Skip to contentSkip to conversation

drm/i915/bdw: Only use 2g GGTT for 32b platforms

PATCH Superseded Ben Widawsky 7 participants 11 messages May 7, 2014 – May 20, 2014

Subsystems (2)

v1 of 5 Superseded Latest is v5 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 drm/i915/bdw: Only use 2g GGTT for 32b platforms

From: Ben Widawsky <[email protected]>

Daniel requested in the bug that I use a 3GB fallback size. Since this is not in the spec as a valid size, I decided against it. We could potentially add a patch to bump it to 3GB on top of this one.

This probably should be CC: stable - but I'll let the powers that be decide that one.

v2: Change ifdef to 32b, instead of ifndef
update comment

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=76619

---
 drivers/gpu/drm/i915/i915_gem_gtt.c | 8 ++++++++
 1 file changed, 8 insertions(+)
drivers/gpu/drm/i915/i915_gem_gtt.c +8 -0
index 846b6ee..d03a540 100644
@@ -1759,6 +1759,14 @@ static inline unsigned int gen8_get_total_gtt_size(u16 bdw_gmch_ctl)
 	bdw_gmch_ctl &= BDW_GMCH_GGMS_MASK;
 	if (bdw_gmch_ctl)
 		bdw_gmch_ctl = 1 << bdw_gmch_ctl;
+
+#ifdef CONFIG_32BIT
+	/* Limit 32B platforms to a 2GB GGTT
+	4 << 20 / pte size * PAGE_SIZE */
+	if (bdw_gmch_ctl > 4)
+		bdw_gmch_ctl = 4;
+#endif
+
 	return bdw_gmch_ctl << 20;
 }
 
Signature · 1 line
1.9.2
#1 lore raw
Quoted 23 lines · Ben Widawsky

From: Ben Widawsky <[email protected]>

Daniel requested in the bug that I use a 3GB fallback size. Since this is not in the spec as a valid size, I decided against it. We could potentially add a patch to bump it to 3GB on top of this one.

This probably should be CC: stable - but I'll let the powers that be decide that one.

v2: Change ifdef to 32b, instead of ifndef
update comment

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=76619

---
 drivers/gpu/drm/i915/i915_gem_gtt.c | 8 ++++++++
 1 file changed, 8 insertions(+)
drivers/gpu/drm/i915/i915_gem_gtt.c +0 -0
index 846b6ee..d03a540 100644

drivers/gpu/drm/i915/i915_gem_gtt.c static inline unsigned int gen8_get_total_gtt_size(u16 bdw_gmch_ctl) · line 1767

Quoted 9 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +6 -0
@@ -1759,6 +1759,14 @@ static inline unsigned int gen8_get_total_gtt_size(u16 bdw_gmch_ctl)
 	bdw_gmch_ctl &= BDW_GMCH_GGMS_MASK;
 	if (bdw_gmch_ctl)
 		bdw_gmch_ctl = 1 << bdw_gmch_ctl;
+
+#ifdef CONFIG_32BIT
+	/* Limit 32B platforms to a 2GB GGTT
+	4 << 20 / pte size * PAGE_SIZE */
+	if (bdw_gmch_ctl > 4)
+		bdw_gmch_ctl = 4;

Comment needs CodingStyle polish and we might as well return the right value directly instead of adjusting bdw_gmch_ctl. With that polish applied this is Reviewed-by: Daniel Vetter <[email protected]>o

And yes most definitely Cc: [email protected] since it's a regression.

Aside: Please add the regression tags when handling bugs, I need those for tracking and stats.
-Daniel

Quoted 12 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +2 -0
+#endif
+
 	return bdw_gmch_ctl << 20;
 }
 
Signature · 6 lines
1.9.2

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 3 lines
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch
#2 lore raw
Quoted 38 lines · Daniel Vetter
Quoted 23 lines · Ben Widawsky

From: Ben Widawsky <[email protected]>

Daniel requested in the bug that I use a 3GB fallback size. Since this is not in the spec as a valid size, I decided against it. We could potentially add a patch to bump it to 3GB on top of this one.

This probably should be CC: stable - but I'll let the powers that be decide that one.

v2: Change ifdef to 32b, instead of ifndef
update comment

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=76619

---
 drivers/gpu/drm/i915/i915_gem_gtt.c | 8 ++++++++
 1 file changed, 8 insertions(+)
drivers/gpu/drm/i915/i915_gem_gtt.c +0 -0
index 846b6ee..d03a540 100644

drivers/gpu/drm/i915/i915_gem_gtt.c static inline unsigned int gen8_get_total_gtt_size(u16 bdw_gmch_ctl) · line 1767

Quoted 9 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +6 -0
@@ -1759,6 +1759,14 @@ static inline unsigned int gen8_get_total_gtt_size(u16 bdw_gmch_ctl)
 	bdw_gmch_ctl &= BDW_GMCH_GGMS_MASK;
 	if (bdw_gmch_ctl)
 		bdw_gmch_ctl = 1 << bdw_gmch_ctl;
+
+#ifdef CONFIG_32BIT
+	/* Limit 32B platforms to a 2GB GGTT
+	4 << 20 / pte size * PAGE_SIZE */
+	if (bdw_gmch_ctl > 4)
+		bdw_gmch_ctl = 4;

Comment needs CodingStyle polish and we might as well return the right value directly instead of adjusting bdw_gmch_ctl. With that polish applied this is Reviewed-by: Daniel Vetter <[email protected]>o

Quoted 7 lines

And yes most definitely Cc: [email protected] since it's a regression.

Aside: Please add the regression tags when handling bugs, I need those for tracking and stats.
-Daniel

I don't know what a regression tag is.

Quoted 21 lines
Quoted 12 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +2 -0
+#endif
+
 	return bdw_gmch_ctl << 20;
 }
 
Signature · 6 lines
1.9.2

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 7 lines
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 1 line
Ben Widawsky, Intel Open Source Technology Center
maintainer
#3 lore raw
Quoted 7 lines · Ben Widawsky
Quoted 4 lines · Daniel Vetter

Aside: Please add the regression tags when handling bugs, I need those for tracking and stats.
-Daniel

I don't know what a regression tag is.

"[regression]" at the front of the bug summary line.

BR,
Jani.

Signature · 1 line
Jani Nikula, Intel Open Source Technology Center
#4 lore raw
Quoted 10 lines · Jani Nikula
Quoted 7 lines · Ben Widawsky
Quoted 4 lines · Daniel Vetter

Aside: Please add the regression tags when handling bugs, I need those for tracking and stats.
-Daniel

I don't know what a regression tag is.

"[regression]" at the front of the bug summary line.

Also valid are [bisected] (used by QA when they supply the bisect result) and on kernel bugzilla there's an explicit boolean for this.

Doing this is about the no. 1 job of doing bug scrub ... -Daniel

Signature · 3 lines
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch
#5 lore raw

PATCH v3 1/1 drm/i915/bdw: Only use 2g GGTT for 32b platforms

In reply to Ben Widawsky

Daniel requested in the bug that I use a 3GB fallback size. Since this is not in the spec as a valid size, I decided against it. We could potentially add a patch to bump it to 3GB on top of this one.

This probably should be CC: stable - but I'll let the powers that be decide that one.

Regression from a revert of the revert:
commit 7907f45bf9f67a1c5e5d4ae05bab428d7c2f43b2

Author: Ben Widawsky <[email protected]>
Date:   Wed Feb 19 22:05:46 2014 -0800

Revert "drm/i915/bdw: Limit GTT to 2GB"

v2: Change ifdef to 32b, instead of ifndef
update comment

v3. Update comment to not wrap (Daniel).
Update commit message

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=76619

---
 drivers/gpu/drm/i915/i915_gem_gtt.c | 7 +++++++
 1 file changed, 7 insertions(+)
drivers/gpu/drm/i915/i915_gem_gtt.c +7 -0
index 4d87bf2..8d6d531 100644
@@ -2640,6 +2640,13 @@ static inline unsigned int gen8_get_total_gtt_size(u16 bdw_gmch_ctl)
 	bdw_gmch_ctl &= BDW_GMCH_GGMS_MASK;
 	if (bdw_gmch_ctl)
 		bdw_gmch_ctl = 1 << bdw_gmch_ctl;
+
+#ifdef CONFIG_32BIT
+	/* Limit 32B platforms to a 2GB GGTT: 4 << 20 / pte size * PAGE_SIZE */
+	if (bdw_gmch_ctl > 4)
+		bdw_gmch_ctl = 4;
+#endif
+
 	return bdw_gmch_ctl << 20;
 }
 
Signature · 1 line
1.9.2
maintainer
#6 lore raw
Quoted 33 lines · Ben Widawsky

Daniel requested in the bug that I use a 3GB fallback size. Since this is not in the spec as a valid size, I decided against it. We could potentially add a patch to bump it to 3GB on top of this one.

This probably should be CC: stable - but I'll let the powers that be decide that one.

Regression from a revert of the revert:
commit 7907f45bf9f67a1c5e5d4ae05bab428d7c2f43b2

Author: Ben Widawsky <[email protected]>
Date:   Wed Feb 19 22:05:46 2014 -0800

Revert "drm/i915/bdw: Limit GTT to 2GB"

v2: Change ifdef to 32b, instead of ifndef
update comment

v3. Update comment to not wrap (Daniel).
Update commit message

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=76619

---
 drivers/gpu/drm/i915/i915_gem_gtt.c | 7 +++++++
 1 file changed, 7 insertions(+)
drivers/gpu/drm/i915/i915_gem_gtt.c +0 -0
index 4d87bf2..8d6d531 100644

drivers/gpu/drm/i915/i915_gem_gtt.c static inline unsigned int gen8_get_total_gtt_size(u16 bdw_gmch_ctl) · line 2644

Quoted 5 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +2 -0
@@ -2640,6 +2640,13 @@ static inline unsigned int gen8_get_total_gtt_size(u16 bdw_gmch_ctl)
 	bdw_gmch_ctl &= BDW_GMCH_GGMS_MASK;
 	if (bdw_gmch_ctl)
 		bdw_gmch_ctl = 1 << bdw_gmch_ctl;
+
+#ifdef CONFIG_32BIT

There's no such config for x86.

There's CONFIG_X86_32 though. Which I think matches the original approach of #ifndef CONFIG_64BIT in the referenced bug. However, "with this patch only, the system still hung while booting." [1].

I want a new patch with Tested-by on it.

BR,
Jani.

[1] https://bugs.freedesktop.org/show_bug.cgi?id=76619#c16

Quoted 15 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +5 -0
+	/* Limit 32B platforms to a 2GB GGTT: 4 << 20 / pte size * PAGE_SIZE */
+	if (bdw_gmch_ctl > 4)
+		bdw_gmch_ctl = 4;
+#endif
+
 	return bdw_gmch_ctl << 20;
 }
 
Signature · 6 lines
1.9.2

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 1 line
Jani Nikula, Intel Open Source Technology Center
#7 lore raw
Quoted 46 lines · On Tue, 13 May 2014 10:41:58 +0300 Jani Nikula <[email protected]>…
Quoted 33 lines · Ben Widawsky

Daniel requested in the bug that I use a 3GB fallback size. Since this is not in the spec as a valid size, I decided against it. We could potentially add a patch to bump it to 3GB on top of this one.

This probably should be CC: stable - but I'll let the powers that be decide that one.

Regression from a revert of the revert:
commit 7907f45bf9f67a1c5e5d4ae05bab428d7c2f43b2

Author: Ben Widawsky <[email protected]>
Date:   Wed Feb 19 22:05:46 2014 -0800

Revert "drm/i915/bdw: Limit GTT to 2GB"

v2: Change ifdef to 32b, instead of ifndef
update comment

v3. Update comment to not wrap (Daniel).
Update commit message

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=76619

---
 drivers/gpu/drm/i915/i915_gem_gtt.c | 7 +++++++
 1 file changed, 7 insertions(+)
drivers/gpu/drm/i915/i915_gem_gtt.c +0 -0
index 4d87bf2..8d6d531 100644

drivers/gpu/drm/i915/i915_gem_gtt.c static inline unsigned int gen8_get_total_gtt_size(u16 bdw_gmch_ctl) · line 2644

Quoted 5 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +2 -0
@@ -2640,6 +2640,13 @@ static inline unsigned int gen8_get_total_gtt_size(u16 bdw_gmch_ctl)
 	bdw_gmch_ctl &= BDW_GMCH_GGMS_MASK;
 	if (bdw_gmch_ctl)
 		bdw_gmch_ctl = 1 << bdw_gmch_ctl;
+
+#ifdef CONFIG_32BIT

There's no such config for x86.

There's CONFIG_X86_32 though. Which I think matches the original approach of #ifndef CONFIG_64BIT in the referenced bug. However, "with this patch only, the system still hung while booting." [1].

Quoted 8 lines

I want a new patch with Tested-by on it.

BR,
Jani.

[1] https://bugs.freedesktop.org/show_bug.cgi?id=76619#c16

Ben, will you re-post this or can you find someone else who can?

Signature · 1 line
Jesse Barnes, Intel Open Source Technology Center
#8 lore raw

PATCH 1/1 drm/i915/bdw: Only use 2g GGTT for 32b platforms

From: Ben Widawsky <[email protected]>

Daniel requested in the bug that I use a 3GB fallback size. Since this is not in the spec as a valid size, I decided against it. We could potentially add a patch to bump it to 3GB on top of this one.

This probably should be CC: stable - but I'll let the powers that be decide that one.

Regression from a revert of the revert:
commit 7907f45bf9f67a1c5e5d4ae05bab428d7c2f43b2

Author: Ben Widawsky <[email protected]>
Date:   Wed Feb 19 22:05:46 2014 -0800

Revert "drm/i915/bdw: Limit GTT to 2GB"

v2: Change ifdef to 32b, instead of ifndef
update comment

v3. Update comment to not wrap (Daniel).
Update commit message

v4: s/CONFIG_32/CONFIG_X86_32 (Jani).

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=76619

---
 drivers/gpu/drm/i915/i915_gem_gtt.c | 7 +++++++
 1 file changed, 7 insertions(+)
drivers/gpu/drm/i915/i915_gem_gtt.c +7 -0
index afd4eef..050e47e 100644
@@ -1773,6 +1773,13 @@ static inline unsigned int gen8_get_total_gtt_size(u16 bdw_gmch_ctl)
 	bdw_gmch_ctl &= BDW_GMCH_GGMS_MASK;
 	if (bdw_gmch_ctl)
 		bdw_gmch_ctl = 1 << bdw_gmch_ctl;
+
+#ifdef CONFIG_X86_32BIT
+	/* Limit 32B platforms to a 2GB GGTT: 4 << 20 / pte size * PAGE_SIZE */
+	if (bdw_gmch_ctl > 4)
+		bdw_gmch_ctl = 4;
+#endif
+
 	return bdw_gmch_ctl << 20;
 }
 
Signature · 1 line
1.9.0
#9 lore raw
Quoted 38 lines · Rodrigo Vivi

From: Ben Widawsky <[email protected]>

Daniel requested in the bug that I use a 3GB fallback size. Since this is not in the spec as a valid size, I decided against it. We could potentially add a patch to bump it to 3GB on top of this one.

This probably should be CC: stable - but I'll let the powers that be decide that one.

Regression from a revert of the revert:
commit 7907f45bf9f67a1c5e5d4ae05bab428d7c2f43b2

Author: Ben Widawsky <[email protected]>
Date:   Wed Feb 19 22:05:46 2014 -0800

Revert "drm/i915/bdw: Limit GTT to 2GB"

v2: Change ifdef to 32b, instead of ifndef
update comment

v3. Update comment to not wrap (Daniel).
Update commit message

v4: s/CONFIG_32/CONFIG_X86_32 (Jani).

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=76619

---
 drivers/gpu/drm/i915/i915_gem_gtt.c | 7 +++++++
 1 file changed, 7 insertions(+)
drivers/gpu/drm/i915/i915_gem_gtt.c +0 -0
index afd4eef..050e47e 100644

drivers/gpu/drm/i915/i915_gem_gtt.c static inline unsigned int gen8_get_total_gtt_size(u16 bdw_gmch_ctl) · line 1778

Quoted 6 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +3 -0
@@ -1773,6 +1773,13 @@ static inline unsigned int gen8_get_total_gtt_size(u16 bdw_gmch_ctl)
 	bdw_gmch_ctl &= BDW_GMCH_GGMS_MASK;
 	if (bdw_gmch_ctl)
 		bdw_gmch_ctl = 1 << bdw_gmch_ctl;
+
+#ifdef CONFIG_X86_32BIT
+	/* Limit 32B platforms to a 2GB GGTT: 4 << 20 / pte size * PAGE_SIZE */

We might as well be picky and write 32b not 32B.
-Chris

Signature · 1 line
Chris Wilson, Intel Open Source Technology Centre
#10 lore raw
In reply to Rodrigo Vivi
Quoted 22 lines · Rodrigo Vivi

From: Ben Widawsky <[email protected]>

Daniel requested in the bug that I use a 3GB fallback size. Since this is not in the spec as a valid size, I decided against it. We could potentially add a patch to bump it to 3GB on top of this one.

This probably should be CC: stable - but I'll let the powers that be decide that one.

Regression from a revert of the revert:
commit 7907f45bf9f67a1c5e5d4ae05bab428d7c2f43b2

Author: Ben Widawsky <[email protected]>
Date:   Wed Feb 19 22:05:46 2014 -0800

Revert "drm/i915/bdw: Limit GTT to 2GB"

v2: Change ifdef to 32b, instead of ifndef
update comment

v3. Update comment to not wrap (Daniel).
Update commit message

Quoted 7 lines

v4: s/CONFIG_32/CONFIG_X86_32 (Jani).

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=76619

Do we have a tested-by on this? Iirc it doesn't actually work, or at least it's an incomplete solution apparently ... QA says things still blow up with this patch applied:

https://bugs.freedesktop.org/show_bug.cgi?id=76619#c30

Cheers, Daniel

Quoted 29 lines
---
 drivers/gpu/drm/i915/i915_gem_gtt.c | 7 +++++++
 1 file changed, 7 insertions(+)
drivers/gpu/drm/i915/i915_gem_gtt.c +7 -0
index afd4eef..050e47e 100644
@@ -1773,6 +1773,13 @@ static inline unsigned int gen8_get_total_gtt_size(u16 bdw_gmch_ctl)
 	bdw_gmch_ctl &= BDW_GMCH_GGMS_MASK;
 	if (bdw_gmch_ctl)
 		bdw_gmch_ctl = 1 << bdw_gmch_ctl;
+
+#ifdef CONFIG_X86_32BIT
+	/* Limit 32B platforms to a 2GB GGTT: 4 << 20 / pte size * PAGE_SIZE */
+	if (bdw_gmch_ctl > 4)
+		bdw_gmch_ctl = 4;
+#endif
+
 	return bdw_gmch_ctl << 20;
 }
 
Signature · 6 lines
1.9.0

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 3 lines
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch