Skip to contentSkip to conversation

drm/i915: There is only one fault register from GEN8 onwards

PATCH In review Michel Thierry 3 participants 7 messages Nov 10, 2017 – Nov 11, 2017

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 drm/i915: There is only one fault register from GEN8 onwards

Until Haswell/Baytrail, the hardware used to have a per engine fault register (e.g. 0x4094 - render fault register, 0x4194 - media fault register and so on). But since Broadwell, all these registers were combined into a singe one and the engine id stored in bits 14:12.

Not only we should not been reading (and writing to) registers that do not exist, in platforms with VCS2 (SKL), the address that would belong this engine (0x4494, VCS2_HW = 4) is already assigned to other register.

References: IHD-OS-BDW-Vol 2c-11.15, page 75.
References: IHD-OS-SKL-Vol 2c-05.16, page 350.

---
 drivers/gpu/drm/i915/i915_gem_gtt.c   | 38 ++++++++++++++++++++++++++++++-----
 drivers/gpu/drm/i915/i915_gpu_error.c |  8 +++++---
 drivers/gpu/drm/i915/i915_reg.h       |  2 ++
 3 files changed, 40 insertions(+), 8 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +33 -5
index 1e40eeb31f9d..66a907330ad2 100644
@@ -2256,16 +2256,13 @@ static bool needs_idle_maps(struct drm_i915_private *dev_priv)
 	return IS_GEN5(dev_priv) && IS_MOBILE(dev_priv) && intel_vtd_active();
 }
 
-void i915_check_and_clear_faults(struct drm_i915_private *dev_priv)
+void __check_and_clear_faults(struct drm_i915_private *dev_priv)
 {
 	struct intel_engine_cs *engine;
 	enum intel_engine_id id;
-
-	if (INTEL_INFO(dev_priv)->gen < 6)
-		return;
+	u32 fault_reg;
 
 	for_each_engine(engine, dev_priv, id) {
-		u32 fault_reg;
 		fault_reg = I915_READ(RING_FAULT_REG(engine));
 		if (fault_reg & RING_FAULT_VALID) {
 			DRM_DEBUG_DRIVER("Unexpected fault\n"
@@ -2281,6 +2278,37 @@ void i915_check_and_clear_faults(struct drm_i915_private *dev_priv)
 				   fault_reg & ~RING_FAULT_VALID);
 		}
 	}
+}
+
+void __gen8_check_and_clear_faults(struct drm_i915_private *dev_priv)
+{
+	u32 fault_reg = I915_READ(GEN8_RING_FAULT_REG);
+
+	if (fault_reg & RING_FAULT_VALID) {
+		DRM_DEBUG_DRIVER("Unexpected fault\n"
+				 "\tAddr: 0x%08lx\n"
+				 "\tEngine ID: %d\n"
+				 "\tSource ID: %d\n"
+				 "\tType: %d\n",
+				 fault_reg & PAGE_MASK,
+				 GEN8_RING_FAULT_ENGINE_ID(fault_reg),
+				 RING_FAULT_SRCID(fault_reg),
+				 RING_FAULT_FAULT_TYPE(fault_reg));
+		I915_WRITE(GEN8_RING_FAULT_REG,
+			   fault_reg & ~RING_FAULT_VALID);
+	}
+}
+
+void i915_check_and_clear_faults(struct drm_i915_private *dev_priv)
+{
+	if (INTEL_GEN(dev_priv) < 6)
+		return;
+
+	/* From GEN8 onwards we only have one 'All Engine Fault Register' */
+	if (INTEL_GEN(dev_priv) >= 8)
+		__gen8_check_and_clear_faults(dev_priv);
+	else
+		__check_and_clear_faults(dev_priv);
 
 	/* Engine specific init may not have been done till this point. */
 	if (dev_priv->engine[RCS])
drivers/gpu/drm/i915/i915_gpu_error.c +5 -3
index 5c2d83a838d8..7481c8e1b5a8 100644
@@ -1217,11 +1217,13 @@ static void error_record_engine_registers(struct i915_gpu_state *error,
 
 	if (INTEL_GEN(dev_priv) >= 6) {
 		ee->rc_psmi = I915_READ(RING_PSMI_CTL(engine->mmio_base));
-		ee->fault_reg = I915_READ(RING_FAULT_REG(engine));
-		if (INTEL_GEN(dev_priv) >= 8)
+		if (INTEL_GEN(dev_priv) >= 8) {
 			gen8_record_semaphore_state(error, engine, ee);
-		else
+			ee->fault_reg = I915_READ(GEN8_RING_FAULT_REG);
+		} else {
 			gen6_record_semaphore_state(engine, ee);
+			ee->fault_reg = I915_READ(RING_FAULT_REG(engine));
+		}
 	}
 
 	if (INTEL_GEN(dev_priv) >= 4) {
drivers/gpu/drm/i915/i915_reg.h +2 -0
index 6ef33422f762..7ef82ba670e9 100644
@@ -2329,6 +2329,8 @@ enum i915_power_well_id {
 #define   ARB_MODE_SWIZZLE_BDW	(1<<1)
 #define RENDER_HWS_PGA_GEN7	_MMIO(0x04080)
 #define RING_FAULT_REG(engine)	_MMIO(0x4094 + 0x100*(engine)->hw_id)
+#define GEN8_RING_FAULT_REG	_MMIO(0x4094)
+#define   GEN8_RING_FAULT_ENGINE_ID(x)	(((x) >> 12) & 0x7)
 #define   RING_FAULT_GTTSEL_MASK (1<<11)
 #define   RING_FAULT_SRCID(x)	(((x) >> 3) & 0xff)
 #define   RING_FAULT_FAULT_TYPE(x) (((x) >> 1) & 0x3)
Signature · 6 lines
2.15.0

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

✓ Fi.CI.BAT: success for drm/i915: There is only one fault register from Gen8 onwards (rev3)

== Series Details ==

== Series Details ==

Series: drm/i915: There is only one fault register from Gen8 onwards (rev3)
URL : https://patchwork.freedesktop.org/series/26317/
State : success

== Summary ==

Series 26317v3 drm/i915: There is only one fault register from Gen8 onwards
https://patchwork.freedesktop.org/api/1.0/series/26317/revisions/3/mbox/

Test gem_exec_suspend:

        Subgroup basic-s3:
                dmesg-fail -> PASS       (fi-kbl-7560u) fdo#103039
        Subgroup basic-s4-devices:
                dmesg-fail -> PASS       (fi-kbl-7560u) fdo#102846 +1
Test gem_flink_basic:
        Subgroup bad-flink:
                dmesg-warn -> PASS       (fi-kbl-7560u) fdo#103049 +4
Test gem_linear_blits:
        Subgroup basic:
                incomplete -> PASS       (fi-kbl-7560u) fdo#103163
Test kms_cursor_legacy:
        Subgroup basic-busy-flip-before-cursor-legacy:
                fail       -> PASS       (fi-gdg-551) fdo#102618

fdo#103039 https://bugs.freedesktop.org/show_bug.cgi?id=103039
fdo#102846 https://bugs.freedesktop.org/show_bug.cgi?id=102846
fdo#103049 https://bugs.freedesktop.org/show_bug.cgi?id=103049
fdo#103163 https://bugs.freedesktop.org/show_bug.cgi?id=103163
fdo#102618 https://bugs.freedesktop.org/show_bug.cgi?id=102618

fi-bdw-5557u     total:289  pass:268  dwarn:0   dfail:0   fail:0   skip:21  time:443s
fi-bdw-gvtdvm    total:289  pass:265  dwarn:0   dfail:0   fail:0   skip:24  time:456s
fi-blb-e6850     total:289  pass:223  dwarn:1   dfail:0   fail:0   skip:65  time:382s
fi-bsw-n3050     total:289  pass:243  dwarn:0   dfail:0   fail:0   skip:46  time:548s
fi-bwr-2160      total:289  pass:183  dwarn:0   dfail:0   fail:0   skip:106 time:274s
fi-bxt-j4205     total:289  pass:260  dwarn:0   dfail:0   fail:0   skip:29  time:510s
fi-byt-j1900     total:289  pass:254  dwarn:0   dfail:0   fail:0   skip:35  time:497s
fi-byt-n2820     total:289  pass:250  dwarn:0   dfail:0   fail:0   skip:39  time:489s
fi-elk-e7500     total:289  pass:229  dwarn:0   dfail:0   fail:0   skip:60  time:428s
fi-gdg-551       total:289  pass:178  dwarn:1   dfail:0   fail:1   skip:109 time:263s
fi-glk-1         total:289  pass:261  dwarn:0   dfail:0   fail:0   skip:28  time:544s
fi-hsw-4770      total:289  pass:262  dwarn:0   dfail:0   fail:0   skip:27  time:436s
fi-hsw-4770r     total:289  pass:262  dwarn:0   dfail:0   fail:0   skip:27  time:443s
fi-ilk-650       total:289  pass:228  dwarn:0   dfail:0   fail:0   skip:61  time:430s
fi-ivb-3520m     total:289  pass:260  dwarn:0   dfail:0   fail:0   skip:29  time:474s
fi-ivb-3770      total:289  pass:260  dwarn:0   dfail:0   fail:0   skip:29  time:461s
fi-kbl-7500u     total:289  pass:264  dwarn:1   dfail:0   fail:0   skip:24  time:489s
fi-kbl-7560u     total:247  pass:230  dwarn:0   dfail:0   fail:0   skip:16 
fi-kbl-7567u     total:289  pass:269  dwarn:0   dfail:0   fail:0   skip:20  time:476s
fi-kbl-r         total:289  pass:262  dwarn:0   dfail:0   fail:0   skip:27  time:544s
fi-pnv-d510      total:289  pass:222  dwarn:1   dfail:0   fail:0   skip:66  time:568s
fi-skl-6260u     total:289  pass:269  dwarn:0   dfail:0   fail:0   skip:20  time:455s
fi-skl-6600u     total:289  pass:262  dwarn:0   dfail:0   fail:0   skip:27  time:549s
fi-skl-6700hq    total:289  pass:263  dwarn:0   dfail:0   fail:0   skip:26  time:566s
fi-skl-6700k     total:289  pass:265  dwarn:0   dfail:0   fail:0   skip:24  time:523s
fi-skl-6770hq    total:289  pass:269  dwarn:0   dfail:0   fail:0   skip:20  time:503s
fi-skl-gvtdvm    total:289  pass:266  dwarn:0   dfail:0   fail:0   skip:23  time:460s
fi-snb-2520m     total:289  pass:250  dwarn:0   dfail:0   fail:0   skip:39  time:565s
fi-snb-2600      total:289  pass:249  dwarn:0   dfail:0   fail:0   skip:40  time:425s
Blacklisted hosts:
fi-cfl-s         total:289  pass:254  dwarn:3   dfail:0   fail:0   skip:32  time:546s
fi-glk-dsi       total:62   pass:45   dwarn:0   dfail:0   fail:0   skip:16 

fi-bxt-dsi failed to connect after reboot

6f3dca38f7a939db9658bd11d9c37357089cabef drm-tip: 2017y-11m-10d-12h-56m-44s UTC integration manifest bf772f904fe3 drm/i915: There is only one fault register from GEN8 onwards

== Logs ==

For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_7067/
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#2 lore raw

✓ Fi.CI.IGT: success for drm/i915: There is only one fault register from Gen8 onwards (rev3)

In reply to Michel Thierry
== Series Details ==

== Series Details ==

Series: drm/i915: There is only one fault register from Gen8 onwards (rev3)
URL : https://patchwork.freedesktop.org/series/26317/
State : success

