From 4811ff4f2388727a161ea49c2b0ddca95e44c7f9 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Ville=20Syrj=C3=A4l=C3=A4?= Date: Sat, 28 Jun 2014 02:04:07 +0300 Subject: drm/i915: Add chv_power_wells[] MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Add chv_power_wells[] so we can start to build up the power well support for chv. Just the "always on" well there initialy. Signed-off-by: Ville Syrjälä Tested-by: Rafael Barbalho Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 11 +++++++++++ 1 file changed, 11 insertions(+) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index 615e341682c3..7dbd7892b968 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -6574,6 +6574,15 @@ static struct i915_power_well vlv_power_wells[] = { }, }; +static struct i915_power_well chv_power_wells[] = { + { + .name = "always-on", + .always_on = 1, + .domains = VLV_ALWAYS_ON_POWER_DOMAINS, + .ops = &i9xx_always_on_power_well_ops, + }, +}; + static struct i915_power_well *lookup_power_well(struct drm_i915_private *dev_priv, enum punit_power_well power_well_id) { @@ -6610,6 +6619,8 @@ int intel_power_domains_init(struct drm_i915_private *dev_priv) } else if (IS_BROADWELL(dev_priv->dev)) { set_power_wells(power_domains, bdw_power_wells); hsw_pwr = power_domains; + } else if (IS_CHERRYVIEW(dev_priv->dev)) { + set_power_wells(power_domains, chv_power_wells); } else if (IS_VALLEYVIEW(dev_priv->dev)) { set_power_wells(power_domains, vlv_power_wells); } else { -- cgit v1.2.3 From 5d6f7ea752228788eddce0b9e268fa1f0eabdd7f Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Ville=20Syrj=C3=A4l=C3=A4?= Date: Sat, 28 Jun 2014 02:04:08 +0300 Subject: drm/i915: Add chv cmnlane power wells MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit CHV has two display PHYs so there are also two cmnlane power wells. Add the approriate code to power the wells up/down. Like on VLV we do the cmnreset assert/deassert and the DPLL refclock enabling at approriate times. This code actually works on my bsw. Signed-off-by: Ville Syrjälä Reviewed-by: Imre Deak Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 89 +++++++++++++++++++++++++++++++++++++++++ 1 file changed, 89 insertions(+) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index 7dbd7892b968..8a78015dd51e 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -6254,6 +6254,64 @@ static void vlv_dpio_cmn_power_well_disable(struct drm_i915_private *dev_priv, vlv_set_power_well(dev_priv, power_well, false); } +static void chv_dpio_cmn_power_well_enable(struct drm_i915_private *dev_priv, + struct i915_power_well *power_well) +{ + enum dpio_phy phy; + + WARN_ON_ONCE(power_well->data != PUNIT_POWER_WELL_DPIO_CMN_BC && + power_well->data != PUNIT_POWER_WELL_DPIO_CMN_D); + + /* + * Enable the CRI clock source so we can get at the + * display and the reference clock for VGA + * hotplug / manual detection. + */ + if (power_well->data == PUNIT_POWER_WELL_DPIO_CMN_BC) { + phy = DPIO_PHY0; + I915_WRITE(DPLL(PIPE_B), I915_READ(DPLL(PIPE_B)) | + DPLL_REFA_CLK_ENABLE_VLV); + I915_WRITE(DPLL(PIPE_B), I915_READ(DPLL(PIPE_B)) | + DPLL_REFA_CLK_ENABLE_VLV | DPLL_INTEGRATED_CRI_CLK_VLV); + } else { + phy = DPIO_PHY1; + I915_WRITE(DPLL(PIPE_C), I915_READ(DPLL(PIPE_C)) | + DPLL_REFA_CLK_ENABLE_VLV | DPLL_INTEGRATED_CRI_CLK_VLV); + } + udelay(1); /* >10ns for cmnreset, >0ns for sidereset */ + vlv_set_power_well(dev_priv, power_well, true); + + /* Poll for phypwrgood signal */ + if (wait_for(I915_READ(DISPLAY_PHY_STATUS) & PHY_POWERGOOD(phy), 1)) + DRM_ERROR("Display PHY %d is not power up\n", phy); + + I915_WRITE(DISPLAY_PHY_CONTROL, + PHY_COM_LANE_RESET_DEASSERT(phy, I915_READ(DISPLAY_PHY_CONTROL))); +} + +static void chv_dpio_cmn_power_well_disable(struct drm_i915_private *dev_priv, + struct i915_power_well *power_well) +{ + enum dpio_phy phy; + + WARN_ON_ONCE(power_well->data != PUNIT_POWER_WELL_DPIO_CMN_BC && + power_well->data != PUNIT_POWER_WELL_DPIO_CMN_D); + + if (power_well->data == PUNIT_POWER_WELL_DPIO_CMN_BC) { + phy = DPIO_PHY0; + assert_pll_disabled(dev_priv, PIPE_A); + assert_pll_disabled(dev_priv, PIPE_B); + } else { + phy = DPIO_PHY1; + assert_pll_disabled(dev_priv, PIPE_C); + } + + I915_WRITE(DISPLAY_PHY_CONTROL, + PHY_COM_LANE_RESET_ASSERT(phy, I915_READ(DISPLAY_PHY_CONTROL))); + + vlv_set_power_well(dev_priv, power_well, false); +} + static void check_power_well_state(struct drm_i915_private *dev_priv, struct i915_power_well *power_well) { @@ -6445,6 +6503,18 @@ EXPORT_SYMBOL_GPL(i915_get_cdclk_freq); BIT(POWER_DOMAIN_PORT_DDI_C_4_LANES) | \ BIT(POWER_DOMAIN_INIT)) +#define CHV_DPIO_CMN_BC_POWER_DOMAINS ( \ + BIT(POWER_DOMAIN_PORT_DDI_B_2_LANES) | \ + BIT(POWER_DOMAIN_PORT_DDI_B_4_LANES) | \ + BIT(POWER_DOMAIN_PORT_DDI_C_2_LANES) | \ + BIT(POWER_DOMAIN_PORT_DDI_C_4_LANES) | \ + BIT(POWER_DOMAIN_INIT)) + +#define CHV_DPIO_CMN_D_POWER_DOMAINS ( \ + BIT(POWER_DOMAIN_PORT_DDI_D_2_LANES) | \ + BIT(POWER_DOMAIN_PORT_DDI_D_4_LANES) | \ + BIT(POWER_DOMAIN_INIT)) + static const struct i915_power_well_ops i9xx_always_on_power_well_ops = { .sync_hw = i9xx_always_on_power_well_noop, .enable = i9xx_always_on_power_well_noop, @@ -6452,6 +6522,13 @@ static const struct i915_power_well_ops i9xx_always_on_power_well_ops = { .is_enabled = i9xx_always_on_power_well_enabled, }; +static const struct i915_power_well_ops chv_dpio_cmn_power_well_ops = { + .sync_hw = vlv_power_well_sync_hw, + .enable = chv_dpio_cmn_power_well_enable, + .disable = chv_dpio_cmn_power_well_disable, + .is_enabled = vlv_power_well_enabled, +}; + static struct i915_power_well i9xx_always_on_power_well[] = { { .name = "always-on", @@ -6581,6 +6658,18 @@ static struct i915_power_well chv_power_wells[] = { .domains = VLV_ALWAYS_ON_POWER_DOMAINS, .ops = &i9xx_always_on_power_well_ops, }, + { + .name = "dpio-common-bc", + .domains = CHV_DPIO_CMN_BC_POWER_DOMAINS, + .data = PUNIT_POWER_WELL_DPIO_CMN_BC, + .ops = &chv_dpio_cmn_power_well_ops, + }, + { + .name = "dpio-common-d", + .domains = CHV_DPIO_CMN_D_POWER_DOMAINS, + .data = PUNIT_POWER_WELL_DPIO_CMN_D, + .ops = &chv_dpio_cmn_power_well_ops, + }, }; static struct i915_power_well *lookup_power_well(struct drm_i915_private *dev_priv, -- cgit v1.2.3 From f07057d13c62c5b925725c6e03a0c4d1c0244bef Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Ville=20Syrj=C3=A4l=C3=A4?= Date: Sat, 28 Jun 2014 02:04:10 +0300 Subject: drm/i915: Add disp2d power well for chv MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Not sure if it's still there since chv has per-pipe power wells. At least with current Punit this doesn't work. Also the display irq handling would need to be adjusted for pipe C. So leave the code iffed out for now. Signed-off-by: Ville Syrjälä Reviewed-by: Imre Deak Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index 8a78015dd51e..dc8719f29d03 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -6658,6 +6658,14 @@ static struct i915_power_well chv_power_wells[] = { .domains = VLV_ALWAYS_ON_POWER_DOMAINS, .ops = &i9xx_always_on_power_well_ops, }, +#if 0 + { + .name = "display", + .domains = VLV_DISPLAY_POWER_DOMAINS, + .data = PUNIT_POWER_WELL_DISP2D, + .ops = &vlv_display_power_well_ops, + }, +#endif { .name = "dpio-common-bc", .domains = CHV_DPIO_CMN_BC_POWER_DOMAINS, -- cgit v1.2.3 From 26972b0a80091ccece1cbd9422772ae625a612f6 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Ville=20Syrj=C3=A4l=C3=A4?= Date: Sat, 28 Jun 2014 02:04:11 +0300 Subject: drm/i915: Add per-pipe power wells for chv MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit CHV has a power well for each pipe. Add the code to deal with them. The Punit in current hardware doesn't seem ready for this yet, so leave it iffed out. Signed-off-by: Ville Syrjälä Reviewed-by: Imre Deak Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 126 ++++++++++++++++++++++++++++++++++++++++ 1 file changed, 126 insertions(+) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index dc8719f29d03..95b3ca5964e9 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -6312,6 +6312,95 @@ static void chv_dpio_cmn_power_well_disable(struct drm_i915_private *dev_priv, vlv_set_power_well(dev_priv, power_well, false); } +static bool chv_pipe_power_well_enabled(struct drm_i915_private *dev_priv, + struct i915_power_well *power_well) +{ + enum pipe pipe = power_well->data; + bool enabled; + u32 state, ctrl; + + mutex_lock(&dev_priv->rps.hw_lock); + + state = vlv_punit_read(dev_priv, PUNIT_REG_DSPFREQ) & DP_SSS_MASK(pipe); + /* + * We only ever set the power-on and power-gate states, anything + * else is unexpected. + */ + WARN_ON(state != DP_SSS_PWR_ON(pipe) && state != DP_SSS_PWR_GATE(pipe)); + enabled = state == DP_SSS_PWR_ON(pipe); + + /* + * A transient state at this point would mean some unexpected party + * is poking at the power controls too. + */ + ctrl = vlv_punit_read(dev_priv, PUNIT_REG_DSPFREQ) & DP_SSC_MASK(pipe); + WARN_ON(ctrl << 16 != state); + + mutex_unlock(&dev_priv->rps.hw_lock); + + return enabled; +} + +static void chv_set_pipe_power_well(struct drm_i915_private *dev_priv, + struct i915_power_well *power_well, + bool enable) +{ + enum pipe pipe = power_well->data; + u32 state; + u32 ctrl; + + state = enable ? DP_SSS_PWR_ON(pipe) : DP_SSS_PWR_GATE(pipe); + + mutex_lock(&dev_priv->rps.hw_lock); + +#define COND \ + ((vlv_punit_read(dev_priv, PUNIT_REG_DSPFREQ) & DP_SSS_MASK(pipe)) == state) + + if (COND) + goto out; + + ctrl = vlv_punit_read(dev_priv, PUNIT_REG_DSPFREQ); + ctrl &= ~DP_SSC_MASK(pipe); + ctrl |= enable ? DP_SSC_PWR_ON(pipe) : DP_SSC_PWR_GATE(pipe); + vlv_punit_write(dev_priv, PUNIT_REG_DSPFREQ, ctrl); + + if (wait_for(COND, 100)) + DRM_ERROR("timout setting power well state %08x (%08x)\n", + state, + vlv_punit_read(dev_priv, PUNIT_REG_DSPFREQ)); + +#undef COND + +out: + mutex_unlock(&dev_priv->rps.hw_lock); +} + +static void chv_pipe_power_well_sync_hw(struct drm_i915_private *dev_priv, + struct i915_power_well *power_well) +{ + chv_set_pipe_power_well(dev_priv, power_well, power_well->count > 0); +} + +static void chv_pipe_power_well_enable(struct drm_i915_private *dev_priv, + struct i915_power_well *power_well) +{ + WARN_ON_ONCE(power_well->data != PIPE_A && + power_well->data != PIPE_B && + power_well->data != PIPE_C); + + chv_set_pipe_power_well(dev_priv, power_well, true); +} + +static void chv_pipe_power_well_disable(struct drm_i915_private *dev_priv, + struct i915_power_well *power_well) +{ + WARN_ON_ONCE(power_well->data != PIPE_A && + power_well->data != PIPE_B && + power_well->data != PIPE_C); + + chv_set_pipe_power_well(dev_priv, power_well, false); +} + static void check_power_well_state(struct drm_i915_private *dev_priv, struct i915_power_well *power_well) { @@ -6503,6 +6592,18 @@ EXPORT_SYMBOL_GPL(i915_get_cdclk_freq); BIT(POWER_DOMAIN_PORT_DDI_C_4_LANES) | \ BIT(POWER_DOMAIN_INIT)) +#define CHV_PIPE_A_POWER_DOMAINS ( \ + BIT(POWER_DOMAIN_PIPE_A) | \ + BIT(POWER_DOMAIN_INIT)) + +#define CHV_PIPE_B_POWER_DOMAINS ( \ + BIT(POWER_DOMAIN_PIPE_B) | \ + BIT(POWER_DOMAIN_INIT)) + +#define CHV_PIPE_C_POWER_DOMAINS ( \ + BIT(POWER_DOMAIN_PIPE_C) | \ + BIT(POWER_DOMAIN_INIT)) + #define CHV_DPIO_CMN_BC_POWER_DOMAINS ( \ BIT(POWER_DOMAIN_PORT_DDI_B_2_LANES) | \ BIT(POWER_DOMAIN_PORT_DDI_B_4_LANES) | \ @@ -6522,6 +6623,13 @@ static const struct i915_power_well_ops i9xx_always_on_power_well_ops = { .is_enabled = i9xx_always_on_power_well_enabled, }; +static const struct i915_power_well_ops chv_pipe_power_well_ops = { + .sync_hw = chv_pipe_power_well_sync_hw, + .enable = chv_pipe_power_well_enable, + .disable = chv_pipe_power_well_disable, + .is_enabled = chv_pipe_power_well_enabled, +}; + static const struct i915_power_well_ops chv_dpio_cmn_power_well_ops = { .sync_hw = vlv_power_well_sync_hw, .enable = chv_dpio_cmn_power_well_enable, @@ -6665,6 +6773,24 @@ static struct i915_power_well chv_power_wells[] = { .data = PUNIT_POWER_WELL_DISP2D, .ops = &vlv_display_power_well_ops, }, + { + .name = "pipe-a", + .domains = CHV_PIPE_A_POWER_DOMAINS, + .data = PIPE_A, + .ops = &chv_pipe_power_well_ops, + }, + { + .name = "pipe-b", + .domains = CHV_PIPE_B_POWER_DOMAINS, + .data = PIPE_B, + .ops = &chv_pipe_power_well_ops, + }, + { + .name = "pipe-c", + .domains = CHV_PIPE_C_POWER_DOMAINS, + .data = PIPE_C, + .ops = &chv_pipe_power_well_ops, + }, #endif { .name = "dpio-common-bc", -- cgit v1.2.3 From 8258356537871cf579868002552a0ed1762b0487 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Ville=20Syrj=C3=A4l=C3=A4?= Date: Sat, 28 Jun 2014 02:04:12 +0300 Subject: drm/i915: Add chv port B and C TX wells MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Add the TX wells for ports B and C just like on VLV. Again Punit doesn't seem ready (or the wells don't even exist anymore) so leave it iffed out. Signed-off-by: Ville Syrjälä Reviewed-by: Imre Deak Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 30 ++++++++++++++++++++++++++++++ 1 file changed, 30 insertions(+) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index 95b3ca5964e9..59157635d418 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -6804,6 +6804,36 @@ static struct i915_power_well chv_power_wells[] = { .data = PUNIT_POWER_WELL_DPIO_CMN_D, .ops = &chv_dpio_cmn_power_well_ops, }, +#if 0 + { + .name = "dpio-tx-b-01", + .domains = VLV_DPIO_TX_B_LANES_01_POWER_DOMAINS | + VLV_DPIO_TX_B_LANES_23_POWER_DOMAINS, + .ops = &vlv_dpio_power_well_ops, + .data = PUNIT_POWER_WELL_DPIO_TX_B_LANES_01, + }, + { + .name = "dpio-tx-b-23", + .domains = VLV_DPIO_TX_B_LANES_01_POWER_DOMAINS | + VLV_DPIO_TX_B_LANES_23_POWER_DOMAINS, + .ops = &vlv_dpio_power_well_ops, + .data = PUNIT_POWER_WELL_DPIO_TX_B_LANES_23, + }, + { + .name = "dpio-tx-c-01", + .domains = VLV_DPIO_TX_C_LANES_01_POWER_DOMAINS | + VLV_DPIO_TX_C_LANES_23_POWER_DOMAINS, + .ops = &vlv_dpio_power_well_ops, + .data = PUNIT_POWER_WELL_DPIO_TX_C_LANES_01, + }, + { + .name = "dpio-tx-c-23", + .domains = VLV_DPIO_TX_C_LANES_01_POWER_DOMAINS | + VLV_DPIO_TX_C_LANES_23_POWER_DOMAINS, + .ops = &vlv_dpio_power_well_ops, + .data = PUNIT_POWER_WELL_DPIO_TX_C_LANES_23, + }, +#endif }; static struct i915_power_well *lookup_power_well(struct drm_i915_private *dev_priv, -- cgit v1.2.3 From 2ce147f36dc5a1f3b49abd8ce3164ad0f04ec863 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Ville=20Syrj=C3=A4l=C3=A4?= Date: Sat, 28 Jun 2014 02:04:13 +0300 Subject: drm/i915: Add chv port D TX wells MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Add the TX wells for port D. The Punit subsystem numbers are a total guess at this time. Also I'm not sure these even exist. Certainly the Punit in current hardware doesn't deal with these. Signed-off-by: Ville Syrjälä Reviewed-by: Imre Deak Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 23 +++++++++++++++++++++++ 1 file changed, 23 insertions(+) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index 59157635d418..f4a1837c231c 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -6616,6 +6616,15 @@ EXPORT_SYMBOL_GPL(i915_get_cdclk_freq); BIT(POWER_DOMAIN_PORT_DDI_D_4_LANES) | \ BIT(POWER_DOMAIN_INIT)) +#define CHV_DPIO_TX_D_LANES_01_POWER_DOMAINS ( \ + BIT(POWER_DOMAIN_PORT_DDI_D_2_LANES) | \ + BIT(POWER_DOMAIN_PORT_DDI_D_4_LANES) | \ + BIT(POWER_DOMAIN_INIT)) + +#define CHV_DPIO_TX_D_LANES_23_POWER_DOMAINS ( \ + BIT(POWER_DOMAIN_PORT_DDI_D_4_LANES) | \ + BIT(POWER_DOMAIN_INIT)) + static const struct i915_power_well_ops i9xx_always_on_power_well_ops = { .sync_hw = i9xx_always_on_power_well_noop, .enable = i9xx_always_on_power_well_noop, @@ -6833,6 +6842,20 @@ static struct i915_power_well chv_power_wells[] = { .ops = &vlv_dpio_power_well_ops, .data = PUNIT_POWER_WELL_DPIO_TX_C_LANES_23, }, + { + .name = "dpio-tx-d-01", + .domains = CHV_DPIO_TX_D_LANES_01_POWER_DOMAINS | + CHV_DPIO_TX_D_LANES_23_POWER_DOMAINS, + .ops = &vlv_dpio_power_well_ops, + .data = PUNIT_POWER_WELL_DPIO_TX_D_LANES_01, + }, + { + .name = "dpio-tx-d-23", + .domains = CHV_DPIO_TX_D_LANES_01_POWER_DOMAINS | + CHV_DPIO_TX_D_LANES_23_POWER_DOMAINS, + .ops = &vlv_dpio_power_well_ops, + .data = PUNIT_POWER_WELL_DPIO_TX_D_LANES_23, + }, #endif }; -- cgit v1.2.3 From a5043453aa2412ece984373294529d177324c901 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Ville=20Syrj=C3=A4l=C3=A4?= Date: Sat, 28 Jun 2014 02:04:18 +0300 Subject: drm/i915: Split a few long debug prints MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Split some WM debug prints to multiple lines. This shouldn't hurt grappability since the important part is at the start and the rest is just repeated stuff for each pipe. Signed-off-by: Ville Syrjälä Reviewed-by: Jesse Barnes Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index f4a1837c231c..a318cd5ad8ed 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -1387,7 +1387,8 @@ static void valleyview_update_wm(struct drm_crtc *crtc) plane_sr = cursor_sr = 0; } - DRM_DEBUG_KMS("Setting FIFO watermarks - A: plane=%d, cursor=%d, B: plane=%d, cursor=%d, SR: plane=%d, cursor=%d\n", + DRM_DEBUG_KMS("Setting FIFO watermarks - A: plane=%d, cursor=%d, " + "B: plane=%d, cursor=%d, SR: plane=%d, cursor=%d\n", planea_wm, cursora_wm, planeb_wm, cursorb_wm, plane_sr, cursor_sr); @@ -1443,7 +1444,8 @@ static void g4x_update_wm(struct drm_crtc *crtc) plane_sr = cursor_sr = 0; } - DRM_DEBUG_KMS("Setting FIFO watermarks - A: plane=%d, cursor=%d, B: plane=%d, cursor=%d, SR: plane=%d, cursor=%d\n", + DRM_DEBUG_KMS("Setting FIFO watermarks - A: plane=%d, cursor=%d, " + "B: plane=%d, cursor=%d, SR: plane=%d, cursor=%d\n", planea_wm, cursora_wm, planeb_wm, cursorb_wm, plane_sr, cursor_sr); -- cgit v1.2.3 From da46f936bb0396f6a0fb87c2786f541e9f19a73c Mon Sep 17 00:00:00 2001 From: Rodrigo Vivi Date: Fri, 1 Aug 2014 02:04:45 -0700 Subject: drm/i915: Introduce FBC False Color for debug purposes. MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit With this bit enabled, HW changes the color when compressing frames for debug purposes. ALthough the simple way to enable a single bit is over intel_reg_write, this value is overwriten on next update_fbc so depending on the workload it is not possible to set this bit with intel-gpu-tools. So this patch introduces a persistent way to enable false color over debugfs. v2: Use DEFINE_SIMPLE_ATTRIBUTE as Daniel suggested v3: (Ville) only do false color for IVB+ since according to spec bit is MBZ before IVB. v4: We don't have FBC on valleyview nor on cherryview (Ben) v5: s/!HAS_PCH_SPLIT/!HAS_FBC (Ville) Cc: Ville Syrjälä Reviewed-by: Ben Widawsky Signed-off-by: Rodrigo Vivi Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index a318cd5ad8ed..ab80df2909e0 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -309,6 +309,9 @@ static void gen7_enable_fbc(struct drm_crtc *crtc) dpfc_ctl |= IVB_DPFC_CTL_FENCE_EN; + if (dev_priv->fbc.false_color) + dpfc_ctl |= FBC_CTL_FALSE_COLOR; + I915_WRITE(ILK_DPFC_CONTROL, dpfc_ctl | DPFC_CTL_EN); if (IS_IVYBRIDGE(dev)) { -- cgit v1.2.3 From 0a56067469bde6662ce7c89a3d290171f878bac4 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Ville=20Syrj=C3=A4l=C3=A4?= Date: Wed, 11 Jun 2014 16:51:18 +0300 Subject: drm/i915: Fill out the FWx watermark register defines MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Add defines for all the watermark registers on modernish gmch platforms. VLV has increased the number of bits available for certain watermaks so expand the masks appropriately. Also vlv and chv have added some extra FW registers. Not sure what happened on chv because a new register called FW9 is now at the offset where FW7 was on vlv, while FW7 and FW8 (another new register) have been moved off somewhere else. Oh well, well just need two defines for FW7 then. v2: Fix DSPHOWM1 offset (Paulo) Reviewed-by: Paulo Zanoni Signed-off-by: Ville Syrjälä Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index ab80df2909e0..0f9164d854de 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -1400,7 +1400,7 @@ static void valleyview_update_wm(struct drm_crtc *crtc) (plane_sr << DSPFW_SR_SHIFT) | (cursorb_wm << DSPFW_CURSORB_SHIFT) | (planeb_wm << DSPFW_PLANEB_SHIFT) | - planea_wm); + (planea_wm << DSPFW_PLANEA_SHIFT)); I915_WRITE(DSPFW2, (I915_READ(DSPFW2) & ~DSPFW_CURSORA_MASK) | (cursora_wm << DSPFW_CURSORA_SHIFT)); @@ -1457,7 +1457,7 @@ static void g4x_update_wm(struct drm_crtc *crtc) (plane_sr << DSPFW_SR_SHIFT) | (cursorb_wm << DSPFW_CURSORB_SHIFT) | (planeb_wm << DSPFW_PLANEB_SHIFT) | - planea_wm); + (planea_wm << DSPFW_PLANEA_SHIFT)); I915_WRITE(DSPFW2, (I915_READ(DSPFW2) & ~DSPFW_CURSORA_MASK) | (cursora_wm << DSPFW_CURSORA_SHIFT)); @@ -1531,8 +1531,11 @@ static void i965_update_wm(struct drm_crtc *unused_crtc) /* 965 has limitations... */ I915_WRITE(DSPFW1, (srwm << DSPFW_SR_SHIFT) | - (8 << 16) | (8 << 8) | (8 << 0)); - I915_WRITE(DSPFW2, (8 << 8) | (8 << 0)); + (8 << DSPFW_CURSORB_SHIFT) | + (8 << DSPFW_PLANEB_SHIFT) | + (8 << DSPFW_PLANEA_SHIFT)); + I915_WRITE(DSPFW2, (8 << DSPFW_CURSORA_SHIFT) | + (8 << DSPFW_PLANEC_SHIFT_OLD)); /* update cursor SR watermark */ I915_WRITE(DSPFW3, (cursor_sr << DSPFW_CURSOR_SR_SHIFT)); -- cgit v1.2.3 From 1abc4dc7e2ca7886b6b06bcf2ff1451702c83f9b Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Ville=20Syrj=C3=A4l=C3=A4?= Date: Thu, 26 Jun 2014 17:02:37 +0300 Subject: drm/i915: Parametrize VLV_DDL registers MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The VLV/CHV DDL registers are uniform, and neatly enough the register offsets are sane so we can easily unify them to a single set of defines and just pass the pipe as the parameter to compute the register offset. Note that we now fill out the drain latency for pipe C on CHV which we didn't do before. The rest of the pipe C watermarks are still untouched but that will be remedied later by adding a proper cherryview_update_wm() function. v2: Add a note about CHV pipe C changes (Paulo) Reviewed-by: Paulo Zanoni Signed-off-by: Ville Syrjälä Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 52 ++++++++++++++++++----------------------- 1 file changed, 23 insertions(+), 29 deletions(-) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index 0f9164d854de..2573c6967559 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -1313,35 +1313,29 @@ static bool vlv_compute_drain_latency(struct drm_device *dev, static void vlv_update_drain_latency(struct drm_device *dev) { struct drm_i915_private *dev_priv = dev->dev_private; - int planea_prec, planea_dl, planeb_prec, planeb_dl; - int cursora_prec, cursora_dl, cursorb_prec, cursorb_dl; - int plane_prec_mult, cursor_prec_mult; /* Precision multiplier is - either 16 or 32 */ - - /* For plane A, Cursor A */ - if (vlv_compute_drain_latency(dev, 0, &plane_prec_mult, &planea_dl, - &cursor_prec_mult, &cursora_dl)) { - cursora_prec = (cursor_prec_mult == DRAIN_LATENCY_PRECISION_32) ? - DDL_CURSORA_PRECISION_32 : DDL_CURSORA_PRECISION_64; - planea_prec = (plane_prec_mult == DRAIN_LATENCY_PRECISION_32) ? - DDL_PLANEA_PRECISION_32 : DDL_PLANEA_PRECISION_64; - - I915_WRITE(VLV_DDL1, cursora_prec | - (cursora_dl << DDL_CURSORA_SHIFT) | - planea_prec | planea_dl); - } - - /* For plane B, Cursor B */ - if (vlv_compute_drain_latency(dev, 1, &plane_prec_mult, &planeb_dl, - &cursor_prec_mult, &cursorb_dl)) { - cursorb_prec = (cursor_prec_mult == DRAIN_LATENCY_PRECISION_32) ? - DDL_CURSORB_PRECISION_32 : DDL_CURSORB_PRECISION_64; - planeb_prec = (plane_prec_mult == DRAIN_LATENCY_PRECISION_32) ? - DDL_PLANEB_PRECISION_32 : DDL_PLANEB_PRECISION_64; - - I915_WRITE(VLV_DDL2, cursorb_prec | - (cursorb_dl << DDL_CURSORB_SHIFT) | - planeb_prec | planeb_dl); + enum pipe pipe; + + for_each_pipe(pipe) { + int plane_prec, plane_dl; + int cursor_prec, cursor_dl; + int plane_prec_mult, cursor_prec_mult; + + if (!vlv_compute_drain_latency(dev, pipe, &plane_prec_mult, &plane_dl, + &cursor_prec_mult, &cursor_dl)) + continue; + + /* + * FIXME CHV spec still lists 16 and 32 as the precision + * values. Need to figure out if spec is outdated or what. + */ + cursor_prec = (cursor_prec_mult == DRAIN_LATENCY_PRECISION_64) ? + DDL_CURSOR_PRECISION_64 : DDL_CURSOR_PRECISION_32; + plane_prec = (plane_prec_mult == DRAIN_LATENCY_PRECISION_64) ? + DDL_PLANE_PRECISION_64 : DDL_PLANE_PRECISION_32; + + I915_WRITE(VLV_DDL(pipe), cursor_prec | + (cursor_dl << DDL_CURSOR_SHIFT) | + plane_prec | (plane_dl << DDL_PLANE_SHIFT)); } } -- cgit v1.2.3 From 41aad816d7061f6cc3d92f39fc655f034bbfb1c0 Mon Sep 17 00:00:00 2001 From: Gajanan Bhat Date: Wed, 16 Jul 2014 18:24:03 +0530 Subject: drm/i915: Update DDL only for current CRTC Instead of looping through all CRTCs, update DDL for current CRTC for which watermark is being updated. CHV is confirmed to have precision of 32/64 which is same as VLV. Reviewed-by: Imre Deak Signed-off-by: Gajanan Bhat Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 25 +++++++++---------------- 1 file changed, 9 insertions(+), 16 deletions(-) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index 2573c6967559..0feeae845f12 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -1310,24 +1310,17 @@ static bool vlv_compute_drain_latency(struct drm_device *dev, * latency value. */ -static void vlv_update_drain_latency(struct drm_device *dev) +static void vlv_update_drain_latency(struct drm_crtc *crtc) { + struct drm_device *dev = crtc->dev; struct drm_i915_private *dev_priv = dev->dev_private; - enum pipe pipe; - - for_each_pipe(pipe) { - int plane_prec, plane_dl; - int cursor_prec, cursor_dl; - int plane_prec_mult, cursor_prec_mult; + enum pipe pipe = to_intel_crtc(crtc)->pipe; + int plane_prec, plane_dl; + int cursor_prec, cursor_dl; + int plane_prec_mult, cursor_prec_mult; - if (!vlv_compute_drain_latency(dev, pipe, &plane_prec_mult, &plane_dl, - &cursor_prec_mult, &cursor_dl)) - continue; - - /* - * FIXME CHV spec still lists 16 and 32 as the precision - * values. Need to figure out if spec is outdated or what. - */ + if (vlv_compute_drain_latency(dev, pipe, &plane_prec_mult, &plane_dl, + &cursor_prec_mult, &cursor_dl)) { cursor_prec = (cursor_prec_mult == DRAIN_LATENCY_PRECISION_64) ? DDL_CURSOR_PRECISION_64 : DDL_CURSOR_PRECISION_32; plane_prec = (plane_prec_mult == DRAIN_LATENCY_PRECISION_64) ? @@ -1352,7 +1345,7 @@ static void valleyview_update_wm(struct drm_crtc *crtc) unsigned int enabled = 0; bool cxsr_enabled; - vlv_update_drain_latency(dev); + vlv_update_drain_latency(crtc); if (g4x_compute_wm0(dev, PIPE_A, &valleyview_wm_info, latency_ns, -- cgit v1.2.3 From 3c2777fd2faa5d1c1d5867baa086f2fd8b05479e Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Ville=20Syrj=C3=A4l=C3=A4?= Date: Thu, 26 Jun 2014 17:03:06 +0300 Subject: drm/i915: Add cherryview_update_wm() MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit CHV has a third pipe so we need to compute the watermarks for its planes. Add cherryview_update_wm() to do just that. v2: Rebase on top of Imre's cxsr changes v3: Pass crtc to vlv_update_drain_latency() Reviewed-by: Paulo Zanoni Signed-off-by: Ville Syrjälä Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 81 ++++++++++++++++++++++++++++++++++++++++- 1 file changed, 80 insertions(+), 1 deletion(-) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index 0feeae845f12..da43e30ad60b 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -1399,6 +1399,85 @@ static void valleyview_update_wm(struct drm_crtc *crtc) intel_set_memory_cxsr(dev_priv, true); } +static void cherryview_update_wm(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + static const int sr_latency_ns = 12000; + struct drm_i915_private *dev_priv = dev->dev_private; + int planea_wm, planeb_wm, planec_wm; + int cursora_wm, cursorb_wm, cursorc_wm; + int plane_sr, cursor_sr; + int ignore_plane_sr, ignore_cursor_sr; + unsigned int enabled = 0; + bool cxsr_enabled; + + vlv_update_drain_latency(crtc); + + if (g4x_compute_wm0(dev, PIPE_A, + &valleyview_wm_info, latency_ns, + &valleyview_cursor_wm_info, latency_ns, + &planea_wm, &cursora_wm)) + enabled |= 1 << PIPE_A; + + if (g4x_compute_wm0(dev, PIPE_B, + &valleyview_wm_info, latency_ns, + &valleyview_cursor_wm_info, latency_ns, + &planeb_wm, &cursorb_wm)) + enabled |= 1 << PIPE_B; + + if (g4x_compute_wm0(dev, PIPE_C, + &valleyview_wm_info, latency_ns, + &valleyview_cursor_wm_info, latency_ns, + &planec_wm, &cursorc_wm)) + enabled |= 1 << PIPE_C; + + if (single_plane_enabled(enabled) && + g4x_compute_srwm(dev, ffs(enabled) - 1, + sr_latency_ns, + &valleyview_wm_info, + &valleyview_cursor_wm_info, + &plane_sr, &ignore_cursor_sr) && + g4x_compute_srwm(dev, ffs(enabled) - 1, + 2*sr_latency_ns, + &valleyview_wm_info, + &valleyview_cursor_wm_info, + &ignore_plane_sr, &cursor_sr)) { + cxsr_enabled = true; + } else { + cxsr_enabled = false; + intel_set_memory_cxsr(dev_priv, false); + plane_sr = cursor_sr = 0; + } + + DRM_DEBUG_KMS("Setting FIFO watermarks - A: plane=%d, cursor=%d, " + "B: plane=%d, cursor=%d, C: plane=%d, cursor=%d, " + "SR: plane=%d, cursor=%d\n", + planea_wm, cursora_wm, + planeb_wm, cursorb_wm, + planec_wm, cursorc_wm, + plane_sr, cursor_sr); + + I915_WRITE(DSPFW1, + (plane_sr << DSPFW_SR_SHIFT) | + (cursorb_wm << DSPFW_CURSORB_SHIFT) | + (planeb_wm << DSPFW_PLANEB_SHIFT) | + (planea_wm << DSPFW_PLANEA_SHIFT)); + I915_WRITE(DSPFW2, + (I915_READ(DSPFW2) & ~DSPFW_CURSORA_MASK) | + (cursora_wm << DSPFW_CURSORA_SHIFT)); + I915_WRITE(DSPFW3, + (I915_READ(DSPFW3) & ~DSPFW_CURSOR_SR_MASK) | + (cursor_sr << DSPFW_CURSOR_SR_SHIFT)); + I915_WRITE(DSPFW9_CHV, + (I915_READ(DSPFW9_CHV) & ~(DSPFW_PLANEC_MASK | + DSPFW_CURSORC_MASK)) | + (planec_wm << DSPFW_PLANEC_SHIFT) | + (cursorc_wm << DSPFW_CURSORC_SHIFT)); + + if (cxsr_enabled) + intel_set_memory_cxsr(dev_priv, true); +} + static void g4x_update_wm(struct drm_crtc *crtc) { struct drm_device *dev = crtc->dev; @@ -7119,7 +7198,7 @@ void intel_init_pm(struct drm_device *dev) else if (INTEL_INFO(dev)->gen == 8) dev_priv->display.init_clock_gating = gen8_init_clock_gating; } else if (IS_CHERRYVIEW(dev)) { - dev_priv->display.update_wm = valleyview_update_wm; + dev_priv->display.update_wm = cherryview_update_wm; dev_priv->display.init_clock_gating = cherryview_init_clock_gating; } else if (IS_VALLEYVIEW(dev)) { -- cgit v1.2.3 From 3dd7b97458e8aa2d8985b46622d226fa635071e7 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Ville=20Syrj=C3=A4l=C3=A4?= Date: Fri, 27 Jun 2014 19:49:57 +0300 Subject: drm/i915: Hack to tie both common lanes together on chv MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit It looks like frobbing the cmnreset line on pne PHY disturbs the other PHY on chv. The result is a black screen. On HDMI it's just a flash of black, but DP usually falls over and can't get back up. As a workaround set up the power domains so that both common lane wells power up and down together. I also tried leaving the cmnreset deasserted even the if the power well goes down but that didn't seem acceptable to the PHY. Reviewed-by: Rafael Barbalho Signed-off-by: Ville Syrjälä Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 14 ++++++++++++-- 1 file changed, 12 insertions(+), 2 deletions(-) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index da43e30ad60b..aaae22815965 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -6877,13 +6877,23 @@ static struct i915_power_well chv_power_wells[] = { #endif { .name = "dpio-common-bc", - .domains = CHV_DPIO_CMN_BC_POWER_DOMAINS, + /* + * XXX: cmnreset for one PHY seems to disturb the other. + * As a workaround keep both powered on at the same + * time for now. + */ + .domains = CHV_DPIO_CMN_BC_POWER_DOMAINS | CHV_DPIO_CMN_D_POWER_DOMAINS, .data = PUNIT_POWER_WELL_DPIO_CMN_BC, .ops = &chv_dpio_cmn_power_well_ops, }, { .name = "dpio-common-d", - .domains = CHV_DPIO_CMN_D_POWER_DOMAINS, + /* + * XXX: cmnreset for one PHY seems to disturb the other. + * As a workaround keep both powered on at the same + * time for now. + */ + .domains = CHV_DPIO_CMN_BC_POWER_DOMAINS | CHV_DPIO_CMN_D_POWER_DOMAINS, .data = PUNIT_POWER_WELL_DPIO_CMN_D, .ops = &chv_dpio_cmn_power_well_ops, }, -- cgit v1.2.3 From efd814b73cfcf4ed1cb561561d8df6daccb3ceb6 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Ville=20Syrj=C3=A4l=C3=A4?= Date: Fri, 27 Jun 2014 19:52:13 +0300 Subject: drm/i915: Polish the chv cmnlane resrt macros MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Replace the semi-funky cmnlane assert/deassert macros with something a bit more conventional. Also protect the macro arguments properly (also for PHY_POWERGOOD()). Reviewed-by: Rafael Barbalho Signed-off-by: Ville Syrjälä Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index aaae22815965..42bb329b2d05 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -6359,8 +6359,8 @@ static void chv_dpio_cmn_power_well_enable(struct drm_i915_private *dev_priv, if (wait_for(I915_READ(DISPLAY_PHY_STATUS) & PHY_POWERGOOD(phy), 1)) DRM_ERROR("Display PHY %d is not power up\n", phy); - I915_WRITE(DISPLAY_PHY_CONTROL, - PHY_COM_LANE_RESET_DEASSERT(phy, I915_READ(DISPLAY_PHY_CONTROL))); + I915_WRITE(DISPLAY_PHY_CONTROL, I915_READ(DISPLAY_PHY_CONTROL) | + PHY_COM_LANE_RESET_DEASSERT(phy)); } static void chv_dpio_cmn_power_well_disable(struct drm_i915_private *dev_priv, @@ -6380,8 +6380,8 @@ static void chv_dpio_cmn_power_well_disable(struct drm_i915_private *dev_priv, assert_pll_disabled(dev_priv, PIPE_C); } - I915_WRITE(DISPLAY_PHY_CONTROL, - PHY_COM_LANE_RESET_ASSERT(phy, I915_READ(DISPLAY_PHY_CONTROL))); + I915_WRITE(DISPLAY_PHY_CONTROL, I915_READ(DISPLAY_PHY_CONTROL) & + ~PHY_COM_LANE_RESET_DEASSERT(phy)); vlv_set_power_well(dev_priv, power_well, false); } -- cgit v1.2.3 From 0948c2651413d56c90d7ee9c99d75bef82d4c351 Mon Sep 17 00:00:00 2001 From: Gajanan Bhat Date: Thu, 7 Aug 2014 01:58:24 +0530 Subject: drm/i915: Generalize drain latency computation MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Modify drain latency computation to use it for any plane. Same function can be used for primary, cursor and sprite planes. v2: Adressed review comments by Imre and Ville. - Moved clock round up in separate patch - Added WARN check for clock and pixel size - Simplified bit masking - Use cursor_base instead of reg read v3: Changed to bitwise shorthand operator for plane_dl assignment. Signed-off-by: Gajanan Bhat Reviewed-by: Ville Syrjälä Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 87 +++++++++++++++++++++++------------------ 1 file changed, 50 insertions(+), 37 deletions(-) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index 42bb329b2d05..de27439636e8 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -1271,33 +1271,24 @@ static bool g4x_compute_srwm(struct drm_device *dev, display, cursor); } -static bool vlv_compute_drain_latency(struct drm_device *dev, - int plane, - int *plane_prec_mult, - int *plane_dl, - int *cursor_prec_mult, - int *cursor_dl) +static bool vlv_compute_drain_latency(struct drm_crtc *crtc, + int pixel_size, + int *prec_mult, + int *drain_latency) { - struct drm_crtc *crtc; - int clock, pixel_size; int entries; + int clock = to_intel_crtc(crtc)->config.adjusted_mode.crtc_clock; - crtc = intel_get_crtc_for_plane(dev, plane); - if (!intel_crtc_active(crtc)) + if (WARN(clock == 0, "Pixel clock is zero!\n")) return false; - clock = to_intel_crtc(crtc)->config.adjusted_mode.crtc_clock; - pixel_size = crtc->primary->fb->bits_per_pixel / 8; /* BPP */ + if (WARN(pixel_size == 0, "Pixel size is zero!\n")) + return false; entries = (clock / 1000) * pixel_size; - *plane_prec_mult = (entries > 128) ? - DRAIN_LATENCY_PRECISION_64 : DRAIN_LATENCY_PRECISION_32; - *plane_dl = (64 * (*plane_prec_mult) * 4) / entries; - - entries = (clock / 1000) * 4; /* BPP is always 4 for cursor */ - *cursor_prec_mult = (entries > 128) ? - DRAIN_LATENCY_PRECISION_64 : DRAIN_LATENCY_PRECISION_32; - *cursor_dl = (64 * (*cursor_prec_mult) * 4) / entries; + *prec_mult = (entries > 128) ? DRAIN_LATENCY_PRECISION_64 : + DRAIN_LATENCY_PRECISION_32; + *drain_latency = (64 * (*prec_mult) * 4) / entries; return true; } @@ -1312,24 +1303,46 @@ static bool vlv_compute_drain_latency(struct drm_device *dev, static void vlv_update_drain_latency(struct drm_crtc *crtc) { - struct drm_device *dev = crtc->dev; - struct drm_i915_private *dev_priv = dev->dev_private; - enum pipe pipe = to_intel_crtc(crtc)->pipe; - int plane_prec, plane_dl; - int cursor_prec, cursor_dl; - int plane_prec_mult, cursor_prec_mult; - - if (vlv_compute_drain_latency(dev, pipe, &plane_prec_mult, &plane_dl, - &cursor_prec_mult, &cursor_dl)) { - cursor_prec = (cursor_prec_mult == DRAIN_LATENCY_PRECISION_64) ? - DDL_CURSOR_PRECISION_64 : DDL_CURSOR_PRECISION_32; - plane_prec = (plane_prec_mult == DRAIN_LATENCY_PRECISION_64) ? - DDL_PLANE_PRECISION_64 : DDL_PLANE_PRECISION_32; - - I915_WRITE(VLV_DDL(pipe), cursor_prec | - (cursor_dl << DDL_CURSOR_SHIFT) | - plane_prec | (plane_dl << DDL_PLANE_SHIFT)); + struct drm_i915_private *dev_priv = crtc->dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int pixel_size; + int drain_latency; + enum pipe pipe = intel_crtc->pipe; + int plane_prec, prec_mult, plane_dl; + + plane_dl = I915_READ(VLV_DDL(pipe)) & ~(DDL_PLANE_PRECISION_64 | + DRAIN_LATENCY_MASK | DDL_CURSOR_PRECISION_64 | + (DRAIN_LATENCY_MASK << DDL_CURSOR_SHIFT)); + + if (!intel_crtc_active(crtc)) { + I915_WRITE(VLV_DDL(pipe), plane_dl); + return; + } + + /* Primary plane Drain Latency */ + pixel_size = crtc->primary->fb->bits_per_pixel / 8; /* BPP */ + if (vlv_compute_drain_latency(crtc, pixel_size, &prec_mult, &drain_latency)) { + plane_prec = (prec_mult == DRAIN_LATENCY_PRECISION_64) ? + DDL_PLANE_PRECISION_64 : + DDL_PLANE_PRECISION_32; + plane_dl |= plane_prec | drain_latency; } + + /* Cursor Drain Latency + * BPP is always 4 for cursor + */ + pixel_size = 4; + + /* Program cursor DL only if it is enabled */ + if (intel_crtc->cursor_base && + vlv_compute_drain_latency(crtc, pixel_size, &prec_mult, &drain_latency)) { + plane_prec = (prec_mult == DRAIN_LATENCY_PRECISION_64) ? + DDL_CURSOR_PRECISION_64 : + DDL_CURSOR_PRECISION_32; + plane_dl |= plane_prec | (drain_latency << DDL_CURSOR_SHIFT); + } + + I915_WRITE(VLV_DDL(pipe), plane_dl); } #define single_plane_enabled(mask) is_power_of_2(mask) -- cgit v1.2.3 From a398e9c79eea74ba2f3f24ac08902661682f008c Mon Sep 17 00:00:00 2001 From: Gajanan Bhat Date: Tue, 5 Aug 2014 23:15:54 +0530 Subject: drm/i915: Round-up clock and limit drain latency MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Round up clock computation and limit drain latency to maximum of 0x7F. Signed-off-by: Gajanan Bhat Reviewed-by: Ville Syrjälä Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index de27439636e8..19bd7212f4a2 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -1285,11 +1285,14 @@ static bool vlv_compute_drain_latency(struct drm_crtc *crtc, if (WARN(pixel_size == 0, "Pixel size is zero!\n")) return false; - entries = (clock / 1000) * pixel_size; + entries = DIV_ROUND_UP(clock, 1000) * pixel_size; *prec_mult = (entries > 128) ? DRAIN_LATENCY_PRECISION_64 : DRAIN_LATENCY_PRECISION_32; *drain_latency = (64 * (*prec_mult) * 4) / entries; + if (*drain_latency > DRAIN_LATENCY_MASK) + *drain_latency = DRAIN_LATENCY_MASK; + return true; } -- cgit v1.2.3 From 01e184cc85d4516cab0ecea7c2c43a2dd3ad432b Mon Sep 17 00:00:00 2001 From: Gajanan Bhat Date: Thu, 7 Aug 2014 17:03:30 +0530 Subject: drm/i915: Add sprite watermark programming for VLV and CHV MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Program DDL register as part of sprite watermark programming for CHV and VLV. v2: Rename DRAIN_LATENCY_MAX by DRAIN_LATENCY_MASK v3: Addressed review comments by Ville - Changed Sprite DDL definitions to more generic to avoid multiple if-else - Changed bit masking to customary form - Changed to bitwise shorthand operator for sprite_dl assignment Signed-off-by: Gajanan Bhat Reviewed-by: Ville Syrjälä Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 33 +++++++++++++++++++++++++++++++++ 1 file changed, 33 insertions(+) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index 19bd7212f4a2..41de760bf1d4 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -1494,6 +1494,37 @@ static void cherryview_update_wm(struct drm_crtc *crtc) intel_set_memory_cxsr(dev_priv, true); } +static void valleyview_update_sprite_wm(struct drm_plane *plane, + struct drm_crtc *crtc, + uint32_t sprite_width, + uint32_t sprite_height, + int pixel_size, + bool enabled, bool scaled) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + int pipe = to_intel_plane(plane)->pipe; + int sprite = to_intel_plane(plane)->plane; + int drain_latency; + int plane_prec; + int sprite_dl; + int prec_mult; + + sprite_dl = I915_READ(VLV_DDL(pipe)) & ~(DDL_SPRITE_PRECISION_64(sprite) | + (DRAIN_LATENCY_MASK << DDL_SPRITE_SHIFT(sprite))); + + if (enabled && vlv_compute_drain_latency(crtc, pixel_size, &prec_mult, + &drain_latency)) { + plane_prec = (prec_mult == DRAIN_LATENCY_PRECISION_64) ? + DDL_SPRITE_PRECISION_64(sprite) : + DDL_SPRITE_PRECISION_32(sprite); + sprite_dl |= plane_prec | + (drain_latency << DDL_SPRITE_SHIFT(sprite)); + } + + I915_WRITE(VLV_DDL(pipe), sprite_dl); +} + static void g4x_update_wm(struct drm_crtc *crtc) { struct drm_device *dev = crtc->dev; @@ -7225,10 +7256,12 @@ void intel_init_pm(struct drm_device *dev) dev_priv->display.init_clock_gating = gen8_init_clock_gating; } else if (IS_CHERRYVIEW(dev)) { dev_priv->display.update_wm = cherryview_update_wm; + dev_priv->display.update_sprite_wm = valleyview_update_sprite_wm; dev_priv->display.init_clock_gating = cherryview_init_clock_gating; } else if (IS_VALLEYVIEW(dev)) { dev_priv->display.update_wm = valleyview_update_wm; + dev_priv->display.update_sprite_wm = valleyview_update_sprite_wm; dev_priv->display.init_clock_gating = valleyview_init_clock_gating; } else if (IS_PINEVIEW(dev)) { -- cgit v1.2.3 From 1bee20175f27b46427f10290fdd4a79334d41a60 Mon Sep 17 00:00:00 2001 From: Damien Lespiau Date: Sat, 9 Aug 2014 23:00:58 +0100 Subject: drm/i915: Remove set but unused 'gt_perf_status' Signed-off-by: Damien Lespiau Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 2 -- 1 file changed, 2 deletions(-) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index 41de760bf1d4..12f4e143328c 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -3719,7 +3719,6 @@ static void gen6_enable_rps(struct drm_device *dev) struct drm_i915_private *dev_priv = dev->dev_private; struct intel_engine_cs *ring; u32 rp_state_cap; - u32 gt_perf_status; u32 rc6vids, pcu_mbox = 0, rc6_mask = 0; u32 gtfifodbg; int rc6_mode; @@ -3744,7 +3743,6 @@ static void gen6_enable_rps(struct drm_device *dev) gen6_gt_force_wake_get(dev_priv, FORCEWAKE_ALL); rp_state_cap = I915_READ(GEN6_RP_STATE_CAP); - gt_perf_status = I915_READ(GEN6_GT_PERF_STATUS); parse_rp_state_cap(dev_priv, rp_state_cap); -- cgit v1.2.3 From 48404c1e53d4e15d7e3e77cd2c8f3f22973ef38e Mon Sep 17 00:00:00 2001 From: Sonika Jindal Date: Fri, 22 Aug 2014 14:06:04 +0530 Subject: drm/i915: Add 180 degree primary plane rotation support Primary planes support 180 degree rotation. Expose the feature through rotation drm property. v2: Calculating linear/tiled offsets based on pipe source width and height. Added 180 degree rotation support in ironlake_update_plane. v3: Checking if CRTC is active before issueing update_plane. Added wait for vblank to make sure we dont overtake page flips. Disabling FBC since it does not work with rotated planes. v4: Updated rotation checks for pending flips, fbc disable. Creating rotation property only for Gen4 onwards. Property resetting as part of lastclose. v5: Resetting property in i915_driver_lastclose properly for planes and crtcs. Fixed linear offset calculation that was off by 1 w.r.t width in i9xx_update_plane and ironlake_update_plane. Removed tab based indentation and unnecessary braces in intel_crtc_set_property and intel_update_fbc. FBC and flip related checks should be done only for valid crtcs. v6: Minor nits in FBC disable checks for comments in intel_crtc_set_property and positioning the disable code in intel_update_fbc. v7: In case rotation property on inactive crtc is updated, we return successfully printing debug log as crtc is inactive and only property change is preserved. v8: update_plane is changed to update_primary_plane, crtc->fb is changed to crtc->primary->fb and return value of update_primary_plane is ignored. v9: added rotation property to primary plane instead of crtc. Removing reset of rotation property from lastclose. rotation_property is moved to drm_mode_config, so drm layer will take care of resetting. Adding updation of fbc when rotation is set to 0. Allowing rotation only if value is different than old one. v10: Calling intel_primary_plane_setplane instead of update_primary_plane in set_property(Daniel). v11: Using same set_property function for both primary and sprite, Adding primary plane specific code in the same function (Matt). v12: Removing disabling/ enabling of fbc from set_property because it is done from intel_pipe_set_base. Other formatting v13: we need to call disable_fbc before changing the rotation to 180, disable_fbc from intel_pipe_set_base gets called very late, that will be used to re-enable fbc if rotation is set to 0 (Ville). Testcase: igt/kms_rotation_crc Signed-off-by: Uma Shankar Signed-off-by: Sagar Kamble Signed-off-by: Sonika Jindal [danvet: Add FIXME to explain why we need the open-coded update_fbc hunk to disable fbc when rotated 180 degree. And make checkpatch happier.] Acked-by: Matt Roper Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index c8f744c418f0..1ab3e117644e 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -581,6 +581,12 @@ void intel_update_fbc(struct drm_device *dev) DRM_DEBUG_KMS("framebuffer not tiled or fenced, disabling compression\n"); goto out_disable; } + if (INTEL_INFO(dev)->gen <= 4 && !IS_G4X(dev) && + to_intel_plane(crtc->primary)->rotation != BIT(DRM_ROTATE_0)) { + if (set_no_fbc_reason(dev_priv, FBC_UNSUPPORTED_MODE)) + DRM_DEBUG_KMS("Rotation unsupported, disabling\n"); + goto out_disable; + } /* If the kernel debugger is active, always disable compression */ if (in_dbg_master()) -- cgit v1.2.3 From 055e393fa3ade8cb91d8229f1c76ca9a7b23b8b3 Mon Sep 17 00:00:00 2001 From: Damien Lespiau Date: Mon, 18 Aug 2014 13:49:10 +0100 Subject: drm/i915: Use dev_priv as first argument of for_each_pipe() Chris has decided that enough is enough. It's time to fixup dev Vs dev_priv. This is a modest contribution to the crusade. v2: Still use INTEL_INFO(), for the (mythical!) case we want to hardcode the info struct with defines (Chris) Rename the macro argument from 'dev' to 'dev_priv' (Jani) v3: Use names unlikely to be used as macro arguments (Chris) Suggested-by: Chris Wilson Signed-off-by: Damien Lespiau Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 17 ++++++++--------- 1 file changed, 8 insertions(+), 9 deletions(-) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index 1ab3e117644e..b1b3a6b892c6 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -2654,7 +2654,7 @@ static struct intel_pipe_wm *ilk_find_best_result(struct drm_device *dev, #define WM_DIRTY_FBC (1 << 24) #define WM_DIRTY_DDB (1 << 25) -static unsigned int ilk_compute_wm_dirty(struct drm_device *dev, +static unsigned int ilk_compute_wm_dirty(struct drm_i915_private *dev_priv, const struct ilk_wm_values *old, const struct ilk_wm_values *new) { @@ -2662,7 +2662,7 @@ static unsigned int ilk_compute_wm_dirty(struct drm_device *dev, enum pipe pipe; int wm_lp; - for_each_pipe(pipe) { + for_each_pipe(dev_priv, pipe) { if (old->wm_linetime[pipe] != new->wm_linetime[pipe]) { dirty |= WM_DIRTY_LINETIME(pipe); /* Must disable LP1+ watermarks too */ @@ -2748,7 +2748,7 @@ static void ilk_write_wm_values(struct drm_i915_private *dev_priv, unsigned int dirty; uint32_t val; - dirty = ilk_compute_wm_dirty(dev, previous, results); + dirty = ilk_compute_wm_dirty(dev_priv, previous, results); if (!dirty) return; @@ -5213,7 +5213,7 @@ static void g4x_disable_trickle_feed(struct drm_device *dev) struct drm_i915_private *dev_priv = dev->dev_private; int pipe; - for_each_pipe(pipe) { + for_each_pipe(dev_priv, pipe) { I915_WRITE(DSPCNTR(pipe), I915_READ(DSPCNTR(pipe)) | DISPPLANE_TRICKLE_FEED_DISABLE); @@ -5328,7 +5328,7 @@ static void cpt_init_clock_gating(struct drm_device *dev) /* The below fixes the weird display corruption, a few pixels shifted * downward, on (only) LVDS of some HP laptops with IVY. */ - for_each_pipe(pipe) { + for_each_pipe(dev_priv, pipe) { val = I915_READ(TRANS_CHICKEN2(pipe)); val |= TRANS_CHICKEN2_TIMING_OVERRIDE; val &= ~TRANS_CHICKEN2_FDI_POLARITY_REVERSED; @@ -5340,7 +5340,7 @@ static void cpt_init_clock_gating(struct drm_device *dev) I915_WRITE(TRANS_CHICKEN2(pipe), val); } /* WADP0ClockGatingDisable */ - for_each_pipe(pipe) { + for_each_pipe(dev_priv, pipe) { I915_WRITE(TRANS_CHICKEN1(pipe), TRANS_CHICKEN1_DP0UNIT_GC_DISABLE); } @@ -5560,7 +5560,7 @@ static void gen8_init_clock_gating(struct drm_device *dev) I915_READ(CHICKEN_PAR1_1) | DPA_MASK_VBLANK_SRD); /* WaPsrDPRSUnmaskVBlankInSRD:bdw */ - for_each_pipe(pipe) { + for_each_pipe(dev_priv, pipe) { I915_WRITE(CHICKEN_PIPESL_1(pipe), I915_READ(CHICKEN_PIPESL_1(pipe)) | BDW_DPRS_MASK_VBLANK_SRD); @@ -6363,12 +6363,11 @@ static void vlv_dpio_cmn_power_well_enable(struct drm_i915_private *dev_priv, static void vlv_dpio_cmn_power_well_disable(struct drm_i915_private *dev_priv, struct i915_power_well *power_well) { - struct drm_device *dev = dev_priv->dev; enum pipe pipe; WARN_ON_ONCE(power_well->data != PUNIT_POWER_WELL_DPIO_CMN_BC); - for_each_pipe(pipe) + for_each_pipe(dev_priv, pipe) assert_pll_disabled(dev_priv, pipe); /* Assert common reset */ -- cgit v1.2.3 From 98a2e5f94275b6aafb12a3650937f6c54222cdc2 Mon Sep 17 00:00:00 2001 From: Deepak S Date: Mon, 18 Aug 2014 10:35:27 -0700 Subject: drm/i915: Bring UP Power Wells before disabling RC6. We need do forcewake before Disabling RC6, This is what the BIOS expects while going into suspend. v2: updated commit message. (Daniel) Cc: Paulo Zanoni Signed-off-by: Deepak S Signed-off-by: Rodrigo Vivi Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index b1b3a6b892c6..a381c9997a25 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -3533,8 +3533,14 @@ static void valleyview_disable_rps(struct drm_device *dev) { struct drm_i915_private *dev_priv = dev->dev_private; + /* we're doing forcewake before Disabling RC6, + * This what the BIOS expects when going into suspend */ + gen6_gt_force_wake_get(dev_priv, FORCEWAKE_ALL); + I915_WRITE(GEN6_RC_CONTROL, 0); + gen6_gt_force_wake_put(dev_priv, FORCEWAKE_ALL); + gen6_disable_rps_interrupts(dev); } -- cgit v1.2.3 From 89d6b2b81fdfa57666188d63dc970f8f14efed71 Mon Sep 17 00:00:00 2001 From: Paulo Zanoni Date: Thu, 21 Aug 2014 17:09:36 -0300 Subject: drm/i915: call lpt_init_clock_gating on BDW too Because BDW has WPT, which is equivalent to LPT. This is just like the CPT/PPT case. Signed-off-by: Paulo Zanoni Reviewed-by: Damien Lespiau Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index a381c9997a25..88944607b3a7 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -5607,6 +5607,8 @@ static void gen8_init_clock_gating(struct drm_device *dev) /* Wa4x4STCOptimizationDisable:bdw */ I915_WRITE(CACHE_MODE_1, _MASKED_BIT_ENABLE(GEN8_4x4_STC_OPTIMIZATION_DISABLE)); + + lpt_init_clock_gating(dev); } static void haswell_init_clock_gating(struct drm_device *dev) -- cgit v1.2.3 From 47c2bd97cf367f712433eece065df276a64d3e66 Mon Sep 17 00:00:00 2001 From: Paulo Zanoni Date: Thu, 21 Aug 2014 17:09:37 -0300 Subject: drm/i915: rename gen8_init_clock_gating to broadwell_init_clock_gating Because CHV uses cherryview_init_clock_gating instead of gen8_init_clock_gating. Signed-off-by: Paulo Zanoni Reviewed-by: Damien Lespiau Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index 88944607b3a7..2253f878adf4 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -5514,7 +5514,7 @@ static void lpt_suspend_hw(struct drm_device *dev) } } -static void gen8_init_clock_gating(struct drm_device *dev) +static void broadwell_init_clock_gating(struct drm_device *dev) { struct drm_i915_private *dev_priv = dev->dev_private; enum pipe pipe; @@ -7262,7 +7262,7 @@ void intel_init_pm(struct drm_device *dev) else if (IS_HASWELL(dev)) dev_priv->display.init_clock_gating = haswell_init_clock_gating; else if (INTEL_INFO(dev)->gen == 8) - dev_priv->display.init_clock_gating = gen8_init_clock_gating; + dev_priv->display.init_clock_gating = broadwell_init_clock_gating; } else if (IS_CHERRYVIEW(dev)) { dev_priv->display.update_wm = cherryview_update_wm; dev_priv->display.update_sprite_wm = valleyview_update_sprite_wm; -- cgit v1.2.3 From c5ad011d7d256ecbe173324029e992817194d2b0 Mon Sep 17 00:00:00 2001 From: Rodrigo Vivi Date: Mon, 4 Aug 2014 03:51:38 -0700 Subject: drm/i915: FBC flush nuke for BDW According to spec FBC on BDW and HSW are identical without any gaps. So let's copy the nuke and let FBC really start compressing stuff. Without this patch we can verify with false color that nothing is being compressed. With the nuke in place and false color it is possible to see false color debugs. Unfortunatelly on some rings like BCS on BDW we have to avoid Bits 22:18 on LRIs due to a high risk of hung. So, when using Blt ring for frontbuffer rend cache would never been cleaned and FBC would stop compressing buffer. One alternative is to cache clean on software frontbuffer tracking. v2: Fix rebase conflict. v3: Do not clean cache on BCS ring. Instead use sw frontbuffer tracking. Signed-off-by: Rodrigo Vivi Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 10 ++++++++++ 1 file changed, 10 insertions(+) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index 2253f878adf4..b9edfd426a19 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -345,6 +345,16 @@ bool intel_fbc_enabled(struct drm_device *dev) return dev_priv->display.fbc_enabled(dev); } +void gen8_fbc_sw_flush(struct drm_device *dev, u32 value) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + if (!IS_GEN8(dev)) + return; + + I915_WRITE(MSG_FBC_REND_STATE, value); +} + static void intel_fbc_work_fn(struct work_struct *__work) { struct intel_fbc_work *work = -- cgit v1.2.3 From 86d7f23842f1bce3ab5e8c8d0c676112bbc4c99b Mon Sep 17 00:00:00 2001 From: Arun Siluvery Date: Tue, 26 Aug 2014 14:44:50 +0100 Subject: drm/i915/bdw: Apply workarounds in render ring init function MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit For BDW workarounds are currently initialized in init_clock_gating() but they are lost during reset, suspend/resume etc; this patch moves the WAs that are part of register state context to render ring init fn otherwise default context ends up with incorrect values as they don't get initialized until init_clock_gating fn. v2: Add workarounds to golden render state This method has its own issues, first of all this is different for each gen and it is generated using a tool so adding new workaround and mainitaining them across gens is not a straightforward process. v3: Use LRIs to emit these workarounds (Ville) Instead of modifying the golden render state the same LRIs are emitted from within the driver. v4: Use abstract name when exporting gen specific routines (Chris) For: VIZ-4092 Signed-off-by: Arun Siluvery Reviewed-by: Ville Syrjälä Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 48 ----------------------------------------- 1 file changed, 48 deletions(-) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index b9edfd426a19..718023859686 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -5536,37 +5536,12 @@ static void broadwell_init_clock_gating(struct drm_device *dev) /* FIXME(BDW): Check all the w/a, some might only apply to * pre-production hw. */ - /* WaDisablePartialInstShootdown:bdw */ - I915_WRITE(GEN8_ROW_CHICKEN, - _MASKED_BIT_ENABLE(PARTIAL_INSTRUCTION_SHOOTDOWN_DISABLE)); - - /* WaDisableThreadStallDopClockGating:bdw */ - /* FIXME: Unclear whether we really need this on production bdw. */ - I915_WRITE(GEN8_ROW_CHICKEN, - _MASKED_BIT_ENABLE(STALL_DOP_GATING_DISABLE)); - /* - * This GEN8_CENTROID_PIXEL_OPT_DIS W/A is only needed for - * pre-production hardware - */ - I915_WRITE(HALF_SLICE_CHICKEN3, - _MASKED_BIT_ENABLE(GEN8_CENTROID_PIXEL_OPT_DIS)); - I915_WRITE(HALF_SLICE_CHICKEN3, - _MASKED_BIT_ENABLE(GEN8_SAMPLER_POWER_BYPASS_DIS)); I915_WRITE(GAMTARBMODE, _MASKED_BIT_ENABLE(ARB_MODE_BWGTLB_DISABLE)); I915_WRITE(_3D_CHICKEN3, _MASKED_BIT_ENABLE(_3D_CHICKEN_SDE_LIMIT_FIFO_POLY_DEPTH(2))); - I915_WRITE(COMMON_SLICE_CHICKEN2, - _MASKED_BIT_ENABLE(GEN8_CSC2_SBE_VUE_CACHE_CONSERVATIVE)); - - I915_WRITE(GEN7_HALF_SLICE_CHICKEN1, - _MASKED_BIT_ENABLE(GEN7_SINGLE_SUBSCAN_DISPATCH_ENABLE)); - - /* WaDisableDopClockGating:bdw May not be needed for production */ - I915_WRITE(GEN7_ROW_CHICKEN2, - _MASKED_BIT_ENABLE(DOP_CLOCK_GATING_DISABLE)); /* WaSwitchSolVfFArbitrationPriority:bdw */ I915_WRITE(GAM_ECOCHK, I915_READ(GAM_ECOCHK) | HSW_ECOCHK_ARB_PRIO_SOL); @@ -5582,31 +5557,12 @@ static void broadwell_init_clock_gating(struct drm_device *dev) BDW_DPRS_MASK_VBLANK_SRD); } - /* Use Force Non-Coherent whenever executing a 3D context. This is a - * workaround for for a possible hang in the unlikely event a TLB - * invalidation occurs during a PSD flush. - */ - I915_WRITE(HDC_CHICKEN0, - I915_READ(HDC_CHICKEN0) | - _MASKED_BIT_ENABLE(HDC_FORCE_NON_COHERENT)); - /* WaVSRefCountFullforceMissDisable:bdw */ /* WaDSRefCountFullforceMissDisable:bdw */ I915_WRITE(GEN7_FF_THREAD_MODE, I915_READ(GEN7_FF_THREAD_MODE) & ~(GEN8_FF_DS_REF_CNT_FFME | GEN7_FF_VS_REF_CNT_FFME)); - /* - * BSpec recommends 8x4 when MSAA is used, - * however in practice 16x4 seems fastest. - * - * Note that PS/WM thread counts depend on the WIZ hashing - * disable bit, which we don't touch here, but it's good - * to keep in mind (see 3DSTATE_PS and 3DSTATE_WM). - */ - I915_WRITE(GEN7_GT_MODE, - GEN6_WIZ_HASHING_MASK | GEN6_WIZ_HASHING_16x4); - I915_WRITE(GEN6_RC_SLEEP_PSMI_CONTROL, _MASKED_BIT_ENABLE(GEN8_RC_SEMA_IDLE_MSG_DISABLE)); @@ -5614,10 +5570,6 @@ static void broadwell_init_clock_gating(struct drm_device *dev) I915_WRITE(GEN8_UCGCTL6, I915_READ(GEN8_UCGCTL6) | GEN8_SDEUNIT_CLOCK_GATE_DISABLE); - /* Wa4x4STCOptimizationDisable:bdw */ - I915_WRITE(CACHE_MODE_1, - _MASKED_BIT_ENABLE(GEN8_4x4_STC_OPTIMIZATION_DISABLE)); - lpt_init_clock_gating(dev); } -- cgit v1.2.3 From 2bb25c17bbe032eecaffa401dbc21809a3f8e44a Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Ville=20Syrj=C3=A4l=C3=A4?= Date: Mon, 18 Aug 2014 14:42:44 +0300 Subject: drm/i915: Populate mem_freq in init_gt_powerwave() MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit init_clock_gating() is too late to read out the mem_freq. We already want to print out the GPU MHz numbers before it's called. Move the mem_freq setup to init_gt_powersave(). v2: Also kill the CHV_CZ_CLOCK_FREQ_MODE_* defines Signed-off-by: Ville Syrjälä Reviewed-by: Deepak S Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 90 ++++++++++++++++++++--------------------- 1 file changed, 43 insertions(+), 47 deletions(-) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index 718023859686..a5c1a6cf5b59 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -4106,11 +4106,27 @@ static void valleyview_cleanup_pctx(struct drm_device *dev) static void valleyview_init_gt_powersave(struct drm_device *dev) { struct drm_i915_private *dev_priv = dev->dev_private; + u32 val; valleyview_setup_pctx(dev); mutex_lock(&dev_priv->rps.hw_lock); + val = vlv_punit_read(dev_priv, PUNIT_REG_GPU_FREQ_STS); + switch ((val >> 6) & 3) { + case 0: + case 1: + dev_priv->mem_freq = 800; + break; + case 2: + dev_priv->mem_freq = 1066; + break; + case 3: + dev_priv->mem_freq = 1333; + break; + } + DRM_DEBUG_DRIVER("DDR speed: %d MHz", dev_priv->mem_freq); + dev_priv->rps.max_freq = valleyview_rps_max_freq(dev_priv); dev_priv->rps.rp0_freq = dev_priv->rps.max_freq; DRM_DEBUG_DRIVER("max GPU freq: %d MHz (%u)\n", @@ -4145,11 +4161,38 @@ static void valleyview_init_gt_powersave(struct drm_device *dev) static void cherryview_init_gt_powersave(struct drm_device *dev) { struct drm_i915_private *dev_priv = dev->dev_private; + u32 val; cherryview_setup_pctx(dev); mutex_lock(&dev_priv->rps.hw_lock); + val = vlv_punit_read(dev_priv, CCK_FUSE_REG); + switch ((val >> 2) & 0x7) { + case 0: + case 1: + dev_priv->rps.cz_freq = 200; + dev_priv->mem_freq = 1600; + break; + case 2: + dev_priv->rps.cz_freq = 267; + dev_priv->mem_freq = 1600; + break; + case 3: + dev_priv->rps.cz_freq = 333; + dev_priv->mem_freq = 2000; + break; + case 4: + dev_priv->rps.cz_freq = 320; + dev_priv->mem_freq = 1600; + break; + case 5: + dev_priv->rps.cz_freq = 400; + dev_priv->mem_freq = 1600; + break; + } + DRM_DEBUG_DRIVER("DDR speed: %d MHz", dev_priv->mem_freq); + dev_priv->rps.max_freq = cherryview_rps_max_freq(dev_priv); dev_priv->rps.rp0_freq = dev_priv->rps.max_freq; DRM_DEBUG_DRIVER("max GPU freq: %d MHz (%u)\n", @@ -5726,24 +5769,6 @@ static void ivybridge_init_clock_gating(struct drm_device *dev) static void valleyview_init_clock_gating(struct drm_device *dev) { struct drm_i915_private *dev_priv = dev->dev_private; - u32 val; - - mutex_lock(&dev_priv->rps.hw_lock); - val = vlv_punit_read(dev_priv, PUNIT_REG_GPU_FREQ_STS); - mutex_unlock(&dev_priv->rps.hw_lock); - switch ((val >> 6) & 3) { - case 0: - case 1: - dev_priv->mem_freq = 800; - break; - case 2: - dev_priv->mem_freq = 1066; - break; - case 3: - dev_priv->mem_freq = 1333; - break; - } - DRM_DEBUG_DRIVER("DDR speed: %d MHz", dev_priv->mem_freq); I915_WRITE(DSPCLK_GATE_D, VRHUNIT_CLOCK_GATE_DISABLE); @@ -5819,35 +5844,6 @@ static void valleyview_init_clock_gating(struct drm_device *dev) static void cherryview_init_clock_gating(struct drm_device *dev) { struct drm_i915_private *dev_priv = dev->dev_private; - u32 val; - - mutex_lock(&dev_priv->rps.hw_lock); - val = vlv_punit_read(dev_priv, CCK_FUSE_REG); - mutex_unlock(&dev_priv->rps.hw_lock); - switch ((val >> 2) & 0x7) { - case 0: - case 1: - dev_priv->rps.cz_freq = CHV_CZ_CLOCK_FREQ_MODE_200; - dev_priv->mem_freq = 1600; - break; - case 2: - dev_priv->rps.cz_freq = CHV_CZ_CLOCK_FREQ_MODE_267; - dev_priv->mem_freq = 1600; - break; - case 3: - dev_priv->rps.cz_freq = CHV_CZ_CLOCK_FREQ_MODE_333; - dev_priv->mem_freq = 2000; - break; - case 4: - dev_priv->rps.cz_freq = CHV_CZ_CLOCK_FREQ_MODE_320; - dev_priv->mem_freq = 1600; - break; - case 5: - dev_priv->rps.cz_freq = CHV_CZ_CLOCK_FREQ_MODE_400; - dev_priv->mem_freq = 1600; - break; - } - DRM_DEBUG_DRIVER("DDR speed: %d MHz", dev_priv->mem_freq); I915_WRITE(DSPCLK_GATE_D, VRHUNIT_CLOCK_GATE_DISABLE); -- cgit v1.2.3 From c76bb61a71083b2d90504cc6d0dda2047c5d63ca Mon Sep 17 00:00:00 2001 From: Daisy Sun Date: Mon, 11 Aug 2014 11:08:38 -0700 Subject: drm/i915/bdw: BDW Software Turbo BDW supports GT C0 residency reporting in constant time unit. Driver calculates GT utilization based on C0 residency and adjusts RP frequency up/down accordingly. For offscreen workload specificly, set frequency to RP0. Offscreen task is not restricted by frame rate, it can be executed as soon as possible. Transcoding and serilized workload between CPU and GPU both need high GT performance, RP0 is a good option in this case. RC6 will kick in to compensate power consumption when GT is not active. v2: Rebase on recent drm-intel-nightly v3: Add flip timerout monitor, when no flip is deteced within 100ms, set frequency to RP0. Signed-off-by: Daisy Sun [torourke: rebased on latest and resolved conflict] Signed-off-by: Tom O'Rourke Acked-by: Jesse Barnes Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 230 +++++++++++++++++++++++++++++++++------- 1 file changed, 191 insertions(+), 39 deletions(-) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index a5c1a6cf5b59..72791a787f88 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -2258,7 +2258,6 @@ int ilk_wm_max_level(const struct drm_device *dev) else return 2; } - static void intel_print_wm_latency(struct drm_device *dev, const char *name, const uint16_t wm[5]) @@ -3227,6 +3226,9 @@ static void gen6_set_rps_thresholds(struct drm_i915_private *dev_priv, u8 val) { int new_power; + if (dev_priv->rps.is_bdw_sw_turbo) + return; + new_power = dev_priv->rps.power; switch (dev_priv->rps.power) { case LOW_POWER: @@ -3434,8 +3436,11 @@ void gen6_rps_idle(struct drm_i915_private *dev_priv) valleyview_set_rps(dev_priv->dev, dev_priv->rps.min_freq_softlimit); else if (IS_VALLEYVIEW(dev)) vlv_set_rps_idle(dev_priv); - else + else if (!dev_priv->rps.is_bdw_sw_turbo + || atomic_read(&dev_priv->rps.sw_turbo.flip_received)){ gen6_set_rps(dev_priv->dev, dev_priv->rps.min_freq_softlimit); + } + dev_priv->rps.last_adj = 0; } mutex_unlock(&dev_priv->rps.hw_lock); @@ -3449,8 +3454,11 @@ void gen6_rps_boost(struct drm_i915_private *dev_priv) if (dev_priv->rps.enabled) { if (IS_VALLEYVIEW(dev)) valleyview_set_rps(dev_priv->dev, dev_priv->rps.max_freq_softlimit); - else + else if (!dev_priv->rps.is_bdw_sw_turbo + || atomic_read(&dev_priv->rps.sw_turbo.flip_received)){ gen6_set_rps(dev_priv->dev, dev_priv->rps.max_freq_softlimit); + } + dev_priv->rps.last_adj = 0; } mutex_unlock(&dev_priv->rps.hw_lock); @@ -3481,21 +3489,26 @@ void valleyview_set_rps(struct drm_device *dev, u8 val) static void gen8_disable_rps_interrupts(struct drm_device *dev) { struct drm_i915_private *dev_priv = dev->dev_private; + if (IS_BROADWELL(dev) && dev_priv->rps.is_bdw_sw_turbo){ + if (atomic_read(&dev_priv->rps.sw_turbo.flip_received)) + del_timer(&dev_priv->rps.sw_turbo.flip_timer); + dev_priv-> rps.is_bdw_sw_turbo = false; + } else { + I915_WRITE(GEN6_PMINTRMSK, ~GEN8_PMINTR_REDIRECT_TO_NON_DISP); + I915_WRITE(GEN8_GT_IER(2), I915_READ(GEN8_GT_IER(2)) & + ~dev_priv->pm_rps_events); + /* Complete PM interrupt masking here doesn't race with the rps work + * item again unmasking PM interrupts because that is using a different + * register (GEN8_GT_IMR(2)) to mask PM interrupts. The only risk is in + * leaving stale bits in GEN8_GT_IIR(2) and GEN8_GT_IMR(2) which + * gen8_enable_rps will clean up. */ - I915_WRITE(GEN6_PMINTRMSK, ~GEN8_PMINTR_REDIRECT_TO_NON_DISP); - I915_WRITE(GEN8_GT_IER(2), I915_READ(GEN8_GT_IER(2)) & - ~dev_priv->pm_rps_events); - /* Complete PM interrupt masking here doesn't race with the rps work - * item again unmasking PM interrupts because that is using a different - * register (GEN8_GT_IMR(2)) to mask PM interrupts. The only risk is in - * leaving stale bits in GEN8_GT_IIR(2) and GEN8_GT_IMR(2) which - * gen8_enable_rps will clean up. */ - - spin_lock_irq(&dev_priv->irq_lock); - dev_priv->rps.pm_iir = 0; - spin_unlock_irq(&dev_priv->irq_lock); + spin_lock_irq(&dev_priv->irq_lock); + dev_priv->rps.pm_iir = 0; + spin_unlock_irq(&dev_priv->irq_lock); - I915_WRITE(GEN8_GT_IIR(2), dev_priv->pm_rps_events); + I915_WRITE(GEN8_GT_IIR(2), dev_priv->pm_rps_events); + } } static void gen6_disable_rps_interrupts(struct drm_device *dev) @@ -3653,13 +3666,111 @@ static void parse_rp_state_cap(struct drm_i915_private *dev_priv, u32 rp_state_c dev_priv->rps.min_freq_softlimit = dev_priv->rps.min_freq; } +static void bdw_sw_calculate_freq(struct drm_device *dev, + struct intel_rps_bdw_cal *c, u32 *cur_time, u32 *c0) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u64 busy = 0; + u32 busyness_pct = 0; + u32 elapsed_time = 0; + u16 new_freq = 0; + + if (!c || !cur_time || !c0) + return; + + if (0 == c->last_c0) + goto out; + + /* Check Evaluation interval */ + elapsed_time = *cur_time - c->last_ts; + if (elapsed_time < c->eval_interval) + return; + + mutex_lock(&dev_priv->rps.hw_lock); + + /* + * c0 unit in 32*1.28 usec, elapsed_time unit in 1 usec. + * Whole busyness_pct calculation should be + * busy = ((u64)(*c0 - c->last_c0) << 5 << 7) / 100; + * busyness_pct = (u32)(busy * 100 / elapsed_time); + * The final formula is to simplify CPU calculation + */ + busy = (u64)(*c0 - c->last_c0) << 12; + do_div(busy, elapsed_time); + busyness_pct = (u32)busy; + + if (c->is_up && busyness_pct >= c->it_threshold_pct) + new_freq = (u16)dev_priv->rps.cur_freq + 3; + if (!c->is_up && busyness_pct <= c->it_threshold_pct) + new_freq = (u16)dev_priv->rps.cur_freq - 1; + + /* Adjust to new frequency busyness and compare with threshold */ + if (0 != new_freq) { + if (new_freq > dev_priv->rps.max_freq_softlimit) + new_freq = dev_priv->rps.max_freq_softlimit; + else if (new_freq < dev_priv->rps.min_freq_softlimit) + new_freq = dev_priv->rps.min_freq_softlimit; + + gen6_set_rps(dev, new_freq); + } + + mutex_unlock(&dev_priv->rps.hw_lock); + +out: + c->last_c0 = *c0; + c->last_ts = *cur_time; +} + +static void gen8_set_frequency_RP0(struct work_struct *work) +{ + struct intel_rps_bdw_turbo *p_bdw_turbo = + container_of(work, struct intel_rps_bdw_turbo, work_max_freq); + struct intel_gen6_power_mgmt *p_power_mgmt = + container_of(p_bdw_turbo, struct intel_gen6_power_mgmt, sw_turbo); + struct drm_i915_private *dev_priv = + container_of(p_power_mgmt, struct drm_i915_private, rps); + + mutex_lock(&dev_priv->rps.hw_lock); + gen6_set_rps(dev_priv->dev, dev_priv->rps.rp0_freq); + mutex_unlock(&dev_priv->rps.hw_lock); +} + +static void flip_active_timeout_handler(unsigned long var) +{ + struct drm_i915_private *dev_priv = (struct drm_i915_private *) var; + + del_timer(&dev_priv->rps.sw_turbo.flip_timer); + atomic_set(&dev_priv->rps.sw_turbo.flip_received, false); + + queue_work(dev_priv->wq, &dev_priv->rps.sw_turbo.work_max_freq); +} + +void bdw_software_turbo(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + u32 current_time = I915_READ(TIMESTAMP_CTR); /* unit in usec */ + u32 current_c0 = I915_READ(MCHBAR_PCU_C0); /* unit in 32*1.28 usec */ + + bdw_sw_calculate_freq(dev, &dev_priv->rps.sw_turbo.up, + ¤t_time, ¤t_c0); + bdw_sw_calculate_freq(dev, &dev_priv->rps.sw_turbo.down, + ¤t_time, ¤t_c0); +} + static void gen8_enable_rps(struct drm_device *dev) { struct drm_i915_private *dev_priv = dev->dev_private; struct intel_engine_cs *ring; uint32_t rc6_mask = 0, rp_state_cap; + uint32_t threshold_up_pct, threshold_down_pct; + uint32_t ei_up, ei_down; /* up and down evaluation interval */ + u32 rp_ctl_flag; int unused; + /* Use software Turbo for BDW */ + dev_priv->rps.is_bdw_sw_turbo = IS_BROADWELL(dev); + /* 1a: Software RC state - RC0 */ I915_WRITE(GEN6_RC_STATE, 0); @@ -3703,35 +3814,74 @@ static void gen8_enable_rps(struct drm_device *dev) HSW_FREQUENCY(dev_priv->rps.rp1_freq)); I915_WRITE(GEN6_RC_VIDEO_FREQ, HSW_FREQUENCY(dev_priv->rps.rp1_freq)); - /* NB: Docs say 1s, and 1000000 - which aren't equivalent */ - I915_WRITE(GEN6_RP_DOWN_TIMEOUT, 100000000 / 128); /* 1 second timeout */ + ei_up = 84480; /* 84.48ms */ + ei_down = 448000; + threshold_up_pct = 90; /* x percent busy */ + threshold_down_pct = 70; + + if (dev_priv->rps.is_bdw_sw_turbo) { + dev_priv->rps.sw_turbo.up.it_threshold_pct = threshold_up_pct; + dev_priv->rps.sw_turbo.up.eval_interval = ei_up; + dev_priv->rps.sw_turbo.up.is_up = true; + dev_priv->rps.sw_turbo.up.last_ts = 0; + dev_priv->rps.sw_turbo.up.last_c0 = 0; + + dev_priv->rps.sw_turbo.down.it_threshold_pct = threshold_down_pct; + dev_priv->rps.sw_turbo.down.eval_interval = ei_down; + dev_priv->rps.sw_turbo.down.is_up = false; + dev_priv->rps.sw_turbo.down.last_ts = 0; + dev_priv->rps.sw_turbo.down.last_c0 = 0; + + /* Start the timer to track if flip comes*/ + dev_priv->rps.sw_turbo.timeout = 200*1000; /* in us */ + + init_timer(&dev_priv->rps.sw_turbo.flip_timer); + dev_priv->rps.sw_turbo.flip_timer.function = flip_active_timeout_handler; + dev_priv->rps.sw_turbo.flip_timer.data = (unsigned long) dev_priv; + dev_priv->rps.sw_turbo.flip_timer.expires = + usecs_to_jiffies(dev_priv->rps.sw_turbo.timeout) + jiffies; + add_timer(&dev_priv->rps.sw_turbo.flip_timer); + INIT_WORK(&dev_priv->rps.sw_turbo.work_max_freq, gen8_set_frequency_RP0); + + atomic_set(&dev_priv->rps.sw_turbo.flip_received, true); + } else { + /* NB: Docs say 1s, and 1000000 - which aren't equivalent + * 1 second timeout*/ + I915_WRITE(GEN6_RP_DOWN_TIMEOUT, FREQ_1_28_US(1000000)); - /* Docs recommend 900MHz, and 300 MHz respectively */ - I915_WRITE(GEN6_RP_INTERRUPT_LIMITS, - dev_priv->rps.max_freq_softlimit << 24 | - dev_priv->rps.min_freq_softlimit << 16); + /* Docs recommend 900MHz, and 300 MHz respectively */ + I915_WRITE(GEN6_RP_INTERRUPT_LIMITS, + dev_priv->rps.max_freq_softlimit << 24 | + dev_priv->rps.min_freq_softlimit << 16); - I915_WRITE(GEN6_RP_UP_THRESHOLD, 7600000 / 128); /* 76ms busyness per EI, 90% */ - I915_WRITE(GEN6_RP_DOWN_THRESHOLD, 31300000 / 128); /* 313ms busyness per EI, 70%*/ - I915_WRITE(GEN6_RP_UP_EI, 66000); /* 84.48ms, XXX: random? */ - I915_WRITE(GEN6_RP_DOWN_EI, 350000); /* 448ms, XXX: random? */ + I915_WRITE(GEN6_RP_UP_THRESHOLD, + FREQ_1_28_US(ei_up * threshold_up_pct / 100)); + I915_WRITE(GEN6_RP_DOWN_THRESHOLD, + FREQ_1_28_US(ei_down * threshold_down_pct / 100)); + I915_WRITE(GEN6_RP_UP_EI, + FREQ_1_28_US(ei_up)); + I915_WRITE(GEN6_RP_DOWN_EI, + FREQ_1_28_US(ei_down)); - I915_WRITE(GEN6_RP_IDLE_HYSTERSIS, 10); + I915_WRITE(GEN6_RP_IDLE_HYSTERSIS, 10); + } /* 5: Enable RPS */ - I915_WRITE(GEN6_RP_CONTROL, - GEN6_RP_MEDIA_TURBO | - GEN6_RP_MEDIA_HW_NORMAL_MODE | - GEN6_RP_MEDIA_IS_GFX | - GEN6_RP_ENABLE | - GEN6_RP_UP_BUSY_AVG | - GEN6_RP_DOWN_IDLE_AVG); - - /* 6: Ring frequency + overclocking (our driver does this later */ - + rp_ctl_flag = GEN6_RP_MEDIA_TURBO | + GEN6_RP_MEDIA_HW_NORMAL_MODE | + GEN6_RP_MEDIA_IS_GFX | + GEN6_RP_UP_BUSY_AVG | + GEN6_RP_DOWN_IDLE_AVG; + if (!dev_priv->rps.is_bdw_sw_turbo) + rp_ctl_flag |= GEN6_RP_ENABLE; + + I915_WRITE(GEN6_RP_CONTROL, rp_ctl_flag); + + /* 6: Ring frequency + overclocking + * (our driver does this later */ gen6_set_rps(dev, (I915_READ(GEN6_GT_PERF_STATUS) & 0xff00) >> 8); - - gen8_enable_rps_interrupts(dev); + if (!dev_priv->rps.is_bdw_sw_turbo) + gen8_enable_rps_interrupts(dev); gen6_gt_force_wake_put(dev_priv, FORCEWAKE_ALL); } @@ -5199,6 +5349,8 @@ static void intel_gen6_powersave_work(struct work_struct *work) rps.delayed_resume_work.work); struct drm_device *dev = dev_priv->dev; + dev_priv->rps.is_bdw_sw_turbo = false; + mutex_lock(&dev_priv->rps.hw_lock); if (IS_CHERRYVIEW(dev)) { -- cgit v1.2.3 From 1c14762d0cc3eecfdc5060bd634d124d32bd9d44 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Ville=20Syrj=C3=A4l=C3=A4?= Date: Mon, 18 Aug 2014 14:42:43 +0300 Subject: drm/i915: Warn about odd rps values on CHV MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit CHV wants even rps opcodes so print a warning of the min/max/rpe/rp1 values are odd, and warn if an odd value slips through to valleyview_set_rps() and truncate it to an even value. Also add a comment to chv_freq_opcode() to make sure no one changes the code without considering this requirement. Signed-off-by: Ville Syrjälä Reviewed-by: Deepak S [danvet: Help git along in applying the patch, somehow it silently ended up in the vlv init_gt_powersave function.] Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 11 +++++++++++ 1 file changed, 11 insertions(+) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index 72791a787f88..6bf533100dad 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -3477,6 +3477,10 @@ void valleyview_set_rps(struct drm_device *dev, u8 val) dev_priv->rps.cur_freq, vlv_gpu_freq(dev_priv, val), val); + if (WARN_ONCE(IS_CHERRYVIEW(dev) && (val & 1), + "Odd GPU freq value\n")) + val &= ~1; + if (val != dev_priv->rps.cur_freq) vlv_punit_write(dev_priv, PUNIT_REG_GPU_FREQ_REQ, val); @@ -4364,6 +4368,12 @@ static void cherryview_init_gt_powersave(struct drm_device *dev) vlv_gpu_freq(dev_priv, dev_priv->rps.min_freq), dev_priv->rps.min_freq); + WARN_ONCE((dev_priv->rps.max_freq | + dev_priv->rps.efficient_freq | + dev_priv->rps.rp1_freq | + dev_priv->rps.min_freq) & 1, + "Odd GPU freq values\n"); + /* Preserve min/max settings in case of re-init */ if (dev_priv->rps.max_freq_softlimit == 0) dev_priv->rps.max_freq_softlimit = dev_priv->rps.max_freq; @@ -7566,6 +7576,7 @@ static int chv_freq_opcode(struct drm_i915_private *dev_priv, int val) return -1; } + /* CHV needs even values */ opcode = (DIV_ROUND_CLOSEST((val * 2 * mul), dev_priv->rps.cz_freq) * 2); return opcode; -- cgit v1.2.3 From 00e1e623e62cd8452e28633182b91ddcbb70cc7c Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Ville=20Syrj=C3=A4l=C3=A4?= Date: Wed, 27 Aug 2014 17:33:12 +0300 Subject: drm/i915: Init some CHV workarounds via LRIs in ring->init_context() MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Follow the BDW example and apply the workarounds touching registers which are saved in the context image through LRIs in the new ring->init_context() hook. This makes Mesa much happier and eg. glxgears doesn't hang after the first frame. Cc: Arun Siluvery Signed-off-by: Ville Syrjälä [danvet: Add missing wa table initialization to avoid a functional conflict with Arun's wa table debugfs support.] Reviewed-by: "Barbalho, Rafael" Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 14 -------------- 1 file changed, 14 deletions(-) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index 6bf533100dad..376cc2c8751a 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -6011,14 +6011,6 @@ static void cherryview_init_clock_gating(struct drm_device *dev) I915_WRITE(MI_ARB_VLV, MI_ARB_DISPLAY_TRICKLE_FEED_DISABLE); - /* WaDisablePartialInstShootdown:chv */ - I915_WRITE(GEN8_ROW_CHICKEN, - _MASKED_BIT_ENABLE(PARTIAL_INSTRUCTION_SHOOTDOWN_DISABLE)); - - /* WaDisableThreadStallDopClockGating:chv */ - I915_WRITE(GEN8_ROW_CHICKEN, - _MASKED_BIT_ENABLE(STALL_DOP_GATING_DISABLE)); - /* WaVSRefCountFullforceMissDisable:chv */ /* WaDSRefCountFullforceMissDisable:chv */ I915_WRITE(GEN7_FF_THREAD_MODE, @@ -6037,10 +6029,6 @@ static void cherryview_init_clock_gating(struct drm_device *dev) I915_WRITE(GEN8_UCGCTL6, I915_READ(GEN8_UCGCTL6) | GEN8_SDEUNIT_CLOCK_GATE_DISABLE); - /* WaDisableSamplerPowerBypass:chv (pre-production hw) */ - I915_WRITE(HALF_SLICE_CHICKEN3, - _MASKED_BIT_ENABLE(GEN8_SAMPLER_POWER_BYPASS_DIS)); - /* WaDisableGunitClockGating:chv (pre-production hw) */ I915_WRITE(VLV_GUNIT_CLOCK_GATE, I915_READ(VLV_GUNIT_CLOCK_GATE) | GINT_DIS); @@ -6050,8 +6038,6 @@ static void cherryview_init_clock_gating(struct drm_device *dev) _MASKED_BIT_ENABLE(GEN8_FF_DOP_CLOCK_GATE_DISABLE)); /* WaDisableDopClockGating:chv (pre-production hw) */ - I915_WRITE(GEN7_ROW_CHICKEN2, - _MASKED_BIT_ENABLE(DOP_CLOCK_GATING_DISABLE)); I915_WRITE(GEN6_UCGCTL1, I915_READ(GEN6_UCGCTL1) | GEN6_EU_TCUNIT_CLOCK_GATE_DISABLE); } -- cgit v1.2.3 From 9d53910580512207186729c94b39cca9a9778c13 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Ville=20Syrj=C3=A4l=C3=A4?= Date: Fri, 15 Aug 2014 01:21:53 +0300 Subject: drm/i915: Fix gen2 planes B and C max watermark value MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The max watermark value for gen2 planes B and C is 0x1f, instead of the 0x3f that plane A uses. Also check against the max even if the pipe is disabled since the FIFO size exceeds the plane B and C max watermark value. Signed-off-by: Ville Syrjälä Tested-by: Thomas Richter Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 24 ++++++++++++++++++++---- 1 file changed, 20 insertions(+), 4 deletions(-) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index 376cc2c8751a..38b61e4d17ae 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -1001,13 +1001,20 @@ static const struct intel_watermark_params i915_wm_info = { .guard_size = 2, .cacheline_size = I915_FIFO_LINE_SIZE, }; -static const struct intel_watermark_params i830_wm_info = { +static const struct intel_watermark_params i830_a_wm_info = { .fifo_size = I855GM_FIFO_SIZE, .max_wm = I915_MAX_WM, .default_wm = 1, .guard_size = 2, .cacheline_size = I830_FIFO_LINE_SIZE, }; +static const struct intel_watermark_params i830_bc_wm_info = { + .fifo_size = I855GM_FIFO_SIZE, + .max_wm = I915_MAX_WM/2, + .default_wm = 1, + .guard_size = 2, + .cacheline_size = I830_FIFO_LINE_SIZE, +}; static const struct intel_watermark_params i845_wm_info = { .fifo_size = I830_FIFO_SIZE, .max_wm = I915_MAX_WM, @@ -1689,7 +1696,7 @@ static void i9xx_update_wm(struct drm_crtc *unused_crtc) else if (!IS_GEN2(dev)) wm_info = &i915_wm_info; else - wm_info = &i830_wm_info; + wm_info = &i830_a_wm_info; fifo_size = dev_priv->display.get_fifo_size(dev, 0); crtc = intel_get_crtc_for_plane(dev, 0); @@ -1704,8 +1711,14 @@ static void i9xx_update_wm(struct drm_crtc *unused_crtc) wm_info, fifo_size, cpp, latency_ns); enabled = crtc; - } else + } else { planea_wm = fifo_size - wm_info->guard_size; + if (planea_wm > (long)wm_info->max_wm) + planea_wm = wm_info->max_wm; + } + + if (IS_GEN2(dev)) + wm_info = &i830_bc_wm_info; fifo_size = dev_priv->display.get_fifo_size(dev, 1); crtc = intel_get_crtc_for_plane(dev, 1); @@ -1723,8 +1736,11 @@ static void i9xx_update_wm(struct drm_crtc *unused_crtc) enabled = crtc; else enabled = NULL; - } else + } else { planeb_wm = fifo_size - wm_info->guard_size; + if (planeb_wm > (long)wm_info->max_wm) + planeb_wm = wm_info->max_wm; + } DRM_DEBUG_KMS("FIFO watermarks - A: %d, B: %d\n", planea_wm, planeb_wm); -- cgit v1.2.3 From 1038392b4dd02a46006d38f33aab71486d477ced Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Ville=20Syrj=C3=A4l=C3=A4?= Date: Fri, 15 Aug 2014 01:21:54 +0300 Subject: drm/i915: Disable trickle feed for gen2/3 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit My 830 is unhappy with trickle feed enabled. The symptom is that the image on the screen shifts a bit to right occasionally. The BIOS initially disables trickle feed, but it gets reset during suspend, so we need to re-disable it ourselves. Juse disable it always. Also disable it for all other gen2/3 platforms since we disable it for all more recent platforms as well (until HSW that is). At least my 855 doesn't seem to mind us doing this. I don't have gen3 hardware to test that. Signed-off-by: Ville Syrjälä Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 10 ++++++++++ 1 file changed, 10 insertions(+) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index 38b61e4d17ae..8541feb893f3 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -6138,6 +6138,9 @@ static void gen3_init_clock_gating(struct drm_device *dev) /* On GEN3 we really need to make sure the ARB C3 LP bit is set */ I915_WRITE(MI_ARB_STATE, _MASKED_BIT_ENABLE(MI_ARB_C3_LP_WRITE_ENABLE)); + + I915_WRITE(MI_ARB_STATE, + _MASKED_BIT_ENABLE(MI_ARB_DISPLAY_TRICKLE_FEED_DISABLE)); } static void i85x_init_clock_gating(struct drm_device *dev) @@ -6149,6 +6152,9 @@ static void i85x_init_clock_gating(struct drm_device *dev) /* interrupts should cause a wake up from C3 */ I915_WRITE(MI_STATE, _MASKED_BIT_ENABLE(MI_AGPBUSY_INT_EN) | _MASKED_BIT_DISABLE(MI_AGPBUSY_830_MODE)); + + I915_WRITE(MEM_MODE, + _MASKED_BIT_ENABLE(MEM_DISPLAY_TRICKLE_FEED_DISABLE)); } static void i830_init_clock_gating(struct drm_device *dev) @@ -6156,6 +6162,10 @@ static void i830_init_clock_gating(struct drm_device *dev) struct drm_i915_private *dev_priv = dev->dev_private; I915_WRITE(DSPCLK_GATE_D, OVRUNIT_CLOCK_GATE_DISABLE); + + I915_WRITE(MEM_MODE, + _MASKED_BIT_ENABLE(MEM_DISPLAY_A_TRICKLE_FEED_DISABLE) | + _MASKED_BIT_ENABLE(MEM_DISPLAY_B_TRICKLE_FEED_DISABLE)); } void intel_init_clock_gating(struct drm_device *dev) -- cgit v1.2.3 From 5aef600321caee7f950842b87f878780f693ca3a Mon Sep 17 00:00:00 2001 From: Chris Wilson Date: Wed, 3 Sep 2014 11:56:07 +0100 Subject: drm/i915: Rename global latency_ns variable MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit We use the variable name latency_ns in both the local lowlevel wm calculation routines and at the global level. Rename the global value to reduce shadow warnings and future confusion. Signed-off-by: Chris Wilson Reviewed-by: Ville Syrjälä Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 36 ++++++++++++++++++------------------ 1 file changed, 18 insertions(+), 18 deletions(-) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index 8541feb893f3..49af81f6b4ad 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -872,7 +872,7 @@ void intel_set_memory_cxsr(struct drm_i915_private *dev_priv, bool enable) * A value of 5us seems to be a good balance; safe for very low end * platforms but not overly aggressive on lower latency configs. */ -static const int latency_ns = 5000; +static const int pessimal_latency_ns = 5000; static int i9xx_get_fifo_size(struct drm_device *dev, int plane) { @@ -1387,14 +1387,14 @@ static void valleyview_update_wm(struct drm_crtc *crtc) vlv_update_drain_latency(crtc); if (g4x_compute_wm0(dev, PIPE_A, - &valleyview_wm_info, latency_ns, - &valleyview_cursor_wm_info, latency_ns, + &valleyview_wm_info, pessimal_latency_ns, + &valleyview_cursor_wm_info, pessimal_latency_ns, &planea_wm, &cursora_wm)) enabled |= 1 << PIPE_A; if (g4x_compute_wm0(dev, PIPE_B, - &valleyview_wm_info, latency_ns, - &valleyview_cursor_wm_info, latency_ns, + &valleyview_wm_info, pessimal_latency_ns, + &valleyview_cursor_wm_info, pessimal_latency_ns, &planeb_wm, &cursorb_wm)) enabled |= 1 << PIPE_B; @@ -1453,20 +1453,20 @@ static void cherryview_update_wm(struct drm_crtc *crtc) vlv_update_drain_latency(crtc); if (g4x_compute_wm0(dev, PIPE_A, - &valleyview_wm_info, latency_ns, - &valleyview_cursor_wm_info, latency_ns, + &valleyview_wm_info, pessimal_latency_ns, + &valleyview_cursor_wm_info, pessimal_latency_ns, &planea_wm, &cursora_wm)) enabled |= 1 << PIPE_A; if (g4x_compute_wm0(dev, PIPE_B, - &valleyview_wm_info, latency_ns, - &valleyview_cursor_wm_info, latency_ns, + &valleyview_wm_info, pessimal_latency_ns, + &valleyview_cursor_wm_info, pessimal_latency_ns, &planeb_wm, &cursorb_wm)) enabled |= 1 << PIPE_B; if (g4x_compute_wm0(dev, PIPE_C, - &valleyview_wm_info, latency_ns, - &valleyview_cursor_wm_info, latency_ns, + &valleyview_wm_info, pessimal_latency_ns, + &valleyview_cursor_wm_info, pessimal_latency_ns, &planec_wm, &cursorc_wm)) enabled |= 1 << PIPE_C; @@ -1559,14 +1559,14 @@ static void g4x_update_wm(struct drm_crtc *crtc) bool cxsr_enabled; if (g4x_compute_wm0(dev, PIPE_A, - &g4x_wm_info, latency_ns, - &g4x_cursor_wm_info, latency_ns, + &g4x_wm_info, pessimal_latency_ns, + &g4x_cursor_wm_info, pessimal_latency_ns, &planea_wm, &cursora_wm)) enabled |= 1 << PIPE_A; if (g4x_compute_wm0(dev, PIPE_B, - &g4x_wm_info, latency_ns, - &g4x_cursor_wm_info, latency_ns, + &g4x_wm_info, pessimal_latency_ns, + &g4x_cursor_wm_info, pessimal_latency_ns, &planeb_wm, &cursorb_wm)) enabled |= 1 << PIPE_B; @@ -1709,7 +1709,7 @@ static void i9xx_update_wm(struct drm_crtc *unused_crtc) adjusted_mode = &to_intel_crtc(crtc)->config.adjusted_mode; planea_wm = intel_calculate_wm(adjusted_mode->crtc_clock, wm_info, fifo_size, cpp, - latency_ns); + pessimal_latency_ns); enabled = crtc; } else { planea_wm = fifo_size - wm_info->guard_size; @@ -1731,7 +1731,7 @@ static void i9xx_update_wm(struct drm_crtc *unused_crtc) adjusted_mode = &to_intel_crtc(crtc)->config.adjusted_mode; planeb_wm = intel_calculate_wm(adjusted_mode->crtc_clock, wm_info, fifo_size, cpp, - latency_ns); + pessimal_latency_ns); if (enabled == NULL) enabled = crtc; else @@ -1827,7 +1827,7 @@ static void i845_update_wm(struct drm_crtc *unused_crtc) planea_wm = intel_calculate_wm(adjusted_mode->crtc_clock, &i845_wm_info, dev_priv->display.get_fifo_size(dev, 0), - 4, latency_ns); + 4, pessimal_latency_ns); fwater_lo = I915_READ(FW_BLC) & ~0xfff; fwater_lo |= (3<<8) | planea_wm; -- cgit v1.2.3 From 773538e86081d146e0020435d614f4b96996c1f9 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Ville=20Syrj=C3=A4l=C3=A4?= Date: Thu, 4 Sep 2014 14:54:56 +0300 Subject: drm/i915: Reset power sequencer pipe tracking when disp2d is off MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The power sequencer loses its state when the disp2d power well is down. Clear the dev_priv->pps_pipe tracking so that the power sequencer state gets reinitialized the next time it's needed. v2: Fix the pps_mutex vs. power_domain mutex deadlock by taking power domain reference first v3: Rename from edp_pps_(un)lock() to just pps_(un)lock() for the future, update due to backlight code changes Reviewed-by: Imre Deak Signed-off-by: Ville Syrjälä Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index 49af81f6b4ad..45f71e6dc544 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -6468,6 +6468,8 @@ static void vlv_display_power_well_disable(struct drm_i915_private *dev_priv, spin_unlock_irq(&dev_priv->irq_lock); vlv_set_power_well(dev_priv, power_well, false); + + vlv_power_sequencer_reset(dev_priv); } static void vlv_dpio_cmn_power_well_enable(struct drm_i915_private *dev_priv, -- cgit v1.2.3 From d6feb1962d08890080cbce080a3d73e1035f0a3d Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Ville=20Syrj=C3=A4l=C3=A4?= Date: Fri, 5 Sep 2014 21:54:13 +0300 Subject: drm/i915: Limit the watermark to at least 8 entries on gen2/3 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit 830 is very unhappy of the watermark value is too low (indicating a very high watermark in fact, ie. memory fetch will occur with an almost full FIFO). Limit the watermark value to at least 8 cache lines. That also matches the burst size we use on most platforms. BSpec seems to indicate we should limit the watermark to 'burst size + 1'. But on gen4 we already use a hardcoded 8 as the watermark value (as the spec says we should), so just use 8 as the limit on gen2/3 as well. Signed-off-by: Ville Syrjälä Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 11 +++++++++++ 1 file changed, 11 insertions(+) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index 45f71e6dc544..675e8a2ce988 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -1070,6 +1070,17 @@ static unsigned long intel_calculate_wm(unsigned long clock_in_khz, wm_size = wm->max_wm; if (wm_size <= 0) wm_size = wm->default_wm; + + /* + * Bspec seems to indicate that the value shouldn't be lower than + * 'burst size + 1'. Certainly 830 is quite unhappy with low values. + * Lets go for 8 which is the burst size since certain platforms + * already use a hardcoded 8 (which is what the spec says should be + * done). + */ + if (wm_size <= 8) + wm_size = 8; + return wm_size; } -- cgit v1.2.3 From 7526ed79b014cbd943cd48b8dda63e74391ddd48 Mon Sep 17 00:00:00 2001 From: Daniel Vetter Date: Mon, 29 Sep 2014 15:07:19 +0200 Subject: Revert "drm/i915/bdw: BDW Software Turbo" This reverts commit c76bb61a71083b2d90504cc6d0dda2047c5d63ca. It's apparently too broken so that Rodrigo submitted a patch to add a config option for it. Given that the design is also ... suboptimal and that I've only merged this to get lead engineers and managers off my back for one second let's just revert this. /me puts on combat gear again It was worth a shot ... References: http://mid.mail-archive.com/1411686380-1953-1-git-send-email-rodrigo.vivi@intel.com Cc: Jesse Barnes Cc: Daisy Sun Cc: Jesse Barnes Cc: Rodrigo Vivi Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 230 +++++++--------------------------------- 1 file changed, 39 insertions(+), 191 deletions(-) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index 675e8a2ce988..45f2aa0b8fe5 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -2285,6 +2285,7 @@ int ilk_wm_max_level(const struct drm_device *dev) else return 2; } + static void intel_print_wm_latency(struct drm_device *dev, const char *name, const uint16_t wm[5]) @@ -3253,9 +3254,6 @@ static void gen6_set_rps_thresholds(struct drm_i915_private *dev_priv, u8 val) { int new_power; - if (dev_priv->rps.is_bdw_sw_turbo) - return; - new_power = dev_priv->rps.power; switch (dev_priv->rps.power) { case LOW_POWER: @@ -3463,11 +3461,8 @@ void gen6_rps_idle(struct drm_i915_private *dev_priv) valleyview_set_rps(dev_priv->dev, dev_priv->rps.min_freq_softlimit); else if (IS_VALLEYVIEW(dev)) vlv_set_rps_idle(dev_priv); - else if (!dev_priv->rps.is_bdw_sw_turbo - || atomic_read(&dev_priv->rps.sw_turbo.flip_received)){ + else gen6_set_rps(dev_priv->dev, dev_priv->rps.min_freq_softlimit); - } - dev_priv->rps.last_adj = 0; } mutex_unlock(&dev_priv->rps.hw_lock); @@ -3481,11 +3476,8 @@ void gen6_rps_boost(struct drm_i915_private *dev_priv) if (dev_priv->rps.enabled) { if (IS_VALLEYVIEW(dev)) valleyview_set_rps(dev_priv->dev, dev_priv->rps.max_freq_softlimit); - else if (!dev_priv->rps.is_bdw_sw_turbo - || atomic_read(&dev_priv->rps.sw_turbo.flip_received)){ + else gen6_set_rps(dev_priv->dev, dev_priv->rps.max_freq_softlimit); - } - dev_priv->rps.last_adj = 0; } mutex_unlock(&dev_priv->rps.hw_lock); @@ -3520,26 +3512,21 @@ void valleyview_set_rps(struct drm_device *dev, u8 val) static void gen8_disable_rps_interrupts(struct drm_device *dev) { struct drm_i915_private *dev_priv = dev->dev_private; - if (IS_BROADWELL(dev) && dev_priv->rps.is_bdw_sw_turbo){ - if (atomic_read(&dev_priv->rps.sw_turbo.flip_received)) - del_timer(&dev_priv->rps.sw_turbo.flip_timer); - dev_priv-> rps.is_bdw_sw_turbo = false; - } else { - I915_WRITE(GEN6_PMINTRMSK, ~GEN8_PMINTR_REDIRECT_TO_NON_DISP); - I915_WRITE(GEN8_GT_IER(2), I915_READ(GEN8_GT_IER(2)) & - ~dev_priv->pm_rps_events); - /* Complete PM interrupt masking here doesn't race with the rps work - * item again unmasking PM interrupts because that is using a different - * register (GEN8_GT_IMR(2)) to mask PM interrupts. The only risk is in - * leaving stale bits in GEN8_GT_IIR(2) and GEN8_GT_IMR(2) which - * gen8_enable_rps will clean up. */ - spin_lock_irq(&dev_priv->irq_lock); - dev_priv->rps.pm_iir = 0; - spin_unlock_irq(&dev_priv->irq_lock); + I915_WRITE(GEN6_PMINTRMSK, ~GEN8_PMINTR_REDIRECT_TO_NON_DISP); + I915_WRITE(GEN8_GT_IER(2), I915_READ(GEN8_GT_IER(2)) & + ~dev_priv->pm_rps_events); + /* Complete PM interrupt masking here doesn't race with the rps work + * item again unmasking PM interrupts because that is using a different + * register (GEN8_GT_IMR(2)) to mask PM interrupts. The only risk is in + * leaving stale bits in GEN8_GT_IIR(2) and GEN8_GT_IMR(2) which + * gen8_enable_rps will clean up. */ - I915_WRITE(GEN8_GT_IIR(2), dev_priv->pm_rps_events); - } + spin_lock_irq(&dev_priv->irq_lock); + dev_priv->rps.pm_iir = 0; + spin_unlock_irq(&dev_priv->irq_lock); + + I915_WRITE(GEN8_GT_IIR(2), dev_priv->pm_rps_events); } static void gen6_disable_rps_interrupts(struct drm_device *dev) @@ -3697,111 +3684,13 @@ static void parse_rp_state_cap(struct drm_i915_private *dev_priv, u32 rp_state_c dev_priv->rps.min_freq_softlimit = dev_priv->rps.min_freq; } -static void bdw_sw_calculate_freq(struct drm_device *dev, - struct intel_rps_bdw_cal *c, u32 *cur_time, u32 *c0) -{ - struct drm_i915_private *dev_priv = dev->dev_private; - u64 busy = 0; - u32 busyness_pct = 0; - u32 elapsed_time = 0; - u16 new_freq = 0; - - if (!c || !cur_time || !c0) - return; - - if (0 == c->last_c0) - goto out; - - /* Check Evaluation interval */ - elapsed_time = *cur_time - c->last_ts; - if (elapsed_time < c->eval_interval) - return; - - mutex_lock(&dev_priv->rps.hw_lock); - - /* - * c0 unit in 32*1.28 usec, elapsed_time unit in 1 usec. - * Whole busyness_pct calculation should be - * busy = ((u64)(*c0 - c->last_c0) << 5 << 7) / 100; - * busyness_pct = (u32)(busy * 100 / elapsed_time); - * The final formula is to simplify CPU calculation - */ - busy = (u64)(*c0 - c->last_c0) << 12; - do_div(busy, elapsed_time); - busyness_pct = (u32)busy; - - if (c->is_up && busyness_pct >= c->it_threshold_pct) - new_freq = (u16)dev_priv->rps.cur_freq + 3; - if (!c->is_up && busyness_pct <= c->it_threshold_pct) - new_freq = (u16)dev_priv->rps.cur_freq - 1; - - /* Adjust to new frequency busyness and compare with threshold */ - if (0 != new_freq) { - if (new_freq > dev_priv->rps.max_freq_softlimit) - new_freq = dev_priv->rps.max_freq_softlimit; - else if (new_freq < dev_priv->rps.min_freq_softlimit) - new_freq = dev_priv->rps.min_freq_softlimit; - - gen6_set_rps(dev, new_freq); - } - - mutex_unlock(&dev_priv->rps.hw_lock); - -out: - c->last_c0 = *c0; - c->last_ts = *cur_time; -} - -static void gen8_set_frequency_RP0(struct work_struct *work) -{ - struct intel_rps_bdw_turbo *p_bdw_turbo = - container_of(work, struct intel_rps_bdw_turbo, work_max_freq); - struct intel_gen6_power_mgmt *p_power_mgmt = - container_of(p_bdw_turbo, struct intel_gen6_power_mgmt, sw_turbo); - struct drm_i915_private *dev_priv = - container_of(p_power_mgmt, struct drm_i915_private, rps); - - mutex_lock(&dev_priv->rps.hw_lock); - gen6_set_rps(dev_priv->dev, dev_priv->rps.rp0_freq); - mutex_unlock(&dev_priv->rps.hw_lock); -} - -static void flip_active_timeout_handler(unsigned long var) -{ - struct drm_i915_private *dev_priv = (struct drm_i915_private *) var; - - del_timer(&dev_priv->rps.sw_turbo.flip_timer); - atomic_set(&dev_priv->rps.sw_turbo.flip_received, false); - - queue_work(dev_priv->wq, &dev_priv->rps.sw_turbo.work_max_freq); -} - -void bdw_software_turbo(struct drm_device *dev) -{ - struct drm_i915_private *dev_priv = dev->dev_private; - - u32 current_time = I915_READ(TIMESTAMP_CTR); /* unit in usec */ - u32 current_c0 = I915_READ(MCHBAR_PCU_C0); /* unit in 32*1.28 usec */ - - bdw_sw_calculate_freq(dev, &dev_priv->rps.sw_turbo.up, - ¤t_time, ¤t_c0); - bdw_sw_calculate_freq(dev, &dev_priv->rps.sw_turbo.down, - ¤t_time, ¤t_c0); -} - static void gen8_enable_rps(struct drm_device *dev) { struct drm_i915_private *dev_priv = dev->dev_private; struct intel_engine_cs *ring; uint32_t rc6_mask = 0, rp_state_cap; - uint32_t threshold_up_pct, threshold_down_pct; - uint32_t ei_up, ei_down; /* up and down evaluation interval */ - u32 rp_ctl_flag; int unused; - /* Use software Turbo for BDW */ - dev_priv->rps.is_bdw_sw_turbo = IS_BROADWELL(dev); - /* 1a: Software RC state - RC0 */ I915_WRITE(GEN6_RC_STATE, 0); @@ -3845,74 +3734,35 @@ static void gen8_enable_rps(struct drm_device *dev) HSW_FREQUENCY(dev_priv->rps.rp1_freq)); I915_WRITE(GEN6_RC_VIDEO_FREQ, HSW_FREQUENCY(dev_priv->rps.rp1_freq)); - ei_up = 84480; /* 84.48ms */ - ei_down = 448000; - threshold_up_pct = 90; /* x percent busy */ - threshold_down_pct = 70; - - if (dev_priv->rps.is_bdw_sw_turbo) { - dev_priv->rps.sw_turbo.up.it_threshold_pct = threshold_up_pct; - dev_priv->rps.sw_turbo.up.eval_interval = ei_up; - dev_priv->rps.sw_turbo.up.is_up = true; - dev_priv->rps.sw_turbo.up.last_ts = 0; - dev_priv->rps.sw_turbo.up.last_c0 = 0; - - dev_priv->rps.sw_turbo.down.it_threshold_pct = threshold_down_pct; - dev_priv->rps.sw_turbo.down.eval_interval = ei_down; - dev_priv->rps.sw_turbo.down.is_up = false; - dev_priv->rps.sw_turbo.down.last_ts = 0; - dev_priv->rps.sw_turbo.down.last_c0 = 0; - - /* Start the timer to track if flip comes*/ - dev_priv->rps.sw_turbo.timeout = 200*1000; /* in us */ - - init_timer(&dev_priv->rps.sw_turbo.flip_timer); - dev_priv->rps.sw_turbo.flip_timer.function = flip_active_timeout_handler; - dev_priv->rps.sw_turbo.flip_timer.data = (unsigned long) dev_priv; - dev_priv->rps.sw_turbo.flip_timer.expires = - usecs_to_jiffies(dev_priv->rps.sw_turbo.timeout) + jiffies; - add_timer(&dev_priv->rps.sw_turbo.flip_timer); - INIT_WORK(&dev_priv->rps.sw_turbo.work_max_freq, gen8_set_frequency_RP0); - - atomic_set(&dev_priv->rps.sw_turbo.flip_received, true); - } else { - /* NB: Docs say 1s, and 1000000 - which aren't equivalent - * 1 second timeout*/ - I915_WRITE(GEN6_RP_DOWN_TIMEOUT, FREQ_1_28_US(1000000)); + /* NB: Docs say 1s, and 1000000 - which aren't equivalent */ + I915_WRITE(GEN6_RP_DOWN_TIMEOUT, 100000000 / 128); /* 1 second timeout */ - /* Docs recommend 900MHz, and 300 MHz respectively */ - I915_WRITE(GEN6_RP_INTERRUPT_LIMITS, - dev_priv->rps.max_freq_softlimit << 24 | - dev_priv->rps.min_freq_softlimit << 16); + /* Docs recommend 900MHz, and 300 MHz respectively */ + I915_WRITE(GEN6_RP_INTERRUPT_LIMITS, + dev_priv->rps.max_freq_softlimit << 24 | + dev_priv->rps.min_freq_softlimit << 16); - I915_WRITE(GEN6_RP_UP_THRESHOLD, - FREQ_1_28_US(ei_up * threshold_up_pct / 100)); - I915_WRITE(GEN6_RP_DOWN_THRESHOLD, - FREQ_1_28_US(ei_down * threshold_down_pct / 100)); - I915_WRITE(GEN6_RP_UP_EI, - FREQ_1_28_US(ei_up)); - I915_WRITE(GEN6_RP_DOWN_EI, - FREQ_1_28_US(ei_down)); + I915_WRITE(GEN6_RP_UP_THRESHOLD, 7600000 / 128); /* 76ms busyness per EI, 90% */ + I915_WRITE(GEN6_RP_DOWN_THRESHOLD, 31300000 / 128); /* 313ms busyness per EI, 70%*/ + I915_WRITE(GEN6_RP_UP_EI, 66000); /* 84.48ms, XXX: random? */ + I915_WRITE(GEN6_RP_DOWN_EI, 350000); /* 448ms, XXX: random? */ - I915_WRITE(GEN6_RP_IDLE_HYSTERSIS, 10); - } + I915_WRITE(GEN6_RP_IDLE_HYSTERSIS, 10); /* 5: Enable RPS */ - rp_ctl_flag = GEN6_RP_MEDIA_TURBO | - GEN6_RP_MEDIA_HW_NORMAL_MODE | - GEN6_RP_MEDIA_IS_GFX | - GEN6_RP_UP_BUSY_AVG | - GEN6_RP_DOWN_IDLE_AVG; - if (!dev_priv->rps.is_bdw_sw_turbo) - rp_ctl_flag |= GEN6_RP_ENABLE; - - I915_WRITE(GEN6_RP_CONTROL, rp_ctl_flag); - - /* 6: Ring frequency + overclocking - * (our driver does this later */ + I915_WRITE(GEN6_RP_CONTROL, + GEN6_RP_MEDIA_TURBO | + GEN6_RP_MEDIA_HW_NORMAL_MODE | + GEN6_RP_MEDIA_IS_GFX | + GEN6_RP_ENABLE | + GEN6_RP_UP_BUSY_AVG | + GEN6_RP_DOWN_IDLE_AVG); + + /* 6: Ring frequency + overclocking (our driver does this later */ + gen6_set_rps(dev, (I915_READ(GEN6_GT_PERF_STATUS) & 0xff00) >> 8); - if (!dev_priv->rps.is_bdw_sw_turbo) - gen8_enable_rps_interrupts(dev); + + gen8_enable_rps_interrupts(dev); gen6_gt_force_wake_put(dev_priv, FORCEWAKE_ALL); } @@ -5386,8 +5236,6 @@ static void intel_gen6_powersave_work(struct work_struct *work) rps.delayed_resume_work.work); struct drm_device *dev = dev_priv->dev; - dev_priv->rps.is_bdw_sw_turbo = false; - mutex_lock(&dev_priv->rps.hw_lock); if (IS_CHERRYVIEW(dev)) { -- cgit v1.2.3 From 67956867aa07c59d6d83628bbc9ee4bd9799a1e1 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Ville=20Syrj=C3=A4l=C3=A4?= Date: Tue, 2 Sep 2014 15:12:17 +0300 Subject: drm/i915: Don't spam dmesg with rps messages on vlv/chv MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit If the GPU frequency isn't going to change don't spam dmesg with debug messages about it. Signed-off-by: Ville Syrjälä Reviewed-by: Jani Nikula Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_pm.c | 13 +++++++------ 1 file changed, 7 insertions(+), 6 deletions(-) (limited to 'drivers/gpu/drm/i915/intel_pm.c') diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index 45f2aa0b8fe5..c27b6140bfd1 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -3491,17 +3491,18 @@ void valleyview_set_rps(struct drm_device *dev, u8 val) WARN_ON(val > dev_priv->rps.max_freq_softlimit); WARN_ON(val < dev_priv->rps.min_freq_softlimit); - DRM_DEBUG_DRIVER("GPU freq request from %d MHz (%u) to %d MHz (%u)\n", - vlv_gpu_freq(dev_priv, dev_priv->rps.cur_freq), - dev_priv->rps.cur_freq, - vlv_gpu_freq(dev_priv, val), val); - if (WARN_ONCE(IS_CHERRYVIEW(dev) && (val & 1), "Odd GPU freq value\n")) val &= ~1; - if (val != dev_priv->rps.cur_freq) + if (val != dev_priv->rps.cur_freq) { + DRM_DEBUG_DRIVER("GPU freq request from %d MHz (%u) to %d MHz (%u)\n", + vlv_gpu_freq(dev_priv, dev_priv->rps.cur_freq), + dev_priv->rps.cur_freq, + vlv_gpu_freq(dev_priv, val), val); + vlv_punit_write(dev_priv, PUNIT_REG_GPU_FREQ_REQ, val); + } I915_WRITE(GEN6_PMINTRMSK, gen6_rps_pm_mask(dev_priv, val)); -- cgit v1.2.3