Skip to contentSkip to conversation

Three pipe support for IVB

RFC In review Jesse Barnes 6 participants 22 messages Oct 5, 2011 – Oct 6, 2011

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

This set of patches lays the groundwork for supporting 3 pipes on IVB hardware. We've only tested it so far with 2 HDMI plus 1 VGA (a configuration that doesn't work), and are awaiting 3 DP port hardware (the officially supported config) to complete our testing.

So this patchset is necessary but may not be sufficient (once we have 3 DP hardware we can debug and see). Review and comments appreciated.

Thanks,
Jesse

#1 lore raw

PATCH 1/4 drm/i915: PLL macro cleanup and pipe assertion check

Add a macro for accessing the two pipe PLLs and add a check to make sure we don't access a non-existent one in the enable/disable functions.

---
 drivers/gpu/drm/i915/i915_reg.h      |    6 +++---
 drivers/gpu/drm/i915/intel_display.c |    6 ++++++
 2 files changed, 9 insertions(+), 3 deletions(-)
drivers/gpu/drm/i915/i915_reg.h +3 -3
index 542453f..7659075 100644
@@ -2945,15 +2945,15 @@
 
 #define _PCH_DPLL_A              0xc6014
 #define _PCH_DPLL_B              0xc6018
-#define PCH_DPLL(pipe) _PIPE(pipe, _PCH_DPLL_A, _PCH_DPLL_B)
+#define PCH_DPLL(pipe) (pipe == 0 ?  _PCH_DPLL_A : _PCH_DPLL_B)
 
 #define _PCH_FPA0                0xc6040
 #define  FP_CB_TUNE		(0x3<<22)
 #define _PCH_FPA1                0xc6044
 #define _PCH_FPB0                0xc6048
 #define _PCH_FPB1                0xc604c
-#define PCH_FP0(pipe) _PIPE(pipe, _PCH_FPA0, _PCH_FPB0)
-#define PCH_FP1(pipe) _PIPE(pipe, _PCH_FPA1, _PCH_FPB1)
+#define PCH_FP0(pipe) (pipe == 0 ? _PCH_FPA0 : _PCH_FPB0)
+#define PCH_FP1(pipe) (pipe == 0 ? _PCH_FPA1 : _PCH_FPB1)
 
 #define PCH_DPLL_TEST           0xc606c
 
drivers/gpu/drm/i915/intel_display.c +6 -0
index 56a8554..df9d394 100644
@@ -1171,6 +1171,9 @@ static void intel_enable_pch_pll(struct drm_i915_private *dev_priv,
 	int reg;
 	u32 val;
 
+	if (pipe > 1)
+		return;
+
 	/* PCH only available on ILK+ */
 	BUG_ON(dev_priv->info->gen < 5);
 
@@ -1191,6 +1194,9 @@ static void intel_disable_pch_pll(struct drm_i915_private *dev_priv,
 	int reg;
 	u32 val;
 
+	if (pipe > 1)
+		return;
+
 	/* PCH only available on ILK+ */
 	BUG_ON(dev_priv->info->gen < 5);
 
Signature · 1 line
1.7.4.1
#2 lore raw

Re: [PATCH 1/4] drm/i915: PLL macro cleanup and pipe assertion check

Quoted 2 lines · On Wed, 5 Oct 2011 10:25:18 -0700 Jesse Barnes <[email protected]>…

Add a macro for accessing the two pipe PLLs and add a check to make sure we don't access a non-existent one in the enable/disable functions.

This reminds me that the IVB 3 pipe support has a big limitation: two of the pipes need to share pipe PLLs. That means they have to be the same output port type and share timings, thus the two DP or identical HDMI config implied in my earlier mail.

Signature · 1 line
Jesse Barnes, Intel Open Source Technology Center

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

#3 lore raw

Re: [PATCH 1/4] drm/i915: PLL macro cleanup and pipe assertion check

In reply to Jesse Barnes
Quoted 2 lines · Jesse Barnes

Add a macro for accessing the two pipe PLLs and add a check to make sure we don't access a non-existent one in the enable/disable functions.

So, I haven't read through the specs, but are the PLLs tied to specific pipes? Or are we really talking about a different object here, and just mis-naming it as a 'pipe'?

Signature · 1 line
[email protected]

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

#4 lore raw

Re: [PATCH 1/4] drm/i915: PLL macro cleanup and pipe assertion check

Quoted 8 lines · On Wed, 05 Oct 2011 11:07:53 -0700 Keith Packard <[email protected]> wrote:
Quoted 2 lines · Jesse Barnes

Add a macro for accessing the two pipe PLLs and add a check to make sure we don't access a non-existent one in the enable/disable functions.

So, I haven't read through the specs, but are the PLLs tied to specific pipes? Or are we really talking about a different object here, and just mis-naming it as a 'pipe'?

They're not hardwired; we can use the DPLL_SEL reg to control which PLL drives which transcoder. Unfortunately our old pipe terminology is a bit outdated with the PCH split. It used to be the pipe was fairly
self-contained: a PLL with pipe and plane controls. Now it's split across the CPU and PCH with FDI in between (sometimes) and transcoder configuration (sometimes with DP).

Signature · 1 line
Jesse Barnes, Intel Open Source Technology Center

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

#5 lore raw

Re: [PATCH 1/4] drm/i915: PLL macro cleanup and pipe assertion check

In reply to Jesse Barnes
Quoted 14 lines · Jesse Barnes

Add a macro for accessing the two pipe PLLs and add a check to make sure we don't access a non-existent one in the enable/disable functions.

---
 drivers/gpu/drm/i915/i915_reg.h      |    6 +++---
 drivers/gpu/drm/i915/intel_display.c |    6 ++++++
 2 files changed, 9 insertions(+), 3 deletions(-)
drivers/gpu/drm/i915/i915_reg.h +0 -0
index 542453f..7659075 100644

drivers/gpu/drm/i915/i915_reg.h line 2948

Quoted 5 lines
drivers/gpu/drm/i915/i915_reg.h +1 -1
@@ -2945,15 +2945,15 @@
 
 #define _PCH_DPLL_A              0xc6014
 #define _PCH_DPLL_B              0xc6018
-#define PCH_DPLL(pipe) _PIPE(pipe, _PCH_DPLL_A, _PCH_DPLL_B)
+#define PCH_DPLL(pipe) (pipe == 0 ?  _PCH_DPLL_A : _PCH_DPLL_B)

If you check that pipe != 2 elsewhere why change these macros? What might be a good addition is to use a compile-time assert in case we do feed these an invalid enum pipe.
-Chris

Signature · 1 line
Chris Wilson, Intel Open Source Technology Centre
#6 lore raw

Re: [PATCH 1/4] drm/i915: PLL macro cleanup and pipe assertion check

Quoted 16 lines · On Wed, 05 Oct 2011 19:43:53 +0100 Chris Wilson <[email protected]>…
Quoted 15 lines · Jesse Barnes

Add a macro for accessing the two pipe PLLs and add a check to make sure we don't access a non-existent one in the enable/disable functions.

---
 drivers/gpu/drm/i915/i915_reg.h      |    6 +++---
 drivers/gpu/drm/i915/intel_display.c |    6 ++++++
 2 files changed, 9 insertions(+), 3 deletions(-)
drivers/gpu/drm/i915/i915_reg.h +0 -0
index 542453f..7659075 100644
@@ -2945,15 +2945,15 @@
 
Quoted 8 lines

drivers/gpu/drm/i915/i915_reg.h line 2948

Quoted 4 lines
drivers/gpu/drm/i915/i915_reg.h +1 -1
 #define _PCH_DPLL_A              0xc6014
 #define _PCH_DPLL_B              0xc6018
-#define PCH_DPLL(pipe) _PIPE(pipe, _PCH_DPLL_A, _PCH_DPLL_B)
+#define PCH_DPLL(pipe) (pipe == 0 ?  _PCH_DPLL_A : _PCH_DPLL_B)

If you check that pipe != 2 elsewhere why change these macros? What might be a good addition is to use a compile-time assert in case we do feed these an invalid enum pipe.

Belt & suspenders; I added the checks and fixes at different times too. But a compile time check would be good for this one. In future PCH parts we'll have more PLLs to drive pipes, but they'll be at a different offset so we'll need new macro anyway.

Signature · 1 line
Jesse Barnes, Intel Open Source Technology Center

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

#7 lore raw

PATCH 2/4 drm/i915: support 3 pipes on IVB+

In reply to Jesse Barnes

Well almost anyway. IVB has 3 planes, pipes, transcoders, and FDI interfaces, but only 2 pipe PLLs. So two of the pipes must use the same pipe timings (e.g. 2 DP plus one other, or two HDMI with the same mode and one other, etc.).

---
 drivers/gpu/drm/i915/i915_dma.c      |    4 +++-
 drivers/gpu/drm/i915/i915_drv.h      |    4 ++--
 drivers/gpu/drm/i915/intel_display.c |    7 +++++++
 drivers/gpu/drm/i915/intel_dp.c      |    2 +-
 drivers/gpu/drm/i915/intel_hdmi.c    |    2 +-
 drivers/gpu/drm/i915/intel_lvds.c    |    8 +++++---
 drivers/gpu/drm/i915/intel_sdvo.c    |    2 +-
 7 files changed, 20 insertions(+), 9 deletions(-)
drivers/gpu/drm/i915/i915_dma.c +3 -1
index 8a3942c..7a2ee0d 100644
@@ -2035,7 +2035,9 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags)
 	spin_lock_init(&dev_priv->error_lock);
 	spin_lock_init(&dev_priv->rps_lock);
 
-	if (IS_MOBILE(dev) || !IS_GEN2(dev))
+	if (IS_IVYBRIDGE(dev))
+		dev_priv->num_pipe = 3;
+	else if (IS_MOBILE(dev) || !IS_GEN2(dev))
 		dev_priv->num_pipe = 2;
 	else
 		dev_priv->num_pipe = 1;
drivers/gpu/drm/i915/i915_drv.h +2 -2
index 7916bd9..e865a3c 100644
@@ -674,8 +674,8 @@ typedef struct drm_i915_private {
 	u32 pch_pf_pos, pch_pf_size;
 	int panel_t3, panel_t12;
 
-	struct drm_crtc *plane_to_crtc_mapping[2];
-	struct drm_crtc *pipe_to_crtc_mapping[2];
+	struct drm_crtc *plane_to_crtc_mapping[3];
+	struct drm_crtc *pipe_to_crtc_mapping[3];
 	wait_queue_head_t pending_flip_queue;
 	bool flip_pending_is_done;
 
drivers/gpu/drm/i915/intel_display.c +7 -0
index df9d394..0dffb52 100644
@@ -2083,6 +2083,7 @@ static int ironlake_update_plane(struct drm_crtc *crtc,
 	switch (plane) {
 	case 0:
 	case 1:
+	case 2:
 		break;
 	default:
 		DRM_ERROR("Can't update plane %d in SAREA\n", plane);
@@ -2182,6 +2183,10 @@ intel_pipe_set_base(struct drm_crtc *crtc, int x, int y,
 	case 0:
 	case 1:
 		break;
+	case 2:
+		if (IS_IVYBRIDGE(dev))
+			break;
+		/* fall through otherwise */
 	default:
 		DRM_ERROR("no plane for crtc\n");
 		return -EINVAL;
@@ -2878,6 +2883,8 @@ static void ironlake_pch_enable(struct drm_crtc *crtc)
 			temp |= (TRANSA_DPLL_ENABLE | TRANSA_DPLLA_SEL);
 		else if (pipe == 1 && (temp & TRANSB_DPLL_ENABLE) == 0)
 			temp |= (TRANSB_DPLL_ENABLE | TRANSB_DPLLB_SEL);
+		else if (pipe == 2 && (temp & TRANSC_DPLL_ENABLE) == 0)
+			temp |= (TRANSC_DPLL_ENABLE | TRANSC_DPLLB_SEL);
 		I915_WRITE(PCH_DPLL_SEL, temp);
 	}
 
drivers/gpu/drm/i915/intel_dp.c +1 -1
index 44fef5e..edc369e 100644
@@ -1996,7 +1996,7 @@ intel_dp_init(struct drm_device *dev, int output_reg)
 	if (is_edp(intel_dp))
 		intel_encoder->clone_mask = (1 << INTEL_EDP_CLONE_BIT);
 
-	intel_encoder->crtc_mask = (1 << 0) | (1 << 1);
+	intel_encoder->crtc_mask = (1 << 0) | (1 << 1) | (1 << 2);
 	connector->interlace_allowed = true;
 	connector->doublescan_allowed = 0;
 
drivers/gpu/drm/i915/intel_hdmi.c +1 -1
index 226ba83..11883c9 100644
@@ -511,7 +511,7 @@ void intel_hdmi_init(struct drm_device *dev, int sdvox_reg)
 	connector->polled = DRM_CONNECTOR_POLL_HPD;
 	connector->interlace_allowed = 0;
 	connector->doublescan_allowed = 0;
-	intel_encoder->crtc_mask = (1 << 0) | (1 << 1);
+	intel_encoder->crtc_mask = (1 << 0) | (1 << 1) | (1 << 2);
 
 	/* Set up the DDC bus. */
 	if (sdvox_reg == SDVOB) {
drivers/gpu/drm/i915/intel_lvds.c +5 -3
index 31da77f..42f165a 100644
@@ -888,9 +888,11 @@ bool intel_lvds_init(struct drm_device *dev)
 	intel_encoder->type = INTEL_OUTPUT_LVDS;
 
 	intel_encoder->clone_mask = (1 << INTEL_LVDS_CLONE_BIT);
-	intel_encoder->crtc_mask = (1 << 1);
-	if (INTEL_INFO(dev)->gen >= 5)
-		intel_encoder->crtc_mask |= (1 << 0);
+	if (HAS_PCH_SPLIT(dev))
+		intel_encoder->crtc_mask = (1 << 0) | (1 << 1) | (1 << 2);
+	else
+		intel_encoder->crtc_mask = (1 << 1);
+
 	drm_encoder_helper_add(encoder, &intel_lvds_helper_funcs);
 	drm_connector_helper_add(connector, &intel_lvds_connector_helper_funcs);
 	connector->display_info.subpixel_order = SubPixelHorizontalRGB;
drivers/gpu/drm/i915/intel_sdvo.c +1 -1
index 30fe554..f6853d6 100644
@@ -2244,7 +2244,7 @@ intel_sdvo_output_setup(struct intel_sdvo *intel_sdvo, uint16_t flags)
 			      bytes[0], bytes[1]);
 		return false;
 	}
-	intel_sdvo->base.crtc_mask = (1 << 0) | (1 << 1);
+	intel_sdvo->base.crtc_mask = (1 << 0) | (1 << 1) | (1 << 2);
 
 	return true;
 }
Signature · 1 line
1.7.4.1
#8 lore raw

PATCH 3/4 drm/i915: split refclk code out of ironlake_crtc_mode_set

In reply to Jesse Barnes

Just a cleanup to make the mode_set function more manageable.

---
 drivers/gpu/drm/i915/intel_display.c |   49 +++++++++++++++++++++++++++-------
 1 files changed, 39 insertions(+), 10 deletions(-)
drivers/gpu/drm/i915/intel_display.c +39 -10
index 0dffb52..058dc05 100644
@@ -5185,6 +5185,44 @@ static void ironlake_update_pch_refclk(struct drm_device *dev)
 	}
 }
 
+static int ironlake_get_refclk(struct drm_crtc *crtc)
+{
+	struct drm_device *dev = crtc->dev;
+	struct drm_i915_private *dev_priv = dev->dev_private;
+	struct intel_encoder *encoder;
+	struct drm_mode_config *mode_config = &dev->mode_config;
+	struct intel_encoder *edp_encoder = NULL;
+	int num_connectors = 0;
+	bool is_lvds = false;
+
+	list_for_each_entry(encoder, &mode_config->encoder_list, base.head) {
+		if (encoder->base.crtc != crtc)
+			continue;
+
+		switch (encoder->type) {
+		case INTEL_OUTPUT_LVDS:
+			is_lvds = true;
+			break;
+		case INTEL_OUTPUT_EDP:
+			edp_encoder = encoder;
+			break;
+		}
+		num_connectors++;
+	}
+
+	if (is_lvds && intel_panel_use_ssc(dev_priv) && num_connectors < 2) {
+		DRM_DEBUG_KMS("using SSC reference clock of %d MHz\n",
+			      dev_priv->lvds_ssc_freq);
+		return dev_priv->lvds_ssc_freq * 1000;
+	}
+
+	/* PCH ports use a 120 MHz refclk */
+	if (!edp_encoder || intel_encoder_is_pch_edp(&edp_encoder->base))
+		return 120000;
+	else
+		return 96000;
+}
+
 static int ironlake_crtc_mode_set(struct drm_crtc *crtc,
 				  struct drm_display_mode *mode,
 				  struct drm_display_mode *adjusted_mode,
@@ -5244,16 +5282,7 @@ static int ironlake_crtc_mode_set(struct drm_crtc *crtc,
 		num_connectors++;
 	}
 
-	if (is_lvds && intel_panel_use_ssc(dev_priv) && num_connectors < 2) {
-		refclk = dev_priv->lvds_ssc_freq * 1000;
-		DRM_DEBUG_KMS("using SSC reference clock of %d MHz\n",
-			      refclk / 1000);
-	} else {
-		refclk = 96000;
-		if (!has_edp_encoder ||
-		    intel_encoder_is_pch_edp(&has_edp_encoder->base))
-			refclk = 120000; /* 120Mhz refclk */
-	}
+	refclk = ironlake_get_refclk(crtc);
 
 	/*
 	 * Returns a set of divisors for the desired target clock with the given
Signature · 1 line
1.7.4.1
#9 lore raw

Re: [PATCH 3/4] drm/i915: split refclk code out of ironlake_crtc_mode_set

drivers/gpu/drm/i915/intel_display.c static void ironlake_update_pch_refclk(struct drm_device *dev) · line 5224

Quoted 6 lines · Jesse Barnes
drivers/gpu/drm/i915/intel_display.c +6 -0
+	/* PCH ports use a 120 MHz refclk */
+	if (!edp_encoder || intel_encoder_is_pch_edp(&edp_encoder->base))
+		return 120000;
+	else
+		return 96000;
+}

I can't find any documentation that says that there is any 96MHz refclk on PCH hardware.

Signature · 1 line
[email protected]

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

#10 lore raw

Re: [PATCH 3/4] drm/i915: split refclk code out of ironlake_crtc_mode_set

Quoted 11 lines · On Wed, 05 Oct 2011 11:09:57 -0700 Keith Packard <[email protected]> wrote:

drivers/gpu/drm/i915/intel_display.c static void ironlake_update_pch_refclk(struct drm_device *dev) · line 5224

Quoted 6 lines · Jesse Barnes
drivers/gpu/drm/i915/intel_display.c +6 -0
+	/* PCH ports use a 120 MHz refclk */
+	if (!edp_encoder || intel_encoder_is_pch_edp(&edp_encoder->base))
+		return 120000;
+	else
+		return 96000;
+}

I can't find any documentation that says that there is any 96MHz refclk on PCH hardware.

Was just preserving the old code; I think you've fixed this issue already in another patch (looks like it's another one that's been there since we first brought up ILK?).

Signature · 1 line
Jesse Barnes, Intel Open Source Technology Center

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

#11 lore raw

Re: [PATCH 3/4] drm/i915: split refclk code out of ironlake_crtc_mode_set

Quoted 3 lines · Jesse Barnes

Was just preserving the old code; I think you've fixed this issue already in another patch (looks like it's another one that's been there since we first brought up ILK?).

I think the only case where we select the '96MHz' clock is the same case where we hard-code the actually clock for DP.

I was just wondering if you had special info that I couldn't find.

Signature · 1 line
[email protected]

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

#12 lore raw

PATCH 4/4 drm/i915: more 3 pipe support

In reply to Jesse Barnes

Handle PLL allocation and transcoder select bits on CPT+.

---
 drivers/gpu/drm/i915/intel_crt.c     |   13 ++---
 drivers/gpu/drm/i915/intel_display.c |   93 +++++++++++++++++++++------------
 drivers/gpu/drm/i915/intel_drv.h     |    3 +
 drivers/gpu/drm/i915/intel_hdmi.c    |   10 ++--
 4 files changed, 70 insertions(+), 49 deletions(-)
drivers/gpu/drm/i915/intel_crt.c +4 -9
index 0979d88..da769e6 100644
@@ -152,17 +152,12 @@ static void intel_crt_mode_set(struct drm_encoder *encoder,
 	if (adjusted_mode->flags & DRM_MODE_FLAG_PVSYNC)
 		adpa |= ADPA_VSYNC_ACTIVE_HIGH;
 
-	if (intel_crtc->pipe == 0) {
-		if (HAS_PCH_CPT(dev))
-			adpa |= PORT_TRANS_A_SEL_CPT;
-		else
+	if (HAS_PCH_CPT(dev))
+		adpa |= PORT_TRANS_SEL_CPT(intel_crtc->pipe);
+	else if (intel_crtc->pipe == 0)
 			adpa |= ADPA_PIPE_A_SELECT;
-	} else {
-		if (HAS_PCH_CPT(dev))
-			adpa |= PORT_TRANS_B_SEL_CPT;
-		else
+	else
 			adpa |= ADPA_PIPE_B_SELECT;
-	}
 
 	if (!HAS_PCH_SPLIT(dev))
 		I915_WRITE(BCLRPAT(intel_crtc->pipe), 0);
drivers/gpu/drm/i915/intel_display.c +59 -34

186 lines not on this page · raw message

drivers/gpu/drm/i915/intel_drv.h +3 -0
index 0b2ee9d..4b5f543 100644
@@ -171,6 +171,9 @@ struct intel_crtc {
 	int16_t cursor_width, cursor_height;
 	bool cursor_visible;
 	unsigned int bpp;
+
+	bool no_pll; /* tertiary pipe for IVB */
+	bool use_pll_a;
 };
 
 #define to_intel_crtc(x) container_of(x, struct intel_crtc, base)
drivers/gpu/drm/i915/intel_hdmi.c +4 -6
index 11883c9..ac621dd 100644
@@ -249,12 +249,10 @@ static void intel_hdmi_mode_set(struct drm_encoder *encoder,
 		sdvox |= SDVO_NULL_PACKETS_DURING_VSYNC;
 	}
 
-	if (intel_crtc->pipe == 1) {
-		if (HAS_PCH_CPT(dev))
-			sdvox |= PORT_TRANS_B_SEL_CPT;
-		else
-			sdvox |= SDVO_PIPE_B_SELECT;
-	}
+	if (HAS_PCH_CPT(dev))
+		sdvox |= PORT_TRANS_SEL_CPT(intel_crtc->pipe);
+	else if (intel_crtc->pipe == 1)
+		sdvox |= SDVO_PIPE_B_SELECT;
 
 	I915_WRITE(intel_hdmi->sdvox_reg, sdvox);
 	POSTING_READ(intel_hdmi->sdvox_reg);
Signature · 1 line
1.7.4.1
#13 lore raw

Re: [PATCH 4/4] drm/i915: more 3 pipe support

Quoted 1 line · On Wed, 5 Oct 2011 10:25:21 -0700 Jesse Barnes <[email protected]>…

Handle PLL allocation and transcoder select bits on CPT+.

Updated patch that fixes the bug with VGA plus two HDMI. Was missing the composite sync bits in the FDI TX side (setting them everywhere here just to be safe).

Signature · 17 lines
Jesse Barnes, Intel Open Source Technology Center

From 3efb84701437f2d1eae3fb68c1259c2bb5a0c518 Mon Sep 17 00:00:00 2001
From: Jesse Barnes <[email protected]>
Date: Fri, 2 Sep 2011 15:28:39 -0700
Subject: [PATCH] drm/i915: more 3 pipe support

Handle PLL allocation and transcoder select bits on CPT+.

Signed-off-by: Jesse Barnes <[email protected]>
---
 drivers/gpu/drm/i915/i915_reg.h      |    1 +
 drivers/gpu/drm/i915/intel_crt.c     |   13 +---
 drivers/gpu/drm/i915/intel_display.c |  101 +++++++++++++++++++++-------------
 drivers/gpu/drm/i915/intel_drv.h     |    3 +
 drivers/gpu/drm/i915/intel_hdmi.c    |   10 +--
 5 files changed, 74 insertions(+), 54 deletions(-)
drivers/gpu/drm/i915/i915_reg.h +1 -0
index 7659075..b31f253 100644
@@ -3167,6 +3167,7 @@
 #define  FDI_LINK_TRAIN_NONE_IVB            (3<<8)
 
 /* both Tx and Rx */
+#define  FDI_COMPOSITE_SYNC		(1<<11)
 #define  FDI_LINK_TRAIN_AUTO		(1<<10)
 #define  FDI_SCRAMBLING_ENABLE          (0<<7)
 #define  FDI_SCRAMBLING_DISABLE         (1<<7)
drivers/gpu/drm/i915/intel_crt.c +4 -9
index 0979d88..da769e6 100644
@@ -152,17 +152,12 @@ static void intel_crt_mode_set(struct drm_encoder *encoder,
 	if (adjusted_mode->flags & DRM_MODE_FLAG_PVSYNC)
 		adpa |= ADPA_VSYNC_ACTIVE_HIGH;
 
-	if (intel_crtc->pipe == 0) {
-		if (HAS_PCH_CPT(dev))
-			adpa |= PORT_TRANS_A_SEL_CPT;
-		else
+	if (HAS_PCH_CPT(dev))
+		adpa |= PORT_TRANS_SEL_CPT(intel_crtc->pipe);
+	else if (intel_crtc->pipe == 0)
 			adpa |= ADPA_PIPE_A_SELECT;
-	} else {
-		if (HAS_PCH_CPT(dev))
-			adpa |= PORT_TRANS_B_SEL_CPT;
-		else
+	else
 			adpa |= ADPA_PIPE_B_SELECT;
-	}
 
 	if (!HAS_PCH_SPLIT(dev))
 		I915_WRITE(BCLRPAT(intel_crtc->pipe), 0);
drivers/gpu/drm/i915/intel_display.c +62 -39

212 lines not on this page · raw message

drivers/gpu/drm/i915/intel_drv.h +3 -0
index 0b2ee9d..4b5f543 100644
@@ -171,6 +171,9 @@ struct intel_crtc {
 	int16_t cursor_width, cursor_height;
 	bool cursor_visible;
 	unsigned int bpp;
+
+	bool no_pll; /* tertiary pipe for IVB */
+	bool use_pll_a;
 };
 
 #define to_intel_crtc(x) container_of(x, struct intel_crtc, base)
drivers/gpu/drm/i915/intel_hdmi.c +4 -6
index 11883c9..ac621dd 100644
@@ -249,12 +249,10 @@ static void intel_hdmi_mode_set(struct drm_encoder *encoder,
 		sdvox |= SDVO_NULL_PACKETS_DURING_VSYNC;
 	}
 
-	if (intel_crtc->pipe == 1) {
-		if (HAS_PCH_CPT(dev))
-			sdvox |= PORT_TRANS_B_SEL_CPT;
-		else
-			sdvox |= SDVO_PIPE_B_SELECT;
-	}
+	if (HAS_PCH_CPT(dev))
+		sdvox |= PORT_TRANS_SEL_CPT(intel_crtc->pipe);
+	else if (intel_crtc->pipe == 1)
+		sdvox |= SDVO_PIPE_B_SELECT;
 
 	I915_WRITE(intel_hdmi->sdvox_reg, sdvox);
 	POSTING_READ(intel_hdmi->sdvox_reg);
Signature · 1 line
1.7.4.1

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

#14 lore raw

Re: [PATCH 4/4] drm/i915: more 3 pipe support

Quoted 12 lines · Jesse Barnes
Quoted 1 line · On Wed, 5 Oct 2011 10:25:21 -0700 Jesse Barnes <[email protected]>…

Handle PLL allocation and transcoder select bits on CPT+.

Updated patch that fixes the bug with VGA plus two HDMI. Was missing the composite sync bits in the FDI TX side (setting them everywhere here just to be safe).

--
Jesse Barnes, Intel Open Source Technology Center

The following chunk which we discussed on IRC earlier today made it work correctly on IVB:

Quoted 1 line

From 8a5e2eb86f62ab14ca0c12b16628e009d1fcbe98 Mon Sep 17 00:00:00 2001

From: Eugeni Dodonov <[email protected]>
Date: Wed, 5 Oct 2011 18:34:09 -0300
Subject: [PATCH] Enable composite sync bit on IVB

---
 drivers/gpu/drm/i915/intel_display.c |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)
(quoted diff) +0 -0

b/drivers/gpu/drm/i915/intel_display.c
index e63a187..26cadff 100644

drivers/gpu/drm/i915/intel_display.c +0 -0
@@ -2608,6 +2608,7 @@ static void ivb_manual_fdi_link_train(struct drm_crtc

*crtc)

     temp |= FDI_LINK_TRAIN_PATTERN_1_IVB;
     temp &= ~FDI_LINK_TRAIN_VOL_EMP_MASK;
     temp |= FDI_LINK_TRAIN_400MV_0DB_SNB_B;
+    temp |= FDI_COMPOSITE_SYNC;
     I915_WRITE(reg, temp | FDI_TX_ENABLE);

     reg = FDI_RX_CTL(pipe);
@@ -2615,6 +2616,7 @@ static void ivb_manual_fdi_link_train(struct drm_crtc
*crtc)
     temp &= ~FDI_LINK_TRAIN_AUTO;
     temp &= ~FDI_LINK_TRAIN_PATTERN_MASK_CPT;
     temp |= FDI_LINK_TRAIN_PATTERN_1_CPT;
+    temp |= FDI_COMPOSITE_SYNC;
     I915_WRITE(reg, temp | FDI_RX_ENABLE);

     POSTING_READ(reg);
Signature · 5 lines
1.7.6.4

-- 
Eugeni Dodonov
<http://eugeni.dodonov.net/>

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

#15 lore raw
In reply to Jesse Barnes
Quoted 7 lines · On Wed, 5 Oct 2011 10:25:17 -0700 Jesse Barnes <[email protected]>…

This set of patches lays the groundwork for supporting 3 pipes on IVB hardware. We've only tested it so far with 2 HDMI plus 1 VGA (a configuration that doesn't work), and are awaiting 3 DP port hardware (the officially supported config) to complete our testing.

So this patchset is necessary but may not be sufficient (once we have 3 DP hardware we can debug and see). Review and comments appreciated.

Gordon, I think this set (with the updated 4/4 patch) is ready for testing. You'll need a configuration that uses the same monitor on two identical ports to get all the pipes running. I'm also interested in the failure cases though. There are some limitations:
1) two heads need to use the same timings and ports
2) FDI B and C share lanes, so are limited to a max of two lanes each

For (1), I'm currently using a VGA attached monitor, an HDMI attached monitor, and another HDMI monitor attached to a DP->HDMI dongle on my SDV (that allows me to drive the port with the HDMI encoder rather than the DP encoder).

Unfortunately, (2) complicates our mode list output. If you query for available modes, you'll definitely see some that you can't drive with 3 pipes enabled. I'm not sure if the best way to handle that...

Signature · 1 line
Jesse Barnes, Intel Open Source Technology Center

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

#16 lore raw
Quoted 3 lines · Jesse Barnes

Unfortunately, (2) complicates our mode list output. If you query for available modes, you'll definitely see some that you can't drive with 3 pipes enabled. I'm not sure if the best way to handle that...

All we can do is say 'no' when someone tries to select that configuration. We've never figured out a better way to list valid settings.

The proposed RandR 1.4 protocol has a 'set everything all at once' API which allows you to say 'no' before even starting the mode setting process, which is kinda nice. We just need to make sure sure this can be mirrored through KMS to the hardware.

Signature · 1 line
[email protected]

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

maintainer
#17 lore raw
Quoted 6 lines · Keith Packard
Quoted 3 lines · Jesse Barnes

Unfortunately, (2) complicates our mode list output.  If you query for available modes, you'll definitely see some that you can't drive with 3 pipes enabled.  I'm not sure if the best way to handle that...

Quoted 8 lines

All we can do is say 'no' when someone tries to select that configuration. We've never figured out a better way to list valid settings.

The proposed RandR 1.4 protocol has a 'set everything all at once' API which allows you to say 'no' before even starting the mode setting process, which is kinda nice. We just need to make sure sure this can be mirrored through KMS to the hardware.

Fine for dynamic stuff, how do you pick a correct initial mode?

You can't say no there, you need to make a decision from the information provided.

Dave.

#18 lore raw
Quoted 16 lines · On Thu, 6 Oct 2011 07:31:44 +0100 Dave Airlie <[email protected]> wrote:
Quoted 6 lines · Keith Packard
Quoted 3 lines · Jesse Barnes

Unfortunately, (2) complicates our mode list output.  If you query for available modes, you'll definitely see some that you can't drive with 3 pipes enabled.  I'm not sure if the best way to handle that...

Quoted 8 lines

All we can do is say 'no' when someone tries to select that configuration. We've never figured out a better way to list valid settings.

The proposed RandR 1.4 protocol has a 'set everything all at once' API which allows you to say 'no' before even starting the mode setting process, which is kinda nice. We just need to make sure sure this can be mirrored through KMS to the hardware.

Quoted 4 lines

Fine for dynamic stuff, how do you pick a correct initial mode?

You can't say no there, you need to make a decision from the information provided.

Yeah that's a good point. There's something weird going on in general with X's default config on my system at least. All 3 heads come up at 1280x1024, but the two identical HDMI heads come up with different refresh rates for some reason (one uses the preferred mode and the other tries to get 75Hz).

Of course I'd prefer an extended config as the default, which might make it easier to choose all the preferred modes, but that doesn't solve the problem of having say two 1920x1200 monitors attached along with a 2560x1600, which won't work on IVB...

Should we add a new mode flag to indicate which modes are conditional on config and which modes can always be supported? That way X and other tools could get all the heads lit up by default at least...

Signature · 1 line
Jesse Barnes, Intel Open Source Technology Center

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

#19 lore raw
Quoted 29 lines · Jesse Barnes
Quoted 16 lines · On Thu, 6 Oct 2011 07:31:44 +0100 Dave Airlie <[email protected]> wrote:
Quoted 6 lines · Keith Packard
Quoted 3 lines · Jesse Barnes

Unfortunately, (2) complicates our mode list output.  If you query for available modes, you'll definitely see some that you can't drive with 3 pipes enabled.  I'm not sure if the best way to handle that...

Quoted 8 lines

All we can do is say 'no' when someone tries to select that configuration. We've never figured out a better way to list valid settings.

The proposed RandR 1.4 protocol has a 'set everything all at once' API which allows you to say 'no' before even starting the mode setting process, which is kinda nice. We just need to make sure sure this can be mirrored through KMS to the hardware.

Quoted 4 lines

Fine for dynamic stuff, how do you pick a correct initial mode?

You can't say no there, you need to make a decision from the information provided.

Yeah that's a good point. There's something weird going on in general with X's default config on my system at least. All 3 heads come up at 1280x1024, but the two identical HDMI heads come up with different refresh rates for some reason (one uses the preferred mode and the other tries to get 75Hz).

Quoted 8 lines

Of course I'd prefer an extended config as the default, which might make it easier to choose all the preferred modes, but that doesn't solve the problem of having say two 1920x1200 monitors attached along with a 2560x1600, which won't work on IVB...

Should we add a new mode flag to indicate which modes are conditional on config and which modes can always be supported? That way X and other tools could get all the heads lit up by default at least...

For the kernel console we have the identical "light up as much as possible in the hope the user can see something" problem. So I think the only thing to do is to expose the kernel console mode somewhere (in case a running X session has totally mangled the modeset config). This way we need to solve the problem only once and the kernel has all the information to make the best out of it.

For everything else userspace just has to remember the users preferred modeset layout and restore it on boot-up. I should be safe to assume that a given mode that once worked should continue to do so, and if it fails userspace can fall back to the kernel console configuration.

-Daniel

Signature · 3 lines
Daniel Vetter
Mail: [email protected]
Mobile: +41 (0)79 365 57 48
#20 lore raw
In reply to Dave Airlie
Quoted 2 lines · Dave Airlie

You can't say no there, you need to make a decision from the information provided.

Yeah, you'd end up having to use two modes with the same clock. Let's hope DisplayPort ends up a lot more popular than it has gotten so far?

Signature · 1 line
[email protected]

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

Reviewed-by Tested-by
#21 lore raw
In reply to Jesse Barnes
Quoted 18 lines · Jesse Barnes
Quoted 7 lines · On Wed, 5 Oct 2011 10:25:17 -0700 Jesse Barnes <[email protected]>…

This set of patches lays the groundwork for supporting 3 pipes on IVB hardware. We've only tested it so far with 2 HDMI plus 1 VGA (a configuration that doesn't work), and are awaiting 3 DP port hardware (the officially supported config) to complete our testing.

So this patchset is necessary but may not be sufficient (once we have 3 DP hardware we can debug and see). Review and comments appreciated.

Gordon, I think this set (with the updated 4/4 patch) is ready for testing. You'll need a configuration that uses the same monitor on two identical ports to get all the pipes running. I'm also interested in the failure cases though. There are some limitations:
1) two heads need to use the same timings and ports
2) FDI B and C share lanes, so are limited to a max of two lanes each

Quoted 9 lines

For (1), I'm currently using a VGA attached monitor, an HDMI attached monitor, and another HDMI monitor attached to a DP->HDMI dongle on my SDV (that allows me to drive the port with the HDMI encoder rather than the DP encoder).

Unfortunately, (2) complicates our mode list output. If you query for available modes, you'll definitely see some that you can't drive with 3 pipes enabled. I'm not sure if the best way to handle that...

It works - with a small change to the FDI_COMPOSITE_SYNC which we discussed on irc with Jesse - I'll send it on the corresponding patch.

With that change, for the series:

Signature · 2 lines
Eugeni Dodonov
<http://eugeni.dodonov.net/>

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