== Summary ==

Test kms_flip:

        Subgroup modeset-vs-vblank-race-interruptible:
                fail       -> PASS       (shard-hsw) fdo#103060
        Subgroup plain-flip-ts-check-interruptible:
                pass       -> FAIL       (shard-hsw) fdo#100368
Test perf:
        Subgroup blocking:
                fail       -> PASS       (shard-hsw) fdo#102252
Test kms_frontbuffer_tracking:
        Subgroup fbc-farfromfence:
                skip       -> PASS       (shard-hsw)
Test kms_busy:
        Subgroup extended-modeset-hang-oldfb-with-reset-render-a:
                pass       -> DMESG-WARN (shard-hsw) fdo#102249

fdo#103060 https://bugs.freedesktop.org/show_bug.cgi?id=103060
fdo#100368 https://bugs.freedesktop.org/show_bug.cgi?id=100368
fdo#102252 https://bugs.freedesktop.org/show_bug.cgi?id=102252
fdo#102249 https://bugs.freedesktop.org/show_bug.cgi?id=102249

shard-hsw total:2584 pass:1471 dwarn:2 dfail:1 fail:11 skip:1099 time:9454s
Blacklisted hosts:

shard-apl        total:2584 pass:1620 dwarn:4   dfail:1   fail:23  skip:936 time:13241s
shard-kbl        total:2556 pass:1688 dwarn:17  dfail:1   fail:23  skip:825 time:9926s
shard-snb        total:2584 pass:1211 dwarn:2   dfail:1   fail:12  skip:1358 time:7801s

== Logs ==

For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_7067/shards.html
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#3 lore raw
In reply to Michel Thierry

Quoting Michel Thierry (2017-11-10 19:01:16)

Quoted 23 lines

Until Haswell/Baytrail, the hardware used to have a per engine fault register (e.g. 0x4094 - render fault register, 0x4194 - media fault register and so on). But since Broadwell, all these registers were combined into a singe one and the engine id stored in bits 14:12.

Not only we should not been reading (and writing to) registers that do not exist, in platforms with VCS2 (SKL), the address that would belong this engine (0x4494, VCS2_HW = 4) is already assigned to other register.

References: IHD-OS-BDW-Vol 2c-11.15, page 75.
References: IHD-OS-SKL-Vol 2c-05.16, page 350.

---
 drivers/gpu/drm/i915/i915_gem_gtt.c   | 38 ++++++++++++++++++++++++++++++-----
 drivers/gpu/drm/i915/i915_gpu_error.c |  8 +++++---
 drivers/gpu/drm/i915/i915_reg.h       |  2 ++
 3 files changed, 40 insertions(+), 8 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +0 -0
index 1e40eeb31f9d..66a907330ad2 100644

drivers/gpu/drm/i915/i915_gem_gtt.c static bool needs_idle_maps(struct drm_i915_private *dev_priv) · line 2259

Quoted 5 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +1 -1
@@ -2256,16 +2256,13 @@ static bool needs_idle_maps(struct drm_i915_private *dev_priv)
        return IS_GEN5(dev_priv) && IS_MOBILE(dev_priv) && intel_vtd_active();
 }
 
-void i915_check_and_clear_faults(struct drm_i915_private *dev_priv)
+void __check_and_clear_faults(struct drm_i915_private *dev_priv)

I am amazed that -Wextra doesn't complain. Try with sparse.

There's an old thread where this was raised and how we are not clearing the faults early enough.
-Chris
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#4 lore raw
Quoted 27 lines · Chris Wilson

Quoting Michel Thierry (2017-11-10 19:01:16)

Quoted 26 lines

Until Haswell/Baytrail, the hardware used to have a per engine fault register (e.g. 0x4094 - render fault register, 0x4194 - media fault register and so on). But since Broadwell, all these registers were combined into a singe one and the engine id stored in bits 14:12.

Not only we should not been reading (and writing to) registers that do not exist, in platforms with VCS2 (SKL), the address that would belong this engine (0x4494, VCS2_HW = 4) is already assigned to other register.

References: IHD-OS-BDW-Vol 2c-11.15, page 75.
References: IHD-OS-SKL-Vol 2c-05.16, page 350.

---
  drivers/gpu/drm/i915/i915_gem_gtt.c   | 38 ++++++++++++++++++++++++++++++-----
  drivers/gpu/drm/i915/i915_gpu_error.c |  8 +++++---
  drivers/gpu/drm/i915/i915_reg.h       |  2 ++
  3 files changed, 40 insertions(+), 8 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +0 -0
index 1e40eeb31f9d..66a907330ad2 100644
@@ -2256,16 +2256,13 @@ static bool needs_idle_maps(struct drm_i915_private *dev_priv)
         return IS_GEN5(dev_priv) && IS_MOBILE(dev_priv) && intel_vtd_active();
  }
 
Quoted 5 lines

drivers/gpu/drm/i915/i915_gem_gtt.c static bool needs_idle_maps(struct drm_i915_private *dev_priv) · line 2259

Quoted 2 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +1 -1
-void i915_check_and_clear_faults(struct drm_i915_private *dev_priv)
+void __check_and_clear_faults(struct drm_i915_private *dev_priv)

I am amazed that -Wextra doesn't complain. Try with sparse.

Hmm, no they didn't... gen6_check_and_clear_faults and gen8_check_and_clear_faults are less controversial.

Quoted 2 lines

There's an old thread where this was raised and how we are not clearing the faults early enough.

Ah, I found that thread [1], and as you said there, right now the for_each_engine is a nop (with this patch at least gen8+ would do the right thing).

Since gen6/gen7 engines are well known and can't change, do we really need for_each_engine?

[1] https://patchwork.freedesktop.org/patch/101331/
_______________________________________________
Intel-gfx mailing list [email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#5 lore raw

Quoting Michel Thierry (2017-11-10 23:42:31)

Quoted 28 lines · Chris Wilson
Quoted 27 lines

Quoting Michel Thierry (2017-11-10 19:01:16)

Quoted 26 lines

Until Haswell/Baytrail, the hardware used to have a per engine fault register (e.g. 0x4094 - render fault register, 0x4194 - media fault register and so on). But since Broadwell, all these registers were combined into a singe one and the engine id stored in bits 14:12.

Not only we should not been reading (and writing to) registers that do not exist, in platforms with VCS2 (SKL), the address that would belong this engine (0x4494, VCS2_HW = 4) is already assigned to other register.

References: IHD-OS-BDW-Vol 2c-11.15, page 75.
References: IHD-OS-SKL-Vol 2c-05.16, page 350.

---
  drivers/gpu/drm/i915/i915_gem_gtt.c   | 38 ++++++++++++++++++++++++++++++-----
  drivers/gpu/drm/i915/i915_gpu_error.c |  8 +++++---
  drivers/gpu/drm/i915/i915_reg.h       |  2 ++
  3 files changed, 40 insertions(+), 8 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +0 -0
index 1e40eeb31f9d..66a907330ad2 100644
@@ -2256,16 +2256,13 @@ static bool needs_idle_maps(struct drm_i915_private *dev_priv)
         return IS_GEN5(dev_priv) && IS_MOBILE(dev_priv) && intel_vtd_active();
  }
 
Quoted 8 lines
Quoted 5 lines

drivers/gpu/drm/i915/i915_gem_gtt.c static bool needs_idle_maps(struct drm_i915_private *dev_priv) · line 2259

Quoted 2 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +1 -1
-void i915_check_and_clear_faults(struct drm_i915_private *dev_priv)
+void __check_and_clear_faults(struct drm_i915_private *dev_priv)

I am amazed that -Wextra doesn't complain. Try with sparse.

Hmm, no they didn't... gen6_check_and_clear_faults and gen8_check_and_clear_faults are less controversial.

The missing static and lack of extern declaration. I know sparse catches it, I guess for K&R the order is just right for it not to complain.

Quoted 9 lines
Quoted 2 lines

There's an old thread where this was raised and how we are not clearing the faults early enough.

Ah, I found that thread [1], and as you said there, right now the for_each_engine is a nop (with this patch at least gen8+ would do the right thing).

Since gen6/gen7 engines are well known and can't change, do we really need for_each_engine?

For consistency, yeah it would be nice to keep for_each_engine. We should be ok to do the clear around i915_driver_init_mmio (it's mmio, it has to be there ;), maybe worth pulling it into intel_engines_init_mmio() or i915_gem_init_mmio(). I'm learning towards sanitization from within intel_engines_init_mmio()
-Chris
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#6 lore raw
Quoted 48 lines · Chris Wilson

Quoting Michel Thierry (2017-11-10 23:42:31)

Quoted 28 lines · Chris Wilson
Quoted 27 lines

Quoting Michel Thierry (2017-11-10 19:01:16)

Quoted 26 lines

Until Haswell/Baytrail, the hardware used to have a per engine fault register (e.g. 0x4094 - render fault register, 0x4194 - media fault register and so on). But since Broadwell, all these registers were combined into a singe one and the engine id stored in bits 14:12.

Not only we should not been reading (and writing to) registers that do not exist, in platforms with VCS2 (SKL), the address that would belong this engine (0x4494, VCS2_HW = 4) is already assigned to other register.

References: IHD-OS-BDW-Vol 2c-11.15, page 75.
References: IHD-OS-SKL-Vol 2c-05.16, page 350.

---
   drivers/gpu/drm/i915/i915_gem_gtt.c   | 38 ++++++++++++++++++++++++++++++-----
   drivers/gpu/drm/i915/i915_gpu_error.c |  8 +++++---
   drivers/gpu/drm/i915/i915_reg.h       |  2 ++
   3 files changed, 40 insertions(+), 8 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +0 -0
index 1e40eeb31f9d..66a907330ad2 100644
@@ -2256,16 +2256,13 @@ static bool needs_idle_maps(struct drm_i915_private *dev_priv)
          return IS_GEN5(dev_priv) && IS_MOBILE(dev_priv) && intel_vtd_active();
   }
 
Quoted 8 lines
Quoted 5 lines

drivers/gpu/drm/i915/i915_gem_gtt.c static bool needs_idle_maps(struct drm_i915_private *dev_priv) · line 2259

Quoted 2 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +1 -1
-void i915_check_and_clear_faults(struct drm_i915_private *dev_priv)
+void __check_and_clear_faults(struct drm_i915_private *dev_priv)

I am amazed that -Wextra doesn't complain. Try with sparse.

Hmm, no they didn't... gen6_check_and_clear_faults and gen8_check_and_clear_faults are less controversial.

The missing static and lack of extern declaration. I know sparse catches it, I guess for K&R the order is just right for it not to complain.

Quoted 7 lines
Quoted 2 lines

There's an old thread where this was raised and how we are not clearing the faults early enough.

Ah, I found that thread [1], and as you said there, right now the for_each_engine is a nop (with this patch at least gen8+ would do the right thing).

Quoted 8 lines
Quoted 2 lines

Since gen6/gen7 engines are well known and can't change, do we really need for_each_engine?

For consistency, yeah it would be nice to keep for_each_engine. We should be ok to do the clear around i915_driver_init_mmio (it's mmio, it has to be there ;), maybe worth pulling it into intel_engines_init_mmio() or i915_gem_init_mmio(). I'm learning towards sanitization from within intel_engines_init_mmio()

Ok, I'll move the check_and_clear_faults from intel_uncore_init to intel_engines_init_mmio (that's the one checking at driver load time).

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