahead.

Otherwise I have a minor style suggestion below, either way :

Quoted 7 lines
drivers/gpu/drm/i915/intel_lrc.c +3 -0
+		slices *= 2;
+	}
+
  	/*
  	 * Starting in Gen9, render power gating can leave
  	 * slice/subslice/EU in a partially enabled state. We

drivers/gpu/drm/i915/intel_lrc.c make_rpcs(struct drm_i915_private *dev_priv) · line 2502

Quoted 7 lines
drivers/gpu/drm/i915/intel_lrc.c +0 -4
@@ -2496,24 +2531,52 @@ make_rpcs(struct drm_i915_private *dev_priv)
  	 * enablement.
  	*/
  	if (INTEL_INFO(dev_priv)->sseu.has_slice_pg) {
-		rpcs |= GEN8_RPCS_S_CNT_ENABLE;
-		rpcs |= hweight8(INTEL_INFO(dev_priv)->sseu.slice_mask) <<
-			GEN8_RPCS_S_CNT_SHIFT;
-		rpcs |= GEN8_RPCS_ENABLE;

I would use a u32 val; like you did for the subslice part below and just OR things together.

Just feels a bit more readable, but up to you.

Quoted 48 lines
drivers/gpu/drm/i915/intel_lrc.c +34 -9
+		u32 mask;
+
+		rpcs = slices;
+
+		if (INTEL_GEN(dev_priv) >= 11) {
+			mask = GEN11_RPCS_S_CNT_MASK;
+			rpcs <<= GEN11_RPCS_S_CNT_SHIFT;
+		} else {
+			mask = GEN8_RPCS_S_CNT_MASK;
+			rpcs <<= GEN8_RPCS_S_CNT_SHIFT;
+		}
+
+		GEM_BUG_ON(rpcs & ~mask);
+		rpcs &= mask;
+
+		rpcs |= GEN8_RPCS_ENABLE | GEN8_RPCS_S_CNT_ENABLE;
  	}
 
-	if (INTEL_INFO(dev_priv)->sseu.has_subslice_pg) {
-		rpcs |= GEN8_RPCS_SS_CNT_ENABLE;
-		rpcs |= hweight8(INTEL_INFO(dev_priv)->sseu.subslice_mask[0]) <<
-			GEN8_RPCS_SS_CNT_SHIFT;
-		rpcs |= GEN8_RPCS_ENABLE;
+	if (subslice_pg) {
+		u32 val = subslices;
+
+		val <<= GEN8_RPCS_SS_CNT_SHIFT;
+
+		GEM_BUG_ON(val & ~GEN8_RPCS_SS_CNT_MASK);
+		val &= GEN8_RPCS_SS_CNT_MASK;
+
+		rpcs |= GEN8_RPCS_ENABLE | GEN8_RPCS_SS_CNT_ENABLE | val;
  	}
 
  	if (INTEL_INFO(dev_priv)->sseu.has_eu_pg) {
-		rpcs |= INTEL_INFO(dev_priv)->sseu.eu_per_subslice <<
-			GEN8_RPCS_EU_MIN_SHIFT;
-		rpcs |= INTEL_INFO(dev_priv)->sseu.eu_per_subslice <<
-			GEN8_RPCS_EU_MAX_SHIFT;
+		u32 val;
+
+		val = INTEL_INFO(dev_priv)->sseu.eu_per_subslice <<
+		      GEN8_RPCS_EU_MIN_SHIFT;
+		GEM_BUG_ON(val & ~GEN8_RPCS_EU_MIN_MASK);
+		val &= GEN8_RPCS_EU_MIN_MASK;
+
+		rpcs |= val;
+

drivers/gpu/drm/i915/intel_lrc.c make_rpcs(struct drm_i915_private *dev_priv) · line 2582

Quoted 10 lines
drivers/gpu/drm/i915/intel_lrc.c +7 -0
+		val = INTEL_INFO(dev_priv)->sseu.eu_per_subslice <<
+		      GEN8_RPCS_EU_MAX_SHIFT;
+		GEM_BUG_ON(val & ~GEN8_RPCS_EU_MAX_MASK);
+		val &= GEN8_RPCS_EU_MAX_MASK;
+
+		rpcs |= val;
+
  		rpcs |= GEN8_RPCS_ENABLE;
  	}
 

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

#5 lore raw
Quoted 134 lines · Lionel Landwerlin
Quoted 116 lines · Tvrtko Ursulin

From: Tvrtko Ursulin <[email protected]>

There are two issues with the current RPCS programming for Icelake:

Expansion of the slice count bitfield has been missed, as well as the required programming workaround for the subslice count bitfield size limitation.

1)

Bitfield width for configuring the active slice count has grown so we need to program the GEN8_R_PWR_CLK_STATE accordingly.

Current code was always requesting eight times the number of slices (due writting to a bitfield starting three bits higher than it should). These requests were luckily a) capped by the hardware to the available number of slices, and b) we haven't yet exported the code to ask for reduced slice configurations.

Due both of the above there was no impact from this incorrect programming but we should still fix it.

2)

Due subslice count bitfield being only three bits wide and furthermore capped to a maximum documented value of four, special programming workaround is needed to enable more than four subslices.

With this programming driver has to consider the GT configuration as 2x4x8, while the hardware internally translates this to 1x8x8.

A limitation stemming from this is that either a subslice count between one and four can be selected, or a subslice count equaling the total number of subslices in all selected slices. In other words, odd subslice counts greater than four are impossible, as are odd subslice counts greater than a single slice subslice count.

This also had no impact in the current code base due breakage from 1) always reqesting more than one slice.

While fixing this we also add some asserts to flag up any future bitfield overflows.

Bspec: 12247

---
  drivers/gpu/drm/i915/i915_reg.h  |  2 + drivers/gpu/drm/i915/intel_lrc.c | 89 +++++++++++++++++++++++++++-----

  2 files changed, 78 insertions(+), 13 deletions(-)
drivers/gpu/drm/i915/i915_reg.h +2 -0
index f2321785cbd6..09bc8e730ee1 100644
@@ -344,6 +344,8 @@ static inline bool i915_mmio_reg_valid(i915_reg_t 
 reg)
   #define   GEN8_RPCS_S_CNT_ENABLE    (1 << 18)
   #define   GEN8_RPCS_S_CNT_SHIFT        15
   #define   GEN8_RPCS_S_CNT_MASK        (0x7 << GEN8_RPCS_S_CNT_SHIFT)
+#define   GEN11_RPCS_S_CNT_SHIFT    12
+#define   GEN11_RPCS_S_CNT_MASK        (0x3f << GEN11_RPCS_S_CNT_SHIFT)
   #define   GEN8_RPCS_SS_CNT_ENABLE    (1 << 11)
   #define   GEN8_RPCS_SS_CNT_SHIFT    8

  #define   GEN8_RPCS_SS_CNT_MASK        (0x7 << GEN8_RPCS_SS_CNT_SHIFT)

drivers/gpu/drm/i915/intel_lrc.c +0 -0
index f8ceb9c99dd6..323c46319cb8 100644
@@ -2480,6 +2480,9 @@ int logical_xcs_ring_init(struct intel_engine_cs 
 *engine)

  static u32
  make_rpcs(struct drm_i915_private *dev_priv)
  {
+    bool subslice_pg = INTEL_INFO(dev_priv)->sseu.has_subslice_pg; +    u8 slices = hweight8(INTEL_INFO(dev_priv)->sseu.slice_mask); +    u8 subslices = hweight8(INTEL_INFO(dev_priv)->sseu.subslice_mask[0]); u32 rpcs = 0;
      /*

drivers/gpu/drm/i915/intel_lrc.c +2 -0
@@ -2489,6 +2492,38 @@ make_rpcs(struct drm_i915_private *dev_priv)
       if (INTEL_GEN(dev_priv) < 9)
           return 0;
+    /*
+     * Since the SScount bitfield in GEN8_R_PWR_CLK_STATE is only 

three bits
+     * wide and Icelake has up to eight subslices, specfial programming is
+     * needed in order to correctly enable all subslices.
+     *
+     * According to documentation software must consider the configuration
+     * as 2x4x8 and hardware will translate this to 1x8x8.
+     *
+     * Furthemore, even though SScount is three bits, maximum documented +     * value for it is four. From this some rules/restrictions follow:
+     *
+     * 1.
+     * If enabled subslice count is greater than four, two whole slices must
+     * be enabled instead.
+     *
+     * 2.
+     * When more than one slice is enabled, hardware ignores the subslice
+     * count altogether.
+     *
+     * From these restrictions it follows that it is not possible to enable
+     * a count of subslices between the SScount maximum of four restriction,
+     * and the maximum available number on a particular SKU. Either all

Quoted 8 lines

+     * subslices are enabled, or a count between one and four on the first
+     * slice.
+     */
+    if (IS_GEN11(dev_priv) && slices == 1 && subslices >= 4) { +        GEM_BUG_ON(subslices & 1);
+
+        subslice_pg = false;

Err... Now I'm looking at the documentation again and I see this for the subslice enable field :

Enable Subslice Count Request.

0 = Use Async subslice count

Quoted 8 lines

1 = Use SScount in this register

Searching for "Async subslice count" leads me to another register (BSpec 9497).

Really confused as to what this does :|

Is the hardware then reading for that second register if you set SSCountEn to 0?

I think it isn't. Well, by transitive property of subslice shutdown not being supported unless slice count is one. So when we set slice count to greater than one, it doesn't matter what we put in any subslice control fields.

As bspec 33575 says programming via R_PWR_CLK_STATE is preferred over the Async mode config, I think we are overall good with this patch.

Quoted 31 lines

We could put max(hweight8(sseu.subslice_mask[0]), 4) in SSCount, but it's unclear what's right...

If you feel like this needs to be investigated, go ahead.

Otherwise I have a minor style suggestion below, either way :

Quoted 7 lines

+        slices *= 2;
+    }
+
      /*
       * Starting in Gen9, render power gating can leave
       * slice/subslice/EU in a partially enabled state. We

drivers/gpu/drm/i915/intel_lrc.c make_rpcs(struct drm_i915_private *dev_priv) · line 2502

Quoted 7 lines
drivers/gpu/drm/i915/intel_lrc.c +0 -4
@@ -2496,24 +2531,52 @@ make_rpcs(struct drm_i915_private *dev_priv)
        * enablement.
       */
       if (INTEL_INFO(dev_priv)->sseu.has_slice_pg) {
-        rpcs |= GEN8_RPCS_S_CNT_ENABLE;
-        rpcs |= hweight8(INTEL_INFO(dev_priv)->sseu.slice_mask) <<
-            GEN8_RPCS_S_CNT_SHIFT;
-        rpcs |= GEN8_RPCS_ENABLE;

I would use a u32 val; like you did for the subslice part below and just OR things together.

Quoted 1 line

Just feels a bit more readable, but up to you.

I know, was 50-50 on this myself. You tipped the scale now so I'll respin. Thanks for the r-b, I'll keep it for this tweak.

Regards,

Tvrtko

Quoted 54 lines
Quoted 54 lines
drivers/gpu/drm/i915/intel_lrc.c +41 -9
+        u32 mask;
+
+        rpcs = slices;
+
+        if (INTEL_GEN(dev_priv) >= 11) {
+            mask = GEN11_RPCS_S_CNT_MASK;
+            rpcs <<= GEN11_RPCS_S_CNT_SHIFT;
+        } else {
+            mask = GEN8_RPCS_S_CNT_MASK;
+            rpcs <<= GEN8_RPCS_S_CNT_SHIFT;
+        }
+
+        GEM_BUG_ON(rpcs & ~mask);
+        rpcs &= mask;
+
+        rpcs |= GEN8_RPCS_ENABLE | GEN8_RPCS_S_CNT_ENABLE;
       }
-    if (INTEL_INFO(dev_priv)->sseu.has_subslice_pg) {
-        rpcs |= GEN8_RPCS_SS_CNT_ENABLE;
-        rpcs |= hweight8(INTEL_INFO(dev_priv)->sseu.subslice_mask[0]) <<
-            GEN8_RPCS_SS_CNT_SHIFT;
-        rpcs |= GEN8_RPCS_ENABLE;
+    if (subslice_pg) {
+        u32 val = subslices;
+
+        val <<= GEN8_RPCS_SS_CNT_SHIFT;
+
+        GEM_BUG_ON(val & ~GEN8_RPCS_SS_CNT_MASK);
+        val &= GEN8_RPCS_SS_CNT_MASK;
+
+        rpcs |= GEN8_RPCS_ENABLE | GEN8_RPCS_SS_CNT_ENABLE | val;
       }
       if (INTEL_INFO(dev_priv)->sseu.has_eu_pg) {
-        rpcs |= INTEL_INFO(dev_priv)->sseu.eu_per_subslice <<
-            GEN8_RPCS_EU_MIN_SHIFT;
-        rpcs |= INTEL_INFO(dev_priv)->sseu.eu_per_subslice <<
-            GEN8_RPCS_EU_MAX_SHIFT;
+        u32 val;
+
+        val = INTEL_INFO(dev_priv)->sseu.eu_per_subslice <<
+              GEN8_RPCS_EU_MIN_SHIFT;
+        GEM_BUG_ON(val & ~GEN8_RPCS_EU_MIN_MASK);
+        val &= GEN8_RPCS_EU_MIN_MASK;
+
+        rpcs |= val;
+
+        val = INTEL_INFO(dev_priv)->sseu.eu_per_subslice <<
+              GEN8_RPCS_EU_MAX_SHIFT;
+        GEM_BUG_ON(val & ~GEN8_RPCS_EU_MAX_MASK);
+        val &= GEN8_RPCS_EU_MAX_MASK;
+
+        rpcs |= val;
+
Quoted 8 lines

drivers/gpu/drm/i915/intel_lrc.c make_rpcs(struct drm_i915_private *dev_priv) · line 2581

Quoted 2 lines
drivers/gpu/drm/i915/intel_lrc.c +0 -0
           rpcs |= GEN8_RPCS_ENABLE;
       }

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

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

author maintainer
#6 lore raw

PATCH v2 1/1 drm/i915/icl: Fix context RPCS programming

From: Tvrtko Ursulin <[email protected]>

There are two issues with the current RPCS programming for Icelake:

Expansion of the slice count bitfield has been missed, as well as the required programming workaround for the subslice count bitfield size limitation.

1)

Bitfield width for configuring the active slice count has grown so we need to program the GEN8_R_PWR_CLK_STATE accordingly.

Current code was always requesting eight times the number of slices (due writting to a bitfield starting three bits higher than it should). These requests were luckily a) capped by the hardware to the available number of slices, and b) we haven't yet exported the code to ask for reduced slice configurations.

Due both of the above there was no impact from this incorrect programming but we should still fix it.

2)

Due subslice count bitfield being only three bits wide and furthermore capped to a maximum documented value of four, special programming workaround is needed to enable more than four subslices.

With this programming driver has to consider the GT configuration as 2x4x8, while the hardware internally translates this to 1x8x8.

A limitation stemming from this is that either a subslice count between one and four can be selected, or a subslice count equaling the total number of subslices in all selected slices. In other words, odd subslice counts greater than four are impossible, as are odd subslice counts greater than a single slice subslice count.

This also had no impact in the current code base due breakage from 1) always reqesting more than one slice.

While fixing this we also add some asserts to flag up any future bitfield overflows.

v2:
* Use a local in all branches for clarity. (Lionel)

Bspec: 12247

---
 drivers/gpu/drm/i915/i915_reg.h  |  2 +
 drivers/gpu/drm/i915/intel_lrc.c | 87 +++++++++++++++++++++++++++-----
 2 files changed, 76 insertions(+), 13 deletions(-)
drivers/gpu/drm/i915/i915_reg.h +2 -0
index f2321785cbd6..09bc8e730ee1 100644
@@ -344,6 +344,8 @@ static inline bool i915_mmio_reg_valid(i915_reg_t reg)
 #define   GEN8_RPCS_S_CNT_ENABLE	(1 << 18)
 #define   GEN8_RPCS_S_CNT_SHIFT		15
 #define   GEN8_RPCS_S_CNT_MASK		(0x7 << GEN8_RPCS_S_CNT_SHIFT)
+#define   GEN11_RPCS_S_CNT_SHIFT	12
+#define   GEN11_RPCS_S_CNT_MASK		(0x3f << GEN11_RPCS_S_CNT_SHIFT)
 #define   GEN8_RPCS_SS_CNT_ENABLE	(1 << 11)
 #define   GEN8_RPCS_SS_CNT_SHIFT	8
 #define   GEN8_RPCS_SS_CNT_MASK		(0x7 << GEN8_RPCS_SS_CNT_SHIFT)
drivers/gpu/drm/i915/intel_lrc.c +74 -13

110 lines not on this page · raw message

Signature · 6 lines
2.17.1

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

✗ Fi.CI.CHECKPATCH: warning for drm/i915/icl: Fix context RPCS programming (rev2)

In reply to Tvrtko Ursulin
== Series Details ==

== Series Details ==

Series: drm/i915/icl: Fix context RPCS programming (rev2)
URL : https://patchwork.freedesktop.org/series/49005/
State : warning

== Summary ==

$ dim checkpatch origin/drm-tip
9a2751c01ca2 drm/i915/icl: Fix context RPCS programming
-:18: WARNING:TYPO_SPELLING: 'writting' may be misspelled - perhaps 'writing'?
#18: 
writting to a bitfield starting three bits higher than it should). These

total: 0 errors, 1 warnings, 0 checks, 118 lines checked

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

#8 lore raw

✓ Fi.CI.BAT: success for drm/i915/icl: Fix context RPCS programming (rev2)

In reply to Tvrtko Ursulin
== Series Details ==

== Series Details ==

Series: drm/i915/icl: Fix context RPCS programming (rev2)
URL : https://patchwork.freedesktop.org/series/49005/
State : success

== Summary ==

= CI Bug Log - changes from CI_DRM_4757 -> Patchwork_10070 =

== Summary - SUCCESS ==

No regressions found.

External URL: https://patchwork.freedesktop.org/api/1.0/series/49005/revisions/2/mbox/

== Possible new issues ==

Here are the unknown changes that may have been introduced in Patchwork_10070:

=== IGT changes ===

    ==== Warnings ====

    {igt@pm_rpm@module-reload}:
      fi-hsw-4770r:       PASS -> SKIP

== Known issues ==

Here are the changes found in Patchwork_10070 that come from known issues:

=== IGT changes ===

    ==== Issues hit ====

    igt@drv_module_reload@basic-reload-inject:
      fi-hsw-4770r:       PASS -> DMESG-WARN (fdo#107425)

    {igt@pm_rpm@module-reload}:
      fi-cnl-psr:         PASS -> WARN (fdo#107708, fdo#107602)

    
    ==== Possible fixes ====

    igt@kms_pipe_crc_basic@nonblocking-crc-pipe-b-frame-sequence:
      {fi-byt-clapper}:   FAIL (fdo#103191, fdo#107362) -> PASS

    igt@kms_pipe_crc_basic@suspend-read-crc-pipe-c:
      fi-bxt-dsi:         INCOMPLETE (fdo#103927) -> PASS

    igt@kms_psr@primary_page_flip:
      fi-cnl-psr:         FAIL (fdo#107336) -> PASS

{name}: This element is suppressed. This means it is ignored when computing
the status of the difference (SUCCESS, WARNING, or FAILURE).

fdo#103191 https://bugs.freedesktop.org/show_bug.cgi?id=103191
fdo#103927 https://bugs.freedesktop.org/show_bug.cgi?id=103927
fdo#107336 https://bugs.freedesktop.org/show_bug.cgi?id=107336
fdo#107362 https://bugs.freedesktop.org/show_bug.cgi?id=107362
fdo#107425 https://bugs.freedesktop.org/show_bug.cgi?id=107425
fdo#107602 https://bugs.freedesktop.org/show_bug.cgi?id=107602
fdo#107708 https://bugs.freedesktop.org/show_bug.cgi?id=107708

== Participating hosts (53 -> 48) ==

Missing (5): fi-ctg-p8600 fi-ilk-m540 fi-byt-squawks fi-bsw-cyan fi-hsw-4200u

== Build changes ==

* Linux: CI_DRM_4757 -> Patchwork_10070

CI_DRM_4757: 1465de895e2b5d9e74e9a85189c9075155efa30d @ git://anongit.freedesktop.org/gfx-ci/linux
IGT_4621: 125eee6e981eac0a004aeb4f327439a132ceac5c @ git://anongit.freedesktop.org/xorg/app/intel-gpu-tools
Patchwork_10070: 9a2751c01ca21789da2a8f95dab23033540ad6c9 @ git://anongit.freedesktop.org/gfx-ci/linux

== Linux commits ==

9a2751c01ca2 drm/i915/icl: Fix context RPCS programming

== Logs ==

For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_10070/issues.html
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#9 lore raw

Re: ✓ Fi.CI.BAT: success for drm/i915/icl: Fix context RPCS programming (rev2)

Quoted 76 lines · Patchwork

== Series Details ==

Series: drm/i915/icl: Fix context RPCS programming (rev2)
URL : https://patchwork.freedesktop.org/series/49005/
State : success

== Summary ==

= CI Bug Log - changes from CI_DRM_4757 -> Patchwork_10070 =

== Summary - SUCCESS ==

No regressions found.

   External URL: https://patchwork.freedesktop.org/api/1.0/series/49005/revisions/2/mbox/

== Possible new issues ==

   Here are the unknown changes that may have been introduced in Patchwork_10070:

   === IGT changes ===

     ==== Warnings ====

     {igt@pm_rpm@module-reload}:
       fi-hsw-4770r:       PASS -> SKIP

== Known issues ==

   Here are the changes found in Patchwork_10070 that come from known issues:

   === IGT changes ===

     ==== Issues hit ====

     igt@drv_module_reload@basic-reload-inject:
       fi-hsw-4770r:       PASS -> DMESG-WARN (fdo#107425)

     {igt@pm_rpm@module-reload}:
       fi-cnl-psr:         PASS -> WARN (fdo#107708, fdo#107602)

     
     ==== Possible fixes ====

     igt@kms_pipe_crc_basic@nonblocking-crc-pipe-b-frame-sequence:
       {fi-byt-clapper}:   FAIL (fdo#103191, fdo#107362) -> PASS

     igt@kms_pipe_crc_basic@suspend-read-crc-pipe-c:
       fi-bxt-dsi:         INCOMPLETE (fdo#103927) -> PASS

     igt@kms_psr@primary_page_flip:
       fi-cnl-psr:         FAIL (fdo#107336) -> PASS

     
   {name}: This element is suppressed. This means it is ignored when computing
           the status of the difference (SUCCESS, WARNING, or FAILURE).

   fdo#103191 https://bugs.freedesktop.org/show_bug.cgi?id=103191
   fdo#103927 https://bugs.freedesktop.org/show_bug.cgi?id=103927
   fdo#107336 https://bugs.freedesktop.org/show_bug.cgi?id=107336
   fdo#107362 https://bugs.freedesktop.org/show_bug.cgi?id=107362
   fdo#107425 https://bugs.freedesktop.org/show_bug.cgi?id=107425
   fdo#107602 https://bugs.freedesktop.org/show_bug.cgi?id=107602
   fdo#107708 https://bugs.freedesktop.org/show_bug.cgi?id=107708

== Participating hosts (53 -> 48) ==

   Missing    (5): fi-ctg-p8600 fi-ilk-m540 fi-byt-squawks fi-bsw-cyan fi-hsw-4200u

== Build changes ==

* Linux: CI_DRM_4757 -> Patchwork_10070

Quoted 8 lines
   CI_DRM_4757: 1465de895e2b5d9e74e9a85189c9075155efa30d @ git://anongit.freedesktop.org/gfx-ci/linux
   IGT_4621: 125eee6e981eac0a004aeb4f327439a132ceac5c @ git://anongit.freedesktop.org/xorg/app/intel-gpu-tools
   Patchwork_10070: 9a2751c01ca21789da2a8f95dab23033540ad6c9 @ git://anongit.freedesktop.org/gfx-ci/linux

== Linux commits ==

9a2751c01ca2 drm/i915/icl: Fix context RPCS programming

Pushed, thanks for the review!

Regards,

Tvrtko

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

#10 lore raw

✓ Fi.CI.IGT: success for drm/i915/icl: Fix context RPCS programming (rev2)

In reply to Tvrtko Ursulin
== Series Details ==

== Series Details ==

Series: drm/i915/icl: Fix context RPCS programming (rev2)
URL : https://patchwork.freedesktop.org/series/49005/
State : success

== Summary ==

= CI Bug Log - changes from CI_DRM_4757_full -> Patchwork_10070_full =

== Summary - WARNING ==

Minor unknown changes coming with Patchwork_10070_full need to be verified manually.

If you think the reported changes have nothing to do with the changes introduced in Patchwork_10070_full, please notify your bug team to allow them to document this new failure mode, which will reduce false positives in CI.

== Possible new issues ==

Here are the unknown changes that may have been introduced in Patchwork_10070_full:

=== IGT changes ===

    ==== Warnings ====

    igt@perf_pmu@rc6:
      shard-kbl:          PASS -> SKIP

== Known issues ==

Here are the changes found in Patchwork_10070_full that come from known issues:

=== IGT changes ===

    ==== Issues hit ====

    igt@kms_cursor_legacy@cursor-vs-flip-toggle:
      shard-hsw:          PASS -> FAIL (fdo#103355)

    igt@kms_setmode@basic:
      shard-kbl:          PASS -> FAIL (fdo#99912)

    
    ==== Possible fixes ====

    igt@gem_exec_await@wide-contexts:
      shard-glk:          FAIL (fdo#105900) -> PASS

    igt@kms_frontbuffer_tracking@fbc-1p-primscrn-spr-indfb-draw-pwrite:
      shard-apl:          DMESG-WARN (fdo#105602, fdo#103558) -> PASS +2

    igt@kms_setmode@basic:
      shard-apl:          FAIL (fdo#99912) -> PASS

    igt@kms_vblank@pipe-b-ts-continuation-idle-hang:
      shard-apl:          DMESG-WARN -> PASS

    igt@pm_rc6_residency@rc6-accuracy:
      shard-glk:          INCOMPLETE (fdo#103359, k.org#198133) -> PASS

fdo#103355 https://bugs.freedesktop.org/show_bug.cgi?id=103355
fdo#103359 https://bugs.freedesktop.org/show_bug.cgi?id=103359
fdo#103558 https://bugs.freedesktop.org/show_bug.cgi?id=103558
fdo#105602 https://bugs.freedesktop.org/show_bug.cgi?id=105602
fdo#105900 https://bugs.freedesktop.org/show_bug.cgi?id=105900
fdo#99912 https://bugs.freedesktop.org/show_bug.cgi?id=99912
k.org#198133 https://bugzilla.kernel.org/show_bug.cgi?id=198133

== Participating hosts (5 -> 5) ==

No changes in participating hosts

== Build changes ==

* Linux: CI_DRM_4757 -> Patchwork_10070

CI_DRM_4757: 1465de895e2b5d9e74e9a85189c9075155efa30d @ git://anongit.freedesktop.org/gfx-ci/linux
IGT_4621: 125eee6e981eac0a004aeb4f327439a132ceac5c @ git://anongit.freedesktop.org/xorg/app/intel-gpu-tools
Patchwork_10070: 9a2751c01ca21789da2a8f95dab23033540ad6c9 @ git://anongit.freedesktop.org/gfx-ci/linux
piglit_4509: fdc5a4ca11124ab8413c7988896eec4c97336694 @ git://anongit.freedesktop.org/piglit

== Logs ==

For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_10070/shards.html
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx