Haswell watermarks, round 1
Subsystems (2)
cover Haswell watermarks, round 1
From: Paulo Zanoni <[email protected]>
Hi
All our Haswell watermarks code does not follow our specifications, and this series only does some small fixes and also fixes the linetime watermarks. For the other watermarks we're still calling the old sandybridge update_wm functions, so the next patches will implement Haswell-specific functions and use them, also using the linetime watermarks fixed in this series. Consider this series as a warm-up for the next ones :)
Cheers,
Paulo
Paulo Zanoni (9):
drm/i915: ILK, SNB and IVB don't have linetime watermarks drm/i915: fix linetime_watermarks code
drm/i915: use the mode->htotal to calculate linetime watermarks drm/i915: fix haswell linetime watermarks calculation drm/i915: use the correct clock when calculating linetime watermarks drm/i915: make intel_ddi_get_cdclk_freq return values in KHz drm/i915: set the IPS linetime watermark
drm/i915: MCH_SSKPD is a 64 bit register on Haswell drm/i915: set FORCE_ARB_IDLE_PLANES workaround
drivers/gpu/drm/i915/i915_drv.h | 2 - drivers/gpu/drm/i915/i915_reg.h | 3 ++ drivers/gpu/drm/i915/intel_ddi.c | 10 ++--- drivers/gpu/drm/i915/intel_display.c | 4 -- drivers/gpu/drm/i915/intel_dp.c | 3 +- drivers/gpu/drm/i915/intel_drv.h | 2 - drivers/gpu/drm/i915/intel_pm.c | 73 ++++++++++++++++++---------------- 7 files changed, 48 insertions(+), 49 deletions(-)
Signature · 1 line
1.7.10.4
PATCH 1/9 drm/i915: ILK, SNB and IVB don't have linetime watermarks
From: Paulo Zanoni <[email protected]>
So don't call intel_update_linetime_watermarks from ironlake_crtc_mode_set. Only Haswell has these watermarks.
--- drivers/gpu/drm/i915/intel_display.c | 2 -- 1 file changed, 2 deletions(-)
drivers/gpu/drm/i915/intel_display.c +0 -2
@@ -5863,8 +5863,6 @@ static int ironlake_crtc_mode_set(struct drm_crtc *crtc,intel_update_watermarks (dev); -intel_update_linetime_watermarks (dev, pipe, adjusted_mode); -return ret; }
Signature · 1 line
1.7.10.4
Re: [PATCH 1/9] drm/i915: ILK, SNB and IVB don't have linetime watermarks
Quoted 6 lines · Paulo Zanoni
From: Paulo Zanoni <[email protected]>
So don't call
intel_update_linetime_watermarksfromironlake_crtc_mode_set. Only Haswell has these watermarks.Signed-off-by Paulo Zanoni <[email protected]>
Quoted 24 lines
--- drivers/gpu/drm/i915/intel_display.c | 2 -- 1 file changed, 2 deletions(-)drivers/gpu/drm/i915/intel_display.c +0
-2@@ -5863,8 +5863,6 @@ static int ironlake_crtc_mode_set(struct drm_crtc *crtc,intel_update_watermarks (dev); -intel_update_linetime_watermarks (dev, pipe, adjusted_mode); -return ret; }Signature · 6 lines
1.7.10.4 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 2 lines
Ville Syrjälä Intel OTC
PATCH 2/9 drm/i915: fix linetime_watermarks code
In reply to Paulo ZanoniFrom: Paulo Zanoni <[email protected]>
The spec says the linetime watermarks must be programmed before enabling any display low power watermarks, but we're currently updating the linetime watermarks after we call intel_update_watermarks (and only at crtc_mode_set, not at crtc_{enable,disable}). So IMHO the best way guarantee the linetime watermarks will be updated before the low power watermarks is inside the update_wm function, because it's the function that enables low power watermarks. And since Haswell is the only platform that has linetime watermarks, let's completely kill the "intel_update_linetime_watermarks" abstraction and just use the intel_update_watermarks abstraction by creating haswell_update_wm.
For now haswell_update_wm is still calling sandybridge_update_wm, but in the future I plan to implement a function specific to Haswell.
--- drivers/gpu/drm/i915/i915_drv.h | 2 -- drivers/gpu/drm/i915/intel_display.c | 2 -- drivers/gpu/drm/i915/intel_drv.h | 2 -- drivers/gpu/drm/i915/intel_pm.c | 37 ++++++++++++++++++++++------------ 4 files changed, 24 insertions(+), 19 deletions(-)
drivers/gpu/drm/i915/i915_drv.h +0 -2
@@ -316,8 +316,6 @@ struct drm_i915_display_funcs {void (*update_wm)(struct drm_device *dev);void (*update_sprite_wm)(struct drm_device *dev,int pipe,uint32_t sprite_width,int pixel_size); -void (*update_linetime_wm)(struct drm_device *dev,int pipe, -struct drm_display_mode *mode);void (*modeset_global_resources)(struct drm_device *dev);/* Returns the active state of the crtc, and if the crtc is active, * fills out the pipe-config with the hw state. */
drivers/gpu/drm/i915/intel_display.c +0 -2
@@ -6013,8 +6013,6 @@ static int haswell_crtc_mode_set(struct drm_crtc *crtc,intel_update_watermarks (dev); -intel_update_linetime_watermarks (dev, pipe, adjusted_mode); -return ret; }
drivers/gpu/drm/i915/intel_drv.h +0 -2
@@ -732,8 +732,6 @@ extern void intel_update_watermarks(struct drm_device *dev);extern void intel_update_sprite_watermarks (struct drm_device *dev,int pipe,uint32_t sprite_width,int pixel_size); -extern void intel_update_linetime_watermarks (struct drm_device *dev,int pipe, -struct drm_display_mode *mode); extern unsigned long intel_gen4_compute_page_offset(int *x, int *y, unsigned int tiling_mode,
drivers/gpu/drm/i915/intel_pm.c +24 -13
@@ -2015,12 +2015,19 @@ static void ivybridge_update_wm(struct drm_device *dev) }static void -haswell_update_linetime_wm (struct drm_device *dev,int pipe, -struct drm_display_mode *mode) +haswell_update_linetime_wm (struct drm_device *dev,struct drm_crtc *crtc) {struct drm_i915_private *dev_priv = dev->dev_private; +struct intel_crtc *intel_crtc =to_intel_crtc (crtc); +enum pipe pipe = intel_crtc->pipe; +struct drm_display_mode *mode = &intel_crtc->config.adjusted_mode;u32 temp; +if (!intel_crtc_active (crtc)) { +I915_WRITE (PIPE_WM_LINETIME (pipe),0 ); +return ; + } + temp =I915_READ (PIPE_WM_LINETIME (pipe)); temp &= ~PIPE_WM_LINETIME_MASK ;
@@ -2041,6 +2048,20 @@ haswell_update_linetime_wm(struct drm_device *dev, int pipe,I915_WRITE (PIPE_WM_LINETIME (pipe), temp); } +static void haswell_update_wm (struct drm_device *dev) +{ +struct drm_i915_private *dev_priv = dev->dev_private; +struct drm_crtc *crtc; +enum pipe pipe; + +for_each_pipe (pipe) { + crtc = dev_priv->pipe_to_crtc_mapping[pipe]; +haswell_update_linetime_wm (dev, crtc); + } + +sandybridge_update_wm (dev); +} + static bool sandybridge_compute_sprite_wm(struct drm_device *dev, int plane, uint32_t sprite_width, int pixel_size,
@@ -2236,15 +2257,6 @@ void intel_update_watermarks(struct drm_device *dev) dev_priv->display.update_wm (dev); } -void intel_update_linetime_watermarks (struct drm_device *dev, -int pipe,struct drm_display_mode *mode) -{ -struct drm_i915_private *dev_priv = dev->dev_private; - -if (dev_priv->display.update_linetime_wm) - dev_priv->display.update_linetime_wm (dev, pipe, mode); -} -void intel_update_sprite_watermarks (struct drm_device *dev,int pipe,uint32_t sprite_width,int pixel_size) {
@@ -4482,9 +4494,8 @@ void intel_init_pm(struct drm_device *dev) dev_priv->display.init_clock_gating = ivybridge_init_clock_gating; }else if (IS_HASWELL (dev)) {if (SNB_READ_WM0_LATENCY ()) { - dev_priv->display.update_wm = sandybridge_update_wm; + dev_priv->display.update_wm = haswell_update_wm; dev_priv->display.update_sprite_wm = sandybridge_update_sprite_wm; - dev_priv->display.update_linetime_wm = haswell_update_linetime_wm; }else {DRM_DEBUG_KMS ("Failed to read display plane latency. " "Disable CxSR\n" );
Signature · 1 line
1.7.10.4
Re: [PATCH 2/9] drm/i915: fix linetime_watermarks code
Quoted 13 lines · Paulo Zanoni
From: Paulo Zanoni <[email protected]>
The spec says the linetime watermarks must be programmed before enabling any display low power watermarks, but we're currently updating the linetime watermarks after we call
intel_update_watermarks(and only atcrtc_mode_set, not at crtc_{enable,disable}). So IMHO the best way guarantee the linetime watermarks will be updated before the low power watermarks is inside theupdate_wmfunction, because it's the function that enables low power watermarks. And since Haswell is the only platform that has linetime watermarks, let's completely kill the "intel_update_linetime_watermarks" abstraction and just use theintel_update_watermarksabstraction by creatinghaswell_update_wm.
drivers/gpu/drm/i915/intel_pm.c haswell_update_linetime_wm(struct drm_device *dev, int pipe, · line 2060
Quoted 10 lines
drivers/gpu/drm/i915/intel_pm.c +10
-0+static void haswell_update_wm (struct drm_device *dev) +{ +struct drm_i915_private *dev_priv = dev->dev_private; +struct drm_crtc *crtc; +enum pipe pipe; + +for_each_pipe (pipe) { + crtc = dev_priv->pipe_to_crtc_mapping[pipe]; +haswell_update_linetime_wm (dev, crtc); + }
The LP watermarks are still enabled at this point.
drivers/gpu/drm/i915/intel_pm.c haswell_update_linetime_wm(struct drm_device *dev, int pipe, · line 2063
Quoted 3 lines
drivers/gpu/drm/i915/intel_pm.c +3
-0+ +sandybridge_update_wm (dev); +}
-Chris
Signature · 1 line
Chris Wilson, Intel Open Source Technology Centre
Re: [PATCH 2/9] drm/i915: fix linetime_watermarks code
2013/5/5 Chris Wilson <[email protected]>:
Quoted 14 lines · Paulo Zanoni
Quoted 12 lines
From: Paulo Zanoni <[email protected]>
The spec says the linetime watermarks must be programmed before enabling any display low power watermarks, but we're currently updating the linetime watermarks after we call
intel_update_watermarks(and only atcrtc_mode_set, not at crtc_{enable,disable}). So IMHO the best way guarantee the linetime watermarks will be updated before the low power watermarks is inside theupdate_wmfunction, because it's the function that enables low power watermarks. And since Haswell is the only platform that has linetime watermarks, let's completely kill the "intel_update_linetime_watermarks" abstraction and just use theintel_update_watermarksabstraction by creatinghaswell_update_wm.
Quoted 12 lines
drivers/gpu/drm/i915/intel_pm.c haswell_update_linetime_wm(struct drm_device *dev, int pipe, · line 2060
Quoted 10 lines
drivers/gpu/drm/i915/intel_pm.c +10
-0+static void haswell_update_wm (struct drm_device *dev) +{ +struct drm_i915_private *dev_priv = dev->dev_private; +struct drm_crtc *crtc; +enum pipe pipe; + +for_each_pipe (pipe) { + crtc = dev_priv->pipe_to_crtc_mapping[pipe]; +haswell_update_linetime_wm (dev, crtc); + }The LP watermarks are still enabled at this point.
Did you mean "they're not disabled yet"? Well, at least now we're programming the linetime registers _before_ we update the LP registers, and every time. If we want to be 100% spec-compliant without any intermediate steps then I guess we can discard this linetime series and merge it all inside the single patch that will fully implement haswell_update_wm. But it will be one single big patch. Just tell me which one you prefer.
Quoted 7 lines
drivers/gpu/drm/i915/intel_pm.c haswell_update_linetime_wm(struct drm_device *dev, int pipe, · line 2063
Quoted 3 lines
drivers/gpu/drm/i915/intel_pm.c +3
-0+ +sandybridge_update_wm (dev); +}-Chris
--
Chris Wilson, Intel Open Source Technology Centre
Signature · 1 line
Paulo Zanoni
Re: [PATCH 2/9] drm/i915: fix linetime_watermarks code
Quoted 28 lines · Paulo Zanoni
2013/5/5 Chris Wilson <[email protected]>:
Quoted 14 lines · Paulo Zanoni
Quoted 12 lines
From: Paulo Zanoni <[email protected]>
The spec says the linetime watermarks must be programmed before enabling any display low power watermarks, but we're currently updating the linetime watermarks after we call
intel_update_watermarks(and only atcrtc_mode_set, not at crtc_{enable,disable}). So IMHO the best way guarantee the linetime watermarks will be updated before the low power watermarks is inside theupdate_wmfunction, because it's the function that enables low power watermarks. And since Haswell is the only platform that has linetime watermarks, let's completely kill the "intel_update_linetime_watermarks" abstraction and just use theintel_update_watermarksabstraction by creatinghaswell_update_wm.Quoted 12 lines
drivers/gpu/drm/i915/intel_pm.c haswell_update_linetime_wm(struct drm_device *dev, int pipe, · line 2060
Quoted 10 lines
drivers/gpu/drm/i915/intel_pm.c +10
-0+static void haswell_update_wm (struct drm_device *dev) +{ +struct drm_i915_private *dev_priv = dev->dev_private; +struct drm_crtc *crtc; +enum pipe pipe; + +for_each_pipe (pipe) { + crtc = dev_priv->pipe_to_crtc_mapping[pipe]; +haswell_update_linetime_wm (dev, crtc); + }The LP watermarks are still enabled at this point.
Quoted 7 lines
Did you mean "they're not disabled yet"? Well, at least now we're programming the linetime registers _before_ we update the LP registers, and every time. If we want to be 100% spec-compliant without any intermediate steps then I guess we can discard this linetime series and merge it all inside the single patch that will fully implement
haswell_update_wm. But it will be one single big patch. Just tell me which one you prefer.
Or you can just disable the 3 LP wm first before updating the linetime wm.
-Chris
Signature · 1 line
Chris Wilson, Intel Open Source Technology Centre
PATCH 2/9 drm/i915: remove intel_update_linetime_watermarks
From: Paulo Zanoni <[email protected]>
The spec says the linetime watermarks must be programmed before enabling any display low power watermarks, but we're currently updating the linetime watermarks after we call intel_update_watermarks (and only at crtc_mode_set, not at crtc_{enable,disable}). So IMHO the best way guarantee the linetime watermarks will be updated before the low power watermarks is inside the update_wm function, because it's the function that enables low power watermarks. And since Haswell is the only platform that has linetime watermarks, let's completely kill the "intel_update_linetime_watermarks" abstraction and just use the intel_update_watermarks abstraction by creating haswell_update_wm.
For now haswell_update_wm is still calling sandybridge_update_wm, but in the future I plan to implement a function specific to Haswell.
v2: - Rename patch
- Disable LP watermarks before changing linetime WMs (Chris)
- Add a comment explaining that this is just temporary code.
--- drivers/gpu/drm/i915/i915_drv.h | 2 -- drivers/gpu/drm/i915/intel_display.c | 2 -- drivers/gpu/drm/i915/intel_drv.h | 2 -- drivers/gpu/drm/i915/intel_pm.c | 43 ++++++++++++++++++++++++---------- 4 files changed, 30 insertions(+), 19 deletions(-)
drivers/gpu/drm/i915/i915_drv.h +0 -2
@@ -316,8 +316,6 @@ struct drm_i915_display_funcs {void (*update_wm)(struct drm_device *dev);void (*update_sprite_wm)(struct drm_device *dev,int pipe,uint32_t sprite_width,int pixel_size); -void (*update_linetime_wm)(struct drm_device *dev,int pipe, -struct drm_display_mode *mode);void (*modeset_global_resources)(struct drm_device *dev);/* Returns the active state of the crtc, and if the crtc is active, * fills out the pipe-config with the hw state. */
drivers/gpu/drm/i915/intel_display.c +0 -2
@@ -6026,8 +6026,6 @@ static int haswell_crtc_mode_set(struct drm_crtc *crtc,intel_update_watermarks (dev); -intel_update_linetime_watermarks (dev, pipe, adjusted_mode); -return ret; }
drivers/gpu/drm/i915/intel_drv.h +0 -2
@@ -727,8 +727,6 @@ extern void intel_update_watermarks(struct drm_device *dev);extern void intel_update_sprite_watermarks (struct drm_device *dev,int pipe,uint32_t sprite_width,int pixel_size); -extern void intel_update_linetime_watermarks (struct drm_device *dev,int pipe, -struct drm_display_mode *mode); extern unsigned long intel_gen4_compute_page_offset(int *x, int *y, unsigned int tiling_mode,
drivers/gpu/drm/i915/intel_pm.c +30 -13
72 lines not on this page · raw message
Signature · 1 line
1.7.10.4
Re: [PATCH 2/9] drm/i915: remove intel_update_linetime_watermarks
Quoted 101 lines · Paulo Zanoni
From: Paulo Zanoni <[email protected]>
The spec says the linetime watermarks must be programmed before enabling any display low power watermarks, but we're currently updating the linetime watermarks after we call
intel_update_watermarks(and only atcrtc_mode_set, not at crtc_{enable,disable}). So IMHO the best way guarantee the linetime watermarks will be updated before the low power watermarks is inside theupdate_wmfunction, because it's the function that enables low power watermarks. And since Haswell is the only platform that has linetime watermarks, let's completely kill the "intel_update_linetime_watermarks" abstraction and just use theintel_update_watermarksabstraction by creatinghaswell_update_wm.For now
haswell_update_wmis still callingsandybridge_update_wm, but in the future I plan to implement a function specific to Haswell.v2: - Rename patch
- Disable LP watermarks before changing linetime WMs (Chris)
- Add a comment explaining that this is just temporary code.Signed-off-by Paulo Zanoni <[email protected]>--- drivers/gpu/drm/i915/i915_drv.h | 2 -- drivers/gpu/drm/i915/intel_display.c | 2 -- drivers/gpu/drm/i915/intel_drv.h | 2 -- drivers/gpu/drm/i915/intel_pm.c | 43 ++++++++++++++++++++++++---------- 4 files changed, 30 insertions(+), 19 deletions(-)drivers/gpu/drm/i915/i915_drv.h +0
-2@@ -316,8 +316,6 @@ struct drm_i915_display_funcs {void (*update_wm)(struct drm_device *dev);void (*update_sprite_wm)(struct drm_device *dev,int pipe,uint32_t sprite_width,int pixel_size); -void (*update_linetime_wm)(struct drm_device *dev,int pipe, -struct drm_display_mode *mode);void (*modeset_global_resources)(struct drm_device *dev);/* Returns the active state of the crtc, and if the crtc is active, * fills out the pipe-config with the hw state. */ drivers/gpu/drm/i915/intel_display.c +0
-2@@ -6026,8 +6026,6 @@ static int haswell_crtc_mode_set(struct drm_crtc *crtc,intel_update_watermarks (dev); -intel_update_linetime_watermarks (dev, pipe, adjusted_mode); -return ret; }drivers/gpu/drm/i915/intel_drv.h +0
-2@@ -727,8 +727,6 @@ extern void intel_update_watermarks(struct drm_device *dev);extern void intel_update_sprite_watermarks (struct drm_device *dev,int pipe,uint32_t sprite_width,int pixel_size); -extern void intel_update_linetime_watermarks (struct drm_device *dev,int pipe, -struct drm_display_mode *mode); extern unsigned long intel_gen4_compute_page_offset(int *x, int *y, unsigned int tiling_mode,drivers/gpu/drm/i915/intel_pm.c +13
-2@@ -2069,12 +2069,19 @@ static void ivybridge_update_wm(struct drm_device *dev) }static void -haswell_update_linetime_wm (struct drm_device *dev,int pipe, -struct drm_display_mode *mode) +haswell_update_linetime_wm (struct drm_device *dev,struct drm_crtc *crtc) {struct drm_i915_private *dev_priv = dev->dev_private; +struct intel_crtc *intel_crtc =to_intel_crtc (crtc); +enum pipe pipe = intel_crtc->pipe; +struct drm_display_mode *mode = &intel_crtc->config.adjusted_mode;u32 temp; +if (!intel_crtc_active (crtc)) { +I915_WRITE (PIPE_WM_LINETIME (pipe),0 ); +return ; + } + temp =I915_READ (PIPE_WM_LINETIME (pipe)); temp &= ~PIPE_WM_LINETIME_MASK ;@@ -2095,6 +2102,26 @@ haswell_update_linetime_wm(struct drm_device *dev, int pipe,I915_WRITE (PIPE_WM_LINETIME (pipe), temp); } +static void haswell_update_wm (struct drm_device *dev) +{ +struct drm_i915_private *dev_priv = dev->dev_private; +struct drm_crtc *crtc;
drivers/gpu/drm/i915/intel_pm.c haswell_update_linetime_wm(struct drm_device *dev, int pipe, · line 2118
Quoted 10 lines
drivers/gpu/drm/i915/intel_pm.c +10
-0+enum pipe pipe; + +/* Disable the LP WMs before changine the linetime registers. This is +* just a temporary code that will be replaced soon. */ +I915_WRITE (WM3_LP_ILK ,0 ); +I915_WRITE (WM2_LP_ILK ,0 ); +I915_WRITE (WM1_LP_ILK ,0 ); + +for_each_pipe (pipe) { + crtc = dev_priv->pipe_to_crtc_mapping[pipe];
list_for_each_entry(crtc, ...)
Otherwise:
Quoted 43 lines
drivers/gpu/drm/i915/intel_pm.c +7
-11+haswell_update_linetime_wm (dev, crtc); + } + +sandybridge_update_wm (dev); +} + static bool sandybridge_compute_sprite_wm(struct drm_device *dev, int plane, uint32_t sprite_width, int pixel_size,@@ -2290,15 +2317,6 @@ void intel_update_watermarks(struct drm_device *dev) dev_priv->display.update_wm (dev); } -void intel_update_linetime_watermarks (struct drm_device *dev, -int pipe,struct drm_display_mode *mode) -{ -struct drm_i915_private *dev_priv = dev->dev_private; - -if (dev_priv->display.update_linetime_wm) - dev_priv->display.update_linetime_wm (dev, pipe, mode); -} -void intel_update_sprite_watermarks (struct drm_device *dev,int pipe,uint32_t sprite_width,int pixel_size) {@@ -4553,9 +4571,8 @@ void intel_init_pm(struct drm_device *dev) dev_priv->display.init_clock_gating = ivybridge_init_clock_gating; }else if (IS_HASWELL (dev)) {if (SNB_READ_WM0_LATENCY ()) { - dev_priv->display.update_wm = sandybridge_update_wm; + dev_priv->display.update_wm = haswell_update_wm; dev_priv->display.update_sprite_wm = sandybridge_update_sprite_wm; - dev_priv->display.update_linetime_wm = haswell_update_linetime_wm; }else {DRM_DEBUG_KMS ("Failed to read display plane latency. " "Disable CxSR\n" );Signature · 6 lines
1.7.10.4 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 2 lines
Ville Syrjälä Intel OTC
PATCH 3/9 drm/i915: use the mode->htotal to calculate linetime watermarks
In reply to Paulo ZanoniFrom: Paulo Zanoni <[email protected]>
... instead of mode->crtc_display. The spec says "pipe horizontal total number of pixels" and the "Haswell Watermark Calculator" tool uses the "Pipe H Total" instead of "Pipe H Src" as the value.
--- drivers/gpu/drm/i915/intel_pm.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/i915/intel_pm.c +1 -1
@@ -2035,7 +2035,7 @@ haswell_update_linetime_wm(struct drm_device *dev, struct drm_crtc *crtc)* row at the given clock rate, multiplied by 8. * */ temp |=PIPE_WM_LINETIME_TIME ( - ((mode->crtc_hdisplay *1000 ) / mode->clock) *8 ); + ((mode->htotal *1000 ) / mode->clock) *8 );/* IPS watermarks are only used by pipe A, and are ignored by * pipes B and C. They are calculated similarly to the common
Signature · 1 line
1.7.10.4
Re: [PATCH 3/9] drm/i915: use the mode->htotal to calculate linetime watermarks
Quoted 7 lines · Paulo Zanoni
From: Paulo Zanoni <[email protected]>
... instead of
mode->crtc_display. The spec says "pipe horizontal total number of pixels" and the "Haswell Watermark Calculator" tool uses the "Pipe H Total" instead of "Pipe H Src" as the value.Signed-off-by Paulo Zanoni <[email protected]>
Quoted 24 lines
--- drivers/gpu/drm/i915/intel_pm.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)drivers/gpu/drm/i915/intel_pm.c +1
-1@@ -2035,7 +2035,7 @@ haswell_update_linetime_wm(struct drm_device *dev, struct drm_crtc *crtc)* row at the given clock rate, multiplied by 8. * */ temp |=PIPE_WM_LINETIME_TIME ( - ((mode->crtc_hdisplay *1000 ) / mode->clock) *8 ); + ((mode->htotal *1000 ) / mode->clock) *8 );/* IPS watermarks are only used by pipe A, and are ignored by * pipes B and C. They are calculated similarly to the common Signature · 6 lines
1.7.10.4 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 2 lines
Ville Syrjälä Intel OTC
Re: [PATCH 3/9] drm/i915: use the mode->htotal to calculate linetime watermarks
In reply to Paulo ZanoniQuoted 16 lines · Paulo Zanoni
From: Paulo Zanoni <[email protected]>
... instead of
mode->crtc_display. The spec says "pipe horizontal total number of pixels" and the "Haswell Watermark Calculator" tool uses the "Pipe H Total" instead of "Pipe H Src" as the value.Signed-off-by Paulo Zanoni <[email protected]>--- drivers/gpu/drm/i915/intel_pm.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)drivers/gpu/drm/i915/intel_pm.c +0
-0
drivers/gpu/drm/i915/intel_pm.c haswell_update_linetime_wm(struct drm_device *dev, struct drm_crtc *crtc) · line 2038
Quoted 5 lines
drivers/gpu/drm/i915/intel_pm.c +1
-1@@ -2035,7 +2035,7 @@ haswell_update_linetime_wm(struct drm_device *dev, struct drm_crtc *crtc)* row at the given clock rate, multiplied by 8. * */ temp |=PIPE_WM_LINETIME_TIME ( - ((mode->crtc_hdisplay *1000 ) / mode->clock) *8 ); + ((mode->htotal *1000 ) / mode->clock) *8 );
Big question: What's the right value for interlaced modes? On progressive mode crtc_ mode values match the non-crtc_ prefixed ones, but not so much for interlaced modes ... So if your tool expects something resembling what we program into the pipe registers, we need to change this again.
Merged for now since the spec explicitly says "pixels", but I'd like someone to double-check this.
-Daniel
Quoted 10 lines
drivers/gpu/drm/i915/intel_pm.c +0
-0/* IPS watermarks are only used by pipe A, and are ignored by * pipes B and C. They are calculated similarly to the common Signature · 6 lines
1.7.10.4 _______________________________________________ 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
Re: [PATCH 3/9] drm/i915: use the mode->htotal to calculate linetime watermarks
2013/5/21 Daniel Vetter <[email protected]>:
Quoted 28 lines · Paulo Zanoni
Quoted 16 lines
From: Paulo Zanoni <[email protected]>
... instead of
mode->crtc_display. The spec says "pipe horizontal total number of pixels" and the "Haswell Watermark Calculator" tool uses the "Pipe H Total" instead of "Pipe H Src" as the value.Signed-off-by Paulo Zanoni <[email protected]>--- drivers/gpu/drm/i915/intel_pm.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)drivers/gpu/drm/i915/intel_pm.c +0
-0drivers/gpu/drm/i915/intel_pm.c haswell_update_linetime_wm(struct drm_device *dev, struct drm_crtc *crtc) · line 2038
Quoted 5 lines
drivers/gpu/drm/i915/intel_pm.c +1
-1@@ -2035,7 +2035,7 @@ haswell_update_linetime_wm(struct drm_device *dev, struct drm_crtc *crtc)* row at the given clock rate, multiplied by 8. * */ temp |=PIPE_WM_LINETIME_TIME ( - ((mode->crtc_hdisplay *1000 ) / mode->clock) *8 ); + ((mode->htotal *1000 ) / mode->clock) *8 );Big question: What's the right value for interlaced modes? On progressive mode
crtc_mode values match the non-crtc_ prefixed ones, but not so much for interlaced modes ... So if your tool expects something resembling what we program into the pipe registers, we need to change this again.
Quoted 2 lines
Merged for now since the spec explicitly says "pixels", but I'd like someone to double-check this.
The WM calculator tool says:
"Pipe H total is found in PIPE_HTOTAL horizontal field".
But mode->htotal and mode->crtc_htotal are the same even when the mode is interlaced. So at least the values we compute are correct. I could write a follow-up patch updating this if you want.
Quoted 16 lines
-Daniel
Quoted 10 lines
drivers/gpu/drm/i915/intel_pm.c +0
-1/* IPS watermarks are only used by pipe A, and are ignored by * pipes B and C. They are calculated similarly to the common -- 1.7.10.4
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx--
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch
Signature · 1 line
Paulo Zanoni
Re: [PATCH 3/9] drm/i915: use the mode->htotal to calculate linetime watermarks
Quoted 32 lines · Paulo Zanoni
2013/5/21 Daniel Vetter <[email protected]>:
Quoted 28 lines · Paulo Zanoni
Quoted 16 lines
From: Paulo Zanoni <[email protected]>
... instead of
mode->crtc_display. The spec says "pipe horizontal total number of pixels" and the "Haswell Watermark Calculator" tool uses the "Pipe H Total" instead of "Pipe H Src" as the value.Signed-off-by Paulo Zanoni <[email protected]>--- drivers/gpu/drm/i915/intel_pm.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)drivers/gpu/drm/i915/intel_pm.c +0
-0drivers/gpu/drm/i915/intel_pm.c haswell_update_linetime_wm(struct drm_device *dev, struct drm_crtc *crtc) · line 2038
Quoted 5 lines
drivers/gpu/drm/i915/intel_pm.c +1
-1@@ -2035,7 +2035,7 @@ haswell_update_linetime_wm(struct drm_device *dev, struct drm_crtc *crtc)* row at the given clock rate, multiplied by 8. * */ temp |=PIPE_WM_LINETIME_TIME ( - ((mode->crtc_hdisplay *1000 ) / mode->clock) *8 ); + ((mode->htotal *1000 ) / mode->clock) *8 );Big question: What's the right value for interlaced modes? On progressive mode
crtc_mode values match the non-crtc_ prefixed ones, but not so much for interlaced modes ... So if your tool expects something resembling what we program into the pipe registers, we need to change this again.Quoted 2 lines
Merged for now since the spec explicitly says "pixels", but I'd like someone to double-check this.
Quoted 6 lines
The WM calculator tool says:
"Pipe H total is found inPIPE_HTOTALhorizontal field".But
mode->htotalandmode->crtc_htotalare the same even when the mode is interlaced. So at least the values we compute are correct. I could write a follow-up patch updating this if you want.
Oh, I've mixed up the horizontal timings with the vertical ones. You're right, everything is working correctly as-is. -Daniel
--
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch
PATCH 4/9 drm/i915: fix haswell linetime watermarks calculation
In reply to Paulo ZanoniFrom: Paulo Zanoni <[email protected]>
Move the "*8" calculation to the left side so we don't propagate rounding errors. Also use DIV_ROUND_CLOSEST because that's what the spec says we need to do.
--- drivers/gpu/drm/i915/intel_pm.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/i915/intel_pm.c +1 -1
@@ -2035,7 +2035,7 @@ haswell_update_linetime_wm(struct drm_device *dev, struct drm_crtc *crtc)* row at the given clock rate, multiplied by 8. * */ temp |=PIPE_WM_LINETIME_TIME ( - ((mode->htotal *1000 ) / mode->clock) *8 ); +DIV_ROUND_CLOSEST (mode->htotal *1000 *8 , mode->clock));/* IPS watermarks are only used by pipe A, and are ignored by * pipes B and C. They are calculated similarly to the common
Signature · 1 line
1.7.10.4
Re: [PATCH 4/9] drm/i915: fix haswell linetime watermarks calculation
Quoted 7 lines · Paulo Zanoni
From: Paulo Zanoni <[email protected]>
Move the "*8" calculation to the left side so we don't propagate rounding errors. Also use
DIV_ROUND_CLOSESTbecause that's what the spec says we need to do.Signed-off-by Paulo Zanoni <[email protected]>
Quoted 24 lines
--- drivers/gpu/drm/i915/intel_pm.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)drivers/gpu/drm/i915/intel_pm.c +1
-1@@ -2035,7 +2035,7 @@ haswell_update_linetime_wm(struct drm_device *dev, struct drm_crtc *crtc)* row at the given clock rate, multiplied by 8. * */ temp |=PIPE_WM_LINETIME_TIME ( - ((mode->htotal *1000 ) / mode->clock) *8 ); +DIV_ROUND_CLOSEST (mode->htotal *1000 *8 , mode->clock));/* IPS watermarks are only used by pipe A, and are ignored by * pipes B and C. They are calculated similarly to the common Signature · 6 lines
1.7.10.4 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 2 lines
Ville Syrjälä Intel OTC
PATCH 5/9 drm/i915: use the correct clock when calculating linetime watermarks
In reply to Paulo ZanoniFrom: Paulo Zanoni <[email protected]>
If we're using DP/eDP, adjusted_mode->clock may be just the port link clock, but we also can't use mode->clock because it's wrong when we're using the using panel fitter.
--- drivers/gpu/drm/i915/intel_pm.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-)
drivers/gpu/drm/i915/intel_pm.c +7 -1
@@ -2021,6 +2021,7 @@ haswell_update_linetime_wm(struct drm_device *dev, struct drm_crtc *crtc)struct intel_crtc *intel_crtc =to_intel_crtc (crtc);enum pipe pipe = intel_crtc->pipe;struct drm_display_mode *mode = &intel_crtc->config.adjusted_mode; +int target_clock;u32 temp;if (!intel_crtc_active (crtc)) {
@@ -2028,6 +2029,11 @@ haswell_update_linetime_wm(struct drm_device *dev, struct drm_crtc *crtc)return ; } +if (intel_crtc->config.pixel_target_clock) + target_clock = intel_crtc->config.pixel_target_clock; +else + target_clock = intel_crtc->config.adjusted_mode.clock; + temp =I915_READ (PIPE_WM_LINETIME (pipe)); temp &= ~PIPE_WM_LINETIME_MASK ;
@@ -2035,7 +2041,7 @@ haswell_update_linetime_wm(struct drm_device *dev, struct drm_crtc *crtc)* row at the given clock rate, multiplied by 8. * */ temp |=PIPE_WM_LINETIME_TIME ( -DIV_ROUND_CLOSEST (mode->htotal *1000 *8 , mode->clock)); +DIV_ROUND_CLOSEST (mode->htotal *1000 *8 , target_clock));/* IPS watermarks are only used by pipe A, and are ignored by * pipes B and C. They are calculated similarly to the common
Signature · 1 line
1.7.10.4
Re: [PATCH 5/9] drm/i915: use the correct clock when calculating linetime watermarks
Quoted 7 lines · Paulo Zanoni
From: Paulo Zanoni <[email protected]>
If we're using DP/eDP,
adjusted_mode->clockmay be just the port link clock, but we also can't usemode->clockbecause it's wrong when we're using the using panel fitter.Signed-off-by Paulo Zanoni <[email protected]>
Quoted 44 lines
--- drivers/gpu/drm/i915/intel_pm.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-)drivers/gpu/drm/i915/intel_pm.c +7
-1@@ -2021,6 +2021,7 @@ haswell_update_linetime_wm(struct drm_device *dev, struct drm_crtc *crtc)struct intel_crtc *intel_crtc =to_intel_crtc (crtc);enum pipe pipe = intel_crtc->pipe;struct drm_display_mode *mode = &intel_crtc->config.adjusted_mode; +int target_clock;u32 temp;if (!intel_crtc_active (crtc)) {@@ -2028,6 +2029,11 @@ haswell_update_linetime_wm(struct drm_device *dev, struct drm_crtc *crtc)return ; } +if (intel_crtc->config.pixel_target_clock) + target_clock = intel_crtc->config.pixel_target_clock; +else + target_clock = intel_crtc->config.adjusted_mode.clock; + temp =I915_READ (PIPE_WM_LINETIME (pipe)); temp &= ~PIPE_WM_LINETIME_MASK ;@@ -2035,7 +2041,7 @@ haswell_update_linetime_wm(struct drm_device *dev, struct drm_crtc *crtc)* row at the given clock rate, multiplied by 8. * */ temp |=PIPE_WM_LINETIME_TIME ( -DIV_ROUND_CLOSEST (mode->htotal *1000 *8 , mode->clock)); +DIV_ROUND_CLOSEST (mode->htotal *1000 *8 , target_clock));/* IPS watermarks are only used by pipe A, and are ignored by * pipes B and C. They are calculated similarly to the common Signature · 6 lines
1.7.10.4 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 2 lines
Ville Syrjälä Intel OTC
Re: [PATCH 5/9] drm/i915: use the correct clock when calculating linetime watermarks
In reply to Paulo ZanoniQuoted 24 lines · Paulo Zanoni
From: Paulo Zanoni <[email protected]>
If we're using DP/eDP,
adjusted_mode->clockmay be just the port link clock, but we also can't usemode->clockbecause it's wrong when we're using the using panel fitter.Signed-off-by Paulo Zanoni <[email protected]>--- drivers/gpu/drm/i915/intel_pm.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-)drivers/gpu/drm/i915/intel_pm.c +1
-0@@ -2021,6 +2021,7 @@ haswell_update_linetime_wm(struct drm_device *dev, struct drm_crtc *crtc)struct intel_crtc *intel_crtc =to_intel_crtc (crtc);enum pipe pipe = intel_crtc->pipe;struct drm_display_mode *mode = &intel_crtc->config.adjusted_mode; +int target_clock;u32 temp;if (!intel_crtc_active (crtc)) {
drivers/gpu/drm/i915/intel_pm.c haswell_update_linetime_wm(struct drm_device *dev, struct drm_crtc *crtc) · line 2035
Quoted 7 lines
drivers/gpu/drm/i915/intel_pm.c +4
-0@@ -2028,6 +2029,11 @@ haswell_update_linetime_wm(struct drm_device *dev, struct drm_crtc *crtc)return ; } +if (intel_crtc->config.pixel_target_clock) + target_clock = intel_crtc->config.pixel_target_clock; +else + target_clock = intel_crtc->config.adjusted_mode.clock;
I'll ignore this one here, since I already have the real fix at
I'll submit the series this patch is a part of asap to intel-gfx.
Cheers, Daniel
Quoted 20 lines
drivers/gpu/drm/i915/intel_pm.c +2
-1+ temp =I915_READ (PIPE_WM_LINETIME (pipe)); temp &= ~PIPE_WM_LINETIME_MASK ;@@ -2035,7 +2041,7 @@ haswell_update_linetime_wm(struct drm_device *dev, struct drm_crtc *crtc)* row at the given clock rate, multiplied by 8. * */ temp |=PIPE_WM_LINETIME_TIME ( -DIV_ROUND_CLOSEST (mode->htotal *1000 *8 , mode->clock)); +DIV_ROUND_CLOSEST (mode->htotal *1000 *8 , target_clock));/* IPS watermarks are only used by pipe A, and are ignored by * pipes B and C. They are calculated similarly to the common Signature · 6 lines
1.7.10.4 _______________________________________________ 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
PATCH 6/9 drm/i915: make intel_ddi_get_cdclk_freq return values in KHz
In reply to Paulo ZanoniFrom: Paulo Zanoni <[email protected]>
With this, that 338 can finally become the correct 337500.
Due to the change we need to adjust the intel_dp_aux_ch function to set the correct value, so adjust the division and also use DIV_ROUND_CLOSEST instead of the old "round down" behavior because the spec says the value "should be programmed to get as close as possible to the ideal rate of 2MHz".
--- drivers/gpu/drm/i915/intel_ddi.c | 10 +++++----- drivers/gpu/drm/i915/intel_dp.c | 3 ++- 2 files changed, 7 insertions(+), 6 deletions(-)
drivers/gpu/drm/i915/intel_ddi.c +5 -5
@@ -1335,14 +1335,14 @@ static void intel_disable_ddi(struct intel_encoder *intel_encoder)int intel_ddi_get_cdclk_freq (struct drm_i915_private *dev_priv) {if (I915_READ (HSW_FUSE_STRAP ) &HSW_CDCLK_LIMIT ) -return 450 ; +return 450000 ;else if ((I915_READ (LCPLL_CTL ) &LCPLL_CLK_FREQ_MASK ) ==LCPLL_CLK_FREQ_450 ) -return 450 ; +return 450000 ;else if (IS_ULT (dev_priv->dev)) -return 338 ; +return 337500 ;else -return 540 ; +return 540000 ; }void intel_ddi_pll_init (struct drm_device *dev)
@@ -1355,7 +1355,7 @@ void intel_ddi_pll_init(struct drm_device *dev)* Don't even try to turn it on. */ -DRM_DEBUG_KMS ("CDCLK running at %dMHz\n" , +DRM_DEBUG_KMS ("CDCLK running at %dKHz\n" ,intel_ddi_get_cdclk_freq (dev_priv));if (val &LCPLL_CD_SOURCE_FCLK )
drivers/gpu/drm/i915/intel_dp.c +2 -1
@@ -346,7 +346,8 @@ intel_dp_aux_ch(struct intel_dp *intel_dp,*/ if (is_cpu_edp (intel_dp)) {if (HAS_DDI (dev)) - aux_clock_divider =intel_ddi_get_cdclk_freq (dev_priv) >>1 ; + aux_clock_divider =DIV_ROUND_CLOSEST ( +intel_ddi_get_cdclk_freq (dev_priv),2000 );else if (IS_VALLEYVIEW (dev)) aux_clock_divider =100 ;else if (IS_GEN6 (dev) ||IS_GEN7 (dev))
Signature · 1 line
1.7.10.4
Re: [PATCH 6/9] drm/i915: make intel_ddi_get_cdclk_freq return values in KHz
Quoted 9 lines · Paulo Zanoni
From: Paulo Zanoni <[email protected]>
With this, that 338 can finally become the correct 337500.
Due to the change we need to adjust the
intel_dp_aux_chfunction to set the correct value, so adjust the division and also useDIV_ROUND_CLOSESTinstead of the old "round down" behavior because the spec says the value "should be programmed to get as close as possible to the ideal rate of 2MHz".
Can you please demonstrate an instance where this code produces a different value? And only then correct the constants. -Chris
Signature · 1 line
Chris Wilson, Intel Open Source Technology Centre
Re: [PATCH 6/9] drm/i915: make intel_ddi_get_cdclk_freq return values in KHz
2013/5/5 Chris Wilson <[email protected]>:
Quoted 5 lines · Paulo Zanoni
Quoted 4 lines
From: Paulo Zanoni <[email protected]>
With this, that 338 can finally become the correct 337500.
Quoted 8 lines
Quoted 5 lines
Due to the change we need to adjust the
intel_dp_aux_chfunction to set the correct value, so adjust the division and also useDIV_ROUND_CLOSESTinstead of the old "round down" behavior because the spec says the value "should be programmed to get as close as possible to the ideal rate of 2MHz".Can you please demonstrate an instance where this code produces a different value? And only then correct the constants.
I use the 337500 value on the next patch, when setting the ips_linetime value. The correct frequency is 337500, not 338000.
ips_linetime = DIV_ROUND_CLOSEST(mode->htotal * 1000 * 8, intel_ddi_get_cdclk_freq);
For a mode with htotal of 2640 [0] we'll have: (i) (2640 * 1000 * 8) / 338000 = 62.48, resulting in 62 and (ii) (2640 * 1000 * 8) / 337500 = 62.57 resulting in 63.
For the case inside intel_dp.c:
Previously we were using 338. So with the old formula we were writing 338/2 = 169 to the register. And 337500 / 169 = 1997.04 (we use 337500 here because it's the real clock value). With the new value of 337500/2000 we'll have 168.75, which is 168 on the round-down case and 169 on the round-closest case. If we write 168 to the register, 337500 / 168 = 2008.92, and 2008.92 is more distant from 2000 than 1997.04. So with this patch we're changing the formula but still writing the same correct value to the DP AUX register.
[0]: That's 1920x1080@50Hz on my DP monitor.
Quoted 4 lines
-Chris
--
Chris Wilson, Intel Open Source Technology Centre
Signature · 1 line
Paulo Zanoni
Re: [PATCH 6/9] drm/i915: make intel_ddi_get_cdclk_freq return values in KHz
In reply to Paulo ZanoniQuoted 11 lines · Paulo Zanoni
From: Paulo Zanoni <[email protected]>
With this, that 338 can finally become the correct 337500.
Due to the change we need to adjust the
intel_dp_aux_chfunction to set the correct value, so adjust the division and also useDIV_ROUND_CLOSESTinstead of the old "round down" behavior because the spec says the value "should be programmed to get as close as possible to the ideal rate of 2MHz".Signed-off-by Paulo Zanoni <[email protected]>
Quoted 58 lines
--- drivers/gpu/drm/i915/intel_ddi.c | 10 +++++----- drivers/gpu/drm/i915/intel_dp.c | 3 ++- 2 files changed, 7 insertions(+), 6 deletions(-)drivers/gpu/drm/i915/intel_ddi.c +5
-5@@ -1335,14 +1335,14 @@ static void intel_disable_ddi(struct intel_encoder *intel_encoder)int intel_ddi_get_cdclk_freq (struct drm_i915_private *dev_priv) {if (I915_READ (HSW_FUSE_STRAP ) &HSW_CDCLK_LIMIT ) -return 450 ; +return 450000 ;else if ((I915_READ (LCPLL_CTL ) &LCPLL_CLK_FREQ_MASK ) ==LCPLL_CLK_FREQ_450 ) -return 450 ; +return 450000 ;else if (IS_ULT (dev_priv->dev)) -return 338 ; +return 337500 ;else -return 540 ; +return 540000 ; }void intel_ddi_pll_init (struct drm_device *dev)@@ -1355,7 +1355,7 @@ void intel_ddi_pll_init(struct drm_device *dev)* Don't even try to turn it on. */ -DRM_DEBUG_KMS ("CDCLK running at %dMHz\n" , +DRM_DEBUG_KMS ("CDCLK running at %dKHz\n" ,intel_ddi_get_cdclk_freq (dev_priv));if (val &LCPLL_CD_SOURCE_FCLK )drivers/gpu/drm/i915/intel_dp.c +2
-1@@ -346,7 +346,8 @@ intel_dp_aux_ch(struct intel_dp *intel_dp,*/ if (is_cpu_edp (intel_dp)) {if (HAS_DDI (dev)) - aux_clock_divider =intel_ddi_get_cdclk_freq (dev_priv) >>1 ; + aux_clock_divider =DIV_ROUND_CLOSEST ( +intel_ddi_get_cdclk_freq (dev_priv),2000 );else if (IS_VALLEYVIEW (dev)) aux_clock_divider =100 ;else if (IS_GEN6 (dev) ||IS_GEN7 (dev))Signature · 6 lines
1.7.10.4 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 2 lines
Ville Syrjälä Intel OTC
PATCH 7/9 drm/i915: set the IPS linetime watermark
In reply to Paulo ZanoniFrom: Paulo Zanoni <[email protected]>
Remove the "placeholder" comment and set the actual value described by the specification. We still don't enable IPS, but it won't hurt to already have the value set here.
While at it, fully set the register value instead of just masking the values we're changing.
--- drivers/gpu/drm/i915/intel_pm.c | 22 +++++++--------------- 1 file changed, 7 insertions(+), 15 deletions(-)
drivers/gpu/drm/i915/intel_pm.c +7 -15
@@ -2022,7 +2022,7 @@ haswell_update_linetime_wm(struct drm_device *dev, struct drm_crtc *crtc)enum pipe pipe = intel_crtc->pipe;struct drm_display_mode *mode = &intel_crtc->config.adjusted_mode;int target_clock; -u32 temp; +u32 linetime, ips_linetime;if (!intel_crtc_active (crtc)) {I915_WRITE (PIPE_WM_LINETIME (pipe),0 );
@@ -2034,24 +2034,16 @@ haswell_update_linetime_wm(struct drm_device *dev, struct drm_crtc *crtc)else target_clock = intel_crtc->config.adjusted_mode.clock; - temp =I915_READ (PIPE_WM_LINETIME (pipe)); - temp &= ~PIPE_WM_LINETIME_MASK ; -/* The WM are computed with base on how long it takes to fill a single * row at the given clock rate, multiplied by 8. * */ - temp |=PIPE_WM_LINETIME_TIME ( -DIV_ROUND_CLOSEST (mode->htotal *1000 *8 , target_clock)); - -/* IPS watermarks are only used by pipe A, and are ignored by -* pipes B and C. They are calculated similarly to the common -* linetime values, except that we are using CD clock frequency -* in MHz instead of pixel rate for the division. -* -* This is a placeholder for the IPS watermark calculation code. -*/ + linetime =DIV_ROUND_CLOSEST (mode->htotal *1000 *8 , target_clock); + ips_linetime =DIV_ROUND_CLOSEST (mode->htotal *1000 *8 , +intel_ddi_get_cdclk_freq (dev_priv)); -I915_WRITE (PIPE_WM_LINETIME (pipe), temp); +I915_WRITE (PIPE_WM_LINETIME (pipe), +PIPE_WM_LINETIME_IPS_LINETIME (ips_linetime) | +PIPE_WM_LINETIME_TIME (linetime)); }static void haswell_update_wm (struct drm_device *dev)
Signature · 1 line
1.7.10.4
Re: [PATCH 7/9] drm/i915: set the IPS linetime watermark
Quoted 10 lines · Paulo Zanoni
From: Paulo Zanoni <[email protected]>
Remove the "placeholder" comment and set the actual value described by the specification. We still don't enable IPS, but it won't hurt to already have the value set here.
While at it, fully set the register value instead of just masking the values we're changing.
Signed-off-by Paulo Zanoni <[email protected]>
Quoted 55 lines
--- drivers/gpu/drm/i915/intel_pm.c | 22 +++++++--------------- 1 file changed, 7 insertions(+), 15 deletions(-)drivers/gpu/drm/i915/intel_pm.c +7
-15@@ -2022,7 +2022,7 @@ haswell_update_linetime_wm(struct drm_device *dev, struct drm_crtc *crtc)enum pipe pipe = intel_crtc->pipe;struct drm_display_mode *mode = &intel_crtc->config.adjusted_mode;int target_clock; -u32 temp; +u32 linetime, ips_linetime;if (!intel_crtc_active (crtc)) {I915_WRITE (PIPE_WM_LINETIME (pipe),0 );@@ -2034,24 +2034,16 @@ haswell_update_linetime_wm(struct drm_device *dev, struct drm_crtc *crtc)else target_clock = intel_crtc->config.adjusted_mode.clock; - temp =I915_READ (PIPE_WM_LINETIME (pipe)); - temp &= ~PIPE_WM_LINETIME_MASK ; -/* The WM are computed with base on how long it takes to fill a single * row at the given clock rate, multiplied by 8. * */ - temp |=PIPE_WM_LINETIME_TIME ( -DIV_ROUND_CLOSEST (mode->htotal *1000 *8 , target_clock)); - -/* IPS watermarks are only used by pipe A, and are ignored by -* pipes B and C. They are calculated similarly to the common -* linetime values, except that we are using CD clock frequency -* in MHz instead of pixel rate for the division. -* -* This is a placeholder for the IPS watermark calculation code. -*/ + linetime =DIV_ROUND_CLOSEST (mode->htotal *1000 *8 , target_clock); + ips_linetime =DIV_ROUND_CLOSEST (mode->htotal *1000 *8 , +intel_ddi_get_cdclk_freq (dev_priv)); -I915_WRITE (PIPE_WM_LINETIME (pipe), temp); +I915_WRITE (PIPE_WM_LINETIME (pipe), +PIPE_WM_LINETIME_IPS_LINETIME (ips_linetime) | +PIPE_WM_LINETIME_TIME (linetime)); }static void haswell_update_wm (struct drm_device *dev)Signature · 6 lines
1.7.10.4 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 2 lines
Ville Syrjälä Intel OTC
PATCH 8/9 drm/i915: MCH_SSKPD is a 64 bit register on Haswell
In reply to Paulo ZanoniFrom: Paulo Zanoni <[email protected]>
And the SNB_READ_WM0_LATENCY macro is not valid anymore because we have the "New WM0" at 63:56, so the "Old WM0" could maybe be zero if the new one is not zero.
--- drivers/gpu/drm/i915/intel_pm.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/i915/intel_pm.c +1 -1
@@ -4491,7 +4491,7 @@ void intel_init_pm(struct drm_device *dev) } dev_priv->display.init_clock_gating = ivybridge_init_clock_gating; }else if (IS_HASWELL (dev)) { -if (SNB_READ_WM0_LATENCY ()) { +if (I915_READ64 (MCH_SSKPD )) { dev_priv->display.update_wm = haswell_update_wm; dev_priv->display.update_sprite_wm = sandybridge_update_sprite_wm; }else {
Signature · 1 line
1.7.10.4
Re: [PATCH 8/9] drm/i915: MCH_SSKPD is a 64 bit register on Haswell
Quoted 7 lines · Paulo Zanoni
From: Paulo Zanoni <[email protected]>
And the
SNB_READ_WM0_LATENCYmacro is not valid anymore because we have the "New WM0" at 63:56, so the "Old WM0" could maybe be zero if the new one is not zero.Signed-off-by Paulo Zanoni <[email protected]>
Quoted 24 lines
--- drivers/gpu/drm/i915/intel_pm.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)drivers/gpu/drm/i915/intel_pm.c +1
-1@@ -4491,7 +4491,7 @@ void intel_init_pm(struct drm_device *dev) } dev_priv->display.init_clock_gating = ivybridge_init_clock_gating; }else if (IS_HASWELL (dev)) { -if (SNB_READ_WM0_LATENCY ()) { +if (I915_READ64 (MCH_SSKPD )) { dev_priv->display.update_wm = haswell_update_wm; dev_priv->display.update_sprite_wm = sandybridge_update_sprite_wm; }else {Signature · 6 lines
1.7.10.4 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 2 lines
Ville Syrjälä Intel OTC
PATCH 9/9 drm/i915: set FORCE_ARB_IDLE_PLANES workaround
In reply to Paulo ZanoniFrom: Paulo Zanoni <[email protected]>
Commit 1544d9d57396d5c0c6b7644ed5ae1f4d6caad07a added a workaround inside haswell_init_clock_gating and mentioned it is "a workaround for early silicon revisions and should be removed later". This workaround is documented in bit 31 of PRI_CTL. I asked Arthur and he mentioned that setting FORCE_ARB_IDLE_PLANES replaces that workaround for the newer machines. So use the new one.
Also notice that there's still another workaround for PRI_CTL that involves WM_DBG, but it's not the one we're reverting. And notice that we were previously setting WM_DBG_DISALLOW_MULTIPIPE_LP which disables the LP watermarks when more than one pipe is used, and we really don't want this because we need the LP watermarks if we want to reach deeper PC states.
--- drivers/gpu/drm/i915/i915_reg.h | 3 +++ drivers/gpu/drm/i915/intel_pm.c | 10 ++-------- 2 files changed, 5 insertions(+), 8 deletions(-)
drivers/gpu/drm/i915/i915_reg.h +3 -0
@@ -3697,6 +3697,9 @@# define CHICKEN3_DGMG_REQ_OUT_FIX_DISABLE (1 << 5)# define CHICKEN3_DGMG_DONE_FIX_DISABLE (1 << 2) +#define CHICKEN_PAR1_1 0x42080 +#define FORCE_ARB_IDLE_PLANES (1 << 14) +#define DISP_ARB_CTL 0x45000#define DISP_TILE_SURFACE_SWIZZLING (1<<13)#define DISP_FBC_WM_DIS (1<<15)
drivers/gpu/drm/i915/intel_pm.c +2 -8
@@ -4042,14 +4042,8 @@ static void haswell_init_clock_gating(struct drm_device *dev)/* WaSwitchSolVfFArbitrationPriority */ I915_WRITE (GAM_ECOCHK ,I915_READ (GAM_ECOCHK ) |HSW_ECOCHK_ARB_PRIO_SOL ); -/* XXX: This is a workaround for early silicon revisions and should be -* removed later. -*/ -I915_WRITE (WM_DBG , -I915_READ (WM_DBG ) | -WM_DBG_DISALLOW_MULTIPLE_LP | -WM_DBG_DISALLOW_SPRITE | -WM_DBG_DISALLOW_MAXFIFO ); +I915_WRITE (CHICKEN_PAR1_1 , +I915_READ (CHICKEN_PAR1_1 ) |FORCE_ARB_IDLE_PLANES );lpt_init_clock_gating (dev); }
Signature · 1 line
1.7.10.4
Re: [PATCH 9/9] drm/i915: set FORCE_ARB_IDLE_PLANES workaround
Quoted 9 lines · Paulo Zanoni
From: Paulo Zanoni <[email protected]>
Commit
1544d9d57396d5c0c6b7644ed5ae1f4d6caad07aadded a workaround insidehaswell_init_clock_gatingand mentioned it is "a workaround for early silicon revisions and should be removed later". This workaround is documented in bit 31 ofPRI_CTL. I asked Arthur and he mentioned that settingFORCE_ARB_IDLE_PLANESreplaces that workaround for the newer machines. So use the new one.
Quoted 8 lines
Also notice that there's still another workaround for
PRI_CTLthat involvesWM_DBG, but it's not the one we're reverting. And notice that we were previously settingWM_DBG_DISALLOW_MULTIPIPE_LPwhich disables the LP watermarks when more than one pipe is used, and we really don't want this because we need the LP watermarks if we want to reach deeper PC states.Signed-off-by Paulo Zanoni <[email protected]>
Quoted 47 lines
--- drivers/gpu/drm/i915/i915_reg.h | 3 +++ drivers/gpu/drm/i915/intel_pm.c | 10 ++-------- 2 files changed, 5 insertions(+), 8 deletions(-)drivers/gpu/drm/i915/i915_reg.h +3
-0@@ -3697,6 +3697,9 @@# define CHICKEN3_DGMG_REQ_OUT_FIX_DISABLE (1 << 5)# define CHICKEN3_DGMG_DONE_FIX_DISABLE (1 << 2) +#define CHICKEN_PAR1_1 0x42080 +#define FORCE_ARB_IDLE_PLANES (1 << 14) +#define DISP_ARB_CTL 0x45000#define DISP_TILE_SURFACE_SWIZZLING (1<<13)#define DISP_FBC_WM_DIS (1<<15)drivers/gpu/drm/i915/intel_pm.c +2
-8@@ -4042,14 +4042,8 @@ static void haswell_init_clock_gating(struct drm_device *dev)/* WaSwitchSolVfFArbitrationPriority */ I915_WRITE (GAM_ECOCHK ,I915_READ (GAM_ECOCHK ) |HSW_ECOCHK_ARB_PRIO_SOL ); -/* XXX: This is a workaround for early silicon revisions and should be -* removed later. -*/ -I915_WRITE (WM_DBG , -I915_READ (WM_DBG ) | -WM_DBG_DISALLOW_MULTIPLE_LP | -WM_DBG_DISALLOW_SPRITE | -WM_DBG_DISALLOW_MAXFIFO ); +I915_WRITE (CHICKEN_PAR1_1 , +I915_READ (CHICKEN_PAR1_1 ) |FORCE_ARB_IDLE_PLANES );lpt_init_clock_gating (dev); }Signature · 6 lines
1.7.10.4 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 2 lines
Ville Syrjälä Intel OTC
Re: [PATCH 9/9] drm/i915: set FORCE_ARB_IDLE_PLANES workaround
In reply to Paulo ZanoniQuoted 9 lines · Paulo Zanoni
From: Paulo Zanoni <[email protected]>
Commit
1544d9d57396d5c0c6b7644ed5ae1f4d6caad07aadded a workaround insidehaswell_init_clock_gatingand mentioned it is "a workaround for early silicon revisions and should be removed later". This workaround is documented in bit 31 ofPRI_CTL. I asked Arthur and he mentioned that settingFORCE_ARB_IDLE_PLANESreplaces that workaround for the newer machines. So use the new one.
Quoted 8 lines
Also notice that there's still another workaround for
PRI_CTLthat involvesWM_DBG, but it's not the one we're reverting. And notice that we were previously settingWM_DBG_DISALLOW_MULTIPIPE_LPwhich disables the LP watermarks when more than one pipe is used, and we really don't want this because we need the LP watermarks if we want to reach deeper PC states.Signed-off-by Paulo Zanoni <[email protected]>
I've applied all the patches in this series safe for the target_clock one. But this patch here is missing the w/a tag, can you please supply that one in a quick reply so that I can smash it into the patch?
Thanks, Daniel
Quoted 47 lines
--- drivers/gpu/drm/i915/i915_reg.h | 3 +++ drivers/gpu/drm/i915/intel_pm.c | 10 ++-------- 2 files changed, 5 insertions(+), 8 deletions(-)drivers/gpu/drm/i915/i915_reg.h +3
-0@@ -3697,6 +3697,9 @@# define CHICKEN3_DGMG_REQ_OUT_FIX_DISABLE (1 << 5)# define CHICKEN3_DGMG_DONE_FIX_DISABLE (1 << 2) +#define CHICKEN_PAR1_1 0x42080 +#define FORCE_ARB_IDLE_PLANES (1 << 14) +#define DISP_ARB_CTL 0x45000#define DISP_TILE_SURFACE_SWIZZLING (1<<13)#define DISP_FBC_WM_DIS (1<<15)drivers/gpu/drm/i915/intel_pm.c +2
-8@@ -4042,14 +4042,8 @@ static void haswell_init_clock_gating(struct drm_device *dev)/* WaSwitchSolVfFArbitrationPriority */ I915_WRITE (GAM_ECOCHK ,I915_READ (GAM_ECOCHK ) |HSW_ECOCHK_ARB_PRIO_SOL ); -/* XXX: This is a workaround for early silicon revisions and should be -* removed later. -*/ -I915_WRITE (WM_DBG , -I915_READ (WM_DBG ) | -WM_DBG_DISALLOW_MULTIPLE_LP | -WM_DBG_DISALLOW_SPRITE | -WM_DBG_DISALLOW_MAXFIFO ); +I915_WRITE (CHICKEN_PAR1_1 , +I915_READ (CHICKEN_PAR1_1 ) |FORCE_ARB_IDLE_PLANES );lpt_init_clock_gating (dev); }Signature · 6 lines
1.7.10.4 _______________________________________________ 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
Re: [PATCH 9/9] drm/i915: set FORCE_ARB_IDLE_PLANES workaround
2013/5/21 Daniel Vetter <[email protected]>:
Quoted 17 lines · Paulo Zanoni
Quoted 16 lines
From: Paulo Zanoni <[email protected]>
Commit
1544d9d57396d5c0c6b7644ed5ae1f4d6caad07aadded a workaround insidehaswell_init_clock_gatingand mentioned it is "a workaround for early silicon revisions and should be removed later". This workaround is documented in bit 31 ofPRI_CTL. I asked Arthur and he mentioned that settingFORCE_ARB_IDLE_PLANESreplaces that workaround for the newer machines. So use the new one.Also notice that there's still another workaround for
PRI_CTLthat involvesWM_DBG, but it's not the one we're reverting. And notice that we were previously settingWM_DBG_DISALLOW_MULTIPIPE_LPwhich disables the LP watermarks when more than one pipe is used, and we really don't want this because we need the LP watermarks if we want to reach deeper PC states.
Quoted 5 lines
Quoted 1 line
Signed-off-by Paulo Zanoni <[email protected]>I've applied all the patches in this series safe for the
target_clockone. But this patch here is missing the w/a tag, can you please supply that one in a quick reply so that I can smash it into the patch?
Just like most display workarounds, this one doesn't have a name.
Quoted 54 lines
Thanks, Daniel
Quoted 43 lines
--- drivers/gpu/drm/i915/i915_reg.h | 3 +++ drivers/gpu/drm/i915/intel_pm.c | 10 ++-------- 2 files changed, 5 insertions(+), 8 deletions(-)drivers/gpu/drm/i915/i915_reg.h +3
-0@@ -3697,6 +3697,9 @@# define CHICKEN3_DGMG_REQ_OUT_FIX_DISABLE (1 << 5)# define CHICKEN3_DGMG_DONE_FIX_DISABLE (1 << 2) +#define CHICKEN_PAR1_1 0x42080 +#define FORCE_ARB_IDLE_PLANES (1 << 14) +#define DISP_ARB_CTL 0x45000#define DISP_TILE_SURFACE_SWIZZLING (1<<13)#define DISP_FBC_WM_DIS (1<<15)drivers/gpu/drm/i915/intel_pm.c +2
-9@@ -4042,14 +4042,8 @@ static void haswell_init_clock_gating(struct drm_device *dev)/* WaSwitchSolVfFArbitrationPriority */ I915_WRITE (GAM_ECOCHK ,I915_READ (GAM_ECOCHK ) |HSW_ECOCHK_ARB_PRIO_SOL ); -/* XXX: This is a workaround for early silicon revisions and should be -* removed later. -*/ -I915_WRITE (WM_DBG , -I915_READ (WM_DBG ) | -WM_DBG_DISALLOW_MULTIPLE_LP | -WM_DBG_DISALLOW_SPRITE | -WM_DBG_DISALLOW_MAXFIFO ); +I915_WRITE (CHICKEN_PAR1_1 , +I915_READ (CHICKEN_PAR1_1 ) |FORCE_ARB_IDLE_PLANES );lpt_init_clock_gating (dev); } --1.7.10.4
Quoted 4 lines
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx--
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch
Signature · 1 line
Paulo Zanoni
Re: [PATCH 9/9] drm/i915: set FORCE_ARB_IDLE_PLANES workaround
Quoted 18 lines · Paulo Zanoni
2013/5/21 Daniel Vetter <[email protected]>:
Quoted 17 lines · Paulo Zanoni
Quoted 16 lines
From: Paulo Zanoni <[email protected]>
Commit
1544d9d57396d5c0c6b7644ed5ae1f4d6caad07aadded a workaround insidehaswell_init_clock_gatingand mentioned it is "a workaround for early silicon revisions and should be removed later". This workaround is documented in bit 31 ofPRI_CTL. I asked Arthur and he mentioned that settingFORCE_ARB_IDLE_PLANESreplaces that workaround for the newer machines. So use the new one.Also notice that there's still another workaround for
PRI_CTLthat involvesWM_DBG, but it's not the one we're reverting. And notice that we were previously settingWM_DBG_DISALLOW_MULTIPIPE_LPwhich disables the LP watermarks when more than one pipe is used, and we really don't want this because we need the LP watermarks if we want to reach deeper PC states.
Quoted 7 lines
Quoted 5 lines
Quoted 1 line
Signed-off-by Paulo Zanoni <[email protected]>I've applied all the patches in this series safe for the
target_clockone. But this patch here is missing the w/a tag, can you please supply that one in a quick reply so that I can smash it into the patch?Just like most display workarounds, this one doesn't have a name.
Actually I found a name for it in BSpec: WaRsPkgCStateDisplayPMReq
Quoted 64 lines
Quoted 55 lines
Thanks, Daniel
Quoted 43 lines
--- drivers/gpu/drm/i915/i915_reg.h | 3 +++ drivers/gpu/drm/i915/intel_pm.c | 10 ++-------- 2 files changed, 5 insertions(+), 8 deletions(-)drivers/gpu/drm/i915/i915_reg.h +3
-0@@ -3697,6 +3697,9 @@# define CHICKEN3_DGMG_REQ_OUT_FIX_DISABLE (1 << 5)# define CHICKEN3_DGMG_DONE_FIX_DISABLE (1 << 2) +#define CHICKEN_PAR1_1 0x42080 +#define FORCE_ARB_IDLE_PLANES (1 << 14) +#define DISP_ARB_CTL 0x45000#define DISP_TILE_SURFACE_SWIZZLING (1<<13)#define DISP_FBC_WM_DIS (1<<15)drivers/gpu/drm/i915/intel_pm.c +2
-9@@ -4042,14 +4042,8 @@ static void haswell_init_clock_gating(struct drm_device *dev)/* WaSwitchSolVfFArbitrationPriority */ I915_WRITE (GAM_ECOCHK ,I915_READ (GAM_ECOCHK ) |HSW_ECOCHK_ARB_PRIO_SOL ); -/* XXX: This is a workaround for early silicon revisions and should be -* removed later. -*/ -I915_WRITE (WM_DBG , -I915_READ (WM_DBG ) | -WM_DBG_DISALLOW_MULTIPLE_LP | -WM_DBG_DISALLOW_SPRITE | -WM_DBG_DISALLOW_MAXFIFO ); +I915_WRITE (CHICKEN_PAR1_1 , +I915_READ (CHICKEN_PAR1_1 ) |FORCE_ARB_IDLE_PLANES );lpt_init_clock_gating (dev); } --1.7.10.4
Quoted 4 lines
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx--
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.chSignature · 5 lines
Paulo Zanoni _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 2 lines
Ville Syrjälä Intel OTC
Re: [PATCH 9/9] drm/i915: set FORCE_ARB_IDLE_PLANES workaround
Quoted 7 lines · Ville Syrjälä
Quoted 5 lines
Quoted 3 lines
I've applied all the patches in this series safe for the
target_clockone. But this patch here is missing the w/a tag, can you please supply that one in a quick reply so that I can smash it into the patch?Just like most display workarounds, this one doesn't have a name.
Actually I found a name for it in BSpec: WaRsPkgCStateDisplayPMReq
Comment added, thanks for digging this one out.
-Daniel
--
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch