Skip to contentSkip to conversation

drm/vc4: Modesetting fixes

PATCH New Eric Anholt 1 participant 7 messages Feb 16, 2016

Subsystems (3)
lore
Share & tools
lore
#1 lore raw

PATCH 1/6 drm/vc4: Fix a framebuffer reference leak on async flip interrupt.

We'd need X to queue up an async pageflip while another is outstanding, and then take a SIGIO. I think X actually avoids sending out the next pageflip while one's already queued, but I'm not sure.

---
 drivers/gpu/drm/vc4/vc4_crtc.c | 1 +
 1 file changed, 1 insertion(+)
drivers/gpu/drm/vc4/vc4_crtc.c +1 -0
index 68227cc..bb74cb9 100644
@@ -527,6 +527,7 @@ static int vc4_async_page_flip(struct drm_crtc *crtc,
 	/* Make sure all other async modesetes have landed. */
 	ret = down_interruptible(&vc4->async_modeset);
 	if (ret) {
+		drm_framebuffer_unreference(fb);
 		kfree(flip_state);
 		return ret;
 	}
Signature · 6 lines
2.7.0

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
#2 lore raw

PATCH 2/6 drm/vc4: Bring HDMI up from power off if necessary.

In reply to Eric Anholt

If the firmware hadn't brought up HDMI for us, we need to do its power-on reset sequence (reset HD and and clear its STANDBY bits, reset HDMI, and leave the PHY disabled).

---
 drivers/gpu/drm/vc4/vc4_hdmi.c | 29 ++++++++++++++++++++++++++++-
 drivers/gpu/drm/vc4/vc4_regs.h |  2 ++
 2 files changed, 30 insertions(+), 1 deletion(-)
drivers/gpu/drm/vc4/vc4_hdmi.c +28 -1
index c49cb44..d189906 100644
@@ -497,6 +497,16 @@ static int vc4_hdmi_bind(struct device *dev, struct device *master, void *data)
 		goto err_put_i2c;
 	}
 
+	/* This is the rate that is set by the firmware.  The number
+	 * needs to be a bit higher than the pixel clock rate
+	 * (generally 148.5Mhz).
+	 */
+	ret = clk_set_rate(hdmi->hsm_clock, 163682864);
+	if (ret) {
+		DRM_ERROR("Failed to set HSM clock rate: %d\n", ret);
+		goto err_unprepare_pix;
+	}
+
 	ret = clk_prepare_enable(hdmi->hsm_clock);
 	if (ret) {
 		DRM_ERROR("Failed to turn on HDMI state machine clock: %d\n",
@@ -518,7 +528,24 @@ static int vc4_hdmi_bind(struct device *dev, struct device *master, void *data)
 	vc4->hdmi = hdmi;
 
 	/* HDMI core must be enabled. */
-	WARN_ON_ONCE((HD_READ(VC4_HD_M_CTL) & VC4_HD_M_ENABLE) == 0);
+	if (!(HD_READ(VC4_HD_M_CTL) & VC4_HD_M_ENABLE)) {
+		HD_WRITE(VC4_HD_M_CTL, VC4_HD_M_SW_RST);
+		udelay(1);
+		HD_WRITE(VC4_HD_M_CTL, 0);
+
+		HD_WRITE(VC4_HD_M_CTL, VC4_HD_M_ENABLE);
+
+		HDMI_WRITE(VC4_HDMI_SW_RESET_CONTROL,
+			   VC4_HDMI_SW_RESET_HDMI |
+			   VC4_HDMI_SW_RESET_FORMAT_DETECT);
+
+		HDMI_WRITE(VC4_HDMI_SW_RESET_CONTROL, 0);
+
+		/* PHY should be in reset, like
+		 * vc4_hdmi_encoder_disable() does.
+		 */
+		HDMI_WRITE(VC4_HDMI_TX_PHY_RESET_CTL, 0xf << 16);
+	}
 
 	drm_encoder_init(drm, hdmi->encoder, &vc4_hdmi_encoder_funcs,
 			 DRM_MODE_ENCODER_TMDS, NULL);
drivers/gpu/drm/vc4/vc4_regs.h +2 -0
index 25df20e..31042a4 100644
@@ -456,6 +456,8 @@
 #define VC4_HDMI_TX_PHY_RESET_CTL		0x2c0
 
 #define VC4_HD_M_CTL				0x00c
+# define VC4_HD_M_REGISTER_FILE_STANDBY		(3 << 6)
+# define VC4_HD_M_RAM_STANDBY			(3 << 4)
 # define VC4_HD_M_SW_RST			BIT(2)
 # define VC4_HD_M_ENABLE			BIT(0)
 
Signature · 6 lines
2.7.0

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
#3 lore raw

PATCH 3/6 drm/vc4: Add another reg to HDMI debug dumping.

In reply to Eric Anholt

This is also involved in the HDMI setup sequence so it's nice to see it.

---

This patch and the next one I'd target for -next, since they aren't functional fixes.

 drivers/gpu/drm/vc4/vc4_hdmi.c | 1 +
 1 file changed, 1 insertion(+)
drivers/gpu/drm/vc4/vc4_hdmi.c +1 -0
index d189906..b4e7b8a 100644
@@ -95,6 +95,7 @@ static const struct {
 	HDMI_REG(VC4_HDMI_SW_RESET_CONTROL),
 	HDMI_REG(VC4_HDMI_HOTPLUG_INT),
 	HDMI_REG(VC4_HDMI_HOTPLUG),
+	HDMI_REG(VC4_HDMI_RAM_PACKET_CONFIG),
 	HDMI_REG(VC4_HDMI_HORZA),
 	HDMI_REG(VC4_HDMI_HORZB),
 	HDMI_REG(VC4_HDMI_FIFO_CTL),
Signature · 6 lines
2.7.0

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
#4 lore raw

PATCH 4/6 drm/vc4: Fix the name of the VSYNCD_EVEN register.

In reply to Eric Anholt

It's used for delaying vsync in interlaced mode.

---
 drivers/gpu/drm/vc4/vc4_crtc.c | 2 +-
 drivers/gpu/drm/vc4/vc4_regs.h | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)
drivers/gpu/drm/vc4/vc4_crtc.c +1 -1
index bb74cb9..5daa824 100644
@@ -88,7 +88,7 @@ static const struct {
 } crtc_regs[] = {
 	CRTC_REG(PV_CONTROL),
 	CRTC_REG(PV_V_CONTROL),
-	CRTC_REG(PV_VSYNCD),
+	CRTC_REG(PV_VSYNCD_EVEN),
 	CRTC_REG(PV_HORZA),
 	CRTC_REG(PV_HORZB),
 	CRTC_REG(PV_VERTA),
drivers/gpu/drm/vc4/vc4_regs.h +1 -1
index 31042a4..58d4cb3 100644
@@ -187,7 +187,7 @@
 # define PV_VCONTROL_CONTINUOUS			BIT(1)
 # define PV_VCONTROL_VIDEN			BIT(0)
 
-#define PV_VSYNCD				0x08
+#define PV_VSYNCD_EVEN				0x08
 
 #define PV_HORZA				0x0c
 # define PV_HORZA_HBP_MASK			VC4_MASK(31, 16)
Signature · 6 lines
2.7.0

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
#5 lore raw

PATCH 5/6 drm/vc4: Fix setting of vertical timings in the CRTC.

In reply to Eric Anholt

It looks like when I went to add the interlaced bits, I just took the existing PV_VERT* block and indented it, instead of copy and pasting it first. Without this, changing resolution never worked.

---
 drivers/gpu/drm/vc4/vc4_crtc.c | 10 ++++++++++
 1 file changed, 10 insertions(+)
drivers/gpu/drm/vc4/vc4_crtc.c +10 -0
index 5daa824..89562904 100644
@@ -217,6 +217,16 @@ static void vc4_crtc_mode_set_nofb(struct drm_crtc *crtc)
 				 PV_HORZB_HFP) |
 		   VC4_SET_FIELD(mode->hdisplay, PV_HORZB_HACTIVE));
 
+	CRTC_WRITE(PV_VERTA,
+		   VC4_SET_FIELD(mode->vtotal - mode->vsync_end,
+				 PV_VERTA_VBP) |
+		   VC4_SET_FIELD(mode->vsync_end - mode->vsync_start,
+				 PV_VERTA_VSYNC));
+	CRTC_WRITE(PV_VERTB,
+		   VC4_SET_FIELD(mode->vsync_start - mode->vdisplay,
+				 PV_VERTB_VFP) |
+		   VC4_SET_FIELD(vactive, PV_VERTB_VACTIVE));
+
 	if (interlace) {
 		CRTC_WRITE(PV_VERTA_EVEN,
 			   VC4_SET_FIELD(mode->vtotal - mode->vsync_end - 1,
Signature · 6 lines
2.7.0

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
#6 lore raw

PATCH 6/6 drm/vc4: Initialize scaler DISPBKGND on modeset.

In reply to Eric Anholt

We weren't updating the interlaced bit, so we'd scan out incorrectly if the firmware had brought up the TV encoder and we were switching to HDMI.

---
 drivers/gpu/drm/vc4/vc4_crtc.c |  6 ++++++
 drivers/gpu/drm/vc4/vc4_regs.h | 14 ++++++++++++++
 2 files changed, 20 insertions(+)
drivers/gpu/drm/vc4/vc4_crtc.c +6 -0
index 89562904..66499e3 100644
@@ -188,6 +188,8 @@ static int vc4_get_clock_select(struct drm_crtc *crtc)
 
 static void vc4_crtc_mode_set_nofb(struct drm_crtc *crtc)
 {
+	struct drm_device *dev = crtc->dev;
+	struct vc4_dev *vc4 = to_vc4_dev(dev);
 	struct vc4_crtc *vc4_crtc = to_vc4_crtc(crtc);
 	struct drm_crtc_state *state = crtc->state;
 	struct drm_display_mode *mode = &state->adjusted_mode;
@@ -256,6 +258,10 @@ static void vc4_crtc_mode_set_nofb(struct drm_crtc *crtc)
 		   PV_CONTROL_FIFO_CLR |
 		   PV_CONTROL_EN);
 
+	HVS_WRITE(SCALER_DISPBKGNDX(vc4_crtc->channel),
+		  SCALER_DISPBKGND_AUTOHS |
+		  (interlace ? SCALER_DISPBKGND_INTERLACE : 0));
+
 	if (debug_dump_regs) {
 		DRM_INFO("CRTC %d regs after:\n", drm_crtc_index(crtc));
 		vc4_crtc_dump_regs(vc4_crtc);
drivers/gpu/drm/vc4/vc4_regs.h +14 -0
index 58d4cb3..bf42a8e 100644
@@ -350,6 +350,17 @@
 # define SCALER_DISPCTRLX_HEIGHT_SHIFT		0
 
 #define SCALER_DISPBKGND0                       0x00000044
+# define SCALER_DISPBKGND_AUTOHS		BIT(31)
+# define SCALER_DISPBKGND_INTERLACE		BIT(30)
+# define SCALER_DISPBKGND_GAMMA			BIT(29)
+# define SCALER_DISPBKGND_TESTMODE_MASK		VC4_MASK(28, 25)
+# define SCALER_DISPBKGND_TESTMODE_SHIFT	25
+/* Enables filling the scaler line with the RGB value in the low 24
+ * bits before compositing.  Costs cycles, so should be skipped if
+ * opaque display planes will cover everything.
+ */
+# define SCALER_DISPBKGND_FILL			BIT(24)
+
 #define SCALER_DISPSTAT0                        0x00000048
 #define SCALER_DISPBASE0                        0x0000004c
 # define SCALER_DISPSTATX_MODE_MASK		VC4_MASK(31, 30)
@@ -362,6 +373,9 @@
 # define SCALER_DISPSTATX_EMPTY			BIT(28)
 #define SCALER_DISPCTRL1                        0x00000050
 #define SCALER_DISPBKGND1                       0x00000054
+#define SCALER_DISPBKGNDX(x)			(SCALER_DISPBKGND0 +        \
+						 (x) * (SCALER_DISPBKGND1 - \
+							SCALER_DISPBKGND0))
 #define SCALER_DISPSTAT1                        0x00000058
 #define SCALER_DISPSTATX(x)			(SCALER_DISPSTAT0 +        \
 						 (x) * (SCALER_DISPSTAT1 - \
Signature · 6 lines
2.7.0

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel