Skip to contentSkip to conversation

drm/i915: Add some more bits to CURSOR_POS_MASK

PATCH In review Robert Fekete 4 participants 5 messages Nov 4, 2015 – Nov 25, 2015

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

PATCH 1/1 drm/i915: Add some more bits to CURSOR_POS_MASK

The old value of 0x7FF will wrap the position at 2048 giving wrong coordinate values on panels larger than 2048 pixels in any direction. Used in i915_debugfs atm. Looking at all hw specs available at 01.org shows that X position is bit 0:11, and even 0:12 on some hw where remaining bits up to bit 14 is MBZ. For Y position it is bits 16-27 where bits 28:30 is MBZ. It should be safe to increase CURSOR_POS_MASK to 13 bits (0x1FFF) making 8192 as a new wrap around value still getting valid cursor positions on platforms with only 12bits available thanks to MBZ on adjacent bits above.

---
 drivers/gpu/drm/i915/i915_reg.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/i915/i915_reg.h +1 -1
index 894253228947..f351f46f8cb9 100644
@@ -4883,7 +4883,7 @@ enum skl_disp_power_wells {
 #define   CURSOR_TRICKLE_FEED_DISABLE	(1 << 14)
 #define _CURABASE		0x70084
 #define _CURAPOS		0x70088
-#define   CURSOR_POS_MASK       0x007FF
+#define   CURSOR_POS_MASK       0x01FFF
 #define   CURSOR_POS_SIGN       0x8000
 #define   CURSOR_X_SHIFT        0
 #define   CURSOR_Y_SHIFT        16
Signature · 6 lines
1.9.1

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Reviewed-by on v1 1/1
#1 lore raw
Quoted 9 lines · Robert Fekete

The old value of 0x7FF will wrap the position at 2048 giving wrong coordinate values on panels larger than 2048 pixels in any direction. Used in i915_debugfs atm. Looking at all hw specs available at 01.org shows that X position is bit 0:11, and even 0:12 on some hw where remaining bits up to bit 14 is MBZ. For Y position it is bits 16-27 where bits 28:30 is MBZ. It should be safe to increase CURSOR_POS_MASK to 13 bits (0x1FFF) making 8192 as a new wrap around value still getting valid cursor positions on platforms with only 12bits available thanks to MBZ on adjacent bits above.

I cannot find documentation for older hardware and this only touches debugfs, so in worst case we get wrong values for really old hardware but good ones for newer. I think that's a fair tradeoff.

Quoted 26 lines
---
 drivers/gpu/drm/i915/i915_reg.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/i915/i915_reg.h +1 -1
index 894253228947..f351f46f8cb9 100644
@@ -4883,7 +4883,7 @@ enum skl_disp_power_wells {
 #define   CURSOR_TRICKLE_FEED_DISABLE	(1 << 14)
 #define _CURABASE		0x70084
 #define _CURAPOS		0x70088
-#define   CURSOR_POS_MASK       0x007FF
+#define   CURSOR_POS_MASK       0x01FFF
 #define   CURSOR_POS_SIGN       0x8000
 #define   CURSOR_X_SHIFT        0
 #define   CURSOR_Y_SHIFT        16
Signature · 6 lines
1.9.1

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

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

#2 lore raw
Quoted 11 lines · Patrik Jakobsson
Quoted 9 lines · Robert Fekete

The old value of 0x7FF will wrap the position at 2048 giving wrong coordinate values on panels larger than 2048 pixels in any direction. Used in i915_debugfs atm. Looking at all hw specs available at 01.org shows that X position is bit 0:11, and even 0:12 on some hw where remaining bits up to bit 14 is MBZ. For Y position it is bits 16-27 where bits 28:30 is MBZ. It should be safe to increase CURSOR_POS_MASK to 13 bits (0x1FFF) making 8192 as a new wrap around value still getting valid cursor positions on platforms with only 12bits available thanks to MBZ on adjacent bits above.

Quoted 5 lines

I cannot find documentation for older hardware and this only touches debugfs, so in worst case we get wrong values for really old hardware but good ones for newer. I think that's a fair tradeoff.

If it's only used in debugfs then imo just drop it. Having a _MASK which isn't valid on all platforms, but where we don't have differnt #defines for the different platforms is really confusing.
-Daniel

Quoted 31 lines
Quoted 27 lines
---
 drivers/gpu/drm/i915/i915_reg.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/i915/i915_reg.h +1 -1
index 894253228947..f351f46f8cb9 100644
@@ -4883,7 +4883,7 @@ enum skl_disp_power_wells {
 #define   CURSOR_TRICKLE_FEED_DISABLE	(1 << 14)
 #define _CURABASE		0x70084
 #define _CURAPOS		0x70088
-#define   CURSOR_POS_MASK       0x007FF
+#define   CURSOR_POS_MASK       0x01FFF
 #define   CURSOR_POS_SIGN       0x8000
 #define   CURSOR_X_SHIFT        0
 #define   CURSOR_Y_SHIFT        16
Signature · 6 lines
1.9.1

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

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

Signature · 7 lines
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
#3 lore raw
Quoted 16 lines · Daniel Vetter
Quoted 15 lines · Patrik Jakobsson
Quoted 9 lines · Robert Fekete

The old value of 0x7FF will wrap the position at 2048 giving wrong coordinate values on panels larger than 2048 pixels in any direction. Used in i915_debugfs atm. Looking at all hw specs available at 01.org shows that X position is bit 0:11, and even 0:12 on some hw where remaining bits up to bit 14 is MBZ. For Y position it is bits 16-27 where bits 28:30 is MBZ. It should be safe to increase CURSOR_POS_MASK to 13 bits (0x1FFF) making 8192 as a new wrap around value still getting valid cursor positions on platforms with only 12bits available thanks to MBZ on adjacent bits above.

I cannot find documentation for older hardware and this only touches debugfs, so in worst case we get wrong values for really old hardware but good ones for newer. I think that's a fair tradeoff.

Quoted 6 lines
Quoted 1 line

If it's only used in debugfs then imo just drop it. Having a _MASK which isn't valid on all platforms, but where we don't have differnt #defines for the different platforms is really confusing.
-Daniel

Well, not the most important patch around, but still the value of 0x7ff is still too conservative and gives wrong cursor pos values on large panels. I have hard times digging up really old register specs so I still can't see which Intel platform this new value of mine isn't valid on.

It is this patch by Chris in debugfs that is broken on large panels wrapping coords at (x_pos/y_pos > 2048)
http://marc.info/?l=git-commits-head&m=139697989108096&w=1

Quoted 32 lines
Quoted 31 lines
Quoted 27 lines
---
 drivers/gpu/drm/i915/i915_reg.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/i915/i915_reg.h +1 -1
index 894253228947..f351f46f8cb9 100644
@@ -4883,7 +4883,7 @@ enum skl_disp_power_wells {
 #define   CURSOR_TRICKLE_FEED_DISABLE	(1 << 14)
 #define _CURABASE		0x70084
 #define _CURAPOS		0x70088
-#define   CURSOR_POS_MASK       0x007FF
+#define   CURSOR_POS_MASK       0x01FFF
 #define   CURSOR_POS_SIGN       0x8000
 #define   CURSOR_X_SHIFT        0
 #define   CURSOR_Y_SHIFT        16
Signature · 6 lines
1.9.1

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

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

Signature · 8 lines
BR
/Robert Fekete
Intel Open Source Technology Center

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
#4 lore raw
Quoted 17 lines · Robert Fekete
Quoted 16 lines · Daniel Vetter
Quoted 15 lines · Patrik Jakobsson
Quoted 9 lines · Robert Fekete

The old value of 0x7FF will wrap the position at 2048 giving wrong coordinate values on panels larger than 2048 pixels in any direction. Used in i915_debugfs atm. Looking at all hw specs available at 01.org shows that X position is bit 0:11, and even 0:12 on some hw where remaining bits up to bit 14 is MBZ. For Y position it is bits 16-27 where bits 28:30 is MBZ. It should be safe to increase CURSOR_POS_MASK to 13 bits (0x1FFF) making 8192 as a new wrap around value still getting valid cursor positions on platforms with only 12bits available thanks to MBZ on adjacent bits above.

I cannot find documentation for older hardware and this only touches debugfs, so in worst case we get wrong values for really old hardware but good ones for newer. I think that's a fair tradeoff.

Quoted 6 lines
Quoted 6 lines
Quoted 1 line

If it's only used in debugfs then imo just drop it. Having a _MASK which isn't valid on all platforms, but where we don't have differnt #defines for the different platforms is really confusing.
-Daniel

Yes, seems fair.

Quoted 6 lines

Well, not the most important patch around, but still the value of 0x7ff is still too conservative and gives wrong cursor pos values on large panels. I have hard times digging up really old register specs so I still can't see which Intel platform this new value of mine isn't valid on.

I think Daniel meant that we can remove the define altogether and hardcode a sane mask in debugfs instead. Keeping the define around might give people the wrong idea what the bspec actually says. Also adding a comment along with the hardcoded mask in debugfs would be nice.

-Patrik

Quoted 43 lines

It is this patch by Chris in debugfs that is broken on large panels wrapping coords at (x_pos/y_pos > 2048)
http://marc.info/?l=git-commits-head&m=139697989108096&w=1

Quoted 23 lines
Quoted 23 lines
Quoted 23 lines
---
 drivers/gpu/drm/i915/i915_reg.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/i915/i915_reg.h +1 -2
index 894253228947..f351f46f8cb9 100644
@@ -4883,7 +4883,7 @@ enum skl_disp_power_wells {
 #define   CURSOR_TRICKLE_FEED_DISABLE    (1 << 14)
 #define _CURABASE                0x70084
 #define _CURAPOS         0x70088
-#define   CURSOR_POS_MASK       0x007FF
+#define   CURSOR_POS_MASK       0x01FFF
 #define   CURSOR_POS_SIGN       0x8000
 #define   CURSOR_X_SHIFT        0
 #define   CURSOR_Y_SHIFT        16
--

1.9.1

Quoted 9 lines
Quoted 8 lines
Quoted 4 lines

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

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

--
BR
/Robert Fekete
Intel Open Source Technology Center

Quoted 4 lines

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

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