Skip to contentSkip to conversation

Updated MIPI sequence for BYT

PATCH In review Shobhit Kumar 2 participants 11 messages Apr 9, 2014

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

cover Updated MIPI sequence for BYT

Hi,
The changes in DSI sequence are as suggested by HW and SV teams. Notable difference apart form few WAs is that for MIPI it is suggetsed that the PORT is enabled before PIPE and PLANE. The patch makes these changes. So few sequence changes, few workarounds and few new feature support like Clockstop.

A generic panel driver to enable MIPI is planned in next patchset

Known issue -
Today the upstream kernel does not have PMIC driver amd these patches works if UEFI BIOS enables MIPI and reuse BKL_EN, PANEL_EN from there, but during suspend/resume things will still fail.

Regards
Shobhit

Shobhit Kumar (7):
drm/i915: Program Rcomp and band gap reset everytime we resume from power gate drm/i915: Enable MIPI port before the plane and pipe enable drm/i915: Disable DPOunit clock gating
drm/i915: Parameterize the Clockstop and escape_clk_div
drm/i915: Panel commands can be sent only when clock is in LP11 drm/i915: Send DPI command explicitely in LP mode
drm/i915: Enable RANDOM resolution support for MIPI panels

 drivers/gpu/drm/i915/intel_dsi.c     | 125 +++++++++++++++++++++++++++--------
 drivers/gpu/drm/i915/intel_dsi.h     |   4 +-
 drivers/gpu/drm/i915/intel_dsi_cmd.c |   4 +-
 drivers/gpu/drm/i915/intel_dsi_cmd.h |   5 +-
 4 files changed, 108 insertions(+), 30 deletions(-)
Signature · 1 line
1.8.3.2
#1 lore raw

PATCH 1/7 drm/i915: Program Rcomp and band gap reset everytime we resume from power gate

---
 drivers/gpu/drm/i915/intel_dsi.c | 12 +++++++++---
 1 file changed, 9 insertions(+), 3 deletions(-)
drivers/gpu/drm/i915/intel_dsi.c +9 -3
index 3365664..7ceb8c6 100644
@@ -110,6 +110,15 @@ static void intel_dsi_device_ready(struct intel_encoder *encoder)
 
 	DRM_DEBUG_KMS("\n");
 
+	mutex_lock(&dev_priv->dpio_lock);
+	/* program rcomp for compliance, reduce from 50 ohms to 45 ohms
+	 * needed everytime after power gate */
+	vlv_flisdsi_write(dev_priv, 0x04, 0x0004);
+	mutex_unlock(&dev_priv->dpio_lock);
+
+	/* bandgap reset is needed after everytime we do power gate */
+	band_gap_reset(dev_priv);
+
 	val = I915_READ(MIPI_PORT_CTRL(pipe));
 	I915_WRITE(MIPI_PORT_CTRL(pipe), val | LP_OUTPUT_HOLD);
 	usleep_range(1000, 1500);
@@ -379,9 +388,6 @@ static void intel_dsi_mode_set(struct intel_encoder *intel_encoder)
 
 	DRM_DEBUG_KMS("pipe %c\n", pipe_name(pipe));
 
-	/* XXX: Location of the call */
-	band_gap_reset(dev_priv);
-
 	/* escape clock divider, 20MHz, shared for A and C. device ready must be
 	 * off when doing this! txclkesc? */
 	tmp = I915_READ(MIPI_CTRL(0));
Signature · 1 line
1.8.3.2
#2 lore raw

PATCH 2/7 drm/i915: Enable MIPI port before the plane and pipe enable

In reply to Shobhit Kumar

As per the hw team's recommendation we need to enable the MIPI port before enabling the plane and pipe. So call MIPI port enable in pre_enable phase itself

---
 drivers/gpu/drm/i915/intel_dsi.c | 51 ++++++++++++++++++++++++++--------------
 1 file changed, 33 insertions(+), 18 deletions(-)
drivers/gpu/drm/i915/intel_dsi.c +33 -18
index 7ceb8c6..569e6c6 100644
@@ -131,21 +131,6 @@ static void intel_dsi_device_ready(struct intel_encoder *encoder)
 	I915_WRITE(MIPI_DEVICE_READY(pipe), DEVICE_READY);
 	usleep_range(2000, 2500);
 }
-static void intel_dsi_pre_enable(struct intel_encoder *encoder)
-{
-	struct intel_dsi *intel_dsi = enc_to_intel_dsi(&encoder->base);
-
-	DRM_DEBUG_KMS("\n");
-
-	if (intel_dsi->dev.dev_ops->panel_reset)
-		intel_dsi->dev.dev_ops->panel_reset(&intel_dsi->dev);
-
-	/* put device in ready state */
-	intel_dsi_device_ready(encoder);
-
-	if (intel_dsi->dev.dev_ops->send_otp_cmds)
-		intel_dsi->dev.dev_ops->send_otp_cmds(&intel_dsi->dev);
-}
 
 static void intel_dsi_enable(struct intel_encoder *encoder)
 {
@@ -165,15 +150,45 @@ static void intel_dsi_enable(struct intel_encoder *encoder)
 		dpi_send_cmd(intel_dsi, TURN_ON);
 		msleep(100);
 
+		if (intel_dsi->dev.dev_ops->enable)
+			intel_dsi->dev.dev_ops->enable(&intel_dsi->dev);
+
 		/* assert ip_tg_enable signal */
 		temp = I915_READ(MIPI_PORT_CTRL(pipe)) & ~LANE_CONFIGURATION_MASK;
 		temp = temp | intel_dsi->port_bits;
 		I915_WRITE(MIPI_PORT_CTRL(pipe), temp | DPI_ENABLE);
 		POSTING_READ(MIPI_PORT_CTRL(pipe));
 	}
+}
+
+static void intel_dsi_pre_enable(struct intel_encoder *encoder)
+{
+	struct intel_dsi *intel_dsi = enc_to_intel_dsi(&encoder->base);
+
+	DRM_DEBUG_KMS("\n");
+
+	if (intel_dsi->dev.dev_ops->panel_reset)
+		intel_dsi->dev.dev_ops->panel_reset(&intel_dsi->dev);
+
+	/* put device in ready state */
+	intel_dsi_device_ready(encoder);
 
-	if (intel_dsi->dev.dev_ops->enable)
-		intel_dsi->dev.dev_ops->enable(&intel_dsi->dev);
+	if (intel_dsi->dev.dev_ops->send_otp_cmds)
+		intel_dsi->dev.dev_ops->send_otp_cmds(&intel_dsi->dev);
+
+	/* Enable port in pre-enable phase itself because as per hw team
+	 * recommendation, port should be enabled befor plane & pipe */
+	intel_dsi_enable(encoder);
+}
+
+static void intel_dsi_enable_nop(struct intel_encoder *encoder)
+{
+	DRM_DEBUG_KMS("\n");
+
+	/* for DSI port enable has to be done before pipe
+	 * and plane enable, so port enable is done in
+	 * pre_enable phase itself unlike other encoders
+	 */
 }
 
 static void intel_dsi_disable(struct intel_encoder *encoder)
@@ -600,7 +615,7 @@ bool intel_dsi_init(struct drm_device *dev)
 	intel_encoder->compute_config = intel_dsi_compute_config;
 	intel_encoder->pre_pll_enable = intel_dsi_pre_pll_enable;
 	intel_encoder->pre_enable = intel_dsi_pre_enable;
-	intel_encoder->enable = intel_dsi_enable;
+	intel_encoder->enable = intel_dsi_enable_nop;
 	intel_encoder->mode_set = intel_dsi_mode_set;
 	intel_encoder->disable = intel_dsi_disable;
 	intel_encoder->post_disable = intel_dsi_post_disable;
Signature · 1 line
1.8.3.2
#3 lore raw

PATCH 3/7 drm/i915: Disable DPOunit clock gating

In reply to Shobhit Kumar

Otherwise, this can stall pipe. We also need DPLL REFA always enabled

---
 drivers/gpu/drm/i915/intel_dsi.c | 26 ++++++++++++++++++++++++--
 1 file changed, 24 insertions(+), 2 deletions(-)
drivers/gpu/drm/i915/intel_dsi.c +24 -2
index 569e6c6..4793a5b 100644
@@ -163,16 +163,31 @@ static void intel_dsi_enable(struct intel_encoder *encoder)
 
 static void intel_dsi_pre_enable(struct intel_encoder *encoder)
 {
+	struct drm_device *dev = encoder->base.dev;
+	struct drm_i915_private *dev_priv = dev->dev_private;
 	struct intel_dsi *intel_dsi = enc_to_intel_dsi(&encoder->base);
+	struct intel_crtc *intel_crtc = to_intel_crtc(encoder->base.crtc);
+	enum pipe pipe = intel_crtc->pipe;
+	u32 tmp;
 
 	DRM_DEBUG_KMS("\n");
 
-	if (intel_dsi->dev.dev_ops->panel_reset)
-		intel_dsi->dev.dev_ops->panel_reset(&intel_dsi->dev);
+	/* Disable DPOunit clock gating, can stall pipe
+	 * and we need DPLL REFA always enabled */
+	tmp = I915_READ(DPLL(pipe));
+	tmp |= DPLL_REFA_CLK_ENABLE_VLV;
+	I915_WRITE(DPLL(pipe), tmp);
+
+	tmp = I915_READ(DSPCLK_GATE_D);
+	tmp |= DPOUNIT_CLOCK_GATE_DISABLE;
+	I915_WRITE(DSPCLK_GATE_D, tmp);
 
 	/* put device in ready state */
 	intel_dsi_device_ready(encoder);
 
+	if (intel_dsi->dev.dev_ops->panel_reset)
+		intel_dsi->dev.dev_ops->panel_reset(&intel_dsi->dev);
+
 	if (intel_dsi->dev.dev_ops->send_otp_cmds)
 		intel_dsi->dev.dev_ops->send_otp_cmds(&intel_dsi->dev);
 
@@ -251,14 +266,21 @@ static void intel_dsi_clear_device_ready(struct intel_encoder *encoder)
 
 	vlv_disable_dsi_pll(encoder);
 }
+
 static void intel_dsi_post_disable(struct intel_encoder *encoder)
 {
+	struct drm_i915_private *dev_priv = encoder->base.dev->dev_private;
 	struct intel_dsi *intel_dsi = enc_to_intel_dsi(&encoder->base);
+	u32 val;
 
 	DRM_DEBUG_KMS("\n");
 
 	intel_dsi_clear_device_ready(encoder);
 
+	val = I915_READ(DSPCLK_GATE_D);
+	val &= ~DPOUNIT_CLOCK_GATE_DISABLE;
+	I915_WRITE(DSPCLK_GATE_D, val);
+
 	if (intel_dsi->dev.dev_ops->disable_panel_power)
 		intel_dsi->dev.dev_ops->disable_panel_power(&intel_dsi->dev);
 }
Signature · 1 line
1.8.3.2
#4 lore raw

PATCH 4/7 drm/i915: Parameterize the Clockstop and escape_clk_div

In reply to Shobhit Kumar

In preparation for Generic driver

---
 drivers/gpu/drm/i915/intel_dsi.c | 11 +++++++++--
 drivers/gpu/drm/i915/intel_dsi.h |  4 +++-
 2 files changed, 12 insertions(+), 3 deletions(-)
drivers/gpu/drm/i915/intel_dsi.c +9 -2
index 4793a5b..dfcdb10 100644
@@ -495,10 +495,17 @@ static void intel_dsi_mode_set(struct intel_encoder *intel_encoder)
 	/* dphy stuff */
 
 	/* in terms of low power clock */
-	I915_WRITE(MIPI_INIT_COUNT(pipe), txclkesc(ESCAPE_CLOCK_DIVIDER_1, 100));
+	I915_WRITE(MIPI_INIT_COUNT(pipe), txclkesc(intel_dsi->escape_clk_div, 100));
+
+	val = 0;
+	if (intel_dsi->eotp_pkt == 0)
+		val |= EOT_DISABLE;
+
+	if (intel_dsi->clock_stop)
+		val |= CLOCKSTOP;
 
 	/* recovery disables */
-	I915_WRITE(MIPI_EOT_DISABLE(pipe), intel_dsi->eot_disable);
+	I915_WRITE(MIPI_EOT_DISABLE(pipe), val);
 
 	/* in terms of txbyteclkhs. actual high to low switch +
 	 * MIPI_STOP_STATE_STALL * MIPI_LP_BYTECLK.
drivers/gpu/drm/i915/intel_dsi.h +3 -1
index b4a27ce..550714c 100644
@@ -95,8 +95,10 @@ struct intel_dsi {
 	u32 video_mode_format;
 
 	/* eot for MIPI_EOT_DISABLE register */
-	u32 eot_disable;
+	u8 eotp_pkt;
+	u8 clock_stop;
 
+	u8 escape_clk_div;
 	u32 port_bits;
 	u32 bw_timer;
 	u32 dphy_reg;
Signature · 1 line
1.8.3.2
#5 lore raw

PATCH 5/7 drm/i915: Panel commands can be sent only when clock is in LP11

In reply to Shobhit Kumar
---
 drivers/gpu/drm/i915/intel_dsi.c | 17 +++++++++++++++++
 1 file changed, 17 insertions(+)
drivers/gpu/drm/i915/intel_dsi.c +17 -0
index dfcdb10..d8eccda 100644
@@ -229,6 +229,23 @@ static void intel_dsi_disable(struct intel_encoder *encoder)
 		msleep(2);
 	}
 
+	/* Panel commands can be sent when clock is in LP11 */
+	I915_WRITE(MIPI_DEVICE_READY(pipe), 0x0);
+
+	temp = I915_READ(MIPI_CTRL(pipe));
+	temp &= ~ESCAPE_CLOCK_DIVIDER_MASK;
+	I915_WRITE(MIPI_CTRL(pipe), temp |
+			intel_dsi->escape_clk_div <<
+			ESCAPE_CLOCK_DIVIDER_SHIFT);
+
+	I915_WRITE(MIPI_EOT_DISABLE(pipe), CLOCKSTOP);
+
+	temp = I915_READ(MIPI_DSI_FUNC_PRG(pipe));
+	temp &= ~VID_MODE_FORMAT_MASK;
+	I915_WRITE(MIPI_DSI_FUNC_PRG(pipe), temp);
+
+	I915_WRITE(MIPI_DEVICE_READY(pipe), 0x1);
+
 	/* if disable packets are sent before sending shutdown packet then in
 	 * some next enable sequence send turn on packet error is observed */
 	if (intel_dsi->dev.dev_ops->disable)
Signature · 1 line
1.8.3.2
#6 lore raw

PATCH 6/7 drm/i915: Send DPI command explicitely in LP mode

In reply to Shobhit Kumar

Though HS mode also should work.

v2: Change parameter as "bool hs" as suggested by Jani

---
 drivers/gpu/drm/i915/intel_dsi.c     | 5 +++--
 drivers/gpu/drm/i915/intel_dsi_cmd.c | 4 ++--
 drivers/gpu/drm/i915/intel_dsi_cmd.h | 5 ++++-
 3 files changed, 9 insertions(+), 5 deletions(-)
drivers/gpu/drm/i915/intel_dsi.c +3 -2
index d8eccda..b5948b7 100644
@@ -147,7 +147,7 @@ static void intel_dsi_enable(struct intel_encoder *encoder)
 		I915_WRITE(MIPI_MAX_RETURN_PKT_SIZE(pipe), 8 * 4);
 	else {
 		msleep(20); /* XXX */
-		dpi_send_cmd(intel_dsi, TURN_ON);
+		dpi_send_cmd(intel_dsi, TURN_ON, DPI_LP_MODE_EN);
 		msleep(100);
 
 		if (intel_dsi->dev.dev_ops->enable)
@@ -218,7 +218,8 @@ static void intel_dsi_disable(struct intel_encoder *encoder)
 	DRM_DEBUG_KMS("\n");
 
 	if (is_vid_mode(intel_dsi)) {
-		dpi_send_cmd(intel_dsi, SHUTDOWN);
+		/* Send Shutdown command to the panel in LP mode */
+		dpi_send_cmd(intel_dsi, SHUTDOWN, DPI_LP_MODE_EN);
 		msleep(10);
 
 		/* de-assert ip_tg_enable signal */
drivers/gpu/drm/i915/intel_dsi_cmd.c +2 -2
index 7c40f98..3eeb21b 100644
@@ -389,7 +389,7 @@ int dsi_vc_generic_read(struct intel_dsi *intel_dsi, int channel,
  *
  * XXX: commands with data in MIPI_DPI_DATA?
  */
-int dpi_send_cmd(struct intel_dsi *intel_dsi, u32 cmd)
+int dpi_send_cmd(struct intel_dsi *intel_dsi, u32 cmd, bool hs)
 {
 	struct drm_encoder *encoder = &intel_dsi->base.base;
 	struct drm_device *dev = encoder->dev;
@@ -399,7 +399,7 @@ int dpi_send_cmd(struct intel_dsi *intel_dsi, u32 cmd)
 	u32 mask;
 
 	/* XXX: pipe, hs */
-	if (intel_dsi->hs)
+	if (hs)
 		cmd &= ~DPI_LP_MODE;
 	else
 		cmd |= DPI_LP_MODE;
drivers/gpu/drm/i915/intel_dsi_cmd.h +4 -1
index 54c8a23..9a18cbf 100644
@@ -33,6 +33,9 @@
 #include "intel_drv.h"
 #include "intel_dsi.h"
 
+#define DPI_LP_MODE_EN	false
+#define DPI_HS_MODE_EN	true
+
 void dsi_hs_mode_enable(struct intel_dsi *intel_dsi, bool enable);
 
 int dsi_vc_dcs_write(struct intel_dsi *intel_dsi, int channel,
@@ -47,7 +50,7 @@ int dsi_vc_dcs_read(struct intel_dsi *intel_dsi, int channel, u8 dcs_cmd,
 int dsi_vc_generic_read(struct intel_dsi *intel_dsi, int channel,
 			u8 *reqdata, int reqlen, u8 *buf, int buflen);
 
-int dpi_send_cmd(struct intel_dsi *intel_dsi, u32 cmd);
+int dpi_send_cmd(struct intel_dsi *intel_dsi, u32 cmd, bool hs);
 
 /* XXX: questionable write helpers */
 static inline int dsi_vc_dcs_write_0(struct intel_dsi *intel_dsi,
Signature · 1 line
1.8.3.2
#7 lore raw

PATCH 7/7 drm/i915: Enable RANDOM resolution support for MIPI panels

In reply to Shobhit Kumar

Some MIPI panels might not have resolution which is a multiple of 64 like 1366x768. Enable this feature for such panels by default

---
 drivers/gpu/drm/i915/intel_dsi.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
drivers/gpu/drm/i915/intel_dsi.c +6 -1
index b5948b7..4e271c7 100644
@@ -552,9 +552,14 @@ static void intel_dsi_mode_set(struct intel_encoder *intel_encoder)
 		   intel_dsi->clk_hs_to_lp_count << HS_LP_PWR_SW_CNT_SHIFT);
 
 	if (is_vid_mode(intel_dsi))
+		/* Some panels might have resolution which is not a multiple of
+		 * 64 like 1366 x 768. Enable RANDOM resolution support for such
+		 * panels by default */
 		I915_WRITE(MIPI_VIDEO_MODE_FORMAT(pipe),
 				intel_dsi->video_frmt_cfg_bits |
-				intel_dsi->video_mode_format);
+				intel_dsi->video_mode_format |
+				IP_TG_CONFIG |
+				RANDOM_DPI_DISPLAY_RESOLUTION);
 }
 
 static enum drm_connector_status
Signature · 1 line
1.8.3.2
#8 lore raw
In reply to Shobhit Kumar
Quoted 7 lines · Shobhit Kumar

Hi,
The changes in DSI sequence are as suggested by HW and SV teams. Notable difference apart form few WAs is that for MIPI it is suggetsed that the PORT is enabled before PIPE and PLANE. The patch makes these changes. So few sequence changes, few workarounds and few new feature support like Clockstop.

Quoted 6 lines

A generic panel driver to enable MIPI is planned in next patchset

Known issue -
Today the upstream kernel does not have PMIC driver amd these patches works if UEFI BIOS enables MIPI and reuse BKL_EN, PANEL_EN from there, but during suspend/resume things will still fail.

Thanks a lot for the patches&review, looking forward to the next round. Hopefully that one will fix our mipi/dsi code to Just Work (tm) so that we can close the asus 100t bug on kernel.org:

https://bugzilla.kernel.org/show_bug.cgi?id=68451

Cheers, Daniel

Quoted 26 lines

Regards
Shobhit

Shobhit Kumar (7):
drm/i915: Program Rcomp and band gap reset everytime we resume from power gate drm/i915: Enable MIPI port before the plane and pipe enable drm/i915: Disable DPOunit clock gating
drm/i915: Parameterize the Clockstop and escape_clk_div
drm/i915: Panel commands can be sent only when clock is in LP11 drm/i915: Send DPI command explicitely in LP mode
drm/i915: Enable RANDOM resolution support for MIPI panels

 drivers/gpu/drm/i915/intel_dsi.c     | 125 +++++++++++++++++++++++++++--------
 drivers/gpu/drm/i915/intel_dsi.h     |   4 +-
 drivers/gpu/drm/i915/intel_dsi_cmd.c |   4 +-
 drivers/gpu/drm/i915/intel_dsi_cmd.h |   5 +-
 4 files changed, 108 insertions(+), 30 deletions(-)
Signature · 6 lines
1.8.3.2

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 3 lines
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch
#9 lore raw
Quoted 15 lines · Daniel Vetter
Quoted 7 lines · Shobhit Kumar

Hi,
The changes in DSI sequence are as suggested by HW and SV teams. Notable difference apart form few WAs is that for MIPI it is suggetsed that the PORT is enabled before PIPE and PLANE. The patch makes these changes. So few sequence changes, few workarounds and few new feature support like Clockstop.

Quoted 6 lines

A generic panel driver to enable MIPI is planned in next patchset

Known issue -
Today the upstream kernel does not have PMIC driver amd these patches works if UEFI BIOS enables MIPI and reuse BKL_EN, PANEL_EN from there, but during suspend/resume things will still fail.

Quoted 5 lines

Thanks a lot for the patches&review, looking forward to the next round. Hopefully that one will fix our mipi/dsi code to Just Work (tm) so that we can close the asus 100t bug on kernel.org:

https://bugzilla.kernel.org/show_bug.cgi?id=68451

Thanks Daniel. Now the panel driver in next patch will definitely get the asus 100t booting at-least as I have got the patches already verified by those who have the device.

Regards
Shobhit

#10 lore raw
Quoted 20 lines · Shobhit
Quoted 19 lines · Daniel Vetter
Quoted 7 lines · Shobhit Kumar

Hi,
The changes in DSI sequence are as suggested by HW and SV teams. Notable difference apart form few WAs is that for MIPI it is suggetsed that the PORT is enabled before PIPE and PLANE. The patch makes these changes. So few sequence changes, few workarounds and few new feature support like Clockstop.

Quoted 6 lines

A generic panel driver to enable MIPI is planned in next patchset

Known issue -
Today the upstream kernel does not have PMIC driver amd these patches works if UEFI BIOS enables MIPI and reuse BKL_EN, PANEL_EN from there, but during suspend/resume things will still fail.

Thanks a lot for the patches&review, looking forward to the next round. Hopefully that one will fix our mipi/dsi code to Just Work (tm) so that we can close the asus 100t bug on kernel.org:

Quoted 5 lines
Quoted 1 line

https://bugzilla.kernel.org/show_bug.cgi?id=68451

Thanks Daniel. Now the panel driver in next patch will definitely get the asus 100t booting at-least as I have got the patches already verified by those who have the device.

Awesome, that should make a lot of people happy!

Cheers, Daniel

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