Skip to contentSkip to conversation

Misc forcewake patches

PATCH In review Tvrtko Ursulin 4 participants 11 messages Sep 28, 2016

Subsystems (2)
lore
Share & tools
lore

Tvrtko Ursulin

author maintainer
#0 lore raw

cover Misc forcewake patches

From: Tvrtko Ursulin <[email protected]>

Apart from 2/3 which I already sent yesterday, there are two new patches here and the end result of the series is approximately 3KiB reduction in i915.ko text size.

Tvrtko Ursulin (3):
drm/i915: Remove redundant hsw_write* mmio functions drm/i915: Keep track of active forcewake domains in a bitmask drm/i915: Do not inline forcewake taking in mmio accessors

 drivers/gpu/drm/i915/i915_drv.h     |  2 +
 drivers/gpu/drm/i915/intel_uncore.c | 90 +++++++++++++------------------------
 2 files changed, 34 insertions(+), 58 deletions(-)
Signature · 6 lines
2.7.4

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

PATCH 1/3 drm/i915: Remove redundant hsw_write* mmio functions

From: Tvrtko Ursulin <[email protected]>

They are completely identical to gen6_write* ones.

---
 drivers/gpu/drm/i915/intel_uncore.c | 25 +------------------------
 1 file changed, 1 insertion(+), 24 deletions(-)
drivers/gpu/drm/i915/intel_uncore.c +1 -24
index a9b6c936aadd..478364057812 100644
@@ -1055,21 +1055,6 @@ gen6_write##x(struct drm_i915_private *dev_priv, i915_reg_t reg, u##x val, bool
 	GEN6_WRITE_FOOTER; \
 }
 
-#define __hsw_write(x) \
-static void \
-hsw_write##x(struct drm_i915_private *dev_priv, i915_reg_t reg, u##x val, bool trace) { \
-	u32 __fifo_ret = 0; \
-	GEN6_WRITE_HEADER; \
-	if (NEEDS_FORCE_WAKE(offset)) { \
-		__fifo_ret = __gen6_gt_wait_for_fifo(dev_priv); \
-	} \
-	__raw_i915_write##x(dev_priv, reg, val); \
-	if (unlikely(__fifo_ret)) { \
-		gen6_gt_check_fifodbg(dev_priv); \
-	} \
-	GEN6_WRITE_FOOTER; \
-}
-
 #define __gen8_write(x) \
 static void \
 gen8_write##x(struct drm_i915_private *dev_priv, i915_reg_t reg, u##x val, bool trace) { \
@@ -1116,9 +1101,6 @@ __chv_write(32)
 __gen8_write(8)
 __gen8_write(16)
 __gen8_write(32)
-__hsw_write(8)
-__hsw_write(16)
-__hsw_write(32)
 __gen6_write(8)
 __gen6_write(16)
 __gen6_write(32)
@@ -1126,7 +1108,6 @@ __gen6_write(32)
 #undef __gen9_write
 #undef __chv_write
 #undef __gen8_write
-#undef __hsw_write
 #undef __gen6_write
 #undef GEN6_WRITE_FOOTER
 #undef GEN6_WRITE_HEADER
@@ -1343,11 +1324,7 @@ void intel_uncore_init(struct drm_i915_private *dev_priv)
 		break;
 	case 7:
 	case 6:
-		if (IS_HASWELL(dev_priv)) {
-			ASSIGN_WRITE_MMIO_VFUNCS(hsw);
-		} else {
-			ASSIGN_WRITE_MMIO_VFUNCS(gen6);
-		}
+		ASSIGN_WRITE_MMIO_VFUNCS(gen6);
 
 		if (IS_VALLEYVIEW(dev_priv)) {
 			ASSIGN_READ_MMIO_VFUNCS(vlv);
Signature · 6 lines
2.7.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
maintainer Reviewed-by on v1 1/3
#2 lore raw

Re: [PATCH 1/3] drm/i915: Remove redundant hsw_write* mmio functions

Quoted 3 lines · Tvrtko Ursulin

From: Tvrtko Ursulin <[email protected]>

They are completely identical to gen6_write* ones.

Seems to be true.

Regards, Joonas

Signature · 7 lines
Joonas Lahtinen
Open Source Technology Center
Intel Corporation
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Reviewed-by on v1 1/3
#3 lore raw

Re: [PATCH 1/3] drm/i915: Remove redundant hsw_write* mmio functions

In reply to Tvrtko Ursulin
Quoted 3 lines · Tvrtko Ursulin

From: Tvrtko Ursulin <[email protected]>

They are completely identical to gen6_write* ones.

iirc, the difference used to be mmiodebug.

Quoted 1 line

-Chris

Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
author maintainer
#4 lore raw

PATCH 2/3 drm/i915: Keep track of active forcewake domains in a bitmask

In reply to Tvrtko Ursulin

From: Tvrtko Ursulin <[email protected]>

There are current places in the code, and there will be more in the future, which iterate the forcewake domains to find out which ones are currently active.

To save them from doing this iteration, we can cheaply keep a mask of active domains in dev_priv->uncore.fw_domains_active.

This has no cost in terms of object size, even manages to shrink it overall by 368 bytes on my config.

---
 drivers/gpu/drm/i915/i915_drv.h     |  2 ++
 drivers/gpu/drm/i915/intel_uncore.c | 54 ++++++++++++++++---------------------
 2 files changed, 25 insertions(+), 31 deletions(-)
drivers/gpu/drm/i915/i915_drv.h +2 -0
index 23bc43d23d2c..4cd727376d1d 100644
@@ -588,7 +588,9 @@ struct intel_uncore {
 	struct intel_uncore_funcs funcs;
 
 	unsigned fifo_count;
+
 	enum forcewake_domains fw_domains;
+	enum forcewake_domains fw_domains_active;
 
 	struct intel_uncore_forcewake_domain {
 		struct drm_i915_private *i915;
drivers/gpu/drm/i915/intel_uncore.c +23 -31

116 lines not on this page · raw message

Signature · 6 lines
2.7.4

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

PATCH 3/3 drm/i915: Do not inline forcewake taking in mmio accessors

In reply to Tvrtko Ursulin

From: Tvrtko Ursulin <[email protected]>

Once we know we need to take new forcewakes, that being a slow operation, it does not make sense to inline that code into every mmio accessor.

Move it to a separate function and save some code.

---
 drivers/gpu/drm/i915/intel_uncore.c | 25 +++++++++++++++----------
 1 file changed, 15 insertions(+), 10 deletions(-)
drivers/gpu/drm/i915/intel_uncore.c +15 -10
index 079102b46fd3..c887ab0947c6 100644
@@ -862,11 +862,22 @@ __gen2_read(64)
 	trace_i915_reg_rw(false, reg, val, sizeof(val), trace); \
 	return val
 
-static inline void __force_wake_auto(struct drm_i915_private *dev_priv,
-				     enum forcewake_domains fw_domains)
+static void ___force_wake_auto(struct drm_i915_private *dev_priv,
+			       enum forcewake_domains fw_domains)
 {
 	struct intel_uncore_forcewake_domain *domain;
 
+	/* Ideally GCC would be constant-fold and eliminate this loop */
+	for_each_fw_domain_masked(domain, fw_domains, dev_priv)
+		fw_domain_arm_timer(domain);
+
+	dev_priv->uncore.funcs.force_wake_get(dev_priv, fw_domains);
+	dev_priv->uncore.fw_domains_active |= fw_domains;
+}
+
+static inline void __force_wake_auto(struct drm_i915_private *dev_priv,
+				     enum forcewake_domains fw_domains)
+{
 	if (WARN_ON(!fw_domains))
 		return;
 
@@ -874,14 +885,8 @@ static inline void __force_wake_auto(struct drm_i915_private *dev_priv,
 	fw_domains &= dev_priv->uncore.fw_domains;
 	fw_domains &= ~dev_priv->uncore.fw_domains_active;
 
-	if (fw_domains) {
-		/* Ideally GCC would be constant-fold and eliminate this loop */
-		for_each_fw_domain_masked(domain, fw_domains, dev_priv)
-			fw_domain_arm_timer(domain);
-
-		dev_priv->uncore.funcs.force_wake_get(dev_priv, fw_domains);
-		dev_priv->uncore.fw_domains_active |= fw_domains;
-	}
+	if (fw_domains)
+		___force_wake_auto(dev_priv, fw_domains);
 }
 
 #define __gen6_read(x) \
Signature · 6 lines
2.7.4

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

Re: [PATCH 3/3] drm/i915: Do not inline forcewake taking in mmio accessors

Quoted 18 lines · Tvrtko Ursulin

From: Tvrtko Ursulin <[email protected]>

Once we know we need to take new forcewakes, that being a slow operation, it does not make sense to inline that code into every mmio accessor.

Move it to a separate function and save some code.

---
 drivers/gpu/drm/i915/intel_uncore.c | 25 +++++++++++++++----------
 1 file changed, 15 insertions(+), 10 deletions(-)
drivers/gpu/drm/i915/intel_uncore.c +0 -0
index 079102b46fd3..c887ab0947c6 100644

drivers/gpu/drm/i915/intel_uncore.c __gen2_read(64) · line 866

Quoted 5 lines
drivers/gpu/drm/i915/intel_uncore.c +0 -2
@@ -862,11 +862,22 @@ __gen2_read(64)
 	trace_i915_reg_rw(false, reg, val, sizeof(val), trace); \
 	return val
 
-static inline void __force_wake_auto(struct drm_i915_private *dev_priv,
-				     enum forcewake_domains fw_domains)

noinline just to reinforce?

drivers/gpu/drm/i915/intel_uncore.c __gen2_read(64) · line 870

Quoted 6 lines
drivers/gpu/drm/i915/intel_uncore.c +3 -0
+static void ___force_wake_auto(struct drm_i915_private *dev_priv,
+			       enum forcewake_domains fw_domains)
 {
 	struct intel_uncore_forcewake_domain *domain;
 
+	/* Ideally GCC would be constant-fold and eliminate this loop */

That comment no longer applies, since we are now out of line.

Quoted 14 lines
drivers/gpu/drm/i915/intel_uncore.c +10 -0
+	for_each_fw_domain_masked(domain, fw_domains, dev_priv)
+		fw_domain_arm_timer(domain);
+
+	dev_priv->uncore.funcs.force_wake_get(dev_priv, fw_domains);
+	dev_priv->uncore.fw_domains_active |= fw_domains;
+}
+
+static inline void __force_wake_auto(struct drm_i915_private *dev_priv,
+				     enum forcewake_domains fw_domains)
+{
 	if (WARN_ON(!fw_domains))
 		return;
 

drivers/gpu/drm/i915/intel_uncore.c static inline void __force_wake_auto(struct drm_i915_private *dev_priv, · line 886

Quoted 2 lines
drivers/gpu/drm/i915/intel_uncore.c +0 -0
@@ -874,14 +885,8 @@ static inline void __force_wake_auto(struct drm_i915_private *dev_priv,
 	fw_domains &= dev_priv->uncore.fw_domains;
 	fw_domains &= ~dev_priv->uncore.fw_domains_active;

Ok, that seems a reasonable justification (and that is what would neuter the /* Ideally... */ comment).
-Chris

Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
author maintainer
#7 lore raw

PATCH v2 1/1 drm/i915: Do not inline forcewake taking in mmio accessors

From: Tvrtko Ursulin <[email protected]>

Once we know we need to take new forcewakes, that being a slow operation, it does not make sense to inline that code into every mmio accessor.

Move it to a separate function and save some code.

v2: Be explicit with noinline and remove stale comment.
(Chris Wilson)

---
 drivers/gpu/drm/i915/intel_uncore.c | 24 ++++++++++++++----------
 1 file changed, 14 insertions(+), 10 deletions(-)
drivers/gpu/drm/i915/intel_uncore.c +14 -10
index 079102b46fd3..60137a164094 100644
@@ -862,11 +862,21 @@ __gen2_read(64)
 	trace_i915_reg_rw(false, reg, val, sizeof(val), trace); \
 	return val
 
-static inline void __force_wake_auto(struct drm_i915_private *dev_priv,
-				     enum forcewake_domains fw_domains)
+static noinline void ___force_wake_auto(struct drm_i915_private *dev_priv,
+					enum forcewake_domains fw_domains)
 {
 	struct intel_uncore_forcewake_domain *domain;
 
+	for_each_fw_domain_masked(domain, fw_domains, dev_priv)
+		fw_domain_arm_timer(domain);
+
+	dev_priv->uncore.funcs.force_wake_get(dev_priv, fw_domains);
+	dev_priv->uncore.fw_domains_active |= fw_domains;
+}
+
+static inline void __force_wake_auto(struct drm_i915_private *dev_priv,
+				     enum forcewake_domains fw_domains)
+{
 	if (WARN_ON(!fw_domains))
 		return;
 
@@ -874,14 +884,8 @@ static inline void __force_wake_auto(struct drm_i915_private *dev_priv,
 	fw_domains &= dev_priv->uncore.fw_domains;
 	fw_domains &= ~dev_priv->uncore.fw_domains_active;
 
-	if (fw_domains) {
-		/* Ideally GCC would be constant-fold and eliminate this loop */
-		for_each_fw_domain_masked(domain, fw_domains, dev_priv)
-			fw_domain_arm_timer(domain);
-
-		dev_priv->uncore.funcs.force_wake_get(dev_priv, fw_domains);
-		dev_priv->uncore.fw_domains_active |= fw_domains;
-	}
+	if (fw_domains)
+		___force_wake_auto(dev_priv, fw_domains);
 }
 
 #define __gen6_read(x) \
Signature · 6 lines
2.7.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Reviewed-by on v2 1/1
#8 lore raw

Re: [PATCH v2] drm/i915: Do not inline forcewake taking in mmio accessors

Quoted 6 lines · Tvrtko Ursulin

From: Tvrtko Ursulin <[email protected]>

Once we know we need to take new forcewakes, that being a slow operation, it does not make sense to inline that code into every mmio accessor.

Quoted 6 lines

Move it to a separate function and save some code.

v2: Be explicit with noinline and remove stale comment.
(Chris Wilson)

-Chris

Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#9 lore raw

✗ Fi.CI.BAT: warning for Misc forcewake patches

In reply to Tvrtko Ursulin
== Series Details ==

== Series Details ==

Series: Misc forcewake patches
URL : https://patchwork.freedesktop.org/series/13016/
State : warning

== Summary ==

Series 13016v1 Misc forcewake patches
https://patchwork.freedesktop.org/api/1.0/series/13016/revisions/1/mbox/

Test kms_pipe_crc_basic:

        Subgroup nonblocking-crc-pipe-b-frame-sequence:
                dmesg-warn -> PASS       (fi-ilk-650)
Test kms_psr_sink_crc:
        Subgroup psr_basic:
                pass       -> DMESG-WARN (fi-skl-6700hq)
Test pm_rpm:
        Subgroup basic-pci-d3-state:
                pass       -> DMESG-WARN (fi-skl-6770hq)
fi-bdw-5557u     total:244  pass:229  dwarn:0   dfail:0   fail:0   skip:15 
fi-bsw-n3050     total:244  pass:202  dwarn:0   dfail:0   fail:0   skip:42 
fi-bxt-t5700     total:244  pass:214  dwarn:0   dfail:0   fail:0   skip:30 
fi-byt-n2820     total:244  pass:208  dwarn:0   dfail:0   fail:1   skip:35 
fi-hsw-4770      total:244  pass:222  dwarn:0   dfail:0   fail:0   skip:22 
fi-hsw-4770r     total:244  pass:222  dwarn:0   dfail:0   fail:0   skip:22 
fi-ilk-650       total:244  pass:182  dwarn:0   dfail:0   fail:2   skip:60 
fi-ivb-3520m     total:244  pass:219  dwarn:0   dfail:0   fail:0   skip:25 
fi-ivb-3770      total:244  pass:207  dwarn:0   dfail:0   fail:0   skip:37 
fi-skl-6260u     total:244  pass:230  dwarn:0   dfail:0   fail:0   skip:14 
fi-skl-6700hq    total:244  pass:221  dwarn:1   dfail:0   fail:0   skip:22 
fi-skl-6700k     total:244  pass:219  dwarn:1   dfail:0   fail:0   skip:24 
fi-skl-6770hq    total:244  pass:227  dwarn:2   dfail:0   fail:1   skip:14 
fi-snb-2520m     total:244  pass:208  dwarn:0   dfail:0   fail:0   skip:36 
fi-snb-2600      total:244  pass:207  dwarn:0   dfail:0   fail:0   skip:37 

Results at /archive/results/CI_IGT_test/Patchwork_2585/

bdb6dbd5e9c1a8dc3ccdd7964092cf5d6bc08ca2 drm-intel-nightly: 2016y-09m-28d-12h-10m-04s UTC integration manifest 3a2cfbd drm/i915: Do not inline forcewake taking in mmio accessors
75ee595 drm/i915: Keep track of active forcewake domains in a bitmask
47cf3c0 drm/i915: Remove redundant hsw_write* mmio functions

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

#10 lore raw

✓ Fi.CI.BAT: success for Misc forcewake patches (rev2)

In reply to Tvrtko Ursulin
== Series Details ==

== Series Details ==

Series: Misc forcewake patches (rev2)
URL : https://patchwork.freedesktop.org/series/13016/
State : success

== Summary ==

Series 13016v2 Misc forcewake patches
https://patchwork.freedesktop.org/api/1.0/series/13016/revisions/2/mbox/

fi-bdw-5557u     total:33   pass:26   dwarn:0   dfail:1   fail:0   skip:5  
fi-bsw-n3050     total:33   pass:24   dwarn:0   dfail:1   fail:0   skip:7  
fi-bxt-t5700     total:33   pass:24   dwarn:0   dfail:1   fail:0   skip:7  
fi-byt-j1900     total:33   pass:27   dwarn:0   dfail:1   fail:0   skip:4  
fi-byt-n2820     total:33   pass:23   dwarn:0   dfail:1   fail:0   skip:8  
fi-hsw-4770      total:33   pass:24   dwarn:0   dfail:1   fail:0   skip:7  
fi-hsw-4770r     total:33   pass:24   dwarn:0   dfail:1   fail:0   skip:7  
fi-ilk-650       total:33   pass:12   dwarn:0   dfail:1   fail:0   skip:19 
fi-ivb-3520m     total:33   pass:27   dwarn:0   dfail:1   fail:0   skip:4  
fi-ivb-3770      total:33   pass:27   dwarn:0   dfail:1   fail:0   skip:4  
fi-skl-6260u     total:33   pass:26   dwarn:0   dfail:1   fail:0   skip:5  
fi-skl-6700hq    total:33   pass:25   dwarn:0   dfail:1   fail:0   skip:6  
fi-skl-6700k     total:33   pass:23   dwarn:1   dfail:1   fail:0   skip:7  
fi-skl-6770hq    total:33   pass:26   dwarn:0   dfail:1   fail:0   skip:5  
fi-snb-2520m     total:33   pass:26   dwarn:0   dfail:1   fail:0   skip:5  
fi-snb-2600      total:33   pass:26   dwarn:0   dfail:1   fail:0   skip:5  

Results at /archive/results/CI_IGT_test/Patchwork_2587/

edf68a3aacccc2ad1ff8915fcc8011ad43d53be9 drm-intel-nightly: 2016y-09m-28d-15h-13m-34s UTC integration manifest 6ea9a57 drm/i915: Do not inline forcewake taking in mmio accessors
ce3df76 drm/i915: Keep track of active forcewake domains in a bitmask
338c6a5 drm/i915: Remove redundant hsw_write* mmio functions

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